| Message ID | 8704e5bd46d9f8dc37cec2781104704fa7213aa3.1739363803.git.nirjhar.roy.lists@gmail.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | Add mount and remount related tests | expand |
On Wed, Feb 12, 2025 at 12:39:56PM +0000, Nirjhar Roy (IBM) wrote: > This test is to verify that repeated warnings are not printed > for default options (attr2, noikeep) and warnings are > printed for non default options (noattr2, ikeep). Remount > with noattr2 fails on a v5 filesystem, so skip the mount option. Why do we care if remount succeeds or fails? That's not what the test is exercising. i.e. We are testing to see if the appropriate deprecation warning for a deprecated mount option has been issued or not, and that should happen regardless of whether the mount option is valid or not for the given filesysetm format.... Hence I don't see any reason for changing the test to exclude noattr2 testing on v5 filesystems... -Dave.
On Wed, Feb 12, 2025 at 12:39:56PM +0000, Nirjhar Roy (IBM) wrote: > This test is to verify that repeated warnings are not printed > for default options (attr2, noikeep) and warnings are > printed for non default options (noattr2, ikeep). Remount > with noattr2 fails on a v5 filesystem, so skip the mount option. > > Signed-off-by: Nirjhar Roy (IBM) <nirjhar.roy.lists@gmail.com> > --- > tests/xfs/539 | 10 ++++++++-- > 1 file changed, 8 insertions(+), 2 deletions(-) > > diff --git a/tests/xfs/539 b/tests/xfs/539 > index b9bb7cc1..58eead67 100755 > --- a/tests/xfs/539 > +++ b/tests/xfs/539 > @@ -42,7 +42,8 @@ echo "Silence is golden." > > # Skip old kernels that did not print the warning yet > log_tag > -_scratch_mkfs > $seqres.full 2>&1 > +is_v5=true > +_scratch_mkfs |& grep -q "crc=0" && is_v5=false >> $seqres.full 2>&1 Usually we do this with something more like: _scratch_mkfs | _filter_mkfs >>$seqres.full 2>$tmp.mkfs . $tmp.mkfs if [ $_fs_has_crcs -eq 1 ]; then # v5 stuff else # v4 stuff endif > _scratch_mount -o attr2 > _scratch_unmount > check_dmesg_for_since_tag "XFS: attr2 mount option is deprecated" || \ > @@ -60,8 +61,13 @@ for VAR in {attr2,noikeep}; do > echo "Should not be able to find deprecation warning for $VAR" > done > for VAR in {noattr2,ikeep}; do > + if [[ "$VAR" == "noattr2" ]] && $is_v5; then > + echo "remount with noattr2 will fail in v5 filesystem. Skip" \ > + >> $seqres.full > + continue /me wonders if it'd be cleaner to do: VARS=(ikeep) test $_fs_has_crcs -eq 0 && VARS+=(noattr2) for VAR in "${VARS[@]}"; do ... done > + fi > log_tag > - _scratch_remount $VAR > + _scratch_remount $VAR >> $seqres.full 2>&1 Nit: Indentation. --D > check_dmesg_for_since_tag "XFS: $VAR mount option is deprecated" || \ > echo "Could not find deprecation warning for $VAR" > done > -- > 2.34.1 > >
On 2/13/25 02:31, Dave Chinner wrote: > On Wed, Feb 12, 2025 at 12:39:56PM +0000, Nirjhar Roy (IBM) wrote: >> This test is to verify that repeated warnings are not printed >> for default options (attr2, noikeep) and warnings are >> printed for non default options (noattr2, ikeep). Remount >> with noattr2 fails on a v5 filesystem, so skip the mount option. > Why do we care if remount succeeds or fails? That's not what the > test is exercising. > > i.e. We are testing to see if the appropriate deprecation warning > for a deprecated mount option has been issued or not, and that > should happen regardless of whether the mount option is valid or not > for the given filesysetm format.... Okay, thank you for the clarification. Also, based on your response on patch 3/3, remount with noattr2(or any other invalid remount options), should be silently ignored, so this patch won't be necessary. However, we have observed failure of the test xfs/539 because remount with noattr2 was failing with CONFIG_XFS_SUPPORT_V4=n on v5 xfs and this failure looks like a kernel bug. More on this on my reply[1] to your comments on patch 3/3. [1] https://lore.kernel.org/all/b43e4cd9-d8aa-4cc0-a5ff-35f2e0553682@gmail.com/ > > Hence I don't see any reason for changing the test to exclude > noattr2 testing on v5 filesystems... > > -Dave.
