* [PATCHSET v2 0/2] fstests: add inode btree blocks counters to the AGI header @ 2021-04-21 0:22 Darrick J. Wong 2021-04-21 0:22 ` [PATCH 1/2] xfs: functional testing of V5-relevant options Darrick J. Wong 2021-04-21 0:22 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 0 siblings, 2 replies; 20+ messages in thread From: Darrick J. Wong @ 2021-04-21 0:22 UTC (permalink / raw) To: djwong, guaneryu; +Cc: linux-xfs, fstests, guan Hi all, Years ago, Christoph diagnosed a problem where freeing an inode on a totally full filesystem could fail due to finobt expansion not being able to allocate enough blocks. He solved the problem by using the per-AG block reservation system to ensure that there are always enough blocks for finobt expansion, but that came at the cost of having to walk the entire finobt at mount time. This new feature solves that performance regression by adding inode btree block counts to the AGI header. The patches in this series amend fstests to handle the new metadata fields and to test that upgrades work properly. v2: move the xfs_admin functional test into a separate patch file and split them into one test for basic functionality and another for the extended functionality testing that modifies the test run config If you're going to start using this mess, you probably ought to just pull from my git trees, which are linked below. This is an extraordinary way to destroy everything. Enjoy! Comments and questions are, as always, welcome. --D kernel git tree: https://git.kernel.org/cgit/linux/kernel/git/djwong/xfs-linux.git/log/?h=inobt-counters xfsprogs git tree: https://git.kernel.org/cgit/linux/kernel/git/djwong/xfsprogs-dev.git/log/?h=inobt-counters fstests git tree: https://git.kernel.org/cgit/linux/kernel/git/djwong/xfstests-dev.git/log/?h=inobt-counters --- common/xfs | 23 +++++++++++ tests/xfs/764 | 93 +++++++++++++++++++++++++++++++++++++++++++ tests/xfs/764.out | 17 ++++++++ tests/xfs/773 | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/773.out | 19 +++++++++ tests/xfs/910 | 98 ++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/910.out | 23 +++++++++++ tests/xfs/group | 3 + 8 files changed, 390 insertions(+) create mode 100755 tests/xfs/764 create mode 100644 tests/xfs/764.out create mode 100755 tests/xfs/773 create mode 100644 tests/xfs/773.out create mode 100755 tests/xfs/910 create mode 100644 tests/xfs/910.out ^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH 1/2] xfs: functional testing of V5-relevant options 2021-04-21 0:22 [PATCHSET v2 0/2] fstests: add inode btree blocks counters to the AGI header Darrick J. Wong @ 2021-04-21 0:22 ` Darrick J. Wong 2021-04-21 18:00 ` Brian Foster ` (2 more replies) 2021-04-21 0:22 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 1 sibling, 3 replies; 20+ messages in thread From: Darrick J. Wong @ 2021-04-21 0:22 UTC (permalink / raw) To: djwong, guaneryu; +Cc: linux-xfs, fstests, guan From: Darrick J. Wong <djwong@kernel.org> Currently, the only functional testing for xfs_admin is xfs/287, which checks that one can add 32-bit project ids to a V4 filesystem. This obviously isn't an exhaustive test of all the CLI arguments, and historically there have been xfs configurations that don't even work. Therefore, introduce a couple of new tests -- one that will test the simple options with the default configuration, and a second test that steps a bit outside of the test run configuration to make sure that we do the right thing for external devices. The second test already caught a nasty bug in xfsprogs 5.11. Signed-off-by: Darrick J. Wong <djwong@kernel.org> --- common/xfs | 21 ++++++++++ tests/xfs/764 | 93 +++++++++++++++++++++++++++++++++++++++++++ tests/xfs/764.out | 17 ++++++++ tests/xfs/773 | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/773.out | 19 +++++++++ tests/xfs/group | 2 + 6 files changed, 266 insertions(+) create mode 100755 tests/xfs/764 create mode 100644 tests/xfs/764.out create mode 100755 tests/xfs/773 create mode 100644 tests/xfs/773.out diff --git a/common/xfs b/common/xfs index c2384146..5abc7034 100644 --- a/common/xfs +++ b/common/xfs @@ -1142,3 +1142,24 @@ _check_scratch_xfs_features() echo "${output[@]}" test "${found}" -eq "$#" } + +# Decide if xfs_repair knows how to set (or clear) a filesystem feature. +_require_xfs_repair_upgrade() +{ + local type="$1" + + $XFS_REPAIR_PROG -c "$type=garbagevalue" 2>&1 | \ + grep -q 'unknown option' && \ + _notrun "xfs_repair does not support upgrading fs with $type" +} + +_require_xfs_scratch_inobtcount() +{ + _require_scratch + + _scratch_mkfs -m inobtcount=1 &> /dev/null || \ + _notrun "mkfs.xfs doesn't have inobtcount feature" + _try_scratch_mount || \ + _notrun "inobtcount not supported by scratch filesystem type: $FSTYP" + _scratch_unmount +} diff --git a/tests/xfs/764 b/tests/xfs/764 new file mode 100755 index 00000000..ebdf8883 --- /dev/null +++ b/tests/xfs/764 @@ -0,0 +1,93 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0-or-later +# Copyright (c) 2021 Oracle. All Rights Reserved. +# +# FS QA Test No. 764 +# +# Functional testing for xfs_admin to make sure that it handles option parsing +# correctly for functionality that's relevant to V5 filesystems. It doesn't +# test the options that apply only to V4 filesystems because that disk format +# is deprecated. + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# real QA test starts here +_supported_fs xfs +_require_scratch +_require_command "$XFS_ADMIN_PROG" "xfs_admin" + +rm -f $seqres.full + +note() { + echo "$@" | tee -a $seqres.full +} + +note "S0: Initialize filesystem" +_scratch_mkfs -L origlabel -m uuid=babababa-baba-baba-baba-babababababa >> $seqres.full +_scratch_xfs_db -c label -c uuid +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S1: Set a filesystem label" +_scratch_xfs_admin -L newlabel >> $seqres.full +_scratch_xfs_db -c label +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S2: Clear filesystem label" +_scratch_xfs_admin -L -- >> $seqres.full +_scratch_xfs_db -c label +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S3: Try to set oversized label" +_scratch_xfs_admin -L thisismuchtoolongforxfstohandle >> $seqres.full +_scratch_xfs_db -c label +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S4: Set filesystem UUID" +_scratch_xfs_admin -U deaddead-dead-dead-dead-deaddeaddead >> $seqres.full +_scratch_xfs_db -c uuid +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S5: Zero out filesystem UUID" +_scratch_xfs_admin -U nil >> $seqres.full +_scratch_xfs_db -c uuid +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S6: Randomize filesystem UUID" +old_uuid="$(_scratch_xfs_db -c uuid)" +_scratch_xfs_admin -U generate >> $seqres.full +new_uuid="$(_scratch_xfs_db -c uuid)" +if [ "$new_uuid" = "$old_uuid" ]; then + echo "UUID randomization failed? $old_uuid == $new_uuid" +fi +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S7: Restore original filesystem UUID" +if _check_scratch_xfs_features V5 >/dev/null; then + # Only V5 supports the metauuid feature that enables us to restore the + # original UUID after a change. + _scratch_xfs_admin -U restore >> $seqres.full + _scratch_xfs_db -c uuid +else + echo "UUID = babababa-baba-baba-baba-babababababa" +fi + +# success, all done +status=0 +exit diff --git a/tests/xfs/764.out b/tests/xfs/764.out new file mode 100644 index 00000000..8da929ec --- /dev/null +++ b/tests/xfs/764.out @@ -0,0 +1,17 @@ +QA output created by 764 +S0: Initialize filesystem +label = "origlabel" +UUID = babababa-baba-baba-baba-babababababa +S1: Set a filesystem label +label = "newlabel" +S2: Clear filesystem label +label = "" +S3: Try to set oversized label +label = "thisismuchto" +S4: Set filesystem UUID +UUID = deaddead-dead-dead-dead-deaddeaddead +S5: Zero out filesystem UUID +UUID = 00000000-0000-0000-0000-000000000000 +S6: Randomize filesystem UUID +S7: Restore original filesystem UUID +UUID = babababa-baba-baba-baba-babababababa diff --git a/tests/xfs/773 b/tests/xfs/773 new file mode 100755 index 00000000..f184962a --- /dev/null +++ b/tests/xfs/773 @@ -0,0 +1,114 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0-or-later +# Copyright (c) 2021 Oracle. All Rights Reserved. +# +# FS QA Test No. 773 +# +# Functional testing for xfs_admin to ensure that it parses arguments correctly +# with regards to data devices that are files, external logs, and realtime +# devices. +# +# Because this test synthesizes log and rt devices (by modifying the test run +# configuration), it does /not/ require the ability to mount the scratch +# filesystem. This increases test coverage while isolating the weird bits to a +# single test. +# +# This is partially a regression test for "xfs_admin: pick up log arguments +# correctly", insofar as the issue fixed by that patch was discovered with an +# earlier revision of this test. + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* $fake_logfile $fake_rtfile $fake_datafile +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# real QA test starts here +_supported_fs xfs +_require_test +_require_scratch_nocheck +_require_command "$XFS_ADMIN_PROG" "xfs_admin" + +rm -f $seqres.full + +# Create some fake sparse files for testing external devices and whatnot +fake_datafile=$TEST_DIR/scratch.data +rm -f $fake_datafile +truncate -s 500m $fake_datafile + +fake_logfile=$TEST_DIR/scratch.log +rm -f $fake_logfile +truncate -s 500m $fake_logfile + +fake_rtfile=$TEST_DIR/scratch.rt +rm -f $fake_rtfile +truncate -s 500m $fake_rtfile + +# Save the original variables +orig_ddev=$SCRATCH_DEV +orig_external=$USE_EXTERNAL +orig_logdev=$SCRATCH_LOGDEV +orig_rtdev=$SCRATCH_RTDEV + +scenario() { + echo "$@" | tee -a $seqres.full + + SCRATCH_DEV=$orig_ddev + USE_EXTERNAL=$orig_external + SCRATCH_LOGDEV=$orig_logdev + SCRATCH_RTDEV=$orig_rtdev +} + +check_label() { + _scratch_mkfs -L oldlabel >> $seqres.full + _scratch_xfs_db -c label + _scratch_xfs_admin -L newlabel "$@" >> $seqres.full + _scratch_xfs_db -c label + _scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" +} + +scenario "S1: Check that label setting with file image" +SCRATCH_DEV=$fake_datafile +check_label -f + +scenario "S2: Check that setting with logdev works" +USE_EXTERNAL=yes +SCRATCH_LOGDEV=$fake_logfile +check_label + +scenario "S3: Check that setting with rtdev works" +USE_EXTERNAL=yes +SCRATCH_RTDEV=$fake_rtfile +check_label + +scenario "S4: Check that setting with rtdev + logdev works" +USE_EXTERNAL=yes +SCRATCH_LOGDEV=$fake_logfile +SCRATCH_RTDEV=$fake_rtfile +check_label + +scenario "S5: Check that setting with nortdev + nologdev works" +USE_EXTERNAL= +SCRATCH_LOGDEV= +SCRATCH_RTDEV= +check_label + +scenario "S6: Check that setting with bdev incorrectly flagged as file works" +check_label -f + +# success, all done +status=0 +exit diff --git a/tests/xfs/773.out b/tests/xfs/773.out new file mode 100644 index 00000000..954bfb85 --- /dev/null +++ b/tests/xfs/773.out @@ -0,0 +1,19 @@ +QA output created by 773 +S1: Check that label setting with file image +label = "oldlabel" +label = "newlabel" +S2: Check that setting with logdev works +label = "oldlabel" +label = "newlabel" +S3: Check that setting with rtdev works +label = "oldlabel" +label = "newlabel" +S4: Check that setting with rtdev + logdev works +label = "oldlabel" +label = "newlabel" +S5: Check that setting with nortdev + nologdev works +label = "oldlabel" +label = "newlabel" +S6: Check that setting with bdev incorrectly flagged as file works +label = "oldlabel" +label = "newlabel" diff --git a/tests/xfs/group b/tests/xfs/group index 461ae2b2..a2309465 100644 --- a/tests/xfs/group +++ b/tests/xfs/group @@ -522,5 +522,7 @@ 537 auto quick 538 auto stress 539 auto quick mount +764 auto quick repair 768 auto quick repair 770 auto repair +773 auto quick repair ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] xfs: functional testing of V5-relevant options 2021-04-21 0:22 ` [PATCH 1/2] xfs: functional testing of V5-relevant options Darrick J. Wong @ 2021-04-21 18:00 ` Brian Foster 2021-04-21 20:39 ` Darrick J. Wong 2021-04-22 0:29 ` Allison Henderson 2021-04-22 0:51 ` [PATCH v2.1 " Darrick J. Wong 2 siblings, 1 reply; 20+ messages in thread From: Brian Foster @ 2021-04-21 18:00 UTC (permalink / raw) To: Darrick J. Wong; +Cc: guaneryu, linux-xfs, fstests, guan On Tue, Apr 20, 2021 at 05:22:50PM -0700, Darrick J. Wong wrote: > From: Darrick J. Wong <djwong@kernel.org> > > Currently, the only functional testing for xfs_admin is xfs/287, which > checks that one can add 32-bit project ids to a V4 filesystem. This > obviously isn't an exhaustive test of all the CLI arguments, and > historically there have been xfs configurations that don't even work. > > Therefore, introduce a couple of new tests -- one that will test the > simple options with the default configuration, and a second test that > steps a bit outside of the test run configuration to make sure that we > do the right thing for external devices. The second test already caught > a nasty bug in xfsprogs 5.11. > > Signed-off-by: Darrick J. Wong <djwong@kernel.org> > --- > common/xfs | 21 ++++++++++ > tests/xfs/764 | 93 +++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/764.out | 17 ++++++++ > tests/xfs/773 | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/773.out | 19 +++++++++ > tests/xfs/group | 2 + > 6 files changed, 266 insertions(+) > create mode 100755 tests/xfs/764 > create mode 100644 tests/xfs/764.out > create mode 100755 tests/xfs/773 > create mode 100644 tests/xfs/773.out > > ... > diff --git a/tests/xfs/773 b/tests/xfs/773 > new file mode 100755 > index 00000000..f184962a > --- /dev/null > +++ b/tests/xfs/773 > @@ -0,0 +1,114 @@ > +#! /bin/bash > +# SPDX-License-Identifier: GPL-2.0-or-later > +# Copyright (c) 2021 Oracle. All Rights Reserved. > +# > +# FS QA Test No. 773 > +# > +# Functional testing for xfs_admin to ensure that it parses arguments correctly > +# with regards to data devices that are files, external logs, and realtime > +# devices. > +# > +# Because this test synthesizes log and rt devices (by modifying the test run > +# configuration), it does /not/ require the ability to mount the scratch > +# filesystem. This increases test coverage while isolating the weird bits to a > +# single test. > +# > +# This is partially a regression test for "xfs_admin: pick up log arguments > +# correctly", insofar as the issue fixed by that patch was discovered with an > +# earlier revision of this test. > + > +seq=`basename $0` > +seqres=$RESULT_DIR/$seq > +echo "QA output created by $seq" > + > +here=`pwd` > +tmp=/tmp/$$ > +status=1 # failure is the default! > +trap "_cleanup; exit \$status" 0 1 2 3 15 > + > +_cleanup() > +{ > + cd / > + rm -f $tmp.