| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| ... | |
| |
|
|
| |
Signed-off-by: David Cantrell <dcantrell@redhat.com>
|
| |
|
|
|
|
|
|
| |
Conflicts:
scripts/mk-images
Signed-off-by: David Cantrell <dcantrell@redhat.com>
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Caveat #1: If the brand package doesn't exist, buildinstall will fail
Caveat #2: If the brand package has uppercase characters in it's name,
it needs to provide ${brand}-{release,logos} in lowercase or
yum thinks the package doesn't exist and buildinstall will fail
An additional patch may be necessary to:
1) check for ${brand}-{release,logos}
2) if those do not exist, fall back by setting the brand to "generic"
This will need to happen *after* the YUM configuration file has been
written, but before upd-instroot is executed.
|
| |
|
|
| |
Signed-off-by: David Cantrell <dcantrell@redhat.com>
|
| |
|
|
| |
Signed-off-by: David Cantrell <dcantrell@redhat.com>
|
| |
|
|
|
| |
Require pyparted >= 2.0.0 and be sure to save the _pedmodule.so
and parted Python modules.
|
| | |
|
| |
|
|
|
|
|
|
| |
Patch from Jeroen van Meeuwen (Fedora Unity)
<kanarip@fedoraunity.org>
NOTE: I would have done a git am, but it kept complaining about
the email address format so I'm merging it this way.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This patch replaces the iscsi.py code with new code using libiscsi, getting
rid of the awefull configfile mangling to set auth info hack, and in general
making it much cleaner.
Fixes:
-Report error when we cannot add a disk instead of silently failing (461830)
-Report error when trying to use iscsi from kickstart without having an
iscsiname directive first (463156)
-Better feedback while scanning iscsi disks
-Make ibft work with chap and 2-way chap
-Make ibft configured disks available for upgrades and in rescue mode
-Make ibft boot work even when iscsi-initiator-utils is installed after the
kernel in the transaction
Note:
libiscsi is available in Fedora now
|
| |
|
|
|
|
|
| |
Moves setuptime step into earliear stage to create files
with correct timestamp (prevent future time skew after boot).
Includes /usr/share/zoneinfo tree in stage 2, it is needed
when hw clock is in localtime (for hwclock --hctosys call).
|
| |
|
|
|
|
|
|
|
|
|
| |
We can use 'modinfo -F firmware' to query the modules that are
going into the initrd about what firmware they require. Some
modules don't (or can't) export that information, so they will
still need to be pulled in manually.
iwl4965 and iwl5000 don't exist anymore, so they were removed from
the case statement; qla2xxx exports modinfo data so it's not
necessary to pull its firmware manually.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Now that btrfs is in mainline, let's let anaconda play with
it too ;) We still need btrfs.ko in the kernel, and btrfs
support in e2fsprogs. I've got the latter patch submitted
upstream, and it's in rawhide now.
Note that the below is 100% totally untested and may even have
typos, my track record with python is not too good, sorry. :)
So review would be appreciated....
Updated to address Radek's review, as well as adding the requisite
witty (commandline-space-eating) boot option...
(And stopped trying to use parted fs identification -- katzj)
Signed-off-by: Eric Sandeen <sandeen@redhat.com>
|
| | |
|
| |
|
|
|
|
|
|
| |
Cracklib is now a non-compiled python module, as well as a
second compiled module. Kinda weird, but with both sets of files
it works.
Signed-off-by: Chris Lumens <clumens@redhat.com>
|
| |
|
|
|
| |
We have to look to where we've expanded the kernel in order to read
the module list sets to expand our module list.
|
| |
|
|
|
|
| |
(1) Remove the extra calls to cat.
(2) Remove an extra paren on the modules.block line.
(3) Use $version instead of $kernel.
|
| | |
|
| |
|
|
|
| |
Based on a patch by Tom Callaway <tcallawa@redhat.com>, this adds supporting
for using ntfs-3g (and thus fuse) to mount filesystems.
|
| |
|
|
|
|
|
|
| |
These (dejavu) fonts are likely done changing names, but more
are likely to change. All this is being worked on in rawhide
right now. But for alpha we want decent fonts.
Signed-off-by: Chris Lumens <clumens@redhat.com>
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
| |
'cp -a' considered harmful. jkeating brought to my attention today that
libgio-2.0.so.0 was not getting copying in to the initrd.img which is
required by libnm-settings-plugin-ifcfg-rh.so.
The problem was caused by my patch to mk-images which changed the plugin
instbin line to a 'cp -a' for the *.so files in /usr/$LIBDIR/NetworkManager.
