| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Finally, no more libdhcp.
This is the first set of changes to take anaconda over to
the wonderful world of NetworkManager. We are no longer
linking with libdhcp to do interface configuration. NM
is started early in the installation and opens the door
to things like WPA installation support and things like
that.
|
| |
|
| |
|
|
|
|
| |
#254032)
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
'RFC 2461' from the text for the automatic neighbor discovery
option.
* loader/net.c (configureTCPIP): Use DHCPV6_METHOD_STR for the
DHCP on IPv6 checkbox.
* loader/net.h: Define DHCPV6_METHOD_STR. Remove 'RFC 2461' from
the AUTO_METHOD_STR macro.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
* loader2/hdinstall.c: Whitespace consistency, loaderData->method
is now an int.
* loader2/loader.c (doLoaderMain): loaderData->method is now an
int, pass methodNum to readNetConfig().
* loader2/loader.h: char * method -> int method.
* loader2/method.c: Set loaderData->method to appropriate METHOD_
value since it's now an int.
* loader2/method.h: Add METHOD_* macros.
* loader2/net.c (cidrCallback): Dialog wording change.
* loader2/net.c (readNetConfig): Take methodNum as a param, pass
methodNum to configureTCPIP().
* loader2/net.c (configureTCPIP): Take methodNum as a param, if
user selects METHOD_NFS and unchecks IPv4, display error dialog
and ask user to try again.
* loader2/net.c (kickstartNetworkUp): Pass methodNum to
readNetConfig().
* loader2/net.h: Prototype updates.
* loader2/nfsinstall.c: loaderData->method is now an int.
* loader2/urlinstall.c: loaderData->method is now an int.
|
|
|
|
|
|
| |
bit.
* loader2/net.{c,h}: Removed NETWORK_DEBUG mode and send data to the
logging function via debugNetworkInfo().
|
|
|
|
| |
the python side of things.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- The manual entry screen makes sense. There is only ONE gateway entry
box, for instance.
- The IPv4 entry field is long enough for an IPv4 dotted quad IP.
- The IPv6 entry field is long enough for an IPv6 colonated IP.
- IPv4 netmask can be specified in CIDR or dotted-quad notation.
- Gateway and nameserver can be IPv4 or IPv6 addresses.
- IPv6 netmask must be entered in CIDR notation.
- The back button will take you to the first network config screen,
which is accomplished by the use of the poorly named configureTCPIP()
and manualNetConfig() functions I added to handle this junk.
- fillInIpInfo() was wronged, so I removed it and replaced it with
correct stuff (wrong because libdhcp is wronging us now).
- cidrCallback() function added to make sure you have entered a valid
cidr mask.
- ipCallback() rewritten to handle new input validation requirements.
What am I working on? Making sure all of the widgets are connected to the
right things so manual configuration works. Making sure input validation
is done. Etc.
This concludes the micro-blog CVS commit message.
|
|
|
|
|
|
|
|
|
| |
Phase 1 of the great network overhaul for loader2, secret of the ooze.
Users can enable or disable IPv4 and IPv6 as they see fit. They can do
manual network configuration (though I haven't hooked up the bits for
IPv6 yet). Phase 2 will include more breakage and more features.
If you have ideas for stupid options, let me know.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
corru���r���z]��@��Z������(:�G5S�s"����.�>�5��c��(��*�m�|[�m���d��ZZ)���wI2�tu& �>�5�zG�
emr�����E�D0�Oi)h�!�CC�����U��j1+d�R`V�����
SD
|���=T^s0s�.��qE������*��[,�������[%<�`�+�=G�E���|�R���/.%M���Y��P��Ne���Qn��?iQ�8Y�h�8��d{��H��AU��g�9�$�van��u�( ��\j�X��{S,4���^^ #W�J!�V�I���)�$��9*�&bL2����KYw1HWY-��g���y��i��������2_a,M1�b���)Pr}��X��ga=Ln���'��q
}p~h1(;�.�S��C���y%���6U1�w�T[��>B�����\&�~Xh������)�5ption as well as some
housekeeping tasks:
* loader2/net.c (setupInterface): Removed and moved those two lines to
the calling function (configureNetwork).