On 2/13/25 02:31, Dave Chinner wrote: > On Wed, Feb 12, 2025 at 12:39:56PM +0000, Nirjhar Roy (IBM) wrote: >> This test is to verify that repeated warnings are not printed >> for default options (attr2, noikeep) and warnings are >> printed for non default options (noattr2, ikeep). Remount >> with noattr2 fails on a v5 filesystem, so skip the mount option. > Why do we care if remount succeeds or fails? That's not what the > test is exercising. > > i.e. We are testing to see if the appropriate deprecation warning > for a deprecated mount option has been issued or not, and that > should happen regardless of whether the mount option is valid or not > for the given filesysetm format.... > > Hence I don't see any reason for changing the test to exclude > noattr2 testing on v5 filesystems... Yes, this makes sense. The test indeed just checks for the dmesg warnings, and they appear even if the remount fails. I wrote the patch because xfs/539 has started failing in one of our fstests CI runs because RHEL 10 has started disabling xfs v4 support i.e, CONFIG_XFS_SUPPORT_V4=n. Do you think modifying this patch in such a way that the test ignores the remount failures with noattr2 and continues the test is an appropriate idea (since the test xfs/539 only intends to check the dmesg warnings)? So something like:, --- a/tests/xfs/539 +++ b/tests/xfs/539 @@ -61,7 +61,11 @@ for VAR in {attr2,noikeep}; do done for VAR in {noattr2,ikeep}; do log_tag - _scratch_remount $VAR + _scratch_remount $VAR >> $seqres.full 2>&1 + if [[ "$VAR" == "noattr2" && "$?" != "0" ]]; then + echo "remount will fail in v5 filesystem but the warning should be printed" \ + >> $seqres.full + fi check_dmesg_for_since_tag "XFS: $VAR mount option is deprecated" || \ echo "Could not find deprecation warning for $VAR" I also suggested something similar in one of my previous replies[1] in this patch series. Can you please let me know your thoughts on this? [1] https://lore.kernel.org/all/90be3350-67e5-4dec-bc65-442762f5f856@gmail.com/ --NR > > -Dave.
diff --git a/tests/xfs/539 b/tests/xfs/539 index b9bb7cc1..58eead67 100755 --- a/tests/xfs/539 +++ b/tests/xfs/539 @@ -42,7 +42,8 @@ echo "Silence is golden." # Skip old kernels that did not print the warning yet log_tag -_scratch_mkfs > $seqres.full 2>&1 +is_v5=true +_scratch_mkfs |& grep -q "crc=0" && is_v5=false >> $seqres.full 2>&1 _scratch_mount -o attr2 _scratch_unmount check_dmesg_for_since_tag "XFS: attr2 mount option is deprecated" || \ @@ -60,8 +61,13 @@ for VAR in {attr2,noikeep}; do echo "Should not be able to find deprecation warning for $VAR" done for VAR in {noattr2,ikeep}; do + if [[ "$VAR" == "noattr2" ]] && $is_v5; then + echo "remount with noattr2 will fail in v5 filesystem. Skip" \ + >> $seqres.full + continue + fi log_tag - _scratch_remount $VAR + _scratch_remount $VAR >> $seqres.full 2>&1 check_dmesg_for_since_tag "XFS: $VAR mount option is deprecated" || \ echo "Could not find deprecation warning for $VAR" done
This test is to verify that repeated warnings are not printed for default options (attr2, noikeep) and warnings are printed for non default options (noattr2, ikeep). Remount with noattr2 fails on a v5 filesystem, so skip the mount option. Signed-off-by: Nirjhar Roy (IBM) <nirjhar.roy.lists@gmail.com> --- tests/xfs/539 | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-)