diff options
Diffstat (limited to 'drivers/dma')
40 files changed, 27972 insertions, 0 deletions
diff --git a/drivers/dma/Kconfig b/drivers/dma/Kconfig new file mode 100644 index 00000000000..e02d74b1e89 --- /dev/null +++ b/drivers/dma/Kconfig @@ -0,0 +1,167 @@ +# +# DMA engine configuration +# + +menuconfig DMADEVICES + bool "DMA Engine support" + depends on HAS_DMA + help + DMA engines can do asynchronous data transfers without + involving the host CPU. Currently, this framework can be + used to offload memory copies in the network stack and + RAID operations in the MD driver. This menu only presents + DMA Device drivers supported by the configured arch, it may + be empty in some cases. + +if DMADEVICES + +comment "DMA Devices" + +config ASYNC_TX_DISABLE_CHANNEL_SWITCH + bool + +config INTEL_IOATDMA + tristate "Intel I/OAT DMA support" + depends on PCI && X86 + select DMA_ENGINE + select DCA + select ASYNC_TX_DISABLE_CHANNEL_SWITCH + select ASYNC_TX_DISABLE_PQ_VAL_DMA + select ASYNC_TX_DISABLE_XOR_VAL_DMA + help + Enable support for the Intel(R) I/OAT DMA engine present + in recent Intel Xeon chipsets. + + Say Y here if you have such a chipset. + + If unsure, say N. + +config INTEL_IOP_ADMA + tristate "Intel IOP ADMA support" + depends on ARCH_IOP32X || ARCH_IOP33X || ARCH_IOP13XX + select DMA_ENGINE + help + Enable support for the Intel(R) IOP Series RAID engines. + +config DW_DMAC + tristate "Synopsys DesignWare AHB DMA support" + depends on AVR32 + select DMA_ENGINE + default y if CPU_AT32AP7000 + help + Support the Synopsys DesignWare AHB DMA controller. This + can be integrated in chips such as the Atmel AT32ap7000. + +config AT_HDMAC + tristate "Atmel AHB DMA support" + depends on ARCH_AT91SAM9RL || ARCH_AT91SAM9G45 + select DMA_ENGINE + help + Support the Atmel AHB DMA controller. This can be integrated in + chips such as the Atmel AT91SAM9RL. + +config FSL_DMA + tristate "Freescale Elo and Elo Plus DMA support" + depends on FSL_SOC + select DMA_ENGINE + ---help--- + Enable support for the Freescale Elo and Elo Plus DMA controllers. + The Elo is the DMA controller on some 82xx and 83xx parts, and the + Elo Plus is the DMA controller on 85xx and 86xx parts. + +config MV_XOR + bool "Marvell XOR engine support" + depends on PLAT_ORION + select DMA_ENGINE + ---help--- + Enable support for the Marvell XOR engine. + +config MX3_IPU + bool "MX3x Image Processing Unit support" + depends on ARCH_MX3 + select DMA_ENGINE + default y + help + If you plan to use the Image Processing unit in the i.MX3x, say + Y here. If unsure, select Y. + +config MX3_IPU_IRQS + int "Number of dynamically mapped interrupts for IPU" + depends on MX3_IPU + range 2 137 + default 4 + help + Out of 137 interrupt sources on i.MX31 IPU only very few are used. + To avoid bloating the irq_desc[] array we allocate a sufficient + number of IRQ slots and map them dynamically to specific sources. + +config TXX9_DMAC + tristate "Toshiba TXx9 SoC DMA support" + depends on MACH_TX49XX || MACH_TX39XX + select DMA_ENGINE + help + Support the TXx9 SoC internal DMA controller. This can be + integrated in chips such as the Toshiba TX4927/38/39. + +config SH_DMAE + tristate "Renesas SuperH DMAC support" + depends on SUPERH && SH_DMA + depends on !SH_DMA_API + select DMA_ENGINE + help + Enable support for the Renesas SuperH DMA controllers. + +config COH901318 + bool "ST-Ericsson COH901318 DMA support" + select DMA_ENGINE + depends on ARCH_U300 + help + Enable support for ST-Ericsson COH 901 318 DMA. + +config AMCC_PPC440SPE_ADMA + tristate "AMCC PPC440SPe ADMA support" + depends on 440SPe || 440SP + select DMA_ENGINE + select ARCH_HAS_ASYNC_TX_FIND_CHANNEL + help + Enable support for the AMCC PPC440SPe RAID engines. + +config ARCH_HAS_ASYNC_TX_FIND_CHANNEL + bool + +config DMA_ENGINE + bool + +comment "DMA Clients" + depends on DMA_ENGINE + +config NET_DMA + bool "Network: TCP receive copy offload" + depends on DMA_ENGINE && NET + default (INTEL_IOATDMA || FSL_DMA) + help + This enables the use of DMA engines in the network stack to + offload receive copy-to-user operations, freeing CPU cycles. + + Say Y here if you enabled INTEL_IOATDMA or FSL_DMA, otherwise + say N. + +config ASYNC_TX_DMA + bool "Async_tx: Offload support for the async_tx api" + depends on DMA_ENGINE + help + This allows the async_tx api to take advantage of offload engines for + memcpy, memset, xor, and raid6 p+q operations. If your platform has + a dma engine that can perform raid operations and you have enabled + MD_RAID456 say Y. + + If unsure, say N. + +config DMATEST + tristate "DMA Test client" + depends on DMA_ENGINE + help + Simple DMA test client. Say N unless you're debugging a + DMA Device driver. + +endif diff --git a/drivers/dma/Makefile b/drivers/dma/Makefile new file mode 100644 index 00000000000..807053d4823 --- /dev/null +++ b/drivers/dma/Makefile @@ -0,0 +1,14 @@ +obj-$(CONFIG_DMA_ENGINE) += dmaengine.o +obj-$(CONFIG_NET_DMA) += iovlock.o +obj-$(CONFIG_DMATEST) += dmatest.o +obj-$(CONFIG_INTEL_IOATDMA) += ioat/ +obj-$(CONFIG_INTEL_IOP_ADMA) += iop-adma.o +obj-$(CONFIG_FSL_DMA) += fsldma.o +obj-$(CONFIG_MV_XOR) += mv_xor.o +obj-$(CONFIG_DW_DMAC) += dw_dmac.o +obj-$(CONFIG_AT_HDMAC) += at_hdmac.o +obj-$(CONFIG_MX3_IPU) += ipu/ +obj-$(CONFIG_TXX9_DMAC) += txx9dmac.o +obj-$(CONFIG_SH_DMAE) += shdma.o +obj-$(CONFIG_COH901318) += coh901318.o coh901318_lli.o +obj-$(CONFIG_AMCC_PPC440SPE_ADMA) += ppc4xx/ diff --git a/drivers/dma/at_hdmac.c b/drivers/dma/at_hdmac.c new file mode 100644 index 00000000000..efc1a61ca23 --- /dev/null +++ b/drivers/dma/at_hdmac.c @@ -0,0 +1,1220 @@ +/* + * Driver for the Atmel AHB DMA Controller (aka HDMA or DMAC on AT91 systems) + * + * Copyright (C) 2008 Atmel Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * + * This supports the Atmel AHB DMA Controller, + * + * The driver has currently been tested with the Atmel AT91SAM9RL + * and AT91SAM9G45 series. + */ + +#include <linux/clk.h> +#include <linux/dmaengine.h> +#include <linux/dma-mapping.h> +#include <linux/dmapool.h> +#include <linux/interrupt.h> +#include <linux/module.h> +#include <linux/platform_device.h> + +#include "at_hdmac_regs.h" + +/* + * Glossary + * -------- + * + * at_hdmac : Name of the ATmel AHB DMA Controller + * at_dma_ / atdma : ATmel DMA controller entity related + * atc_ / atchan : ATmel DMA Channel entity related + */ + +#define ATC_DEFAULT_CFG (ATC_FIFOCFG_HALFFIFO) +#define ATC_DEFAULT_CTRLA (0) +#define ATC_DEFAULT_CTRLB (ATC_SIF(0) \ + |ATC_DIF(1)) + +/* + * Initial number of descriptors to allocate for each channel. This could + * be increased during dma usage. + */ +static unsigned int init_nr_desc_per_channel = 64; +module_param(init_nr_desc_per_channel, uint, 0644); +MODULE_PARM_DESC(init_nr_desc_per_channel, + "initial descriptors per channel (default: 64)"); + + +/* prototypes */ +static dma_cookie_t atc_tx_submit(struct dma_async_tx_descriptor *tx); + + +/*----------------------------------------------------------------------*/ + +static struct at_desc *atc_first_active(struct at_dma_chan *atchan) +{ + return list_first_entry(&atchan->active_list, + struct at_desc, desc_node); +} + +static struct at_desc *atc_first_queued(struct at_dma_chan *atchan) +{ + return list_first_entry(&atchan->queue, + struct at_desc, desc_node); +} + +/** + * atc_alloc_descriptor - allocate and return an initilized descriptor + * @chan: the channel to allocate descriptors for + * @gfp_flags: GFP allocation flags + * + * Note: The ack-bit is positioned in the descriptor flag at creation time + * to make initial allocation more convenient. This bit will be cleared + * and control will be given to client at usage time (during + * preparation functions). + */ +static struct at_desc *atc_alloc_descriptor(struct dma_chan *chan, + gfp_t gfp_flags) +{ + struct at_desc *desc = NULL; + struct at_dma *atdma = to_at_dma(chan->device); + dma_addr_t phys; + + desc = dma_pool_alloc(atdma->dma_desc_pool, gfp_flags, &phys); + if (desc) { + memset(desc, 0, sizeof(struct at_desc)); + INIT_LIST_HEAD(&desc->tx_list); + dma_async_tx_descriptor_init(&desc->txd, chan); + /* txd.flags will be overwritten in prep functions */ + desc->txd.flags = DMA_CTRL_ACK; + desc->txd.tx_submit = atc_tx_submit; + desc->txd.phys = phys; + } + + return desc; +} + +/** + * atc_desc_get - get an unused descriptor from free_list + * @atchan: channel we want a new descriptor for + */ +static struct at_desc *atc_desc_get(struct at_dma_chan *atchan) +{ + struct at_desc *desc, *_desc; + struct at_desc *ret = NULL; + unsigned int i = 0; + LIST_HEAD(tmp_list); + + spin_lock_bh(&atchan->lock); + list_for_each_entry_safe(desc, _desc, &atchan->free_list, desc_node) { + i++; + if (async_tx_test_ack(&desc->txd)) { + list_del(&desc->desc_node); + ret = desc; + break; + } + dev_dbg(chan2dev(&atchan->chan_common), + "desc %p not ACKed\n", desc); + } + spin_unlock_bh(&atchan->lock); + dev_vdbg(chan2dev(&atchan->chan_common), + "scanned %u descriptors on freelist\n", i); + + /* no more descriptor available in initial pool: create one more */ + if (!ret) { + ret = atc_alloc_descriptor(&atchan->chan_common, GFP_ATOMIC); + if (ret) { + spin_lock_bh(&atchan->lock); + atchan->descs_allocated++; + spin_unlock_bh(&atchan->lock); + } else { + dev_err(chan2dev(&atchan->chan_common), + "not enough descriptors available\n"); + } + } + + return ret; +} + +/** + * atc_desc_put - move a descriptor, including any children, to the free list + * @atchan: channel we work on + * @desc: descriptor, at the head of a chain, to move to free list + */ +static void atc_desc_put(struct at_dma_chan *atchan, struct at_desc *desc) +{ + if (desc) { + struct at_desc *child; + + spin_lock_bh(&atchan->lock); + list_for_each_entry(child, &desc->tx_list, desc_node) + dev_vdbg(chan2dev(&atchan->chan_common), + "moving child desc %p to freelist\n", + child); + list_splice_init(&desc->tx_list, &atchan->free_list); + dev_vdbg(chan2dev(&atchan->chan_common), + "moving desc %p to freelist\n", desc); + list_add(&desc->desc_node, &atchan->free_list); + spin_unlock_bh(&atchan->lock); + } +} + +/** + * atc_assign_cookie - compute and assign new cookie + * @atchan: channel we work on + * @desc: descriptor to asign cookie for + * + * Called with atchan->lock held and bh disabled + */ +static dma_cookie_t +atc_assign_cookie(struct at_dma_chan *atchan, struct at_desc *desc) +{ + dma_cookie_t cookie = atchan->chan_common.cookie; + + if (++cookie < 0) + cookie = 1; + + atchan->chan_common.cookie = cookie; + desc->txd.cookie = cookie; + + return cookie; +} + +/** + * atc_dostart - starts the DMA engine for real + * @atchan: the channel we want to start + * @first: first descriptor in the list we want to begin with + * + * Called with atchan->lock held and bh disabled + */ +static void atc_dostart(struct at_dma_chan *atchan, struct at_desc *first) +{ + struct at_dma *atdma = to_at_dma(atchan->chan_common.device); + + /* ASSERT: channel is idle */ + if (atc_chan_is_enabled(atchan)) { + dev_err(chan2dev(&atchan->chan_common), + "BUG: Attempted to start non-idle channel\n"); + dev_err(chan2dev(&atchan->chan_common), + " channel: s0x%x d0x%x ctrl0x%x:0x%x l0x%x\n", + channel_readl(atchan, SADDR), + channel_readl(atchan, DADDR), + channel_readl(atchan, CTRLA), + channel_readl(atchan, CTRLB), + channel_readl(atchan, DSCR)); + + /* The tasklet will hopefully advance the queue... */ + return; + } + + vdbg_dump_regs(atchan); + + /* clear any pending interrupt */ + while (dma_readl(atdma, EBCISR)) + cpu_relax(); + + channel_writel(atchan, SADDR, 0); + channel_writel(atchan, DADDR, 0); + channel_writel(atchan, CTRLA, 0); + channel_writel(atchan, CTRLB, 0); + channel_writel(atchan, DSCR, first->txd.phys); + dma_writel(atdma, CHER, atchan->mask); + + vdbg_dump_regs(atchan); +} + +/** + * atc_chain_complete - finish work for one transaction chain + * @atchan: channel we work on + * @desc: descriptor at the head of the chain we want do complete + * + * Called with atchan->lock held and bh disabled */ +static void +atc_chain_complete(struct at_dma_chan *atchan, struct at_desc *desc) +{ + dma_async_tx_callback callback; + void *param; + struct dma_async_tx_descriptor *txd = &desc->txd; + + dev_vdbg(chan2dev(&atchan->chan_common), + "descriptor %u complete\n", txd->cookie); + + atchan->completed_cookie = txd->cookie; + callback = txd->callback; + param = txd->callback_param; + + /* move children to free_list */ + list_splice_init(&desc->tx_list, &atchan->free_list); + /* move myself to free_list */ + list_move(&desc->desc_node, &atchan->free_list); + + /* unmap dma addresses */ + if (!atchan->chan_common.private) { + struct device *parent = chan2parent(&atchan->chan_common); + if (!(txd->flags & DMA_COMPL_SKIP_DEST_UNMAP)) { + if (txd->flags & DMA_COMPL_DEST_UNMAP_SINGLE) + dma_unmap_single(parent, + desc->lli.daddr, + desc->len, DMA_FROM_DEVICE); + else + dma_unmap_page(parent, + desc->lli.daddr, + desc->len, DMA_FROM_DEVICE); + } + if (!(txd->flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + if (txd->flags & DMA_COMPL_SRC_UNMAP_SINGLE) + dma_unmap_single(parent, + desc->lli.saddr, + desc->len, DMA_TO_DEVICE); + else + dma_unmap_page(parent, + desc->lli.saddr, + desc->len, DMA_TO_DEVICE); + } + } + + /* + * The API requires that no submissions are done from a + * callback, so we don't need to drop the lock here + */ + if (callback) + callback(param); + + dma_run_dependencies(txd); +} + +/** + * atc_complete_all - finish work for all transactions + * @atchan: channel to complete transactions for + * + * Eventually submit queued descriptors if any + * + * Assume channel is idle while calling this function + * Called with atchan->lock held and bh disabled + */ +static void atc_complete_all(struct at_dma_chan *atchan) +{ + struct at_desc *desc, *_desc; + LIST_HEAD(list); + + dev_vdbg(chan2dev(&atchan->chan_common), "complete all\n"); + + BUG_ON(atc_chan_is_enabled(atchan)); + + /* + * Submit queued descriptors ASAP, i.e. before we go through + * the completed ones. + */ + if (!list_empty(&atchan->queue)) + atc_dostart(atchan, atc_first_queued(atchan)); + /* empty active_list now it is completed */ + list_splice_init(&atchan->active_list, &list); + /* empty queue list by moving descriptors (if any) to active_list */ + list_splice_init(&atchan->queue, &atchan->active_list); + + list_for_each_entry_safe(desc, _desc, &list, desc_node) + atc_chain_complete(atchan, desc); +} + +/** + * atc_cleanup_descriptors - cleanup up finished descriptors in active_list + * @atchan: channel to be cleaned up + * + * Called with atchan->lock held and bh disabled + */ +static void atc_cleanup_descriptors(struct at_dma_chan *atchan) +{ + struct at_desc *desc, *_desc; + struct at_desc *child; + + dev_vdbg(chan2dev(&atchan->chan_common), "cleanup descriptors\n"); + + list_for_each_entry_safe(desc, _desc, &atchan->active_list, desc_node) { + if (!(desc->lli.ctrla & ATC_DONE)) + /* This one is currently in progress */ + return; + + list_for_each_entry(child, &desc->tx_list, desc_node) + if (!(child->lli.ctrla & ATC_DONE)) + /* Currently in progress */ + return; + + /* + * No descriptors so far seem to be in progress, i.e. + * this chain must be done. + */ + atc_chain_complete(atchan, desc); + } +} + +/** + * atc_advance_work - at the end of a transaction, move forward + * @atchan: channel where the transaction ended + * + * Called with atchan->lock held and bh disabled + */ +static void atc_advance_work(struct at_dma_chan *atchan) +{ + dev_vdbg(chan2dev(&atchan->chan_common), "advance_work\n"); + + if (list_empty(&atchan->active_list) || + list_is_singular(&atchan->active_list)) { + atc_complete_all(atchan); + } else { + atc_chain_complete(atchan, atc_first_active(atchan)); + /* advance work */ + atc_dostart(atchan, atc_first_active(atchan)); + } +} + + +/** + * atc_handle_error - handle errors reported by DMA controller + * @atchan: channel where error occurs + * + * Called with atchan->lock held and bh disabled + */ +static void atc_handle_error(struct at_dma_chan *atchan) +{ + struct at_desc *bad_desc; + struct at_desc *child; + + /* + * The descriptor currently at the head of the active list is + * broked. Since we don't have any way to report errors, we'll + * just have to scream loudly and try to carry on. + */ + bad_desc = atc_first_active(atchan); + list_del_init(&bad_desc->desc_node); + + /* As we are stopped, take advantage to push queued descriptors + * in active_list */ + list_splice_init(&atchan->queue, atchan->active_list.prev); + + /* Try to restart the controller */ + if (!list_empty(&atchan->active_list)) + atc_dostart(atchan, atc_first_active(atchan)); + + /* + * KERN_CRITICAL may seem harsh, but since this only happens + * when someone submits a bad physical address in a + * descriptor, we should consider ourselves lucky that the + * controller flagged an error instead of scribbling over + * random memory locations. + */ + dev_crit(chan2dev(&atchan->chan_common), + "Bad descriptor submitted for DMA!\n"); + dev_crit(chan2dev(&atchan->chan_common), + " cookie: %d\n", bad_desc->txd.cookie); + atc_dump_lli(atchan, &bad_desc->lli); + list_for_each_entry(child, &bad_desc->tx_list, desc_node) + atc_dump_lli(atchan, &child->lli); + + /* Pretend the descriptor completed successfully */ + atc_chain_complete(atchan, bad_desc); +} + + +/*-- IRQ & Tasklet ---------------------------------------------------*/ + +static void atc_tasklet(unsigned long data) +{ + struct at_dma_chan *atchan = (struct at_dma_chan *)data; + + /* Channel cannot be enabled here */ + if (atc_chan_is_enabled(atchan)) { + dev_err(chan2dev(&atchan->chan_common), + "BUG: channel enabled in tasklet\n"); + return; + } + + spin_lock(&atchan->lock); + if (test_and_clear_bit(0, &atchan->error_status)) + atc_handle_error(atchan); + else + atc_advance_work(atchan); + + spin_unlock(&atchan->lock); +} + +static irqreturn_t at_dma_interrupt(int irq, void *dev_id) +{ + struct at_dma *atdma = (struct at_dma *)dev_id; + struct at_dma_chan *atchan; + int i; + u32 status, pending, imr; + int ret = IRQ_NONE; + + do { + imr = dma_readl(atdma, EBCIMR); + status = dma_readl(atdma, EBCISR); + pending = status & imr; + + if (!pending) + break; + + dev_vdbg(atdma->dma_common.dev, + "interrupt: status = 0x%08x, 0x%08x, 0x%08x\n", + status, imr, pending); + + for (i = 0; i < atdma->dma_common.chancnt; i++) { + atchan = &atdma->chan[i]; + if (pending & (AT_DMA_CBTC(i) | AT_DMA_ERR(i))) { + if (pending & AT_DMA_ERR(i)) { + /* Disable channel on AHB error */ + dma_writel(atdma, CHDR, atchan->mask); + /* Give information to tasklet */ + set_bit(0, &atchan->error_status); + } + tasklet_schedule(&atchan->tasklet); + ret = IRQ_HANDLED; + } + } + + } while (pending); + + return ret; +} + + +/*-- DMA Engine API --------------------------------------------------*/ + +/** + * atc_tx_submit - set the prepared descriptor(s) to be executed by the engine + * @desc: descriptor at the head of the transaction chain + * + * Queue chain if DMA engine is working already + * + * Cookie increment and adding to active_list or queue must be atomic + */ +static dma_cookie_t atc_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct at_desc *desc = txd_to_at_desc(tx); + struct at_dma_chan *atchan = to_at_dma_chan(tx->chan); + dma_cookie_t cookie; + + spin_lock_bh(&atchan->lock); + cookie = atc_assign_cookie(atchan, desc); + + if (list_empty(&atchan->active_list)) { + dev_vdbg(chan2dev(tx->chan), "tx_submit: started %u\n", + desc->txd.cookie); + atc_dostart(atchan, desc); + list_add_tail(&desc->desc_node, &atchan->active_list); + } else { + dev_vdbg(chan2dev(tx->chan), "tx_submit: queued %u\n", + desc->txd.cookie); + list_add_tail(&desc->desc_node, &atchan->queue); + } + + spin_unlock_bh(&atchan->lock); + + return cookie; +} + +/** + * atc_prep_dma_memcpy - prepare a memcpy operation + * @chan: the channel to prepare operation on + * @dest: operation virtual destination address + * @src: operation virtual source address + * @len: operation length + * @flags: tx descriptor status flags + */ +static struct dma_async_tx_descriptor * +atc_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src, + size_t len, unsigned long flags) +{ + struct at_dma_chan *atchan = to_at_dma_chan(chan); + struct at_desc *desc = NULL; + struct at_desc *first = NULL; + struct at_desc *prev = NULL; + size_t xfer_count; + size_t offset; + unsigned int src_width; + unsigned int dst_width; + u32 ctrla; + u32 ctrlb; + + dev_vdbg(chan2dev(chan), "prep_dma_memcpy: d0x%x s0x%x l0x%zx f0x%lx\n", + dest, src, len, flags); + + if (unlikely(!len)) { + dev_dbg(chan2dev(chan), "prep_dma_memcpy: length is zero!\n"); + return NULL; + } + + ctrla = ATC_DEFAULT_CTRLA; + ctrlb = ATC_DEFAULT_CTRLB + | ATC_SRC_ADDR_MODE_INCR + | ATC_DST_ADDR_MODE_INCR + | ATC_FC_MEM2MEM; + + /* + * We can be a lot more clever here, but this should take care + * of the most common optimization. + */ + if (!((src | dest | len) & 3)) { + ctrla |= ATC_SRC_WIDTH_WORD | ATC_DST_WIDTH_WORD; + src_width = dst_width = 2; + } else if (!((src | dest | len) & 1)) { + ctrla |= ATC_SRC_WIDTH_HALFWORD | ATC_DST_WIDTH_HALFWORD; + src_width = dst_width = 1; + } else { + ctrla |= ATC_SRC_WIDTH_BYTE | ATC_DST_WIDTH_BYTE; + src_width = dst_width = 0; + } + + for (offset = 0; offset < len; offset += xfer_count << src_width) { + xfer_count = min_t(size_t, (len - offset) >> src_width, + ATC_BTSIZE_MAX); + + desc = atc_desc_get(atchan); + if (!desc) + goto err_desc_get; + + desc->lli.saddr = src + offset; + desc->lli.daddr = dest + offset; + desc->lli.ctrla = ctrla | xfer_count; + desc->lli.ctrlb = ctrlb; + + desc->txd.cookie = 0; + async_tx_ack(&desc->txd); + + if (!first) { + first = desc; + } else { + /* inform the HW lli about chaining */ + prev->lli.dscr = desc->txd.phys; + /* insert the link descriptor to the LD ring */ + list_add_tail(&desc->desc_node, + &first->tx_list); + } + prev = desc; + } + + /* First descriptor of the chain embedds additional information */ + first->txd.cookie = -EBUSY; + first->len = len; + + /* set end-of-link to the last link descriptor of list*/ + set_desc_eol(desc); + + desc->txd.flags = flags; /* client is in control of this ack */ + + return &first->txd; + +err_desc_get: + atc_desc_put(atchan, first); + return NULL; +} + + +/** + * atc_prep_slave_sg - prepare descriptors for a DMA_SLAVE transaction + * @chan: DMA channel + * @sgl: scatterlist to transfer to/from + * @sg_len: number of entries in @scatterlist + * @direction: DMA direction + * @flags: tx descriptor status flags + */ +static struct dma_async_tx_descriptor * +atc_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, + unsigned int sg_len, enum dma_data_direction direction, + unsigned long flags) +{ + struct at_dma_chan *atchan = to_at_dma_chan(chan); + struct at_dma_slave *atslave = chan->private; + struct at_desc *first = NULL; + struct at_desc *prev = NULL; + u32 ctrla; + u32 ctrlb; + dma_addr_t reg; + unsigned int reg_width; + unsigned int mem_width; + unsigned int i; + struct scatterlist *sg; + size_t total_len = 0; + + dev_vdbg(chan2dev(chan), "prep_slave_sg: %s f0x%lx\n", + direction == DMA_TO_DEVICE ? "TO DEVICE" : "FROM DEVICE", + flags); + + if (unlikely(!atslave || !sg_len)) { + dev_dbg(chan2dev(chan), "prep_dma_memcpy: length is zero!\n"); + return NULL; + } + + reg_width = atslave->reg_width; + + ctrla = ATC_DEFAULT_CTRLA | atslave->ctrla; + ctrlb = ATC_DEFAULT_CTRLB | ATC_IEN; + + switch (direction) { + case DMA_TO_DEVICE: + ctrla |= ATC_DST_WIDTH(reg_width); + ctrlb |= ATC_DST_ADDR_MODE_FIXED + | ATC_SRC_ADDR_MODE_INCR + | ATC_FC_MEM2PER; + reg = atslave->tx_reg; + for_each_sg(sgl, sg, sg_len, i) { + struct at_desc *desc; + u32 len; + u32 mem; + + desc = atc_desc_get(atchan); + if (!desc) + goto err_desc_get; + + mem = sg_phys(sg); + len = sg_dma_len(sg); + mem_width = 2; + if (unlikely(mem & 3 || len & 3)) + mem_width = 0; + + desc->lli.saddr = mem; + desc->lli.daddr = reg; + desc->lli.ctrla = ctrla + | ATC_SRC_WIDTH(mem_width) + | len >> mem_width; + desc->lli.ctrlb = ctrlb; + + if (!first) { + first = desc; + } else { + /* inform the HW lli about chaining */ + prev->lli.dscr = desc->txd.phys; + /* insert the link descriptor to the LD ring */ + list_add_tail(&desc->desc_node, + &first->tx_list); + } + prev = desc; + total_len += len; + } + break; + case DMA_FROM_DEVICE: + ctrla |= ATC_SRC_WIDTH(reg_width); + ctrlb |= ATC_DST_ADDR_MODE_INCR + | ATC_SRC_ADDR_MODE_FIXED + | ATC_FC_PER2MEM; + + reg = atslave->rx_reg; + for_each_sg(sgl, sg, sg_len, i) { + struct at_desc *desc; + u32 len; + u32 mem; + + desc = atc_desc_get(atchan); + if (!desc) + goto err_desc_get; + + mem = sg_phys(sg); + len = sg_dma_len(sg); + mem_width = 2; + if (unlikely(mem & 3 || len & 3)) + mem_width = 0; + + desc->lli.saddr = reg; + desc->lli.daddr = mem; + desc->lli.ctrla = ctrla + | ATC_DST_WIDTH(mem_width) + | len >> mem_width; + desc->lli.ctrlb = ctrlb; + + if (!first) { + first = desc; + } else { + /* inform the HW lli about chaining */ + prev->lli.dscr = desc->txd.phys; + /* insert the link descriptor to the LD ring */ + list_add_tail(&desc->desc_node, + &first->tx_list); + } + prev = desc; + total_len += len; + } + break; + default: + return NULL; + } + + /* set end-of-link to the last link descriptor of list*/ + set_desc_eol(prev); + + /* First descriptor of the chain embedds additional information */ + first->txd.cookie = -EBUSY; + first->len = total_len; + + /* last link descriptor of list is responsible of flags */ + prev->txd.flags = flags; /* client is in control of this ack */ + + return &first->txd; + +err_desc_get: + dev_err(chan2dev(chan), "not enough descriptors available\n"); + atc_desc_put(atchan, first); + return NULL; +} + +static void atc_terminate_all(struct dma_chan *chan) +{ + struct at_dma_chan *atchan = to_at_dma_chan(chan); + struct at_dma *atdma = to_at_dma(chan->device); + struct at_desc *desc, *_desc; + LIST_HEAD(list); + + /* + * This is only called when something went wrong elsewhere, so + * we don't really care about the data. Just disable the + * channel. We still have to poll the channel enable bit due + * to AHB/HSB limitations. + */ + spin_lock_bh(&atchan->lock); + + dma_writel(atdma, CHDR, atchan->mask); + + /* confirm that this channel is disabled */ + while (dma_readl(atdma, CHSR) & atchan->mask) + cpu_relax(); + + /* active_list entries will end up before queued entries */ + list_splice_init(&atchan->queue, &list); + list_splice_init(&atchan->active_list, &list); + + spin_unlock_bh(&atchan->lock); + + /* Flush all pending and queued descriptors */ + list_for_each_entry_safe(desc, _desc, &list, desc_node) + atc_chain_complete(atchan, desc); +} + +/** + * atc_is_tx_complete - poll for transaction completion + * @chan: DMA channel + * @cookie: transaction identifier to check status of + * @done: if not %NULL, updated with last completed transaction + * @used: if not %NULL, updated with last used transaction + * + * If @done and @used are passed in, upon return they reflect the driver + * internal state and can be used with dma_async_is_complete() to check + * the status of multiple cookies without re-checking hardware state. + */ +static enum dma_status +atc_is_tx_complete(struct dma_chan *chan, + dma_cookie_t cookie, + dma_cookie_t *done, dma_cookie_t *used) +{ + struct at_dma_chan *atchan = to_at_dma_chan(chan); + dma_cookie_t last_used; + dma_cookie_t last_complete; + enum dma_status ret; + + dev_vdbg(chan2dev(chan), "is_tx_complete: %d (d%d, u%d)\n", + cookie, done ? *done : 0, used ? *used : 0); + + spin_lock_bh(&atchan->lock); + + last_complete = atchan->completed_cookie; + last_used = chan->cookie; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + if (ret != DMA_SUCCESS) { + atc_cleanup_descriptors(atchan); + + last_complete = atchan->completed_cookie; + last_used = chan->cookie; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + } + + spin_unlock_bh(&atchan->lock); + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + return ret; +} + +/** + * atc_issue_pending - try to finish work + * @chan: target DMA channel + */ +static void atc_issue_pending(struct dma_chan *chan) +{ + struct at_dma_chan *atchan = to_at_dma_chan(chan); + + dev_vdbg(chan2dev(chan), "issue_pending\n"); + + if (!atc_chan_is_enabled(atchan)) { + spin_lock_bh(&atchan->lock); + atc_advance_work(atchan); + spin_unlock_bh(&atchan->lock); + } +} + +/** + * atc_alloc_chan_resources - allocate resources for DMA channel + * @chan: allocate descriptor resources for this channel + * @client: current client requesting the channel be ready for requests + * + * return - the number of allocated descriptors + */ +static int atc_alloc_chan_resources(struct dma_chan *chan) +{ + struct at_dma_chan *atchan = to_at_dma_chan(chan); + struct at_dma *atdma = to_at_dma(chan->device); + struct at_desc *desc; + struct at_dma_slave *atslave; + int i; + u32 cfg; + LIST_HEAD(tmp_list); + + dev_vdbg(chan2dev(chan), "alloc_chan_resources\n"); + + /* ASSERT: channel is idle */ + if (atc_chan_is_enabled(atchan)) { + dev_dbg(chan2dev(chan), "DMA channel not idle ?\n"); + return -EIO; + } + + cfg = ATC_DEFAULT_CFG; + + atslave = chan->private; + if (atslave) { + /* + * We need controller-specific data to set up slave + * transfers. + */ + BUG_ON(!atslave->dma_dev || atslave->dma_dev != atdma->dma_common.dev); + + /* if cfg configuration specified take it instad of default */ + if (atslave->cfg) + cfg = atslave->cfg; + } + + /* have we already been set up? + * reconfigure channel but no need to reallocate descriptors */ + if (!list_empty(&atchan->free_list)) + return atchan->descs_allocated; + + /* Allocate initial pool of descriptors */ + for (i = 0; i < init_nr_desc_per_channel; i++) { + desc = atc_alloc_descriptor(chan, GFP_KERNEL); + if (!desc) { + dev_err(atdma->dma_common.dev, + "Only %d initial descriptors\n", i); + break; + } + list_add_tail(&desc->desc_node, &tmp_list); + } + + spin_lock_bh(&atchan->lock); + atchan->descs_allocated = i; + list_splice(&tmp_list, &atchan->free_list); + atchan->completed_cookie = chan->cookie = 1; + spin_unlock_bh(&atchan->lock); + + /* channel parameters */ + channel_writel(atchan, CFG, cfg); + + dev_dbg(chan2dev(chan), + "alloc_chan_resources: allocated %d descriptors\n", + atchan->descs_allocated); + + return atchan->descs_allocated; +} + +/** + * atc_free_chan_resources - free all channel resources + * @chan: DMA channel + */ +static void atc_free_chan_resources(struct dma_chan *chan) +{ + struct at_dma_chan *atchan = to_at_dma_chan(chan); + struct at_dma *atdma = to_at_dma(chan->device); + struct at_desc *desc, *_desc; + LIST_HEAD(list); + + dev_dbg(chan2dev(chan), "free_chan_resources: (descs allocated=%u)\n", + atchan->descs_allocated); + + /* ASSERT: channel is idle */ + BUG_ON(!list_empty(&atchan->active_list)); + BUG_ON(!list_empty(&atchan->queue)); + BUG_ON(atc_chan_is_enabled(atchan)); + + list_for_each_entry_safe(desc, _desc, &atchan->free_list, desc_node) { + dev_vdbg(chan2dev(chan), " freeing descriptor %p\n", desc); + list_del(&desc->desc_node); + /* free link descriptor */ + dma_pool_free(atdma->dma_desc_pool, desc, desc->txd.phys); + } + list_splice_init(&atchan->free_list, &list); + atchan->descs_allocated = 0; + + dev_vdbg(chan2dev(chan), "free_chan_resources: done\n"); +} + + +/*-- Module Management -----------------------------------------------*/ + +/** + * at_dma_off - disable DMA controller + * @atdma: the Atmel HDAMC device + */ +static void at_dma_off(struct at_dma *atdma) +{ + dma_writel(atdma, EN, 0); + + /* disable all interrupts */ + dma_writel(atdma, EBCIDR, -1L); + + /* confirm that all channels are disabled */ + while (dma_readl(atdma, CHSR) & atdma->all_chan_mask) + cpu_relax(); +} + +static int __init at_dma_probe(struct platform_device *pdev) +{ + struct at_dma_platform_data *pdata; + struct resource *io; + struct at_dma *atdma; + size_t size; + int irq; + int err; + int i; + + /* get DMA Controller parameters from platform */ + pdata = pdev->dev.platform_data; + if (!pdata || pdata->nr_channels > AT_DMA_MAX_NR_CHANNELS) + return -EINVAL; + + io = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!io) + return -EINVAL; + + irq = platform_get_irq(pdev, 0); + if (irq < 0) + return irq; + + size = sizeof(struct at_dma); + size += pdata->nr_channels * sizeof(struct at_dma_chan); + atdma = kzalloc(size, GFP_KERNEL); + if (!atdma) + return -ENOMEM; + + /* discover transaction capabilites from the platform data */ + atdma->dma_common.cap_mask = pdata->cap_mask; + atdma->all_chan_mask = (1 << pdata->nr_channels) - 1; + + size = io->end - io->start + 1; + if (!request_mem_region(io->start, size, pdev->dev.driver->name)) { + err = -EBUSY; + goto err_kfree; + } + + atdma->regs = ioremap(io->start, size); + if (!atdma->regs) { + err = -ENOMEM; + goto err_release_r; + } + + atdma->clk = clk_get(&pdev->dev, "dma_clk"); + if (IS_ERR(atdma->clk)) { + err = PTR_ERR(atdma->clk); + goto err_clk; + } + clk_enable(atdma->clk); + + /* force dma off, just in case */ + at_dma_off(atdma); + + err = request_irq(irq, at_dma_interrupt, 0, "at_hdmac", atdma); + if (err) + goto err_irq; + + platform_set_drvdata(pdev, atdma); + + /* create a pool of consistent memory blocks for hardware descriptors */ + atdma->dma_desc_pool = dma_pool_create("at_hdmac_desc_pool", + &pdev->dev, sizeof(struct at_desc), + 4 /* word alignment */, 0); + if (!atdma->dma_desc_pool) { + dev_err(&pdev->dev, "No memory for descriptors dma pool\n"); + err = -ENOMEM; + goto err_pool_create; + } + + /* clear any pending interrupt */ + while (dma_readl(atdma, EBCISR)) + cpu_relax(); + + /* initialize channels related values */ + INIT_LIST_HEAD(&atdma->dma_common.channels); + for (i = 0; i < pdata->nr_channels; i++, atdma->dma_common.chancnt++) { + struct at_dma_chan *atchan = &atdma->chan[i]; + + atchan->chan_common.device = &atdma->dma_common; + atchan->chan_common.cookie = atchan->completed_cookie = 1; + atchan->chan_common.chan_id = i; + list_add_tail(&atchan->chan_common.device_node, + &atdma->dma_common.channels); + + atchan->ch_regs = atdma->regs + ch_regs(i); + spin_lock_init(&atchan->lock); + atchan->mask = 1 << i; + + INIT_LIST_HEAD(&atchan->active_list); + INIT_LIST_HEAD(&atchan->queue); + INIT_LIST_HEAD(&atchan->free_list); + + tasklet_init(&atchan->tasklet, atc_tasklet, + (unsigned long)atchan); + atc_enable_irq(atchan); + } + + /* set base routines */ + atdma->dma_common.device_alloc_chan_resources = atc_alloc_chan_resources; + atdma->dma_common.device_free_chan_resources = atc_free_chan_resources; + atdma->dma_common.device_is_tx_complete = atc_is_tx_complete; + atdma->dma_common.device_issue_pending = atc_issue_pending; + atdma->dma_common.dev = &pdev->dev; + + /* set prep routines based on capability */ + if (dma_has_cap(DMA_MEMCPY, atdma->dma_common.cap_mask)) + atdma->dma_common.device_prep_dma_memcpy = atc_prep_dma_memcpy; + + if (dma_has_cap(DMA_SLAVE, atdma->dma_common.cap_mask)) { + atdma->dma_common.device_prep_slave_sg = atc_prep_slave_sg; + atdma->dma_common.device_terminate_all = atc_terminate_all; + } + + dma_writel(atdma, EN, AT_DMA_ENABLE); + + dev_info(&pdev->dev, "Atmel AHB DMA Controller ( %s%s), %d channels\n", + dma_has_cap(DMA_MEMCPY, atdma->dma_common.cap_mask) ? "cpy " : "", + dma_has_cap(DMA_SLAVE, atdma->dma_common.cap_mask) ? "slave " : "", + atdma->dma_common.chancnt); + + dma_async_device_register(&atdma->dma_common); + + return 0; + +err_pool_create: + platform_set_drvdata(pdev, NULL); + free_irq(platform_get_irq(pdev, 0), atdma); +err_irq: + clk_disable(atdma->clk); + clk_put(atdma->clk); +err_clk: + iounmap(atdma->regs); + atdma->regs = NULL; +err_release_r: + release_mem_region(io->start, size); +err_kfree: + kfree(atdma); + return err; +} + +static int __exit at_dma_remove(struct platform_device *pdev) +{ + struct at_dma *atdma = platform_get_drvdata(pdev); + struct dma_chan *chan, *_chan; + struct resource *io; + + at_dma_off(atdma); + dma_async_device_unregister(&atdma->dma_common); + + dma_pool_destroy(atdma->dma_desc_pool); + platform_set_drvdata(pdev, NULL); + free_irq(platform_get_irq(pdev, 0), atdma); + + list_for_each_entry_safe(chan, _chan, &atdma->dma_common.channels, + device_node) { + struct at_dma_chan *atchan = to_at_dma_chan(chan); + + /* Disable interrupts */ + atc_disable_irq(atchan); + tasklet_disable(&atchan->tasklet); + + tasklet_kill(&atchan->tasklet); + list_del(&chan->device_node); + } + + clk_disable(atdma->clk); + clk_put(atdma->clk); + + iounmap(atdma->regs); + atdma->regs = NULL; + + io = platform_get_resource(pdev, IORESOURCE_MEM, 0); + release_mem_region(io->start, io->end - io->start + 1); + + kfree(atdma); + + return 0; +} + +static void at_dma_shutdown(struct platform_device *pdev) +{ + struct at_dma *atdma = platform_get_drvdata(pdev); + + at_dma_off(platform_get_drvdata(pdev)); + clk_disable(atdma->clk); +} + +static int at_dma_suspend_noirq(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct at_dma *atdma = platform_get_drvdata(pdev); + + at_dma_off(platform_get_drvdata(pdev)); + clk_disable(atdma->clk); + return 0; +} + +static int at_dma_resume_noirq(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct at_dma *atdma = platform_get_drvdata(pdev); + + clk_enable(atdma->clk); + dma_writel(atdma, EN, AT_DMA_ENABLE); + return 0; +} + +static const struct dev_pm_ops at_dma_dev_pm_ops = { + .suspend_noirq = at_dma_suspend_noirq, + .resume_noirq = at_dma_resume_noirq, +}; + +static struct platform_driver at_dma_driver = { + .remove = __exit_p(at_dma_remove), + .shutdown = at_dma_shutdown, + .driver = { + .name = "at_hdmac", + .pm = &at_dma_dev_pm_ops, + }, +}; + +static int __init at_dma_init(void) +{ + return platform_driver_probe(&at_dma_driver, at_dma_probe); +} +module_init(at_dma_init); + +static void __exit at_dma_exit(void) +{ + platform_driver_unregister(&at_dma_driver); +} +module_exit(at_dma_exit); + +MODULE_DESCRIPTION("Atmel AHB DMA Controller driver"); +MODULE_AUTHOR("Nicolas Ferre <nicolas.ferre@atmel.com>"); +MODULE_LICENSE("GPL"); +MODULE_ALIAS("platform:at_hdmac"); diff --git a/drivers/dma/at_hdmac_regs.h b/drivers/dma/at_hdmac_regs.h new file mode 100644 index 00000000000..495457e3dc4 --- /dev/null +++ b/drivers/dma/at_hdmac_regs.h @@ -0,0 +1,354 @@ +/* + * Header file for the Atmel AHB DMA Controller driver + * + * Copyright (C) 2008 Atmel Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + */ +#ifndef AT_HDMAC_REGS_H +#define AT_HDMAC_REGS_H + +#include <mach/at_hdmac.h> + +#define AT_DMA_MAX_NR_CHANNELS 8 + + +#define AT_DMA_GCFG 0x00 /* Global Configuration Register */ +#define AT_DMA_IF_BIGEND(i) (0x1 << (i)) /* AHB-Lite Interface i in Big-endian mode */ +#define AT_DMA_ARB_CFG (0x1 << 4) /* Arbiter mode. */ +#define AT_DMA_ARB_CFG_FIXED (0x0 << 4) +#define AT_DMA_ARB_CFG_ROUND_ROBIN (0x1 << 4) + +#define AT_DMA_EN 0x04 /* Controller Enable Register */ +#define AT_DMA_ENABLE (0x1 << 0) + +#define AT_DMA_SREQ 0x08 /* Software Single Request Register */ +#define AT_DMA_SSREQ(x) (0x1 << ((x) << 1)) /* Request a source single transfer on channel x */ +#define AT_DMA_DSREQ(x) (0x1 << (1 + ((x) << 1))) /* Request a destination single transfer on channel x */ + +#define AT_DMA_CREQ 0x0C /* Software Chunk Transfer Request Register */ +#define AT_DMA_SCREQ(x) (0x1 << ((x) << 1)) /* Request a source chunk transfer on channel x */ +#define AT_DMA_DCREQ(x) (0x1 << (1 + ((x) << 1))) /* Request a destination chunk transfer on channel x */ + +#define AT_DMA_LAST 0x10 /* Software Last Transfer Flag Register */ +#define AT_DMA_SLAST(x) (0x1 << ((x) << 1)) /* This src rq is last tx of buffer on channel x */ +#define AT_DMA_DLAST(x) (0x1 << (1 + ((x) << 1))) /* This dst rq is last tx of buffer on channel x */ + +#define AT_DMA_SYNC 0x14 /* Request Synchronization Register */ +#define AT_DMA_SYR(h) (0x1 << (h)) /* Synchronize handshake line h */ + +/* Error, Chained Buffer transfer completed and Buffer transfer completed Interrupt registers */ +#define AT_DMA_EBCIER 0x18 /* Enable register */ +#define AT_DMA_EBCIDR 0x1C /* Disable register */ +#define AT_DMA_EBCIMR 0x20 /* Mask Register */ +#define AT_DMA_EBCISR 0x24 /* Status Register */ +#define AT_DMA_CBTC_OFFSET 8 +#define AT_DMA_ERR_OFFSET 16 +#define AT_DMA_BTC(x) (0x1 << (x)) +#define AT_DMA_CBTC(x) (0x1 << (AT_DMA_CBTC_OFFSET + (x))) +#define AT_DMA_ERR(x) (0x1 << (AT_DMA_ERR_OFFSET + (x))) + +#define AT_DMA_CHER 0x28 /* Channel Handler Enable Register */ +#define AT_DMA_ENA(x) (0x1 << (x)) +#define AT_DMA_SUSP(x) (0x1 << ( 8 + (x))) +#define AT_DMA_KEEP(x) (0x1 << (24 + (x))) + +#define AT_DMA_CHDR 0x2C /* Channel Handler Disable Register */ +#define AT_DMA_DIS(x) (0x1 << (x)) +#define AT_DMA_RES(x) (0x1 << ( 8 + (x))) + +#define AT_DMA_CHSR 0x30 /* Channel Handler Status Register */ +#define AT_DMA_EMPT(x) (0x1 << (16 + (x))) +#define AT_DMA_STAL(x) (0x1 << (24 + (x))) + + +#define AT_DMA_CH_REGS_BASE 0x3C /* Channel registers base address */ +#define ch_regs(x) (AT_DMA_CH_REGS_BASE + (x) * 0x28) /* Channel x base addr */ + +/* Hardware register offset for each channel */ +#define ATC_SADDR_OFFSET 0x00 /* Source Address Register */ +#define ATC_DADDR_OFFSET 0x04 /* Destination Address Register */ +#define ATC_DSCR_OFFSET 0x08 /* Descriptor Address Register */ +#define ATC_CTRLA_OFFSET 0x0C /* Control A Register */ +#define ATC_CTRLB_OFFSET 0x10 /* Control B Register */ +#define ATC_CFG_OFFSET 0x14 /* Configuration Register */ +#define ATC_SPIP_OFFSET 0x18 /* Src PIP Configuration Register */ +#define ATC_DPIP_OFFSET 0x1C /* Dst PIP Configuration Register */ + + +/* Bitfield definitions */ + +/* Bitfields in DSCR */ +#define ATC_DSCR_IF(i) (0x3 & (i)) /* Dsc feched via AHB-Lite Interface i */ + +/* Bitfields in CTRLA */ +#define ATC_BTSIZE_MAX 0xFFFFUL /* Maximum Buffer Transfer Size */ +#define ATC_BTSIZE(x) (ATC_BTSIZE_MAX & (x)) /* Buffer Transfer Size */ +/* Chunck Tranfer size definitions are in at_hdmac.h */ +#define ATC_SRC_WIDTH_MASK (0x3 << 24) /* Source Single Transfer Size */ +#define ATC_SRC_WIDTH(x) ((x) << 24) +#define ATC_SRC_WIDTH_BYTE (0x0 << 24) +#define ATC_SRC_WIDTH_HALFWORD (0x1 << 24) +#define ATC_SRC_WIDTH_WORD (0x2 << 24) +#define ATC_DST_WIDTH_MASK (0x3 << 28) /* Destination Single Transfer Size */ +#define ATC_DST_WIDTH(x) ((x) << 28) +#define ATC_DST_WIDTH_BYTE (0x0 << 28) +#define ATC_DST_WIDTH_HALFWORD (0x1 << 28) +#define ATC_DST_WIDTH_WORD (0x2 << 28) +#define ATC_DONE (0x1 << 31) /* Tx Done (only written back in descriptor) */ + +/* Bitfields in CTRLB */ +#define ATC_SIF(i) (0x3 & (i)) /* Src tx done via AHB-Lite Interface i */ +#define ATC_DIF(i) ((0x3 & (i)) << 4) /* Dst tx done via AHB-Lite Interface i */ +#define ATC_SRC_PIP (0x1 << 8) /* Source Picture-in-Picture enabled */ +#define ATC_DST_PIP (0x1 << 12) /* Destination Picture-in-Picture enabled */ +#define ATC_SRC_DSCR_DIS (0x1 << 16) /* Src Descriptor fetch disable */ +#define ATC_DST_DSCR_DIS (0x1 << 20) /* Dst Descriptor fetch disable */ +#define ATC_FC_MASK (0x7 << 21) /* Choose Flow Controller */ +#define ATC_FC_MEM2MEM (0x0 << 21) /* Mem-to-Mem (DMA) */ +#define ATC_FC_MEM2PER (0x1 << 21) /* Mem-to-Periph (DMA) */ +#define ATC_FC_PER2MEM (0x2 << 21) /* Periph-to-Mem (DMA) */ +#define ATC_FC_PER2PER (0x3 << 21) /* Periph-to-Periph (DMA) */ +#define ATC_FC_PER2MEM_PER (0x4 << 21) /* Periph-to-Mem (Peripheral) */ +#define ATC_FC_MEM2PER_PER (0x5 << 21) /* Mem-to-Periph (Peripheral) */ +#define ATC_FC_PER2PER_SRCPER (0x6 << 21) /* Periph-to-Periph (Src Peripheral) */ +#define ATC_FC_PER2PER_DSTPER (0x7 << 21) /* Periph-to-Periph (Dst Peripheral) */ +#define ATC_SRC_ADDR_MODE_MASK (0x3 << 24) +#define ATC_SRC_ADDR_MODE_INCR (0x0 << 24) /* Incrementing Mode */ +#define ATC_SRC_ADDR_MODE_DECR (0x1 << 24) /* Decrementing Mode */ +#define ATC_SRC_ADDR_MODE_FIXED (0x2 << 24) /* Fixed Mode */ +#define ATC_DST_ADDR_MODE_MASK (0x3 << 28) +#define ATC_DST_ADDR_MODE_INCR (0x0 << 28) /* Incrementing Mode */ +#define ATC_DST_ADDR_MODE_DECR (0x1 << 28) /* Decrementing Mode */ +#define ATC_DST_ADDR_MODE_FIXED (0x2 << 28) /* Fixed Mode */ +#define ATC_IEN (0x1 << 30) /* BTC interrupt enable (active low) */ +#define ATC_AUTO (0x1 << 31) /* Auto multiple buffer tx enable */ + +/* Bitfields in CFG */ +/* are in at_hdmac.h */ + +/* Bitfields in SPIP */ +#define ATC_SPIP_HOLE(x) (0xFFFFU & (x)) +#define ATC_SPIP_BOUNDARY(x) ((0x3FF & (x)) << 16) + +/* Bitfields in DPIP */ +#define ATC_DPIP_HOLE(x) (0xFFFFU & (x)) +#define ATC_DPIP_BOUNDARY(x) ((0x3FF & (x)) << 16) + + +/*-- descriptors -----------------------------------------------------*/ + +/* LLI == Linked List Item; aka DMA buffer descriptor */ +struct at_lli { + /* values that are not changed by hardware */ + dma_addr_t saddr; + dma_addr_t daddr; + /* value that may get written back: */ + u32 ctrla; + /* more values that are not changed by hardware */ + u32 ctrlb; + dma_addr_t dscr; /* chain to next lli */ +}; + +/** + * struct at_desc - software descriptor + * @at_lli: hardware lli structure + * @txd: support for the async_tx api + * @desc_node: node on the channed descriptors list + * @len: total transaction bytecount + */ +struct at_desc { + /* FIRST values the hardware uses */ + struct at_lli lli; + + /* THEN values for driver housekeeping */ + struct list_head tx_list; + struct dma_async_tx_descriptor txd; + struct list_head desc_node; + size_t len; +}; + +static inline struct at_desc * +txd_to_at_desc(struct dma_async_tx_descriptor *txd) +{ + return container_of(txd, struct at_desc, txd); +} + + +/*-- Channels --------------------------------------------------------*/ + +/** + * struct at_dma_chan - internal representation of an Atmel HDMAC channel + * @chan_common: common dmaengine channel object members + * @device: parent device + * @ch_regs: memory mapped register base + * @mask: channel index in a mask + * @error_status: transmit error status information from irq handler + * to tasklet (use atomic operations) + * @tasklet: bottom half to finish transaction work + * @lock: serializes enqueue/dequeue operations to descriptors lists + * @completed_cookie: identifier for the most recently completed operation + * @active_list: list of descriptors dmaengine is being running on + * @queue: list of descriptors ready to be submitted to engine + * @free_list: list of descriptors usable by the channel + * @descs_allocated: records the actual size of the descriptor pool + */ +struct at_dma_chan { + struct dma_chan chan_common; + struct at_dma *device; + void __iomem *ch_regs; + u8 mask; + unsigned long error_status; + struct tasklet_struct tasklet; + + spinlock_t lock; + + /* these other elements are all protected by lock */ + dma_cookie_t completed_cookie; + struct list_head active_list; + struct list_head queue; + struct list_head free_list; + unsigned int descs_allocated; +}; + +#define channel_readl(atchan, name) \ + __raw_readl((atchan)->ch_regs + ATC_##name##_OFFSET) + +#define channel_writel(atchan, name, val) \ + __raw_writel((val), (atchan)->ch_regs + ATC_##name##_OFFSET) + +static inline struct at_dma_chan *to_at_dma_chan(struct dma_chan *dchan) +{ + return container_of(dchan, struct at_dma_chan, chan_common); +} + + +/*-- Controller ------------------------------------------------------*/ + +/** + * struct at_dma - internal representation of an Atmel HDMA Controller + * @chan_common: common dmaengine dma_device object members + * @ch_regs: memory mapped register base + * @clk: dma controller clock + * @all_chan_mask: all channels availlable in a mask + * @dma_desc_pool: base of DMA descriptor region (DMA address) + * @chan: channels table to store at_dma_chan structures + */ +struct at_dma { + struct dma_device dma_common; + void __iomem *regs; + struct clk *clk; + + u8 all_chan_mask; + + struct dma_pool *dma_desc_pool; + /* AT THE END channels table */ + struct at_dma_chan chan[0]; +}; + +#define dma_readl(atdma, name) \ + __raw_readl((atdma)->regs + AT_DMA_##name) +#define dma_writel(atdma, name, val) \ + __raw_writel((val), (atdma)->regs + AT_DMA_##name) + +static inline struct at_dma *to_at_dma(struct dma_device *ddev) +{ + return container_of(ddev, struct at_dma, dma_common); +} + + +/*-- Helper functions ------------------------------------------------*/ + +static struct device *chan2dev(struct dma_chan *chan) +{ + return &chan->dev->device; +} +static struct device *chan2parent(struct dma_chan *chan) +{ + return chan->dev->device.parent; +} + +#if defined(VERBOSE_DEBUG) +static void vdbg_dump_regs(struct at_dma_chan *atchan) +{ + struct at_dma *atdma = to_at_dma(atchan->chan_common.device); + + dev_err(chan2dev(&atchan->chan_common), + " channel %d : imr = 0x%x, chsr = 0x%x\n", + atchan->chan_common.chan_id, + dma_readl(atdma, EBCIMR), + dma_readl(atdma, CHSR)); + + dev_err(chan2dev(&atchan->chan_common), + " channel: s0x%x d0x%x ctrl0x%x:0x%x cfg0x%x l0x%x\n", + channel_readl(atchan, SADDR), + channel_readl(atchan, DADDR), + channel_readl(atchan, CTRLA), + channel_readl(atchan, CTRLB), + channel_readl(atchan, CFG), + channel_readl(atchan, DSCR)); +} +#else +static void vdbg_dump_regs(struct at_dma_chan *atchan) {} +#endif + +static void atc_dump_lli(struct at_dma_chan *atchan, struct at_lli *lli) +{ + dev_printk(KERN_CRIT, chan2dev(&atchan->chan_common), + " desc: s0x%x d0x%x ctrl0x%x:0x%x l0x%x\n", + lli->saddr, lli->daddr, + lli->ctrla, lli->ctrlb, lli->dscr); +} + + +static void atc_setup_irq(struct at_dma_chan *atchan, int on) +{ + struct at_dma *atdma = to_at_dma(atchan->chan_common.device); + u32 ebci; + + /* enable interrupts on buffer chain completion & error */ + ebci = AT_DMA_CBTC(atchan->chan_common.chan_id) + | AT_DMA_ERR(atchan->chan_common.chan_id); + if (on) + dma_writel(atdma, EBCIER, ebci); + else + dma_writel(atdma, EBCIDR, ebci); +} + +static inline void atc_enable_irq(struct at_dma_chan *atchan) +{ + atc_setup_irq(atchan, 1); +} + +static inline void atc_disable_irq(struct at_dma_chan *atchan) +{ + atc_setup_irq(atchan, 0); +} + + +/** + * atc_chan_is_enabled - test if given channel is enabled + * @atchan: channel we want to test status + */ +static inline int atc_chan_is_enabled(struct at_dma_chan *atchan) +{ + struct at_dma *atdma = to_at_dma(atchan->chan_common.device); + + return !!(dma_readl(atdma, CHSR) & atchan->mask); +} + + +/** + * set_desc_eol - set end-of-link to descriptor so it will end transfer + * @desc: descriptor, signle or at the end of a chain, to end chain on + */ +static void set_desc_eol(struct at_desc *desc) +{ + desc->lli.ctrlb |= ATC_SRC_DSCR_DIS | ATC_DST_DSCR_DIS; + desc->lli.dscr = 0; +} + +#endif /* AT_HDMAC_REGS_H */ diff --git a/drivers/dma/coh901318.c b/drivers/dma/coh901318.c new file mode 100644 index 00000000000..64a937262a4 --- /dev/null +++ b/drivers/dma/coh901318.c @@ -0,0 +1,1323 @@ +/* + * driver/dma/coh901318.c + * + * Copyright (C) 2007-2009 ST-Ericsson + * License terms: GNU General Public License (GPL) version 2 + * DMA driver for COH 901 318 + * Author: Per Friden <per.friden@stericsson.com> + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/kernel.h> /* printk() */ +#include <linux/fs.h> /* everything... */ +#include <linux/slab.h> /* kmalloc() */ +#include <linux/dmaengine.h> +#include <linux/platform_device.h> +#include <linux/device.h> +#include <linux/irqreturn.h> +#include <linux/interrupt.h> +#include <linux/io.h> +#include <linux/uaccess.h> +#include <linux/debugfs.h> +#include <mach/coh901318.h> + +#include "coh901318_lli.h" + +#define COHC_2_DEV(cohc) (&cohc->chan.dev->device) + +#ifdef VERBOSE_DEBUG +#define COH_DBG(x) ({ if (1) x; 0; }) +#else +#define COH_DBG(x) ({ if (0) x; 0; }) +#endif + +struct coh901318_desc { + struct dma_async_tx_descriptor desc; + struct list_head node; + struct scatterlist *sg; + unsigned int sg_len; + struct coh901318_lli *data; + enum dma_data_direction dir; + int pending_irqs; + unsigned long flags; +}; + +struct coh901318_base { + struct device *dev; + void __iomem *virtbase; + struct coh901318_pool pool; + struct powersave pm; + struct dma_device dma_slave; + struct dma_device dma_memcpy; + struct coh901318_chan *chans; + struct coh901318_platform *platform; +}; + +struct coh901318_chan { + spinlock_t lock; + int allocated; + int completed; + int id; + int stopped; + + struct work_struct free_work; + struct dma_chan chan; + + struct tasklet_struct tasklet; + + struct list_head active; + struct list_head queue; + struct list_head free; + + unsigned long nbr_active_done; + unsigned long busy; + int pending_irqs; + + struct coh901318_base *base; +}; + +static void coh901318_list_print(struct coh901318_chan *cohc, + struct coh901318_lli *lli) +{ + struct coh901318_lli *l; + dma_addr_t addr = virt_to_phys(lli); + int i = 0; + + while (addr) { + l = phys_to_virt(addr); + dev_vdbg(COHC_2_DEV(cohc), "i %d, lli %p, ctrl 0x%x, src 0x%x" + ", dst 0x%x, link 0x%x link_virt 0x%p\n", + i, l, l->control, l->src_addr, l->dst_addr, + l->link_addr, phys_to_virt(l->link_addr)); + i++; + addr = l->link_addr; + } +} + +#ifdef CONFIG_DEBUG_FS + +#define COH901318_DEBUGFS_ASSIGN(x, y) (x = y) + +static struct coh901318_base *debugfs_dma_base; +static struct dentry *dma_dentry; + +static int coh901318_debugfs_open(struct inode *inode, struct file *file) +{ + + file->private_data = inode->i_private; + return 0; +} + +static int coh901318_debugfs_read(struct file *file, char __user *buf, + size_t count, loff_t *f_pos) +{ + u64 started_channels = debugfs_dma_base->pm.started_channels; + int pool_count = debugfs_dma_base->pool.debugfs_pool_counter; + int i; + int ret = 0; + char *dev_buf; + char *tmp; + int dev_size; + + dev_buf = kmalloc(4*1024, GFP_KERNEL); + if (dev_buf == NULL) + goto err_kmalloc; + tmp = dev_buf; + + tmp += sprintf(tmp, "DMA -- enable dma channels\n"); + + for (i = 0; i < debugfs_dma_base->platform->max_channels; i++) + if (started_channels & (1 << i)) + tmp += sprintf(tmp, "channel %d\n", i); + + tmp += sprintf(tmp, "Pool alloc nbr %d\n", pool_count); + dev_size = tmp - dev_buf; + + /* No more to read if offset != 0 */ + if (*f_pos > dev_size) + goto out; + + if (count > dev_size - *f_pos) + count = dev_size - *f_pos; + + if (copy_to_user(buf, dev_buf + *f_pos, count)) + ret = -EINVAL; + ret = count; + *f_pos += count; + + out: + kfree(dev_buf); + return ret; + + err_kmalloc: + return 0; +} + +static const struct file_operations coh901318_debugfs_status_operations = { + .owner = THIS_MODULE, + .open = coh901318_debugfs_open, + .read = coh901318_debugfs_read, +}; + + +static int __init init_coh901318_debugfs(void) +{ + + dma_dentry = debugfs_create_dir("dma", NULL); + + (void) debugfs_create_file("status", + S_IFREG | S_IRUGO, + dma_dentry, NULL, + &coh901318_debugfs_status_operations); + return 0; +} + +static void __exit exit_coh901318_debugfs(void) +{ + debugfs_remove_recursive(dma_dentry); +} + +module_init(init_coh901318_debugfs); +module_exit(exit_coh901318_debugfs); +#else + +#define COH901318_DEBUGFS_ASSIGN(x, y) + +#endif /* CONFIG_DEBUG_FS */ + +static inline struct coh901318_chan *to_coh901318_chan(struct dma_chan *chan) +{ + return container_of(chan, struct coh901318_chan, chan); +} + +static inline dma_addr_t +cohc_dev_addr(struct coh901318_chan *cohc) +{ + return cohc->base->platform->chan_conf[cohc->id].dev_addr; +} + +static inline const struct coh901318_params * +cohc_chan_param(struct coh901318_chan *cohc) +{ + return &cohc->base->platform->chan_conf[cohc->id].param; +} + +static inline const struct coh_dma_channel * +cohc_chan_conf(struct coh901318_chan *cohc) +{ + return &cohc->base->platform->chan_conf[cohc->id]; +} + +static void enable_powersave(struct coh901318_chan *cohc) +{ + unsigned long flags; + struct powersave *pm = &cohc->base->pm; + + spin_lock_irqsave(&pm->lock, flags); + + pm->started_channels &= ~(1ULL << cohc->id); + + if (!pm->started_channels) { + /* DMA no longer intends to access memory */ + cohc->base->platform->access_memory_state(cohc->base->dev, + false); + } + + spin_unlock_irqrestore(&pm->lock, flags); +} +static void disable_powersave(struct coh901318_chan *cohc) +{ + unsigned long flags; + struct powersave *pm = &cohc->base->pm; + + spin_lock_irqsave(&pm->lock, flags); + + if (!pm->started_channels) { + /* DMA intends to access memory */ + cohc->base->platform->access_memory_state(cohc->base->dev, + true); + } + + pm->started_channels |= (1ULL << cohc->id); + + spin_unlock_irqrestore(&pm->lock, flags); +} + +static inline int coh901318_set_ctrl(struct coh901318_chan *cohc, u32 control) +{ + int channel = cohc->id; + void __iomem *virtbase = cohc->base->virtbase; + + writel(control, + virtbase + COH901318_CX_CTRL + + COH901318_CX_CTRL_SPACING * channel); + return 0; +} + +static inline int coh901318_set_conf(struct coh901318_chan *cohc, u32 conf) +{ + int channel = cohc->id; + void __iomem *virtbase = cohc->base->virtbase; + + writel(conf, + virtbase + COH901318_CX_CFG + + COH901318_CX_CFG_SPACING*channel); + return 0; +} + + +static int coh901318_start(struct coh901318_chan *cohc) +{ + u32 val; + int channel = cohc->id; + void __iomem *virtbase = cohc->base->virtbase; + + disable_powersave(cohc); + + val = readl(virtbase + COH901318_CX_CFG + + COH901318_CX_CFG_SPACING * channel); + + /* Enable channel */ + val |= COH901318_CX_CFG_CH_ENABLE; + writel(val, virtbase + COH901318_CX_CFG + + COH901318_CX_CFG_SPACING * channel); + + return 0; +} + +static int coh901318_prep_linked_list(struct coh901318_chan *cohc, + struct coh901318_lli *data) +{ + int channel = cohc->id; + void __iomem *virtbase = cohc->base->virtbase; + + BUG_ON(readl(virtbase + COH901318_CX_STAT + + COH901318_CX_STAT_SPACING*channel) & + COH901318_CX_STAT_ACTIVE); + + writel(data->src_addr, + virtbase + COH901318_CX_SRC_ADDR + + COH901318_CX_SRC_ADDR_SPACING * channel); + + writel(data->dst_addr, virtbase + + COH901318_CX_DST_ADDR + + COH901318_CX_DST_ADDR_SPACING * channel); + + writel(data->link_addr, virtbase + COH901318_CX_LNK_ADDR + + COH901318_CX_LNK_ADDR_SPACING * channel); + + writel(data->control, virtbase + COH901318_CX_CTRL + + COH901318_CX_CTRL_SPACING * channel); + + return 0; +} +static dma_cookie_t +coh901318_assign_cookie(struct coh901318_chan *cohc, + struct coh901318_desc *cohd) +{ + dma_cookie_t cookie = cohc->chan.cookie; + + if (++cookie < 0) + cookie = 1; + + cohc->chan.cookie = cookie; + cohd->desc.cookie = cookie; + + return cookie; +} + +static struct coh901318_desc * +coh901318_desc_get(struct coh901318_chan *cohc) +{ + struct coh901318_desc *desc; + + if (list_empty(&cohc->free)) { + /* alloc new desc because we're out of used ones + * TODO: alloc a pile of descs instead of just one, + * avoid many small allocations. + */ + desc = kmalloc(sizeof(struct coh901318_desc), GFP_NOWAIT); + if (desc == NULL) + goto out; + INIT_LIST_HEAD(&desc->node); + } else { + /* Reuse an old desc. */ + desc = list_first_entry(&cohc->free, + struct coh901318_desc, + node); + list_del(&desc->node); + } + + out: + return desc; +} + +static void +coh901318_desc_free(struct coh901318_chan *cohc, struct coh901318_desc *cohd) +{ + list_add_tail(&cohd->node, &cohc->free); +} + +/* call with irq lock held */ +static void +coh901318_desc_submit(struct coh901318_chan *cohc, struct coh901318_desc *desc) +{ + list_add_tail(&desc->node, &cohc->active); + + BUG_ON(cohc->pending_irqs != 0); + + cohc->pending_irqs = desc->pending_irqs; +} + +static struct coh901318_desc * +coh901318_first_active_get(struct coh901318_chan *cohc) +{ + struct coh901318_desc *d; + + if (list_empty(&cohc->active)) + return NULL; + + d = list_first_entry(&cohc->active, + struct coh901318_desc, + node); + return d; +} + +static void +coh901318_desc_remove(struct coh901318_desc *cohd) +{ + list_del(&cohd->node); +} + +static void +coh901318_desc_queue(struct coh901318_chan *cohc, struct coh901318_desc *desc) +{ + list_add_tail(&desc->node, &cohc->queue); +} + +static struct coh901318_desc * +coh901318_first_queued(struct coh901318_chan *cohc) +{ + struct coh901318_desc *d; + + if (list_empty(&cohc->queue)) + return NULL; + + d = list_first_entry(&cohc->queue, + struct coh901318_desc, + node); + return d; +} + +/* + * DMA start/stop controls + */ +u32 coh901318_get_bytes_left(struct dma_chan *chan) +{ + unsigned long flags; + u32 ret; + struct coh901318_chan *cohc = to_coh901318_chan(chan); + + spin_lock_irqsave(&cohc->lock, flags); + + /* Read transfer count value */ + ret = readl(cohc->base->virtbase + + COH901318_CX_CTRL+COH901318_CX_CTRL_SPACING * + cohc->id) & COH901318_CX_CTRL_TC_VALUE_MASK; + + spin_unlock_irqrestore(&cohc->lock, flags); + + return ret; +} +EXPORT_SYMBOL(coh901318_get_bytes_left); + + +/* Stops a transfer without losing data. Enables power save. + Use this function in conjunction with coh901318_continue(..) +*/ +void coh901318_stop(struct dma_chan *chan) +{ + u32 val; + unsigned long flags; + struct coh901318_chan *cohc = to_coh901318_chan(chan); + int channel = cohc->id; + void __iomem *virtbase = cohc->base->virtbase; + + spin_lock_irqsave(&cohc->lock, flags); + + /* Disable channel in HW */ + val = readl(virtbase + COH901318_CX_CFG + + COH901318_CX_CFG_SPACING * channel); + + /* Stopping infinit transfer */ + if ((val & COH901318_CX_CTRL_TC_ENABLE) == 0 && + (val & COH901318_CX_CFG_CH_ENABLE)) + cohc->stopped = 1; + + + val &= ~COH901318_CX_CFG_CH_ENABLE; + /* Enable twice, HW bug work around */ + writel(val, virtbase + COH901318_CX_CFG + + COH901318_CX_CFG_SPACING * channel); + writel(val, virtbase + COH901318_CX_CFG + + COH901318_CX_CFG_SPACING * channel); + + /* Spin-wait for it to actually go inactive */ + while (readl(virtbase + COH901318_CX_STAT+COH901318_CX_STAT_SPACING * + channel) & COH901318_CX_STAT_ACTIVE) + cpu_relax(); + + /* Check if we stopped an active job */ + if ((readl(virtbase + COH901318_CX_CTRL+COH901318_CX_CTRL_SPACING * + channel) & COH901318_CX_CTRL_TC_VALUE_MASK) > 0) + cohc->stopped = 1; + + enable_powersave(cohc); + + spin_unlock_irqrestore(&cohc->lock, flags); +} +EXPORT_SYMBOL(coh901318_stop); + +/* Continues a transfer that has been stopped via 300_dma_stop(..). + Power save is handled. +*/ +void coh901318_continue(struct dma_chan *chan) +{ + u32 val; + unsigned long flags; + struct coh901318_chan *cohc = to_coh901318_chan(chan); + int channel = cohc->id; + + spin_lock_irqsave(&cohc->lock, flags); + + disable_powersave(cohc); + + if (cohc->stopped) { + /* Enable channel in HW */ + val = readl(cohc->base->virtbase + COH901318_CX_CFG + + COH901318_CX_CFG_SPACING * channel); + + val |= COH901318_CX_CFG_CH_ENABLE; + + writel(val, cohc->base->virtbase + COH901318_CX_CFG + + COH901318_CX_CFG_SPACING*channel); + + cohc->stopped = 0; + } + + spin_unlock_irqrestore(&cohc->lock, flags); +} +EXPORT_SYMBOL(coh901318_continue); + +bool coh901318_filter_id(struct dma_chan *chan, void *chan_id) +{ + unsigned int ch_nr = (unsigned int) chan_id; + + if (ch_nr == to_coh901318_chan(chan)->id) + return true; + + return false; +} +EXPORT_SYMBOL(coh901318_filter_id); + +/* + * DMA channel allocation + */ +static int coh901318_config(struct coh901318_chan *cohc, + struct coh901318_params *param) +{ + unsigned long flags; + const struct coh901318_params *p; + int channel = cohc->id; + void __iomem *virtbase = cohc->base->virtbase; + + spin_lock_irqsave(&cohc->lock, flags); + + if (param) + p = param; + else + p = &cohc->base->platform->chan_conf[channel].param; + + /* Clear any pending BE or TC interrupt */ + if (channel < 32) { + writel(1 << channel, virtbase + COH901318_BE_INT_CLEAR1); + writel(1 << channel, virtbase + COH901318_TC_INT_CLEAR1); + } else { + writel(1 << (channel - 32), virtbase + + COH901318_BE_INT_CLEAR2); + writel(1 << (channel - 32), virtbase + + COH901318_TC_INT_CLEAR2); + } + + coh901318_set_conf(cohc, p->config); + coh901318_set_ctrl(cohc, p->ctrl_lli_last); + + spin_unlock_irqrestore(&cohc->lock, flags); + + return 0; +} + +/* must lock when calling this function + * start queued jobs, if any + * TODO: start all queued jobs in one go + * + * Returns descriptor if queued job is started otherwise NULL. + * If the queue is empty NULL is returned. + */ +static struct coh901318_desc *coh901318_queue_start(struct coh901318_chan *cohc) +{ + struct coh901318_desc *cohd_que; + + /* start queued jobs, if any + * TODO: transmit all queued jobs in one go + */ + cohd_que = coh901318_first_queued(cohc); + + if (cohd_que != NULL) { + /* Remove from queue */ + coh901318_desc_remove(cohd_que); + /* initiate DMA job */ + cohc->busy = 1; + + coh901318_desc_submit(cohc, cohd_que); + + coh901318_prep_linked_list(cohc, cohd_que->data); + + /* start dma job */ + coh901318_start(cohc); + + } + + return cohd_que; +} + +static void dma_tasklet(unsigned long data) +{ + struct coh901318_chan *cohc = (struct coh901318_chan *) data; + struct coh901318_desc *cohd_fin; + unsigned long flags; + dma_async_tx_callback callback; + void *callback_param; + + spin_lock_irqsave(&cohc->lock, flags); + + /* get first active entry from list */ + cohd_fin = coh901318_first_active_get(cohc); + + BUG_ON(cohd_fin->pending_irqs == 0); + + if (cohd_fin == NULL) + goto err; + + cohd_fin->pending_irqs--; + cohc->completed = cohd_fin->desc.cookie; + + if (cohc->nbr_active_done == 0) + return; + + if (!cohd_fin->pending_irqs) { + /* release the lli allocation*/ + coh901318_lli_free(&cohc->base->pool, &cohd_fin->data); + } + + dev_vdbg(COHC_2_DEV(cohc), "[%s] chan_id %d pending_irqs %d" + " nbr_active_done %ld\n", __func__, + cohc->id, cohc->pending_irqs, cohc->nbr_active_done); + + /* callback to client */ + callback = cohd_fin->desc.callback; + callback_param = cohd_fin->desc.callback_param; + + if (!cohd_fin->pending_irqs) { + coh901318_desc_remove(cohd_fin); + + /* return desc to free-list */ + coh901318_desc_free(cohc, cohd_fin); + } + + if (cohc->nbr_active_done) + cohc->nbr_active_done--; + + if (cohc->nbr_active_done) { + if (cohc_chan_conf(cohc)->priority_high) + tasklet_hi_schedule(&cohc->tasklet); + else + tasklet_schedule(&cohc->tasklet); + } + spin_unlock_irqrestore(&cohc->lock, flags); + + if (callback) + callback(callback_param); + + return; + + err: + spin_unlock_irqrestore(&cohc->lock, flags); + dev_err(COHC_2_DEV(cohc), "[%s] No active dma desc\n", __func__); +} + + +/* called from interrupt context */ +static void dma_tc_handle(struct coh901318_chan *cohc) +{ + BUG_ON(!cohc->allocated && (list_empty(&cohc->active) || + list_empty(&cohc->queue))); + + if (!cohc->allocated) + return; + + BUG_ON(cohc->pending_irqs == 0); + + cohc->pending_irqs--; + cohc->nbr_active_done++; + + if (cohc->pending_irqs == 0 && coh901318_queue_start(cohc) == NULL) + cohc->busy = 0; + + BUG_ON(list_empty(&cohc->active)); + + if (cohc_chan_conf(cohc)->priority_high) + tasklet_hi_schedule(&cohc->tasklet); + else + tasklet_schedule(&cohc->tasklet); +} + + +static irqreturn_t dma_irq_handler(int irq, void *dev_id) +{ + u32 status1; + u32 status2; + int i; + int ch; + struct coh901318_base *base = dev_id; + struct coh901318_chan *cohc; + void __iomem *virtbase = base->virtbase; + + status1 = readl(virtbase + COH901318_INT_STATUS1); + status2 = readl(virtbase + COH901318_INT_STATUS2); + + if (unlikely(status1 == 0 && status2 == 0)) { + dev_warn(base->dev, "spurious DMA IRQ from no channel!\n"); + return IRQ_HANDLED; + } + + /* TODO: consider handle IRQ in tasklet here to + * minimize interrupt latency */ + + /* Check the first 32 DMA channels for IRQ */ + while (status1) { + /* Find first bit set, return as a number. */ + i = ffs(status1) - 1; + ch = i; + + cohc = &base->chans[ch]; + spin_lock(&cohc->lock); + + /* Mask off this bit */ + status1 &= ~(1 << i); + /* Check the individual channel bits */ + if (test_bit(i, virtbase + COH901318_BE_INT_STATUS1)) { + dev_crit(COHC_2_DEV(cohc), + "DMA bus error on channel %d!\n", ch); + BUG_ON(1); + /* Clear BE interrupt */ + __set_bit(i, virtbase + COH901318_BE_INT_CLEAR1); + } else { + /* Caused by TC, really? */ + if (unlikely(!test_bit(i, virtbase + + COH901318_TC_INT_STATUS1))) { + dev_warn(COHC_2_DEV(cohc), + "ignoring interrupt not caused by terminal count on channel %d\n", ch); + /* Clear TC interrupt */ + BUG_ON(1); + __set_bit(i, virtbase + COH901318_TC_INT_CLEAR1); + } else { + /* Enable powersave if transfer has finished */ + if (!(readl(virtbase + COH901318_CX_STAT + + COH901318_CX_STAT_SPACING*ch) & + COH901318_CX_STAT_ENABLED)) { + enable_powersave(cohc); + } + + /* Must clear TC interrupt before calling + * dma_tc_handle + * in case tc_handle initate a new dma job + */ + __set_bit(i, virtbase + COH901318_TC_INT_CLEAR1); + + dma_tc_handle(cohc); + } + } + spin_unlock(&cohc->lock); + } + + /* Check the remaining 32 DMA channels for IRQ */ + while (status2) { + /* Find first bit set, return as a number. */ + i = ffs(status2) - 1; + ch = i + 32; + cohc = &base->chans[ch]; + spin_lock(&cohc->lock); + + /* Mask off this bit */ + status2 &= ~(1 << i); + /* Check the individual channel bits */ + if (test_bit(i, virtbase + COH901318_BE_INT_STATUS2)) { + dev_crit(COHC_2_DEV(cohc), + "DMA bus error on channel %d!\n", ch); + /* Clear BE interrupt */ + BUG_ON(1); + __set_bit(i, virtbase + COH901318_BE_INT_CLEAR2); + } else { + /* Caused by TC, really? */ + if (unlikely(!test_bit(i, virtbase + + COH901318_TC_INT_STATUS2))) { + dev_warn(COHC_2_DEV(cohc), + "ignoring interrupt not caused by terminal count on channel %d\n", ch); + /* Clear TC interrupt */ + __set_bit(i, virtbase + COH901318_TC_INT_CLEAR2); + BUG_ON(1); + } else { + /* Enable powersave if transfer has finished */ + if (!(readl(virtbase + COH901318_CX_STAT + + COH901318_CX_STAT_SPACING*ch) & + COH901318_CX_STAT_ENABLED)) { + enable_powersave(cohc); + } + /* Must clear TC interrupt before calling + * dma_tc_handle + * in case tc_handle initate a new dma job + */ + __set_bit(i, virtbase + COH901318_TC_INT_CLEAR2); + + dma_tc_handle(cohc); + } + } + spin_unlock(&cohc->lock); + } + + return IRQ_HANDLED; +} + +static int coh901318_alloc_chan_resources(struct dma_chan *chan) +{ + struct coh901318_chan *cohc = to_coh901318_chan(chan); + + dev_vdbg(COHC_2_DEV(cohc), "[%s] DMA channel %d\n", + __func__, cohc->id); + + if (chan->client_count > 1) + return -EBUSY; + + coh901318_config(cohc, NULL); + + cohc->allocated = 1; + cohc->completed = chan->cookie = 1; + + return 1; +} + +static void +coh901318_free_chan_resources(struct dma_chan *chan) +{ + struct coh901318_chan *cohc = to_coh901318_chan(chan); + int channel = cohc->id; + unsigned long flags; + + spin_lock_irqsave(&cohc->lock, flags); + + /* Disable HW */ + writel(0x00000000U, cohc->base->virtbase + COH901318_CX_CFG + + COH901318_CX_CFG_SPACING*channel); + writel(0x00000000U, cohc->base->virtbase + COH901318_CX_CTRL + + COH901318_CX_CTRL_SPACING*channel); + + cohc->allocated = 0; + + spin_unlock_irqrestore(&cohc->lock, flags); + + chan->device->device_terminate_all(chan); +} + + +static dma_cookie_t +coh901318_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct coh901318_desc *cohd = container_of(tx, struct coh901318_desc, + desc); + struct coh901318_chan *cohc = to_coh901318_chan(tx->chan); + unsigned long flags; + + spin_lock_irqsave(&cohc->lock, flags); + + tx->cookie = coh901318_assign_cookie(cohc, cohd); + + coh901318_desc_queue(cohc, cohd); + + spin_unlock_irqrestore(&cohc->lock, flags); + + return tx->cookie; +} + +static struct dma_async_tx_descriptor * +coh901318_prep_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src, + size_t size, unsigned long flags) +{ + struct coh901318_lli *data; + struct coh901318_desc *cohd; + unsigned long flg; + struct coh901318_chan *cohc = to_coh901318_chan(chan); + int lli_len; + u32 ctrl_last = cohc_chan_param(cohc)->ctrl_lli_last; + + spin_lock_irqsave(&cohc->lock, flg); + + dev_vdbg(COHC_2_DEV(cohc), + "[%s] channel %d src 0x%x dest 0x%x size %d\n", + __func__, cohc->id, src, dest, size); + + if (flags & DMA_PREP_INTERRUPT) + /* Trigger interrupt after last lli */ + ctrl_last |= COH901318_CX_CTRL_TC_IRQ_ENABLE; + + lli_len = size >> MAX_DMA_PACKET_SIZE_SHIFT; + if ((lli_len << MAX_DMA_PACKET_SIZE_SHIFT) < size) + lli_len++; + + data = coh901318_lli_alloc(&cohc->base->pool, lli_len); + + if (data == NULL) + goto err; + + cohd = coh901318_desc_get(cohc); + cohd->sg = NULL; + cohd->sg_len = 0; + cohd->data = data; + + cohd->pending_irqs = + coh901318_lli_fill_memcpy( + &cohc->base->pool, data, src, size, dest, + cohc_chan_param(cohc)->ctrl_lli_chained, + ctrl_last); + cohd->flags = flags; + + COH_DBG(coh901318_list_print(cohc, data)); + + dma_async_tx_descriptor_init(&cohd->desc, chan); + + cohd->desc.tx_submit = coh901318_tx_submit; + + spin_unlock_irqrestore(&cohc->lock, flg); + + return &cohd->desc; + err: + spin_unlock_irqrestore(&cohc->lock, flg); + return NULL; +} + +static struct dma_async_tx_descriptor * +coh901318_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, + unsigned int sg_len, enum dma_data_direction direction, + unsigned long flags) +{ + struct coh901318_chan *cohc = to_coh901318_chan(chan); + struct coh901318_lli *data; + struct coh901318_desc *cohd; + struct scatterlist *sg; + int len = 0; + int size; + int i; + u32 ctrl_chained = cohc_chan_param(cohc)->ctrl_lli_chained; + u32 ctrl = cohc_chan_param(cohc)->ctrl_lli; + u32 ctrl_last = cohc_chan_param(cohc)->ctrl_lli_last; + unsigned long flg; + + if (!sgl) + goto out; + if (sgl->length == 0) + goto out; + + spin_lock_irqsave(&cohc->lock, flg); + + dev_vdbg(COHC_2_DEV(cohc), "[%s] sg_len %d dir %d\n", + __func__, sg_len, direction); + + if (flags & DMA_PREP_INTERRUPT) + /* Trigger interrupt after last lli */ + ctrl_last |= COH901318_CX_CTRL_TC_IRQ_ENABLE; + + cohd = coh901318_desc_get(cohc); + cohd->sg = NULL; + cohd->sg_len = 0; + cohd->dir = direction; + + if (direction == DMA_TO_DEVICE) { + u32 tx_flags = COH901318_CX_CTRL_PRDD_SOURCE | + COH901318_CX_CTRL_SRC_ADDR_INC_ENABLE; + + ctrl_chained |= tx_flags; + ctrl_last |= tx_flags; + ctrl |= tx_flags; + } else if (direction == DMA_FROM_DEVICE) { + u32 rx_flags = COH901318_CX_CTRL_PRDD_DEST | + COH901318_CX_CTRL_DST_ADDR_INC_ENABLE; + + ctrl_chained |= rx_flags; + ctrl_last |= rx_flags; + ctrl |= rx_flags; + } else + goto err_direction; + + dma_async_tx_descriptor_init(&cohd->desc, chan); + + cohd->desc.tx_submit = coh901318_tx_submit; + + + /* The dma only supports transmitting packages up to + * MAX_DMA_PACKET_SIZE. Calculate to total number of + * dma elemts required to send the entire sg list + */ + for_each_sg(sgl, sg, sg_len, i) { + unsigned int factor; + size = sg_dma_len(sg); + + if (size <= MAX_DMA_PACKET_SIZE) { + len++; + continue; + } + + factor = size >> MAX_DMA_PACKET_SIZE_SHIFT; + if ((factor << MAX_DMA_PACKET_SIZE_SHIFT) < size) + factor++; + + len += factor; + } + + data = coh901318_lli_alloc(&cohc->base->pool, len); + + if (data == NULL) + goto err_dma_alloc; + + /* initiate allocated data list */ + cohd->pending_irqs = + coh901318_lli_fill_sg(&cohc->base->pool, data, sgl, sg_len, + cohc_dev_addr(cohc), + ctrl_chained, + ctrl, + ctrl_last, + direction, COH901318_CX_CTRL_TC_IRQ_ENABLE); + cohd->data = data; + + cohd->flags = flags; + + COH_DBG(coh901318_list_print(cohc, data)); + + spin_unlock_irqrestore(&cohc->lock, flg); + + return &cohd->desc; + err_dma_alloc: + err_direction: + coh901318_desc_remove(cohd); + coh901318_desc_free(cohc, cohd); + spin_unlock_irqrestore(&cohc->lock, flg); + out: + return NULL; +} + +static enum dma_status +coh901318_is_tx_complete(struct dma_chan *chan, + dma_cookie_t cookie, dma_cookie_t *done, + dma_cookie_t *used) +{ + struct coh901318_chan *cohc = to_coh901318_chan(chan); + dma_cookie_t last_used; + dma_cookie_t last_complete; + int ret; + + last_complete = cohc->completed; + last_used = chan->cookie; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + return ret; +} + +static void +coh901318_issue_pending(struct dma_chan *chan) +{ + struct coh901318_chan *cohc = to_coh901318_chan(chan); + unsigned long flags; + + spin_lock_irqsave(&cohc->lock, flags); + + /* Busy means that pending jobs are already being processed */ + if (!cohc->busy) + coh901318_queue_start(cohc); + + spin_unlock_irqrestore(&cohc->lock, flags); +} + +static void +coh901318_terminate_all(struct dma_chan *chan) +{ + unsigned long flags; + struct coh901318_chan *cohc = to_coh901318_chan(chan); + struct coh901318_desc *cohd; + void __iomem *virtbase = cohc->base->virtbase; + + coh901318_stop(chan); + + spin_lock_irqsave(&cohc->lock, flags); + + /* Clear any pending BE or TC interrupt */ + if (cohc->id < 32) { + writel(1 << cohc->id, virtbase + COH901318_BE_INT_CLEAR1); + writel(1 << cohc->id, virtbase + COH901318_TC_INT_CLEAR1); + } else { + writel(1 << (cohc->id - 32), virtbase + + COH901318_BE_INT_CLEAR2); + writel(1 << (cohc->id - 32), virtbase + + COH901318_TC_INT_CLEAR2); + } + + enable_powersave(cohc); + + while ((cohd = coh901318_first_active_get(cohc))) { + /* release the lli allocation*/ + coh901318_lli_free(&cohc->base->pool, &cohd->data); + + coh901318_desc_remove(cohd); + + /* return desc to free-list */ + coh901318_desc_free(cohc, cohd); + } + + while ((cohd = coh901318_first_queued(cohc))) { + /* release the lli allocation*/ + coh901318_lli_free(&cohc->base->pool, &cohd->data); + + coh901318_desc_remove(cohd); + + /* return desc to free-list */ + coh901318_desc_free(cohc, cohd); + } + + + cohc->nbr_active_done = 0; + cohc->busy = 0; + cohc->pending_irqs = 0; + + spin_unlock_irqrestore(&cohc->lock, flags); +} +void coh901318_base_init(struct dma_device *dma, const int *pick_chans, + struct coh901318_base *base) +{ + int chans_i; + int i = 0; + struct coh901318_chan *cohc; + + INIT_LIST_HEAD(&dma->channels); + + for (chans_i = 0; pick_chans[chans_i] != -1; chans_i += 2) { + for (i = pick_chans[chans_i]; i <= pick_chans[chans_i+1]; i++) { + cohc = &base->chans[i]; + + cohc->base = base; + cohc->chan.device = dma; + cohc->id = i; + + /* TODO: do we really need this lock if only one + * client is connected to each channel? + */ + + spin_lock_init(&cohc->lock); + + cohc->pending_irqs = 0; + cohc->nbr_active_done = 0; + cohc->busy = 0; + INIT_LIST_HEAD(&cohc->free); + INIT_LIST_HEAD(&cohc->active); + INIT_LIST_HEAD(&cohc->queue); + + tasklet_init(&cohc->tasklet, dma_tasklet, + (unsigned long) cohc); + + list_add_tail(&cohc->chan.device_node, + &dma->channels); + } + } +} + +static int __init coh901318_probe(struct platform_device *pdev) +{ + int err = 0; + struct coh901318_platform *pdata; + struct coh901318_base *base; + int irq; + struct resource *io; + + io = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!io) + goto err_get_resource; + + /* Map DMA controller registers to virtual memory */ + if (request_mem_region(io->start, + resource_size(io), + pdev->dev.driver->name) == NULL) { + err = -EBUSY; + goto err_request_mem; + } + + pdata = pdev->dev.platform_data; + if (!pdata) + goto err_no_platformdata; + + base = kmalloc(ALIGN(sizeof(struct coh901318_base), 4) + + pdata->max_channels * + sizeof(struct coh901318_chan), + GFP_KERNEL); + if (!base) + goto err_alloc_coh_dma_channels; + + base->chans = ((void *)base) + ALIGN(sizeof(struct coh901318_base), 4); + + base->virtbase = ioremap(io->start, resource_size(io)); + if (!base->virtbase) { + err = -ENOMEM; + goto err_no_ioremap; + } + + base->dev = &pdev->dev; + base->platform = pdata; + spin_lock_init(&base->pm.lock); + base->pm.started_channels = 0; + + COH901318_DEBUGFS_ASSIGN(debugfs_dma_base, base); + + platform_set_drvdata(pdev, base); + + irq = platform_get_irq(pdev, 0); + if (irq < 0) + goto err_no_irq; + + err = request_irq(irq, dma_irq_handler, IRQF_DISABLED, + "coh901318", base); + if (err) { + dev_crit(&pdev->dev, + "Cannot allocate IRQ for DMA controller!\n"); + goto err_request_irq; + } + + err = coh901318_pool_create(&base->pool, &pdev->dev, + sizeof(struct coh901318_lli), + 32); + if (err) + goto err_pool_create; + + /* init channels for device transfers */ + coh901318_base_init(&base->dma_slave, base->platform->chans_slave, + base); + + dma_cap_zero(base->dma_slave.cap_mask); + dma_cap_set(DMA_SLAVE, base->dma_slave.cap_mask); + + base->dma_slave.device_alloc_chan_resources = coh901318_alloc_chan_resources; + base->dma_slave.device_free_chan_resources = coh901318_free_chan_resources; + base->dma_slave.device_prep_slave_sg = coh901318_prep_slave_sg; + base->dma_slave.device_is_tx_complete = coh901318_is_tx_complete; + base->dma_slave.device_issue_pending = coh901318_issue_pending; + base->dma_slave.device_terminate_all = coh901318_terminate_all; + base->dma_slave.dev = &pdev->dev; + + err = dma_async_device_register(&base->dma_slave); + + if (err) + goto err_register_slave; + + /* init channels for memcpy */ + coh901318_base_init(&base->dma_memcpy, base->platform->chans_memcpy, + base); + + dma_cap_zero(base->dma_memcpy.cap_mask); + dma_cap_set(DMA_MEMCPY, base->dma_memcpy.cap_mask); + + base->dma_memcpy.device_alloc_chan_resources = coh901318_alloc_chan_resources; + base->dma_memcpy.device_free_chan_resources = coh901318_free_chan_resources; + base->dma_memcpy.device_prep_dma_memcpy = coh901318_prep_memcpy; + base->dma_memcpy.device_is_tx_complete = coh901318_is_tx_complete; + base->dma_memcpy.device_issue_pending = coh901318_issue_pending; + base->dma_memcpy.device_terminate_all = coh901318_terminate_all; + base->dma_memcpy.dev = &pdev->dev; + err = dma_async_device_register(&base->dma_memcpy); + + if (err) + goto err_register_memcpy; + + dev_dbg(&pdev->dev, "Initialized COH901318 DMA on virtual base 0x%08x\n", + (u32) base->virtbase); + + return err; + + err_register_memcpy: + dma_async_device_unregister(&base->dma_slave); + err_register_slave: + coh901318_pool_destroy(&base->pool); + err_pool_create: + free_irq(platform_get_irq(pdev, 0), base); + err_request_irq: + err_no_irq: + iounmap(base->virtbase); + err_no_ioremap: + kfree(base); + err_alloc_coh_dma_channels: + err_no_platformdata: + release_mem_region(pdev->resource->start, + resource_size(pdev->resource)); + err_request_mem: + err_get_resource: + return err; +} + +static int __exit coh901318_remove(struct platform_device *pdev) +{ + struct coh901318_base *base = platform_get_drvdata(pdev); + + dma_async_device_unregister(&base->dma_memcpy); + dma_async_device_unregister(&base->dma_slave); + coh901318_pool_destroy(&base->pool); + free_irq(platform_get_irq(pdev, 0), base); + iounmap(base->virtbase); + kfree(base); + release_mem_region(pdev->resource->start, + resource_size(pdev->resource)); + return 0; +} + + +static struct platform_driver coh901318_driver = { + .remove = __exit_p(coh901318_remove), + .driver = { + .name = "coh901318", + }, +}; + +int __init coh901318_init(void) +{ + return platform_driver_probe(&coh901318_driver, coh901318_probe); +} +subsys_initcall(coh901318_init); + +void __exit coh901318_exit(void) +{ + platform_driver_unregister(&coh901318_driver); +} +module_exit(coh901318_exit); + +MODULE_LICENSE("GPL"); +MODULE_AUTHOR("Per Friden"); diff --git a/drivers/dma/coh901318_lli.c b/drivers/dma/coh901318_lli.c new file mode 100644 index 00000000000..f5120f238a4 --- /dev/null +++ b/drivers/dma/coh901318_lli.c @@ -0,0 +1,318 @@ +/* + * driver/dma/coh901318_lli.c + * + * Copyright (C) 2007-2009 ST-Ericsson + * License terms: GNU General Public License (GPL) version 2 + * Support functions for handling lli for dma + * Author: Per Friden <per.friden@stericsson.com> + */ + +#include <linux/dma-mapping.h> +#include <linux/spinlock.h> +#include <linux/dmapool.h> +#include <linux/memory.h> +#include <mach/coh901318.h> + +#include "coh901318_lli.h" + +#if (defined(CONFIG_DEBUG_FS) && defined(CONFIG_U300_DEBUG)) +#define DEBUGFS_POOL_COUNTER_RESET(pool) (pool->debugfs_pool_counter = 0) +#define DEBUGFS_POOL_COUNTER_ADD(pool, add) (pool->debugfs_pool_counter += add) +#else +#define DEBUGFS_POOL_COUNTER_RESET(pool) +#define DEBUGFS_POOL_COUNTER_ADD(pool, add) +#endif + +static struct coh901318_lli * +coh901318_lli_next(struct coh901318_lli *data) +{ + if (data == NULL || data->link_addr == 0) + return NULL; + + return (struct coh901318_lli *) data->virt_link_addr; +} + +int coh901318_pool_create(struct coh901318_pool *pool, + struct device *dev, + size_t size, size_t align) +{ + spin_lock_init(&pool->lock); + pool->dev = dev; + pool->dmapool = dma_pool_create("lli_pool", dev, size, align, 0); + + DEBUGFS_POOL_COUNTER_RESET(pool); + return 0; +} + +int coh901318_pool_destroy(struct coh901318_pool *pool) +{ + + dma_pool_destroy(pool->dmapool); + return 0; +} + +struct coh901318_lli * +coh901318_lli_alloc(struct coh901318_pool *pool, unsigned int len) +{ + int i; + struct coh901318_lli *head; + struct coh901318_lli *lli; + struct coh901318_lli *lli_prev; + dma_addr_t phy; + + if (len == 0) + goto err; + + spin_lock(&pool->lock); + + head = dma_pool_alloc(pool->dmapool, GFP_NOWAIT, &phy); + + if (head == NULL) + goto err; + + DEBUGFS_POOL_COUNTER_ADD(pool, 1); + + lli = head; + lli->phy_this = phy; + + for (i = 1; i < len; i++) { + lli_prev = lli; + + lli = dma_pool_alloc(pool->dmapool, GFP_NOWAIT, &phy); + + if (lli == NULL) + goto err_clean_up; + + DEBUGFS_POOL_COUNTER_ADD(pool, 1); + lli->phy_this = phy; + + lli_prev->link_addr = phy; + lli_prev->virt_link_addr = lli; + } + + lli->link_addr = 0x00000000U; + + spin_unlock(&pool->lock); + + return head; + + err: + spin_unlock(&pool->lock); + return NULL; + + err_clean_up: + lli_prev->link_addr = 0x00000000U; + spin_unlock(&pool->lock); + coh901318_lli_free(pool, &head); + return NULL; +} + +void coh901318_lli_free(struct coh901318_pool *pool, + struct coh901318_lli **lli) +{ + struct coh901318_lli *l; + struct coh901318_lli *next; + + if (lli == NULL) + return; + + l = *lli; + + if (l == NULL) + return; + + spin_lock(&pool->lock); + + while (l->link_addr) { + next = l->virt_link_addr; + dma_pool_free(pool->dmapool, l, l->phy_this); + DEBUGFS_POOL_COUNTER_ADD(pool, -1); + l = next; + } + dma_pool_free(pool->dmapool, l, l->phy_this); + DEBUGFS_POOL_COUNTER_ADD(pool, -1); + + spin_unlock(&pool->lock); + *lli = NULL; +} + +int +coh901318_lli_fill_memcpy(struct coh901318_pool *pool, + struct coh901318_lli *lli, + dma_addr_t source, unsigned int size, + dma_addr_t destination, u32 ctrl_chained, + u32 ctrl_eom) +{ + int s = size; + dma_addr_t src = source; + dma_addr_t dst = destination; + + lli->src_addr = src; + lli->dst_addr = dst; + + while (lli->link_addr) { + lli->control = ctrl_chained | MAX_DMA_PACKET_SIZE; + lli->src_addr = src; + lli->dst_addr = dst; + + s -= MAX_DMA_PACKET_SIZE; + lli = coh901318_lli_next(lli); + + src += MAX_DMA_PACKET_SIZE; + dst += MAX_DMA_PACKET_SIZE; + } + + lli->control = ctrl_eom | s; + lli->src_addr = src; + lli->dst_addr = dst; + + /* One irq per single transfer */ + return 1; +} + +int +coh901318_lli_fill_single(struct coh901318_pool *pool, + struct coh901318_lli *lli, + dma_addr_t buf, unsigned int size, + dma_addr_t dev_addr, u32 ctrl_chained, u32 ctrl_eom, + enum dma_data_direction dir) +{ + int s = size; + dma_addr_t src; + dma_addr_t dst; + + + if (dir == DMA_TO_DEVICE) { + src = buf; + dst = dev_addr; + + } else if (dir == DMA_FROM_DEVICE) { + + src = dev_addr; + dst = buf; + } else { + return -EINVAL; + } + + while (lli->link_addr) { + size_t block_size = MAX_DMA_PACKET_SIZE; + lli->control = ctrl_chained | MAX_DMA_PACKET_SIZE; + + /* If we are on the next-to-final block and there will + * be less than half a DMA packet left for the last + * block, then we want to make this block a little + * smaller to balance the sizes. This is meant to + * avoid too small transfers if the buffer size is + * (MAX_DMA_PACKET_SIZE*N + 1) */ + if (s < (MAX_DMA_PACKET_SIZE + MAX_DMA_PACKET_SIZE/2)) + block_size = MAX_DMA_PACKET_SIZE/2; + + s -= block_size; + lli->src_addr = src; + lli->dst_addr = dst; + + lli = coh901318_lli_next(lli); + + if (dir == DMA_TO_DEVICE) + src += block_size; + else if (dir == DMA_FROM_DEVICE) + dst += block_size; + } + + lli->control = ctrl_eom | s; + lli->src_addr = src; + lli->dst_addr = dst; + + /* One irq per single transfer */ + return 1; +} + +int +coh901318_lli_fill_sg(struct coh901318_pool *pool, + struct coh901318_lli *lli, + struct scatterlist *sgl, unsigned int nents, + dma_addr_t dev_addr, u32 ctrl_chained, u32 ctrl, + u32 ctrl_last, + enum dma_data_direction dir, u32 ctrl_irq_mask) +{ + int i; + struct scatterlist *sg; + u32 ctrl_sg; + dma_addr_t src = 0; + dma_addr_t dst = 0; + int nbr_of_irq = 0; + u32 bytes_to_transfer; + u32 elem_size; + + if (lli == NULL) + goto err; + + spin_lock(&pool->lock); + + if (dir == DMA_TO_DEVICE) + dst = dev_addr; + else if (dir == DMA_FROM_DEVICE) + src = dev_addr; + else + goto err; + + for_each_sg(sgl, sg, nents, i) { + if (sg_is_chain(sg)) { + /* sg continues to the next sg-element don't + * send ctrl_finish until the last + * sg-element in the chain + */ + ctrl_sg = ctrl_chained; + } else if (i == nents - 1) + ctrl_sg = ctrl_last; + else + ctrl_sg = ctrl ? ctrl : ctrl_last; + + + if ((ctrl_sg & ctrl_irq_mask)) + nbr_of_irq++; + + if (dir == DMA_TO_DEVICE) + /* increment source address */ + src = sg_dma_address(sg); + else + /* increment destination address */ + dst = sg_dma_address(sg); + + bytes_to_transfer = sg_dma_len(sg); + + while (bytes_to_transfer) { + u32 val; + + if (bytes_to_transfer > MAX_DMA_PACKET_SIZE) { + elem_size = MAX_DMA_PACKET_SIZE; + val = ctrl_chained; + } else { + elem_size = bytes_to_transfer; + val = ctrl_sg; + } + + lli->control = val | elem_size; + lli->src_addr = src; + lli->dst_addr = dst; + + if (dir == DMA_FROM_DEVICE) + dst += elem_size; + else + src += elem_size; + + BUG_ON(lli->link_addr & 3); + + bytes_to_transfer -= elem_size; + lli = coh901318_lli_next(lli); + } + + } + spin_unlock(&pool->lock); + + /* There can be many IRQs per sg transfer */ + return nbr_of_irq; + err: + spin_unlock(&pool->lock); + return -EINVAL; +} diff --git a/drivers/dma/coh901318_lli.h b/drivers/dma/coh901318_lli.h new file mode 100644 index 00000000000..7bf713b79c6 --- /dev/null +++ b/drivers/dma/coh901318_lli.h @@ -0,0 +1,124 @@ +/* + * driver/dma/coh901318_lli.h + * + * Copyright (C) 2007-2009 ST-Ericsson + * License terms: GNU General Public License (GPL) version 2 + * Support functions for handling lli for coh901318 + * Author: Per Friden <per.friden@stericsson.com> + */ + +#ifndef COH901318_LLI_H +#define COH901318_LLI_H + +#include <mach/coh901318.h> + +struct device; + +struct coh901318_pool { + spinlock_t lock; + struct dma_pool *dmapool; + struct device *dev; + +#ifdef CONFIG_DEBUG_FS + int debugfs_pool_counter; +#endif +}; + +struct device; +/** + * coh901318_pool_create() - Creates an dma pool for lli:s + * @pool: pool handle + * @dev: dma device + * @lli_nbr: number of lli:s in the pool + * @algin: adress alignemtn of lli:s + * returns 0 on success otherwise none zero + */ +int coh901318_pool_create(struct coh901318_pool *pool, + struct device *dev, + size_t lli_nbr, size_t align); + +/** + * coh901318_pool_destroy() - Destroys the dma pool + * @pool: pool handle + * returns 0 on success otherwise none zero + */ +int coh901318_pool_destroy(struct coh901318_pool *pool); + +/** + * coh901318_lli_alloc() - Allocates a linked list + * + * @pool: pool handle + * @len: length to list + * return: none NULL if success otherwise NULL + */ +struct coh901318_lli * +coh901318_lli_alloc(struct coh901318_pool *pool, + unsigned int len); + +/** + * coh901318_lli_free() - Returns the linked list items to the pool + * @pool: pool handle + * @lli: reference to lli pointer to be freed + */ +void coh901318_lli_free(struct coh901318_pool *pool, + struct coh901318_lli **lli); + +/** + * coh901318_lli_fill_memcpy() - Prepares the lli:s for dma memcpy + * @pool: pool handle + * @lli: allocated lli + * @src: src address + * @size: transfer size + * @dst: destination address + * @ctrl_chained: ctrl for chained lli + * @ctrl_last: ctrl for the last lli + * returns number of CPU interrupts for the lli, negative on error. + */ +int +coh901318_lli_fill_memcpy(struct coh901318_pool *pool, + struct coh901318_lli *lli, + dma_addr_t src, unsigned int size, + dma_addr_t dst, u32 ctrl_chained, u32 ctrl_last); + +/** + * coh901318_lli_fill_single() - Prepares the lli:s for dma single transfer + * @pool: pool handle + * @lli: allocated lli + * @buf: transfer buffer + * @size: transfer size + * @dev_addr: address of periphal + * @ctrl_chained: ctrl for chained lli + * @ctrl_last: ctrl for the last lli + * @dir: direction of transfer (to or from device) + * returns number of CPU interrupts for the lli, negative on error. + */ +int +coh901318_lli_fill_single(struct coh901318_pool *pool, + struct coh901318_lli *lli, + dma_addr_t buf, unsigned int size, + dma_addr_t dev_addr, u32 ctrl_chained, u32 ctrl_last, + enum dma_data_direction dir); + +/** + * coh901318_lli_fill_single() - Prepares the lli:s for dma scatter list transfer + * @pool: pool handle + * @lli: allocated lli + * @sg: scatter gather list + * @nents: number of entries in sg + * @dev_addr: address of periphal + * @ctrl_chained: ctrl for chained lli + * @ctrl: ctrl of middle lli + * @ctrl_last: ctrl for the last lli + * @dir: direction of transfer (to or from device) + * @ctrl_irq_mask: ctrl mask for CPU interrupt + * returns number of CPU interrupts for the lli, negative on error. + */ +int +coh901318_lli_fill_sg(struct coh901318_pool *pool, + struct coh901318_lli *lli, + struct scatterlist *sg, unsigned int nents, + dma_addr_t dev_addr, u32 ctrl_chained, + u32 ctrl, u32 ctrl_last, + enum dma_data_direction dir, u32 ctrl_irq_mask); + +#endif /* COH901318_LLI_H */ diff --git a/drivers/dma/dmaengine.c b/drivers/dma/dmaengine.c new file mode 100644 index 00000000000..e7a3230fb7d --- /dev/null +++ b/drivers/dma/dmaengine.c @@ -0,0 +1,1053 @@ +/* + * Copyright(c) 2004 - 2006 Intel Corporation. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the Free + * Software Foundation; either version 2 of the License, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 + * Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * The full GNU General Public License is included in this distribution in the + * file called COPYING. + */ + +/* + * This code implements the DMA subsystem. It provides a HW-neutral interface + * for other kernel code to use asynchronous memory copy capabilities, + * if present, and allows different HW DMA drivers to register as providing + * this capability. + * + * Due to the fact we are accelerating what is already a relatively fast + * operation, the code goes to great lengths to avoid additional overhead, + * such as locking. + * + * LOCKING: + * + * The subsystem keeps a global list of dma_device structs it is protected by a + * mutex, dma_list_mutex. + * + * A subsystem can get access to a channel by calling dmaengine_get() followed + * by dma_find_channel(), or if it has need for an exclusive channel it can call + * dma_request_channel(). Once a channel is allocated a reference is taken + * against its corresponding driver to disable removal. + * + * Each device has a channels list, which runs unlocked but is never modified + * once the device is registered, it's just setup by the driver. + * + * See Documentation/dmaengine.txt for more details + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/mm.h> +#include <linux/device.h> +#include <linux/dmaengine.h> +#include <linux/hardirq.h> +#include <linux/spinlock.h> +#include <linux/percpu.h> +#include <linux/rcupdate.h> +#include <linux/mutex.h> +#include <linux/jiffies.h> +#include <linux/rculist.h> +#include <linux/idr.h> + +static DEFINE_MUTEX(dma_list_mutex); +static LIST_HEAD(dma_device_list); +static long dmaengine_ref_count; +static struct idr dma_idr; + +/* --- sysfs implementation --- */ + +/** + * dev_to_dma_chan - convert a device pointer to the its sysfs container object + * @dev - device node + * + * Must be called under dma_list_mutex + */ +static struct dma_chan *dev_to_dma_chan(struct device *dev) +{ + struct dma_chan_dev *chan_dev; + + chan_dev = container_of(dev, typeof(*chan_dev), device); + return chan_dev->chan; +} + +static ssize_t show_memcpy_count(struct device *dev, struct device_attribute *attr, char *buf) +{ + struct dma_chan *chan; + unsigned long count = 0; + int i; + int err; + + mutex_lock(&dma_list_mutex); + chan = dev_to_dma_chan(dev); + if (chan) { + for_each_possible_cpu(i) + count += per_cpu_ptr(chan->local, i)->memcpy_count; + err = sprintf(buf, "%lu\n", count); + } else + err = -ENODEV; + mutex_unlock(&dma_list_mutex); + + return err; +} + +static ssize_t show_bytes_transferred(struct device *dev, struct device_attribute *attr, + char *buf) +{ + struct dma_chan *chan; + unsigned long count = 0; + int i; + int err; + + mutex_lock(&dma_list_mutex); + chan = dev_to_dma_chan(dev); + if (chan) { + for_each_possible_cpu(i) + count += per_cpu_ptr(chan->local, i)->bytes_transferred; + err = sprintf(buf, "%lu\n", count); + } else + err = -ENODEV; + mutex_unlock(&dma_list_mutex); + + return err; +} + +static ssize_t show_in_use(struct device *dev, struct device_attribute *attr, char *buf) +{ + struct dma_chan *chan; + int err; + + mutex_lock(&dma_list_mutex); + chan = dev_to_dma_chan(dev); + if (chan) + err = sprintf(buf, "%d\n", chan->client_count); + else + err = -ENODEV; + mutex_unlock(&dma_list_mutex); + + return err; +} + +static struct device_attribute dma_attrs[] = { + __ATTR(memcpy_count, S_IRUGO, show_memcpy_count, NULL), + __ATTR(bytes_transferred, S_IRUGO, show_bytes_transferred, NULL), + __ATTR(in_use, S_IRUGO, show_in_use, NULL), + __ATTR_NULL +}; + +static void chan_dev_release(struct device *dev) +{ + struct dma_chan_dev *chan_dev; + + chan_dev = container_of(dev, typeof(*chan_dev), device); + if (atomic_dec_and_test(chan_dev->idr_ref)) { + mutex_lock(&dma_list_mutex); + idr_remove(&dma_idr, chan_dev->dev_id); + mutex_unlock(&dma_list_mutex); + kfree(chan_dev->idr_ref); + } + kfree(chan_dev); +} + +static struct class dma_devclass = { + .name = "dma", + .dev_attrs = dma_attrs, + .dev_release = chan_dev_release, +}; + +/* --- client and device registration --- */ + +#define dma_device_satisfies_mask(device, mask) \ + __dma_device_satisfies_mask((device), &(mask)) +static int +__dma_device_satisfies_mask(struct dma_device *device, dma_cap_mask_t *want) +{ + dma_cap_mask_t has; + + bitmap_and(has.bits, want->bits, device->cap_mask.bits, + DMA_TX_TYPE_END); + return bitmap_equal(want->bits, has.bits, DMA_TX_TYPE_END); +} + +static struct module *dma_chan_to_owner(struct dma_chan *chan) +{ + return chan->device->dev->driver->owner; +} + +/** + * balance_ref_count - catch up the channel reference count + * @chan - channel to balance ->client_count versus dmaengine_ref_count + * + * balance_ref_count must be called under dma_list_mutex + */ +static void balance_ref_count(struct dma_chan *chan) +{ + struct module *owner = dma_chan_to_owner(chan); + + while (chan->client_count < dmaengine_ref_count) { + __module_get(owner); + chan->client_count++; + } +} + +/** + * dma_chan_get - try to grab a dma channel's parent driver module + * @chan - channel to grab + * + * Must be called under dma_list_mutex + */ +static int dma_chan_get(struct dma_chan *chan) +{ + int err = -ENODEV; + struct module *owner = dma_chan_to_owner(chan); + + if (chan->client_count) { + __module_get(owner); + err = 0; + } else if (try_module_get(owner)) + err = 0; + + if (err == 0) + chan->client_count++; + + /* allocate upon first client reference */ + if (chan->client_count == 1 && err == 0) { + int desc_cnt = chan->device->device_alloc_chan_resources(chan); + + if (desc_cnt < 0) { + err = desc_cnt; + chan->client_count = 0; + module_put(owner); + } else if (!dma_has_cap(DMA_PRIVATE, chan->device->cap_mask)) + balance_ref_count(chan); + } + + return err; +} + +/** + * dma_chan_put - drop a reference to a dma channel's parent driver module + * @chan - channel to release + * + * Must be called under dma_list_mutex + */ +static void dma_chan_put(struct dma_chan *chan) +{ + if (!chan->client_count) + return; /* this channel failed alloc_chan_resources */ + chan->client_count--; + module_put(dma_chan_to_owner(chan)); + if (chan->client_count == 0) + chan->device->device_free_chan_resources(chan); +} + +enum dma_status dma_sync_wait(struct dma_chan *chan, dma_cookie_t cookie) +{ + enum dma_status status; + unsigned long dma_sync_wait_timeout = jiffies + msecs_to_jiffies(5000); + + dma_async_issue_pending(chan); + do { + status = dma_async_is_tx_complete(chan, cookie, NULL, NULL); + if (time_after_eq(jiffies, dma_sync_wait_timeout)) { + printk(KERN_ERR "dma_sync_wait_timeout!\n"); + return DMA_ERROR; + } + } while (status == DMA_IN_PROGRESS); + + return status; +} +EXPORT_SYMBOL(dma_sync_wait); + +/** + * dma_cap_mask_all - enable iteration over all operation types + */ +static dma_cap_mask_t dma_cap_mask_all; + +/** + * dma_chan_tbl_ent - tracks channel allocations per core/operation + * @chan - associated channel for this entry + */ +struct dma_chan_tbl_ent { + struct dma_chan *chan; +}; + +/** + * channel_table - percpu lookup table for memory-to-memory offload providers + */ +static struct dma_chan_tbl_ent *channel_table[DMA_TX_TYPE_END]; + +static int __init dma_channel_table_init(void) +{ + enum dma_transaction_type cap; + int err = 0; + + bitmap_fill(dma_cap_mask_all.bits, DMA_TX_TYPE_END); + + /* 'interrupt', 'private', and 'slave' are channel capabilities, + * but are not associated with an operation so they do not need + * an entry in the channel_table + */ + clear_bit(DMA_INTERRUPT, dma_cap_mask_all.bits); + clear_bit(DMA_PRIVATE, dma_cap_mask_all.bits); + clear_bit(DMA_SLAVE, dma_cap_mask_all.bits); + + for_each_dma_cap_mask(cap, dma_cap_mask_all) { + channel_table[cap] = alloc_percpu(struct dma_chan_tbl_ent); + if (!channel_table[cap]) { + err = -ENOMEM; + break; + } + } + + if (err) { + pr_err("dmaengine: initialization failure\n"); + for_each_dma_cap_mask(cap, dma_cap_mask_all) + if (channel_table[cap]) + free_percpu(channel_table[cap]); + } + + return err; +} +arch_initcall(dma_channel_table_init); + +/** + * dma_find_channel - find a channel to carry out the operation + * @tx_type: transaction type + */ +struct dma_chan *dma_find_channel(enum dma_transaction_type tx_type) +{ + return this_cpu_read(channel_table[tx_type]->chan); +} +EXPORT_SYMBOL(dma_find_channel); + +/** + * dma_issue_pending_all - flush all pending operations across all channels + */ +void dma_issue_pending_all(void) +{ + struct dma_device *device; + struct dma_chan *chan; + + rcu_read_lock(); + list_for_each_entry_rcu(device, &dma_device_list, global_node) { + if (dma_has_cap(DMA_PRIVATE, device->cap_mask)) + continue; + list_for_each_entry(chan, &device->channels, device_node) + if (chan->client_count) + device->device_issue_pending(chan); + } + rcu_read_unlock(); +} +EXPORT_SYMBOL(dma_issue_pending_all); + +/** + * nth_chan - returns the nth channel of the given capability + * @cap: capability to match + * @n: nth channel desired + * + * Defaults to returning the channel with the desired capability and the + * lowest reference count when 'n' cannot be satisfied. Must be called + * under dma_list_mutex. + */ +static struct dma_chan *nth_chan(enum dma_transaction_type cap, int n) +{ + struct dma_device *device; + struct dma_chan *chan; + struct dma_chan *ret = NULL; + struct dma_chan *min = NULL; + + list_for_each_entry(device, &dma_device_list, global_node) { + if (!dma_has_cap(cap, device->cap_mask) || + dma_has_cap(DMA_PRIVATE, device->cap_mask)) + continue; + list_for_each_entry(chan, &device->channels, device_node) { + if (!chan->client_count) + continue; + if (!min) + min = chan; + else if (chan->table_count < min->table_count) + min = chan; + + if (n-- == 0) { + ret = chan; + break; /* done */ + } + } + if (ret) + break; /* done */ + } + + if (!ret) + ret = min; + + if (ret) + ret->table_count++; + + return ret; +} + +/** + * dma_channel_rebalance - redistribute the available channels + * + * Optimize for cpu isolation (each cpu gets a dedicated channel for an + * operation type) in the SMP case, and operation isolation (avoid + * multi-tasking channels) in the non-SMP case. Must be called under + * dma_list_mutex. + */ +static void dma_channel_rebalance(void) +{ + struct dma_chan *chan; + struct dma_device *device; + int cpu; + int cap; + int n; + + /* undo the last distribution */ + for_each_dma_cap_mask(cap, dma_cap_mask_all) + for_each_possible_cpu(cpu) + per_cpu_ptr(channel_table[cap], cpu)->chan = NULL; + + list_for_each_entry(device, &dma_device_list, global_node) { + if (dma_has_cap(DMA_PRIVATE, device->cap_mask)) + continue; + list_for_each_entry(chan, &device->channels, device_node) + chan->table_count = 0; + } + + /* don't populate the channel_table if no clients are available */ + if (!dmaengine_ref_count) + return; + + /* redistribute available channels */ + n = 0; + for_each_dma_cap_mask(cap, dma_cap_mask_all) + for_each_online_cpu(cpu) { + if (num_possible_cpus() > 1) + chan = nth_chan(cap, n++); + else + chan = nth_chan(cap, -1); + + per_cpu_ptr(channel_table[cap], cpu)->chan = chan; + } +} + +static struct dma_chan *private_candidate(dma_cap_mask_t *mask, struct dma_device *dev, + dma_filter_fn fn, void *fn_param) +{ + struct dma_chan *chan; + + if (!__dma_device_satisfies_mask(dev, mask)) { + pr_debug("%s: wrong capabilities\n", __func__); + return NULL; + } + /* devices with multiple channels need special handling as we need to + * ensure that all channels are either private or public. + */ + if (dev->chancnt > 1 && !dma_has_cap(DMA_PRIVATE, dev->cap_mask)) + list_for_each_entry(chan, &dev->channels, device_node) { + /* some channels are already publicly allocated */ + if (chan->client_count) + return NULL; + } + + list_for_each_entry(chan, &dev->channels, device_node) { + if (chan->client_count) { + pr_debug("%s: %s busy\n", + __func__, dma_chan_name(chan)); + continue; + } + if (fn && !fn(chan, fn_param)) { + pr_debug("%s: %s filter said false\n", + __func__, dma_chan_name(chan)); + continue; + } + return chan; + } + + return NULL; +} + +/** + * dma_request_channel - try to allocate an exclusive channel + * @mask: capabilities that the channel must satisfy + * @fn: optional callback to disposition available channels + * @fn_param: opaque parameter to pass to dma_filter_fn + */ +struct dma_chan *__dma_request_channel(dma_cap_mask_t *mask, dma_filter_fn fn, void *fn_param) +{ + struct dma_device *device, *_d; + struct dma_chan *chan = NULL; + int err; + + /* Find a channel */ + mutex_lock(&dma_list_mutex); + list_for_each_entry_safe(device, _d, &dma_device_list, global_node) { + chan = private_candidate(mask, device, fn, fn_param); + if (chan) { + /* Found a suitable channel, try to grab, prep, and + * return it. We first set DMA_PRIVATE to disable + * balance_ref_count as this channel will not be + * published in the general-purpose allocator + */ + dma_cap_set(DMA_PRIVATE, device->cap_mask); + device->privatecnt++; + err = dma_chan_get(chan); + + if (err == -ENODEV) { + pr_debug("%s: %s module removed\n", __func__, + dma_chan_name(chan)); + list_del_rcu(&device->global_node); + } else if (err) + pr_err("dmaengine: failed to get %s: (%d)\n", + dma_chan_name(chan), err); + else + break; + if (--device->privatecnt == 0) + dma_cap_clear(DMA_PRIVATE, device->cap_mask); + chan->private = NULL; + chan = NULL; + } + } + mutex_unlock(&dma_list_mutex); + + pr_debug("%s: %s (%s)\n", __func__, chan ? "success" : "fail", + chan ? dma_chan_name(chan) : NULL); + + return chan; +} +EXPORT_SYMBOL_GPL(__dma_request_channel); + +void dma_release_channel(struct dma_chan *chan) +{ + mutex_lock(&dma_list_mutex); + WARN_ONCE(chan->client_count != 1, + "chan reference count %d != 1\n", chan->client_count); + dma_chan_put(chan); + /* drop PRIVATE cap enabled by __dma_request_channel() */ + if (--chan->device->privatecnt == 0) + dma_cap_clear(DMA_PRIVATE, chan->device->cap_mask); + chan->private = NULL; + mutex_unlock(&dma_list_mutex); +} +EXPORT_SYMBOL_GPL(dma_release_channel); + +/** + * dmaengine_get - register interest in dma_channels + */ +void dmaengine_get(void) +{ + struct dma_device *device, *_d; + struct dma_chan *chan; + int err; + + mutex_lock(&dma_list_mutex); + dmaengine_ref_count++; + + /* try to grab channels */ + list_for_each_entry_safe(device, _d, &dma_device_list, global_node) { + if (dma_has_cap(DMA_PRIVATE, device->cap_mask)) + continue; + list_for_each_entry(chan, &device->channels, device_node) { + err = dma_chan_get(chan); + if (err == -ENODEV) { + /* module removed before we could use it */ + list_del_rcu(&device->global_node); + break; + } else if (err) + pr_err("dmaengine: failed to get %s: (%d)\n", + dma_chan_name(chan), err); + } + } + + /* if this is the first reference and there were channels + * waiting we need to rebalance to get those channels + * incorporated into the channel table + */ + if (dmaengine_ref_count == 1) + dma_channel_rebalance(); + mutex_unlock(&dma_list_mutex); +} +EXPORT_SYMBOL(dmaengine_get); + +/** + * dmaengine_put - let dma drivers be removed when ref_count == 0 + */ +void dmaengine_put(void) +{ + struct dma_device *device; + struct dma_chan *chan; + + mutex_lock(&dma_list_mutex); + dmaengine_ref_count--; + BUG_ON(dmaengine_ref_count < 0); + /* drop channel references */ + list_for_each_entry(device, &dma_device_list, global_node) { + if (dma_has_cap(DMA_PRIVATE, device->cap_mask)) + continue; + list_for_each_entry(chan, &device->channels, device_node) + dma_chan_put(chan); + } + mutex_unlock(&dma_list_mutex); +} +EXPORT_SYMBOL(dmaengine_put); + +static bool device_has_all_tx_types(struct dma_device *device) +{ + /* A device that satisfies this test has channels that will never cause + * an async_tx channel switch event as all possible operation types can + * be handled. + */ + #ifdef CONFIG_ASYNC_TX_DMA + if (!dma_has_cap(DMA_INTERRUPT, device->cap_mask)) + return false; + #endif + + #if defined(CONFIG_ASYNC_MEMCPY) || defined(CONFIG_ASYNC_MEMCPY_MODULE) + if (!dma_has_cap(DMA_MEMCPY, device->cap_mask)) + return false; + #endif + + #if defined(CONFIG_ASYNC_MEMSET) || defined(CONFIG_ASYNC_MEMSET_MODULE) + if (!dma_has_cap(DMA_MEMSET, device->cap_mask)) + return false; + #endif + + #if defined(CONFIG_ASYNC_XOR) || defined(CONFIG_ASYNC_XOR_MODULE) + if (!dma_has_cap(DMA_XOR, device->cap_mask)) + return false; + + #ifndef CONFIG_ASYNC_TX_DISABLE_XOR_VAL_DMA + if (!dma_has_cap(DMA_XOR_VAL, device->cap_mask)) + return false; + #endif + #endif + + #if defined(CONFIG_ASYNC_PQ) || defined(CONFIG_ASYNC_PQ_MODULE) + if (!dma_has_cap(DMA_PQ, device->cap_mask)) + return false; + + #ifndef CONFIG_ASYNC_TX_DISABLE_PQ_VAL_DMA + if (!dma_has_cap(DMA_PQ_VAL, device->cap_mask)) + return false; + #endif + #endif + + return true; +} + +static int get_dma_id(struct dma_device *device) +{ + int rc; + + idr_retry: + if (!idr_pre_get(&dma_idr, GFP_KERNEL)) + return -ENOMEM; + mutex_lock(&dma_list_mutex); + rc = idr_get_new(&dma_idr, NULL, &device->dev_id); + mutex_unlock(&dma_list_mutex); + if (rc == -EAGAIN) + goto idr_retry; + else if (rc != 0) + return rc; + + return 0; +} + +/** + * dma_async_device_register - registers DMA devices found + * @device: &dma_device + */ +int dma_async_device_register(struct dma_device *device) +{ + int chancnt = 0, rc; + struct dma_chan* chan; + atomic_t *idr_ref; + + if (!device) + return -ENODEV; + + /* validate device routines */ + BUG_ON(dma_has_cap(DMA_MEMCPY, device->cap_mask) && + !device->device_prep_dma_memcpy); + BUG_ON(dma_has_cap(DMA_XOR, device->cap_mask) && + !device->device_prep_dma_xor); + BUG_ON(dma_has_cap(DMA_XOR_VAL, device->cap_mask) && + !device->device_prep_dma_xor_val); + BUG_ON(dma_has_cap(DMA_PQ, device->cap_mask) && + !device->device_prep_dma_pq); + BUG_ON(dma_has_cap(DMA_PQ_VAL, device->cap_mask) && + !device->device_prep_dma_pq_val); + BUG_ON(dma_has_cap(DMA_MEMSET, device->cap_mask) && + !device->device_prep_dma_memset); + BUG_ON(dma_has_cap(DMA_INTERRUPT, device->cap_mask) && + !device->device_prep_dma_interrupt); + BUG_ON(dma_has_cap(DMA_SLAVE, device->cap_mask) && + !device->device_prep_slave_sg); + BUG_ON(dma_has_cap(DMA_SLAVE, device->cap_mask) && + !device->device_terminate_all); + + BUG_ON(!device->device_alloc_chan_resources); + BUG_ON(!device->device_free_chan_resources); + BUG_ON(!device->device_is_tx_complete); + BUG_ON(!device->device_issue_pending); + BUG_ON(!device->dev); + + /* note: this only matters in the + * CONFIG_ASYNC_TX_DISABLE_CHANNEL_SWITCH=y case + */ + if (device_has_all_tx_types(device)) + dma_cap_set(DMA_ASYNC_TX, device->cap_mask); + + idr_ref = kmalloc(sizeof(*idr_ref), GFP_KERNEL); + if (!idr_ref) + return -ENOMEM; + rc = get_dma_id(device); + if (rc != 0) { + kfree(idr_ref); + return rc; + } + + atomic_set(idr_ref, 0); + + /* represent channels in sysfs. Probably want devs too */ + list_for_each_entry(chan, &device->channels, device_node) { + rc = -ENOMEM; + chan->local = alloc_percpu(typeof(*chan->local)); + if (chan->local == NULL) + goto err_out; + chan->dev = kzalloc(sizeof(*chan->dev), GFP_KERNEL); + if (chan->dev == NULL) { + free_percpu(chan->local); + chan->local = NULL; + goto err_out; + } + + chan->chan_id = chancnt++; + chan->dev->device.class = &dma_devclass; + chan->dev->device.parent = device->dev; + chan->dev->chan = chan; + chan->dev->idr_ref = idr_ref; + chan->dev->dev_id = device->dev_id; + atomic_inc(idr_ref); + dev_set_name(&chan->dev->device, "dma%dchan%d", + device->dev_id, chan->chan_id); + + rc = device_register(&chan->dev->device); + if (rc) { + free_percpu(chan->local); + chan->local = NULL; + kfree(chan->dev); + atomic_dec(idr_ref); + goto err_out; + } + chan->client_count = 0; + } + device->chancnt = chancnt; + + mutex_lock(&dma_list_mutex); + /* take references on public channels */ + if (dmaengine_ref_count && !dma_has_cap(DMA_PRIVATE, device->cap_mask)) + list_for_each_entry(chan, &device->channels, device_node) { + /* if clients are already waiting for channels we need + * to take references on their behalf + */ + if (dma_chan_get(chan) == -ENODEV) { + /* note we can only get here for the first + * channel as the remaining channels are + * guaranteed to get a reference + */ + rc = -ENODEV; + mutex_unlock(&dma_list_mutex); + goto err_out; + } + } + list_add_tail_rcu(&device->global_node, &dma_device_list); + if (dma_has_cap(DMA_PRIVATE, device->cap_mask)) + device->privatecnt++; /* Always private */ + dma_channel_rebalance(); + mutex_unlock(&dma_list_mutex); + + return 0; + +err_out: + /* if we never registered a channel just release the idr */ + if (atomic_read(idr_ref) == 0) { + mutex_lock(&dma_list_mutex); + idr_remove(&dma_idr, device->dev_id); + mutex_unlock(&dma_list_mutex); + kfree(idr_ref); + return rc; + } + + list_for_each_entry(chan, &device->channels, device_node) { + if (chan->local == NULL) + continue; + mutex_lock(&dma_list_mutex); + chan->dev->chan = NULL; + mutex_unlock(&dma_list_mutex); + device_unregister(&chan->dev->device); + free_percpu(chan->local); + } + return rc; +} +EXPORT_SYMBOL(dma_async_device_register); + +/** + * dma_async_device_unregister - unregister a DMA device + * @device: &dma_device + * + * This routine is called by dma driver exit routines, dmaengine holds module + * references to prevent it being called while channels are in use. + */ +void dma_async_device_unregister(struct dma_device *device) +{ + struct dma_chan *chan; + + mutex_lock(&dma_list_mutex); + list_del_rcu(&device->global_node); + dma_channel_rebalance(); + mutex_unlock(&dma_list_mutex); + + list_for_each_entry(chan, &device->channels, device_node) { + WARN_ONCE(chan->client_count, + "%s called while %d clients hold a reference\n", + __func__, chan->client_count); + mutex_lock(&dma_list_mutex); + chan->dev->chan = NULL; + mutex_unlock(&dma_list_mutex); + device_unregister(&chan->dev->device); + free_percpu(chan->local); + } +} +EXPORT_SYMBOL(dma_async_device_unregister); + +/** + * dma_async_memcpy_buf_to_buf - offloaded copy between virtual addresses + * @chan: DMA channel to offload copy to + * @dest: destination address (virtual) + * @src: source address (virtual) + * @len: length + * + * Both @dest and @src must be mappable to a bus address according to the + * DMA mapping API rules for streaming mappings. + * Both @dest and @src must stay memory resident (kernel memory or locked + * user space pages). + */ +dma_cookie_t +dma_async_memcpy_buf_to_buf(struct dma_chan *chan, void *dest, + void *src, size_t len) +{ + struct dma_device *dev = chan->device; + struct dma_async_tx_descriptor *tx; + dma_addr_t dma_dest, dma_src; + dma_cookie_t cookie; + unsigned long flags; + + dma_src = dma_map_single(dev->dev, src, len, DMA_TO_DEVICE); + dma_dest = dma_map_single(dev->dev, dest, len, DMA_FROM_DEVICE); + flags = DMA_CTRL_ACK | + DMA_COMPL_SRC_UNMAP_SINGLE | + DMA_COMPL_DEST_UNMAP_SINGLE; + tx = dev->device_prep_dma_memcpy(chan, dma_dest, dma_src, len, flags); + + if (!tx) { + dma_unmap_single(dev->dev, dma_src, len, DMA_TO_DEVICE); + dma_unmap_single(dev->dev, dma_dest, len, DMA_FROM_DEVICE); + return -ENOMEM; + } + + tx->callback = NULL; + cookie = tx->tx_submit(tx); + + preempt_disable(); + __this_cpu_add(chan->local->bytes_transferred, len); + __this_cpu_inc(chan->local->memcpy_count); + preempt_enable(); + + return cookie; +} +EXPORT_SYMBOL(dma_async_memcpy_buf_to_buf); + +/** + * dma_async_memcpy_buf_to_pg - offloaded copy from address to page + * @chan: DMA channel to offload copy to + * @page: destination page + * @offset: offset in page to copy to + * @kdata: source address (virtual) + * @len: length + * + * Both @page/@offset and @kdata must be mappable to a bus address according + * to the DMA mapping API rules for streaming mappings. + * Both @page/@offset and @kdata must stay memory resident (kernel memory or + * locked user space pages) + */ +dma_cookie_t +dma_async_memcpy_buf_to_pg(struct dma_chan *chan, struct page *page, + unsigned int offset, void *kdata, size_t len) +{ + struct dma_device *dev = chan->device; + struct dma_async_tx_descriptor *tx; + dma_addr_t dma_dest, dma_src; + dma_cookie_t cookie; + unsigned long flags; + + dma_src = dma_map_single(dev->dev, kdata, len, DMA_TO_DEVICE); + dma_dest = dma_map_page(dev->dev, page, offset, len, DMA_FROM_DEVICE); + flags = DMA_CTRL_ACK | DMA_COMPL_SRC_UNMAP_SINGLE; + tx = dev->device_prep_dma_memcpy(chan, dma_dest, dma_src, len, flags); + + if (!tx) { + dma_unmap_single(dev->dev, dma_src, len, DMA_TO_DEVICE); + dma_unmap_page(dev->dev, dma_dest, len, DMA_FROM_DEVICE); + return -ENOMEM; + } + + tx->callback = NULL; + cookie = tx->tx_submit(tx); + + preempt_disable(); + __this_cpu_add(chan->local->bytes_transferred, len); + __this_cpu_inc(chan->local->memcpy_count); + preempt_enable(); + + return cookie; +} +EXPORT_SYMBOL(dma_async_memcpy_buf_to_pg); + +/** + * dma_async_memcpy_pg_to_pg - offloaded copy from page to page + * @chan: DMA channel to offload copy to + * @dest_pg: destination page + * @dest_off: offset in page to copy to + * @src_pg: source page + * @src_off: offset in page to copy from + * @len: length + * + * Both @dest_page/@dest_off and @src_page/@src_off must be mappable to a bus + * address according to the DMA mapping API rules for streaming mappings. + * Both @dest_page/@dest_off and @src_page/@src_off must stay memory resident + * (kernel memory or locked user space pages). + */ +dma_cookie_t +dma_async_memcpy_pg_to_pg(struct dma_chan *chan, struct page *dest_pg, + unsigned int dest_off, struct page *src_pg, unsigned int src_off, + size_t len) +{ + struct dma_device *dev = chan->device; + struct dma_async_tx_descriptor *tx; + dma_addr_t dma_dest, dma_src; + dma_cookie_t cookie; + unsigned long flags; + + dma_src = dma_map_page(dev->dev, src_pg, src_off, len, DMA_TO_DEVICE); + dma_dest = dma_map_page(dev->dev, dest_pg, dest_off, len, + DMA_FROM_DEVICE); + flags = DMA_CTRL_ACK; + tx = dev->device_prep_dma_memcpy(chan, dma_dest, dma_src, len, flags); + + if (!tx) { + dma_unmap_page(dev->dev, dma_src, len, DMA_TO_DEVICE); + dma_unmap_page(dev->dev, dma_dest, len, DMA_FROM_DEVICE); + return -ENOMEM; + } + + tx->callback = NULL; + cookie = tx->tx_submit(tx); + + preempt_disable(); + __this_cpu_add(chan->local->bytes_transferred, len); + __this_cpu_inc(chan->local->memcpy_count); + preempt_enable(); + + return cookie; +} +EXPORT_SYMBOL(dma_async_memcpy_pg_to_pg); + +void dma_async_tx_descriptor_init(struct dma_async_tx_descriptor *tx, + struct dma_chan *chan) +{ + tx->chan = chan; + spin_lock_init(&tx->lock); +} +EXPORT_SYMBOL(dma_async_tx_descriptor_init); + +/* dma_wait_for_async_tx - spin wait for a transaction to complete + * @tx: in-flight transaction to wait on + */ +enum dma_status +dma_wait_for_async_tx(struct dma_async_tx_descriptor *tx) +{ + unsigned long dma_sync_wait_timeout = jiffies + msecs_to_jiffies(5000); + + if (!tx) + return DMA_SUCCESS; + + while (tx->cookie == -EBUSY) { + if (time_after_eq(jiffies, dma_sync_wait_timeout)) { + pr_err("%s timeout waiting for descriptor submission\n", + __func__); + return DMA_ERROR; + } + cpu_relax(); + } + return dma_sync_wait(tx->chan, tx->cookie); +} +EXPORT_SYMBOL_GPL(dma_wait_for_async_tx); + +/* dma_run_dependencies - helper routine for dma drivers to process + * (start) dependent operations on their target channel + * @tx: transaction with dependencies + */ +void dma_run_dependencies(struct dma_async_tx_descriptor *tx) +{ + struct dma_async_tx_descriptor *dep = tx->next; + struct dma_async_tx_descriptor *dep_next; + struct dma_chan *chan; + + if (!dep) + return; + + /* we'll submit tx->next now, so clear the link */ + tx->next = NULL; + chan = dep->chan; + + /* keep submitting up until a channel switch is detected + * in that case we will be called again as a result of + * processing the interrupt from async_tx_channel_switch + */ + for (; dep; dep = dep_next) { + spin_lock_bh(&dep->lock); + dep->parent = NULL; + dep_next = dep->next; + if (dep_next && dep_next->chan == chan) + dep->next = NULL; /* ->next will be submitted */ + else + dep_next = NULL; /* submit current dep and terminate */ + spin_unlock_bh(&dep->lock); + + dep->tx_submit(dep); + } + + chan->device->device_issue_pending(chan); +} +EXPORT_SYMBOL_GPL(dma_run_dependencies); + +static int __init dma_bus_init(void) +{ + idr_init(&dma_idr); + mutex_init(&dma_list_mutex); + return class_register(&dma_devclass); +} +arch_initcall(dma_bus_init); + + diff --git a/drivers/dma/dmatest.c b/drivers/dma/dmatest.c new file mode 100644 index 00000000000..948d563941c --- /dev/null +++ b/drivers/dma/dmatest.c @@ -0,0 +1,627 @@ +/* + * DMA Engine test module + * + * Copyright (C) 2007 Atmel Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ +#include <linux/delay.h> +#include <linux/dmaengine.h> +#include <linux/init.h> +#include <linux/kthread.h> +#include <linux/module.h> +#include <linux/moduleparam.h> +#include <linux/random.h> +#include <linux/wait.h> + +static unsigned int test_buf_size = 16384; +module_param(test_buf_size, uint, S_IRUGO); +MODULE_PARM_DESC(test_buf_size, "Size of the memcpy test buffer"); + +static char test_channel[20]; +module_param_string(channel, test_channel, sizeof(test_channel), S_IRUGO); +MODULE_PARM_DESC(channel, "Bus ID of the channel to test (default: any)"); + +static char test_device[20]; +module_param_string(device, test_device, sizeof(test_device), S_IRUGO); +MODULE_PARM_DESC(device, "Bus ID of the DMA Engine to test (default: any)"); + +static unsigned int threads_per_chan = 1; +module_param(threads_per_chan, uint, S_IRUGO); +MODULE_PARM_DESC(threads_per_chan, + "Number of threads to start per channel (default: 1)"); + +static unsigned int max_channels; +module_param(max_channels, uint, S_IRUGO); +MODULE_PARM_DESC(max_channels, + "Maximum number of channels to use (default: all)"); + +static unsigned int iterations; +module_param(iterations, uint, S_IRUGO); +MODULE_PARM_DESC(iterations, + "Iterations before stopping test (default: infinite)"); + +static unsigned int xor_sources = 3; +module_param(xor_sources, uint, S_IRUGO); +MODULE_PARM_DESC(xor_sources, + "Number of xor source buffers (default: 3)"); + +static unsigned int pq_sources = 3; +module_param(pq_sources, uint, S_IRUGO); +MODULE_PARM_DESC(pq_sources, + "Number of p+q source buffers (default: 3)"); + +/* + * Initialization patterns. All bytes in the source buffer has bit 7 + * set, all bytes in the destination buffer has bit 7 cleared. + * + * Bit 6 is set for all bytes which are to be copied by the DMA + * engine. Bit 5 is set for all bytes which are to be overwritten by + * the DMA engine. + * + * The remaining bits are the inverse of a counter which increments by + * one for each byte address. + */ +#define PATTERN_SRC 0x80 +#define PATTERN_DST 0x00 +#define PATTERN_COPY 0x40 +#define PATTERN_OVERWRITE 0x20 +#define PATTERN_COUNT_MASK 0x1f + +struct dmatest_thread { + struct list_head node; + struct task_struct *task; + struct dma_chan *chan; + u8 **srcs; + u8 **dsts; + enum dma_transaction_type type; +}; + +struct dmatest_chan { + struct list_head node; + struct dma_chan *chan; + struct list_head threads; +}; + +/* + * These are protected by dma_list_mutex since they're only used by + * the DMA filter function callback + */ +static LIST_HEAD(dmatest_channels); +static unsigned int nr_channels; + +static bool dmatest_match_channel(struct dma_chan *chan) +{ + if (test_channel[0] == '\0') + return true; + return strcmp(dma_chan_name(chan), test_channel) == 0; +} + +static bool dmatest_match_device(struct dma_device *device) +{ + if (test_device[0] == '\0') + return true; + return strcmp(dev_name(device->dev), test_device) == 0; +} + +static unsigned long dmatest_random(void) +{ + unsigned long buf; + + get_random_bytes(&buf, sizeof(buf)); + return buf; +} + +static void dmatest_init_srcs(u8 **bufs, unsigned int start, unsigned int len) +{ + unsigned int i; + u8 *buf; + + for (; (buf = *bufs); bufs++) { + for (i = 0; i < start; i++) + buf[i] = PATTERN_SRC | (~i & PATTERN_COUNT_MASK); + for ( ; i < start + len; i++) + buf[i] = PATTERN_SRC | PATTERN_COPY + | (~i & PATTERN_COUNT_MASK); + for ( ; i < test_buf_size; i++) + buf[i] = PATTERN_SRC | (~i & PATTERN_COUNT_MASK); + buf++; + } +} + +static void dmatest_init_dsts(u8 **bufs, unsigned int start, unsigned int len) +{ + unsigned int i; + u8 *buf; + + for (; (buf = *bufs); bufs++) { + for (i = 0; i < start; i++) + buf[i] = PATTERN_DST | (~i & PATTERN_COUNT_MASK); + for ( ; i < start + len; i++) + buf[i] = PATTERN_DST | PATTERN_OVERWRITE + | (~i & PATTERN_COUNT_MASK); + for ( ; i < test_buf_size; i++) + buf[i] = PATTERN_DST | (~i & PATTERN_COUNT_MASK); + } +} + +static void dmatest_mismatch(u8 actual, u8 pattern, unsigned int index, + unsigned int counter, bool is_srcbuf) +{ + u8 diff = actual ^ pattern; + u8 expected = pattern | (~counter & PATTERN_COUNT_MASK); + const char *thread_name = current->comm; + + if (is_srcbuf) + pr_warning("%s: srcbuf[0x%x] overwritten!" + " Expected %02x, got %02x\n", + thread_name, index, expected, actual); + else if ((pattern & PATTERN_COPY) + && (diff & (PATTERN_COPY | PATTERN_OVERWRITE))) + pr_warning("%s: dstbuf[0x%x] not copied!" + " Expected %02x, got %02x\n", + thread_name, index, expected, actual); + else if (diff & PATTERN_SRC) + pr_warning("%s: dstbuf[0x%x] was copied!" + " Expected %02x, got %02x\n", + thread_name, index, expected, actual); + else + pr_warning("%s: dstbuf[0x%x] mismatch!" + " Expected %02x, got %02x\n", + thread_name, index, expected, actual); +} + +static unsigned int dmatest_verify(u8 **bufs, unsigned int start, + unsigned int end, unsigned int counter, u8 pattern, + bool is_srcbuf) +{ + unsigned int i; + unsigned int error_count = 0; + u8 actual; + u8 expected; + u8 *buf; + unsigned int counter_orig = counter; + + for (; (buf = *bufs); bufs++) { + counter = counter_orig; + for (i = start; i < end; i++) { + actual = buf[i]; + expected = pattern | (~counter & PATTERN_COUNT_MASK); + if (actual != expected) { + if (error_count < 32) + dmatest_mismatch(actual, pattern, i, + counter, is_srcbuf); + error_count++; + } + counter++; + } + } + + if (error_count > 32) + pr_warning("%s: %u errors suppressed\n", + current->comm, error_count - 32); + + return error_count; +} + +static void dmatest_callback(void *completion) +{ + complete(completion); +} + +/* + * This function repeatedly tests DMA transfers of various lengths and + * offsets for a given operation type until it is told to exit by + * kthread_stop(). There may be multiple threads running this function + * in parallel for a single channel, and there may be multiple channels + * being tested in parallel. + * + * Before each test, the source and destination buffer is initialized + * with a known pattern. This pattern is different depending on + * whether it's in an area which is supposed to be copied or + * overwritten, and different in the source and destination buffers. + * So if the DMA engine doesn't copy exactly what we tell it to copy, + * we'll notice. + */ +static int dmatest_func(void *data) +{ + struct dmatest_thread *thread = data; + struct dma_chan *chan; + const char *thread_name; + unsigned int src_off, dst_off, len; + unsigned int error_count; + unsigned int failed_tests = 0; + unsigned int total_tests = 0; + dma_cookie_t cookie; + enum dma_status status; + enum dma_ctrl_flags flags; + u8 pq_coefs[pq_sources]; + int ret; + int src_cnt; + int dst_cnt; + int i; + + thread_name = current->comm; + + ret = -ENOMEM; + + smp_rmb(); + chan = thread->chan; + if (thread->type == DMA_MEMCPY) + src_cnt = dst_cnt = 1; + else if (thread->type == DMA_XOR) { + src_cnt = xor_sources | 1; /* force odd to ensure dst = src */ + dst_cnt = 1; + } else if (thread->type == DMA_PQ) { + src_cnt = pq_sources | 1; /* force odd to ensure dst = src */ + dst_cnt = 2; + for (i = 0; i < pq_sources; i++) + pq_coefs[i] = 1; + } else + goto err_srcs; + + thread->srcs = kcalloc(src_cnt+1, sizeof(u8 *), GFP_KERNEL); + if (!thread->srcs) + goto err_srcs; + for (i = 0; i < src_cnt; i++) { + thread->srcs[i] = kmalloc(test_buf_size, GFP_KERNEL); + if (!thread->srcs[i]) + goto err_srcbuf; + } + thread->srcs[i] = NULL; + + thread->dsts = kcalloc(dst_cnt+1, sizeof(u8 *), GFP_KERNEL); + if (!thread->dsts) + goto err_dsts; + for (i = 0; i < dst_cnt; i++) { + thread->dsts[i] = kmalloc(test_buf_size, GFP_KERNEL); + if (!thread->dsts[i]) + goto err_dstbuf; + } + thread->dsts[i] = NULL; + + set_user_nice(current, 10); + + flags = DMA_CTRL_ACK | DMA_COMPL_SKIP_DEST_UNMAP | DMA_PREP_INTERRUPT; + + while (!kthread_should_stop() + && !(iterations && total_tests >= iterations)) { + struct dma_device *dev = chan->device; + struct dma_async_tx_descriptor *tx = NULL; + dma_addr_t dma_srcs[src_cnt]; + dma_addr_t dma_dsts[dst_cnt]; + struct completion cmp; + unsigned long tmo = msecs_to_jiffies(3000); + u8 align = 0; + + total_tests++; + + /* honor alignment restrictions */ + if (thread->type == DMA_MEMCPY) + align = dev->copy_align; + else if (thread->type == DMA_XOR) + align = dev->xor_align; + else if (thread->type == DMA_PQ) + align = dev->pq_align; + + if (1 << align > test_buf_size) { + pr_err("%u-byte buffer too small for %d-byte alignment\n", + test_buf_size, 1 << align); + break; + } + + len = dmatest_random() % test_buf_size + 1; + len = (len >> align) << align; + if (!len) + len = 1 << align; + src_off = dmatest_random() % (test_buf_size - len + 1); + dst_off = dmatest_random() % (test_buf_size - len + 1); + + src_off = (src_off >> align) << align; + dst_off = (dst_off >> align) << align; + + dmatest_init_srcs(thread->srcs, src_off, len); + dmatest_init_dsts(thread->dsts, dst_off, len); + + for (i = 0; i < src_cnt; i++) { + u8 *buf = thread->srcs[i] + src_off; + + dma_srcs[i] = dma_map_single(dev->dev, buf, len, + DMA_TO_DEVICE); + } + /* map with DMA_BIDIRECTIONAL to force writeback/invalidate */ + for (i = 0; i < dst_cnt; i++) { + dma_dsts[i] = dma_map_single(dev->dev, thread->dsts[i], + test_buf_size, + DMA_BIDIRECTIONAL); + } + + + if (thread->type == DMA_MEMCPY) + tx = dev->device_prep_dma_memcpy(chan, + dma_dsts[0] + dst_off, + dma_srcs[0], len, + flags); + else if (thread->type == DMA_XOR) + tx = dev->device_prep_dma_xor(chan, + dma_dsts[0] + dst_off, + dma_srcs, xor_sources, + len, flags); + else if (thread->type == DMA_PQ) { + dma_addr_t dma_pq[dst_cnt]; + + for (i = 0; i < dst_cnt; i++) + dma_pq[i] = dma_dsts[i] + dst_off; + tx = dev->device_prep_dma_pq(chan, dma_pq, dma_srcs, + pq_sources, pq_coefs, + len, flags); + } + + if (!tx) { + for (i = 0; i < src_cnt; i++) + dma_unmap_single(dev->dev, dma_srcs[i], len, + DMA_TO_DEVICE); + for (i = 0; i < dst_cnt; i++) + dma_unmap_single(dev->dev, dma_dsts[i], + test_buf_size, + DMA_BIDIRECTIONAL); + pr_warning("%s: #%u: prep error with src_off=0x%x " + "dst_off=0x%x len=0x%x\n", + thread_name, total_tests - 1, + src_off, dst_off, len); + msleep(100); + failed_tests++; + continue; + } + + init_completion(&cmp); + tx->callback = dmatest_callback; + tx->callback_param = &cmp; + cookie = tx->tx_submit(tx); + + if (dma_submit_error(cookie)) { + pr_warning("%s: #%u: submit error %d with src_off=0x%x " + "dst_off=0x%x len=0x%x\n", + thread_name, total_tests - 1, cookie, + src_off, dst_off, len); + msleep(100); + failed_tests++; + continue; + } + dma_async_issue_pending(chan); + + tmo = wait_for_completion_timeout(&cmp, tmo); + status = dma_async_is_tx_complete(chan, cookie, NULL, NULL); + + if (tmo == 0) { + pr_warning("%s: #%u: test timed out\n", + thread_name, total_tests - 1); + failed_tests++; + continue; + } else if (status != DMA_SUCCESS) { + pr_warning("%s: #%u: got completion callback," + " but status is \'%s\'\n", + thread_name, total_tests - 1, + status == DMA_ERROR ? "error" : "in progress"); + failed_tests++; + continue; + } + + /* Unmap by myself (see DMA_COMPL_SKIP_DEST_UNMAP above) */ + for (i = 0; i < dst_cnt; i++) + dma_unmap_single(dev->dev, dma_dsts[i], test_buf_size, + DMA_BIDIRECTIONAL); + + error_count = 0; + + pr_debug("%s: verifying source buffer...\n", thread_name); + error_count += dmatest_verify(thread->srcs, 0, src_off, + 0, PATTERN_SRC, true); + error_count += dmatest_verify(thread->srcs, src_off, + src_off + len, src_off, + PATTERN_SRC | PATTERN_COPY, true); + error_count += dmatest_verify(thread->srcs, src_off + len, + test_buf_size, src_off + len, + PATTERN_SRC, true); + + pr_debug("%s: verifying dest buffer...\n", + thread->task->comm); + error_count += dmatest_verify(thread->dsts, 0, dst_off, + 0, PATTERN_DST, false); + error_count += dmatest_verify(thread->dsts, dst_off, + dst_off + len, src_off, + PATTERN_SRC | PATTERN_COPY, false); + error_count += dmatest_verify(thread->dsts, dst_off + len, + test_buf_size, dst_off + len, + PATTERN_DST, false); + + if (error_count) { + pr_warning("%s: #%u: %u errors with " + "src_off=0x%x dst_off=0x%x len=0x%x\n", + thread_name, total_tests - 1, error_count, + src_off, dst_off, len); + failed_tests++; + } else { + pr_debug("%s: #%u: No errors with " + "src_off=0x%x dst_off=0x%x len=0x%x\n", + thread_name, total_tests - 1, + src_off, dst_off, len); + } + } + + ret = 0; + for (i = 0; thread->dsts[i]; i++) + kfree(thread->dsts[i]); +err_dstbuf: + kfree(thread->dsts); +err_dsts: + for (i = 0; thread->srcs[i]; i++) + kfree(thread->srcs[i]); +err_srcbuf: + kfree(thread->srcs); +err_srcs: + pr_notice("%s: terminating after %u tests, %u failures (status %d)\n", + thread_name, total_tests, failed_tests, ret); + + if (iterations > 0) + while (!kthread_should_stop()) { + DECLARE_WAIT_QUEUE_HEAD_ONSTACK(wait_dmatest_exit); + interruptible_sleep_on(&wait_dmatest_exit); + } + + return ret; +} + +static void dmatest_cleanup_channel(struct dmatest_chan *dtc) +{ + struct dmatest_thread *thread; + struct dmatest_thread *_thread; + int ret; + + list_for_each_entry_safe(thread, _thread, &dtc->threads, node) { + ret = kthread_stop(thread->task); + pr_debug("dmatest: thread %s exited with status %d\n", + thread->task->comm, ret); + list_del(&thread->node); + kfree(thread); + } + kfree(dtc); +} + +static int dmatest_add_threads(struct dmatest_chan *dtc, enum dma_transaction_type type) +{ + struct dmatest_thread *thread; + struct dma_chan *chan = dtc->chan; + char *op; + unsigned int i; + + if (type == DMA_MEMCPY) + op = "copy"; + else if (type == DMA_XOR) + op = "xor"; + else if (type == DMA_PQ) + op = "pq"; + else + return -EINVAL; + + for (i = 0; i < threads_per_chan; i++) { + thread = kzalloc(sizeof(struct dmatest_thread), GFP_KERNEL); + if (!thread) { + pr_warning("dmatest: No memory for %s-%s%u\n", + dma_chan_name(chan), op, i); + + break; + } + thread->chan = dtc->chan; + thread->type = type; + smp_wmb(); + thread->task = kthread_run(dmatest_func, thread, "%s-%s%u", + dma_chan_name(chan), op, i); + if (IS_ERR(thread->task)) { + pr_warning("dmatest: Failed to run thread %s-%s%u\n", + dma_chan_name(chan), op, i); + kfree(thread); + break; + } + + /* srcbuf and dstbuf are allocated by the thread itself */ + + list_add_tail(&thread->node, &dtc->threads); + } + + return i; +} + +static int dmatest_add_channel(struct dma_chan *chan) +{ + struct dmatest_chan *dtc; + struct dma_device *dma_dev = chan->device; + unsigned int thread_count = 0; + unsigned int cnt; + + dtc = kmalloc(sizeof(struct dmatest_chan), GFP_KERNEL); + if (!dtc) { + pr_warning("dmatest: No memory for %s\n", dma_chan_name(chan)); + return -ENOMEM; + } + + dtc->chan = chan; + INIT_LIST_HEAD(&dtc->threads); + + if (dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask)) { + cnt = dmatest_add_threads(dtc, DMA_MEMCPY); + thread_count += cnt > 0 ? cnt : 0; + } + if (dma_has_cap(DMA_XOR, dma_dev->cap_mask)) { + cnt = dmatest_add_threads(dtc, DMA_XOR); + thread_count += cnt > 0 ? cnt : 0; + } + if (dma_has_cap(DMA_PQ, dma_dev->cap_mask)) { + cnt = dmatest_add_threads(dtc, DMA_PQ); + thread_count += cnt > 0 ?: 0; + } + + pr_info("dmatest: Started %u threads using %s\n", + thread_count, dma_chan_name(chan)); + + list_add_tail(&dtc->node, &dmatest_channels); + nr_channels++; + + return 0; +} + +static bool filter(struct dma_chan *chan, void *param) +{ + if (!dmatest_match_channel(chan) || !dmatest_match_device(chan->device)) + return false; + else + return true; +} + +static int __init dmatest_init(void) +{ + dma_cap_mask_t mask; + struct dma_chan *chan; + int err = 0; + + dma_cap_zero(mask); + dma_cap_set(DMA_MEMCPY, mask); + for (;;) { + chan = dma_request_channel(mask, filter, NULL); + if (chan) { + err = dmatest_add_channel(chan); + if (err) { + dma_release_channel(chan); + break; /* add_channel failed, punt */ + } + } else + break; /* no more channels available */ + if (max_channels && nr_channels >= max_channels) + break; /* we have all we need */ + } + + return err; +} +/* when compiled-in wait for drivers to load first */ +late_initcall(dmatest_init); + +static void __exit dmatest_exit(void) +{ + struct dmatest_chan *dtc, *_dtc; + struct dma_chan *chan; + + list_for_each_entry_safe(dtc, _dtc, &dmatest_channels, node) { + list_del(&dtc->node); + chan = dtc->chan; + dmatest_cleanup_channel(dtc); + pr_debug("dmatest: dropped channel %s\n", + dma_chan_name(chan)); + dma_release_channel(chan); + } +} +module_exit(dmatest_exit); + +MODULE_AUTHOR("Haavard Skinnemoen <hskinnemoen@atmel.com>"); +MODULE_LICENSE("GPL v2"); diff --git a/drivers/dma/dw_dmac.c b/drivers/dma/dw_dmac.c new file mode 100644 index 00000000000..d28369f7afd --- /dev/null +++ b/drivers/dma/dw_dmac.c @@ -0,0 +1,1456 @@ +/* + * Driver for the Synopsys DesignWare DMA Controller (aka DMACA on + * AVR32 systems.) + * + * Copyright (C) 2007-2008 Atmel Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ +#include <linux/clk.h> +#include <linux/delay.h> +#include <linux/dmaengine.h> +#include <linux/dma-mapping.h> +#include <linux/init.h> +#include <linux/interrupt.h> +#include <linux/io.h> +#include <linux/mm.h> +#include <linux/module.h> +#include <linux/platform_device.h> +#include <linux/slab.h> + +#include "dw_dmac_regs.h" + +/* + * This supports the Synopsys "DesignWare AHB Central DMA Controller", + * (DW_ahb_dmac) which is used with various AMBA 2.0 systems (not all + * of which use ARM any more). See the "Databook" from Synopsys for + * information beyond what licensees probably provide. + * + * The driver has currently been tested only with the Atmel AT32AP7000, + * which does not support descriptor writeback. + */ + +/* NOTE: DMS+SMS is system-specific. We should get this information + * from the platform code somehow. + */ +#define DWC_DEFAULT_CTLLO (DWC_CTLL_DST_MSIZE(0) \ + | DWC_CTLL_SRC_MSIZE(0) \ + | DWC_CTLL_DMS(0) \ + | DWC_CTLL_SMS(1) \ + | DWC_CTLL_LLP_D_EN \ + | DWC_CTLL_LLP_S_EN) + +/* + * This is configuration-dependent and usually a funny size like 4095. + * Let's round it down to the nearest power of two. + * + * Note that this is a transfer count, i.e. if we transfer 32-bit + * words, we can do 8192 bytes per descriptor. + * + * This parameter is also system-specific. + */ +#define DWC_MAX_COUNT 2048U + +/* + * Number of descriptors to allocate for each channel. This should be + * made configurable somehow; preferably, the clients (at least the + * ones using slave transfers) should be able to give us a hint. + */ +#define NR_DESCS_PER_CHANNEL 64 + +/*----------------------------------------------------------------------*/ + +/* + * Because we're not relying on writeback from the controller (it may not + * even be configured into the core!) we don't need to use dma_pool. These + * descriptors -- and associated data -- are cacheable. We do need to make + * sure their dcache entries are written back before handing them off to + * the controller, though. + */ + +static struct device *chan2dev(struct dma_chan *chan) +{ + return &chan->dev->device; +} +static struct device *chan2parent(struct dma_chan *chan) +{ + return chan->dev->device.parent; +} + +static struct dw_desc *dwc_first_active(struct dw_dma_chan *dwc) +{ + return list_entry(dwc->active_list.next, struct dw_desc, desc_node); +} + +static struct dw_desc *dwc_first_queued(struct dw_dma_chan *dwc) +{ + return list_entry(dwc->queue.next, struct dw_desc, desc_node); +} + +static struct dw_desc *dwc_desc_get(struct dw_dma_chan *dwc) +{ + struct dw_desc *desc, *_desc; + struct dw_desc *ret = NULL; + unsigned int i = 0; + + spin_lock_bh(&dwc->lock); + list_for_each_entry_safe(desc, _desc, &dwc->free_list, desc_node) { + if (async_tx_test_ack(&desc->txd)) { + list_del(&desc->desc_node); + ret = desc; + break; + } + dev_dbg(chan2dev(&dwc->chan), "desc %p not ACKed\n", desc); + i++; + } + spin_unlock_bh(&dwc->lock); + + dev_vdbg(chan2dev(&dwc->chan), "scanned %u descriptors on freelist\n", i); + + return ret; +} + +static void dwc_sync_desc_for_cpu(struct dw_dma_chan *dwc, struct dw_desc *desc) +{ + struct dw_desc *child; + + list_for_each_entry(child, &desc->tx_list, desc_node) + dma_sync_single_for_cpu(chan2parent(&dwc->chan), + child->txd.phys, sizeof(child->lli), + DMA_TO_DEVICE); + dma_sync_single_for_cpu(chan2parent(&dwc->chan), + desc->txd.phys, sizeof(desc->lli), + DMA_TO_DEVICE); +} + +/* + * Move a descriptor, including any children, to the free list. + * `desc' must not be on any lists. + */ +static void dwc_desc_put(struct dw_dma_chan *dwc, struct dw_desc *desc) +{ + if (desc) { + struct dw_desc *child; + + dwc_sync_desc_for_cpu(dwc, desc); + + spin_lock_bh(&dwc->lock); + list_for_each_entry(child, &desc->tx_list, desc_node) + dev_vdbg(chan2dev(&dwc->chan), + "moving child desc %p to freelist\n", + child); + list_splice_init(&desc->tx_list, &dwc->free_list); + dev_vdbg(chan2dev(&dwc->chan), "moving desc %p to freelist\n", desc); + list_add(&desc->desc_node, &dwc->free_list); + spin_unlock_bh(&dwc->lock); + } +} + +/* Called with dwc->lock held and bh disabled */ +static dma_cookie_t +dwc_assign_cookie(struct dw_dma_chan *dwc, struct dw_desc *desc) +{ + dma_cookie_t cookie = dwc->chan.cookie; + + if (++cookie < 0) + cookie = 1; + + dwc->chan.cookie = cookie; + desc->txd.cookie = cookie; + + return cookie; +} + +/*----------------------------------------------------------------------*/ + +/* Called with dwc->lock held and bh disabled */ +static void dwc_dostart(struct dw_dma_chan *dwc, struct dw_desc *first) +{ + struct dw_dma *dw = to_dw_dma(dwc->chan.device); + + /* ASSERT: channel is idle */ + if (dma_readl(dw, CH_EN) & dwc->mask) { + dev_err(chan2dev(&dwc->chan), + "BUG: Attempted to start non-idle channel\n"); + dev_err(chan2dev(&dwc->chan), + " SAR: 0x%x DAR: 0x%x LLP: 0x%x CTL: 0x%x:%08x\n", + channel_readl(dwc, SAR), + channel_readl(dwc, DAR), + channel_readl(dwc, LLP), + channel_readl(dwc, CTL_HI), + channel_readl(dwc, CTL_LO)); + + /* The tasklet will hopefully advance the queue... */ + return; + } + + channel_writel(dwc, LLP, first->txd.phys); + channel_writel(dwc, CTL_LO, + DWC_CTLL_LLP_D_EN | DWC_CTLL_LLP_S_EN); + channel_writel(dwc, CTL_HI, 0); + channel_set_bit(dw, CH_EN, dwc->mask); +} + +/*----------------------------------------------------------------------*/ + +static void +dwc_descriptor_complete(struct dw_dma_chan *dwc, struct dw_desc *desc) +{ + dma_async_tx_callback callback; + void *param; + struct dma_async_tx_descriptor *txd = &desc->txd; + + dev_vdbg(chan2dev(&dwc->chan), "descriptor %u complete\n", txd->cookie); + + dwc->completed = txd->cookie; + callback = txd->callback; + param = txd->callback_param; + + dwc_sync_desc_for_cpu(dwc, desc); + list_splice_init(&desc->tx_list, &dwc->free_list); + list_move(&desc->desc_node, &dwc->free_list); + + if (!dwc->chan.private) { + struct device *parent = chan2parent(&dwc->chan); + if (!(txd->flags & DMA_COMPL_SKIP_DEST_UNMAP)) { + if (txd->flags & DMA_COMPL_DEST_UNMAP_SINGLE) + dma_unmap_single(parent, desc->lli.dar, + desc->len, DMA_FROM_DEVICE); + else + dma_unmap_page(parent, desc->lli.dar, + desc->len, DMA_FROM_DEVICE); + } + if (!(txd->flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + if (txd->flags & DMA_COMPL_SRC_UNMAP_SINGLE) + dma_unmap_single(parent, desc->lli.sar, + desc->len, DMA_TO_DEVICE); + else + dma_unmap_page(parent, desc->lli.sar, + desc->len, DMA_TO_DEVICE); + } + } + + /* + * The API requires that no submissions are done from a + * callback, so we don't need to drop the lock here + */ + if (callback) + callback(param); +} + +static void dwc_complete_all(struct dw_dma *dw, struct dw_dma_chan *dwc) +{ + struct dw_desc *desc, *_desc; + LIST_HEAD(list); + + if (dma_readl(dw, CH_EN) & dwc->mask) { + dev_err(chan2dev(&dwc->chan), + "BUG: XFER bit set, but channel not idle!\n"); + + /* Try to continue after resetting the channel... */ + channel_clear_bit(dw, CH_EN, dwc->mask); + while (dma_readl(dw, CH_EN) & dwc->mask) + cpu_relax(); + } + + /* + * Submit queued descriptors ASAP, i.e. before we go through + * the completed ones. + */ + if (!list_empty(&dwc->queue)) + dwc_dostart(dwc, dwc_first_queued(dwc)); + list_splice_init(&dwc->active_list, &list); + list_splice_init(&dwc->queue, &dwc->active_list); + + list_for_each_entry_safe(desc, _desc, &list, desc_node) + dwc_descriptor_complete(dwc, desc); +} + +static void dwc_scan_descriptors(struct dw_dma *dw, struct dw_dma_chan *dwc) +{ + dma_addr_t llp; + struct dw_desc *desc, *_desc; + struct dw_desc *child; + u32 status_xfer; + + /* + * Clear block interrupt flag before scanning so that we don't + * miss any, and read LLP before RAW_XFER to ensure it is + * valid if we decide to scan the list. + */ + dma_writel(dw, CLEAR.BLOCK, dwc->mask); + llp = channel_readl(dwc, LLP); + status_xfer = dma_readl(dw, RAW.XFER); + + if (status_xfer & dwc->mask) { + /* Everything we've submitted is done */ + dma_writel(dw, CLEAR.XFER, dwc->mask); + dwc_complete_all(dw, dwc); + return; + } + + dev_vdbg(chan2dev(&dwc->chan), "scan_descriptors: llp=0x%x\n", llp); + + list_for_each_entry_safe(desc, _desc, &dwc->active_list, desc_node) { + if (desc->lli.llp == llp) + /* This one is currently in progress */ + return; + + list_for_each_entry(child, &desc->tx_list, desc_node) + if (child->lli.llp == llp) + /* Currently in progress */ + return; + + /* + * No descriptors so far seem to be in progress, i.e. + * this one must be done. + */ + dwc_descriptor_complete(dwc, desc); + } + + dev_err(chan2dev(&dwc->chan), + "BUG: All descriptors done, but channel not idle!\n"); + + /* Try to continue after resetting the channel... */ + channel_clear_bit(dw, CH_EN, dwc->mask); + while (dma_readl(dw, CH_EN) & dwc->mask) + cpu_relax(); + + if (!list_empty(&dwc->queue)) { + dwc_dostart(dwc, dwc_first_queued(dwc)); + list_splice_init(&dwc->queue, &dwc->active_list); + } +} + +static void dwc_dump_lli(struct dw_dma_chan *dwc, struct dw_lli *lli) +{ + dev_printk(KERN_CRIT, chan2dev(&dwc->chan), + " desc: s0x%x d0x%x l0x%x c0x%x:%x\n", + lli->sar, lli->dar, lli->llp, + lli->ctlhi, lli->ctllo); +} + +static void dwc_handle_error(struct dw_dma *dw, struct dw_dma_chan *dwc) +{ + struct dw_desc *bad_desc; + struct dw_desc *child; + + dwc_scan_descriptors(dw, dwc); + + /* + * The descriptor currently at the head of the active list is + * borked. Since we don't have any way to report errors, we'll + * just have to scream loudly and try to carry on. + */ + bad_desc = dwc_first_active(dwc); + list_del_init(&bad_desc->desc_node); + list_splice_init(&dwc->queue, dwc->active_list.prev); + + /* Clear the error flag and try to restart the controller */ + dma_writel(dw, CLEAR.ERROR, dwc->mask); + if (!list_empty(&dwc->active_list)) + dwc_dostart(dwc, dwc_first_active(dwc)); + + /* + * KERN_CRITICAL may seem harsh, but since this only happens + * when someone submits a bad physical address in a + * descriptor, we should consider ourselves lucky that the + * controller flagged an error instead of scribbling over + * random memory locations. + */ + dev_printk(KERN_CRIT, chan2dev(&dwc->chan), + "Bad descriptor submitted for DMA!\n"); + dev_printk(KERN_CRIT, chan2dev(&dwc->chan), + " cookie: %d\n", bad_desc->txd.cookie); + dwc_dump_lli(dwc, &bad_desc->lli); + list_for_each_entry(child, &bad_desc->tx_list, desc_node) + dwc_dump_lli(dwc, &child->lli); + + /* Pretend the descriptor completed successfully */ + dwc_descriptor_complete(dwc, bad_desc); +} + +/* --------------------- Cyclic DMA API extensions -------------------- */ + +inline dma_addr_t dw_dma_get_src_addr(struct dma_chan *chan) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + return channel_readl(dwc, SAR); +} +EXPORT_SYMBOL(dw_dma_get_src_addr); + +inline dma_addr_t dw_dma_get_dst_addr(struct dma_chan *chan) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + return channel_readl(dwc, DAR); +} +EXPORT_SYMBOL(dw_dma_get_dst_addr); + +/* called with dwc->lock held and all DMAC interrupts disabled */ +static void dwc_handle_cyclic(struct dw_dma *dw, struct dw_dma_chan *dwc, + u32 status_block, u32 status_err, u32 status_xfer) +{ + if (status_block & dwc->mask) { + void (*callback)(void *param); + void *callback_param; + + dev_vdbg(chan2dev(&dwc->chan), "new cyclic period llp 0x%08x\n", + channel_readl(dwc, LLP)); + dma_writel(dw, CLEAR.BLOCK, dwc->mask); + + callback = dwc->cdesc->period_callback; + callback_param = dwc->cdesc->period_callback_param; + if (callback) { + spin_unlock(&dwc->lock); + callback(callback_param); + spin_lock(&dwc->lock); + } + } + + /* + * Error and transfer complete are highly unlikely, and will most + * likely be due to a configuration error by the user. + */ + if (unlikely(status_err & dwc->mask) || + unlikely(status_xfer & dwc->mask)) { + int i; + + dev_err(chan2dev(&dwc->chan), "cyclic DMA unexpected %s " + "interrupt, stopping DMA transfer\n", + status_xfer ? "xfer" : "error"); + dev_err(chan2dev(&dwc->chan), + " SAR: 0x%x DAR: 0x%x LLP: 0x%x CTL: 0x%x:%08x\n", + channel_readl(dwc, SAR), + channel_readl(dwc, DAR), + channel_readl(dwc, LLP), + channel_readl(dwc, CTL_HI), + channel_readl(dwc, CTL_LO)); + + channel_clear_bit(dw, CH_EN, dwc->mask); + while (dma_readl(dw, CH_EN) & dwc->mask) + cpu_relax(); + + /* make sure DMA does not restart by loading a new list */ + channel_writel(dwc, LLP, 0); + channel_writel(dwc, CTL_LO, 0); + channel_writel(dwc, CTL_HI, 0); + + dma_writel(dw, CLEAR.BLOCK, dwc->mask); + dma_writel(dw, CLEAR.ERROR, dwc->mask); + dma_writel(dw, CLEAR.XFER, dwc->mask); + + for (i = 0; i < dwc->cdesc->periods; i++) + dwc_dump_lli(dwc, &dwc->cdesc->desc[i]->lli); + } +} + +/* ------------------------------------------------------------------------- */ + +static void dw_dma_tasklet(unsigned long data) +{ + struct dw_dma *dw = (struct dw_dma *)data; + struct dw_dma_chan *dwc; + u32 status_block; + u32 status_xfer; + u32 status_err; + int i; + + status_block = dma_readl(dw, RAW.BLOCK); + status_xfer = dma_readl(dw, RAW.XFER); + status_err = dma_readl(dw, RAW.ERROR); + + dev_vdbg(dw->dma.dev, "tasklet: status_block=%x status_err=%x\n", + status_block, status_err); + + for (i = 0; i < dw->dma.chancnt; i++) { + dwc = &dw->chan[i]; + spin_lock(&dwc->lock); + if (test_bit(DW_DMA_IS_CYCLIC, &dwc->flags)) + dwc_handle_cyclic(dw, dwc, status_block, status_err, + status_xfer); + else if (status_err & (1 << i)) + dwc_handle_error(dw, dwc); + else if ((status_block | status_xfer) & (1 << i)) + dwc_scan_descriptors(dw, dwc); + spin_unlock(&dwc->lock); + } + + /* + * Re-enable interrupts. Block Complete interrupts are only + * enabled if the INT_EN bit in the descriptor is set. This + * will trigger a scan before the whole list is done. + */ + channel_set_bit(dw, MASK.XFER, dw->all_chan_mask); + channel_set_bit(dw, MASK.BLOCK, dw->all_chan_mask); + channel_set_bit(dw, MASK.ERROR, dw->all_chan_mask); +} + +static irqreturn_t dw_dma_interrupt(int irq, void *dev_id) +{ + struct dw_dma *dw = dev_id; + u32 status; + + dev_vdbg(dw->dma.dev, "interrupt: status=0x%x\n", + dma_readl(dw, STATUS_INT)); + + /* + * Just disable the interrupts. We'll turn them back on in the + * softirq handler. + */ + channel_clear_bit(dw, MASK.XFER, dw->all_chan_mask); + channel_clear_bit(dw, MASK.BLOCK, dw->all_chan_mask); + channel_clear_bit(dw, MASK.ERROR, dw->all_chan_mask); + + status = dma_readl(dw, STATUS_INT); + if (status) { + dev_err(dw->dma.dev, + "BUG: Unexpected interrupts pending: 0x%x\n", + status); + + /* Try to recover */ + channel_clear_bit(dw, MASK.XFER, (1 << 8) - 1); + channel_clear_bit(dw, MASK.BLOCK, (1 << 8) - 1); + channel_clear_bit(dw, MASK.SRC_TRAN, (1 << 8) - 1); + channel_clear_bit(dw, MASK.DST_TRAN, (1 << 8) - 1); + channel_clear_bit(dw, MASK.ERROR, (1 << 8) - 1); + } + + tasklet_schedule(&dw->tasklet); + + return IRQ_HANDLED; +} + +/*----------------------------------------------------------------------*/ + +static dma_cookie_t dwc_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct dw_desc *desc = txd_to_dw_desc(tx); + struct dw_dma_chan *dwc = to_dw_dma_chan(tx->chan); + dma_cookie_t cookie; + + spin_lock_bh(&dwc->lock); + cookie = dwc_assign_cookie(dwc, desc); + + /* + * REVISIT: We should attempt to chain as many descriptors as + * possible, perhaps even appending to those already submitted + * for DMA. But this is hard to do in a race-free manner. + */ + if (list_empty(&dwc->active_list)) { + dev_vdbg(chan2dev(tx->chan), "tx_submit: started %u\n", + desc->txd.cookie); + dwc_dostart(dwc, desc); + list_add_tail(&desc->desc_node, &dwc->active_list); + } else { + dev_vdbg(chan2dev(tx->chan), "tx_submit: queued %u\n", + desc->txd.cookie); + + list_add_tail(&desc->desc_node, &dwc->queue); + } + + spin_unlock_bh(&dwc->lock); + + return cookie; +} + +static struct dma_async_tx_descriptor * +dwc_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src, + size_t len, unsigned long flags) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + struct dw_desc *desc; + struct dw_desc *first; + struct dw_desc *prev; + size_t xfer_count; + size_t offset; + unsigned int src_width; + unsigned int dst_width; + u32 ctllo; + + dev_vdbg(chan2dev(chan), "prep_dma_memcpy d0x%x s0x%x l0x%zx f0x%lx\n", + dest, src, len, flags); + + if (unlikely(!len)) { + dev_dbg(chan2dev(chan), "prep_dma_memcpy: length is zero!\n"); + return NULL; + } + + /* + * We can be a lot more clever here, but this should take care + * of the most common optimization. + */ + if (!((src | dest | len) & 3)) + src_width = dst_width = 2; + else if (!((src | dest | len) & 1)) + src_width = dst_width = 1; + else + src_width = dst_width = 0; + + ctllo = DWC_DEFAULT_CTLLO + | DWC_CTLL_DST_WIDTH(dst_width) + | DWC_CTLL_SRC_WIDTH(src_width) + | DWC_CTLL_DST_INC + | DWC_CTLL_SRC_INC + | DWC_CTLL_FC_M2M; + prev = first = NULL; + + for (offset = 0; offset < len; offset += xfer_count << src_width) { + xfer_count = min_t(size_t, (len - offset) >> src_width, + DWC_MAX_COUNT); + + desc = dwc_desc_get(dwc); + if (!desc) + goto err_desc_get; + + desc->lli.sar = src + offset; + desc->lli.dar = dest + offset; + desc->lli.ctllo = ctllo; + desc->lli.ctlhi = xfer_count; + + if (!first) { + first = desc; + } else { + prev->lli.llp = desc->txd.phys; + dma_sync_single_for_device(chan2parent(chan), + prev->txd.phys, sizeof(prev->lli), + DMA_TO_DEVICE); + list_add_tail(&desc->desc_node, + &first->tx_list); + } + prev = desc; + } + + + if (flags & DMA_PREP_INTERRUPT) + /* Trigger interrupt after last block */ + prev->lli.ctllo |= DWC_CTLL_INT_EN; + + prev->lli.llp = 0; + dma_sync_single_for_device(chan2parent(chan), + prev->txd.phys, sizeof(prev->lli), + DMA_TO_DEVICE); + + first->txd.flags = flags; + first->len = len; + + return &first->txd; + +err_desc_get: + dwc_desc_put(dwc, first); + return NULL; +} + +static struct dma_async_tx_descriptor * +dwc_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, + unsigned int sg_len, enum dma_data_direction direction, + unsigned long flags) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + struct dw_dma_slave *dws = chan->private; + struct dw_desc *prev; + struct dw_desc *first; + u32 ctllo; + dma_addr_t reg; + unsigned int reg_width; + unsigned int mem_width; + unsigned int i; + struct scatterlist *sg; + size_t total_len = 0; + + dev_vdbg(chan2dev(chan), "prep_dma_slave\n"); + + if (unlikely(!dws || !sg_len)) + return NULL; + + reg_width = dws->reg_width; + prev = first = NULL; + + switch (direction) { + case DMA_TO_DEVICE: + ctllo = (DWC_DEFAULT_CTLLO + | DWC_CTLL_DST_WIDTH(reg_width) + | DWC_CTLL_DST_FIX + | DWC_CTLL_SRC_INC + | DWC_CTLL_FC_M2P); + reg = dws->tx_reg; + for_each_sg(sgl, sg, sg_len, i) { + struct dw_desc *desc; + u32 len; + u32 mem; + + desc = dwc_desc_get(dwc); + if (!desc) { + dev_err(chan2dev(chan), + "not enough descriptors available\n"); + goto err_desc_get; + } + + mem = sg_phys(sg); + len = sg_dma_len(sg); + mem_width = 2; + if (unlikely(mem & 3 || len & 3)) + mem_width = 0; + + desc->lli.sar = mem; + desc->lli.dar = reg; + desc->lli.ctllo = ctllo | DWC_CTLL_SRC_WIDTH(mem_width); + desc->lli.ctlhi = len >> mem_width; + + if (!first) { + first = desc; + } else { + prev->lli.llp = desc->txd.phys; + dma_sync_single_for_device(chan2parent(chan), + prev->txd.phys, + sizeof(prev->lli), + DMA_TO_DEVICE); + list_add_tail(&desc->desc_node, + &first->tx_list); + } + prev = desc; + total_len += len; + } + break; + case DMA_FROM_DEVICE: + ctllo = (DWC_DEFAULT_CTLLO + | DWC_CTLL_SRC_WIDTH(reg_width) + | DWC_CTLL_DST_INC + | DWC_CTLL_SRC_FIX + | DWC_CTLL_FC_P2M); + + reg = dws->rx_reg; + for_each_sg(sgl, sg, sg_len, i) { + struct dw_desc *desc; + u32 len; + u32 mem; + + desc = dwc_desc_get(dwc); + if (!desc) { + dev_err(chan2dev(chan), + "not enough descriptors available\n"); + goto err_desc_get; + } + + mem = sg_phys(sg); + len = sg_dma_len(sg); + mem_width = 2; + if (unlikely(mem & 3 || len & 3)) + mem_width = 0; + + desc->lli.sar = reg; + desc->lli.dar = mem; + desc->lli.ctllo = ctllo | DWC_CTLL_DST_WIDTH(mem_width); + desc->lli.ctlhi = len >> reg_width; + + if (!first) { + first = desc; + } else { + prev->lli.llp = desc->txd.phys; + dma_sync_single_for_device(chan2parent(chan), + prev->txd.phys, + sizeof(prev->lli), + DMA_TO_DEVICE); + list_add_tail(&desc->desc_node, + &first->tx_list); + } + prev = desc; + total_len += len; + } + break; + default: + return NULL; + } + + if (flags & DMA_PREP_INTERRUPT) + /* Trigger interrupt after last block */ + prev->lli.ctllo |= DWC_CTLL_INT_EN; + + prev->lli.llp = 0; + dma_sync_single_for_device(chan2parent(chan), + prev->txd.phys, sizeof(prev->lli), + DMA_TO_DEVICE); + + first->len = total_len; + + return &first->txd; + +err_desc_get: + dwc_desc_put(dwc, first); + return NULL; +} + +static void dwc_terminate_all(struct dma_chan *chan) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + struct dw_dma *dw = to_dw_dma(chan->device); + struct dw_desc *desc, *_desc; + LIST_HEAD(list); + + /* + * This is only called when something went wrong elsewhere, so + * we don't really care about the data. Just disable the + * channel. We still have to poll the channel enable bit due + * to AHB/HSB limitations. + */ + spin_lock_bh(&dwc->lock); + + channel_clear_bit(dw, CH_EN, dwc->mask); + + while (dma_readl(dw, CH_EN) & dwc->mask) + cpu_relax(); + + /* active_list entries will end up before queued entries */ + list_splice_init(&dwc->queue, &list); + list_splice_init(&dwc->active_list, &list); + + spin_unlock_bh(&dwc->lock); + + /* Flush all pending and queued descriptors */ + list_for_each_entry_safe(desc, _desc, &list, desc_node) + dwc_descriptor_complete(dwc, desc); +} + +static enum dma_status +dwc_is_tx_complete(struct dma_chan *chan, + dma_cookie_t cookie, + dma_cookie_t *done, dma_cookie_t *used) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + dma_cookie_t last_used; + dma_cookie_t last_complete; + int ret; + + last_complete = dwc->completed; + last_used = chan->cookie; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + if (ret != DMA_SUCCESS) { + dwc_scan_descriptors(to_dw_dma(chan->device), dwc); + + last_complete = dwc->completed; + last_used = chan->cookie; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + } + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + return ret; +} + +static void dwc_issue_pending(struct dma_chan *chan) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + + spin_lock_bh(&dwc->lock); + if (!list_empty(&dwc->queue)) + dwc_scan_descriptors(to_dw_dma(chan->device), dwc); + spin_unlock_bh(&dwc->lock); +} + +static int dwc_alloc_chan_resources(struct dma_chan *chan) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + struct dw_dma *dw = to_dw_dma(chan->device); + struct dw_desc *desc; + struct dw_dma_slave *dws; + int i; + u32 cfghi; + u32 cfglo; + + dev_vdbg(chan2dev(chan), "alloc_chan_resources\n"); + + /* ASSERT: channel is idle */ + if (dma_readl(dw, CH_EN) & dwc->mask) { + dev_dbg(chan2dev(chan), "DMA channel not idle?\n"); + return -EIO; + } + + dwc->completed = chan->cookie = 1; + + cfghi = DWC_CFGH_FIFO_MODE; + cfglo = 0; + + dws = chan->private; + if (dws) { + /* + * We need controller-specific data to set up slave + * transfers. + */ + BUG_ON(!dws->dma_dev || dws->dma_dev != dw->dma.dev); + + cfghi = dws->cfg_hi; + cfglo = dws->cfg_lo; + } + channel_writel(dwc, CFG_LO, cfglo); + channel_writel(dwc, CFG_HI, cfghi); + + /* + * NOTE: some controllers may have additional features that we + * need to initialize here, like "scatter-gather" (which + * doesn't mean what you think it means), and status writeback. + */ + + spin_lock_bh(&dwc->lock); + i = dwc->descs_allocated; + while (dwc->descs_allocated < NR_DESCS_PER_CHANNEL) { + spin_unlock_bh(&dwc->lock); + + desc = kzalloc(sizeof(struct dw_desc), GFP_KERNEL); + if (!desc) { + dev_info(chan2dev(chan), + "only allocated %d descriptors\n", i); + spin_lock_bh(&dwc->lock); + break; + } + + INIT_LIST_HEAD(&desc->tx_list); + dma_async_tx_descriptor_init(&desc->txd, chan); + desc->txd.tx_submit = dwc_tx_submit; + desc->txd.flags = DMA_CTRL_ACK; + desc->txd.phys = dma_map_single(chan2parent(chan), &desc->lli, + sizeof(desc->lli), DMA_TO_DEVICE); + dwc_desc_put(dwc, desc); + + spin_lock_bh(&dwc->lock); + i = ++dwc->descs_allocated; + } + + /* Enable interrupts */ + channel_set_bit(dw, MASK.XFER, dwc->mask); + channel_set_bit(dw, MASK.BLOCK, dwc->mask); + channel_set_bit(dw, MASK.ERROR, dwc->mask); + + spin_unlock_bh(&dwc->lock); + + dev_dbg(chan2dev(chan), + "alloc_chan_resources allocated %d descriptors\n", i); + + return i; +} + +static void dwc_free_chan_resources(struct dma_chan *chan) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + struct dw_dma *dw = to_dw_dma(chan->device); + struct dw_desc *desc, *_desc; + LIST_HEAD(list); + + dev_dbg(chan2dev(chan), "free_chan_resources (descs allocated=%u)\n", + dwc->descs_allocated); + + /* ASSERT: channel is idle */ + BUG_ON(!list_empty(&dwc->active_list)); + BUG_ON(!list_empty(&dwc->queue)); + BUG_ON(dma_readl(to_dw_dma(chan->device), CH_EN) & dwc->mask); + + spin_lock_bh(&dwc->lock); + list_splice_init(&dwc->free_list, &list); + dwc->descs_allocated = 0; + + /* Disable interrupts */ + channel_clear_bit(dw, MASK.XFER, dwc->mask); + channel_clear_bit(dw, MASK.BLOCK, dwc->mask); + channel_clear_bit(dw, MASK.ERROR, dwc->mask); + + spin_unlock_bh(&dwc->lock); + + list_for_each_entry_safe(desc, _desc, &list, desc_node) { + dev_vdbg(chan2dev(chan), " freeing descriptor %p\n", desc); + dma_unmap_single(chan2parent(chan), desc->txd.phys, + sizeof(desc->lli), DMA_TO_DEVICE); + kfree(desc); + } + + dev_vdbg(chan2dev(chan), "free_chan_resources done\n"); +} + +/* --------------------- Cyclic DMA API extensions -------------------- */ + +/** + * dw_dma_cyclic_start - start the cyclic DMA transfer + * @chan: the DMA channel to start + * + * Must be called with soft interrupts disabled. Returns zero on success or + * -errno on failure. + */ +int dw_dma_cyclic_start(struct dma_chan *chan) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + struct dw_dma *dw = to_dw_dma(dwc->chan.device); + + if (!test_bit(DW_DMA_IS_CYCLIC, &dwc->flags)) { + dev_err(chan2dev(&dwc->chan), "missing prep for cyclic DMA\n"); + return -ENODEV; + } + + spin_lock(&dwc->lock); + + /* assert channel is idle */ + if (dma_readl(dw, CH_EN) & dwc->mask) { + dev_err(chan2dev(&dwc->chan), + "BUG: Attempted to start non-idle channel\n"); + dev_err(chan2dev(&dwc->chan), + " SAR: 0x%x DAR: 0x%x LLP: 0x%x CTL: 0x%x:%08x\n", + channel_readl(dwc, SAR), + channel_readl(dwc, DAR), + channel_readl(dwc, LLP), + channel_readl(dwc, CTL_HI), + channel_readl(dwc, CTL_LO)); + spin_unlock(&dwc->lock); + return -EBUSY; + } + + dma_writel(dw, CLEAR.BLOCK, dwc->mask); + dma_writel(dw, CLEAR.ERROR, dwc->mask); + dma_writel(dw, CLEAR.XFER, dwc->mask); + + /* setup DMAC channel registers */ + channel_writel(dwc, LLP, dwc->cdesc->desc[0]->txd.phys); + channel_writel(dwc, CTL_LO, DWC_CTLL_LLP_D_EN | DWC_CTLL_LLP_S_EN); + channel_writel(dwc, CTL_HI, 0); + + channel_set_bit(dw, CH_EN, dwc->mask); + + spin_unlock(&dwc->lock); + + return 0; +} +EXPORT_SYMBOL(dw_dma_cyclic_start); + +/** + * dw_dma_cyclic_stop - stop the cyclic DMA transfer + * @chan: the DMA channel to stop + * + * Must be called with soft interrupts disabled. + */ +void dw_dma_cyclic_stop(struct dma_chan *chan) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + struct dw_dma *dw = to_dw_dma(dwc->chan.device); + + spin_lock(&dwc->lock); + + channel_clear_bit(dw, CH_EN, dwc->mask); + while (dma_readl(dw, CH_EN) & dwc->mask) + cpu_relax(); + + spin_unlock(&dwc->lock); +} +EXPORT_SYMBOL(dw_dma_cyclic_stop); + +/** + * dw_dma_cyclic_prep - prepare the cyclic DMA transfer + * @chan: the DMA channel to prepare + * @buf_addr: physical DMA address where the buffer starts + * @buf_len: total number of bytes for the entire buffer + * @period_len: number of bytes for each period + * @direction: transfer direction, to or from device + * + * Must be called before trying to start the transfer. Returns a valid struct + * dw_cyclic_desc if successful or an ERR_PTR(-errno) if not successful. + */ +struct dw_cyclic_desc *dw_dma_cyclic_prep(struct dma_chan *chan, + dma_addr_t buf_addr, size_t buf_len, size_t period_len, + enum dma_data_direction direction) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + struct dw_cyclic_desc *cdesc; + struct dw_cyclic_desc *retval = NULL; + struct dw_desc *desc; + struct dw_desc *last = NULL; + struct dw_dma_slave *dws = chan->private; + unsigned long was_cyclic; + unsigned int reg_width; + unsigned int periods; + unsigned int i; + + spin_lock_bh(&dwc->lock); + if (!list_empty(&dwc->queue) || !list_empty(&dwc->active_list)) { + spin_unlock_bh(&dwc->lock); + dev_dbg(chan2dev(&dwc->chan), + "queue and/or active list are not empty\n"); + return ERR_PTR(-EBUSY); + } + + was_cyclic = test_and_set_bit(DW_DMA_IS_CYCLIC, &dwc->flags); + spin_unlock_bh(&dwc->lock); + if (was_cyclic) { + dev_dbg(chan2dev(&dwc->chan), + "channel already prepared for cyclic DMA\n"); + return ERR_PTR(-EBUSY); + } + + retval = ERR_PTR(-EINVAL); + reg_width = dws->reg_width; + periods = buf_len / period_len; + + /* Check for too big/unaligned periods and unaligned DMA buffer. */ + if (period_len > (DWC_MAX_COUNT << reg_width)) + goto out_err; + if (unlikely(period_len & ((1 << reg_width) - 1))) + goto out_err; + if (unlikely(buf_addr & ((1 << reg_width) - 1))) + goto out_err; + if (unlikely(!(direction & (DMA_TO_DEVICE | DMA_FROM_DEVICE)))) + goto out_err; + + retval = ERR_PTR(-ENOMEM); + + if (periods > NR_DESCS_PER_CHANNEL) + goto out_err; + + cdesc = kzalloc(sizeof(struct dw_cyclic_desc), GFP_KERNEL); + if (!cdesc) + goto out_err; + + cdesc->desc = kzalloc(sizeof(struct dw_desc *) * periods, GFP_KERNEL); + if (!cdesc->desc) + goto out_err_alloc; + + for (i = 0; i < periods; i++) { + desc = dwc_desc_get(dwc); + if (!desc) + goto out_err_desc_get; + + switch (direction) { + case DMA_TO_DEVICE: + desc->lli.dar = dws->tx_reg; + desc->lli.sar = buf_addr + (period_len * i); + desc->lli.ctllo = (DWC_DEFAULT_CTLLO + | DWC_CTLL_DST_WIDTH(reg_width) + | DWC_CTLL_SRC_WIDTH(reg_width) + | DWC_CTLL_DST_FIX + | DWC_CTLL_SRC_INC + | DWC_CTLL_FC_M2P + | DWC_CTLL_INT_EN); + break; + case DMA_FROM_DEVICE: + desc->lli.dar = buf_addr + (period_len * i); + desc->lli.sar = dws->rx_reg; + desc->lli.ctllo = (DWC_DEFAULT_CTLLO + | DWC_CTLL_SRC_WIDTH(reg_width) + | DWC_CTLL_DST_WIDTH(reg_width) + | DWC_CTLL_DST_INC + | DWC_CTLL_SRC_FIX + | DWC_CTLL_FC_P2M + | DWC_CTLL_INT_EN); + break; + default: + break; + } + + desc->lli.ctlhi = (period_len >> reg_width); + cdesc->desc[i] = desc; + + if (last) { + last->lli.llp = desc->txd.phys; + dma_sync_single_for_device(chan2parent(chan), + last->txd.phys, sizeof(last->lli), + DMA_TO_DEVICE); + } + + last = desc; + } + + /* lets make a cyclic list */ + last->lli.llp = cdesc->desc[0]->txd.phys; + dma_sync_single_for_device(chan2parent(chan), last->txd.phys, + sizeof(last->lli), DMA_TO_DEVICE); + + dev_dbg(chan2dev(&dwc->chan), "cyclic prepared buf 0x%08x len %zu " + "period %zu periods %d\n", buf_addr, buf_len, + period_len, periods); + + cdesc->periods = periods; + dwc->cdesc = cdesc; + + return cdesc; + +out_err_desc_get: + while (i--) + dwc_desc_put(dwc, cdesc->desc[i]); +out_err_alloc: + kfree(cdesc); +out_err: + clear_bit(DW_DMA_IS_CYCLIC, &dwc->flags); + return (struct dw_cyclic_desc *)retval; +} +EXPORT_SYMBOL(dw_dma_cyclic_prep); + +/** + * dw_dma_cyclic_free - free a prepared cyclic DMA transfer + * @chan: the DMA channel to free + */ +void dw_dma_cyclic_free(struct dma_chan *chan) +{ + struct dw_dma_chan *dwc = to_dw_dma_chan(chan); + struct dw_dma *dw = to_dw_dma(dwc->chan.device); + struct dw_cyclic_desc *cdesc = dwc->cdesc; + int i; + + dev_dbg(chan2dev(&dwc->chan), "cyclic free\n"); + + if (!cdesc) + return; + + spin_lock_bh(&dwc->lock); + + channel_clear_bit(dw, CH_EN, dwc->mask); + while (dma_readl(dw, CH_EN) & dwc->mask) + cpu_relax(); + + dma_writel(dw, CLEAR.BLOCK, dwc->mask); + dma_writel(dw, CLEAR.ERROR, dwc->mask); + dma_writel(dw, CLEAR.XFER, dwc->mask); + + spin_unlock_bh(&dwc->lock); + + for (i = 0; i < cdesc->periods; i++) + dwc_desc_put(dwc, cdesc->desc[i]); + + kfree(cdesc->desc); + kfree(cdesc); + + clear_bit(DW_DMA_IS_CYCLIC, &dwc->flags); +} +EXPORT_SYMBOL(dw_dma_cyclic_free); + +/*----------------------------------------------------------------------*/ + +static void dw_dma_off(struct dw_dma *dw) +{ + dma_writel(dw, CFG, 0); + + channel_clear_bit(dw, MASK.XFER, dw->all_chan_mask); + channel_clear_bit(dw, MASK.BLOCK, dw->all_chan_mask); + channel_clear_bit(dw, MASK.SRC_TRAN, dw->all_chan_mask); + channel_clear_bit(dw, MASK.DST_TRAN, dw->all_chan_mask); + channel_clear_bit(dw, MASK.ERROR, dw->all_chan_mask); + + while (dma_readl(dw, CFG) & DW_CFG_DMA_EN) + cpu_relax(); +} + +static int __init dw_probe(struct platform_device *pdev) +{ + struct dw_dma_platform_data *pdata; + struct resource *io; + struct dw_dma *dw; + size_t size; + int irq; + int err; + int i; + + pdata = pdev->dev.platform_data; + if (!pdata || pdata->nr_channels > DW_DMA_MAX_NR_CHANNELS) + return -EINVAL; + + io = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!io) + return -EINVAL; + + irq = platform_get_irq(pdev, 0); + if (irq < 0) + return irq; + + size = sizeof(struct dw_dma); + size += pdata->nr_channels * sizeof(struct dw_dma_chan); + dw = kzalloc(size, GFP_KERNEL); + if (!dw) + return -ENOMEM; + + if (!request_mem_region(io->start, DW_REGLEN, pdev->dev.driver->name)) { + err = -EBUSY; + goto err_kfree; + } + + dw->regs = ioremap(io->start, DW_REGLEN); + if (!dw->regs) { + err = -ENOMEM; + goto err_release_r; + } + + dw->clk = clk_get(&pdev->dev, "hclk"); + if (IS_ERR(dw->clk)) { + err = PTR_ERR(dw->clk); + goto err_clk; + } + clk_enable(dw->clk); + + /* force dma off, just in case */ + dw_dma_off(dw); + + err = request_irq(irq, dw_dma_interrupt, 0, "dw_dmac", dw); + if (err) + goto err_irq; + + platform_set_drvdata(pdev, dw); + + tasklet_init(&dw->tasklet, dw_dma_tasklet, (unsigned long)dw); + + dw->all_chan_mask = (1 << pdata->nr_channels) - 1; + + INIT_LIST_HEAD(&dw->dma.channels); + for (i = 0; i < pdata->nr_channels; i++, dw->dma.chancnt++) { + struct dw_dma_chan *dwc = &dw->chan[i]; + + dwc->chan.device = &dw->dma; + dwc->chan.cookie = dwc->completed = 1; + dwc->chan.chan_id = i; + list_add_tail(&dwc->chan.device_node, &dw->dma.channels); + + dwc->ch_regs = &__dw_regs(dw)->CHAN[i]; + spin_lock_init(&dwc->lock); + dwc->mask = 1 << i; + + INIT_LIST_HEAD(&dwc->active_list); + INIT_LIST_HEAD(&dwc->queue); + INIT_LIST_HEAD(&dwc->free_list); + + channel_clear_bit(dw, CH_EN, dwc->mask); + } + + /* Clear/disable all interrupts on all channels. */ + dma_writel(dw, CLEAR.XFER, dw->all_chan_mask); + dma_writel(dw, CLEAR.BLOCK, dw->all_chan_mask); + dma_writel(dw, CLEAR.SRC_TRAN, dw->all_chan_mask); + dma_writel(dw, CLEAR.DST_TRAN, dw->all_chan_mask); + dma_writel(dw, CLEAR.ERROR, dw->all_chan_mask); + + channel_clear_bit(dw, MASK.XFER, dw->all_chan_mask); + channel_clear_bit(dw, MASK.BLOCK, dw->all_chan_mask); + channel_clear_bit(dw, MASK.SRC_TRAN, dw->all_chan_mask); + channel_clear_bit(dw, MASK.DST_TRAN, dw->all_chan_mask); + channel_clear_bit(dw, MASK.ERROR, dw->all_chan_mask); + + dma_cap_set(DMA_MEMCPY, dw->dma.cap_mask); + dma_cap_set(DMA_SLAVE, dw->dma.cap_mask); + dw->dma.dev = &pdev->dev; + dw->dma.device_alloc_chan_resources = dwc_alloc_chan_resources; + dw->dma.device_free_chan_resources = dwc_free_chan_resources; + + dw->dma.device_prep_dma_memcpy = dwc_prep_dma_memcpy; + + dw->dma.device_prep_slave_sg = dwc_prep_slave_sg; + dw->dma.device_terminate_all = dwc_terminate_all; + + dw->dma.device_is_tx_complete = dwc_is_tx_complete; + dw->dma.device_issue_pending = dwc_issue_pending; + + dma_writel(dw, CFG, DW_CFG_DMA_EN); + + printk(KERN_INFO "%s: DesignWare DMA Controller, %d channels\n", + dev_name(&pdev->dev), dw->dma.chancnt); + + dma_async_device_register(&dw->dma); + + return 0; + +err_irq: + clk_disable(dw->clk); + clk_put(dw->clk); +err_clk: + iounmap(dw->regs); + dw->regs = NULL; +err_release_r: + release_resource(io); +err_kfree: + kfree(dw); + return err; +} + +static int __exit dw_remove(struct platform_device *pdev) +{ + struct dw_dma *dw = platform_get_drvdata(pdev); + struct dw_dma_chan *dwc, *_dwc; + struct resource *io; + + dw_dma_off(dw); + dma_async_device_unregister(&dw->dma); + + free_irq(platform_get_irq(pdev, 0), dw); + tasklet_kill(&dw->tasklet); + + list_for_each_entry_safe(dwc, _dwc, &dw->dma.channels, + chan.device_node) { + list_del(&dwc->chan.device_node); + channel_clear_bit(dw, CH_EN, dwc->mask); + } + + clk_disable(dw->clk); + clk_put(dw->clk); + + iounmap(dw->regs); + dw->regs = NULL; + + io = platform_get_resource(pdev, IORESOURCE_MEM, 0); + release_mem_region(io->start, DW_REGLEN); + + kfree(dw); + + return 0; +} + +static void dw_shutdown(struct platform_device *pdev) +{ + struct dw_dma *dw = platform_get_drvdata(pdev); + + dw_dma_off(platform_get_drvdata(pdev)); + clk_disable(dw->clk); +} + +static int dw_suspend_noirq(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct dw_dma *dw = platform_get_drvdata(pdev); + + dw_dma_off(platform_get_drvdata(pdev)); + clk_disable(dw->clk); + return 0; +} + +static int dw_resume_noirq(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct dw_dma *dw = platform_get_drvdata(pdev); + + clk_enable(dw->clk); + dma_writel(dw, CFG, DW_CFG_DMA_EN); + return 0; +} + +static const struct dev_pm_ops dw_dev_pm_ops = { + .suspend_noirq = dw_suspend_noirq, + .resume_noirq = dw_resume_noirq, +}; + +static struct platform_driver dw_driver = { + .remove = __exit_p(dw_remove), + .shutdown = dw_shutdown, + .driver = { + .name = "dw_dmac", + .pm = &dw_dev_pm_ops, + }, +}; + +static int __init dw_init(void) +{ + return platform_driver_probe(&dw_driver, dw_probe); +} +module_init(dw_init); + +static void __exit dw_exit(void) +{ + platform_driver_unregister(&dw_driver); +} +module_exit(dw_exit); + +MODULE_LICENSE("GPL v2"); +MODULE_DESCRIPTION("Synopsys DesignWare DMA Controller driver"); +MODULE_AUTHOR("Haavard Skinnemoen <haavard.skinnemoen@atmel.com>"); diff --git a/drivers/dma/dw_dmac_regs.h b/drivers/dma/dw_dmac_regs.h new file mode 100644 index 00000000000..d9a939f67f4 --- /dev/null +++ b/drivers/dma/dw_dmac_regs.h @@ -0,0 +1,229 @@ +/* + * Driver for the Synopsys DesignWare AHB DMA Controller + * + * Copyright (C) 2005-2007 Atmel Corporation + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include <linux/dw_dmac.h> + +#define DW_DMA_MAX_NR_CHANNELS 8 + +/* + * Redefine this macro to handle differences between 32- and 64-bit + * addressing, big vs. little endian, etc. + */ +#define DW_REG(name) u32 name; u32 __pad_##name + +/* Hardware register definitions. */ +struct dw_dma_chan_regs { + DW_REG(SAR); /* Source Address Register */ + DW_REG(DAR); /* Destination Address Register */ + DW_REG(LLP); /* Linked List Pointer */ + u32 CTL_LO; /* Control Register Low */ + u32 CTL_HI; /* Control Register High */ + DW_REG(SSTAT); + DW_REG(DSTAT); + DW_REG(SSTATAR); + DW_REG(DSTATAR); + u32 CFG_LO; /* Configuration Register Low */ + u32 CFG_HI; /* Configuration Register High */ + DW_REG(SGR); + DW_REG(DSR); +}; + +struct dw_dma_irq_regs { + DW_REG(XFER); + DW_REG(BLOCK); + DW_REG(SRC_TRAN); + DW_REG(DST_TRAN); + DW_REG(ERROR); +}; + +struct dw_dma_regs { + /* per-channel registers */ + struct dw_dma_chan_regs CHAN[DW_DMA_MAX_NR_CHANNELS]; + + /* irq handling */ + struct dw_dma_irq_regs RAW; /* r */ + struct dw_dma_irq_regs STATUS; /* r (raw & mask) */ + struct dw_dma_irq_regs MASK; /* rw (set = irq enabled) */ + struct dw_dma_irq_regs CLEAR; /* w (ack, affects "raw") */ + + DW_REG(STATUS_INT); /* r */ + + /* software handshaking */ + DW_REG(REQ_SRC); + DW_REG(REQ_DST); + DW_REG(SGL_REQ_SRC); + DW_REG(SGL_REQ_DST); + DW_REG(LAST_SRC); + DW_REG(LAST_DST); + + /* miscellaneous */ + DW_REG(CFG); + DW_REG(CH_EN); + DW_REG(ID); + DW_REG(TEST); + + /* optional encoded params, 0x3c8..0x3 */ +}; + +/* Bitfields in CTL_LO */ +#define DWC_CTLL_INT_EN (1 << 0) /* irqs enabled? */ +#define DWC_CTLL_DST_WIDTH(n) ((n)<<1) /* bytes per element */ +#define DWC_CTLL_SRC_WIDTH(n) ((n)<<4) +#define DWC_CTLL_DST_INC (0<<7) /* DAR update/not */ +#define DWC_CTLL_DST_DEC (1<<7) +#define DWC_CTLL_DST_FIX (2<<7) +#define DWC_CTLL_SRC_INC (0<<7) /* SAR update/not */ +#define DWC_CTLL_SRC_DEC (1<<9) +#define DWC_CTLL_SRC_FIX (2<<9) +#define DWC_CTLL_DST_MSIZE(n) ((n)<<11) /* burst, #elements */ +#define DWC_CTLL_SRC_MSIZE(n) ((n)<<14) +#define DWC_CTLL_S_GATH_EN (1 << 17) /* src gather, !FIX */ +#define DWC_CTLL_D_SCAT_EN (1 << 18) /* dst scatter, !FIX */ +#define DWC_CTLL_FC_M2M (0 << 20) /* mem-to-mem */ +#define DWC_CTLL_FC_M2P (1 << 20) /* mem-to-periph */ +#define DWC_CTLL_FC_P2M (2 << 20) /* periph-to-mem */ +#define DWC_CTLL_FC_P2P (3 << 20) /* periph-to-periph */ +/* plus 4 transfer types for peripheral-as-flow-controller */ +#define DWC_CTLL_DMS(n) ((n)<<23) /* dst master select */ +#define DWC_CTLL_SMS(n) ((n)<<25) /* src master select */ +#define DWC_CTLL_LLP_D_EN (1 << 27) /* dest block chain */ +#define DWC_CTLL_LLP_S_EN (1 << 28) /* src block chain */ + +/* Bitfields in CTL_HI */ +#define DWC_CTLH_DONE 0x00001000 +#define DWC_CTLH_BLOCK_TS_MASK 0x00000fff + +/* Bitfields in CFG_LO. Platform-configurable bits are in <linux/dw_dmac.h> */ +#define DWC_CFGL_CH_SUSP (1 << 8) /* pause xfer */ +#define DWC_CFGL_FIFO_EMPTY (1 << 9) /* pause xfer */ +#define DWC_CFGL_HS_DST (1 << 10) /* handshake w/dst */ +#define DWC_CFGL_HS_SRC (1 << 11) /* handshake w/src */ +#define DWC_CFGL_MAX_BURST(x) ((x) << 20) +#define DWC_CFGL_RELOAD_SAR (1 << 30) +#define DWC_CFGL_RELOAD_DAR (1 << 31) + +/* Bitfields in CFG_HI. Platform-configurable bits are in <linux/dw_dmac.h> */ +#define DWC_CFGH_DS_UPD_EN (1 << 5) +#define DWC_CFGH_SS_UPD_EN (1 << 6) + +/* Bitfields in SGR */ +#define DWC_SGR_SGI(x) ((x) << 0) +#define DWC_SGR_SGC(x) ((x) << 20) + +/* Bitfields in DSR */ +#define DWC_DSR_DSI(x) ((x) << 0) +#define DWC_DSR_DSC(x) ((x) << 20) + +/* Bitfields in CFG */ +#define DW_CFG_DMA_EN (1 << 0) + +#define DW_REGLEN 0x400 + +enum dw_dmac_flags { + DW_DMA_IS_CYCLIC = 0, +}; + +struct dw_dma_chan { + struct dma_chan chan; + void __iomem *ch_regs; + u8 mask; + + spinlock_t lock; + + /* these other elements are all protected by lock */ + unsigned long flags; + dma_cookie_t completed; + struct list_head active_list; + struct list_head queue; + struct list_head free_list; + struct dw_cyclic_desc *cdesc; + + unsigned int descs_allocated; +}; + +static inline struct dw_dma_chan_regs __iomem * +__dwc_regs(struct dw_dma_chan *dwc) +{ + return dwc->ch_regs; +} + +#define channel_readl(dwc, name) \ + __raw_readl(&(__dwc_regs(dwc)->name)) +#define channel_writel(dwc, name, val) \ + __raw_writel((val), &(__dwc_regs(dwc)->name)) + +static inline struct dw_dma_chan *to_dw_dma_chan(struct dma_chan *chan) +{ + return container_of(chan, struct dw_dma_chan, chan); +} + +struct dw_dma { + struct dma_device dma; + void __iomem *regs; + struct tasklet_struct tasklet; + struct clk *clk; + + u8 all_chan_mask; + + struct dw_dma_chan chan[0]; +}; + +static inline struct dw_dma_regs __iomem *__dw_regs(struct dw_dma *dw) +{ + return dw->regs; +} + +#define dma_readl(dw, name) \ + __raw_readl(&(__dw_regs(dw)->name)) +#define dma_writel(dw, name, val) \ + __raw_writel((val), &(__dw_regs(dw)->name)) + +#define channel_set_bit(dw, reg, mask) \ + dma_writel(dw, reg, ((mask) << 8) | (mask)) +#define channel_clear_bit(dw, reg, mask) \ + dma_writel(dw, reg, ((mask) << 8) | 0) + +static inline struct dw_dma *to_dw_dma(struct dma_device *ddev) +{ + return container_of(ddev, struct dw_dma, dma); +} + +/* LLI == Linked List Item; a.k.a. DMA block descriptor */ +struct dw_lli { + /* values that are not changed by hardware */ + dma_addr_t sar; + dma_addr_t dar; + dma_addr_t llp; /* chain to next lli */ + u32 ctllo; + /* values that may get written back: */ + u32 ctlhi; + /* sstat and dstat can snapshot peripheral register state. + * silicon config may discard either or both... + */ + u32 sstat; + u32 dstat; +}; + +struct dw_desc { + /* FIRST values the hardware uses */ + struct dw_lli lli; + + /* THEN values for driver housekeeping */ + struct list_head desc_node; + struct list_head tx_list; + struct dma_async_tx_descriptor txd; + size_t len; +}; + +static inline struct dw_desc * +txd_to_dw_desc(struct dma_async_tx_descriptor *txd) +{ + return container_of(txd, struct dw_desc, txd); +} diff --git a/drivers/dma/fsldma.c b/drivers/dma/fsldma.c new file mode 100644 index 00000000000..296f9e747fa --- /dev/null +++ b/drivers/dma/fsldma.c @@ -0,0 +1,1312 @@ +/* + * Freescale MPC85xx, MPC83xx DMA Engine support + * + * Copyright (C) 2007 Freescale Semiconductor, Inc. All rights reserved. + * + * Author: + * Zhang Wei <wei.zhang@freescale.com>, Jul 2007 + * Ebony Zhu <ebony.zhu@freescale.com>, May 2007 + * + * Description: + * DMA engine driver for Freescale MPC8540 DMA controller, which is + * also fit for MPC8560, MPC8555, MPC8548, MPC8641, and etc. + * The support for MPC8349 DMA contorller is also added. + * + * This driver instructs the DMA controller to issue the PCI Read Multiple + * command for PCI read operations, instead of using the default PCI Read Line + * command. Please be aware that this setting may result in read pre-fetching + * on some platforms. + * + * This is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/interrupt.h> +#include <linux/dmaengine.h> +#include <linux/delay.h> +#include <linux/dma-mapping.h> +#include <linux/dmapool.h> +#include <linux/of_platform.h> + +#include <asm/fsldma.h> +#include "fsldma.h" + +static void dma_init(struct fsl_dma_chan *fsl_chan) +{ + /* Reset the channel */ + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, 0, 32); + + switch (fsl_chan->feature & FSL_DMA_IP_MASK) { + case FSL_DMA_IP_85XX: + /* Set the channel to below modes: + * EIE - Error interrupt enable + * EOSIE - End of segments interrupt enable (basic mode) + * EOLNIE - End of links interrupt enable + */ + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, FSL_DMA_MR_EIE + | FSL_DMA_MR_EOLNIE | FSL_DMA_MR_EOSIE, 32); + break; + case FSL_DMA_IP_83XX: + /* Set the channel to below modes: + * EOTIE - End-of-transfer interrupt enable + * PRC_RM - PCI read multiple + */ + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, FSL_DMA_MR_EOTIE + | FSL_DMA_MR_PRC_RM, 32); + break; + } + +} + +static void set_sr(struct fsl_dma_chan *fsl_chan, u32 val) +{ + DMA_OUT(fsl_chan, &fsl_chan->reg_base->sr, val, 32); +} + +static u32 get_sr(struct fsl_dma_chan *fsl_chan) +{ + return DMA_IN(fsl_chan, &fsl_chan->reg_base->sr, 32); +} + +static void set_desc_cnt(struct fsl_dma_chan *fsl_chan, + struct fsl_dma_ld_hw *hw, u32 count) +{ + hw->count = CPU_TO_DMA(fsl_chan, count, 32); +} + +static void set_desc_src(struct fsl_dma_chan *fsl_chan, + struct fsl_dma_ld_hw *hw, dma_addr_t src) +{ + u64 snoop_bits; + + snoop_bits = ((fsl_chan->feature & FSL_DMA_IP_MASK) == FSL_DMA_IP_85XX) + ? ((u64)FSL_DMA_SATR_SREADTYPE_SNOOP_READ << 32) : 0; + hw->src_addr = CPU_TO_DMA(fsl_chan, snoop_bits | src, 64); +} + +static void set_desc_dest(struct fsl_dma_chan *fsl_chan, + struct fsl_dma_ld_hw *hw, dma_addr_t dest) +{ + u64 snoop_bits; + + snoop_bits = ((fsl_chan->feature & FSL_DMA_IP_MASK) == FSL_DMA_IP_85XX) + ? ((u64)FSL_DMA_DATR_DWRITETYPE_SNOOP_WRITE << 32) : 0; + hw->dst_addr = CPU_TO_DMA(fsl_chan, snoop_bits | dest, 64); +} + +static void set_desc_next(struct fsl_dma_chan *fsl_chan, + struct fsl_dma_ld_hw *hw, dma_addr_t next) +{ + u64 snoop_bits; + + snoop_bits = ((fsl_chan->feature & FSL_DMA_IP_MASK) == FSL_DMA_IP_83XX) + ? FSL_DMA_SNEN : 0; + hw->next_ln_addr = CPU_TO_DMA(fsl_chan, snoop_bits | next, 64); +} + +static void set_cdar(struct fsl_dma_chan *fsl_chan, dma_addr_t addr) +{ + DMA_OUT(fsl_chan, &fsl_chan->reg_base->cdar, addr | FSL_DMA_SNEN, 64); +} + +static dma_addr_t get_cdar(struct fsl_dma_chan *fsl_chan) +{ + return DMA_IN(fsl_chan, &fsl_chan->reg_base->cdar, 64) & ~FSL_DMA_SNEN; +} + +static void set_ndar(struct fsl_dma_chan *fsl_chan, dma_addr_t addr) +{ + DMA_OUT(fsl_chan, &fsl_chan->reg_base->ndar, addr, 64); +} + +static dma_addr_t get_ndar(struct fsl_dma_chan *fsl_chan) +{ + return DMA_IN(fsl_chan, &fsl_chan->reg_base->ndar, 64); +} + +static u32 get_bcr(struct fsl_dma_chan *fsl_chan) +{ + return DMA_IN(fsl_chan, &fsl_chan->reg_base->bcr, 32); +} + +static int dma_is_idle(struct fsl_dma_chan *fsl_chan) +{ + u32 sr = get_sr(fsl_chan); + return (!(sr & FSL_DMA_SR_CB)) || (sr & FSL_DMA_SR_CH); +} + +static void dma_start(struct fsl_dma_chan *fsl_chan) +{ + u32 mr_set = 0; + + if (fsl_chan->feature & FSL_DMA_CHAN_PAUSE_EXT) { + DMA_OUT(fsl_chan, &fsl_chan->reg_base->bcr, 0, 32); + mr_set |= FSL_DMA_MR_EMP_EN; + } else if ((fsl_chan->feature & FSL_DMA_IP_MASK) == FSL_DMA_IP_85XX) { + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, + DMA_IN(fsl_chan, &fsl_chan->reg_base->mr, 32) + & ~FSL_DMA_MR_EMP_EN, 32); + } + + if (fsl_chan->feature & FSL_DMA_CHAN_START_EXT) + mr_set |= FSL_DMA_MR_EMS_EN; + else + mr_set |= FSL_DMA_MR_CS; + + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, + DMA_IN(fsl_chan, &fsl_chan->reg_base->mr, 32) + | mr_set, 32); +} + +static void dma_halt(struct fsl_dma_chan *fsl_chan) +{ + int i; + + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, + DMA_IN(fsl_chan, &fsl_chan->reg_base->mr, 32) | FSL_DMA_MR_CA, + 32); + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, + DMA_IN(fsl_chan, &fsl_chan->reg_base->mr, 32) & ~(FSL_DMA_MR_CS + | FSL_DMA_MR_EMS_EN | FSL_DMA_MR_CA), 32); + + for (i = 0; i < 100; i++) { + if (dma_is_idle(fsl_chan)) + break; + udelay(10); + } + if (i >= 100 && !dma_is_idle(fsl_chan)) + dev_err(fsl_chan->dev, "DMA halt timeout!