* $fake_logfile $fake_rtfile $fake_datafile > +} > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/filter > + > +# real QA test starts here > +_supported_fs xfs > +_require_test > +_require_scratch_nocheck > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > + > +rm -f $seqres.full > + > +# Create some fake sparse files for testing external devices and whatnot > +fake_datafile=$TEST_DIR/scratch.data > +rm -f $fake_datafile > +truncate -s 500m $fake_datafile > + > +fake_logfile=$TEST_DIR/scratch.log > +rm -f $fake_logfile > +truncate -s 500m $fake_logfile > + > +fake_rtfile=$TEST_DIR/scratch.rt > +rm -f $fake_rtfile > +truncate -s 500m $fake_rtfile > + I think it's usually good practice to incorporate $seq into filenames created on the test device. > +# Save the original variables > +orig_ddev=$SCRATCH_DEV > +orig_external=$USE_EXTERNAL > +orig_logdev=$SCRATCH_LOGDEV > +orig_rtdev=$SCRATCH_RTDEV > + > +scenario() { > + echo "$@" | tee -a $seqres.full > + > + SCRATCH_DEV=$orig_ddev > + USE_EXTERNAL=$orig_external > + SCRATCH_LOGDEV=$orig_logdev > + SCRATCH_RTDEV=$orig_rtdev > +} > + > +check_label() { > + _scratch_mkfs -L oldlabel >> $seqres.full > + _scratch_xfs_db -c label > + _scratch_xfs_admin -L newlabel "$@" >> $seqres.full > + _scratch_xfs_db -c label > + _scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > +} > + > +scenario "S1: Check that label setting with file image" > +SCRATCH_DEV=$fake_datafile > +check_label -f > + > +scenario "S2: Check that setting with logdev works" > +USE_EXTERNAL=yes > +SCRATCH_LOGDEV=$fake_logfile > +check_label > + > +scenario "S3: Check that setting with rtdev works" > +USE_EXTERNAL=yes > +SCRATCH_RTDEV=$fake_rtfile > +check_label > + > +scenario "S4: Check that setting with rtdev + logdev works" > +USE_EXTERNAL=yes > +SCRATCH_LOGDEV=$fake_logfile > +SCRATCH_RTDEV=$fake_rtfile > +check_label > + > +scenario "S5: Check that setting with nortdev + nologdev works" > +USE_EXTERNAL= > +SCRATCH_LOGDEV= > +SCRATCH_RTDEV= > +check_label > + > +scenario "S6: Check that setting with bdev incorrectly flagged as file works" > +check_label -f > + > +# success, all done > +status=0 > +exit > diff --git a/tests/xfs/773.out b/tests/xfs/773.out > new file mode 100644 > index 00000000..954bfb85 > --- /dev/null > +++ b/tests/xfs/773.out > @@ -0,0 +1,19 @@ > +QA output created by 773 > +S1: Check that label setting with file image > +label = "oldlabel" > +label = "newlabel" > +S2: Check that setting with logdev works > +label = "oldlabel" > +label = "newlabel" > +S3: Check that setting with rtdev works > +label = "oldlabel" > +label = "newlabel" > +S4: Check that setting with rtdev + logdev works > +label = "oldlabel" > +label = "newlabel" > +S5: Check that setting with nortdev + nologdev works > +label = "oldlabel" > +label = "newlabel" > +S6: Check that setting with bdev incorrectly flagged as file works > +label = "oldlabel" > +label = "newlabel" > diff --git a/tests/xfs/group b/tests/xfs/group > index 461ae2b2..a2309465 100644 > --- a/tests/xfs/group > +++ b/tests/xfs/group > @@ -522,5 +522,7 @@ > 537 auto quick > 538 auto stress > 539 auto quick mount > +764 auto quick repair > 768 auto quick repair > 770 auto repair > +773 auto quick repair > Both of these tests are primarily targeted at xfs_admin, right? If so, I'm not sure the repair group makes much sense. With the nits fixed up: Reviewed-by: Brian Foster <bfoster@redhat.com> ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] xfs: functional testing of V5-relevant options 2021-04-21 18:00 ` Brian Foster @ 2021-04-21 20:39 ` Darrick J. Wong 0 siblings, 0 replies; 20+ messages in thread From: Darrick J. Wong @ 2021-04-21 20:39 UTC (permalink / raw) To: Brian Foster; +Cc: guaneryu, linux-xfs, fstests, guan On Wed, Apr 21, 2021 at 02:00:38PM -0400, Brian Foster wrote: > On Tue, Apr 20, 2021 at 05:22:50PM -0700, Darrick J. Wong wrote: > > From: Darrick J. Wong <djwong@kernel.org> > > > > Currently, the only functional testing for xfs_admin is xfs/287, which > > checks that one can add 32-bit project ids to a V4 filesystem. This > > obviously isn't an exhaustive test of all the CLI arguments, and > > historically there have been xfs configurations that don't even work. > > > > Therefore, introduce a couple of new tests -- one that will test the > > simple options with the default configuration, and a second test that > > steps a bit outside of the test run configuration to make sure that we > > do the right thing for external devices. The second test already caught > > a nasty bug in xfsprogs 5.11. > > > > Signed-off-by: Darrick J. Wong <djwong@kernel.org> > > --- > > common/xfs | 21 ++++++++++ > > tests/xfs/764 | 93 +++++++++++++++++++++++++++++++++++++++++++ > > tests/xfs/764.out | 17 ++++++++ > > tests/xfs/773 | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > > tests/xfs/773.out | 19 +++++++++ > > tests/xfs/group | 2 + > > 6 files changed, 266 insertions(+) > > create mode 100755 tests/xfs/764 > > create mode 100644 tests/xfs/764.out > > create mode 100755 tests/xfs/773 > > create mode 100644 tests/xfs/773.out > > > > > ... > > diff --git a/tests/xfs/773 b/tests/xfs/773 > > new file mode 100755 > > index 00000000..f184962a > > --- /dev/null > > +++ b/tests/xfs/773 > > @@ -0,0 +1,114 @@ > > +#! /bin/bash > > +# SPDX-License-Identifier: GPL-2.0-or-later > > +# Copyright (c) 2021 Oracle. All Rights Reserved. > > +# > > +# FS QA Test No. 773 > > +# > > +# Functional testing for xfs_admin to ensure that it parses arguments correctly > > +# with regards to data devices that are files, external logs, and realtime > > +# devices. > > +# > > +# Because this test synthesizes log and rt devices (by modifying the test run > > +# configuration), it does /not/ require the ability to mount the scratch > > +# filesystem. This increases test coverage while isolating the weird bits to a > > +# single test. > > +# > > +# This is partially a regression test for "xfs_admin: pick up log arguments > > +# correctly", insofar as the issue fixed by that patch was discovered with an > > +# earlier revision of this test. > > + > > +seq=`basename $0` > > +seqres=$RESULT_DIR/$seq > > +echo "QA output created by $seq" > > + > > +here=`pwd` > > +tmp=/tmp/$$ > > +status=1 # failure is the default! > > +trap "_cleanup; exit \$status" 0 1 2 3 15 > > + > > +_cleanup() > > +{ > > + cd / > > + rm -f $tmp.* $fake_logfile $fake_rtfile $fake_datafile > > +} > > + > > +# get standard environment, filters and checks > > +. ./common/rc > > +. ./common/filter > > + > > +# real QA test starts here > > +_supported_fs xfs > > +_require_test > > +_require_scratch_nocheck > > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > > + > > +rm -f $seqres.full > > + > > +# Create some fake sparse files for testing external devices and whatnot > > +fake_datafile=$TEST_DIR/scratch.data > > +rm -f $fake_datafile > > +truncate -s 500m $fake_datafile > > + > > +fake_logfile=$TEST_DIR/scratch.log > > +rm -f $fake_logfile > > +truncate -s 500m $fake_logfile > > + > > +fake_rtfile=$TEST_DIR/scratch.rt > > +rm -f $fake_rtfile > > +truncate -s 500m $fake_rtfile > > + > > I think it's usually good practice to incorporate $seq into filenames > created on the test device. Ok fixed. > > +# Save the original variables > > +orig_ddev=$SCRATCH_DEV > > +orig_external=$USE_EXTERNAL > > +orig_logdev=$SCRATCH_LOGDEV > > +orig_rtdev=$SCRATCH_RTDEV > > + > > +scenario() { > > + echo "$@" | tee -a $seqres.full > > + > > + SCRATCH_DEV=$orig_ddev > > + USE_EXTERNAL=$orig_external > > + SCRATCH_LOGDEV=$orig_logdev > > + SCRATCH_RTDEV=$orig_rtdev > > +} > > + > > +check_label() { > > + _scratch_mkfs -L oldlabel >> $seqres.full > > + _scratch_xfs_db -c label > > + _scratch_xfs_admin -L newlabel "$@" >> $seqres.full > > + _scratch_xfs_db -c label > > + _scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > > +} > > + > > +scenario "S1: Check that label setting with file image" > > +SCRATCH_DEV=$fake_datafile > > +check_label -f > > + > > +scenario "S2: Check that setting with logdev works" > > +USE_EXTERNAL=yes > > +SCRATCH_LOGDEV=$fake_logfile > > +check_label > > + > > +scenario "S3: Check that setting with rtdev works" > > +USE_EXTERNAL=yes > > +SCRATCH_RTDEV=$fake_rtfile > > +check_label > > + > > +scenario "S4: Check that setting with rtdev + logdev works" > > +USE_EXTERNAL=yes > > +SCRATCH_LOGDEV=$fake_logfile > > +SCRATCH_RTDEV=$fake_rtfile > > +check_label > > + > > +scenario "S5: Check that setting with nortdev + nologdev works" > > +USE_EXTERNAL= > > +SCRATCH_LOGDEV= > > +SCRATCH_RTDEV= > > +check_label > > + > > +scenario "S6: Check that setting with bdev incorrectly flagged as file works" > > +check_label -f > > + > > +# success, all done > > +status=0 > > +exit > > diff --git a/tests/xfs/773.out b/tests/xfs/773.out > > new file mode 100644 > > index 00000000..954bfb85 > > --- /dev/null > > +++ b/tests/xfs/773.out > > @@ -0,0 +1,19 @@ > > +QA output created by 773 > > +S1: Check that label setting with file image > > +label = "oldlabel" > > +label = "newlabel" > > +S2: Check that setting with logdev works > > +label = "oldlabel" > > +label = "newlabel" > > +S3: Check that setting with rtdev works > > +label = "oldlabel" > > +label = "newlabel" > > +S4: Check that setting with rtdev + logdev works > > +label = "oldlabel" > > +label = "newlabel" > > +S5: Check that setting with nortdev + nologdev works > > +label = "oldlabel" > > +label = "newlabel" > > +S6: Check that setting with bdev incorrectly flagged as file works > > +label = "oldlabel" > > +label = "newlabel" > > diff --git a/tests/xfs/group b/tests/xfs/group > > index 461ae2b2..a2309465 100644 > > --- a/tests/xfs/group > > +++ b/tests/xfs/group > > @@ -522,5 +522,7 @@ > > 537 auto quick > > 538 auto stress > > 539 auto quick mount > > +764 auto quick repair > > 768 auto quick repair > > 770 auto repair > > +773 auto quick repair > > > > Both of these tests are primarily targeted at xfs_admin, right? If so, > I'm not sure the repair group makes much sense. With the nits fixed up: > > Reviewed-by: Brian Foster <bfoster@redhat.com> Will fix the groups; thanks for the review! --D ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 1/2] xfs: functional testing of V5-relevant options 2021-04-21 0:22 ` [PATCH 1/2] xfs: functional testing of V5-relevant options Darrick J. Wong 2021-04-21 18:00 ` Brian Foster @ 2021-04-22 0:29 ` Allison Henderson 2021-04-22 0:51 ` [PATCH v2.1 " Darrick J. Wong 2 siblings, 0 replies; 20+ messages in thread From: Allison Henderson @ 2021-04-22 0:29 UTC (permalink / raw) To: Darrick J. Wong, guaneryu; +Cc: linux-xfs, fstests, guan On 4/20/21 5:22 PM, Darrick J. Wong wrote: > From: Darrick J. Wong <djwong@kernel.org> > > Currently, the only functional testing for xfs_admin is xfs/287, which > checks that one can add 32-bit project ids to a V4 filesystem. This > obviously isn't an exhaustive test of all the CLI arguments, and > historically there have been xfs configurations that don't even work. > > Therefore, introduce a couple of new tests -- one that will test the > simple options with the default configuration, and a second test that > steps a bit outside of the test run configuration to make sure that we > do the right thing for external devices. The second test already caught > a nasty bug in xfsprogs 5.11. > > Signed-off-by: Darrick J. Wong <djwong@kernel.org> Ok, I followed it through and I think it makes sense. With Brians comments fixed: Reviewed-by: Allison Henderson <allison.henderson@oracle.com> > --- > common/xfs | 21 ++++++++++ > tests/xfs/764 | 93 +++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/764.out | 17 ++++++++ > tests/xfs/773 | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/773.out | 19 +++++++++ > tests/xfs/group | 2 + > 6 files changed, 266 insertions(+) > create mode 100755 tests/xfs/764 > create mode 100644 tests/xfs/764.out > create mode 100755 tests/xfs/773 > create mode 100644 tests/xfs/773.out > > > diff --git a/common/xfs b/common/xfs > index c2384146..5abc7034 100644 > --- a/common/xfs > +++ b/common/xfs > @@ -1142,3 +1142,24 @@ _check_scratch_xfs_features() > echo "${output[@]}" > test "${found}" -eq "$#" > } > + > +# Decide if xfs_repair knows how to set (or clear) a filesystem feature. > +_require_xfs_repair_upgrade() > +{ > + local type="$1" > + > + $XFS_REPAIR_PROG -c "$type=garbagevalue" 2>&1 | \ > + grep -q 'unknown option' && \ > + _notrun "xfs_repair does not support upgrading fs with $type" > +} > + > +_require_xfs_scratch_inobtcount() > +{ > + _require_scratch > + > + _scratch_mkfs -m inobtcount=1 &> /dev/null || \ > + _notrun "mkfs.xfs doesn't have inobtcount feature" > + _try_scratch_mount || \ > + _notrun "inobtcount not supported by scratch filesystem type: $FSTYP" > + _scratch_unmount > +} > diff --git a/tests/xfs/764 b/tests/xfs/764 > new file mode 100755 > index 00000000..ebdf8883 > --- /dev/null > +++ b/tests/xfs/764 > @@ -0,0 +1,93 @@ > +#! /bin/bash > +# SPDX-License-Identifier: GPL-2.0-or-later > +# Copyright (c) 2021 Oracle. All Rights Reserved. > +# > +# FS QA Test No. 764 > +# > +# Functional testing for xfs_admin to make sure that it handles option parsing > +# correctly for functionality that's relevant to V5 filesystems. It doesn't > +# test the options that apply only to V4 filesystems because that disk format > +# is deprecated. > + > +seq=`basename $0` > +seqres=$RESULT_DIR/$seq > +echo "QA output created by $seq" > + > +here=`pwd` > +tmp=/tmp/$$ > +status=1 # failure is the default! > +trap "_cleanup; exit \$status" 0 1 2 3 15 > + > +_cleanup() > +{ > + cd / > + rm -f $tmp.* > +} > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/filter > + > +# real QA test starts here > +_supported_fs xfs > +_require_scratch > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > + > +rm -f $seqres.full > + > +note() { > + echo "$@" | tee -a $seqres.full > +} > + > +note "S0: Initialize filesystem" > +_scratch_mkfs -L origlabel -m uuid=babababa-baba-baba-baba-babababababa >> $seqres.full > +_scratch_xfs_db -c label -c uuid > +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > + > +note "S1: Set a filesystem label" > +_scratch_xfs_admin -L newlabel >> $seqres.full > +_scratch_xfs_db -c label > +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > + > +note "S2: Clear filesystem label" > +_scratch_xfs_admin -L -- >> $seqres.full > +_scratch_xfs_db -c label > +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > + > +note "S3: Try to set oversized label" > +_scratch_xfs_admin -L thisismuchtoolongforxfstohandle >> $seqres.full > +_scratch_xfs_db -c label > +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > + > +note "S4: Set filesystem UUID" > +_scratch_xfs_admin -U deaddead-dead-dead-dead-deaddeaddead >> $seqres.full > +_scratch_xfs_db -c uuid > +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > + > +note "S5: Zero out filesystem UUID" > +_scratch_xfs_admin -U nil >> $seqres.full > +_scratch_xfs_db -c uuid > +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > + > +note "S6: Randomize filesystem UUID" > +old_uuid="$(_scratch_xfs_db -c uuid)" > +_scratch_xfs_admin -U generate >> $seqres.full > +new_uuid="$(_scratch_xfs_db -c uuid)" > +if [ "$new_uuid" = "$old_uuid" ]; then > + echo "UUID randomization failed? $old_uuid == $new_uuid" > +fi > +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > + > +note "S7: Restore original filesystem UUID" > +if _check_scratch_xfs_features V5 >/dev/null; then > + # Only V5 supports the metauuid feature that enables us to restore the > + # original UUID after a change. > + _scratch_xfs_admin -U restore >> $seqres.full > + _scratch_xfs_db -c uuid > +else > + echo "UUID = babababa-baba-baba-baba-babababababa" > +fi > + > +# success, all done > +status=0 > +exit > diff --git a/tests/xfs/764.out b/tests/xfs/764.out > new file mode 100644 > index 00000000..8da929ec > --- /dev/null > +++ b/tests/xfs/764.out > @@ -0,0 +1,17 @@ > +QA output created by 764 > +S0: Initialize filesystem > +label = "origlabel" > +UUID = babababa-baba-baba-baba-babababababa > +S1: Set a filesystem label > +label = "newlabel" > +S2: Clear filesystem label > +label = "" > +S3: Try to set oversized label > +label = "thisismuchto" > +S4: Set filesystem UUID > +UUID = deaddead-dead-dead-dead-deaddeaddead > +S5: Zero out filesystem UUID > +UUID = 00000000-0000-0000-0000-000000000000 > +S6: Randomize filesystem UUID > +S7: Restore original filesystem UUID > +UUID = babababa-baba-baba-baba-babababababa > diff --git a/tests/xfs/773 b/tests/xfs/773 > new file mode 100755 > index 00000000..f184962a > --- /dev/null > +++ b/tests/xfs/773 > @@ -0,0 +1,114 @@ > +#! /bin/bash > +# SPDX-License-Identifier: GPL-2.0-or-later > +# Copyright (c) 2021 Oracle. All Rights Reserved. > +# > +# FS QA Test No. 773 > +# > +# Functional testing for xfs_admin to ensure that it parses arguments correctly > +# with regards to data devices that are files, external logs, and realtime > +# devices. > +# > +# Because this test synthesizes log and rt devices (by modifying the test run > +# configuration), it does /not/ require the ability to mount the scratch > +# filesystem. This increases test coverage while isolating the weird bits to a > +# single test. > +# > +# This is partially a regression test for "xfs_admin: pick up log arguments > +# correctly", insofar as the issue fixed by that patch was discovered with an > +# earlier revision of this test. > + > +seq=`basename $0` > +seqres=$RESULT_DIR/$seq > +echo "QA output created by $seq" > + > +here=`pwd` > +tmp=/tmp/$$ > +status=1 # failure is the default! > +trap "_cleanup; exit \$status" 0 1 2 3 15 > + > +_cleanup() > +{ > + cd / > + rm -f $tmp.