dracut/backport-fix-shutdown-add-timeout-to-umount-calls.patch
hongjinghao 2663ebded8 backport patchs from upstream
(cherry picked from commit e37aecb42764d47b923dd3280385bd5438f466b6)
2024-02-22 14:56:32 +08:00

150 lines
5.5 KiB
Diff

From 38ef31939056dea30017d782988ba1a46df4784a Mon Sep 17 00:00:00 2001
From: Petr Pavlu <petr.pavlu@suse.com>
Date: Tue, 9 Feb 2021 13:13:25 +0100
Subject: [PATCH] fix(shutdown): add timeout to umount calls
When terminating a system, the shutdown module attempts to unmount all
file systems from under /oldroot. This reaps remaining file systems that
systemd cannot unmount and detaches /oldroot itself.
In case that running umount for some file system repeatedly fails, the
module reports this error and continues the processing in order to
shutdown the system. This handles a condition when the umount command
actually terminates but it can happen in some cases that it waits
indefinitely.
An example with NFS mounts:
# mount -t nfs 192.168.0.1:/srv/nfs/dir /mnt/nfs
# mkdir /mnt/nfs/dir2
# mount -t nfs 192.168.0.1:/srv/nfs/dir2 /mnt/nfs/dir2
# touch /mnt/nfs/dir2/file
# systemd-run -pKillMode=none -pSendSIGKILL=no tail -f /mnt/nfs/dir2/file
Running as unit: run-r367825c967ca4d88a793ae4793c02f8b.service
# systemctl poweroff
The invoked tail command escapes normal termination by systemd and
prevents stopping mnt-nfs.mount and mnt-nfs-dir2.mount as it makes the
mounts busy. Systemd then again attempts to unmount these file systems
in systemd-shutdown but this fails as well. The utility tries to unmount
/mnt/nfs/dir2 but the kernel waits indefinitely doing a path lookup for
/mnt/nfs because network is no longer available at that point. The
systemd-shutdown gives up after 90 seconds. Finally, the control is
transferred to dracut which tries to unmount the file systems in the
same way and ends up indefinitely waiting on umount to finish.
This situation causes that the system hangs during shutdown. The patch
improves the shutdown module to add a timeout of 90 seconds for the
umount operation and continue with the shutdown if it gets reached,
similarly to what systemd-shutdown does.
Conflict:NA
Reference:https://github.com/dracutdevs/dracut/commit/38ef31939056dea30017d782988ba1a46df4784a
---
dracut.cmdline.7.asc | 6 ++++
modules.d/99shutdown/module-setup.sh | 2 +-
modules.d/99shutdown/shutdown.sh | 44 +++++++++++++++++++++++-----
3 files changed, 43 insertions(+), 9 deletions(-)
diff --git a/dracut.cmdline.7.asc b/dracut.cmdline.7.asc
index 0951c0e0..4e598b6b 100644
--- a/dracut.cmdline.7.asc
+++ b/dracut.cmdline.7.asc
@@ -167,6 +167,12 @@ Misc
specify the controlling terminal for the console.
This is useful, if you have multiple "console=" arguments.
+**rd.shutdown.timeout.umount=**__<seconds>__::
+ specify how long dracut should wait for an individual umount to finish
+ during shutdown. This avoids the system from blocking when unmounting a file
+ system cannot complete and waits indefinitely. Value '0' means to wait
+ 'forever'. The default is 90 seconds.
+
[[dracutkerneldebug]]
Debug
~~~~~
diff --git a/modules.d/99shutdown/module-setup.sh b/modules.d/99shutdown/module-setup.sh
index dfd6caa2..ff9b7734 100755
--- a/modules.d/99shutdown/module-setup.sh
+++ b/modules.d/99shutdown/module-setup.sh
@@ -14,7 +14,7 @@ depends() {
# called by dracut
install() {
local _d
- inst_multiple umount poweroff reboot halt losetup stat sleep
+ inst_multiple umount poweroff reboot halt losetup stat sleep timeout
inst_multiple -o kexec
inst "$moddir/shutdown.sh" "$prefix/shutdown"
[ -e "${initdir}/lib" ] || mkdir -m 0755 -p ${initdir}/lib
diff --git a/modules.d/99shutdown/shutdown.sh b/modules.d/99shutdown/shutdown.sh
index a30a126f..90a0b9e1 100755
--- a/modules.d/99shutdown/shutdown.sh
+++ b/modules.d/99shutdown/shutdown.sh
@@ -46,28 +46,56 @@ warn "Killing all remaining processes"
killall_proc_mountpoint /oldroot || sleep 0.2
+# Timeout for umount calls. The value can be set to 0 to wait forever.
+_umount_timeout=$(getarg rd.shutdown.timeout.umount)
+_umount_timeout=${_umount_timeout:-90s}
+_timed_out_umounts=""
+
umount_a() {
+ local _verbose="n"
+ if [ "$1" = "-v" ]; then
+ _verbose="y"; shift
+ exec 7>&2
+ else
+ exec 7>/dev/null
+ fi
+
local _did_umount="n"
while read a mp a || [ -n "$mp" ]; do
- if strstr "$mp" oldroot; then
- if umount "$mp"; then
- _did_umount="y"
- warn "Unmounted $mp."
- fi
+ strstr "$mp" oldroot || continue
+ strstr "$_timed_out_umounts" " $mp " && continue
+
+ # Unmount the file system. The operation uses a timeout to avoid waiting
+ # indefinitely if this is e.g. a stuck NFS mount. The command is
+ # invoked in a subshell to silence also the "Killed" message that might
+ # be produced by the shell.
+ (set +m; timeout --signal=KILL "$_umount_timeout" umount "$mp") 2>&7
+ local ret=$?
+ if [ $ret -eq 0 ]; then
+ _did_umount="y"
+ warn "Unmounted $mp."
+ elif [ $ret -eq 137 ]; then
+ _timed_out_umounts="$_timed_out_umounts $mp "
+ warn "Unmounting $mp timed out."
+ elif [ "$_verbose" = "y" ]; then
+ warn "Unmounting $mp failed with status $ret."
fi
done </proc/mounts
- losetup -D
+
+ losetup -D 2>&7
+
+ exec 7>&-
[ "$_did_umount" = "y" ] && return 0
return 1
}
_cnt=0
while [ $_cnt -le 40 ]; do
- umount_a 2>/dev/null || break
+ umount_a || break
_cnt=$(($_cnt+1))
done
-[ $_cnt -ge 40 ] && umount_a
+[ $_cnt -ge 40 ] && umount_a -v
if strstr "$(cat /proc/mounts)" "/oldroot"; then
warn "Cannot umount /oldroot"
--
2.19.1