\n"); +} + +static void set_ld_eol(struct fsl_dma_chan *fsl_chan, + struct fsl_desc_sw *desc) +{ + u64 snoop_bits; + + snoop_bits = ((fsl_chan->feature & FSL_DMA_IP_MASK) == FSL_DMA_IP_83XX) + ? FSL_DMA_SNEN : 0; + + desc->hw.next_ln_addr = CPU_TO_DMA(fsl_chan, + DMA_TO_CPU(fsl_chan, desc->hw.next_ln_addr, 64) | FSL_DMA_EOL + | snoop_bits, 64); +} + +static void append_ld_queue(struct fsl_dma_chan *fsl_chan, + struct fsl_desc_sw *new_desc) +{ + struct fsl_desc_sw *queue_tail = to_fsl_desc(fsl_chan->ld_queue.prev); + + if (list_empty(&fsl_chan->ld_queue)) + return; + + /* Link to the new descriptor physical address and + * Enable End-of-segment interrupt for + * the last link descriptor. + * (the previous node's next link descriptor) + * + * For FSL_DMA_IP_83xx, the snoop enable bit need be set. + */ + queue_tail->hw.next_ln_addr = CPU_TO_DMA(fsl_chan, + new_desc->async_tx.phys | FSL_DMA_EOSIE | + (((fsl_chan->feature & FSL_DMA_IP_MASK) + == FSL_DMA_IP_83XX) ? FSL_DMA_SNEN : 0), 64); +} + +/** + * fsl_chan_set_src_loop_size - Set source address hold transfer size + * @fsl_chan : Freescale DMA channel + * @size : Address loop size, 0 for disable loop + * + * The set source address hold transfer size. The source + * address hold or loop transfer size is when the DMA transfer + * data from source address (SA), if the loop size is 4, the DMA will + * read data from SA, SA + 1, SA + 2, SA + 3, then loop back to SA, + * SA + 1 ... and so on. + */ +static void fsl_chan_set_src_loop_size(struct fsl_dma_chan *fsl_chan, int size) +{ + switch (size) { + case 0: + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, + DMA_IN(fsl_chan, &fsl_chan->reg_base->mr, 32) & + (~FSL_DMA_MR_SAHE), 32); + break; + case 1: + case 2: + case 4: + case 8: + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, + DMA_IN(fsl_chan, &fsl_chan->reg_base->mr, 32) | + FSL_DMA_MR_SAHE | (__ilog2(size) << 14), + 32); + break; + } +} + +/** + * fsl_chan_set_dest_loop_size - Set destination address hold transfer size + * @fsl_chan : Freescale DMA channel + * @size : Address loop size, 0 for disable loop + * + * The set destination address hold transfer size. The destination + * address hold or loop transfer size is when the DMA transfer + * data to destination address (TA), if the loop size is 4, the DMA will + * write data to TA, TA + 1, TA + 2, TA + 3, then loop back to TA, + * TA + 1 ... and so on. + */ +static void fsl_chan_set_dest_loop_size(struct fsl_dma_chan *fsl_chan, int size) +{ + switch (size) { + case 0: + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, + DMA_IN(fsl_chan, &fsl_chan->reg_base->mr, 32) & + (~FSL_DMA_MR_DAHE), 32); + break; + case 1: + case 2: + case 4: + case 8: + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, + DMA_IN(fsl_chan, &fsl_chan->reg_base->mr, 32) | + FSL_DMA_MR_DAHE | (__ilog2(size) << 16), + 32); + break; + } +} + +/** + * fsl_chan_set_request_count - Set DMA Request Count for external control + * @fsl_chan : Freescale DMA channel + * @size : Number of bytes to transfer in a single request + * + * The Freescale DMA channel can be controlled by the external signal DREQ#. + * The DMA request count is how many bytes are allowed to transfer before + * pausing the channel, after which a new assertion of DREQ# resumes channel + * operation. + * + * A size of 0 disables external pause control. The maximum size is 1024. + */ +static void fsl_chan_set_request_count(struct fsl_dma_chan *fsl_chan, int size) +{ + BUG_ON(size > 1024); + DMA_OUT(fsl_chan, &fsl_chan->reg_base->mr, + DMA_IN(fsl_chan, &fsl_chan->reg_base->mr, 32) + | ((__ilog2(size) << 24) & 0x0f000000), + 32); +} + +/** + * fsl_chan_toggle_ext_pause - Toggle channel external pause status + * @fsl_chan : Freescale DMA channel + * @enable : 0 is disabled, 1 is enabled. + * + * The Freescale DMA channel can be controlled by the external signal DREQ#. + * The DMA Request Count feature should be used in addition to this feature + * to set the number of bytes to transfer before pausing the channel. + */ +static void fsl_chan_toggle_ext_pause(struct fsl_dma_chan *fsl_chan, int enable) +{ + if (enable) + fsl_chan->feature |= FSL_DMA_CHAN_PAUSE_EXT; + else + fsl_chan->feature &= ~FSL_DMA_CHAN_PAUSE_EXT; +} + +/** + * fsl_chan_toggle_ext_start - Toggle channel external start status + * @fsl_chan : Freescale DMA channel + * @enable : 0 is disabled, 1 is enabled. + * + * If enable the external start, the channel can be started by an + * external DMA start pin. So the dma_start() does not start the + * transfer immediately. The DMA channel will wait for the + * control pin asserted. + */ +static void fsl_chan_toggle_ext_start(struct fsl_dma_chan *fsl_chan, int enable) +{ + if (enable) + fsl_chan->feature |= FSL_DMA_CHAN_START_EXT; + else + fsl_chan->feature &= ~FSL_DMA_CHAN_START_EXT; +} + +static dma_cookie_t fsl_dma_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct fsl_dma_chan *fsl_chan = to_fsl_chan(tx->chan); + struct fsl_desc_sw *desc = tx_to_fsl_desc(tx); + struct fsl_desc_sw *child; + unsigned long flags; + dma_cookie_t cookie; + + /* cookie increment and adding to ld_queue must be atomic */ + spin_lock_irqsave(&fsl_chan->desc_lock, flags); + + cookie = fsl_chan->common.cookie; + list_for_each_entry(child, &desc->tx_list, node) { + cookie++; + if (cookie < 0) + cookie = 1; + + desc->async_tx.cookie = cookie; + } + + fsl_chan->common.cookie = cookie; + append_ld_queue(fsl_chan, desc); + list_splice_init(&desc->tx_list, fsl_chan->ld_queue.prev); + + spin_unlock_irqrestore(&fsl_chan->desc_lock, flags); + + return cookie; +} + +/** + * fsl_dma_alloc_descriptor - Allocate descriptor from channel's DMA pool. + * @fsl_chan : Freescale DMA channel + * + * Return - The descriptor allocated. NULL for failed. + */ +static struct fsl_desc_sw *fsl_dma_alloc_descriptor( + struct fsl_dma_chan *fsl_chan) +{ + dma_addr_t pdesc; + struct fsl_desc_sw *desc_sw; + + desc_sw = dma_pool_alloc(fsl_chan->desc_pool, GFP_ATOMIC, &pdesc); + if (desc_sw) { + memset(desc_sw, 0, sizeof(struct fsl_desc_sw)); + INIT_LIST_HEAD(&desc_sw->tx_list); + dma_async_tx_descriptor_init(&desc_sw->async_tx, + &fsl_chan->common); + desc_sw->async_tx.tx_submit = fsl_dma_tx_submit; + desc_sw->async_tx.phys = pdesc; + } + + return desc_sw; +} + + +/** + * fsl_dma_alloc_chan_resources - Allocate resources for DMA channel. + * @fsl_chan : Freescale DMA channel + * + * This function will create a dma pool for descriptor allocation. + * + * Return - The number of descriptors allocated. + */ +static int fsl_dma_alloc_chan_resources(struct dma_chan *chan) +{ + struct fsl_dma_chan *fsl_chan = to_fsl_chan(chan); + + /* Has this channel already been allocated? */ + if (fsl_chan->desc_pool) + return 1; + + /* We need the descriptor to be aligned to 32bytes + * for meeting FSL DMA specification requirement. + */ + fsl_chan->desc_pool = dma_pool_create("fsl_dma_engine_desc_pool", + fsl_chan->dev, sizeof(struct fsl_desc_sw), + 32, 0); + if (!fsl_chan->desc_pool) { + dev_err(fsl_chan->dev, "No memory for channel %d " + "descriptor dma pool.\n", fsl_chan->id); + return 0; + } + + return 1; +} + +/** + * fsl_dma_free_chan_resources - Free all resources of the channel. + * @fsl_chan : Freescale DMA channel + */ +static void fsl_dma_free_chan_resources(struct dma_chan *chan) +{ + struct fsl_dma_chan *fsl_chan = to_fsl_chan(chan); + struct fsl_desc_sw *desc, *_desc; + unsigned long flags; + + dev_dbg(fsl_chan->dev, "Free all channel resources.\n"); + spin_lock_irqsave(&fsl_chan->desc_lock, flags); + list_for_each_entry_safe(desc, _desc, &fsl_chan->ld_queue, node) { +#ifdef FSL_DMA_LD_DEBUG + dev_dbg(fsl_chan->dev, + "LD %p will be released.\n", desc); +#endif + list_del(&desc->node); + /* free link descriptor */ + dma_pool_free(fsl_chan->desc_pool, desc, desc->async_tx.phys); + } + spin_unlock_irqrestore(&fsl_chan->desc_lock, flags); + dma_pool_destroy(fsl_chan->desc_pool); + + fsl_chan->desc_pool = NULL; +} + +static struct dma_async_tx_descriptor * +fsl_dma_prep_interrupt(struct dma_chan *chan, unsigned long flags) +{ + struct fsl_dma_chan *fsl_chan; + struct fsl_desc_sw *new; + + if (!chan) + return NULL; + + fsl_chan = to_fsl_chan(chan); + + new = fsl_dma_alloc_descriptor(fsl_chan); + if (!new) { + dev_err(fsl_chan->dev, "No free memory for link descriptor\n"); + return NULL; + } + + new->async_tx.cookie = -EBUSY; + new->async_tx.flags = flags; + + /* Insert the link descriptor to the LD ring */ + list_add_tail(&new->node, &new->tx_list); + + /* Set End-of-link to the last link descriptor of new list*/ + set_ld_eol(fsl_chan, new); + + return &new->async_tx; +} + +static struct dma_async_tx_descriptor *fsl_dma_prep_memcpy( + struct dma_chan *chan, dma_addr_t dma_dest, dma_addr_t dma_src, + size_t len, unsigned long flags) +{ + struct fsl_dma_chan *fsl_chan; + struct fsl_desc_sw *first = NULL, *prev = NULL, *new; + struct list_head *list; + size_t copy; + + if (!chan) + return NULL; + + if (!len) + return NULL; + + fsl_chan = to_fsl_chan(chan); + + do { + + /* Allocate the link descriptor from DMA pool */ + new = fsl_dma_alloc_descriptor(fsl_chan); + if (!new) { + dev_err(fsl_chan->dev, + "No free memory for link descriptor\n"); + goto fail; + } +#ifdef FSL_DMA_LD_DEBUG + dev_dbg(fsl_chan->dev, "new link desc alloc %p\n", new); +#endif + + copy = min(len, (size_t)FSL_DMA_BCR_MAX_CNT); + + set_desc_cnt(fsl_chan, &new->hw, copy); + set_desc_src(fsl_chan, &new->hw, dma_src); + set_desc_dest(fsl_chan, &new->hw, dma_dest); + + if (!first) + first = new; + else + set_desc_next(fsl_chan, &prev->hw, new->async_tx.phys); + + new->async_tx.cookie = 0; + async_tx_ack(&new->async_tx); + + prev = new; + len -= copy; + dma_src += copy; + dma_dest += copy; + + /* Insert the link descriptor to the LD ring */ + list_add_tail(&new->node, &first->tx_list); + } while (len); + + new->async_tx.flags = flags; /* client is in control of this ack */ + new->async_tx.cookie = -EBUSY; + + /* Set End-of-link to the last link descriptor of new list*/ + set_ld_eol(fsl_chan, new); + + return &first->async_tx; + +fail: + if (!first) + return NULL; + + list = &first->tx_list; + list_for_each_entry_safe_reverse(new, prev, list, node) { + list_del(&new->node); + dma_pool_free(fsl_chan->desc_pool, new, new->async_tx.phys); + } + + return NULL; +} + +/** + * fsl_dma_prep_slave_sg - prepare descriptors for a DMA_SLAVE transaction + * @chan: DMA channel + * @sgl: scatterlist to transfer to/from + * @sg_len: number of entries in @scatterlist + * @direction: DMA direction + * @flags: DMAEngine flags + * + * Prepare a set of descriptors for a DMA_SLAVE transaction. Following the + * DMA_SLAVE API, this gets the device-specific information from the + * chan->private variable. + */ +static struct dma_async_tx_descriptor *fsl_dma_prep_slave_sg( + struct dma_chan *chan, struct scatterlist *sgl, unsigned int sg_len, + enum dma_data_direction direction, unsigned long flags) +{ + struct fsl_dma_chan *fsl_chan; + struct fsl_desc_sw *first = NULL, *prev = NULL, *new = NULL; + struct fsl_dma_slave *slave; + struct list_head *tx_list; + size_t copy; + + int i; + struct scatterlist *sg; + size_t sg_used; + size_t hw_used; + struct fsl_dma_hw_addr *hw; + dma_addr_t dma_dst, dma_src; + + if (!chan) + return NULL; + + if (!chan->private) + return NULL; + + fsl_chan = to_fsl_chan(chan); + slave = chan->private; + + if (list_empty(&slave->addresses)) + return NULL; + + hw = list_first_entry(&slave->addresses, struct fsl_dma_hw_addr, entry); + hw_used = 0; + + /* + * Build the hardware transaction to copy from the scatterlist to + * the hardware, or from the hardware to the scatterlist + * + * If you are copying from the hardware to the scatterlist and it + * takes two hardware entries to fill an entire page, then both + * hardware entries will be coalesced into the same page + * + * If you are copying from the scatterlist to the hardware and a + * single page can fill two hardware entries, then the data will + * be read out of the page into the first hardware entry, and so on + */ + for_each_sg(sgl, sg, sg_len, i) { + sg_used = 0; + + /* Loop until the entire scatterlist entry is used */ + while (sg_used < sg_dma_len(sg)) { + + /* + * If we've used up the current hardware address/length + * pair, we need to load a new one + * + * This is done in a while loop so that descriptors with + * length == 0 will be skipped + */ + while (hw_used >= hw->length) { + + /* + * If the current hardware entry is the last + * entry in the list, we're finished + */ + if (list_is_last(&hw->entry, &slave->addresses)) + goto finished; + + /* Get the next hardware address/length pair */ + hw = list_entry(hw->entry.next, + struct fsl_dma_hw_addr, entry); + hw_used = 0; + } + + /* Allocate the link descriptor from DMA pool */ + new = fsl_dma_alloc_descriptor(fsl_chan); + if (!new) { + dev_err(fsl_chan->dev, "No free memory for " + "link descriptor\n"); + goto fail; + } +#ifdef FSL_DMA_LD_DEBUG + dev_dbg(fsl_chan->dev, "new link desc alloc %p\n", new); +#endif + + /* + * Calculate the maximum number of bytes to transfer, + * making sure it is less than the DMA controller limit + */ + copy = min_t(size_t, sg_dma_len(sg) - sg_used, + hw->length - hw_used); + copy = min_t(size_t, copy, FSL_DMA_BCR_MAX_CNT); + + /* + * DMA_FROM_DEVICE + * from the hardware to the scatterlist + * + * DMA_TO_DEVICE + * from the scatterlist to the hardware + */ + if (direction == DMA_FROM_DEVICE) { + dma_src = hw->address + hw_used; + dma_dst = sg_dma_address(sg) + sg_used; + } else { + dma_src = sg_dma_address(sg) + sg_used; + dma_dst = hw->address + hw_used; + } + + /* Fill in the descriptor */ + set_desc_cnt(fsl_chan, &new->hw, copy); + set_desc_src(fsl_chan, &new->hw, dma_src); + set_desc_dest(fsl_chan, &new->hw, dma_dst); + + /* + * If this is not the first descriptor, chain the + * current descriptor after the previous descriptor + */ + if (!first) { + first = new; + } else { + set_desc_next(fsl_chan, &prev->hw, + new->async_tx.phys); + } + + new->async_tx.cookie = 0; + async_tx_ack(&new->async_tx); + + prev = new; + sg_used += copy; + hw_used += copy; + + /* Insert the link descriptor into the LD ring */ + list_add_tail(&new->node, &first->tx_list); + } + } + +finished: + + /* All of the hardware address/length pairs had length == 0 */ + if (!first || !new) + return NULL; + + new->async_tx.flags = flags; + new->async_tx.cookie = -EBUSY; + + /* Set End-of-link to the last link descriptor of new list */ + set_ld_eol(fsl_chan, new); + + /* Enable extra controller features */ + if (fsl_chan->set_src_loop_size) + fsl_chan->set_src_loop_size(fsl_chan, slave->src_loop_size); + + if (fsl_chan->set_dest_loop_size) + fsl_chan->set_dest_loop_size(fsl_chan, slave->dst_loop_size); + + if (fsl_chan->toggle_ext_start) + fsl_chan->toggle_ext_start(fsl_chan, slave->external_start); + + if (fsl_chan->toggle_ext_pause) + fsl_chan->toggle_ext_pause(fsl_chan, slave->external_pause); + + if (fsl_chan->set_request_count) + fsl_chan->set_request_count(fsl_chan, slave->request_count); + + return &first->async_tx; + +fail: + /* If first was not set, then we failed to allocate the very first + * descriptor, and we're done */ + if (!first) + return NULL; + + /* + * First is set, so all of the descriptors we allocated have been added + * to first->tx_list, INCLUDING "first" itself. Therefore we + * must traverse the list backwards freeing each descriptor in turn + * + * We're re-using variables for the loop, oh well + */ + tx_list = &first->tx_list; + list_for_each_entry_safe_reverse(new, prev, tx_list, node) { + list_del_init(&new->node); + dma_pool_free(fsl_chan->desc_pool, new, new->async_tx.phys); + } + + return NULL; +} + +static void fsl_dma_device_terminate_all(struct dma_chan *chan) +{ + struct fsl_dma_chan *fsl_chan; + struct fsl_desc_sw *desc, *tmp; + unsigned long flags; + + if (!chan) + return; + + fsl_chan = to_fsl_chan(chan); + + /* Halt the DMA engine */ + dma_halt(fsl_chan); + + spin_lock_irqsave(&fsl_chan->desc_lock, flags); + + /* Remove and free all of the descriptors in the LD queue */ + list_for_each_entry_safe(desc, tmp, &fsl_chan->ld_queue, node) { + list_del(&desc->node); + dma_pool_free(fsl_chan->desc_pool, desc, desc->async_tx.phys); + } + + spin_unlock_irqrestore(&fsl_chan->desc_lock, flags); +} + +/** + * fsl_dma_update_completed_cookie - Update the completed cookie. + * @fsl_chan : Freescale DMA channel + */ +static void fsl_dma_update_completed_cookie(struct fsl_dma_chan *fsl_chan) +{ + struct fsl_desc_sw *cur_desc, *desc; + dma_addr_t ld_phy; + + ld_phy = get_cdar(fsl_chan) & FSL_DMA_NLDA_MASK; + + if (ld_phy) { + cur_desc = NULL; + list_for_each_entry(desc, &fsl_chan->ld_queue, node) + if (desc->async_tx.phys == ld_phy) { + cur_desc = desc; + break; + } + + if (cur_desc && cur_desc->async_tx.cookie) { + if (dma_is_idle(fsl_chan)) + fsl_chan->completed_cookie = + cur_desc->async_tx.cookie; + else + fsl_chan->completed_cookie = + cur_desc->async_tx.cookie - 1; + } + } +} + +/** + * fsl_chan_ld_cleanup - Clean up link descriptors + * @fsl_chan : Freescale DMA channel + * + * This function clean up the ld_queue of DMA channel. + * If 'in_intr' is set, the function will move the link descriptor to + * the recycle list. Otherwise, free it directly. + */ +static void fsl_chan_ld_cleanup(struct fsl_dma_chan *fsl_chan) +{ + struct fsl_desc_sw *desc, *_desc; + unsigned long flags; + + spin_lock_irqsave(&fsl_chan->desc_lock, flags); + + dev_dbg(fsl_chan->dev, "chan completed_cookie = %d\n", + fsl_chan->completed_cookie); + list_for_each_entry_safe(desc, _desc, &fsl_chan->ld_queue, node) { + dma_async_tx_callback callback; + void *callback_param; + + if (dma_async_is_complete(desc->async_tx.cookie, + fsl_chan->completed_cookie, fsl_chan->common.cookie) + == DMA_IN_PROGRESS) + break; + + callback = desc->async_tx.callback; + callback_param = desc->async_tx.callback_param; + + /* Remove from ld_queue list */ + list_del(&desc->node); + + dev_dbg(fsl_chan->dev, "link descriptor %p will be recycle.\n", + desc); + dma_pool_free(fsl_chan->desc_pool, desc, desc->async_tx.phys); + + /* Run the link descriptor callback function */ + if (callback) { + spin_unlock_irqrestore(&fsl_chan->desc_lock, flags); + dev_dbg(fsl_chan->dev, "link descriptor %p callback\n", + desc); + callback(callback_param); + spin_lock_irqsave(&fsl_chan->desc_lock, flags); + } + } + spin_unlock_irqrestore(&fsl_chan->desc_lock, flags); +} + +/** + * fsl_chan_xfer_ld_queue - Transfer link descriptors in channel ld_queue. + * @fsl_chan : Freescale DMA channel + */ +static void fsl_chan_xfer_ld_queue(struct fsl_dma_chan *fsl_chan) +{ + struct list_head *ld_node; + dma_addr_t next_dest_addr; + unsigned long flags; + + spin_lock_irqsave(&fsl_chan->desc_lock, flags); + + if (!dma_is_idle(fsl_chan)) + goto out_unlock; + + dma_halt(fsl_chan); + + /* If there are some link descriptors + * not transfered in queue. We need to start it. + */ + + /* Find the first un-transfer desciptor */ + for (ld_node = fsl_chan->ld_queue.next; + (ld_node != &fsl_chan->ld_queue) + && (dma_async_is_complete( + to_fsl_desc(ld_node)->async_tx.cookie, + fsl_chan->completed_cookie, + fsl_chan->common.cookie) == DMA_SUCCESS); + ld_node = ld_node->next); + + if (ld_node != &fsl_chan->ld_queue) { + /* Get the ld start address from ld_queue */ + next_dest_addr = to_fsl_desc(ld_node)->async_tx.phys; + dev_dbg(fsl_chan->dev, "xfer LDs staring from 0x%llx\n", + (unsigned long long)next_dest_addr); + set_cdar(fsl_chan, next_dest_addr); + dma_start(fsl_chan); + } else { + set_cdar(fsl_chan, 0); + set_ndar(fsl_chan, 0); + } + +out_unlock: + spin_unlock_irqrestore(&fsl_chan->desc_lock, flags); +} + +/** + * fsl_dma_memcpy_issue_pending - Issue the DMA start command + * @fsl_chan : Freescale DMA channel + */ +static void fsl_dma_memcpy_issue_pending(struct dma_chan *chan) +{ + struct fsl_dma_chan *fsl_chan = to_fsl_chan(chan); + +#ifdef FSL_DMA_LD_DEBUG + struct fsl_desc_sw *ld; + unsigned long flags; + + spin_lock_irqsave(&fsl_chan->desc_lock, flags); + if (list_empty(&fsl_chan->ld_queue)) { + spin_unlock_irqrestore(&fsl_chan->desc_lock, flags); + return; + } + + dev_dbg(fsl_chan->dev, "--memcpy issue--\n"); + list_for_each_entry(ld, &fsl_chan->ld_queue, node) { + int i; + dev_dbg(fsl_chan->dev, "Ch %d, LD %08x\n", + fsl_chan->id, ld->async_tx.phys); + for (i = 0; i < 8; i++) + dev_dbg(fsl_chan->dev, "LD offset %d: %08x\n", + i, *(((u32 *)&ld->hw) + i)); + } + dev_dbg(fsl_chan->dev, "----------------\n"); + spin_unlock_irqrestore(&fsl_chan->desc_lock, flags); +#endif + + fsl_chan_xfer_ld_queue(fsl_chan); +} + +/** + * fsl_dma_is_complete - Determine the DMA status + * @fsl_chan : Freescale DMA channel + */ +static enum dma_status fsl_dma_is_complete(struct dma_chan *chan, + dma_cookie_t cookie, + dma_cookie_t *done, + dma_cookie_t *used) +{ + struct fsl_dma_chan *fsl_chan = to_fsl_chan(chan); + dma_cookie_t last_used; + dma_cookie_t last_complete; + + fsl_chan_ld_cleanup(fsl_chan); + + last_used = chan->cookie; + last_complete = fsl_chan->completed_cookie; + + if (done) + *done = last_complete; + + if (used) + *used = last_used; + + return dma_async_is_complete(cookie, last_complete, last_used); +} + +static irqreturn_t fsl_dma_chan_do_interrupt(int irq, void *data) +{ + struct fsl_dma_chan *fsl_chan = (struct fsl_dma_chan *)data; + u32 stat; + int update_cookie = 0; + int xfer_ld_q = 0; + + stat = get_sr(fsl_chan); + dev_dbg(fsl_chan->dev, "event: channel %d, stat = 0x%x\n", + fsl_chan->id, stat); + set_sr(fsl_chan, stat); /* Clear the event register */ + + stat &= ~(FSL_DMA_SR_CB | FSL_DMA_SR_CH); + if (!stat) + return IRQ_NONE; + + if (stat & FSL_DMA_SR_TE) + dev_err(fsl_chan->dev, "Transfer Error!\n"); + + /* Programming Error + * The DMA_INTERRUPT async_tx is a NULL transfer, which will + * triger a PE interrupt. + */ + if (stat & FSL_DMA_SR_PE) { + dev_dbg(fsl_chan->dev, "event: Programming Error INT\n"); + if (get_bcr(fsl_chan) == 0) { + /* BCR register is 0, this is a DMA_INTERRUPT async_tx. + * Now, update the completed cookie, and continue the + * next uncompleted transfer. + */ + update_cookie = 1; + xfer_ld_q = 1; + } + stat &= ~FSL_DMA_SR_PE; + } + + /* If the link descriptor segment transfer finishes, + * we will recycle the used descriptor. + */ + if (stat & FSL_DMA_SR_EOSI) { + dev_dbg(fsl_chan->dev, "event: End-of-segments INT\n"); + dev_dbg(fsl_chan->dev, "event: clndar 0x%llx, nlndar 0x%llx\n", + (unsigned long long)get_cdar(fsl_chan), + (unsigned long long)get_ndar(fsl_chan)); + stat &= ~FSL_DMA_SR_EOSI; + update_cookie = 1; + } + + /* For MPC8349, EOCDI event need to update cookie + * and start the next transfer if it exist. + */ + if (stat & FSL_DMA_SR_EOCDI) { + dev_dbg(fsl_chan->dev, "event: End-of-Chain link INT\n"); + stat &= ~FSL_DMA_SR_EOCDI; + update_cookie = 1; + xfer_ld_q = 1; + } + + /* If it current transfer is the end-of-transfer, + * we should clear the Channel Start bit for + * prepare next transfer. + */ + if (stat & FSL_DMA_SR_EOLNI) { + dev_dbg(fsl_chan->dev, "event: End-of-link INT\n"); + stat &= ~FSL_DMA_SR_EOLNI; + xfer_ld_q = 1; + } + + if (update_cookie) + fsl_dma_update_completed_cookie(fsl_chan); + if (xfer_ld_q) + fsl_chan_xfer_ld_queue(fsl_chan); + if (stat) + dev_dbg(fsl_chan->dev, "event: unhandled sr 0x%02x\n", + stat); + + dev_dbg(fsl_chan->dev, "event: Exit\n"); + tasklet_schedule(&fsl_chan->tasklet); + return IRQ_HANDLED; +} + +static irqreturn_t fsl_dma_do_interrupt(int irq, void *data) +{ + struct fsl_dma_device *fdev = (struct fsl_dma_device *)data; + u32 gsr; + int ch_nr; + + gsr = (fdev->feature & FSL_DMA_BIG_ENDIAN) ? in_be32(fdev->reg_base) + : in_le32(fdev->reg_base); + ch_nr = (32 - ffs(gsr)) / 8; + + return fdev->chan[ch_nr] ? fsl_dma_chan_do_interrupt(irq, + fdev->chan[ch_nr]) : IRQ_NONE; +} + +static void dma_do_tasklet(unsigned long data) +{ + struct fsl_dma_chan *fsl_chan = (struct fsl_dma_chan *)data; + fsl_chan_ld_cleanup(fsl_chan); +} + +static int __devinit fsl_dma_chan_probe(struct fsl_dma_device *fdev, + struct device_node *node, u32 feature, const char *compatible) +{ + struct fsl_dma_chan *new_fsl_chan; + int err; + + /* alloc channel */ + new_fsl_chan = kzalloc(sizeof(struct fsl_dma_chan), GFP_KERNEL); + if (!new_fsl_chan) { + dev_err(fdev->dev, "No free memory for allocating " + "dma channels!\n"); + return -ENOMEM; + } + + /* get dma channel register base */ + err = of_address_to_resource(node, 0, &new_fsl_chan->reg); + if (err) { + dev_err(fdev->dev, "Can't get %s property 'reg'\n", + node->full_name); + goto err_no_reg; + } + + new_fsl_chan->feature = feature; + + if (!fdev->feature) + fdev->feature = new_fsl_chan->feature; + + /* If the DMA device's feature is different than its channels', + * report the bug. + */ + WARN_ON(fdev->feature != new_fsl_chan->feature); + + new_fsl_chan->dev = fdev->dev; + new_fsl_chan->reg_base = ioremap(new_fsl_chan->reg.start, + new_fsl_chan->reg.end - new_fsl_chan->reg.start + 1); + + new_fsl_chan->id = ((new_fsl_chan->reg.start - 0x100) & 0xfff) >> 7; + if (new_fsl_chan->id >= FSL_DMA_MAX_CHANS_PER_DEVICE) { + dev_err(fdev->dev, "There is no %d channel!\n", + new_fsl_chan->id); + err = -EINVAL; + goto err_no_chan; + } + fdev->chan[new_fsl_chan->id] = new_fsl_chan; + tasklet_init(&new_fsl_chan->tasklet, dma_do_tasklet, + (unsigned long)new_fsl_chan); + + /* Init the channel */ + dma_init(new_fsl_chan); + + /* Clear cdar registers */ + set_cdar(new_fsl_chan, 0); + + switch (new_fsl_chan->feature & FSL_DMA_IP_MASK) { + case FSL_DMA_IP_85XX: + new_fsl_chan->toggle_ext_pause = fsl_chan_toggle_ext_pause; + case FSL_DMA_IP_83XX: + new_fsl_chan->toggle_ext_start = fsl_chan_toggle_ext_start; + new_fsl_chan->set_src_loop_size = fsl_chan_set_src_loop_size; + new_fsl_chan->set_dest_loop_size = fsl_chan_set_dest_loop_size; + new_fsl_chan->set_request_count = fsl_chan_set_request_count; + } + + spin_lock_init(&new_fsl_chan->desc_lock); + INIT_LIST_HEAD(&new_fsl_chan->ld_queue); + + new_fsl_chan->common.device = &fdev->common; + + /* Add the channel to DMA device channel list */ + list_add_tail(&new_fsl_chan->common.device_node, + &fdev->common.channels); + fdev->common.chancnt++; + + new_fsl_chan->irq = irq_of_parse_and_map(node, 0); + if (new_fsl_chan->irq != NO_IRQ) { + err = request_irq(new_fsl_chan->irq, + &fsl_dma_chan_do_interrupt, IRQF_SHARED, + "fsldma-channel", new_fsl_chan); + if (err) { + dev_err(fdev->dev, "DMA channel %s request_irq error " + "with return %d\n", node->full_name, err); + goto err_no_irq; + } + } + + dev_info(fdev->dev, "#%d (%s), irq %d\n", new_fsl_chan->id, + compatible, + new_fsl_chan->irq != NO_IRQ ? new_fsl_chan->irq : fdev->irq); + + return 0; + +err_no_irq: + list_del(&new_fsl_chan->common.device_node); +err_no_chan: + iounmap(new_fsl_chan->reg_base); +err_no_reg: + kfree(new_fsl_chan); + return err; +} + +static void fsl_dma_chan_remove(struct fsl_dma_chan *fchan) +{ + if (fchan->irq != NO_IRQ) + free_irq(fchan->irq, fchan); + list_del(&fchan->common.device_node); + iounmap(fchan->reg_base); + kfree(fchan); +} + +static int __devinit of_fsl_dma_probe(struct of_device *dev, + const struct of_device_id *match) +{ + int err; + struct fsl_dma_device *fdev; + struct device_node *child; + + fdev = kzalloc(sizeof(struct fsl_dma_device), GFP_KERNEL); + if (!fdev) { + dev_err(&dev->dev, "No enough memory for 'priv'\n"); + return -ENOMEM; + } + fdev->dev = &dev->dev; + INIT_LIST_HEAD(&fdev->common.channels); + + /* get DMA controller register base */ + err = of_address_to_resource(dev->node, 0, &fdev->reg); + if (err) { + dev_err(&dev->dev, "Can't get %s property 'reg'\n", + dev->node->full_name); + goto err_no_reg; + } + + dev_info(&dev->dev, "Probe the Freescale DMA driver for %s " + "controller at 0x%llx...\n", + match->compatible, (unsigned long long)fdev->reg.start); + fdev->reg_base = ioremap(fdev->reg.start, fdev->reg.end + - fdev->reg.start + 1); + + dma_cap_set(DMA_MEMCPY, fdev->common.cap_mask); + dma_cap_set(DMA_INTERRUPT, fdev->common.cap_mask); + dma_cap_set(DMA_SLAVE, fdev->common.cap_mask); + fdev->common.device_alloc_chan_resources = fsl_dma_alloc_chan_resources; + fdev->common.device_free_chan_resources = fsl_dma_free_chan_resources; + fdev->common.device_prep_dma_interrupt = fsl_dma_prep_interrupt; + fdev->common.device_prep_dma_memcpy = fsl_dma_prep_memcpy; + fdev->common.device_is_tx_complete = fsl_dma_is_complete; + fdev->common.device_issue_pending = fsl_dma_memcpy_issue_pending; + fdev->common.device_prep_slave_sg = fsl_dma_prep_slave_sg; + fdev->common.device_terminate_all = fsl_dma_device_terminate_all; + fdev->common.dev = &dev->dev; + + fdev->irq = irq_of_parse_and_map(dev->node, 0); + if (fdev->irq != NO_IRQ) { + err = request_irq(fdev->irq, &fsl_dma_do_interrupt, IRQF_SHARED, + "fsldma-device", fdev); + if (err) { + dev_err(&dev->dev, "DMA device request_irq error " + "with return %d\n", err); + goto err; + } + } + + dev_set_drvdata(&(dev->dev), fdev); + + /* We cannot use of_platform_bus_probe() because there is no + * of_platform_bus_remove. Instead, we manually instantiate every DMA + * channel object. + */ + for_each_child_of_node(dev->node, child) { + if (of_device_is_compatible(child, "fsl,eloplus-dma-channel")) + fsl_dma_chan_probe(fdev, child, + FSL_DMA_IP_85XX | FSL_DMA_BIG_ENDIAN, + "fsl,eloplus-dma-channel"); + if (of_device_is_compatible(child, "fsl,elo-dma-channel")) + fsl_dma_chan_probe(fdev, child, + FSL_DMA_IP_83XX | FSL_DMA_LITTLE_ENDIAN, + "fsl,elo-dma-channel"); + } + + dma_async_device_register(&fdev->common); + return 0; + +err: + iounmap(fdev->reg_base); +err_no_reg: + kfree(fdev); + return err; +} + +static int of_fsl_dma_remove(struct of_device *of_dev) +{ + struct fsl_dma_device *fdev; + unsigned int i; + + fdev = dev_get_drvdata(&of_dev->dev); + + dma_async_device_unregister(&fdev->common); + + for (i = 0; i < FSL_DMA_MAX_CHANS_PER_DEVICE; i++) + if (fdev->chan[i]) + fsl_dma_chan_remove(fdev->chan[i]); + + if (fdev->irq != NO_IRQ) + free_irq(fdev->irq, fdev); + + iounmap(fdev->reg_base); + + kfree(fdev); + dev_set_drvdata(&of_dev->dev, NULL); + + return 0; +} + +static struct of_device_id of_fsl_dma_ids[] = { + { .compatible = "fsl,eloplus-dma", }, + { .compatible = "fsl,elo-dma", }, + {} +}; + +static struct of_platform_driver of_fsl_dma_driver = { + .name = "fsl-elo-dma", + .match_table = of_fsl_dma_ids, + .probe = of_fsl_dma_probe, + .remove = of_fsl_dma_remove, +}; + +static __init int of_fsl_dma_init(void) +{ + int ret; + + pr_info("Freescale Elo / Elo Plus DMA driver\n"); + + ret = of_register_platform_driver(&of_fsl_dma_driver); + if (ret) + pr_err("fsldma: failed to register platform driver\n"); + + return ret; +} + +static void __exit of_fsl_dma_exit(void) +{ + of_unregister_platform_driver(&of_fsl_dma_driver); +} + +subsys_initcall(of_fsl_dma_init); +module_exit(of_fsl_dma_exit); + +MODULE_DESCRIPTION("Freescale Elo / Elo Plus DMA driver"); +MODULE_LICENSE("GPL"); diff --git a/drivers/dma/fsldma.h b/drivers/dma/fsldma.h new file mode 100644 index 00000000000..0df14cbb8ca --- /dev/null +++ b/drivers/dma/fsldma.h @@ -0,0 +1,201 @@ +/* + * Copyright (C) 2007 Freescale Semiconductor, Inc. All rights reserved. + * + * Author: + * Zhang Wei <wei.zhang@freescale.com>, Jul 2007 + * Ebony Zhu <ebony.zhu@freescale.com>, May 2007 + * + * This is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + */ +#ifndef __DMA_FSLDMA_H +#define __DMA_FSLDMA_H + +#include <linux/device.h> +#include <linux/dmapool.h> +#include <linux/dmaengine.h> + +/* Define data structures needed by Freescale + * MPC8540 and MPC8349 DMA controller. + */ +#define FSL_DMA_MR_CS 0x00000001 +#define FSL_DMA_MR_CC 0x00000002 +#define FSL_DMA_MR_CA 0x00000008 +#define FSL_DMA_MR_EIE 0x00000040 +#define FSL_DMA_MR_XFE 0x00000020 +#define FSL_DMA_MR_EOLNIE 0x00000100 +#define FSL_DMA_MR_EOLSIE 0x00000080 +#define FSL_DMA_MR_EOSIE 0x00000200 +#define FSL_DMA_MR_CDSM 0x00000010 +#define FSL_DMA_MR_CTM 0x00000004 +#define FSL_DMA_MR_EMP_EN 0x00200000 +#define FSL_DMA_MR_EMS_EN 0x00040000 +#define FSL_DMA_MR_DAHE 0x00002000 +#define FSL_DMA_MR_SAHE 0x00001000 + +/* Special MR definition for MPC8349 */ +#define FSL_DMA_MR_EOTIE 0x00000080 +#define FSL_DMA_MR_PRC_RM 0x00000800 + +#define FSL_DMA_SR_CH 0x00000020 +#define FSL_DMA_SR_PE 0x00000010 +#define FSL_DMA_SR_CB 0x00000004 +#define FSL_DMA_SR_TE 0x00000080 +#define FSL_DMA_SR_EOSI 0x00000002 +#define FSL_DMA_SR_EOLSI 0x00000001 +#define FSL_DMA_SR_EOCDI 0x00000001 +#define FSL_DMA_SR_EOLNI 0x00000008 + +#define FSL_DMA_SATR_SBPATMU 0x20000000 +#define FSL_DMA_SATR_STRANSINT_RIO 0x00c00000 +#define FSL_DMA_SATR_SREADTYPE_SNOOP_READ 0x00050000 +#define FSL_DMA_SATR_SREADTYPE_BP_IORH 0x00020000 +#define FSL_DMA_SATR_SREADTYPE_BP_NREAD 0x00040000 +#define FSL_DMA_SATR_SREADTYPE_BP_MREAD 0x00070000 + +#define FSL_DMA_DATR_DBPATMU 0x20000000 +#define FSL_DMA_DATR_DTRANSINT_RIO 0x00c00000 +#define FSL_DMA_DATR_DWRITETYPE_SNOOP_WRITE 0x00050000 +#define FSL_DMA_DATR_DWRITETYPE_BP_FLUSH 0x00010000 + +#define FSL_DMA_EOL ((u64)0x1) +#define FSL_DMA_SNEN ((u64)0x10) +#define FSL_DMA_EOSIE 0x8 +#define FSL_DMA_NLDA_MASK (~(u64)0x1f) + +#define FSL_DMA_BCR_MAX_CNT 0x03ffffffu + +#define FSL_DMA_DGSR_TE 0x80 +#define FSL_DMA_DGSR_CH 0x20 +#define FSL_DMA_DGSR_PE 0x10 +#define FSL_DMA_DGSR_EOLNI 0x08 +#define FSL_DMA_DGSR_CB 0x04 +#define FSL_DMA_DGSR_EOSI 0x02 +#define FSL_DMA_DGSR_EOLSI 0x01 + +typedef u64 __bitwise v64; +typedef u32 __bitwise v32; + +struct fsl_dma_ld_hw { + v64 src_addr; + v64 dst_addr; + v64 next_ln_addr; + v32 count; + v32 reserve; +} __attribute__((aligned(32))); + +struct fsl_desc_sw { + struct fsl_dma_ld_hw hw; + struct list_head node; + struct list_head tx_list; + struct dma_async_tx_descriptor async_tx; + struct list_head *ld; + void *priv; +} __attribute__((aligned(32))); + +struct fsl_dma_chan_regs { + u32 mr; /* 0x00 - Mode Register */ + u32 sr; /* 0x04 - Status Register */ + u64 cdar; /* 0x08 - Current descriptor address register */ + u64 sar; /* 0x10 - Source Address Register */ + u64 dar; /* 0x18 - Destination Address Register */ + u32 bcr; /* 0x20 - Byte Count Register */ + u64 ndar; /* 0x24 - Next Descriptor Address Register */ +}; + +struct fsl_dma_chan; +#define FSL_DMA_MAX_CHANS_PER_DEVICE 4 + +struct fsl_dma_device { + void __iomem *reg_base; /* DGSR register base */ + struct resource reg; /* Resource for register */ + struct device *dev; + struct dma_device common; + struct fsl_dma_chan *chan[FSL_DMA_MAX_CHANS_PER_DEVICE]; + u32 feature; /* The same as DMA channels */ + int irq; /* Channel IRQ */ +}; + +/* Define macros for fsl_dma_chan->feature property */ +#define FSL_DMA_LITTLE_ENDIAN 0x00000000 +#define FSL_DMA_BIG_ENDIAN 0x00000001 + +#define FSL_DMA_IP_MASK 0x00000ff0 +#define FSL_DMA_IP_85XX 0x00000010 +#define FSL_DMA_IP_83XX 0x00000020 + +#define FSL_DMA_CHAN_PAUSE_EXT 0x00001000 +#define FSL_DMA_CHAN_START_EXT 0x00002000 + +struct fsl_dma_chan { + struct fsl_dma_chan_regs __iomem *reg_base; + dma_cookie_t completed_cookie; /* The maximum cookie completed */ + spinlock_t desc_lock; /* Descriptor operation lock */ + struct list_head ld_queue; /* Link descriptors queue */ + struct dma_chan common; /* DMA common channel */ + struct dma_pool *desc_pool; /* Descriptors pool */ + struct device *dev; /* Channel device */ + struct resource reg; /* Resource for register */ + int irq; /* Channel IRQ */ + int id; /* Raw id of this channel */ + struct tasklet_struct tasklet; + u32 feature; + + void (*toggle_ext_pause)(struct fsl_dma_chan *fsl_chan, int enable); + void (*toggle_ext_start)(struct fsl_dma_chan *fsl_chan, int enable); + void (*set_src_loop_size)(struct fsl_dma_chan *fsl_chan, int size); + void (*set_dest_loop_size)(struct fsl_dma_chan *fsl_chan, int size); + void (*set_request_count)(struct fsl_dma_chan *fsl_chan, int size); +}; + +#define to_fsl_chan(chan) container_of(chan, struct fsl_dma_chan, common) +#define to_fsl_desc(lh) container_of(lh, struct fsl_desc_sw, node) +#define tx_to_fsl_desc(tx) container_of(tx, struct fsl_desc_sw, async_tx) + +#ifndef __powerpc64__ +static u64 in_be64(const u64 __iomem *addr) +{ + return ((u64)in_be32((u32 __iomem *)addr) << 32) | + (in_be32((u32 __iomem *)addr + 1)); +} + +static void out_be64(u64 __iomem *addr, u64 val) +{ + out_be32((u32 __iomem *)addr, val >> 32); + out_be32((u32 __iomem *)addr + 1, (u32)val); +} + +/* There is no asm instructions for 64 bits reverse loads and stores */ +static u64 in_le64(const u64 __iomem *addr) +{ + return ((u64)in_le32((u32 __iomem *)addr + 1) << 32) | + (in_le32((u32 __iomem *)addr)); +} + +static void out_le64(u64 __iomem *addr, u64 val) +{ + out_le32((u32 __iomem *)addr + 1, val >> 32); + out_le32((u32 __iomem *)addr, (u32)val); +} +#endif + +#define DMA_IN(fsl_chan, addr, width) \ + (((fsl_chan)->feature & FSL_DMA_BIG_ENDIAN) ? \ + in_be##width(addr) : in_le##width(addr)) +#define DMA_OUT(fsl_chan, addr, val, width) \ + (((fsl_chan)->feature & FSL_DMA_BIG_ENDIAN) ? \ + out_be##width(addr, val) : out_le##width(addr, val)) + +#define DMA_TO_CPU(fsl_chan, d, width) \ + (((fsl_chan)->feature & FSL_DMA_BIG_ENDIAN) ? \ + be##width##_to_cpu((__force __be##width)(v##width)d) : \ + le##width##_to_cpu((__force __le##width)(v##width)d)) +#define CPU_TO_DMA(fsl_chan, c, width) \ + (((fsl_chan)->feature & FSL_DMA_BIG_ENDIAN) ? \ + (__force v##width)cpu_to_be##width(c) : \ + (__force v##width)cpu_to_le##width(c)) + +#endif /* __DMA_FSLDMA_H */ diff --git a/drivers/dma/ioat/Makefile b/drivers/dma/ioat/Makefile new file mode 100644 index 00000000000..8997d3fb905 --- /dev/null +++ b/drivers/dma/ioat/Makefile @@ -0,0 +1,2 @@ +obj-$(CONFIG_INTEL_IOATDMA) += ioatdma.o +ioatdma-objs := pci.o dma.o dma_v2.o dma_v3.o dca.o diff --git a/drivers/dma/ioat/dca.c b/drivers/dma/ioat/dca.c new file mode 100644 index 00000000000..abd9038e06b --- /dev/null +++ b/drivers/dma/ioat/dca.c @@ -0,0 +1,684 @@ +/* + * Intel I/OAT DMA Linux driver + * Copyright(c) 2007 - 2009 Intel Corporation. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + * + * The full GNU General Public License is included in this distribution in + * the file called "COPYING". + * + */ + +#include <linux/kernel.h> +#include <linux/pci.h> +#include <linux/smp.h> +#include <linux/interrupt.h> +#include <linux/dca.h> + +/* either a kernel change is needed, or we need something like this in kernel */ +#ifndef CONFIG_SMP +#include <asm/smp.h> +#undef cpu_physical_id +#define cpu_physical_id(cpu) (cpuid_ebx(1) >> 24) +#endif + +#include "dma.h" +#include "registers.h" + +/* + * Bit 7 of a tag map entry is the "valid" bit, if it is set then bits 0:6 + * contain the bit number of the APIC ID to map into the DCA tag. If the valid + * bit is not set, then the value must be 0 or 1 and defines the bit in the tag. + */ +#define DCA_TAG_MAP_VALID 0x80 + +#define DCA3_TAG_MAP_BIT_TO_INV 0x80 +#define DCA3_TAG_MAP_BIT_TO_SEL 0x40 +#define DCA3_TAG_MAP_LITERAL_VAL 0x1 + +#define DCA_TAG_MAP_MASK 0xDF + +/* expected tag map bytes for I/OAT ver.2 */ +#define DCA2_TAG_MAP_BYTE0 0x80 +#define DCA2_TAG_MAP_BYTE1 0x0 +#define DCA2_TAG_MAP_BYTE2 0x81 +#define DCA2_TAG_MAP_BYTE3 0x82 +#define DCA2_TAG_MAP_BYTE4 0x82 + +/* verify if tag map matches expected values */ +static inline int dca2_tag_map_valid(u8 *tag_map) +{ + return ((tag_map[0] == DCA2_TAG_MAP_BYTE0) && + (tag_map[1] == DCA2_TAG_MAP_BYTE1) && + (tag_map[2] == DCA2_TAG_MAP_BYTE2) && + (tag_map[3] == DCA2_TAG_MAP_BYTE3) && + (tag_map[4] == DCA2_TAG_MAP_BYTE4)); +} + +/* + * "Legacy" DCA systems do not implement the DCA register set in the + * I/OAT device. Software needs direct support for their tag mappings. + */ + +#define APICID_BIT(x) (DCA_TAG_MAP_VALID | (x)) +#define IOAT_TAG_MAP_LEN 8 + +static u8 ioat_tag_map_BNB[IOAT_TAG_MAP_LEN] = { + 1, APICID_BIT(1), APICID_BIT(2), APICID_BIT(2), }; +static u8 ioat_tag_map_SCNB[IOAT_TAG_MAP_LEN] = { + 1, APICID_BIT(1), APICID_BIT(2), APICID_BIT(2), }; +static u8 ioat_tag_map_CNB[IOAT_TAG_MAP_LEN] = { + 1, APICID_BIT(1), APICID_BIT(3), APICID_BIT(4), APICID_BIT(2), }; +static u8 ioat_tag_map_UNISYS[IOAT_TAG_MAP_LEN] = { 0 }; + +/* pack PCI B/D/F into a u16 */ +static inline u16 dcaid_from_pcidev(struct pci_dev *pci) +{ + return (pci->bus->number << 8) | pci->devfn; +} + +static int dca_enabled_in_bios(struct pci_dev *pdev) +{ + /* CPUID level 9 returns DCA configuration */ + /* Bit 0 indicates DCA enabled by the BIOS */ + unsigned long cpuid_level_9; + int res; + + cpuid_level_9 = cpuid_eax(9); + res = test_bit(0, &cpuid_level_9); + if (!res) + dev_dbg(&pdev->dev, "DCA is disabled in BIOS\n"); + + return res; +} + +int system_has_dca_enabled(struct pci_dev *pdev) +{ + if (boot_cpu_has(X86_FEATURE_DCA)) + return dca_enabled_in_bios(pdev); + + dev_dbg(&pdev->dev, "boot cpu doesn't have X86_FEATURE_DCA\n"); + return 0; +} + +struct ioat_dca_slot { + struct pci_dev *pdev; /* requester device */ + u16 rid; /* requester id, as used by IOAT */ +}; + +#define IOAT_DCA_MAX_REQ 6 +#define IOAT3_DCA_MAX_REQ 2 + +struct ioat_dca_priv { + void __iomem *iobase; + void __iomem *dca_base; + int max_requesters; + int requester_count; + u8 tag_map[IOAT_TAG_MAP_LEN]; + struct ioat_dca_slot req_slots[0]; +}; + +/* 5000 series chipset DCA Port Requester ID Table Entry Format + * [15:8] PCI-Express Bus Number + * [7:3] PCI-Express Device Number + * [2:0] PCI-Express Function Number + * + * 5000 series chipset DCA control register format + * [7:1] Reserved (0) + * [0] Ignore Function Number + */ + +static int ioat_dca_add_requester(struct dca_provider *dca, struct device *dev) +{ + struct ioat_dca_priv *ioatdca = dca_priv(dca); + struct pci_dev *pdev; + int i; + u16 id; + + /* This implementation only supports PCI-Express */ + if (dev->bus != &pci_bus_type) + return -ENODEV; + pdev = to_pci_dev(dev); + id = dcaid_from_pcidev(pdev); + + if (ioatdca->requester_count == ioatdca->max_requesters) + return -ENODEV; + + for (i = 0; i < ioatdca->max_requesters; i++) { + if (ioatdca->req_slots[i].pdev == NULL) { + /* found an empty slot */ + ioatdca->requester_count++; + ioatdca->req_slots[i].pdev = pdev; + ioatdca->req_slots[i].rid = id; + writew(id, ioatdca->dca_base + (i * 4)); + /* make sure the ignore function bit is off */ + writeb(0, ioatdca->dca_base + (i * 4) + 2); + return i; + } + } + /* Error, ioatdma->requester_count is out of whack */ + return -EFAULT; +} + +static int ioat_dca_remove_requester(struct dca_provider *dca, + struct device *dev) +{ + struct ioat_dca_priv *ioatdca = dca_priv(dca); + struct pci_dev *pdev; + int i; + + /* This implementation only supports PCI-Express */ + if (dev->bus != &pci_bus_type) + return -ENODEV; + pdev = to_pci_dev(dev); + + for (i = 0; i < ioatdca->max_requesters; i++) { + if (ioatdca->req_slots[i].pdev == pdev) { + writew(0, ioatdca->dca_base + (i * 4)); + ioatdca->req_slots[i].pdev = NULL; + ioatdca->req_slots[i].rid = 0; + ioatdca->requester_count--; + return i; + } + } + return -ENODEV; +} + +static u8 ioat_dca_get_tag(struct dca_provider *dca, + struct device *dev, + int cpu) +{ + struct ioat_dca_priv *ioatdca = dca_priv(dca); + int i, apic_id, bit, value; + u8 entry, tag; + + tag = 0; + apic_id = cpu_physical_id(cpu); + + for (i = 0; i < IOAT_TAG_MAP_LEN; i++) { + entry = ioatdca->tag_map[i]; + if (entry & DCA_TAG_MAP_VALID) { + bit = entry & ~DCA_TAG_MAP_VALID; + value = (apic_id & (1 << bit)) ? 1 : 0; + } else { + value = entry ? 1 : 0; + } + tag |= (value << i); + } + return tag; +} + +static int ioat_dca_dev_managed(struct dca_provider *dca, + struct device *dev) +{ + struct ioat_dca_priv *ioatdca = dca_priv(dca); + struct pci_dev *pdev; + int i; + + pdev = to_pci_dev(dev); + for (i = 0; i < ioatdca->max_requesters; i++) { + if (ioatdca->req_slots[i].pdev == pdev) + return 1; + } + return 0; +} + +static struct dca_ops ioat_dca_ops = { + .add_requester = ioat_dca_add_requester, + .remove_requester = ioat_dca_remove_requester, + .get_tag = ioat_dca_get_tag, + .dev_managed = ioat_dca_dev_managed, +}; + + +struct dca_provider * __devinit +ioat_dca_init(struct pci_dev *pdev, void __iomem *iobase) +{ + struct dca_provider *dca; + struct ioat_dca_priv *ioatdca; + u8 *tag_map = NULL; + int i; + int err; + u8 version; + u8 max_requesters; + + if (!system_has_dca_enabled(pdev)) + return NULL; + + /* I/OAT v1 systems must have a known tag_map to support DCA */ + switch (pdev->vendor) { + case PCI_VENDOR_ID_INTEL: + switch (pdev->device) { + case PCI_DEVICE_ID_INTEL_IOAT: + tag_map = ioat_tag_map_BNB; + break; + case PCI_DEVICE_ID_INTEL_IOAT_CNB: + tag_map = ioat_tag_map_CNB; + break; + case PCI_DEVICE_ID_INTEL_IOAT_SCNB: + tag_map = ioat_tag_map_SCNB; + break; + } + break; + case PCI_VENDOR_ID_UNISYS: + switch (pdev->device) { + case PCI_DEVICE_ID_UNISYS_DMA_DIRECTOR: + tag_map = ioat_tag_map_UNISYS; + break; + } + break; + } + if (tag_map == NULL) + return NULL; + + version = readb(iobase + IOAT_VER_OFFSET); + if (version == IOAT_VER_3_0) + max_requesters = IOAT3_DCA_MAX_REQ; + else + max_requesters = IOAT_DCA_MAX_REQ; + + dca = alloc_dca_provider(&ioat_dca_ops, + sizeof(*ioatdca) + + (sizeof(struct ioat_dca_slot) * max_requesters)); + if (!dca) + return NULL; + + ioatdca = dca_priv(dca); + ioatdca->max_requesters = max_requesters; + ioatdca->dca_base = iobase + 0x54; + + /* copy over the APIC ID to DCA tag mapping */ + for (i = 0; i < IOAT_TAG_MAP_LEN; i++) + ioatdca->tag_map[i] = tag_map[i]; + + err = register_dca_provider(dca, &pdev->dev); + if (err) { + free_dca_provider(dca); + return NULL; + } + + return dca; +} + + +static int ioat2_dca_add_requester(struct dca_provider *dca, struct device *dev) +{ + struct ioat_dca_priv *ioatdca = dca_priv(dca); + struct pci_dev *pdev; + int i; + u16 id; + u16 global_req_table; + + /* This implementation only supports PCI-Express */ + if (dev->bus != &pci_bus_type) + return -ENODEV; + pdev = to_pci_dev(dev); + id = dcaid_from_pcidev(pdev); + + if (ioatdca->requester_count == ioatdca->max_requesters) + return -ENODEV; + + for (i = 0; i < ioatdca->max_requesters; i++) { + if (ioatdca->req_slots[i].pdev == NULL) { + /* found an empty slot */ + ioatdca->requester_count++; + ioatdca->req_slots[i].pdev = pdev; + ioatdca->req_slots[i].rid = id; + global_req_table = + readw(ioatdca->dca_base + IOAT_DCA_GREQID_OFFSET); + writel(id | IOAT_DCA_GREQID_VALID, + ioatdca->iobase + global_req_table + (i * 4)); + return i; + } + } + /* Error, ioatdma->requester_count is out of whack */ + return -EFAULT; +} + +static int ioat2_dca_remove_requester(struct dca_provider *dca, + struct device *dev) +{ + struct ioat_dca_priv *ioatdca = dca_priv(dca); + struct pci_dev *pdev; + int i; + u16 global_req_table; + + /* This implementation only supports PCI-Express */ + if (dev->bus != &pci_bus_type) + return -ENODEV; + pdev = to_pci_dev(dev); + + for (i = 0; i < ioatdca->max_requesters; i++) { + if (ioatdca->req_slots[i].pdev == pdev) { + global_req_table = + readw(ioatdca->dca_base + IOAT_DCA_GREQID_OFFSET); + writel(0, ioatdca->iobase + global_req_table + (i * 4)); + ioatdca->req_slots[i].pdev = NULL; + ioatdca->req_slots[i].rid = 0; + ioatdca->requester_count--; + return i; + } + } + return -ENODEV; +} + +static u8 ioat2_dca_get_tag(struct dca_provider *dca, + struct device *dev, + int cpu) +{ + u8 tag; + + tag = ioat_dca_get_tag(dca, dev, cpu); + tag = (~tag) & 0x1F; + return tag; +} + +static struct dca_ops ioat2_dca_ops = { + .add_requester = ioat2_dca_add_requester, + .remove_requester = ioat2_dca_remove_requester, + .get_tag = ioat2_dca_get_tag, + .dev_managed = ioat_dca_dev_managed, +}; + +static int ioat2_dca_count_dca_slots(void __iomem *iobase, u16 dca_offset) +{ + int slots = 0; + u32 req; + u16 global_req_table; + + global_req_table = readw(iobase + dca_offset + IOAT_DCA_GREQID_OFFSET); + if (global_req_table == 0) + return 0; + do { + req = readl(iobase + global_req_table + (slots * sizeof(u32))); + slots++; + } while ((req & IOAT_DCA_GREQID_LASTID) == 0); + + return slots; +} + +struct dca_provider * __devinit +ioat2_dca_init(struct pci_dev *pdev, void __iomem *iobase) +{ + struct dca_provider *dca; + struct ioat_dca_priv *ioatdca; + int slots; + int i; + int err; + u32 tag_map; + u16 dca_offset; + u16 csi_fsb_control; + u16 pcie_control; + u8 bit; + + if (!system_has_dca_enabled(pdev)) + return NULL; + + dca_offset = readw(iobase + IOAT_DCAOFFSET_OFFSET); + if (dca_offset == 0) + return NULL; + + slots = ioat2_dca_count_dca_slots(iobase, dca_offset); + if (slots == 0) + return NULL; + + dca = alloc_dca_provider(&ioat2_dca_ops, + sizeof(*ioatdca) + + (sizeof(struct ioat_dca_slot) * slots)); + if (!dca) + return NULL; + + ioatdca = dca_priv(dca); + ioatdca->iobase = iobase; + ioatdca->dca_base = iobase + dca_offset; + ioatdca->max_requesters = slots; + + /* some bios might not know to turn these on */ + csi_fsb_control = readw(ioatdca->dca_base + IOAT_FSB_CAP_ENABLE_OFFSET); + if ((csi_fsb_control & IOAT_FSB_CAP_ENABLE_PREFETCH) == 0) { + csi_fsb_control |= IOAT_FSB_CAP_ENABLE_PREFETCH; + writew(csi_fsb_control, + ioatdca->dca_base + IOAT_FSB_CAP_ENABLE_OFFSET); + } + pcie_control = readw(ioatdca->dca_base + IOAT_PCI_CAP_ENABLE_OFFSET); + if ((pcie_control & IOAT_PCI_CAP_ENABLE_MEMWR) == 0) { + pcie_control |= IOAT_PCI_CAP_ENABLE_MEMWR; + writew(pcie_control, + ioatdca->dca_base + IOAT_PCI_CAP_ENABLE_OFFSET); + } + + + /* TODO version, compatibility and configuration checks */ + + /* copy out the APIC to DCA tag map */ + tag_map = readl(ioatdca->dca_base + IOAT_APICID_TAG_MAP_OFFSET); + for (i = 0; i < 5; i++) { + bit = (tag_map >> (4 * i)) & 0x0f; + if (bit < 8) + ioatdca->tag_map[i] = bit | DCA_TAG_MAP_VALID; + else + ioatdca->tag_map[i] = 0; + } + + if (!dca2_tag_map_valid(ioatdca->tag_map)) { + dev_err(&pdev->dev, "APICID_TAG_MAP set incorrectly by BIOS, " + "disabling DCA\n"); + free_dca_provider(dca); + return NULL; + } + + err = register_dca_provider(dca, &pdev->dev); + if (err) { + free_dca_provider(dca); + return NULL; + } + + return dca; +} + +static int ioat3_dca_add_requester(struct dca_provider *dca, struct device *dev) +{ + struct ioat_dca_priv *ioatdca = dca_priv(dca); + struct pci_dev *pdev; + int i; + u16 id; + u16 global_req_table; + + /* This implementation only supports PCI-Express */ + if (dev->bus != &pci_bus_type) + return -ENODEV; + pdev = to_pci_dev(dev); + id = dcaid_from_pcidev(pdev); + + if (ioatdca->requester_count == ioatdca->max_requesters) + return -ENODEV; + + for (i = 0; i < ioatdca->max_requesters; i++) { + if (ioatdca->req_slots[i].pdev == NULL) { + /* found an empty slot */ + ioatdca->requester_count++; + ioatdca->req_slots[i].pdev = pdev; + ioatdca->req_slots[i].rid = id; + global_req_table = + readw(ioatdca->dca_base + IOAT3_DCA_GREQID_OFFSET); + writel(id | IOAT_DCA_GREQID_VALID, + ioatdca->iobase + global_req_table + (i * 4)); + return i; + } + } + /* Error, ioatdma->requester_count is out of whack */ + return -EFAULT; +} + +static int ioat3_dca_remove_requester(struct dca_provider *dca, + struct device *dev) +{ + struct ioat_dca_priv *ioatdca = dca_priv(dca); + struct pci_dev *pdev; + int i; + u16 global_req_table; + + /* This implementation only supports PCI-Express */ + if (dev->bus != &pci_bus_type) + return -ENODEV; + pdev = to_pci_dev(dev); + + for (i = 0; i < ioatdca->max_requesters; i++) { + if (ioatdca->req_slots[i].pdev == pdev) { + global_req_table = + readw(ioatdca->dca_base + IOAT3_DCA_GREQID_OFFSET); + writel(0, ioatdca->iobase + global_req_table + (i * 4)); + ioatdca->req_slots[i].pdev = NULL; + ioatdca->req_slots[i].rid = 0; + ioatdca->requester_count--; + return i; + } + } + return -ENODEV; +} + +static u8 ioat3_dca_get_tag(struct dca_provider *dca, + struct device *dev, + int cpu) +{ + u8 tag; + + struct ioat_dca_priv *ioatdca = dca_priv(dca); + int i, apic_id, bit, value; + u8 entry; + + tag = 0; + apic_id = cpu_physical_id(cpu); + + for (i = 0; i < IOAT_TAG_MAP_LEN; i++) { + entry = ioatdca->tag_map[i]; + if (entry & DCA3_TAG_MAP_BIT_TO_SEL) { + bit = entry & + ~(DCA3_TAG_MAP_BIT_TO_SEL | DCA3_TAG_MAP_BIT_TO_INV); + value = (apic_id & (1 << bit)) ? 1 : 0; + } else if (entry & DCA3_TAG_MAP_BIT_TO_INV) { + bit = entry & ~DCA3_TAG_MAP_BIT_TO_INV; + value = (apic_id & (1 << bit)) ? 0 : 1; + } else { + value = (entry & DCA3_TAG_MAP_LITERAL_VAL) ? 1 : 0; + } + tag |= (value << i); + } + + return tag; +} + +static struct dca_ops ioat3_dca_ops = { + .add_requester = ioat3_dca_add_requester, + .remove_requester = ioat3_dca_remove_requester, + .get_tag = ioat3_dca_get_tag, + .dev_managed = ioat_dca_dev_managed, +}; + +static int ioat3_dca_count_dca_slots(void *iobase, u16 dca_offset) +{ + int slots = 0; + u32 req; + u16 global_req_table; + + global_req_table = readw(iobase + dca_offset + IOAT3_DCA_GREQID_OFFSET); + if (global_req_table == 0) + return 0; + + do { + req = readl(iobase + global_req_table + (slots * sizeof(u32))); + slots++; + } while ((req & IOAT_DCA_GREQID_LASTID) == 0); + + return slots; +} + +struct dca_provider * __devinit +ioat3_dca_init(struct pci_dev *pdev, void __iomem *iobase) +{ + struct dca_provider *dca; + struct ioat_dca_priv *ioatdca; + int slots; + int i; + int err; + u16 dca_offset; + u16 csi_fsb_control; + u16 pcie_control; + u8 bit; + + union { + u64 full; + struct { + u32 low; + u32 high; + }; + } tag_map; + + if (!system_has_dca_enabled(pdev)) + return NULL; + + dca_offset = readw(iobase + IOAT_DCAOFFSET_OFFSET); + if (dca_offset == 0) + return NULL; + + slots = ioat3_dca_count_dca_slots(iobase, dca_offset); + if (slots == 0) + return NULL; + + dca = alloc_dca_provider(&ioat3_dca_ops, + sizeof(*ioatdca) + + (sizeof(struct ioat_dca_slot) * slots)); + if (!dca) + return NULL; + + ioatdca = dca_priv(dca); + ioatdca->iobase = iobase; + ioatdca->dca_base = iobase + dca_offset; + ioatdca->max_requesters = slots; + + /* some bios might not know to turn these on */ + csi_fsb_control = readw(ioatdca->dca_base + IOAT3_CSI_CONTROL_OFFSET); + if ((csi_fsb_control & IOAT3_CSI_CONTROL_PREFETCH) == 0) { + csi_fsb_control |= IOAT3_CSI_CONTROL_PREFETCH; + writew(csi_fsb_control, + ioatdca->dca_base + IOAT3_CSI_CONTROL_OFFSET); + } + pcie_control = readw(ioatdca->dca_base + IOAT3_PCI_CONTROL_OFFSET); + if ((pcie_control & IOAT3_PCI_CONTROL_MEMWR) == 0) { + pcie_control |= IOAT3_PCI_CONTROL_MEMWR; + writew(pcie_control, + ioatdca->dca_base + IOAT3_PCI_CONTROL_OFFSET); + } + + + /* TODO version, compatibility and configuration checks */ + + /* copy out the APIC to DCA tag map */ + tag_map.low = + readl(ioatdca->dca_base + IOAT3_APICID_TAG_MAP_OFFSET_LOW); + tag_map.high = + readl(ioatdca->dca_base + IOAT3_APICID_TAG_MAP_OFFSET_HIGH); + for (i = 0; i < 8; i++) { + bit = tag_map.full >> (8 * i); + ioatdca->tag_map[i] = bit & DCA_TAG_MAP_MASK; + } + + err = register_dca_provider(dca, &pdev->dev); + if (err) { + free_dca_provider(dca); + return NULL; + } + + return dca; +} diff --git a/drivers/dma/ioat/dma.c b/drivers/dma/ioat/dma.c new file mode 100644 index 00000000000..dcc4ab78b32 --- /dev/null +++ b/drivers/dma/ioat/dma.c @@ -0,0 +1,1238 @@ +/* + * Intel I/OAT DMA Linux driver + * Copyright(c) 2004 - 2009 Intel Corporation. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + * + * The full GNU General Public License is included in this distribution in + * the file called "COPYING". + * + */ + +/* + * This driver supports an Intel I/OAT DMA engine, which does asynchronous + * copy operations. + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/interrupt.h> +#include <linux/dmaengine.h> +#include <linux/delay.h> +#include <linux/dma-mapping.h> +#include <linux/workqueue.h> +#include <linux/i7300_idle.h> +#include "dma.h" +#include "registers.h" +#include "hw.h" + +int ioat_pending_level = 4; +module_param(ioat_pending_level, int, 0644); +MODULE_PARM_DESC(ioat_pending_level, + "high-water mark for pushing ioat descriptors (default: 4)"); + +/* internal functions */ +static void ioat1_cleanup(struct ioat_dma_chan *ioat); +static void ioat1_dma_start_null_desc(struct ioat_dma_chan *ioat); + +/** + * ioat_dma_do_interrupt - handler used for single vector interrupt mode + * @irq: interrupt id + * @data: interrupt data + */ +static irqreturn_t ioat_dma_do_interrupt(int irq, void *data) +{ + struct ioatdma_device *instance = data; + struct ioat_chan_common *chan; + unsigned long attnstatus; + int bit; + u8 intrctrl; + + intrctrl = readb(instance->reg_base + IOAT_INTRCTRL_OFFSET); + + if (!(intrctrl & IOAT_INTRCTRL_MASTER_INT_EN)) + return IRQ_NONE; + + if (!(intrctrl & IOAT_INTRCTRL_INT_STATUS)) { + writeb(intrctrl, instance->reg_base + IOAT_INTRCTRL_OFFSET); + return IRQ_NONE; + } + + attnstatus = readl(instance->reg_base + IOAT_ATTNSTATUS_OFFSET); + for_each_bit(bit, &attnstatus, BITS_PER_LONG) { + chan = ioat_chan_by_index(instance, bit); + tasklet_schedule(&chan->cleanup_task); + } + + writeb(intrctrl, instance->reg_base + IOAT_INTRCTRL_OFFSET); + return IRQ_HANDLED; +} + +/** + * ioat_dma_do_interrupt_msix - handler used for vector-per-channel interrupt mode + * @irq: interrupt id + * @data: interrupt data + */ +static irqreturn_t ioat_dma_do_interrupt_msix(int irq, void *data) +{ + struct ioat_chan_common *chan = data; + + tasklet_schedule(&chan->cleanup_task); + + return IRQ_HANDLED; +} + +static void ioat1_cleanup_tasklet(unsigned long data); + +/* common channel initialization */ +void ioat_init_channel(struct ioatdma_device *device, + struct ioat_chan_common *chan, int idx, + void (*timer_fn)(unsigned long), + void (*tasklet)(unsigned long), + unsigned long ioat) +{ + struct dma_device *dma = &device->common; + + chan->device = device; + chan->reg_base = device->reg_base + (0x80 * (idx + 1)); + spin_lock_init(&chan->cleanup_lock); + chan->common.device = dma; + list_add_tail(&chan->common.device_node, &dma->channels); + device->idx[idx] = chan; + init_timer(&chan->timer); + chan->timer.function = timer_fn; + chan->timer.data = ioat; + tasklet_init(&chan->cleanup_task, tasklet, ioat); + tasklet_disable(&chan->cleanup_task); +} + +static void ioat1_timer_event(unsigned long data); + +/** + * ioat1_dma_enumerate_channels - find and initialize the device's channels + * @device: the device to be enumerated + */ +static int ioat1_enumerate_channels(struct ioatdma_device *device) +{ + u8 xfercap_scale; + u32 xfercap; + int i; + struct ioat_dma_chan *ioat; + struct device *dev = &device->pdev->dev; + struct dma_device *dma = &device->common; + + INIT_LIST_HEAD(&dma->channels); + dma->chancnt = readb(device->reg_base + IOAT_CHANCNT_OFFSET); + dma->chancnt &= 0x1f; /* bits [4:0] valid */ + if (dma->chancnt > ARRAY_SIZE(device->idx)) { + dev_warn(dev, "(%d) exceeds max supported channels (%zu)\n", + dma->chancnt, ARRAY_SIZE(device->idx)); + dma->chancnt = ARRAY_SIZE(device->idx); + } + xfercap_scale = readb(device->reg_base + IOAT_XFERCAP_OFFSET); + xfercap_scale &= 0x1f; /* bits [4:0] valid */ + xfercap = (xfercap_scale == 0 ? -1 : (1UL << xfercap_scale)); + dev_dbg(dev, "%s: xfercap = %d\n", __func__, xfercap); + +#ifdef CONFIG_I7300_IDLE_IOAT_CHANNEL + if (i7300_idle_platform_probe(NULL, NULL, 1) == 0) + dma->chancnt--; +#endif + for (i = 0; i < dma->chancnt; i++) { + ioat = devm_kzalloc(dev, sizeof(*ioat), GFP_KERNEL); + if (!ioat) + break; + + ioat_init_channel(device, &ioat->base, i, + ioat1_timer_event, + ioat1_cleanup_tasklet, + (unsigned long) ioat); + ioat->xfercap = xfercap; + spin_lock_init(&ioat->desc_lock); + INIT_LIST_HEAD(&ioat->free_desc); + INIT_LIST_HEAD(&ioat->used_desc); + } + dma->chancnt = i; + return i; +} + +/** + * ioat_dma_memcpy_issue_pending - push potentially unrecognized appended + * descriptors to hw + * @chan: DMA channel handle + */ +static inline void +__ioat1_dma_memcpy_issue_pending(struct ioat_dma_chan *ioat) +{ + void __iomem *reg_base = ioat->base.reg_base; + + dev_dbg(to_dev(&ioat->base), "%s: pending: %d\n", + __func__, ioat->pending); + ioat->pending = 0; + writeb(IOAT_CHANCMD_APPEND, reg_base + IOAT1_CHANCMD_OFFSET); +} + +static void ioat1_dma_memcpy_issue_pending(struct dma_chan *chan) +{ + struct ioat_dma_chan *ioat = to_ioat_chan(chan); + + if (ioat->pending > 0) { + spin_lock_bh(&ioat->desc_lock); + __ioat1_dma_memcpy_issue_pending(ioat); + spin_unlock_bh(&ioat->desc_lock); + } +} + +/** + * ioat1_reset_channel - restart a channel + * @ioat: IOAT DMA channel handle + */ +static void ioat1_reset_channel(struct ioat_dma_chan *ioat) +{ + struct ioat_chan_common *chan = &ioat->base; + void __iomem *reg_base = chan->reg_base; + u32 chansts, chanerr; + + dev_warn(to_dev(chan), "reset\n"); + chanerr = readl(reg_base + IOAT_CHANERR_OFFSET); + chansts = *chan->completion & IOAT_CHANSTS_STATUS; + if (chanerr) { + dev_err(to_dev(chan), + "chan%d, CHANSTS = 0x%08x CHANERR = 0x%04x, clearing\n", + chan_num(chan), chansts, chanerr); + writel(chanerr, reg_base + IOAT_CHANERR_OFFSET); + } + + /* + * whack it upside the head with a reset + * and wait for things to settle out. + * force the pending count to a really big negative + * to make sure no one forces an issue_pending + * while we're waiting. + */ + + ioat->pending = INT_MIN; + writeb(IOAT_CHANCMD_RESET, + reg_base + IOAT_CHANCMD_OFFSET(chan->device->version)); + set_bit(IOAT_RESET_PENDING, &chan->state); + mod_timer(&chan->timer, jiffies + RESET_DELAY); +} + +static dma_cookie_t ioat1_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct dma_chan *c = tx->chan; + struct ioat_dma_chan *ioat = to_ioat_chan(c); + struct ioat_desc_sw *desc = tx_to_ioat_desc(tx); + struct ioat_chan_common *chan = &ioat->base; + struct ioat_desc_sw *first; + struct ioat_desc_sw *chain_tail; + dma_cookie_t cookie; + + spin_lock_bh(&ioat->desc_lock); + /* cookie incr and addition to used_list must be atomic */ + cookie = c->cookie; + cookie++; + if (cookie < 0) + cookie = 1; + c->cookie = cookie; + tx->cookie = cookie; + dev_dbg(to_dev(&ioat->base), "%s: cookie: %d\n", __func__, cookie); + + /* write address into NextDescriptor field of last desc in chain */ + first = to_ioat_desc(desc->tx_list.next); + chain_tail = to_ioat_desc(ioat->used_desc.prev); + /* make descriptor updates globally visible before chaining */ + wmb(); + chain_tail->hw->next = first->txd.phys; + list_splice_tail_init(&desc->tx_list, &ioat->used_desc); + dump_desc_dbg(ioat, chain_tail); + dump_desc_dbg(ioat, first); + + if (!test_and_set_bit(IOAT_COMPLETION_PENDING, &chan->state)) + mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT); + + ioat->active += desc->hw->tx_cnt; + ioat->pending += desc->hw->tx_cnt; + if (ioat->pending >= ioat_pending_level) + __ioat1_dma_memcpy_issue_pending(ioat); + spin_unlock_bh(&ioat->desc_lock); + + return cookie; +} + +/** + * ioat_dma_alloc_descriptor - allocate and return a sw and hw descriptor pair + * @ioat: the channel supplying the memory pool for the descriptors + * @flags: allocation flags + */ +static struct ioat_desc_sw * +ioat_dma_alloc_descriptor(struct ioat_dma_chan *ioat, gfp_t flags) +{ + struct ioat_dma_descriptor *desc; + struct ioat_desc_sw *desc_sw; + struct ioatdma_device *ioatdma_device; + dma_addr_t phys; + + ioatdma_device = ioat->base.device; + desc = pci_pool_alloc(ioatdma_device->dma_pool, flags, &phys); + if (unlikely(!desc)) + return NULL; + + desc_sw = kzalloc(sizeof(*desc_sw), flags); + if (unlikely(!desc_sw)) { + pci_pool_free(ioatdma_device->dma_pool, desc, phys); + return NULL; + } + + memset(desc, 0, sizeof(*desc)); + + INIT_LIST_HEAD(&desc_sw->tx_list); + dma_async_tx_descriptor_init(&desc_sw->txd, &ioat->base.common); + desc_sw->txd.tx_submit = ioat1_tx_submit; + desc_sw->hw = desc; + desc_sw->txd.phys = phys; + set_desc_id(desc_sw, -1); + + return desc_sw; +} + +static int ioat_initial_desc_count = 256; +module_param(ioat_initial_desc_count, int, 0644); +MODULE_PARM_DESC(ioat_initial_desc_count, + "ioat1: initial descriptors per channel (default: 256)"); +/** + * ioat1_dma_alloc_chan_resources - returns the number of allocated descriptors + * @chan: the channel to be filled out + */ +static int ioat1_dma_alloc_chan_resources(struct dma_chan *c) +{ + struct ioat_dma_chan *ioat = to_ioat_chan(c); + struct ioat_chan_common *chan = &ioat->base; + struct ioat_desc_sw *desc; + u32 chanerr; + int i; + LIST_HEAD(tmp_list); + + /* have we already been set up? */ + if (!list_empty(&ioat->free_desc)) + return ioat->desccount; + + /* Setup register to interrupt and write completion status on error */ + writew(IOAT_CHANCTRL_RUN, chan->reg_base + IOAT_CHANCTRL_OFFSET); + + chanerr = readl(chan->reg_base + IOAT_CHANERR_OFFSET); + if (chanerr) { + dev_err(to_dev(chan), "CHANERR = %x, clearing\n", chanerr); + writel(chanerr, chan->reg_base + IOAT_CHANERR_OFFSET); + } + + /* Allocate descriptors */ + for (i = 0; i < ioat_initial_desc_count; i++) { + desc = ioat_dma_alloc_descriptor(ioat, GFP_KERNEL); + if (!desc) { + dev_err(to_dev(chan), "Only %d initial descriptors\n", i); + break; + } + set_desc_id(desc, i); + list_add_tail(&desc->node, &tmp_list); + } + spin_lock_bh(&ioat->desc_lock); + ioat->desccount = i; + list_splice(&tmp_list, &ioat->free_desc); + spin_unlock_bh(&ioat->desc_lock); + + /* allocate a completion writeback area */ + /* doing 2 32bit writes to mmio since 1 64b write doesn't work */ + chan->completion = pci_pool_alloc(chan->device->completion_pool, + GFP_KERNEL, &chan->completion_dma); + memset(chan->completion, 0, sizeof(*chan->completion)); + writel(((u64) chan->completion_dma) & 0x00000000FFFFFFFF, + chan->reg_base + IOAT_CHANCMP_OFFSET_LOW); + writel(((u64) chan->completion_dma) >> 32, + chan->reg_base + IOAT_CHANCMP_OFFSET_HIGH); + + tasklet_enable(&chan->cleanup_task); + ioat1_dma_start_null_desc(ioat); /* give chain to dma device */ + dev_dbg(to_dev(chan), "%s: allocated %d descriptors\n", + __func__, ioat->desccount); + return ioat->desccount; +} + +/** + * ioat1_dma_free_chan_resources - release all the descriptors + * @chan: the channel to be cleaned + */ +static void ioat1_dma_free_chan_resources(struct dma_chan *c) +{ + struct ioat_dma_chan *ioat = to_ioat_chan(c); + struct ioat_chan_common *chan = &ioat->base; + struct ioatdma_device *ioatdma_device = chan->device; + struct ioat_desc_sw *desc, *_desc; + int in_use_descs = 0; + + /* Before freeing channel resources first check + * if they have been previously allocated for this channel. + */ + if (ioat->desccount == 0) + return; + + tasklet_disable(&chan->cleanup_task); + del_timer_sync(&chan->timer); + ioat1_cleanup(ioat); + + /* Delay 100ms after reset to allow internal DMA logic to quiesce + * before removing DMA descriptor resources. + */ + writeb(IOAT_CHANCMD_RESET, + chan->reg_base + IOAT_CHANCMD_OFFSET(chan->device->version)); + mdelay(100); + + spin_lock_bh(&ioat->desc_lock); + list_for_each_entry_safe(desc, _desc, &ioat->used_desc, node) { + dev_dbg(to_dev(chan), "%s: freeing %d from used list\n", + __func__, desc_id(desc)); + dump_desc_dbg(ioat, desc); + in_use_descs++; + list_del(&desc->node); + pci_pool_free(ioatdma_device->dma_pool, desc->hw, + desc->txd.phys); + kfree(desc); + } + list_for_each_entry_safe(desc, _desc, + &ioat->free_desc, node) { + list_del(&desc->node); + pci_pool_free(ioatdma_device->dma_pool, desc->hw, + desc->txd.phys); + kfree(desc); + } + spin_unlock_bh(&ioat->desc_lock); + + pci_pool_free(ioatdma_device->completion_pool, + chan->completion, + chan->completion_dma); + + /* one is ok since we left it on there on purpose */ + if (in_use_descs > 1) + dev_err(to_dev(chan), "Freeing %d in use descriptors!\n", + in_use_descs - 1); + + chan->last_completion = 0; + chan->completion_dma = 0; + ioat->pending = 0; + ioat->desccount = 0; +} + +/** + * ioat1_dma_get_next_descriptor - return the next available descriptor + * @ioat: IOAT DMA channel handle + * + * Gets the next descriptor from the chain, and must be called with the + * channel's desc_lock held. Allocates more descriptors if the channel + * has run out. + */ +static struct ioat_desc_sw * +ioat1_dma_get_next_descriptor(struct ioat_dma_chan *ioat) +{ + struct ioat_desc_sw *new; + + if (!list_empty(&ioat->free_desc)) { + new = to_ioat_desc(ioat->free_desc.next); + list_del(&new->node); + } else { + /* try to get another desc */ + new = ioat_dma_alloc_descriptor(ioat, GFP_ATOMIC); + if (!new) { + dev_err(to_dev(&ioat->base), "alloc failed\n"); + return NULL; + } + } + dev_dbg(to_dev(&ioat->base), "%s: allocated: %d\n", + __func__, desc_id(new)); + prefetch(new->hw); + return new; +} + +static struct dma_async_tx_descriptor * +ioat1_dma_prep_memcpy(struct dma_chan *c, dma_addr_t dma_dest, + dma_addr_t dma_src, size_t len, unsigned long flags) +{ + struct ioat_dma_chan *ioat = to_ioat_chan(c); + struct ioat_desc_sw *desc; + size_t copy; + LIST_HEAD(chain); + dma_addr_t src = dma_src; + dma_addr_t dest = dma_dest; + size_t total_len = len; + struct ioat_dma_descriptor *hw = NULL; + int tx_cnt = 0; + + spin_lock_bh(&ioat->desc_lock); + desc = ioat1_dma_get_next_descriptor(ioat); + do { + if (!desc) + break; + + tx_cnt++; + copy = min_t(size_t, len, ioat->xfercap); + + hw = desc->hw; + hw->size = copy; + hw->ctl = 0; + hw->src_addr = src; + hw->dst_addr = dest; + + list_add_tail(&desc->node, &chain); + + len -= copy; + dest += copy; + src += copy; + if (len) { + struct ioat_desc_sw *next; + + async_tx_ack(&desc->txd); + next = ioat1_dma_get_next_descriptor(ioat); + hw->next = next ? next->txd.phys : 0; + dump_desc_dbg(ioat, desc); + desc = next; + } else + hw->next = 0; + } while (len); + + if (!desc) { + struct ioat_chan_common *chan = &ioat->base; + + dev_err(to_dev(chan), + "chan%d - get_next_desc failed\n", chan_num(chan)); + list_splice(&chain, &ioat->free_desc); + spin_unlock_bh(&ioat->desc_lock); + return NULL; + } + spin_unlock_bh(&ioat->desc_lock); + + desc->txd.flags = flags; + desc->len = total_len; + list_splice(&chain, &desc->tx_list); + hw->ctl_f.int_en = !!(flags & DMA_PREP_INTERRUPT); + hw->ctl_f.compl_write = 1; + hw->tx_cnt = tx_cnt; + dump_desc_dbg(ioat, desc); + + return &desc->txd; +} + +static void ioat1_cleanup_tasklet(unsigned long data) +{ + struct ioat_dma_chan *chan = (void *)data; + + ioat1_cleanup(chan); + writew(IOAT_CHANCTRL_RUN, chan->base.reg_base + IOAT_CHANCTRL_OFFSET); +} + +void ioat_dma_unmap(struct ioat_chan_common *chan, enum dma_ctrl_flags flags, + size_t len, struct ioat_dma_descriptor *hw) +{ + struct pci_dev *pdev = chan->device->pdev; + size_t offset = len - hw->size; + + if (!(flags & DMA_COMPL_SKIP_DEST_UNMAP)) + ioat_unmap(pdev, hw->dst_addr - offset, len, + PCI_DMA_FROMDEVICE, flags, 1); + + if (!(flags & DMA_COMPL_SKIP_SRC_UNMAP)) + ioat_unmap(pdev, hw->src_addr - offset, len, + PCI_DMA_TODEVICE, flags, 0); +} + +unsigned long ioat_get_current_completion(struct ioat_chan_common *chan) +{ + unsigned long phys_complete; + u64 completion; + + completion = *chan->completion; + phys_complete = ioat_chansts_to_addr(completion); + + dev_dbg(to_dev(chan), "%s: phys_complete: %#llx\n", __func__, + (unsigned long long) phys_complete); + + if (is_ioat_halted(completion)) { + u32 chanerr = readl(chan->reg_base + IOAT_CHANERR_OFFSET); + dev_err(to_dev(chan), "Channel halted, chanerr = %x\n", + chanerr); + + /* TODO do something to salvage the situation */ + } + + return phys_complete; +} + +bool ioat_cleanup_preamble(struct ioat_chan_common *chan, + unsigned long *phys_complete) +{ + *phys_complete = ioat_get_current_completion(chan); + if (*phys_complete == chan->last_completion) + return false; + clear_bit(IOAT_COMPLETION_ACK, &chan->state); + mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT); + + return true; +} + +static void __cleanup(struct ioat_dma_chan *ioat, unsigned long phys_complete) +{ + struct ioat_chan_common *chan = &ioat->base; + struct list_head *_desc, *n; + struct dma_async_tx_descriptor *tx; + + dev_dbg(to_dev(chan), "%s: phys_complete: %lx\n", + __func__, phys_complete); + list_for_each_safe(_desc, n, &ioat->used_desc) { + struct ioat_desc_sw *desc; + + prefetch(n); + desc = list_entry(_desc, typeof(*desc), node); + tx = &desc->txd; + /* + * Incoming DMA requests may use multiple descriptors, + * due to exceeding xfercap, perhaps. If so, only the + * last one will have a cookie, and require unmapping. + */ + dump_desc_dbg(ioat, desc); + if (tx->cookie) { + chan->completed_cookie = tx->cookie; + tx->cookie = 0; + ioat_dma_unmap(chan, tx->flags, desc->len, desc->hw); + ioat->active -= desc->hw->tx_cnt; + if (tx->callback) { + tx->callback(tx->callback_param); + tx->callback = NULL; + } + } + + if (tx->phys != phys_complete) { + /* + * a completed entry, but not the last, so clean + * up if the client is done with the descriptor + */ + if (async_tx_test_ack(tx)) + list_move_tail(&desc->node, &ioat->free_desc); + } else { + /* + * last used desc. Do not remove, so we can + * append from it. + */ + + /* if nothing else is pending, cancel the + * completion timeout + */ + if (n == &ioat->used_desc) { + dev_dbg(to_dev(chan), + "%s cancel completion timeout\n", + __func__); + clear_bit(IOAT_COMPLETION_PENDING, &chan->state); + } + + /* TODO check status bits? */ + break; + } + } + + chan->last_completion = phys_complete; +} + +/** + * ioat1_cleanup - cleanup up finished descriptors + * @chan: ioat channel to be cleaned up + * + * To prevent lock contention we defer cleanup when the locks are + * contended with a terminal timeout that forces cleanup and catches + * completion notification errors. + */ +static void ioat1_cleanup(struct ioat_dma_chan *ioat) +{ + struct ioat_chan_common *chan = &ioat->base; + unsigned long phys_complete; + + prefetch(chan->completion); + + if (!spin_trylock_bh(&chan->cleanup_lock)) + return; + + if (!ioat_cleanup_preamble(chan, &phys_complete)) { + spin_unlock_bh(&chan->cleanup_lock); + return; + } + + if (!spin_trylock_bh(&ioat->desc_lock)) { + spin_unlock_bh(&chan->cleanup_lock); + return; + } + + __cleanup(ioat, phys_complete); + + spin_unlock_bh(&ioat->desc_lock); + spin_unlock_bh(&chan->cleanup_lock); +} + +static void ioat1_timer_event(unsigned long data) +{ + struct ioat_dma_chan *ioat = (void *) data; + struct ioat_chan_common *chan = &ioat->base; + + dev_dbg(to_dev(chan), "%s: state: %lx\n", __func__, chan->state); + + spin_lock_bh(&chan->cleanup_lock); + if (test_and_clear_bit(IOAT_RESET_PENDING, &chan->state)) { + struct ioat_desc_sw *desc; + + spin_lock_bh(&ioat->desc_lock); + + /* restart active descriptors */ + desc = to_ioat_desc(ioat->used_desc.prev); + ioat_set_chainaddr(ioat, desc->txd.phys); + ioat_start(chan); + + ioat->pending = 0; + set_bit(IOAT_COMPLETION_PENDING, &chan->state); + mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT); + spin_unlock_bh(&ioat->desc_lock); + } else if (test_bit(IOAT_COMPLETION_PENDING, &chan->state)) { + unsigned long phys_complete; + + spin_lock_bh(&ioat->desc_lock); + /* if we haven't made progress and we have already + * acknowledged a pending completion once, then be more + * forceful with a restart + */ + if (ioat_cleanup_preamble(chan, &phys_complete)) + __cleanup(ioat, phys_complete); + else if (test_bit(IOAT_COMPLETION_ACK, &chan->state)) + ioat1_reset_channel(ioat); + else { + u64 status = ioat_chansts(chan); + + /* manually update the last completion address */ + if (ioat_chansts_to_addr(status) != 0) + *chan->completion = status; + + set_bit(IOAT_COMPLETION_ACK, &chan->state); + mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT); + } + spin_unlock_bh(&ioat->desc_lock); + } + spin_unlock_bh(&chan->cleanup_lock); +} + +static enum dma_status +ioat1_dma_is_complete(struct dma_chan *c, dma_cookie_t cookie, + dma_cookie_t *done, dma_cookie_t *used) +{ + struct ioat_dma_chan *ioat = to_ioat_chan(c); + + if (ioat_is_complete(c, cookie, done, used) == DMA_SUCCESS) + return DMA_SUCCESS; + + ioat1_cleanup(ioat); + + return ioat_is_complete(c, cookie, done, used); +} + +static void ioat1_dma_start_null_desc(struct ioat_dma_chan *ioat) +{ + struct ioat_chan_common *chan = &ioat->base; + struct ioat_desc_sw *desc; + struct ioat_dma_descriptor *hw; + + spin_lock_bh(&ioat->desc_lock); + + desc = ioat1_dma_get_next_descriptor(ioat); + + if (!desc) { + dev_err(to_dev(chan), + "Unable to start null desc - get next desc failed\n"); + spin_unlock_bh(&ioat->desc_lock); + return; + } + + hw = desc->hw; + hw->ctl = 0; + hw->ctl_f.null = 1; + hw->ctl_f.int_en = 1; + hw->ctl_f.compl_write = 1; + /* set size to non-zero value (channel returns error when size is 0) */ + hw->size = NULL_DESC_BUFFER_SIZE; + hw->src_addr = 0; + hw->dst_addr = 0; + async_tx_ack(&desc->txd); + hw->next = 0; + list_add_tail(&desc->node, &ioat->used_desc); + dump_desc_dbg(ioat, desc); + + ioat_set_chainaddr(ioat, desc->txd.phys); + ioat_start(chan); + spin_unlock_bh(&ioat->desc_lock); +} + +/* + * Perform a IOAT transaction to verify the HW works. + */ +#define IOAT_TEST_SIZE 2000 + +static void __devinit ioat_dma_test_callback(void *dma_async_param) +{ + struct completion *cmp = dma_async_param; + + complete(cmp); +} + +/** + * ioat_dma_self_test - Perform a IOAT transaction to verify the HW works. + * @device: device to be tested + */ +int __devinit ioat_dma_self_test(struct ioatdma_device *device) +{ + int i; + u8 *src; + u8 *dest; + struct dma_device *dma = &device->common; + struct device *dev = &device->pdev->dev; + struct dma_chan *dma_chan; + struct dma_async_tx_descriptor *tx; + dma_addr_t dma_dest, dma_src; + dma_cookie_t cookie; + int err = 0; + struct completion cmp; + unsigned long tmo; + unsigned long flags; + + src = kzalloc(sizeof(u8) * IOAT_TEST_SIZE, GFP_KERNEL); + if (!src) + return -ENOMEM; + dest = kzalloc(sizeof(u8) * IOAT_TEST_SIZE, GFP_KERNEL); + if (!dest) { + kfree(src); + return -ENOMEM; + } + + /* Fill in src buffer */ + for (i = 0; i < IOAT_TEST_SIZE; i++) + src[i] = (u8)i; + + /* Start copy, using first DMA channel */ + dma_chan = container_of(dma->channels.next, struct dma_chan, + device_node); + if (dma->device_alloc_chan_resources(dma_chan) < 1) { + dev_err(dev, "selftest cannot allocate chan resource\n"); + err = -ENODEV; + goto out; + } + + dma_src = dma_map_single(dev, src, IOAT_TEST_SIZE, DMA_TO_DEVICE); + dma_dest = dma_map_single(dev, dest, IOAT_TEST_SIZE, DMA_FROM_DEVICE); + flags = DMA_COMPL_SRC_UNMAP_SINGLE | DMA_COMPL_DEST_UNMAP_SINGLE | + DMA_PREP_INTERRUPT; + tx = device->common.device_prep_dma_memcpy(dma_chan, dma_dest, dma_src, + IOAT_TEST_SIZE, flags); + if (!tx) { + dev_err(dev, "Self-test prep failed, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + async_tx_ack(tx); + init_completion(&cmp); + tx->callback = ioat_dma_test_callback; + tx->callback_param = &cmp; + cookie = tx->tx_submit(tx); + if (cookie < 0) { + dev_err(dev, "Self-test setup failed, disabling\n"); + err = -ENODEV; + goto free_resources; + } + dma->device_issue_pending(dma_chan); + + tmo = wait_for_completion_timeout(&cmp, msecs_to_jiffies(3000)); + + if (tmo == 0 || + dma->device_is_tx_complete(dma_chan, cookie, NULL, NULL) + != DMA_SUCCESS) { + dev_err(dev, "Self-test copy timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + if (memcmp(src, dest, IOAT_TEST_SIZE)) { + dev_err(dev, "Self-test copy failed compare, disabling\n"); + err = -ENODEV; + goto free_resources; + } + +free_resources: + dma->device_free_chan_resources(dma_chan); +out: + kfree(src); + kfree(dest); + return err; +} + +static char ioat_interrupt_style[32] = "msix"; +module_param_string(ioat_interrupt_style, ioat_interrupt_style, + sizeof(ioat_interrupt_style), 0644); +MODULE_PARM_DESC(ioat_interrupt_style, + "set ioat interrupt style: msix (default), " + "msix-single-vector, msi, intx)"); + +/** + * ioat_dma_setup_interrupts - setup interrupt handler + * @device: ioat device + */ +static int ioat_dma_setup_interrupts(struct ioatdma_device *device) +{ + struct ioat_chan_common *chan; + struct pci_dev *pdev = device->pdev; + struct device *dev = &pdev->dev; + struct msix_entry *msix; + int i, j, msixcnt; + int err = -EINVAL; + u8 intrctrl = 0; + + if (!strcmp(ioat_interrupt_style, "msix")) + goto msix; + if (!strcmp(ioat_interrupt_style, "msix-single-vector")) + goto msix_single_vector; + if (!strcmp(ioat_interrupt_style, "msi")) + goto msi; + if (!strcmp(ioat_interrupt_style, "intx")) + goto intx; + dev_err(dev, "invalid ioat_interrupt_style %s\n", ioat_interrupt_style); + goto err_no_irq; + +msix: + /* The number of MSI-X vectors should equal the number of channels */ + msixcnt = device->common.chancnt; + for (i = 0; i < msixcnt; i++) + device->msix_entries[i].entry = i; + + err = pci_enable_msix(pdev, device->msix_entries, msixcnt); + if (err < 0) + goto msi; + if (err > 0) + goto msix_single_vector; + + for (i = 0; i < msixcnt; i++) { + msix = &device->msix_entries[i]; + chan = ioat_chan_by_index(device, i); + err = devm_request_irq(dev, msix->vector, + ioat_dma_do_interrupt_msix, 0, + "ioat-msix", chan); + if (err) { + for (j = 0; j < i; j++) { + msix = &device->msix_entries[j]; + chan = ioat_chan_by_index(device, j); + devm_free_irq(dev, msix->vector, chan); + } + goto msix_single_vector; + } + } + intrctrl |= IOAT_INTRCTRL_MSIX_VECTOR_CONTROL; + goto done; + +msix_single_vector: + msix = &device->msix_entries[0]; + msix->entry = 0; + err = pci_enable_msix(pdev, device->msix_entries, 1); + if (err) + goto msi; + + err = devm_request_irq(dev, msix->vector, ioat_dma_do_interrupt, 0, + "ioat-msix", device); + if (err) { + pci_disable_msix(pdev); + goto msi; + } + goto done; + +msi: + err = pci_enable_msi(pdev); + if (err) + goto intx; + + err = devm_request_irq(dev, pdev->irq, ioat_dma_do_interrupt, 0, + "ioat-msi", device); + if (err) { + pci_disable_msi(pdev); + goto intx; + } + goto done; + +intx: + err = devm_request_irq(dev, pdev->irq, ioat_dma_do_interrupt, + IRQF_SHARED, "ioat-intx", device); + if (err) + goto err_no_irq; + +done: + if (device->intr_quirk) + device->intr_quirk(device); + intrctrl |= IOAT_INTRCTRL_MASTER_INT_EN; + writeb(intrctrl, device->reg_base + IOAT_INTRCTRL_OFFSET); + return 0; + +err_no_irq: + /* Disable all interrupt generation */ + writeb(0, device->reg_base + IOAT_INTRCTRL_OFFSET); + dev_err(dev, "no usable interrupts\n"); + return err; +} + +static void ioat_disable_interrupts(struct ioatdma_device *device) +{ + /* Disable all interrupt generation */ + writeb(0, device->reg_base + IOAT_INTRCTRL_OFFSET); +} + +int __devinit ioat_probe(struct ioatdma_device *device) +{ + int err = -ENODEV; + struct dma_device *dma = &device->common; + struct pci_dev *pdev = device->pdev; + struct device *dev = &pdev->dev; + + /* DMA coherent memory pool for DMA descriptor allocations */ + device->dma_pool = pci_pool_create("dma_desc_pool", pdev, + sizeof(struct ioat_dma_descriptor), + 64, 0); + if (!device->dma_pool) { + err = -ENOMEM; + goto err_dma_pool; + } + + device->completion_pool = pci_pool_create("completion_pool", pdev, + sizeof(u64), SMP_CACHE_BYTES, + SMP_CACHE_BYTES); + + if (!device->completion_pool) { + err = -ENOMEM; + goto err_completion_pool; + } + + device->enumerate_channels(device); + + dma_cap_set(DMA_MEMCPY, dma->cap_mask); + dma->dev = &pdev->dev; + + if (!dma->chancnt) { + dev_err(dev, "channel enumeration error\n"); + goto err_setup_interrupts; + } + + err = ioat_dma_setup_interrupts(device); + if (err) + goto err_setup_interrupts; + + err = device->self_test(device); + if (err) + goto err_self_test; + + return 0; + +err_self_test: + ioat_disable_interrupts(device); +err_setup_interrupts: + pci_pool_destroy(device->completion_pool); +err_completion_pool: + pci_pool_destroy(device->dma_pool); +err_dma_pool: + return err; +} + +int __devinit ioat_register(struct ioatdma_device *device) +{ + int err = dma_async_device_register(&device->common); + + if (err) { + ioat_disable_interrupts(device); + pci_pool_destroy(device->completion_pool); + pci_pool_destroy(device->dma_pool); + } + + return err; +} + +/* ioat1_intr_quirk - fix up dma ctrl register to enable / disable msi */ +static void ioat1_intr_quirk(struct ioatdma_device *device) +{ + struct pci_dev *pdev = device->pdev; + u32 dmactrl; + + pci_read_config_dword(pdev, IOAT_PCI_DMACTRL_OFFSET, &dmactrl); + if (pdev->msi_enabled) + dmactrl |= IOAT_PCI_DMACTRL_MSI_EN; + else + dmactrl &= ~IOAT_PCI_DMACTRL_MSI_EN; + pci_write_config_dword(pdev, IOAT_PCI_DMACTRL_OFFSET, dmactrl); +} + +static ssize_t ring_size_show(struct dma_chan *c, char *page) +{ + struct ioat_dma_chan *ioat = to_ioat_chan(c); + + return sprintf(page, "%d\n", ioat->desccount); +} +static struct ioat_sysfs_entry ring_size_attr = __ATTR_RO(ring_size); + +static ssize_t ring_active_show(struct dma_chan *c, char *page) +{ + struct ioat_dma_chan *ioat = to_ioat_chan(c); + + return sprintf(page, "%d\n", ioat->active); +} +static struct ioat_sysfs_entry ring_active_attr = __ATTR_RO(ring_active); + +static ssize_t cap_show(struct dma_chan *c, char *page) +{ + struct dma_device *dma = c->device; + + return sprintf(page, "copy%s%s%s%s%s%s\n", + dma_has_cap(DMA_PQ, dma->cap_mask) ? " pq" : "", + dma_has_cap(DMA_PQ_VAL, dma->cap_mask) ? " pq_val" : "", + dma_has_cap(DMA_XOR, dma->cap_mask) ? " xor" : "", + dma_has_cap(DMA_XOR_VAL, dma->cap_mask) ? " xor_val" : "", + dma_has_cap(DMA_MEMSET, dma->cap_mask) ? " fill" : "", + dma_has_cap(DMA_INTERRUPT, dma->cap_mask) ? " intr" : ""); + +} +struct ioat_sysfs_entry ioat_cap_attr = __ATTR_RO(cap); + +static ssize_t version_show(struct dma_chan *c, char *page) +{ + struct dma_device *dma = c->device; + struct ioatdma_device *device = to_ioatdma_device(dma); + + return sprintf(page, "%d.%d\n", + device->version >> 4, device->version & 0xf); +} +struct ioat_sysfs_entry ioat_version_attr = __ATTR_RO(version); + +static struct attribute *ioat1_attrs[] = { + &ring_size_attr.attr, + &ring_active_attr.attr, + &ioat_cap_attr.attr, + &ioat_version_attr.attr, + NULL, +}; + +static ssize_t +ioat_attr_show(struct kobject *kobj, struct attribute *attr, char *page) +{ + struct ioat_sysfs_entry *entry; + struct ioat_chan_common *chan; + + entry = container_of(attr, struct ioat_sysfs_entry, attr); + chan = container_of(kobj, struct ioat_chan_common, kobj); + + if (!entry->show) + return -EIO; + return entry->show(&chan->common, page); +} + +struct sysfs_ops ioat_sysfs_ops = { + .show = ioat_attr_show, +}; + +static struct kobj_type ioat1_ktype = { + .sysfs_ops = &ioat_sysfs_ops, + .default_attrs = ioat1_attrs, +}; + +void ioat_kobject_add(struct ioatdma_device *device, struct kobj_type *type) +{ + struct dma_device *dma = &device->common; + struct dma_chan *c; + + list_for_each_entry(c, &dma->channels, device_node) { + struct ioat_chan_common *chan = to_chan_common(c); + struct kobject *parent = &c->dev->device.kobj; + int err; + + err = kobject_init_and_add(&chan->kobj, type, parent, "quickdata"); + if (err) { + dev_warn(to_dev(chan), + "sysfs init error (%d), continuing...\n", err); + kobject_put(&chan->kobj); + set_bit(IOAT_KOBJ_INIT_FAIL, &chan->state); + } + } +} + +void ioat_kobject_del(struct ioatdma_device *device) +{ + struct dma_device *dma = &device->common; + struct dma_chan *c; + + list_for_each_entry(c, &dma->channels, device_node) { + struct ioat_chan_common *chan = to_chan_common(c); + + if (!test_bit(IOAT_KOBJ_INIT_FAIL, &chan->state)) { + kobject_del(&chan->kobj); + kobject_put(&chan->kobj); + } + } +} + +int __devinit ioat1_dma_probe(struct ioatdma_device *device, int dca) +{ + struct pci_dev *pdev = device->pdev; + struct dma_device *dma; + int err; + + device->intr_quirk = ioat1_intr_quirk; + device->enumerate_channels = ioat1_enumerate_channels; + device->self_test = ioat_dma_self_test; + dma = &device->common; + dma->device_prep_dma_memcpy = ioat1_dma_prep_memcpy; + dma->device_issue_pending = ioat1_dma_memcpy_issue_pending; + dma->device_alloc_chan_resources = ioat1_dma_alloc_chan_resources; + dma->device_free_chan_resources = ioat1_dma_free_chan_resources; + dma->device_is_tx_complete = ioat1_dma_is_complete; + + err = ioat_probe(device); + if (err) + return err; + ioat_set_tcp_copy_break(4096); + err = ioat_register(device); + if (err) + return err; + ioat_kobject_add(device, &ioat1_ktype); + + if (dca) + device->dca = ioat_dca_init(pdev, device->reg_base); + + return err; +} + +void __devexit ioat_dma_remove(struct ioatdma_device *device) +{ + struct dma_device *dma = &device->common; + + ioat_disable_interrupts(device); + + ioat_kobject_del(device); + + dma_async_device_unregister(dma); + + pci_pool_destroy(device->dma_pool); + pci_pool_destroy(device->completion_pool); + + INIT_LIST_HEAD(&dma->channels); +} diff --git a/drivers/dma/ioat/dma.h b/drivers/dma/ioat/dma.h new file mode 100644 index 00000000000..bbc3e78ef33 --- /dev/null +++ b/drivers/dma/ioat/dma.h @@ -0,0 +1,353 @@ +/* + * Copyright(c) 2004 - 2009 Intel Corporation. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the Free + * Software Foundation; either version 2 of the License, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 + * Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * The full GNU General Public License is included in this distribution in the + * file called COPYING. + */ +#ifndef IOATDMA_H +#define IOATDMA_H + +#include <linux/dmaengine.h> +#include "hw.h" +#include "registers.h" +#include <linux/init.h> +#include <linux/dmapool.h> +#include <linux/cache.h> +#include <linux/pci_ids.h> +#include <net/tcp.h> + +#define IOAT_DMA_VERSION "4.00" + +#define IOAT_LOW_COMPLETION_MASK 0xffffffc0 +#define IOAT_DMA_DCA_ANY_CPU ~0 + +#define to_ioatdma_device(dev) container_of(dev, struct ioatdma_device, common) +#define to_ioat_desc(lh) container_of(lh, struct ioat_desc_sw, node) +#define tx_to_ioat_desc(tx) container_of(tx, struct ioat_desc_sw, txd) +#define to_dev(ioat_chan) (&(ioat_chan)->device->pdev->dev) + +#define chan_num(ch) ((int)((ch)->reg_base - (ch)->device->reg_base) / 0x80) + +/* + * workaround for IOAT ver.3.0 null descriptor issue + * (channel returns error when size is 0) + */ +#define NULL_DESC_BUFFER_SIZE 1 + +/** + * struct ioatdma_device - internal representation of a IOAT device + * @pdev: PCI-Express device + * @reg_base: MMIO register space base address + * @dma_pool: for allocating DMA descriptors + * @common: embedded struct dma_device + * @version: version of ioatdma device + * @msix_entries: irq handlers + * @idx: per channel data + * @dca: direct cache access context + * @intr_quirk: interrupt setup quirk (for ioat_v1 devices) + * @enumerate_channels: hw version specific channel enumeration + * @reset_hw: hw version specific channel (re)initialization + * @cleanup_tasklet: select between the v2 and v3 cleanup routines + * @timer_fn: select between the v2 and v3 timer watchdog routines + * @self_test: hardware version specific self test for each supported op type + * + * Note: the v3 cleanup routine supports raid operations + */ +struct ioatdma_device { + struct pci_dev *pdev; + void __iomem *reg_base; + struct pci_pool *dma_pool; + struct pci_pool *completion_pool; + struct dma_device common; + u8 version; + struct msix_entry msix_entries[4]; + struct ioat_chan_common *idx[4]; + struct dca_provider *dca; + void (*intr_quirk)(struct ioatdma_device *device); + int (*enumerate_channels)(struct ioatdma_device *device); + int (*reset_hw)(struct ioat_chan_common *chan); + void (*cleanup_tasklet)(unsigned long data); + void (*timer_fn)(unsigned long data); + int (*self_test)(struct ioatdma_device *device); +}; + +struct ioat_chan_common { + struct dma_chan common; + void __iomem *reg_base; + unsigned long last_completion; + spinlock_t cleanup_lock; + dma_cookie_t completed_cookie; + unsigned long state; + #define IOAT_COMPLETION_PENDING 0 + #define IOAT_COMPLETION_ACK 1 + #define IOAT_RESET_PENDING 2 + #define IOAT_KOBJ_INIT_FAIL 3 + struct timer_list timer; + #define COMPLETION_TIMEOUT msecs_to_jiffies(100) + #define IDLE_TIMEOUT msecs_to_jiffies(2000) + #define RESET_DELAY msecs_to_jiffies(100) + struct ioatdma_device *device; + dma_addr_t completion_dma; + u64 *completion; + struct tasklet_struct cleanup_task; + struct kobject kobj; +}; + +struct ioat_sysfs_entry { + struct attribute attr; + ssize_t (*show)(struct dma_chan *, char *); +}; + +/** + * struct ioat_dma_chan - internal representation of a DMA channel + */ +struct ioat_dma_chan { + struct ioat_chan_common base; + + size_t xfercap; /* XFERCAP register value expanded out */ + + spinlock_t desc_lock; + struct list_head free_desc; + struct list_head used_desc; + + int pending; + u16 desccount; + u16 active; +}; + +static inline struct ioat_chan_common *to_chan_common(struct dma_chan *c) +{ + return container_of(c, struct ioat_chan_common, common); +} + +static inline struct ioat_dma_chan *to_ioat_chan(struct dma_chan *c) +{ + struct ioat_chan_common *chan = to_chan_common(c); + + return container_of(chan, struct ioat_dma_chan, base); +} + +/** + * ioat_is_complete - poll the status of an ioat transaction + * @c: channel handle + * @cookie: transaction identifier + * @done: if set, updated with last completed transaction + * @used: if set, updated with last used transaction + */ +static inline enum dma_status +ioat_is_complete(struct dma_chan *c, dma_cookie_t cookie, + dma_cookie_t *done, dma_cookie_t *used) +{ + struct ioat_chan_common *chan = to_chan_common(c); + dma_cookie_t last_used; + dma_cookie_t last_complete; + + last_used = c->cookie; + last_complete = chan->completed_cookie; + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + return dma_async_is_complete(cookie, last_complete, last_used); +} + +/* wrapper around hardware descriptor format + additional software fields */ + +/** + * struct ioat_desc_sw - wrapper around hardware descriptor + * @hw: hardware DMA descriptor (for memcpy) + * @node: this descriptor will either be on the free list, + * or attached to a transaction list (tx_list) + * @txd: the generic software descriptor for all engines + * @id: identifier for debug + */ +struct ioat_desc_sw { + struct ioat_dma_descriptor *hw; + struct list_head node; + size_t len; + struct list_head tx_list; + struct dma_async_tx_descriptor txd; + #ifdef DEBUG + int id; + #endif +}; + +#ifdef DEBUG +#define set_desc_id(desc, i) ((desc)->id = (i)) +#define desc_id(desc) ((desc)->id) +#else +#define set_desc_id(desc, i) +#define desc_id(desc) (0) +#endif + +static inline void +__dump_desc_dbg(struct ioat_chan_common *chan, struct ioat_dma_descriptor *hw, + struct dma_async_tx_descriptor *tx, int id) +{ + struct device *dev = to_dev(chan); + + dev_dbg(dev, "desc[%d]: (%#llx->%#llx) cookie: %d flags: %#x" + " ctl: %#x (op: %d int_en: %d compl: %d)\n", id, + (unsigned long long) tx->phys, + (unsigned long long) hw->next, tx->cookie, tx->flags, + hw->ctl, hw->ctl_f.op, hw->ctl_f.int_en, hw->ctl_f.compl_write); +} + +#define dump_desc_dbg(c, d) \ + ({ if (d) __dump_desc_dbg(&c->base, d->hw, &d->txd, desc_id(d)); 0; }) + +static inline void ioat_set_tcp_copy_break(unsigned long copybreak) +{ + #ifdef CONFIG_NET_DMA + sysctl_tcp_dma_copybreak = copybreak; + #endif +} + +static inline struct ioat_chan_common * +ioat_chan_by_index(struct ioatdma_device *device, int index) +{ + return device->idx[index]; +} + +static inline u64 ioat_chansts(struct ioat_chan_common *chan) +{ + u8 ver = chan->device->version; + u64 status; + u32 status_lo; + + /* We need to read the low address first as this causes the + * chipset to latch the upper bits for the subsequent read + */ + status_lo = readl(chan->reg_base + IOAT_CHANSTS_OFFSET_LOW(ver)); + status = readl(chan->reg_base + IOAT_CHANSTS_OFFSET_HIGH(ver)); + status <<= 32; + status |= status_lo; + + return status; +} + +static inline void ioat_start(struct ioat_chan_common *chan) +{ + u8 ver = chan->device->version; + + writeb(IOAT_CHANCMD_START, chan->reg_base + IOAT_CHANCMD_OFFSET(ver)); +} + +static inline u64 ioat_chansts_to_addr(u64 status) +{ + return status & IOAT_CHANSTS_COMPLETED_DESCRIPTOR_ADDR; +} + +static inline u32 ioat_chanerr(struct ioat_chan_common *chan) +{ + return readl(chan->reg_base + IOAT_CHANERR_OFFSET); +} + +static inline void ioat_suspend(struct ioat_chan_common *chan) +{ + u8 ver = chan->device->version; + + writeb(IOAT_CHANCMD_SUSPEND, chan->reg_base + IOAT_CHANCMD_OFFSET(ver)); +} + +static inline void ioat_reset(struct ioat_chan_common *chan) +{ + u8 ver = chan->device->version; + + writeb(IOAT_CHANCMD_RESET, chan->reg_base + IOAT_CHANCMD_OFFSET(ver)); +} + +static inline bool ioat_reset_pending(struct ioat_chan_common *chan) +{ + u8 ver = chan->device->version; + u8 cmd; + + cmd = readb(chan->reg_base + IOAT_CHANCMD_OFFSET(ver)); + return (cmd & IOAT_CHANCMD_RESET) == IOAT_CHANCMD_RESET; +} + +static inline void ioat_set_chainaddr(struct ioat_dma_chan *ioat, u64 addr) +{ + struct ioat_chan_common *chan = &ioat->base; + + writel(addr & 0x00000000FFFFFFFF, + chan->reg_base + IOAT1_CHAINADDR_OFFSET_LOW); + writel(addr >> 32, + chan->reg_base + IOAT1_CHAINADDR_OFFSET_HIGH); +} + +static inline bool is_ioat_active(unsigned long status) +{ + return ((status & IOAT_CHANSTS_STATUS) == IOAT_CHANSTS_ACTIVE); +} + +static inline bool is_ioat_idle(unsigned long status) +{ + return ((status & IOAT_CHANSTS_STATUS) == IOAT_CHANSTS_DONE); +} + +static inline bool is_ioat_halted(unsigned long status) +{ + return ((status & IOAT_CHANSTS_STATUS) == IOAT_CHANSTS_HALTED); +} + +static inline bool is_ioat_suspended(unsigned long status) +{ + return ((status & IOAT_CHANSTS_STATUS) == IOAT_CHANSTS_SUSPENDED); +} + +/* channel was fatally programmed */ +static inline bool is_ioat_bug(unsigned long err) +{ + return !!err; +} + +static inline void ioat_unmap(struct pci_dev *pdev, dma_addr_t addr, size_t len, + int direction, enum dma_ctrl_flags flags, bool dst) +{ + if ((dst && (flags & DMA_COMPL_DEST_UNMAP_SINGLE)) || + (!dst && (flags & DMA_COMPL_SRC_UNMAP_SINGLE))) + pci_unmap_single(pdev, addr, len, direction); + else + pci_unmap_page(pdev, addr, len, direction); +} + +int __devinit ioat_probe(struct ioatdma_device *device); +int __devinit ioat_register(struct ioatdma_device *device); +int __devinit ioat1_dma_probe(struct ioatdma_device *dev, int dca); +int __devinit ioat_dma_self_test(struct ioatdma_device *device); +void __devexit ioat_dma_remove(struct ioatdma_device *device); +struct dca_provider * __devinit ioat_dca_init(struct pci_dev *pdev, + void __iomem *iobase); +unsigned long ioat_get_current_completion(struct ioat_chan_common *chan); +void ioat_init_channel(struct ioatdma_device *device, + struct ioat_chan_common *chan, int idx, + void (*timer_fn)(unsigned long), + void (*tasklet)(unsigned long), + unsigned long ioat); +void ioat_dma_unmap(struct ioat_chan_common *chan, enum dma_ctrl_flags flags, + size_t len, struct ioat_dma_descriptor *hw); +bool ioat_cleanup_preamble(struct ioat_chan_common *chan, + unsigned long *phys_complete); +void ioat_kobject_add(struct ioatdma_device *device, struct kobj_type *type); +void ioat_kobject_del(struct ioatdma_device *device); +extern struct sysfs_ops ioat_sysfs_ops; +extern struct ioat_sysfs_entry ioat_version_attr; +extern struct ioat_sysfs_entry ioat_cap_attr; +#endif /* IOATDMA_H */ diff --git a/drivers/dma/ioat/dma_v2.c b/drivers/dma/ioat/dma_v2.c new file mode 100644 index 00000000000..5cc37afe2bc --- /dev/null +++ b/drivers/dma/ioat/dma_v2.c @@ -0,0 +1,908 @@ +/* + * Intel I/OAT DMA Linux driver + * Copyright(c) 2004 - 2009 Intel Corporation. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + * + * The full GNU General Public License is included in this distribution in + * the file called "COPYING". + * + */ + +/* + * This driver supports an Intel I/OAT DMA engine (versions >= 2), which + * does asynchronous data movement and checksumming operations. + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/interrupt.h> +#include <linux/dmaengine.h> +#include <linux/delay.h> +#include <linux/dma-mapping.h> +#include <linux/workqueue.h> +#include <linux/i7300_idle.h> +#include "dma.h" +#include "dma_v2.h" +#include "registers.h" +#include "hw.h" + +int ioat_ring_alloc_order = 8; +module_param(ioat_ring_alloc_order, int, 0644); +MODULE_PARM_DESC(ioat_ring_alloc_order, + "ioat2+: allocate 2^n descriptors per channel" + " (default: 8 max: 16)"); +static int ioat_ring_max_alloc_order = IOAT_MAX_ORDER; +module_param(ioat_ring_max_alloc_order, int, 0644); +MODULE_PARM_DESC(ioat_ring_max_alloc_order, + "ioat2+: upper limit for ring size (default: 16)"); + +void __ioat2_issue_pending(struct ioat2_dma_chan *ioat) +{ + void * __iomem reg_base = ioat->base.reg_base; + + ioat->pending = 0; + ioat->dmacount += ioat2_ring_pending(ioat); + ioat->issued = ioat->head; + /* make descriptor updates globally visible before notifying channel */ + wmb(); + writew(ioat->dmacount, reg_base + IOAT_CHAN_DMACOUNT_OFFSET); + dev_dbg(to_dev(&ioat->base), + "%s: head: %#x tail: %#x issued: %#x count: %#x\n", + __func__, ioat->head, ioat->tail, ioat->issued, ioat->dmacount); +} + +void ioat2_issue_pending(struct dma_chan *chan) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(chan); + + spin_lock_bh(&ioat->ring_lock); + if (ioat->pending == 1) + __ioat2_issue_pending(ioat); + spin_unlock_bh(&ioat->ring_lock); +} + +/** + * ioat2_update_pending - log pending descriptors + * @ioat: ioat2+ channel + * + * set pending to '1' unless pending is already set to '2', pending == 2 + * indicates that submission is temporarily blocked due to an in-flight + * reset. If we are already above the ioat_pending_level threshold then + * just issue pending. + * + * called with ring_lock held + */ +static void ioat2_update_pending(struct ioat2_dma_chan *ioat) +{ + if (unlikely(ioat->pending == 2)) + return; + else if (ioat2_ring_pending(ioat) > ioat_pending_level) + __ioat2_issue_pending(ioat); + else + ioat->pending = 1; +} + +static void __ioat2_start_null_desc(struct ioat2_dma_chan *ioat) +{ + struct ioat_ring_ent *desc; + struct ioat_dma_descriptor *hw; + int idx; + + if (ioat2_ring_space(ioat) < 1) { + dev_err(to_dev(&ioat->base), + "Unable to start null desc - ring full\n"); + return; + } + + dev_dbg(to_dev(&ioat->base), "%s: head: %#x tail: %#x issued: %#x\n", + __func__, ioat->head, ioat->tail, ioat->issued); + idx = ioat2_desc_alloc(ioat, 1); + desc = ioat2_get_ring_ent(ioat, idx); + + hw = desc->hw; + hw->ctl = 0; + hw->ctl_f.null = 1; + hw->ctl_f.int_en = 1; + hw->ctl_f.compl_write = 1; + /* set size to non-zero value (channel returns error when size is 0) */ + hw->size = NULL_DESC_BUFFER_SIZE; + hw->src_addr = 0; + hw->dst_addr = 0; + async_tx_ack(&desc->txd); + ioat2_set_chainaddr(ioat, desc->txd.phys); + dump_desc_dbg(ioat, desc); + __ioat2_issue_pending(ioat); +} + +static void ioat2_start_null_desc(struct ioat2_dma_chan *ioat) +{ + spin_lock_bh(&ioat->ring_lock); + __ioat2_start_null_desc(ioat); + spin_unlock_bh(&ioat->ring_lock); +} + +static void __cleanup(struct ioat2_dma_chan *ioat, unsigned long phys_complete) +{ + struct ioat_chan_common *chan = &ioat->base; + struct dma_async_tx_descriptor *tx; + struct ioat_ring_ent *desc; + bool seen_current = false; + u16 active; + int i; + + dev_dbg(to_dev(chan), "%s: head: %#x tail: %#x issued: %#x\n", + __func__, ioat->head, ioat->tail, ioat->issued); + + active = ioat2_ring_active(ioat); + for (i = 0; i < active && !seen_current; i++) { + prefetch(ioat2_get_ring_ent(ioat, ioat->tail + i + 1)); + desc = ioat2_get_ring_ent(ioat, ioat->tail + i); + tx = &desc->txd; + dump_desc_dbg(ioat, desc); + if (tx->cookie) { + ioat_dma_unmap(chan, tx->flags, desc->len, desc->hw); + chan->completed_cookie = tx->cookie; + tx->cookie = 0; + if (tx->callback) { + tx->callback(tx->callback_param); + tx->callback = NULL; + } + } + + if (tx->phys == phys_complete) + seen_current = true; + } + ioat->tail += i; + BUG_ON(!seen_current); /* no active descs have written a completion? */ + + chan->last_completion = phys_complete; + if (ioat->head == ioat->tail) { + dev_dbg(to_dev(chan), "%s: cancel completion timeout\n", + __func__); + clear_bit(IOAT_COMPLETION_PENDING, &chan->state); + mod_timer(&chan->timer, jiffies + IDLE_TIMEOUT); + } +} + +/** + * ioat2_cleanup - clean finished descriptors (advance tail pointer) + * @chan: ioat channel to be cleaned up + */ +static void ioat2_cleanup(struct ioat2_dma_chan *ioat) +{ + struct ioat_chan_common *chan = &ioat->base; + unsigned long phys_complete; + + prefetch(chan->completion); + + if (!spin_trylock_bh(&chan->cleanup_lock)) + return; + + if (!ioat_cleanup_preamble(chan, &phys_complete)) { + spin_unlock_bh(&chan->cleanup_lock); + return; + } + + if (!spin_trylock_bh(&ioat->ring_lock)) { + spin_unlock_bh(&chan->cleanup_lock); + return; + } + + __cleanup(ioat, phys_complete); + + spin_unlock_bh(&ioat->ring_lock); + spin_unlock_bh(&chan->cleanup_lock); +} + +void ioat2_cleanup_tasklet(unsigned long data) +{ + struct ioat2_dma_chan *ioat = (void *) data; + + ioat2_cleanup(ioat); + writew(IOAT_CHANCTRL_RUN, ioat->base.reg_base + IOAT_CHANCTRL_OFFSET); +} + +void __ioat2_restart_chan(struct ioat2_dma_chan *ioat) +{ + struct ioat_chan_common *chan = &ioat->base; + + /* set the tail to be re-issued */ + ioat->issued = ioat->tail; + ioat->dmacount = 0; + set_bit(IOAT_COMPLETION_PENDING, &chan->state); + mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT); + + dev_dbg(to_dev(chan), + "%s: head: %#x tail: %#x issued: %#x count: %#x\n", + __func__, ioat->head, ioat->tail, ioat->issued, ioat->dmacount); + + if (ioat2_ring_pending(ioat)) { + struct ioat_ring_ent *desc; + + desc = ioat2_get_ring_ent(ioat, ioat->tail); + ioat2_set_chainaddr(ioat, desc->txd.phys); + __ioat2_issue_pending(ioat); + } else + __ioat2_start_null_desc(ioat); +} + +int ioat2_quiesce(struct ioat_chan_common *chan, unsigned long tmo) +{ + unsigned long end = jiffies + tmo; + int err = 0; + u32 status; + + status = ioat_chansts(chan); + if (is_ioat_active(status) || is_ioat_idle(status)) + ioat_suspend(chan); + while (is_ioat_active(status) || is_ioat_idle(status)) { + if (tmo && time_after(jiffies, end)) { + err = -ETIMEDOUT; + break; + } + status = ioat_chansts(chan); + cpu_relax(); + } + + return err; +} + +int ioat2_reset_sync(struct ioat_chan_common *chan, unsigned long tmo) +{ + unsigned long end = jiffies + tmo; + int err = 0; + + ioat_reset(chan); + while (ioat_reset_pending(chan)) { + if (end && time_after(jiffies, end)) { + err = -ETIMEDOUT; + break; + } + cpu_relax(); + } + + return err; +} + +static void ioat2_restart_channel(struct ioat2_dma_chan *ioat) +{ + struct ioat_chan_common *chan = &ioat->base; + unsigned long phys_complete; + + ioat2_quiesce(chan, 0); + if (ioat_cleanup_preamble(chan, &phys_complete)) + __cleanup(ioat, phys_complete); + + __ioat2_restart_chan(ioat); +} + +void ioat2_timer_event(unsigned long data) +{ + struct ioat2_dma_chan *ioat = (void *) data; + struct ioat_chan_common *chan = &ioat->base; + + spin_lock_bh(&chan->cleanup_lock); + if (test_bit(IOAT_COMPLETION_PENDING, &chan->state)) { + unsigned long phys_complete; + u64 status; + + spin_lock_bh(&ioat->ring_lock); + status = ioat_chansts(chan); + + /* when halted due to errors check for channel + * programming errors before advancing the completion state + */ + if (is_ioat_halted(status)) { + u32 chanerr; + + chanerr = readl(chan->reg_base + IOAT_CHANERR_OFFSET); + dev_err(to_dev(chan), "%s: Channel halted (%x)\n", + __func__, chanerr); + BUG_ON(is_ioat_bug(chanerr)); + } + + /* if we haven't made progress and we have already + * acknowledged a pending completion once, then be more + * forceful with a restart + */ + if (ioat_cleanup_preamble(chan, &phys_complete)) + __cleanup(ioat, phys_complete); + else if (test_bit(IOAT_COMPLETION_ACK, &chan->state)) + ioat2_restart_channel(ioat); + else { + set_bit(IOAT_COMPLETION_ACK, &chan->state); + mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT); + } + spin_unlock_bh(&ioat->ring_lock); + } else { + u16 active; + + /* if the ring is idle, empty, and oversized try to step + * down the size + */ + spin_lock_bh(&ioat->ring_lock); + active = ioat2_ring_active(ioat); + if (active == 0 && ioat->alloc_order > ioat_get_alloc_order()) + reshape_ring(ioat, ioat->alloc_order-1); + spin_unlock_bh(&ioat->ring_lock); + + /* keep shrinking until we get back to our minimum + * default size + */ + if (ioat->alloc_order > ioat_get_alloc_order()) + mod_timer(&chan->timer, jiffies + IDLE_TIMEOUT); + } + spin_unlock_bh(&chan->cleanup_lock); +} + +static int ioat2_reset_hw(struct ioat_chan_common *chan) +{ + /* throw away whatever the channel was doing and get it initialized */ + u32 chanerr; + + ioat2_quiesce(chan, msecs_to_jiffies(100)); + + chanerr = readl(chan->reg_base + IOAT_CHANERR_OFFSET); + writel(chanerr, chan->reg_base + IOAT_CHANERR_OFFSET); + + return ioat2_reset_sync(chan, msecs_to_jiffies(200)); +} + +/** + * ioat2_enumerate_channels - find and initialize the device's channels + * @device: the device to be enumerated + */ +int ioat2_enumerate_channels(struct ioatdma_device *device) +{ + struct ioat2_dma_chan *ioat; + struct device *dev = &device->pdev->dev; + struct dma_device *dma = &device->common; + u8 xfercap_log; + int i; + + INIT_LIST_HEAD(&dma->channels); + dma->chancnt = readb(device->reg_base + IOAT_CHANCNT_OFFSET); + dma->chancnt &= 0x1f; /* bits [4:0] valid */ + if (dma->chancnt > ARRAY_SIZE(device->idx)) { + dev_warn(dev, "(%d) exceeds max supported channels (%zu)\n", + dma->chancnt, ARRAY_SIZE(device->idx)); + dma->chancnt = ARRAY_SIZE(device->idx); + } + xfercap_log = readb(device->reg_base + IOAT_XFERCAP_OFFSET); + xfercap_log &= 0x1f; /* bits [4:0] valid */ + if (xfercap_log == 0) + return 0; + dev_dbg(dev, "%s: xfercap = %d\n", __func__, 1 << xfercap_log); + + /* FIXME which i/oat version is i7300? */ +#ifdef CONFIG_I7300_IDLE_IOAT_CHANNEL + if (i7300_idle_platform_probe(NULL, NULL, 1) == 0) + dma->chancnt--; +#endif + for (i = 0; i < dma->chancnt; i++) { + ioat = devm_kzalloc(dev, sizeof(*ioat), GFP_KERNEL); + if (!ioat) + break; + + ioat_init_channel(device, &ioat->base, i, + device->timer_fn, + device->cleanup_tasklet, + (unsigned long) ioat); + ioat->xfercap_log = xfercap_log; + spin_lock_init(&ioat->ring_lock); + if (device->reset_hw(&ioat->base)) { + i = 0; + break; + } + } + dma->chancnt = i; + return i; +} + +static dma_cookie_t ioat2_tx_submit_unlock(struct dma_async_tx_descriptor *tx) +{ + struct dma_chan *c = tx->chan; + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + struct ioat_chan_common *chan = &ioat->base; + dma_cookie_t cookie = c->cookie; + + cookie++; + if (cookie < 0) + cookie = 1; + tx->cookie = cookie; + c->cookie = cookie; + dev_dbg(to_dev(&ioat->base), "%s: cookie: %d\n", __func__, cookie); + + if (!test_and_set_bit(IOAT_COMPLETION_PENDING, &chan->state)) + mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT); + ioat2_update_pending(ioat); + spin_unlock_bh(&ioat->ring_lock); + + return cookie; +} + +static struct ioat_ring_ent *ioat2_alloc_ring_ent(struct dma_chan *chan, gfp_t flags) +{ + struct ioat_dma_descriptor *hw; + struct ioat_ring_ent *desc; + struct ioatdma_device *dma; + dma_addr_t phys; + + dma = to_ioatdma_device(chan->device); + hw = pci_pool_alloc(dma->dma_pool, flags, &phys); + if (!hw) + return NULL; + memset(hw, 0, sizeof(*hw)); + + desc = kmem_cache_alloc(ioat2_cache, flags); + if (!desc) { + pci_pool_free(dma->dma_pool, hw, phys); + return NULL; + } + memset(desc, 0, sizeof(*desc)); + + dma_async_tx_descriptor_init(&desc->txd, chan); + desc->txd.tx_submit = ioat2_tx_submit_unlock; + desc->hw = hw; + desc->txd.phys = phys; + return desc; +} + +static void ioat2_free_ring_ent(struct ioat_ring_ent *desc, struct dma_chan *chan) +{ + struct ioatdma_device *dma; + + dma = to_ioatdma_device(chan->device); + pci_pool_free(dma->dma_pool, desc->hw, desc->txd.phys); + kmem_cache_free(ioat2_cache, desc); +} + +static struct ioat_ring_ent **ioat2_alloc_ring(struct dma_chan *c, int order, gfp_t flags) +{ + struct ioat_ring_ent **ring; + int descs = 1 << order; + int i; + + if (order > ioat_get_max_alloc_order()) + return NULL; + + /* allocate the array to hold the software ring */ + ring = kcalloc(descs, sizeof(*ring), flags); + if (!ring) + return NULL; + for (i = 0; i < descs; i++) { + ring[i] = ioat2_alloc_ring_ent(c, flags); + if (!ring[i]) { + while (i--) + ioat2_free_ring_ent(ring[i], c); + kfree(ring); + return NULL; + } + set_desc_id(ring[i], i); + } + + /* link descs */ + for (i = 0; i < descs-1; i++) { + struct ioat_ring_ent *next = ring[i+1]; + struct ioat_dma_descriptor *hw = ring[i]->hw; + + hw->next = next->txd.phys; + } + ring[i]->hw->next = ring[0]->txd.phys; + + return ring; +} + +/* ioat2_alloc_chan_resources - allocate/initialize ioat2 descriptor ring + * @chan: channel to be initialized + */ +int ioat2_alloc_chan_resources(struct dma_chan *c) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + struct ioat_chan_common *chan = &ioat->base; + struct ioat_ring_ent **ring; + int order; + + /* have we already been set up? */ + if (ioat->ring) + return 1 << ioat->alloc_order; + + /* Setup register to interrupt and write completion status on error */ + writew(IOAT_CHANCTRL_RUN, chan->reg_base + IOAT_CHANCTRL_OFFSET); + + /* allocate a completion writeback area */ + /* doing 2 32bit writes to mmio since 1 64b write doesn't work */ + chan->completion = pci_pool_alloc(chan->device->completion_pool, + GFP_KERNEL, &chan->completion_dma); + if (!chan->completion) + return -ENOMEM; + + memset(chan->completion, 0, sizeof(*chan->completion)); + writel(((u64) chan->completion_dma) & 0x00000000FFFFFFFF, + chan->reg_base + IOAT_CHANCMP_OFFSET_LOW); + writel(((u64) chan->completion_dma) >> 32, + chan->reg_base + IOAT_CHANCMP_OFFSET_HIGH); + + order = ioat_get_alloc_order(); + ring = ioat2_alloc_ring(c, order, GFP_KERNEL); + if (!ring) + return -ENOMEM; + + spin_lock_bh(&ioat->ring_lock); + ioat->ring = ring; + ioat->head = 0; + ioat->issued = 0; + ioat->tail = 0; + ioat->pending = 0; + ioat->alloc_order = order; + spin_unlock_bh(&ioat->ring_lock); + + tasklet_enable(&chan->cleanup_task); + ioat2_start_null_desc(ioat); + + return 1 << ioat->alloc_order; +} + +bool reshape_ring(struct ioat2_dma_chan *ioat, int order) +{ + /* reshape differs from normal ring allocation in that we want + * to allocate a new software ring while only + * extending/truncating the hardware ring + */ + struct ioat_chan_common *chan = &ioat->base; + struct dma_chan *c = &chan->common; + const u16 curr_size = ioat2_ring_mask(ioat) + 1; + const u16 active = ioat2_ring_active(ioat); + const u16 new_size = 1 << order; + struct ioat_ring_ent **ring; + u16 i; + + if (order > ioat_get_max_alloc_order()) + return false; + + /* double check that we have at least 1 free descriptor */ + if (active == curr_size) + return false; + + /* when shrinking, verify that we can hold the current active + * set in the new ring + */ + if (active >= new_size) + return false; + + /* allocate the array to hold the software ring */ + ring = kcalloc(new_size, sizeof(*ring), GFP_NOWAIT); + if (!ring) + return false; + + /* allocate/trim descriptors as needed */ + if (new_size > curr_size) { + /* copy current descriptors to the new ring */ + for (i = 0; i < curr_size; i++) { + u16 curr_idx = (ioat->tail+i) & (curr_size-1); + u16 new_idx = (ioat->tail+i) & (new_size-1); + + ring[new_idx] = ioat->ring[curr_idx]; + set_desc_id(ring[new_idx], new_idx); + } + + /* add new descriptors to the ring */ + for (i = curr_size; i < new_size; i++) { + u16 new_idx = (ioat->tail+i) & (new_size-1); + + ring[new_idx] = ioat2_alloc_ring_ent(c, GFP_NOWAIT); + if (!ring[new_idx]) { + while (i--) { + u16 new_idx = (ioat->tail+i) & (new_size-1); + + ioat2_free_ring_ent(ring[new_idx], c); + } + kfree(ring); + return false; + } + set_desc_id(ring[new_idx], new_idx); + } + + /* hw link new descriptors */ + for (i = curr_size-1; i < new_size; i++) { + u16 new_idx = (ioat->tail+i) & (new_size-1); + struct ioat_ring_ent *next = ring[(new_idx+1) & (new_size-1)]; + struct ioat_dma_descriptor *hw = ring[new_idx]->hw; + + hw->next = next->txd.phys; + } + } else { + struct ioat_dma_descriptor *hw; + struct ioat_ring_ent *next; + + /* copy current descriptors to the new ring, dropping the + * removed descriptors + */ + for (i = 0; i < new_size; i++) { + u16 curr_idx = (ioat->tail+i) & (curr_size-1); + u16 new_idx = (ioat->tail+i) & (new_size-1); + + ring[new_idx] = ioat->ring[curr_idx]; + set_desc_id(ring[new_idx], new_idx); + } + + /* free deleted descriptors */ + for (i = new_size; i < curr_size; i++) { + struct ioat_ring_ent *ent; + + ent = ioat2_get_ring_ent(ioat, ioat->tail+i); + ioat2_free_ring_ent(ent, c); + } + + /* fix up hardware ring */ + hw = ring[(ioat->tail+new_size-1) & (new_size-1)]->hw; + next = ring[(ioat->tail+new_size) & (new_size-1)]; + hw->next = next->txd.phys; + } + + dev_dbg(to_dev(chan), "%s: allocated %d descriptors\n", + __func__, new_size); + + kfree(ioat->ring); + ioat->ring = ring; + ioat->alloc_order = order; + + return true; +} + +/** + * ioat2_alloc_and_lock - common descriptor alloc boilerplate for ioat2,3 ops + * @idx: gets starting descriptor index on successful allocation + * @ioat: ioat2,3 channel (ring) to operate on + * @num_descs: allocation length + */ +int ioat2_alloc_and_lock(u16 *idx, struct ioat2_dma_chan *ioat, int num_descs) +{ + struct ioat_chan_common *chan = &ioat->base; + + spin_lock_bh(&ioat->ring_lock); + /* never allow the last descriptor to be consumed, we need at + * least one free at all times to allow for on-the-fly ring + * resizing. + */ + while (unlikely(ioat2_ring_space(ioat) <= num_descs)) { + if (reshape_ring(ioat, ioat->alloc_order + 1) && + ioat2_ring_space(ioat) > num_descs) + break; + + if (printk_ratelimit()) + dev_dbg(to_dev(chan), + "%s: ring full! num_descs: %d (%x:%x:%x)\n", + __func__, num_descs, ioat->head, ioat->tail, + ioat->issued); + spin_unlock_bh(&ioat->ring_lock); + + /* progress reclaim in the allocation failure case we + * may be called under bh_disabled so we need to trigger + * the timer event directly + */ + spin_lock_bh(&chan->cleanup_lock); + if (jiffies > chan->timer.expires && + timer_pending(&chan->timer)) { + struct ioatdma_device *device = chan->device; + + mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT); + spin_unlock_bh(&chan->cleanup_lock); + device->timer_fn((unsigned long) ioat); + } else + spin_unlock_bh(&chan->cleanup_lock); + return -ENOMEM; + } + + dev_dbg(to_dev(chan), "%s: num_descs: %d (%x:%x:%x)\n", + __func__, num_descs, ioat->head, ioat->tail, ioat->issued); + + *idx = ioat2_desc_alloc(ioat, num_descs); + return 0; /* with ioat->ring_lock held */ +} + +struct dma_async_tx_descriptor * +ioat2_dma_prep_memcpy_lock(struct dma_chan *c, dma_addr_t dma_dest, + dma_addr_t dma_src, size_t len, unsigned long flags) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + struct ioat_dma_descriptor *hw; + struct ioat_ring_ent *desc; + dma_addr_t dst = dma_dest; + dma_addr_t src = dma_src; + size_t total_len = len; + int num_descs; + u16 idx; + int i; + + num_descs = ioat2_xferlen_to_descs(ioat, len); + if (likely(num_descs) && + ioat2_alloc_and_lock(&idx, ioat, num_descs) == 0) + /* pass */; + else + return NULL; + i = 0; + do { + size_t copy = min_t(size_t, len, 1 << ioat->xfercap_log); + + desc = ioat2_get_ring_ent(ioat, idx + i); + hw = desc->hw; + + hw->size = copy; + hw->ctl = 0; + hw->src_addr = src; + hw->dst_addr = dst; + + len -= copy; + dst += copy; + src += copy; + dump_desc_dbg(ioat, desc); + } while (++i < num_descs); + + desc->txd.flags = flags; + desc->len = total_len; + hw->ctl_f.int_en = !!(flags & DMA_PREP_INTERRUPT); + hw->ctl_f.fence = !!(flags & DMA_PREP_FENCE); + hw->ctl_f.compl_write = 1; + dump_desc_dbg(ioat, desc); + /* we leave the channel locked to ensure in order submission */ + + return &desc->txd; +} + +/** + * ioat2_free_chan_resources - release all the descriptors + * @chan: the channel to be cleaned + */ +void ioat2_free_chan_resources(struct dma_chan *c) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + struct ioat_chan_common *chan = &ioat->base; + struct ioatdma_device *device = chan->device; + struct ioat_ring_ent *desc; + const u16 total_descs = 1 << ioat->alloc_order; + int descs; + int i; + + /* Before freeing channel resources first check + * if they have been previously allocated for this channel. + */ + if (!ioat->ring) + return; + + tasklet_disable(&chan->cleanup_task); + del_timer_sync(&chan->timer); + device->cleanup_tasklet((unsigned long) ioat); + device->reset_hw(chan); + + spin_lock_bh(&ioat->ring_lock); + descs = ioat2_ring_space(ioat); + dev_dbg(to_dev(chan), "freeing %d idle descriptors\n", descs); + for (i = 0; i < descs; i++) { + desc = ioat2_get_ring_ent(ioat, ioat->head + i); + ioat2_free_ring_ent(desc, c); + } + + if (descs < total_descs) + dev_err(to_dev(chan), "Freeing %d in use descriptors!\n", + total_descs - descs); + + for (i = 0; i < total_descs - descs; i++) { + desc = ioat2_get_ring_ent(ioat, ioat->tail + i); + dump_desc_dbg(ioat, desc); + ioat2_free_ring_ent(desc, c); + } + + kfree(ioat->ring); + ioat->ring = NULL; + ioat->alloc_order = 0; + pci_pool_free(device->completion_pool, chan->completion, + chan->completion_dma); + spin_unlock_bh(&ioat->ring_lock); + + chan->last_completion = 0; + chan->completion_dma = 0; + ioat->pending = 0; + ioat->dmacount = 0; +} + +enum dma_status +ioat2_is_complete(struct dma_chan *c, dma_cookie_t cookie, + dma_cookie_t *done, dma_cookie_t *used) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + struct ioatdma_device *device = ioat->base.device; + + if (ioat_is_complete(c, cookie, done, used) == DMA_SUCCESS) + return DMA_SUCCESS; + + device->cleanup_tasklet((unsigned long) ioat); + + return ioat_is_complete(c, cookie, done, used); +} + +static ssize_t ring_size_show(struct dma_chan *c, char *page) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + + return sprintf(page, "%d\n", (1 << ioat->alloc_order) & ~1); +} +static struct ioat_sysfs_entry ring_size_attr = __ATTR_RO(ring_size); + +static ssize_t ring_active_show(struct dma_chan *c, char *page) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + + /* ...taken outside the lock, no need to be precise */ + return sprintf(page, "%d\n", ioat2_ring_active(ioat)); +} +static struct ioat_sysfs_entry ring_active_attr = __ATTR_RO(ring_active); + +static struct attribute *ioat2_attrs[] = { + &ring_size_attr.attr, + &ring_active_attr.attr, + &ioat_cap_attr.attr, + &ioat_version_attr.attr, + NULL, +}; + +struct kobj_type ioat2_ktype = { + .sysfs_ops = &ioat_sysfs_ops, + .default_attrs = ioat2_attrs, +}; + +int __devinit ioat2_dma_probe(struct ioatdma_device *device, int dca) +{ + struct pci_dev *pdev = device->pdev; + struct dma_device *dma; + struct dma_chan *c; + struct ioat_chan_common *chan; + int err; + + device->enumerate_channels = ioat2_enumerate_channels; + device->reset_hw = ioat2_reset_hw; + device->cleanup_tasklet = ioat2_cleanup_tasklet; + device->timer_fn = ioat2_timer_event; + device->self_test = ioat_dma_self_test; + dma = &device->common; + dma->device_prep_dma_memcpy = ioat2_dma_prep_memcpy_lock; + dma->device_issue_pending = ioat2_issue_pending; + dma->device_alloc_chan_resources = ioat2_alloc_chan_resources; + dma->device_free_chan_resources = ioat2_free_chan_resources; + dma->device_is_tx_complete = ioat2_is_complete; + + err = ioat_probe(device); + if (err) + return err; + ioat_set_tcp_copy_break(2048); + + list_for_each_entry(c, &dma->channels, device_node) { + chan = to_chan_common(c); + writel(IOAT_DCACTRL_CMPL_WRITE_ENABLE | IOAT_DMA_DCA_ANY_CPU, + chan->reg_base + IOAT_DCACTRL_OFFSET); + } + + err = ioat_register(device); + if (err) + return err; + + ioat_kobject_add(device, &ioat2_ktype); + + if (dca) + device->dca = ioat2_dca_init(pdev, device->reg_base); + + return err; +} diff --git a/drivers/dma/ioat/dma_v2.h b/drivers/dma/ioat/dma_v2.h new file mode 100644 index 00000000000..3afad8da43c --- /dev/null +++ b/drivers/dma/ioat/dma_v2.h @@ -0,0 +1,192 @@ +/* + * Copyright(c) 2004 - 2009 Intel Corporation. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the Free + * Software Foundation; either version 2 of the License, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 + * Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * The full GNU General Public License is included in this distribution in the + * file called COPYING. + */ +#ifndef IOATDMA_V2_H +#define IOATDMA_V2_H + +#include <linux/dmaengine.h> +#include "dma.h" +#include "hw.h" + + +extern int ioat_pending_level; +extern int ioat_ring_alloc_order; + +/* + * workaround for IOAT ver.3.0 null descriptor issue + * (channel returns error when size is 0) + */ +#define NULL_DESC_BUFFER_SIZE 1 + +#define IOAT_MAX_ORDER 16 +#define ioat_get_alloc_order() \ + (min(ioat_ring_alloc_order, IOAT_MAX_ORDER)) +#define ioat_get_max_alloc_order() \ + (min(ioat_ring_max_alloc_order, IOAT_MAX_ORDER)) + +/* struct ioat2_dma_chan - ioat v2 / v3 channel attributes + * @base: common ioat channel parameters + * @xfercap_log; log2 of channel max transfer length (for fast division) + * @head: allocated index + * @issued: hardware notification point + * @tail: cleanup index + * @pending: lock free indicator for issued != head + * @dmacount: identical to 'head' except for occasionally resetting to zero + * @alloc_order: log2 of the number of allocated descriptors + * @ring: software ring buffer implementation of hardware ring + * @ring_lock: protects ring attributes + */ +struct ioat2_dma_chan { + struct ioat_chan_common base; + size_t xfercap_log; + u16 head; + u16 issued; + u16 tail; + u16 dmacount; + u16 alloc_order; + int pending; + struct ioat_ring_ent **ring; + spinlock_t ring_lock; +}; + +static inline struct ioat2_dma_chan *to_ioat2_chan(struct dma_chan *c) +{ + struct ioat_chan_common *chan = to_chan_common(c); + + return container_of(chan, struct ioat2_dma_chan, base); +} + +static inline u16 ioat2_ring_mask(struct ioat2_dma_chan *ioat) +{ + return (1 << ioat->alloc_order) - 1; +} + +/* count of descriptors in flight with the engine */ +static inline u16 ioat2_ring_active(struct ioat2_dma_chan *ioat) +{ + return (ioat->head - ioat->tail) & ioat2_ring_mask(ioat); +} + +/* count of descriptors pending submission to hardware */ +static inline u16 ioat2_ring_pending(struct ioat2_dma_chan *ioat) +{ + return (ioat->head - ioat->issued) & ioat2_ring_mask(ioat); +} + +static inline u16 ioat2_ring_space(struct ioat2_dma_chan *ioat) +{ + u16 num_descs = ioat2_ring_mask(ioat) + 1; + u16 active = ioat2_ring_active(ioat); + + BUG_ON(active > num_descs); + + return num_descs - active; +} + +/* assumes caller already checked space */ +static inline u16 ioat2_desc_alloc(struct ioat2_dma_chan *ioat, u16 len) +{ + ioat->head += len; + return ioat->head - len; +} + +static inline u16 ioat2_xferlen_to_descs(struct ioat2_dma_chan *ioat, size_t len) +{ + u16 num_descs = len >> ioat->xfercap_log; + + num_descs += !!(len & ((1 << ioat->xfercap_log) - 1)); + return num_descs; +} + +/** + * struct ioat_ring_ent - wrapper around hardware descriptor + * @hw: hardware DMA descriptor (for memcpy) + * @fill: hardware fill descriptor + * @xor: hardware xor descriptor + * @xor_ex: hardware xor extension descriptor + * @pq: hardware pq descriptor + * @pq_ex: hardware pq extension descriptor + * @pqu: hardware pq update descriptor + * @raw: hardware raw (un-typed) descriptor + * @txd: the generic software descriptor for all engines + * @len: total transaction length for unmap + * @result: asynchronous result of validate operations + * @id: identifier for debug + */ + +struct ioat_ring_ent { + union { + struct ioat_dma_descriptor *hw; + struct ioat_fill_descriptor *fill; + struct ioat_xor_descriptor *xor; + struct ioat_xor_ext_descriptor *xor_ex; + struct ioat_pq_descriptor *pq; + struct ioat_pq_ext_descriptor *pq_ex; + struct ioat_pq_update_descriptor *pqu; + struct ioat_raw_descriptor *raw; + }; + size_t len; + struct dma_async_tx_descriptor txd; + enum sum_check_flags *result; + #ifdef DEBUG + int id; + #endif +}; + +static inline struct ioat_ring_ent * +ioat2_get_ring_ent(struct ioat2_dma_chan *ioat, u16 idx) +{ + return ioat->ring[idx & ioat2_ring_mask(ioat)]; +} + +static inline void ioat2_set_chainaddr(struct ioat2_dma_chan *ioat, u64 addr) +{ + struct ioat_chan_common *chan = &ioat->base; + + writel(addr & 0x00000000FFFFFFFF, + chan->reg_base + IOAT2_CHAINADDR_OFFSET_LOW); + writel(addr >> 32, + chan->reg_base + IOAT2_CHAINADDR_OFFSET_HIGH); +} + +int __devinit ioat2_dma_probe(struct ioatdma_device *dev, int dca); +int __devinit ioat3_dma_probe(struct ioatdma_device *dev, int dca); +struct dca_provider * __devinit ioat2_dca_init(struct pci_dev *pdev, void __iomem *iobase); +struct dca_provider * __devinit ioat3_dca_init(struct pci_dev *pdev, void __iomem *iobase); +int ioat2_alloc_and_lock(u16 *idx, struct ioat2_dma_chan *ioat, int num_descs); +int ioat2_enumerate_channels(struct ioatdma_device *device); +struct dma_async_tx_descriptor * +ioat2_dma_prep_memcpy_lock(struct dma_chan *c, dma_addr_t dma_dest, + dma_addr_t dma_src, size_t len, unsigned long flags); +void ioat2_issue_pending(struct dma_chan *chan); +int ioat2_alloc_chan_resources(struct dma_chan *c); +void ioat2_free_chan_resources(struct dma_chan *c); +enum dma_status ioat2_is_complete(struct dma_chan *c, dma_cookie_t cookie, + dma_cookie_t *done, dma_cookie_t *used); +void __ioat2_restart_chan(struct ioat2_dma_chan *ioat); +bool reshape_ring(struct ioat2_dma_chan *ioat, int order); +void __ioat2_issue_pending(struct ioat2_dma_chan *ioat); +void ioat2_cleanup_tasklet(unsigned long data); +void ioat2_timer_event(unsigned long data); +int ioat2_quiesce(struct ioat_chan_common *chan, unsigned long tmo); +int ioat2_reset_sync(struct ioat_chan_common *chan, unsigned long tmo); +extern struct kobj_type ioat2_ktype; +extern struct kmem_cache *ioat2_cache; +#endif /* IOATDMA_V2_H */ diff --git a/drivers/dma/ioat/dma_v3.c b/drivers/dma/ioat/dma_v3.c new file mode 100644 index 00000000000..9908c9e94b2 --- /dev/null +++ b/drivers/dma/ioat/dma_v3.c @@ -0,0 +1,1281 @@ +/* + * This file is provided under a dual BSD/GPLv2 license. When using or + * redistributing this file, you may do so under either license. + * + * GPL LICENSE SUMMARY + * + * Copyright(c) 2004 - 2009 Intel Corporation. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + * + * The full GNU General Public License is included in this distribution in + * the file called "COPYING". + * + * BSD LICENSE + * + * Copyright(c) 2004-2009 Intel Corporation. All rights reserved. + * + * Redistribution and use in source and binary forms, with or without + * modification, are permitted provided that the following conditions are met: + * + * * Redistributions of source code must retain the above copyright + * notice, this list of conditions and the following disclaimer. + * * Redistributions in binary form must reproduce the above copyright + * notice, this list of conditions and the following disclaimer in + * the documentation and/or other materials provided with the + * distribution. + * * Neither the name of Intel Corporation nor the names of its + * contributors may be used to endorse or promote products derived + * from this software without specific prior written permission. + * + * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" + * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE + * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE + * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE + * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR + * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF + * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS + * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN + * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) + * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE + * POSSIBILITY OF SUCH DAMAGE. + */ + +/* + * Support routines for v3+ hardware + */ + +#include <linux/pci.h> +#include <linux/dmaengine.h> +#include <linux/dma-mapping.h> +#include "registers.h" +#include "hw.h" +#include "dma.h" +#include "dma_v2.h" + +/* ioat hardware assumes at least two sources for raid operations */ +#define src_cnt_to_sw(x) ((x) + 2) +#define src_cnt_to_hw(x) ((x) - 2) + +/* provide a lookup table for setting the source address in the base or + * extended descriptor of an xor or pq descriptor + */ +static const u8 xor_idx_to_desc __read_mostly = 0xd0; +static const u8 xor_idx_to_field[] __read_mostly = { 1, 4, 5, 6, 7, 0, 1, 2 }; +static const u8 pq_idx_to_desc __read_mostly = 0xf8; +static const u8 pq_idx_to_field[] __read_mostly = { 1, 4, 5, 0, 1, 2, 4, 5 }; + +static dma_addr_t xor_get_src(struct ioat_raw_descriptor *descs[2], int idx) +{ + struct ioat_raw_descriptor *raw = descs[xor_idx_to_desc >> idx & 1]; + + return raw->field[xor_idx_to_field[idx]]; +} + +static void xor_set_src(struct ioat_raw_descriptor *descs[2], + dma_addr_t addr, u32 offset, int idx) +{ + struct ioat_raw_descriptor *raw = descs[xor_idx_to_desc >> idx & 1]; + + raw->field[xor_idx_to_field[idx]] = addr + offset; +} + +static dma_addr_t pq_get_src(struct ioat_raw_descriptor *descs[2], int idx) +{ + struct ioat_raw_descriptor *raw = descs[pq_idx_to_desc >> idx & 1]; + + return raw->field[pq_idx_to_field[idx]]; +} + +static void pq_set_src(struct ioat_raw_descriptor *descs[2], + dma_addr_t addr, u32 offset, u8 coef, int idx) +{ + struct ioat_pq_descriptor *pq = (struct ioat_pq_descriptor *) descs[0]; + struct ioat_raw_descriptor *raw = descs[pq_idx_to_desc >> idx & 1]; + + raw->field[pq_idx_to_field[idx]] = addr + offset; + pq->coef[idx] = coef; +} + +static void ioat3_dma_unmap(struct ioat2_dma_chan *ioat, + struct ioat_ring_ent *desc, int idx) +{ + struct ioat_chan_common *chan = &ioat->base; + struct pci_dev *pdev = chan->device->pdev; + size_t len = desc->len; + size_t offset = len - desc->hw->size; + struct dma_async_tx_descriptor *tx = &desc->txd; + enum dma_ctrl_flags flags = tx->flags; + + switch (desc->hw->ctl_f.op) { + case IOAT_OP_COPY: + if (!desc->hw->ctl_f.null) /* skip 'interrupt' ops */ + ioat_dma_unmap(chan, flags, len, desc->hw); + break; + case IOAT_OP_FILL: { + struct ioat_fill_descriptor *hw = desc->fill; + + if (!(flags & DMA_COMPL_SKIP_DEST_UNMAP)) + ioat_unmap(pdev, hw->dst_addr - offset, len, + PCI_DMA_FROMDEVICE, flags, 1); + break; + } + case IOAT_OP_XOR_VAL: + case IOAT_OP_XOR: { + struct ioat_xor_descriptor *xor = desc->xor; + struct ioat_ring_ent *ext; + struct ioat_xor_ext_descriptor *xor_ex = NULL; + int src_cnt = src_cnt_to_sw(xor->ctl_f.src_cnt); + struct ioat_raw_descriptor *descs[2]; + int i; + + if (src_cnt > 5) { + ext = ioat2_get_ring_ent(ioat, idx + 1); + xor_ex = ext->xor_ex; + } + + if (!(flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + descs[0] = (struct ioat_raw_descriptor *) xor; + descs[1] = (struct ioat_raw_descriptor *) xor_ex; + for (i = 0; i < src_cnt; i++) { + dma_addr_t src = xor_get_src(descs, i); + + ioat_unmap(pdev, src - offset, len, + PCI_DMA_TODEVICE, flags, 0); + } + + /* dest is a source in xor validate operations */ + if (xor->ctl_f.op == IOAT_OP_XOR_VAL) { + ioat_unmap(pdev, xor->dst_addr - offset, len, + PCI_DMA_TODEVICE, flags, 1); + break; + } + } + + if (!(flags & DMA_COMPL_SKIP_DEST_UNMAP)) + ioat_unmap(pdev, xor->dst_addr - offset, len, + PCI_DMA_FROMDEVICE, flags, 1); + break; + } + case IOAT_OP_PQ_VAL: + case IOAT_OP_PQ: { + struct ioat_pq_descriptor *pq = desc->pq; + struct ioat_ring_ent *ext; + struct ioat_pq_ext_descriptor *pq_ex = NULL; + int src_cnt = src_cnt_to_sw(pq->ctl_f.src_cnt); + struct ioat_raw_descriptor *descs[2]; + int i; + + if (src_cnt > 3) { + ext = ioat2_get_ring_ent(ioat, idx + 1); + pq_ex = ext->pq_ex; + } + + /* in the 'continue' case don't unmap the dests as sources */ + if (dmaf_p_disabled_continue(flags)) + src_cnt--; + else if (dmaf_continue(flags)) + src_cnt -= 3; + + if (!(flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + descs[0] = (struct ioat_raw_descriptor *) pq; + descs[1] = (struct ioat_raw_descriptor *) pq_ex; + for (i = 0; i < src_cnt; i++) { + dma_addr_t src = pq_get_src(descs, i); + + ioat_unmap(pdev, src - offset, len, + PCI_DMA_TODEVICE, flags, 0); + } + + /* the dests are sources in pq validate operations */ + if (pq->ctl_f.op == IOAT_OP_XOR_VAL) { + if (!(flags & DMA_PREP_PQ_DISABLE_P)) + ioat_unmap(pdev, pq->p_addr - offset, + len, PCI_DMA_TODEVICE, flags, 0); + if (!(flags & DMA_PREP_PQ_DISABLE_Q)) + ioat_unmap(pdev, pq->q_addr - offset, + len, PCI_DMA_TODEVICE, flags, 0); + break; + } + } + + if (!(flags & DMA_COMPL_SKIP_DEST_UNMAP)) { + if (!(flags & DMA_PREP_PQ_DISABLE_P)) + ioat_unmap(pdev, pq->p_addr - offset, len, + PCI_DMA_BIDIRECTIONAL, flags, 1); + if (!(flags & DMA_PREP_PQ_DISABLE_Q)) + ioat_unmap(pdev, pq->q_addr - offset, len, + PCI_DMA_BIDIRECTIONAL, flags, 1); + } + break; + } + default: + dev_err(&pdev->dev, "%s: unknown op type: %#x\n", + __func__, desc->hw->ctl_f.op); + } +} + +static bool desc_has_ext(struct ioat_ring_ent *desc) +{ + struct ioat_dma_descriptor *hw = desc->hw; + + if (hw->ctl_f.op == IOAT_OP_XOR || + hw->ctl_f.op == IOAT_OP_XOR_VAL) { + struct ioat_xor_descriptor *xor = desc->xor; + + if (src_cnt_to_sw(xor->ctl_f.src_cnt) > 5) + return true; + } else if (hw->ctl_f.op == IOAT_OP_PQ || + hw->ctl_f.op == IOAT_OP_PQ_VAL) { + struct ioat_pq_descriptor *pq = desc->pq; + + if (src_cnt_to_sw(pq->ctl_f.src_cnt) > 3) + return true; + } + + return false; +} + +/** + * __cleanup - reclaim used descriptors + * @ioat: channel (ring) to clean + * + * The difference from the dma_v2.c __cleanup() is that this routine + * handles extended descriptors and dma-unmapping raid operations. + */ +static void __cleanup(struct ioat2_dma_chan *ioat, unsigned long phys_complete) +{ + struct ioat_chan_common *chan = &ioat->base; + struct ioat_ring_ent *desc; + bool seen_current = false; + u16 active; + int i; + + dev_dbg(to_dev(chan), "%s: head: %#x tail: %#x issued: %#x\n", + __func__, ioat->head, ioat->tail, ioat->issued); + + active = ioat2_ring_active(ioat); + for (i = 0; i < active && !seen_current; i++) { + struct dma_async_tx_descriptor *tx; + + prefetch(ioat2_get_ring_ent(ioat, ioat->tail + i + 1)); + desc = ioat2_get_ring_ent(ioat, ioat->tail + i); + dump_desc_dbg(ioat, desc); + tx = &desc->txd; + if (tx->cookie) { + chan->completed_cookie = tx->cookie; + ioat3_dma_unmap(ioat, desc, ioat->tail + i); + tx->cookie = 0; + if (tx->callback) { + tx->callback(tx->callback_param); + tx->callback = NULL; + } + } + + if (tx->phys == phys_complete) + seen_current = true; + + /* skip extended descriptors */ + if (desc_has_ext(desc)) { + BUG_ON(i + 1 >= active); + i++; + } + } + ioat->tail += i; + BUG_ON(!seen_current); /* no active descs have written a completion? */ + chan->last_completion = phys_complete; + if (ioat->head == ioat->tail) { + dev_dbg(to_dev(chan), "%s: cancel completion timeout\n", + __func__); + clear_bit(IOAT_COMPLETION_PENDING, &chan->state); + mod_timer(&chan->timer, jiffies + IDLE_TIMEOUT); + } +} + +static void ioat3_cleanup(struct ioat2_dma_chan *ioat) +{ + struct ioat_chan_common *chan = &ioat->base; + unsigned long phys_complete; + + prefetch(chan->completion); + + if (!spin_trylock_bh(&chan->cleanup_lock)) + return; + + if (!ioat_cleanup_preamble(chan, &phys_complete)) { + spin_unlock_bh(&chan->cleanup_lock); + return; + } + + if (!spin_trylock_bh(&ioat->ring_lock)) { + spin_unlock_bh(&chan->cleanup_lock); + return; + } + + __cleanup(ioat, phys_complete); + + spin_unlock_bh(&ioat->ring_lock); + spin_unlock_bh(&chan->cleanup_lock); +} + +static void ioat3_cleanup_tasklet(unsigned long data) +{ + struct ioat2_dma_chan *ioat = (void *) data; + + ioat3_cleanup(ioat); + writew(IOAT_CHANCTRL_RUN | IOAT3_CHANCTRL_COMPL_DCA_EN, + ioat->base.reg_base + IOAT_CHANCTRL_OFFSET); +} + +static void ioat3_restart_channel(struct ioat2_dma_chan *ioat) +{ + struct ioat_chan_common *chan = &ioat->base; + unsigned long phys_complete; + u32 status; + + status = ioat_chansts(chan); + if (is_ioat_active(status) || is_ioat_idle(status)) + ioat_suspend(chan); + while (is_ioat_active(status) || is_ioat_idle(status)) { + status = ioat_chansts(chan); + cpu_relax(); + } + + if (ioat_cleanup_preamble(chan, &phys_complete)) + __cleanup(ioat, phys_complete); + + __ioat2_restart_chan(ioat); +} + +static void ioat3_timer_event(unsigned long data) +{ + struct ioat2_dma_chan *ioat = (void *) data; + struct ioat_chan_common *chan = &ioat->base; + + spin_lock_bh(&chan->cleanup_lock); + if (test_bit(IOAT_COMPLETION_PENDING, &chan->state)) { + unsigned long phys_complete; + u64 status; + + spin_lock_bh(&ioat->ring_lock); + status = ioat_chansts(chan); + + /* when halted due to errors check for channel + * programming errors before advancing the completion state + */ + if (is_ioat_halted(status)) { + u32 chanerr; + + chanerr = readl(chan->reg_base + IOAT_CHANERR_OFFSET); + dev_err(to_dev(chan), "%s: Channel halted (%x)\n", + __func__, chanerr); + BUG_ON(is_ioat_bug(chanerr)); + } + + /* if we haven't made progress and we have already + * acknowledged a pending completion once, then be more + * forceful with a restart + */ + if (ioat_cleanup_preamble(chan, &phys_complete)) + __cleanup(ioat, phys_complete); + else if (test_bit(IOAT_COMPLETION_ACK, &chan->state)) + ioat3_restart_channel(ioat); + else { + set_bit(IOAT_COMPLETION_ACK, &chan->state); + mod_timer(&chan->timer, jiffies + COMPLETION_TIMEOUT); + } + spin_unlock_bh(&ioat->ring_lock); + } else { + u16 active; + + /* if the ring is idle, empty, and oversized try to step + * down the size + */ + spin_lock_bh(&ioat->ring_lock); + active = ioat2_ring_active(ioat); + if (active == 0 && ioat->alloc_order > ioat_get_alloc_order()) + reshape_ring(ioat, ioat->alloc_order-1); + spin_unlock_bh(&ioat->ring_lock); + + /* keep shrinking until we get back to our minimum + * default size + */ + if (ioat->alloc_order > ioat_get_alloc_order()) + mod_timer(&chan->timer, jiffies + IDLE_TIMEOUT); + } + spin_unlock_bh(&chan->cleanup_lock); +} + +static enum dma_status +ioat3_is_complete(struct dma_chan *c, dma_cookie_t cookie, + dma_cookie_t *done, dma_cookie_t *used) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + + if (ioat_is_complete(c, cookie, done, used) == DMA_SUCCESS) + return DMA_SUCCESS; + + ioat3_cleanup(ioat); + + return ioat_is_complete(c, cookie, done, used); +} + +static struct dma_async_tx_descriptor * +ioat3_prep_memset_lock(struct dma_chan *c, dma_addr_t dest, int value, + size_t len, unsigned long flags) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + struct ioat_ring_ent *desc; + size_t total_len = len; + struct ioat_fill_descriptor *fill; + int num_descs; + u64 src_data = (0x0101010101010101ULL) * (value & 0xff); + u16 idx; + int i; + + num_descs = ioat2_xferlen_to_descs(ioat, len); + if (likely(num_descs) && + ioat2_alloc_and_lock(&idx, ioat, num_descs) == 0) + /* pass */; + else + return NULL; + i = 0; + do { + size_t xfer_size = min_t(size_t, len, 1 << ioat->xfercap_log); + + desc = ioat2_get_ring_ent(ioat, idx + i); + fill = desc->fill; + + fill->size = xfer_size; + fill->src_data = src_data; + fill->dst_addr = dest; + fill->ctl = 0; + fill->ctl_f.op = IOAT_OP_FILL; + + len -= xfer_size; + dest += xfer_size; + dump_desc_dbg(ioat, desc); + } while (++i < num_descs); + + desc->txd.flags = flags; + desc->len = total_len; + fill->ctl_f.int_en = !!(flags & DMA_PREP_INTERRUPT); + fill->ctl_f.fence = !!(flags & DMA_PREP_FENCE); + fill->ctl_f.compl_write = 1; + dump_desc_dbg(ioat, desc); + + /* we leave the channel locked to ensure in order submission */ + return &desc->txd; +} + +static struct dma_async_tx_descriptor * +__ioat3_prep_xor_lock(struct dma_chan *c, enum sum_check_flags *result, + dma_addr_t dest, dma_addr_t *src, unsigned int src_cnt, + size_t len, unsigned long flags) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + struct ioat_ring_ent *compl_desc; + struct ioat_ring_ent *desc; + struct ioat_ring_ent *ext; + size_t total_len = len; + struct ioat_xor_descriptor *xor; + struct ioat_xor_ext_descriptor *xor_ex = NULL; + struct ioat_dma_descriptor *hw; + u32 offset = 0; + int num_descs; + int with_ext; + int i; + u16 idx; + u8 op = result ? IOAT_OP_XOR_VAL : IOAT_OP_XOR; + + BUG_ON(src_cnt < 2); + + num_descs = ioat2_xferlen_to_descs(ioat, len); + /* we need 2x the number of descriptors to cover greater than 5 + * sources + */ + if (src_cnt > 5) { + with_ext = 1; + num_descs *= 2; + } else + with_ext = 0; + + /* completion writes from the raid engine may pass completion + * writes from the legacy engine, so we need one extra null + * (legacy) descriptor to ensure all completion writes arrive in + * order. + */ + if (likely(num_descs) && + ioat2_alloc_and_lock(&idx, ioat, num_descs+1) == 0) + /* pass */; + else + return NULL; + i = 0; + do { + struct ioat_raw_descriptor *descs[2]; + size_t xfer_size = min_t(size_t, len, 1 << ioat->xfercap_log); + int s; + + desc = ioat2_get_ring_ent(ioat, idx + i); + xor = desc->xor; + + /* save a branch by unconditionally retrieving the + * extended descriptor xor_set_src() knows to not write + * to it in the single descriptor case + */ + ext = ioat2_get_ring_ent(ioat, idx + i + 1); + xor_ex = ext->xor_ex; + + descs[0] = (struct ioat_raw_descriptor *) xor; + descs[1] = (struct ioat_raw_descriptor *) xor_ex; + for (s = 0; s < src_cnt; s++) + xor_set_src(descs, src[s], offset, s); + xor->size = xfer_size; + xor->dst_addr = dest + offset; + xor->ctl = 0; + xor->ctl_f.op = op; + xor->ctl_f.src_cnt = src_cnt_to_hw(src_cnt); + + len -= xfer_size; + offset += xfer_size; + dump_desc_dbg(ioat, desc); + } while ((i += 1 + with_ext) < num_descs); + + /* last xor descriptor carries the unmap parameters and fence bit */ + desc->txd.flags = flags; + desc->len = total_len; + if (result) + desc->result = result; + xor->ctl_f.fence = !!(flags & DMA_PREP_FENCE); + + /* completion descriptor carries interrupt bit */ + compl_desc = ioat2_get_ring_ent(ioat, idx + i); + compl_desc->txd.flags = flags & DMA_PREP_INTERRUPT; + hw = compl_desc->hw; + hw->ctl = 0; + hw->ctl_f.null = 1; + hw->ctl_f.int_en = !!(flags & DMA_PREP_INTERRUPT); + hw->ctl_f.compl_write = 1; + hw->size = NULL_DESC_BUFFER_SIZE; + dump_desc_dbg(ioat, compl_desc); + + /* we leave the channel locked to ensure in order submission */ + return &compl_desc->txd; +} + +static struct dma_async_tx_descriptor * +ioat3_prep_xor(struct dma_chan *chan, dma_addr_t dest, dma_addr_t *src, + unsigned int src_cnt, size_t len, unsigned long flags) +{ + return __ioat3_prep_xor_lock(chan, NULL, dest, src, src_cnt, len, flags); +} + +struct dma_async_tx_descriptor * +ioat3_prep_xor_val(struct dma_chan *chan, dma_addr_t *src, + unsigned int src_cnt, size_t len, + enum sum_check_flags *result, unsigned long flags) +{ + /* the cleanup routine only sets bits on validate failure, it + * does not clear bits on validate success... so clear it here + */ + *result = 0; + + return __ioat3_prep_xor_lock(chan, result, src[0], &src[1], + src_cnt - 1, len, flags); +} + +static void +dump_pq_desc_dbg(struct ioat2_dma_chan *ioat, struct ioat_ring_ent *desc, struct ioat_ring_ent *ext) +{ + struct device *dev = to_dev(&ioat->base); + struct ioat_pq_descriptor *pq = desc->pq; + struct ioat_pq_ext_descriptor *pq_ex = ext ? ext->pq_ex : NULL; + struct ioat_raw_descriptor *descs[] = { (void *) pq, (void *) pq_ex }; + int src_cnt = src_cnt_to_sw(pq->ctl_f.src_cnt); + int i; + + dev_dbg(dev, "desc[%d]: (%#llx->%#llx) flags: %#x" + " sz: %#x ctl: %#x (op: %d int: %d compl: %d pq: '%s%s' src_cnt: %d)\n", + desc_id(desc), (unsigned long long) desc->txd.phys, + (unsigned long long) (pq_ex ? pq_ex->next : pq->next), + desc->txd.flags, pq->size, pq->ctl, pq->ctl_f.op, pq->ctl_f.int_en, + pq->ctl_f.compl_write, + pq->ctl_f.p_disable ? "" : "p", pq->ctl_f.q_disable ? "" : "q", + pq->ctl_f.src_cnt); + for (i = 0; i < src_cnt; i++) + dev_dbg(dev, "\tsrc[%d]: %#llx coef: %#x\n", i, + (unsigned long long) pq_get_src(descs, i), pq->coef[i]); + dev_dbg(dev, "\tP: %#llx\n", pq->p_addr); + dev_dbg(dev, "\tQ: %#llx\n", pq->q_addr); +} + +static struct dma_async_tx_descriptor * +__ioat3_prep_pq_lock(struct dma_chan *c, enum sum_check_flags *result, + const dma_addr_t *dst, const dma_addr_t *src, + unsigned int src_cnt, const unsigned char *scf, + size_t len, unsigned long flags) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + struct ioat_chan_common *chan = &ioat->base; + struct ioat_ring_ent *compl_desc; + struct ioat_ring_ent *desc; + struct ioat_ring_ent *ext; + size_t total_len = len; + struct ioat_pq_descriptor *pq; + struct ioat_pq_ext_descriptor *pq_ex = NULL; + struct ioat_dma_descriptor *hw; + u32 offset = 0; + int num_descs; + int with_ext; + int i, s; + u16 idx; + u8 op = result ? IOAT_OP_PQ_VAL : IOAT_OP_PQ; + + dev_dbg(to_dev(chan), "%s\n", __func__); + /* the engine requires at least two sources (we provide + * at least 1 implied source in the DMA_PREP_CONTINUE case) + */ + BUG_ON(src_cnt + dmaf_continue(flags) < 2); + + num_descs = ioat2_xferlen_to_descs(ioat, len); + /* we need 2x the number of descriptors to cover greater than 3 + * sources (we need 1 extra source in the q-only continuation + * case and 3 extra sources in the p+q continuation case. + */ + if (src_cnt + dmaf_p_disabled_continue(flags) > 3 || + (dmaf_continue(flags) && !dmaf_p_disabled_continue(flags))) { + with_ext = 1; + num_descs *= 2; + } else + with_ext = 0; + + /* completion writes from the raid engine may pass completion + * writes from the legacy engine, so we need one extra null + * (legacy) descriptor to ensure all completion writes arrive in + * order. + */ + if (likely(num_descs) && + ioat2_alloc_and_lock(&idx, ioat, num_descs+1) == 0) + /* pass */; + else + return NULL; + i = 0; + do { + struct ioat_raw_descriptor *descs[2]; + size_t xfer_size = min_t(size_t, len, 1 << ioat->xfercap_log); + + desc = ioat2_get_ring_ent(ioat, idx + i); + pq = desc->pq; + + /* save a branch by unconditionally retrieving the + * extended descriptor pq_set_src() knows to not write + * to it in the single descriptor case + */ + ext = ioat2_get_ring_ent(ioat, idx + i + with_ext); + pq_ex = ext->pq_ex; + + descs[0] = (struct ioat_raw_descriptor *) pq; + descs[1] = (struct ioat_raw_descriptor *) pq_ex; + + for (s = 0; s < src_cnt; s++) + pq_set_src(descs, src[s], offset, scf[s], s); + + /* see the comment for dma_maxpq in include/linux/dmaengine.h */ + if (dmaf_p_disabled_continue(flags)) + pq_set_src(descs, dst[1], offset, 1, s++); + else if (dmaf_continue(flags)) { + pq_set_src(descs, dst[0], offset, 0, s++); + pq_set_src(descs, dst[1], offset, 1, s++); + pq_set_src(descs, dst[1], offset, 0, s++); + } + pq->size = xfer_size; + pq->p_addr = dst[0] + offset; + pq->q_addr = dst[1] + offset; + pq->ctl = 0; + pq->ctl_f.op = op; + pq->ctl_f.src_cnt = src_cnt_to_hw(s); + pq->ctl_f.p_disable = !!(flags & DMA_PREP_PQ_DISABLE_P); + pq->ctl_f.q_disable = !!(flags & DMA_PREP_PQ_DISABLE_Q); + + len -= xfer_size; + offset += xfer_size; + } while ((i += 1 + with_ext) < num_descs); + + /* last pq descriptor carries the unmap parameters and fence bit */ + desc->txd.flags = flags; + desc->len = total_len; + if (result) + desc->result = result; + pq->ctl_f.fence = !!(flags & DMA_PREP_FENCE); + dump_pq_desc_dbg(ioat, desc, ext); + + /* completion descriptor carries interrupt bit */ + compl_desc = ioat2_get_ring_ent(ioat, idx + i); + compl_desc->txd.flags = flags & DMA_PREP_INTERRUPT; + hw = compl_desc->hw; + hw->ctl = 0; + hw->ctl_f.null = 1; + hw->ctl_f.int_en = !!(flags & DMA_PREP_INTERRUPT); + hw->ctl_f.compl_write = 1; + hw->size = NULL_DESC_BUFFER_SIZE; + dump_desc_dbg(ioat, compl_desc); + + /* we leave the channel locked to ensure in order submission */ + return &compl_desc->txd; +} + +static struct dma_async_tx_descriptor * +ioat3_prep_pq(struct dma_chan *chan, dma_addr_t *dst, dma_addr_t *src, + unsigned int src_cnt, const unsigned char *scf, size_t len, + unsigned long flags) +{ + /* specify valid address for disabled result */ + if (flags & DMA_PREP_PQ_DISABLE_P) + dst[0] = dst[1]; + if (flags & DMA_PREP_PQ_DISABLE_Q) + dst[1] = dst[0]; + + /* handle the single source multiply case from the raid6 + * recovery path + */ + if ((flags & DMA_PREP_PQ_DISABLE_P) && src_cnt == 1) { + dma_addr_t single_source[2]; + unsigned char single_source_coef[2]; + + BUG_ON(flags & DMA_PREP_PQ_DISABLE_Q); + single_source[0] = src[0]; + single_source[1] = src[0]; + single_source_coef[0] = scf[0]; + single_source_coef[1] = 0; + + return __ioat3_prep_pq_lock(chan, NULL, dst, single_source, 2, + single_source_coef, len, flags); + } else + return __ioat3_prep_pq_lock(chan, NULL, dst, src, src_cnt, scf, + len, flags); +} + +struct dma_async_tx_descriptor * +ioat3_prep_pq_val(struct dma_chan *chan, dma_addr_t *pq, dma_addr_t *src, + unsigned int src_cnt, const unsigned char *scf, size_t len, + enum sum_check_flags *pqres, unsigned long flags) +{ + /* specify valid address for disabled result */ + if (flags & DMA_PREP_PQ_DISABLE_P) + pq[0] = pq[1]; + if (flags & DMA_PREP_PQ_DISABLE_Q) + pq[1] = pq[0]; + + /* the cleanup routine only sets bits on validate failure, it + * does not clear bits on validate success... so clear it here + */ + *pqres = 0; + + return __ioat3_prep_pq_lock(chan, pqres, pq, src, src_cnt, scf, len, + flags); +} + +static struct dma_async_tx_descriptor * +ioat3_prep_pqxor(struct dma_chan *chan, dma_addr_t dst, dma_addr_t *src, + unsigned int src_cnt, size_t len, unsigned long flags) +{ + unsigned char scf[src_cnt]; + dma_addr_t pq[2]; + + memset(scf, 0, src_cnt); + pq[0] = dst; + flags |= DMA_PREP_PQ_DISABLE_Q; + pq[1] = dst; /* specify valid address for disabled result */ + + return __ioat3_prep_pq_lock(chan, NULL, pq, src, src_cnt, scf, len, + flags); +} + +struct dma_async_tx_descriptor * +ioat3_prep_pqxor_val(struct dma_chan *chan, dma_addr_t *src, + unsigned int src_cnt, size_t len, + enum sum_check_flags *result, unsigned long flags) +{ + unsigned char scf[src_cnt]; + dma_addr_t pq[2]; + + /* the cleanup routine only sets bits on validate failure, it + * does not clear bits on validate success... so clear it here + */ + *result = 0; + + memset(scf, 0, src_cnt); + pq[0] = src[0]; + flags |= DMA_PREP_PQ_DISABLE_Q; + pq[1] = pq[0]; /* specify valid address for disabled result */ + + return __ioat3_prep_pq_lock(chan, result, pq, &src[1], src_cnt - 1, scf, + len, flags); +} + +static struct dma_async_tx_descriptor * +ioat3_prep_interrupt_lock(struct dma_chan *c, unsigned long flags) +{ + struct ioat2_dma_chan *ioat = to_ioat2_chan(c); + struct ioat_ring_ent *desc; + struct ioat_dma_descriptor *hw; + u16 idx; + + if (ioat2_alloc_and_lock(&idx, ioat, 1) == 0) + desc = ioat2_get_ring_ent(ioat, idx); + else + return NULL; + + hw = desc->hw; + hw->ctl = 0; + hw->ctl_f.null = 1; + hw->ctl_f.int_en = 1; + hw->ctl_f.fence = !!(flags & DMA_PREP_FENCE); + hw->ctl_f.compl_write = 1; + hw->size = NULL_DESC_BUFFER_SIZE; + hw->src_addr = 0; + hw->dst_addr = 0; + + desc->txd.flags = flags; + desc->len = 1; + + dump_desc_dbg(ioat, desc); + + /* we leave the channel locked to ensure in order submission */ + return &desc->txd; +} + +static void __devinit ioat3_dma_test_callback(void *dma_async_param) +{ + struct completion *cmp = dma_async_param; + + complete(cmp); +} + +#define IOAT_NUM_SRC_TEST 6 /* must be <= 8 */ +static int __devinit ioat_xor_val_self_test(struct ioatdma_device *device) +{ + int i, src_idx; + struct page *dest; + struct page *xor_srcs[IOAT_NUM_SRC_TEST]; + struct page *xor_val_srcs[IOAT_NUM_SRC_TEST + 1]; + dma_addr_t dma_srcs[IOAT_NUM_SRC_TEST + 1]; + dma_addr_t dma_addr, dest_dma; + struct dma_async_tx_descriptor *tx; + struct dma_chan *dma_chan; + dma_cookie_t cookie; + u8 cmp_byte = 0; + u32 cmp_word; + u32 xor_val_result; + int err = 0; + struct completion cmp; + unsigned long tmo; + struct device *dev = &device->pdev->dev; + struct dma_device *dma = &device->common; + + dev_dbg(dev, "%s\n", __func__); + + if (!dma_has_cap(DMA_XOR, dma->cap_mask)) + return 0; + + for (src_idx = 0; src_idx < IOAT_NUM_SRC_TEST; src_idx++) { + xor_srcs[src_idx] = alloc_page(GFP_KERNEL); + if (!xor_srcs[src_idx]) { + while (src_idx--) + __free_page(xor_srcs[src_idx]); + return -ENOMEM; + } + } + + dest = alloc_page(GFP_KERNEL); + if (!dest) { + while (src_idx--) + __free_page(xor_srcs[src_idx]); + return -ENOMEM; + } + + /* Fill in src buffers */ + for (src_idx = 0; src_idx < IOAT_NUM_SRC_TEST; src_idx++) { + u8 *ptr = page_address(xor_srcs[src_idx]); + for (i = 0; i < PAGE_SIZE; i++) + ptr[i] = (1 << src_idx); + } + + for (src_idx = 0; src_idx < IOAT_NUM_SRC_TEST; src_idx++) + cmp_byte ^= (u8) (1 << src_idx); + + cmp_word = (cmp_byte << 24) | (cmp_byte << 16) | + (cmp_byte << 8) | cmp_byte; + + memset(page_address(dest), 0, PAGE_SIZE); + + dma_chan = container_of(dma->channels.next, struct dma_chan, + device_node); + if (dma->device_alloc_chan_resources(dma_chan) < 1) { + err = -ENODEV; + goto out; + } + + /* test xor */ + dest_dma = dma_map_page(dev, dest, 0, PAGE_SIZE, DMA_FROM_DEVICE); + for (i = 0; i < IOAT_NUM_SRC_TEST; i++) + dma_srcs[i] = dma_map_page(dev, xor_srcs[i], 0, PAGE_SIZE, + DMA_TO_DEVICE); + tx = dma->device_prep_dma_xor(dma_chan, dest_dma, dma_srcs, + IOAT_NUM_SRC_TEST, PAGE_SIZE, + DMA_PREP_INTERRUPT); + + if (!tx) { + dev_err(dev, "Self-test xor prep failed\n"); + err = -ENODEV; + goto free_resources; + } + + async_tx_ack(tx); + init_completion(&cmp); + tx->callback = ioat3_dma_test_callback; + tx->callback_param = &cmp; + cookie = tx->tx_submit(tx); + if (cookie < 0) { + dev_err(dev, "Self-test xor setup failed\n"); + err = -ENODEV; + goto free_resources; + } + dma->device_issue_pending(dma_chan); + + tmo = wait_for_completion_timeout(&cmp, msecs_to_jiffies(3000)); + + if (dma->device_is_tx_complete(dma_chan, cookie, NULL, NULL) != DMA_SUCCESS) { + dev_err(dev, "Self-test xor timed out\n"); + err = -ENODEV; + goto free_resources; + } + + dma_sync_single_for_cpu(dev, dest_dma, PAGE_SIZE, DMA_FROM_DEVICE); + for (i = 0; i < (PAGE_SIZE / sizeof(u32)); i++) { + u32 *ptr = page_address(dest); + if (ptr[i] != cmp_word) { + dev_err(dev, "Self-test xor failed compare\n"); + err = -ENODEV; + goto free_resources; + } + } + dma_sync_single_for_device(dev, dest_dma, PAGE_SIZE, DMA_TO_DEVICE); + + /* skip validate if the capability is not present */ + if (!dma_has_cap(DMA_XOR_VAL, dma_chan->device->cap_mask)) + goto free_resources; + + /* validate the sources with the destintation page */ + for (i = 0; i < IOAT_NUM_SRC_TEST; i++) + xor_val_srcs[i] = xor_srcs[i]; + xor_val_srcs[i] = dest; + + xor_val_result = 1; + + for (i = 0; i < IOAT_NUM_SRC_TEST + 1; i++) + dma_srcs[i] = dma_map_page(dev, xor_val_srcs[i], 0, PAGE_SIZE, + DMA_TO_DEVICE); + tx = dma->device_prep_dma_xor_val(dma_chan, dma_srcs, + IOAT_NUM_SRC_TEST + 1, PAGE_SIZE, + &xor_val_result, DMA_PREP_INTERRUPT); + if (!tx) { + dev_err(dev, "Self-test zero prep failed\n"); + err = -ENODEV; + goto free_resources; + } + + async_tx_ack(tx); + init_completion(&cmp); + tx->callback = ioat3_dma_test_callback; + tx->callback_param = &cmp; + cookie = tx->tx_submit(tx); + if (cookie < 0) { + dev_err(dev, "Self-test zero setup failed\n"); + err = -ENODEV; + goto free_resources; + } + dma->device_issue_pending(dma_chan); + + tmo = wait_for_completion_timeout(&cmp, msecs_to_jiffies(3000)); + + if (dma->device_is_tx_complete(dma_chan, cookie, NULL, NULL) != DMA_SUCCESS) { + dev_err(dev, "Self-test validate timed out\n"); + err = -ENODEV; + goto free_resources; + } + + if (xor_val_result != 0) { + dev_err(dev, "Self-test validate failed compare\n"); + err = -ENODEV; + goto free_resources; + } + + /* skip memset if the capability is not present */ + if (!dma_has_cap(DMA_MEMSET, dma_chan->device->cap_mask)) + goto free_resources; + + /* test memset */ + dma_addr = dma_map_page(dev, dest, 0, + PAGE_SIZE, DMA_FROM_DEVICE); + tx = dma->device_prep_dma_memset(dma_chan, dma_addr, 0, PAGE_SIZE, + DMA_PREP_INTERRUPT); + if (!tx) { + dev_err(dev, "Self-test memset prep failed\n"); + err = -ENODEV; + goto free_resources; + } + + async_tx_ack(tx); + init_completion(&cmp); + tx->callback = ioat3_dma_test_callback; + tx->callback_param = &cmp; + cookie = tx->tx_submit(tx); + if (cookie < 0) { + dev_err(dev, "Self-test memset setup failed\n"); + err = -ENODEV; + goto free_resources; + } + dma->device_issue_pending(dma_chan); + + tmo = wait_for_completion_timeout(&cmp, msecs_to_jiffies(3000)); + + if (dma->device_is_tx_complete(dma_chan, cookie, NULL, NULL) != DMA_SUCCESS) { + dev_err(dev, "Self-test memset timed out\n"); + err = -ENODEV; + goto free_resources; + } + + for (i = 0; i < PAGE_SIZE/sizeof(u32); i++) { + u32 *ptr = page_address(dest); + if (ptr[i]) { + dev_err(dev, "Self-test memset failed compare\n"); + err = -ENODEV; + goto free_resources; + } + } + + /* test for non-zero parity sum */ + xor_val_result = 0; + for (i = 0; i < IOAT_NUM_SRC_TEST + 1; i++) + dma_srcs[i] = dma_map_page(dev, xor_val_srcs[i], 0, PAGE_SIZE, + DMA_TO_DEVICE); + tx = dma->device_prep_dma_xor_val(dma_chan, dma_srcs, + IOAT_NUM_SRC_TEST + 1, PAGE_SIZE, + &xor_val_result, DMA_PREP_INTERRUPT); + if (!tx) { + dev_err(dev, "Self-test 2nd zero prep failed\n"); + err = -ENODEV; + goto free_resources; + } + + async_tx_ack(tx); + init_completion(&cmp); + tx->callback = ioat3_dma_test_callback; + tx->callback_param = &cmp; + cookie = tx->tx_submit(tx); + if (cookie < 0) { + dev_err(dev, "Self-test 2nd zero setup failed\n"); + err = -ENODEV; + goto free_resources; + } + dma->device_issue_pending(dma_chan); + + tmo = wait_for_completion_timeout(&cmp, msecs_to_jiffies(3000)); + + if (dma->device_is_tx_complete(dma_chan, cookie, NULL, NULL) != DMA_SUCCESS) { + dev_err(dev, "Self-test 2nd validate timed out\n"); + err = -ENODEV; + goto free_resources; + } + + if (xor_val_result != SUM_CHECK_P_RESULT) { + dev_err(dev, "Self-test validate failed compare\n"); + err = -ENODEV; + goto free_resources; + } + +free_resources: + dma->device_free_chan_resources(dma_chan); +out: + src_idx = IOAT_NUM_SRC_TEST; + while (src_idx--) + __free_page(xor_srcs[src_idx]); + __free_page(dest); + return err; +} + +static int __devinit ioat3_dma_self_test(struct ioatdma_device *device) +{ + int rc = ioat_dma_self_test(device); + + if (rc) + return rc; + + rc = ioat_xor_val_self_test(device); + if (rc) + return rc; + + return 0; +} + +static int ioat3_reset_hw(struct ioat_chan_common *chan) +{ + /* throw away whatever the channel was doing and get it + * initialized, with ioat3 specific workarounds + */ + struct ioatdma_device *device = chan->device; + struct pci_dev *pdev = device->pdev; + u32 chanerr; + u16 dev_id; + int err; + + ioat2_quiesce(chan, msecs_to_jiffies(100)); + + chanerr = readl(chan->reg_base + IOAT_CHANERR_OFFSET); + writel(chanerr, chan->reg_base + IOAT_CHANERR_OFFSET); + + /* -= IOAT ver.3 workarounds =- */ + /* Write CHANERRMSK_INT with 3E07h to mask out the errors + * that can cause stability issues for IOAT ver.3, and clear any + * pending errors + */ + pci_write_config_dword(pdev, IOAT_PCI_CHANERRMASK_INT_OFFSET, 0x3e07); + err = pci_read_config_dword(pdev, IOAT_PCI_CHANERR_INT_OFFSET, &chanerr); + if (err) { + dev_err(&pdev->dev, "channel error register unreachable\n"); + return err; + } + pci_write_config_dword(pdev, IOAT_PCI_CHANERR_INT_OFFSET, chanerr); + + /* Clear DMAUNCERRSTS Cfg-Reg Parity Error status bit + * (workaround for spurious config parity error after restart) + */ + pci_read_config_word(pdev, IOAT_PCI_DEVICE_ID_OFFSET, &dev_id); + if (dev_id == PCI_DEVICE_ID_INTEL_IOAT_TBG0) + pci_write_config_dword(pdev, IOAT_PCI_DMAUNCERRSTS_OFFSET, 0x10); + + return ioat2_reset_sync(chan, msecs_to_jiffies(200)); +} + +int __devinit ioat3_dma_probe(struct ioatdma_device *device, int dca) +{ + struct pci_dev *pdev = device->pdev; + int dca_en = system_has_dca_enabled(pdev); + struct dma_device *dma; + struct dma_chan *c; + struct ioat_chan_common *chan; + bool is_raid_device = false; + int err; + u32 cap; + + device->enumerate_channels = ioat2_enumerate_channels; + device->reset_hw = ioat3_reset_hw; + device->self_test = ioat3_dma_self_test; + dma = &device->common; + dma->device_prep_dma_memcpy = ioat2_dma_prep_memcpy_lock; + dma->device_issue_pending = ioat2_issue_pending; + dma->device_alloc_chan_resources = ioat2_alloc_chan_resources; + dma->device_free_chan_resources = ioat2_free_chan_resources; + + dma_cap_set(DMA_INTERRUPT, dma->cap_mask); + dma->device_prep_dma_interrupt = ioat3_prep_interrupt_lock; + + cap = readl(device->reg_base + IOAT_DMA_CAP_OFFSET); + + /* dca is incompatible with raid operations */ + if (dca_en && (cap & (IOAT_CAP_XOR|IOAT_CAP_PQ))) + cap &= ~(IOAT_CAP_XOR|IOAT_CAP_PQ); + + if (cap & IOAT_CAP_XOR) { + is_raid_device = true; + dma->max_xor = 8; + dma->xor_align = 2; + + dma_cap_set(DMA_XOR, dma->cap_mask); + dma->device_prep_dma_xor = ioat3_prep_xor; + + dma_cap_set(DMA_XOR_VAL, dma->cap_mask); + dma->device_prep_dma_xor_val = ioat3_prep_xor_val; + } + if (cap & IOAT_CAP_PQ) { + is_raid_device = true; + dma_set_maxpq(dma, 8, 0); + dma->pq_align = 2; + + dma_cap_set(DMA_PQ, dma->cap_mask); + dma->device_prep_dma_pq = ioat3_prep_pq; + + dma_cap_set(DMA_PQ_VAL, dma->cap_mask); + dma->device_prep_dma_pq_val = ioat3_prep_pq_val; + + if (!(cap & IOAT_CAP_XOR)) { + dma->max_xor = 8; + dma->xor_align = 2; + + dma_cap_set(DMA_XOR, dma->cap_mask); + dma->device_prep_dma_xor = ioat3_prep_pqxor; + + dma_cap_set(DMA_XOR_VAL, dma->cap_mask); + dma->device_prep_dma_xor_val = ioat3_prep_pqxor_val; + } + } + if (is_raid_device && (cap & IOAT_CAP_FILL_BLOCK)) { + dma_cap_set(DMA_MEMSET, dma->cap_mask); + dma->device_prep_dma_memset = ioat3_prep_memset_lock; + } + + + if (is_raid_device) { + dma->device_is_tx_complete = ioat3_is_complete; + device->cleanup_tasklet = ioat3_cleanup_tasklet; + device->timer_fn = ioat3_timer_event; + } else { + dma->device_is_tx_complete = ioat2_is_complete; + device->cleanup_tasklet = ioat2_cleanup_tasklet; + device->timer_fn = ioat2_timer_event; + } + + #ifdef CONFIG_ASYNC_TX_DISABLE_PQ_VAL_DMA + dma_cap_clear(DMA_PQ_VAL, dma->cap_mask); + dma->device_prep_dma_pq_val = NULL; + #endif + + #ifdef CONFIG_ASYNC_TX_DISABLE_XOR_VAL_DMA + dma_cap_clear(DMA_XOR_VAL, dma->cap_mask); + dma->device_prep_dma_xor_val = NULL; + #endif + + err = ioat_probe(device); + if (err) + return err; + ioat_set_tcp_copy_break(262144); + + list_for_each_entry(c, &dma->channels, device_node) { + chan = to_chan_common(c); + writel(IOAT_DMA_DCA_ANY_CPU, + chan->reg_base + IOAT_DCACTRL_OFFSET); + } + + err = ioat_register(device); + if (err) + return err; + + ioat_kobject_add(device, &ioat2_ktype); + + if (dca) + device->dca = ioat3_dca_init(pdev, device->reg_base); + + return 0; +} diff --git a/drivers/dma/ioat/hw.h b/drivers/dma/ioat/hw.h new file mode 100644 index 00000000000..60e675455b6 --- /dev/null +++ b/drivers/dma/ioat/hw.h @@ -0,0 +1,217 @@ +/* + * Copyright(c) 2004 - 2009 Intel Corporation. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the Free + * Software Foundation; either version 2 of the License, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 + * Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * The full GNU General Public License is included in this distribution in the + * file called COPYING. + */ +#ifndef _IOAT_HW_H_ +#define _IOAT_HW_H_ + +/* PCI Configuration Space Values */ +#define IOAT_PCI_VID 0x8086 +#define IOAT_MMIO_BAR 0 + +/* CB device ID's */ +#define IOAT_PCI_DID_5000 0x1A38 +#define IOAT_PCI_DID_CNB 0x360B +#define IOAT_PCI_DID_SCNB 0x65FF +#define IOAT_PCI_DID_SNB 0x402F + +#define IOAT_PCI_RID 0x00 +#define IOAT_PCI_SVID 0x8086 +#define IOAT_PCI_SID 0x8086 +#define IOAT_VER_1_2 0x12 /* Version 1.2 */ +#define IOAT_VER_2_0 0x20 /* Version 2.0 */ +#define IOAT_VER_3_0 0x30 /* Version 3.0 */ +#define IOAT_VER_3_2 0x32 /* Version 3.2 */ + +int system_has_dca_enabled(struct pci_dev *pdev); + +struct ioat_dma_descriptor { + uint32_t size; + union { + uint32_t ctl; + struct { + unsigned int int_en:1; + unsigned int src_snoop_dis:1; + unsigned int dest_snoop_dis:1; + unsigned int compl_write:1; + unsigned int fence:1; + unsigned int null:1; + unsigned int src_brk:1; + unsigned int dest_brk:1; + unsigned int bundle:1; + unsigned int dest_dca:1; + unsigned int hint:1; + unsigned int rsvd2:13; + #define IOAT_OP_COPY 0x00 + unsigned int op:8; + } ctl_f; + }; + uint64_t src_addr; + uint64_t dst_addr; + uint64_t next; + uint64_t rsv1; + uint64_t rsv2; + /* store some driver data in an unused portion of the descriptor */ + union { + uint64_t user1; + uint64_t tx_cnt; + }; + uint64_t user2; +}; + +struct ioat_fill_descriptor { + uint32_t size; + union { + uint32_t ctl; + struct { + unsigned int int_en:1; + unsigned int rsvd:1; + unsigned int dest_snoop_dis:1; + unsigned int compl_write:1; + unsigned int fence:1; + unsigned int rsvd2:2; + unsigned int dest_brk:1; + unsigned int bundle:1; + unsigned int rsvd4:15; + #define IOAT_OP_FILL 0x01 + unsigned int op:8; + } ctl_f; + }; + uint64_t src_data; + uint64_t dst_addr; + uint64_t next; + uint64_t rsv1; + uint64_t next_dst_addr; + uint64_t user1; + uint64_t user2; +}; + +struct ioat_xor_descriptor { + uint32_t size; + union { + uint32_t ctl; + struct { + unsigned int int_en:1; + unsigned int src_snoop_dis:1; + unsigned int dest_snoop_dis:1; + unsigned int compl_write:1; + unsigned int fence:1; + unsigned int src_cnt:3; + unsigned int bundle:1; + unsigned int dest_dca:1; + unsigned int hint:1; + unsigned int rsvd:13; + #define IOAT_OP_XOR 0x87 + #define IOAT_OP_XOR_VAL 0x88 + unsigned int op:8; + } ctl_f; + }; + uint64_t src_addr; + uint64_t dst_addr; + uint64_t next; + uint64_t src_addr2; + uint64_t src_addr3; + uint64_t src_addr4; + uint64_t src_addr5; +}; + +struct ioat_xor_ext_descriptor { + uint64_t src_addr6; + uint64_t src_addr7; + uint64_t src_addr8; + uint64_t next; + uint64_t rsvd[4]; +}; + +struct ioat_pq_descriptor { + uint32_t size; + union { + uint32_t ctl; + struct { + unsigned int int_en:1; + unsigned int src_snoop_dis:1; + unsigned int dest_snoop_dis:1; + unsigned int compl_write:1; + unsigned int fence:1; + unsigned int src_cnt:3; + unsigned int bundle:1; + unsigned int dest_dca:1; + unsigned int hint:1; + unsigned int p_disable:1; + unsigned int q_disable:1; + unsigned int rsvd:11; + #define IOAT_OP_PQ 0x89 + #define IOAT_OP_PQ_VAL 0x8a + unsigned int op:8; + } ctl_f; + }; + uint64_t src_addr; + uint64_t p_addr; + uint64_t next; + uint64_t src_addr2; + uint64_t src_addr3; + uint8_t coef[8]; + uint64_t q_addr; +}; + +struct ioat_pq_ext_descriptor { + uint64_t src_addr4; + uint64_t src_addr5; + uint64_t src_addr6; + uint64_t next; + uint64_t src_addr7; + uint64_t src_addr8; + uint64_t rsvd[2]; +}; + +struct ioat_pq_update_descriptor { + uint32_t size; + union { + uint32_t ctl; + struct { + unsigned int int_en:1; + unsigned int src_snoop_dis:1; + unsigned int dest_snoop_dis:1; + unsigned int compl_write:1; + unsigned int fence:1; + unsigned int src_cnt:3; + unsigned int bundle:1; + unsigned int dest_dca:1; + unsigned int hint:1; + unsigned int p_disable:1; + unsigned int q_disable:1; + unsigned int rsvd:3; + unsigned int coef:8; + #define IOAT_OP_PQ_UP 0x8b + unsigned int op:8; + } ctl_f; + }; + uint64_t src_addr; + uint64_t p_addr; + uint64_t next; + uint64_t src_addr2; + uint64_t p_src; + uint64_t q_src; + uint64_t q_addr; +}; + +struct ioat_raw_descriptor { + uint64_t field[8]; +}; +#endif diff --git a/drivers/dma/ioat/pci.c b/drivers/dma/ioat/pci.c new file mode 100644 index 00000000000..d545fae30f3 --- /dev/null +++ b/drivers/dma/ioat/pci.c @@ -0,0 +1,210 @@ +/* + * Intel I/OAT DMA Linux driver + * Copyright(c) 2007 - 2009 Intel Corporation. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + * + * The full GNU General Public License is included in this distribution in + * the file called "COPYING". + * + */ + +/* + * This driver supports an Intel I/OAT DMA engine, which does asynchronous + * copy operations. + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/pci.h> +#include <linux/interrupt.h> +#include <linux/dca.h> +#include "dma.h" +#include "dma_v2.h" +#include "registers.h" +#include "hw.h" + +MODULE_VERSION(IOAT_DMA_VERSION); +MODULE_LICENSE("Dual BSD/GPL"); +MODULE_AUTHOR("Intel Corporation"); + +static struct pci_device_id ioat_pci_tbl[] = { + /* I/OAT v1 platforms */ + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_CNB) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_SCNB) }, + { PCI_VDEVICE(UNISYS, PCI_DEVICE_ID_UNISYS_DMA_DIRECTOR) }, + + /* I/OAT v2 platforms */ + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_SNB) }, + + /* I/OAT v3 platforms */ + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_TBG0) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_TBG1) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_TBG2) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_TBG3) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_TBG4) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_TBG5) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_TBG6) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_TBG7) }, + + /* I/OAT v3.2 platforms */ + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF0) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF1) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF2) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF3) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF4) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF5) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF6) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF7) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF8) }, + { PCI_VDEVICE(INTEL, PCI_DEVICE_ID_INTEL_IOAT_JSF9) }, + + { 0, } +}; +MODULE_DEVICE_TABLE(pci, ioat_pci_tbl); + +static int __devinit ioat_pci_probe(struct pci_dev *pdev, + const struct pci_device_id *id); +static void __devexit ioat_remove(struct pci_dev *pdev); + +static int ioat_dca_enabled = 1; +module_param(ioat_dca_enabled, int, 0644); +MODULE_PARM_DESC(ioat_dca_enabled, "control support of dca service (default: 1)"); + +struct kmem_cache *ioat2_cache; + +#define DRV_NAME "ioatdma" + +static struct pci_driver ioat_pci_driver = { + .name = DRV_NAME, + .id_table = ioat_pci_tbl, + .probe = ioat_pci_probe, + .remove = __devexit_p(ioat_remove), +}; + +static struct ioatdma_device * +alloc_ioatdma(struct pci_dev *pdev, void __iomem *iobase) +{ + struct device *dev = &pdev->dev; + struct ioatdma_device *d = devm_kzalloc(dev, sizeof(*d), GFP_KERNEL); + + if (!d) + return NULL; + d->pdev = pdev; + d->reg_base = iobase; + return d; +} + +static int __devinit ioat_pci_probe(struct pci_dev *pdev, const struct pci_device_id *id) +{ + void __iomem * const *iomap; + struct device *dev = &pdev->dev; + struct ioatdma_device *device; + int err; + + err = pcim_enable_device(pdev); + if (err) + return err; + + err = pcim_iomap_regions(pdev, 1 << IOAT_MMIO_BAR, DRV_NAME); + if (err) + return err; + iomap = pcim_iomap_table(pdev); + if (!iomap) + return -ENOMEM; + + err = pci_set_dma_mask(pdev, DMA_BIT_MASK(64)); + if (err) + err = pci_set_dma_mask(pdev, DMA_BIT_MASK(32)); + if (err) + return err; + + err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(64)); + if (err) + err = pci_set_consistent_dma_mask(pdev, DMA_BIT_MASK(32)); + if (err) + return err; + + device = devm_kzalloc(dev, sizeof(*device), GFP_KERNEL); + if (!device) + return -ENOMEM; + + pci_set_master(pdev); + + device = alloc_ioatdma(pdev, iomap[IOAT_MMIO_BAR]); + if (!device) + return -ENOMEM; + pci_set_drvdata(pdev, device); + + device->version = readb(device->reg_base + IOAT_VER_OFFSET); + if (device->version == IOAT_VER_1_2) + err = ioat1_dma_probe(device, ioat_dca_enabled); + else if (device->version == IOAT_VER_2_0) + err = ioat2_dma_probe(device, ioat_dca_enabled); + else if (device->version >= IOAT_VER_3_0) + err = ioat3_dma_probe(device, ioat_dca_enabled); + else + return -ENODEV; + + if (err) { + dev_err(dev, "Intel(R) I/OAT DMA Engine init failed\n"); + return -ENODEV; + } + + return 0; +} + +static void __devexit ioat_remove(struct pci_dev *pdev) +{ + struct ioatdma_device *device = pci_get_drvdata(pdev); + + if (!device) + return; + + dev_err(&pdev->dev, "Removing dma and dca services\n"); + if (device->dca) { + unregister_dca_provider(device->dca, &pdev->dev); + free_dca_provider(device->dca); + device->dca = NULL; + } + ioat_dma_remove(device); +} + +static int __init ioat_init_module(void) +{ + int err; + + pr_info("%s: Intel(R) QuickData Technology Driver %s\n", + DRV_NAME, IOAT_DMA_VERSION); + + ioat2_cache = kmem_cache_create("ioat2", sizeof(struct ioat_ring_ent), + 0, SLAB_HWCACHE_ALIGN, NULL); + if (!ioat2_cache) + return -ENOMEM; + + err = pci_register_driver(&ioat_pci_driver); + if (err) + kmem_cache_destroy(ioat2_cache); + + return err; +} +module_init(ioat_init_module); + +static void __exit ioat_exit_module(void) +{ + pci_unregister_driver(&ioat_pci_driver); + kmem_cache_destroy(ioat2_cache); +} +module_exit(ioat_exit_module); diff --git a/drivers/dma/ioat/registers.h b/drivers/dma/ioat/registers.h new file mode 100644 index 00000000000..e8ae63baf58 --- /dev/null +++ b/drivers/dma/ioat/registers.h @@ -0,0 +1,249 @@ +/* + * Copyright(c) 2004 - 2009 Intel Corporation. All rights reserved. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the Free + * Software Foundation; either version 2 of the License, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 + * Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * The full GNU General Public License is included in this distribution in the + * file called COPYING. + */ +#ifndef _IOAT_REGISTERS_H_ +#define _IOAT_REGISTERS_H_ + +#define IOAT_PCI_DMACTRL_OFFSET 0x48 +#define IOAT_PCI_DMACTRL_DMA_EN 0x00000001 +#define IOAT_PCI_DMACTRL_MSI_EN 0x00000002 + +#define IOAT_PCI_DEVICE_ID_OFFSET 0x02 +#define IOAT_PCI_DMAUNCERRSTS_OFFSET 0x148 +#define IOAT_PCI_CHANERR_INT_OFFSET 0x180 +#define IOAT_PCI_CHANERRMASK_INT_OFFSET 0x184 + +/* MMIO Device Registers */ +#define IOAT_CHANCNT_OFFSET 0x00 /* 8-bit */ + +#define IOAT_XFERCAP_OFFSET 0x01 /* 8-bit */ +#define IOAT_XFERCAP_4KB 12 +#define IOAT_XFERCAP_8KB 13 +#define IOAT_XFERCAP_16KB 14 +#define IOAT_XFERCAP_32KB 15 +#define IOAT_XFERCAP_32GB 0 + +#define IOAT_GENCTRL_OFFSET 0x02 /* 8-bit */ +#define IOAT_GENCTRL_DEBUG_EN 0x01 + +#define IOAT_INTRCTRL_OFFSET 0x03 /* 8-bit */ +#define IOAT_INTRCTRL_MASTER_INT_EN 0x01 /* Master Interrupt Enable */ +#define IOAT_INTRCTRL_INT_STATUS 0x02 /* ATTNSTATUS -or- Channel Int */ +#define IOAT_INTRCTRL_INT 0x04 /* INT_STATUS -and- MASTER_INT_EN */ +#define IOAT_INTRCTRL_MSIX_VECTOR_CONTROL 0x08 /* Enable all MSI-X vectors */ + +#define IOAT_ATTNSTATUS_OFFSET 0x04 /* Each bit is a channel */ + +#define IOAT_VER_OFFSET 0x08 /* 8-bit */ +#define IOAT_VER_MAJOR_MASK 0xF0 +#define IOAT_VER_MINOR_MASK 0x0F +#define GET_IOAT_VER_MAJOR(x) (((x) & IOAT_VER_MAJOR_MASK) >> 4) +#define GET_IOAT_VER_MINOR(x) ((x) & IOAT_VER_MINOR_MASK) + +#define IOAT_PERPORTOFFSET_OFFSET 0x0A /* 16-bit */ + +#define IOAT_INTRDELAY_OFFSET 0x0C /* 16-bit */ +#define IOAT_INTRDELAY_INT_DELAY_MASK 0x3FFF /* Interrupt Delay Time */ +#define IOAT_INTRDELAY_COALESE_SUPPORT 0x8000 /* Interrupt Coalescing Supported */ + +#define IOAT_DEVICE_STATUS_OFFSET 0x0E /* 16-bit */ +#define IOAT_DEVICE_STATUS_DEGRADED_MODE 0x0001 +#define IOAT_DEVICE_MMIO_RESTRICTED 0x0002 +#define IOAT_DEVICE_MEMORY_BYPASS 0x0004 +#define IOAT_DEVICE_ADDRESS_REMAPPING 0x0008 + +#define IOAT_DMA_CAP_OFFSET 0x10 /* 32-bit */ +#define IOAT_CAP_PAGE_BREAK 0x00000001 +#define IOAT_CAP_CRC 0x00000002 +#define IOAT_CAP_SKIP_MARKER 0x00000004 +#define IOAT_CAP_DCA 0x00000010 +#define IOAT_CAP_CRC_MOVE 0x00000020 +#define IOAT_CAP_FILL_BLOCK 0x00000040 +#define IOAT_CAP_APIC 0x00000080 +#define IOAT_CAP_XOR 0x00000100 +#define IOAT_CAP_PQ 0x00000200 + +#define IOAT_CHANNEL_MMIO_SIZE 0x80 /* Each Channel MMIO space is this size */ + +/* DMA Channel Registers */ +#define IOAT_CHANCTRL_OFFSET 0x00 /* 16-bit Channel Control Register */ +#define IOAT_CHANCTRL_CHANNEL_PRIORITY_MASK 0xF000 +#define IOAT3_CHANCTRL_COMPL_DCA_EN 0x0200 +#define IOAT_CHANCTRL_CHANNEL_IN_USE 0x0100 +#define IOAT_CHANCTRL_DESCRIPTOR_ADDR_SNOOP_CONTROL 0x0020 +#define IOAT_CHANCTRL_ERR_INT_EN 0x0010 +#define IOAT_CHANCTRL_ANY_ERR_ABORT_EN 0x0008 +#define IOAT_CHANCTRL_ERR_COMPLETION_EN 0x0004 +#define IOAT_CHANCTRL_INT_REARM 0x0001 +#define IOAT_CHANCTRL_RUN (IOAT_CHANCTRL_INT_REARM |\ + IOAT_CHANCTRL_ANY_ERR_ABORT_EN) + +#define IOAT_DMA_COMP_OFFSET 0x02 /* 16-bit DMA channel compatibility */ +#define IOAT_DMA_COMP_V1 0x0001 /* Compatibility with DMA version 1 */ +#define IOAT_DMA_COMP_V2 0x0002 /* Compatibility with DMA version 2 */ + + +#define IOAT1_CHANSTS_OFFSET 0x04 /* 64-bit Channel Status Register */ +#define IOAT2_CHANSTS_OFFSET 0x08 /* 64-bit Channel Status Register */ +#define IOAT_CHANSTS_OFFSET(ver) ((ver) < IOAT_VER_2_0 \ + ? IOAT1_CHANSTS_OFFSET : IOAT2_CHANSTS_OFFSET) +#define IOAT1_CHANSTS_OFFSET_LOW 0x04 +#define IOAT2_CHANSTS_OFFSET_LOW 0x08 +#define IOAT_CHANSTS_OFFSET_LOW(ver) ((ver) < IOAT_VER_2_0 \ + ? IOAT1_CHANSTS_OFFSET_LOW : IOAT2_CHANSTS_OFFSET_LOW) +#define IOAT1_CHANSTS_OFFSET_HIGH 0x08 +#define IOAT2_CHANSTS_OFFSET_HIGH 0x0C +#define IOAT_CHANSTS_OFFSET_HIGH(ver) ((ver) < IOAT_VER_2_0 \ + ? IOAT1_CHANSTS_OFFSET_HIGH : IOAT2_CHANSTS_OFFSET_HIGH) +#define IOAT_CHANSTS_COMPLETED_DESCRIPTOR_ADDR (~0x3fULL) +#define IOAT_CHANSTS_SOFT_ERR 0x10ULL +#define IOAT_CHANSTS_UNAFFILIATED_ERR 0x8ULL +#define IOAT_CHANSTS_STATUS 0x7ULL +#define IOAT_CHANSTS_ACTIVE 0x0 +#define IOAT_CHANSTS_DONE 0x1 +#define IOAT_CHANSTS_SUSPENDED 0x2 +#define IOAT_CHANSTS_HALTED 0x3 + + + +#define IOAT_CHAN_DMACOUNT_OFFSET 0x06 /* 16-bit DMA Count register */ + +#define IOAT_DCACTRL_OFFSET 0x30 /* 32 bit Direct Cache Access Control Register */ +#define IOAT_DCACTRL_CMPL_WRITE_ENABLE 0x10000 +#define IOAT_DCACTRL_TARGET_CPU_MASK 0xFFFF /* APIC ID */ + +/* CB DCA Memory Space Registers */ +#define IOAT_DCAOFFSET_OFFSET 0x14 +/* CB_BAR + IOAT_DCAOFFSET value */ +#define IOAT_DCA_VER_OFFSET 0x00 +#define IOAT_DCA_VER_MAJOR_MASK 0xF0 +#define IOAT_DCA_VER_MINOR_MASK 0x0F + +#define IOAT_DCA_COMP_OFFSET 0x02 +#define IOAT_DCA_COMP_V1 0x1 + +#define IOAT_FSB_CAPABILITY_OFFSET 0x04 +#define IOAT_FSB_CAPABILITY_PREFETCH 0x1 + +#define IOAT_PCI_CAPABILITY_OFFSET 0x06 +#define IOAT_PCI_CAPABILITY_MEMWR 0x1 + +#define IOAT_FSB_CAP_ENABLE_OFFSET 0x08 +#define IOAT_FSB_CAP_ENABLE_PREFETCH 0x1 + +#define IOAT_PCI_CAP_ENABLE_OFFSET 0x0A +#define IOAT_PCI_CAP_ENABLE_MEMWR 0x1 + +#define IOAT_APICID_TAG_MAP_OFFSET 0x0C +#define IOAT_APICID_TAG_MAP_TAG0 0x0000000F +#define IOAT_APICID_TAG_MAP_TAG0_SHIFT 0 +#define IOAT_APICID_TAG_MAP_TAG1 0x000000F0 +#define IOAT_APICID_TAG_MAP_TAG1_SHIFT 4 +#define IOAT_APICID_TAG_MAP_TAG2 0x00000F00 +#define IOAT_APICID_TAG_MAP_TAG2_SHIFT 8 +#define IOAT_APICID_TAG_MAP_TAG3 0x0000F000 +#define IOAT_APICID_TAG_MAP_TAG3_SHIFT 12 +#define IOAT_APICID_TAG_MAP_TAG4 0x000F0000 +#define IOAT_APICID_TAG_MAP_TAG4_SHIFT 16 +#define IOAT_APICID_TAG_CB2_VALID 0x8080808080 + +#define IOAT_DCA_GREQID_OFFSET 0x10 +#define IOAT_DCA_GREQID_SIZE 0x04 +#define IOAT_DCA_GREQID_MASK 0xFFFF +#define IOAT_DCA_GREQID_IGNOREFUN 0x10000000 +#define IOAT_DCA_GREQID_VALID 0x20000000 +#define IOAT_DCA_GREQID_LASTID 0x80000000 + +#define IOAT3_CSI_CAPABILITY_OFFSET 0x08 +#define IOAT3_CSI_CAPABILITY_PREFETCH 0x1 + +#define IOAT3_PCI_CAPABILITY_OFFSET 0x0A +#define IOAT3_PCI_CAPABILITY_MEMWR 0x1 + +#define IOAT3_CSI_CONTROL_OFFSET 0x0C +#define IOAT3_CSI_CONTROL_PREFETCH 0x1 + +#define IOAT3_PCI_CONTROL_OFFSET 0x0E +#define IOAT3_PCI_CONTROL_MEMWR 0x1 + +#define IOAT3_APICID_TAG_MAP_OFFSET 0x10 +#define IOAT3_APICID_TAG_MAP_OFFSET_LOW 0x10 +#define IOAT3_APICID_TAG_MAP_OFFSET_HIGH 0x14 + +#define IOAT3_DCA_GREQID_OFFSET 0x02 + +#define IOAT1_CHAINADDR_OFFSET 0x0C /* 64-bit Descriptor Chain Address Register */ +#define IOAT2_CHAINADDR_OFFSET 0x10 /* 64-bit Descriptor Chain Address Register */ +#define IOAT_CHAINADDR_OFFSET(ver) ((ver) < IOAT_VER_2_0 \ + ? IOAT1_CHAINADDR_OFFSET : IOAT2_CHAINADDR_OFFSET) +#define IOAT1_CHAINADDR_OFFSET_LOW 0x0C +#define IOAT2_CHAINADDR_OFFSET_LOW 0x10 +#define IOAT_CHAINADDR_OFFSET_LOW(ver) ((ver) < IOAT_VER_2_0 \ + ? IOAT1_CHAINADDR_OFFSET_LOW : IOAT2_CHAINADDR_OFFSET_LOW) +#define IOAT1_CHAINADDR_OFFSET_HIGH 0x10 +#define IOAT2_CHAINADDR_OFFSET_HIGH 0x14 +#define IOAT_CHAINADDR_OFFSET_HIGH(ver) ((ver) < IOAT_VER_2_0 \ + ? IOAT1_CHAINADDR_OFFSET_HIGH : IOAT2_CHAINADDR_OFFSET_HIGH) + +#define IOAT1_CHANCMD_OFFSET 0x14 /* 8-bit DMA Channel Command Register */ +#define IOAT2_CHANCMD_OFFSET 0x04 /* 8-bit DMA Channel Command Register */ +#define IOAT_CHANCMD_OFFSET(ver) ((ver) < IOAT_VER_2_0 \ + ? IOAT1_CHANCMD_OFFSET : IOAT2_CHANCMD_OFFSET) +#define IOAT_CHANCMD_RESET 0x20 +#define IOAT_CHANCMD_RESUME 0x10 +#define IOAT_CHANCMD_ABORT 0x08 +#define IOAT_CHANCMD_SUSPEND 0x04 +#define IOAT_CHANCMD_APPEND 0x02 +#define IOAT_CHANCMD_START 0x01 + +#define IOAT_CHANCMP_OFFSET 0x18 /* 64-bit Channel Completion Address Register */ +#define IOAT_CHANCMP_OFFSET_LOW 0x18 +#define IOAT_CHANCMP_OFFSET_HIGH 0x1C + +#define IOAT_CDAR_OFFSET 0x20 /* 64-bit Current Descriptor Address Register */ +#define IOAT_CDAR_OFFSET_LOW 0x20 +#define IOAT_CDAR_OFFSET_HIGH 0x24 + +#define IOAT_CHANERR_OFFSET 0x28 /* 32-bit Channel Error Register */ +#define IOAT_CHANERR_SRC_ADDR_ERR 0x0001 +#define IOAT_CHANERR_DEST_ADDR_ERR 0x0002 +#define IOAT_CHANERR_NEXT_ADDR_ERR 0x0004 +#define IOAT_CHANERR_NEXT_DESC_ALIGN_ERR 0x0008 +#define IOAT_CHANERR_CHAIN_ADDR_VALUE_ERR 0x0010 +#define IOAT_CHANERR_CHANCMD_ERR 0x0020 +#define IOAT_CHANERR_CHIPSET_UNCORRECTABLE_DATA_INTEGRITY_ERR 0x0040 +#define IOAT_CHANERR_DMA_UNCORRECTABLE_DATA_INTEGRITY_ERR 0x0080 +#define IOAT_CHANERR_READ_DATA_ERR 0x0100 +#define IOAT_CHANERR_WRITE_DATA_ERR 0x0200 +#define IOAT_CHANERR_CONTROL_ERR 0x0400 +#define IOAT_CHANERR_LENGTH_ERR 0x0800 +#define IOAT_CHANERR_COMPLETION_ADDR_ERR 0x1000 +#define IOAT_CHANERR_INT_CONFIGURATION_ERR 0x2000 +#define IOAT_CHANERR_SOFT_ERR 0x4000 +#define IOAT_CHANERR_UNAFFILIATED_ERR 0x8000 +#define IOAT_CHANERR_XOR_P_OR_CRC_ERR 0x10000 +#define IOAT_CHANERR_XOR_Q_ERR 0x20000 +#define IOAT_CHANERR_DESCRIPTOR_COUNT_ERR 0x40000 + +#define IOAT_CHANERR_HANDLE_MASK (IOAT_CHANERR_XOR_P_OR_CRC_ERR | IOAT_CHANERR_XOR_Q_ERR) + +#define IOAT_CHANERR_MASK_OFFSET 0x2C /* 32-bit Channel Error Register */ + +#endif /* _IOAT_REGISTERS_H_ */ diff --git a/drivers/dma/iop-adma.c b/drivers/dma/iop-adma.c new file mode 100644 index 00000000000..ca6e6a0cb79 --- /dev/null +++ b/drivers/dma/iop-adma.c @@ -0,0 +1,1772 @@ +/* + * offload engine driver for the Intel Xscale series of i/o processors + * Copyright © 2006, Intel Corporation. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + * + */ + +/* + * This driver supports the asynchrounous DMA copy and RAID engines available + * on the Intel Xscale(R) family of I/O Processors (IOP 32x, 33x, 134x) + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/delay.h> +#include <linux/dma-mapping.h> +#include <linux/spinlock.h> +#include <linux/interrupt.h> +#include <linux/platform_device.h> +#include <linux/memory.h> +#include <linux/ioport.h> +#include <linux/raid/pq.h> + +#include <mach/adma.h> + +#define to_iop_adma_chan(chan) container_of(chan, struct iop_adma_chan, common) +#define to_iop_adma_device(dev) \ + container_of(dev, struct iop_adma_device, common) +#define tx_to_iop_adma_slot(tx) \ + container_of(tx, struct iop_adma_desc_slot, async_tx) + +/** + * iop_adma_free_slots - flags descriptor slots for reuse + * @slot: Slot to free + * Caller must hold &iop_chan->lock while calling this function + */ +static void iop_adma_free_slots(struct iop_adma_desc_slot *slot) +{ + int stride = slot->slots_per_op; + + while (stride--) { + slot->slots_per_op = 0; + slot = list_entry(slot->slot_node.next, + struct iop_adma_desc_slot, + slot_node); + } +} + +static void +iop_desc_unmap(struct iop_adma_chan *iop_chan, struct iop_adma_desc_slot *desc) +{ + struct dma_async_tx_descriptor *tx = &desc->async_tx; + struct iop_adma_desc_slot *unmap = desc->group_head; + struct device *dev = &iop_chan->device->pdev->dev; + u32 len = unmap->unmap_len; + enum dma_ctrl_flags flags = tx->flags; + u32 src_cnt; + dma_addr_t addr; + dma_addr_t dest; + + src_cnt = unmap->unmap_src_cnt; + dest = iop_desc_get_dest_addr(unmap, iop_chan); + if (!(flags & DMA_COMPL_SKIP_DEST_UNMAP)) { + enum dma_data_direction dir; + + if (src_cnt > 1) /* is xor? */ + dir = DMA_BIDIRECTIONAL; + else + dir = DMA_FROM_DEVICE; + + dma_unmap_page(dev, dest, len, dir); + } + + if (!(flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + while (src_cnt--) { + addr = iop_desc_get_src_addr(unmap, iop_chan, src_cnt); + if (addr == dest) + continue; + dma_unmap_page(dev, addr, len, DMA_TO_DEVICE); + } + } + desc->group_head = NULL; +} + +static void +iop_desc_unmap_pq(struct iop_adma_chan *iop_chan, struct iop_adma_desc_slot *desc) +{ + struct dma_async_tx_descriptor *tx = &desc->async_tx; + struct iop_adma_desc_slot *unmap = desc->group_head; + struct device *dev = &iop_chan->device->pdev->dev; + u32 len = unmap->unmap_len; + enum dma_ctrl_flags flags = tx->flags; + u32 src_cnt = unmap->unmap_src_cnt; + dma_addr_t pdest = iop_desc_get_dest_addr(unmap, iop_chan); + dma_addr_t qdest = iop_desc_get_qdest_addr(unmap, iop_chan); + int i; + + if (tx->flags & DMA_PREP_CONTINUE) + src_cnt -= 3; + + if (!(flags & DMA_COMPL_SKIP_DEST_UNMAP) && !desc->pq_check_result) { + dma_unmap_page(dev, pdest, len, DMA_BIDIRECTIONAL); + dma_unmap_page(dev, qdest, len, DMA_BIDIRECTIONAL); + } + + if (!(flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + dma_addr_t addr; + + for (i = 0; i < src_cnt; i++) { + addr = iop_desc_get_src_addr(unmap, iop_chan, i); + dma_unmap_page(dev, addr, len, DMA_TO_DEVICE); + } + if (desc->pq_check_result) { + dma_unmap_page(dev, pdest, len, DMA_TO_DEVICE); + dma_unmap_page(dev, qdest, len, DMA_TO_DEVICE); + } + } + + desc->group_head = NULL; +} + + +static dma_cookie_t +iop_adma_run_tx_complete_actions(struct iop_adma_desc_slot *desc, + struct iop_adma_chan *iop_chan, dma_cookie_t cookie) +{ + struct dma_async_tx_descriptor *tx = &desc->async_tx; + + BUG_ON(tx->cookie < 0); + if (tx->cookie > 0) { + cookie = tx->cookie; + tx->cookie = 0; + + /* call the callback (must not sleep or submit new + * operations to this channel) + */ + if (tx->callback) + tx->callback(tx->callback_param); + + /* unmap dma addresses + * (unmap_single vs unmap_page?) + */ + if (desc->group_head && desc->unmap_len) { + if (iop_desc_is_pq(desc)) + iop_desc_unmap_pq(iop_chan, desc); + else + iop_desc_unmap(iop_chan, desc); + } + } + + /* run dependent operations */ + dma_run_dependencies(tx); + + return cookie; +} + +static int +iop_adma_clean_slot(struct iop_adma_desc_slot *desc, + struct iop_adma_chan *iop_chan) +{ + /* the client is allowed to attach dependent operations + * until 'ack' is set + */ + if (!async_tx_test_ack(&desc->async_tx)) + return 0; + + /* leave the last descriptor in the chain + * so we can append to it + */ + if (desc->chain_node.next == &iop_chan->chain) + return 1; + + dev_dbg(iop_chan->device->common.dev, + "\tfree slot: %d slots_per_op: %d\n", + desc->idx, desc->slots_per_op); + + list_del(&desc->chain_node); + iop_adma_free_slots(desc); + + return 0; +} + +static void __iop_adma_slot_cleanup(struct iop_adma_chan *iop_chan) +{ + struct iop_adma_desc_slot *iter, *_iter, *grp_start = NULL; + dma_cookie_t cookie = 0; + u32 current_desc = iop_chan_get_current_descriptor(iop_chan); + int busy = iop_chan_is_busy(iop_chan); + int seen_current = 0, slot_cnt = 0, slots_per_op = 0; + + dev_dbg(iop_chan->device->common.dev, "%s\n", __func__); + /* free completed slots from the chain starting with + * the oldest descriptor + */ + list_for_each_entry_safe(iter, _iter, &iop_chan->chain, + chain_node) { + pr_debug("\tcookie: %d slot: %d busy: %d " + "this_desc: %#x next_desc: %#x ack: %d\n", + iter->async_tx.cookie, iter->idx, busy, + iter->async_tx.phys, iop_desc_get_next_desc(iter), + async_tx_test_ack(&iter->async_tx)); + prefetch(_iter); + prefetch(&_iter->async_tx); + + /* do not advance past the current descriptor loaded into the + * hardware channel, subsequent descriptors are either in + * process or have not been submitted + */ + if (seen_current) + break; + + /* stop the search if we reach the current descriptor and the + * channel is busy, or if it appears that the current descriptor + * needs to be re-read (i.e. has been appended to) + */ + if (iter->async_tx.phys == current_desc) { + BUG_ON(seen_current++); + if (busy || iop_desc_get_next_desc(iter)) + break; + } + + /* detect the start of a group transaction */ + if (!slot_cnt && !slots_per_op) { + slot_cnt = iter->slot_cnt; + slots_per_op = iter->slots_per_op; + if (slot_cnt <= slots_per_op) { + slot_cnt = 0; + slots_per_op = 0; + } + } + + if (slot_cnt) { + pr_debug("\tgroup++\n"); + if (!grp_start) + grp_start = iter; + slot_cnt -= slots_per_op; + } + + /* all the members of a group are complete */ + if (slots_per_op != 0 && slot_cnt == 0) { + struct iop_adma_desc_slot *grp_iter, *_grp_iter; + int end_of_chain = 0; + pr_debug("\tgroup end\n"); + + /* collect the total results */ + if (grp_start->xor_check_result) { + u32 zero_sum_result = 0; + slot_cnt = grp_start->slot_cnt; + grp_iter = grp_start; + + list_for_each_entry_from(grp_iter, + &iop_chan->chain, chain_node) { + zero_sum_result |= + iop_desc_get_zero_result(grp_iter); + pr_debug("\titer%d result: %d\n", + grp_iter->idx, zero_sum_result); + slot_cnt -= slots_per_op; + if (slot_cnt == 0) + break; + } + pr_debug("\tgrp_start->xor_check_result: %p\n", + grp_start->xor_check_result); + *grp_start->xor_check_result = zero_sum_result; + } + + /* clean up the group */ + slot_cnt = grp_start->slot_cnt; + grp_iter = grp_start; + list_for_each_entry_safe_from(grp_iter, _grp_iter, + &iop_chan->chain, chain_node) { + cookie = iop_adma_run_tx_complete_actions( + grp_iter, iop_chan, cookie); + + slot_cnt -= slots_per_op; + end_of_chain = iop_adma_clean_slot(grp_iter, + iop_chan); + + if (slot_cnt == 0 || end_of_chain) + break; + } + + /* the group should be complete at this point */ + BUG_ON(slot_cnt); + + slots_per_op = 0; + grp_start = NULL; + if (end_of_chain) + break; + else + continue; + } else if (slots_per_op) /* wait for group completion */ + continue; + + /* write back zero sum results (single descriptor case) */ + if (iter->xor_check_result && iter->async_tx.cookie) + *iter->xor_check_result = + iop_desc_get_zero_result(iter); + + cookie = iop_adma_run_tx_complete_actions( + iter, iop_chan, cookie); + + if (iop_adma_clean_slot(iter, iop_chan)) + break; + } + + if (cookie > 0) { + iop_chan->completed_cookie = cookie; + pr_debug("\tcompleted cookie %d\n", cookie); + } +} + +static void +iop_adma_slot_cleanup(struct iop_adma_chan *iop_chan) +{ + spin_lock_bh(&iop_chan->lock); + __iop_adma_slot_cleanup(iop_chan); + spin_unlock_bh(&iop_chan->lock); +} + +static void iop_adma_tasklet(unsigned long data) +{ + struct iop_adma_chan *iop_chan = (struct iop_adma_chan *) data; + + /* lockdep will flag depedency submissions as potentially + * recursive locking, this is not the case as a dependency + * submission will never recurse a channels submit routine. + * There are checks in async_tx.c to prevent this. + */ + spin_lock_nested(&iop_chan->lock, SINGLE_DEPTH_NESTING); + __iop_adma_slot_cleanup(iop_chan); + spin_unlock(&iop_chan->lock); +} + +static struct iop_adma_desc_slot * +iop_adma_alloc_slots(struct iop_adma_chan *iop_chan, int num_slots, + int slots_per_op) +{ + struct iop_adma_desc_slot *iter, *_iter, *alloc_start = NULL; + LIST_HEAD(chain); + int slots_found, retry = 0; + + /* start search from the last allocated descrtiptor + * if a contiguous allocation can not be found start searching + * from the beginning of the list + */ +retry: + slots_found = 0; + if (retry == 0) + iter = iop_chan->last_used; + else + iter = list_entry(&iop_chan->all_slots, + struct iop_adma_desc_slot, + slot_node); + + list_for_each_entry_safe_continue( + iter, _iter, &iop_chan->all_slots, slot_node) { + prefetch(_iter); + prefetch(&_iter->async_tx); + if (iter->slots_per_op) { + /* give up after finding the first busy slot + * on the second pass through the list + */ + if (retry) + break; + + slots_found = 0; + continue; + } + + /* start the allocation if the slot is correctly aligned */ + if (!slots_found++) { + if (iop_desc_is_aligned(iter, slots_per_op)) + alloc_start = iter; + else { + slots_found = 0; + continue; + } + } + + if (slots_found == num_slots) { + struct iop_adma_desc_slot *alloc_tail = NULL; + struct iop_adma_desc_slot *last_used = NULL; + iter = alloc_start; + while (num_slots) { + int i; + dev_dbg(iop_chan->device->common.dev, + "allocated slot: %d " + "(desc %p phys: %#x) slots_per_op %d\n", + iter->idx, iter->hw_desc, + iter->async_tx.phys, slots_per_op); + + /* pre-ack all but the last descriptor */ + if (num_slots != slots_per_op) + async_tx_ack(&iter->async_tx); + + list_add_tail(&iter->chain_node, &chain); + alloc_tail = iter; + iter->async_tx.cookie = 0; + iter->slot_cnt = num_slots; + iter->xor_check_result = NULL; + for (i = 0; i < slots_per_op; i++) { + iter->slots_per_op = slots_per_op - i; + last_used = iter; + iter = list_entry(iter->slot_node.next, + struct iop_adma_desc_slot, + slot_node); + } + num_slots -= slots_per_op; + } + alloc_tail->group_head = alloc_start; + alloc_tail->async_tx.cookie = -EBUSY; + list_splice(&chain, &alloc_tail->tx_list); + iop_chan->last_used = last_used; + iop_desc_clear_next_desc(alloc_start); + iop_desc_clear_next_desc(alloc_tail); + return alloc_tail; + } + } + if (!retry++) + goto retry; + + /* perform direct reclaim if the allocation fails */ + __iop_adma_slot_cleanup(iop_chan); + + return NULL; +} + +static dma_cookie_t +iop_desc_assign_cookie(struct iop_adma_chan *iop_chan, + struct iop_adma_desc_slot *desc) +{ + dma_cookie_t cookie = iop_chan->common.cookie; + cookie++; + if (cookie < 0) + cookie = 1; + iop_chan->common.cookie = desc->async_tx.cookie = cookie; + return cookie; +} + +static void iop_adma_check_threshold(struct iop_adma_chan *iop_chan) +{ + dev_dbg(iop_chan->device->common.dev, "pending: %d\n", + iop_chan->pending); + + if (iop_chan->pending >= IOP_ADMA_THRESHOLD) { + iop_chan->pending = 0; + iop_chan_append(iop_chan); + } +} + +static dma_cookie_t +iop_adma_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct iop_adma_desc_slot *sw_desc = tx_to_iop_adma_slot(tx); + struct iop_adma_chan *iop_chan = to_iop_adma_chan(tx->chan); + struct iop_adma_desc_slot *grp_start, *old_chain_tail; + int slot_cnt; + int slots_per_op; + dma_cookie_t cookie; + dma_addr_t next_dma; + + grp_start = sw_desc->group_head; + slot_cnt = grp_start->slot_cnt; + slots_per_op = grp_start->slots_per_op; + + spin_lock_bh(&iop_chan->lock); + cookie = iop_desc_assign_cookie(iop_chan, sw_desc); + + old_chain_tail = list_entry(iop_chan->chain.prev, + struct iop_adma_desc_slot, chain_node); + list_splice_init(&sw_desc->tx_list, + &old_chain_tail->chain_node); + + /* fix up the hardware chain */ + next_dma = grp_start->async_tx.phys; + iop_desc_set_next_desc(old_chain_tail, next_dma); + BUG_ON(iop_desc_get_next_desc(old_chain_tail) != next_dma); /* flush */ + + /* check for pre-chained descriptors */ + iop_paranoia(iop_desc_get_next_desc(sw_desc)); + + /* increment the pending count by the number of slots + * memcpy operations have a 1:1 (slot:operation) relation + * other operations are heavier and will pop the threshold + * more often. + */ + iop_chan->pending += slot_cnt; + iop_adma_check_threshold(iop_chan); + spin_unlock_bh(&iop_chan->lock); + + dev_dbg(iop_chan->device->common.dev, "%s cookie: %d slot: %d\n", + __func__, sw_desc->async_tx.cookie, sw_desc->idx); + + return cookie; +} + +static void iop_chan_start_null_memcpy(struct iop_adma_chan *iop_chan); +static void iop_chan_start_null_xor(struct iop_adma_chan *iop_chan); + +/** + * iop_adma_alloc_chan_resources - returns the number of allocated descriptors + * @chan - allocate descriptor resources for this channel + * @client - current client requesting the channel be ready for requests + * + * Note: We keep the slots for 1 operation on iop_chan->chain at all times. To + * avoid deadlock, via async_xor, num_descs_in_pool must at a minimum be + * greater than 2x the number slots needed to satisfy a device->max_xor + * request. + * */ +static int iop_adma_alloc_chan_resources(struct dma_chan *chan) +{ + char *hw_desc; + int idx; + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + struct iop_adma_desc_slot *slot = NULL; + int init = iop_chan->slots_allocated ? 0 : 1; + struct iop_adma_platform_data *plat_data = + iop_chan->device->pdev->dev.platform_data; + int num_descs_in_pool = plat_data->pool_size/IOP_ADMA_SLOT_SIZE; + + /* Allocate descriptor slots */ + do { + idx = iop_chan->slots_allocated; + if (idx == num_descs_in_pool) + break; + + slot = kzalloc(sizeof(*slot), GFP_KERNEL); + if (!slot) { + printk(KERN_INFO "IOP ADMA Channel only initialized" + " %d descriptor slots", idx); + break; + } + hw_desc = (char *) iop_chan->device->dma_desc_pool_virt; + slot->hw_desc = (void *) &hw_desc[idx * IOP_ADMA_SLOT_SIZE]; + + dma_async_tx_descriptor_init(&slot->async_tx, chan); + slot->async_tx.tx_submit = iop_adma_tx_submit; + INIT_LIST_HEAD(&slot->tx_list); + INIT_LIST_HEAD(&slot->chain_node); + INIT_LIST_HEAD(&slot->slot_node); + hw_desc = (char *) iop_chan->device->dma_desc_pool; + slot->async_tx.phys = + (dma_addr_t) &hw_desc[idx * IOP_ADMA_SLOT_SIZE]; + slot->idx = idx; + + spin_lock_bh(&iop_chan->lock); + iop_chan->slots_allocated++; + list_add_tail(&slot->slot_node, &iop_chan->all_slots); + spin_unlock_bh(&iop_chan->lock); + } while (iop_chan->slots_allocated < num_descs_in_pool); + + if (idx && !iop_chan->last_used) + iop_chan->last_used = list_entry(iop_chan->all_slots.next, + struct iop_adma_desc_slot, + slot_node); + + dev_dbg(iop_chan->device->common.dev, + "allocated %d descriptor slots last_used: %p\n", + iop_chan->slots_allocated, iop_chan->last_used); + + /* initialize the channel and the chain with a null operation */ + if (init) { + if (dma_has_cap(DMA_MEMCPY, + iop_chan->device->common.cap_mask)) + iop_chan_start_null_memcpy(iop_chan); + else if (dma_has_cap(DMA_XOR, + iop_chan->device->common.cap_mask)) + iop_chan_start_null_xor(iop_chan); + else + BUG(); + } + + return (idx > 0) ? idx : -ENOMEM; +} + +static struct dma_async_tx_descriptor * +iop_adma_prep_dma_interrupt(struct dma_chan *chan, unsigned long flags) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + struct iop_adma_desc_slot *sw_desc, *grp_start; + int slot_cnt, slots_per_op; + + dev_dbg(iop_chan->device->common.dev, "%s\n", __func__); + + spin_lock_bh(&iop_chan->lock); + slot_cnt = iop_chan_interrupt_slot_count(&slots_per_op, iop_chan); + sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op); + if (sw_desc) { + grp_start = sw_desc->group_head; + iop_desc_init_interrupt(grp_start, iop_chan); + grp_start->unmap_len = 0; + sw_desc->async_tx.flags = flags; + } + spin_unlock_bh(&iop_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static struct dma_async_tx_descriptor * +iop_adma_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dma_dest, + dma_addr_t dma_src, size_t len, unsigned long flags) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + struct iop_adma_desc_slot *sw_desc, *grp_start; + int slot_cnt, slots_per_op; + + if (unlikely(!len)) + return NULL; + BUG_ON(unlikely(len > IOP_ADMA_MAX_BYTE_COUNT)); + + dev_dbg(iop_chan->device->common.dev, "%s len: %u\n", + __func__, len); + + spin_lock_bh(&iop_chan->lock); + slot_cnt = iop_chan_memcpy_slot_count(len, &slots_per_op); + sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op); + if (sw_desc) { + grp_start = sw_desc->group_head; + iop_desc_init_memcpy(grp_start, flags); + iop_desc_set_byte_count(grp_start, iop_chan, len); + iop_desc_set_dest_addr(grp_start, iop_chan, dma_dest); + iop_desc_set_memcpy_src_addr(grp_start, dma_src); + sw_desc->unmap_src_cnt = 1; + sw_desc->unmap_len = len; + sw_desc->async_tx.flags = flags; + } + spin_unlock_bh(&iop_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static struct dma_async_tx_descriptor * +iop_adma_prep_dma_memset(struct dma_chan *chan, dma_addr_t dma_dest, + int value, size_t len, unsigned long flags) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + struct iop_adma_desc_slot *sw_desc, *grp_start; + int slot_cnt, slots_per_op; + + if (unlikely(!len)) + return NULL; + BUG_ON(unlikely(len > IOP_ADMA_MAX_BYTE_COUNT)); + + dev_dbg(iop_chan->device->common.dev, "%s len: %u\n", + __func__, len); + + spin_lock_bh(&iop_chan->lock); + slot_cnt = iop_chan_memset_slot_count(len, &slots_per_op); + sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op); + if (sw_desc) { + grp_start = sw_desc->group_head; + iop_desc_init_memset(grp_start, flags); + iop_desc_set_byte_count(grp_start, iop_chan, len); + iop_desc_set_block_fill_val(grp_start, value); + iop_desc_set_dest_addr(grp_start, iop_chan, dma_dest); + sw_desc->unmap_src_cnt = 1; + sw_desc->unmap_len = len; + sw_desc->async_tx.flags = flags; + } + spin_unlock_bh(&iop_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static struct dma_async_tx_descriptor * +iop_adma_prep_dma_xor(struct dma_chan *chan, dma_addr_t dma_dest, + dma_addr_t *dma_src, unsigned int src_cnt, size_t len, + unsigned long flags) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + struct iop_adma_desc_slot *sw_desc, *grp_start; + int slot_cnt, slots_per_op; + + if (unlikely(!len)) + return NULL; + BUG_ON(unlikely(len > IOP_ADMA_XOR_MAX_BYTE_COUNT)); + + dev_dbg(iop_chan->device->common.dev, + "%s src_cnt: %d len: %u flags: %lx\n", + __func__, src_cnt, len, flags); + + spin_lock_bh(&iop_chan->lock); + slot_cnt = iop_chan_xor_slot_count(len, src_cnt, &slots_per_op); + sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op); + if (sw_desc) { + grp_start = sw_desc->group_head; + iop_desc_init_xor(grp_start, src_cnt, flags); + iop_desc_set_byte_count(grp_start, iop_chan, len); + iop_desc_set_dest_addr(grp_start, iop_chan, dma_dest); + sw_desc->unmap_src_cnt = src_cnt; + sw_desc->unmap_len = len; + sw_desc->async_tx.flags = flags; + while (src_cnt--) + iop_desc_set_xor_src_addr(grp_start, src_cnt, + dma_src[src_cnt]); + } + spin_unlock_bh(&iop_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static struct dma_async_tx_descriptor * +iop_adma_prep_dma_xor_val(struct dma_chan *chan, dma_addr_t *dma_src, + unsigned int src_cnt, size_t len, u32 *result, + unsigned long flags) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + struct iop_adma_desc_slot *sw_desc, *grp_start; + int slot_cnt, slots_per_op; + + if (unlikely(!len)) + return NULL; + + dev_dbg(iop_chan->device->common.dev, "%s src_cnt: %d len: %u\n", + __func__, src_cnt, len); + + spin_lock_bh(&iop_chan->lock); + slot_cnt = iop_chan_zero_sum_slot_count(len, src_cnt, &slots_per_op); + sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op); + if (sw_desc) { + grp_start = sw_desc->group_head; + iop_desc_init_zero_sum(grp_start, src_cnt, flags); + iop_desc_set_zero_sum_byte_count(grp_start, len); + grp_start->xor_check_result = result; + pr_debug("\t%s: grp_start->xor_check_result: %p\n", + __func__, grp_start->xor_check_result); + sw_desc->unmap_src_cnt = src_cnt; + sw_desc->unmap_len = len; + sw_desc->async_tx.flags = flags; + while (src_cnt--) + iop_desc_set_zero_sum_src_addr(grp_start, src_cnt, + dma_src[src_cnt]); + } + spin_unlock_bh(&iop_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static struct dma_async_tx_descriptor * +iop_adma_prep_dma_pq(struct dma_chan *chan, dma_addr_t *dst, dma_addr_t *src, + unsigned int src_cnt, const unsigned char *scf, size_t len, + unsigned long flags) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + struct iop_adma_desc_slot *sw_desc, *g; + int slot_cnt, slots_per_op; + int continue_srcs; + + if (unlikely(!len)) + return NULL; + BUG_ON(len > IOP_ADMA_XOR_MAX_BYTE_COUNT); + + dev_dbg(iop_chan->device->common.dev, + "%s src_cnt: %d len: %u flags: %lx\n", + __func__, src_cnt, len, flags); + + if (dmaf_p_disabled_continue(flags)) + continue_srcs = 1+src_cnt; + else if (dmaf_continue(flags)) + continue_srcs = 3+src_cnt; + else + continue_srcs = 0+src_cnt; + + spin_lock_bh(&iop_chan->lock); + slot_cnt = iop_chan_pq_slot_count(len, continue_srcs, &slots_per_op); + sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op); + if (sw_desc) { + int i; + + g = sw_desc->group_head; + iop_desc_set_byte_count(g, iop_chan, len); + + /* even if P is disabled its destination address (bits + * [3:0]) must match Q. It is ok if P points to an + * invalid address, it won't be written. + */ + if (flags & DMA_PREP_PQ_DISABLE_P) + dst[0] = dst[1] & 0x7; + + iop_desc_set_pq_addr(g, dst); + sw_desc->unmap_src_cnt = src_cnt; + sw_desc->unmap_len = len; + sw_desc->async_tx.flags = flags; + for (i = 0; i < src_cnt; i++) + iop_desc_set_pq_src_addr(g, i, src[i], scf[i]); + + /* if we are continuing a previous operation factor in + * the old p and q values, see the comment for dma_maxpq + * in include/linux/dmaengine.h + */ + if (dmaf_p_disabled_continue(flags)) + iop_desc_set_pq_src_addr(g, i++, dst[1], 1); + else if (dmaf_continue(flags)) { + iop_desc_set_pq_src_addr(g, i++, dst[0], 0); + iop_desc_set_pq_src_addr(g, i++, dst[1], 1); + iop_desc_set_pq_src_addr(g, i++, dst[1], 0); + } + iop_desc_init_pq(g, i, flags); + } + spin_unlock_bh(&iop_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static struct dma_async_tx_descriptor * +iop_adma_prep_dma_pq_val(struct dma_chan *chan, dma_addr_t *pq, dma_addr_t *src, + unsigned int src_cnt, const unsigned char *scf, + size_t len, enum sum_check_flags *pqres, + unsigned long flags) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + struct iop_adma_desc_slot *sw_desc, *g; + int slot_cnt, slots_per_op; + + if (unlikely(!len)) + return NULL; + BUG_ON(len > IOP_ADMA_XOR_MAX_BYTE_COUNT); + + dev_dbg(iop_chan->device->common.dev, "%s src_cnt: %d len: %u\n", + __func__, src_cnt, len); + + spin_lock_bh(&iop_chan->lock); + slot_cnt = iop_chan_pq_zero_sum_slot_count(len, src_cnt + 2, &slots_per_op); + sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op); + if (sw_desc) { + /* for validate operations p and q are tagged onto the + * end of the source list + */ + int pq_idx = src_cnt; + + g = sw_desc->group_head; + iop_desc_init_pq_zero_sum(g, src_cnt+2, flags); + iop_desc_set_pq_zero_sum_byte_count(g, len); + g->pq_check_result = pqres; + pr_debug("\t%s: g->pq_check_result: %p\n", + __func__, g->pq_check_result); + sw_desc->unmap_src_cnt = src_cnt+2; + sw_desc->unmap_len = len; + sw_desc->async_tx.flags = flags; + while (src_cnt--) + iop_desc_set_pq_zero_sum_src_addr(g, src_cnt, + src[src_cnt], + scf[src_cnt]); + iop_desc_set_pq_zero_sum_addr(g, pq_idx, src); + } + spin_unlock_bh(&iop_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static void iop_adma_free_chan_resources(struct dma_chan *chan) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + struct iop_adma_desc_slot *iter, *_iter; + int in_use_descs = 0; + + iop_adma_slot_cleanup(iop_chan); + + spin_lock_bh(&iop_chan->lock); + list_for_each_entry_safe(iter, _iter, &iop_chan->chain, + chain_node) { + in_use_descs++; + list_del(&iter->chain_node); + } + list_for_each_entry_safe_reverse( + iter, _iter, &iop_chan->all_slots, slot_node) { + list_del(&iter->slot_node); + kfree(iter); + iop_chan->slots_allocated--; + } + iop_chan->last_used = NULL; + + dev_dbg(iop_chan->device->common.dev, "%s slots_allocated %d\n", + __func__, iop_chan->slots_allocated); + spin_unlock_bh(&iop_chan->lock); + + /* one is ok since we left it on there on purpose */ + if (in_use_descs > 1) + printk(KERN_ERR "IOP: Freeing %d in use descriptors!\n", + in_use_descs - 1); +} + +/** + * iop_adma_is_complete - poll the status of an ADMA transaction + * @chan: ADMA channel handle + * @cookie: ADMA transaction identifier + */ +static enum dma_status iop_adma_is_complete(struct dma_chan *chan, + dma_cookie_t cookie, + dma_cookie_t *done, + dma_cookie_t *used) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + dma_cookie_t last_used; + dma_cookie_t last_complete; + enum dma_status ret; + + last_used = chan->cookie; + last_complete = iop_chan->completed_cookie; + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + if (ret == DMA_SUCCESS) + return ret; + + iop_adma_slot_cleanup(iop_chan); + + last_used = chan->cookie; + last_complete = iop_chan->completed_cookie; + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + return dma_async_is_complete(cookie, last_complete, last_used); +} + +static irqreturn_t iop_adma_eot_handler(int irq, void *data) +{ + struct iop_adma_chan *chan = data; + + dev_dbg(chan->device->common.dev, "%s\n", __func__); + + tasklet_schedule(&chan->irq_tasklet); + + iop_adma_device_clear_eot_status(chan); + + return IRQ_HANDLED; +} + +static irqreturn_t iop_adma_eoc_handler(int irq, void *data) +{ + struct iop_adma_chan *chan = data; + + dev_dbg(chan->device->common.dev, "%s\n", __func__); + + tasklet_schedule(&chan->irq_tasklet); + + iop_adma_device_clear_eoc_status(chan); + + return IRQ_HANDLED; +} + +static irqreturn_t iop_adma_err_handler(int irq, void *data) +{ + struct iop_adma_chan *chan = data; + unsigned long status = iop_chan_get_status(chan); + + dev_printk(KERN_ERR, chan->device->common.dev, + "error ( %s%s%s%s%s%s%s)\n", + iop_is_err_int_parity(status, chan) ? "int_parity " : "", + iop_is_err_mcu_abort(status, chan) ? "mcu_abort " : "", + iop_is_err_int_tabort(status, chan) ? "int_tabort " : "", + iop_is_err_int_mabort(status, chan) ? "int_mabort " : "", + iop_is_err_pci_tabort(status, chan) ? "pci_tabort " : "", + iop_is_err_pci_mabort(status, chan) ? "pci_mabort " : "", + iop_is_err_split_tx(status, chan) ? "split_tx " : ""); + + iop_adma_device_clear_err_status(chan); + + BUG(); + + return IRQ_HANDLED; +} + +static void iop_adma_issue_pending(struct dma_chan *chan) +{ + struct iop_adma_chan *iop_chan = to_iop_adma_chan(chan); + + if (iop_chan->pending) { + iop_chan->pending = 0; + iop_chan_append(iop_chan); + } +} + +/* + * Perform a transaction to verify the HW works. + */ +#define IOP_ADMA_TEST_SIZE 2000 + +static int __devinit iop_adma_memcpy_self_test(struct iop_adma_device *device) +{ + int i; + void *src, *dest; + dma_addr_t src_dma, dest_dma; + struct dma_chan *dma_chan; + dma_cookie_t cookie; + struct dma_async_tx_descriptor *tx; + int err = 0; + struct iop_adma_chan *iop_chan; + + dev_dbg(device->common.dev, "%s\n", __func__); + + src = kmalloc(IOP_ADMA_TEST_SIZE, GFP_KERNEL); + if (!src) + return -ENOMEM; + dest = kzalloc(IOP_ADMA_TEST_SIZE, GFP_KERNEL); + if (!dest) { + kfree(src); + return -ENOMEM; + } + + /* Fill in src buffer */ + for (i = 0; i < IOP_ADMA_TEST_SIZE; i++) + ((u8 *) src)[i] = (u8)i; + + /* Start copy, using first DMA channel */ + dma_chan = container_of(device->common.channels.next, + struct dma_chan, + device_node); + if (iop_adma_alloc_chan_resources(dma_chan) < 1) { + err = -ENODEV; + goto out; + } + + dest_dma = dma_map_single(dma_chan->device->dev, dest, + IOP_ADMA_TEST_SIZE, DMA_FROM_DEVICE); + src_dma = dma_map_single(dma_chan->device->dev, src, + IOP_ADMA_TEST_SIZE, DMA_TO_DEVICE); + tx = iop_adma_prep_dma_memcpy(dma_chan, dest_dma, src_dma, + IOP_ADMA_TEST_SIZE, + DMA_PREP_INTERRUPT | DMA_CTRL_ACK); + + cookie = iop_adma_tx_submit(tx); + iop_adma_issue_pending(dma_chan); + msleep(1); + + if (iop_adma_is_complete(dma_chan, cookie, NULL, NULL) != + DMA_SUCCESS) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test copy timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + iop_chan = to_iop_adma_chan(dma_chan); + dma_sync_single_for_cpu(&iop_chan->device->pdev->dev, dest_dma, + IOP_ADMA_TEST_SIZE, DMA_FROM_DEVICE); + if (memcmp(src, dest, IOP_ADMA_TEST_SIZE)) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test copy failed compare, disabling\n"); + err = -ENODEV; + goto free_resources; + } + +free_resources: + iop_adma_free_chan_resources(dma_chan); +out: + kfree(src); + kfree(dest); + return err; +} + +#define IOP_ADMA_NUM_SRC_TEST 4 /* must be <= 15 */ +static int __devinit +iop_adma_xor_val_self_test(struct iop_adma_device *device) +{ + int i, src_idx; + struct page *dest; + struct page *xor_srcs[IOP_ADMA_NUM_SRC_TEST]; + struct page *zero_sum_srcs[IOP_ADMA_NUM_SRC_TEST + 1]; + dma_addr_t dma_srcs[IOP_ADMA_NUM_SRC_TEST + 1]; + dma_addr_t dma_addr, dest_dma; + struct dma_async_tx_descriptor *tx; + struct dma_chan *dma_chan; + dma_cookie_t cookie; + u8 cmp_byte = 0; + u32 cmp_word; + u32 zero_sum_result; + int err = 0; + struct iop_adma_chan *iop_chan; + + dev_dbg(device->common.dev, "%s\n", __func__); + + for (src_idx = 0; src_idx < IOP_ADMA_NUM_SRC_TEST; src_idx++) { + xor_srcs[src_idx] = alloc_page(GFP_KERNEL); + if (!xor_srcs[src_idx]) { + while (src_idx--) + __free_page(xor_srcs[src_idx]); + return -ENOMEM; + } + } + + dest = alloc_page(GFP_KERNEL); + if (!dest) { + while (src_idx--) + __free_page(xor_srcs[src_idx]); + return -ENOMEM; + } + + /* Fill in src buffers */ + for (src_idx = 0; src_idx < IOP_ADMA_NUM_SRC_TEST; src_idx++) { + u8 *ptr = page_address(xor_srcs[src_idx]); + for (i = 0; i < PAGE_SIZE; i++) + ptr[i] = (1 << src_idx); + } + + for (src_idx = 0; src_idx < IOP_ADMA_NUM_SRC_TEST; src_idx++) + cmp_byte ^= (u8) (1 << src_idx); + + cmp_word = (cmp_byte << 24) | (cmp_byte << 16) | + (cmp_byte << 8) | cmp_byte; + + memset(page_address(dest), 0, PAGE_SIZE); + + dma_chan = container_of(device->common.channels.next, + struct dma_chan, + device_node); + if (iop_adma_alloc_chan_resources(dma_chan) < 1) { + err = -ENODEV; + goto out; + } + + /* test xor */ + dest_dma = dma_map_page(dma_chan->device->dev, dest, 0, + PAGE_SIZE, DMA_FROM_DEVICE); + for (i = 0; i < IOP_ADMA_NUM_SRC_TEST; i++) + dma_srcs[i] = dma_map_page(dma_chan->device->dev, xor_srcs[i], + 0, PAGE_SIZE, DMA_TO_DEVICE); + tx = iop_adma_prep_dma_xor(dma_chan, dest_dma, dma_srcs, + IOP_ADMA_NUM_SRC_TEST, PAGE_SIZE, + DMA_PREP_INTERRUPT | DMA_CTRL_ACK); + + cookie = iop_adma_tx_submit(tx); + iop_adma_issue_pending(dma_chan); + msleep(8); + + if (iop_adma_is_complete(dma_chan, cookie, NULL, NULL) != + DMA_SUCCESS) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test xor timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + iop_chan = to_iop_adma_chan(dma_chan); + dma_sync_single_for_cpu(&iop_chan->device->pdev->dev, dest_dma, + PAGE_SIZE, DMA_FROM_DEVICE); + for (i = 0; i < (PAGE_SIZE / sizeof(u32)); i++) { + u32 *ptr = page_address(dest); + if (ptr[i] != cmp_word) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test xor failed compare, disabling\n"); + err = -ENODEV; + goto free_resources; + } + } + dma_sync_single_for_device(&iop_chan->device->pdev->dev, dest_dma, + PAGE_SIZE, DMA_TO_DEVICE); + + /* skip zero sum if the capability is not present */ + if (!dma_has_cap(DMA_XOR_VAL, dma_chan->device->cap_mask)) + goto free_resources; + + /* zero sum the sources with the destintation page */ + for (i = 0; i < IOP_ADMA_NUM_SRC_TEST; i++) + zero_sum_srcs[i] = xor_srcs[i]; + zero_sum_srcs[i] = dest; + + zero_sum_result = 1; + + for (i = 0; i < IOP_ADMA_NUM_SRC_TEST + 1; i++) + dma_srcs[i] = dma_map_page(dma_chan->device->dev, + zero_sum_srcs[i], 0, PAGE_SIZE, + DMA_TO_DEVICE); + tx = iop_adma_prep_dma_xor_val(dma_chan, dma_srcs, + IOP_ADMA_NUM_SRC_TEST + 1, PAGE_SIZE, + &zero_sum_result, + DMA_PREP_INTERRUPT | DMA_CTRL_ACK); + + cookie = iop_adma_tx_submit(tx); + iop_adma_issue_pending(dma_chan); + msleep(8); + + if (iop_adma_is_complete(dma_chan, cookie, NULL, NULL) != DMA_SUCCESS) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test zero sum timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + if (zero_sum_result != 0) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test zero sum failed compare, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + /* test memset */ + dma_addr = dma_map_page(dma_chan->device->dev, dest, 0, + PAGE_SIZE, DMA_FROM_DEVICE); + tx = iop_adma_prep_dma_memset(dma_chan, dma_addr, 0, PAGE_SIZE, + DMA_PREP_INTERRUPT | DMA_CTRL_ACK); + + cookie = iop_adma_tx_submit(tx); + iop_adma_issue_pending(dma_chan); + msleep(8); + + if (iop_adma_is_complete(dma_chan, cookie, NULL, NULL) != DMA_SUCCESS) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test memset timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + for (i = 0; i < PAGE_SIZE/sizeof(u32); i++) { + u32 *ptr = page_address(dest); + if (ptr[i]) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test memset failed compare, disabling\n"); + err = -ENODEV; + goto free_resources; + } + } + + /* test for non-zero parity sum */ + zero_sum_result = 0; + for (i = 0; i < IOP_ADMA_NUM_SRC_TEST + 1; i++) + dma_srcs[i] = dma_map_page(dma_chan->device->dev, + zero_sum_srcs[i], 0, PAGE_SIZE, + DMA_TO_DEVICE); + tx = iop_adma_prep_dma_xor_val(dma_chan, dma_srcs, + IOP_ADMA_NUM_SRC_TEST + 1, PAGE_SIZE, + &zero_sum_result, + DMA_PREP_INTERRUPT | DMA_CTRL_ACK); + + cookie = iop_adma_tx_submit(tx); + iop_adma_issue_pending(dma_chan); + msleep(8); + + if (iop_adma_is_complete(dma_chan, cookie, NULL, NULL) != DMA_SUCCESS) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test non-zero sum timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + if (zero_sum_result != 1) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test non-zero sum failed compare, disabling\n"); + err = -ENODEV; + goto free_resources; + } + +free_resources: + iop_adma_free_chan_resources(dma_chan); +out: + src_idx = IOP_ADMA_NUM_SRC_TEST; + while (src_idx--) + __free_page(xor_srcs[src_idx]); + __free_page(dest); + return err; +} + +#ifdef CONFIG_MD_RAID6_PQ +static int __devinit +iop_adma_pq_zero_sum_self_test(struct iop_adma_device *device) +{ + /* combined sources, software pq results, and extra hw pq results */ + struct page *pq[IOP_ADMA_NUM_SRC_TEST+2+2]; + /* ptr to the extra hw pq buffers defined above */ + struct page **pq_hw = &pq[IOP_ADMA_NUM_SRC_TEST+2]; + /* address conversion buffers (dma_map / page_address) */ + void *pq_sw[IOP_ADMA_NUM_SRC_TEST+2]; + dma_addr_t pq_src[IOP_ADMA_NUM_SRC_TEST]; + dma_addr_t pq_dest[2]; + + int i; + struct dma_async_tx_descriptor *tx; + struct dma_chan *dma_chan; + dma_cookie_t cookie; + u32 zero_sum_result; + int err = 0; + struct device *dev; + + dev_dbg(device->common.dev, "%s\n", __func__); + + for (i = 0; i < ARRAY_SIZE(pq); i++) { + pq[i] = alloc_page(GFP_KERNEL); + if (!pq[i]) { + while (i--) + __free_page(pq[i]); + return -ENOMEM; + } + } + + /* Fill in src buffers */ + for (i = 0; i < IOP_ADMA_NUM_SRC_TEST; i++) { + pq_sw[i] = page_address(pq[i]); + memset(pq_sw[i], 0x11111111 * (1<<i), PAGE_SIZE); + } + pq_sw[i] = page_address(pq[i]); + pq_sw[i+1] = page_address(pq[i+1]); + + dma_chan = container_of(device->common.channels.next, + struct dma_chan, + device_node); + if (iop_adma_alloc_chan_resources(dma_chan) < 1) { + err = -ENODEV; + goto out; + } + + dev = dma_chan->device->dev; + + /* initialize the dests */ + memset(page_address(pq_hw[0]), 0 , PAGE_SIZE); + memset(page_address(pq_hw[1]), 0 , PAGE_SIZE); + + /* test pq */ + pq_dest[0] = dma_map_page(dev, pq_hw[0], 0, PAGE_SIZE, DMA_FROM_DEVICE); + pq_dest[1] = dma_map_page(dev, pq_hw[1], 0, PAGE_SIZE, DMA_FROM_DEVICE); + for (i = 0; i < IOP_ADMA_NUM_SRC_TEST; i++) + pq_src[i] = dma_map_page(dev, pq[i], 0, PAGE_SIZE, + DMA_TO_DEVICE); + + tx = iop_adma_prep_dma_pq(dma_chan, pq_dest, pq_src, + IOP_ADMA_NUM_SRC_TEST, (u8 *)raid6_gfexp, + PAGE_SIZE, + DMA_PREP_INTERRUPT | + DMA_CTRL_ACK); + + cookie = iop_adma_tx_submit(tx); + iop_adma_issue_pending(dma_chan); + msleep(8); + + if (iop_adma_is_complete(dma_chan, cookie, NULL, NULL) != + DMA_SUCCESS) { + dev_err(dev, "Self-test pq timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + raid6_call.gen_syndrome(IOP_ADMA_NUM_SRC_TEST+2, PAGE_SIZE, pq_sw); + + if (memcmp(pq_sw[IOP_ADMA_NUM_SRC_TEST], + page_address(pq_hw[0]), PAGE_SIZE) != 0) { + dev_err(dev, "Self-test p failed compare, disabling\n"); + err = -ENODEV; + goto free_resources; + } + if (memcmp(pq_sw[IOP_ADMA_NUM_SRC_TEST+1], + page_address(pq_hw[1]), PAGE_SIZE) != 0) { + dev_err(dev, "Self-test q failed compare, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + /* test correct zero sum using the software generated pq values */ + for (i = 0; i < IOP_ADMA_NUM_SRC_TEST + 2; i++) + pq_src[i] = dma_map_page(dev, pq[i], 0, PAGE_SIZE, + DMA_TO_DEVICE); + + zero_sum_result = ~0; + tx = iop_adma_prep_dma_pq_val(dma_chan, &pq_src[IOP_ADMA_NUM_SRC_TEST], + pq_src, IOP_ADMA_NUM_SRC_TEST, + raid6_gfexp, PAGE_SIZE, &zero_sum_result, + DMA_PREP_INTERRUPT|DMA_CTRL_ACK); + + cookie = iop_adma_tx_submit(tx); + iop_adma_issue_pending(dma_chan); + msleep(8); + + if (iop_adma_is_complete(dma_chan, cookie, NULL, NULL) != + DMA_SUCCESS) { + dev_err(dev, "Self-test pq-zero-sum timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + if (zero_sum_result != 0) { + dev_err(dev, "Self-test pq-zero-sum failed to validate: %x\n", + zero_sum_result); + err = -ENODEV; + goto free_resources; + } + + /* test incorrect zero sum */ + i = IOP_ADMA_NUM_SRC_TEST; + memset(pq_sw[i] + 100, 0, 100); + memset(pq_sw[i+1] + 200, 0, 200); + for (i = 0; i < IOP_ADMA_NUM_SRC_TEST + 2; i++) + pq_src[i] = dma_map_page(dev, pq[i], 0, PAGE_SIZE, + DMA_TO_DEVICE); + + zero_sum_result = 0; + tx = iop_adma_prep_dma_pq_val(dma_chan, &pq_src[IOP_ADMA_NUM_SRC_TEST], + pq_src, IOP_ADMA_NUM_SRC_TEST, + raid6_gfexp, PAGE_SIZE, &zero_sum_result, + DMA_PREP_INTERRUPT|DMA_CTRL_ACK); + + cookie = iop_adma_tx_submit(tx); + iop_adma_issue_pending(dma_chan); + msleep(8); + + if (iop_adma_is_complete(dma_chan, cookie, NULL, NULL) != + DMA_SUCCESS) { + dev_err(dev, "Self-test !pq-zero-sum timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + if (zero_sum_result != (SUM_CHECK_P_RESULT | SUM_CHECK_Q_RESULT)) { + dev_err(dev, "Self-test !pq-zero-sum failed to validate: %x\n", + zero_sum_result); + err = -ENODEV; + goto free_resources; + } + +free_resources: + iop_adma_free_chan_resources(dma_chan); +out: + i = ARRAY_SIZE(pq); + while (i--) + __free_page(pq[i]); + return err; +} +#endif + +static int __devexit iop_adma_remove(struct platform_device *dev) +{ + struct iop_adma_device *device = platform_get_drvdata(dev); + struct dma_chan *chan, *_chan; + struct iop_adma_chan *iop_chan; + struct iop_adma_platform_data *plat_data = dev->dev.platform_data; + + dma_async_device_unregister(&device->common); + + dma_free_coherent(&dev->dev, plat_data->pool_size, + device->dma_desc_pool_virt, device->dma_desc_pool); + + list_for_each_entry_safe(chan, _chan, &device->common.channels, + device_node) { + iop_chan = to_iop_adma_chan(chan); + list_del(&chan->device_node); + kfree(iop_chan); + } + kfree(device); + + return 0; +} + +static int __devinit iop_adma_probe(struct platform_device *pdev) +{ + struct resource *res; + int ret = 0, i; + struct iop_adma_device *adev; + struct iop_adma_chan *iop_chan; + struct dma_device *dma_dev; + struct iop_adma_platform_data *plat_data = pdev->dev.platform_data; + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!res) + return -ENODEV; + + if (!devm_request_mem_region(&pdev->dev, res->start, + resource_size(res), pdev->name)) + return -EBUSY; + + adev = kzalloc(sizeof(*adev), GFP_KERNEL); + if (!adev) + return -ENOMEM; + dma_dev = &adev->common; + + /* allocate coherent memory for hardware descriptors + * note: writecombine gives slightly better performance, but + * requires that we explicitly flush the writes + */ + if ((adev->dma_desc_pool_virt = dma_alloc_writecombine(&pdev->dev, + plat_data->pool_size, + &adev->dma_desc_pool, + GFP_KERNEL)) == NULL) { + ret = -ENOMEM; + goto err_free_adev; + } + + dev_dbg(&pdev->dev, "%s: allocted descriptor pool virt %p phys %p\n", + __func__, adev->dma_desc_pool_virt, + (void *) adev->dma_desc_pool); + + adev->id = plat_data->hw_id; + + /* discover transaction capabilites from the platform data */ + dma_dev->cap_mask = plat_data->cap_mask; + + adev->pdev = pdev; + platform_set_drvdata(pdev, adev); + + INIT_LIST_HEAD(&dma_dev->channels); + + /* set base routines */ + dma_dev->device_alloc_chan_resources = iop_adma_alloc_chan_resources; + dma_dev->device_free_chan_resources = iop_adma_free_chan_resources; + dma_dev->device_is_tx_complete = iop_adma_is_complete; + dma_dev->device_issue_pending = iop_adma_issue_pending; + dma_dev->dev = &pdev->dev; + + /* set prep routines based on capability */ + if (dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask)) + dma_dev->device_prep_dma_memcpy = iop_adma_prep_dma_memcpy; + if (dma_has_cap(DMA_MEMSET, dma_dev->cap_mask)) + dma_dev->device_prep_dma_memset = iop_adma_prep_dma_memset; + if (dma_has_cap(DMA_XOR, dma_dev->cap_mask)) { + dma_dev->max_xor = iop_adma_get_max_xor(); + dma_dev->device_prep_dma_xor = iop_adma_prep_dma_xor; + } + if (dma_has_cap(DMA_XOR_VAL, dma_dev->cap_mask)) + dma_dev->device_prep_dma_xor_val = + iop_adma_prep_dma_xor_val; + if (dma_has_cap(DMA_PQ, dma_dev->cap_mask)) { + dma_set_maxpq(dma_dev, iop_adma_get_max_pq(), 0); + dma_dev->device_prep_dma_pq = iop_adma_prep_dma_pq; + } + if (dma_has_cap(DMA_PQ_VAL, dma_dev->cap_mask)) + dma_dev->device_prep_dma_pq_val = + iop_adma_prep_dma_pq_val; + if (dma_has_cap(DMA_INTERRUPT, dma_dev->cap_mask)) + dma_dev->device_prep_dma_interrupt = + iop_adma_prep_dma_interrupt; + + iop_chan = kzalloc(sizeof(*iop_chan), GFP_KERNEL); + if (!iop_chan) { + ret = -ENOMEM; + goto err_free_dma; + } + iop_chan->device = adev; + + iop_chan->mmr_base = devm_ioremap(&pdev->dev, res->start, + resource_size(res)); + if (!iop_chan->mmr_base) { + ret = -ENOMEM; + goto err_free_iop_chan; + } + tasklet_init(&iop_chan->irq_tasklet, iop_adma_tasklet, (unsigned long) + iop_chan); + + /* clear errors before enabling interrupts */ + iop_adma_device_clear_err_status(iop_chan); + + for (i = 0; i < 3; i++) { + irq_handler_t handler[] = { iop_adma_eot_handler, + iop_adma_eoc_handler, + iop_adma_err_handler }; + int irq = platform_get_irq(pdev, i); + if (irq < 0) { + ret = -ENXIO; + goto err_free_iop_chan; + } else { + ret = devm_request_irq(&pdev->dev, irq, + handler[i], 0, pdev->name, iop_chan); + if (ret) + goto err_free_iop_chan; + } + } + + spin_lock_init(&iop_chan->lock); + INIT_LIST_HEAD(&iop_chan->chain); + INIT_LIST_HEAD(&iop_chan->all_slots); + iop_chan->common.device = dma_dev; + list_add_tail(&iop_chan->common.device_node, &dma_dev->channels); + + if (dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask)) { + ret = iop_adma_memcpy_self_test(adev); + dev_dbg(&pdev->dev, "memcpy self test returned %d\n", ret); + if (ret) + goto err_free_iop_chan; + } + + if (dma_has_cap(DMA_XOR, dma_dev->cap_mask) || + dma_has_cap(DMA_MEMSET, dma_dev->cap_mask)) { + ret = iop_adma_xor_val_self_test(adev); + dev_dbg(&pdev->dev, "xor self test returned %d\n", ret); + if (ret) + goto err_free_iop_chan; + } + + if (dma_has_cap(DMA_PQ, dma_dev->cap_mask) && + dma_has_cap(DMA_PQ_VAL, dma_dev->cap_mask)) { + #ifdef CONFIG_MD_RAID6_PQ + ret = iop_adma_pq_zero_sum_self_test(adev); + dev_dbg(&pdev->dev, "pq self test returned %d\n", ret); + #else + /* can not test raid6, so do not publish capability */ + dma_cap_clear(DMA_PQ, dma_dev->cap_mask); + dma_cap_clear(DMA_PQ_VAL, dma_dev->cap_mask); + ret = 0; + #endif + if (ret) + goto err_free_iop_chan; + } + + dev_printk(KERN_INFO, &pdev->dev, "Intel(R) IOP: " + "( %s%s%s%s%s%s%s)\n", + dma_has_cap(DMA_PQ, dma_dev->cap_mask) ? "pq " : "", + dma_has_cap(DMA_PQ_VAL, dma_dev->cap_mask) ? "pq_val " : "", + dma_has_cap(DMA_XOR, dma_dev->cap_mask) ? "xor " : "", + dma_has_cap(DMA_XOR_VAL, dma_dev->cap_mask) ? "xor_val " : "", + dma_has_cap(DMA_MEMSET, dma_dev->cap_mask) ? "fill " : "", + dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask) ? "cpy " : "", + dma_has_cap(DMA_INTERRUPT, dma_dev->cap_mask) ? "intr " : ""); + + dma_async_device_register(dma_dev); + goto out; + + err_free_iop_chan: + kfree(iop_chan); + err_free_dma: + dma_free_coherent(&adev->pdev->dev, plat_data->pool_size, + adev->dma_desc_pool_virt, adev->dma_desc_pool); + err_free_adev: + kfree(adev); + out: + return ret; +} + +static void iop_chan_start_null_memcpy(struct iop_adma_chan *iop_chan) +{ + struct iop_adma_desc_slot *sw_desc, *grp_start; + dma_cookie_t cookie; + int slot_cnt, slots_per_op; + + dev_dbg(iop_chan->device->common.dev, "%s\n", __func__); + + spin_lock_bh(&iop_chan->lock); + slot_cnt = iop_chan_memcpy_slot_count(0, &slots_per_op); + sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op); + if (sw_desc) { + grp_start = sw_desc->group_head; + + list_splice_init(&sw_desc->tx_list, &iop_chan->chain); + async_tx_ack(&sw_desc->async_tx); + iop_desc_init_memcpy(grp_start, 0); + iop_desc_set_byte_count(grp_start, iop_chan, 0); + iop_desc_set_dest_addr(grp_start, iop_chan, 0); + iop_desc_set_memcpy_src_addr(grp_start, 0); + + cookie = iop_chan->common.cookie; + cookie++; + if (cookie <= 1) + cookie = 2; + + /* initialize the completed cookie to be less than + * the most recently used cookie + */ + iop_chan->completed_cookie = cookie - 1; + iop_chan->common.cookie = sw_desc->async_tx.cookie = cookie; + + /* channel should not be busy */ + BUG_ON(iop_chan_is_busy(iop_chan)); + + /* clear any prior error-status bits */ + iop_adma_device_clear_err_status(iop_chan); + + /* disable operation */ + iop_chan_disable(iop_chan); + + /* set the descriptor address */ + iop_chan_set_next_descriptor(iop_chan, sw_desc->async_tx.phys); + + /* 1/ don't add pre-chained descriptors + * 2/ dummy read to flush next_desc write + */ + BUG_ON(iop_desc_get_next_desc(sw_desc)); + + /* run the descriptor */ + iop_chan_enable(iop_chan); + } else + dev_printk(KERN_ERR, iop_chan->device->common.dev, + "failed to allocate null descriptor\n"); + spin_unlock_bh(&iop_chan->lock); +} + +static void iop_chan_start_null_xor(struct iop_adma_chan *iop_chan) +{ + struct iop_adma_desc_slot *sw_desc, *grp_start; + dma_cookie_t cookie; + int slot_cnt, slots_per_op; + + dev_dbg(iop_chan->device->common.dev, "%s\n", __func__); + + spin_lock_bh(&iop_chan->lock); + slot_cnt = iop_chan_xor_slot_count(0, 2, &slots_per_op); + sw_desc = iop_adma_alloc_slots(iop_chan, slot_cnt, slots_per_op); + if (sw_desc) { + grp_start = sw_desc->group_head; + list_splice_init(&sw_desc->tx_list, &iop_chan->chain); + async_tx_ack(&sw_desc->async_tx); + iop_desc_init_null_xor(grp_start, 2, 0); + iop_desc_set_byte_count(grp_start, iop_chan, 0); + iop_desc_set_dest_addr(grp_start, iop_chan, 0); + iop_desc_set_xor_src_addr(grp_start, 0, 0); + iop_desc_set_xor_src_addr(grp_start, 1, 0); + + cookie = iop_chan->common.cookie; + cookie++; + if (cookie <= 1) + cookie = 2; + + /* initialize the completed cookie to be less than + * the most recently used cookie + */ + iop_chan->completed_cookie = cookie - 1; + iop_chan->common.cookie = sw_desc->async_tx.cookie = cookie; + + /* channel should not be busy */ + BUG_ON(iop_chan_is_busy(iop_chan)); + + /* clear any prior error-status bits */ + iop_adma_device_clear_err_status(iop_chan); + + /* disable operation */ + iop_chan_disable(iop_chan); + + /* set the descriptor address */ + iop_chan_set_next_descriptor(iop_chan, sw_desc->async_tx.phys); + + /* 1/ don't add pre-chained descriptors + * 2/ dummy read to flush next_desc write + */ + BUG_ON(iop_desc_get_next_desc(sw_desc)); + + /* run the descriptor */ + iop_chan_enable(iop_chan); + } else + dev_printk(KERN_ERR, iop_chan->device->common.dev, + "failed to allocate null descriptor\n"); + spin_unlock_bh(&iop_chan->lock); +} + +MODULE_ALIAS("platform:iop-adma"); + +static struct platform_driver iop_adma_driver = { + .probe = iop_adma_probe, + .remove = __devexit_p(iop_adma_remove), + .driver = { + .owner = THIS_MODULE, + .name = "iop-adma", + }, +}; + +static int __init iop_adma_init (void) +{ + return platform_driver_register(&iop_adma_driver); +} + +static void __exit iop_adma_exit (void) +{ + platform_driver_unregister(&iop_adma_driver); + return; +} +module_exit(iop_adma_exit); +module_init(iop_adma_init); + +MODULE_AUTHOR("Intel Corporation"); +MODULE_DESCRIPTION("IOP ADMA Engine Driver"); +MODULE_LICENSE("GPL"); diff --git a/drivers/dma/iovlock.c b/drivers/dma/iovlock.c new file mode 100644 index 00000000000..c0a272c7368 --- /dev/null +++ b/drivers/dma/iovlock.c @@ -0,0 +1,279 @@ +/* + * Copyright(c) 2004 - 2006 Intel Corporation. All rights reserved. + * Portions based on net/core/datagram.c and copyrighted by their authors. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the Free + * Software Foundation; either version 2 of the License, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 + * Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * The full GNU General Public License is included in this distribution in the + * file called COPYING. + */ + +/* + * This code allows the net stack to make use of a DMA engine for + * skb to iovec copies. + */ + +#include <linux/dmaengine.h> +#include <linux/pagemap.h> +#include <net/tcp.h> /* for memcpy_toiovec */ +#include <asm/io.h> +#include <asm/uaccess.h> + +static int num_pages_spanned(struct iovec *iov) +{ + return + ((PAGE_ALIGN((unsigned long)iov->iov_base + iov->iov_len) - + ((unsigned long)iov->iov_base & PAGE_MASK)) >> PAGE_SHIFT); +} + +/* + * Pin down all the iovec pages needed for len bytes. + * Return a struct dma_pinned_list to keep track of pages pinned down. + * + * We are allocating a single chunk of memory, and then carving it up into + * 3 sections, the latter 2 whose size depends on the number of iovecs and the + * total number of pages, respectively. + */ +struct dma_pinned_list *dma_pin_iovec_pages(struct iovec *iov, size_t len) +{ + struct dma_pinned_list *local_list; + struct page **pages; + int i; + int ret; + int nr_iovecs = 0; + int iovec_len_used = 0; + int iovec_pages_used = 0; + + /* don't pin down non-user-based iovecs */ + if (segment_eq(get_fs(), KERNEL_DS)) + return NULL; + + /* determine how many iovecs/pages there are, up front */ + do { + iovec_len_used += iov[nr_iovecs].iov_len; + iovec_pages_used += num_pages_spanned(&iov[nr_iovecs]); + nr_iovecs++; + } while (iovec_len_used < len); + + /* single kmalloc for pinned list, page_list[], and the page arrays */ + local_list = kmalloc(sizeof(*local_list) + + (nr_iovecs * sizeof (struct dma_page_list)) + + (iovec_pages_used * sizeof (struct page*)), GFP_KERNEL); + if (!local_list) + goto out; + + /* list of pages starts right after the page list array */ + pages = (struct page **) &local_list->page_list[nr_iovecs]; + + local_list->nr_iovecs = 0; + + for (i = 0; i < nr_iovecs; i++) { + struct dma_page_list *page_list = &local_list->page_list[i]; + + len -= iov[i].iov_len; + + if (!access_ok(VERIFY_WRITE, iov[i].iov_base, iov[i].iov_len)) + goto unpin; + + page_list->nr_pages = num_pages_spanned(&iov[i]); + page_list->base_address = iov[i].iov_base; + + page_list->pages = pages; + pages += page_list->nr_pages; + + /* pin pages down */ + down_read(¤t->mm->mmap_sem); + ret = get_user_pages( + current, + current->mm, + (unsigned long) iov[i].iov_base, + page_list->nr_pages, + 1, /* write */ + 0, /* force */ + page_list->pages, + NULL); + up_read(¤t->mm->mmap_sem); + + if (ret != page_list->nr_pages) + goto unpin; + + local_list->nr_iovecs = i + 1; + } + + return local_list; + +unpin: + dma_unpin_iovec_pages(local_list); +out: + return NULL; +} + +void dma_unpin_iovec_pages(struct dma_pinned_list *pinned_list) +{ + int i, j; + + if (!pinned_list) + return; + + for (i = 0; i < pinned_list->nr_iovecs; i++) { + struct dma_page_list *page_list = &pinned_list->page_list[i]; + for (j = 0; j < page_list->nr_pages; j++) { + set_page_dirty_lock(page_list->pages[j]); + page_cache_release(page_list->pages[j]); + } + } + + kfree(pinned_list); +} + + +/* + * We have already pinned down the pages we will be using in the iovecs. + * Each entry in iov array has corresponding entry in pinned_list->page_list. + * Using array indexing to keep iov[] and page_list[] in sync. + * Initial elements in iov array's iov->iov_len will be 0 if already copied into + * by another call. + * iov array length remaining guaranteed to be bigger than len. + */ +dma_cookie_t dma_memcpy_to_iovec(struct dma_chan *chan, struct iovec *iov, + struct dma_pinned_list *pinned_list, unsigned char *kdata, size_t len) +{ + int iov_byte_offset; + int copy; + dma_cookie_t dma_cookie = 0; + int iovec_idx; + int page_idx; + + if (!chan) + return memcpy_toiovec(iov, kdata, len); + + iovec_idx = 0; + while (iovec_idx < pinned_list->nr_iovecs) { + struct dma_page_list *page_list; + + /* skip already used-up iovecs */ + while (!iov[iovec_idx].iov_len) + iovec_idx++; + + page_list = &pinned_list->page_list[iovec_idx]; + + iov_byte_offset = ((unsigned long)iov[iovec_idx].iov_base & ~PAGE_MASK); + page_idx = (((unsigned long)iov[iovec_idx].iov_base & PAGE_MASK) + - ((unsigned long)page_list->base_address & PAGE_MASK)) >> PAGE_SHIFT; + + /* break up copies to not cross page boundary */ + while (iov[iovec_idx].iov_len) { + copy = min_t(int, PAGE_SIZE - iov_byte_offset, len); + copy = min_t(int, copy, iov[iovec_idx].iov_len); + + dma_cookie = dma_async_memcpy_buf_to_pg(chan, + page_list->pages[page_idx], + iov_byte_offset, + kdata, + copy); + /* poll for a descriptor slot */ + if (unlikely(dma_cookie < 0)) { + dma_async_issue_pending(chan); + continue; + } + + len -= copy; + iov[iovec_idx].iov_len -= copy; + iov[iovec_idx].iov_base += copy; + + if (!len) + return dma_cookie; + + kdata += copy; + iov_byte_offset = 0; + page_idx++; + } + iovec_idx++; + } + + /* really bad if we ever run out of iovecs */ + BUG(); + return -EFAULT; +} + +dma_cookie_t dma_memcpy_pg_to_iovec(struct dma_chan *chan, struct iovec *iov, + struct dma_pinned_list *pinned_list, struct page *page, + unsigned int offset, size_t len) +{ + int iov_byte_offset; + int copy; + dma_cookie_t dma_cookie = 0; + int iovec_idx; + int page_idx; + int err; + + /* this needs as-yet-unimplemented buf-to-buff, so punt. */ + /* TODO: use dma for this */ + if (!chan || !pinned_list) { + u8 *vaddr = kmap(page); + err = memcpy_toiovec(iov, vaddr + offset, len); + kunmap(page); + return err; + } + + iovec_idx = 0; + while (iovec_idx < pinned_list->nr_iovecs) { + struct dma_page_list *page_list; + + /* skip already used-up iovecs */ + while (!iov[iovec_idx].iov_len) + iovec_idx++; + + page_list = &pinned_list->page_list[iovec_idx]; + + iov_byte_offset = ((unsigned long)iov[iovec_idx].iov_base & ~PAGE_MASK); + page_idx = (((unsigned long)iov[iovec_idx].iov_base & PAGE_MASK) + - ((unsigned long)page_list->base_address & PAGE_MASK)) >> PAGE_SHIFT; + + /* break up copies to not cross page boundary */ + while (iov[iovec_idx].iov_len) { + copy = min_t(int, PAGE_SIZE - iov_byte_offset, len); + copy = min_t(int, copy, iov[iovec_idx].iov_len); + + dma_cookie = dma_async_memcpy_pg_to_pg(chan, + page_list->pages[page_idx], + iov_byte_offset, + page, + offset, + copy); + /* poll for a descriptor slot */ + if (unlikely(dma_cookie < 0)) { + dma_async_issue_pending(chan); + continue; + } + + len -= copy; + iov[iovec_idx].iov_len -= copy; + iov[iovec_idx].iov_base += copy; + + if (!len) + return dma_cookie; + + offset += copy; + iov_byte_offset = 0; + page_idx++; + } + iovec_idx++; + } + + /* really bad if we ever run out of iovecs */ + BUG(); + return -EFAULT; +} diff --git a/drivers/dma/ipu/Makefile b/drivers/dma/ipu/Makefile new file mode 100644 index 00000000000..6704cf48326 --- /dev/null +++ b/drivers/dma/ipu/Makefile @@ -0,0 +1 @@ +obj-y += ipu_irq.o ipu_idmac.o diff --git a/drivers/dma/ipu/ipu_idmac.c b/drivers/dma/ipu/ipu_idmac.c new file mode 100644 index 00000000000..e80bae1673f --- /dev/null +++ b/drivers/dma/ipu/ipu_idmac.c @@ -0,0 +1,1867 @@ +/* + * Copyright (C) 2008 + * Guennadi Liakhovetski, DENX Software Engineering, <lg@denx.de> + * + * Copyright (C) 2005-2007 Freescale Semiconductor, Inc. All Rights Reserved. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include <linux/init.h> +#include <linux/platform_device.h> +#include <linux/err.h> +#include <linux/spinlock.h> +#include <linux/delay.h> +#include <linux/list.h> +#include <linux/clk.h> +#include <linux/vmalloc.h> +#include <linux/string.h> +#include <linux/interrupt.h> +#include <linux/io.h> + +#include <mach/ipu.h> + +#include "ipu_intern.h" + +#define FS_VF_IN_VALID 0x00000002 +#define FS_ENC_IN_VALID 0x00000001 + +static int ipu_disable_channel(struct idmac *idmac, struct idmac_channel *ichan, + bool wait_for_stop); + +/* + * There can be only one, we could allocate it dynamically, but then we'd have + * to add an extra parameter to some functions, and use something as ugly as + * struct ipu *ipu = to_ipu(to_idmac(ichan->dma_chan.device)); + * in the ISR + */ +static struct ipu ipu_data; + +#define to_ipu(id) container_of(id, struct ipu, idmac) + +static u32 __idmac_read_icreg(struct ipu *ipu, unsigned long reg) +{ + return __raw_readl(ipu->reg_ic + reg); +} + +#define idmac_read_icreg(ipu, reg) __idmac_read_icreg(ipu, reg - IC_CONF) + +static void __idmac_write_icreg(struct ipu *ipu, u32 value, unsigned long reg) +{ + __raw_writel(value, ipu->reg_ic + reg); +} + +#define idmac_write_icreg(ipu, v, reg) __idmac_write_icreg(ipu, v, reg - IC_CONF) + +static u32 idmac_read_ipureg(struct ipu *ipu, unsigned long reg) +{ + return __raw_readl(ipu->reg_ipu + reg); +} + +static void idmac_write_ipureg(struct ipu *ipu, u32 value, unsigned long reg) +{ + __raw_writel(value, ipu->reg_ipu + reg); +} + +/***************************************************************************** + * IPU / IC common functions + */ +static void dump_idmac_reg(struct ipu *ipu) +{ + dev_dbg(ipu->dev, "IDMAC_CONF 0x%x, IC_CONF 0x%x, IDMAC_CHA_EN 0x%x, " + "IDMAC_CHA_PRI 0x%x, IDMAC_CHA_BUSY 0x%x\n", + idmac_read_icreg(ipu, IDMAC_CONF), + idmac_read_icreg(ipu, IC_CONF), + idmac_read_icreg(ipu, IDMAC_CHA_EN), + idmac_read_icreg(ipu, IDMAC_CHA_PRI), + idmac_read_icreg(ipu, IDMAC_CHA_BUSY)); + dev_dbg(ipu->dev, "BUF0_RDY 0x%x, BUF1_RDY 0x%x, CUR_BUF 0x%x, " + "DB_MODE 0x%x, TASKS_STAT 0x%x\n", + idmac_read_ipureg(ipu, IPU_CHA_BUF0_RDY), + idmac_read_ipureg(ipu, IPU_CHA_BUF1_RDY), + idmac_read_ipureg(ipu, IPU_CHA_CUR_BUF), + idmac_read_ipureg(ipu, IPU_CHA_DB_MODE_SEL), + idmac_read_ipureg(ipu, IPU_TASKS_STAT)); +} + +static uint32_t bytes_per_pixel(enum pixel_fmt fmt) +{ + switch (fmt) { + case IPU_PIX_FMT_GENERIC: /* generic data */ + case IPU_PIX_FMT_RGB332: + case IPU_PIX_FMT_YUV420P: + case IPU_PIX_FMT_YUV422P: + default: + return 1; + case IPU_PIX_FMT_RGB565: + case IPU_PIX_FMT_YUYV: + case IPU_PIX_FMT_UYVY: + return 2; + case IPU_PIX_FMT_BGR24: + case IPU_PIX_FMT_RGB24: + return 3; + case IPU_PIX_FMT_GENERIC_32: /* generic data */ + case IPU_PIX_FMT_BGR32: + case IPU_PIX_FMT_RGB32: + case IPU_PIX_FMT_ABGR32: + return 4; + } +} + +/* Enable direct write to memory by the Camera Sensor Interface */ +static void ipu_ic_enable_task(struct ipu *ipu, enum ipu_channel channel) +{ + uint32_t ic_conf, mask; + + switch (channel) { + case IDMAC_IC_0: + mask = IC_CONF_PRPENC_EN; + break; + case IDMAC_IC_7: + mask = IC_CONF_RWS_EN | IC_CONF_PRPENC_EN; + break; + default: + return; + } + ic_conf = idmac_read_icreg(ipu, IC_CONF) | mask; + idmac_write_icreg(ipu, ic_conf, IC_CONF); +} + +/* Called under spin_lock_irqsave(&ipu_data.lock) */ +static void ipu_ic_disable_task(struct ipu *ipu, enum ipu_channel channel) +{ + uint32_t ic_conf, mask; + + switch (channel) { + case IDMAC_IC_0: + mask = IC_CONF_PRPENC_EN; + break; + case IDMAC_IC_7: + mask = IC_CONF_RWS_EN | IC_CONF_PRPENC_EN; + break; + default: + return; + } + ic_conf = idmac_read_icreg(ipu, IC_CONF) & ~mask; + idmac_write_icreg(ipu, ic_conf, IC_CONF); +} + +static uint32_t ipu_channel_status(struct ipu *ipu, enum ipu_channel channel) +{ + uint32_t stat = TASK_STAT_IDLE; + uint32_t task_stat_reg = idmac_read_ipureg(ipu, IPU_TASKS_STAT); + + switch (channel) { + case IDMAC_IC_7: + stat = (task_stat_reg & TSTAT_CSI2MEM_MASK) >> + TSTAT_CSI2MEM_OFFSET; + break; + case IDMAC_IC_0: + case IDMAC_SDC_0: + case IDMAC_SDC_1: + default: + break; + } + return stat; +} + +struct chan_param_mem_planar { + /* Word 0 */ + u32 xv:10; + u32 yv:10; + u32 xb:12; + + u32 yb:12; + u32 res1:2; + u32 nsb:1; + u32 lnpb:6; + u32 ubo_l:11; + + u32 ubo_h:15; + u32 vbo_l:17; + + u32 vbo_h:9; + u32 res2:3; + u32 fw:12; + u32 fh_l:8; + + u32 fh_h:4; + u32 res3:28; + + /* Word 1 */ + u32 eba0; + + u32 eba1; + + u32 bpp:3; + u32 sl:14; + u32 pfs:3; + u32 bam:3; + u32 res4:2; + u32 npb:6; + u32 res5:1; + + u32 sat:2; + u32 res6:30; +} __attribute__ ((packed)); + +struct chan_param_mem_interleaved { + /* Word 0 */ + u32 xv:10; + u32 yv:10; + u32 xb:12; + + u32 yb:12; + u32 sce:1; + u32 res1:1; + u32 nsb:1; + u32 lnpb:6; + u32 sx:10; + u32 sy_l:1; + + u32 sy_h:9; + u32 ns:10; + u32 sm:10; + u32 sdx_l:3; + + u32 sdx_h:2; + u32 sdy:5; + u32 sdrx:1; + u32 sdry:1; + u32 sdr1:1; + u32 res2:2; + u32 fw:12; + u32 fh_l:8; + + u32 fh_h:4; + u32 res3:28; + + /* Word 1 */ + u32 eba0; + + u32 eba1; + + u32 bpp:3; + u32 sl:14; + u32 pfs:3; + u32 bam:3; + u32 res4:2; + u32 npb:6; + u32 res5:1; + + u32 sat:2; + u32 scc:1; + u32 ofs0:5; + u32 ofs1:5; + u32 ofs2:5; + u32 ofs3:5; + u32 wid0:3; + u32 wid1:3; + u32 wid2:3; + + u32 wid3:3; + u32 dec_sel:1; + u32 res6:28; +} __attribute__ ((packed)); + +union chan_param_mem { + struct chan_param_mem_planar pp; + struct chan_param_mem_interleaved ip; +}; + +static void ipu_ch_param_set_plane_offset(union chan_param_mem *params, + u32 u_offset, u32 v_offset) +{ + params->pp.ubo_l = u_offset & 0x7ff; + params->pp.ubo_h = u_offset >> 11; + params->pp.vbo_l = v_offset & 0x1ffff; + params->pp.vbo_h = v_offset >> 17; +} + +static void ipu_ch_param_set_size(union chan_param_mem *params, + uint32_t pixel_fmt, uint16_t width, + uint16_t height, uint16_t stride) +{ + u32 u_offset; + u32 v_offset; + + params->pp.fw = width - 1; + params->pp.fh_l = height - 1; + params->pp.fh_h = (height - 1) >> 8; + params->pp.sl = stride - 1; + + switch (pixel_fmt) { + case IPU_PIX_FMT_GENERIC: + /*Represents 8-bit Generic data */ + params->pp.bpp = 3; + params->pp.pfs = 7; + params->pp.npb = 31; + params->pp.sat = 2; /* SAT = use 32-bit access */ + break; + case IPU_PIX_FMT_GENERIC_32: + /*Represents 32-bit Generic data */ + params->pp.bpp = 0; + params->pp.pfs = 7; + params->pp.npb = 7; + params->pp.sat = 2; /* SAT = use 32-bit access */ + break; + case IPU_PIX_FMT_RGB565: + params->ip.bpp = 2; + params->ip.pfs = 4; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + params->ip.ofs0 = 0; /* Red bit offset */ + params->ip.ofs1 = 5; /* Green bit offset */ + params->ip.ofs2 = 11; /* Blue bit offset */ + params->ip.ofs3 = 16; /* Alpha bit offset */ + params->ip.wid0 = 4; /* Red bit width - 1 */ + params->ip.wid1 = 5; /* Green bit width - 1 */ + params->ip.wid2 = 4; /* Blue bit width - 1 */ + break; + case IPU_PIX_FMT_BGR24: + params->ip.bpp = 1; /* 24 BPP & RGB PFS */ + params->ip.pfs = 4; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + params->ip.ofs0 = 0; /* Red bit offset */ + params->ip.ofs1 = 8; /* Green bit offset */ + params->ip.ofs2 = 16; /* Blue bit offset */ + params->ip.ofs3 = 24; /* Alpha bit offset */ + params->ip.wid0 = 7; /* Red bit width - 1 */ + params->ip.wid1 = 7; /* Green bit width - 1 */ + params->ip.wid2 = 7; /* Blue bit width - 1 */ + break; + case IPU_PIX_FMT_RGB24: + params->ip.bpp = 1; /* 24 BPP & RGB PFS */ + params->ip.pfs = 4; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + params->ip.ofs0 = 16; /* Red bit offset */ + params->ip.ofs1 = 8; /* Green bit offset */ + params->ip.ofs2 = 0; /* Blue bit offset */ + params->ip.ofs3 = 24; /* Alpha bit offset */ + params->ip.wid0 = 7; /* Red bit width - 1 */ + params->ip.wid1 = 7; /* Green bit width - 1 */ + params->ip.wid2 = 7; /* Blue bit width - 1 */ + break; + case IPU_PIX_FMT_BGRA32: + case IPU_PIX_FMT_BGR32: + params->ip.bpp = 0; + params->ip.pfs = 4; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + params->ip.ofs0 = 8; /* Red bit offset */ + params->ip.ofs1 = 16; /* Green bit offset */ + params->ip.ofs2 = 24; /* Blue bit offset */ + params->ip.ofs3 = 0; /* Alpha bit offset */ + params->ip.wid0 = 7; /* Red bit width - 1 */ + params->ip.wid1 = 7; /* Green bit width - 1 */ + params->ip.wid2 = 7; /* Blue bit width - 1 */ + params->ip.wid3 = 7; /* Alpha bit width - 1 */ + break; + case IPU_PIX_FMT_RGBA32: + case IPU_PIX_FMT_RGB32: + params->ip.bpp = 0; + params->ip.pfs = 4; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + params->ip.ofs0 = 24; /* Red bit offset */ + params->ip.ofs1 = 16; /* Green bit offset */ + params->ip.ofs2 = 8; /* Blue bit offset */ + params->ip.ofs3 = 0; /* Alpha bit offset */ + params->ip.wid0 = 7; /* Red bit width - 1 */ + params->ip.wid1 = 7; /* Green bit width - 1 */ + params->ip.wid2 = 7; /* Blue bit width - 1 */ + params->ip.wid3 = 7; /* Alpha bit width - 1 */ + break; + case IPU_PIX_FMT_ABGR32: + params->ip.bpp = 0; + params->ip.pfs = 4; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + params->ip.ofs0 = 8; /* Red bit offset */ + params->ip.ofs1 = 16; /* Green bit offset */ + params->ip.ofs2 = 24; /* Blue bit offset */ + params->ip.ofs3 = 0; /* Alpha bit offset */ + params->ip.wid0 = 7; /* Red bit width - 1 */ + params->ip.wid1 = 7; /* Green bit width - 1 */ + params->ip.wid2 = 7; /* Blue bit width - 1 */ + params->ip.wid3 = 7; /* Alpha bit width - 1 */ + break; + case IPU_PIX_FMT_UYVY: + params->ip.bpp = 2; + params->ip.pfs = 6; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + break; + case IPU_PIX_FMT_YUV420P2: + case IPU_PIX_FMT_YUV420P: + params->ip.bpp = 3; + params->ip.pfs = 3; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + u_offset = stride * height; + v_offset = u_offset + u_offset / 4; + ipu_ch_param_set_plane_offset(params, u_offset, v_offset); + break; + case IPU_PIX_FMT_YVU422P: + params->ip.bpp = 3; + params->ip.pfs = 2; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + v_offset = stride * height; + u_offset = v_offset + v_offset / 2; + ipu_ch_param_set_plane_offset(params, u_offset, v_offset); + break; + case IPU_PIX_FMT_YUV422P: + params->ip.bpp = 3; + params->ip.pfs = 2; + params->ip.npb = 7; + params->ip.sat = 2; /* SAT = 32-bit access */ + u_offset = stride * height; + v_offset = u_offset + u_offset / 2; + ipu_ch_param_set_plane_offset(params, u_offset, v_offset); + break; + default: + dev_err(ipu_data.dev, + "mx3 ipu: unimplemented pixel format %d\n", pixel_fmt); + break; + } + + params->pp.nsb = 1; +} + +static void ipu_ch_param_set_burst_size(union chan_param_mem *params, + uint16_t burst_pixels) +{ + params->pp.npb = burst_pixels - 1; +} + +static void ipu_ch_param_set_buffer(union chan_param_mem *params, + dma_addr_t buf0, dma_addr_t buf1) +{ + params->pp.eba0 = buf0; + params->pp.eba1 = buf1; +} + +static void ipu_ch_param_set_rotation(union chan_param_mem *params, + enum ipu_rotate_mode rotate) +{ + params->pp.bam = rotate; +} + +static void ipu_write_param_mem(uint32_t addr, uint32_t *data, + uint32_t num_words) +{ + for (; num_words > 0; num_words--) { + dev_dbg(ipu_data.dev, + "write param mem - addr = 0x%08X, data = 0x%08X\n", + addr, *data); + idmac_write_ipureg(&ipu_data, addr, IPU_IMA_ADDR); + idmac_write_ipureg(&ipu_data, *data++, IPU_IMA_DATA); + addr++; + if ((addr & 0x7) == 5) { + addr &= ~0x7; /* set to word 0 */ + addr += 8; /* increment to next row */ + } + } +} + +static int calc_resize_coeffs(uint32_t in_size, uint32_t out_size, + uint32_t *resize_coeff, + uint32_t *downsize_coeff) +{ + uint32_t temp_size; + uint32_t temp_downsize; + + *resize_coeff = 1 << 13; + *downsize_coeff = 1 << 13; + + /* Cannot downsize more than 8:1 */ + if (out_size << 3 < in_size) + return -EINVAL; + + /* compute downsizing coefficient */ + temp_downsize = 0; + temp_size = in_size; + while (temp_size >= out_size * 2 && temp_downsize < 2) { + temp_size >>= 1; + temp_downsize++; + } + *downsize_coeff = temp_downsize; + + /* + * compute resizing coefficient using the following formula: + * resize_coeff = M*(SI -1)/(SO - 1) + * where M = 2^13, SI - input size, SO - output size + */ + *resize_coeff = (8192L * (temp_size - 1)) / (out_size - 1); + if (*resize_coeff >= 16384L) { + dev_err(ipu_data.dev, "Warning! Overflow on resize coeff.\n"); + *resize_coeff = 0x3FFF; + } + + dev_dbg(ipu_data.dev, "resizing from %u -> %u pixels, " + "downsize=%u, resize=%u.%lu (reg=%u)\n", in_size, out_size, + *downsize_coeff, *resize_coeff >= 8192L ? 1 : 0, + ((*resize_coeff & 0x1FFF) * 10000L) / 8192L, *resize_coeff); + + return 0; +} + +static enum ipu_color_space format_to_colorspace(enum pixel_fmt fmt) +{ + switch (fmt) { + case IPU_PIX_FMT_RGB565: + case IPU_PIX_FMT_BGR24: + case IPU_PIX_FMT_RGB24: + case IPU_PIX_FMT_BGR32: + case IPU_PIX_FMT_RGB32: + return IPU_COLORSPACE_RGB; + default: + return IPU_COLORSPACE_YCBCR; + } +} + +static int ipu_ic_init_prpenc(struct ipu *ipu, + union ipu_channel_param *params, bool src_is_csi) +{ + uint32_t reg, ic_conf; + uint32_t downsize_coeff, resize_coeff; + enum ipu_color_space in_fmt, out_fmt; + + /* Setup vertical resizing */ + calc_resize_coeffs(params->video.in_height, + params->video.out_height, + &resize_coeff, &downsize_coeff); + reg = (downsize_coeff << 30) | (resize_coeff << 16); + + /* Setup horizontal resizing */ + calc_resize_coeffs(params->video.in_width, + params->video.out_width, + &resize_coeff, &downsize_coeff); + reg |= (downsize_coeff << 14) | resize_coeff; + + /* Setup color space conversion */ + in_fmt = format_to_colorspace(params->video.in_pixel_fmt); + out_fmt = format_to_colorspace(params->video.out_pixel_fmt); + + /* + * Colourspace conversion unsupported yet - see _init_csc() in + * Freescale sources + */ + if (in_fmt != out_fmt) { + dev_err(ipu->dev, "Colourspace conversion unsupported!\n"); + return -EOPNOTSUPP; + } + + idmac_write_icreg(ipu, reg, IC_PRP_ENC_RSC); + + ic_conf = idmac_read_icreg(ipu, IC_CONF); + + if (src_is_csi) + ic_conf &= ~IC_CONF_RWS_EN; + else + ic_conf |= IC_CONF_RWS_EN; + + idmac_write_icreg(ipu, ic_conf, IC_CONF); + + return 0; +} + +static uint32_t dma_param_addr(uint32_t dma_ch) +{ + /* Channel Parameter Memory */ + return 0x10000 | (dma_ch << 4); +} + +static void ipu_channel_set_priority(struct ipu *ipu, enum ipu_channel channel, + bool prio) +{ + u32 reg = idmac_read_icreg(ipu, IDMAC_CHA_PRI); + + if (prio) + reg |= 1UL << channel; + else + reg &= ~(1UL << channel); + + idmac_write_icreg(ipu, reg, IDMAC_CHA_PRI); + + dump_idmac_reg(ipu); +} + +static uint32_t ipu_channel_conf_mask(enum ipu_channel channel) +{ + uint32_t mask; + + switch (channel) { + case IDMAC_IC_0: + case IDMAC_IC_7: + mask = IPU_CONF_CSI_EN | IPU_CONF_IC_EN; + break; + case IDMAC_SDC_0: + case IDMAC_SDC_1: + mask = IPU_CONF_SDC_EN | IPU_CONF_DI_EN; + break; + default: + mask = 0; + break; + } + + return mask; +} + +/** + * ipu_enable_channel() - enable an IPU channel. + * @idmac: IPU DMAC context. + * @ichan: IDMAC channel. + * @return: 0 on success or negative error code on failure. + */ +static int ipu_enable_channel(struct idmac *idmac, struct idmac_channel *ichan) +{ + struct ipu *ipu = to_ipu(idmac); + enum ipu_channel channel = ichan->dma_chan.chan_id; + uint32_t reg; + unsigned long flags; + + spin_lock_irqsave(&ipu->lock, flags); + + /* Reset to buffer 0 */ + idmac_write_ipureg(ipu, 1UL << channel, IPU_CHA_CUR_BUF); + ichan->active_buffer = 0; + ichan->status = IPU_CHANNEL_ENABLED; + + switch (channel) { + case IDMAC_SDC_0: + case IDMAC_SDC_1: + case IDMAC_IC_7: + ipu_channel_set_priority(ipu, channel, true); + default: + break; + } + + reg = idmac_read_icreg(ipu, IDMAC_CHA_EN); + + idmac_write_icreg(ipu, reg | (1UL << channel), IDMAC_CHA_EN); + + ipu_ic_enable_task(ipu, channel); + + spin_unlock_irqrestore(&ipu->lock, flags); + return 0; +} + +/** + * ipu_init_channel_buffer() - initialize a buffer for logical IPU channel. + * @ichan: IDMAC channel. + * @pixel_fmt: pixel format of buffer. Pixel format is a FOURCC ASCII code. + * @width: width of buffer in pixels. + * @height: height of buffer in pixels. + * @stride: stride length of buffer in pixels. + * @rot_mode: rotation mode of buffer. A rotation setting other than + * IPU_ROTATE_VERT_FLIP should only be used for input buffers of + * rotation channels. + * @phyaddr_0: buffer 0 physical address. + * @phyaddr_1: buffer 1 physical address. Setting this to a value other than + * NULL enables double buffering mode. + * @return: 0 on success or negative error code on failure. + */ +static int ipu_init_channel_buffer(struct idmac_channel *ichan, + enum pixel_fmt pixel_fmt, + uint16_t width, uint16_t height, + uint32_t stride, + enum ipu_rotate_mode rot_mode, + dma_addr_t phyaddr_0, dma_addr_t phyaddr_1) +{ + enum ipu_channel channel = ichan->dma_chan.chan_id; + struct idmac *idmac = to_idmac(ichan->dma_chan.device); + struct ipu *ipu = to_ipu(idmac); + union chan_param_mem params = {}; + unsigned long flags; + uint32_t reg; + uint32_t stride_bytes; + + stride_bytes = stride * bytes_per_pixel(pixel_fmt); + + if (stride_bytes % 4) { + dev_err(ipu->dev, + "Stride length must be 32-bit aligned, stride = %d, bytes = %d\n", + stride, stride_bytes); + return -EINVAL; + } + + /* IC channel's stride must be a multiple of 8 pixels */ + if ((channel <= IDMAC_IC_13) && (stride % 8)) { + dev_err(ipu->dev, "Stride must be 8 pixel multiple\n"); + return -EINVAL; + } + + /* Build parameter memory data for DMA channel */ + ipu_ch_param_set_size(¶ms, pixel_fmt, width, height, stride_bytes); + ipu_ch_param_set_buffer(¶ms, phyaddr_0, phyaddr_1); + ipu_ch_param_set_rotation(¶ms, rot_mode); + /* Some channels (rotation) have restriction on burst length */ + switch (channel) { + case IDMAC_IC_7: /* Hangs with burst 8, 16, other values + invalid - Table 44-30 */ +/* + ipu_ch_param_set_burst_size(¶ms, 8); + */ + break; + case IDMAC_SDC_0: + case IDMAC_SDC_1: + /* In original code only IPU_PIX_FMT_RGB565 was setting burst */ + ipu_ch_param_set_burst_size(¶ms, 16); + break; + case IDMAC_IC_0: + default: + break; + } + + spin_lock_irqsave(&ipu->lock, flags); + + ipu_write_param_mem(dma_param_addr(channel), (uint32_t *)¶ms, 10); + + reg = idmac_read_ipureg(ipu, IPU_CHA_DB_MODE_SEL); + + if (phyaddr_1) + reg |= 1UL << channel; + else + reg &= ~(1UL << channel); + + idmac_write_ipureg(ipu, reg, IPU_CHA_DB_MODE_SEL); + + ichan->status = IPU_CHANNEL_READY; + + spin_unlock_irqrestore(&ipu->lock, flags); + + return 0; +} + +/** + * ipu_select_buffer() - mark a channel's buffer as ready. + * @channel: channel ID. + * @buffer_n: buffer number to mark ready. + */ +static void ipu_select_buffer(enum ipu_channel channel, int buffer_n) +{ + /* No locking - this is a write-one-to-set register, cleared by IPU */ + if (buffer_n == 0) + /* Mark buffer 0 as ready. */ + idmac_write_ipureg(&ipu_data, 1UL << channel, IPU_CHA_BUF0_RDY); + else + /* Mark buffer 1 as ready. */ + idmac_write_ipureg(&ipu_data, 1UL << channel, IPU_CHA_BUF1_RDY); +} + +/** + * ipu_update_channel_buffer() - update physical address of a channel buffer. + * @ichan: IDMAC channel. + * @buffer_n: buffer number to update. + * 0 or 1 are the only valid values. + * @phyaddr: buffer physical address. + */ +/* Called under spin_lock(_irqsave)(&ichan->lock) */ +static void ipu_update_channel_buffer(struct idmac_channel *ichan, + int buffer_n, dma_addr_t phyaddr) +{ + enum ipu_channel channel = ichan->dma_chan.chan_id; + uint32_t reg; + unsigned long flags; + + spin_lock_irqsave(&ipu_data.lock, flags); + + if (buffer_n == 0) { + reg = idmac_read_ipureg(&ipu_data, IPU_CHA_BUF0_RDY); + if (reg & (1UL << channel)) { + ipu_ic_disable_task(&ipu_data, channel); + ichan->status = IPU_CHANNEL_READY; + } + + /* 44.3.3.1.9 - Row Number 1 (WORD1, offset 0) */ + idmac_write_ipureg(&ipu_data, dma_param_addr(channel) + + 0x0008UL, IPU_IMA_ADDR); + idmac_write_ipureg(&ipu_data, phyaddr, IPU_IMA_DATA); + } else { + reg = idmac_read_ipureg(&ipu_data, IPU_CHA_BUF1_RDY); + if (reg & (1UL << channel)) { + ipu_ic_disable_task(&ipu_data, channel); + ichan->status = IPU_CHANNEL_READY; + } + + /* Check if double-buffering is already enabled */ + reg = idmac_read_ipureg(&ipu_data, IPU_CHA_DB_MODE_SEL); + + if (!(reg & (1UL << channel))) + idmac_write_ipureg(&ipu_data, reg | (1UL << channel), + IPU_CHA_DB_MODE_SEL); + + /* 44.3.3.1.9 - Row Number 1 (WORD1, offset 1) */ + idmac_write_ipureg(&ipu_data, dma_param_addr(channel) + + 0x0009UL, IPU_IMA_ADDR); + idmac_write_ipureg(&ipu_data, phyaddr, IPU_IMA_DATA); + } + + spin_unlock_irqrestore(&ipu_data.lock, flags); +} + +/* Called under spin_lock_irqsave(&ichan->lock) */ +static int ipu_submit_buffer(struct idmac_channel *ichan, + struct idmac_tx_desc *desc, struct scatterlist *sg, int buf_idx) +{ + unsigned int chan_id = ichan->dma_chan.chan_id; + struct device *dev = &ichan->dma_chan.dev->device; + + if (async_tx_test_ack(&desc->txd)) + return -EINTR; + + /* + * On first invocation this shouldn't be necessary, the call to + * ipu_init_channel_buffer() above will set addresses for us, so we + * could make it conditional on status >= IPU_CHANNEL_ENABLED, but + * doing it again shouldn't hurt either. + */ + ipu_update_channel_buffer(ichan, buf_idx, sg_dma_address(sg)); + + ipu_select_buffer(chan_id, buf_idx); + dev_dbg(dev, "Updated sg %p on channel 0x%x buffer %d\n", + sg, chan_id, buf_idx); + + return 0; +} + +/* Called under spin_lock_irqsave(&ichan->lock) */ +static int ipu_submit_channel_buffers(struct idmac_channel *ichan, + struct idmac_tx_desc *desc) +{ + struct scatterlist *sg; + int i, ret = 0; + + for (i = 0, sg = desc->sg; i < 2 && sg; i++) { + if (!ichan->sg[i]) { + ichan->sg[i] = sg; + + ret = ipu_submit_buffer(ichan, desc, sg, i); + if (ret < 0) + return ret; + + sg = sg_next(sg); + } + } + + return ret; +} + +static dma_cookie_t idmac_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct idmac_tx_desc *desc = to_tx_desc(tx); + struct idmac_channel *ichan = to_idmac_chan(tx->chan); + struct idmac *idmac = to_idmac(tx->chan->device); + struct ipu *ipu = to_ipu(idmac); + struct device *dev = &ichan->dma_chan.dev->device; + dma_cookie_t cookie; + unsigned long flags; + int ret; + + /* Sanity check */ + if (!list_empty(&desc->list)) { + /* The descriptor doesn't belong to client */ + dev_err(dev, "Descriptor %p not prepared!\n", tx); + return -EBUSY; + } + + mutex_lock(&ichan->chan_mutex); + + async_tx_clear_ack(tx); + + if (ichan->status < IPU_CHANNEL_READY) { + struct idmac_video_param *video = &ichan->params.video; + /* + * Initial buffer assignment - the first two sg-entries from + * the descriptor will end up in the IDMAC buffers + */ + dma_addr_t dma_1 = sg_is_last(desc->sg) ? 0 : + sg_dma_address(&desc->sg[1]); + + WARN_ON(ichan->sg[0] || ichan->sg[1]); + + cookie = ipu_init_channel_buffer(ichan, + video->out_pixel_fmt, + video->out_width, + video->out_height, + video->out_stride, + IPU_ROTATE_NONE, + sg_dma_address(&desc->sg[0]), + dma_1); + if (cookie < 0) + goto out; + } + + dev_dbg(dev, "Submitting sg %p\n", &desc->sg[0]); + + cookie = ichan->dma_chan.cookie; + + if (++cookie < 0) + cookie = 1; + + /* from dmaengine.h: "last cookie value returned to client" */ + ichan->dma_chan.cookie = cookie; + tx->cookie = cookie; + + /* ipu->lock can be taken under ichan->lock, but not v.v. */ + spin_lock_irqsave(&ichan->lock, flags); + + list_add_tail(&desc->list, &ichan->queue); + /* submit_buffers() atomically verifies and fills empty sg slots */ + ret = ipu_submit_channel_buffers(ichan, desc); + + spin_unlock_irqrestore(&ichan->lock, flags); + + if (ret < 0) { + cookie = ret; + goto dequeue; + } + + if (ichan->status < IPU_CHANNEL_ENABLED) { + ret = ipu_enable_channel(idmac, ichan); + if (ret < 0) { + cookie = ret; + goto dequeue; + } + } + + dump_idmac_reg(ipu); + +dequeue: + if (cookie < 0) { + spin_lock_irqsave(&ichan->lock, flags); + list_del_init(&desc->list); + spin_unlock_irqrestore(&ichan->lock, flags); + tx->cookie = cookie; + ichan->dma_chan.cookie = cookie; + } + +out: + mutex_unlock(&ichan->chan_mutex); + + return cookie; +} + +/* Called with ichan->chan_mutex held */ +static int idmac_desc_alloc(struct idmac_channel *ichan, int n) +{ + struct idmac_tx_desc *desc = vmalloc(n * sizeof(struct idmac_tx_desc)); + struct idmac *idmac = to_idmac(ichan->dma_chan.device); + + if (!desc) + return -ENOMEM; + + /* No interrupts, just disable the tasklet for a moment */ + tasklet_disable(&to_ipu(idmac)->tasklet); + + ichan->n_tx_desc = n; + ichan->desc = desc; + INIT_LIST_HEAD(&ichan->queue); + INIT_LIST_HEAD(&ichan->free_list); + + while (n--) { + struct dma_async_tx_descriptor *txd = &desc->txd; + + memset(txd, 0, sizeof(*txd)); + dma_async_tx_descriptor_init(txd, &ichan->dma_chan); + txd->tx_submit = idmac_tx_submit; + + list_add(&desc->list, &ichan->free_list); + + desc++; + } + + tasklet_enable(&to_ipu(idmac)->tasklet); + + return 0; +} + +/** + * ipu_init_channel() - initialize an IPU channel. + * @idmac: IPU DMAC context. + * @ichan: pointer to the channel object. + * @return 0 on success or negative error code on failure. + */ +static int ipu_init_channel(struct idmac *idmac, struct idmac_channel *ichan) +{ + union ipu_channel_param *params = &ichan->params; + uint32_t ipu_conf; + enum ipu_channel channel = ichan->dma_chan.chan_id; + unsigned long flags; + uint32_t reg; + struct ipu *ipu = to_ipu(idmac); + int ret = 0, n_desc = 0; + + dev_dbg(ipu->dev, "init channel = %d\n", channel); + + if (channel != IDMAC_SDC_0 && channel != IDMAC_SDC_1 && + channel != IDMAC_IC_7) + return -EINVAL; + + spin_lock_irqsave(&ipu->lock, flags); + + switch (channel) { + case IDMAC_IC_7: + n_desc = 16; + reg = idmac_read_icreg(ipu, IC_CONF); + idmac_write_icreg(ipu, reg & ~IC_CONF_CSI_MEM_WR_EN, IC_CONF); + break; + case IDMAC_IC_0: + n_desc = 16; + reg = idmac_read_ipureg(ipu, IPU_FS_PROC_FLOW); + idmac_write_ipureg(ipu, reg & ~FS_ENC_IN_VALID, IPU_FS_PROC_FLOW); + ret = ipu_ic_init_prpenc(ipu, params, true); + break; + case IDMAC_SDC_0: + case IDMAC_SDC_1: + n_desc = 4; + default: + break; + } + + ipu->channel_init_mask |= 1L << channel; + + /* Enable IPU sub module */ + ipu_conf = idmac_read_ipureg(ipu, IPU_CONF) | + ipu_channel_conf_mask(channel); + idmac_write_ipureg(ipu, ipu_conf, IPU_CONF); + + spin_unlock_irqrestore(&ipu->lock, flags); + + if (n_desc && !ichan->desc) + ret = idmac_desc_alloc(ichan, n_desc); + + dump_idmac_reg(ipu); + + return ret; +} + +/** + * ipu_uninit_channel() - uninitialize an IPU channel. + * @idmac: IPU DMAC context. + * @ichan: pointer to the channel object. + */ +static void ipu_uninit_channel(struct idmac *idmac, struct idmac_channel *ichan) +{ + enum ipu_channel channel = ichan->dma_chan.chan_id; + unsigned long flags; + uint32_t reg; + unsigned long chan_mask = 1UL << channel; + uint32_t ipu_conf; + struct ipu *ipu = to_ipu(idmac); + + spin_lock_irqsave(&ipu->lock, flags); + + if (!(ipu->channel_init_mask & chan_mask)) { + dev_err(ipu->dev, "Channel already uninitialized %d\n", + channel); + spin_unlock_irqrestore(&ipu->lock, flags); + return; + } + + /* Reset the double buffer */ + reg = idmac_read_ipureg(ipu, IPU_CHA_DB_MODE_SEL); + idmac_write_ipureg(ipu, reg & ~chan_mask, IPU_CHA_DB_MODE_SEL); + + ichan->sec_chan_en = false; + + switch (channel) { + case IDMAC_IC_7: + reg = idmac_read_icreg(ipu, IC_CONF); + idmac_write_icreg(ipu, reg & ~(IC_CONF_RWS_EN | IC_CONF_PRPENC_EN), + IC_CONF); + break; + case IDMAC_IC_0: + reg = idmac_read_icreg(ipu, IC_CONF); + idmac_write_icreg(ipu, reg & ~(IC_CONF_PRPENC_EN | IC_CONF_PRPENC_CSC1), + IC_CONF); + break; + case IDMAC_SDC_0: + case IDMAC_SDC_1: + default: + break; + } + + ipu->channel_init_mask &= ~(1L << channel); + + ipu_conf = idmac_read_ipureg(ipu, IPU_CONF) & + ~ipu_channel_conf_mask(channel); + idmac_write_ipureg(ipu, ipu_conf, IPU_CONF); + + spin_unlock_irqrestore(&ipu->lock, flags); + + ichan->n_tx_desc = 0; + vfree(ichan->desc); + ichan->desc = NULL; +} + +/** + * ipu_disable_channel() - disable an IPU channel. + * @idmac: IPU DMAC context. + * @ichan: channel object pointer. + * @wait_for_stop: flag to set whether to wait for channel end of frame or + * return immediately. + * @return: 0 on success or negative error code on failure. + */ +static int ipu_disable_channel(struct idmac *idmac, struct idmac_channel *ichan, + bool wait_for_stop) +{ + enum ipu_channel channel = ichan->dma_chan.chan_id; + struct ipu *ipu = to_ipu(idmac); + uint32_t reg; + unsigned long flags; + unsigned long chan_mask = 1UL << channel; + unsigned int timeout; + + if (wait_for_stop && channel != IDMAC_SDC_1 && channel != IDMAC_SDC_0) { + timeout = 40; + /* This waiting always fails. Related to spurious irq problem */ + while ((idmac_read_icreg(ipu, IDMAC_CHA_BUSY) & chan_mask) || + (ipu_channel_status(ipu, channel) == TASK_STAT_ACTIVE)) { + timeout--; + msleep(10); + + if (!timeout) { + dev_dbg(ipu->dev, + "Warning: timeout waiting for channel %u to " + "stop: buf0_rdy = 0x%08X, buf1_rdy = 0x%08X, " + "busy = 0x%08X, tstat = 0x%08X\n", channel, + idmac_read_ipureg(ipu, IPU_CHA_BUF0_RDY), + idmac_read_ipureg(ipu, IPU_CHA_BUF1_RDY), + idmac_read_icreg(ipu, IDMAC_CHA_BUSY), + idmac_read_ipureg(ipu, IPU_TASKS_STAT)); + break; + } + } + dev_dbg(ipu->dev, "timeout = %d * 10ms\n", 40 - timeout); + } + /* SDC BG and FG must be disabled before DMA is disabled */ + if (wait_for_stop && (channel == IDMAC_SDC_0 || + channel == IDMAC_SDC_1)) { + for (timeout = 5; + timeout && !ipu_irq_status(ichan->eof_irq); timeout--) + msleep(5); + } + + spin_lock_irqsave(&ipu->lock, flags); + + /* Disable IC task */ + ipu_ic_disable_task(ipu, channel); + + /* Disable DMA channel(s) */ + reg = idmac_read_icreg(ipu, IDMAC_CHA_EN); + idmac_write_icreg(ipu, reg & ~chan_mask, IDMAC_CHA_EN); + + /* + * Problem (observed with channel DMAIC_7): after enabling the channel + * and initialising buffers, there comes an interrupt with current still + * pointing at buffer 0, whereas it should use buffer 0 first and only + * generate an interrupt when it is done, then current should already + * point to buffer 1. This spurious interrupt also comes on channel + * DMASDC_0. With DMAIC_7 normally, is we just leave the ISR after the + * first interrupt, there comes the second with current correctly + * pointing to buffer 1 this time. But sometimes this second interrupt + * doesn't come and the channel hangs. Clearing BUFx_RDY when disabling + * the channel seems to prevent the channel from hanging, but it doesn't + * prevent the spurious interrupt. This might also be unsafe. Think + * about the IDMAC controller trying to switch to a buffer, when we + * clear the ready bit, and re-enable it a moment later. + */ + reg = idmac_read_ipureg(ipu, IPU_CHA_BUF0_RDY); + idmac_write_ipureg(ipu, 0, IPU_CHA_BUF0_RDY); + idmac_write_ipureg(ipu, reg & ~(1UL << channel), IPU_CHA_BUF0_RDY); + + reg = idmac_read_ipureg(ipu, IPU_CHA_BUF1_RDY); + idmac_write_ipureg(ipu, 0, IPU_CHA_BUF1_RDY); + idmac_write_ipureg(ipu, reg & ~(1UL << channel), IPU_CHA_BUF1_RDY); + + spin_unlock_irqrestore(&ipu->lock, flags); + + return 0; +} + +static struct scatterlist *idmac_sg_next(struct idmac_channel *ichan, + struct idmac_tx_desc **desc, struct scatterlist *sg) +{ + struct scatterlist *sgnew = sg ? sg_next(sg) : NULL; + + if (sgnew) + /* next sg-element in this list */ + return sgnew; + + if ((*desc)->list.next == &ichan->queue) + /* No more descriptors on the queue */ + return NULL; + + /* Fetch next descriptor */ + *desc = list_entry((*desc)->list.next, struct idmac_tx_desc, list); + return (*desc)->sg; +} + +/* + * We have several possibilities here: + * current BUF next BUF + * + * not last sg next not last sg + * not last sg next last sg + * last sg first sg from next descriptor + * last sg NULL + * + * Besides, the descriptor queue might be empty or not. We process all these + * cases carefully. + */ +static irqreturn_t idmac_interrupt(int irq, void *dev_id) +{ + struct idmac_channel *ichan = dev_id; + struct device *dev = &ichan->dma_chan.dev->device; + unsigned int chan_id = ichan->dma_chan.chan_id; + struct scatterlist **sg, *sgnext, *sgnew = NULL; + /* Next transfer descriptor */ + struct idmac_tx_desc *desc, *descnew; + dma_async_tx_callback callback; + void *callback_param; + bool done = false; + u32 ready0, ready1, curbuf, err; + unsigned long flags; + + /* IDMAC has cleared the respective BUFx_RDY bit, we manage the buffer */ + + dev_dbg(dev, "IDMAC irq %d, buf %d\n", irq, ichan->active_buffer); + + spin_lock_irqsave(&ipu_data.lock, flags); + + ready0 = idmac_read_ipureg(&ipu_data, IPU_CHA_BUF0_RDY); + ready1 = idmac_read_ipureg(&ipu_data, IPU_CHA_BUF1_RDY); + curbuf = idmac_read_ipureg(&ipu_data, IPU_CHA_CUR_BUF); + err = idmac_read_ipureg(&ipu_data, IPU_INT_STAT_4); + + if (err & (1 << chan_id)) { + idmac_write_ipureg(&ipu_data, 1 << chan_id, IPU_INT_STAT_4); + spin_unlock_irqrestore(&ipu_data.lock, flags); + /* + * Doing this + * ichan->sg[0] = ichan->sg[1] = NULL; + * you can force channel re-enable on the next tx_submit(), but + * this is dirty - think about descriptors with multiple + * sg elements. + */ + dev_warn(dev, "NFB4EOF on channel %d, ready %x, %x, cur %x\n", + chan_id, ready0, ready1, curbuf); + return IRQ_HANDLED; + } + spin_unlock_irqrestore(&ipu_data.lock, flags); + + /* Other interrupts do not interfere with this channel */ + spin_lock(&ichan->lock); + if (unlikely(chan_id != IDMAC_SDC_0 && chan_id != IDMAC_SDC_1 && + ((curbuf >> chan_id) & 1) == ichan->active_buffer && + !list_is_last(ichan->queue.next, &ichan->queue))) { + int i = 100; + + /* This doesn't help. See comment in ipu_disable_channel() */ + while (--i) { + curbuf = idmac_read_ipureg(&ipu_data, IPU_CHA_CUR_BUF); + if (((curbuf >> chan_id) & 1) != ichan->active_buffer) + break; + cpu_relax(); + } + + if (!i) { + spin_unlock(&ichan->lock); + dev_dbg(dev, + "IRQ on active buffer on channel %x, active " + "%d, ready %x, %x, current %x!\n", chan_id, + ichan->active_buffer, ready0, ready1, curbuf); + return IRQ_NONE; + } else + dev_dbg(dev, + "Buffer deactivated on channel %x, active " + "%d, ready %x, %x, current %x, rest %d!\n", chan_id, + ichan->active_buffer, ready0, ready1, curbuf, i); + } + + if (unlikely((ichan->active_buffer && (ready1 >> chan_id) & 1) || + (!ichan->active_buffer && (ready0 >> chan_id) & 1) + )) { + spin_unlock(&ichan->lock); + dev_dbg(dev, + "IRQ with active buffer still ready on channel %x, " + "active %d, ready %x, %x!\n", chan_id, + ichan->active_buffer, ready0, ready1); + return IRQ_NONE; + } + + if (unlikely(list_empty(&ichan->queue))) { + ichan->sg[ichan->active_buffer] = NULL; + spin_unlock(&ichan->lock); + dev_err(dev, + "IRQ without queued buffers on channel %x, active %d, " + "ready %x, %x!\n", chan_id, + ichan->active_buffer, ready0, ready1); + return IRQ_NONE; + } + + /* + * active_buffer is a software flag, it shows which buffer we are + * currently expecting back from the hardware, IDMAC should be + * processing the other buffer already + */ + sg = &ichan->sg[ichan->active_buffer]; + sgnext = ichan->sg[!ichan->active_buffer]; + + if (!*sg) { + spin_unlock(&ichan->lock); + return IRQ_HANDLED; + } + + desc = list_entry(ichan->queue.next, struct idmac_tx_desc, list); + descnew = desc; + + dev_dbg(dev, "IDMAC irq %d, dma 0x%08x, next dma 0x%08x, current %d, curbuf 0x%08x\n", + irq, sg_dma_address(*sg), sgnext ? sg_dma_address(sgnext) : 0, ichan->active_buffer, curbuf); + + /* Find the descriptor of sgnext */ + sgnew = idmac_sg_next(ichan, &descnew, *sg); + if (sgnext != sgnew) + dev_err(dev, "Submitted buffer %p, next buffer %p\n", sgnext, sgnew); + + /* + * if sgnext == NULL sg must be the last element in a scatterlist and + * queue must be empty + */ + if (unlikely(!sgnext)) { + if (!WARN_ON(sg_next(*sg))) + dev_dbg(dev, "Underrun on channel %x\n", chan_id); + ichan->sg[!ichan->active_buffer] = sgnew; + + if (unlikely(sgnew)) { + ipu_submit_buffer(ichan, descnew, sgnew, !ichan->active_buffer); + } else { + spin_lock_irqsave(&ipu_data.lock, flags); + ipu_ic_disable_task(&ipu_data, chan_id); + spin_unlock_irqrestore(&ipu_data.lock, flags); + ichan->status = IPU_CHANNEL_READY; + /* Continue to check for complete descriptor */ + } + } + + /* Calculate and submit the next sg element */ + sgnew = idmac_sg_next(ichan, &descnew, sgnew); + + if (unlikely(!sg_next(*sg)) || !sgnext) { + /* + * Last element in scatterlist done, remove from the queue, + * _init for debugging + */ + list_del_init(&desc->list); + done = true; + } + + *sg = sgnew; + + if (likely(sgnew) && + ipu_submit_buffer(ichan, descnew, sgnew, ichan->active_buffer) < 0) { + callback = descnew->txd.callback; + callback_param = descnew->txd.callback_param; + spin_unlock(&ichan->lock); + if (callback) + callback(callback_param); + spin_lock(&ichan->lock); + } + + /* Flip the active buffer - even if update above failed */ + ichan->active_buffer = !ichan->active_buffer; + if (done) + ichan->completed = desc->txd.cookie; + + callback = desc->txd.callback; + callback_param = desc->txd.callback_param; + + spin_unlock(&ichan->lock); + + if (done && (desc->txd.flags & DMA_PREP_INTERRUPT) && callback) + callback(callback_param); + + return IRQ_HANDLED; +} + +static void ipu_gc_tasklet(unsigned long arg) +{ + struct ipu *ipu = (struct ipu *)arg; + int i; + + for (i = 0; i < IPU_CHANNELS_NUM; i++) { + struct idmac_channel *ichan = ipu->channel + i; + struct idmac_tx_desc *desc; + unsigned long flags; + struct scatterlist *sg; + int j, k; + + for (j = 0; j < ichan->n_tx_desc; j++) { + desc = ichan->desc + j; + spin_lock_irqsave(&ichan->lock, flags); + if (async_tx_test_ack(&desc->txd)) { + list_move(&desc->list, &ichan->free_list); + for_each_sg(desc->sg, sg, desc->sg_len, k) { + if (ichan->sg[0] == sg) + ichan->sg[0] = NULL; + else if (ichan->sg[1] == sg) + ichan->sg[1] = NULL; + } + async_tx_clear_ack(&desc->txd); + } + spin_unlock_irqrestore(&ichan->lock, flags); + } + } +} + +/* Allocate and initialise a transfer descriptor. */ +static struct dma_async_tx_descriptor *idmac_prep_slave_sg(struct dma_chan *chan, + struct scatterlist *sgl, unsigned int sg_len, + enum dma_data_direction direction, unsigned long tx_flags) +{ + struct idmac_channel *ichan = to_idmac_chan(chan); + struct idmac_tx_desc *desc = NULL; + struct dma_async_tx_descriptor *txd = NULL; + unsigned long flags; + + /* We only can handle these three channels so far */ + if (chan->chan_id != IDMAC_SDC_0 && chan->chan_id != IDMAC_SDC_1 && + chan->chan_id != IDMAC_IC_7) + return NULL; + + if (direction != DMA_FROM_DEVICE && direction != DMA_TO_DEVICE) { + dev_err(chan->device->dev, "Invalid DMA direction %d!