| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
|
|
|
| |
The merge.pl script and initial configuration files for ppc, ppc64 x86
and x86_64 are from RHEL6. They will be replaced as development
progresses.
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
|
| |
#508362
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
CONFIG_AUDIT is already enabled on the most common architectures. It is
ancient, yet it is required by SELinux. Lets enable it on all
architectures for SELinux compatibility.
It is not clear that ./arch/ppc64/kernel-2.6-pSeries and
./arch/ppc64/kernel-2.6.g5 even still work. I am omitting them from this
commit because it is not clear to me that they even still work. Someone
with appropriate hardware will need to update the config file(s).
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
|
|
| |
Recent versions of systemd-udev require CONFIG_FHANDLE=y and provide no
fallback for kernels that lack it.
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
|
|
| |
<axs@gentoo.org>.
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Jorge Manuel B. S. Vicetto (jmbsvicetto) <jmbsvicetto@gentoo.org>
|
|
|
|
| |
Signed-off-by: Jorge Manuel B. S. Vicetto (jmbsvicetto) <jmbsvicetto@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previous commit 76e801c8 did not actually go far enough.
The existing install from kerncache code that is normally used with
passing --install used a form of kernel-${KNAME}-${ARCH}-${KV}.
Instead of just the kernel-${ARCH}-${KV} we were using in commit
76e801c8, or kernel-${KV} that we used before that, let's use the same
naming per the regular path.
Also add safety checks that the kernels exist, as you could get an empty
tarball otherwise!
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
|
|
|
|
|
|
| |
If you pass --minkernpackage=X, --modulespackage=X or --kerncache=X with
a relative path, then it's creates the output file in a temporary
working dir that is later removed during cleanup.
Resolve the paths during argument parsing to prevent this.
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
|
|
|
|
| |
The files are not created in the PWD when genkernel run without
specifying absolute paths for those arguments and I need to debug where
they are actually going.
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
|
|
|
| |
Move DISTDIR to /usr/share/genkernel/distfiles: /var/cache might be on
tmpfs and wiped each boot.
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Jorge Manuel B. S. Vicetto (jmbsvicetto) <jmbsvicetto@gentoo.org>
|
|
|
|
|
|
| |
(kerncache).
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
|
|
|
|
|
| |
- Improve the debug-cleanup code to use isTrue and match other code
style.
- add explicit return 0 after debug-cleanup to avoid passing a false
negative failure code up to a higher level function.
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
| |
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
|
|
| |
Clean up LVM install to work with non-root and handle errors.
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
|
|
|
|
| |
This reverts commit 2fa0a9beaa069a61f907f0b06c9bd6abeddb1ca7.
Signed-off-by: Richard Yao <ryao@gentoo.org>
Conflicts:
defaults/initrd.scripts
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Our switch to busybox modprobe broke ZFS module loading where busybox
modprobe would load two modules and then fail. Limited developer time
resulted in a hack being put into place to repeat modprobe until ZFS
appeared. However, this was never a real long term solution.
Recent analysis with strace suggests that loading two modules corrupts
busybox's current working directory inside the kernel. Consequently,
subsequent tests where absolute paths were used instead of relative ones
made the problem disappear.
Modifying busybox to use full paths when loading modules makes module
loading work on all affected kernels. While the long term plan is to fix
the kernel, this workaround will be needed indefinitely for affected
kernels, even after mainline Linux is fixed.
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
|
|
|
|
| |
When LOGLEVEL is unset or empty then genkernel loops endlessly in
gen_funcs.sh.
The solution is to set a proper default loglevel in $USRBIN/genkernel.
Signed-off-by: Martin Dummer <martin.dummer@gmx.net>
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
|
|
|
|
|
|
| |
There is a mistake in the parameter check of function prompt_user()
which causes to exit the initramfs unconditionally when the function is
called with *correct* number of parameters (which then causes a kernel
panic)
This patch puts the parameter check into a readable if..then..fi block.
Signed-off-by: Martin Dummer <martin.dummer@gmx.net>
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
| |
Signed-off-by: Martin Dummer <martin.dummer@gmx.net>
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
|
|
|
| |
This is meant to facilite automatic detection of root filesystems that
require additional support. The main one being ZFS, but btrfs will also
qualify once support has been added.
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
|
|
|
|
| |
ZFS pool import on non-exported pools is only permitted when the hostid
of the system matches the host id of the last pool to import it. Store
the hostid of the system to ensure that it matches at boot.
Signed-off-by: Richard Yao <ryao@gentoo.org>
|
|
|
|
| |
Signed-off-by: Richard Yao <ryao@gentoo.org>
|