summaryrefslogtreecommitdiffstats
path: root/loader
Commit message (Collapse)AuthorAgeFilesLines
...
* Fixes problems in the manual network configuration screen in loader with IPv6.Ales Kozumplik2010-01-071-1/+1
| | | | | | A crash occurs after the IPv6 GUI has been enabled when we chose to configure the network device manually. When the crash is fixed the screen still does not appear. This patch fixes both the problems.
* Bring back missing IPv6 pieces that were lost in time.David Cantrell2010-01-073-1/+6
| | | | | | strcmp() -> strncmp() for ipv6 'auto' comparison. Add back missing gateway6 member in loaderData. Initialize ipv6 and gateway6 in loaderData to NULL.
* Add configuration option to enable/disable IPv6 support.David Cantrell2010-01-071-1/+1
| | | | | Default is IPv6 is enabled, but people can pass --disable-ipv6 to the configure script to disable the IPv6 bits in loader.
* Kill yet another unused lodaer flag.Chris Lumens2009-12-061-1/+1
|
* stage1 (init): Make /tmp tmpfs large enough to hold install.img (#540146)Hans de Goede2009-12-041-1/+1
| | | | | | | | | | Be default mount makes tmpfs mounts 1/2 RAM size, on 256MB machines, where part of the RAM is taken up by a shared framebuffer, this results in a /tmp which is to small to hold install.img, resulting in install failure when trying to do for example a network installation. This patch fixes this by specifying a size for /tmp, it uses 250MB which should be plenty for now and the forseeable future.
* Remove test mode from the loader, too.Chris Lumens2009-12-0312-240/+154
| | | | | All previous comments about test mode being untested apply to loader, but doubly so.
* Make loader log time with milliseconds (part of #524980).Ales Kozumplik2009-12-021-12/+17
| | | | | This is to make the anaconda.log have consistent entries from loader and from stage 2.
* Adds interactive install support for NFS options (#537764)Ales Kozumplik2009-11-191-13/+36
| | | | This is a migration of the patch 22e2548cbb41a0c32fa3a0a2e67e67bbf3109155 from rhel5-branch, BZ#493052.
* Introduces check_asprintf macro that checks asprintfs return value and ↵Ales Kozumplik2009-11-1913-347/+138
| | | | | | | | | terminates program in OOM scenarios. This is to avoid having to copy-paste the asprintf-log-abort if branch all the time. This commit also modifies existing asprintf() calls to use the new macro in places where the change wouldn't modify program's semantics.
* Sleep if the kickstart file read fails (#537361)Ales Kozumplik2009-11-193-2/+27
| | | | | | | | | | | | | If a read from a block device (like USB key) or a cdrom (USB cdrom) fails, sleep a bit to give the device some time to initialize. Return status of the mount command is checked so waiting occurs only if there is a reasonable chance that the device might eventually come online. This required extending our existing mounting methods to report the correct mount error code. I tested that * the installer still runs * reading the kickstart from a usb works * starting the installer with no USB key inserted but a command line argument pointing to a USB drive will trigger the waiting cycle and that when the key is inserted while waiting the kickstart is eventually found on the key.
* Move libcurl initialization to urlinstTransfer() (#537870).David Cantrell2009-11-183-23/+20
| | | | | | | | | | | | | | | | | | Recent trees have been giving errors when we get to the stage2 download point, such as this one: Unable to retrieve http://download.fedoraproject.org/pub/fedora/linux/development/x86_64/os//images/install.img. libcurl needs to be initialized inside urlinstTransfer() so it has the latest network state for the system. The above error is caused by old DNS settings. libcurl won't automatically do a res_init(), so we need to set up a new curl instance and use that. Since this patch moves curl usage to be exclusively within urlinstTransfer(), move 'curl' to that function as well and remove it from loaderData. Add curl cleanup calls at the end of urlinstTransfer().
* Replace all popt use with glib's option parsing code.David Cantrell2009-11-1810-194/+225
| | | | | We have glib for other things, so we can use it for option parsing and dump another library dependency.
* Clean up initProductInfo() in loader.c.David Cantrell2009-11-181-37/+45
| | | | | Use g_file_get_contents() and g_strsplit() to read in .buildstamp data and split it up. Remove static buffers.
* Use glib string parsing functions in driverselect.c.David Cantrell2009-11-181-31/+9
|
* Get correct boot device in reIPL code for s390 (#537390).Mark Hamzy2009-11-171-0/+10
| | | | | | Account for the /boot existing on its own partition or as part of /. Output the values of the reipl configuration in linuxrc.s390 before reboot.
* Expand the proxy table a little bit to reduce clutter (#537878).Chris Lumens2009-11-171-7/+7
|
* Use glib data structures in loader's module handling code.David Cantrell2009-11-165-266/+276
| | | | | Use glib's data structures and string functions in modules.c since we already have glib. Add in some safety checks as well.