\n", direction); + return NULL; + } + + mutex_lock(&ichan->chan_mutex); + + spin_lock_irqsave(&ichan->lock, flags); + if (!list_empty(&ichan->free_list)) { + desc = list_entry(ichan->free_list.next, + struct idmac_tx_desc, list); + + list_del_init(&desc->list); + + desc->sg_len = sg_len; + desc->sg = sgl; + txd = &desc->txd; + txd->flags = tx_flags; + } + spin_unlock_irqrestore(&ichan->lock, flags); + + mutex_unlock(&ichan->chan_mutex); + + tasklet_schedule(&to_ipu(to_idmac(chan->device))->tasklet); + + return txd; +} + +/* Re-select the current buffer and re-activate the channel */ +static void idmac_issue_pending(struct dma_chan *chan) +{ + struct idmac_channel *ichan = to_idmac_chan(chan); + struct idmac *idmac = to_idmac(chan->device); + struct ipu *ipu = to_ipu(idmac); + unsigned long flags; + + /* This is not always needed, but doesn't hurt either */ + spin_lock_irqsave(&ipu->lock, flags); + ipu_select_buffer(chan->chan_id, ichan->active_buffer); + spin_unlock_irqrestore(&ipu->lock, flags); + + /* + * Might need to perform some parts of initialisation from + * ipu_enable_channel(), but not all, we do not want to reset to buffer + * 0, don't need to set priority again either, but re-enabling the task + * and the channel might be a good idea. + */ +} + +static void __idmac_terminate_all(struct dma_chan *chan) +{ + struct idmac_channel *ichan = to_idmac_chan(chan); + struct idmac *idmac = to_idmac(chan->device); + unsigned long flags; + int i; + + ipu_disable_channel(idmac, ichan, + ichan->status >= IPU_CHANNEL_ENABLED); + + tasklet_disable(&to_ipu(idmac)->tasklet); + + /* ichan->queue is modified in ISR, have to spinlock */ + spin_lock_irqsave(&ichan->lock, flags); + list_splice_init(&ichan->queue, &ichan->free_list); + + if (ichan->desc) + for (i = 0; i < ichan->n_tx_desc; i++) { + struct idmac_tx_desc *desc = ichan->desc + i; + if (list_empty(&desc->list)) + /* Descriptor was prepared, but not submitted */ + list_add(&desc->list, &ichan->free_list); + + async_tx_clear_ack(&desc->txd); + } + + ichan->sg[0] = NULL; + ichan->sg[1] = NULL; + spin_unlock_irqrestore(&ichan->lock, flags); + + tasklet_enable(&to_ipu(idmac)->tasklet); + + ichan->status = IPU_CHANNEL_INITIALIZED; +} + +static void idmac_terminate_all(struct dma_chan *chan) +{ + struct idmac_channel *ichan = to_idmac_chan(chan); + + mutex_lock(&ichan->chan_mutex); + + __idmac_terminate_all(chan); + + mutex_unlock(&ichan->chan_mutex); +} + +#ifdef DEBUG +static irqreturn_t ic_sof_irq(int irq, void *dev_id) +{ + struct idmac_channel *ichan = dev_id; + printk(KERN_DEBUG "Got SOF IRQ %d on Channel %d\n", + irq, ichan->dma_chan.chan_id); + disable_irq_nosync(irq); + return IRQ_HANDLED; +} + +static irqreturn_t ic_eof_irq(int irq, void *dev_id) +{ + struct idmac_channel *ichan = dev_id; + printk(KERN_DEBUG "Got EOF IRQ %d on Channel %d\n", + irq, ichan->dma_chan.chan_id); + disable_irq_nosync(irq); + return IRQ_HANDLED; +} + +static int ic_sof = -EINVAL, ic_eof = -EINVAL; +#endif + +static int idmac_alloc_chan_resources(struct dma_chan *chan) +{ + struct idmac_channel *ichan = to_idmac_chan(chan); + struct idmac *idmac = to_idmac(chan->device); + int ret; + + /* dmaengine.c now guarantees to only offer free channels */ + BUG_ON(chan->client_count > 1); + WARN_ON(ichan->status != IPU_CHANNEL_FREE); + + chan->cookie = 1; + ichan->completed = -ENXIO; + + ret = ipu_irq_map(chan->chan_id); + if (ret < 0) + goto eimap; + + ichan->eof_irq = ret; + + /* + * Important to first disable the channel, because maybe someone + * used it before us, e.g., the bootloader + */ + ipu_disable_channel(idmac, ichan, true); + + ret = ipu_init_channel(idmac, ichan); + if (ret < 0) + goto eichan; + + ret = request_irq(ichan->eof_irq, idmac_interrupt, 0, + ichan->eof_name, ichan); + if (ret < 0) + goto erirq; + +#ifdef DEBUG + if (chan->chan_id == IDMAC_IC_7) { + ic_sof = ipu_irq_map(69); + if (ic_sof > 0) + request_irq(ic_sof, ic_sof_irq, 0, "IC SOF", ichan); + ic_eof = ipu_irq_map(70); + if (ic_eof > 0) + request_irq(ic_eof, ic_eof_irq, 0, "IC EOF", ichan); + } +#endif + + ichan->status = IPU_CHANNEL_INITIALIZED; + + dev_dbg(&chan->dev->device, "Found channel 0x%x, irq %d\n", + chan->chan_id, ichan->eof_irq); + + return ret; + +erirq: + ipu_uninit_channel(idmac, ichan); +eichan: + ipu_irq_unmap(chan->chan_id); +eimap: + return ret; +} + +static void idmac_free_chan_resources(struct dma_chan *chan) +{ + struct idmac_channel *ichan = to_idmac_chan(chan); + struct idmac *idmac = to_idmac(chan->device); + + mutex_lock(&ichan->chan_mutex); + + __idmac_terminate_all(chan); + + if (ichan->status > IPU_CHANNEL_FREE) { +#ifdef DEBUG + if (chan->chan_id == IDMAC_IC_7) { + if (ic_sof > 0) { + free_irq(ic_sof, ichan); + ipu_irq_unmap(69); + ic_sof = -EINVAL; + } + if (ic_eof > 0) { + free_irq(ic_eof, ichan); + ipu_irq_unmap(70); + ic_eof = -EINVAL; + } + } +#endif + free_irq(ichan->eof_irq, ichan); + ipu_irq_unmap(chan->chan_id); + } + + ichan->status = IPU_CHANNEL_FREE; + + ipu_uninit_channel(idmac, ichan); + + mutex_unlock(&ichan->chan_mutex); + + tasklet_schedule(&to_ipu(idmac)->tasklet); +} + +static enum dma_status idmac_is_tx_complete(struct dma_chan *chan, + dma_cookie_t cookie, dma_cookie_t *done, dma_cookie_t *used) +{ + struct idmac_channel *ichan = to_idmac_chan(chan); + + if (done) + *done = ichan->completed; + if (used) + *used = chan->cookie; + if (cookie != chan->cookie) + return DMA_ERROR; + return DMA_SUCCESS; +} + +static int __init ipu_idmac_init(struct ipu *ipu) +{ + struct idmac *idmac = &ipu->idmac; + struct dma_device *dma = &idmac->dma; + int i; + + dma_cap_set(DMA_SLAVE, dma->cap_mask); + dma_cap_set(DMA_PRIVATE, dma->cap_mask); + + /* Compulsory common fields */ + dma->dev = ipu->dev; + dma->device_alloc_chan_resources = idmac_alloc_chan_resources; + dma->device_free_chan_resources = idmac_free_chan_resources; + dma->device_is_tx_complete = idmac_is_tx_complete; + dma->device_issue_pending = idmac_issue_pending; + + /* Compulsory for DMA_SLAVE fields */ + dma->device_prep_slave_sg = idmac_prep_slave_sg; + dma->device_terminate_all = idmac_terminate_all; + + INIT_LIST_HEAD(&dma->channels); + for (i = 0; i < IPU_CHANNELS_NUM; i++) { + struct idmac_channel *ichan = ipu->channel + i; + struct dma_chan *dma_chan = &ichan->dma_chan; + + spin_lock_init(&ichan->lock); + mutex_init(&ichan->chan_mutex); + + ichan->status = IPU_CHANNEL_FREE; + ichan->sec_chan_en = false; + ichan->completed = -ENXIO; + snprintf(ichan->eof_name, sizeof(ichan->eof_name), "IDMAC EOF %d", i); + + dma_chan->device = &idmac->dma; + dma_chan->cookie = 1; + dma_chan->chan_id = i; + list_add_tail(&dma_chan->device_node, &dma->channels); + } + + idmac_write_icreg(ipu, 0x00000070, IDMAC_CONF); + + return dma_async_device_register(&idmac->dma); +} + +static void __exit ipu_idmac_exit(struct ipu *ipu) +{ + int i; + struct idmac *idmac = &ipu->idmac; + + for (i = 0; i < IPU_CHANNELS_NUM; i++) { + struct idmac_channel *ichan = ipu->channel + i; + + idmac_terminate_all(&ichan->dma_chan); + idmac_prep_slave_sg(&ichan->dma_chan, NULL, 0, DMA_NONE, 0); + } + + dma_async_device_unregister(&idmac->dma); +} + +/***************************************************************************** + * IPU common probe / remove + */ + +static int __init ipu_probe(struct platform_device *pdev) +{ + struct ipu_platform_data *pdata = pdev->dev.platform_data; + struct resource *mem_ipu, *mem_ic; + int ret; + + spin_lock_init(&ipu_data.lock); + + mem_ipu = platform_get_resource(pdev, IORESOURCE_MEM, 0); + mem_ic = platform_get_resource(pdev, IORESOURCE_MEM, 1); + if (!pdata || !mem_ipu || !mem_ic) + return -EINVAL; + + ipu_data.dev = &pdev->dev; + + platform_set_drvdata(pdev, &ipu_data); + + ret = platform_get_irq(pdev, 0); + if (ret < 0) + goto err_noirq; + + ipu_data.irq_fn = ret; + ret = platform_get_irq(pdev, 1); + if (ret < 0) + goto err_noirq; + + ipu_data.irq_err = ret; + ipu_data.irq_base = pdata->irq_base; + + dev_dbg(&pdev->dev, "fn irq %u, err irq %u, irq-base %u\n", + ipu_data.irq_fn, ipu_data.irq_err, ipu_data.irq_base); + + /* Remap IPU common registers */ + ipu_data.reg_ipu = ioremap(mem_ipu->start, + mem_ipu->end - mem_ipu->start + 1); + if (!ipu_data.reg_ipu) { + ret = -ENOMEM; + goto err_ioremap_ipu; + } + + /* Remap Image Converter and Image DMA Controller registers */ + ipu_data.reg_ic = ioremap(mem_ic->start, + mem_ic->end - mem_ic->start + 1); + if (!ipu_data.reg_ic) { + ret = -ENOMEM; + goto err_ioremap_ic; + } + + /* Get IPU clock */ + ipu_data.ipu_clk = clk_get(&pdev->dev, NULL); + if (IS_ERR(ipu_data.ipu_clk)) { + ret = PTR_ERR(ipu_data.ipu_clk); + goto err_clk_get; + } + + /* Make sure IPU HSP clock is running */ + clk_enable(ipu_data.ipu_clk); + + /* Disable all interrupts */ + idmac_write_ipureg(&ipu_data, 0, IPU_INT_CTRL_1); + idmac_write_ipureg(&ipu_data, 0, IPU_INT_CTRL_2); + idmac_write_ipureg(&ipu_data, 0, IPU_INT_CTRL_3); + idmac_write_ipureg(&ipu_data, 0, IPU_INT_CTRL_4); + idmac_write_ipureg(&ipu_data, 0, IPU_INT_CTRL_5); + + dev_dbg(&pdev->dev, "%s @ 0x%08lx, fn irq %u, err irq %u\n", pdev->name, + (unsigned long)mem_ipu->start, ipu_data.irq_fn, ipu_data.irq_err); + + ret = ipu_irq_attach_irq(&ipu_data, pdev); + if (ret < 0) + goto err_attach_irq; + + /* Initialize DMA engine */ + ret = ipu_idmac_init(&ipu_data); + if (ret < 0) + goto err_idmac_init; + + tasklet_init(&ipu_data.tasklet, ipu_gc_tasklet, (unsigned long)&ipu_data); + + ipu_data.dev = &pdev->dev; + + dev_dbg(ipu_data.dev, "IPU initialized\n"); + + return 0; + +err_idmac_init: +err_attach_irq: + ipu_irq_detach_irq(&ipu_data, pdev); + clk_disable(ipu_data.ipu_clk); + clk_put(ipu_data.ipu_clk); +err_clk_get: + iounmap(ipu_data.reg_ic); +err_ioremap_ic: + iounmap(ipu_data.reg_ipu); +err_ioremap_ipu: +err_noirq: + dev_err(&pdev->dev, "Failed to probe IPU: %d\n", ret); + return ret; +} + +static int __exit ipu_remove(struct platform_device *pdev) +{ + struct ipu *ipu = platform_get_drvdata(pdev); + + ipu_idmac_exit(ipu); + ipu_irq_detach_irq(ipu, pdev); + clk_disable(ipu->ipu_clk); + clk_put(ipu->ipu_clk); + iounmap(ipu->reg_ic); + iounmap(ipu->reg_ipu); + tasklet_kill(&ipu->tasklet); + platform_set_drvdata(pdev, NULL); + + return 0; +} + +/* + * We need two MEM resources - with IPU-common and Image Converter registers, + * including PF_CONF and IDMAC_* registers, and two IRQs - function and error + */ +static struct platform_driver ipu_platform_driver = { + .driver = { + .name = "ipu-core", + .owner = THIS_MODULE, + }, + .remove = __exit_p(ipu_remove), +}; + +static int __init ipu_init(void) +{ + return platform_driver_probe(&ipu_platform_driver, ipu_probe); +} +subsys_initcall(ipu_init); + +MODULE_DESCRIPTION("IPU core driver"); +MODULE_LICENSE("GPL v2"); +MODULE_AUTHOR("Guennadi Liakhovetski <lg@denx.de>"); +MODULE_ALIAS("platform:ipu-core"); diff --git a/drivers/dma/ipu/ipu_intern.h b/drivers/dma/ipu/ipu_intern.h new file mode 100644 index 00000000000..545cf11a94a --- /dev/null +++ b/drivers/dma/ipu/ipu_intern.h @@ -0,0 +1,176 @@ +/* + * Copyright (C) 2008 + * Guennadi Liakhovetski, DENX Software Engineering, <lg@denx.de> + * + * Copyright (C) 2005-2007 Freescale Semiconductor, Inc. All Rights Reserved. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#ifndef _IPU_INTERN_H_ +#define _IPU_INTERN_H_ + +#include <linux/dmaengine.h> +#include <linux/platform_device.h> +#include <linux/interrupt.h> + +/* IPU Common registers */ +#define IPU_CONF 0x00 +#define IPU_CHA_BUF0_RDY 0x04 +#define IPU_CHA_BUF1_RDY 0x08 +#define IPU_CHA_DB_MODE_SEL 0x0C +#define IPU_CHA_CUR_BUF 0x10 +#define IPU_FS_PROC_FLOW 0x14 +#define IPU_FS_DISP_FLOW 0x18 +#define IPU_TASKS_STAT 0x1C +#define IPU_IMA_ADDR 0x20 +#define IPU_IMA_DATA 0x24 +#define IPU_INT_CTRL_1 0x28 +#define IPU_INT_CTRL_2 0x2C +#define IPU_INT_CTRL_3 0x30 +#define IPU_INT_CTRL_4 0x34 +#define IPU_INT_CTRL_5 0x38 +#define IPU_INT_STAT_1 0x3C +#define IPU_INT_STAT_2 0x40 +#define IPU_INT_STAT_3 0x44 +#define IPU_INT_STAT_4 0x48 +#define IPU_INT_STAT_5 0x4C +#define IPU_BRK_CTRL_1 0x50 +#define IPU_BRK_CTRL_2 0x54 +#define IPU_BRK_STAT 0x58 +#define IPU_DIAGB_CTRL 0x5C + +/* IPU_CONF Register bits */ +#define IPU_CONF_CSI_EN 0x00000001 +#define IPU_CONF_IC_EN 0x00000002 +#define IPU_CONF_ROT_EN 0x00000004 +#define IPU_CONF_PF_EN 0x00000008 +#define IPU_CONF_SDC_EN 0x00000010 +#define IPU_CONF_ADC_EN 0x00000020 +#define IPU_CONF_DI_EN 0x00000040 +#define IPU_CONF_DU_EN 0x00000080 +#define IPU_CONF_PXL_ENDIAN 0x00000100 + +/* Image Converter Registers */ +#define IC_CONF 0x88 +#define IC_PRP_ENC_RSC 0x8C +#define IC_PRP_VF_RSC 0x90 +#define IC_PP_RSC 0x94 +#define IC_CMBP_1 0x98 +#define IC_CMBP_2 0x9C +#define PF_CONF 0xA0 +#define IDMAC_CONF 0xA4 +#define IDMAC_CHA_EN 0xA8 +#define IDMAC_CHA_PRI 0xAC +#define IDMAC_CHA_BUSY 0xB0 + +/* Image Converter Register bits */ +#define IC_CONF_PRPENC_EN 0x00000001 +#define IC_CONF_PRPENC_CSC1 0x00000002 +#define IC_CONF_PRPENC_ROT_EN 0x00000004 +#define IC_CONF_PRPVF_EN 0x00000100 +#define IC_CONF_PRPVF_CSC1 0x00000200 +#define IC_CONF_PRPVF_CSC2 0x00000400 +#define IC_CONF_PRPVF_CMB 0x00000800 +#define IC_CONF_PRPVF_ROT_EN 0x00001000 +#define IC_CONF_PP_EN 0x00010000 +#define IC_CONF_PP_CSC1 0x00020000 +#define IC_CONF_PP_CSC2 0x00040000 +#define IC_CONF_PP_CMB 0x00080000 +#define IC_CONF_PP_ROT_EN 0x00100000 +#define IC_CONF_IC_GLB_LOC_A 0x10000000 +#define IC_CONF_KEY_COLOR_EN 0x20000000 +#define IC_CONF_RWS_EN 0x40000000 +#define IC_CONF_CSI_MEM_WR_EN 0x80000000 + +#define IDMA_CHAN_INVALID 0x000000FF +#define IDMA_IC_0 0x00000001 +#define IDMA_IC_1 0x00000002 +#define IDMA_IC_2 0x00000004 +#define IDMA_IC_3 0x00000008 +#define IDMA_IC_4 0x00000010 +#define IDMA_IC_5 0x00000020 +#define IDMA_IC_6 0x00000040 +#define IDMA_IC_7 0x00000080 +#define IDMA_IC_8 0x00000100 +#define IDMA_IC_9 0x00000200 +#define IDMA_IC_10 0x00000400 +#define IDMA_IC_11 0x00000800 +#define IDMA_IC_12 0x00001000 +#define IDMA_IC_13 0x00002000 +#define IDMA_SDC_BG 0x00004000 +#define IDMA_SDC_FG 0x00008000 +#define IDMA_SDC_MASK 0x00010000 +#define IDMA_SDC_PARTIAL 0x00020000 +#define IDMA_ADC_SYS1_WR 0x00040000 +#define IDMA_ADC_SYS2_WR 0x00080000 +#define IDMA_ADC_SYS1_CMD 0x00100000 +#define IDMA_ADC_SYS2_CMD 0x00200000 +#define IDMA_ADC_SYS1_RD 0x00400000 +#define IDMA_ADC_SYS2_RD 0x00800000 +#define IDMA_PF_QP 0x01000000 +#define IDMA_PF_BSP 0x02000000 +#define IDMA_PF_Y_IN 0x04000000 +#define IDMA_PF_U_IN 0x08000000 +#define IDMA_PF_V_IN 0x10000000 +#define IDMA_PF_Y_OUT 0x20000000 +#define IDMA_PF_U_OUT 0x40000000 +#define IDMA_PF_V_OUT 0x80000000 + +#define TSTAT_PF_H264_PAUSE 0x00000001 +#define TSTAT_CSI2MEM_MASK 0x0000000C +#define TSTAT_CSI2MEM_OFFSET 2 +#define TSTAT_VF_MASK 0x00000600 +#define TSTAT_VF_OFFSET 9 +#define TSTAT_VF_ROT_MASK 0x000C0000 +#define TSTAT_VF_ROT_OFFSET 18 +#define TSTAT_ENC_MASK 0x00000180 +#define TSTAT_ENC_OFFSET 7 +#define TSTAT_ENC_ROT_MASK 0x00030000 +#define TSTAT_ENC_ROT_OFFSET 16 +#define TSTAT_PP_MASK 0x00001800 +#define TSTAT_PP_OFFSET 11 +#define TSTAT_PP_ROT_MASK 0x00300000 +#define TSTAT_PP_ROT_OFFSET 20 +#define TSTAT_PF_MASK 0x00C00000 +#define TSTAT_PF_OFFSET 22 +#define TSTAT_ADCSYS1_MASK 0x03000000 +#define TSTAT_ADCSYS1_OFFSET 24 +#define TSTAT_ADCSYS2_MASK 0x0C000000 +#define TSTAT_ADCSYS2_OFFSET 26 + +#define TASK_STAT_IDLE 0 +#define TASK_STAT_ACTIVE 1 +#define TASK_STAT_WAIT4READY 2 + +struct idmac { + struct dma_device dma; +}; + +struct ipu { + void __iomem *reg_ipu; + void __iomem *reg_ic; + unsigned int irq_fn; /* IPU Function IRQ to the CPU */ + unsigned int irq_err; /* IPU Error IRQ to the CPU */ + unsigned int irq_base; /* Beginning of the IPU IRQ range */ + unsigned long channel_init_mask; + spinlock_t lock; + struct clk *ipu_clk; + struct device *dev; + struct idmac idmac; + struct idmac_channel channel[IPU_CHANNELS_NUM]; + struct tasklet_struct tasklet; +}; + +#define to_idmac(d) container_of(d, struct idmac, dma) + +extern int ipu_irq_attach_irq(struct ipu *ipu, struct platform_device *dev); +extern void ipu_irq_detach_irq(struct ipu *ipu, struct platform_device *dev); + +extern bool ipu_irq_status(uint32_t irq); +extern int ipu_irq_map(unsigned int source); +extern int ipu_irq_unmap(unsigned int source); + +#endif diff --git a/drivers/dma/ipu/ipu_irq.c b/drivers/dma/ipu/ipu_irq.c new file mode 100644 index 00000000000..dd8ebc75b66 --- /dev/null +++ b/drivers/dma/ipu/ipu_irq.c @@ -0,0 +1,413 @@ +/* + * Copyright (C) 2008 + * Guennadi Liakhovetski, DENX Software Engineering, <lg@denx.de> + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include <linux/init.h> +#include <linux/err.h> +#include <linux/spinlock.h> +#include <linux/delay.h> +#include <linux/clk.h> +#include <linux/irq.h> +#include <linux/io.h> + +#include <mach/ipu.h> + +#include "ipu_intern.h" + +/* + * Register read / write - shall be inlined by the compiler + */ +static u32 ipu_read_reg(struct ipu *ipu, unsigned long reg) +{ + return __raw_readl(ipu->reg_ipu + reg); +} + +static void ipu_write_reg(struct ipu *ipu, u32 value, unsigned long reg) +{ + __raw_writel(value, ipu->reg_ipu + reg); +} + + +/* + * IPU IRQ chip driver + */ + +#define IPU_IRQ_NR_FN_BANKS 3 +#define IPU_IRQ_NR_ERR_BANKS 2 +#define IPU_IRQ_NR_BANKS (IPU_IRQ_NR_FN_BANKS + IPU_IRQ_NR_ERR_BANKS) + +struct ipu_irq_bank { + unsigned int control; + unsigned int status; + spinlock_t lock; + struct ipu *ipu; +}; + +static struct ipu_irq_bank irq_bank[IPU_IRQ_NR_BANKS] = { + /* 3 groups of functional interrupts */ + { + .control = IPU_INT_CTRL_1, + .status = IPU_INT_STAT_1, + }, { + .control = IPU_INT_CTRL_2, + .status = IPU_INT_STAT_2, + }, { + .control = IPU_INT_CTRL_3, + .status = IPU_INT_STAT_3, + }, + /* 2 groups of error interrupts */ + { + .control = IPU_INT_CTRL_4, + .status = IPU_INT_STAT_4, + }, { + .control = IPU_INT_CTRL_5, + .status = IPU_INT_STAT_5, + }, +}; + +struct ipu_irq_map { + unsigned int irq; + int source; + struct ipu_irq_bank *bank; + struct ipu *ipu; +}; + +static struct ipu_irq_map irq_map[CONFIG_MX3_IPU_IRQS]; +/* Protects allocations from the above array of maps */ +static DEFINE_MUTEX(map_lock); +/* Protects register accesses and individual mappings */ +static DEFINE_SPINLOCK(bank_lock); + +static struct ipu_irq_map *src2map(unsigned int src) +{ + int i; + + for (i = 0; i < CONFIG_MX3_IPU_IRQS; i++) + if (irq_map[i].source == src) + return irq_map + i; + + return NULL; +} + +static void ipu_irq_unmask(unsigned int irq) +{ + struct ipu_irq_map *map = get_irq_chip_data(irq); + struct ipu_irq_bank *bank; + uint32_t reg; + unsigned long lock_flags; + + spin_lock_irqsave(&bank_lock, lock_flags); + + bank = map->bank; + if (!bank) { + spin_unlock_irqrestore(&bank_lock, lock_flags); + pr_err("IPU: %s(%u) - unmapped!\n", __func__, irq); + return; + } + + reg = ipu_read_reg(bank->ipu, bank->control); + reg |= (1UL << (map->source & 31)); + ipu_write_reg(bank->ipu, reg, bank->control); + + spin_unlock_irqrestore(&bank_lock, lock_flags); +} + +static void ipu_irq_mask(unsigned int irq) +{ + struct ipu_irq_map *map = get_irq_chip_data(irq); + struct ipu_irq_bank *bank; + uint32_t reg; + unsigned long lock_flags; + + spin_lock_irqsave(&bank_lock, lock_flags); + + bank = map->bank; + if (!bank) { + spin_unlock_irqrestore(&bank_lock, lock_flags); + pr_err("IPU: %s(%u) - unmapped!\n", __func__, irq); + return; + } + + reg = ipu_read_reg(bank->ipu, bank->control); + reg &= ~(1UL << (map->source & 31)); + ipu_write_reg(bank->ipu, reg, bank->control); + + spin_unlock_irqrestore(&bank_lock, lock_flags); +} + +static void ipu_irq_ack(unsigned int irq) +{ + struct ipu_irq_map *map = get_irq_chip_data(irq); + struct ipu_irq_bank *bank; + unsigned long lock_flags; + + spin_lock_irqsave(&bank_lock, lock_flags); + + bank = map->bank; + if (!bank) { + spin_unlock_irqrestore(&bank_lock, lock_flags); + pr_err("IPU: %s(%u) - unmapped!\n", __func__, irq); + return; + } + + ipu_write_reg(bank->ipu, 1UL << (map->source & 31), bank->status); + spin_unlock_irqrestore(&bank_lock, lock_flags); +} + +/** + * ipu_irq_status() - returns the current interrupt status of the specified IRQ. + * @irq: interrupt line to get status for. + * @return: true if the interrupt is pending/asserted or false if the + * interrupt is not pending. + */ +bool ipu_irq_status(unsigned int irq) +{ + struct ipu_irq_map *map = get_irq_chip_data(irq); + struct ipu_irq_bank *bank; + unsigned long lock_flags; + bool ret; + + spin_lock_irqsave(&bank_lock, lock_flags); + bank = map->bank; + ret = bank && ipu_read_reg(bank->ipu, bank->status) & + (1UL << (map->source & 31)); + spin_unlock_irqrestore(&bank_lock, lock_flags); + + return ret; +} + +/** + * ipu_irq_map() - map an IPU interrupt source to an IRQ number + * @source: interrupt source bit position (see below) + * @return: mapped IRQ number or negative error code + * + * The source parameter has to be explained further. On i.MX31 IPU has 137 IRQ + * sources, they are broken down in 5 32-bit registers, like 32, 32, 24, 32, 17. + * However, the source argument of this function is not the sequence number of + * the possible IRQ, but rather its bit position. So, first interrupt in fourth + * register has source number 96, and not 88. This makes calculations easier, + * and also provides forward compatibility with any future IPU implementations + * with any interrupt bit assignments. + */ +int ipu_irq_map(unsigned int source) +{ + int i, ret = -ENOMEM; + struct ipu_irq_map *map; + + might_sleep(); + + mutex_lock(&map_lock); + map = src2map(source); + if (map) { + pr_err("IPU: Source %u already mapped to IRQ %u\n", source, map->irq); + ret = -EBUSY; + goto out; + } + + for (i = 0; i < CONFIG_MX3_IPU_IRQS; i++) { + if (irq_map[i].source < 0) { + unsigned long lock_flags; + + spin_lock_irqsave(&bank_lock, lock_flags); + irq_map[i].source = source; + irq_map[i].bank = irq_bank + source / 32; + spin_unlock_irqrestore(&bank_lock, lock_flags); + + ret = irq_map[i].irq; + pr_debug("IPU: mapped source %u to IRQ %u\n", + source, ret); + break; + } + } +out: + mutex_unlock(&map_lock); + + if (ret < 0) + pr_err("IPU: couldn't map source %u: %d\n", source, ret); + + return ret; +} + +/** + * ipu_irq_map() - map an IPU interrupt source to an IRQ number + * @source: interrupt source bit position (see ipu_irq_map()) + * @return: 0 or negative error code + */ +int ipu_irq_unmap(unsigned int source) +{ + int i, ret = -EINVAL; + + might_sleep(); + + mutex_lock(&map_lock); + for (i = 0; i < CONFIG_MX3_IPU_IRQS; i++) { + if (irq_map[i].source == source) { + unsigned long lock_flags; + + pr_debug("IPU: unmapped source %u from IRQ %u\n", + source, irq_map[i].irq); + + spin_lock_irqsave(&bank_lock, lock_flags); + irq_map[i].source = -EINVAL; + irq_map[i].bank = NULL; + spin_unlock_irqrestore(&bank_lock, lock_flags); + + ret = 0; + break; + } + } + mutex_unlock(&map_lock); + + return ret; +} + +/* Chained IRQ handler for IPU error interrupt */ +static void ipu_irq_err(unsigned int irq, struct irq_desc *desc) +{ + struct ipu *ipu = get_irq_data(irq); + u32 status; + int i, line; + + for (i = IPU_IRQ_NR_FN_BANKS; i < IPU_IRQ_NR_BANKS; i++) { + struct ipu_irq_bank *bank = irq_bank + i; + + spin_lock(&bank_lock); + status = ipu_read_reg(ipu, bank->status); + /* + * Don't think we have to clear all interrupts here, they will + * be acked by ->handle_irq() (handle_level_irq). However, we + * might want to clear unhandled interrupts after the loop... + */ + status &= ipu_read_reg(ipu, bank->control); + spin_unlock(&bank_lock); + while ((line = ffs(status))) { + struct ipu_irq_map *map; + + line--; + status &= ~(1UL << line); + + spin_lock(&bank_lock); + map = src2map(32 * i + line); + if (map) + irq = map->irq; + spin_unlock(&bank_lock); + + if (!map) { + pr_err("IPU: Interrupt on unmapped source %u bank %d\n", + line, i); + continue; + } + generic_handle_irq(irq); + } + } +} + +/* Chained IRQ handler for IPU function interrupt */ +static void ipu_irq_fn(unsigned int irq, struct irq_desc *desc) +{ + struct ipu *ipu = get_irq_data(irq); + u32 status; + int i, line; + + for (i = 0; i < IPU_IRQ_NR_FN_BANKS; i++) { + struct ipu_irq_bank *bank = irq_bank + i; + + spin_lock(&bank_lock); + status = ipu_read_reg(ipu, bank->status); + /* Not clearing all interrupts, see above */ + status &= ipu_read_reg(ipu, bank->control); + spin_unlock(&bank_lock); + while ((line = ffs(status))) { + struct ipu_irq_map *map; + + line--; + status &= ~(1UL << line); + + spin_lock(&bank_lock); + map = src2map(32 * i + line); + if (map) + irq = map->irq; + spin_unlock(&bank_lock); + + if (!map) { + pr_err("IPU: Interrupt on unmapped source %u bank %d\n", + line, i); + continue; + } + generic_handle_irq(irq); + } + } +} + +static struct irq_chip ipu_irq_chip = { + .name = "ipu_irq", + .ack = ipu_irq_ack, + .mask = ipu_irq_mask, + .unmask = ipu_irq_unmask, +}; + +/* Install the IRQ handler */ +int __init ipu_irq_attach_irq(struct ipu *ipu, struct platform_device *dev) +{ + struct ipu_platform_data *pdata = dev->dev.platform_data; + unsigned int irq, irq_base, i; + + irq_base = pdata->irq_base; + + for (i = 0; i < IPU_IRQ_NR_BANKS; i++) + irq_bank[i].ipu = ipu; + + for (i = 0; i < CONFIG_MX3_IPU_IRQS; i++) { + int ret; + + irq = irq_base + i; + ret = set_irq_chip(irq, &ipu_irq_chip); + if (ret < 0) + return ret; + ret = set_irq_chip_data(irq, irq_map + i); + if (ret < 0) + return ret; + irq_map[i].ipu = ipu; + irq_map[i].irq = irq; + irq_map[i].source = -EINVAL; + set_irq_handler(irq, handle_level_irq); +#ifdef CONFIG_ARM + set_irq_flags(irq, IRQF_VALID | IRQF_PROBE); +#endif + } + + set_irq_data(ipu->irq_fn, ipu); + set_irq_chained_handler(ipu->irq_fn, ipu_irq_fn); + + set_irq_data(ipu->irq_err, ipu); + set_irq_chained_handler(ipu->irq_err, ipu_irq_err); + + return 0; +} + +void ipu_irq_detach_irq(struct ipu *ipu, struct platform_device *dev) +{ + struct ipu_platform_data *pdata = dev->dev.platform_data; + unsigned int irq, irq_base; + + irq_base = pdata->irq_base; + + set_irq_chained_handler(ipu->irq_fn, NULL); + set_irq_data(ipu->irq_fn, NULL); + + set_irq_chained_handler(ipu->irq_err, NULL); + set_irq_data(ipu->irq_err, NULL); + + for (irq = irq_base; irq < irq_base + CONFIG_MX3_IPU_IRQS; irq++) { +#ifdef CONFIG_ARM + set_irq_flags(irq, 0); +#endif + set_irq_chip(irq, NULL); + set_irq_chip_data(irq, NULL); + } +} diff --git a/drivers/dma/mv_xor.c b/drivers/dma/mv_xor.c new file mode 100644 index 00000000000..466ab10c1ff --- /dev/null +++ b/drivers/dma/mv_xor.c @@ -0,0 +1,1381 @@ +/* + * offload engine driver for the Marvell XOR engine + * Copyright (C) 2007, 2008, Marvell International Ltd. + * + * This program is free software; you can redistribute it and/or modify it + * under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., + * 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/delay.h> +#include <linux/dma-mapping.h> +#include <linux/spinlock.h> +#include <linux/interrupt.h> +#include <linux/platform_device.h> +#include <linux/memory.h> +#include <plat/mv_xor.h> +#include "mv_xor.h" + +static void mv_xor_issue_pending(struct dma_chan *chan); + +#define to_mv_xor_chan(chan) \ + container_of(chan, struct mv_xor_chan, common) + +#define to_mv_xor_device(dev) \ + container_of(dev, struct mv_xor_device, common) + +#define to_mv_xor_slot(tx) \ + container_of(tx, struct mv_xor_desc_slot, async_tx) + +static void mv_desc_init(struct mv_xor_desc_slot *desc, unsigned long flags) +{ + struct mv_xor_desc *hw_desc = desc->hw_desc; + + hw_desc->status = (1 << 31); + hw_desc->phy_next_desc = 0; + hw_desc->desc_command = (1 << 31); +} + +static u32 mv_desc_get_dest_addr(struct mv_xor_desc_slot *desc) +{ + struct mv_xor_desc *hw_desc = desc->hw_desc; + return hw_desc->phy_dest_addr; +} + +static u32 mv_desc_get_src_addr(struct mv_xor_desc_slot *desc, + int src_idx) +{ + struct mv_xor_desc *hw_desc = desc->hw_desc; + return hw_desc->phy_src_addr[src_idx]; +} + + +static void mv_desc_set_byte_count(struct mv_xor_desc_slot *desc, + u32 byte_count) +{ + struct mv_xor_desc *hw_desc = desc->hw_desc; + hw_desc->byte_count = byte_count; +} + +static void mv_desc_set_next_desc(struct mv_xor_desc_slot *desc, + u32 next_desc_addr) +{ + struct mv_xor_desc *hw_desc = desc->hw_desc; + BUG_ON(hw_desc->phy_next_desc); + hw_desc->phy_next_desc = next_desc_addr; +} + +static void mv_desc_clear_next_desc(struct mv_xor_desc_slot *desc) +{ + struct mv_xor_desc *hw_desc = desc->hw_desc; + hw_desc->phy_next_desc = 0; +} + +static void mv_desc_set_block_fill_val(struct mv_xor_desc_slot *desc, u32 val) +{ + desc->value = val; +} + +static void mv_desc_set_dest_addr(struct mv_xor_desc_slot *desc, + dma_addr_t addr) +{ + struct mv_xor_desc *hw_desc = desc->hw_desc; + hw_desc->phy_dest_addr = addr; +} + +static int mv_chan_memset_slot_count(size_t len) +{ + return 1; +} + +#define mv_chan_memcpy_slot_count(c) mv_chan_memset_slot_count(c) + +static void mv_desc_set_src_addr(struct mv_xor_desc_slot *desc, + int index, dma_addr_t addr) +{ + struct mv_xor_desc *hw_desc = desc->hw_desc; + hw_desc->phy_src_addr[index] = addr; + if (desc->type == DMA_XOR) + hw_desc->desc_command |= (1 << index); +} + +static u32 mv_chan_get_current_desc(struct mv_xor_chan *chan) +{ + return __raw_readl(XOR_CURR_DESC(chan)); +} + +static void mv_chan_set_next_descriptor(struct mv_xor_chan *chan, + u32 next_desc_addr) +{ + __raw_writel(next_desc_addr, XOR_NEXT_DESC(chan)); +} + +static void mv_chan_set_dest_pointer(struct mv_xor_chan *chan, u32 desc_addr) +{ + __raw_writel(desc_addr, XOR_DEST_POINTER(chan)); +} + +static void mv_chan_set_block_size(struct mv_xor_chan *chan, u32 block_size) +{ + __raw_writel(block_size, XOR_BLOCK_SIZE(chan)); +} + +static void mv_chan_set_value(struct mv_xor_chan *chan, u32 value) +{ + __raw_writel(value, XOR_INIT_VALUE_LOW(chan)); + __raw_writel(value, XOR_INIT_VALUE_HIGH(chan)); +} + +static void mv_chan_unmask_interrupts(struct mv_xor_chan *chan) +{ + u32 val = __raw_readl(XOR_INTR_MASK(chan)); + val |= XOR_INTR_MASK_VALUE << (chan->idx * 16); + __raw_writel(val, XOR_INTR_MASK(chan)); +} + +static u32 mv_chan_get_intr_cause(struct mv_xor_chan *chan) +{ + u32 intr_cause = __raw_readl(XOR_INTR_CAUSE(chan)); + intr_cause = (intr_cause >> (chan->idx * 16)) & 0xFFFF; + return intr_cause; +} + +static int mv_is_err_intr(u32 intr_cause) +{ + if (intr_cause & ((1<<4)|(1<<5)|(1<<6)|(1<<7)|(1<<8)|(1<<9))) + return 1; + + return 0; +} + +static void mv_xor_device_clear_eoc_cause(struct mv_xor_chan *chan) +{ + u32 val = (1 << (1 + (chan->idx * 16))); + dev_dbg(chan->device->common.dev, "%s, val 0x%08x\n", __func__, val); + __raw_writel(val, XOR_INTR_CAUSE(chan)); +} + +static void mv_xor_device_clear_err_status(struct mv_xor_chan *chan) +{ + u32 val = 0xFFFF0000 >> (chan->idx * 16); + __raw_writel(val, XOR_INTR_CAUSE(chan)); +} + +static int mv_can_chain(struct mv_xor_desc_slot *desc) +{ + struct mv_xor_desc_slot *chain_old_tail = list_entry( + desc->chain_node.prev, struct mv_xor_desc_slot, chain_node); + + if (chain_old_tail->type != desc->type) + return 0; + if (desc->type == DMA_MEMSET) + return 0; + + return 1; +} + +static void mv_set_mode(struct mv_xor_chan *chan, + enum dma_transaction_type type) +{ + u32 op_mode; + u32 config = __raw_readl(XOR_CONFIG(chan)); + + switch (type) { + case DMA_XOR: + op_mode = XOR_OPERATION_MODE_XOR; + break; + case DMA_MEMCPY: + op_mode = XOR_OPERATION_MODE_MEMCPY; + break; + case DMA_MEMSET: + op_mode = XOR_OPERATION_MODE_MEMSET; + break; + default: + dev_printk(KERN_ERR, chan->device->common.dev, + "error: unsupported operation %d.\n", + type); + BUG(); + return; + } + + config &= ~0x7; + config |= op_mode; + __raw_writel(config, XOR_CONFIG(chan)); + chan->current_type = type; +} + +static void mv_chan_activate(struct mv_xor_chan *chan) +{ + u32 activation; + + dev_dbg(chan->device->common.dev, " activate chan.\n"); + activation = __raw_readl(XOR_ACTIVATION(chan)); + activation |= 0x1; + __raw_writel(activation, XOR_ACTIVATION(chan)); +} + +static char mv_chan_is_busy(struct mv_xor_chan *chan) +{ + u32 state = __raw_readl(XOR_ACTIVATION(chan)); + + state = (state >> 4) & 0x3; + + return (state == 1) ? 1 : 0; +} + +static int mv_chan_xor_slot_count(size_t len, int src_cnt) +{ + return 1; +} + +/** + * mv_xor_free_slots - flags descriptor slots for reuse + * @slot: Slot to free + * Caller must hold &mv_chan->lock while calling this function + */ +static void mv_xor_free_slots(struct mv_xor_chan *mv_chan, + struct mv_xor_desc_slot *slot) +{ + dev_dbg(mv_chan->device->common.dev, "%s %d slot %p\n", + __func__, __LINE__, slot); + + slot->slots_per_op = 0; + +} + +/* + * mv_xor_start_new_chain - program the engine to operate on new chain headed by + * sw_desc + * Caller must hold &mv_chan->lock while calling this function + */ +static void mv_xor_start_new_chain(struct mv_xor_chan *mv_chan, + struct mv_xor_desc_slot *sw_desc) +{ + dev_dbg(mv_chan->device->common.dev, "%s %d: sw_desc %p\n", + __func__, __LINE__, sw_desc); + if (sw_desc->type != mv_chan->current_type) + mv_set_mode(mv_chan, sw_desc->type); + + if (sw_desc->type == DMA_MEMSET) { + /* for memset requests we need to program the engine, no + * descriptors used. + */ + struct mv_xor_desc *hw_desc = sw_desc->hw_desc; + mv_chan_set_dest_pointer(mv_chan, hw_desc->phy_dest_addr); + mv_chan_set_block_size(mv_chan, sw_desc->unmap_len); + mv_chan_set_value(mv_chan, sw_desc->value); + } else { + /* set the hardware chain */ + mv_chan_set_next_descriptor(mv_chan, sw_desc->async_tx.phys); + } + mv_chan->pending += sw_desc->slot_cnt; + mv_xor_issue_pending(&mv_chan->common); +} + +static dma_cookie_t +mv_xor_run_tx_complete_actions(struct mv_xor_desc_slot *desc, + struct mv_xor_chan *mv_chan, dma_cookie_t cookie) +{ + BUG_ON(desc->async_tx.cookie < 0); + + if (desc->async_tx.cookie > 0) { + cookie = desc->async_tx.cookie; + + /* call the callback (must not sleep or submit new + * operations to this channel) + */ + if (desc->async_tx.callback) + desc->async_tx.callback( + desc->async_tx.callback_param); + + /* unmap dma addresses + * (unmap_single vs unmap_page?) + */ + if (desc->group_head && desc->unmap_len) { + struct mv_xor_desc_slot *unmap = desc->group_head; + struct device *dev = + &mv_chan->device->pdev->dev; + u32 len = unmap->unmap_len; + enum dma_ctrl_flags flags = desc->async_tx.flags; + u32 src_cnt; + dma_addr_t addr; + dma_addr_t dest; + + src_cnt = unmap->unmap_src_cnt; + dest = mv_desc_get_dest_addr(unmap); + if (!(flags & DMA_COMPL_SKIP_DEST_UNMAP)) { + enum dma_data_direction dir; + + if (src_cnt > 1) /* is xor ? */ + dir = DMA_BIDIRECTIONAL; + else + dir = DMA_FROM_DEVICE; + dma_unmap_page(dev, dest, len, dir); + } + + if (!(flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + while (src_cnt--) { + addr = mv_desc_get_src_addr(unmap, + src_cnt); + if (addr == dest) + continue; + dma_unmap_page(dev, addr, len, + DMA_TO_DEVICE); + } + } + desc->group_head = NULL; + } + } + + /* run dependent operations */ + dma_run_dependencies(&desc->async_tx); + + return cookie; +} + +static int +mv_xor_clean_completed_slots(struct mv_xor_chan *mv_chan) +{ + struct mv_xor_desc_slot *iter, *_iter; + + dev_dbg(mv_chan->device->common.dev, "%s %d\n", __func__, __LINE__); + list_for_each_entry_safe(iter, _iter, &mv_chan->completed_slots, + completed_node) { + + if (async_tx_test_ack(&iter->async_tx)) { + list_del(&iter->completed_node); + mv_xor_free_slots(mv_chan, iter); + } + } + return 0; +} + +static int +mv_xor_clean_slot(struct mv_xor_desc_slot *desc, + struct mv_xor_chan *mv_chan) +{ + dev_dbg(mv_chan->device->common.dev, "%s %d: desc %p flags %d\n", + __func__, __LINE__, desc, desc->async_tx.flags); + list_del(&desc->chain_node); + /* the client is allowed to attach dependent operations + * until 'ack' is set + */ + if (!async_tx_test_ack(&desc->async_tx)) { + /* move this slot to the completed_slots */ + list_add_tail(&desc->completed_node, &mv_chan->completed_slots); + return 0; + } + + mv_xor_free_slots(mv_chan, desc); + return 0; +} + +static void __mv_xor_slot_cleanup(struct mv_xor_chan *mv_chan) +{ + struct mv_xor_desc_slot *iter, *_iter; + dma_cookie_t cookie = 0; + int busy = mv_chan_is_busy(mv_chan); + u32 current_desc = mv_chan_get_current_desc(mv_chan); + int seen_current = 0; + + dev_dbg(mv_chan->device->common.dev, "%s %d\n", __func__, __LINE__); + dev_dbg(mv_chan->device->common.dev, "current_desc %x\n", current_desc); + mv_xor_clean_completed_slots(mv_chan); + + /* free completed slots from the chain starting with + * the oldest descriptor + */ + + list_for_each_entry_safe(iter, _iter, &mv_chan->chain, + chain_node) { + prefetch(_iter); + prefetch(&_iter->async_tx); + + /* do not advance past the current descriptor loaded into the + * hardware channel, subsequent descriptors are either in + * process or have not been submitted + */ + if (seen_current) + break; + + /* stop the search if we reach the current descriptor and the + * channel is busy + */ + if (iter->async_tx.phys == current_desc) { + seen_current = 1; + if (busy) + break; + } + + cookie = mv_xor_run_tx_complete_actions(iter, mv_chan, cookie); + + if (mv_xor_clean_slot(iter, mv_chan)) + break; + } + + if ((busy == 0) && !list_empty(&mv_chan->chain)) { + struct mv_xor_desc_slot *chain_head; + chain_head = list_entry(mv_chan->chain.next, + struct mv_xor_desc_slot, + chain_node); + + mv_xor_start_new_chain(mv_chan, chain_head); + } + + if (cookie > 0) + mv_chan->completed_cookie = cookie; +} + +static void +mv_xor_slot_cleanup(struct mv_xor_chan *mv_chan) +{ + spin_lock_bh(&mv_chan->lock); + __mv_xor_slot_cleanup(mv_chan); + spin_unlock_bh(&mv_chan->lock); +} + +static void mv_xor_tasklet(unsigned long data) +{ + struct mv_xor_chan *chan = (struct mv_xor_chan *) data; + __mv_xor_slot_cleanup(chan); +} + +static struct mv_xor_desc_slot * +mv_xor_alloc_slots(struct mv_xor_chan *mv_chan, int num_slots, + int slots_per_op) +{ + struct mv_xor_desc_slot *iter, *_iter, *alloc_start = NULL; + LIST_HEAD(chain); + int slots_found, retry = 0; + + /* start search from the last allocated descrtiptor + * if a contiguous allocation can not be found start searching + * from the beginning of the list + */ +retry: + slots_found = 0; + if (retry == 0) + iter = mv_chan->last_used; + else + iter = list_entry(&mv_chan->all_slots, + struct mv_xor_desc_slot, + slot_node); + + list_for_each_entry_safe_continue( + iter, _iter, &mv_chan->all_slots, slot_node) { + prefetch(_iter); + prefetch(&_iter->async_tx); + if (iter->slots_per_op) { + /* give up after finding the first busy slot + * on the second pass through the list + */ + if (retry) + break; + + slots_found = 0; + continue; + } + + /* start the allocation if the slot is correctly aligned */ + if (!slots_found++) + alloc_start = iter; + + if (slots_found == num_slots) { + struct mv_xor_desc_slot *alloc_tail = NULL; + struct mv_xor_desc_slot *last_used = NULL; + iter = alloc_start; + while (num_slots) { + int i; + + /* pre-ack all but the last descriptor */ + async_tx_ack(&iter->async_tx); + + list_add_tail(&iter->chain_node, &chain); + alloc_tail = iter; + iter->async_tx.cookie = 0; + iter->slot_cnt = num_slots; + iter->xor_check_result = NULL; + for (i = 0; i < slots_per_op; i++) { + iter->slots_per_op = slots_per_op - i; + last_used = iter; + iter = list_entry(iter->slot_node.next, + struct mv_xor_desc_slot, + slot_node); + } + num_slots -= slots_per_op; + } + alloc_tail->group_head = alloc_start; + alloc_tail->async_tx.cookie = -EBUSY; + list_splice(&chain, &alloc_tail->tx_list); + mv_chan->last_used = last_used; + mv_desc_clear_next_desc(alloc_start); + mv_desc_clear_next_desc(alloc_tail); + return alloc_tail; + } + } + if (!retry++) + goto retry; + + /* try to free some slots if the allocation fails */ + tasklet_schedule(&mv_chan->irq_tasklet); + + return NULL; +} + +static dma_cookie_t +mv_desc_assign_cookie(struct mv_xor_chan *mv_chan, + struct mv_xor_desc_slot *desc) +{ + dma_cookie_t cookie = mv_chan->common.cookie; + + if (++cookie < 0) + cookie = 1; + mv_chan->common.cookie = desc->async_tx.cookie = cookie; + return cookie; +} + +/************************ DMA engine API functions ****************************/ +static dma_cookie_t +mv_xor_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct mv_xor_desc_slot *sw_desc = to_mv_xor_slot(tx); + struct mv_xor_chan *mv_chan = to_mv_xor_chan(tx->chan); + struct mv_xor_desc_slot *grp_start, *old_chain_tail; + dma_cookie_t cookie; + int new_hw_chain = 1; + + dev_dbg(mv_chan->device->common.dev, + "%s sw_desc %p: async_tx %p\n", + __func__, sw_desc, &sw_desc->async_tx); + + grp_start = sw_desc->group_head; + + spin_lock_bh(&mv_chan->lock); + cookie = mv_desc_assign_cookie(mv_chan, sw_desc); + + if (list_empty(&mv_chan->chain)) + list_splice_init(&sw_desc->tx_list, &mv_chan->chain); + else { + new_hw_chain = 0; + + old_chain_tail = list_entry(mv_chan->chain.prev, + struct mv_xor_desc_slot, + chain_node); + list_splice_init(&grp_start->tx_list, + &old_chain_tail->chain_node); + + if (!mv_can_chain(grp_start)) + goto submit_done; + + dev_dbg(mv_chan->device->common.dev, "Append to last desc %x\n", + old_chain_tail->async_tx.phys); + + /* fix up the hardware chain */ + mv_desc_set_next_desc(old_chain_tail, grp_start->async_tx.phys); + + /* if the channel is not busy */ + if (!mv_chan_is_busy(mv_chan)) { + u32 current_desc = mv_chan_get_current_desc(mv_chan); + /* + * and the curren desc is the end of the chain before + * the append, then we need to start the channel + */ + if (current_desc == old_chain_tail->async_tx.phys) + new_hw_chain = 1; + } + } + + if (new_hw_chain) + mv_xor_start_new_chain(mv_chan, grp_start); + +submit_done: + spin_unlock_bh(&mv_chan->lock); + + return cookie; +} + +/* returns the number of allocated descriptors */ +static int mv_xor_alloc_chan_resources(struct dma_chan *chan) +{ + char *hw_desc; + int idx; + struct mv_xor_chan *mv_chan = to_mv_xor_chan(chan); + struct mv_xor_desc_slot *slot = NULL; + struct mv_xor_platform_data *plat_data = + mv_chan->device->pdev->dev.platform_data; + int num_descs_in_pool = plat_data->pool_size/MV_XOR_SLOT_SIZE; + + /* Allocate descriptor slots */ + idx = mv_chan->slots_allocated; + while (idx < num_descs_in_pool) { + slot = kzalloc(sizeof(*slot), GFP_KERNEL); + if (!slot) { + printk(KERN_INFO "MV XOR Channel only initialized" + " %d descriptor slots", idx); + break; + } + hw_desc = (char *) mv_chan->device->dma_desc_pool_virt; + slot->hw_desc = (void *) &hw_desc[idx * MV_XOR_SLOT_SIZE]; + + dma_async_tx_descriptor_init(&slot->async_tx, chan); + slot->async_tx.tx_submit = mv_xor_tx_submit; + INIT_LIST_HEAD(&slot->chain_node); + INIT_LIST_HEAD(&slot->slot_node); + INIT_LIST_HEAD(&slot->tx_list); + hw_desc = (char *) mv_chan->device->dma_desc_pool; + slot->async_tx.phys = + (dma_addr_t) &hw_desc[idx * MV_XOR_SLOT_SIZE]; + slot->idx = idx++; + + spin_lock_bh(&mv_chan->lock); + mv_chan->slots_allocated = idx; + list_add_tail(&slot->slot_node, &mv_chan->all_slots); + spin_unlock_bh(&mv_chan->lock); + } + + if (mv_chan->slots_allocated && !mv_chan->last_used) + mv_chan->last_used = list_entry(mv_chan->all_slots.next, + struct mv_xor_desc_slot, + slot_node); + + dev_dbg(mv_chan->device->common.dev, + "allocated %d descriptor slots last_used: %p\n", + mv_chan->slots_allocated, mv_chan->last_used); + + return mv_chan->slots_allocated ? : -ENOMEM; +} + +static struct dma_async_tx_descriptor * +mv_xor_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src, + size_t len, unsigned long flags) +{ + struct mv_xor_chan *mv_chan = to_mv_xor_chan(chan); + struct mv_xor_desc_slot *sw_desc, *grp_start; + int slot_cnt; + + dev_dbg(mv_chan->device->common.dev, + "%s dest: %x src %x len: %u flags: %ld\n", + __func__, dest, src, len, flags); + if (unlikely(len < MV_XOR_MIN_BYTE_COUNT)) + return NULL; + + BUG_ON(unlikely(len > MV_XOR_MAX_BYTE_COUNT)); + + spin_lock_bh(&mv_chan->lock); + slot_cnt = mv_chan_memcpy_slot_count(len); + sw_desc = mv_xor_alloc_slots(mv_chan, slot_cnt, 1); + if (sw_desc) { + sw_desc->type = DMA_MEMCPY; + sw_desc->async_tx.flags = flags; + grp_start = sw_desc->group_head; + mv_desc_init(grp_start, flags); + mv_desc_set_byte_count(grp_start, len); + mv_desc_set_dest_addr(sw_desc->group_head, dest); + mv_desc_set_src_addr(grp_start, 0, src); + sw_desc->unmap_src_cnt = 1; + sw_desc->unmap_len = len; + } + spin_unlock_bh(&mv_chan->lock); + + dev_dbg(mv_chan->device->common.dev, + "%s sw_desc %p async_tx %p\n", + __func__, sw_desc, sw_desc ? &sw_desc->async_tx : 0); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static struct dma_async_tx_descriptor * +mv_xor_prep_dma_memset(struct dma_chan *chan, dma_addr_t dest, int value, + size_t len, unsigned long flags) +{ + struct mv_xor_chan *mv_chan = to_mv_xor_chan(chan); + struct mv_xor_desc_slot *sw_desc, *grp_start; + int slot_cnt; + + dev_dbg(mv_chan->device->common.dev, + "%s dest: %x len: %u flags: %ld\n", + __func__, dest, len, flags); + if (unlikely(len < MV_XOR_MIN_BYTE_COUNT)) + return NULL; + + BUG_ON(unlikely(len > MV_XOR_MAX_BYTE_COUNT)); + + spin_lock_bh(&mv_chan->lock); + slot_cnt = mv_chan_memset_slot_count(len); + sw_desc = mv_xor_alloc_slots(mv_chan, slot_cnt, 1); + if (sw_desc) { + sw_desc->type = DMA_MEMSET; + sw_desc->async_tx.flags = flags; + grp_start = sw_desc->group_head; + mv_desc_init(grp_start, flags); + mv_desc_set_byte_count(grp_start, len); + mv_desc_set_dest_addr(sw_desc->group_head, dest); + mv_desc_set_block_fill_val(grp_start, value); + sw_desc->unmap_src_cnt = 1; + sw_desc->unmap_len = len; + } + spin_unlock_bh(&mv_chan->lock); + dev_dbg(mv_chan->device->common.dev, + "%s sw_desc %p async_tx %p \n", + __func__, sw_desc, &sw_desc->async_tx); + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static struct dma_async_tx_descriptor * +mv_xor_prep_dma_xor(struct dma_chan *chan, dma_addr_t dest, dma_addr_t *src, + unsigned int src_cnt, size_t len, unsigned long flags) +{ + struct mv_xor_chan *mv_chan = to_mv_xor_chan(chan); + struct mv_xor_desc_slot *sw_desc, *grp_start; + int slot_cnt; + + if (unlikely(len < MV_XOR_MIN_BYTE_COUNT)) + return NULL; + + BUG_ON(unlikely(len > MV_XOR_MAX_BYTE_COUNT)); + + dev_dbg(mv_chan->device->common.dev, + "%s src_cnt: %d len: dest %x %u flags: %ld\n", + __func__, src_cnt, len, dest, flags); + + spin_lock_bh(&mv_chan->lock); + slot_cnt = mv_chan_xor_slot_count(len, src_cnt); + sw_desc = mv_xor_alloc_slots(mv_chan, slot_cnt, 1); + if (sw_desc) { + sw_desc->type = DMA_XOR; + sw_desc->async_tx.flags = flags; + grp_start = sw_desc->group_head; + mv_desc_init(grp_start, flags); + /* the byte count field is the same as in memcpy desc*/ + mv_desc_set_byte_count(grp_start, len); + mv_desc_set_dest_addr(sw_desc->group_head, dest); + sw_desc->unmap_src_cnt = src_cnt; + sw_desc->unmap_len = len; + while (src_cnt--) + mv_desc_set_src_addr(grp_start, src_cnt, src[src_cnt]); + } + spin_unlock_bh(&mv_chan->lock); + dev_dbg(mv_chan->device->common.dev, + "%s sw_desc %p async_tx %p \n", + __func__, sw_desc, &sw_desc->async_tx); + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static void mv_xor_free_chan_resources(struct dma_chan *chan) +{ + struct mv_xor_chan *mv_chan = to_mv_xor_chan(chan); + struct mv_xor_desc_slot *iter, *_iter; + int in_use_descs = 0; + + mv_xor_slot_cleanup(mv_chan); + + spin_lock_bh(&mv_chan->lock); + list_for_each_entry_safe(iter, _iter, &mv_chan->chain, + chain_node) { + in_use_descs++; + list_del(&iter->chain_node); + } + list_for_each_entry_safe(iter, _iter, &mv_chan->completed_slots, + completed_node) { + in_use_descs++; + list_del(&iter->completed_node); + } + list_for_each_entry_safe_reverse( + iter, _iter, &mv_chan->all_slots, slot_node) { + list_del(&iter->slot_node); + kfree(iter); + mv_chan->slots_allocated--; + } + mv_chan->last_used = NULL; + + dev_dbg(mv_chan->device->common.dev, "%s slots_allocated %d\n", + __func__, mv_chan->slots_allocated); + spin_unlock_bh(&mv_chan->lock); + + if (in_use_descs) + dev_err(mv_chan->device->common.dev, + "freeing %d in use descriptors!\n", in_use_descs); +} + +/** + * mv_xor_is_complete - poll the status of an XOR transaction + * @chan: XOR channel handle + * @cookie: XOR transaction identifier + */ +static enum dma_status mv_xor_is_complete(struct dma_chan *chan, + dma_cookie_t cookie, + dma_cookie_t *done, + dma_cookie_t *used) +{ + struct mv_xor_chan *mv_chan = to_mv_xor_chan(chan); + dma_cookie_t last_used; + dma_cookie_t last_complete; + enum dma_status ret; + + last_used = chan->cookie; + last_complete = mv_chan->completed_cookie; + mv_chan->is_complete_cookie = cookie; + if (done) + *done = last_complete; + if (used) + *used = last_used; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + if (ret == DMA_SUCCESS) { + mv_xor_clean_completed_slots(mv_chan); + return ret; + } + mv_xor_slot_cleanup(mv_chan); + + last_used = chan->cookie; + last_complete = mv_chan->completed_cookie; + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + return dma_async_is_complete(cookie, last_complete, last_used); +} + +static void mv_dump_xor_regs(struct mv_xor_chan *chan) +{ + u32 val; + + val = __raw_readl(XOR_CONFIG(chan)); + dev_printk(KERN_ERR, chan->device->common.dev, + "config 0x%08x.\n", val); + + val = __raw_readl(XOR_ACTIVATION(chan)); + dev_printk(KERN_ERR, chan->device->common.dev, + "activation 0x%08x.\n", val); + + val = __raw_readl(XOR_INTR_CAUSE(chan)); + dev_printk(KERN_ERR, chan->device->common.dev, + "intr cause 0x%08x.\n", val); + + val = __raw_readl(XOR_INTR_MASK(chan)); + dev_printk(KERN_ERR, chan->device->common.dev, + "intr mask 0x%08x.\n", val); + + val = __raw_readl(XOR_ERROR_CAUSE(chan)); + dev_printk(KERN_ERR, chan->device->common.dev, + "error cause 0x%08x.\n", val); + + val = __raw_readl(XOR_ERROR_ADDR(chan)); + dev_printk(KERN_ERR, chan->device->common.dev, + "error addr 0x%08x.\n", val); +} + +static void mv_xor_err_interrupt_handler(struct mv_xor_chan *chan, + u32 intr_cause) +{ + if (intr_cause & (1 << 4)) { + dev_dbg(chan->device->common.dev, + "ignore this error\n"); + return; + } + + dev_printk(KERN_ERR, chan->device->common.dev, + "error on chan %d. intr cause 0x%08x.\n", + chan->idx, intr_cause); + + mv_dump_xor_regs(chan); + BUG(); +} + +static irqreturn_t mv_xor_interrupt_handler(int irq, void *data) +{ + struct mv_xor_chan *chan = data; + u32 intr_cause = mv_chan_get_intr_cause(chan); + + dev_dbg(chan->device->common.dev, "intr cause %x\n", intr_cause); + + if (mv_is_err_intr(intr_cause)) + mv_xor_err_interrupt_handler(chan, intr_cause); + + tasklet_schedule(&chan->irq_tasklet); + + mv_xor_device_clear_eoc_cause(chan); + + return IRQ_HANDLED; +} + +static void mv_xor_issue_pending(struct dma_chan *chan) +{ + struct mv_xor_chan *mv_chan = to_mv_xor_chan(chan); + + if (mv_chan->pending >= MV_XOR_THRESHOLD) { + mv_chan->pending = 0; + mv_chan_activate(mv_chan); + } +} + +/* + * Perform a transaction to verify the HW works. + */ +#define MV_XOR_TEST_SIZE 2000 + +static int __devinit mv_xor_memcpy_self_test(struct mv_xor_device *device) +{ + int i; + void *src, *dest; + dma_addr_t src_dma, dest_dma; + struct dma_chan *dma_chan; + dma_cookie_t cookie; + struct dma_async_tx_descriptor *tx; + int err = 0; + struct mv_xor_chan *mv_chan; + + src = kmalloc(sizeof(u8) * MV_XOR_TEST_SIZE, GFP_KERNEL); + if (!src) + return -ENOMEM; + + dest = kzalloc(sizeof(u8) * MV_XOR_TEST_SIZE, GFP_KERNEL); + if (!dest) { + kfree(src); + return -ENOMEM; + } + + /* Fill in src buffer */ + for (i = 0; i < MV_XOR_TEST_SIZE; i++) + ((u8 *) src)[i] = (u8)i; + + /* Start copy, using first DMA channel */ + dma_chan = container_of(device->common.channels.next, + struct dma_chan, + device_node); + if (mv_xor_alloc_chan_resources(dma_chan) < 1) { + err = -ENODEV; + goto out; + } + + dest_dma = dma_map_single(dma_chan->device->dev, dest, + MV_XOR_TEST_SIZE, DMA_FROM_DEVICE); + + src_dma = dma_map_single(dma_chan->device->dev, src, + MV_XOR_TEST_SIZE, DMA_TO_DEVICE); + + tx = mv_xor_prep_dma_memcpy(dma_chan, dest_dma, src_dma, + MV_XOR_TEST_SIZE, 0); + cookie = mv_xor_tx_submit(tx); + mv_xor_issue_pending(dma_chan); + async_tx_ack(tx); + msleep(1); + + if (mv_xor_is_complete(dma_chan, cookie, NULL, NULL) != + DMA_SUCCESS) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test copy timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + mv_chan = to_mv_xor_chan(dma_chan); + dma_sync_single_for_cpu(&mv_chan->device->pdev->dev, dest_dma, + MV_XOR_TEST_SIZE, DMA_FROM_DEVICE); + if (memcmp(src, dest, MV_XOR_TEST_SIZE)) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test copy failed compare, disabling\n"); + err = -ENODEV; + goto free_resources; + } + +free_resources: + mv_xor_free_chan_resources(dma_chan); +out: + kfree(src); + kfree(dest); + return err; +} + +#define MV_XOR_NUM_SRC_TEST 4 /* must be <= 15 */ +static int __devinit +mv_xor_xor_self_test(struct mv_xor_device *device) +{ + int i, src_idx; + struct page *dest; + struct page *xor_srcs[MV_XOR_NUM_SRC_TEST]; + dma_addr_t dma_srcs[MV_XOR_NUM_SRC_TEST]; + dma_addr_t dest_dma; + struct dma_async_tx_descriptor *tx; + struct dma_chan *dma_chan; + dma_cookie_t cookie; + u8 cmp_byte = 0; + u32 cmp_word; + int err = 0; + struct mv_xor_chan *mv_chan; + + for (src_idx = 0; src_idx < MV_XOR_NUM_SRC_TEST; src_idx++) { + xor_srcs[src_idx] = alloc_page(GFP_KERNEL); + if (!xor_srcs[src_idx]) { + while (src_idx--) + __free_page(xor_srcs[src_idx]); + return -ENOMEM; + } + } + + dest = alloc_page(GFP_KERNEL); + if (!dest) { + while (src_idx--) + __free_page(xor_srcs[src_idx]); + return -ENOMEM; + } + + /* Fill in src buffers */ + for (src_idx = 0; src_idx < MV_XOR_NUM_SRC_TEST; src_idx++) { + u8 *ptr = page_address(xor_srcs[src_idx]); + for (i = 0; i < PAGE_SIZE; i++) + ptr[i] = (1 << src_idx); + } + + for (src_idx = 0; src_idx < MV_XOR_NUM_SRC_TEST; src_idx++) + cmp_byte ^= (u8) (1 << src_idx); + + cmp_word = (cmp_byte << 24) | (cmp_byte << 16) | + (cmp_byte << 8) | cmp_byte; + + memset(page_address(dest), 0, PAGE_SIZE); + + dma_chan = container_of(device->common.channels.next, + struct dma_chan, + device_node); + if (mv_xor_alloc_chan_resources(dma_chan) < 1) { + err = -ENODEV; + goto out; + } + + /* test xor */ + dest_dma = dma_map_page(dma_chan->device->dev, dest, 0, PAGE_SIZE, + DMA_FROM_DEVICE); + + for (i = 0; i < MV_XOR_NUM_SRC_TEST; i++) + dma_srcs[i] = dma_map_page(dma_chan->device->dev, xor_srcs[i], + 0, PAGE_SIZE, DMA_TO_DEVICE); + + tx = mv_xor_prep_dma_xor(dma_chan, dest_dma, dma_srcs, + MV_XOR_NUM_SRC_TEST, PAGE_SIZE, 0); + + cookie = mv_xor_tx_submit(tx); + mv_xor_issue_pending(dma_chan); + async_tx_ack(tx); + msleep(8); + + if (mv_xor_is_complete(dma_chan, cookie, NULL, NULL) != + DMA_SUCCESS) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test xor timed out, disabling\n"); + err = -ENODEV; + goto free_resources; + } + + mv_chan = to_mv_xor_chan(dma_chan); + dma_sync_single_for_cpu(&mv_chan->device->pdev->dev, dest_dma, + PAGE_SIZE, DMA_FROM_DEVICE); + for (i = 0; i < (PAGE_SIZE / sizeof(u32)); i++) { + u32 *ptr = page_address(dest); + if (ptr[i] != cmp_word) { + dev_printk(KERN_ERR, dma_chan->device->dev, + "Self-test xor failed compare, disabling." + " index %d, data %x, expected %x\n", i, + ptr[i], cmp_word); + err = -ENODEV; + goto free_resources; + } + } + +free_resources: + mv_xor_free_chan_resources(dma_chan); +out: + src_idx = MV_XOR_NUM_SRC_TEST; + while (src_idx--) + __free_page(xor_srcs[src_idx]); + __free_page(dest); + return err; +} + +static int __devexit mv_xor_remove(struct platform_device *dev) +{ + struct mv_xor_device *device = platform_get_drvdata(dev); + struct dma_chan *chan, *_chan; + struct mv_xor_chan *mv_chan; + struct mv_xor_platform_data *plat_data = dev->dev.platform_data; + + dma_async_device_unregister(&device->common); + + dma_free_coherent(&dev->dev, plat_data->pool_size, + device->dma_desc_pool_virt, device->dma_desc_pool); + + list_for_each_entry_safe(chan, _chan, &device->common.channels, + device_node) { + mv_chan = to_mv_xor_chan(chan); + list_del(&chan->device_node); + } + + return 0; +} + +static int __devinit mv_xor_probe(struct platform_device *pdev) +{ + int ret = 0; + int irq; + struct mv_xor_device *adev; + struct mv_xor_chan *mv_chan; + struct dma_device *dma_dev; + struct mv_xor_platform_data *plat_data = pdev->dev.platform_data; + + + adev = devm_kzalloc(&pdev->dev, sizeof(*adev), GFP_KERNEL); + if (!adev) + return -ENOMEM; + + dma_dev = &adev->common; + + /* allocate coherent memory for hardware descriptors + * note: writecombine gives slightly better performance, but + * requires that we explicitly flush the writes + */ + adev->dma_desc_pool_virt = dma_alloc_writecombine(&pdev->dev, + plat_data->pool_size, + &adev->dma_desc_pool, + GFP_KERNEL); + if (!adev->dma_desc_pool_virt) + return -ENOMEM; + + adev->id = plat_data->hw_id; + + /* discover transaction capabilites from the platform data */ + dma_dev->cap_mask = plat_data->cap_mask; + adev->pdev = pdev; + platform_set_drvdata(pdev, adev); + + adev->shared = platform_get_drvdata(plat_data->shared); + + INIT_LIST_HEAD(&dma_dev->channels); + + /* set base routines */ + dma_dev->device_alloc_chan_resources = mv_xor_alloc_chan_resources; + dma_dev->device_free_chan_resources = mv_xor_free_chan_resources; + dma_dev->device_is_tx_complete = mv_xor_is_complete; + dma_dev->device_issue_pending = mv_xor_issue_pending; + dma_dev->dev = &pdev->dev; + + /* set prep routines based on capability */ + if (dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask)) + dma_dev->device_prep_dma_memcpy = mv_xor_prep_dma_memcpy; + if (dma_has_cap(DMA_MEMSET, dma_dev->cap_mask)) + dma_dev->device_prep_dma_memset = mv_xor_prep_dma_memset; + if (dma_has_cap(DMA_XOR, dma_dev->cap_mask)) { + dma_dev->max_xor = 8; + dma_dev->device_prep_dma_xor = mv_xor_prep_dma_xor; + } + + mv_chan = devm_kzalloc(&pdev->dev, sizeof(*mv_chan), GFP_KERNEL); + if (!mv_chan) { + ret = -ENOMEM; + goto err_free_dma; + } + mv_chan->device = adev; + mv_chan->idx = plat_data->hw_id; + mv_chan->mmr_base = adev->shared->xor_base; + + if (!mv_chan->mmr_base) { + ret = -ENOMEM; + goto err_free_dma; + } + tasklet_init(&mv_chan->irq_tasklet, mv_xor_tasklet, (unsigned long) + mv_chan); + + /* clear errors before enabling interrupts */ + mv_xor_device_clear_err_status(mv_chan); + + irq = platform_get_irq(pdev, 0); + if (irq < 0) { + ret = irq; + goto err_free_dma; + } + ret = devm_request_irq(&pdev->dev, irq, + mv_xor_interrupt_handler, + 0, dev_name(&pdev->dev), mv_chan); + if (ret) + goto err_free_dma; + + mv_chan_unmask_interrupts(mv_chan); + + mv_set_mode(mv_chan, DMA_MEMCPY); + + spin_lock_init(&mv_chan->lock); + INIT_LIST_HEAD(&mv_chan->chain); + INIT_LIST_HEAD(&mv_chan->completed_slots); + INIT_LIST_HEAD(&mv_chan->all_slots); + mv_chan->common.device = dma_dev; + + list_add_tail(&mv_chan->common.device_node, &dma_dev->channels); + + if (dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask)) { + ret = mv_xor_memcpy_self_test(adev); + dev_dbg(&pdev->dev, "memcpy self test returned %d\n", ret); + if (ret) + goto err_free_dma; + } + + if (dma_has_cap(DMA_XOR, dma_dev->cap_mask)) { + ret = mv_xor_xor_self_test(adev); + dev_dbg(&pdev->dev, "xor self test returned %d\n", ret); + if (ret) + goto err_free_dma; + } + + dev_printk(KERN_INFO, &pdev->dev, "Marvell XOR: " + "( %s%s%s%s)\n", + dma_has_cap(DMA_XOR, dma_dev->cap_mask) ? "xor " : "", + dma_has_cap(DMA_MEMSET, dma_dev->cap_mask) ? "fill " : "", + dma_has_cap(DMA_MEMCPY, dma_dev->cap_mask) ? "cpy " : "", + dma_has_cap(DMA_INTERRUPT, dma_dev->cap_mask) ? "intr " : ""); + + dma_async_device_register(dma_dev); + goto out; + + err_free_dma: + dma_free_coherent(&adev->pdev->dev, plat_data->pool_size, + adev->dma_desc_pool_virt, adev->dma_desc_pool); + out: + return ret; +} + +static void +mv_xor_conf_mbus_windows(struct mv_xor_shared_private *msp, + struct mbus_dram_target_info *dram) +{ + void __iomem *base = msp->xor_base; + u32 win_enable = 0; + int i; + + for (i = 0; i < 8; i++) { + writel(0, base + WINDOW_BASE(i)); + writel(0, base + WINDOW_SIZE(i)); + if (i < 4) + writel(0, base + WINDOW_REMAP_HIGH(i)); + } + + for (i = 0; i < dram->num_cs; i++) { + struct mbus_dram_window *cs = dram->cs + i; + + writel((cs->base & 0xffff0000) | + (cs->mbus_attr << 8) | + dram->mbus_dram_target_id, base + WINDOW_BASE(i)); + writel((cs->size - 1) & 0xffff0000, base + WINDOW_SIZE(i)); + + win_enable |= (1 << i); + win_enable |= 3 << (16 + (2 * i)); + } + + writel(win_enable, base + WINDOW_BAR_ENABLE(0)); + writel(win_enable, base + WINDOW_BAR_ENABLE(1)); +} + +static struct platform_driver mv_xor_driver = { + .probe = mv_xor_probe, + .remove = __devexit_p(mv_xor_remove), + .driver = { + .owner = THIS_MODULE, + .name = MV_XOR_NAME, + }, +}; + +static int mv_xor_shared_probe(struct platform_device *pdev) +{ + struct mv_xor_platform_shared_data *msd = pdev->dev.platform_data; + struct mv_xor_shared_private *msp; + struct resource *res; + + dev_printk(KERN_NOTICE, &pdev->dev, "Marvell shared XOR driver\n"); + + msp = devm_kzalloc(&pdev->dev, sizeof(*msp), GFP_KERNEL); + if (!msp) + return -ENOMEM; + + res = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!res) + return -ENODEV; + + msp->xor_base = devm_ioremap(&pdev->dev, res->start, + res->end - res->start + 1); + if (!msp->xor_base) + return -EBUSY; + + res = platform_get_resource(pdev, IORESOURCE_MEM, 1); + if (!res) + return -ENODEV; + + msp->xor_high_base = devm_ioremap(&pdev->dev, res->start, + res->end - res->start + 1); + if (!msp->xor_high_base) + return -EBUSY; + + platform_set_drvdata(pdev, msp); + + /* + * (Re-)program MBUS remapping windows if we are asked to. + */ + if (msd != NULL && msd->dram != NULL) + mv_xor_conf_mbus_windows(msp, msd->dram); + + return 0; +} + +static int mv_xor_shared_remove(struct platform_device *pdev) +{ + return 0; +} + +static struct platform_driver mv_xor_shared_driver = { + .probe = mv_xor_shared_probe, + .remove = mv_xor_shared_remove, + .driver = { + .owner = THIS_MODULE, + .name = MV_XOR_SHARED_NAME, + }, +}; + + +static int __init mv_xor_init(void) +{ + int rc; + + rc = platform_driver_register(&mv_xor_shared_driver); + if (!rc) { + rc = platform_driver_register(&mv_xor_driver); + if (rc) + platform_driver_unregister(&mv_xor_shared_driver); + } + return rc; +} +module_init(mv_xor_init); + +/* it's currently unsafe to unload this module */ +#if 0 +static void __exit mv_xor_exit(void) +{ + platform_driver_unregister(&mv_xor_driver); + platform_driver_unregister(&mv_xor_shared_driver); + return; +} + +module_exit(mv_xor_exit); +#endif + +MODULE_AUTHOR("Saeed Bishara <saeed@marvell.com>"); +MODULE_DESCRIPTION("DMA engine driver for Marvell's XOR engine"); +MODULE_LICENSE("GPL"); diff --git a/drivers/dma/mv_xor.h b/drivers/dma/mv_xor.h new file mode 100644 index 00000000000..977b592e976 --- /dev/null +++ b/drivers/dma/mv_xor.h @@ -0,0 +1,183 @@ +/* + * Copyright (C) 2007, 2008, Marvell International Ltd. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms and conditions of the GNU General Public License, + * version 2, as published by the Free Software Foundation. + * + * This program is distributed in the hope it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License + * for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software Foundation, + * Inc., 51 Franklin St - Fifth Floor, Boston, MA 02110-1301 USA. + */ + +#ifndef MV_XOR_H +#define MV_XOR_H + +#include <linux/types.h> +#include <linux/io.h> +#include <linux/dmaengine.h> +#include <linux/interrupt.h> + +#define USE_TIMER +#define MV_XOR_SLOT_SIZE 64 +#define MV_XOR_THRESHOLD 1 + +#define XOR_OPERATION_MODE_XOR 0 +#define XOR_OPERATION_MODE_MEMCPY 2 +#define XOR_OPERATION_MODE_MEMSET 4 + +#define XOR_CURR_DESC(chan) (chan->mmr_base + 0x210 + (chan->idx * 4)) +#define XOR_NEXT_DESC(chan) (chan->mmr_base + 0x200 + (chan->idx * 4)) +#define XOR_BYTE_COUNT(chan) (chan->mmr_base + 0x220 + (chan->idx * 4)) +#define XOR_DEST_POINTER(chan) (chan->mmr_base + 0x2B0 + (chan->idx * 4)) +#define XOR_BLOCK_SIZE(chan) (chan->mmr_base + 0x2C0 + (chan->idx * 4)) +#define XOR_INIT_VALUE_LOW(chan) (chan->mmr_base + 0x2E0) +#define XOR_INIT_VALUE_HIGH(chan) (chan->mmr_base + 0x2E4) + +#define XOR_CONFIG(chan) (chan->mmr_base + 0x10 + (chan->idx * 4)) +#define XOR_ACTIVATION(chan) (chan->mmr_base + 0x20 + (chan->idx * 4)) +#define XOR_INTR_CAUSE(chan) (chan->mmr_base + 0x30) +#define XOR_INTR_MASK(chan) (chan->mmr_base + 0x40) +#define XOR_ERROR_CAUSE(chan) (chan->mmr_base + 0x50) +#define XOR_ERROR_ADDR(chan) (chan->mmr_base + 0x60) +#define XOR_INTR_MASK_VALUE 0x3F5 + +#define WINDOW_BASE(w) (0x250 + ((w) << 2)) +#define WINDOW_SIZE(w) (0x270 + ((w) << 2)) +#define WINDOW_REMAP_HIGH(w) (0x290 + ((w) << 2)) +#define WINDOW_BAR_ENABLE(chan) (0x240 + ((chan) << 2)) + +struct mv_xor_shared_private { + void __iomem *xor_base; + void __iomem *xor_high_base; +}; + + +/** + * struct mv_xor_device - internal representation of a XOR device + * @pdev: Platform device + * @id: HW XOR Device selector + * @dma_desc_pool: base of DMA descriptor region (DMA address) + * @dma_desc_pool_virt: base of DMA descriptor region (CPU address) + * @common: embedded struct dma_device + */ +struct mv_xor_device { + struct platform_device *pdev; + int id; + dma_addr_t dma_desc_pool; + void *dma_desc_pool_virt; + struct dma_device common; + struct mv_xor_shared_private *shared; +}; + +/** + * struct mv_xor_chan - internal representation of a XOR channel + * @pending: allows batching of hardware operations + * @completed_cookie: identifier for the most recently completed operation + * @lock: serializes enqueue/dequeue operations to the descriptors pool + * @mmr_base: memory mapped register base + * @idx: the index of the xor channel + * @chain: device chain view of the descriptors + * @completed_slots: slots completed by HW but still need to be acked + * @device: parent device + * @common: common dmaengine channel object members + * @last_used: place holder for allocation to continue from where it left off + * @all_slots: complete domain of slots usable by the channel + * @slots_allocated: records the actual size of the descriptor slot pool + * @irq_tasklet: bottom half where mv_xor_slot_cleanup runs + */ +struct mv_xor_chan { + int pending; + dma_cookie_t completed_cookie; + spinlock_t lock; /* protects the descriptor slot pool */ + void __iomem *mmr_base; + unsigned int idx; + enum dma_transaction_type current_type; + struct list_head chain; + struct list_head completed_slots; + struct mv_xor_device *device; + struct dma_chan common; + struct mv_xor_desc_slot *last_used; + struct list_head all_slots; + int slots_allocated; + struct tasklet_struct irq_tasklet; +#ifdef USE_TIMER + unsigned long cleanup_time; + u32 current_on_last_cleanup; + dma_cookie_t is_complete_cookie; +#endif +}; + +/** + * struct mv_xor_desc_slot - software descriptor + * @slot_node: node on the mv_xor_chan.all_slots list + * @chain_node: node on the mv_xor_chan.chain list + * @completed_node: node on the mv_xor_chan.completed_slots list + * @hw_desc: virtual address of the hardware descriptor chain + * @phys: hardware address of the hardware descriptor chain + * @group_head: first operation in a transaction + * @slot_cnt: total slots used in an transaction (group of operations) + * @slots_per_op: number of slots per operation + * @idx: pool index + * @unmap_src_cnt: number of xor sources + * @unmap_len: transaction bytecount + * @tx_list: list of slots that make up a multi-descriptor transaction + * @async_tx: support for the async_tx api + * @xor_check_result: result of zero sum + * @crc32_result: result crc calculation + */ +struct mv_xor_desc_slot { + struct list_head slot_node; + struct list_head chain_node; + struct list_head completed_node; + enum dma_transaction_type type; + void *hw_desc; + struct mv_xor_desc_slot *group_head; + u16 slot_cnt; + u16 slots_per_op; + u16 idx; + u16 unmap_src_cnt; + u32 value; + size_t unmap_len; + struct list_head tx_list; + struct dma_async_tx_descriptor async_tx; + union { + u32 *xor_check_result; + u32 *crc32_result; + }; +#ifdef USE_TIMER + unsigned long arrival_time; + struct timer_list timeout; +#endif +}; + +/* This structure describes XOR descriptor size 64bytes */ +struct mv_xor_desc { + u32 status; /* descriptor execution status */ + u32 crc32_result; /* result of CRC-32 calculation */ + u32 desc_command; /* type of operation to be carried out */ + u32 phy_next_desc; /* next descriptor address pointer */ + u32 byte_count; /* size of src/dst blocks in bytes */ + u32 phy_dest_addr; /* destination block address */ + u32 phy_src_addr[8]; /* source block addresses */ + u32 reserved0; + u32 reserved1; +}; + +#define to_mv_sw_desc(addr_hw_desc) \ + container_of(addr_hw_desc, struct mv_xor_desc_slot, hw_desc) + +#define mv_hw_desc_slot_idx(hw_desc, idx) \ + ((void *)(((unsigned long)hw_desc) + ((idx) << 5))) + +#define MV_XOR_MIN_BYTE_COUNT (128) +#define XOR_MAX_BYTE_COUNT ((16 * 1024 * 1024) - 1) +#define MV_XOR_MAX_BYTE_COUNT XOR_MAX_BYTE_COUNT + + +#endif diff --git a/drivers/dma/ppc4xx/Makefile b/drivers/dma/ppc4xx/Makefile new file mode 100644 index 00000000000..b3d259b3e52 --- /dev/null +++ b/drivers/dma/ppc4xx/Makefile @@ -0,0 +1 @@ +obj-$(CONFIG_AMCC_PPC440SPE_ADMA) += adma.o diff --git a/drivers/dma/ppc4xx/adma.c b/drivers/dma/ppc4xx/adma.c new file mode 100644 index 00000000000..0a3478e910f --- /dev/null +++ b/drivers/dma/ppc4xx/adma.c @@ -0,0 +1,5027 @@ +/* + * Copyright (C) 2006-2009 DENX Software Engineering. + * + * Author: Yuri Tikhonov <yur@emcraft.com> + * + * Further porting to arch/powerpc by + * Anatolij Gustschin <agust@denx.de> + * + * This program is free software; you can redistribute it and/or modify it + * under the terms of the GNU General Public License as published by the Free + * Software Foundation; either version 2 of the License, or (at your option) + * any later version. + * + * This program is distributed in the hope that it will be useful, but WITHOUT + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for + * more details. + * + * You should have received a copy of the GNU General Public License along with + * this program; if not, write to the Free Software Foundation, Inc., 59 + * Temple Place - Suite 330, Boston, MA 02111-1307, USA. + * + * The full GNU General Public License is included in this distribution in the + * file called COPYING. + */ + +/* + * This driver supports the asynchrounous DMA copy and RAID engines available + * on the AMCC PPC440SPe Processors. + * Based on the Intel Xscale(R) family of I/O Processors (IOP 32x, 33x, 134x) + * ADMA driver written by D.Williams. + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/async_tx.h> +#include <linux/delay.h> +#include <linux/dma-mapping.h> +#include <linux/spinlock.h> +#include <linux/interrupt.h> +#include <linux/uaccess.h> +#include <linux/proc_fs.h> +#include <linux/of.h> +#include <linux/of_platform.h> +#include <asm/dcr.h> +#include <asm/dcr-regs.h> +#include "adma.h" + +enum ppc_adma_init_code { + PPC_ADMA_INIT_OK = 0, + PPC_ADMA_INIT_MEMRES, + PPC_ADMA_INIT_MEMREG, + PPC_ADMA_INIT_ALLOC, + PPC_ADMA_INIT_COHERENT, + PPC_ADMA_INIT_CHANNEL, + PPC_ADMA_INIT_IRQ1, + PPC_ADMA_INIT_IRQ2, + PPC_ADMA_INIT_REGISTER +}; + +static char *ppc_adma_errors[] = { + [PPC_ADMA_INIT_OK] = "ok", + [PPC_ADMA_INIT_MEMRES] = "failed to get memory resource", + [PPC_ADMA_INIT_MEMREG] = "failed to request memory region", + [PPC_ADMA_INIT_ALLOC] = "failed to allocate memory for adev " + "structure", + [PPC_ADMA_INIT_COHERENT] = "failed to allocate coherent memory for " + "hardware descriptors", + [PPC_ADMA_INIT_CHANNEL] = "failed to allocate memory for channel", + [PPC_ADMA_INIT_IRQ1] = "failed to request first irq", + [PPC_ADMA_INIT_IRQ2] = "failed to request second irq", + [PPC_ADMA_INIT_REGISTER] = "failed to register dma async device", +}; + +static enum ppc_adma_init_code +ppc440spe_adma_devices[PPC440SPE_ADMA_ENGINES_NUM]; + +struct ppc_dma_chan_ref { + struct dma_chan *chan; + struct list_head node; +}; + +/* The list of channels exported by ppc440spe ADMA */ +struct list_head +ppc440spe_adma_chan_list = LIST_HEAD_INIT(ppc440spe_adma_chan_list); + +/* This flag is set when want to refetch the xor chain in the interrupt + * handler + */ +static u32 do_xor_refetch; + +/* Pointer to DMA0, DMA1 CP/CS FIFO */ +static void *ppc440spe_dma_fifo_buf; + +/* Pointers to last submitted to DMA0, DMA1 CDBs */ +static struct ppc440spe_adma_desc_slot *chan_last_sub[3]; +static struct ppc440spe_adma_desc_slot *chan_first_cdb[3]; + +/* Pointer to last linked and submitted xor CB */ +static struct ppc440spe_adma_desc_slot *xor_last_linked; +static struct ppc440spe_adma_desc_slot *xor_last_submit; + +/* This array is used in data-check operations for storing a pattern */ +static char ppc440spe_qword[16]; + +static atomic_t ppc440spe_adma_err_irq_ref; +static dcr_host_t ppc440spe_mq_dcr_host; +static unsigned int ppc440spe_mq_dcr_len; + +/* Since RXOR operations use the common register (MQ0_CF2H) for setting-up + * the block size in transactions, then we do not allow to activate more than + * only one RXOR transactions simultaneously. So use this var to store + * the information about is RXOR currently active (PPC440SPE_RXOR_RUN bit is + * set) or not (PPC440SPE_RXOR_RUN is clear). + */ +static unsigned long ppc440spe_rxor_state; + +/* These are used in enable & check routines + */ +static u32 ppc440spe_r6_enabled; +static struct ppc440spe_adma_chan *ppc440spe_r6_tchan; +static struct completion ppc440spe_r6_test_comp; + +static int ppc440spe_adma_dma2rxor_prep_src( + struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_rxor *cursor, int index, + int src_cnt, u32 addr); +static void ppc440spe_adma_dma2rxor_set_src( + struct ppc440spe_adma_desc_slot *desc, + int index, dma_addr_t addr); +static void ppc440spe_adma_dma2rxor_set_mult( + struct ppc440spe_adma_desc_slot *desc, + int index, u8 mult); + +#ifdef ADMA_LL_DEBUG +#define ADMA_LL_DBG(x) ({ if (1) x; 0; }) +#else +#define ADMA_LL_DBG(x) ({ if (0) x; 0; }) +#endif + +static void print_cb(struct ppc440spe_adma_chan *chan, void *block) +{ + struct dma_cdb *cdb; + struct xor_cb *cb; + int i; + + switch (chan->device->id) { + case 0: + case 1: + cdb = block; + + pr_debug("CDB at %p [%d]:\n" + "\t attr 0x%02x opc 0x%02x cnt 0x%08x\n" + "\t sg1u 0x%08x sg1l 0x%08x\n" + "\t sg2u 0x%08x sg2l 0x%08x\n" + "\t sg3u 0x%08x sg3l 0x%08x\n", + cdb, chan->device->id, + cdb->attr, cdb->opc, le32_to_cpu(cdb->cnt), + le32_to_cpu(cdb->sg1u), le32_to_cpu(cdb->sg1l), + le32_to_cpu(cdb->sg2u), le32_to_cpu(cdb->sg2l), + le32_to_cpu(cdb->sg3u), le32_to_cpu(cdb->sg3l) + ); + break; + case 2: + cb = block; + + pr_debug("CB at %p [%d]:\n" + "\t cbc 0x%08x cbbc 0x%08x cbs 0x%08x\n" + "\t cbtah 0x%08x cbtal 0x%08x\n" + "\t cblah 0x%08x cblal 0x%08x\n", + cb, chan->device->id, + cb->cbc, cb->cbbc, cb->cbs, + cb->cbtah, cb->cbtal, + cb->cblah, cb->cblal); + for (i = 0; i < 16; i++) { + if (i && !cb->ops[i].h && !cb->ops[i].l) + continue; + pr_debug("\t ops[%2d]: h 0x%08x l 0x%08x\n", + i, cb->ops[i].h, cb->ops[i].l); + } + break; + } +} + +static void print_cb_list(struct ppc440spe_adma_chan *chan, + struct ppc440spe_adma_desc_slot *iter) +{ + for (; iter; iter = iter->hw_next) + print_cb(chan, iter->hw_desc); +} + +static void prep_dma_xor_dbg(int id, dma_addr_t dst, dma_addr_t *src, + unsigned int src_cnt) +{ + int i; + + pr_debug("\n%s(%d):\nsrc: ", __func__, id); + for (i = 0; i < src_cnt; i++) + pr_debug("\t0x%016llx ", src[i]); + pr_debug("dst:\n\t0x%016llx\n", dst); +} + +static void prep_dma_pq_dbg(int id, dma_addr_t *dst, dma_addr_t *src, + unsigned int src_cnt) +{ + int i; + + pr_debug("\n%s(%d):\nsrc: ", __func__, id); + for (i = 0; i < src_cnt; i++) + pr_debug("\t0x%016llx ", src[i]); + pr_debug("dst: "); + for (i = 0; i < 2; i++) + pr_debug("\t0x%016llx ", dst[i]); +} + +static void prep_dma_pqzero_sum_dbg(int id, dma_addr_t *src, + unsigned int src_cnt, + const unsigned char *scf) +{ + int i; + + pr_debug("\n%s(%d):\nsrc(coef): ", __func__, id); + if (scf) { + for (i = 0; i < src_cnt; i++) + pr_debug("\t0x%016llx(0x%02x) ", src[i], scf[i]); + } else { + for (i = 0; i < src_cnt; i++) + pr_debug("\t0x%016llx(no) ", src[i]); + } + + pr_debug("dst: "); + for (i = 0; i < 2; i++) + pr_debug("\t0x%016llx ", src[src_cnt + i]); +} + +/****************************************************************************** + * Command (Descriptor) Blocks low-level routines + ******************************************************************************/ +/** + * ppc440spe_desc_init_interrupt - initialize the descriptor for INTERRUPT + * pseudo operation + */ +static void ppc440spe_desc_init_interrupt(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan) +{ + struct xor_cb *p; + + switch (chan->device->id) { + case PPC440SPE_XOR_ID: + p = desc->hw_desc; + memset(desc->hw_desc, 0, sizeof(struct xor_cb)); + /* NOP with Command Block Complete Enable */ + p->cbc = XOR_CBCR_CBCE_BIT; + break; + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + memset(desc->hw_desc, 0, sizeof(struct dma_cdb)); + /* NOP with interrupt */ + set_bit(PPC440SPE_DESC_INT, &desc->flags); + break; + default: + printk(KERN_ERR "Unsupported id %d in %s\n", chan->device->id, + __func__); + break; + } +} + +/** + * ppc440spe_desc_init_null_xor - initialize the descriptor for NULL XOR + * pseudo operation + */ +static void ppc440spe_desc_init_null_xor(struct ppc440spe_adma_desc_slot *desc) +{ + memset(desc->hw_desc, 0, sizeof(struct xor_cb)); + desc->hw_next = NULL; + desc->src_cnt = 0; + desc->dst_cnt = 1; +} + +/** + * ppc440spe_desc_init_xor - initialize the descriptor for XOR operation + */ +static void ppc440spe_desc_init_xor(struct ppc440spe_adma_desc_slot *desc, + int src_cnt, unsigned long flags) +{ + struct xor_cb *hw_desc = desc->hw_desc; + + memset(desc->hw_desc, 0, sizeof(struct xor_cb)); + desc->hw_next = NULL; + desc->src_cnt = src_cnt; + desc->dst_cnt = 1; + + hw_desc->cbc = XOR_CBCR_TGT_BIT | src_cnt; + if (flags & DMA_PREP_INTERRUPT) + /* Enable interrupt on completion */ + hw_desc->cbc |= XOR_CBCR_CBCE_BIT; +} + +/** + * ppc440spe_desc_init_dma2pq - initialize the descriptor for PQ + * operation in DMA2 controller + */ +static void ppc440spe_desc_init_dma2pq(struct ppc440spe_adma_desc_slot *desc, + int dst_cnt, int src_cnt, unsigned long flags) +{ + struct xor_cb *hw_desc = desc->hw_desc; + + memset(desc->hw_desc, 0, sizeof(struct xor_cb)); + desc->hw_next = NULL; + desc->src_cnt = src_cnt; + desc->dst_cnt = dst_cnt; + memset(desc->reverse_flags, 0, sizeof(desc->reverse_flags)); + desc->descs_per_op = 0; + + hw_desc->cbc = XOR_CBCR_TGT_BIT; + if (flags & DMA_PREP_INTERRUPT) + /* Enable interrupt on completion */ + hw_desc->cbc |= XOR_CBCR_CBCE_BIT; +} + +#define DMA_CTRL_FLAGS_LAST DMA_PREP_FENCE +#define DMA_PREP_ZERO_P (DMA_CTRL_FLAGS_LAST << 1) +#define DMA_PREP_ZERO_Q (DMA_PREP_ZERO_P << 1) + +/** + * ppc440spe_desc_init_dma01pq - initialize the descriptors for PQ operation + * with DMA0/1 + */ +static void ppc440spe_desc_init_dma01pq(struct ppc440spe_adma_desc_slot *desc, + int dst_cnt, int src_cnt, unsigned long flags, + unsigned long op) +{ + struct dma_cdb *hw_desc; + struct ppc440spe_adma_desc_slot *iter; + u8 dopc; + + /* Common initialization of a PQ descriptors chain */ + set_bits(op, &desc->flags); + desc->src_cnt = src_cnt; + desc->dst_cnt = dst_cnt; + + /* WXOR MULTICAST if both P and Q are being computed + * MV_SG1_SG2 if Q only + */ + dopc = (desc->dst_cnt == DMA_DEST_MAX_NUM) ? + DMA_CDB_OPC_MULTICAST : DMA_CDB_OPC_MV_SG1_SG2; + + list_for_each_entry(iter, &desc->group_list, chain_node) { + hw_desc = iter->hw_desc; + memset(iter->hw_desc, 0, sizeof(struct dma_cdb)); + + if (likely(!list_is_last(&iter->chain_node, + &desc->group_list))) { + /* set 'next' pointer */ + iter->hw_next = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, chain_node); + clear_bit(PPC440SPE_DESC_INT, &iter->flags); + } else { + /* this is the last descriptor. + * this slot will be pasted from ADMA level + * each time it wants to configure parameters + * of the transaction (src, dst, ...) + */ + iter->hw_next = NULL; + if (flags & DMA_PREP_INTERRUPT) + set_bit(PPC440SPE_DESC_INT, &iter->flags); + else + clear_bit(PPC440SPE_DESC_INT, &iter->flags); + } + } + + /* Set OPS depending on WXOR/RXOR type of operation */ + if (!test_bit(PPC440SPE_DESC_RXOR, &desc->flags)) { + /* This is a WXOR only chain: + * - first descriptors are for zeroing destinations + * if PPC440SPE_ZERO_P/Q set; + * - descriptors remained are for GF-XOR operations. + */ + iter = list_first_entry(&desc->group_list, + struct ppc440spe_adma_desc_slot, + chain_node); + + if (test_bit(PPC440SPE_ZERO_P, &desc->flags)) { + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; + iter = list_first_entry(&iter->chain_node, + struct ppc440spe_adma_desc_slot, + chain_node); + } + + if (test_bit(PPC440SPE_ZERO_Q, &desc->flags)) { + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; + iter = list_first_entry(&iter->chain_node, + struct ppc440spe_adma_desc_slot, + chain_node); + } + + list_for_each_entry_from(iter, &desc->group_list, chain_node) { + hw_desc = iter->hw_desc; + hw_desc->opc = dopc; + } + } else { + /* This is either RXOR-only or mixed RXOR/WXOR */ + + /* The first 1 or 2 slots in chain are always RXOR, + * if need to calculate P & Q, then there are two + * RXOR slots; if only P or only Q, then there is one + */ + iter = list_first_entry(&desc->group_list, + struct ppc440spe_adma_desc_slot, + chain_node); + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; + + if (desc->dst_cnt == DMA_DEST_MAX_NUM) { + iter = list_first_entry(&iter->chain_node, + struct ppc440spe_adma_desc_slot, + chain_node); + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; + } + + /* The remaining descs (if any) are WXORs */ + if (test_bit(PPC440SPE_DESC_WXOR, &desc->flags)) { + iter = list_first_entry(&iter->chain_node, + struct ppc440spe_adma_desc_slot, + chain_node); + list_for_each_entry_from(iter, &desc->group_list, + chain_node) { + hw_desc = iter->hw_desc; + hw_desc->opc = dopc; + } + } + } +} + +/** + * ppc440spe_desc_init_dma01pqzero_sum - initialize the descriptor + * for PQ_ZERO_SUM operation + */ +static void ppc440spe_desc_init_dma01pqzero_sum( + struct ppc440spe_adma_desc_slot *desc, + int dst_cnt, int src_cnt) +{ + struct dma_cdb *hw_desc; + struct ppc440spe_adma_desc_slot *iter; + int i = 0; + u8 dopc = (dst_cnt == 2) ? DMA_CDB_OPC_MULTICAST : + DMA_CDB_OPC_MV_SG1_SG2; + /* + * Initialize starting from 2nd or 3rd descriptor dependent + * on dst_cnt. First one or two slots are for cloning P + * and/or Q to chan->pdest and/or chan->qdest as we have + * to preserve original P/Q. + */ + iter = list_first_entry(&desc->group_list, + struct ppc440spe_adma_desc_slot, chain_node); + iter = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, chain_node); + + if (dst_cnt > 1) { + iter = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, chain_node); + } + /* initialize each source descriptor in chain */ + list_for_each_entry_from(iter, &desc->group_list, chain_node) { + hw_desc = iter->hw_desc; + memset(iter->hw_desc, 0, sizeof(struct dma_cdb)); + iter->src_cnt = 0; + iter->dst_cnt = 0; + + /* This is a ZERO_SUM operation: + * - <src_cnt> descriptors starting from 2nd or 3rd + * descriptor are for GF-XOR operations; + * - remaining <dst_cnt> descriptors are for checking the result + */ + if (i++ < src_cnt) + /* MV_SG1_SG2 if only Q is being verified + * MULTICAST if both P and Q are being verified + */ + hw_desc->opc = dopc; + else + /* DMA_CDB_OPC_DCHECK128 operation */ + hw_desc->opc = DMA_CDB_OPC_DCHECK128; + + if (likely(!list_is_last(&iter->chain_node, + &desc->group_list))) { + /* set 'next' pointer */ + iter->hw_next = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + } else { + /* this is the last descriptor. + * this slot will be pasted from ADMA level + * each time it wants to configure parameters + * of the transaction (src, dst, ...) + */ + iter->hw_next = NULL; + /* always enable interrupt generation since we get + * the status of pqzero from the handler + */ + set_bit(PPC440SPE_DESC_INT, &iter->flags); + } + } + desc->src_cnt = src_cnt; + desc->dst_cnt = dst_cnt; +} + +/** + * ppc440spe_desc_init_memcpy - initialize the descriptor for MEMCPY operation + */ +static void ppc440spe_desc_init_memcpy(struct ppc440spe_adma_desc_slot *desc, + unsigned long flags) +{ + struct dma_cdb *hw_desc = desc->hw_desc; + + memset(desc->hw_desc, 0, sizeof(struct dma_cdb)); + desc->hw_next = NULL; + desc->src_cnt = 1; + desc->dst_cnt = 1; + + if (flags & DMA_PREP_INTERRUPT) + set_bit(PPC440SPE_DESC_INT, &desc->flags); + else + clear_bit(PPC440SPE_DESC_INT, &desc->flags); + + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; +} + +/** + * ppc440spe_desc_init_memset - initialize the descriptor for MEMSET operation + */ +static void ppc440spe_desc_init_memset(struct ppc440spe_adma_desc_slot *desc, + int value, unsigned long flags) +{ + struct dma_cdb *hw_desc = desc->hw_desc; + + memset(desc->hw_desc, 0, sizeof(struct dma_cdb)); + desc->hw_next = NULL; + desc->src_cnt = 1; + desc->dst_cnt = 1; + + if (flags & DMA_PREP_INTERRUPT) + set_bit(PPC440SPE_DESC_INT, &desc->flags); + else + clear_bit(PPC440SPE_DESC_INT, &desc->flags); + + hw_desc->sg1u = hw_desc->sg1l = cpu_to_le32((u32)value); + hw_desc->sg3u = hw_desc->sg3l = cpu_to_le32((u32)value); + hw_desc->opc = DMA_CDB_OPC_DFILL128; +} + +/** + * ppc440spe_desc_set_src_addr - set source address into the descriptor + */ +static void ppc440spe_desc_set_src_addr(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan, + int src_idx, dma_addr_t addrh, + dma_addr_t addrl) +{ + struct dma_cdb *dma_hw_desc; + struct xor_cb *xor_hw_desc; + phys_addr_t addr64, tmplow, tmphi; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + if (!addrh) { + addr64 = addrl; + tmphi = (addr64 >> 32); + tmplow = (addr64 & 0xFFFFFFFF); + } else { + tmphi = addrh; + tmplow = addrl; + } + dma_hw_desc = desc->hw_desc; + dma_hw_desc->sg1l = cpu_to_le32((u32)tmplow); + dma_hw_desc->sg1u |= cpu_to_le32((u32)tmphi); + break; + case PPC440SPE_XOR_ID: + xor_hw_desc = desc->hw_desc; + xor_hw_desc->ops[src_idx].l = addrl; + xor_hw_desc->ops[src_idx].h |= addrh; + break; + } +} + +/** + * ppc440spe_desc_set_src_mult - set source address mult into the descriptor + */ +static void ppc440spe_desc_set_src_mult(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan, u32 mult_index, + int sg_index, unsigned char mult_value) +{ + struct dma_cdb *dma_hw_desc; + struct xor_cb *xor_hw_desc; + u32 *psgu; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + dma_hw_desc = desc->hw_desc; + + switch (sg_index) { + /* for RXOR operations set multiplier + * into source cued address + */ + case DMA_CDB_SG_SRC: + psgu = &dma_hw_desc->sg1u; + break; + /* for WXOR operations set multiplier + * into destination cued address(es) + */ + case DMA_CDB_SG_DST1: + psgu = &dma_hw_desc->sg2u; + break; + case DMA_CDB_SG_DST2: + psgu = &dma_hw_desc->sg3u; + break; + default: + BUG(); + } + + *psgu |= cpu_to_le32(mult_value << mult_index); + break; + case PPC440SPE_XOR_ID: + xor_hw_desc = desc->hw_desc; + break; + default: + BUG(); + } +} + +/** + * ppc440spe_desc_set_dest_addr - set destination address into the descriptor + */ +static void ppc440spe_desc_set_dest_addr(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan, + dma_addr_t addrh, dma_addr_t addrl, + u32 dst_idx) +{ + struct dma_cdb *dma_hw_desc; + struct xor_cb *xor_hw_desc; + phys_addr_t addr64, tmphi, tmplow; + u32 *psgu, *psgl; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + if (!addrh) { + addr64 = addrl; + tmphi = (addr64 >> 32); + tmplow = (addr64 & 0xFFFFFFFF); + } else { + tmphi = addrh; + tmplow = addrl; + } + dma_hw_desc = desc->hw_desc; + + psgu = dst_idx ? &dma_hw_desc->sg3u : &dma_hw_desc->sg2u; + psgl = dst_idx ? &dma_hw_desc->sg3l : &dma_hw_desc->sg2l; + + *psgl = cpu_to_le32((u32)tmplow); + *psgu |= cpu_to_le32((u32)tmphi); + break; + case PPC440SPE_XOR_ID: + xor_hw_desc = desc->hw_desc; + xor_hw_desc->cbtal = addrl; + xor_hw_desc->cbtah |= addrh; + break; + } +} + +/** + * ppc440spe_desc_set_byte_count - set number of data bytes involved + * into the operation + */ +static void ppc440spe_desc_set_byte_count(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan, + u32 byte_count) +{ + struct dma_cdb *dma_hw_desc; + struct xor_cb *xor_hw_desc; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + dma_hw_desc = desc->hw_desc; + dma_hw_desc->cnt = cpu_to_le32(byte_count); + break; + case PPC440SPE_XOR_ID: + xor_hw_desc = desc->hw_desc; + xor_hw_desc->cbbc = byte_count; + break; + } +} + +/** + * ppc440spe_desc_set_rxor_block_size - set RXOR block size + */ +static inline void ppc440spe_desc_set_rxor_block_size(u32 byte_count) +{ + /* assume that byte_count is aligned on the 512-boundary; + * thus write it directly to the register (bits 23:31 are + * reserved there). + */ + dcr_write(ppc440spe_mq_dcr_host, DCRN_MQ0_CF2H, byte_count); +} + +/** + * ppc440spe_desc_set_dcheck - set CHECK pattern + */ +static void ppc440spe_desc_set_dcheck(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan, u8 *qword) +{ + struct dma_cdb *dma_hw_desc; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + dma_hw_desc = desc->hw_desc; + iowrite32(qword[0], &dma_hw_desc->sg3l); + iowrite32(qword[4], &dma_hw_desc->sg3u); + iowrite32(qword[8], &dma_hw_desc->sg2l); + iowrite32(qword[12], &dma_hw_desc->sg2u); + break; + default: + BUG(); + } +} + +/** + * ppc440spe_xor_set_link - set link address in xor CB + */ +static void ppc440spe_xor_set_link(struct ppc440spe_adma_desc_slot *prev_desc, + struct ppc440spe_adma_desc_slot *next_desc) +{ + struct xor_cb *xor_hw_desc = prev_desc->hw_desc; + + if (unlikely(!next_desc || !(next_desc->phys))) { + printk(KERN_ERR "%s: next_desc=0x%p; next_desc->phys=0x%llx\n", + __func__, next_desc, + next_desc ? next_desc->phys : 0); + BUG(); + } + + xor_hw_desc->cbs = 0; + xor_hw_desc->cblal = next_desc->phys; + xor_hw_desc->cblah = 0; + xor_hw_desc->cbc |= XOR_CBCR_LNK_BIT; +} + +/** + * ppc440spe_desc_set_link - set the address of descriptor following this + * descriptor in chain + */ +static void ppc440spe_desc_set_link(struct ppc440spe_adma_chan *chan, + struct ppc440spe_adma_desc_slot *prev_desc, + struct ppc440spe_adma_desc_slot *next_desc) +{ + unsigned long flags; + struct ppc440spe_adma_desc_slot *tail = next_desc; + + if (unlikely(!prev_desc || !next_desc || + (prev_desc->hw_next && prev_desc->hw_next != next_desc))) { + /* If previous next is overwritten something is wrong. + * though we may refetch from append to initiate list + * processing; in this case - it's ok. + */ + printk(KERN_ERR "%s: prev_desc=0x%p; next_desc=0x%p; " + "prev->hw_next=0x%p\n", __func__, prev_desc, + next_desc, prev_desc ? prev_desc->hw_next : 0); + BUG(); + } + + local_irq_save(flags); + + /* do s/w chaining both for DMA and XOR descriptors */ + prev_desc->hw_next = next_desc; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + break; + case PPC440SPE_XOR_ID: + /* bind descriptor to the chain */ + while (tail->hw_next) + tail = tail->hw_next; + xor_last_linked = tail; + + if (prev_desc == xor_last_submit) + /* do not link to the last submitted CB */ + break; + ppc440spe_xor_set_link(prev_desc, next_desc); + break; + } + + local_irq_restore(flags); +} + +/** + * ppc440spe_desc_get_src_addr - extract the source address from the descriptor + */ +static u32 ppc440spe_desc_get_src_addr(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan, int src_idx) +{ + struct dma_cdb *dma_hw_desc; + struct xor_cb *xor_hw_desc; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + dma_hw_desc = desc->hw_desc; + /* May have 0, 1, 2, or 3 sources */ + switch (dma_hw_desc->opc) { + case DMA_CDB_OPC_NO_OP: + case DMA_CDB_OPC_DFILL128: + return 0; + case DMA_CDB_OPC_DCHECK128: + if (unlikely(src_idx)) { + printk(KERN_ERR "%s: try to get %d source for" + " DCHECK128\n", __func__, src_idx); + BUG(); + } + return le32_to_cpu(dma_hw_desc->sg1l); + case DMA_CDB_OPC_MULTICAST: + case DMA_CDB_OPC_MV_SG1_SG2: + if (unlikely(src_idx > 2)) { + printk(KERN_ERR "%s: try to get %d source from" + " DMA descr\n", __func__, src_idx); + BUG(); + } + if (src_idx) { + if (le32_to_cpu(dma_hw_desc->sg1u) & + DMA_CUED_XOR_WIN_MSK) { + u8 region; + + if (src_idx == 1) + return le32_to_cpu( + dma_hw_desc->sg1l) + + desc->unmap_len; + + region = (le32_to_cpu( + dma_hw_desc->sg1u)) >> + DMA_CUED_REGION_OFF; + + region &= DMA_CUED_REGION_MSK; + switch (region) { + case DMA_RXOR123: + return le32_to_cpu( + dma_hw_desc->sg1l) + + (desc->unmap_len << 1); + case DMA_RXOR124: + return le32_to_cpu( + dma_hw_desc->sg1l) + + (desc->unmap_len * 3); + case DMA_RXOR125: + return le32_to_cpu( + dma_hw_desc->sg1l) + + (desc->unmap_len << 2); + default: + printk(KERN_ERR + "%s: try to" + " get src3 for region %02x" + "PPC440SPE_DESC_RXOR12?\n", + __func__, region); + BUG(); + } + } else { + printk(KERN_ERR + "%s: try to get %d" + " source for non-cued descr\n", + __func__, src_idx); + BUG(); + } + } + return le32_to_cpu(dma_hw_desc->sg1l); + default: + printk(KERN_ERR "%s: unknown OPC 0x%02x\n", + __func__, dma_hw_desc->opc); + BUG(); + } + return le32_to_cpu(dma_hw_desc->sg1l); + case PPC440SPE_XOR_ID: + /* May have up to 16 sources */ + xor_hw_desc = desc->hw_desc; + return xor_hw_desc->ops[src_idx].l; + } + return 0; +} + +/** + * ppc440spe_desc_get_dest_addr - extract the destination address from the + * descriptor + */ +static u32 ppc440spe_desc_get_dest_addr(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan, int idx) +{ + struct dma_cdb *dma_hw_desc; + struct xor_cb *xor_hw_desc; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + dma_hw_desc = desc->hw_desc; + + if (likely(!idx)) + return le32_to_cpu(dma_hw_desc->sg2l); + return le32_to_cpu(dma_hw_desc->sg3l); + case PPC440SPE_XOR_ID: + xor_hw_desc = desc->hw_desc; + return xor_hw_desc->cbtal; + } + return 0; +} + +/** + * ppc440spe_desc_get_src_num - extract the number of source addresses from + * the descriptor + */ +static u32 ppc440spe_desc_get_src_num(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan) +{ + struct dma_cdb *dma_hw_desc; + struct xor_cb *xor_hw_desc; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + dma_hw_desc = desc->hw_desc; + + switch (dma_hw_desc->opc) { + case DMA_CDB_OPC_NO_OP: + case DMA_CDB_OPC_DFILL128: + return 0; + case DMA_CDB_OPC_DCHECK128: + return 1; + case DMA_CDB_OPC_MV_SG1_SG2: + case DMA_CDB_OPC_MULTICAST: + /* + * Only for RXOR operations we have more than + * one source + */ + if (le32_to_cpu(dma_hw_desc->sg1u) & + DMA_CUED_XOR_WIN_MSK) { + /* RXOR op, there are 2 or 3 sources */ + if (((le32_to_cpu(dma_hw_desc->sg1u) >> + DMA_CUED_REGION_OFF) & + DMA_CUED_REGION_MSK) == DMA_RXOR12) { + /* RXOR 1-2 */ + return 2; + } else { + /* RXOR 1-2-3/1-2-4/1-2-5 */ + return 3; + } + } + return 1; + default: + printk(KERN_ERR "%s: unknown OPC 0x%02x\n", + __func__, dma_hw_desc->opc); + BUG(); + } + case PPC440SPE_XOR_ID: + /* up to 16 sources */ + xor_hw_desc = desc->hw_desc; + return xor_hw_desc->cbc & XOR_CDCR_OAC_MSK; + default: + BUG(); + } + return 0; +} + +/** + * ppc440spe_desc_get_dst_num - get the number of destination addresses in + * this descriptor + */ +static u32 ppc440spe_desc_get_dst_num(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan) +{ + struct dma_cdb *dma_hw_desc; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + /* May be 1 or 2 destinations */ + dma_hw_desc = desc->hw_desc; + switch (dma_hw_desc->opc) { + case DMA_CDB_OPC_NO_OP: + case DMA_CDB_OPC_DCHECK128: + return 0; + case DMA_CDB_OPC_MV_SG1_SG2: + case DMA_CDB_OPC_DFILL128: + return 1; + case DMA_CDB_OPC_MULTICAST: + if (desc->dst_cnt == 2) + return 2; + else + return 1; + default: + printk(KERN_ERR "%s: unknown OPC 0x%02x\n", + __func__, dma_hw_desc->opc); + BUG(); + } + case PPC440SPE_XOR_ID: + /* Always only 1 destination */ + return 1; + default: + BUG(); + } + return 0; +} + +/** + * ppc440spe_desc_get_link - get the address of the descriptor that + * follows this one + */ +static inline u32 ppc440spe_desc_get_link(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan) +{ + if (!desc->hw_next) + return 0; + + return desc->hw_next->phys; +} + +/** + * ppc440spe_desc_is_aligned - check alignment + */ +static inline int ppc440spe_desc_is_aligned( + struct ppc440spe_adma_desc_slot *desc, int num_slots) +{ + return (desc->idx & (num_slots - 1)) ? 0 : 1; +} + +/** + * ppc440spe_chan_xor_slot_count - get the number of slots necessary for + * XOR operation + */ +static int ppc440spe_chan_xor_slot_count(size_t len, int src_cnt, + int *slots_per_op) +{ + int slot_cnt; + + /* each XOR descriptor provides up to 16 source operands */ + slot_cnt = *slots_per_op = (src_cnt + XOR_MAX_OPS - 1)/XOR_MAX_OPS; + + if (likely(len <= PPC440SPE_ADMA_XOR_MAX_BYTE_COUNT)) + return slot_cnt; + + printk(KERN_ERR "%s: len %d > max %d !!