btrfs: Add new test for qgroup assign functionality
PATCH v2 1/1 btrfs: Add new test for qgroup assign functionality
This new test will test btrfs's qgroup assign functionality. The test has 3 cases.
- assign, no shared extents
- assign, shared extents
- snapshot -i, shared extents
Each cases create subvolumes and assign qgroup in their own way and check with the command "btrfs check".
---
v2: create new test and use the cases
--- tests/btrfs/221 | 121 ++++++++++++++++++++++++++++++++++++++++++++ tests/btrfs/221.out | 2 + tests/btrfs/group | 1 + 3 files changed, 124 insertions(+) create mode 100755 tests/btrfs/221 create mode 100644 tests/btrfs/221.out
tests/btrfs/221 new file +121 -0
@@ -0,0 +1,121 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2020 YOUR NAME HERE. All Rights Reserved. +# +# FS QA Test 221 +# +# Test the assign functionality of qgroups +# +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 +. ./common/reflink + +# remove previous $seqres.full before test +rm -f $seqres.full + +# real QA test starts here + +# Modify as appropriate. +_supported_fs generic +_supported_os Linux + +_require_test +_require_scratch +_require_btrfs_qgroup_report +_require_cp_reflink + +# Test assign qgroup for submodule with shared extents by reflink +assign_shared_test() +{ + echo "=== qgroup assign shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + + _ddt of="$SCRATCH_MNT"/a/file1 bs=1M count=1 >> $seqres.full 2>&1 + cp --reflink=always "$SCRATCH_MNT"/a/file1 "$SCRATCH_MNT"/b/file1 >> $seqres.full 2>&1 + + _scratch_unmount + _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test assign qgroup for submodule without shared extents +assign_no_shared_test() +{ + echo "=== qgroup assign no shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test snapshot with assigning qgroup for submodule +snapshot_test() +{ + echo "=== qgroup snapshot test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + + _run_btrfs_util_prog subvolume snapshot -i 0/$subvolid $SCRATCH_MNT/a $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_no_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +snapshot_test + +# success, all done +echo "Silence is golden" +status=0 +exit
tests/btrfs/221.out new file +2 -0
@@ -0,0 +1,2 @@ +QA output created by 221 +Silence is golden
tests/btrfs/group +1 -0
@@ -222,3 +222,4 @@ 218 auto quick volume 219 auto quick volume 220 auto quick +221 auto quick qgroup
Signature · 1 line
2.25.1
Quoted 30 lines · Sidong Yang
This new test will test btrfs's qgroup assign functionality. The test has 3 cases.
- assign, no shared extents
- assign, shared extents
- snapshot -i, shared extentsEach cases create subvolumes and assign qgroup in their own way and check with the command "btrfs check".
Cc Qu Wenruo <[email protected]>Cc Eryu Guan <[email protected]>Signed-off-by Sidong Yang <[email protected]>---
v2: create new test and use the cases--- tests/btrfs/221 | 121 ++++++++++++++++++++++++++++++++++++++++++++ tests/btrfs/221.out | 2 + tests/btrfs/group | 1 + 3 files changed, 124 insertions(+) create mode 100755 tests/btrfs/221 create mode 100644 tests/btrfs/221.outtests/btrfs/221 new file +0
-0
tests/btrfs/221 line 3
Quoted 3 lines
tests/btrfs/221 new file +3
-0@@ -0,0 +1,121 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2020 YOUR NAME HERE. All Rights Reserved.
So, "YOUR NAME HERE" is your name? :)
Quoted 22 lines
tests/btrfs/221 new file +22
-0+# +# FS QA Test 221 +# +# Test the assign functionality of qgroups +# +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
tests/btrfs/221 line 35
Quoted 10 lines
tests/btrfs/221 new file +10
-0+. ./common/filter +. ./common/reflink + +# remove previous $seqres.full before test +rm -f $seqres.full + +# real QA test starts here + +# Modify as appropriate. +_supported_fs generic
It's a btrfs specific test.
tests/btrfs/221 line 38
Quoted 3 lines
tests/btrfs/221 new file +3
-0+_supported_os Linux + +_require_test
You don't need test_dev at all.
tests/btrfs/221 line 47
Quoted 9 lines
tests/btrfs/221 new file +9
-0+_require_scratch +_require_btrfs_qgroup_report +_require_cp_reflink + +# Test assign qgroup for submodule with shared extents by reflink +assign_shared_test() +{ + echo "=== qgroup assign shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT
I'm not sure if _run_btrfs_util_prog is still recommended. IIRC nowadays we recommend to call $BTRFS_UTIL_PROG directly.
