all: Refactor the whole kdump/pstore folder setting

This is somewhat a big refactor. The early design of kdump/pstore was
meant to support the A/B scheme of Steam Deck and a dracut-based initrd
only. In this scheme, we had a DEVNODE (like nvme partition or a devlink)
that represented the device node to be mounted early in boot when kdump
was in use. Also, we had a folder defined in the config file on top of
such dev node, and a temporary file used to "carry" the composition of
the full kdump path across boot time scripts. Yeah, kinda complex setup.

We are now moving to a more generic approach, and for that, the design
choice was a more convenient/simple one for the common cases, that
requires some operations to properly work on SteamOS dracut-based initrd.
Now we have only a single path on config file, which should be accessible
in a R/W filesystem by both scripts executed in the systemd service. No
devnode information or temp file is used anymore.

But with that, comes the need of discovering the proper devnode and base
folder for kdump'ing early in boot, from the initrd. Using the findmnt
tool we manage to derive all the necessary data during the initrd
preparation phase. Also, while at it we manage to fix an "inconsistency"
of our dracut initrd creation script: installkernel() should be responsible
to deal with DRM modules removal, not install().

On top of this (already big) change, now our dracut initrd excludes not
only amdgpu driver/FWs, but radeon, nvidia and i915 as well. And due to
our refactor of the mount point information (using findmnt to collect info
during dracut initrd creation), we also allow now arbitrary filesystem
drivers to be included, i.e., we don't hardcode/limit for ext4 only.

Again, mea culpa for not splitting this in multiple atomic/simple commits,
the burden to keep a pretty git log is starting to consume precious time.

Signed-off-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
This commit is contained in:
Guilherme G. Piccoli
2022-12-28 13:15:36 -03:00
parent f6fa4976d1
commit a3ab8c421b
5 changed files with 57 additions and 52 deletions

View File

@ -22,17 +22,12 @@ PSTORE_RECORD_SZ=2097152
#
#
# Mount-related options
# MOUNT_DEVNODE is the desired mount-point unique link in the /dev/disk
# directory link; this is used to derive the numerical devnode for kdump,
# since the link is not present so early in the system boot. MOUNT_FOLDER
# should live in the filesystem of the devnode (and will be created if
# doesn't exist); this is the place in which both pstore and kdump logs
# will be stored, as well as the kdump initrd and some ancillary data.
# Finally, MNT_TMP is just a temporary file that carries the full path
# of the mounted folder across boot-time scripts.
MOUNT_DEVNODE="/dev/disk/by-partsets/shared/home"
MOUNT_FOLDER="/.steamos/offload/var/kdump"
MNT_TMP="/tmp/.kdump_load.mnt"
# MOUNT_FOLDER is the the place in which both pstore and kdump logs will
# be stored, as well as the kdump initrd and some ancillary data. This
# directory should be in an accessible filesystem (read/write) and if such
# folder doesn't exist, it'll be created.
MOUNT_FOLDER="/home/.steamos/offload/var/kdump"
#
#
# Kdump controlling settings

View File

@ -13,6 +13,13 @@
# output for the user.
#
# Bail out in case we don't have a vmcore, i.e. either we're not kdumping
# or something is pretty wrong and we wouldn't be able to progress.
VMCORE="/proc/vmcore"
if [ ! -f $VMCORE ]; then
reboot -f
fi
# We have a more controlled situation with regards the config
# files here, since we manually added them in the initrd and
# the validation also happened there, during such addition,
@ -21,21 +28,13 @@ for cfg in "/usr/share/kdump.d"/*; do
. "$cfg"
done
VMCORE="/proc/vmcore"
KDUMP_TIMESTAMP=$(date -u +"%Y%m%d%H%M")
KDUMP_FOLDER="/kdump_path/${MOUNT_FOLDER}/crash/${KDUMP_TIMESTAMP}"
MOUNT_POINT="$(cat /usr/lib/kdump/kdump.mnt)"
BASE_FOLDER="$(cat /usr/lib/kdump/kdump.dir)"
KDUMP_FOLDER="/kdump_path/${BASE_FOLDER}/crash/${KDUMP_TIMESTAMP}"
# Bail out in case we don't have a vmcore, i.e. either we're not kdumping
# or something is pretty wrong and we wouldn't be able to progress.
#
if [ ! -f $VMCORE ]; then
reboot -f
fi
DEVN="$(cat /usr/lib/kdump/kdump.devnode)"
mkdir -p "/kdump_path"
if ! mount "${DEVN}" /kdump_path; then
if ! mount "${MOUNT_POINT}" /kdump_path; then
reboot -f
fi
@ -52,7 +51,7 @@ if [ "${FULL_COREDUMP}" -ne 0 ]; then
sync "${KDUMP_FOLDER}/vmcore.compressed"
fi
umount "${DEVN}"
umount "${MOUNT_POINT}"
sync
reboot -f

