Commit 3a054014e8 replaced our modprobe
with busybox's modprobe. Unfortunately, busybox's modprobe appears to be
unable to properly load modules with more than 1 level of dependencies.
The zfs and zpool commands will invoke modprobe if /dev/zvol is missing,
which concealed this problem. However, this caused problems because some
invocations would fail and under certain circumstances, init would be
killed, causing a kernel panic. This issue was made clear by commit
c812c35100771bb527f6b03853fa6d8ef66a48fe, which ensured that the zpool
and zfs commands were not run until the ZFS module was loaded.
busybox modprobe's failure to load module dependencies correctly appears
to occur because busybox modprobe does not wait until until a module is
loaded before loading a module that depends on it, which is a race. It
would be best to correct this race by waiting until the module has
properly loaded, but it is not clear that the race is the only thing
going wrong and developer time is a premium.
We implement a workaround by modifying the busy loop added in the
previous commit to explicit call `modprobe zfs` on each iteration. While
the first few calls fail due to bugs in busybox modprobe, it will
eventually work, after which each call is a noop. This lets us keep
looping until either the loop exit condition that /dev/zvol exist is
reached or the 5 second timeout is reached.
Once the busybox modprobe issue is fixed, this workaround should be safe
to revert.
Signed-off-by: Richard Yao <ryao@gentoo.org>
udev may still be processing rules and this can cause very bad
things. For instance, modules_load may have loaded an usb host
controller driver and we must wait for the udev rules to terminate.
However, this may lead to other race conditions, but we have
observed that adding scandelay=n where n >= 5 actually fixes the
issue of booting off USB under certain scenarios.
"quiet" is already used by the kernel to avoid printing messages on the
console unless they are errors or warnings. Genkernel should do the same
wrt its initramfs initialization code.
This has also the advantage of improving the boot speed.
scandelay=<secs> should be always preferred. Moreover, nowadays
automatically sleeping waiting for USB storage to come up in
the way we were used to do it doesn't work as one expects, because
USB is very common and USB storage devices are very likely to be
present on a system.
Furthermore, the initialization code after setup_slowusb is already
waiting for devices to come up.
Nearly every general documentation including the one in the
kernel tree and the scripts for GRUB2 all expects this flag
to work.
Signed-off-by: Peter Hjalmarsson <xake@rymdraket.net>
If INIT_OPTS='' is not specified before the argument loop, scoping rules
will prevent it from being seen by the switch_root invocation.
Signed-off-by: Toralf Förster <toralf.foerster@gmx.de>
Signed-off-by: Richard Yao <ryao@gentoo.org>
Clean up documentation and functionality of "part" argument to
initramfs. It is meant to imitate the same option to the kernel.
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
We do not support ramdisk, only initramfs nowdays.
So init= is a dead command, and we may as well have it do the same as
it does for dracut and in the Linux kernel documentation.
(Use rdinit to choose the initramfs init file)
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
Somewhere in the mists of time we lost part of the multipath patch, so
we were bundling the userspace, but not the kernel modules, or ever
using it.
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
Running `mdev -s` will ensure that device nodes are properly initialized.
It does not function as a netlink hotplug daemon. We must run it after
the modules are loaded to ensure that /dev/zfs has been created before
userland programs attempt to use it.
Effective for genkernel's modprobe script or busybox "not small"
modprobe.
Signed-off-by: Dzianis Kahanovich <mahatma@eu.by>
Updated-by: Richard Yao <ryao@cs.stonybrook.edu>
Make ZFS only import the specified pool
Set zfsutil when mounting non-legacy rootfs
Check for /sbin/zpool in addition to /sbin/zfs
Print information messages when ZFS pools are imported
Prompt user when mounting a ZFS filesystem without ZFS support
Verify that <dataset> in real_root=ZFS=<dataset> is a filesystem
Dogfooding of the /usr mount code with the matching OpenRC change to
make the mounts RW revealed some bugs and gotchas in our prior code, now
fixed and verified to work.
Signed-off-by: Robin H. Johnson <robbat2@gentoo.org>
instead. At the same time, mount cdrom into /mnt/cdrom instead of
/newroot/mnt/cdrom (which is now just a bind mount), this avoids
losetup to expose unavailable paths inside the live system, breaking
mkfs.btrfs (next upstream version, which does silly things with
/proc/mounts).
This makes the ramdisk always mount some kind of /dev,
and then try to move it to the new root.
This unbreaks handling of nodes created pre-udev likefor example lvm does.
This can be considered good practice, likely less chance that nodes
created in the ramdisk disapperes when switching root.
Some applications (mostly debugging/benchmarking) also needs this to
survive the switch_root.
Change vgscan and vgchange so they are done in one go.
Faster since less work for the lvm command.
vgscan is still needed for some cases like dmcrypt where
/etc/lvm/cache is created, but not updated with the unlocked devices.
Older systems featured full path to tty inside console=. This
is no longer valid and causes issues with fbcondecor. The new
expected format of console= is tty name with the directory part
stripped off: console=tty1 for example.
This patch has been in sabayon's genkernel for <2 years, but
should be tested out on non-standard environments.
RAID0 btrfs volumes can be detected only if btrfsctl is started.
Since embedding extra executables in initramfs is already possible,
add minimal btrfs support inside linuxrc.
There were several places where a mixture of sdelay and "sleep 10"
were called without worrying too much, resulting in slower boot
time, mainly. This commit should fix the mess and make some order
in the slowusb saga.
Fabio Erculiani
Richard Yao
Peter Hjalmarsson
Toralf Förster
Robin H. Johnson
Dustin Frisch
Richard Yao
Dzianis Kahanovich
Kenny Cheng
Sebastian Pipping
Christian Kruse
craig
devsk
Andrew Gaffney