Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

tests: fix fsck on boot on arm devices #9655

Merged
merged 10 commits into from
Nov 18, 2020
Merged

tests: fix fsck on boot on arm devices #9655

Changes from 4 commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
80b2208
Support arm devices
sergiocazzolato Nov 16, 2020
0120736
Update hot dirty byte is udpated
sergiocazzolato Nov 16, 2020
492853d
Support uc16 and uc18 for arm devices
sergiocazzolato Nov 17, 2020
2314645
Adding restore for test and making the code more simple
sergiocazzolato Nov 17, 2020
49ef844
revert dd to previous implemetation
sergiocazzolato Nov 17, 2020
2c73b4b
change skip -> seek
sergiocazzolato Nov 17, 2020
c610ea4
Merge branch 'master' into tests-fix-fsck-on-boot
sergiocazzolato Nov 18, 2020
c2501e5
Merge remote-tracking branch 'upstream/master' into tests-fix-fsck-on…
sergiocazzolato Nov 18, 2020
b5676fb
Umount /boot/uboot on arm devices for uc20
sergiocazzolato Nov 18, 2020
5ea1c54
Fix the scenario for uc20 on amd64
sergiocazzolato Nov 18, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 42 additions & 11 deletions tests/core/fsck-on-boot/task.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,24 +1,40 @@
summary: the boot base provides essential fsck programs

details: |
Snapd uses vfat on certain essential boot partitions, due to external
requirements imposed by the bootloader architecture. This test verifies that
the boot process is capable of detecting unclean vfat and fixing it before
such file system is mounted. This is an essential property to ensure
longevity of devices that rely on write to vfat to operate.

prepare: |
tests.cleanup prepare

restore: |
tests.cleanup restore

execute: |
unmount_vfat() {
if os.query is-core16; then
# Refer to the core 16 PC gadget for details:
# Refer to the core 16 gadgets for details:
# https://github.com/snapcore/pc-amd64-gadget/blob/16/gadget.yaml
umount /boot/efi
umount /boot/grub
# https://github.com/snapcore/pi2-gadget/blob/master/gadget.yaml
if [[ "$SPREAD_SYSTEM" == ubuntu-core-16-arm-* ]]; then
umount /boot/uboot
else
umount /boot/efi
umount /boot/grub
fi
elif os.query is-core18; then
# Refer to the core 18 PC gadget for details:
# Refer to the core 18 gadgets for details:
# https://github.com/snapcore/pc-amd64-gadget/blob/18/gadget.yaml
umount /boot/efi
umount /boot/grub
# https://github.com/snapcore/pi2-gadget/blob/18/gadget.yaml
if [[ "$SPREAD_SYSTEM" == ubuntu-core-18-arm-* ]]; then
umount /boot/uboot
else
umount /boot/efi
umount /boot/grub
fi
elif os.query is-core20; then
# TODO:UC20 The property of having to keep a mounted vfat at all time
# is not the most fortunate. Any power loss will result in a dirty
mvo5 marked this conversation as resolved.
Show resolved Hide resolved
Expand All @@ -30,8 +46,16 @@ execute: |
# on the first boot - subsequent boots do not mount it, because we will
# unset RecoverySystem in the modeenv, and it's not necessary for
# seeding anymore

# Refer to the core 20 gadgets for details:
# https://github.com/snapcore/pc-amd64-gadget/blob/20/gadget.yaml
# https://github.com/snapcore/pi-gadget/blob/20-arm64/gadget.yaml
if mountpoint /run/mnt/snapd >/dev/null; then
umount /run/mnt/ubuntu-seed/systems/*/snaps/snapd_*.snap
if [[ "$SPREAD_SYSTEM" == ubuntu-core-20-arm-* ]]; then
umount /run/mnt/ubuntu-seed/snaps/snapd_*.snap
else
umount /run/mnt/ubuntu-seed/systems/*/snaps/snapd_*.snap
fi
fi
umount /var/lib/snapd/seed
umount /run/mnt/ubuntu-seed
Expand All @@ -41,6 +65,15 @@ execute: |
fi
}

if os.query is-core16 || os.query is-core18; then
LABEL=system-boot
elif os.query is-core20; then
LABEL=ubuntu-seed
else
echo "unknown core system, please update test"
exit 1
fi

case "$SPREAD_REBOOT" in
0)
echo "We can corrupt the boot partition"
Expand All @@ -54,7 +87,7 @@ execute: |
# Use offset 65 as FAT32 kicks in for devices larger than 32MB
printf "\x01" > one
tests.cleanup defer rm -f one
dd if=one of=/dev/sda2 seek=65 bs=1 count=1 conv=notrunc
dd if="/dev/disk/by-label/$LABEL" of=dirty skip=65 bs=1 count=1 conv=notrunc
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this is now wrong, it was

dd \
  if=one \
  of=/dev/... \
  seek=65 \
  ...

and it has now become:

dd \
  if=/dev/...
  of=dirty \
  skip=65 \
  ...

perhaps this line was mixed up with the one below?

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

yes, thanks for checking

Copy link
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@anonymouse64 updated

tests.cleanup pop

# Reboot to give the early boot process a chance to fix the corruption.
Expand All @@ -76,12 +109,10 @@ execute: |
# chain.
unmount_vfat
cat /proc/self/mountinfo >boot1-after-umount.log
dd if=/dev/sda2 of=dirty skip=65 bs=1 count=1 conv=notrunc
dd if="/dev/disk/by-label/$LABEL" of=dirty skip=65 bs=1 count=1 conv=notrunc
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

this looks correct

test "$(od -t x1 -A n dirty)" = " 00" # NOTE: the leading space is relevant

# Reboot to restore mount points.
REBOOT
;;
esac
restore: |
tests.cleanup restore
mvo5 marked this conversation as resolved.
Show resolved Hide resolved