* $fake_logfile $fake_rtfile $fake_datafile > +} > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/filter > + > +# real QA test starts here > +_supported_fs xfs > +_require_test > +_require_scratch_nocheck > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > + > +rm -f $seqres.full > + > +# Create some fake sparse files for testing external devices and whatnot > +fake_datafile=$TEST_DIR/scratch.data > +rm -f $fake_datafile > +truncate -s 500m $fake_datafile > + > +fake_logfile=$TEST_DIR/scratch.log > +rm -f $fake_logfile > +truncate -s 500m $fake_logfile > + > +fake_rtfile=$TEST_DIR/scratch.rt > +rm -f $fake_rtfile > +truncate -s 500m $fake_rtfile > + > +# Save the original variables > +orig_ddev=$SCRATCH_DEV > +orig_external=$USE_EXTERNAL > +orig_logdev=$SCRATCH_LOGDEV > +orig_rtdev=$SCRATCH_RTDEV > + > +scenario() { > + echo "$@" | tee -a $seqres.full > + > + SCRATCH_DEV=$orig_ddev > + USE_EXTERNAL=$orig_external > + SCRATCH_LOGDEV=$orig_logdev > + SCRATCH_RTDEV=$orig_rtdev > +} > + > +check_label() { > + _scratch_mkfs -L oldlabel >> $seqres.full > + _scratch_xfs_db -c label > + _scratch_xfs_admin -L newlabel "$@" >> $seqres.full > + _scratch_xfs_db -c label > + _scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" > +} > + > +scenario "S1: Check that label setting with file image" > +SCRATCH_DEV=$fake_datafile > +check_label -f > + > +scenario "S2: Check that setting with logdev works" > +USE_EXTERNAL=yes > +SCRATCH_LOGDEV=$fake_logfile > +check_label > + > +scenario "S3: Check that setting with rtdev works" > +USE_EXTERNAL=yes > +SCRATCH_RTDEV=$fake_rtfile > +check_label > + > +scenario "S4: Check that setting with rtdev + logdev works" > +USE_EXTERNAL=yes > +SCRATCH_LOGDEV=$fake_logfile > +SCRATCH_RTDEV=$fake_rtfile > +check_label > + > +scenario "S5: Check that setting with nortdev + nologdev works" > +USE_EXTERNAL= > +SCRATCH_LOGDEV= > +SCRATCH_RTDEV= > +check_label > + > +scenario "S6: Check that setting with bdev incorrectly flagged as file works" > +check_label -f > + > +# success, all done > +status=0 > +exit > diff --git a/tests/xfs/773.out b/tests/xfs/773.out > new file mode 100644 > index 00000000..954bfb85 > --- /dev/null > +++ b/tests/xfs/773.out > @@ -0,0 +1,19 @@ > +QA output created by 773 > +S1: Check that label setting with file image > +label = "oldlabel" > +label = "newlabel" > +S2: Check that setting with logdev works > +label = "oldlabel" > +label = "newlabel" > +S3: Check that setting with rtdev works > +label = "oldlabel" > +label = "newlabel" > +S4: Check that setting with rtdev + logdev works > +label = "oldlabel" > +label = "newlabel" > +S5: Check that setting with nortdev + nologdev works > +label = "oldlabel" > +label = "newlabel" > +S6: Check that setting with bdev incorrectly flagged as file works > +label = "oldlabel" > +label = "newlabel" > diff --git a/tests/xfs/group b/tests/xfs/group > index 461ae2b2..a2309465 100644 > --- a/tests/xfs/group > +++ b/tests/xfs/group > @@ -522,5 +522,7 @@ > 537 auto quick > 538 auto stress > 539 auto quick mount > +764 auto quick repair > 768 auto quick repair > 770 auto repair > +773 auto quick repair > ^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH v2.1 1/2] xfs: functional testing of V5-relevant options 2021-04-21 0:22 ` [PATCH 1/2] xfs: functional testing of V5-relevant options Darrick J. Wong 2021-04-21 18:00 ` Brian Foster 2021-04-22 0:29 ` Allison Henderson @ 2021-04-22 0:51 ` Darrick J. Wong 2 siblings, 0 replies; 20+ messages in thread From: Darrick J. Wong @ 2021-04-22 0:51 UTC (permalink / raw) To: guaneryu; +Cc: linux-xfs, fstests, guan, bfoster, allison.henderson From: Darrick J. Wong <djwong@kernel.org> Currently, the only functional testing for xfs_admin is xfs/287, which checks that one can add 32-bit project ids to a V4 filesystem. This obviously isn't an exhaustive test of all the CLI arguments, and historically there have been xfs configurations that don't even work. Therefore, introduce a couple of new tests -- one that will test the simple options with the default configuration, and a second test that steps a bit outside of the test run configuration to make sure that we do the right thing for external devices. The second test already caught a nasty bug in xfsprogs 5.11. Signed-off-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Allison Henderson <allison.henderson@oracle.com> --- v2.1: add sequence number to synthesized external devices, change groups --- common/xfs | 10 +++++ tests/xfs/764 | 93 +++++++++++++++++++++++++++++++++++++++++++ tests/xfs/764.out | 17 ++++++++ tests/xfs/773 | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/773.out | 19 +++++++++ tests/xfs/group | 2 + 6 files changed, 255 insertions(+) create mode 100755 tests/xfs/764 create mode 100644 tests/xfs/764.out create mode 100755 tests/xfs/773 create mode 100644 tests/xfs/773.out diff --git a/common/xfs b/common/xfs index ba6523ad..4f1a4dba 100644 --- a/common/xfs +++ b/common/xfs @@ -1128,3 +1128,13 @@ _check_scratch_xfs_features() echo "${output[@]}" test "${found}" -eq "$#" } + +# Decide if xfs_repair knows how to set (or clear) a filesystem feature. +_require_xfs_repair_upgrade() +{ + local type="$1" + + $XFS_REPAIR_PROG -c "$type=garbagevalue" 2>&1 | \ + grep -q 'unknown option' && \ + _notrun "xfs_repair does not support upgrading fs with $type" +} diff --git a/tests/xfs/764 b/tests/xfs/764 new file mode 100755 index 00000000..ebdf8883 --- /dev/null +++ b/tests/xfs/764 @@ -0,0 +1,93 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0-or-later +# Copyright (c) 2021 Oracle. All Rights Reserved. +# +# FS QA Test No. 764 +# +# Functional testing for xfs_admin to make sure that it handles option parsing +# correctly for functionality that's relevant to V5 filesystems. It doesn't +# test the options that apply only to V4 filesystems because that disk format +# is deprecated. + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# real QA test starts here +_supported_fs xfs +_require_scratch +_require_command "$XFS_ADMIN_PROG" "xfs_admin" + +rm -f $seqres.full + +note() { + echo "$@" | tee -a $seqres.full +} + +note "S0: Initialize filesystem" +_scratch_mkfs -L origlabel -m uuid=babababa-baba-baba-baba-babababababa >> $seqres.full +_scratch_xfs_db -c label -c uuid +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S1: Set a filesystem label" +_scratch_xfs_admin -L newlabel >> $seqres.full +_scratch_xfs_db -c label +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S2: Clear filesystem label" +_scratch_xfs_admin -L -- >> $seqres.full +_scratch_xfs_db -c label +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S3: Try to set oversized label" +_scratch_xfs_admin -L thisismuchtoolongforxfstohandle >> $seqres.full +_scratch_xfs_db -c label +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S4: Set filesystem UUID" +_scratch_xfs_admin -U deaddead-dead-dead-dead-deaddeaddead >> $seqres.full +_scratch_xfs_db -c uuid +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S5: Zero out filesystem UUID" +_scratch_xfs_admin -U nil >> $seqres.full +_scratch_xfs_db -c uuid +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S6: Randomize filesystem UUID" +old_uuid="$(_scratch_xfs_db -c uuid)" +_scratch_xfs_admin -U generate >> $seqres.full +new_uuid="$(_scratch_xfs_db -c uuid)" +if [ "$new_uuid" = "$old_uuid" ]; then + echo "UUID randomization failed? $old_uuid == $new_uuid" +fi +_scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" + +note "S7: Restore original filesystem UUID" +if _check_scratch_xfs_features V5 >/dev/null; then + # Only V5 supports the metauuid feature that enables us to restore the + # original UUID after a change. + _scratch_xfs_admin -U restore >> $seqres.full + _scratch_xfs_db -c uuid +else + echo "UUID = babababa-baba-baba-baba-babababababa" +fi + +# success, all done +status=0 +exit diff --git a/tests/xfs/764.out b/tests/xfs/764.out new file mode 100644 index 00000000..8da929ec --- /dev/null +++ b/tests/xfs/764.out @@ -0,0 +1,17 @@ +QA output created by 764 +S0: Initialize filesystem +label = "origlabel" +UUID = babababa-baba-baba-baba-babababababa +S1: Set a filesystem label +label = "newlabel" +S2: Clear filesystem label +label = "" +S3: Try to set oversized label +label = "thisismuchto" +S4: Set filesystem UUID +UUID = deaddead-dead-dead-dead-deaddeaddead +S5: Zero out filesystem UUID +UUID = 00000000-0000-0000-0000-000000000000 +S6: Randomize filesystem UUID +S7: Restore original filesystem UUID +UUID = babababa-baba-baba-baba-babababababa diff --git a/tests/xfs/773 b/tests/xfs/773 new file mode 100755 index 00000000..f7a340f6 --- /dev/null +++ b/tests/xfs/773 @@ -0,0 +1,114 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0-or-later +# Copyright (c) 2021 Oracle. All Rights Reserved. +# +# FS QA Test No. 773 +# +# Functional testing for xfs_admin to ensure that it parses arguments correctly +# with regards to data devices that are files, external logs, and realtime +# devices. +# +# Because this test synthesizes log and rt devices (by modifying the test run +# configuration), it does /not/ require the ability to mount the scratch +# filesystem. This increases test coverage while isolating the weird bits to a +# single test. +# +# This is partially a regression test for "xfs_admin: pick up log arguments +# correctly", insofar as the issue fixed by that patch was discovered with an +# earlier revision of this test. + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* $fake_logfile $fake_rtfile $fake_datafile +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# real QA test starts here +_supported_fs xfs +_require_test +_require_scratch_nocheck +_require_command "$XFS_ADMIN_PROG" "xfs_admin" + +rm -f $seqres.full + +# Create some fake sparse files for testing external devices and whatnot +fake_datafile=$TEST_DIR/$seq.scratch.data +rm -f $fake_datafile +truncate -s 500m $fake_datafile + +fake_logfile=$TEST_DIR/$seq.scratch.log +rm -f $fake_logfile +truncate -s 500m $fake_logfile + +fake_rtfile=$TEST_DIR/$seq.scratch.rt +rm -f $fake_rtfile +truncate -s 500m $fake_rtfile + +# Save the original variables +orig_ddev=$SCRATCH_DEV +orig_external=$USE_EXTERNAL +orig_logdev=$SCRATCH_LOGDEV +orig_rtdev=$SCRATCH_RTDEV + +scenario() { + echo "$@" | tee -a $seqres.full + + SCRATCH_DEV=$orig_ddev + USE_EXTERNAL=$orig_external + SCRATCH_LOGDEV=$orig_logdev + SCRATCH_RTDEV=$orig_rtdev +} + +check_label() { + _scratch_mkfs -L oldlabel >> $seqres.full + _scratch_xfs_db -c label + _scratch_xfs_admin -L newlabel "$@" >> $seqres.full + _scratch_xfs_db -c label + _scratch_xfs_repair -n &>> $seqres.full || echo "Check failed?" +} + +scenario "S1: Check that label setting with file image" +SCRATCH_DEV=$fake_datafile +check_label -f + +scenario "S2: Check that setting with logdev works" +USE_EXTERNAL=yes +SCRATCH_LOGDEV=$fake_logfile +check_label + +scenario "S3: Check that setting with rtdev works" +USE_EXTERNAL=yes +SCRATCH_RTDEV=$fake_rtfile +check_label + +scenario "S4: Check that setting with rtdev + logdev works" +USE_EXTERNAL=yes +SCRATCH_LOGDEV=$fake_logfile +SCRATCH_RTDEV=$fake_rtfile +check_label + +scenario "S5: Check that setting with nortdev + nologdev works" +USE_EXTERNAL= +SCRATCH_LOGDEV= +SCRATCH_RTDEV= +check_label + +scenario "S6: Check that setting with bdev incorrectly flagged as file works" +check_label -f + +# success, all done +status=0 +exit diff --git a/tests/xfs/773.out b/tests/xfs/773.out new file mode 100644 index 00000000..954bfb85 --- /dev/null +++ b/tests/xfs/773.out @@ -0,0 +1,19 @@ +QA output created by 773 +S1: Check that label setting with file image +label = "oldlabel" +label = "newlabel" +S2: Check that setting with logdev works +label = "oldlabel" +label = "newlabel" +S3: Check that setting with rtdev works +label = "oldlabel" +label = "newlabel" +S4: Check that setting with rtdev + logdev works +label = "oldlabel" +label = "newlabel" +S5: Check that setting with nortdev + nologdev works +label = "oldlabel" +label = "newlabel" +S6: Check that setting with bdev incorrectly flagged as file works +label = "oldlabel" +label = "newlabel" diff --git a/tests/xfs/group b/tests/xfs/group index 461ae2b2..4231d1d5 100644 --- a/tests/xfs/group +++ b/tests/xfs/group @@ -522,5 +522,7 @@ 537 auto quick 538 auto stress 539 auto quick mount +764 auto quick admin 768 auto quick repair 770 auto repair +773 auto quick admin ^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PATCH 2/2] xfs: test inobtcount upgrade 2021-04-21 0:22 [PATCHSET v2 0/2] fstests: add inode btree blocks counters to the AGI header Darrick J. Wong 2021-04-21 0:22 ` [PATCH 1/2] xfs: functional testing of V5-relevant options Darrick J. Wong @ 2021-04-21 0:22 ` Darrick J. Wong 2021-04-21 5:52 ` Amir Goldstein ` (3 more replies) 1 sibling, 4 replies; 20+ messages in thread From: Darrick J. Wong @ 2021-04-21 0:22 UTC (permalink / raw) To: djwong, guaneryu; +Cc: linux-xfs, fstests, guan From: Darrick J. Wong <djwong@kernel.org> Make sure we can actually upgrade filesystems to support inode btree counters. Signed-off-by: Darrick J. Wong <djwong@kernel.org> --- common/xfs | 8 +++- tests/xfs/910 | 98 +++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/910.out | 23 ++++++++++++ tests/xfs/group | 1 + 4 files changed, 127 insertions(+), 3 deletions(-) create mode 100755 tests/xfs/910 create mode 100644 tests/xfs/910.out diff --git a/common/xfs b/common/xfs index 5abc7034..3d660858 100644 --- a/common/xfs +++ b/common/xfs @@ -1153,13 +1153,15 @@ _require_xfs_repair_upgrade() _notrun "xfs_repair does not support upgrading fs with $type" } -_require_xfs_scratch_inobtcount() +# Require that the scratch device exists, that mkfs can format with inobtcount +# enabled, and that the kernel can mount such a filesystem. +_require_scratch_xfs_inobtcount() { _require_scratch _scratch_mkfs -m inobtcount=1 &> /dev/null || \ - _notrun "mkfs.xfs doesn't have inobtcount feature" + _notrun "mkfs.xfs doesn't support inobtcount feature" _try_scratch_mount || \ - _notrun "inobtcount not supported by scratch filesystem type: $FSTYP" + _notrun "kernel doesn't support xfs inobtcount feature" _scratch_unmount } diff --git a/tests/xfs/910 b/tests/xfs/910 new file mode 100755 index 00000000..237d0a35 --- /dev/null +++ b/tests/xfs/910 @@ -0,0 +1,98 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0-or-later +# Copyright (c) 2021 Oracle. All Rights Reserved. +# +# FS QA Test No. 910 +# +# Check that we can upgrade a filesystem to support inobtcount and that +# everything works properly after the upgrade. + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# real QA test starts here +_supported_fs xfs +_require_scratch_xfs_inobtcount +_require_command "$XFS_ADMIN_PROG" "xfs_admin" +_require_xfs_repair_upgrade inobtcount + +rm -f $seqres.full + +# Make sure we can't format a filesystem with inobtcount and not finobt. +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ + echo "Should not be able to format with inobtcount but not finobt." + +# Make sure we can't upgrade a V4 filesystem +_scratch_mkfs -m crc=0,inobtcount=0,finobt=0 >> $seqres.full +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full +_check_scratch_xfs_features INOBTCNT + +# Make sure we can't upgrade a filesystem to inobtcount without finobt. +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 >> $seqres.full +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full +_check_scratch_xfs_features INOBTCNT + +# Format V5 filesystem without inode btree counter support and populate it. +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full +_scratch_mount + +mkdir $SCRATCH_MNT/stress +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full +echo moo > $SCRATCH_MNT/urk + +_scratch_unmount + +# Upgrade filesystem to have the counters and inject failure into repair and +# make sure that the only path forward is to re-run repair on the filesystem. +echo "Fail partway through upgrading" +XFS_REPAIR_FAIL_AFTER_PHASE=2 _scratch_xfs_repair -c inobtcount=1 2>> $seqres.full +test $? -eq 137 || echo "repair should have been killed??" +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT +_try_scratch_mount &> $tmp.mount +res=$? +_filter_scratch < $tmp.mount +if [ $res -eq 0 ]; then + echo "needsrepair should have prevented mount" + _scratch_unmount +fi + +echo "Re-run repair to finish upgrade" +_scratch_xfs_repair 2>> $seqres.full +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT + +echo "Filesystem should be usable again" +_scratch_mount +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full +_scratch_unmount +_check_scratch_fs +_check_scratch_xfs_features INOBTCNT + +echo "Make sure we have nonzero counters" +_scratch_xfs_db -c 'agi 0' -c 'print ino_blocks fino_blocks' | \ + sed -e 's/= [1-9]*/= NONZERO/g' + +echo "Make sure we can't re-add inobtcount" +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full + +echo "Mount again, look at our files" +_scratch_mount >> $seqres.full +cat $SCRATCH_MNT/urk + +status=0 +exit diff --git a/tests/xfs/910.out b/tests/xfs/910.out new file mode 100644 index 00000000..1bf040d5 --- /dev/null +++ b/tests/xfs/910.out @@ -0,0 +1,23 @@ +QA output created by 910 +Running xfs_repair to upgrade filesystem. +Inode btree count feature only supported on V5 filesystems. +FEATURES: INOBTCNT:NO +Running xfs_repair to upgrade filesystem. +Inode btree count feature requires free inode btree. +FEATURES: INOBTCNT:NO +Fail partway through upgrading +Adding inode btree counts to filesystem. +FEATURES: NEEDSREPAIR:YES INOBTCNT:YES +mount: SCRATCH_MNT: mount(2) system call failed: Structure needs cleaning. +Re-run repair to finish upgrade +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES +Filesystem should be usable again +FEATURES: INOBTCNT:YES +Make sure we have nonzero counters +ino_blocks = NONZERO +fino_blocks = NONZERO +Make sure we can't re-add inobtcount +Running xfs_repair to upgrade filesystem. +Filesystem already has inode btree counts. +Mount again, look at our files +moo diff --git a/tests/xfs/group b/tests/xfs/group index a2309465..bd47333c 100644 --- a/tests/xfs/group +++ b/tests/xfs/group @@ -526,3 +526,4 @@ 768 auto quick repair 770 auto repair 773 auto quick repair +910 auto quick inobtcount ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2021-04-21 0:22 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong @ 2021-04-21 5:52 ` Amir Goldstein 2021-04-22 0:50 ` Darrick J. Wong 2021-04-21 18:00 ` Brian Foster ` (2 subsequent siblings) 3 siblings, 1 reply; 20+ messages in thread From: Amir Goldstein @ 2021-04-21 5:52 UTC (permalink / raw) To: Darrick J. Wong; +Cc: Eryu Guan, linux-xfs, fstests, Eryu Guan On Wed, Apr 21, 2021 at 3:23 AM Darrick J. Wong <djwong@kernel.org> wrote: > > From: Darrick J. Wong <djwong@kernel.org> > > Make sure we can actually upgrade filesystems to support inode btree > counters. > > Signed-off-by: Darrick J. Wong <djwong@kernel.org> > --- > common/xfs | 8 +++- > tests/xfs/910 | 98 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/910.out | 23 ++++++++++++ > tests/xfs/group | 1 + > 4 files changed, 127 insertions(+), 3 deletions(-) > create mode 100755 tests/xfs/910 > create mode 100644 tests/xfs/910.out > > > diff --git a/common/xfs b/common/xfs > index 5abc7034..3d660858 100644 > --- a/common/xfs > +++ b/common/xfs > @@ -1153,13 +1153,15 @@ _require_xfs_repair_upgrade() > _notrun "xfs_repair does not support upgrading fs with $type" > } > > -_require_xfs_scratch_inobtcount() > +# Require that the scratch device exists, that mkfs can format with inobtcount > +# enabled, and that the kernel can mount such a filesystem. > +_require_scratch_xfs_inobtcount() > { > _require_scratch > > _scratch_mkfs -m inobtcount=1 &> /dev/null || \ > - _notrun "mkfs.xfs doesn't have inobtcount feature" > + _notrun "mkfs.xfs doesn't support inobtcount feature" > _try_scratch_mount || \ > - _notrun "inobtcount not supported by scratch filesystem type: $FSTYP" > + _notrun "kernel doesn't support xfs inobtcount feature" > _scratch_unmount > } > diff --git a/tests/xfs/910 b/tests/xfs/910 > new file mode 100755 > index 00000000..237d0a35 > --- /dev/null > +++ b/tests/xfs/910 > @@ -0,0 +1,98 @@ > +#! /bin/bash > +# SPDX-License-Identifier: GPL-2.0-or-later > +# Copyright (c) 2021 Oracle. All Rights Reserved. > +# > +# FS QA Test No. 910 > +# > +# Check that we can upgrade a filesystem to support inobtcount and that > +# everything works properly after the upgrade. > + > +seq=`basename $0` > +seqres=$RESULT_DIR/$seq > +echo "QA output created by $seq" > + > +here=`pwd` > +tmp=/tmp/$$ > +status=1 # failure is the default! > +trap "_cleanup; exit \$status" 0 1 2 3 15 > + > +_cleanup() > +{ > + cd / > + rm -f $tmp.* > +} > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/filter > + > +# real QA test starts here > +_supported_fs xfs > +_require_scratch_xfs_inobtcount > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > +_require_xfs_repair_upgrade inobtcount > + > +rm -f $seqres.full > + > +# Make sure we can't format a filesystem with inobtcount and not finobt. > +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ > + echo "Should not be able to format with inobtcount but not finobt." > + > +# Make sure we can't upgrade a V4 filesystem > +_scratch_mkfs -m crc=0,inobtcount=0,finobt=0 >> $seqres.full > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_check_scratch_xfs_features INOBTCNT > + > +# Make sure we can't upgrade a filesystem to inobtcount without finobt. > +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 >> $seqres.full > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_check_scratch_xfs_features INOBTCNT > + > +# Format V5 filesystem without inode btree counter support and populate it. > +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full > +_scratch_mount > + > +mkdir $SCRATCH_MNT/stress > +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full > +echo moo > $SCRATCH_MNT/urk > + > +_scratch_unmount > + > +# Upgrade filesystem to have the counters and inject failure into repair and > +# make sure that the only path forward is to re-run repair on the filesystem. > +echo "Fail partway through upgrading" > +XFS_REPAIR_FAIL_AFTER_PHASE=2 _scratch_xfs_repair -c inobtcount=1 2>> $seqres.full > +test $? -eq 137 || echo "repair should have been killed??" > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > +_try_scratch_mount &> $tmp.mount > +res=$? > +_filter_scratch < $tmp.mount > +if [ $res -eq 0 ]; then > + echo "needsrepair should have prevented mount" > + _scratch_unmount > +fi > + > +echo "Re-run repair to finish upgrade" > +_scratch_xfs_repair 2>> $seqres.full > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > + > +echo "Filesystem should be usable again" > +_scratch_mount > +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full > +_scratch_unmount > +_check_scratch_fs > +_check_scratch_xfs_features INOBTCNT > + > +echo "Make sure we have nonzero counters" > +_scratch_xfs_db -c 'agi 0' -c 'print ino_blocks fino_blocks' | \ > + sed -e 's/= [1-9]*/= NONZERO/g' > + I don't think = 100 translates to = NONZERO... > +echo "Make sure we can't re-add inobtcount" > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > + > +echo "Mount again, look at our files" > +_scratch_mount >> $seqres.full > +cat $SCRATCH_MNT/urk > + > +status=0 > +exit > diff --git a/tests/xfs/910.out b/tests/xfs/910.out > new file mode 100644 > index 00000000..1bf040d5 > --- /dev/null > +++ b/tests/xfs/910.out > @@ -0,0 +1,23 @@ > +QA output created by 910 > +Running xfs_repair to upgrade filesystem. > +Inode btree count feature only supported on V5 filesystems. > +FEATURES: INOBTCNT:NO > +Running xfs_repair to upgrade filesystem. > +Inode btree count feature requires free inode btree. > +FEATURES: INOBTCNT:NO > +Fail partway through upgrading > +Adding inode btree counts to filesystem. > +FEATURES: NEEDSREPAIR:YES INOBTCNT:YES > +mount: SCRATCH_MNT: mount(2) system call failed: Structure needs cleaning. > +Re-run repair to finish upgrade > +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES > +Filesystem should be usable again > +FEATURES: INOBTCNT:YES > +Make sure we have nonzero counters > +ino_blocks = NONZERO > +fino_blocks = NONZERO > +Make sure we can't re-add inobtcount > +Running xfs_repair to upgrade filesystem. > +Filesystem already has inode btree counts. > +Mount again, look at our files > +moo This test is quite rigid in the format of messages expected from xfs_repair. I suppose there is no precedent here... > diff --git a/tests/xfs/group b/tests/xfs/group > index a2309465..bd47333c 100644 > --- a/tests/xfs/group > +++ b/tests/xfs/group > @@ -526,3 +526,4 @@ > 768 auto quick repair > 770 auto repair > 773 auto quick repair > +910 auto quick inobtcount > +repair? Thanks, Amir. ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2021-04-21 5:52 ` Amir Goldstein @ 2021-04-22 0:50 ` Darrick J. Wong 0 siblings, 0 replies; 20+ messages in thread From: Darrick J. Wong @ 2021-04-22 0:50 UTC (permalink / raw) To: Amir Goldstein; +Cc: Eryu Guan, linux-xfs, fstests, Eryu Guan On Wed, Apr 21, 2021 at 08:52:39AM +0300, Amir Goldstein wrote: > On Wed, Apr 21, 2021 at 3:23 AM Darrick J. Wong <djwong@kernel.org> wrote: > > > > From: Darrick J. Wong <djwong@kernel.org> > > > > Make sure we can actually upgrade filesystems to support inode btree > > counters. > > > > Signed-off-by: Darrick J. Wong <djwong@kernel.org> > > --- > > common/xfs | 8 +++- > > tests/xfs/910 | 98 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > > tests/xfs/910.out | 23 ++++++++++++ > > tests/xfs/group | 1 + > > 4 files changed, 127 insertions(+), 3 deletions(-) > > create mode 100755 tests/xfs/910 > > create mode 100644 tests/xfs/910.out > > > > > > diff --git a/common/xfs b/common/xfs > > index 5abc7034..3d660858 100644 > > --- a/common/xfs > > +++ b/common/xfs > > @@ -1153,13 +1153,15 @@ _require_xfs_repair_upgrade() > > _notrun "xfs_repair does not support upgrading fs with $type" > > } > > > > -_require_xfs_scratch_inobtcount() > > +# Require that the scratch device exists, that mkfs can format with inobtcount > > +# enabled, and that the kernel can mount such a filesystem. > > +_require_scratch_xfs_inobtcount() <groan> I clearly changed the wrong patch in response to reviewer comments, sorry, will resubmit... > > { > > _require_scratch > > > > _scratch_mkfs -m inobtcount=1 &> /dev/null || \ > > - _notrun "mkfs.xfs doesn't have inobtcount feature" > > + _notrun "mkfs.xfs doesn't support inobtcount feature" > > _try_scratch_mount || \ > > - _notrun "inobtcount not supported by scratch filesystem type: $FSTYP" > > + _notrun "kernel doesn't support xfs inobtcount feature" > > _scratch_unmount > > } > > diff --git a/tests/xfs/910 b/tests/xfs/910 > > new file mode 100755 > > index 00000000..237d0a35 > > --- /dev/null > > +++ b/tests/xfs/910 > > @@ -0,0 +1,98 @@ > > +#! /bin/bash > > +# SPDX-License-Identifier: GPL-2.0-or-later > > +# Copyright (c) 2021 Oracle. All Rights Reserved. > > +# > > +# FS QA Test No. 910 > > +# > > +# Check that we can upgrade a filesystem to support inobtcount and that > > +# everything works properly after the upgrade. > > + > > +seq=`basename $0` > > +seqres=$RESULT_DIR/$seq > > +echo "QA output created by $seq" > > + > > +here=`pwd` > > +tmp=/tmp/$$ > > +status=1 # failure is the default! > > +trap "_cleanup; exit \$status" 0 1 2 3 15 > > + > > +_cleanup() > > +{ > > + cd / > > + rm -f $tmp.