Test case btrfs/193 would be the example.
tests/btrfs/221 line 52
Quoted 5 lines
tests/btrfs/221 new file +5
-0+ _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT
"btrfs qgroup assign" can take path directly. This would save your some lines. E.g:
# btrfs qgroup create 1/100 /mnt/btrfs/ # btrfs qgroup assign /mnt/btrfs/ 1/100 /mnt/btrfs/ # btrfs qgroup show -pc /mnt/btrfs/ qgroupid rfer excl parent child -------- ---- ---- ------ ----- 0/5 16.00KiB 16.00KiB 1/100 --- 1/100 16.00KiB 16.00KiB --- 0/5
tests/btrfs/221 line 56
Quoted 4 lines
tests/btrfs/221 new file +4
-0+ + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT
Shouldn't this assign happens when we have shared extents between the two subvolumes?
Now you're just testing the basic qgroup functionality of accounting, not assign.
For real assign tests, what we want is either:
- After assign, qgroup accounting is still correct We don't need even to rescan.
And "btrfs check" will verify the numbers are correct.
- After assign, qgroup accounting is inconsistent
At least we should either have qgroup inconsistent bit set, or qgroup rescan kicked in automatically.
And "btrfs check" will skip the qgroup numbers.
But in your case, we're assigning two empty subovlumes into the same qgroup, then do some operations.
This only covers the "assign, no shared extents" case.
tests/btrfs/221 line 62
Quoted 6 lines
tests/btrfs/221 new file +6
-0+ _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + + _ddt of="$SCRATCH_MNT"/a/file1 bs=1M count=1 >> $seqres.full 2>&1 + cp --reflink=always "$SCRATCH_MNT"/a/file1 "$SCRATCH_MNT"/b/file1 >> $seqres.full 2>&1 + + _scratch_unmount
Since you're unmounting here, why not keep the _scratch_mkfs and _scratch_unmount pair in the same function?
Quoted 9 lines
tests/btrfs/221 new file +9
-0+ _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test assign qgroup for submodule without shared extents +assign_no_shared_test() +{ + echo "=== qgroup assign no shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNT
tests/btrfs/221 line 81
Quoted 10 lines
tests/btrfs/221 new file +10
-0+ + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT
No, we don't want rescan.
And the timing is still wrong.
Quoted 6 lines
tests/btrfs/221 new file +6
-0+ _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test snapshot with assigning qgroup for submodule
tests/btrfs/221 line 97
Quoted 10 lines
tests/btrfs/221 new file +10
-0+snapshot_test() +{ + echo "=== qgroup snapshot test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + + _run_btrfs_util_prog subvolume snapshot -i 0/$subvolid $SCRATCH_MNT/a $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b)
Even we're snapshotting on the source subvolume, since it's empty, we will only copy the root item, resulting two empty subvolumes without sharing anything.
You need to at least fill the source subvolumes with some data. It's better to bump the tree level with some inline extents.
Thanks,
Qu
Quoted 42 lines
tests/btrfs/221 new file +24
-0+ + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_no_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +snapshot_test + +# success, all done +echo "Silence is golden" +status=0 +exittests/btrfs/221.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 221 +Silence is goldentests/btrfs/group +1
-0@@ -222,3 +222,4 @@ 218 auto quick volume 219 auto quick volume 220 auto quick +221 auto quick qgroup
Quoted 27 lines · Qu Wenruo
Quoted 24 lines · Sidong Yang
This new test will test btrfs's qgroup assign functionality. The test has 3 cases.