\n", + __func__, len, PPC440SPE_ADMA_XOR_MAX_BYTE_COUNT); + BUG(); + return slot_cnt; +} + +/** + * ppc440spe_dma2_pq_slot_count - get the number of slots necessary for + * DMA2 PQ operation + */ +static int ppc440spe_dma2_pq_slot_count(dma_addr_t *srcs, + int src_cnt, size_t len) +{ + signed long long order = 0; + int state = 0; + int addr_count = 0; + int i; + for (i = 1; i < src_cnt; i++) { + dma_addr_t cur_addr = srcs[i]; + dma_addr_t old_addr = srcs[i-1]; + switch (state) { + case 0: + if (cur_addr == old_addr + len) { + /* direct RXOR */ + order = 1; + state = 1; + if (i == src_cnt-1) + addr_count++; + } else if (old_addr == cur_addr + len) { + /* reverse RXOR */ + order = -1; + state = 1; + if (i == src_cnt-1) + addr_count++; + } else { + state = 3; + } + break; + case 1: + if (i == src_cnt-2 || (order == -1 + && cur_addr != old_addr - len)) { + order = 0; + state = 0; + addr_count++; + } else if (cur_addr == old_addr + len*order) { + state = 2; + if (i == src_cnt-1) + addr_count++; + } else if (cur_addr == old_addr + 2*len) { + state = 2; + if (i == src_cnt-1) + addr_count++; + } else if (cur_addr == old_addr + 3*len) { + state = 2; + if (i == src_cnt-1) + addr_count++; + } else { + order = 0; + state = 0; + addr_count++; + } + break; + case 2: + order = 0; + state = 0; + addr_count++; + break; + } + if (state == 3) + break; + } + if (src_cnt <= 1 || (state != 1 && state != 2)) { + pr_err("%s: src_cnt=%d, state=%d, addr_count=%d, order=%lld\n", + __func__, src_cnt, state, addr_count, order); + for (i = 0; i < src_cnt; i++) + pr_err("\t[%d] 0x%llx \n", i, srcs[i]); + BUG(); + } + + return (addr_count + XOR_MAX_OPS - 1) / XOR_MAX_OPS; +} + + +/****************************************************************************** + * ADMA channel low-level routines + ******************************************************************************/ + +static u32 +ppc440spe_chan_get_current_descriptor(struct ppc440spe_adma_chan *chan); +static void ppc440spe_chan_append(struct ppc440spe_adma_chan *chan); + +/** + * ppc440spe_adma_device_clear_eot_status - interrupt ack to XOR or DMA engine + */ +static void ppc440spe_adma_device_clear_eot_status( + struct ppc440spe_adma_chan *chan) +{ + struct dma_regs *dma_reg; + struct xor_regs *xor_reg; + u8 *p = chan->device->dma_desc_pool_virt; + struct dma_cdb *cdb; + u32 rv, i; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + /* read FIFO to ack */ + dma_reg = chan->device->dma_reg; + while ((rv = ioread32(&dma_reg->csfpl))) { + i = rv & DMA_CDB_ADDR_MSK; + cdb = (struct dma_cdb *)&p[i - + (u32)chan->device->dma_desc_pool]; + + /* Clear opcode to ack. This is necessary for + * ZeroSum operations only + */ + cdb->opc = 0; + + if (test_bit(PPC440SPE_RXOR_RUN, + &ppc440spe_rxor_state)) { + /* probably this is a completed RXOR op, + * get pointer to CDB using the fact that + * physical and virtual addresses of CDB + * in pools have the same offsets + */ + if (le32_to_cpu(cdb->sg1u) & + DMA_CUED_XOR_BASE) { + /* this is a RXOR */ + clear_bit(PPC440SPE_RXOR_RUN, + &ppc440spe_rxor_state); + } + } + + if (rv & DMA_CDB_STATUS_MSK) { + /* ZeroSum check failed + */ + struct ppc440spe_adma_desc_slot *iter; + dma_addr_t phys = rv & ~DMA_CDB_MSK; + + /* + * Update the status of corresponding + * descriptor. + */ + list_for_each_entry(iter, &chan->chain, + chain_node) { + if (iter->phys == phys) + break; + } + /* + * if cannot find the corresponding + * slot it's a bug + */ + BUG_ON(&iter->chain_node == &chan->chain); + + if (iter->xor_check_result) { + if (test_bit(PPC440SPE_DESC_PCHECK, + &iter->flags)) { + *iter->xor_check_result |= + SUM_CHECK_P_RESULT; + } else + if (test_bit(PPC440SPE_DESC_QCHECK, + &iter->flags)) { + *iter->xor_check_result |= + SUM_CHECK_Q_RESULT; + } else + BUG(); + } + } + } + + rv = ioread32(&dma_reg->dsts); + if (rv) { + pr_err("DMA%d err status: 0x%x\n", + chan->device->id, rv); + /* write back to clear */ + iowrite32(rv, &dma_reg->dsts); + } + break; + case PPC440SPE_XOR_ID: + /* reset status bits to ack */ + xor_reg = chan->device->xor_reg; + rv = ioread32be(&xor_reg->sr); + iowrite32be(rv, &xor_reg->sr); + + if (rv & (XOR_IE_ICBIE_BIT|XOR_IE_ICIE_BIT|XOR_IE_RPTIE_BIT)) { + if (rv & XOR_IE_RPTIE_BIT) { + /* Read PLB Timeout Error. + * Try to resubmit the CB + */ + u32 val = ioread32be(&xor_reg->ccbalr); + + iowrite32be(val, &xor_reg->cblalr); + + val = ioread32be(&xor_reg->crsr); + iowrite32be(val | XOR_CRSR_XAE_BIT, + &xor_reg->crsr); + } else + pr_err("XOR ERR 0x%x status\n", rv); + break; + } + + /* if the XORcore is idle, but there are unprocessed CBs + * then refetch the s/w chain here + */ + if (!(ioread32be(&xor_reg->sr) & XOR_SR_XCP_BIT) && + do_xor_refetch) + ppc440spe_chan_append(chan); + break; + } +} + +/** + * ppc440spe_chan_is_busy - get the channel status + */ +static int ppc440spe_chan_is_busy(struct ppc440spe_adma_chan *chan) +{ + struct dma_regs *dma_reg; + struct xor_regs *xor_reg; + int busy = 0; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + dma_reg = chan->device->dma_reg; + /* if command FIFO's head and tail pointers are equal and + * status tail is the same as command, then channel is free + */ + if (ioread16(&dma_reg->cpfhp) != ioread16(&dma_reg->cpftp) || + ioread16(&dma_reg->cpftp) != ioread16(&dma_reg->csftp)) + busy = 1; + break; + case PPC440SPE_XOR_ID: + /* use the special status bit for the XORcore + */ + xor_reg = chan->device->xor_reg; + busy = (ioread32be(&xor_reg->sr) & XOR_SR_XCP_BIT) ? 1 : 0; + break; + } + + return busy; +} + +/** + * ppc440spe_chan_set_first_xor_descriptor - init XORcore chain + */ +static void ppc440spe_chan_set_first_xor_descriptor( + struct ppc440spe_adma_chan *chan, + struct ppc440spe_adma_desc_slot *next_desc) +{ + struct xor_regs *xor_reg = chan->device->xor_reg; + + if (ioread32be(&xor_reg->sr) & XOR_SR_XCP_BIT) + printk(KERN_INFO "%s: Warn: XORcore is running " + "when try to set the first CDB!\n", + __func__); + + xor_last_submit = xor_last_linked = next_desc; + + iowrite32be(XOR_CRSR_64BA_BIT, &xor_reg->crsr); + + iowrite32be(next_desc->phys, &xor_reg->cblalr); + iowrite32be(0, &xor_reg->cblahr); + iowrite32be(ioread32be(&xor_reg->cbcr) | XOR_CBCR_LNK_BIT, + &xor_reg->cbcr); + + chan->hw_chain_inited = 1; +} + +/** + * ppc440spe_dma_put_desc - put DMA0,1 descriptor to FIFO. + * called with irqs disabled + */ +static void ppc440spe_dma_put_desc(struct ppc440spe_adma_chan *chan, + struct ppc440spe_adma_desc_slot *desc) +{ + u32 pcdb; + struct dma_regs *dma_reg = chan->device->dma_reg; + + pcdb = desc->phys; + if (!test_bit(PPC440SPE_DESC_INT, &desc->flags)) + pcdb |= DMA_CDB_NO_INT; + + chan_last_sub[chan->device->id] = desc; + + ADMA_LL_DBG(print_cb(chan, desc->hw_desc)); + + iowrite32(pcdb, &dma_reg->cpfpl); +} + +/** + * ppc440spe_chan_append - update the h/w chain in the channel + */ +static void ppc440spe_chan_append(struct ppc440spe_adma_chan *chan) +{ + struct xor_regs *xor_reg; + struct ppc440spe_adma_desc_slot *iter; + struct xor_cb *xcb; + u32 cur_desc; + unsigned long flags; + + local_irq_save(flags); + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + cur_desc = ppc440spe_chan_get_current_descriptor(chan); + + if (likely(cur_desc)) { + iter = chan_last_sub[chan->device->id]; + BUG_ON(!iter); + } else { + /* first peer */ + iter = chan_first_cdb[chan->device->id]; + BUG_ON(!iter); + ppc440spe_dma_put_desc(chan, iter); + chan->hw_chain_inited = 1; + } + + /* is there something new to append */ + if (!iter->hw_next) + break; + + /* flush descriptors from the s/w queue to fifo */ + list_for_each_entry_continue(iter, &chan->chain, chain_node) { + ppc440spe_dma_put_desc(chan, iter); + if (!iter->hw_next) + break; + } + break; + case PPC440SPE_XOR_ID: + /* update h/w links and refetch */ + if (!xor_last_submit->hw_next) + break; + + xor_reg = chan->device->xor_reg; + /* the last linked CDB has to generate an interrupt + * that we'd be able to append the next lists to h/w + * regardless of the XOR engine state at the moment of + * appending of these next lists + */ + xcb = xor_last_linked->hw_desc; + xcb->cbc |= XOR_CBCR_CBCE_BIT; + + if (!(ioread32be(&xor_reg->sr) & XOR_SR_XCP_BIT)) { + /* XORcore is idle. Refetch now */ + do_xor_refetch = 0; + ppc440spe_xor_set_link(xor_last_submit, + xor_last_submit->hw_next); + + ADMA_LL_DBG(print_cb_list(chan, + xor_last_submit->hw_next)); + + xor_last_submit = xor_last_linked; + iowrite32be(ioread32be(&xor_reg->crsr) | + XOR_CRSR_RCBE_BIT | XOR_CRSR_64BA_BIT, + &xor_reg->crsr); + } else { + /* XORcore is running. Refetch later in the handler */ + do_xor_refetch = 1; + } + + break; + } + + local_irq_restore(flags); +} + +/** + * ppc440spe_chan_get_current_descriptor - get the currently executed descriptor + */ +static u32 +ppc440spe_chan_get_current_descriptor(struct ppc440spe_adma_chan *chan) +{ + struct dma_regs *dma_reg; + struct xor_regs *xor_reg; + + if (unlikely(!chan->hw_chain_inited)) + /* h/w descriptor chain is not initialized yet */ + return 0; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + dma_reg = chan->device->dma_reg; + return ioread32(&dma_reg->acpl) & (~DMA_CDB_MSK); + case PPC440SPE_XOR_ID: + xor_reg = chan->device->xor_reg; + return ioread32be(&xor_reg->ccbalr); + } + return 0; +} + +/** + * ppc440spe_chan_run - enable the channel + */ +static void ppc440spe_chan_run(struct ppc440spe_adma_chan *chan) +{ + struct xor_regs *xor_reg; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + /* DMAs are always enabled, do nothing */ + break; + case PPC440SPE_XOR_ID: + /* drain write buffer */ + xor_reg = chan->device->xor_reg; + + /* fetch descriptor pointed to in <link> */ + iowrite32be(XOR_CRSR_64BA_BIT | XOR_CRSR_XAE_BIT, + &xor_reg->crsr); + break; + } +} + +/****************************************************************************** + * ADMA device level + ******************************************************************************/ + +static void ppc440spe_chan_start_null_xor(struct ppc440spe_adma_chan *chan); +static int ppc440spe_adma_alloc_chan_resources(struct dma_chan *chan); + +static dma_cookie_t +ppc440spe_adma_tx_submit(struct dma_async_tx_descriptor *tx); + +static void ppc440spe_adma_set_dest(struct ppc440spe_adma_desc_slot *tx, + dma_addr_t addr, int index); +static void +ppc440spe_adma_memcpy_xor_set_src(struct ppc440spe_adma_desc_slot *tx, + dma_addr_t addr, int index); + +static void +ppc440spe_adma_pq_set_dest(struct ppc440spe_adma_desc_slot *tx, + dma_addr_t *paddr, unsigned long flags); +static void +ppc440spe_adma_pq_set_src(struct ppc440spe_adma_desc_slot *tx, + dma_addr_t addr, int index); +static void +ppc440spe_adma_pq_set_src_mult(struct ppc440spe_adma_desc_slot *tx, + unsigned char mult, int index, int dst_pos); +static void +ppc440spe_adma_pqzero_sum_set_dest(struct ppc440spe_adma_desc_slot *tx, + dma_addr_t paddr, dma_addr_t qaddr); + +static struct page *ppc440spe_rxor_srcs[32]; + +/** + * ppc440spe_can_rxor - check if the operands may be processed with RXOR + */ +static int ppc440spe_can_rxor(struct page **srcs, int src_cnt, size_t len) +{ + int i, order = 0, state = 0; + int idx = 0; + + if (unlikely(!(src_cnt > 1))) + return 0; + + BUG_ON(src_cnt > ARRAY_SIZE(ppc440spe_rxor_srcs)); + + /* Skip holes in the source list before checking */ + for (i = 0; i < src_cnt; i++) { + if (!srcs[i]) + continue; + ppc440spe_rxor_srcs[idx++] = srcs[i]; + } + src_cnt = idx; + + for (i = 1; i < src_cnt; i++) { + char *cur_addr = page_address(ppc440spe_rxor_srcs[i]); + char *old_addr = page_address(ppc440spe_rxor_srcs[i - 1]); + + switch (state) { + case 0: + if (cur_addr == old_addr + len) { + /* direct RXOR */ + order = 1; + state = 1; + } else if (old_addr == cur_addr + len) { + /* reverse RXOR */ + order = -1; + state = 1; + } else + goto out; + break; + case 1: + if ((i == src_cnt - 2) || + (order == -1 && cur_addr != old_addr - len)) { + order = 0; + state = 0; + } else if ((cur_addr == old_addr + len * order) || + (cur_addr == old_addr + 2 * len) || + (cur_addr == old_addr + 3 * len)) { + state = 2; + } else { + order = 0; + state = 0; + } + break; + case 2: + order = 0; + state = 0; + break; + } + } + +out: + if (state == 1 || state == 2) + return 1; + + return 0; +} + +/** + * ppc440spe_adma_device_estimate - estimate the efficiency of processing + * the operation given on this channel. It's assumed that 'chan' is + * capable to process 'cap' type of operation. + * @chan: channel to use + * @cap: type of transaction + * @dst_lst: array of destination pointers + * @dst_cnt: number of destination operands + * @src_lst: array of source pointers + * @src_cnt: number of source operands + * @src_sz: size of each source operand + */ +static int ppc440spe_adma_estimate(struct dma_chan *chan, + enum dma_transaction_type cap, struct page **dst_lst, int dst_cnt, + struct page **src_lst, int src_cnt, size_t src_sz) +{ + int ef = 1; + + if (cap == DMA_PQ || cap == DMA_PQ_VAL) { + /* If RAID-6 capabilities were not activated don't try + * to use them + */ + if (unlikely(!ppc440spe_r6_enabled)) + return -1; + } + /* In the current implementation of ppc440spe ADMA driver it + * makes sense to pick out only pq case, because it may be + * processed: + * (1) either using Biskup method on DMA2; + * (2) or on DMA0/1. + * Thus we give a favour to (1) if the sources are suitable; + * else let it be processed on one of the DMA0/1 engines. + * In the sum_product case where destination is also the + * source process it on DMA0/1 only. + */ + if (cap == DMA_PQ && chan->chan_id == PPC440SPE_XOR_ID) { + + if (dst_cnt == 1 && src_cnt == 2 && dst_lst[0] == src_lst[1]) + ef = 0; /* sum_product case, process on DMA0/1 */ + else if (ppc440spe_can_rxor(src_lst, src_cnt, src_sz)) + ef = 3; /* override (DMA0/1 + idle) */ + else + ef = 0; /* can't process on DMA2 if !rxor */ + } + + /* channel idleness increases the priority */ + if (likely(ef) && + !ppc440spe_chan_is_busy(to_ppc440spe_adma_chan(chan))) + ef++; + + return ef; +} + +struct dma_chan * +ppc440spe_async_tx_find_best_channel(enum dma_transaction_type cap, + struct page **dst_lst, int dst_cnt, struct page **src_lst, + int src_cnt, size_t src_sz) +{ + struct dma_chan *best_chan = NULL; + struct ppc_dma_chan_ref *ref; + int best_rank = -1; + + if (unlikely(!src_sz)) + return NULL; + if (src_sz > PAGE_SIZE) { + /* + * should a user of the api ever pass > PAGE_SIZE requests + * we sort out cases where temporary page-sized buffers + * are used. + */ + switch (cap) { + case DMA_PQ: + if (src_cnt == 1 && dst_lst[1] == src_lst[0]) + return NULL; + if (src_cnt == 2 && dst_lst[1] == src_lst[1]) + return NULL; + break; + case DMA_PQ_VAL: + case DMA_XOR_VAL: + return NULL; + default: + break; + } + } + + list_for_each_entry(ref, &ppc440spe_adma_chan_list, node) { + if (dma_has_cap(cap, ref->chan->device->cap_mask)) { + int rank; + + rank = ppc440spe_adma_estimate(ref->chan, cap, dst_lst, + dst_cnt, src_lst, src_cnt, src_sz); + if (rank > best_rank) { + best_rank = rank; + best_chan = ref->chan; + } + } + } + + return best_chan; +} +EXPORT_SYMBOL_GPL(ppc440spe_async_tx_find_best_channel); + +/** + * ppc440spe_get_group_entry - get group entry with index idx + * @tdesc: is the last allocated slot in the group. + */ +static struct ppc440spe_adma_desc_slot * +ppc440spe_get_group_entry(struct ppc440spe_adma_desc_slot *tdesc, u32 entry_idx) +{ + struct ppc440spe_adma_desc_slot *iter = tdesc->group_head; + int i = 0; + + if (entry_idx < 0 || entry_idx >= (tdesc->src_cnt + tdesc->dst_cnt)) { + printk("%s: entry_idx %d, src_cnt %d, dst_cnt %d\n", + __func__, entry_idx, tdesc->src_cnt, tdesc->dst_cnt); + BUG(); + } + + list_for_each_entry(iter, &tdesc->group_list, chain_node) { + if (i++ == entry_idx) + break; + } + return iter; +} + +/** + * ppc440spe_adma_free_slots - flags descriptor slots for reuse + * @slot: Slot to free + * Caller must hold &ppc440spe_chan->lock while calling this function + */ +static void ppc440spe_adma_free_slots(struct ppc440spe_adma_desc_slot *slot, + struct ppc440spe_adma_chan *chan) +{ + int stride = slot->slots_per_op; + + while (stride--) { + slot->slots_per_op = 0; + slot = list_entry(slot->slot_node.next, + struct ppc440spe_adma_desc_slot, + slot_node); + } +} + +static void ppc440spe_adma_unmap(struct ppc440spe_adma_chan *chan, + struct ppc440spe_adma_desc_slot *desc) +{ + u32 src_cnt, dst_cnt; + dma_addr_t addr; + + /* + * get the number of sources & destination + * included in this descriptor and unmap + * them all + */ + src_cnt = ppc440spe_desc_get_src_num(desc, chan); + dst_cnt = ppc440spe_desc_get_dst_num(desc, chan); + + /* unmap destinations */ + if (!(desc->async_tx.flags & DMA_COMPL_SKIP_DEST_UNMAP)) { + while (dst_cnt--) { + addr = ppc440spe_desc_get_dest_addr( + desc, chan, dst_cnt); + dma_unmap_page(chan->device->dev, + addr, desc->unmap_len, + DMA_FROM_DEVICE); + } + } + + /* unmap sources */ + if (!(desc->async_tx.flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + while (src_cnt--) { + addr = ppc440spe_desc_get_src_addr( + desc, chan, src_cnt); + dma_unmap_page(chan->device->dev, + addr, desc->unmap_len, + DMA_TO_DEVICE); + } + } +} + +/** + * ppc440spe_adma_run_tx_complete_actions - call functions to be called + * upon completion + */ +static dma_cookie_t ppc440spe_adma_run_tx_complete_actions( + struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan, + dma_cookie_t cookie) +{ + int i; + + BUG_ON(desc->async_tx.cookie < 0); + if (desc->async_tx.cookie > 0) { + cookie = desc->async_tx.cookie; + desc->async_tx.cookie = 0; + + /* call the callback (must not sleep or submit new + * operations to this channel) + */ + if (desc->async_tx.callback) + desc->async_tx.callback( + desc->async_tx.callback_param); + + /* unmap dma addresses + * (unmap_single vs unmap_page?) + * + * actually, ppc's dma_unmap_page() functions are empty, so + * the following code is just for the sake of completeness + */ + if (chan && chan->needs_unmap && desc->group_head && + desc->unmap_len) { + struct ppc440spe_adma_desc_slot *unmap = + desc->group_head; + /* assume 1 slot per op always */ + u32 slot_count = unmap->slot_cnt; + + /* Run through the group list and unmap addresses */ + for (i = 0; i < slot_count; i++) { + BUG_ON(!unmap); + ppc440spe_adma_unmap(chan, unmap); + unmap = unmap->hw_next; + } + } + } + + /* run dependent operations */ + dma_run_dependencies(&desc->async_tx); + + return cookie; +} + +/** + * ppc440spe_adma_clean_slot - clean up CDB slot (if ack is set) + */ +static int ppc440spe_adma_clean_slot(struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_adma_chan *chan) +{ + /* the client is allowed to attach dependent operations + * until 'ack' is set + */ + if (!async_tx_test_ack(&desc->async_tx)) + return 0; + + /* leave the last descriptor in the chain + * so we can append to it + */ + if (list_is_last(&desc->chain_node, &chan->chain) || + desc->phys == ppc440spe_chan_get_current_descriptor(chan)) + return 1; + + if (chan->device->id != PPC440SPE_XOR_ID) { + /* our DMA interrupt handler clears opc field of + * each processed descriptor. For all types of + * operations except for ZeroSum we do not actually + * need ack from the interrupt handler. ZeroSum is a + * special case since the result of this operation + * is available from the handler only, so if we see + * such type of descriptor (which is unprocessed yet) + * then leave it in chain. + */ + struct dma_cdb *cdb = desc->hw_desc; + if (cdb->opc == DMA_CDB_OPC_DCHECK128) + return 1; + } + + dev_dbg(chan->device->common.dev, "\tfree slot %llx: %d stride: %d\n", + desc->phys, desc->idx, desc->slots_per_op); + + list_del(&desc->chain_node); + ppc440spe_adma_free_slots(desc, chan); + return 0; +} + +/** + * __ppc440spe_adma_slot_cleanup - this is the common clean-up routine + * which runs through the channel CDBs list until reach the descriptor + * currently processed. When routine determines that all CDBs of group + * are completed then corresponding callbacks (if any) are called and slots + * are freed. + */ +static void __ppc440spe_adma_slot_cleanup(struct ppc440spe_adma_chan *chan) +{ + struct ppc440spe_adma_desc_slot *iter, *_iter, *group_start = NULL; + dma_cookie_t cookie = 0; + u32 current_desc = ppc440spe_chan_get_current_descriptor(chan); + int busy = ppc440spe_chan_is_busy(chan); + int seen_current = 0, slot_cnt = 0, slots_per_op = 0; + + dev_dbg(chan->device->common.dev, "ppc440spe adma%d: %s\n", + chan->device->id, __func__); + + if (!current_desc) { + /* There were no transactions yet, so + * nothing to clean + */ + return; + } + + /* free completed slots from the chain starting with + * the oldest descriptor + */ + list_for_each_entry_safe(iter, _iter, &chan->chain, + chain_node) { + dev_dbg(chan->device->common.dev, "\tcookie: %d slot: %d " + "busy: %d this_desc: %#llx next_desc: %#x " + "cur: %#x ack: %d\n", + iter->async_tx.cookie, iter->idx, busy, iter->phys, + ppc440spe_desc_get_link(iter, chan), current_desc, + async_tx_test_ack(&iter->async_tx)); + prefetch(_iter); + prefetch(&_iter->async_tx); + + /* do not advance past the current descriptor loaded into the + * hardware channel,subsequent descriptors are either in process + * or have not been submitted + */ + if (seen_current) + break; + + /* stop the search if we reach the current descriptor and the + * channel is busy, or if it appears that the current descriptor + * needs to be re-read (i.e. has been appended to) + */ + if (iter->phys == current_desc) { + BUG_ON(seen_current++); + if (busy || ppc440spe_desc_get_link(iter, chan)) { + /* not all descriptors of the group have + * been completed; exit. + */ + break; + } + } + + /* detect the start of a group transaction */ + if (!slot_cnt && !slots_per_op) { + slot_cnt = iter->slot_cnt; + slots_per_op = iter->slots_per_op; + if (slot_cnt <= slots_per_op) { + slot_cnt = 0; + slots_per_op = 0; + } + } + + if (slot_cnt) { + if (!group_start) + group_start = iter; + slot_cnt -= slots_per_op; + } + + /* all the members of a group are complete */ + if (slots_per_op != 0 && slot_cnt == 0) { + struct ppc440spe_adma_desc_slot *grp_iter, *_grp_iter; + int end_of_chain = 0; + + /* clean up the group */ + slot_cnt = group_start->slot_cnt; + grp_iter = group_start; + list_for_each_entry_safe_from(grp_iter, _grp_iter, + &chan->chain, chain_node) { + + cookie = ppc440spe_adma_run_tx_complete_actions( + grp_iter, chan, cookie); + + slot_cnt -= slots_per_op; + end_of_chain = ppc440spe_adma_clean_slot( + grp_iter, chan); + if (end_of_chain && slot_cnt) { + /* Should wait for ZeroSum completion */ + if (cookie > 0) + chan->completed_cookie = cookie; + return; + } + + if (slot_cnt == 0 || end_of_chain) + break; + } + + /* the group should be complete at this point */ + BUG_ON(slot_cnt); + + slots_per_op = 0; + group_start = NULL; + if (end_of_chain) + break; + else + continue; + } else if (slots_per_op) /* wait for group completion */ + continue; + + cookie = ppc440spe_adma_run_tx_complete_actions(iter, chan, + cookie); + + if (ppc440spe_adma_clean_slot(iter, chan)) + break; + } + + BUG_ON(!seen_current); + + if (cookie > 0) { + chan->completed_cookie = cookie; + pr_debug("\tcompleted cookie %d\n", cookie); + } + +} + +/** + * ppc440spe_adma_tasklet - clean up watch-dog initiator + */ +static void ppc440spe_adma_tasklet(unsigned long data) +{ + struct ppc440spe_adma_chan *chan = (struct ppc440spe_adma_chan *) data; + + spin_lock_nested(&chan->lock, SINGLE_DEPTH_NESTING); + __ppc440spe_adma_slot_cleanup(chan); + spin_unlock(&chan->lock); +} + +/** + * ppc440spe_adma_slot_cleanup - clean up scheduled initiator + */ +static void ppc440spe_adma_slot_cleanup(struct ppc440spe_adma_chan *chan) +{ + spin_lock_bh(&chan->lock); + __ppc440spe_adma_slot_cleanup(chan); + spin_unlock_bh(&chan->lock); +} + +/** + * ppc440spe_adma_alloc_slots - allocate free slots (if any) + */ +static struct ppc440spe_adma_desc_slot *ppc440spe_adma_alloc_slots( + struct ppc440spe_adma_chan *chan, int num_slots, + int slots_per_op) +{ + struct ppc440spe_adma_desc_slot *iter = NULL, *_iter; + struct ppc440spe_adma_desc_slot *alloc_start = NULL; + struct list_head chain = LIST_HEAD_INIT(chain); + int slots_found, retry = 0; + + + BUG_ON(!num_slots || !slots_per_op); + /* start search from the last allocated descrtiptor + * if a contiguous allocation can not be found start searching + * from the beginning of the list + */ +retry: + slots_found = 0; + if (retry == 0) + iter = chan->last_used; + else + iter = list_entry(&chan->all_slots, + struct ppc440spe_adma_desc_slot, + slot_node); + list_for_each_entry_safe_continue(iter, _iter, &chan->all_slots, + slot_node) { + prefetch(_iter); + prefetch(&_iter->async_tx); + if (iter->slots_per_op) { + slots_found = 0; + continue; + } + + /* start the allocation if the slot is correctly aligned */ + if (!slots_found++) + alloc_start = iter; + + if (slots_found == num_slots) { + struct ppc440spe_adma_desc_slot *alloc_tail = NULL; + struct ppc440spe_adma_desc_slot *last_used = NULL; + + iter = alloc_start; + while (num_slots) { + int i; + /* pre-ack all but the last descriptor */ + if (num_slots != slots_per_op) + async_tx_ack(&iter->async_tx); + + list_add_tail(&iter->chain_node, &chain); + alloc_tail = iter; + iter->async_tx.cookie = 0; + iter->hw_next = NULL; + iter->flags = 0; + iter->slot_cnt = num_slots; + iter->xor_check_result = NULL; + for (i = 0; i < slots_per_op; i++) { + iter->slots_per_op = slots_per_op - i; + last_used = iter; + iter = list_entry(iter->slot_node.next, + struct ppc440spe_adma_desc_slot, + slot_node); + } + num_slots -= slots_per_op; + } + alloc_tail->group_head = alloc_start; + alloc_tail->async_tx.cookie = -EBUSY; + list_splice(&chain, &alloc_tail->group_list); + chan->last_used = last_used; + return alloc_tail; + } + } + if (!retry++) + goto retry; + + /* try to free some slots if the allocation fails */ + tasklet_schedule(&chan->irq_tasklet); + return NULL; +} + +/** + * ppc440spe_adma_alloc_chan_resources - allocate pools for CDB slots + */ +static int ppc440spe_adma_alloc_chan_resources(struct dma_chan *chan) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + struct ppc440spe_adma_desc_slot *slot = NULL; + char *hw_desc; + int i, db_sz; + int init; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + init = ppc440spe_chan->slots_allocated ? 0 : 1; + chan->chan_id = ppc440spe_chan->device->id; + + /* Allocate descriptor slots */ + i = ppc440spe_chan->slots_allocated; + if (ppc440spe_chan->device->id != PPC440SPE_XOR_ID) + db_sz = sizeof(struct dma_cdb); + else + db_sz = sizeof(struct xor_cb); + + for (; i < (ppc440spe_chan->device->pool_size / db_sz); i++) { + slot = kzalloc(sizeof(struct ppc440spe_adma_desc_slot), + GFP_KERNEL); + if (!slot) { + printk(KERN_INFO "SPE ADMA Channel only initialized" + " %d descriptor slots", i--); + break; + } + + hw_desc = (char *) ppc440spe_chan->device->dma_desc_pool_virt; + slot->hw_desc = (void *) &hw_desc[i * db_sz]; + dma_async_tx_descriptor_init(&slot->async_tx, chan); + slot->async_tx.tx_submit = ppc440spe_adma_tx_submit; + INIT_LIST_HEAD(&slot->chain_node); + INIT_LIST_HEAD(&slot->slot_node); + INIT_LIST_HEAD(&slot->group_list); + slot->phys = ppc440spe_chan->device->dma_desc_pool + i * db_sz; + slot->idx = i; + + spin_lock_bh(&ppc440spe_chan->lock); + ppc440spe_chan->slots_allocated++; + list_add_tail(&slot->slot_node, &ppc440spe_chan->all_slots); + spin_unlock_bh(&ppc440spe_chan->lock); + } + + if (i && !ppc440spe_chan->last_used) { + ppc440spe_chan->last_used = + list_entry(ppc440spe_chan->all_slots.next, + struct ppc440spe_adma_desc_slot, + slot_node); + } + + dev_dbg(ppc440spe_chan->device->common.dev, + "ppc440spe adma%d: allocated %d descriptor slots\n", + ppc440spe_chan->device->id, i); + + /* initialize the channel and the chain with a null operation */ + if (init) { + switch (ppc440spe_chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + ppc440spe_chan->hw_chain_inited = 0; + /* Use WXOR for self-testing */ + if (!ppc440spe_r6_tchan) + ppc440spe_r6_tchan = ppc440spe_chan; + break; + case PPC440SPE_XOR_ID: + ppc440spe_chan_start_null_xor(ppc440spe_chan); + break; + default: + BUG(); + } + ppc440spe_chan->needs_unmap = 1; + } + + return (i > 0) ? i : -ENOMEM; +} + +/** + * ppc440spe_desc_assign_cookie - assign a cookie + */ +static dma_cookie_t ppc440spe_desc_assign_cookie( + struct ppc440spe_adma_chan *chan, + struct ppc440spe_adma_desc_slot *desc) +{ + dma_cookie_t cookie = chan->common.cookie; + + cookie++; + if (cookie < 0) + cookie = 1; + chan->common.cookie = desc->async_tx.cookie = cookie; + return cookie; +} + +/** + * ppc440spe_rxor_set_region_data - + */ +static void ppc440spe_rxor_set_region(struct ppc440spe_adma_desc_slot *desc, + u8 xor_arg_no, u32 mask) +{ + struct xor_cb *xcb = desc->hw_desc; + + xcb->ops[xor_arg_no].h |= mask; +} + +/** + * ppc440spe_rxor_set_src - + */ +static void ppc440spe_rxor_set_src(struct ppc440spe_adma_desc_slot *desc, + u8 xor_arg_no, dma_addr_t addr) +{ + struct xor_cb *xcb = desc->hw_desc; + + xcb->ops[xor_arg_no].h |= DMA_CUED_XOR_BASE; + xcb->ops[xor_arg_no].l = addr; +} + +/** + * ppc440spe_rxor_set_mult - + */ +static void ppc440spe_rxor_set_mult(struct ppc440spe_adma_desc_slot *desc, + u8 xor_arg_no, u8 idx, u8 mult) +{ + struct xor_cb *xcb = desc->hw_desc; + + xcb->ops[xor_arg_no].h |= mult << (DMA_CUED_MULT1_OFF + idx * 8); +} + +/** + * ppc440spe_adma_check_threshold - append CDBs to h/w chain if threshold + * has been achieved + */ +static void ppc440spe_adma_check_threshold(struct ppc440spe_adma_chan *chan) +{ + dev_dbg(chan->device->common.dev, "ppc440spe adma%d: pending: %d\n", + chan->device->id, chan->pending); + + if (chan->pending >= PPC440SPE_ADMA_THRESHOLD) { + chan->pending = 0; + ppc440spe_chan_append(chan); + } +} + +/** + * ppc440spe_adma_tx_submit - submit new descriptor group to the channel + * (it's not necessary that descriptors will be submitted to the h/w + * chains too right now) + */ +static dma_cookie_t ppc440spe_adma_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct ppc440spe_adma_desc_slot *sw_desc; + struct ppc440spe_adma_chan *chan = to_ppc440spe_adma_chan(tx->chan); + struct ppc440spe_adma_desc_slot *group_start, *old_chain_tail; + int slot_cnt; + int slots_per_op; + dma_cookie_t cookie; + + sw_desc = tx_to_ppc440spe_adma_slot(tx); + + group_start = sw_desc->group_head; + slot_cnt = group_start->slot_cnt; + slots_per_op = group_start->slots_per_op; + + spin_lock_bh(&chan->lock); + + cookie = ppc440spe_desc_assign_cookie(chan, sw_desc); + + if (unlikely(list_empty(&chan->chain))) { + /* first peer */ + list_splice_init(&sw_desc->group_list, &chan->chain); + chan_first_cdb[chan->device->id] = group_start; + } else { + /* isn't first peer, bind CDBs to chain */ + old_chain_tail = list_entry(chan->chain.prev, + struct ppc440spe_adma_desc_slot, + chain_node); + list_splice_init(&sw_desc->group_list, + &old_chain_tail->chain_node); + /* fix up the hardware chain */ + ppc440spe_desc_set_link(chan, old_chain_tail, group_start); + } + + /* increment the pending count by the number of operations */ + chan->pending += slot_cnt / slots_per_op; + ppc440spe_adma_check_threshold(chan); + spin_unlock_bh(&chan->lock); + + dev_dbg(chan->device->common.dev, + "ppc440spe adma%d: %s cookie: %d slot: %d tx %p\n", + chan->device->id, __func__, + sw_desc->async_tx.cookie, sw_desc->idx, sw_desc); + + return cookie; +} + +/** + * ppc440spe_adma_prep_dma_interrupt - prepare CDB for a pseudo DMA operation + */ +static struct dma_async_tx_descriptor *ppc440spe_adma_prep_dma_interrupt( + struct dma_chan *chan, unsigned long flags) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + struct ppc440spe_adma_desc_slot *sw_desc, *group_start; + int slot_cnt, slots_per_op; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + + dev_dbg(ppc440spe_chan->device->common.dev, + "ppc440spe adma%d: %s\n", ppc440spe_chan->device->id, + __func__); + + spin_lock_bh(&ppc440spe_chan->lock); + slot_cnt = slots_per_op = 1; + sw_desc = ppc440spe_adma_alloc_slots(ppc440spe_chan, slot_cnt, + slots_per_op); + if (sw_desc) { + group_start = sw_desc->group_head; + ppc440spe_desc_init_interrupt(group_start, ppc440spe_chan); + group_start->unmap_len = 0; + sw_desc->async_tx.flags = flags; + } + spin_unlock_bh(&ppc440spe_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +/** + * ppc440spe_adma_prep_dma_memcpy - prepare CDB for a MEMCPY operation + */ +static struct dma_async_tx_descriptor *ppc440spe_adma_prep_dma_memcpy( + struct dma_chan *chan, dma_addr_t dma_dest, + dma_addr_t dma_src, size_t len, unsigned long flags) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + struct ppc440spe_adma_desc_slot *sw_desc, *group_start; + int slot_cnt, slots_per_op; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + + if (unlikely(!len)) + return NULL; + + BUG_ON(unlikely(len > PPC440SPE_ADMA_DMA_MAX_BYTE_COUNT)); + + spin_lock_bh(&ppc440spe_chan->lock); + + dev_dbg(ppc440spe_chan->device->common.dev, + "ppc440spe adma%d: %s len: %u int_en %d\n", + ppc440spe_chan->device->id, __func__, len, + flags & DMA_PREP_INTERRUPT ? 1 : 0); + slot_cnt = slots_per_op = 1; + sw_desc = ppc440spe_adma_alloc_slots(ppc440spe_chan, slot_cnt, + slots_per_op); + if (sw_desc) { + group_start = sw_desc->group_head; + ppc440spe_desc_init_memcpy(group_start, flags); + ppc440spe_adma_set_dest(group_start, dma_dest, 0); + ppc440spe_adma_memcpy_xor_set_src(group_start, dma_src, 0); + ppc440spe_desc_set_byte_count(group_start, ppc440spe_chan, len); + sw_desc->unmap_len = len; + sw_desc->async_tx.flags = flags; + } + spin_unlock_bh(&ppc440spe_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +/** + * ppc440spe_adma_prep_dma_memset - prepare CDB for a MEMSET operation + */ +static struct dma_async_tx_descriptor *ppc440spe_adma_prep_dma_memset( + struct dma_chan *chan, dma_addr_t dma_dest, int value, + size_t len, unsigned long flags) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + struct ppc440spe_adma_desc_slot *sw_desc, *group_start; + int slot_cnt, slots_per_op; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + + if (unlikely(!len)) + return NULL; + + BUG_ON(unlikely(len > PPC440SPE_ADMA_DMA_MAX_BYTE_COUNT)); + + spin_lock_bh(&ppc440spe_chan->lock); + + dev_dbg(ppc440spe_chan->device->common.dev, + "ppc440spe adma%d: %s cal: %u len: %u int_en %d\n", + ppc440spe_chan->device->id, __func__, value, len, + flags & DMA_PREP_INTERRUPT ? 1 : 0); + + slot_cnt = slots_per_op = 1; + sw_desc = ppc440spe_adma_alloc_slots(ppc440spe_chan, slot_cnt, + slots_per_op); + if (sw_desc) { + group_start = sw_desc->group_head; + ppc440spe_desc_init_memset(group_start, value, flags); + ppc440spe_adma_set_dest(group_start, dma_dest, 0); + ppc440spe_desc_set_byte_count(group_start, ppc440spe_chan, len); + sw_desc->unmap_len = len; + sw_desc->async_tx.flags = flags; + } + spin_unlock_bh(&ppc440spe_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +/** + * ppc440spe_adma_prep_dma_xor - prepare CDB for a XOR operation + */ +static struct dma_async_tx_descriptor *ppc440spe_adma_prep_dma_xor( + struct dma_chan *chan, dma_addr_t dma_dest, + dma_addr_t *dma_src, u32 src_cnt, size_t len, + unsigned long flags) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + struct ppc440spe_adma_desc_slot *sw_desc, *group_start; + int slot_cnt, slots_per_op; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + + ADMA_LL_DBG(prep_dma_xor_dbg(ppc440spe_chan->device->id, + dma_dest, dma_src, src_cnt)); + if (unlikely(!len)) + return NULL; + BUG_ON(unlikely(len > PPC440SPE_ADMA_XOR_MAX_BYTE_COUNT)); + + dev_dbg(ppc440spe_chan->device->common.dev, + "ppc440spe adma%d: %s src_cnt: %d len: %u int_en: %d\n", + ppc440spe_chan->device->id, __func__, src_cnt, len, + flags & DMA_PREP_INTERRUPT ? 1 : 0); + + spin_lock_bh(&ppc440spe_chan->lock); + slot_cnt = ppc440spe_chan_xor_slot_count(len, src_cnt, &slots_per_op); + sw_desc = ppc440spe_adma_alloc_slots(ppc440spe_chan, slot_cnt, + slots_per_op); + if (sw_desc) { + group_start = sw_desc->group_head; + ppc440spe_desc_init_xor(group_start, src_cnt, flags); + ppc440spe_adma_set_dest(group_start, dma_dest, 0); + while (src_cnt--) + ppc440spe_adma_memcpy_xor_set_src(group_start, + dma_src[src_cnt], src_cnt); + ppc440spe_desc_set_byte_count(group_start, ppc440spe_chan, len); + sw_desc->unmap_len = len; + sw_desc->async_tx.flags = flags; + } + spin_unlock_bh(&ppc440spe_chan->lock); + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +static inline void +ppc440spe_desc_set_xor_src_cnt(struct ppc440spe_adma_desc_slot *desc, + int src_cnt); +static void ppc440spe_init_rxor_cursor(struct ppc440spe_rxor *cursor); + +/** + * ppc440spe_adma_init_dma2rxor_slot - + */ +static void ppc440spe_adma_init_dma2rxor_slot( + struct ppc440spe_adma_desc_slot *desc, + dma_addr_t *src, int src_cnt) +{ + int i; + + /* initialize CDB */ + for (i = 0; i < src_cnt; i++) { + ppc440spe_adma_dma2rxor_prep_src(desc, &desc->rxor_cursor, i, + desc->src_cnt, (u32)src[i]); + } +} + +/** + * ppc440spe_dma01_prep_mult - + * for Q operation where destination is also the source + */ +static struct ppc440spe_adma_desc_slot *ppc440spe_dma01_prep_mult( + struct ppc440spe_adma_chan *ppc440spe_chan, + dma_addr_t *dst, int dst_cnt, dma_addr_t *src, int src_cnt, + const unsigned char *scf, size_t len, unsigned long flags) +{ + struct ppc440spe_adma_desc_slot *sw_desc = NULL; + unsigned long op = 0; + int slot_cnt; + + set_bit(PPC440SPE_DESC_WXOR, &op); + slot_cnt = 2; + + spin_lock_bh(&ppc440spe_chan->lock); + + /* use WXOR, each descriptor occupies one slot */ + sw_desc = ppc440spe_adma_alloc_slots(ppc440spe_chan, slot_cnt, 1); + if (sw_desc) { + struct ppc440spe_adma_chan *chan; + struct ppc440spe_adma_desc_slot *iter; + struct dma_cdb *hw_desc; + + chan = to_ppc440spe_adma_chan(sw_desc->async_tx.chan); + set_bits(op, &sw_desc->flags); + sw_desc->src_cnt = src_cnt; + sw_desc->dst_cnt = dst_cnt; + /* First descriptor, zero data in the destination and copy it + * to q page using MULTICAST transfer. + */ + iter = list_first_entry(&sw_desc->group_list, + struct ppc440spe_adma_desc_slot, + chain_node); + memset(iter->hw_desc, 0, sizeof(struct dma_cdb)); + /* set 'next' pointer */ + iter->hw_next = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + clear_bit(PPC440SPE_DESC_INT, &iter->flags); + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MULTICAST; + + ppc440spe_desc_set_dest_addr(iter, chan, + DMA_CUED_XOR_BASE, dst[0], 0); + ppc440spe_desc_set_dest_addr(iter, chan, 0, dst[1], 1); + ppc440spe_desc_set_src_addr(iter, chan, 0, DMA_CUED_XOR_HB, + src[0]); + ppc440spe_desc_set_byte_count(iter, ppc440spe_chan, len); + iter->unmap_len = len; + + /* + * Second descriptor, multiply data from the q page + * and store the result in real destination. + */ + iter = list_first_entry(&iter->chain_node, + struct ppc440spe_adma_desc_slot, + chain_node); + memset(iter->hw_desc, 0, sizeof(struct dma_cdb)); + iter->hw_next = NULL; + if (flags & DMA_PREP_INTERRUPT) + set_bit(PPC440SPE_DESC_INT, &iter->flags); + else + clear_bit(PPC440SPE_DESC_INT, &iter->flags); + + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; + ppc440spe_desc_set_src_addr(iter, chan, 0, + DMA_CUED_XOR_HB, dst[1]); + ppc440spe_desc_set_dest_addr(iter, chan, + DMA_CUED_XOR_BASE, dst[0], 0); + + ppc440spe_desc_set_src_mult(iter, chan, DMA_CUED_MULT1_OFF, + DMA_CDB_SG_DST1, scf[0]); + ppc440spe_desc_set_byte_count(iter, ppc440spe_chan, len); + iter->unmap_len = len; + sw_desc->async_tx.flags = flags; + } + + spin_unlock_bh(&ppc440spe_chan->lock); + + return sw_desc; +} + +/** + * ppc440spe_dma01_prep_sum_product - + * Dx = A*(P+Pxy) + B*(Q+Qxy) operation where destination is also + * the source. + */ +static struct ppc440spe_adma_desc_slot *ppc440spe_dma01_prep_sum_product( + struct ppc440spe_adma_chan *ppc440spe_chan, + dma_addr_t *dst, dma_addr_t *src, int src_cnt, + const unsigned char *scf, size_t len, unsigned long flags) +{ + struct ppc440spe_adma_desc_slot *sw_desc = NULL; + unsigned long op = 0; + int slot_cnt; + + set_bit(PPC440SPE_DESC_WXOR, &op); + slot_cnt = 3; + + spin_lock_bh(&ppc440spe_chan->lock); + + /* WXOR, each descriptor occupies one slot */ + sw_desc = ppc440spe_adma_alloc_slots(ppc440spe_chan, slot_cnt, 1); + if (sw_desc) { + struct ppc440spe_adma_chan *chan; + struct ppc440spe_adma_desc_slot *iter; + struct dma_cdb *hw_desc; + + chan = to_ppc440spe_adma_chan(sw_desc->async_tx.chan); + set_bits(op, &sw_desc->flags); + sw_desc->src_cnt = src_cnt; + sw_desc->dst_cnt = 1; + /* 1st descriptor, src[1] data to q page and zero destination */ + iter = list_first_entry(&sw_desc->group_list, + struct ppc440spe_adma_desc_slot, + chain_node); + memset(iter->hw_desc, 0, sizeof(struct dma_cdb)); + iter->hw_next = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + clear_bit(PPC440SPE_DESC_INT, &iter->flags); + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MULTICAST; + + ppc440spe_desc_set_dest_addr(iter, chan, DMA_CUED_XOR_BASE, + *dst, 0); + ppc440spe_desc_set_dest_addr(iter, chan, 0, + ppc440spe_chan->qdest, 1); + ppc440spe_desc_set_src_addr(iter, chan, 0, DMA_CUED_XOR_HB, + src[1]); + ppc440spe_desc_set_byte_count(iter, ppc440spe_chan, len); + iter->unmap_len = len; + + /* 2nd descriptor, multiply src[1] data and store the + * result in destination */ + iter = list_first_entry(&iter->chain_node, + struct ppc440spe_adma_desc_slot, + chain_node); + memset(iter->hw_desc, 0, sizeof(struct dma_cdb)); + /* set 'next' pointer */ + iter->hw_next = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + if (flags & DMA_PREP_INTERRUPT) + set_bit(PPC440SPE_DESC_INT, &iter->flags); + else + clear_bit(PPC440SPE_DESC_INT, &iter->flags); + + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; + ppc440spe_desc_set_src_addr(iter, chan, 0, DMA_CUED_XOR_HB, + ppc440spe_chan->qdest); + ppc440spe_desc_set_dest_addr(iter, chan, DMA_CUED_XOR_BASE, + *dst, 0); + ppc440spe_desc_set_src_mult(iter, chan, DMA_CUED_MULT1_OFF, + DMA_CDB_SG_DST1, scf[1]); + ppc440spe_desc_set_byte_count(iter, ppc440spe_chan, len); + iter->unmap_len = len; + + /* + * 3rd descriptor, multiply src[0] data and xor it + * with destination + */ + iter = list_first_entry(&iter->chain_node, + struct ppc440spe_adma_desc_slot, + chain_node); + memset(iter->hw_desc, 0, sizeof(struct dma_cdb)); + iter->hw_next = NULL; + if (flags & DMA_PREP_INTERRUPT) + set_bit(PPC440SPE_DESC_INT, &iter->flags); + else + clear_bit(PPC440SPE_DESC_INT, &iter->flags); + + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; + ppc440spe_desc_set_src_addr(iter, chan, 0, DMA_CUED_XOR_HB, + src[0]); + ppc440spe_desc_set_dest_addr(iter, chan, DMA_CUED_XOR_BASE, + *dst, 0); + ppc440spe_desc_set_src_mult(iter, chan, DMA_CUED_MULT1_OFF, + DMA_CDB_SG_DST1, scf[0]); + ppc440spe_desc_set_byte_count(iter, ppc440spe_chan, len); + iter->unmap_len = len; + sw_desc->async_tx.flags = flags; + } + + spin_unlock_bh(&ppc440spe_chan->lock); + + return sw_desc; +} + +static struct ppc440spe_adma_desc_slot *ppc440spe_dma01_prep_pq( + struct ppc440spe_adma_chan *ppc440spe_chan, + dma_addr_t *dst, int dst_cnt, dma_addr_t *src, int src_cnt, + const unsigned char *scf, size_t len, unsigned long flags) +{ + int slot_cnt; + struct ppc440spe_adma_desc_slot *sw_desc = NULL, *iter; + unsigned long op = 0; + unsigned char mult = 1; + + pr_debug("%s: dst_cnt %d, src_cnt %d, len %d\n", + __func__, dst_cnt, src_cnt, len); + /* select operations WXOR/RXOR depending on the + * source addresses of operators and the number + * of destinations (RXOR support only Q-parity calculations) + */ + set_bit(PPC440SPE_DESC_WXOR, &op); + if (!test_and_set_bit(PPC440SPE_RXOR_RUN, &ppc440spe_rxor_state)) { + /* no active RXOR; + * do RXOR if: + * - there are more than 1 source, + * - len is aligned on 512-byte boundary, + * - source addresses fit to one of 4 possible regions. + */ + if (src_cnt > 1 && + !(len & MQ0_CF2H_RXOR_BS_MASK) && + (src[0] + len) == src[1]) { + /* may do RXOR R1 R2 */ + set_bit(PPC440SPE_DESC_RXOR, &op); + if (src_cnt != 2) { + /* may try to enhance region of RXOR */ + if ((src[1] + len) == src[2]) { + /* do RXOR R1 R2 R3 */ + set_bit(PPC440SPE_DESC_RXOR123, + &op); + } else if ((src[1] + len * 2) == src[2]) { + /* do RXOR R1 R2 R4 */ + set_bit(PPC440SPE_DESC_RXOR124, &op); + } else if ((src[1] + len * 3) == src[2]) { + /* do RXOR R1 R2 R5 */ + set_bit(PPC440SPE_DESC_RXOR125, + &op); + } else { + /* do RXOR R1 R2 */ + set_bit(PPC440SPE_DESC_RXOR12, + &op); + } + } else { + /* do RXOR R1 R2 */ + set_bit(PPC440SPE_DESC_RXOR12, &op); + } + } + + if (!test_bit(PPC440SPE_DESC_RXOR, &op)) { + /* can not do this operation with RXOR */ + clear_bit(PPC440SPE_RXOR_RUN, + &ppc440spe_rxor_state); + } else { + /* can do; set block size right now */ + ppc440spe_desc_set_rxor_block_size(len); + } + } + + /* Number of necessary slots depends on operation type selected */ + if (!test_bit(PPC440SPE_DESC_RXOR, &op)) { + /* This is a WXOR only chain. Need descriptors for each + * source to GF-XOR them with WXOR, and need descriptors + * for each destination to zero them with WXOR + */ + slot_cnt = src_cnt; + + if (flags & DMA_PREP_ZERO_P) { + slot_cnt++; + set_bit(PPC440SPE_ZERO_P, &op); + } + if (flags & DMA_PREP_ZERO_Q) { + slot_cnt++; + set_bit(PPC440SPE_ZERO_Q, &op); + } + } else { + /* Need 1/2 descriptor for RXOR operation, and + * need (src_cnt - (2 or 3)) for WXOR of sources + * remained (if any) + */ + slot_cnt = dst_cnt; + + if (flags & DMA_PREP_ZERO_P) + set_bit(PPC440SPE_ZERO_P, &op); + if (flags & DMA_PREP_ZERO_Q) + set_bit(PPC440SPE_ZERO_Q, &op); + + if (test_bit(PPC440SPE_DESC_RXOR12, &op)) + slot_cnt += src_cnt - 2; + else + slot_cnt += src_cnt - 3; + + /* Thus we have either RXOR only chain or + * mixed RXOR/WXOR + */ + if (slot_cnt == dst_cnt) + /* RXOR only chain */ + clear_bit(PPC440SPE_DESC_WXOR, &op); + } + + spin_lock_bh(&ppc440spe_chan->lock); + /* for both RXOR/WXOR each descriptor occupies one slot */ + sw_desc = ppc440spe_adma_alloc_slots(ppc440spe_chan, slot_cnt, 1); + if (sw_desc) { + ppc440spe_desc_init_dma01pq(sw_desc, dst_cnt, src_cnt, + flags, op); + + /* setup dst/src/mult */ + pr_debug("%s: set dst descriptor 0, 1: 0x%016llx, 0x%016llx\n", + __func__, dst[0], dst[1]); + ppc440spe_adma_pq_set_dest(sw_desc, dst, flags); + while (src_cnt--) { + ppc440spe_adma_pq_set_src(sw_desc, src[src_cnt], + src_cnt); + + /* NOTE: "Multi = 0 is equivalent to = 1" as it + * stated in 440SPSPe_RAID6_Addendum_UM_1_17.pdf + * doesn't work for RXOR with DMA0/1! Instead, multi=0 + * leads to zeroing source data after RXOR. + * So, for P case set-up mult=1 explicitly. + */ + if (!(flags & DMA_PREP_PQ_DISABLE_Q)) + mult = scf[src_cnt]; + ppc440spe_adma_pq_set_src_mult(sw_desc, + mult, src_cnt, dst_cnt - 1); + } + + /* Setup byte count foreach slot just allocated */ + sw_desc->async_tx.flags = flags; + list_for_each_entry(iter, &sw_desc->group_list, + chain_node) { + ppc440spe_desc_set_byte_count(iter, + ppc440spe_chan, len); + iter->unmap_len = len; + } + } + spin_unlock_bh(&ppc440spe_chan->lock); + + return sw_desc; +} + +static struct ppc440spe_adma_desc_slot *ppc440spe_dma2_prep_pq( + struct ppc440spe_adma_chan *ppc440spe_chan, + dma_addr_t *dst, int dst_cnt, dma_addr_t *src, int src_cnt, + const unsigned char *scf, size_t len, unsigned long flags) +{ + int slot_cnt, descs_per_op; + struct ppc440spe_adma_desc_slot *sw_desc = NULL, *iter; + unsigned long op = 0; + unsigned char mult = 1; + + BUG_ON(!dst_cnt); + /*pr_debug("%s: dst_cnt %d, src_cnt %d, len %d\n", + __func__, dst_cnt, src_cnt, len);*/ + + spin_lock_bh(&ppc440spe_chan->lock); + descs_per_op = ppc440spe_dma2_pq_slot_count(src, src_cnt, len); + if (descs_per_op < 0) { + spin_unlock_bh(&ppc440spe_chan->lock); + return NULL; + } + + /* depending on number of sources we have 1 or 2 RXOR chains */ + slot_cnt = descs_per_op * dst_cnt; + + sw_desc = ppc440spe_adma_alloc_slots(ppc440spe_chan, slot_cnt, 1); + if (sw_desc) { + op = slot_cnt; + sw_desc->async_tx.flags = flags; + list_for_each_entry(iter, &sw_desc->group_list, chain_node) { + ppc440spe_desc_init_dma2pq(iter, dst_cnt, src_cnt, + --op ? 0 : flags); + ppc440spe_desc_set_byte_count(iter, ppc440spe_chan, + len); + iter->unmap_len = len; + + ppc440spe_init_rxor_cursor(&(iter->rxor_cursor)); + iter->rxor_cursor.len = len; + iter->descs_per_op = descs_per_op; + } + op = 0; + list_for_each_entry(iter, &sw_desc->group_list, chain_node) { + op++; + if (op % descs_per_op == 0) + ppc440spe_adma_init_dma2rxor_slot(iter, src, + src_cnt); + if (likely(!list_is_last(&iter->chain_node, + &sw_desc->group_list))) { + /* set 'next' pointer */ + iter->hw_next = + list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + ppc440spe_xor_set_link(iter, iter->hw_next); + } else { + /* this is the last descriptor. */ + iter->hw_next = NULL; + } + } + + /* fixup head descriptor */ + sw_desc->dst_cnt = dst_cnt; + if (flags & DMA_PREP_ZERO_P) + set_bit(PPC440SPE_ZERO_P, &sw_desc->flags); + if (flags & DMA_PREP_ZERO_Q) + set_bit(PPC440SPE_ZERO_Q, &sw_desc->flags); + + /* setup dst/src/mult */ + ppc440spe_adma_pq_set_dest(sw_desc, dst, flags); + + while (src_cnt--) { + /* handle descriptors (if dst_cnt == 2) inside + * the ppc440spe_adma_pq_set_srcxxx() functions + */ + ppc440spe_adma_pq_set_src(sw_desc, src[src_cnt], + src_cnt); + if (!(flags & DMA_PREP_PQ_DISABLE_Q)) + mult = scf[src_cnt]; + ppc440spe_adma_pq_set_src_mult(sw_desc, + mult, src_cnt, dst_cnt - 1); + } + } + spin_unlock_bh(&ppc440spe_chan->lock); + ppc440spe_desc_set_rxor_block_size(len); + return sw_desc; +} + +/** + * ppc440spe_adma_prep_dma_pq - prepare CDB (group) for a GF-XOR operation + */ +static struct dma_async_tx_descriptor *ppc440spe_adma_prep_dma_pq( + struct dma_chan *chan, dma_addr_t *dst, dma_addr_t *src, + unsigned int src_cnt, const unsigned char *scf, + size_t len, unsigned long flags) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + struct ppc440spe_adma_desc_slot *sw_desc = NULL; + int dst_cnt = 0; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + + ADMA_LL_DBG(prep_dma_pq_dbg(ppc440spe_chan->device->id, + dst, src, src_cnt)); + BUG_ON(!len); + BUG_ON(unlikely(len > PPC440SPE_ADMA_XOR_MAX_BYTE_COUNT)); + BUG_ON(!src_cnt); + + if (src_cnt == 1 && dst[1] == src[0]) { + dma_addr_t dest[2]; + + /* dst[1] is real destination (Q) */ + dest[0] = dst[1]; + /* this is the page to multicast source data to */ + dest[1] = ppc440spe_chan->qdest; + sw_desc = ppc440spe_dma01_prep_mult(ppc440spe_chan, + dest, 2, src, src_cnt, scf, len, flags); + return sw_desc ? &sw_desc->async_tx : NULL; + } + + if (src_cnt == 2 && dst[1] == src[1]) { + sw_desc = ppc440spe_dma01_prep_sum_product(ppc440spe_chan, + &dst[1], src, 2, scf, len, flags); + return sw_desc ? &sw_desc->async_tx : NULL; + } + + if (!(flags & DMA_PREP_PQ_DISABLE_P)) { + BUG_ON(!dst[0]); + dst_cnt++; + flags |= DMA_PREP_ZERO_P; + } + + if (!(flags & DMA_PREP_PQ_DISABLE_Q)) { + BUG_ON(!dst[1]); + dst_cnt++; + flags |= DMA_PREP_ZERO_Q; + } + + BUG_ON(!dst_cnt); + + dev_dbg(ppc440spe_chan->device->common.dev, + "ppc440spe adma%d: %s src_cnt: %d len: %u int_en: %d\n", + ppc440spe_chan->device->id, __func__, src_cnt, len, + flags & DMA_PREP_INTERRUPT ? 1 : 0); + + switch (ppc440spe_chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + sw_desc = ppc440spe_dma01_prep_pq(ppc440spe_chan, + dst, dst_cnt, src, src_cnt, scf, + len, flags); + break; + + case PPC440SPE_XOR_ID: + sw_desc = ppc440spe_dma2_prep_pq(ppc440spe_chan, + dst, dst_cnt, src, src_cnt, scf, + len, flags); + break; + } + + return sw_desc ? &sw_desc->async_tx : NULL; +} + +/** + * ppc440spe_adma_prep_dma_pqzero_sum - prepare CDB group for + * a PQ_ZERO_SUM operation + */ +static struct dma_async_tx_descriptor *ppc440spe_adma_prep_dma_pqzero_sum( + struct dma_chan *chan, dma_addr_t *pq, dma_addr_t *src, + unsigned int src_cnt, const unsigned char *scf, size_t len, + enum sum_check_flags *pqres, unsigned long flags) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + struct ppc440spe_adma_desc_slot *sw_desc, *iter; + dma_addr_t pdest, qdest; + int slot_cnt, slots_per_op, idst, dst_cnt; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + + if (flags & DMA_PREP_PQ_DISABLE_P) + pdest = 0; + else + pdest = pq[0]; + + if (flags & DMA_PREP_PQ_DISABLE_Q) + qdest = 0; + else + qdest = pq[1]; + + ADMA_LL_DBG(prep_dma_pqzero_sum_dbg(ppc440spe_chan->device->id, + src, src_cnt, scf)); + + /* Always use WXOR for P/Q calculations (two destinations). + * Need 1 or 2 extra slots to verify results are zero. + */ + idst = dst_cnt = (pdest && qdest) ? 2 : 1; + + /* One additional slot per destination to clone P/Q + * before calculation (we have to preserve destinations). + */ + slot_cnt = src_cnt + dst_cnt * 2; + slots_per_op = 1; + + spin_lock_bh(&ppc440spe_chan->lock); + sw_desc = ppc440spe_adma_alloc_slots(ppc440spe_chan, slot_cnt, + slots_per_op); + if (sw_desc) { + ppc440spe_desc_init_dma01pqzero_sum(sw_desc, dst_cnt, src_cnt); + + /* Setup byte count for each slot just allocated */ + sw_desc->async_tx.flags = flags; + list_for_each_entry(iter, &sw_desc->group_list, chain_node) { + ppc440spe_desc_set_byte_count(iter, ppc440spe_chan, + len); + iter->unmap_len = len; + } + + if (pdest) { + struct dma_cdb *hw_desc; + struct ppc440spe_adma_chan *chan; + + iter = sw_desc->group_head; + chan = to_ppc440spe_adma_chan(iter->async_tx.chan); + memset(iter->hw_desc, 0, sizeof(struct dma_cdb)); + iter->hw_next = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; + iter->src_cnt = 0; + iter->dst_cnt = 0; + ppc440spe_desc_set_dest_addr(iter, chan, 0, + ppc440spe_chan->pdest, 0); + ppc440spe_desc_set_src_addr(iter, chan, 0, 0, pdest); + ppc440spe_desc_set_byte_count(iter, ppc440spe_chan, + len); + iter->unmap_len = 0; + /* override pdest to preserve original P */ + pdest = ppc440spe_chan->pdest; + } + if (qdest) { + struct dma_cdb *hw_desc; + struct ppc440spe_adma_chan *chan; + + iter = list_first_entry(&sw_desc->group_list, + struct ppc440spe_adma_desc_slot, + chain_node); + chan = to_ppc440spe_adma_chan(iter->async_tx.chan); + + if (pdest) { + iter = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + } + + memset(iter->hw_desc, 0, sizeof(struct dma_cdb)); + iter->hw_next = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + hw_desc = iter->hw_desc; + hw_desc->opc = DMA_CDB_OPC_MV_SG1_SG2; + iter->src_cnt = 0; + iter->dst_cnt = 0; + ppc440spe_desc_set_dest_addr(iter, chan, 0, + ppc440spe_chan->qdest, 0); + ppc440spe_desc_set_src_addr(iter, chan, 0, 0, qdest); + ppc440spe_desc_set_byte_count(iter, ppc440spe_chan, + len); + iter->unmap_len = 0; + /* override qdest to preserve original Q */ + qdest = ppc440spe_chan->qdest; + } + + /* Setup destinations for P/Q ops */ + ppc440spe_adma_pqzero_sum_set_dest(sw_desc, pdest, qdest); + + /* Setup zero QWORDs into DCHECK CDBs */ + idst = dst_cnt; + list_for_each_entry_reverse(iter, &sw_desc->group_list, + chain_node) { + /* + * The last CDB corresponds to Q-parity check, + * the one before last CDB corresponds + * P-parity check + */ + if (idst == DMA_DEST_MAX_NUM) { + if (idst == dst_cnt) { + set_bit(PPC440SPE_DESC_QCHECK, + &iter->flags); + } else { + set_bit(PPC440SPE_DESC_PCHECK, + &iter->flags); + } + } else { + if (qdest) { + set_bit(PPC440SPE_DESC_QCHECK, + &iter->flags); + } else { + set_bit(PPC440SPE_DESC_PCHECK, + &iter->flags); + } + } + iter->xor_check_result = pqres; + + /* + * set it to zero, if check fail then result will + * be updated + */ + *iter->xor_check_result = 0; + ppc440spe_desc_set_dcheck(iter, ppc440spe_chan, + ppc440spe_qword); + + if (!(--dst_cnt)) + break; + } + + /* Setup sources and mults for P/Q ops */ + list_for_each_entry_continue_reverse(iter, &sw_desc->group_list, + chain_node) { + struct ppc440spe_adma_chan *chan; + u32 mult_dst; + + chan = to_ppc440spe_adma_chan(iter->async_tx.chan); + ppc440spe_desc_set_src_addr(iter, chan, 0, + DMA_CUED_XOR_HB, + src[src_cnt - 1]); + if (qdest) { + mult_dst = (dst_cnt - 1) ? DMA_CDB_SG_DST2 : + DMA_CDB_SG_DST1; + ppc440spe_desc_set_src_mult(iter, chan, + DMA_CUED_MULT1_OFF, + mult_dst, + scf[src_cnt - 1]); + } + if (!(--src_cnt)) + break; + } + } + spin_unlock_bh(&ppc440spe_chan->lock); + return sw_desc ? &sw_desc->async_tx : NULL; +} + +/** + * ppc440spe_adma_prep_dma_xor_zero_sum - prepare CDB group for + * XOR ZERO_SUM operation + */ +static struct dma_async_tx_descriptor *ppc440spe_adma_prep_dma_xor_zero_sum( + struct dma_chan *chan, dma_addr_t *src, unsigned int src_cnt, + size_t len, enum sum_check_flags *result, unsigned long flags) +{ + struct dma_async_tx_descriptor *tx; + dma_addr_t pq[2]; + + /* validate P, disable Q */ + pq[0] = src[0]; + pq[1] = 0; + flags |= DMA_PREP_PQ_DISABLE_Q; + + tx = ppc440spe_adma_prep_dma_pqzero_sum(chan, pq, &src[1], + src_cnt - 1, 0, len, + result, flags); + return tx; +} + +/** + * ppc440spe_adma_set_dest - set destination address into descriptor + */ +static void ppc440spe_adma_set_dest(struct ppc440spe_adma_desc_slot *sw_desc, + dma_addr_t addr, int index) +{ + struct ppc440spe_adma_chan *chan; + + BUG_ON(index >= sw_desc->dst_cnt); + + chan = to_ppc440spe_adma_chan(sw_desc->async_tx.chan); + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + /* to do: support transfers lengths > + * PPC440SPE_ADMA_DMA/XOR_MAX_BYTE_COUNT + */ + ppc440spe_desc_set_dest_addr(sw_desc->group_head, + chan, 0, addr, index); + break; + case PPC440SPE_XOR_ID: + sw_desc = ppc440spe_get_group_entry(sw_desc, index); + ppc440spe_desc_set_dest_addr(sw_desc, + chan, 0, addr, index); + break; + } +} + +static void ppc440spe_adma_pq_zero_op(struct ppc440spe_adma_desc_slot *iter, + struct ppc440spe_adma_chan *chan, dma_addr_t addr) +{ + /* To clear destinations update the descriptor + * (P or Q depending on index) as follows: + * addr is destination (0 corresponds to SG2): + */ + ppc440spe_desc_set_dest_addr(iter, chan, DMA_CUED_XOR_BASE, addr, 0); + + /* ... and the addr is source: */ + ppc440spe_desc_set_src_addr(iter, chan, 0, DMA_CUED_XOR_HB, addr); + + /* addr is always SG2 then the mult is always DST1 */ + ppc440spe_desc_set_src_mult(iter, chan, DMA_CUED_MULT1_OFF, + DMA_CDB_SG_DST1, 1); +} + +/** + * ppc440spe_adma_pq_set_dest - set destination address into descriptor + * for the PQXOR operation + */ +static void ppc440spe_adma_pq_set_dest(struct ppc440spe_adma_desc_slot *sw_desc, + dma_addr_t *addrs, unsigned long flags) +{ + struct ppc440spe_adma_desc_slot *iter; + struct ppc440spe_adma_chan *chan; + dma_addr_t paddr, qaddr; + dma_addr_t addr = 0, ppath, qpath; + int index = 0, i; + + chan = to_ppc440spe_adma_chan(sw_desc->async_tx.chan); + + if (flags & DMA_PREP_PQ_DISABLE_P) + paddr = 0; + else + paddr = addrs[0]; + + if (flags & DMA_PREP_PQ_DISABLE_Q) + qaddr = 0; + else + qaddr = addrs[1]; + + if (!paddr || !qaddr) + addr = paddr ? paddr : qaddr; + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + /* walk through the WXOR source list and set P/Q-destinations + * for each slot: + */ + if (!test_bit(PPC440SPE_DESC_RXOR, &sw_desc->flags)) { + /* This is WXOR-only chain; may have 1/2 zero descs */ + if (test_bit(PPC440SPE_ZERO_P, &sw_desc->flags)) + index++; + if (test_bit(PPC440SPE_ZERO_Q, &sw_desc->flags)) + index++; + + iter = ppc440spe_get_group_entry(sw_desc, index); + if (addr) { + /* one destination */ + list_for_each_entry_from(iter, + &sw_desc->group_list, chain_node) + ppc440spe_desc_set_dest_addr(iter, chan, + DMA_CUED_XOR_BASE, addr, 0); + } else { + /* two destinations */ + list_for_each_entry_from(iter, + &sw_desc->group_list, chain_node) { + ppc440spe_desc_set_dest_addr(iter, chan, + DMA_CUED_XOR_BASE, paddr, 0); + ppc440spe_desc_set_dest_addr(iter, chan, + DMA_CUED_XOR_BASE, qaddr, 1); + } + } + + if (index) { + /* To clear destinations update the descriptor + * (1st,2nd, or both depending on flags) + */ + index = 0; + if (test_bit(PPC440SPE_ZERO_P, + &sw_desc->flags)) { + iter = ppc440spe_get_group_entry( + sw_desc, index++); + ppc440spe_adma_pq_zero_op(iter, chan, + paddr); + } + + if (test_bit(PPC440SPE_ZERO_Q, + &sw_desc->flags)) { + iter = ppc440spe_get_group_entry( + sw_desc, index++); + ppc440spe_adma_pq_zero_op(iter, chan, + qaddr); + } + + return; + } + } else { + /* This is RXOR-only or RXOR/WXOR mixed chain */ + + /* If we want to include destination into calculations, + * then make dest addresses cued with mult=1 (XOR). + */ + ppath = test_bit(PPC440SPE_ZERO_P, &sw_desc->flags) ? + DMA_CUED_XOR_HB : + DMA_CUED_XOR_BASE | + (1 << DMA_CUED_MULT1_OFF); + qpath = test_bit(PPC440SPE_ZERO_Q, &sw_desc->flags) ? + DMA_CUED_XOR_HB : + DMA_CUED_XOR_BASE | + (1 << DMA_CUED_MULT1_OFF); + + /* Setup destination(s) in RXOR slot(s) */ + iter = ppc440spe_get_group_entry(sw_desc, index++); + ppc440spe_desc_set_dest_addr(iter, chan, + paddr ? ppath : qpath, + paddr ? paddr : qaddr, 0); + if (!addr) { + /* two destinations */ + iter = ppc440spe_get_group_entry(sw_desc, + index++); + ppc440spe_desc_set_dest_addr(iter, chan, + qpath, qaddr, 0); + } + + if (test_bit(PPC440SPE_DESC_WXOR, &sw_desc->flags)) { + /* Setup destination(s) in remaining WXOR + * slots + */ + iter = ppc440spe_get_group_entry(sw_desc, + index); + if (addr) { + /* one destination */ + list_for_each_entry_from(iter, + &sw_desc->group_list, + chain_node) + ppc440spe_desc_set_dest_addr( + iter, chan, + DMA_CUED_XOR_BASE, + addr, 0); + + } else { + /* two destinations */ + list_for_each_entry_from(iter, + &sw_desc->group_list, + chain_node) { + ppc440spe_desc_set_dest_addr( + iter, chan, + DMA_CUED_XOR_BASE, + paddr, 0); + ppc440spe_desc_set_dest_addr( + iter, chan, + DMA_CUED_XOR_BASE, + qaddr, 1); + } + } + } + + } + break; + + case PPC440SPE_XOR_ID: + /* DMA2 descriptors have only 1 destination, so there are + * two chains - one for each dest. + * If we want to include destination into calculations, + * then make dest addresses cued with mult=1 (XOR). + */ + ppath = test_bit(PPC440SPE_ZERO_P, &sw_desc->flags) ? + DMA_CUED_XOR_HB : + DMA_CUED_XOR_BASE | + (1 << DMA_CUED_MULT1_OFF); + + qpath = test_bit(PPC440SPE_ZERO_Q, &sw_desc->flags) ? + DMA_CUED_XOR_HB : + DMA_CUED_XOR_BASE | + (1 << DMA_CUED_MULT1_OFF); + + iter = ppc440spe_get_group_entry(sw_desc, 0); + for (i = 0; i < sw_desc->descs_per_op; i++) { + ppc440spe_desc_set_dest_addr(iter, chan, + paddr ? ppath : qpath, + paddr ? paddr : qaddr, 0); + iter = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + } + + if (!addr) { + /* Two destinations; setup Q here */ + iter = ppc440spe_get_group_entry(sw_desc, + sw_desc->descs_per_op); + for (i = 0; i < sw_desc->descs_per_op; i++) { + ppc440spe_desc_set_dest_addr(iter, + chan, qpath, qaddr, 0); + iter = list_entry(iter->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + } + } + + break; + } +} + +/** + * ppc440spe_adma_pq_zero_sum_set_dest - set destination address into descriptor + * for the PQ_ZERO_SUM operation + */ +static void ppc440spe_adma_pqzero_sum_set_dest( + struct ppc440spe_adma_desc_slot *sw_desc, + dma_addr_t paddr, dma_addr_t qaddr) +{ + struct ppc440spe_adma_desc_slot *iter, *end; + struct ppc440spe_adma_chan *chan; + dma_addr_t addr = 0; + int idx; + + chan = to_ppc440spe_adma_chan(sw_desc->async_tx.chan); + + /* walk through the WXOR source list and set P/Q-destinations + * for each slot + */ + idx = (paddr && qaddr) ? 2 : 1; + /* set end */ + list_for_each_entry_reverse(end, &sw_desc->group_list, + chain_node) { + if (!(--idx)) + break; + } + /* set start */ + idx = (paddr && qaddr) ? 2 : 1; + iter = ppc440spe_get_group_entry(sw_desc, idx); + + if (paddr && qaddr) { + /* two destinations */ + list_for_each_entry_from(iter, &sw_desc->group_list, + chain_node) { + if (unlikely(iter == end)) + break; + ppc440spe_desc_set_dest_addr(iter, chan, + DMA_CUED_XOR_BASE, paddr, 0); + ppc440spe_desc_set_dest_addr(iter, chan, + DMA_CUED_XOR_BASE, qaddr, 1); + } + } else { + /* one destination */ + addr = paddr ? paddr : qaddr; + list_for_each_entry_from(iter, &sw_desc->group_list, + chain_node) { + if (unlikely(iter == end)) + break; + ppc440spe_desc_set_dest_addr(iter, chan, + DMA_CUED_XOR_BASE, addr, 0); + } + } + + /* The remaining descriptors are DATACHECK. These have no need in + * destination. Actually, these destinations are used there + * as sources for check operation. So, set addr as source. + */ + ppc440spe_desc_set_src_addr(end, chan, 0, 0, addr ? addr : paddr); + + if (!addr) { + end = list_entry(end->chain_node.next, + struct ppc440spe_adma_desc_slot, chain_node); + ppc440spe_desc_set_src_addr(end, chan, 0, 0, qaddr); + } +} + +/** + * ppc440spe_desc_set_xor_src_cnt - set source count into descriptor + */ +static inline void ppc440spe_desc_set_xor_src_cnt( + struct ppc440spe_adma_desc_slot *desc, + int src_cnt) +{ + struct xor_cb *hw_desc = desc->hw_desc; + + hw_desc->cbc &= ~XOR_CDCR_OAC_MSK; + hw_desc->cbc |= src_cnt; +} + +/** + * ppc440spe_adma_pq_set_src - set source address into descriptor + */ +static void ppc440spe_adma_pq_set_src(struct ppc440spe_adma_desc_slot *sw_desc, + dma_addr_t addr, int index) +{ + struct ppc440spe_adma_chan *chan; + dma_addr_t haddr = 0; + struct ppc440spe_adma_desc_slot *iter = NULL; + + chan = to_ppc440spe_adma_chan(sw_desc->async_tx.chan); + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + /* DMA0,1 may do: WXOR, RXOR, RXOR+WXORs chain + */ + if (test_bit(PPC440SPE_DESC_RXOR, &sw_desc->flags)) { + /* RXOR-only or RXOR/WXOR operation */ + int iskip = test_bit(PPC440SPE_DESC_RXOR12, + &sw_desc->flags) ? 2 : 3; + + if (index == 0) { + /* 1st slot (RXOR) */ + /* setup sources region (R1-2-3, R1-2-4, + * or R1-2-5) + */ + if (test_bit(PPC440SPE_DESC_RXOR12, + &sw_desc->flags)) + haddr = DMA_RXOR12 << + DMA_CUED_REGION_OFF; + else if (test_bit(PPC440SPE_DESC_RXOR123, + &sw_desc->flags)) + haddr = DMA_RXOR123 << + DMA_CUED_REGION_OFF; + else if (test_bit(PPC440SPE_DESC_RXOR124, + &sw_desc->flags)) + haddr = DMA_RXOR124 << + DMA_CUED_REGION_OFF; + else if (test_bit(PPC440SPE_DESC_RXOR125, + &sw_desc->flags)) + haddr = DMA_RXOR125 << + DMA_CUED_REGION_OFF; + else + BUG(); + haddr |= DMA_CUED_XOR_BASE; + iter = ppc440spe_get_group_entry(sw_desc, 0); + } else if (index < iskip) { + /* 1st slot (RXOR) + * shall actually set source address only once + * instead of first <iskip> + */ + iter = NULL; + } else { + /* 2nd/3d and next slots (WXOR); + * skip first slot with RXOR + */ + haddr = DMA_CUED_XOR_HB; + iter = ppc440spe_get_group_entry(sw_desc, + index - iskip + sw_desc->dst_cnt); + } + } else { + int znum = 0; + + /* WXOR-only operation; skip first slots with + * zeroing destinations + */ + if (test_bit(PPC440SPE_ZERO_P, &sw_desc->flags)) + znum++; + if (test_bit(PPC440SPE_ZERO_Q, &sw_desc->flags)) + znum++; + + haddr = DMA_CUED_XOR_HB; + iter = ppc440spe_get_group_entry(sw_desc, + index + znum); + } + + if (likely(iter)) { + ppc440spe_desc_set_src_addr(iter, chan, 0, haddr, addr); + + if (!index && + test_bit(PPC440SPE_DESC_RXOR, &sw_desc->flags) && + sw_desc->dst_cnt == 2) { + /* if we have two destinations for RXOR, then + * setup source in the second descr too + */ + iter = ppc440spe_get_group_entry(sw_desc, 1); + ppc440spe_desc_set_src_addr(iter, chan, 0, + haddr, addr); + } + } + break; + + case PPC440SPE_XOR_ID: + /* DMA2 may do Biskup */ + iter = sw_desc->group_head; + if (iter->dst_cnt == 2) { + /* both P & Q calculations required; set P src here */ + ppc440spe_adma_dma2rxor_set_src(iter, index, addr); + + /* this is for Q */ + iter = ppc440spe_get_group_entry(sw_desc, + sw_desc->descs_per_op); + } + ppc440spe_adma_dma2rxor_set_src(iter, index, addr); + break; + } +} + +/** + * ppc440spe_adma_memcpy_xor_set_src - set source address into descriptor + */ +static void ppc440spe_adma_memcpy_xor_set_src( + struct ppc440spe_adma_desc_slot *sw_desc, + dma_addr_t addr, int index) +{ + struct ppc440spe_adma_chan *chan; + + chan = to_ppc440spe_adma_chan(sw_desc->async_tx.chan); + sw_desc = sw_desc->group_head; + + if (likely(sw_desc)) + ppc440spe_desc_set_src_addr(sw_desc, chan, index, 0, addr); +} + +/** + * ppc440spe_adma_dma2rxor_inc_addr - + */ +static void ppc440spe_adma_dma2rxor_inc_addr( + struct ppc440spe_adma_desc_slot *desc, + struct ppc440spe_rxor *cursor, int index, int src_cnt) +{ + cursor->addr_count++; + if (index == src_cnt - 1) { + ppc440spe_desc_set_xor_src_cnt(desc, cursor->addr_count); + } else if (cursor->addr_count == XOR_MAX_OPS) { + ppc440spe_desc_set_xor_src_cnt(desc, cursor->addr_count); + cursor->addr_count = 0; + cursor->desc_count++; + } +} + +/** + * ppc440spe_adma_dma2rxor_prep_src - setup RXOR types in DMA2 CDB + */ +static int ppc440spe_adma_dma2rxor_prep_src( + struct ppc440spe_adma_desc_slot *hdesc, + struct ppc440spe_rxor *cursor, int index, + int src_cnt, u32 addr) +{ + int rval = 0; + u32 sign; + struct ppc440spe_adma_desc_slot *desc = hdesc; + int i; + + for (i = 0; i < cursor->desc_count; i++) { + desc = list_entry(hdesc->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + } + + switch (cursor->state) { + case 0: + if (addr == cursor->addrl + cursor->len) { + /* direct RXOR */ + cursor->state = 1; + cursor->xor_count++; + if (index == src_cnt-1) { + ppc440spe_rxor_set_region(desc, + cursor->addr_count, + DMA_RXOR12 << DMA_CUED_REGION_OFF); + ppc440spe_adma_dma2rxor_inc_addr( + desc, cursor, index, src_cnt); + } + } else if (cursor->addrl == addr + cursor->len) { + /* reverse RXOR */ + cursor->state = 1; + cursor->xor_count++; + set_bit(cursor->addr_count, &desc->reverse_flags[0]); + if (index == src_cnt-1) { + ppc440spe_rxor_set_region(desc, + cursor->addr_count, + DMA_RXOR12 << DMA_CUED_REGION_OFF); + ppc440spe_adma_dma2rxor_inc_addr( + desc, cursor, index, src_cnt); + } + } else { + printk(KERN_ERR "Cannot build " + "DMA2 RXOR command block.\n"); + BUG(); + } + break; + case 1: + sign = test_bit(cursor->addr_count, + desc->reverse_flags) + ? -1 : 1; + if (index == src_cnt-2 || (sign == -1 + && addr != cursor->addrl - 2*cursor->len)) { + cursor->state = 0; + cursor->xor_count = 1; + cursor->addrl = addr; + ppc440spe_rxor_set_region(desc, + cursor->addr_count, + DMA_RXOR12 << DMA_CUED_REGION_OFF); + ppc440spe_adma_dma2rxor_inc_addr( + desc, cursor, index, src_cnt); + } else if (addr == cursor->addrl + 2*sign*cursor->len) { + cursor->state = 2; + cursor->xor_count = 0; + ppc440spe_rxor_set_region(desc, + cursor->addr_count, + DMA_RXOR123 << DMA_CUED_REGION_OFF); + if (index == src_cnt-1) { + ppc440spe_adma_dma2rxor_inc_addr( + desc, cursor, index, src_cnt); + } + } else if (addr == cursor->addrl + 3*cursor->len) { + cursor->state = 2; + cursor->xor_count = 0; + ppc440spe_rxor_set_region(desc, + cursor->addr_count, + DMA_RXOR124 << DMA_CUED_REGION_OFF); + if (index == src_cnt-1) { + ppc440spe_adma_dma2rxor_inc_addr( + desc, cursor, index, src_cnt); + } + } else if (addr == cursor->addrl + 4*cursor->len) { + cursor->state = 2; + cursor->xor_count = 0; + ppc440spe_rxor_set_region(desc, + cursor->addr_count, + DMA_RXOR125 << DMA_CUED_REGION_OFF); + if (index == src_cnt-1) { + ppc440spe_adma_dma2rxor_inc_addr( + desc, cursor, index, src_cnt); + } + } else { + cursor->state = 0; + cursor->xor_count = 1; + cursor->addrl = addr; + ppc440spe_rxor_set_region(desc, + cursor->addr_count, + DMA_RXOR12 << DMA_CUED_REGION_OFF); + ppc440spe_adma_dma2rxor_inc_addr( + desc, cursor, index, src_cnt); + } + break; + case 2: + cursor->state = 0; + cursor->addrl = addr; + cursor->xor_count++; + if (index) { + ppc440spe_adma_dma2rxor_inc_addr( + desc, cursor, index, src_cnt); + } + break; + } + + return rval; +} + +/** + * ppc440spe_adma_dma2rxor_set_src - set RXOR source address; it's assumed that + * ppc440spe_adma_dma2rxor_prep_src() has already done prior this call + */ +static void ppc440spe_adma_dma2rxor_set_src( + struct ppc440spe_adma_desc_slot *desc, + int index, dma_addr_t addr) +{ + struct xor_cb *xcb = desc->hw_desc; + int k = 0, op = 0, lop = 0; + + /* get the RXOR operand which corresponds to index addr */ + while (op <= index) { + lop = op; + if (k == XOR_MAX_OPS) { + k = 0; + desc = list_entry(desc->chain_node.next, + struct ppc440spe_adma_desc_slot, chain_node); + xcb = desc->hw_desc; + + } + if ((xcb->ops[k++].h & (DMA_RXOR12 << DMA_CUED_REGION_OFF)) == + (DMA_RXOR12 << DMA_CUED_REGION_OFF)) + op += 2; + else + op += 3; + } + + BUG_ON(k < 1); + + if (test_bit(k-1, desc->reverse_flags)) { + /* reverse operand order; put last op in RXOR group */ + if (index == op - 1) + ppc440spe_rxor_set_src(desc, k - 1, addr); + } else { + /* direct operand order; put first op in RXOR group */ + if (index == lop) + ppc440spe_rxor_set_src(desc, k - 1, addr); + } +} + +/** + * ppc440spe_adma_dma2rxor_set_mult - set RXOR multipliers; it's assumed that + * ppc440spe_adma_dma2rxor_prep_src() has already done prior this call + */ +static void ppc440spe_adma_dma2rxor_set_mult( + struct ppc440spe_adma_desc_slot *desc, + int index, u8 mult) +{ + struct xor_cb *xcb = desc->hw_desc; + int k = 0, op = 0, lop = 0; + + /* get the RXOR operand which corresponds to index mult */ + while (op <= index) { + lop = op; + if (k == XOR_MAX_OPS) { + k = 0; + desc = list_entry(desc->chain_node.next, + struct ppc440spe_adma_desc_slot, + chain_node); + xcb = desc->hw_desc; + + } + if ((xcb->ops[k++].h & (DMA_RXOR12 << DMA_CUED_REGION_OFF)) == + (DMA_RXOR12 << DMA_CUED_REGION_OFF)) + op += 2; + else + op += 3; + } + + BUG_ON(k < 1); + if (test_bit(k-1, desc->reverse_flags)) { + /* reverse order */ + ppc440spe_rxor_set_mult(desc, k - 1, op - index - 1, mult); + } else { + /* direct order */ + ppc440spe_rxor_set_mult(desc, k - 1, index - lop, mult); + } +} + +/** + * ppc440spe_init_rxor_cursor - + */ +static void ppc440spe_init_rxor_cursor(struct ppc440spe_rxor *cursor) +{ + memset(cursor, 0, sizeof(struct ppc440spe_rxor)); + cursor->state = 2; +} + +/** + * ppc440spe_adma_pq_set_src_mult - set multiplication coefficient into + * descriptor for the PQXOR operation + */ +static void ppc440spe_adma_pq_set_src_mult( + struct ppc440spe_adma_desc_slot *sw_desc, + unsigned char mult, int index, int dst_pos) +{ + struct ppc440spe_adma_chan *chan; + u32 mult_idx, mult_dst; + struct ppc440spe_adma_desc_slot *iter = NULL, *iter1 = NULL; + + chan = to_ppc440spe_adma_chan(sw_desc->async_tx.chan); + + switch (chan->device->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + if (test_bit(PPC440SPE_DESC_RXOR, &sw_desc->flags)) { + int region = test_bit(PPC440SPE_DESC_RXOR12, + &sw_desc->flags) ? 2 : 3; + + if (index < region) { + /* RXOR multipliers */ + iter = ppc440spe_get_group_entry(sw_desc, + sw_desc->dst_cnt - 1); + if (sw_desc->dst_cnt == 2) + iter1 = ppc440spe_get_group_entry( + sw_desc, 0); + + mult_idx = DMA_CUED_MULT1_OFF + (index << 3); + mult_dst = DMA_CDB_SG_SRC; + } else { + /* WXOR multiplier */ + iter = ppc440spe_get_group_entry(sw_desc, + index - region + + sw_desc->dst_cnt); + mult_idx = DMA_CUED_MULT1_OFF; + mult_dst = dst_pos ? DMA_CDB_SG_DST2 : + DMA_CDB_SG_DST1; + } + } else { + int znum = 0; + + /* WXOR-only; + * skip first slots with destinations (if ZERO_DST has + * place) + */ + if (test_bit(PPC440SPE_ZERO_P, &sw_desc->flags)) + znum++; + if (test_bit(PPC440SPE_ZERO_Q, &sw_desc->flags)) + znum++; + + iter = ppc440spe_get_group_entry(sw_desc, index + znum); + mult_idx = DMA_CUED_MULT1_OFF; + mult_dst = dst_pos ? DMA_CDB_SG_DST2 : DMA_CDB_SG_DST1; + } + + if (likely(iter)) { + ppc440spe_desc_set_src_mult(iter, chan, + mult_idx, mult_dst, mult); + + if (unlikely(iter1)) { + /* if we have two destinations for RXOR, then + * we've just set Q mult. Set-up P now. + */ + ppc440spe_desc_set_src_mult(iter1, chan, + mult_idx, mult_dst, 1); + } + + } + break; + + case PPC440SPE_XOR_ID: + iter = sw_desc->group_head; + if (sw_desc->dst_cnt == 2) { + /* both P & Q calculations required; set P mult here */ + ppc440spe_adma_dma2rxor_set_mult(iter, index, 1); + + /* and then set Q mult */ + iter = ppc440spe_get_group_entry(sw_desc, + sw_desc->descs_per_op); + } + ppc440spe_adma_dma2rxor_set_mult(iter, index, mult); + break; + } +} + +/** + * ppc440spe_adma_free_chan_resources - free the resources allocated + */ +static void ppc440spe_adma_free_chan_resources(struct dma_chan *chan) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + struct ppc440spe_adma_desc_slot *iter, *_iter; + int in_use_descs = 0; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + ppc440spe_adma_slot_cleanup(ppc440spe_chan); + + spin_lock_bh(&ppc440spe_chan->lock); + list_for_each_entry_safe(iter, _iter, &ppc440spe_chan->chain, + chain_node) { + in_use_descs++; + list_del(&iter->chain_node); + } + list_for_each_entry_safe_reverse(iter, _iter, + &ppc440spe_chan->all_slots, slot_node) { + list_del(&iter->slot_node); + kfree(iter); + ppc440spe_chan->slots_allocated--; + } + ppc440spe_chan->last_used = NULL; + + dev_dbg(ppc440spe_chan->device->common.dev, + "ppc440spe adma%d %s slots_allocated %d\n", + ppc440spe_chan->device->id, + __func__, ppc440spe_chan->slots_allocated); + spin_unlock_bh(&ppc440spe_chan->lock); + + /* one is ok since we left it on there on purpose */ + if (in_use_descs > 1) + printk(KERN_ERR "SPE: Freeing %d in use descriptors!\n", + in_use_descs - 1); +} + +/** + * ppc440spe_adma_is_complete - poll the status of an ADMA transaction + * @chan: ADMA channel handle + * @cookie: ADMA transaction identifier + */ +static enum dma_status ppc440spe_adma_is_complete(struct dma_chan *chan, + dma_cookie_t cookie, dma_cookie_t *done, dma_cookie_t *used) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + dma_cookie_t last_used; + dma_cookie_t last_complete; + enum dma_status ret; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + last_used = chan->cookie; + last_complete = ppc440spe_chan->completed_cookie; + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + if (ret == DMA_SUCCESS) + return ret; + + ppc440spe_adma_slot_cleanup(ppc440spe_chan); + + last_used = chan->cookie; + last_complete = ppc440spe_chan->completed_cookie; + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + return dma_async_is_complete(cookie, last_complete, last_used); +} + +/** + * ppc440spe_adma_eot_handler - end of transfer interrupt handler + */ +static irqreturn_t ppc440spe_adma_eot_handler(int irq, void *data) +{ + struct ppc440spe_adma_chan *chan = data; + + dev_dbg(chan->device->common.dev, + "ppc440spe adma%d: %s\n", chan->device->id, __func__); + + tasklet_schedule(&chan->irq_tasklet); + ppc440spe_adma_device_clear_eot_status(chan); + + return IRQ_HANDLED; +} + +/** + * ppc440spe_adma_err_handler - DMA error interrupt handler; + * do the same things as a eot handler + */ +static irqreturn_t ppc440spe_adma_err_handler(int irq, void *data) +{ + struct ppc440spe_adma_chan *chan = data; + + dev_dbg(chan->device->common.dev, + "ppc440spe adma%d: %s\n", chan->device->id, __func__); + + tasklet_schedule(&chan->irq_tasklet); + ppc440spe_adma_device_clear_eot_status(chan); + + return IRQ_HANDLED; +} + +/** + * ppc440spe_test_callback - called when test operation has been done + */ +static void ppc440spe_test_callback(void *unused) +{ + complete(&ppc440spe_r6_test_comp); +} + +/** + * ppc440spe_adma_issue_pending - flush all pending descriptors to h/w + */ +static void ppc440spe_adma_issue_pending(struct dma_chan *chan) +{ + struct ppc440spe_adma_chan *ppc440spe_chan; + + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + dev_dbg(ppc440spe_chan->device->common.dev, + "ppc440spe adma%d: %s %d \n", ppc440spe_chan->device->id, + __func__, ppc440spe_chan->pending); + + if (ppc440spe_chan->pending) { + ppc440spe_chan->pending = 0; + ppc440spe_chan_append(ppc440spe_chan); + } +} + +/** + * ppc440spe_chan_start_null_xor - initiate the first XOR operation (DMA engines + * use FIFOs (as opposite to chains used in XOR) so this is a XOR + * specific operation) + */ +static void ppc440spe_chan_start_null_xor(struct ppc440spe_adma_chan *chan) +{ + struct ppc440spe_adma_desc_slot *sw_desc, *group_start; + dma_cookie_t cookie; + int slot_cnt, slots_per_op; + + dev_dbg(chan->device->common.dev, + "ppc440spe adma%d: %s\n", chan->device->id, __func__); + + spin_lock_bh(&chan->lock); + slot_cnt = ppc440spe_chan_xor_slot_count(0, 2, &slots_per_op); + sw_desc = ppc440spe_adma_alloc_slots(chan, slot_cnt, slots_per_op); + if (sw_desc) { + group_start = sw_desc->group_head; + list_splice_init(&sw_desc->group_list, &chan->chain); + async_tx_ack(&sw_desc->async_tx); + ppc440spe_desc_init_null_xor(group_start); + + cookie = chan->common.cookie; + cookie++; + if (cookie <= 1) + cookie = 2; + + /* initialize the completed cookie to be less than + * the most recently used cookie + */ + chan->completed_cookie = cookie - 1; + chan->common.cookie = sw_desc->async_tx.cookie = cookie; + + /* channel should not be busy */ + BUG_ON(ppc440spe_chan_is_busy(chan)); + + /* set the descriptor address */ + ppc440spe_chan_set_first_xor_descriptor(chan, sw_desc); + + /* run the descriptor */ + ppc440spe_chan_run(chan); + } else + printk(KERN_ERR "ppc440spe adma%d" + " failed to allocate null descriptor\n", + chan->device->id); + spin_unlock_bh(&chan->lock); +} + +/** + * ppc440spe_test_raid6 - test are RAID-6 capabilities enabled successfully. + * For this we just perform one WXOR operation with the same source + * and destination addresses, the GF-multiplier is 1; so if RAID-6 + * capabilities are enabled then we'll get src/dst filled with zero. + */ +static int ppc440spe_test_raid6(struct ppc440spe_adma_chan *chan) +{ + struct ppc440spe_adma_desc_slot *sw_desc, *iter; + struct page *pg; + char *a; + dma_addr_t dma_addr, addrs[2]; + unsigned long op = 0; + int rval = 0; + + set_bit(PPC440SPE_DESC_WXOR, &op); + + pg = alloc_page(GFP_KERNEL); + if (!pg) + return -ENOMEM; + + spin_lock_bh(&chan->lock); + sw_desc = ppc440spe_adma_alloc_slots(chan, 1, 1); + if (sw_desc) { + /* 1 src, 1 dsr, int_ena, WXOR */ + ppc440spe_desc_init_dma01pq(sw_desc, 1, 1, 1, op); + list_for_each_entry(iter, &sw_desc->group_list, chain_node) { + ppc440spe_desc_set_byte_count(iter, chan, PAGE_SIZE); + iter->unmap_len = PAGE_SIZE; + } + } else { + rval = -EFAULT; + spin_unlock_bh(&chan->lock); + goto exit; + } + spin_unlock_bh(&chan->lock); + + /* Fill the test page with ones */ + memset(page_address(pg), 0xFF, PAGE_SIZE); + dma_addr = dma_map_page(chan->device->dev, pg, 0, + PAGE_SIZE, DMA_BIDIRECTIONAL); + + /* Setup addresses */ + ppc440spe_adma_pq_set_src(sw_desc, dma_addr, 0); + ppc440spe_adma_pq_set_src_mult(sw_desc, 1, 0, 0); + addrs[0] = dma_addr; + addrs[1] = 0; + ppc440spe_adma_pq_set_dest(sw_desc, addrs, DMA_PREP_PQ_DISABLE_Q); + + async_tx_ack(&sw_desc->async_tx); + sw_desc->async_tx.callback = ppc440spe_test_callback; + sw_desc->async_tx.callback_param = NULL; + + init_completion(&ppc440spe_r6_test_comp); + + ppc440spe_adma_tx_submit(&sw_desc->async_tx); + ppc440spe_adma_issue_pending(&chan->common); + + wait_for_completion(&ppc440spe_r6_test_comp); + + /* Now check if the test page is zeroed */ + a = page_address(pg); + if ((*(u32 *)a) == 0 && memcmp(a, a+4, PAGE_SIZE-4) == 0) { + /* page is zero - RAID-6 enabled */ + rval = 0; + } else { + /* RAID-6 was not enabled */ + rval = -EINVAL; + } +exit: + __free_page(pg); + return rval; +} + +static void ppc440spe_adma_init_capabilities(struct ppc440spe_adma_device *adev) +{ + switch (adev->id) { + case PPC440SPE_DMA0_ID: + case PPC440SPE_DMA1_ID: + dma_cap_set(DMA_MEMCPY, adev->common.cap_mask); + dma_cap_set(DMA_INTERRUPT, adev->common.cap_mask); + dma_cap_set(DMA_MEMSET, adev->common.cap_mask); + dma_cap_set(DMA_PQ, adev->common.cap_mask); + dma_cap_set(DMA_PQ_VAL, adev->common.cap_mask); + dma_cap_set(DMA_XOR_VAL, adev->common.cap_mask); + break; + case PPC440SPE_XOR_ID: + dma_cap_set(DMA_XOR, adev->common.cap_mask); + dma_cap_set(DMA_PQ, adev->common.cap_mask); + dma_cap_set(DMA_INTERRUPT, adev->common.cap_mask); + adev->common.cap_mask = adev->common.cap_mask; + break; + } + + /* Set base routines */ + adev->common.device_alloc_chan_resources = + ppc440spe_adma_alloc_chan_resources; + adev->common.device_free_chan_resources = + ppc440spe_adma_free_chan_resources; + adev->common.device_is_tx_complete = ppc440spe_adma_is_complete; + adev->common.device_issue_pending = ppc440spe_adma_issue_pending; + + /* Set prep routines based on capability */ + if (dma_has_cap(DMA_MEMCPY, adev->common.cap_mask)) { + adev->common.device_prep_dma_memcpy = + ppc440spe_adma_prep_dma_memcpy; + } + if (dma_has_cap(DMA_MEMSET, adev->common.cap_mask)) { + adev->common.device_prep_dma_memset = + ppc440spe_adma_prep_dma_memset; + } + if (dma_has_cap(DMA_XOR, adev->common.cap_mask)) { + adev->common.max_xor = XOR_MAX_OPS; + adev->common.device_prep_dma_xor = + ppc440spe_adma_prep_dma_xor; + } + if (dma_has_cap(DMA_PQ, adev->common.cap_mask)) { + switch (adev->id) { + case PPC440SPE_DMA0_ID: + dma_set_maxpq(&adev->common, + DMA0_FIFO_SIZE / sizeof(struct dma_cdb), 0); + break; + case PPC440SPE_DMA1_ID: + dma_set_maxpq(&adev->common, + DMA1_FIFO_SIZE / sizeof(struct dma_cdb), 0); + break; + case PPC440SPE_XOR_ID: + adev->common.max_pq = XOR_MAX_OPS * 3; + break; + } + adev->common.device_prep_dma_pq = + ppc440spe_adma_prep_dma_pq; + } + if (dma_has_cap(DMA_PQ_VAL, adev->common.cap_mask)) { + switch (adev->id) { + case PPC440SPE_DMA0_ID: + adev->common.max_pq = DMA0_FIFO_SIZE / + sizeof(struct dma_cdb); + break; + case PPC440SPE_DMA1_ID: + adev->common.max_pq = DMA1_FIFO_SIZE / + sizeof(struct dma_cdb); + break; + } + adev->common.device_prep_dma_pq_val = + ppc440spe_adma_prep_dma_pqzero_sum; + } + if (dma_has_cap(DMA_XOR_VAL, adev->common.cap_mask)) { + switch (adev->id) { + case PPC440SPE_DMA0_ID: + adev->common.max_xor = DMA0_FIFO_SIZE / + sizeof(struct dma_cdb); + break; + case PPC440SPE_DMA1_ID: + adev->common.max_xor = DMA1_FIFO_SIZE / + sizeof(struct dma_cdb); + break; + } + adev->common.device_prep_dma_xor_val = + ppc440spe_adma_prep_dma_xor_zero_sum; + } + if (dma_has_cap(DMA_INTERRUPT, adev->common.cap_mask)) { + adev->common.device_prep_dma_interrupt = + ppc440spe_adma_prep_dma_interrupt; + } + pr_info("%s: AMCC(R) PPC440SP(E) ADMA Engine: " + "( %s%s%s%s%s%s%s)\n", + dev_name(adev->dev), + dma_has_cap(DMA_PQ, adev->common.cap_mask) ? "pq " : "", + dma_has_cap(DMA_PQ_VAL, adev->common.cap_mask) ? "pq_val " : "", + dma_has_cap(DMA_XOR, adev->common.cap_mask) ? "xor " : "", + dma_has_cap(DMA_XOR_VAL, adev->common.cap_mask) ? "xor_val " : "", + dma_has_cap(DMA_MEMCPY, adev->common.cap_mask) ? "memcpy " : "", + dma_has_cap(DMA_MEMSET, adev->common.cap_mask) ? "memset " : "", + dma_has_cap(DMA_INTERRUPT, adev->common.cap_mask) ? "intr " : ""); +} + +static int ppc440spe_adma_setup_irqs(struct ppc440spe_adma_device *adev, + struct ppc440spe_adma_chan *chan, + int *initcode) +{ + struct device_node *np; + int ret; + + np = container_of(adev->dev, struct of_device, dev)->node; + if (adev->id != PPC440SPE_XOR_ID) { + adev->err_irq = irq_of_parse_and_map(np, 1); + if (adev->err_irq == NO_IRQ) { + dev_warn(adev->dev, "no err irq resource?\n"); + *initcode = PPC_ADMA_INIT_IRQ2; + adev->err_irq = -ENXIO; + } else + atomic_inc(&ppc440spe_adma_err_irq_ref); + } else { + adev->err_irq = -ENXIO; + } + + adev->irq = irq_of_parse_and_map(np, 0); + if (adev->irq == NO_IRQ) { + dev_err(adev->dev, "no irq resource\n"); + *initcode = PPC_ADMA_INIT_IRQ1; + ret = -ENXIO; + goto err_irq_map; + } + dev_dbg(adev->dev, "irq %d, err irq %d\n", + adev->irq, adev->err_irq); + + ret = request_irq(adev->irq, ppc440spe_adma_eot_handler, + 0, dev_driver_string(adev->dev), chan); + if (ret) { + dev_err(adev->dev, "can't request irq %d\n", + adev->irq); + *initcode = PPC_ADMA_INIT_IRQ1; + ret = -EIO; + goto err_req1; + } + + /* only DMA engines have a separate error IRQ + * so it's Ok if err_irq < 0 in XOR engine case. + */ + if (adev->err_irq > 0) { + /* both DMA engines share common error IRQ */ + ret = request_irq(adev->err_irq, + ppc440spe_adma_err_handler, + IRQF_SHARED, + dev_driver_string(adev->dev), + chan); + if (ret) { + dev_err(adev->dev, "can't request irq %d\n", + adev->err_irq); + *initcode = PPC_ADMA_INIT_IRQ2; + ret = -EIO; + goto err_req2; + } + } + + if (adev->id == PPC440SPE_XOR_ID) { + /* enable XOR engine interrupts */ + iowrite32be(XOR_IE_CBCIE_BIT | XOR_IE_ICBIE_BIT | + XOR_IE_ICIE_BIT | XOR_IE_RPTIE_BIT, + &adev->xor_reg->ier); + } else { + u32 mask, enable; + + np = of_find_compatible_node(NULL, NULL, "ibm,i2o-440spe"); + if (!np) { + pr_err("%s: can't find I2O device tree node\n", + __func__); + ret = -ENODEV; + goto err_req2; + } + adev->i2o_reg = of_iomap(np, 0); + if (!adev->i2o_reg) { + pr_err("%s: failed to map I2O registers\n", __func__); + of_node_put(np); + ret = -EINVAL; + goto err_req2; + } + of_node_put(np); + /* Unmask 'CS FIFO Attention' interrupts and + * enable generating interrupts on errors + */ + enable = (adev->id == PPC440SPE_DMA0_ID) ? + ~(I2O_IOPIM_P0SNE | I2O_IOPIM_P0EM) : + ~(I2O_IOPIM_P1SNE | I2O_IOPIM_P1EM); + mask = ioread32(&adev->i2o_reg->iopim) & enable; + iowrite32(mask, &adev->i2o_reg->iopim); + } + return 0; + +err_req2: + free_irq(adev->irq, chan); +err_req1: + irq_dispose_mapping(adev->irq); +err_irq_map: + if (adev->err_irq > 0) { + if (atomic_dec_and_test(&ppc440spe_adma_err_irq_ref)) + irq_dispose_mapping(adev->err_irq); + } + return ret; +} + +static void ppc440spe_adma_release_irqs(struct ppc440spe_adma_device *adev, + struct ppc440spe_adma_chan *chan) +{ + u32 mask, disable; + + if (adev->id == PPC440SPE_XOR_ID) { + /* disable XOR engine interrupts */ + mask = ioread32be(&adev->xor_reg->ier); + mask &= ~(XOR_IE_CBCIE_BIT | XOR_IE_ICBIE_BIT | + XOR_IE_ICIE_BIT | XOR_IE_RPTIE_BIT); + iowrite32be(mask, &adev->xor_reg->ier); + } else { + /* disable DMAx engine interrupts */ + disable = (adev->id == PPC440SPE_DMA0_ID) ? + (I2O_IOPIM_P0SNE | I2O_IOPIM_P0EM) : + (I2O_IOPIM_P1SNE | I2O_IOPIM_P1EM); + mask = ioread32(&adev->i2o_reg->iopim) | disable; + iowrite32(mask, &adev->i2o_reg->iopim); + } + free_irq(adev->irq, chan); + irq_dispose_mapping(adev->irq); + if (adev->err_irq > 0) { + free_irq(adev->err_irq, chan); + if (atomic_dec_and_test(&ppc440spe_adma_err_irq_ref)) { + irq_dispose_mapping(adev->err_irq); + iounmap(adev->i2o_reg); + } + } +} + +/** + * ppc440spe_adma_probe - probe the asynch device + */ +static int __devinit ppc440spe_adma_probe(struct of_device *ofdev, + const struct of_device_id *match) +{ + struct device_node *np = ofdev->node; + struct resource res; + struct ppc440spe_adma_device *adev; + struct ppc440spe_adma_chan *chan; + struct ppc_dma_chan_ref *ref, *_ref; + int ret = 0, initcode = PPC_ADMA_INIT_OK; + const u32 *idx; + int len; + void *regs; + u32 id, pool_size; + + if (of_device_is_compatible(np, "amcc,xor-accelerator")) { + id = PPC440SPE_XOR_ID; + /* As far as the XOR engine is concerned, it does not + * use FIFOs but uses linked list. So there is no dependency + * between pool size to allocate and the engine configuration. + */ + pool_size = PAGE_SIZE << 1; + } else { + /* it is DMA0 or DMA1 */ + idx = of_get_property(np, "cell-index", &len); + if (!idx || (len != sizeof(u32))) { + dev_err(&ofdev->dev, "Device node %s has missing " + "or invalid cell-index property\n", + np->full_name); + return -EINVAL; + } + id = *idx; + /* DMA0,1 engines use FIFO to maintain CDBs, so we + * should allocate the pool accordingly to size of this + * FIFO. Thus, the pool size depends on the FIFO depth: + * how much CDBs pointers the FIFO may contain then so + * much CDBs we should provide in the pool. + * That is + * CDB size = 32B; + * CDBs number = (DMA0_FIFO_SIZE >> 3); + * Pool size = CDBs number * CDB size = + * = (DMA0_FIFO_SIZE >> 3) << 5 = DMA0_FIFO_SIZE << 2. + */ + pool_size = (id == PPC440SPE_DMA0_ID) ? + DMA0_FIFO_SIZE : DMA1_FIFO_SIZE; + pool_size <<= 2; + } + + if (of_address_to_resource(np, 0, &res)) { + dev_err(&ofdev->dev, "failed to get memory resource\n"); + initcode = PPC_ADMA_INIT_MEMRES; + ret = -ENODEV; + goto out; + } + + if (!request_mem_region(res.start, resource_size(&res), + dev_driver_string(&ofdev->dev))) { + dev_err(&ofdev->dev, "failed to request memory region " + "(0x%016llx-0x%016llx)\n", + (u64)res.start, (u64)res.end); + initcode = PPC_ADMA_INIT_MEMREG; + ret = -EBUSY; + goto out; + } + + /* create a device */ + adev = kzalloc(sizeof(*adev), GFP_KERNEL); + if (!adev) { + dev_err(&ofdev->dev, "failed to allocate device\n"); + initcode = PPC_ADMA_INIT_ALLOC; + ret = -ENOMEM; + goto err_adev_alloc; + } + + adev->id = id; + adev->pool_size = pool_size; + /* allocate coherent memory for hardware descriptors */ + adev->dma_desc_pool_virt = dma_alloc_coherent(&ofdev->dev, + adev->pool_size, &adev->dma_desc_pool, + GFP_KERNEL); + if (adev->dma_desc_pool_virt == NULL) { + dev_err(&ofdev->dev, "failed to allocate %d bytes of coherent " + "memory for hardware descriptors\n", + adev->pool_size); + initcode = PPC_ADMA_INIT_COHERENT; + ret = -ENOMEM; + goto err_dma_alloc; + } + dev_dbg(&ofdev->dev, "allocted descriptor pool virt 0x%p phys 0x%llx\n", + adev->dma_desc_pool_virt, (u64)adev->dma_desc_pool); + + regs = ioremap(res.start, resource_size(&res)); + if (!regs) { + dev_err(&ofdev->dev, "failed to ioremap regs!\n"); + goto err_regs_alloc; + } + + if (adev->id == PPC440SPE_XOR_ID) { + adev->xor_reg = regs; + /* Reset XOR */ + iowrite32be(XOR_CRSR_XASR_BIT, &adev->xor_reg->crsr); + iowrite32be(XOR_CRSR_64BA_BIT, &adev->xor_reg->crrr); + } else { + size_t fifo_size = (adev->id == PPC440SPE_DMA0_ID) ? + DMA0_FIFO_SIZE : DMA1_FIFO_SIZE; + adev->dma_reg = regs; + /* DMAx_FIFO_SIZE is defined in bytes, + * <fsiz> - is defined in number of CDB pointers (8byte). + * DMA FIFO Length = CSlength + CPlength, where + * CSlength = CPlength = (fsiz + 1) * 8. + */ + iowrite32(DMA_FIFO_ENABLE | ((fifo_size >> 3) - 2), + &adev->dma_reg->fsiz); + /* Configure DMA engine */ + iowrite32(DMA_CFG_DXEPR_HP | DMA_CFG_DFMPP_HP | DMA_CFG_FALGN, + &adev->dma_reg->cfg); + /* Clear Status */ + iowrite32(~0, &adev->dma_reg->dsts); + } + + adev->dev = &ofdev->dev; + adev->common.dev = &ofdev->dev; + INIT_LIST_HEAD(&adev->common.channels); + dev_set_drvdata(&ofdev->dev, adev); + + /* create a channel */ + chan = kzalloc(sizeof(*chan), GFP_KERNEL); + if (!chan) { + dev_err(&ofdev->dev, "can't allocate channel structure\n"); + initcode = PPC_ADMA_INIT_CHANNEL; + ret = -ENOMEM; + goto err_chan_alloc; + } + + spin_lock_init(&chan->lock); + INIT_LIST_HEAD(&chan->chain); + INIT_LIST_HEAD(&chan->all_slots); + chan->device = adev; + chan->common.device = &adev->common; + list_add_tail(&chan->common.device_node, &adev->common.channels); + tasklet_init(&chan->irq_tasklet, ppc440spe_adma_tasklet, + (unsigned long)chan); + + /* allocate and map helper pages for async validation or + * async_mult/async_sum_product operations on DMA0/1. + */ + if (adev->id != PPC440SPE_XOR_ID) { + chan->pdest_page = alloc_page(GFP_KERNEL); + chan->qdest_page = alloc_page(GFP_KERNEL); + if (!chan->pdest_page || + !chan->qdest_page) { + if (chan->pdest_page) + __free_page(chan->pdest_page); + if (chan->qdest_page) + __free_page(chan->qdest_page); + ret = -ENOMEM; + goto err_page_alloc; + } + chan->pdest = dma_map_page(&ofdev->dev, chan->pdest_page, 0, + PAGE_SIZE, DMA_BIDIRECTIONAL); + chan->qdest = dma_map_page(&ofdev->dev, chan->qdest_page, 0, + PAGE_SIZE, DMA_BIDIRECTIONAL); + } + + ref = kmalloc(sizeof(*ref), GFP_KERNEL); + if (ref) { + ref->chan = &chan->common; + INIT_LIST_HEAD(&ref->node); + list_add_tail(&ref->node, &ppc440spe_adma_chan_list); + } else { + dev_err(&ofdev->dev, "failed to allocate channel reference!