* > > +} > > + > > +# get standard environment, filters and checks > > +. ./common/rc > > +. ./common/filter > > + > > +# real QA test starts here > > +_supported_fs xfs > > +_require_scratch_xfs_inobtcount > > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > > +_require_xfs_repair_upgrade inobtcount > > + > > +rm -f $seqres.full > > + > > +# Make sure we can't format a filesystem with inobtcount and not finobt. > > +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ > > + echo "Should not be able to format with inobtcount but not finobt." > > + > > +# Make sure we can't upgrade a V4 filesystem > > +_scratch_mkfs -m crc=0,inobtcount=0,finobt=0 >> $seqres.full > > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > > +_check_scratch_xfs_features INOBTCNT > > + > > +# Make sure we can't upgrade a filesystem to inobtcount without finobt. > > +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 >> $seqres.full > > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > > +_check_scratch_xfs_features INOBTCNT > > + > > +# Format V5 filesystem without inode btree counter support and populate it. > > +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full > > +_scratch_mount > > + > > +mkdir $SCRATCH_MNT/stress > > +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full > > +echo moo > $SCRATCH_MNT/urk > > + > > +_scratch_unmount > > + > > +# Upgrade filesystem to have the counters and inject failure into repair and > > +# make sure that the only path forward is to re-run repair on the filesystem. > > +echo "Fail partway through upgrading" > > +XFS_REPAIR_FAIL_AFTER_PHASE=2 _scratch_xfs_repair -c inobtcount=1 2>> $seqres.full > > +test $? -eq 137 || echo "repair should have been killed??" > > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > > +_try_scratch_mount &> $tmp.mount > > +res=$? > > +_filter_scratch < $tmp.mount > > +if [ $res -eq 0 ]; then > > + echo "needsrepair should have prevented mount" > > + _scratch_unmount > > +fi > > + > > +echo "Re-run repair to finish upgrade" > > +_scratch_xfs_repair 2>> $seqres.full > > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > > + > > +echo "Filesystem should be usable again" > > +_scratch_mount > > +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full > > +_scratch_unmount > > +_check_scratch_fs > > +_check_scratch_xfs_features INOBTCNT > > + > > +echo "Make sure we have nonzero counters" > > +_scratch_xfs_db -c 'agi 0' -c 'print ino_blocks fino_blocks' | \ > > + sed -e 's/= [1-9]*/= NONZERO/g' > > + > > I don't think = 100 translates to = NONZERO... Urk, yes, will fix. > > +echo "Make sure we can't re-add inobtcount" > > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > > + > > +echo "Mount again, look at our files" > > +_scratch_mount >> $seqres.full > > +cat $SCRATCH_MNT/urk > > + > > +status=0 > > +exit > > diff --git a/tests/xfs/910.out b/tests/xfs/910.out > > new file mode 100644 > > index 00000000..1bf040d5 > > --- /dev/null > > +++ b/tests/xfs/910.out > > @@ -0,0 +1,23 @@ > > +QA output created by 910 > > +Running xfs_repair to upgrade filesystem. > > +Inode btree count feature only supported on V5 filesystems. > > +FEATURES: INOBTCNT:NO > > +Running xfs_repair to upgrade filesystem. > > +Inode btree count feature requires free inode btree. > > +FEATURES: INOBTCNT:NO > > +Fail partway through upgrading > > +Adding inode btree counts to filesystem. > > +FEATURES: NEEDSREPAIR:YES INOBTCNT:YES > > +mount: SCRATCH_MNT: mount(2) system call failed: Structure needs cleaning. > > +Re-run repair to finish upgrade > > +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES > > +Filesystem should be usable again > > +FEATURES: INOBTCNT:YES > > +Make sure we have nonzero counters > > +ino_blocks = NONZERO > > +fino_blocks = NONZERO > > +Make sure we can't re-add inobtcount > > +Running xfs_repair to upgrade filesystem. > > +Filesystem already has inode btree counts. > > +Mount again, look at our files > > +moo > > This test is quite rigid in the format of messages expected from > xfs_repair. I suppose there is no precedent here... <shrug> There /is/ precedent -- if we ever need to change the stdout, we filter the golden output. --D > > > diff --git a/tests/xfs/group b/tests/xfs/group > > index a2309465..bd47333c 100644 > > --- a/tests/xfs/group > > +++ b/tests/xfs/group > > @@ -526,3 +526,4 @@ > > 768 auto quick repair > > 770 auto repair > > 773 auto quick repair > > +910 auto quick inobtcount > > > +repair? > > Thanks, > Amir. ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2021-04-21 0:22 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 2021-04-21 5:52 ` Amir Goldstein @ 2021-04-21 18:00 ` Brian Foster 2021-04-22 0:29 ` Allison Henderson 2021-04-22 0:53 ` [PATCH v2.1 " Darrick J. Wong 3 siblings, 0 replies; 20+ messages in thread From: Brian Foster @ 2021-04-21 18:00 UTC (permalink / raw) To: Darrick J. Wong; +Cc: guaneryu, linux-xfs, fstests, guan On Tue, Apr 20, 2021 at 05:22:57PM -0700, Darrick J. Wong wrote: > From: Darrick J. Wong <djwong@kernel.org> > > Make sure we can actually upgrade filesystems to support inode btree > counters. > > Signed-off-by: Darrick J. Wong <djwong@kernel.org> > --- LGTM: Reviewed-by: Brian Foster <bfoster@redhat.com> > common/xfs | 8 +++- > tests/xfs/910 | 98 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/910.out | 23 ++++++++++++ > tests/xfs/group | 1 + > 4 files changed, 127 insertions(+), 3 deletions(-) > create mode 100755 tests/xfs/910 > create mode 100644 tests/xfs/910.out > > > diff --git a/common/xfs b/common/xfs > index 5abc7034..3d660858 100644 > --- a/common/xfs > +++ b/common/xfs > @@ -1153,13 +1153,15 @@ _require_xfs_repair_upgrade() > _notrun "xfs_repair does not support upgrading fs with $type" > } > > -_require_xfs_scratch_inobtcount() > +# Require that the scratch device exists, that mkfs can format with inobtcount > +# enabled, and that the kernel can mount such a filesystem. > +_require_scratch_xfs_inobtcount() > { > _require_scratch > > _scratch_mkfs -m inobtcount=1 &> /dev/null || \ > - _notrun "mkfs.xfs doesn't have inobtcount feature" > + _notrun "mkfs.xfs doesn't support inobtcount feature" > _try_scratch_mount || \ > - _notrun "inobtcount not supported by scratch filesystem type: $FSTYP" > + _notrun "kernel doesn't support xfs inobtcount feature" > _scratch_unmount > } > diff --git a/tests/xfs/910 b/tests/xfs/910 > new file mode 100755 > index 00000000..237d0a35 > --- /dev/null > +++ b/tests/xfs/910 > @@ -0,0 +1,98 @@ > +#! /bin/bash > +# SPDX-License-Identifier: GPL-2.0-or-later > +# Copyright (c) 2021 Oracle. All Rights Reserved. > +# > +# FS QA Test No. 910 > +# > +# Check that we can upgrade a filesystem to support inobtcount and that > +# everything works properly after the upgrade. > + > +seq=`basename $0` > +seqres=$RESULT_DIR/$seq > +echo "QA output created by $seq" > + > +here=`pwd` > +tmp=/tmp/$$ > +status=1 # failure is the default! > +trap "_cleanup; exit \$status" 0 1 2 3 15 > + > +_cleanup() > +{ > + cd / > + rm -f $tmp.* > +} > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/filter > + > +# real QA test starts here > +_supported_fs xfs > +_require_scratch_xfs_inobtcount > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > +_require_xfs_repair_upgrade inobtcount > + > +rm -f $seqres.full > + > +# Make sure we can't format a filesystem with inobtcount and not finobt. > +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ > + echo "Should not be able to format with inobtcount but not finobt." > + > +# Make sure we can't upgrade a V4 filesystem > +_scratch_mkfs -m crc=0,inobtcount=0,finobt=0 >> $seqres.full > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_check_scratch_xfs_features INOBTCNT > + > +# Make sure we can't upgrade a filesystem to inobtcount without finobt. > +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 >> $seqres.full > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_check_scratch_xfs_features INOBTCNT > + > +# Format V5 filesystem without inode btree counter support and populate it. > +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full > +_scratch_mount > + > +mkdir $SCRATCH_MNT/stress > +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full > +echo moo > $SCRATCH_MNT/urk > + > +_scratch_unmount > + > +# Upgrade filesystem to have the counters and inject failure into repair and > +# make sure that the only path forward is to re-run repair on the filesystem. > +echo "Fail partway through upgrading" > +XFS_REPAIR_FAIL_AFTER_PHASE=2 _scratch_xfs_repair -c inobtcount=1 2>> $seqres.full > +test $? -eq 137 || echo "repair should have been killed??" > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > +_try_scratch_mount &> $tmp.mount > +res=$? > +_filter_scratch < $tmp.mount > +if [ $res -eq 0 ]; then > + echo "needsrepair should have prevented mount" > + _scratch_unmount > +fi > + > +echo "Re-run repair to finish upgrade" > +_scratch_xfs_repair 2>> $seqres.full > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > + > +echo "Filesystem should be usable again" > +_scratch_mount > +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full > +_scratch_unmount > +_check_scratch_fs > +_check_scratch_xfs_features INOBTCNT > + > +echo "Make sure we have nonzero counters" > +_scratch_xfs_db -c 'agi 0' -c 'print ino_blocks fino_blocks' | \ > + sed -e 's/= [1-9]*/= NONZERO/g' > + > +echo "Make sure we can't re-add inobtcount" > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > + > +echo "Mount again, look at our files" > +_scratch_mount >> $seqres.full > +cat $SCRATCH_MNT/urk > + > +status=0 > +exit > diff --git a/tests/xfs/910.out b/tests/xfs/910.out > new file mode 100644 > index 00000000..1bf040d5 > --- /dev/null > +++ b/tests/xfs/910.out > @@ -0,0 +1,23 @@ > +QA output created by 910 > +Running xfs_repair to upgrade filesystem. > +Inode btree count feature only supported on V5 filesystems. > +FEATURES: INOBTCNT:NO > +Running xfs_repair to upgrade filesystem. > +Inode btree count feature requires free inode btree. > +FEATURES: INOBTCNT:NO > +Fail partway through upgrading > +Adding inode btree counts to filesystem. > +FEATURES: NEEDSREPAIR:YES INOBTCNT:YES > +mount: SCRATCH_MNT: mount(2) system call failed: Structure needs cleaning. > +Re-run repair to finish upgrade > +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES > +Filesystem should be usable again > +FEATURES: INOBTCNT:YES > +Make sure we have nonzero counters > +ino_blocks = NONZERO > +fino_blocks = NONZERO > +Make sure we can't re-add inobtcount > +Running xfs_repair to upgrade filesystem. > +Filesystem already has inode btree counts. > +Mount again, look at our files > +moo > diff --git a/tests/xfs/group b/tests/xfs/group > index a2309465..bd47333c 100644 > --- a/tests/xfs/group > +++ b/tests/xfs/group > @@ -526,3 +526,4 @@ > 768 auto quick repair > 770 auto repair > 773 auto quick repair > +910 auto quick inobtcount > ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2021-04-21 0:22 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 2021-04-21 5:52 ` Amir Goldstein 2021-04-21 18:00 ` Brian Foster @ 2021-04-22 0:29 ` Allison Henderson 2021-04-22 0:53 ` [PATCH v2.1 " Darrick J. Wong 3 siblings, 0 replies; 20+ messages in thread From: Allison Henderson @ 2021-04-22 0:29 UTC (permalink / raw) To: Darrick J. Wong, guaneryu; +Cc: linux-xfs, fstests, guan On 4/20/21 5:22 PM, Darrick J. Wong wrote: > From: Darrick J. Wong <djwong@kernel.org> > > Make sure we can actually upgrade filesystems to support inode btree > counters. > > Signed-off-by: Darrick J. Wong <djwong@kernel.org> Seems ok to me Reviewed-by: Allison Henderson <allison.henderson@oracle.com> > --- > common/xfs | 8 +++- > tests/xfs/910 | 98 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/910.out | 23 ++++++++++++ > tests/xfs/group | 1 + > 4 files changed, 127 insertions(+), 3 deletions(-) > create mode 100755 tests/xfs/910 > create mode 100644 tests/xfs/910.out > > > diff --git a/common/xfs b/common/xfs > index 5abc7034..3d660858 100644 > --- a/common/xfs > +++ b/common/xfs > @@ -1153,13 +1153,15 @@ _require_xfs_repair_upgrade() > _notrun "xfs_repair does not support upgrading fs with $type" > } > > -_require_xfs_scratch_inobtcount() > +# Require that the scratch device exists, that mkfs can format with inobtcount > +# enabled, and that the kernel can mount such a filesystem. > +_require_scratch_xfs_inobtcount() > { > _require_scratch > > _scratch_mkfs -m inobtcount=1 &> /dev/null || \ > - _notrun "mkfs.xfs doesn't have inobtcount feature" > + _notrun "mkfs.xfs doesn't support inobtcount feature" > _try_scratch_mount || \ > - _notrun "inobtcount not supported by scratch filesystem type: $FSTYP" > + _notrun "kernel doesn't support xfs inobtcount feature" > _scratch_unmount > } > diff --git a/tests/xfs/910 b/tests/xfs/910 > new file mode 100755 > index 00000000..237d0a35 > --- /dev/null > +++ b/tests/xfs/910 > @@ -0,0 +1,98 @@ > +#! /bin/bash > +# SPDX-License-Identifier: GPL-2.0-or-later > +# Copyright (c) 2021 Oracle. All Rights Reserved. > +# > +# FS QA Test No. 910 > +# > +# Check that we can upgrade a filesystem to support inobtcount and that > +# everything works properly after the upgrade. > + > +seq=`basename $0` > +seqres=$RESULT_DIR/$seq > +echo "QA output created by $seq" > + > +here=`pwd` > +tmp=/tmp/$$ > +status=1 # failure is the default! > +trap "_cleanup; exit \$status" 0 1 2 3 15 > + > +_cleanup() > +{ > + cd / > + rm -f $tmp.* > +} > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/filter > + > +# real QA test starts here > +_supported_fs xfs > +_require_scratch_xfs_inobtcount > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > +_require_xfs_repair_upgrade inobtcount > + > +rm -f $seqres.full > + > +# Make sure we can't format a filesystem with inobtcount and not finobt. > +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ > + echo "Should not be able to format with inobtcount but not finobt." > + > +# Make sure we can't upgrade a V4 filesystem > +_scratch_mkfs -m crc=0,inobtcount=0,finobt=0 >> $seqres.full > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_check_scratch_xfs_features INOBTCNT > + > +# Make sure we can't upgrade a filesystem to inobtcount without finobt. > +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 >> $seqres.full > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_check_scratch_xfs_features INOBTCNT > + > +# Format V5 filesystem without inode btree counter support and populate it. > +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full > +_scratch_mount > + > +mkdir $SCRATCH_MNT/stress > +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full > +echo moo > $SCRATCH_MNT/urk > + > +_scratch_unmount > + > +# Upgrade filesystem to have the counters and inject failure into repair and > +# make sure that the only path forward is to re-run repair on the filesystem. > +echo "Fail partway through upgrading" > +XFS_REPAIR_FAIL_AFTER_PHASE=2 _scratch_xfs_repair -c inobtcount=1 2>> $seqres.full > +test $? -eq 137 || echo "repair should have been killed??" > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > +_try_scratch_mount &> $tmp.mount > +res=$? > +_filter_scratch < $tmp.mount > +if [ $res -eq 0 ]; then > + echo "needsrepair should have prevented mount" > + _scratch_unmount > +fi > + > +echo "Re-run repair to finish upgrade" > +_scratch_xfs_repair 2>> $seqres.full > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > + > +echo "Filesystem should be usable again" > +_scratch_mount > +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full > +_scratch_unmount > +_check_scratch_fs > +_check_scratch_xfs_features INOBTCNT > + > +echo "Make sure we have nonzero counters" > +_scratch_xfs_db -c 'agi 0' -c 'print ino_blocks fino_blocks' | \ > + sed -e 's/= [1-9]*/= NONZERO/g' > + > +echo "Make sure we can't re-add inobtcount" > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > + > +echo "Mount again, look at our files" > +_scratch_mount >> $seqres.full > +cat $SCRATCH_MNT/urk > + > +status=0 > +exit > diff --git a/tests/xfs/910.out b/tests/xfs/910.out > new file mode 100644 > index 00000000..1bf040d5 > --- /dev/null > +++ b/tests/xfs/910.out > @@ -0,0 +1,23 @@ > +QA output created by 910 > +Running xfs_repair to upgrade filesystem. > +Inode btree count feature only supported on V5 filesystems. > +FEATURES: INOBTCNT:NO > +Running xfs_repair to upgrade filesystem. > +Inode btree count feature requires free inode btree. > +FEATURES: INOBTCNT:NO > +Fail partway through upgrading > +Adding inode btree counts to filesystem. > +FEATURES: NEEDSREPAIR:YES INOBTCNT:YES > +mount: SCRATCH_MNT: mount(2) system call failed: Structure needs cleaning. > +Re-run repair to finish upgrade > +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES > +Filesystem should be usable again > +FEATURES: INOBTCNT:YES > +Make sure we have nonzero counters > +ino_blocks = NONZERO > +fino_blocks = NONZERO > +Make sure we can't re-add inobtcount > +Running xfs_repair to upgrade filesystem. > +Filesystem already has inode btree counts. > +Mount again, look at our files > +moo > diff --git a/tests/xfs/group b/tests/xfs/group > index a2309465..bd47333c 100644 > --- a/tests/xfs/group > +++ b/tests/xfs/group > @@ -526,3 +526,4 @@ > 768 auto quick repair > 770 auto repair > 773 auto quick repair > +910 auto quick inobtcount > ^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH v2.1 2/2] xfs: test inobtcount upgrade 2021-04-21 0:22 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong ` (2 preceding siblings ...) 