- assign, no shared extents
- assign, shared extents
- snapshot -i, shared extentsEach cases create subvolumes and assign qgroup in their own way and check with the command "btrfs check".
Cc Qu Wenruo <[email protected]>Cc Eryu Guan <[email protected]>Signed-off-by Sidong Yang <[email protected]>---
v2: create new test and use the cases--- tests/btrfs/221 | 121 ++++++++++++++++++++++++++++++++++++++++++++ tests/btrfs/221.out | 2 + tests/btrfs/group | 1 + 3 files changed, 124 insertions(+) create mode 100755 tests/btrfs/221 create mode 100644 tests/btrfs/221.out
Quoted 11 lines
tests/btrfs/221 line 3
Quoted 9 lines
tests/btrfs/221 new file +3
-0@@ -0,0 +1,121 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2020 YOUR NAME HERE. All Rights Reserved.So, "YOUR NAME HERE" is your name? :)
Oops! I missed it.
Quoted 27 lines
Quoted 27 lines
tests/btrfs/221 new file +26
-0+# +# FS QA Test 221 +# +# Test the assign functionality of qgroups +# +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 +. ./common/reflink + +# remove previous $seqres.full before test
Quoted 8 lines
tests/btrfs/221 line 35
Quoted 6 lines
tests/btrfs/221 new file +6
-0+rm -f $seqres.full + +# real QA test starts here + +# Modify as appropriate. +_supported_fs genericIt's a btrfs specific test.
Thanks!
Quoted 6 lines
tests/btrfs/221 line 38
Quoted 4 lines
tests/btrfs/221 new file +3
-0+_supported_os Linux + +_require_testYou don't need
test_devat all.
Yeah, I just realized that I used only scratch dev noy test dev. Thanks!
Quoted 14 lines
tests/btrfs/221 line 47
Quoted 10 lines
tests/btrfs/221 new file +9
-0+_require_scratch +_require_btrfs_qgroup_report +_require_cp_reflink + +# Test assign qgroup for submodule with shared extents by reflink +assign_shared_test() +{ + echo "=== qgroup assign shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNTI'm not sure if
_run_btrfs_util_progis still recommended. IIRC nowadays we recommend to call $BTRFS_UTIL_PROG directly.
Quoted 1 line
Test case btrfs/193 would be the example.
It's good to replace it! Thanks.
Quoted 10 lines
tests/btrfs/221 line 52
Quoted 6 lines
tests/btrfs/221 new file +5
-0+ _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT"btrfs qgroup assign" can take path directly. This would save your some lines. E.g:
Quoted 7 lines
# btrfs qgroup create 1/100 /mnt/btrfs/ # btrfs qgroup assign /mnt/btrfs/ 1/100 /mnt/btrfs/ # btrfs qgroup show -pc /mnt/btrfs/ qgroupid rfer excl parent child -------- ---- ---- ------ ----- 0/5 16.00KiB 16.00KiB 1/100 --- 1/100 16.00KiB 16.00KiB --- 0/5
Thanks for good tip!
Quoted 17 lines
tests/btrfs/221 line 56
Quoted 5 lines
tests/btrfs/221 new file +4
-0+ + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNTShouldn't this assign happens when we have shared extents between the two subvolumes?
Now you're just testing the basic qgroup functionality of accounting, not assign.
For real assign tests, what we want is either:
- After assign, qgroup accounting is still correct We don't need even to rescan.
And "btrfs check" will verify the numbers are correct.
Quoted 8 lines
- After assign, qgroup accounting is inconsistent
At least we should either have qgroup inconsistent bit set, or qgroup rescan kicked in automatically.
And "btrfs check" will skip the qgroup numbers.But in your case, we're assigning two empty subovlumes into the same qgroup, then do some operations.
This only covers the "assign, no shared extents" case.