\n"); + ret = -ENOMEM; + goto err_ref_alloc; + } + + ret = ppc440spe_adma_setup_irqs(adev, chan, &initcode); + if (ret) + goto err_irq; + + ppc440spe_adma_init_capabilities(adev); + + ret = dma_async_device_register(&adev->common); + if (ret) { + initcode = PPC_ADMA_INIT_REGISTER; + dev_err(&ofdev->dev, "failed to register dma device\n"); + goto err_dev_reg; + } + + goto out; + +err_dev_reg: + ppc440spe_adma_release_irqs(adev, chan); +err_irq: + list_for_each_entry_safe(ref, _ref, &ppc440spe_adma_chan_list, node) { + if (chan == to_ppc440spe_adma_chan(ref->chan)) { + list_del(&ref->node); + kfree(ref); + } + } +err_ref_alloc: + if (adev->id != PPC440SPE_XOR_ID) { + dma_unmap_page(&ofdev->dev, chan->pdest, + PAGE_SIZE, DMA_BIDIRECTIONAL); + dma_unmap_page(&ofdev->dev, chan->qdest, + PAGE_SIZE, DMA_BIDIRECTIONAL); + __free_page(chan->pdest_page); + __free_page(chan->qdest_page); + } +err_page_alloc: + kfree(chan); +err_chan_alloc: + if (adev->id == PPC440SPE_XOR_ID) + iounmap(adev->xor_reg); + else + iounmap(adev->dma_reg); +err_regs_alloc: + dma_free_coherent(adev->dev, adev->pool_size, + adev->dma_desc_pool_virt, + adev->dma_desc_pool); +err_dma_alloc: + kfree(adev); +err_adev_alloc: + release_mem_region(res.start, resource_size(&res)); +out: + if (id < PPC440SPE_ADMA_ENGINES_NUM) + ppc440spe_adma_devices[id] = initcode; + + return ret; +} + +/** + * ppc440spe_adma_remove - remove the asynch device + */ +static int __devexit ppc440spe_adma_remove(struct of_device *ofdev) +{ + struct ppc440spe_adma_device *adev = dev_get_drvdata(&ofdev->dev); + struct device_node *np = ofdev->node; + struct resource res; + struct dma_chan *chan, *_chan; + struct ppc_dma_chan_ref *ref, *_ref; + struct ppc440spe_adma_chan *ppc440spe_chan; + + dev_set_drvdata(&ofdev->dev, NULL); + if (adev->id < PPC440SPE_ADMA_ENGINES_NUM) + ppc440spe_adma_devices[adev->id] = -1; + + dma_async_device_unregister(&adev->common); + + list_for_each_entry_safe(chan, _chan, &adev->common.channels, + device_node) { + ppc440spe_chan = to_ppc440spe_adma_chan(chan); + ppc440spe_adma_release_irqs(adev, ppc440spe_chan); + tasklet_kill(&ppc440spe_chan->irq_tasklet); + if (adev->id != PPC440SPE_XOR_ID) { + dma_unmap_page(&ofdev->dev, ppc440spe_chan->pdest, + PAGE_SIZE, DMA_BIDIRECTIONAL); + dma_unmap_page(&ofdev->dev, ppc440spe_chan->qdest, + PAGE_SIZE, DMA_BIDIRECTIONAL); + __free_page(ppc440spe_chan->pdest_page); + __free_page(ppc440spe_chan->qdest_page); + } + list_for_each_entry_safe(ref, _ref, &ppc440spe_adma_chan_list, + node) { + if (ppc440spe_chan == + to_ppc440spe_adma_chan(ref->chan)) { + list_del(&ref->node); + kfree(ref); + } + } + list_del(&chan->device_node); + kfree(ppc440spe_chan); + } + + dma_free_coherent(adev->dev, adev->pool_size, + adev->dma_desc_pool_virt, adev->dma_desc_pool); + if (adev->id == PPC440SPE_XOR_ID) + iounmap(adev->xor_reg); + else + iounmap(adev->dma_reg); + of_address_to_resource(np, 0, &res); + release_mem_region(res.start, resource_size(&res)); + kfree(adev); + return 0; +} + +/* + * /sys driver interface to enable h/w RAID-6 capabilities + * Files created in e.g. /sys/devices/plb.0/400100100.dma0/driver/ + * directory are "devices", "enable" and "poly". + * "devices" shows available engines. + * "enable" is used to enable RAID-6 capabilities or to check + * whether these has been activated. + * "poly" allows setting/checking used polynomial (for PPC440SPe only). + */ + +static ssize_t show_ppc440spe_devices(struct device_driver *dev, char *buf) +{ + ssize_t size = 0; + int i; + + for (i = 0; i < PPC440SPE_ADMA_ENGINES_NUM; i++) { + if (ppc440spe_adma_devices[i] == -1) + continue; + size += snprintf(buf + size, PAGE_SIZE - size, + "PPC440SP(E)-ADMA.%d: %s\n", i, + ppc_adma_errors[ppc440spe_adma_devices[i]]); + } + return size; +} + +static ssize_t show_ppc440spe_r6enable(struct device_driver *dev, char *buf) +{ + return snprintf(buf, PAGE_SIZE, + "PPC440SP(e) RAID-6 capabilities are %sABLED.\n", + ppc440spe_r6_enabled ? "EN" : "DIS"); +} + +static ssize_t store_ppc440spe_r6enable(struct device_driver *dev, + const char *buf, size_t count) +{ + unsigned long val; + + if (!count || count > 11) + return -EINVAL; + + if (!ppc440spe_r6_tchan) + return -EFAULT; + + /* Write a key */ + sscanf(buf, "%lx", &val); + dcr_write(ppc440spe_mq_dcr_host, DCRN_MQ0_XORBA, val); + isync(); + + /* Verify whether it really works now */ + if (ppc440spe_test_raid6(ppc440spe_r6_tchan) == 0) { + pr_info("PPC440SP(e) RAID-6 has been activated " + "successfully\n"); + ppc440spe_r6_enabled = 1; + } else { + pr_info("PPC440SP(e) RAID-6 hasn't been activated!" + " Error key ?\n"); + ppc440spe_r6_enabled = 0; + } + return count; +} + +static ssize_t show_ppc440spe_r6poly(struct device_driver *dev, char *buf) +{ + ssize_t size = 0; + u32 reg; + +#ifdef CONFIG_440SP + /* 440SP has fixed polynomial */ + reg = 0x4d; +#else + reg = dcr_read(ppc440spe_mq_dcr_host, DCRN_MQ0_CFBHL); + reg >>= MQ0_CFBHL_POLY; + reg &= 0xFF; +#endif + + size = snprintf(buf, PAGE_SIZE, "PPC440SP(e) RAID-6 driver " + "uses 0x1%02x polynomial.\n", reg); + return size; +} + +static ssize_t store_ppc440spe_r6poly(struct device_driver *dev, + const char *buf, size_t count) +{ + unsigned long reg, val; + +#ifdef CONFIG_440SP + /* 440SP uses default 0x14D polynomial only */ + return -EINVAL; +#endif + + if (!count || count > 6) + return -EINVAL; + + /* e.g., 0x14D or 0x11D */ + sscanf(buf, "%lx", &val); + + if (val & ~0x1FF) + return -EINVAL; + + val &= 0xFF; + reg = dcr_read(ppc440spe_mq_dcr_host, DCRN_MQ0_CFBHL); + reg &= ~(0xFF << MQ0_CFBHL_POLY); + reg |= val << MQ0_CFBHL_POLY; + dcr_write(ppc440spe_mq_dcr_host, DCRN_MQ0_CFBHL, reg); + + return count; +} + +static DRIVER_ATTR(devices, S_IRUGO, show_ppc440spe_devices, NULL); +static DRIVER_ATTR(enable, S_IRUGO | S_IWUSR, show_ppc440spe_r6enable, + store_ppc440spe_r6enable); +static DRIVER_ATTR(poly, S_IRUGO | S_IWUSR, show_ppc440spe_r6poly, + store_ppc440spe_r6poly); + +/* + * Common initialisation for RAID engines; allocate memory for + * DMAx FIFOs, perform configuration common for all DMA engines. + * Further DMA engine specific configuration is done at probe time. + */ +static int ppc440spe_configure_raid_devices(void) +{ + struct device_node *np; + struct resource i2o_res; + struct i2o_regs __iomem *i2o_reg; + dcr_host_t i2o_dcr_host; + unsigned int dcr_base, dcr_len; + int i, ret; + + np = of_find_compatible_node(NULL, NULL, "ibm,i2o-440spe"); + if (!np) { + pr_err("%s: can't find I2O device tree node\n", + __func__); + return -ENODEV; + } + + if (of_address_to_resource(np, 0, &i2o_res)) { + of_node_put(np); + return -EINVAL; + } + + i2o_reg = of_iomap(np, 0); + if (!i2o_reg) { + pr_err("%s: failed to map I2O registers\n", __func__); + of_node_put(np); + return -EINVAL; + } + + /* Get I2O DCRs base */ + dcr_base = dcr_resource_start(np, 0); + dcr_len = dcr_resource_len(np, 0); + if (!dcr_base && !dcr_len) { + pr_err("%s: can't get DCR registers base/len!\n", + np->full_name); + of_node_put(np); + iounmap(i2o_reg); + return -ENODEV; + } + + i2o_dcr_host = dcr_map(np, dcr_base, dcr_len); + if (!DCR_MAP_OK(i2o_dcr_host)) { + pr_err("%s: failed to map DCRs!\n", np->full_name); + of_node_put(np); + iounmap(i2o_reg); + return -ENODEV; + } + of_node_put(np); + + /* Provide memory regions for DMA's FIFOs: I2O, DMA0 and DMA1 share + * the base address of FIFO memory space. + * Actually we need twice more physical memory than programmed in the + * <fsiz> register (because there are two FIFOs for each DMA: CP and CS) + */ + ppc440spe_dma_fifo_buf = kmalloc((DMA0_FIFO_SIZE + DMA1_FIFO_SIZE) << 1, + GFP_KERNEL); + if (!ppc440spe_dma_fifo_buf) { + pr_err("%s: DMA FIFO buffer allocation failed.\n", __func__); + iounmap(i2o_reg); + dcr_unmap(i2o_dcr_host, dcr_len); + return -ENOMEM; + } + + /* + * Configure h/w + */ + /* Reset I2O/DMA */ + mtdcri(SDR0, DCRN_SDR0_SRST, DCRN_SDR0_SRST_I2ODMA); + mtdcri(SDR0, DCRN_SDR0_SRST, 0); + + /* Setup the base address of mmaped registers */ + dcr_write(i2o_dcr_host, DCRN_I2O0_IBAH, (u32)(i2o_res.start >> 32)); + dcr_write(i2o_dcr_host, DCRN_I2O0_IBAL, (u32)(i2o_res.start) | + I2O_REG_ENABLE); + dcr_unmap(i2o_dcr_host, dcr_len); + + /* Setup FIFO memory space base address */ + iowrite32(0, &i2o_reg->ifbah); + iowrite32(((u32)__pa(ppc440spe_dma_fifo_buf)), &i2o_reg->ifbal); + + /* set zero FIFO size for I2O, so the whole + * ppc440spe_dma_fifo_buf is used by DMAs. + * DMAx_FIFOs will be configured while probe. + */ + iowrite32(0, &i2o_reg->ifsiz); + iounmap(i2o_reg); + + /* To prepare WXOR/RXOR functionality we need access to + * Memory Queue Module DCRs (finally it will be enabled + * via /sys interface of the ppc440spe ADMA driver). + */ + np = of_find_compatible_node(NULL, NULL, "ibm,mq-440spe"); + if (!np) { + pr_err("%s: can't find MQ device tree node\n", + __func__); + ret = -ENODEV; + goto out_free; + } + + /* Get MQ DCRs base */ + dcr_base = dcr_resource_start(np, 0); + dcr_len = dcr_resource_len(np, 0); + if (!dcr_base && !dcr_len) { + pr_err("%s: can't get DCR registers base/len!\n", + np->full_name); + ret = -ENODEV; + goto out_mq; + } + + ppc440spe_mq_dcr_host = dcr_map(np, dcr_base, dcr_len); + if (!DCR_MAP_OK(ppc440spe_mq_dcr_host)) { + pr_err("%s: failed to map DCRs!\n", np->full_name); + ret = -ENODEV; + goto out_mq; + } + of_node_put(np); + ppc440spe_mq_dcr_len = dcr_len; + + /* Set HB alias */ + dcr_write(ppc440spe_mq_dcr_host, DCRN_MQ0_BAUH, DMA_CUED_XOR_HB); + + /* Set: + * - LL transaction passing limit to 1; + * - Memory controller cycle limit to 1; + * - Galois Polynomial to 0x14d (default) + */ + dcr_write(ppc440spe_mq_dcr_host, DCRN_MQ0_CFBHL, + (1 << MQ0_CFBHL_TPLM) | (1 << MQ0_CFBHL_HBCL) | + (PPC440SPE_DEFAULT_POLY << MQ0_CFBHL_POLY)); + + atomic_set(&ppc440spe_adma_err_irq_ref, 0); + for (i = 0; i < PPC440SPE_ADMA_ENGINES_NUM; i++) + ppc440spe_adma_devices[i] = -1; + + return 0; + +out_mq: + of_node_put(np); +out_free: + kfree(ppc440spe_dma_fifo_buf); + return ret; +} + +static struct of_device_id __devinitdata ppc440spe_adma_of_match[] = { + { .compatible = "ibm,dma-440spe", }, + { .compatible = "amcc,xor-accelerator", }, + {}, +}; +MODULE_DEVICE_TABLE(of, ppc440spe_adma_of_match); + +static struct of_platform_driver ppc440spe_adma_driver = { + .match_table = ppc440spe_adma_of_match, + .probe = ppc440spe_adma_probe, + .remove = __devexit_p(ppc440spe_adma_remove), + .driver = { + .name = "PPC440SP(E)-ADMA", + .owner = THIS_MODULE, + }, +}; + +static __init int ppc440spe_adma_init(void) +{ + int ret; + + ret = ppc440spe_configure_raid_devices(); + if (ret) + return ret; + + ret = of_register_platform_driver(&ppc440spe_adma_driver); + if (ret) { + pr_err("%s: failed to register platform driver\n", + __func__); + goto out_reg; + } + + /* Initialization status */ + ret = driver_create_file(&ppc440spe_adma_driver.driver, + &driver_attr_devices); + if (ret) + goto out_dev; + + /* RAID-6 h/w enable entry */ + ret = driver_create_file(&ppc440spe_adma_driver.driver, + &driver_attr_enable); + if (ret) + goto out_en; + + /* GF polynomial to use */ + ret = driver_create_file(&ppc440spe_adma_driver.driver, + &driver_attr_poly); + if (!ret) + return ret; + + driver_remove_file(&ppc440spe_adma_driver.driver, + &driver_attr_enable); +out_en: + driver_remove_file(&ppc440spe_adma_driver.driver, + &driver_attr_devices); +out_dev: + /* User will not be able to enable h/w RAID-6 */ + pr_err("%s: failed to create RAID-6 driver interface\n", + __func__); + of_unregister_platform_driver(&ppc440spe_adma_driver); +out_reg: + dcr_unmap(ppc440spe_mq_dcr_host, ppc440spe_mq_dcr_len); + kfree(ppc440spe_dma_fifo_buf); + return ret; +} + +static void __exit ppc440spe_adma_exit(void) +{ + driver_remove_file(&ppc440spe_adma_driver.driver, + &driver_attr_poly); + driver_remove_file(&ppc440spe_adma_driver.driver, + &driver_attr_enable); + driver_remove_file(&ppc440spe_adma_driver.driver, + &driver_attr_devices); + of_unregister_platform_driver(&ppc440spe_adma_driver); + dcr_unmap(ppc440spe_mq_dcr_host, ppc440spe_mq_dcr_len); + kfree(ppc440spe_dma_fifo_buf); +} + +arch_initcall(ppc440spe_adma_init); +module_exit(ppc440spe_adma_exit); + +MODULE_AUTHOR("Yuri Tikhonov <yur@emcraft.com>"); +MODULE_DESCRIPTION("PPC440SPE ADMA Engine Driver"); +MODULE_LICENSE("GPL"); diff --git a/drivers/dma/ppc4xx/adma.h b/drivers/dma/ppc4xx/adma.h new file mode 100644 index 00000000000..8ada5a812e3 --- /dev/null +++ b/drivers/dma/ppc4xx/adma.h @@ -0,0 +1,195 @@ +/* + * 2006-2009 (C) DENX Software Engineering. + * + * Author: Yuri Tikhonov <yur@emcraft.com> + * + * This file is licensed under the terms of the GNU General Public License + * version 2. This program is licensed "as is" without any warranty of + * any kind, whether express or implied. + */ + +#ifndef _PPC440SPE_ADMA_H +#define _PPC440SPE_ADMA_H + +#include <linux/types.h> +#include "dma.h" +#include "xor.h" + +#define to_ppc440spe_adma_chan(chan) \ + container_of(chan, struct ppc440spe_adma_chan, common) +#define to_ppc440spe_adma_device(dev) \ + container_of(dev, struct ppc440spe_adma_device, common) +#define tx_to_ppc440spe_adma_slot(tx) \ + container_of(tx, struct ppc440spe_adma_desc_slot, async_tx) + +/* Default polynomial (for 440SP is only available) */ +#define PPC440SPE_DEFAULT_POLY 0x4d + +#define PPC440SPE_ADMA_ENGINES_NUM (XOR_ENGINES_NUM + DMA_ENGINES_NUM) + +#define PPC440SPE_ADMA_WATCHDOG_MSEC 3 +#define PPC440SPE_ADMA_THRESHOLD 1 + +#define PPC440SPE_DMA0_ID 0 +#define PPC440SPE_DMA1_ID 1 +#define PPC440SPE_XOR_ID 2 + +#define PPC440SPE_ADMA_DMA_MAX_BYTE_COUNT 0xFFFFFFUL +/* this is the XOR_CBBCR width */ +#define PPC440SPE_ADMA_XOR_MAX_BYTE_COUNT (1 << 31) +#define PPC440SPE_ADMA_ZERO_SUM_MAX_BYTE_COUNT PPC440SPE_ADMA_XOR_MAX_BYTE_COUNT + +#define PPC440SPE_RXOR_RUN 0 + +#define MQ0_CF2H_RXOR_BS_MASK 0x1FF + +#undef ADMA_LL_DEBUG + +/** + * struct ppc440spe_adma_device - internal representation of an ADMA device + * @dev: device + * @dma_reg: base for DMAx register access + * @xor_reg: base for XOR register access + * @i2o_reg: base for I2O register access + * @id: HW ADMA Device selector + * @dma_desc_pool_virt: base of DMA descriptor region (CPU address) + * @dma_desc_pool: base of DMA descriptor region (DMA address) + * @pool_size: size of the pool + * @irq: DMAx or XOR irq number + * @err_irq: DMAx error irq number + * @common: embedded struct dma_device + */ +struct ppc440spe_adma_device { + struct device *dev; + struct dma_regs __iomem *dma_reg; + struct xor_regs __iomem *xor_reg; + struct i2o_regs __iomem *i2o_reg; + int id; + void *dma_desc_pool_virt; + dma_addr_t dma_desc_pool; + size_t pool_size; + int irq; + int err_irq; + struct dma_device common; +}; + +/** + * struct ppc440spe_adma_chan - internal representation of an ADMA channel + * @lock: serializes enqueue/dequeue operations to the slot pool + * @device: parent device + * @chain: device chain view of the descriptors + * @common: common dmaengine channel object members + * @all_slots: complete domain of slots usable by the channel + * @pending: allows batching of hardware operations + * @completed_cookie: identifier for the most recently completed operation + * @slots_allocated: records the actual size of the descriptor slot pool + * @hw_chain_inited: h/w descriptor chain initialization flag + * @irq_tasklet: bottom half where ppc440spe_adma_slot_cleanup runs + * @needs_unmap: if buffers should not be unmapped upon final processing + * @pdest_page: P destination page for async validate operation + * @qdest_page: Q destination page for async validate operation + * @pdest: P dma addr for async validate operation + * @qdest: Q dma addr for async validate operation + */ +struct ppc440spe_adma_chan { + spinlock_t lock; + struct ppc440spe_adma_device *device; + struct list_head chain; + struct dma_chan common; + struct list_head all_slots; + struct ppc440spe_adma_desc_slot *last_used; + int pending; + dma_cookie_t completed_cookie; + int slots_allocated; + int hw_chain_inited; + struct tasklet_struct irq_tasklet; + u8 needs_unmap; + struct page *pdest_page; + struct page *qdest_page; + dma_addr_t pdest; + dma_addr_t qdest; +}; + +struct ppc440spe_rxor { + u32 addrl; + u32 addrh; + int len; + int xor_count; + int addr_count; + int desc_count; + int state; +}; + +/** + * struct ppc440spe_adma_desc_slot - PPC440SPE-ADMA software descriptor + * @phys: hardware address of the hardware descriptor chain + * @group_head: first operation in a transaction + * @hw_next: pointer to the next descriptor in chain + * @async_tx: support for the async_tx api + * @slot_node: node on the iop_adma_chan.all_slots list + * @chain_node: node on the op_adma_chan.chain list + * @group_list: list of slots that make up a multi-descriptor transaction + * for example transfer lengths larger than the supported hw max + * @unmap_len: transaction bytecount + * @hw_desc: virtual address of the hardware descriptor chain + * @stride: currently chained or not + * @idx: pool index + * @slot_cnt: total slots used in an transaction (group of operations) + * @src_cnt: number of sources set in this descriptor + * @dst_cnt: number of destinations set in the descriptor + * @slots_per_op: number of slots per operation + * @descs_per_op: number of slot per P/Q operation see comment + * for ppc440spe_prep_dma_pqxor function + * @flags: desc state/type + * @reverse_flags: 1 if a corresponding rxor address uses reversed address order + * @xor_check_result: result of zero sum + * @crc32_result: result crc calculation + */ +struct ppc440spe_adma_desc_slot { + dma_addr_t phys; + struct ppc440spe_adma_desc_slot *group_head; + struct ppc440spe_adma_desc_slot *hw_next; + struct dma_async_tx_descriptor async_tx; + struct list_head slot_node; + struct list_head chain_node; /* node in channel ops list */ + struct list_head group_list; /* list */ + unsigned int unmap_len; + void *hw_desc; + u16 stride; + u16 idx; + u16 slot_cnt; + u8 src_cnt; + u8 dst_cnt; + u8 slots_per_op; + u8 descs_per_op; + unsigned long flags; + unsigned long reverse_flags[8]; + +#define PPC440SPE_DESC_INT 0 /* generate interrupt on complete */ +#define PPC440SPE_ZERO_P 1 /* clear P destionaion */ +#define PPC440SPE_ZERO_Q 2 /* clear Q destination */ +#define PPC440SPE_COHERENT 3 /* src/dst are coherent */ + +#define PPC440SPE_DESC_WXOR 4 /* WXORs are in chain */ +#define PPC440SPE_DESC_RXOR 5 /* RXOR is in chain */ + +#define PPC440SPE_DESC_RXOR123 8 /* CDB for RXOR123 operation */ +#define PPC440SPE_DESC_RXOR124 9 /* CDB for RXOR124 operation */ +#define PPC440SPE_DESC_RXOR125 10 /* CDB for RXOR125 operation */ +#define PPC440SPE_DESC_RXOR12 11 /* CDB for RXOR12 operation */ +#define PPC440SPE_DESC_RXOR_REV 12 /* CDB has srcs in reversed order */ + +#define PPC440SPE_DESC_PCHECK 13 +#define PPC440SPE_DESC_QCHECK 14 + +#define PPC440SPE_DESC_RXOR_MSK 0x3 + + struct ppc440spe_rxor rxor_cursor; + + union { + u32 *xor_check_result; + u32 *crc32_result; + }; +}; + +#endif /* _PPC440SPE_ADMA_H */ diff --git a/drivers/dma/ppc4xx/dma.h b/drivers/dma/ppc4xx/dma.h new file mode 100644 index 00000000000..bcde2df2f37 --- /dev/null +++ b/drivers/dma/ppc4xx/dma.h @@ -0,0 +1,223 @@ +/* + * 440SPe's DMA engines support header file + * + * 2006-2009 (C) DENX Software Engineering. + * + * Author: Yuri Tikhonov <yur@emcraft.com> + * + * This file is licensed under the term of the GNU General Public License + * version 2. The program licensed "as is" without any warranty of any + * kind, whether express or implied. + */ + +#ifndef _PPC440SPE_DMA_H +#define _PPC440SPE_DMA_H + +#include <linux/types.h> + +/* Number of elements in the array with statical CDBs */ +#define MAX_STAT_DMA_CDBS 16 +/* Number of DMA engines available on the contoller */ +#define DMA_ENGINES_NUM 2 + +/* Maximum h/w supported number of destinations */ +#define DMA_DEST_MAX_NUM 2 + +/* FIFO's params */ +#define DMA0_FIFO_SIZE 0x1000 +#define DMA1_FIFO_SIZE 0x1000 +#define DMA_FIFO_ENABLE (1<<12) + +/* DMA Configuration Register. Data Transfer Engine PLB Priority: */ +#define DMA_CFG_DXEPR_LP (0<<26) +#define DMA_CFG_DXEPR_HP (3<<26) +#define DMA_CFG_DXEPR_HHP (2<<26) +#define DMA_CFG_DXEPR_HHHP (1<<26) + +/* DMA Configuration Register. DMA FIFO Manager PLB Priority: */ +#define DMA_CFG_DFMPP_LP (0<<23) +#define DMA_CFG_DFMPP_HP (3<<23) +#define DMA_CFG_DFMPP_HHP (2<<23) +#define DMA_CFG_DFMPP_HHHP (1<<23) + +/* DMA Configuration Register. Force 64-byte Alignment */ +#define DMA_CFG_FALGN (1 << 19) + +/*UIC0:*/ +#define D0CPF_INT (1<<12) +#define D0CSF_INT (1<<11) +#define D1CPF_INT (1<<10) +#define D1CSF_INT (1<<9) +/*UIC1:*/ +#define DMAE_INT (1<<9) + +/* I2O IOP Interrupt Mask Register */ +#define I2O_IOPIM_P0SNE (1<<3) +#define I2O_IOPIM_P0EM (1<<5) +#define I2O_IOPIM_P1SNE (1<<6) +#define I2O_IOPIM_P1EM (1<<8) + +/* DMA CDB fields */ +#define DMA_CDB_MSK (0xF) +#define DMA_CDB_64B_ADDR (1<<2) +#define DMA_CDB_NO_INT (1<<3) +#define DMA_CDB_STATUS_MSK (0x3) +#define DMA_CDB_ADDR_MSK (0xFFFFFFF0) + +/* DMA CDB OpCodes */ +#define DMA_CDB_OPC_NO_OP (0x00) +#define DMA_CDB_OPC_MV_SG1_SG2 (0x01) +#define DMA_CDB_OPC_MULTICAST (0x05) +#define DMA_CDB_OPC_DFILL128 (0x24) +#define DMA_CDB_OPC_DCHECK128 (0x23) + +#define DMA_CUED_XOR_BASE (0x10000000) +#define DMA_CUED_XOR_HB (0x00000008) + +#ifdef CONFIG_440SP +#define DMA_CUED_MULT1_OFF 0 +#define DMA_CUED_MULT2_OFF 8 +#define DMA_CUED_MULT3_OFF 16 +#define DMA_CUED_REGION_OFF 24 +#define DMA_CUED_XOR_WIN_MSK (0xFC000000) +#else +#define DMA_CUED_MULT1_OFF 2 +#define DMA_CUED_MULT2_OFF 10 +#define DMA_CUED_MULT3_OFF 18 +#define DMA_CUED_REGION_OFF 26 +#define DMA_CUED_XOR_WIN_MSK (0xF0000000) +#endif + +#define DMA_CUED_REGION_MSK 0x3 +#define DMA_RXOR123 0x0 +#define DMA_RXOR124 0x1 +#define DMA_RXOR125 0x2 +#define DMA_RXOR12 0x3 + +/* S/G addresses */ +#define DMA_CDB_SG_SRC 1 +#define DMA_CDB_SG_DST1 2 +#define DMA_CDB_SG_DST2 3 + +/* + * DMAx engines Command Descriptor Block Type + */ +struct dma_cdb { + /* + * Basic CDB structure (Table 20-17, p.499, 440spe_um_1_22.pdf) + */ + u8 pad0[2]; /* reserved */ + u8 attr; /* attributes */ + u8 opc; /* opcode */ + u32 sg1u; /* upper SG1 address */ + u32 sg1l; /* lower SG1 address */ + u32 cnt; /* SG count, 3B used */ + u32 sg2u; /* upper SG2 address */ + u32 sg2l; /* lower SG2 address */ + u32 sg3u; /* upper SG3 address */ + u32 sg3l; /* lower SG3 address */ +}; + +/* + * DMAx hardware registers (p.515 in 440SPe UM 1.22) + */ +struct dma_regs { + u32 cpfpl; + u32 cpfph; + u32 csfpl; + u32 csfph; + u32 dsts; + u32 cfg; + u8 pad0[0x8]; + u16 cpfhp; + u16 cpftp; + u16 csfhp; + u16 csftp; + u8 pad1[0x8]; + u32 acpl; + u32 acph; + u32 s1bpl; + u32 s1bph; + u32 s2bpl; + u32 s2bph; + u32 s3bpl; + u32 s3bph; + u8 pad2[0x10]; + u32 earl; + u32 earh; + u8 pad3[0x8]; + u32 seat; + u32 sead; + u32 op; + u32 fsiz; +}; + +/* + * I2O hardware registers (p.528 in 440SPe UM 1.22) + */ +struct i2o_regs { + u32 ists; + u32 iseat; + u32 isead; + u8 pad0[0x14]; + u32 idbel; + u8 pad1[0xc]; + u32 ihis; + u32 ihim; + u8 pad2[0x8]; + u32 ihiq; + u32 ihoq; + u8 pad3[0x8]; + u32 iopis; + u32 iopim; + u32 iopiq; + u8 iopoq; + u8 pad4[3]; + u16 iiflh; + u16 iiflt; + u16 iiplh; + u16 iiplt; + u16 ioflh; + u16 ioflt; + u16 ioplh; + u16 ioplt; + u32 iidc; + u32 ictl; + u32 ifcpp; + u8 pad5[0x4]; + u16 mfac0; + u16 mfac1; + u16 mfac2; + u16 mfac3; + u16 mfac4; + u16 mfac5; + u16 mfac6; + u16 mfac7; + u16 ifcfh; + u16 ifcht; + u8 pad6[0x4]; + u32 iifmc; + u32 iodb; + u32 iodbc; + u32 ifbal; + u32 ifbah; + u32 ifsiz; + u32 ispd0; + u32 ispd1; + u32 ispd2; + u32 ispd3; + u32 ihipl; + u32 ihiph; + u32 ihopl; + u32 ihoph; + u32 iiipl; + u32 iiiph; + u32 iiopl; + u32 iioph; + u32 ifcpl; + u32 ifcph; + u8 pad7[0x8]; + u32 iopt; +}; + +#endif /* _PPC440SPE_DMA_H */ diff --git a/drivers/dma/ppc4xx/xor.h b/drivers/dma/ppc4xx/xor.h new file mode 100644 index 00000000000..daed7384daa --- /dev/null +++ b/drivers/dma/ppc4xx/xor.h @@ -0,0 +1,110 @@ +/* + * 440SPe's XOR engines support header file + * + * 2006-2009 (C) DENX Software Engineering. + * + * Author: Yuri Tikhonov <yur@emcraft.com> + * + * This file is licensed under the term of the GNU General Public License + * version 2. The program licensed "as is" without any warranty of any + * kind, whether express or implied. + */ + +#ifndef _PPC440SPE_XOR_H +#define _PPC440SPE_XOR_H + +#include <linux/types.h> + +/* Number of XOR engines available on the contoller */ +#define XOR_ENGINES_NUM 1 + +/* Number of operands supported in the h/w */ +#define XOR_MAX_OPS 16 + +/* + * XOR Command Block Control Register bits + */ +#define XOR_CBCR_LNK_BIT (1<<31) /* link present */ +#define XOR_CBCR_TGT_BIT (1<<30) /* target present */ +#define XOR_CBCR_CBCE_BIT (1<<29) /* command block compete enable */ +#define XOR_CBCR_RNZE_BIT (1<<28) /* result not zero enable */ +#define XOR_CBCR_XNOR_BIT (1<<15) /* XOR/XNOR */ +#define XOR_CDCR_OAC_MSK (0x7F) /* operand address count */ + +/* + * XORCore Status Register bits + */ +#define XOR_SR_XCP_BIT (1<<31) /* core processing */ +#define XOR_SR_ICB_BIT (1<<17) /* invalid CB */ +#define XOR_SR_IC_BIT (1<<16) /* invalid command */ +#define XOR_SR_IPE_BIT (1<<15) /* internal parity error */ +#define XOR_SR_RNZ_BIT (1<<2) /* result not Zero */ +#define XOR_SR_CBC_BIT (1<<1) /* CB complete */ +#define XOR_SR_CBLC_BIT (1<<0) /* CB list complete */ + +/* + * XORCore Control Set and Reset Register bits + */ +#define XOR_CRSR_XASR_BIT (1<<31) /* soft reset */ +#define XOR_CRSR_XAE_BIT (1<<30) /* enable */ +#define XOR_CRSR_RCBE_BIT (1<<29) /* refetch CB enable */ +#define XOR_CRSR_PAUS_BIT (1<<28) /* pause */ +#define XOR_CRSR_64BA_BIT (1<<27) /* 64/32 CB format */ +#define XOR_CRSR_CLP_BIT (1<<25) /* continue list processing */ + +/* + * XORCore Interrupt Enable Register + */ +#define XOR_IE_ICBIE_BIT (1<<17) /* Invalid Command Block IRQ Enable */ +#define XOR_IE_ICIE_BIT (1<<16) /* Invalid Command IRQ Enable */ +#define XOR_IE_RPTIE_BIT (1<<14) /* Read PLB Timeout Error IRQ Enable */ +#define XOR_IE_CBCIE_BIT (1<<1) /* CB complete interrupt enable */ +#define XOR_IE_CBLCI_BIT (1<<0) /* CB list complete interrupt enable */ + +/* + * XOR Accelerator engine Command Block Type + */ +struct xor_cb { + /* + * Basic 64-bit format XOR CB (Table 19-1, p.463, 440spe_um_1_22.pdf) + */ + u32 cbc; /* control */ + u32 cbbc; /* byte count */ + u32 cbs; /* status */ + u8 pad0[4]; /* reserved */ + u32 cbtah; /* target address high */ + u32 cbtal; /* target address low */ + u32 cblah; /* link address high */ + u32 cblal; /* link address low */ + struct { + u32 h; + u32 l; + } __attribute__ ((packed)) ops[16]; +} __attribute__ ((packed)); + +/* + * XOR hardware registers Table 19-3, UM 1.22 + */ +struct xor_regs { + u32 op_ar[16][2]; /* operand address[0]-high,[1]-low registers */ + u8 pad0[352]; /* reserved */ + u32 cbcr; /* CB control register */ + u32 cbbcr; /* CB byte count register */ + u32 cbsr; /* CB status register */ + u8 pad1[4]; /* reserved */ + u32 cbtahr; /* operand target address high register */ + u32 cbtalr; /* operand target address low register */ + u32 cblahr; /* CB link address high register */ + u32 cblalr; /* CB link address low register */ + u32 crsr; /* control set register */ + u32 crrr; /* control reset register */ + u32 ccbahr; /* current CB address high register */ + u32 ccbalr; /* current CB address low register */ + u32 plbr; /* PLB configuration register */ + u32 ier; /* interrupt enable register */ + u32 pecr; /* parity error count register */ + u32 sr; /* status register */ + u32 revidr; /* revision ID register */ +}; + +#endif /* _PPC440SPE_XOR_H */ diff --git a/drivers/dma/shdma.c b/drivers/dma/shdma.c new file mode 100644 index 00000000000..d10cc899c46 --- /dev/null +++ b/drivers/dma/shdma.c @@ -0,0 +1,876 @@ +/* + * Renesas SuperH DMA Engine support + * + * base is drivers/dma/flsdma.c + * + * Copyright (C) 2009 Nobuhiro Iwamatsu <iwamatsu.nobuhiro@renesas.com> + * Copyright (C) 2009 Renesas Solutions, Inc. All rights reserved. + * Copyright (C) 2007 Freescale Semiconductor, Inc. All rights reserved. + * + * This is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * - DMA of SuperH does not have Hardware DMA chain mode. + * - MAX DMA size is 16MB. + * + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/interrupt.h> +#include <linux/dmaengine.h> +#include <linux/delay.h> +#include <linux/dma-mapping.h> +#include <linux/platform_device.h> +#include <cpu/dma.h> +#include <asm/dma-sh.h> +#include "shdma.h" + +/* DMA descriptor control */ +enum sh_dmae_desc_status { + DESC_IDLE, + DESC_PREPARED, + DESC_SUBMITTED, + DESC_COMPLETED, /* completed, have to call callback */ + DESC_WAITING, /* callback called, waiting for ack / re-submit */ +}; + +#define NR_DESCS_PER_CHANNEL 32 +/* + * Define the default configuration for dual address memory-memory transfer. + * The 0x400 value represents auto-request, external->external. + * + * And this driver set 4byte burst mode. + * If you want to change mode, you need to change RS_DEFAULT of value. + * (ex 1byte burst mode -> (RS_DUAL & ~TS_32) + */ +#define RS_DEFAULT (RS_DUAL) + +static void sh_dmae_chan_ld_cleanup(struct sh_dmae_chan *sh_chan, bool all); + +#define SH_DMAC_CHAN_BASE(id) (dma_base_addr[id]) +static void sh_dmae_writel(struct sh_dmae_chan *sh_dc, u32 data, u32 reg) +{ + ctrl_outl(data, (SH_DMAC_CHAN_BASE(sh_dc->id) + reg)); +} + +static u32 sh_dmae_readl(struct sh_dmae_chan *sh_dc, u32 reg) +{ + return ctrl_inl((SH_DMAC_CHAN_BASE(sh_dc->id) + reg)); +} + +static void dmae_init(struct sh_dmae_chan *sh_chan) +{ + u32 chcr = RS_DEFAULT; /* default is DUAL mode */ + sh_dmae_writel(sh_chan, chcr, CHCR); +} + +/* + * Reset DMA controller + * + * SH7780 has two DMAOR register + */ +static void sh_dmae_ctl_stop(int id) +{ + unsigned short dmaor = dmaor_read_reg(id); + + dmaor &= ~(DMAOR_NMIF | DMAOR_AE); + dmaor_write_reg(id, dmaor); +} + +static int sh_dmae_rst(int id) +{ + unsigned short dmaor; + + sh_dmae_ctl_stop(id); + dmaor = dmaor_read_reg(id) | DMAOR_INIT; + + dmaor_write_reg(id, dmaor); + if (dmaor_read_reg(id) & (DMAOR_AE | DMAOR_NMIF)) { + pr_warning(KERN_ERR "dma-sh: Can't initialize DMAOR.\n"); + return -EINVAL; + } + return 0; +} + +static int dmae_is_busy(struct sh_dmae_chan *sh_chan) +{ + u32 chcr = sh_dmae_readl(sh_chan, CHCR); + if (chcr & CHCR_DE) { + if (!(chcr & CHCR_TE)) + return -EBUSY; /* working */ + } + return 0; /* waiting */ +} + +static inline unsigned int calc_xmit_shift(struct sh_dmae_chan *sh_chan) +{ + u32 chcr = sh_dmae_readl(sh_chan, CHCR); + return ts_shift[(chcr & CHCR_TS_MASK) >> CHCR_TS_SHIFT]; +} + +static void dmae_set_reg(struct sh_dmae_chan *sh_chan, struct sh_dmae_regs *hw) +{ + sh_dmae_writel(sh_chan, hw->sar, SAR); + sh_dmae_writel(sh_chan, hw->dar, DAR); + sh_dmae_writel(sh_chan, hw->tcr >> calc_xmit_shift(sh_chan), TCR); +} + +static void dmae_start(struct sh_dmae_chan *sh_chan) +{ + u32 chcr = sh_dmae_readl(sh_chan, CHCR); + + chcr |= CHCR_DE | CHCR_IE; + sh_dmae_writel(sh_chan, chcr, CHCR); +} + +static void dmae_halt(struct sh_dmae_chan *sh_chan) +{ + u32 chcr = sh_dmae_readl(sh_chan, CHCR); + + chcr &= ~(CHCR_DE | CHCR_TE | CHCR_IE); + sh_dmae_writel(sh_chan, chcr, CHCR); +} + +static int dmae_set_chcr(struct sh_dmae_chan *sh_chan, u32 val) +{ + int ret = dmae_is_busy(sh_chan); + /* When DMA was working, can not set data to CHCR */ + if (ret) + return ret; + + sh_dmae_writel(sh_chan, val, CHCR); + return 0; +} + +#define DMARS1_ADDR 0x04 +#define DMARS2_ADDR 0x08 +#define DMARS_SHIFT 8 +#define DMARS_CHAN_MSK 0x01 +static int dmae_set_dmars(struct sh_dmae_chan *sh_chan, u16 val) +{ + u32 addr; + int shift = 0; + int ret = dmae_is_busy(sh_chan); + if (ret) + return ret; + + if (sh_chan->id & DMARS_CHAN_MSK) + shift = DMARS_SHIFT; + + switch (sh_chan->id) { + /* DMARS0 */ + case 0: + case 1: + addr = SH_DMARS_BASE; + break; + /* DMARS1 */ + case 2: + case 3: + addr = (SH_DMARS_BASE + DMARS1_ADDR); + break; + /* DMARS2 */ + case 4: + case 5: + addr = (SH_DMARS_BASE + DMARS2_ADDR); + break; + default: + return -EINVAL; + } + + ctrl_outw((val << shift) | + (ctrl_inw(addr) & (shift ? 0xFF00 : 0x00FF)), + addr); + + return 0; +} + +static dma_cookie_t sh_dmae_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct sh_desc *desc = tx_to_sh_desc(tx), *chunk, *last = desc, *c; + struct sh_dmae_chan *sh_chan = to_sh_chan(tx->chan); + dma_async_tx_callback callback = tx->callback; + dma_cookie_t cookie; + + spin_lock_bh(&sh_chan->desc_lock); + + cookie = sh_chan->common.cookie; + cookie++; + if (cookie < 0) + cookie = 1; + + sh_chan->common.cookie = cookie; + tx->cookie = cookie; + + /* Mark all chunks of this descriptor as submitted, move to the queue */ + list_for_each_entry_safe(chunk, c, desc->node.prev, node) { + /* + * All chunks are on the global ld_free, so, we have to find + * the end of the chain ourselves + */ + if (chunk != desc && (chunk->mark == DESC_IDLE || + chunk->async_tx.cookie > 0 || + chunk->async_tx.cookie == -EBUSY || + &chunk->node == &sh_chan->ld_free)) + break; + chunk->mark = DESC_SUBMITTED; + /* Callback goes to the last chunk */ + chunk->async_tx.callback = NULL; + chunk->cookie = cookie; + list_move_tail(&chunk->node, &sh_chan->ld_queue); + last = chunk; + } + + last->async_tx.callback = callback; + last->async_tx.callback_param = tx->callback_param; + + dev_dbg(sh_chan->dev, "submit #%d@%p on %d: %x[%d] -> %x\n", + tx->cookie, &last->async_tx, sh_chan->id, + desc->hw.sar, desc->hw.tcr, desc->hw.dar); + + spin_unlock_bh(&sh_chan->desc_lock); + + return cookie; +} + +/* Called with desc_lock held */ +static struct sh_desc *sh_dmae_get_desc(struct sh_dmae_chan *sh_chan) +{ + struct sh_desc *desc; + + list_for_each_entry(desc, &sh_chan->ld_free, node) + if (desc->mark != DESC_PREPARED) { + BUG_ON(desc->mark != DESC_IDLE); + list_del(&desc->node); + return desc; + } + + return NULL; +} + +static int sh_dmae_alloc_chan_resources(struct dma_chan *chan) +{ + struct sh_dmae_chan *sh_chan = to_sh_chan(chan); + struct sh_desc *desc; + + spin_lock_bh(&sh_chan->desc_lock); + while (sh_chan->descs_allocated < NR_DESCS_PER_CHANNEL) { + spin_unlock_bh(&sh_chan->desc_lock); + desc = kzalloc(sizeof(struct sh_desc), GFP_KERNEL); + if (!desc) { + spin_lock_bh(&sh_chan->desc_lock); + break; + } + dma_async_tx_descriptor_init(&desc->async_tx, + &sh_chan->common); + desc->async_tx.tx_submit = sh_dmae_tx_submit; + desc->mark = DESC_IDLE; + + spin_lock_bh(&sh_chan->desc_lock); + list_add(&desc->node, &sh_chan->ld_free); + sh_chan->descs_allocated++; + } + spin_unlock_bh(&sh_chan->desc_lock); + + return sh_chan->descs_allocated; +} + +/* + * sh_dma_free_chan_resources - Free all resources of the channel. + */ +static void sh_dmae_free_chan_resources(struct dma_chan *chan) +{ + struct sh_dmae_chan *sh_chan = to_sh_chan(chan); + struct sh_desc *desc, *_desc; + LIST_HEAD(list); + + /* Prepared and not submitted descriptors can still be on the queue */ + if (!list_empty(&sh_chan->ld_queue)) + sh_dmae_chan_ld_cleanup(sh_chan, true); + + spin_lock_bh(&sh_chan->desc_lock); + + list_splice_init(&sh_chan->ld_free, &list); + sh_chan->descs_allocated = 0; + + spin_unlock_bh(&sh_chan->desc_lock); + + list_for_each_entry_safe(desc, _desc, &list, node) + kfree(desc); +} + +static struct dma_async_tx_descriptor *sh_dmae_prep_memcpy( + struct dma_chan *chan, dma_addr_t dma_dest, dma_addr_t dma_src, + size_t len, unsigned long flags) +{ + struct sh_dmae_chan *sh_chan; + struct sh_desc *first = NULL, *prev = NULL, *new; + size_t copy_size; + LIST_HEAD(tx_list); + int chunks = (len + SH_DMA_TCR_MAX) / (SH_DMA_TCR_MAX + 1); + + if (!chan) + return NULL; + + if (!len) + return NULL; + + sh_chan = to_sh_chan(chan); + + /* Have to lock the whole loop to protect against concurrent release */ + spin_lock_bh(&sh_chan->desc_lock); + + /* + * Chaining: + * first descriptor is what user is dealing with in all API calls, its + * cookie is at first set to -EBUSY, at tx-submit to a positive + * number + * if more than one chunk is needed further chunks have cookie = -EINVAL + * the last chunk, if not equal to the first, has cookie = -ENOSPC + * all chunks are linked onto the tx_list head with their .node heads + * only during this function, then they are immediately spliced + * back onto the free list in form of a chain + */ + do { + /* Allocate the link descriptor from the free list */ + new = sh_dmae_get_desc(sh_chan); + if (!new) { + dev_err(sh_chan->dev, + "No free memory for link descriptor\n"); + list_for_each_entry(new, &tx_list, node) + new->mark = DESC_IDLE; + list_splice(&tx_list, &sh_chan->ld_free); + spin_unlock_bh(&sh_chan->desc_lock); + return NULL; + } + + copy_size = min(len, (size_t)SH_DMA_TCR_MAX + 1); + + new->hw.sar = dma_src; + new->hw.dar = dma_dest; + new->hw.tcr = copy_size; + if (!first) { + /* First desc */ + new->async_tx.cookie = -EBUSY; + first = new; + } else { + /* Other desc - invisible to the user */ + new->async_tx.cookie = -EINVAL; + } + + dev_dbg(sh_chan->dev, + "chaining %u of %u with %p, dst %x, cookie %d\n", + copy_size, len, &new->async_tx, dma_dest, + new->async_tx.cookie); + + new->mark = DESC_PREPARED; + new->async_tx.flags = flags; + new->chunks = chunks--; + + prev = new; + len -= copy_size; + dma_src += copy_size; + dma_dest += copy_size; + /* Insert the link descriptor to the LD ring */ + list_add_tail(&new->node, &tx_list); + } while (len); + + if (new != first) + new->async_tx.cookie = -ENOSPC; + + /* Put them back on the free list, so, they don't get lost */ + list_splice_tail(&tx_list, &sh_chan->ld_free); + + spin_unlock_bh(&sh_chan->desc_lock); + + return &first->async_tx; +} + +static dma_async_tx_callback __ld_cleanup(struct sh_dmae_chan *sh_chan, bool all) +{ + struct sh_desc *desc, *_desc; + /* Is the "exposed" head of a chain acked? */ + bool head_acked = false; + dma_cookie_t cookie = 0; + dma_async_tx_callback callback = NULL; + void *param = NULL; + + spin_lock_bh(&sh_chan->desc_lock); + list_for_each_entry_safe(desc, _desc, &sh_chan->ld_queue, node) { + struct dma_async_tx_descriptor *tx = &desc->async_tx; + + BUG_ON(tx->cookie > 0 && tx->cookie != desc->cookie); + BUG_ON(desc->mark != DESC_SUBMITTED && + desc->mark != DESC_COMPLETED && + desc->mark != DESC_WAITING); + + /* + * queue is ordered, and we use this loop to (1) clean up all + * completed descriptors, and to (2) update descriptor flags of + * any chunks in a (partially) completed chain + */ + if (!all && desc->mark == DESC_SUBMITTED && + desc->cookie != cookie) + break; + + if (tx->cookie > 0) + cookie = tx->cookie; + + if (desc->mark == DESC_COMPLETED && desc->chunks == 1) { + BUG_ON(sh_chan->completed_cookie != desc->cookie - 1); + sh_chan->completed_cookie = desc->cookie; + } + + /* Call callback on the last chunk */ + if (desc->mark == DESC_COMPLETED && tx->callback) { + desc->mark = DESC_WAITING; + callback = tx->callback; + param = tx->callback_param; + dev_dbg(sh_chan->dev, "descriptor #%d@%p on %d callback\n", + tx->cookie, tx, sh_chan->id); + BUG_ON(desc->chunks != 1); + break; + } + + if (tx->cookie > 0 || tx->cookie == -EBUSY) { + if (desc->mark == DESC_COMPLETED) { + BUG_ON(tx->cookie < 0); + desc->mark = DESC_WAITING; + } + head_acked = async_tx_test_ack(tx); + } else { + switch (desc->mark) { + case DESC_COMPLETED: + desc->mark = DESC_WAITING; + /* Fall through */ + case DESC_WAITING: + if (head_acked) + async_tx_ack(&desc->async_tx); + } + } + + dev_dbg(sh_chan->dev, "descriptor %p #%d completed.\n", + tx, tx->cookie); + + if (((desc->mark == DESC_COMPLETED || + desc->mark == DESC_WAITING) && + async_tx_test_ack(&desc->async_tx)) || all) { + /* Remove from ld_queue list */ + desc->mark = DESC_IDLE; + list_move(&desc->node, &sh_chan->ld_free); + } + } + spin_unlock_bh(&sh_chan->desc_lock); + + if (callback) + callback(param); + + return callback; +} + +/* + * sh_chan_ld_cleanup - Clean up link descriptors + * + * This function cleans up the ld_queue of DMA channel. + */ +static void sh_dmae_chan_ld_cleanup(struct sh_dmae_chan *sh_chan, bool all) +{ + while (__ld_cleanup(sh_chan, all)) + ; +} + +static void sh_chan_xfer_ld_queue(struct sh_dmae_chan *sh_chan) +{ + struct sh_desc *sd; + + spin_lock_bh(&sh_chan->desc_lock); + /* DMA work check */ + if (dmae_is_busy(sh_chan)) { + spin_unlock_bh(&sh_chan->desc_lock); + return; + } + + /* Find the first un-transfer desciptor */ + list_for_each_entry(sd, &sh_chan->ld_queue, node) + if (sd->mark == DESC_SUBMITTED) { + /* Get the ld start address from ld_queue */ + dmae_set_reg(sh_chan, &sd->hw); + dmae_start(sh_chan); + break; + } + + spin_unlock_bh(&sh_chan->desc_lock); +} + +static void sh_dmae_memcpy_issue_pending(struct dma_chan *chan) +{ + struct sh_dmae_chan *sh_chan = to_sh_chan(chan); + sh_chan_xfer_ld_queue(sh_chan); +} + +static enum dma_status sh_dmae_is_complete(struct dma_chan *chan, + dma_cookie_t cookie, + dma_cookie_t *done, + dma_cookie_t *used) +{ + struct sh_dmae_chan *sh_chan = to_sh_chan(chan); + dma_cookie_t last_used; + dma_cookie_t last_complete; + + sh_dmae_chan_ld_cleanup(sh_chan, false); + + last_used = chan->cookie; + last_complete = sh_chan->completed_cookie; + BUG_ON(last_complete < 0); + + if (done) + *done = last_complete; + + if (used) + *used = last_used; + + return dma_async_is_complete(cookie, last_complete, last_used); +} + +static irqreturn_t sh_dmae_interrupt(int irq, void *data) +{ + irqreturn_t ret = IRQ_NONE; + struct sh_dmae_chan *sh_chan = (struct sh_dmae_chan *)data; + u32 chcr = sh_dmae_readl(sh_chan, CHCR); + + if (chcr & CHCR_TE) { + /* DMA stop */ + dmae_halt(sh_chan); + + ret = IRQ_HANDLED; + tasklet_schedule(&sh_chan->tasklet); + } + + return ret; +} + +#if defined(CONFIG_CPU_SH4) +static irqreturn_t sh_dmae_err(int irq, void *data) +{ + int err = 0; + struct sh_dmae_device *shdev = (struct sh_dmae_device *)data; + + /* IRQ Multi */ + if (shdev->pdata.mode & SHDMA_MIX_IRQ) { + int cnt = 0; + switch (irq) { +#if defined(DMTE6_IRQ) && defined(DMAE1_IRQ) + case DMTE6_IRQ: + cnt++; +#endif + case DMTE0_IRQ: + if (dmaor_read_reg(cnt) & (DMAOR_NMIF | DMAOR_AE)) { + disable_irq(irq); + return IRQ_HANDLED; + } + default: + return IRQ_NONE; + } + } else { + /* reset dma controller */ + err = sh_dmae_rst(0); + if (err) + return err; +#ifdef SH_DMAC_BASE1 + if (shdev->pdata.mode & SHDMA_DMAOR1) { + err = sh_dmae_rst(1); + if (err) + return err; + } +#endif + disable_irq(irq); + return IRQ_HANDLED; + } +} +#endif + +static void dmae_do_tasklet(unsigned long data) +{ + struct sh_dmae_chan *sh_chan = (struct sh_dmae_chan *)data; + struct sh_desc *desc; + u32 sar_buf = sh_dmae_readl(sh_chan, SAR); + + spin_lock(&sh_chan->desc_lock); + list_for_each_entry(desc, &sh_chan->ld_queue, node) { + if ((desc->hw.sar + desc->hw.tcr) == sar_buf && + desc->mark == DESC_SUBMITTED) { + dev_dbg(sh_chan->dev, "done #%d@%p dst %u\n", + desc->async_tx.cookie, &desc->async_tx, + desc->hw.dar); + desc->mark = DESC_COMPLETED; + break; + } + } + spin_unlock(&sh_chan->desc_lock); + + /* Next desc */ + sh_chan_xfer_ld_queue(sh_chan); + sh_dmae_chan_ld_cleanup(sh_chan, false); +} + +static unsigned int get_dmae_irq(unsigned int id) +{ + unsigned int irq = 0; + if (id < ARRAY_SIZE(dmte_irq_map)) + irq = dmte_irq_map[id]; + return irq; +} + +static int __devinit sh_dmae_chan_probe(struct sh_dmae_device *shdev, int id) +{ + int err; + unsigned int irq = get_dmae_irq(id); + unsigned long irqflags = IRQF_DISABLED; + struct sh_dmae_chan *new_sh_chan; + + /* alloc channel */ + new_sh_chan = kzalloc(sizeof(struct sh_dmae_chan), GFP_KERNEL); + if (!new_sh_chan) { + dev_err(shdev->common.dev, + "No free memory for allocating dma channels!\n"); + return -ENOMEM; + } + + new_sh_chan->dev = shdev->common.dev; + new_sh_chan->id = id; + + /* Init DMA tasklet */ + tasklet_init(&new_sh_chan->tasklet, dmae_do_tasklet, + (unsigned long)new_sh_chan); + + /* Init the channel */ + dmae_init(new_sh_chan); + + spin_lock_init(&new_sh_chan->desc_lock); + + /* Init descripter manage list */ + INIT_LIST_HEAD(&new_sh_chan->ld_queue); + INIT_LIST_HEAD(&new_sh_chan->ld_free); + + /* copy struct dma_device */ + new_sh_chan->common.device = &shdev->common; + + /* Add the channel to DMA device channel list */ + list_add_tail(&new_sh_chan->common.device_node, + &shdev->common.channels); + shdev->common.chancnt++; + + if (shdev->pdata.mode & SHDMA_MIX_IRQ) { + irqflags = IRQF_SHARED; +#if defined(DMTE6_IRQ) + if (irq >= DMTE6_IRQ) + irq = DMTE6_IRQ; + else +#endif + irq = DMTE0_IRQ; + } + + snprintf(new_sh_chan->dev_id, sizeof(new_sh_chan->dev_id), + "sh-dmae%d", new_sh_chan->id); + + /* set up channel irq */ + err = request_irq(irq, &sh_dmae_interrupt, irqflags, + new_sh_chan->dev_id, new_sh_chan); + if (err) { + dev_err(shdev->common.dev, "DMA channel %d request_irq error " + "with return %d\n", id, err); + goto err_no_irq; + } + + /* CHCR register control function */ + new_sh_chan->set_chcr = dmae_set_chcr; + /* DMARS register control function */ + new_sh_chan->set_dmars = dmae_set_dmars; + + shdev->chan[id] = new_sh_chan; + return 0; + +err_no_irq: + /* remove from dmaengine device node */ + list_del(&new_sh_chan->common.device_node); + kfree(new_sh_chan); + return err; +} + +static void sh_dmae_chan_remove(struct sh_dmae_device *shdev) +{ + int i; + + for (i = shdev->common.chancnt - 1 ; i >= 0 ; i--) { + if (shdev->chan[i]) { + struct sh_dmae_chan *shchan = shdev->chan[i]; + if (!(shdev->pdata.mode & SHDMA_MIX_IRQ)) + free_irq(dmte_irq_map[i], shchan); + + list_del(&shchan->common.device_node); + kfree(shchan); + shdev->chan[i] = NULL; + } + } + shdev->common.chancnt = 0; +} + +static int __init sh_dmae_probe(struct platform_device *pdev) +{ + int err = 0, cnt, ecnt; + unsigned long irqflags = IRQF_DISABLED; +#if defined(CONFIG_CPU_SH4) + int eirq[] = { DMAE0_IRQ, +#if defined(DMAE1_IRQ) + DMAE1_IRQ +#endif + }; +#endif + struct sh_dmae_device *shdev; + + /* get platform data */ + if (!pdev->dev.platform_data) + return -ENODEV; + + shdev = kzalloc(sizeof(struct sh_dmae_device), GFP_KERNEL); + if (!shdev) { + dev_err(&pdev->dev, "No enough memory\n"); + return -ENOMEM; + } + + /* platform data */ + memcpy(&shdev->pdata, pdev->dev.platform_data, + sizeof(struct sh_dmae_pdata)); + + /* reset dma controller */ + err = sh_dmae_rst(0); + if (err) + goto rst_err; + + /* SH7780/85/23 has DMAOR1 */ + if (shdev->pdata.mode & SHDMA_DMAOR1) { + err = sh_dmae_rst(1); + if (err) + goto rst_err; + } + + INIT_LIST_HEAD(&shdev->common.channels); + + dma_cap_set(DMA_MEMCPY, shdev->common.cap_mask); + shdev->common.device_alloc_chan_resources + = sh_dmae_alloc_chan_resources; + shdev->common.device_free_chan_resources = sh_dmae_free_chan_resources; + shdev->common.device_prep_dma_memcpy = sh_dmae_prep_memcpy; + shdev->common.device_is_tx_complete = sh_dmae_is_complete; + shdev->common.device_issue_pending = sh_dmae_memcpy_issue_pending; + shdev->common.dev = &pdev->dev; + /* Default transfer size of 32 bytes requires 32-byte alignment */ + shdev->common.copy_align = 5; + +#if defined(CONFIG_CPU_SH4) + /* Non Mix IRQ mode SH7722/SH7730 etc... */ + if (shdev->pdata.mode & SHDMA_MIX_IRQ) { + irqflags = IRQF_SHARED; + eirq[0] = DMTE0_IRQ; +#if defined(DMTE6_IRQ) && defined(DMAE1_IRQ) + eirq[1] = DMTE6_IRQ; +#endif + } + + for (ecnt = 0 ; ecnt < ARRAY_SIZE(eirq); ecnt++) { + err = request_irq(eirq[ecnt], sh_dmae_err, irqflags, + "DMAC Address Error", shdev); + if (err) { + dev_err(&pdev->dev, "DMA device request_irq" + "error (irq %d) with return %d\n", + eirq[ecnt], err); + goto eirq_err; + } + } +#endif /* CONFIG_CPU_SH4 */ + + /* Create DMA Channel */ + for (cnt = 0 ; cnt < MAX_DMA_CHANNELS ; cnt++) { + err = sh_dmae_chan_probe(shdev, cnt); + if (err) + goto chan_probe_err; + } + + platform_set_drvdata(pdev, shdev); + dma_async_device_register(&shdev->common); + + return err; + +chan_probe_err: + sh_dmae_chan_remove(shdev); + +eirq_err: + for (ecnt-- ; ecnt >= 0; ecnt--) + free_irq(eirq[ecnt], shdev); + +rst_err: + kfree(shdev); + + return err; +} + +static int __exit sh_dmae_remove(struct platform_device *pdev) +{ + struct sh_dmae_device *shdev = platform_get_drvdata(pdev); + + dma_async_device_unregister(&shdev->common); + + if (shdev->pdata.mode & SHDMA_MIX_IRQ) { + free_irq(DMTE0_IRQ, shdev); +#if defined(DMTE6_IRQ) + free_irq(DMTE6_IRQ, shdev); +#endif + } + + /* channel data remove */ + sh_dmae_chan_remove(shdev); + + if (!(shdev->pdata.mode & SHDMA_MIX_IRQ)) { + free_irq(DMAE0_IRQ, shdev); +#if defined(DMAE1_IRQ) + free_irq(DMAE1_IRQ, shdev); +#endif + } + kfree(shdev); + + return 0; +} + +static void sh_dmae_shutdown(struct platform_device *pdev) +{ + struct sh_dmae_device *shdev = platform_get_drvdata(pdev); + sh_dmae_ctl_stop(0); + if (shdev->pdata.mode & SHDMA_DMAOR1) + sh_dmae_ctl_stop(1); +} + +static struct platform_driver sh_dmae_driver = { + .remove = __exit_p(sh_dmae_remove), + .shutdown = sh_dmae_shutdown, + .driver = { + .name = "sh-dma-engine", + }, +}; + +static int __init sh_dmae_init(void) +{ + return platform_driver_probe(&sh_dmae_driver, sh_dmae_probe); +} +module_init(sh_dmae_init); + +static void __exit sh_dmae_exit(void) +{ + platform_driver_unregister(&sh_dmae_driver); +} +module_exit(sh_dmae_exit); + +MODULE_AUTHOR("Nobuhiro Iwamatsu <iwamatsu.nobuhiro@renesas.com>"); +MODULE_DESCRIPTION("Renesas SH DMA Engine driver"); +MODULE_LICENSE("GPL"); diff --git a/drivers/dma/shdma.h b/drivers/dma/shdma.h new file mode 100644 index 00000000000..108f1cffb6f --- /dev/null +++ b/drivers/dma/shdma.h @@ -0,0 +1,67 @@ +/* + * Renesas SuperH DMA Engine support + * + * Copyright (C) 2009 Nobuhiro Iwamatsu <iwamatsu.nobuhiro@renesas.com> + * Copyright (C) 2009 Renesas Solutions, Inc. All rights reserved. + * + * This is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + */ +#ifndef __DMA_SHDMA_H +#define __DMA_SHDMA_H + +#include <linux/dmaengine.h> +#include <linux/interrupt.h> +#include <linux/list.h> + +#define SH_DMA_TCR_MAX 0x00FFFFFF /* 16MB */ + +struct sh_dmae_regs { + u32 sar; /* SAR / source address */ + u32 dar; /* DAR / destination address */ + u32 tcr; /* TCR / transfer count */ +}; + +struct sh_desc { + struct sh_dmae_regs hw; + struct list_head node; + struct dma_async_tx_descriptor async_tx; + dma_cookie_t cookie; + int chunks; + int mark; +}; + +struct device; + +struct sh_dmae_chan { + dma_cookie_t completed_cookie; /* The maximum cookie completed */ + spinlock_t desc_lock; /* Descriptor operation lock */ + struct list_head ld_queue; /* Link descriptors queue */ + struct list_head ld_free; /* Link descriptors free */ + struct dma_chan common; /* DMA common channel */ + struct device *dev; /* Channel device */ + struct tasklet_struct tasklet; /* Tasklet */ + int descs_allocated; /* desc count */ + int id; /* Raw id of this channel */ + char dev_id[16]; /* unique name per DMAC of channel */ + + /* Set chcr */ + int (*set_chcr)(struct sh_dmae_chan *sh_chan, u32 regs); + /* Set DMA resource */ + int (*set_dmars)(struct sh_dmae_chan *sh_chan, u16 res); +}; + +struct sh_dmae_device { + struct dma_device common; + struct sh_dmae_chan *chan[MAX_DMA_CHANNELS]; + struct sh_dmae_pdata pdata; +}; + +#define to_sh_chan(chan) container_of(chan, struct sh_dmae_chan, common) +#define to_sh_desc(lh) container_of(lh, struct sh_desc, node) +#define tx_to_sh_desc(tx) container_of(tx, struct sh_desc, async_tx) + +#endif /* __DMA_SHDMA_H */ diff --git a/drivers/dma/txx9dmac.c b/drivers/dma/txx9dmac.c new file mode 100644 index 00000000000..3ebc61067e5 --- /dev/null +++ b/drivers/dma/txx9dmac.c @@ -0,0 +1,1361 @@ +/* + * Driver for the TXx9 SoC DMA Controller + * + * Copyright (C) 2009 Atsushi Nemoto + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ +#include <linux/dma-mapping.h> +#include <linux/init.h> +#include <linux/interrupt.h> +#include <linux/io.h> +#include <linux/module.h> +#include <linux/platform_device.h> +#include <linux/slab.h> +#include <linux/scatterlist.h> +#include "txx9dmac.h" + +static struct txx9dmac_chan *to_txx9dmac_chan(struct dma_chan *chan) +{ + return container_of(chan, struct txx9dmac_chan, chan); +} + +static struct txx9dmac_cregs __iomem *__dma_regs(const struct txx9dmac_chan *dc) +{ + return dc->ch_regs; +} + +static struct txx9dmac_cregs32 __iomem *__dma_regs32( + const struct txx9dmac_chan *dc) +{ + return dc->ch_regs; +} + +#define channel64_readq(dc, name) \ + __raw_readq(&(__dma_regs(dc)->name)) +#define channel64_writeq(dc, name, val) \ + __raw_writeq((val), &(__dma_regs(dc)->name)) +#define channel64_readl(dc, name) \ + __raw_readl(&(__dma_regs(dc)->name)) +#define channel64_writel(dc, name, val) \ + __raw_writel((val), &(__dma_regs(dc)->name)) + +#define channel32_readl(dc, name) \ + __raw_readl(&(__dma_regs32(dc)->name)) +#define channel32_writel(dc, name, val) \ + __raw_writel((val), &(__dma_regs32(dc)->name)) + +#define channel_readq(dc, name) channel64_readq(dc, name) +#define channel_writeq(dc, name, val) channel64_writeq(dc, name, val) +#define channel_readl(dc, name) \ + (is_dmac64(dc) ? \ + channel64_readl(dc, name) : channel32_readl(dc, name)) +#define channel_writel(dc, name, val) \ + (is_dmac64(dc) ? \ + channel64_writel(dc, name, val) : channel32_writel(dc, name, val)) + +static dma_addr_t channel64_read_CHAR(const struct txx9dmac_chan *dc) +{ + if (sizeof(__dma_regs(dc)->CHAR) == sizeof(u64)) + return channel64_readq(dc, CHAR); + else + return channel64_readl(dc, CHAR); +} + +static void channel64_write_CHAR(const struct txx9dmac_chan *dc, dma_addr_t val) +{ + if (sizeof(__dma_regs(dc)->CHAR) == sizeof(u64)) + channel64_writeq(dc, CHAR, val); + else + channel64_writel(dc, CHAR, val); +} + +static void channel64_clear_CHAR(const struct txx9dmac_chan *dc) +{ +#if defined(CONFIG_32BIT) && !defined(CONFIG_64BIT_PHYS_ADDR) + channel64_writel(dc, CHAR, 0); + channel64_writel(dc, __pad_CHAR, 0); +#else + channel64_writeq(dc, CHAR, 0); +#endif +} + +static dma_addr_t channel_read_CHAR(const struct txx9dmac_chan *dc) +{ + if (is_dmac64(dc)) + return channel64_read_CHAR(dc); + else + return channel32_readl(dc, CHAR); +} + +static void channel_write_CHAR(const struct txx9dmac_chan *dc, dma_addr_t val) +{ + if (is_dmac64(dc)) + channel64_write_CHAR(dc, val); + else + channel32_writel(dc, CHAR, val); +} + +static struct txx9dmac_regs __iomem *__txx9dmac_regs( + const struct txx9dmac_dev *ddev) +{ + return ddev->regs; +} + +static struct txx9dmac_regs32 __iomem *__txx9dmac_regs32( + const struct txx9dmac_dev *ddev) +{ + return ddev->regs; +} + +#define dma64_readl(ddev, name) \ + __raw_readl(&(__txx9dmac_regs(ddev)->name)) +#define dma64_writel(ddev, name, val) \ + __raw_writel((val), &(__txx9dmac_regs(ddev)->name)) + +#define dma32_readl(ddev, name) \ + __raw_readl(&(__txx9dmac_regs32(ddev)->name)) +#define dma32_writel(ddev, name, val) \ + __raw_writel((val), &(__txx9dmac_regs32(ddev)->name)) + +#define dma_readl(ddev, name) \ + (__is_dmac64(ddev) ? \ + dma64_readl(ddev, name) : dma32_readl(ddev, name)) +#define dma_writel(ddev, name, val) \ + (__is_dmac64(ddev) ? \ + dma64_writel(ddev, name, val) : dma32_writel(ddev, name, val)) + +static struct device *chan2dev(struct dma_chan *chan) +{ + return &chan->dev->device; +} +static struct device *chan2parent(struct dma_chan *chan) +{ + return chan->dev->device.parent; +} + +static struct txx9dmac_desc * +txd_to_txx9dmac_desc(struct dma_async_tx_descriptor *txd) +{ + return container_of(txd, struct txx9dmac_desc, txd); +} + +static dma_addr_t desc_read_CHAR(const struct txx9dmac_chan *dc, + const struct txx9dmac_desc *desc) +{ + return is_dmac64(dc) ? desc->hwdesc.CHAR : desc->hwdesc32.CHAR; +} + +static void desc_write_CHAR(const struct txx9dmac_chan *dc, + struct txx9dmac_desc *desc, dma_addr_t val) +{ + if (is_dmac64(dc)) + desc->hwdesc.CHAR = val; + else + desc->hwdesc32.CHAR = val; +} + +#define TXX9_DMA_MAX_COUNT 0x04000000 + +#define TXX9_DMA_INITIAL_DESC_COUNT 64 + +static struct txx9dmac_desc *txx9dmac_first_active(struct txx9dmac_chan *dc) +{ + return list_entry(dc->active_list.next, + struct txx9dmac_desc, desc_node); +} + +static struct txx9dmac_desc *txx9dmac_last_active(struct txx9dmac_chan *dc) +{ + return list_entry(dc->active_list.prev, + struct txx9dmac_desc, desc_node); +} + +static struct txx9dmac_desc *txx9dmac_first_queued(struct txx9dmac_chan *dc) +{ + return list_entry(dc->queue.next, struct txx9dmac_desc, desc_node); +} + +static struct txx9dmac_desc *txx9dmac_last_child(struct txx9dmac_desc *desc) +{ + if (!list_empty(&desc->tx_list)) + desc = list_entry(desc->tx_list.prev, typeof(*desc), desc_node); + return desc; +} + +static dma_cookie_t txx9dmac_tx_submit(struct dma_async_tx_descriptor *tx); + +static struct txx9dmac_desc *txx9dmac_desc_alloc(struct txx9dmac_chan *dc, + gfp_t flags) +{ + struct txx9dmac_dev *ddev = dc->ddev; + struct txx9dmac_desc *desc; + + desc = kzalloc(sizeof(*desc), flags); + if (!desc) + return NULL; + INIT_LIST_HEAD(&desc->tx_list); + dma_async_tx_descriptor_init(&desc->txd, &dc->chan); + desc->txd.tx_submit = txx9dmac_tx_submit; + /* txd.flags will be overwritten in prep funcs */ + desc->txd.flags = DMA_CTRL_ACK; + desc->txd.phys = dma_map_single(chan2parent(&dc->chan), &desc->hwdesc, + ddev->descsize, DMA_TO_DEVICE); + return desc; +} + +static struct txx9dmac_desc *txx9dmac_desc_get(struct txx9dmac_chan *dc) +{ + struct txx9dmac_desc *desc, *_desc; + struct txx9dmac_desc *ret = NULL; + unsigned int i = 0; + + spin_lock_bh(&dc->lock); + list_for_each_entry_safe(desc, _desc, &dc->free_list, desc_node) { + if (async_tx_test_ack(&desc->txd)) { + list_del(&desc->desc_node); + ret = desc; + break; + } + dev_dbg(chan2dev(&dc->chan), "desc %p not ACKed\n", desc); + i++; + } + spin_unlock_bh(&dc->lock); + + dev_vdbg(chan2dev(&dc->chan), "scanned %u descriptors on freelist\n", + i); + if (!ret) { + ret = txx9dmac_desc_alloc(dc, GFP_ATOMIC); + if (ret) { + spin_lock_bh(&dc->lock); + dc->descs_allocated++; + spin_unlock_bh(&dc->lock); + } else + dev_err(chan2dev(&dc->chan), + "not enough descriptors available\n"); + } + return ret; +} + +static void txx9dmac_sync_desc_for_cpu(struct txx9dmac_chan *dc, + struct txx9dmac_desc *desc) +{ + struct txx9dmac_dev *ddev = dc->ddev; + struct txx9dmac_desc *child; + + list_for_each_entry(child, &desc->tx_list, desc_node) + dma_sync_single_for_cpu(chan2parent(&dc->chan), + child->txd.phys, ddev->descsize, + DMA_TO_DEVICE); + dma_sync_single_for_cpu(chan2parent(&dc->chan), + desc->txd.phys, ddev->descsize, + DMA_TO_DEVICE); +} + +/* + * Move a descriptor, including any children, to the free list. + * `desc' must not be on any lists. + */ +static void txx9dmac_desc_put(struct txx9dmac_chan *dc, + struct txx9dmac_desc *desc) +{ + if (desc) { + struct txx9dmac_desc *child; + + txx9dmac_sync_desc_for_cpu(dc, desc); + + spin_lock_bh(&dc->lock); + list_for_each_entry(child, &desc->tx_list, desc_node) + dev_vdbg(chan2dev(&dc->chan), + "moving child desc %p to freelist\n", + child); + list_splice_init(&desc->tx_list, &dc->free_list); + dev_vdbg(chan2dev(&dc->chan), "moving desc %p to freelist\n", + desc); + list_add(&desc->desc_node, &dc->free_list); + spin_unlock_bh(&dc->lock); + } +} + +/* Called with dc->lock held and bh disabled */ +static dma_cookie_t +txx9dmac_assign_cookie(struct txx9dmac_chan *dc, struct txx9dmac_desc *desc) +{ + dma_cookie_t cookie = dc->chan.cookie; + + if (++cookie < 0) + cookie = 1; + + dc->chan.cookie = cookie; + desc->txd.cookie = cookie; + + return cookie; +} + +/*----------------------------------------------------------------------*/ + +static void txx9dmac_dump_regs(struct txx9dmac_chan *dc) +{ + if (is_dmac64(dc)) + dev_err(chan2dev(&dc->chan), + " CHAR: %#llx SAR: %#llx DAR: %#llx CNTR: %#x" + " SAIR: %#x DAIR: %#x CCR: %#x CSR: %#x\n", + (u64)channel64_read_CHAR(dc), + channel64_readq(dc, SAR), + channel64_readq(dc, DAR), + channel64_readl(dc, CNTR), + channel64_readl(dc, SAIR), + channel64_readl(dc, DAIR), + channel64_readl(dc, CCR), + channel64_readl(dc, CSR)); + else + dev_err(chan2dev(&dc->chan), + " CHAR: %#x SAR: %#x DAR: %#x CNTR: %#x" + " SAIR: %#x DAIR: %#x CCR: %#x CSR: %#x\n", + channel32_readl(dc, CHAR), + channel32_readl(dc, SAR), + channel32_readl(dc, DAR), + channel32_readl(dc, CNTR), + channel32_readl(dc, SAIR), + channel32_readl(dc, DAIR), + channel32_readl(dc, CCR), + channel32_readl(dc, CSR)); +} + +static void txx9dmac_reset_chan(struct txx9dmac_chan *dc) +{ + channel_writel(dc, CCR, TXX9_DMA_CCR_CHRST); + if (is_dmac64(dc)) { + channel64_clear_CHAR(dc); + channel_writeq(dc, SAR, 0); + channel_writeq(dc, DAR, 0); + } else { + channel_writel(dc, CHAR, 0); + channel_writel(dc, SAR, 0); + channel_writel(dc, DAR, 0); + } + channel_writel(dc, CNTR, 0); + channel_writel(dc, SAIR, 0); + channel_writel(dc, DAIR, 0); + channel_writel(dc, CCR, 0); + mmiowb(); +} + +/* Called with dc->lock held and bh disabled */ +static void txx9dmac_dostart(struct txx9dmac_chan *dc, + struct txx9dmac_desc *first) +{ + struct txx9dmac_slave *ds = dc->chan.private; + u32 sai, dai; + + dev_vdbg(chan2dev(&dc->chan), "dostart %u %p\n", + first->txd.cookie, first); + /* ASSERT: channel is idle */ + if (channel_readl(dc, CSR) & TXX9_DMA_CSR_XFACT) { + dev_err(chan2dev(&dc->chan), + "BUG: Attempted to start non-idle channel\n"); + txx9dmac_dump_regs(dc); + /* The tasklet will hopefully advance the queue... */ + return; + } + + if (is_dmac64(dc)) { + channel64_writel(dc, CNTR, 0); + channel64_writel(dc, CSR, 0xffffffff); + if (ds) { + if (ds->tx_reg) { + sai = ds->reg_width; + dai = 0; + } else { + sai = 0; + dai = ds->reg_width; + } + } else { + sai = 8; + dai = 8; + } + channel64_writel(dc, SAIR, sai); + channel64_writel(dc, DAIR, dai); + /* All 64-bit DMAC supports SMPCHN */ + channel64_writel(dc, CCR, dc->ccr); + /* Writing a non zero value to CHAR will assert XFACT */ + channel64_write_CHAR(dc, first->txd.phys); + } else { + channel32_writel(dc, CNTR, 0); + channel32_writel(dc, CSR, 0xffffffff); + if (ds) { + if (ds->tx_reg) { + sai = ds->reg_width; + dai = 0; + } else { + sai = 0; + dai = ds->reg_width; + } + } else { + sai = 4; + dai = 4; + } + channel32_writel(dc, SAIR, sai); + channel32_writel(dc, DAIR, dai); + if (txx9_dma_have_SMPCHN()) { + channel32_writel(dc, CCR, dc->ccr); + /* Writing a non zero value to CHAR will assert XFACT */ + channel32_writel(dc, CHAR, first->txd.phys); + } else { + channel32_writel(dc, CHAR, first->txd.phys); + channel32_writel(dc, CCR, dc->ccr); + } + } +} + +/*----------------------------------------------------------------------*/ + +static void +txx9dmac_descriptor_complete(struct txx9dmac_chan *dc, + struct txx9dmac_desc *desc) +{ + dma_async_tx_callback callback; + void *param; + struct dma_async_tx_descriptor *txd = &desc->txd; + struct txx9dmac_slave *ds = dc->chan.private; + + dev_vdbg(chan2dev(&dc->chan), "descriptor %u %p complete\n", + txd->cookie, desc); + + dc->completed = txd->cookie; + callback = txd->callback; + param = txd->callback_param; + + txx9dmac_sync_desc_for_cpu(dc, desc); + list_splice_init(&desc->tx_list, &dc->free_list); + list_move(&desc->desc_node, &dc->free_list); + + if (!ds) { + dma_addr_t dmaaddr; + if (!(txd->flags & DMA_COMPL_SKIP_DEST_UNMAP)) { + dmaaddr = is_dmac64(dc) ? + desc->hwdesc.DAR : desc->hwdesc32.DAR; + if (txd->flags & DMA_COMPL_DEST_UNMAP_SINGLE) + dma_unmap_single(chan2parent(&dc->chan), + dmaaddr, desc->len, DMA_FROM_DEVICE); + else + dma_unmap_page(chan2parent(&dc->chan), + dmaaddr, desc->len, DMA_FROM_DEVICE); + } + if (!(txd->flags & DMA_COMPL_SKIP_SRC_UNMAP)) { + dmaaddr = is_dmac64(dc) ? + desc->hwdesc.SAR : desc->hwdesc32.SAR; + if (txd->flags & DMA_COMPL_SRC_UNMAP_SINGLE) + dma_unmap_single(chan2parent(&dc->chan), + dmaaddr, desc->len, DMA_TO_DEVICE); + else + dma_unmap_page(chan2parent(&dc->chan), + dmaaddr, desc->len, DMA_TO_DEVICE); + } + } + + /* + * The API requires that no submissions are done from a + * callback, so we don't need to drop the lock here + */ + if (callback) + callback(param); + dma_run_dependencies(txd); +} + +static void txx9dmac_dequeue(struct txx9dmac_chan *dc, struct list_head *list) +{ + struct txx9dmac_dev *ddev = dc->ddev; + struct txx9dmac_desc *desc; + struct txx9dmac_desc *prev = NULL; + + BUG_ON(!list_empty(list)); + do { + desc = txx9dmac_first_queued(dc); + if (prev) { + desc_write_CHAR(dc, prev, desc->txd.phys); + dma_sync_single_for_device(chan2parent(&dc->chan), + prev->txd.phys, ddev->descsize, + DMA_TO_DEVICE); + } + prev = txx9dmac_last_child(desc); + list_move_tail(&desc->desc_node, list); + /* Make chain-completion interrupt happen */ + if ((desc->txd.flags & DMA_PREP_INTERRUPT) && + !txx9dmac_chan_INTENT(dc)) + break; + } while (!list_empty(&dc->queue)); +} + +static void txx9dmac_complete_all(struct txx9dmac_chan *dc) +{ + struct txx9dmac_desc *desc, *_desc; + LIST_HEAD(list); + + /* + * Submit queued descriptors ASAP, i.e. before we go through + * the completed ones. + */ + list_splice_init(&dc->active_list, &list); + if (!list_empty(&dc->queue)) { + txx9dmac_dequeue(dc, &dc->active_list); + txx9dmac_dostart(dc, txx9dmac_first_active(dc)); + } + + list_for_each_entry_safe(desc, _desc, &list, desc_node) + txx9dmac_descriptor_complete(dc, desc); +} + +static void txx9dmac_dump_desc(struct txx9dmac_chan *dc, + struct txx9dmac_hwdesc *desc) +{ + if (is_dmac64(dc)) { +#ifdef TXX9_DMA_USE_SIMPLE_CHAIN + dev_crit(chan2dev(&dc->chan), + " desc: ch%#llx s%#llx d%#llx c%#x\n", + (u64)desc->CHAR, desc->SAR, desc->DAR, desc->CNTR); +#else + dev_crit(chan2dev(&dc->chan), + " desc: ch%#llx s%#llx d%#llx c%#x" + " si%#x di%#x cc%#x cs%#x\n", + (u64)desc->CHAR, desc->SAR, desc->DAR, desc->CNTR, + desc->SAIR, desc->DAIR, desc->CCR, desc->CSR); +#endif + } else { + struct txx9dmac_hwdesc32 *d = (struct txx9dmac_hwdesc32 *)desc; +#ifdef TXX9_DMA_USE_SIMPLE_CHAIN + dev_crit(chan2dev(&dc->chan), + " desc: ch%#x s%#x d%#x c%#x\n", + d->CHAR, d->SAR, d->DAR, d->CNTR); +#else + dev_crit(chan2dev(&dc->chan), + " desc: ch%#x s%#x d%#x c%#x" + " si%#x di%#x cc%#x cs%#x\n", + d->CHAR, d->SAR, d->DAR, d->CNTR, + d->SAIR, d->DAIR, d->CCR, d->CSR); +#endif + } +} + +static void txx9dmac_handle_error(struct txx9dmac_chan *dc, u32 csr) +{ + struct txx9dmac_desc *bad_desc; + struct txx9dmac_desc *child; + u32 errors; + + /* + * The descriptor currently at the head of the active list is + * borked. Since we don't have any way to report errors, we'll + * just have to scream loudly and try to carry on. + */ + dev_crit(chan2dev(&dc->chan), "Abnormal Chain Completion\n"); + txx9dmac_dump_regs(dc); + + bad_desc = txx9dmac_first_active(dc); + list_del_init(&bad_desc->desc_node); + + /* Clear all error flags and try to restart the controller */ + errors = csr & (TXX9_DMA_CSR_ABCHC | + TXX9_DMA_CSR_CFERR | TXX9_DMA_CSR_CHERR | + TXX9_DMA_CSR_DESERR | TXX9_DMA_CSR_SORERR); + channel_writel(dc, CSR, errors); + + if (list_empty(&dc->active_list) && !list_empty(&dc->queue)) + txx9dmac_dequeue(dc, &dc->active_list); + if (!list_empty(&dc->active_list)) + txx9dmac_dostart(dc, txx9dmac_first_active(dc)); + + dev_crit(chan2dev(&dc->chan), + "Bad descriptor submitted for DMA! (cookie: %d)\n", + bad_desc->txd.cookie); + txx9dmac_dump_desc(dc, &bad_desc->hwdesc); + list_for_each_entry(child, &bad_desc->tx_list, desc_node) + txx9dmac_dump_desc(dc, &child->hwdesc); + /* Pretend the descriptor completed successfully */ + txx9dmac_descriptor_complete(dc, bad_desc); +} + +static void txx9dmac_scan_descriptors(struct txx9dmac_chan *dc) +{ + dma_addr_t chain; + struct txx9dmac_desc *desc, *_desc; + struct txx9dmac_desc *child; + u32 csr; + + if (is_dmac64(dc)) { + chain = channel64_read_CHAR(dc); + csr = channel64_readl(dc, CSR); + channel64_writel(dc, CSR, csr); + } else { + chain = channel32_readl(dc, CHAR); + csr = channel32_readl(dc, CSR); + channel32_writel(dc, CSR, csr); + } + /* For dynamic chain, we should look at XFACT instead of NCHNC */ + if (!(csr & (TXX9_DMA_CSR_XFACT | TXX9_DMA_CSR_ABCHC))) { + /* Everything we've submitted is done */ + txx9dmac_complete_all(dc); + return; + } + if (!(csr & TXX9_DMA_CSR_CHNEN)) + chain = 0; /* last descriptor of this chain */ + + dev_vdbg(chan2dev(&dc->chan), "scan_descriptors: char=%#llx\n", + (u64)chain); + + list_for_each_entry_safe(desc, _desc, &dc->active_list, desc_node) { + if (desc_read_CHAR(dc, desc) == chain) { + /* This one is currently in progress */ + if (csr & TXX9_DMA_CSR_ABCHC) + goto scan_done; + return; + } + + list_for_each_entry(child, &desc->tx_list, desc_node) + if (desc_read_CHAR(dc, child) == chain) { + /* Currently in progress */ + if (csr & TXX9_DMA_CSR_ABCHC) + goto scan_done; + return; + } + + /* + * No descriptors so far seem to be in progress, i.e. + * this one must be done. + */ + txx9dmac_descriptor_complete(dc, desc); + } +scan_done: + if (csr & TXX9_DMA_CSR_ABCHC) { + txx9dmac_handle_error(dc, csr); + return; + } + + dev_err(chan2dev(&dc->chan), + "BUG: All descriptors done, but channel not idle!\n"); + + /* Try to continue after resetting the channel... */ + txx9dmac_reset_chan(dc); + + if (!list_empty(&dc->queue)) { + txx9dmac_dequeue(dc, &dc->active_list); + txx9dmac_dostart(dc, txx9dmac_first_active(dc)); + } +} + +static void txx9dmac_chan_tasklet(unsigned long data) +{ + int irq; + u32 csr; + struct txx9dmac_chan *dc; + + dc = (struct txx9dmac_chan *)data; + csr = channel_readl(dc, CSR); + dev_vdbg(chan2dev(&dc->chan), "tasklet: status=%x\n", csr); + + spin_lock(&dc->lock); + if (csr & (TXX9_DMA_CSR_ABCHC | TXX9_DMA_CSR_NCHNC | + TXX9_DMA_CSR_NTRNFC)) + txx9dmac_scan_descriptors(dc); + spin_unlock(&dc->lock); + irq = dc->irq; + + enable_irq(irq); +} + +static irqreturn_t txx9dmac_chan_interrupt(int irq, void *dev_id) +{ + struct txx9dmac_chan *dc = dev_id; + + dev_vdbg(chan2dev(&dc->chan), "interrupt: status=%#x\n", + channel_readl(dc, CSR)); + + tasklet_schedule(&dc->tasklet); + /* + * Just disable the interrupts. We'll turn them back on in the + * softirq handler. + */ + disable_irq_nosync(irq); + + return IRQ_HANDLED; +} + +static void txx9dmac_tasklet(unsigned long data) +{ + int irq; + u32 csr; + struct txx9dmac_chan *dc; + + struct txx9dmac_dev *ddev = (struct txx9dmac_dev *)data; + u32 mcr; + int i; + + mcr = dma_readl(ddev, MCR); + dev_vdbg(ddev->chan[0]->dma.dev, "tasklet: mcr=%x\n", mcr); + for (i = 0; i < TXX9_DMA_MAX_NR_CHANNELS; i++) { + if ((mcr >> (24 + i)) & 0x11) { + dc = ddev->chan[i]; + csr = channel_readl(dc, CSR); + dev_vdbg(chan2dev(&dc->chan), "tasklet: status=%x\n", + csr); + spin_lock(&dc->lock); + if (csr & (TXX9_DMA_CSR_ABCHC | TXX9_DMA_CSR_NCHNC | + TXX9_DMA_CSR_NTRNFC)) + txx9dmac_scan_descriptors(dc); + spin_unlock(&dc->lock); + } + } + irq = ddev->irq; + + enable_irq(irq); +} + +static irqreturn_t txx9dmac_interrupt(int irq, void *dev_id) +{ + struct txx9dmac_dev *ddev = dev_id; + + dev_vdbg(ddev->chan[0]->dma.dev, "interrupt: status=%#x\n", + dma_readl(ddev, MCR)); + + tasklet_schedule(&ddev->tasklet); + /* + * Just disable the interrupts. We'll turn them back on in the + * softirq handler. + */ + disable_irq_nosync(irq); + + return IRQ_HANDLED; +} + +/*----------------------------------------------------------------------*/ + +static dma_cookie_t txx9dmac_tx_submit(struct dma_async_tx_descriptor *tx) +{ + struct txx9dmac_desc *desc = txd_to_txx9dmac_desc(tx); + struct txx9dmac_chan *dc = to_txx9dmac_chan(tx->chan); + dma_cookie_t cookie; + + spin_lock_bh(&dc->lock); + cookie = txx9dmac_assign_cookie(dc, desc); + + dev_vdbg(chan2dev(tx->chan), "tx_submit: queued %u %p\n", + desc->txd.cookie, desc); + + list_add_tail(&desc->desc_node, &dc->queue); + spin_unlock_bh(&dc->lock); + + return cookie; +} + +static struct dma_async_tx_descriptor * +txx9dmac_prep_dma_memcpy(struct dma_chan *chan, dma_addr_t dest, dma_addr_t src, + size_t len, unsigned long flags) +{ + struct txx9dmac_chan *dc = to_txx9dmac_chan(chan); + struct txx9dmac_dev *ddev = dc->ddev; + struct txx9dmac_desc *desc; + struct txx9dmac_desc *first; + struct txx9dmac_desc *prev; + size_t xfer_count; + size_t offset; + + dev_vdbg(chan2dev(chan), "prep_dma_memcpy d%#llx s%#llx l%#zx f%#lx\n", + (u64)dest, (u64)src, len, flags); + + if (unlikely(!len)) { + dev_dbg(chan2dev(chan), "prep_dma_memcpy: length is zero!\n"); + return NULL; + } + + prev = first = NULL; + + for (offset = 0; offset < len; offset += xfer_count) { + xfer_count = min_t(size_t, len - offset, TXX9_DMA_MAX_COUNT); + /* + * Workaround for ERT-TX49H2-033, ERT-TX49H3-020, + * ERT-TX49H4-016 (slightly conservative) + */ + if (__is_dmac64(ddev)) { + if (xfer_count > 0x100 && + (xfer_count & 0xff) >= 0xfa && + (xfer_count & 0xff) <= 0xff) + xfer_count -= 0x20; + } else { + if (xfer_count > 0x80 && + (xfer_count & 0x7f) >= 0x7e && + (xfer_count & 0x7f) <= 0x7f) + xfer_count -= 0x20; + } + + desc = txx9dmac_desc_get(dc); + if (!desc) { + txx9dmac_desc_put(dc, first); + return NULL; + } + + if (__is_dmac64(ddev)) { + desc->hwdesc.SAR = src + offset; + desc->hwdesc.DAR = dest + offset; + desc->hwdesc.CNTR = xfer_count; + txx9dmac_desc_set_nosimple(ddev, desc, 8, 8, + dc->ccr | TXX9_DMA_CCR_XFACT); + } else { + desc->hwdesc32.SAR = src + offset; + desc->hwdesc32.DAR = dest + offset; + desc->hwdesc32.CNTR = xfer_count; + txx9dmac_desc_set_nosimple(ddev, desc, 4, 4, + dc->ccr | TXX9_DMA_CCR_XFACT); + } + + /* + * The descriptors on tx_list are not reachable from + * the dc->queue list or dc->active_list after a + * submit. If we put all descriptors on active_list, + * calling of callback on the completion will be more + * complex. + */ + if (!first) { + first = desc; + } else { + desc_write_CHAR(dc, prev, desc->txd.phys); + dma_sync_single_for_device(chan2parent(&dc->chan), + prev->txd.phys, ddev->descsize, + DMA_TO_DEVICE); + list_add_tail(&desc->desc_node, &first->tx_list); + } + prev = desc; + } + + /* Trigger interrupt after last block */ + if (flags & DMA_PREP_INTERRUPT) + txx9dmac_desc_set_INTENT(ddev, prev); + + desc_write_CHAR(dc, prev, 0); + dma_sync_single_for_device(chan2parent(&dc->chan), + prev->txd.phys, ddev->descsize, + DMA_TO_DEVICE); + + first->txd.flags = flags; + first->len = len; + + return &first->txd; +} + +static struct dma_async_tx_descriptor * +txx9dmac_prep_slave_sg(struct dma_chan *chan, struct scatterlist *sgl, + unsigned int sg_len, enum dma_data_direction direction, + unsigned long flags) +{ + struct txx9dmac_chan *dc = to_txx9dmac_chan(chan); + struct txx9dmac_dev *ddev = dc->ddev; + struct txx9dmac_slave *ds = chan->private; + struct txx9dmac_desc *prev; + struct txx9dmac_desc *first; + unsigned int i; + struct scatterlist *sg; + + dev_vdbg(chan2dev(chan), "prep_dma_slave\n"); + + BUG_ON(!ds || !ds->reg_width); + if (ds->tx_reg) + BUG_ON(direction != DMA_TO_DEVICE); + else + BUG_ON(direction != DMA_FROM_DEVICE); + if (unlikely(!sg_len)) + return NULL; + + prev = first = NULL; + + for_each_sg(sgl, sg, sg_len, i) { + struct txx9dmac_desc *desc; + dma_addr_t mem; + u32 sai, dai; + + desc = txx9dmac_desc_get(dc); + if (!desc) { + txx9dmac_desc_put(dc, first); + return NULL; + } + + mem = sg_dma_address(sg); + + if (__is_dmac64(ddev)) { + if (direction == DMA_TO_DEVICE) { + desc->hwdesc.SAR = mem; + desc->hwdesc.DAR = ds->tx_reg; + } else { + desc->hwdesc.SAR = ds->rx_reg; + desc->hwdesc.DAR = mem; + } + desc->hwdesc.CNTR = sg_dma_len(sg); + } else { + if (direction == DMA_TO_DEVICE) { + desc->hwdesc32.SAR = mem; + desc->hwdesc32.DAR = ds->tx_reg; + } else { + desc->hwdesc32.SAR = ds->rx_reg; + desc->hwdesc32.DAR = mem; + } + desc->hwdesc32.CNTR = sg_dma_len(sg); + } + if (direction == DMA_TO_DEVICE) { + sai = ds->reg_width; + dai = 0; + } else { + sai = 0; + dai = ds->reg_width; + } + txx9dmac_desc_set_nosimple(ddev, desc, sai, dai, + dc->ccr | TXX9_DMA_CCR_XFACT); + + if (!first) { + first = desc; + } else { + desc_write_CHAR(dc, prev, desc->txd.phys); + dma_sync_single_for_device(chan2parent(&dc->chan), + prev->txd.phys, + ddev->descsize, + DMA_TO_DEVICE); + list_add_tail(&desc->desc_node, &first->tx_list); + } + prev = desc; + } + + /* Trigger interrupt after last block */ + if (flags & DMA_PREP_INTERRUPT) + txx9dmac_desc_set_INTENT(ddev, prev); + + desc_write_CHAR(dc, prev, 0); + dma_sync_single_for_device(chan2parent(&dc->chan), + prev->txd.phys, ddev->descsize, + DMA_TO_DEVICE); + + first->txd.flags = flags; + first->len = 0; + + return &first->txd; +} + +static void txx9dmac_terminate_all(struct dma_chan *chan) +{ + struct txx9dmac_chan *dc = to_txx9dmac_chan(chan); + struct txx9dmac_desc *desc, *_desc; + LIST_HEAD(list); + + dev_vdbg(chan2dev(chan), "terminate_all\n"); + spin_lock_bh(&dc->lock); + + txx9dmac_reset_chan(dc); + + /* active_list entries will end up before queued entries */ + list_splice_init(&dc->queue, &list); + list_splice_init(&dc->active_list, &list); + + spin_unlock_bh(&dc->lock); + + /* Flush all pending and queued descriptors */ + list_for_each_entry_safe(desc, _desc, &list, desc_node) + txx9dmac_descriptor_complete(dc, desc); +} + +static enum dma_status +txx9dmac_is_tx_complete(struct dma_chan *chan, + dma_cookie_t cookie, + dma_cookie_t *done, dma_cookie_t *used) +{ + struct txx9dmac_chan *dc = to_txx9dmac_chan(chan); + dma_cookie_t last_used; + dma_cookie_t last_complete; + int ret; + + last_complete = dc->completed; + last_used = chan->cookie; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + if (ret != DMA_SUCCESS) { + spin_lock_bh(&dc->lock); + txx9dmac_scan_descriptors(dc); + spin_unlock_bh(&dc->lock); + + last_complete = dc->completed; + last_used = chan->cookie; + + ret = dma_async_is_complete(cookie, last_complete, last_used); + } + + if (done) + *done = last_complete; + if (used) + *used = last_used; + + return ret; +} + +static void txx9dmac_chain_dynamic(struct txx9dmac_chan *dc, + struct txx9dmac_desc *prev) +{ + struct txx9dmac_dev *ddev = dc->ddev; + struct txx9dmac_desc *desc; + LIST_HEAD(list); + + prev = txx9dmac_last_child(prev); + txx9dmac_dequeue(dc, &list); + desc = list_entry(list.next, struct txx9dmac_desc, desc_node); + desc_write_CHAR(dc, prev, desc->txd.phys); + dma_sync_single_for_device(chan2parent(&dc->chan), + prev->txd.phys, ddev->descsize, + DMA_TO_DEVICE); + mmiowb(); + if (!(channel_readl(dc, CSR) & TXX9_DMA_CSR_CHNEN) && + channel_read_CHAR(dc) == prev->txd.phys) + /* Restart chain DMA */ + channel_write_CHAR(dc, desc->txd.phys); + list_splice_tail(&list, &dc->active_list); +} + +static void txx9dmac_issue_pending(struct dma_chan *chan) +{ + struct txx9dmac_chan *dc = to_txx9dmac_chan(chan); + + spin_lock_bh(&dc->lock); + + if (!list_empty(&dc->active_list)) + txx9dmac_scan_descriptors(dc); + if (!list_empty(&dc->queue)) { + if (list_empty(&dc->active_list)) { + txx9dmac_dequeue(dc, &dc->active_list); + txx9dmac_dostart(dc, txx9dmac_first_active(dc)); + } else if (txx9_dma_have_SMPCHN()) { + struct txx9dmac_desc *prev = txx9dmac_last_active(dc); + + if (!(prev->txd.flags & DMA_PREP_INTERRUPT) || + txx9dmac_chan_INTENT(dc)) + txx9dmac_chain_dynamic(dc, prev); + } + } + + spin_unlock_bh(&dc->lock); +} + +static int txx9dmac_alloc_chan_resources(struct dma_chan *chan) +{ + struct txx9dmac_chan *dc = to_txx9dmac_chan(chan); + struct txx9dmac_slave *ds = chan->private; + struct txx9dmac_desc *desc; + int i; + + dev_vdbg(chan2dev(chan), "alloc_chan_resources\n"); + + /* ASSERT: channel is idle */ + if (channel_readl(dc, CSR) & TXX9_DMA_CSR_XFACT) { + dev_dbg(chan2dev(chan), "DMA channel not idle?\n"); + return -EIO; + } + + dc->completed = chan->cookie = 1; + + dc->ccr = TXX9_DMA_CCR_IMMCHN | TXX9_DMA_CCR_INTENE | CCR_LE; + txx9dmac_chan_set_SMPCHN(dc); + if (!txx9_dma_have_SMPCHN() || (dc->ccr & TXX9_DMA_CCR_SMPCHN)) + dc->ccr |= TXX9_DMA_CCR_INTENC; + if (chan->device->device_prep_dma_memcpy) { + if (ds) + return -EINVAL; + dc->ccr |= TXX9_DMA_CCR_XFSZ_X8; + } else { + if (!ds || + (ds->tx_reg && ds->rx_reg) || (!ds->tx_reg && !ds->rx_reg)) + return -EINVAL; + dc->ccr |= TXX9_DMA_CCR_EXTRQ | + TXX9_DMA_CCR_XFSZ(__ffs(ds->reg_width)); + txx9dmac_chan_set_INTENT(dc); + } + + spin_lock_bh(&dc->lock); + i = dc->descs_allocated; + while (dc->descs_allocated < TXX9_DMA_INITIAL_DESC_COUNT) { + spin_unlock_bh(&dc->lock); + + desc = txx9dmac_desc_alloc(dc, GFP_KERNEL); + if (!desc) { + dev_info(chan2dev(chan), + "only allocated %d descriptors\n", i); + spin_lock_bh(&dc->lock); + break; + } + txx9dmac_desc_put(dc, desc); + + spin_lock_bh(&dc->lock); + i = ++dc->descs_allocated; + } + spin_unlock_bh(&dc->lock); + + dev_dbg(chan2dev(chan), + "alloc_chan_resources allocated %d descriptors\n", i); + + return i; +} + +static void txx9dmac_free_chan_resources(struct dma_chan *chan) +{ + struct txx9dmac_chan *dc = to_txx9dmac_chan(chan); + struct txx9dmac_dev *ddev = dc->ddev; + struct txx9dmac_desc *desc, *_desc; + LIST_HEAD(list); + + dev_dbg(chan2dev(chan), "free_chan_resources (descs allocated=%u)\n", + dc->descs_allocated); + + /* ASSERT: channel is idle */ + BUG_ON(!list_empty(&dc->active_list)); + BUG_ON(!list_empty(&dc->queue)); + BUG_ON(channel_readl(dc, CSR) & TXX9_DMA_CSR_XFACT); + + spin_lock_bh(&dc->lock); + list_splice_init(&dc->free_list, &list); + dc->descs_allocated = 0; + spin_unlock_bh(&dc->lock); + + list_for_each_entry_safe(desc, _desc, &list, desc_node) { + dev_vdbg(chan2dev(chan), " freeing descriptor %p\n", desc); + dma_unmap_single(chan2parent(chan), desc->txd.phys, + ddev->descsize, DMA_TO_DEVICE); + kfree(desc); + } + + dev_vdbg(chan2dev(chan), "free_chan_resources done\n"); +} + +/*----------------------------------------------------------------------*/ + +static void txx9dmac_off(struct txx9dmac_dev *ddev) +{ + dma_writel(ddev, MCR, 0); + mmiowb(); +} + +static int __init txx9dmac_chan_probe(struct platform_device *pdev) +{ + struct txx9dmac_chan_platform_data *cpdata = pdev->dev.platform_data; + struct platform_device *dmac_dev = cpdata->dmac_dev; + struct txx9dmac_platform_data *pdata = dmac_dev->dev.platform_data; + struct txx9dmac_chan *dc; + int err; + int ch = pdev->id % TXX9_DMA_MAX_NR_CHANNELS; + int irq; + + dc = devm_kzalloc(&pdev->dev, sizeof(*dc), GFP_KERNEL); + if (!dc) + return -ENOMEM; + + dc->dma.dev = &pdev->dev; + dc->dma.device_alloc_chan_resources = txx9dmac_alloc_chan_resources; + dc->dma.device_free_chan_resources = txx9dmac_free_chan_resources; + dc->dma.device_terminate_all = txx9dmac_terminate_all; + dc->dma.device_is_tx_complete = txx9dmac_is_tx_complete; + dc->dma.device_issue_pending = txx9dmac_issue_pending; + if (pdata && pdata->memcpy_chan == ch) { + dc->dma.device_prep_dma_memcpy = txx9dmac_prep_dma_memcpy; + dma_cap_set(DMA_MEMCPY, dc->dma.cap_mask); + } else { + dc->dma.device_prep_slave_sg = txx9dmac_prep_slave_sg; + dma_cap_set(DMA_SLAVE, dc->dma.cap_mask); + dma_cap_set(DMA_PRIVATE, dc->dma.cap_mask); + } + + INIT_LIST_HEAD(&dc->dma.channels); + dc->ddev = platform_get_drvdata(dmac_dev); + if (dc->ddev->irq < 0) { + irq = platform_get_irq(pdev, 0); + if (irq < 0) + return irq; + tasklet_init(&dc->tasklet, txx9dmac_chan_tasklet, + (unsigned long)dc); + dc->irq = irq; + err = devm_request_irq(&pdev->dev, dc->irq, + txx9dmac_chan_interrupt, 0, dev_name(&pdev->dev), dc); + if (err) + return err; + } else + dc->irq = -1; + dc->ddev->chan[ch] = dc; + dc->chan.device = &dc->dma; + list_add_tail(&dc->chan.device_node, &dc->chan.device->channels); + dc->chan.cookie = dc->completed = 1; + + if (is_dmac64(dc)) + dc->ch_regs = &__txx9dmac_regs(dc->ddev)->CHAN[ch]; + else + dc->ch_regs = &__txx9dmac_regs32(dc->ddev)->CHAN[ch]; + spin_lock_init(&dc->lock); + + INIT_LIST_HEAD(&dc->active_list); + INIT_LIST_HEAD(&dc->queue); + INIT_LIST_HEAD(&dc->free_list); + + txx9dmac_reset_chan(dc); + + platform_set_drvdata(pdev, dc); + + err = dma_async_device_register(&dc->dma); + if (err) + return err; + dev_dbg(&pdev->dev, "TXx9 DMA Channel (dma%d%s%s)\n", + dc->dma.dev_id, + dma_has_cap(DMA_MEMCPY, dc->dma.cap_mask) ? " memcpy" : "", + dma_has_cap(DMA_SLAVE, dc->dma.cap_mask) ? " slave" : ""); + + return 0; +} + +static int __exit txx9dmac_chan_remove(struct platform_device *pdev) +{ + struct txx9dmac_chan *dc = platform_get_drvdata(pdev); + + dma_async_device_unregister(&dc->dma); + if (dc->irq >= 0) + tasklet_kill(&dc->tasklet); + dc->ddev->chan[pdev->id % TXX9_DMA_MAX_NR_CHANNELS] = NULL; + return 0; +} + +static int __init txx9dmac_probe(struct platform_device *pdev) +{ + struct txx9dmac_platform_data *pdata = pdev->dev.platform_data; + struct resource *io; + struct txx9dmac_dev *ddev; + u32 mcr; + int err; + + io = platform_get_resource(pdev, IORESOURCE_MEM, 0); + if (!io) + return -EINVAL; + + ddev = devm_kzalloc(&pdev->dev, sizeof(*ddev), GFP_KERNEL); + if (!ddev) + return -ENOMEM; + + if (!devm_request_mem_region(&pdev->dev, io->start, resource_size(io), + dev_name(&pdev->dev))) + return -EBUSY; + + ddev->regs = devm_ioremap(&pdev->dev, io->start, resource_size(io)); + if (!ddev->regs) + return -ENOMEM; + ddev->have_64bit_regs = pdata->have_64bit_regs; + if (__is_dmac64(ddev)) + ddev->descsize = sizeof(struct txx9dmac_hwdesc); + else + ddev->descsize = sizeof(struct txx9dmac_hwdesc32); + + /* force dma off, just in case */ + txx9dmac_off(ddev); + + ddev->irq = platform_get_irq(pdev, 0); + if (ddev->irq >= 0) { + tasklet_init(&ddev->tasklet, txx9dmac_tasklet, + (unsigned long)ddev); + err = devm_request_irq(&pdev->dev, ddev->irq, + txx9dmac_interrupt, 0, dev_name(&pdev->dev), ddev); + if (err) + return err; + } + + mcr = TXX9_DMA_MCR_MSTEN | MCR_LE; + if (pdata && pdata->memcpy_chan >= 0) + mcr |= TXX9_DMA_MCR_FIFUM(pdata->memcpy_chan); + dma_writel(ddev, MCR, mcr); + + platform_set_drvdata(pdev, ddev); + return 0; +} + +static int __exit txx9dmac_remove(struct platform_device *pdev) +{ + struct txx9dmac_dev *ddev = platform_get_drvdata(pdev); + + txx9dmac_off(ddev); + if (ddev->irq >= 0) + tasklet_kill(&ddev->tasklet); + return 0; +} + +static void txx9dmac_shutdown(struct platform_device *pdev) +{ + struct txx9dmac_dev *ddev = platform_get_drvdata(pdev); + + txx9dmac_off(ddev); +} + +static int txx9dmac_suspend_noirq(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct txx9dmac_dev *ddev = platform_get_drvdata(pdev); + + txx9dmac_off(ddev); + return 0; +} + +static int txx9dmac_resume_noirq(struct device *dev) +{ + struct platform_device *pdev = to_platform_device(dev); + struct txx9dmac_dev *ddev = platform_get_drvdata(pdev); + struct txx9dmac_platform_data *pdata = pdev->dev.platform_data; + u32 mcr; + + mcr = TXX9_DMA_MCR_MSTEN | MCR_LE; + if (pdata && pdata->memcpy_chan >= 0) + mcr |= TXX9_DMA_MCR_FIFUM(pdata->memcpy_chan); + dma_writel(ddev, MCR, mcr); + return 0; + +} + +static const struct dev_pm_ops txx9dmac_dev_pm_ops = { + .suspend_noirq = txx9dmac_suspend_noirq, + .resume_noirq = txx9dmac_resume_noirq, +}; + +static struct platform_driver txx9dmac_chan_driver = { + .remove = __exit_p(txx9dmac_chan_remove), + .driver = { + .name = "txx9dmac-chan", + }, +}; + +static struct platform_driver txx9dmac_driver = { + .remove = __exit_p(txx9dmac_remove), + .shutdown = txx9dmac_shutdown, + .driver = { + .name = "txx9dmac", + .pm = &txx9dmac_dev_pm_ops, + }, +}; + +static int __init txx9dmac_init(void) +{ + int rc; + + rc = platform_driver_probe(&txx9dmac_driver, txx9dmac_probe); + if (!rc) { + rc = platform_driver_probe(&txx9dmac_chan_driver, + txx9dmac_chan_probe); + if (rc) + platform_driver_unregister(&txx9dmac_driver); + } + return rc; +} +module_init(txx9dmac_init); + +static void __exit txx9dmac_exit(void) +{ + platform_driver_unregister(&txx9dmac_chan_driver); + platform_driver_unregister(&txx9dmac_driver); +} +module_exit(txx9dmac_exit); + +MODULE_LICENSE("GPL"); +MODULE_DESCRIPTION("TXx9 DMA Controller driver"); +MODULE_AUTHOR("Atsushi Nemoto <anemo@mba.ocn.ne.jp>"); diff --git a/drivers/dma/txx9dmac.h b/drivers/dma/txx9dmac.h new file mode 100644 index 00000000000..365d42366b9 --- /dev/null +++ b/drivers/dma/txx9dmac.h @@ -0,0 +1,308 @@ +/* + * Driver for the TXx9 SoC DMA Controller + * + * Copyright (C) 2009 Atsushi Nemoto + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ +#ifndef TXX9DMAC_H +#define TXX9DMAC_H + +#include <linux/dmaengine.h> +#include <asm/txx9/dmac.h> + +/* + * Design Notes: + * + * This DMAC have four channels and one FIFO buffer. Each channel can + * be configured for memory-memory or device-memory transfer, but only + * one channel can do alignment-free memory-memory transfer at a time + * while the channel should occupy the FIFO buffer for effective + * transfers. + * + * Instead of dynamically assign the FIFO buffer to channels, I chose + * make one dedicated channel for memory-memory transfer. The + * dedicated channel is public. Other channels are private and used + * for slave transfer. Some devices in the SoC are wired to certain + * DMA channel. + */ + +#ifdef CONFIG_MACH_TX49XX +static inline bool txx9_dma_have_SMPCHN(void) +{ + return true; +} +#define TXX9_DMA_USE_SIMPLE_CHAIN +#else +static inline bool txx9_dma_have_SMPCHN(void) +{ + return false; +} +#endif + +#ifdef __LITTLE_ENDIAN +#ifdef CONFIG_MACH_TX49XX +#define CCR_LE TXX9_DMA_CCR_LE +#define MCR_LE 0 +#else +#define CCR_LE 0 +#define MCR_LE TXX9_DMA_MCR_LE +#endif +#else +#define CCR_LE 0 +#define MCR_LE 0 +#endif + +/* + * Redefine this macro to handle differences between 32- and 64-bit + * addressing, big vs. little endian, etc. + */ +#ifdef __BIG_ENDIAN +#define TXX9_DMA_REG32(name) u32 __pad_##name; u32 name +#else +#define TXX9_DMA_REG32(name) u32 name; u32 __pad_##name +#endif + +/* Hardware register definitions. */ +struct txx9dmac_cregs { +#if defined(CONFIG_32BIT) && !defined(CONFIG_64BIT_PHYS_ADDR) + TXX9_DMA_REG32(CHAR); /* Chain Address Register */ +#else + u64 CHAR; /* Chain Address Register */ +#endif + u64 SAR; /* Source Address Register */ + u64 DAR; /* Destination Address Register */ + TXX9_DMA_REG32(CNTR); /* Count Register */ + TXX9_DMA_REG32(SAIR); /* Source Address Increment Register */ + TXX9_DMA_REG32(DAIR); /* Destination Address Increment Register */ + TXX9_DMA_REG32(CCR); /* Channel Control Register */ + TXX9_DMA_REG32(CSR); /* Channel Status Register */ +}; +struct txx9dmac_cregs32 { + u32 CHAR; + u32 SAR; + u32 DAR; + u32 CNTR; + u32 SAIR; + u32 DAIR; + u32 CCR; + u32 CSR; +}; + +struct txx9dmac_regs { + /* per-channel registers */ + struct txx9dmac_cregs CHAN[TXX9_DMA_MAX_NR_CHANNELS]; + u64 __pad[9]; + u64 MFDR; /* Memory Fill Data Register */ + TXX9_DMA_REG32(MCR); /* Master Control Register */ +}; +struct txx9dmac_regs32 { + struct txx9dmac_cregs32 CHAN[TXX9_DMA_MAX_NR_CHANNELS]; + u32 __pad[9]; + u32 MFDR; + u32 MCR; +}; + +/* bits for MCR */ +#define TXX9_DMA_MCR_EIS(ch) (0x10000000<<(ch)) +#define TXX9_DMA_MCR_DIS(ch) (0x01000000<<(ch)) +#define TXX9_DMA_MCR_RSFIF 0x00000080 +#define TXX9_DMA_MCR_FIFUM(ch) (0x00000008<<(ch)) +#define TXX9_DMA_MCR_LE 0x00000004 +#define TXX9_DMA_MCR_RPRT 0x00000002 +#define TXX9_DMA_MCR_MSTEN 0x00000001 + +/* bits for CCRn */ +#define TXX9_DMA_CCR_IMMCHN 0x20000000 +#define TXX9_DMA_CCR_USEXFSZ 0x10000000 +#define TXX9_DMA_CCR_LE 0x08000000 +#define TXX9_DMA_CCR_DBINH 0x04000000 +#define TXX9_DMA_CCR_SBINH 0x02000000 +#define TXX9_DMA_CCR_CHRST 0x01000000 +#define TXX9_DMA_CCR_RVBYTE 0x00800000 +#define TXX9_DMA_CCR_ACKPOL 0x00400000 +#define TXX9_DMA_CCR_REQPL 0x00200000 +#define TXX9_DMA_CCR_EGREQ 0x00100000 +#define TXX9_DMA_CCR_CHDN 0x00080000 +#define TXX9_DMA_CCR_DNCTL 0x00060000 +#define TXX9_DMA_CCR_EXTRQ 0x00010000 +#define TXX9_DMA_CCR_INTRQD 0x0000e000 +#define TXX9_DMA_CCR_INTENE 0x00001000 +#define TXX9_DMA_CCR_INTENC 0x00000800 +#define TXX9_DMA_CCR_INTENT 0x00000400 +#define TXX9_DMA_CCR_CHNEN 0x00000200 +#define TXX9_DMA_CCR_XFACT 0x00000100 +#define TXX9_DMA_CCR_SMPCHN 0x00000020 +#define TXX9_DMA_CCR_XFSZ(order) (((order) << 2) & 0x0000001c) +#define TXX9_DMA_CCR_XFSZ_1 TXX9_DMA_CCR_XFSZ(0) +#define TXX9_DMA_CCR_XFSZ_2 TXX9_DMA_CCR_XFSZ(1) +#define TXX9_DMA_CCR_XFSZ_4 TXX9_DMA_CCR_XFSZ(2) +#define TXX9_DMA_CCR_XFSZ_8 TXX9_DMA_CCR_XFSZ(3) +#define TXX9_DMA_CCR_XFSZ_X4 TXX9_DMA_CCR_XFSZ(4) +#define TXX9_DMA_CCR_XFSZ_X8 TXX9_DMA_CCR_XFSZ(5) +#define TXX9_DMA_CCR_XFSZ_X16 TXX9_DMA_CCR_XFSZ(6) +#define TXX9_DMA_CCR_XFSZ_X32 TXX9_DMA_CCR_XFSZ(7) +#define TXX9_DMA_CCR_MEMIO 0x00000002 +#define TXX9_DMA_CCR_SNGAD 0x00000001 + +/* bits for CSRn */ +#define TXX9_DMA_CSR_CHNEN 0x00000400 +#define TXX9_DMA_CSR_STLXFER 0x00000200 +#define TXX9_DMA_CSR_XFACT 0x00000100 +#define TXX9_DMA_CSR_ABCHC 0x00000080 +#define TXX9_DMA_CSR_NCHNC 0x00000040 +#define TXX9_DMA_CSR_NTRNFC 0x00000020 +#define TXX9_DMA_CSR_EXTDN 0x00000010 +#define TXX9_DMA_CSR_CFERR 0x00000008 +#define TXX9_DMA_CSR_CHERR 0x00000004 +#define TXX9_DMA_CSR_DESERR 0x00000002 +#define TXX9_DMA_CSR_SORERR 0x00000001 + +struct txx9dmac_chan { + struct dma_chan chan; + struct dma_device dma; + struct txx9dmac_dev *ddev; + void __iomem *ch_regs; + struct tasklet_struct tasklet; + int irq; + u32 ccr; + + spinlock_t lock; + + /* these other elements are all protected by lock */ + dma_cookie_t completed; + struct list_head active_list; + struct list_head queue; + struct list_head free_list; + + unsigned int descs_allocated; +}; + +struct txx9dmac_dev { + void __iomem *regs; + struct tasklet_struct tasklet; + int irq; + struct txx9dmac_chan *chan[TXX9_DMA_MAX_NR_CHANNELS]; + bool have_64bit_regs; + unsigned int descsize; +}; + +static inline bool __is_dmac64(const struct txx9dmac_dev *ddev) +{ + return ddev->have_64bit_regs; +} + +static inline bool is_dmac64(const struct txx9dmac_chan *dc) +{ + return __is_dmac64(dc->ddev); +} + +#ifdef TXX9_DMA_USE_SIMPLE_CHAIN +/* Hardware descriptor definition. (for simple-chain) */ +struct txx9dmac_hwdesc { +#if defined(CONFIG_32BIT) && !defined(CONFIG_64BIT_PHYS_ADDR) + TXX9_DMA_REG32(CHAR); +#else + u64 CHAR; +#endif + u64 SAR; + u64 DAR; + TXX9_DMA_REG32(CNTR); +}; +struct txx9dmac_hwdesc32 { + u32 CHAR; + u32 SAR; + u32 DAR; + u32 CNTR; +}; +#else +#define txx9dmac_hwdesc txx9dmac_cregs +#define txx9dmac_hwdesc32 txx9dmac_cregs32 +#endif + +struct txx9dmac_desc { + /* FIRST values the hardware uses */ + union { + struct txx9dmac_hwdesc hwdesc; + struct txx9dmac_hwdesc32 hwdesc32; + }; + + /* THEN values for driver housekeeping */ + struct list_head desc_node ____cacheline_aligned; + struct list_head tx_list; + struct dma_async_tx_descriptor txd; + size_t len; +}; + +#ifdef TXX9_DMA_USE_SIMPLE_CHAIN + +static inline bool txx9dmac_chan_INTENT(struct txx9dmac_chan *dc) +{ + return (dc->ccr & TXX9_DMA_CCR_INTENT) != 0; +} + +static inline void txx9dmac_chan_set_INTENT(struct txx9dmac_chan *dc) +{ + dc->ccr |= TXX9_DMA_CCR_INTENT; +} + +static inline void txx9dmac_desc_set_INTENT(struct txx9dmac_dev *ddev, + struct txx9dmac_desc *desc) +{ +} + +static inline void txx9dmac_chan_set_SMPCHN(struct txx9dmac_chan *dc) +{ + dc->ccr |= TXX9_DMA_CCR_SMPCHN; +} + +static inline void txx9dmac_desc_set_nosimple(struct txx9dmac_dev *ddev, + struct txx9dmac_desc *desc, + u32 sair, u32 dair, u32 ccr) +{ +} + +#else /* TXX9_DMA_USE_SIMPLE_CHAIN */ + +static inline bool txx9dmac_chan_INTENT(struct txx9dmac_chan *dc) +{ + return true; +} + +static void txx9dmac_chan_set_INTENT(struct txx9dmac_chan *dc) +{ +} + +static inline void txx9dmac_desc_set_INTENT(struct txx9dmac_dev *ddev, + struct txx9dmac_desc *desc) +{ + if (__is_dmac64(ddev)) + desc->hwdesc.CCR |= TXX9_DMA_CCR_INTENT; + else + desc->hwdesc32.CCR |= TXX9_DMA_CCR_INTENT; +} + +static inline void txx9dmac_chan_set_SMPCHN(struct txx9dmac_chan *dc) +{ +} + +static inline void txx9dmac_desc_set_nosimple(struct txx9dmac_dev *ddev, + struct txx9dmac_desc *desc, + u32 sai, u32 dai, u32 ccr) +{ + if (__is_dmac64(ddev)) { + desc->hwdesc.SAIR = sai; + desc->hwdesc.DAIR = dai; + desc->hwdesc.CCR = ccr; + } else { + desc->hwdesc32.SAIR = sai; + desc->hwdesc32.DAIR = dai; + desc->hwdesc32.CCR = ccr; + } +} + +#endif /* TXX9_DMA_USE_SIMPLE_CHAIN */ + +#endif /* TXX9DMAC_H */ |