2021-04-22 0:29 ` Allison Henderson @ 2021-04-22 0:53 ` Darrick J. Wong 3 siblings, 0 replies; 20+ messages in thread From: Darrick J. Wong @ 2021-04-22 0:53 UTC (permalink / raw) To: guaneryu; +Cc: linux-xfs, fstests, guan, bfoster, allison.henderson, amir73il From: Darrick J. Wong <djwong@kernel.org> Make sure we can actually upgrade filesystems to support inode btree counters. Signed-off-by: Darrick J. Wong <djwong@kernel.org> Reviewed-by: Brian Foster <bfoster@redhat.com> Reviewed-by: Allison Henderson <allison.henderson@oracle.com> --- v2.1: fix helper bogosity, broken regexps. --- common/xfs | 13 +++++++ tests/xfs/910 | 98 +++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/910.out | 23 ++++++++++++ tests/xfs/group | 1 + 4 files changed, 135 insertions(+) create mode 100755 tests/xfs/910 create mode 100644 tests/xfs/910.out diff --git a/common/xfs b/common/xfs index 4f1a4dba..848311b4 100644 --- a/common/xfs +++ b/common/xfs @@ -1138,3 +1138,16 @@ _require_xfs_repair_upgrade() grep -q 'unknown option' && \ _notrun "xfs_repair does not support upgrading fs with $type" } + +# Require that the scratch device exists, that mkfs can format with inobtcount +# enabled, and that the kernel can mount such a filesystem. +_require_scratch_xfs_inobtcount() +{ + _require_scratch + + _scratch_mkfs -m inobtcount=1 &> /dev/null || \ + _notrun "mkfs.xfs doesn't support inobtcount feature" + _try_scratch_mount || \ + _notrun "kernel doesn't support xfs inobtcount feature" + _scratch_unmount +} diff --git a/tests/xfs/910 b/tests/xfs/910 new file mode 100755 index 00000000..0ed2dc4f --- /dev/null +++ b/tests/xfs/910 @@ -0,0 +1,98 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0-or-later +# Copyright (c) 2021 Oracle. All Rights Reserved. +# +# FS QA Test No. 910 +# +# Check that we can upgrade a filesystem to support inobtcount and that +# everything works properly after the upgrade. + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# real QA test starts here +_supported_fs xfs +_require_scratch_xfs_inobtcount +_require_command "$XFS_ADMIN_PROG" "xfs_admin" +_require_xfs_repair_upgrade inobtcount + +rm -f $seqres.full + +# Make sure we can't format a filesystem with inobtcount and not finobt. +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ + echo "Should not be able to format with inobtcount but not finobt." + +# Make sure we can't upgrade a V4 filesystem +_scratch_mkfs -m crc=0,inobtcount=0,finobt=0 >> $seqres.full +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full +_check_scratch_xfs_features INOBTCNT + +# Make sure we can't upgrade a filesystem to inobtcount without finobt. +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 >> $seqres.full +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full +_check_scratch_xfs_features INOBTCNT + +# Format V5 filesystem without inode btree counter support and populate it. +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full +_scratch_mount + +mkdir $SCRATCH_MNT/stress +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full +echo moo > $SCRATCH_MNT/urk + +_scratch_unmount + +# Upgrade filesystem to have the counters and inject failure into repair and +# make sure that the only path forward is to re-run repair on the filesystem. +echo "Fail partway through upgrading" +XFS_REPAIR_FAIL_AFTER_PHASE=2 _scratch_xfs_repair -c inobtcount=1 2>> $seqres.full +test $? -eq 137 || echo "repair should have been killed??" +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT +_try_scratch_mount &> $tmp.mount +res=$? +_filter_scratch < $tmp.mount +if [ $res -eq 0 ]; then + echo "needsrepair should have prevented mount" + _scratch_unmount +fi + +echo "Re-run repair to finish upgrade" +_scratch_xfs_repair 2>> $seqres.full +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT + +echo "Filesystem should be usable again" +_scratch_mount +$FSSTRESS_PROG -d $SCRATCH_MNT/stress -n 1000 >> $seqres.full +_scratch_unmount +_check_scratch_fs +_check_scratch_xfs_features INOBTCNT + +echo "Make sure we have nonzero counters" +_scratch_xfs_db -c 'agi 0' -c 'print ino_blocks fino_blocks' | \ + sed -e 's/= 0$/= ZERO/g' -e 's/= [0-9]*/= NONZERO/g' + +echo "Make sure we can't re-add inobtcount" +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full + +echo "Mount again, look at our files" +_scratch_mount >> $seqres.full +cat $SCRATCH_MNT/urk + +status=0 +exit diff --git a/tests/xfs/910.out b/tests/xfs/910.out new file mode 100644 index 00000000..1bf040d5 --- /dev/null +++ b/tests/xfs/910.out @@ -0,0 +1,23 @@ +QA output created by 910 +Running xfs_repair to upgrade filesystem. +Inode btree count feature only supported on V5 filesystems. +FEATURES: INOBTCNT:NO +Running xfs_repair to upgrade filesystem. +Inode btree count feature requires free inode btree. +FEATURES: INOBTCNT:NO +Fail partway through upgrading +Adding inode btree counts to filesystem. +FEATURES: NEEDSREPAIR:YES INOBTCNT:YES +mount: SCRATCH_MNT: mount(2) system call failed: Structure needs cleaning. +Re-run repair to finish upgrade +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES +Filesystem should be usable again +FEATURES: INOBTCNT:YES +Make sure we have nonzero counters +ino_blocks = NONZERO +fino_blocks = NONZERO +Make sure we can't re-add inobtcount +Running xfs_repair to upgrade filesystem. +Filesystem already has inode btree counts. +Mount again, look at our files +moo diff --git a/tests/xfs/group b/tests/xfs/group index 4231d1d5..4c70cff1 100644 --- a/tests/xfs/group +++ b/tests/xfs/group @@ -526,3 +526,4 @@ 768 auto quick repair 770 auto repair 773 auto quick admin +910 auto quick inobtcount ^ permalink raw reply related [flat|nested] 20+ messages in thread
* [PATCHSET v2 0/2] fstests: add inode btree blocks counters to the AGI header @ 2021-03-31 1:08 Darrick J. Wong 2021-03-31 1:08 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 0 siblings, 1 reply; 20+ messages in thread From: Darrick J. Wong @ 2021-03-31 1:08 UTC (permalink / raw) To: djwong, guaneryu; +Cc: linux-xfs, fstests, guan Hi all, Years ago, Christoph diagnosed a problem where freeing an inode on a totally full filesystem could fail due to finobt expansion not being able to allocate enough blocks. He solved the problem by using the per-AG block reservation system to ensure that there are always enough blocks for finobt expansion, but that came at the cost of having to walk the entire finobt at mount time. This new feature solves that performance regression by adding inode btree block counts to the AGI header. The patches in this series amend fstests to handle the new metadata fields and to test that upgrades work properly. v2: move the xfs_admin functional test into a separate patch file and split them into one test for basic functionality and another for the extended functionality testing that modifies the test run config If you're going to start using this mess, you probably ought to just pull from my git trees, which are linked below. This is an extraordinary way to destroy everything. Enjoy! Comments and questions are, as always, welcome. --D kernel git tree: https://git.kernel.org/cgit/linux/kernel/git/djwong/xfs-linux.git/log/?h=inobt-counters xfsprogs git tree: https://git.kernel.org/cgit/linux/kernel/git/djwong/xfsprogs-dev.git/log/?h=inobt-counters fstests git tree: https://git.kernel.org/cgit/linux/kernel/git/djwong/xfstests-dev.git/log/?h=inobt-counters --- common/xfs | 21 ++++++++++ tests/xfs/764 | 91 ++++++++++++++++++++++++++++++++++++++++++ tests/xfs/764.out | 17 ++++++++ tests/xfs/773 | 114 +++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/773.out | 19 +++++++++ tests/xfs/910 | 112 ++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/910.out | 23 +++++++++++ tests/xfs/group | 3 + 8 files changed, 400 insertions(+) create mode 100755 tests/xfs/764 create mode 100644 tests/xfs/764.out create mode 100755 tests/xfs/773 create mode 100644 tests/xfs/773.out create mode 100755 tests/xfs/910 create mode 100644 tests/xfs/910.out ^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH 2/2] xfs: test inobtcount upgrade 2021-03-31 1:08 [PATCHSET v2 0/2] fstests: add inode btree blocks counters to the AGI header Darrick J. Wong @ 2021-03-31 1:08 ` Darrick J. Wong 2021-03-31 17:36 ` Brian Foster 0 siblings, 1 reply; 20+ messages in thread From: Darrick J. Wong @ 2021-03-31 1:08 UTC (permalink / raw) To: djwong, guaneryu; +Cc: linux-xfs, fstests, guan From: Darrick J. Wong <djwong@kernel.org> Make sure we can actually upgrade filesystems to support inode btree counters. Signed-off-by: Darrick J. Wong <djwong@kernel.org> --- tests/xfs/910 | 112 +++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/910.out | 23 +++++++++++ tests/xfs/group | 1 3 files changed, 136 insertions(+) create mode 100755 tests/xfs/910 create mode 100644 tests/xfs/910.out diff --git a/tests/xfs/910 b/tests/xfs/910 new file mode 100755 index 00000000..5f095324 --- /dev/null +++ b/tests/xfs/910 @@ -0,0 +1,112 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0-or-later +# Copyright (c) 2021 Oracle. All Rights Reserved. +# +# FS QA Test No. 910 +# +# Check that we can upgrade a filesystem to support inobtcount and that +# everything works properly after the upgrade. + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# real QA test starts here +_supported_fs xfs +_require_xfs_scratch_inobtcount +_require_command "$XFS_ADMIN_PROG" "xfs_admin" +_require_xfs_repair_upgrade inobtcount + +rm -f $seqres.full + +# Make sure we can't format a filesystem with inobtcount and not finobt. +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ + echo "Should not be able to format with inobtcount but not finobt." + +# Make sure we can't upgrade a V4 filesystem +_scratch_mkfs -m crc=0,inobtcount=0,finobt=0 >> $seqres.full +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full +_check_scratch_xfs_features INOBTCNT + +# Make sure we can't upgrade a filesystem to inobtcount without finobt. +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 >> $seqres.full +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full +_check_scratch_xfs_features INOBTCNT + +# Format V5 filesystem without inode btree counter support and upgrade it. +# Inject failure into repair and make sure that the only path forward is +# to re-run repair on the filesystem. +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full +echo "Fail partway through upgrading" +XFS_REPAIR_FAIL_AFTER_PHASE=2 _scratch_xfs_repair -c inobtcount=1 2>> $seqres.full +test $? -eq 137 || echo "repair should have been killed??" +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT +_try_scratch_mount &> $tmp.mount +res=$? +_filter_scratch < $tmp.mount +if [ $res -eq 0 ]; then + echo "needsrepair should have prevented mount" + _scratch_unmount +fi + +echo "Re-run repair to finish upgrade" +_scratch_xfs_repair 2>> $seqres.full +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT + +echo "Filesystem should be usable again" +_try_scratch_mount &> $tmp.mount +res=$? +_filter_scratch < $tmp.mount +if [ $res -eq 0 ]; then + _scratch_unmount +else + echo "mount should succeed after second repair" +fi +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT + +# Format V5 filesystem without inode btree counter support and upgrade it. +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' >> $seqres.full +_scratch_mount >> $seqres.full + +echo moo > $SCRATCH_MNT/urk + +_scratch_unmount +_check_scratch_fs + +# Now upgrade to inobtcount support +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full +_check_scratch_xfs_features INOBTCNT +_check_scratch_fs +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' -c 'agi 0' -c 'p' >> $seqres.full + +# Make sure we have nonzero counters +_scratch_xfs_db -c 'agi 0' -c 'print ino_blocks fino_blocks' | \ + sed -e 's/= [1-9]*/= NONZERO/g' + +# Mount again, look at our files +_scratch_mount >> $seqres.full +cat $SCRATCH_MNT/urk + +# Make sure we can't re-add inobtcount +_scratch_unmount +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full +_scratch_mount >> $seqres.full + +status=0 +exit diff --git a/tests/xfs/910.out b/tests/xfs/910.out new file mode 100644 index 00000000..ed78d88f --- /dev/null +++ b/tests/xfs/910.out @@ -0,0 +1,23 @@ +QA output created by 910 +Running xfs_repair to upgrade filesystem. +Inode btree count feature only supported on V5 filesystems. +FEATURES: INOBTCNT:NO +Running xfs_repair to upgrade filesystem. +Inode btree count feature requires free inode btree. +FEATURES: INOBTCNT:NO +Fail partway through upgrading +Adding inode btree counts to filesystem. +FEATURES: NEEDSREPAIR:YES INOBTCNT:YES +mount: SCRATCH_MNT: mount(2) system call failed: Structure needs cleaning. +Re-run repair to finish upgrade +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES +Filesystem should be usable again +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES +Running xfs_repair to upgrade filesystem. +Adding inode btree counts to filesystem. +FEATURES: INOBTCNT:YES +ino_blocks = NONZERO +fino_blocks = NONZERO +moo +Running xfs_repair to upgrade filesystem. +Filesystem already has inode btree counts. diff --git a/tests/xfs/group b/tests/xfs/group index 5801471b..0dc8038a 100644 --- a/tests/xfs/group +++ b/tests/xfs/group @@ -524,3 +524,4 @@ 768 auto quick repair 770 auto repair 773 auto quick repair +910 auto quick inobtcount ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2021-03-31 1:08 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong @ 2021-03-31 17:36 ` Brian Foster 2021-04-03 0:29 ` Darrick J. Wong 0 siblings, 1 reply; 20+ messages in thread From: Brian Foster @ 2021-03-31 17:36 UTC (permalink / raw) To: Darrick J. Wong; +Cc: guaneryu, linux-xfs, fstests, guan On Tue, Mar 30, 2021 at 06:08:34PM -0700, Darrick J. Wong wrote: > From: Darrick J. Wong <djwong@kernel.org> > > Make sure we can actually upgrade filesystems to support inode btree > counters. > > Signed-off-by: Darrick J. Wong <djwong@kernel.org> > --- > tests/xfs/910 | 112 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/910.out | 23 +++++++++++ > tests/xfs/group | 1 > 3 files changed, 136 insertions(+) > create mode 100755 tests/xfs/910 > create mode 100644 tests/xfs/910.out > > > diff --git a/tests/xfs/910 b/tests/xfs/910 > new file mode 100755 > index 00000000..5f095324 > --- /dev/null > +++ b/tests/xfs/910 > @@ -0,0 +1,112 @@ > +#! /bin/bash > +# SPDX-License-Identifier: GPL-2.0-or-later > +# Copyright (c) 2021 Oracle. All Rights Reserved. > +# > +# FS QA Test No. 910 > +# > +# Check that we can upgrade a filesystem to support inobtcount and that > +# everything works properly after the upgrade. > + > +seq=`basename $0` > +seqres=$RESULT_DIR/$seq > +echo "QA output created by $seq" > + > +here=`pwd` > +tmp=/tmp/$$ > +status=1 # failure is the default! > +trap "_cleanup; exit \$status" 0 1 2 3 15 > + > +_cleanup() > +{ > + cd / > + rm -f $tmp.