You mean that there should be some data with reflink before assigning? If so, the code below should be executed before assigning qgroups. Should test process be like this?
make submodules -> make data -> copy with reflink -> assign qgroup
Quoted 33 lines
tests/btrfs/221 line 62
Quoted 7 lines
tests/btrfs/221 new file +6
-0+ _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + + _ddt of="$SCRATCH_MNT"/a/file1 bs=1M count=1 >> $seqres.full 2>&1 + cp --reflink=always "$SCRATCH_MNT"/a/file1 "$SCRATCH_MNT"/b/file1 >> $seqres.full 2>&1 + + _scratch_unmountSince you're unmounting here, why not keep the
_scratch_mkfsand_scratch_unmountpair in the same function?Quoted 9 lines
tests/btrfs/221 new file +9
-0+ _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test assign qgroup for submodule without shared extents +assign_no_shared_test() +{ + echo "=== qgroup assign no shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNTtests/btrfs/221 line 81
Quoted 10 lines
tests/btrfs/221 new file +10
-0+ + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNTNo, we don't want rescan.
Quoted 1 line
And the timing is still wrong.
Yeah, I'll delete it.
Quoted 22 lines
Quoted 7 lines
tests/btrfs/221 new file +6
-0+ _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test snapshot with assigning qgroup for submoduletests/btrfs/221 line 97
Quoted 10 lines
tests/btrfs/221 new file +10
-0+snapshot_test() +{ + echo "=== qgroup snapshot test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + + _run_btrfs_util_prog subvolume snapshot -i 0/$subvolid $SCRATCH_MNT/a $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b)Even we're snapshotting on the source subvolume, since it's empty, we will only copy the root item, resulting two empty subvolumes without sharing anything.
Quoted 2 lines
You need to at least fill the source subvolumes with some data. It's better to bump the tree level with some inline extents.
I should write some data before snapshot. is it right? Thanks for all your comments.
Thanks,
Sidong
Quoted 47 lines
Thanks,
QuQuoted 42 lines
tests/btrfs/221 new file +24
-0+ + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_no_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +snapshot_test + +# success, all done +echo "Silence is golden" +status=0 +exittests/btrfs/221.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 221 +Silence is goldentests/btrfs/group +1
-0@@ -222,3 +222,4 @@ 218 auto quick volume 219 auto quick volume 220 auto quick +221 auto quick qgroup
Quoted 143 lines · Sidong Yang
Quoted 27 lines · Qu Wenruo
Quoted 24 lines · Sidong Yang
This new test will test btrfs's qgroup assign functionality. The test has 3 cases.
- assign, no shared extents
- assign, shared extents
- snapshot -i, shared extentsEach cases create subvolumes and assign qgroup in their own way and check with the command "btrfs check".
Cc Qu Wenruo <[email protected]>Cc Eryu Guan <[email protected]>Signed-off-by Sidong Yang <[email protected]>---
v2: create new test and use the cases--- tests/btrfs/221 | 121 ++++++++++++++++++++++++++++++++++++++++++++ tests/btrfs/221.out | 2 + tests/btrfs/group | 1 + 3 files changed, 124 insertions(+) create mode 100755 tests/btrfs/221 create mode 100644 tests/btrfs/221.outQuoted 11 lines
tests/btrfs/221 line 3
Quoted 9 lines
tests/btrfs/221 new file +3
-0@@ -0,0 +1,121 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2020 YOUR NAME HERE. All Rights Reserved.So, "YOUR NAME HERE" is your name? :)
Oops! I missed it.
Quoted 27 lines
Quoted 27 lines
tests/btrfs/221 new file +26
-0+# +# FS QA Test 221 +# +# Test the assign functionality of qgroups +# +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 +. ./common/reflink + +# remove previous $seqres.full before testQuoted 8 lines
tests/btrfs/221 line 35
Quoted 6 lines
tests/btrfs/221 new file +6
-0+rm -f $seqres.full + +# real QA test starts here + +# Modify as appropriate. +_supported_fs genericIt's a btrfs specific test.
Thanks!
Quoted 6 lines
tests/btrfs/221 line 38
Quoted 4 lines
tests/btrfs/221 new file +3
-0+_supported_os Linux + +_require_testYou don't need
test_devat all.Yeah, I just realized that I used only scratch dev noy test dev. Thanks!