* Remove last references to hal.David Cantrell2009-11-121-8/+0
| | | | | We don't need hal anymore, so remove it from the initrd.img generation and stop loader from starting up hald.
* Honor existing RUNKS conf file variable on s390 (#513951).David Cantrell2009-11-101-0/+4
| | | | | | The linuxrc.s390 rewrite changed the behavior of RUNKS on RHEL-5. We've done away with RUNKS on RHEL-6, but we still need to maintain existing functionality on RHEL-5.
* Add "Hipersockets" to qeth NETTYPE description (#511962).David Cantrell2009-11-101-1/+1
|
* rescue: Don't copy install.ing to /tmp when not enough RAM (#531304, #529392)Jerry Vonau2009-11-091-1/+1
| | | | | | | With rescue copying install.img to /tmp when available_memory>128000k, this left the machine unable to hold install.img in ram if there is under 256M of ram. Patch to change that to be based on MIN_GUI_RAM instead.
* Correct modopts initialization in loader (take 2) (#531932).David Cantrell2009-11-061-10/+33
| | | | | | | | | | | | | The edc665e6fa2ba71e89eb83412738622e916c3a05 commit prevented SIGSEGV but changed behavior of modopts to where we'd never get any values in modopts. What needed to happen is the else clause in addOption() needed to initalize the .options array and add in the option value. That wasn't happening, so strcat() calls in writeModulesConf() were causing SIGSEGV. This patch also adds some realloc() and malloc() checks. Tested with "radeon.nomodeset=1" and without that parameter on x86_64.
* Get rid of dead code, and fix gettimespecofday's math.Peter Jones2009-11-065-76/+2
| | | | | We've got a lot of dead code here. This removes a bunch of it. Also, it fixes gettimespecofday's math to yield the correct result.
* Don't exec without forking first when calling udevadm.Peter Jones2009-11-051-7/+32
| | | | | | Since we don't want init to exit when we call udevadm, fork first. But before we do that, may as well wait for our forked udevd to exit, so there's some chance it's listening for connections.
* If init or loader exit unexpectedly, traceback.Peter Jones2009-11-052-27/+127
| | | | | | | | | | | Currently if init exits unexpectedly, we get "Kernel panic - not syncing: Attempted to kill init!" and then a useless kernel traceback (hey, it went through do_exit()!, who'd'a thunk?) but nothing meaningful about what went wrong in init. This patch attempts to rectify this situation by adding the traceback handler from loader to init, as well as triggering it (in both loader and init) if we exit without going through a path we've specifically marked, by using doExit() instead of exit(3) or _exit(2), as not needing a traceback.
* Fix the vim magic in this file to work.Peter Jones2009-11-051-1/+2
| | | | | | This fixes the vim magic to be something it'll actually do something with. I probably got it wrong the first time, but hey, better usable than not.
* Correctly initialize modopts in loader (#531932).David Cantrell2009-11-041-10/+12
| | | | | | Under certain conditions, modopts is never initialized. This patch fixes up that while preserving the existing functionality introduced with 29e18c35.
* Only try to split proxy commands out if there's actually one specified.Peter Jones2009-11-031-2/+6
| | | | | If we try to regexec() on NULL it segfaults, so let's try not to do that...
* Consolidate the parsing of nfs: locations for ks= and stage2= (#529197)Stijn Hoop2009-11-033-28/+35
| | | | | | Makes sure that NFS options can be used in both parameters. Previously this was only parsed for ks= and not for stage2= using two separate code paths. Move this to a shared function in loader/nfsinstall.c.
* Remove support for IUCV networking devices on s390. (#531494)David Cantrell2009-10-281-170/+58
| | | | | CONFIG_NETIUCV has been disabled in the kernel, so remove support for these devices during installation.
* Add sshd support for non-s390 platforms.Peter Jones2009-10-281-1/+1
| | | | | | This leaves the sshd support on the s390 intact and functional (testing needed), and at the same time add the ability to ssh in and get a terminal on non-s390 platforms.
* Re-enable running udevadm.Chris Lumens2009-10-261-2/+0
| | | | This snuck in as part of another commit when it shouldn't have.
* Reintegrate reipl to trigger reboot/halt on s390x correctly. (#528380)Steffen Maier2009-10-161-2/+9
| | | | | | | | | | | Because of 8bc669549ddcc7040a5ba56b8afafa18684bac4e and maybe others, triggering reipl at the end of anaconda is no longer sufficient to be hit in all different anaconda install paths. Therefore, move the triggering right after having configured reipl in sysfs and do all in one place. Now loader must no longer immediately relay SIGUSR1/2 to init, since anaconda is not nearly finished with installation, but only remember the state requested by anaconda. Only right before loader terminates, it kills init with the remembered state to really reboot/halt.