* > +} > + > +# get standard environment, filters and checks > +. ./common/rc > +. ./common/filter > + > +# real QA test starts here > +_supported_fs xfs > +_require_xfs_scratch_inobtcount > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > +_require_xfs_repair_upgrade inobtcount > + > +rm -f $seqres.full > + > +# Make sure we can't format a filesystem with inobtcount and not finobt. > +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ > + echo "Should not be able to format with inobtcount but not finobt." > + > +# Make sure we can't upgrade a V4 filesystem > +_scratch_mkfs -m crc=0,inobtcount=0,finobt=0 >> $seqres.full > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_check_scratch_xfs_features INOBTCNT > + > +# Make sure we can't upgrade a filesystem to inobtcount without finobt. > +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 >> $seqres.full > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_check_scratch_xfs_features INOBTCNT > + > +# Format V5 filesystem without inode btree counter support and upgrade it. > +# Inject failure into repair and make sure that the only path forward is > +# to re-run repair on the filesystem. > +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full > +echo "Fail partway through upgrading" > +XFS_REPAIR_FAIL_AFTER_PHASE=2 _scratch_xfs_repair -c inobtcount=1 2>> $seqres.full > +test $? -eq 137 || echo "repair should have been killed??" > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > +_try_scratch_mount &> $tmp.mount > +res=$? > +_filter_scratch < $tmp.mount > +if [ $res -eq 0 ]; then > + echo "needsrepair should have prevented mount" > + _scratch_unmount > +fi > + > +echo "Re-run repair to finish upgrade" > +_scratch_xfs_repair 2>> $seqres.full > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > + > +echo "Filesystem should be usable again" > +_try_scratch_mount &> $tmp.mount > +res=$? > +_filter_scratch < $tmp.mount > +if [ $res -eq 0 ]; then > + _scratch_unmount > +else > + echo "mount should succeed after second repair" > +fi > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > + As you're probably aware I'm not a huge fan of sprinkling these kind of repair/mount checks throughout the feature level tests. I don't think it's reasonable to expect this to become a consistent pattern, so over time this will likely just obfuscate the purpose of some of these tests. That said, it seems to be harmless here so I'll just note that as my .02. > +# Format V5 filesystem without inode btree counter support and upgrade it. > +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full > +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' >> $seqres.full > +_scratch_mount >> $seqres.full > + > +echo moo > $SCRATCH_MNT/urk > + > +_scratch_unmount > +_check_scratch_fs > + > +# Now upgrade to inobtcount support > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_check_scratch_xfs_features INOBTCNT > +_check_scratch_fs > +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' -c 'agi 0' -c 'p' >> $seqres.full > + This looks nearly the same as the previous test with the exception of not doing repair failure injection and instead creating a file and checking the resulting counters. Could we combine these two sequences and simultaneously run a slightly more randomized test? E.g., a logical sequence along the lines of: mkfs inobtcount=0 mount run fsstress -n 500 or otherwise (quickly) populate w/ some randomized content umount repair w/ fail injection repair w/o fail injection check resulting counters Hm? Brian > +# Make sure we have nonzero counters > +_scratch_xfs_db -c 'agi 0' -c 'print ino_blocks fino_blocks' | \ > + sed -e 's/= [1-9]*/= NONZERO/g' > + > +# Mount again, look at our files > +_scratch_mount >> $seqres.full > +cat $SCRATCH_MNT/urk > + > +# Make sure we can't re-add inobtcount > +_scratch_unmount > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > +_scratch_mount >> $seqres.full > + > +status=0 > +exit > diff --git a/tests/xfs/910.out b/tests/xfs/910.out > new file mode 100644 > index 00000000..ed78d88f > --- /dev/null > +++ b/tests/xfs/910.out > @@ -0,0 +1,23 @@ > +QA output created by 910 > +Running xfs_repair to upgrade filesystem. > +Inode btree count feature only supported on V5 filesystems. > +FEATURES: INOBTCNT:NO > +Running xfs_repair to upgrade filesystem. > +Inode btree count feature requires free inode btree. > +FEATURES: INOBTCNT:NO > +Fail partway through upgrading > +Adding inode btree counts to filesystem. > +FEATURES: NEEDSREPAIR:YES INOBTCNT:YES > +mount: SCRATCH_MNT: mount(2) system call failed: Structure needs cleaning. > +Re-run repair to finish upgrade > +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES > +Filesystem should be usable again > +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES > +Running xfs_repair to upgrade filesystem. > +Adding inode btree counts to filesystem. > +FEATURES: INOBTCNT:YES > +ino_blocks = NONZERO > +fino_blocks = NONZERO > +moo > +Running xfs_repair to upgrade filesystem. > +Filesystem already has inode btree counts. > diff --git a/tests/xfs/group b/tests/xfs/group > index 5801471b..0dc8038a 100644 > --- a/tests/xfs/group > +++ b/tests/xfs/group > @@ -524,3 +524,4 @@ > 768 auto quick repair > 770 auto repair > 773 auto quick repair > +910 auto quick inobtcount > ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2021-03-31 17:36 ` Brian Foster @ 2021-04-03 0:29 ` Darrick J. Wong 0 siblings, 0 replies; 20+ messages in thread From: Darrick J. Wong @ 2021-04-03 0:29 UTC (permalink / raw) To: Brian Foster; +Cc: guaneryu, linux-xfs, fstests, guan On Wed, Mar 31, 2021 at 01:36:47PM -0400, Brian Foster wrote: > On Tue, Mar 30, 2021 at 06:08:34PM -0700, Darrick J. Wong wrote: > > From: Darrick J. Wong <djwong@kernel.org> > > > > Make sure we can actually upgrade filesystems to support inode btree > > counters. > > > > Signed-off-by: Darrick J. Wong <djwong@kernel.org> > > --- > > tests/xfs/910 | 112 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > > tests/xfs/910.out | 23 +++++++++++ > > tests/xfs/group | 1 > > 3 files changed, 136 insertions(+) > > create mode 100755 tests/xfs/910 > > create mode 100644 tests/xfs/910.out > > > > > > diff --git a/tests/xfs/910 b/tests/xfs/910 > > new file mode 100755 > > index 00000000..5f095324 > > --- /dev/null > > +++ b/tests/xfs/910 > > @@ -0,0 +1,112 @@ > > +#! /bin/bash > > +# SPDX-License-Identifier: GPL-2.0-or-later > > +# Copyright (c) 2021 Oracle. All Rights Reserved. > > +# > > +# FS QA Test No. 910 > > +# > > +# Check that we can upgrade a filesystem to support inobtcount and that > > +# everything works properly after the upgrade. > > + > > +seq=`basename $0` > > +seqres=$RESULT_DIR/$seq > > +echo "QA output created by $seq" > > + > > +here=`pwd` > > +tmp=/tmp/$$ > > +status=1 # failure is the default! > > +trap "_cleanup; exit \$status" 0 1 2 3 15 > > + > > +_cleanup() > > +{ > > + cd / > > + rm -f $tmp.* > > +} > > + > > +# get standard environment, filters and checks > > +. ./common/rc > > +. ./common/filter > > + > > +# real QA test starts here > > +_supported_fs xfs > > +_require_xfs_scratch_inobtcount > > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > > +_require_xfs_repair_upgrade inobtcount > > + > > +rm -f $seqres.full > > + > > +# Make sure we can't format a filesystem with inobtcount and not finobt. > > +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ > > + echo "Should not be able to format with inobtcount but not finobt." > > + > > +# Make sure we can't upgrade a V4 filesystem > > +_scratch_mkfs -m crc=0,inobtcount=0,finobt=0 >> $seqres.full > > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > > +_check_scratch_xfs_features INOBTCNT > > + > > +# Make sure we can't upgrade a filesystem to inobtcount without finobt. > > +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 >> $seqres.full > > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > > +_check_scratch_xfs_features INOBTCNT > > + > > +# Format V5 filesystem without inode btree counter support and upgrade it. > > +# Inject failure into repair and make sure that the only path forward is > > +# to re-run repair on the filesystem. > > +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full > > +echo "Fail partway through upgrading" > > +XFS_REPAIR_FAIL_AFTER_PHASE=2 _scratch_xfs_repair -c inobtcount=1 2>> $seqres.full > > +test $? -eq 137 || echo "repair should have been killed??" > > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > > +_try_scratch_mount &> $tmp.mount > > +res=$? > > +_filter_scratch < $tmp.mount > > +if [ $res -eq 0 ]; then > > + echo "needsrepair should have prevented mount" > > + _scratch_unmount > > +fi > > + > > +echo "Re-run repair to finish upgrade" > > +_scratch_xfs_repair 2>> $seqres.full > > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > > + > > +echo "Filesystem should be usable again" > > +_try_scratch_mount &> $tmp.mount > > +res=$? > > +_filter_scratch < $tmp.mount > > +if [ $res -eq 0 ]; then > > + _scratch_unmount > > +else > > + echo "mount should succeed after second repair" > > +fi > > +_check_scratch_xfs_features NEEDSREPAIR INOBTCNT > > + > > As you're probably aware I'm not a huge fan of sprinkling these kind of > repair/mount checks throughout the feature level tests. I don't think > it's reasonable to expect this to become a consistent pattern, so over > time this will likely just obfuscate the purpose of some of these tests. > That said, it seems to be harmless here so I'll just note that as my > .02. Ok. I think inobtcount is the only feature upgrade test that checks it. > > +# Format V5 filesystem without inode btree counter support and upgrade it. > > +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full > > +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' >> $seqres.full > > +_scratch_mount >> $seqres.full > > + > > +echo moo > $SCRATCH_MNT/urk > > + > > +_scratch_unmount > > +_check_scratch_fs > > + > > +# Now upgrade to inobtcount support > > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > > +_check_scratch_xfs_features INOBTCNT > > +_check_scratch_fs > > +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' -c 'agi 0' -c 'p' >> $seqres.full > > + > > This looks nearly the same as the previous test with the exception of > not doing repair failure injection and instead creating a file and > checking the resulting counters. Could we combine these two sequences > and simultaneously run a slightly more randomized test? E.g., a logical > sequence along the lines of: > > mkfs inobtcount=0 > mount > run fsstress -n 500 or otherwise (quickly) populate w/ some randomized content > umount > repair w/ fail injection > repair w/o fail injection > check resulting counters > > Hm? Yes, the two cases can be combined. Will do, thanks for the feedback! --D > > Brian > > > +# Make sure we have nonzero counters > > +_scratch_xfs_db -c 'agi 0' -c 'print ino_blocks fino_blocks' | \ > > + sed -e 's/= [1-9]*/= NONZERO/g' > > + > > +# Mount again, look at our files > > +_scratch_mount >> $seqres.full > > +cat $SCRATCH_MNT/urk > > + > > +# Make sure we can't re-add inobtcount > > +_scratch_unmount > > +_scratch_xfs_admin -O inobtcount=1 2>> $seqres.full > > +_scratch_mount >> $seqres.full > > + > > +status=0 > > +exit > > diff --git a/tests/xfs/910.out b/tests/xfs/910.out > > new file mode 100644 > > index 00000000..ed78d88f > > --- /dev/null > > +++ b/tests/xfs/910.out > > @@ -0,0 +1,23 @@ > > +QA output created by 910 > > +Running xfs_repair to upgrade filesystem. > > +Inode btree count feature only supported on V5 filesystems. > > +FEATURES: INOBTCNT:NO > > +Running xfs_repair to upgrade filesystem. > > +Inode btree count feature requires free inode btree. > > +FEATURES: INOBTCNT:NO > > +Fail partway through upgrading > > +Adding inode btree counts to filesystem. > > +FEATURES: NEEDSREPAIR:YES INOBTCNT:YES > > +mount: SCRATCH_MNT: mount(2) system call failed: Structure needs cleaning. > > +Re-run repair to finish upgrade > > +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES > > +Filesystem should be usable again > > +FEATURES: NEEDSREPAIR:NO INOBTCNT:YES > > +Running xfs_repair to upgrade filesystem. > > +Adding inode btree counts to filesystem. > > +FEATURES: INOBTCNT:YES > > +ino_blocks = NONZERO > > +fino_blocks = NONZERO > > +moo > > +Running xfs_repair to upgrade filesystem. > > +Filesystem already has inode btree counts. > > diff --git a/tests/xfs/group b/tests/xfs/group > > index 5801471b..0dc8038a 100644 > > --- a/tests/xfs/group > > +++ b/tests/xfs/group > > @@ -524,3 +524,4 @@ > > 768 auto quick repair > > 770 auto repair > > 773 auto quick repair > > +910 auto quick inobtcount > > > ^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH RFCv4 0/2] xfstests: add inode btree blocks counters to the AGI header @ 2020-10-27 19:03 Darrick J. Wong 2020-10-27 19:03 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 0 siblings, 1 reply; 20+ messages in thread From: Darrick J. Wong @ 2020-10-27 19:03 UTC (permalink / raw) To: darrick.wong, guaneryu; +Cc: linux-xfs, fstests Hi all, Years ago, Christoph diagnosed a problem where freeing an inode on a totally full filesystem could fail due to finobt expansion not being able to allocate enough blocks. He solved the problem by using the per-AG block reservation system to ensure that there are always enough blocks for finobt expansion, but that came at the cost of having to walk the entire finobt at mount time. This new feature solves that performance regression by adding inode btree block counts to the AGI header. The patches in this series amend fstests to handle the new metadata fields and to test that upgrades work properly. v2: rebase kernel to 5.9 v3: split logical changes into separate patches v4: support inobtcounts && !finobt properly If you're going to start using this mess, you probably ought to just pull from my git trees, which are linked below. This is an extraordinary way to destroy everything. Enjoy! Comments and questions are, as always, welcome. --D kernel git tree: https://git.kernel.org/cgit/linux/kernel/git/djwong/xfs-linux.git/log/?h=inobt-counters xfsprogs git tree: https://git.kernel.org/cgit/linux/kernel/git/djwong/xfsprogs-dev.git/log/?h=inobt-counters fstests git tree: https://git.kernel.org/cgit/linux/kernel/git/djwong/xfstests-dev.git/log/?h=inobt-counters --- common/xfs | 16 ++++++++++++ tests/xfs/010 | 3 +- tests/xfs/030 | 2 + tests/xfs/122.out | 2 + tests/xfs/910 | 72 +++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/910.out | 3 ++ tests/xfs/group | 1 + 7 files changed, 97 insertions(+), 2 deletions(-) create mode 100755 tests/xfs/910 create mode 100644 tests/xfs/910.out ^ permalink raw reply [flat|nested] 20+ messages in thread