Quoted 14 lines
tests/btrfs/221 line 47
Quoted 10 lines
tests/btrfs/221 new file +9
-0+_require_scratch +_require_btrfs_qgroup_report +_require_cp_reflink + +# Test assign qgroup for submodule with shared extents by reflink +assign_shared_test() +{ + echo "=== qgroup assign shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNTI'm not sure if
_run_btrfs_util_progis still recommended. IIRC nowadays we recommend to call $BTRFS_UTIL_PROG directly.Quoted 1 line
Test case btrfs/193 would be the example.
It's good to replace it! Thanks.
Quoted 10 lines
tests/btrfs/221 line 52
Quoted 6 lines
tests/btrfs/221 new file +5
-0+ _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT"btrfs qgroup assign" can take path directly. This would save your some lines. E.g:
Quoted 7 lines
# btrfs qgroup create 1/100 /mnt/btrfs/ # btrfs qgroup assign /mnt/btrfs/ 1/100 /mnt/btrfs/ # btrfs qgroup show -pc /mnt/btrfs/ qgroupid rfer excl parent child -------- ---- ---- ------ ----- 0/5 16.00KiB 16.00KiB 1/100 --- 1/100 16.00KiB 16.00KiB --- 0/5Thanks for good tip!
Quoted 22 lines
tests/btrfs/221 line 56
Quoted 5 lines
tests/btrfs/221 new file +4
-0+ + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNTShouldn't this assign happens when we have shared extents between the two subvolumes?
Now you're just testing the basic qgroup functionality of accounting, not assign.
For real assign tests, what we want is either:
- After assign, qgroup accounting is still correct We don't need even to rescan.
And "btrfs check" will verify the numbers are correct.- After assign, qgroup accounting is inconsistent
At least we should either have qgroup inconsistent bit set, or qgroup rescan kicked in automatically.
And "btrfs check" will skip the qgroup numbers.
Quoted 8 lines
Quoted 3 lines
But in your case, we're assigning two empty subovlumes into the same qgroup, then do some operations.
This only covers the "assign, no shared extents" case.You mean that there should be some data with reflink before assigning? If so, the code below should be executed before assigning qgroups. Should test process be like this?
make submodules -> make data -> copy with reflink -> assign qgroup
Yep.
Furthermore, we should have qgroup enabled/rescanned before make subvolumes.
Quoted 58 lines
Quoted 34 lines
tests/btrfs/221 line 62
Quoted 8 lines
tests/btrfs/221 new file +6
-0+ _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + + _ddt of="$SCRATCH_MNT"/a/file1 bs=1M count=1 >> $seqres.full 2>&1 + cp --reflink=always "$SCRATCH_MNT"/a/file1 "$SCRATCH_MNT"/b/file1 >> $seqres.full 2>&1 + + _scratch_unmountSince you're unmounting here, why not keep the
_scratch_mkfsand_scratch_unmountpair in the same function?Quoted 9 lines
tests/btrfs/221 new file +9
-0+ _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test assign qgroup for submodule without shared extents +assign_no_shared_test() +{ + echo "=== qgroup assign no shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNTtests/btrfs/221 line 81
Quoted 10 lines
tests/btrfs/221 new file +10
-0+ + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNTNo, we don't want rescan.
Quoted 1 line
And the timing is still wrong.
Yeah, I'll delete it.
Quoted 22 lines
Quoted 7 lines
tests/btrfs/221 new file +6
-0+ _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test snapshot with assigning qgroup for submoduletests/btrfs/221 line 97
Quoted 10 lines
tests/btrfs/221 new file +10
-0+snapshot_test() +{ + echo "=== qgroup snapshot test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + + _run_btrfs_util_prog subvolume snapshot -i 0/$subvolid $SCRATCH_MNT/a $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b)Even we're snapshotting on the source subvolume, since it's empty, we will only copy the root item, resulting two empty subvolumes without sharing anything.
Quoted 4 lines
Quoted 2 lines
You need to at least fill the source subvolumes with some data. It's better to bump the tree level with some inline extents.
I should write some data before snapshot. is it right?
Right. That's the bare minimal.