* Make LOADER_FLAGS_NOSHELL default also for s390x not just s390 (#527063)Steffen Maier2009-10-151-2/+2
| | | | | | | | | | This has always been broken for s390x since it was introduced in 67a5a0afb3100e7986ce127b3c2684e01c97304e. The fix ensures that both s390 and s390x do not start a second shell on the console that collides with init=linuxrc.s390 blocking on console input and potentially providing a rescue shell after hitting return. Apparently LOADER_FLAGS_NOUSB has gone meanwhile but nobody noticed since this code path referencing it was never compiled on s390x.
* Adapt standalone shutdown to nokill changes so s390x can use it. (#528380)Steffen Maier2009-10-151-3/+8
| | | | | | | | | The support for boot option "nokill" in shutdown by ade8105a8515afee24220fd0cc1199e30c1c3142 did not adapt the standalone version of shutdown (#ifdef AS_SHUTDOWN) so doKill was always 0. Introduce a new command line option "--nokill" and otherwise default to doKill=1. While we're at it also provide a command line option "-P" for poweroff.
* Write LAYER2 and PORTNO correctly as parts of OPTIONS to ifcfg for s390xSteffen Maier2009-10-151-5/+10
| | | | | | | | | | | | | | | | | | On s390x, linuxrc.s390 brings up the network completely and writes an NM compliant /etc/sysconfig/network-scripts/ifcfg-<DEVICE>. Apparently, loader never hits STEP_IP or writeEnabledNetworkInfo(). The latter would convert LAYER2= and PORTNO= into OPTIONS="". Since it does not seem to be called, linuxrc.s390 now writes OPTIONS="". Fixed readIfcfgContents() to not swallow OPTIONS which contains '='. With this in place, network.py writes a correct ifcfg to /mnt/sysimage. This patch completes ifcfg production on s390x of the following commits: 25f58fe7c701c453d39d2a9c5c0850eefce07f76 9249e40f42ffbbdcf42cd1caad72e3d622c7a75b 5f0fcf6688d08f83826c2892bb9fc97d6b4d7dd0 81163960a6137d39a5f2082e9f8d9f0a7b2ada57 This patch is intended for both devel and F-12.
* Support upgrading when the language isn't in lang-table (#528317).Chris Lumens2009-10-131-0/+2
|
* Don't run 70-anaconda.rules on an installed system (#527781).Chris Lumens2009-10-132-0/+4
| | | | | | | | The livecd includes anaconda, for obvious reasons. However this means that 70-anaconda.rules will be installed on the finished system and running our rules has some unusual side effects. While we slowly push our udev rules changes into upstream packages, this should make them a no-op on the installed system.
* Fix a segfault when stage2= boot parameter and kickstart url method is used ↵Radek Vykydal2009-10-121-3/+2
| | | | | | (#524417). Concerns cases when stage2= can't be parsed as http or ftp url.
* Stop /lib/udev/rules.d/65-md-incremental.rules from messing with mdraid setsHans de Goede2009-09-301-0/+5
| | | | | | | | | | Touch /dev/.in_sysinit, as that stops /lib/udev/rules.d/65-md-incremental.rules from messing with mdraid sets. This patch adds the touching twice, once to our own init, for when running as standalone installer, and once in python for when running from a livecd, to stop the udev trigger "block" we do will cause /lib/udev/rules.d/65-md-incremental.rules to trigger in the livecd case.
* Wait for mdraid arrays to become clean before reboot / haltHans de Goede2009-09-301-0/+5
| | | | | | | With the new external metadata mdraid support we need to wait for raid sets to become clean (iow for mdmon to have finished updating the metadata) before rebooting. Otherwise we get a full array sync on the first boot after installation.
* Pass the proxy config information to stage2.Chris Lumens2009-09-301-0/+32
| | | | | | | We pass the URL and port as the --proxy= argument, and any authentication information through /tmp/proxy. This file will be removed by stage2. The reason for this song-and-dance is so passwords aren't on the anaconda command line.
* Add support for proxies to the command line.Chris Lumens2009-09-301-0/+5
|
* Add proxy support to kickstart in the loader.Chris Lumens2009-09-303-9/+12
|
* Add a function to split up a proxy parameter into its parts.Chris Lumens2009-09-302-1/+47
| | | | | | | | | The proxy parameter to be used on the cmdline and in kickstart is of the form: proxy=[protocol://][user[:password]@]host[:port] This is also used as url --proxy= and repo --proxy=.
* libcurl supports https in addition to http, so change our tests.Chris Lumens2009-09-305-6/+6
|
* getHostAndPath is only used by the nfs code, so move it.Chris Lumens2009-09-303-69/+35
| | | | | Also, rip out all the username/password stuff too. That's not needed for nfs.
* Add initial loader UI support for proxies (#125917, #484788, #499085).Chris Lumens2009-09-303-8/+92
|
* We no longer need our own FTP/HTTP protocol support code.Chris Lumens2009-09-304-892/+1
|
* Get rid of the convertURL/UI functions, make iurlinfo just store a string.Chris Lumens2009-09-302-148/+7
|
generated by cgit (git 2.34.1) at 2025-08-31 14:38:19 +0000