* [PATCH 2/2] xfs: test inobtcount upgrade 2020-10-27 19:03 [PATCH RFCv4 0/2] xfstests: add inode btree blocks counters to the AGI header Darrick J. Wong @ 2020-10-27 19:03 ` Darrick J. Wong 2020-10-29 17:40 ` Brian Foster 2020-11-18 16:44 ` Eric Sandeen 0 siblings, 2 replies; 20+ messages in thread From: Darrick J. Wong @ 2020-10-27 19:03 UTC (permalink / raw) To: darrick.wong, guaneryu; +Cc: linux-xfs, fstests From: Darrick J. Wong <darrick.wong@oracle.com> Make sure we can actually upgrade filesystems to support inobtcounts. Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> --- common/xfs | 16 ++++++++++++ tests/xfs/910 | 72 +++++++++++++++++++++++++++++++++++++++++++++++++++++ tests/xfs/910.out | 3 ++ tests/xfs/group | 1 + 4 files changed, 92 insertions(+) create mode 100755 tests/xfs/910 create mode 100644 tests/xfs/910.out diff --git a/common/xfs b/common/xfs index 3f5c14ba..e548a0a1 100644 --- a/common/xfs +++ b/common/xfs @@ -978,3 +978,19 @@ _require_xfs_copy() [ "$USE_EXTERNAL" = yes ] && \ _notrun "Cannot xfs_copy with external devices" } + +_require_xfs_mkfs_inobtcount() +{ + _scratch_mkfs_xfs_supported -m inobtcount=1 >/dev/null 2>&1 \ + || _notrun "mkfs.xfs doesn't have inobtcount feature" +} + +_require_xfs_scratch_inobtcount() +{ + _require_scratch + + _scratch_mkfs -m inobtcount=1 > /dev/null + _try_scratch_mount || \ + _notrun "inobtcount not supported by scratch filesystem type: $FSTYP" + _scratch_unmount +} diff --git a/tests/xfs/910 b/tests/xfs/910 new file mode 100755 index 00000000..1924d9ea --- /dev/null +++ b/tests/xfs/910 @@ -0,0 +1,72 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0-or-later +# Copyright (c) 2020, Oracle and/or its affiliates. All Rights Reserved. +# +# FS QA Test No. 910 +# +# Check that we can upgrade a filesystem to support inobtcount and that +# everything works properly after the upgrade. + +seq=`basename $0` +seqres=$RESULT_DIR/$seq +echo "QA output created by $seq" + +here=`pwd` +tmp=/tmp/$$ +status=1 # failure is the default! +trap "_cleanup; exit \$status" 0 1 2 3 15 + +_cleanup() +{ + cd / + rm -f $tmp.* +} + +# get standard environment, filters and checks +. ./common/rc +. ./common/filter + +# We have very specific formatting parameters, so don't let things get complex +# with realtime devices and external logs. +unset USE_EXTERNAL + +# real QA test starts here +_supported_fs xfs +_require_command "$XFS_ADMIN_PROG" "xfs_admin" +_require_xfs_mkfs_inobtcount +_require_xfs_scratch_inobtcount + +rm -f $seqres.full + +# Make sure we can't format a filesystem with inobtcount and not finobt. +_scratch_mkfs -m crc=1,inobtcount=1,finobt=0 &> $seqres.full && \ + echo "Should not be able to format with inobtcount but not finobt." + +# Make sure we can't upgrade a filesystem to inobtcount without finobt. +_scratch_mkfs -m crc=1,inobtcount=0,finobt=0 &> $seqres.full +_scratch_xfs_admin -O inobtcount >> $seqres.full && \ + echo "Should not be able to upgrade to inobtcount without finobt." + +# Format V5 filesystem without inode btree counter support and populate it +_scratch_mkfs -m crc=1,inobtcount=0 >> $seqres.full +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' >> $seqres.full +_scratch_mount >> $seqres.full + +echo moo > $SCRATCH_MNT/urk + +_scratch_unmount +_check_scratch_fs + +# Now upgrade to inobtcount support +_scratch_xfs_admin -O inobtcount >> $seqres.full +_check_scratch_fs +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' -c 'agi 0' -c 'p' >> $seqres.full + +# Mount again, look at our files +_scratch_mount >> $seqres.full +cat $SCRATCH_MNT/urk + +# success, all done +echo Silence is golden. +status=0 +exit diff --git a/tests/xfs/910.out b/tests/xfs/910.out new file mode 100644 index 00000000..83992f49 --- /dev/null +++ b/tests/xfs/910.out @@ -0,0 +1,3 @@ +QA output created by 910 +moo +Silence is golden. diff --git a/tests/xfs/group b/tests/xfs/group index 4b0caea4..862df3be 100644 --- a/tests/xfs/group +++ b/tests/xfs/group @@ -524,6 +524,7 @@ 760 auto quick rw collapse punch insert zero prealloc 761 auto quick realtime 763 auto quick rw realtime +910 auto quick inobtcount 915 auto quick quota 917 auto quick db 918 auto quick db ^ permalink raw reply related [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2020-10-27 19:03 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong @ 2020-10-29 17:40 ` Brian Foster 2020-10-29 20:31 ` Darrick J. Wong 2020-11-18 16:44 ` Eric Sandeen 1 sibling, 1 reply; 20+ messages in thread From: Brian Foster @ 2020-10-29 17:40 UTC (permalink / raw) To: Darrick J. Wong; +Cc: guaneryu, linux-xfs, fstests On Tue, Oct 27, 2020 at 12:03:48PM -0700, Darrick J. Wong wrote: > From: Darrick J. Wong <darrick.wong@oracle.com> > > Make sure we can actually upgrade filesystems to support inobtcounts. > > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> > --- > common/xfs | 16 ++++++++++++ > tests/xfs/910 | 72 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/910.out | 3 ++ > tests/xfs/group | 1 + > 4 files changed, 92 insertions(+) > create mode 100755 tests/xfs/910 > create mode 100644 tests/xfs/910.out > > ... > diff --git a/tests/xfs/910 b/tests/xfs/910 > new file mode 100755 > index 00000000..1924d9ea > --- /dev/null > +++ b/tests/xfs/910 > @@ -0,0 +1,72 @@ ... > + > +# Now upgrade to inobtcount support > +_scratch_xfs_admin -O inobtcount >> $seqres.full > +_check_scratch_fs > +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' -c 'agi 0' -c 'p' >> $seqres.full > + > +# Mount again, look at our files > +_scratch_mount >> $seqres.full > +cat $SCRATCH_MNT/urk > + I think we probably want some more explicit form of validation here. Perhaps dump the inobt block counters from the above xfs_db command to the golden output..? As it is, we can comment out the xfs_admin command and the test still passes. > +# success, all done > +echo Silence is golden. We can also probably drop this if we have some other form of output from the test. Brian > +status=0 > +exit > diff --git a/tests/xfs/910.out b/tests/xfs/910.out > new file mode 100644 > index 00000000..83992f49 > --- /dev/null > +++ b/tests/xfs/910.out > @@ -0,0 +1,3 @@ > +QA output created by 910 > +moo > +Silence is golden. > diff --git a/tests/xfs/group b/tests/xfs/group > index 4b0caea4..862df3be 100644 > --- a/tests/xfs/group > +++ b/tests/xfs/group > @@ -524,6 +524,7 @@ > 760 auto quick rw collapse punch insert zero prealloc > 761 auto quick realtime > 763 auto quick rw realtime > +910 auto quick inobtcount > 915 auto quick quota > 917 auto quick db > 918 auto quick db > ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2020-10-29 17:40 ` Brian Foster @ 2020-10-29 20:31 ` Darrick J. Wong 0 siblings, 0 replies; 20+ messages in thread From: Darrick J. Wong @ 2020-10-29 20:31 UTC (permalink / raw) To: Brian Foster; +Cc: guaneryu, linux-xfs, fstests On Thu, Oct 29, 2020 at 01:40:23PM -0400, Brian Foster wrote: > On Tue, Oct 27, 2020 at 12:03:48PM -0700, Darrick J. Wong wrote: > > From: Darrick J. Wong <darrick.wong@oracle.com> > > > > Make sure we can actually upgrade filesystems to support inobtcounts. > > > > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> > > --- > > common/xfs | 16 ++++++++++++ > > tests/xfs/910 | 72 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > > tests/xfs/910.out | 3 ++ > > tests/xfs/group | 1 + > > 4 files changed, 92 insertions(+) > > create mode 100755 tests/xfs/910 > > create mode 100644 tests/xfs/910.out > > > > > ... > > diff --git a/tests/xfs/910 b/tests/xfs/910 > > new file mode 100755 > > index 00000000..1924d9ea > > --- /dev/null > > +++ b/tests/xfs/910 > > @@ -0,0 +1,72 @@ > ... > > + > > +# Now upgrade to inobtcount support > > +_scratch_xfs_admin -O inobtcount >> $seqres.full > > +_check_scratch_fs > > +_scratch_xfs_db -c 'version' -c 'sb 0' -c 'p' -c 'agi 0' -c 'p' >> $seqres.full > > + > > +# Mount again, look at our files > > +_scratch_mount >> $seqres.full > > +cat $SCRATCH_MNT/urk > > + > > I think we probably want some more explicit form of validation here. > Perhaps dump the inobt block counters from the above xfs_db command to > the golden output..? As it is, we can comment out the xfs_admin command > and the test still passes. Ok. > > +# success, all done > > +echo Silence is golden. > > We can also probably drop this if we have some other form of output from > the test. Will do. Thanks for reviewing! > Brian > > > +status=0 > > +exit > > diff --git a/tests/xfs/910.out b/tests/xfs/910.out > > new file mode 100644 > > index 00000000..83992f49 > > --- /dev/null > > +++ b/tests/xfs/910.out > > @@ -0,0 +1,3 @@ > > +QA output created by 910 > > +moo > > +Silence is golden. > > diff --git a/tests/xfs/group b/tests/xfs/group > > index 4b0caea4..862df3be 100644 > > --- a/tests/xfs/group > > +++ b/tests/xfs/group > > @@ -524,6 +524,7 @@ > > 760 auto quick rw collapse punch insert zero prealloc > > 761 auto quick realtime > > 763 auto quick rw realtime > > +910 auto quick inobtcount > > 915 auto quick quota > > 917 auto quick db > > 918 auto quick db > > > ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2020-10-27 19:03 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 2020-10-29 17:40 ` Brian Foster @ 2020-11-18 16:44 ` Eric Sandeen 2020-11-20 1:53 ` Darrick J. Wong 1 sibling, 1 reply; 20+ messages in thread From: Eric Sandeen @ 2020-11-18 16:44 UTC (permalink / raw) To: Darrick J. Wong, guaneryu; +Cc: linux-xfs, fstests On 10/27/20 2:03 PM, Darrick J. Wong wrote: > From: Darrick J. Wong <darrick.wong@oracle.com> > > Make sure we can actually upgrade filesystems to support inobtcounts. > > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> > --- > common/xfs | 16 ++++++++++++ > tests/xfs/910 | 72 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > tests/xfs/910.out | 3 ++ > tests/xfs/group | 1 + > 4 files changed, 92 insertions(+) > create mode 100755 tests/xfs/910 > create mode 100644 tests/xfs/910.out > > > diff --git a/common/xfs b/common/xfs > index 3f5c14ba..e548a0a1 100644 > --- a/common/xfs > +++ b/common/xfs > @@ -978,3 +978,19 @@ _require_xfs_copy() > [ "$USE_EXTERNAL" = yes ] && \ > _notrun "Cannot xfs_copy with external devices" > } > + > +_require_xfs_mkfs_inobtcount() > +{ > + _scratch_mkfs_xfs_supported -m inobtcount=1 >/dev/null 2>&1 \ > + || _notrun "mkfs.xfs doesn't have inobtcount feature" > +} I'd like to also add: +_require_xfs_admin_upgrade() +{ + local feature="$1" + + _require_scratch + # Catch missing "-O" or missing feature handling + _scratch_xfs_admin -O $feature 2>&1 | grep "illegal option\|Cannot change" \ + && _notrun "xfs_admin does not support upgrading $1" +} and then: ... > +# real QA test starts here > +_supported_fs xfs > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > +_require_xfs_mkfs_inobtcount > +_require_xfs_scratch_inobtcount +_require_xfs_admin_upgrade "inobtcount" to be sure that the upgrade command is also supported by xfs_admin. By the time we get to release, both mkfs & xfs_admin should both support it, but I'm hedging my bets on the upgrade path just a little,and it seems best to explicitly test both requirements. that helper can be re-used on the bigtime upgrade test as well and anything else that requires an xfs_admin upgrade path ... Thanks, -Eric ^ permalink raw reply [flat|nested] 20+ messages in thread
* Re: [PATCH 2/2] xfs: test inobtcount upgrade 2020-11-18 16:44 ` Eric Sandeen @ 2020-11-20 1:53 ` Darrick J. Wong 0 siblings, 0 replies; 20+ messages in thread From: Darrick J. Wong @ 2020-11-20 1:53 UTC (permalink / raw) To: Eric Sandeen; +Cc: guaneryu, linux-xfs, fstests On Wed, Nov 18, 2020 at 10:44:40AM -0600, Eric Sandeen wrote: > On 10/27/20 2:03 PM, Darrick J. Wong wrote: > > From: Darrick J. Wong <darrick.wong@oracle.com> > > > > Make sure we can actually upgrade filesystems to support inobtcounts. > > > > Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com> > > --- > > common/xfs | 16 ++++++++++++ > > tests/xfs/910 | 72 +++++++++++++++++++++++++++++++++++++++++++++++++++++ > > tests/xfs/910.out | 3 ++ > > tests/xfs/group | 1 + > > 4 files changed, 92 insertions(+) > > create mode 100755 tests/xfs/910 > > create mode 100644 tests/xfs/910.out > > > > > > diff --git a/common/xfs b/common/xfs > > index 3f5c14ba..e548a0a1 100644 > > --- a/common/xfs > > +++ b/common/xfs > > @@ -978,3 +978,19 @@ _require_xfs_copy() > > [ "$USE_EXTERNAL" = yes ] && \ > > _notrun "Cannot xfs_copy with external devices" > > } > > + > > +_require_xfs_mkfs_inobtcount() > > +{ > > + _scratch_mkfs_xfs_supported -m inobtcount=1 >/dev/null 2>&1 \ > > + || _notrun "mkfs.xfs doesn't have inobtcount feature" > > +} > > I'd like to also add: > > +_require_xfs_admin_upgrade() > +{ > + local feature="$1" > + > + _require_scratch > + # Catch missing "-O" or missing feature handling > + _scratch_xfs_admin -O $feature 2>&1 | grep "illegal option\|Cannot change" \ > + && _notrun "xfs_admin does not support upgrading $1" > +} Well it occurred to me that the xfs_db version command has a help screen that lists all of the things that it knows how to upgrade. So, it should be easy enough to detect whether or not it makes sense to test an upgrade path. Thanks for the suggestion, sorry I was rambling about this being harder a few days ago... :/ --D > > and then: > > ... > > > +# real QA test starts here > > +_supported_fs xfs > > +_require_command "$XFS_ADMIN_PROG" "xfs_admin" > > +_require_xfs_mkfs_inobtcount > > +_require_xfs_scratch_inobtcount > > +_require_xfs_admin_upgrade "inobtcount" > > to be sure that the upgrade command is also supported by xfs_admin. By the time > we get to release, both mkfs & xfs_admin should both support it, but I'm hedging > my bets on the upgrade path just a little,and it seems best to explicitly test > both requirements. > > that helper can be re-used on the bigtime upgrade test as well and anything else > that requires an xfs_admin upgrade path ... > > Thanks, > -Eric ^ permalink raw reply [flat|nested] 20+ messages in thread
end of thread, other threads:[~2021-04-22 1:01 UTC | newest] Thread overview: 20+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2021-04-21 0:22 [PATCHSET v2 0/2] fstests: add inode btree blocks counters to the AGI header Darrick J. Wong 2021-04-21 0:22 ` [PATCH 1/2] xfs: functional testing of V5-relevant options Darrick J. Wong 2021-04-21 18:00 ` Brian Foster 2021-04-21 20:39 ` Darrick J. Wong 2021-04-22 0:29 ` Allison Henderson 2021-04-22 0:51 ` [PATCH v2.1 " Darrick J. Wong 2021-04-21 0:22 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 2021-04-21 5:52 ` Amir Goldstein 2021-04-22 0:50 ` Darrick J. Wong 2021-04-21 18:00 ` Brian Foster 2021-04-22 0:29 ` Allison Henderson 2021-04-22 0:53 ` [PATCH v2.1 " Darrick J. Wong -- strict thread matches above, loose matches on Subject: below -- 2021-03-31 1:08 [PATCHSET v2 0/2] fstests: add inode btree blocks counters to the AGI header Darrick J. Wong 2021-03-31 1:08 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 2021-03-31 17:36 ` Brian Foster 2021-04-03 0:29 ` Darrick J. Wong 2020-10-27 19:03 [PATCH RFCv4 0/2] xfstests: add inode btree blocks counters to the AGI header Darrick J. Wong 2020-10-27 19:03 ` [PATCH 2/2] xfs: test inobtcount upgrade Darrick J. Wong 2020-10-29 17:40 ` Brian Foster 2020-10-29 20:31 ` Darrick J. Wong 2020-11-18 16:44 ` Eric Sandeen 2020-11-20 1:53 ` Darrick J. Wong
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).