Thanks,
qu
Quoted 55 lines
Thanks for all your comments.
Thanks,
SidongQuoted 47 lines
Thanks,
QuQuoted 42 lines
tests/btrfs/221 new file +24
-0+ + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_no_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +snapshot_test + +# success, all done +echo "Silence is golden" +status=0 +exittests/btrfs/221.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 221 +Silence is goldentests/btrfs/group +1
-0@@ -222,3 +222,4 @@ 218 auto quick volume 219 auto quick volume 220 auto quick +221 auto quick qgroup
Quoted 219 lines · Qu Wenruo
Quoted 143 lines · Sidong Yang
Quoted 27 lines · Qu Wenruo
Quoted 24 lines · Sidong Yang
This new test will test btrfs's qgroup assign functionality. The test has 3 cases.
- assign, no shared extents
- assign, shared extents
- snapshot -i, shared extentsEach cases create subvolumes and assign qgroup in their own way and check with the command "btrfs check".
Cc Qu Wenruo <[email protected]>Cc Eryu Guan <[email protected]>Signed-off-by Sidong Yang <[email protected]>---
v2: create new test and use the cases--- tests/btrfs/221 | 121 ++++++++++++++++++++++++++++++++++++++++++++ tests/btrfs/221.out | 2 + tests/btrfs/group | 1 + 3 files changed, 124 insertions(+) create mode 100755 tests/btrfs/221 create mode 100644 tests/btrfs/221.outQuoted 11 lines
tests/btrfs/221 line 3
Quoted 9 lines
tests/btrfs/221 new file +3
-0@@ -0,0 +1,121 @@ +#! /bin/bash +# SPDX-License-Identifier: GPL-2.0 +# Copyright (c) 2020 YOUR NAME HERE. All Rights Reserved.So, "YOUR NAME HERE" is your name? :)
Oops! I missed it.
Quoted 27 lines
Quoted 27 lines
tests/btrfs/221 new file +26
-0+# +# FS QA Test 221 +# +# Test the assign functionality of qgroups +# +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 +. ./common/reflink + +# remove previous $seqres.full before testQuoted 8 lines
tests/btrfs/221 line 35
Quoted 6 lines
tests/btrfs/221 new file +6
-0+rm -f $seqres.full + +# real QA test starts here + +# Modify as appropriate. +_supported_fs genericIt's a btrfs specific test.
Thanks!
Quoted 6 lines
tests/btrfs/221 line 38
Quoted 4 lines
tests/btrfs/221 new file +3
-0+_supported_os Linux + +_require_testYou don't need
test_devat all.Yeah, I just realized that I used only scratch dev noy test dev. Thanks!
Quoted 14 lines
tests/btrfs/221 line 47
Quoted 10 lines
tests/btrfs/221 new file +9
-0+_require_scratch +_require_btrfs_qgroup_report +_require_cp_reflink + +# Test assign qgroup for submodule with shared extents by reflink +assign_shared_test() +{ + echo "=== qgroup assign shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNTI'm not sure if
_run_btrfs_util_progis still recommended. IIRC nowadays we recommend to call $BTRFS_UTIL_PROG directly.Quoted 1 line
Test case btrfs/193 would be the example.
It's good to replace it! Thanks.
Quoted 10 lines
tests/btrfs/221 line 52
Quoted 6 lines
tests/btrfs/221 new file +5
-0+ _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT"btrfs qgroup assign" can take path directly. This would save your some lines. E.g:
Quoted 7 lines
# btrfs qgroup create 1/100 /mnt/btrfs/ # btrfs qgroup assign /mnt/btrfs/ 1/100 /mnt/btrfs/ # btrfs qgroup show -pc /mnt/btrfs/ qgroupid rfer excl parent child -------- ---- ---- ------ ----- 0/5 16.00KiB 16.00KiB 1/100 --- 1/100 16.00KiB 16.00KiB --- 0/5Thanks for good tip!
Quoted 22 lines
tests/btrfs/221 line 56
Quoted 5 lines
tests/btrfs/221 new file +4
-0+ + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNTShouldn't this assign happens when we have shared extents between the two subvolumes?