We need to use the instbin function so that get_dso_deps() is run on the
plugin files and we pick up dependent libraries.
|
| |
|
|
|
|
|
| |
NetworkManager renamed libnm-settings-plugin-ifcfg-fedora.so to
libnm-settings-plugin-ifcfg-rh.so. To possibly/hopefully prevent such
failbuilds in the future, just copy in every .so from the
/usr/lib[64]/NetworkManager directory to the image.
|
| | |
|
| |
|
|
| |
Various packaging fixed from review (#225246)
|
| | |
|
| |
|
|
|
|
|
|
| |
Log file had a message telling us to use 'udevadm settle' instead of
'udevsettle' and that udevsettle support as argv[0] will be removed
in the future. This patch changes loader to call 'udevadm settle'.
[rawhide]
|
| |
|
|
|
|
| |
Use the modules.* files for finding modules of a type rather than the
modinfo and then pull in drm modules using it rather than
having to explicitly list them. (#446126)
|
| |
|
|
|
| |
"ld-*.*.*.so" matches against ld-2.8.90.so but not ld-2.9.so, so the
initrd in rawhide doesn't work today. Wooooo.
|
| | |
|
| |
|
|
|
|
|
| |
IBM would like us to provide mainframe users with a sample punchcard
reader script along with the other boot files on s390x.
I'd just like to say that again: punch cards.
|
| |
|
|
|
|
|
| |
We've backed this out on RHEL 5.3 for now because it was far more
complex than IBM had told us, so it's back to the drawing board
for this one. Keeping rawhide in sync with what's happening on
the RHEL side with this feature.
|
| | |
|
| |
|
|
|
|
|
| |
This reverts commit ff6f05a1a68f93c85170503c4e930b4c2828f2f9.
In the case of NM not working quite right, dhclient-script is still needed
if you want to run dhclient by hand on tty2.
|
| | |
|
| |
|
|
|
|
|
|
|
|
|
|
| |
IBM had requested that on s390x we add the cio_ignore=all parameter
to the default boot file. Well, cio_ignore all except for the 3270
terminal. The motivation for this change was to decrease device
discovery time on systems with thousands of subchannels.
The fix was only part of the solution. When you use cio_ignore to
disable everything, you need to explicitly free the device from
cio_ignore and then enable it using the magic of /sbin/chccwdev.
This patch adds the un-ignore capability to linuxrc.s390.
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Add modules for sd/mmc subsystem and support the block device name where we
need to do munging given the 'p' separator
|
| |
|
|
|
|
|
|
|
|
| |
Generating lang-name depended on old behavior of rhpl.translate for looking
in po/ for the translation files. The regular gettext module doesn't do this
which means we don't get translated language names unless you already
had anaconda installed. So use gettext.GNUTranslation to directly open
the .mo file
For LANG=C, we need to ensure that the output charset is set to UTF-8
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This reverts commit a6ba02f4099c40f3dd2c2783cf0bac2c1900ade4.
This currently breaks lang-names generation. We see this in the
build.log in koji:
PYTHONPATH="." python2.5 scripts/getlangnames.py > lang-names
Traceback (most recent call last):
File "scripts/getlangnames.py", line 37, in <module>
names[langs.localeInfo[k][0]] = cat.lgettext(langs.localeInfo[k][0])
File "/usr/lib/python2.5/gettext.py", line 368, in lgettext
return tmsg.encode(locale.getpreferredencoding())
UnicodeEncodeError: 'ascii' codec can't encode characters in position 0-5: ordinal not in range(128)
I'm just a simple back country hyperchicken, not a Unicode expert.
|
| |
|
|
|
|
|
|
| |
Generating lang-name depended on old behavior of rhpl.translate for looking
in po/ for the translation files. The regular gettext module doesn't do this
which means we don't get translated language names unless you already
had anaconda installed. So use gettext.GNUTranslation to directly open
the .mo file
|
| |
|
|
|
|
| |
Do not install dhclient-script, ifconfig, ip, or arping in the
initrd.img file. NetworkManager uses nm-dhcp-client.action for
its dhclient-script.
|
| | |
|
| | |
|
| |
|
|
|
|
|
| |
Add mk-images changes forgotten in previous commit:
Add support for iSCSI iBFT table to anaconda and use the configured targets
in it (#307761), forward port of rhel-5 patch from Martin Sivak, commit id:
96255750791e859dd06004fe1fbcc101a26e9e47
|