* loader2/net.c (readNetConfig): Spelling error fix. Calculate IPv4
broadcast if user did a static IP config, close any possibly open
dhcp_nic handle if doing a static IP config.
* loader2/net.h: Removed setupInterface prototype.
|
|
|
|
|
|
|
|
|
|
| |
We don't need to set netDev.useipv6 anymore because flags are global.
* loader2/net.h: Removed useipv6 from networkDeviceConfig struct.
* loader2/net.c (waitForLink): Note which interface we are waiting on.
* loader2/net.c (readNetConfig): Initialize newCfg (#194882).
* loader2/net.c (doDhcp): Read global flags rather than useipv6.
|
|
|
|
| |
to it) all over the place.
|
|
|
|
|
|
|
| |
Set useipv6 in the netDev structure depending on 'noipv6' param.
* loader2/net.c: Sure can't make up my mind. Put doDhcp back so that I
can conditionally do the IPv6 thing or not from one place.
* loader2/net.h: doDhcp() prototype.
|
|
|
|
|
| |
catch libdhcp log messages. libdhcp-1.4 removes the forced debugging output
that we are seeing right now.
|
|
|
|
|
|
|
|
| |
convert loader to using the libdhcp API rather than the pump one (it's hard
to tell which one is worse). And there's some libdhcp noise right now, but
just ignore it.
You can ping6 cutlet now (specify cutlet's IPv6 address) and it will work.
|
|
|
|
|
|
| |
Ok, so what exactly are we looking at here? A slapped-together pile of stuff that uses libdhcp instead of pump. I call it a work in progress. Right now we have support for regular IPv4-based installs. IPv6 is sort-of halfway there. I'm still wedging it in to place (i.e., dodging sigsegvs). I wanted to get an IPv4-libdhcp-move committed now so that it's in CVS. The pump-compatible API in libdhcp is mostly useless, so I'm moving things from using the core-dumping pump API functions to just using the libdhcp stuff directly. I had to back out a lot of IPv6 stuff I did in the interim.
Please avoid making huge changes to the loader2 network code because I'm still tearing through that.
|
| |
|
|
|
|
|
|
|
| |
qeth can hav eth, hsi and tr interfaces, lcs can have eth and tr interfaces,
too. We can't tell apart if p.e. eth0 is qeth or lcs, but need to know it for
sysfs configuration. Remove TYPE=QETH hack and use a more generic NETTYPE
instead
|
|
|
|
| |
instead, required for IUCV setup
|
| |
|
| |
|
|
|
|
|
|
|
| |
- read additional parameters for ccw devices from netinfo and
write them to ifcfg-xxx
(mtu and ptpaddr can now be removed from pump.h)
|
|
|
|
|
| |
a little bit of work so that back/next will work properly through the wireless
config. also things don't get preserved for the actual boot
|
| |
|
|
|
|
|
| |
list everywhere. stop passing around kd. link with libpci_loader (built
with diet) on diet arches to avoid dirent incompatibilities
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
summary of looking through the diff is
* clean up warnings, we build with -Wall -Werror here too
* product.img stuff
* max logical partitions enforcement
* 1 TB max fs size
* ethtool stuff
* autopart in kickstart
* driver disk fixes
* RHEL upgrade stuff
* network driver disks
* variant pkgorder/tree splitting
|
| |
|
| |
|
|
|
|
|
| |
kickstart basically works now. need to hook up sources other than nfs for
installing and getting kickstart file from other than floppy and file
|
|
still a lot to do, but this gets to stage2 for cd, nfs, http, and ftp booting
from either the cd initrd or the pxe initrd
|