Now you're just testing the basic qgroup functionality of accounting, not assign.
For real assign tests, what we want is either:
- After assign, qgroup accounting is still correct We don't need even to rescan.
And "btrfs check" will verify the numbers are correct.- After assign, qgroup accounting is inconsistent
At least we should either have qgroup inconsistent bit set, or qgroup rescan kicked in automatically.
And "btrfs check" will skip the qgroup numbers.Quoted 8 lines
Quoted 3 lines
But in your case, we're assigning two empty subovlumes into the same qgroup, then do some operations.
This only covers the "assign, no shared extents" case.You mean that there should be some data with reflink before assigning? If so, the code below should be executed before assigning qgroups. Should test process be like this?
make submodules -> make data -> copy with reflink -> assign qgroupYep.
Furthermore, we should have qgroup enabled/rescanned before make subvolumes.
Quoted 61 lines
Quoted 34 lines
tests/btrfs/221 line 62
Quoted 8 lines
tests/btrfs/221 new file +6
-0+ _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + + _ddt of="$SCRATCH_MNT"/a/file1 bs=1M count=1 >> $seqres.full 2>&1 + cp --reflink=always "$SCRATCH_MNT"/a/file1 "$SCRATCH_MNT"/b/file1 >> $seqres.full 2>&1 + + _scratch_unmountSince you're unmounting here, why not keep the
_scratch_mkfsand_scratch_unmountpair in the same function?Quoted 9 lines
tests/btrfs/221 new file +9
-0+ _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test assign qgroup for submodule without shared extents +assign_no_shared_test() +{ + echo "=== qgroup assign no shared test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + _run_btrfs_util_prog qgroup create 1/100 $SCRATCH_MNTtests/btrfs/221 line 81
Quoted 10 lines
tests/btrfs/221 new file +10
-0+ + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b) + _run_btrfs_util_prog qgroup assign 0/$subvolid 1/100 $SCRATCH_MNT + + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNTNo, we don't want rescan.
Quoted 1 line
And the timing is still wrong.
Yeah, I'll delete it.
Quoted 24 lines
Quoted 7 lines
tests/btrfs/221 new file +6
-0+ _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + +# Test snapshot with assigning qgroup for submoduletests/btrfs/221 line 97
Quoted 10 lines
tests/btrfs/221 new file +10
-0+snapshot_test() +{ + echo "=== qgroup snapshot test ===" >> $seqres.full + _run_btrfs_util_prog quota enable $SCRATCH_MNT + + _run_btrfs_util_prog subvolume create $SCRATCH_MNT/a + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT a) + + _run_btrfs_util_prog subvolume snapshot -i 0/$subvolid $SCRATCH_MNT/a $SCRATCH_MNT/b + subvolid=$(_btrfs_get_subvolid $SCRATCH_MNT b)Even we're snapshotting on the source subvolume, since it's empty, we will only copy the root item, resulting two empty subvolumes without sharing anything.
You need to at least fill the source subvolumes with some data. It's better to bump the tree level with some inline extents.
Quoted 6 lines
Quoted 1 line
I should write some data before snapshot. is it right?
Right. That's the bare minimal.
Thanks,
qu
Thanks so much! I've written a new patch just before. Could you review it please?
Thanks,
Sidong
Quoted 57 lines
Quoted 56 lines
Thanks for all your comments.
Thanks,
SidongQuoted 47 lines
Thanks,
QuQuoted 42 lines
tests/btrfs/221 new file +24
-0+ + _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT + _scratch_unmount + + _run_btrfs_util_prog check $SCRATCH_DEV +} + + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_no_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +assign_shared_test + +_scratch_mkfs > /dev/null 2>&1 +_scratch_mount +snapshot_test + +# success, all done +echo "Silence is golden" +status=0 +exittests/btrfs/221.out new file +2
-0@@ -0,0 +1,2 @@ +QA output created by 221 +Silence is goldentests/btrfs/group +1
-0@@ -222,3 +222,4 @@ 218 auto quick volume 219 auto quick volume 220 auto quick +221 auto quick qgroup