View File

@ -73,19 +73,10 @@ cleanup_unused_initrd() {
load_kdump_config
# Find the proper mount point expected for kdump collection:
DEVN_MOUNTED="$(findmnt "${MOUNT_DEVNODE}" -fno TARGET)"
# Create the kdump main folder here, as soon as possible, given
# the importance of such directory in all kdump/pstore steps.
MOUNT_FOLDER="${DEVN_MOUNTED}/${MOUNT_FOLDER}"
# In case the kdump main folder doesn't exist, create it
# here, as soon as possible.
mkdir -p "${MOUNT_FOLDER}"
echo "${MOUNT_FOLDER}" > "${MNT_TMP}"
sync "${MNT_TMP}"
# Notice that at this point it's required to have the full
# MOUNT_FOLDER, so this must remain after the DEVNODE operations above.
if [ "$1" = "initrd" ]; then
create_initrd
exit 0

View File

@ -4,39 +4,60 @@ check() {
}
installkernel() {
hostonly='' instmods ext4
load_kdump_config
# First clear all unnecessary firmwares/drivers added by drm in order
# to reduce the size of the minimal initramfs being created - kdump
# is an non-graphical environment. This should have been already done
# via dracut cmdline arguments, but play safe and delete here as well.
# Our list include the most common FWs/drivers (amd, i915, nvidia)
rm -rf "$initdir"/usr/lib/firmware/{amdgpu,i915,nvidia,radeon}
rm -rf "$initdir"/usr/lib/modules/*/kernel/drivers/gpu/drm/{amd,i915,nouveau,radeon}
FSMOD="$(findmnt -n -o FSTYPE --target "${MOUNT_FOLDER}")"
if [ -z "${FSMOD}" ]; then
logger "kdump: error on filesystem discovery"
exit 1
fi
hostonly='' instmods "${FSMOD}"
}
install() {
# A valid makedumpfile is essential for the kdump initrd creation.
if [ ! -x "$(command -v makedumpfile)" ]; then
logger "kdump: failed to create initrd, makedumpfile is missing"
logger "kdump: failed to create dracut initrd, makedumpfile is missing"
exit 1
fi
load_kdump_config
# First clear all unnecessary firmwares/drivers added by drm in order
# to reduce the size of the minimal initramfs being created. This
# should be already done via dracut cmdline arguments, but let's play
# safe and delete from here as well just in case.
rm -rf "$initdir"/usr/lib/firmware/amdgpu/
rm -rf "$initdir"/usr/lib/modules/*/kernel/drivers/gpu/drm/amd/*
# Install necessary binaries
inst date
inst sync
inst makedumpfile
mkdir -p "$initdir"/usr/share/kdump.d/
cp -LR --preserve=all /usr/share/kdump.d/* "$initdir"/usr/share/kdump.d/
# Copying kdump config/lib files is essential for a functional kdump.
cp -LR --preserve=all /usr/share/kdump.d/ "$initdir"/usr/share/
cp -LR --preserve=all /usr/lib/kdump/ "$initdir"/usr/lib/
# Determine the numerical devnode for kdump, and save it on initrd;
# notice that partset link is not available that early in boot time.
DEVN="$(readlink -f "${MOUNT_DEVNODE}")"
echo "${DEVN}" > "$initdir"/usr/lib/kdump/kdump.devnode
# Finally, we need to derive the proper place to save the dump from the
# config files, in a way that makes possible to mount it in early boot.
DEVNODE="$(findmnt -n -o SOURCE --target "${MOUNT_FOLDER}")"
if [ -z "${DEVNODE}" ]; then
logger "kdump: error on devnode discovery"
exit 1
fi
echo "${DEVNODE}" > "$initdir"/usr/lib/kdump/kdump.mnt
cp -LR --preserve=all /usr/lib/kdump/* "$initdir"/usr/lib/kdump/
TGT="$(findmnt -n -o TARGET --target "${MOUNT_FOLDER}")"
if [ -z "${TGT}" ]; then
logger "kdump: error on base folder discovery"
exit 1
fi
BASE_FLD="${MOUNT_FOLDER#*$TGT}"
echo "${BASE_FLD}" > "$initdir"/usr/lib/kdump/kdump.dir
inst_hook pre-mount 01 "$moddir/kdump-collect.sh"
}

View File

@ -1,7 +1,6 @@
load_kdump_config
MAIN_FOLDER="$(cat "${MNT_TMP}")"
rm -f "${MNT_TMP}"
MAIN_FOLDER="${MOUNT_FOLDER}"
if [ ! -d "${MAIN_FOLDER}" ]; then
logger "kdump: invalid folder (${MAIN_FOLDER}) - aborting..."