btrfs/049: remove the test
PATCH 1/2 btrfs/049: remove the test
inode_cache is deprecated and will never appear in mount options; remove it entirely
--- tests/btrfs/049 | 85 ------------------------------------------------- 1 file changed, 85 deletions(-) delete mode 100755 tests/btrfs/049
tests/btrfs/049 deleted +0 -85
@@ -1,85 +0,0 @@ -#! /bin/bash -# SPDX-License-Identifier: GPL-2.0 -# Copyright (c) 2014 Fujitsu. All Rights Reserved. -# -# FS QA Test No. btrfs/049 -# -# Regression test for btrfs inode caching vs tree log which was -# addressed by the following kernel patch. -# -# Btrfs: fix inode caching vs tree log -# -. ./common/preamble -_begin_fstest auto quick - -# Override the default cleanup function. -_cleanup() -{ - _cleanup_flakey - rm -rf $tmp -} - -# Import common functions. -. ./common/filter -. ./common/dmflakey - -# real QA test starts here -_supported_fs btrfs -_require_scratch -_require_dm_target flakey -# Zoned btrfs does not support inode cache -_require_non_zoned_device "$SCRATCH_DEV" - -_scratch_mkfs >> $seqres.full 2>&1 - -SAVE_MOUNT_OPTIONS="$MOUNT_OPTIONS" -MOUNT_OPTIONS="$MOUNT_OPTIONS -o inode_cache,commit=100" - -# create a basic flakey device that will never error out -_init_flakey -_mount_flakey - -_get_inode_id() -{ - local inode_id - inode_id=`stat $1 | grep Inode: | $AWK_PROG '{print $4}'` - echo $inode_id -} - -$XFS_IO_PROG -f -c "pwrite 0 10M" -c "fsync" \ - $SCRATCH_MNT/data >& /dev/null - -inode_id=`_get_inode_id "$SCRATCH_MNT/data"` -rm -f $SCRATCH_MNT/data - -for i in `seq 1 5`; -do - mkdir $SCRATCH_MNT/dir_$i - new_inode_id=`_get_inode_id $SCRATCH_MNT/dir_$i` - if [ $new_inode_id -eq $inode_id ] - then - $XFS_IO_PROG -f -c "pwrite 0 1M" -c "fsync" \ - $SCRATCH_MNT/dir_$i/data1 >& /dev/null - _load_flakey_table 1 - _unmount_flakey - need_umount=1 - break - fi - sleep 1 -done - -# restore previous mount options -export MOUNT_OPTIONS="$SAVE_MOUNT_OPTIONS" - -# ok mount so that any recovery that needs to happen is done -if [ $new_inode_id -eq $inode_id ];then - _load_flakey_table $FLAKEY_ALLOW_WRITES - _mount_flakey - _unmount_flakey -fi - -# make sure we got a valid fs after replay -_check_scratch_fs $FLAKEY_DEV - -status=0 -exit
Signature · 1 line
2.20.1
PATCH 2/2 btrfs/091: remove noinode_cache option
inode cache feature has been removed
--- tests/btrfs/091 | 21 +-------------------- 1 file changed, 1 insertion(+), 20 deletions(-)
tests/btrfs/091 +1 -20
@@ -23,21 +23,12 @@ _require_cp_reflink # use largest node/leaf size (64K) to allow the test to be run on arch with # page size > 4k. NODESIZE=65536 -SUPPORT_NOINODE_CACHE="yes" run_check _scratch_mkfs "--nodesize $NODESIZE" -# inode cache will also take space in fs tree, disable them to get consistent -# result. # discard error output since we will check return value manually. # also disable all compression, or output will mismatch with golden output -_try_scratch_mount "-o noinode_cache,compress=no,compress-force=no" 2> /dev/null - -# Check for old kernel which doesn't support 'noinode_cache' mount option -if [ $? -ne 0 ]; then - support_noinode_cache="no" - _scratch_mount -fi +_try_scratch_mount "-o compress=no,compress-force=no" 2> /dev/null _run_btrfs_util_prog subvolume create $SCRATCH_MNT/subv1 _run_btrfs_util_prog subvolume create $SCRATCH_MNT/subv2
@@ -46,16 +37,6 @@ _run_btrfs_util_prog subvolume create $SCRATCH_MNT/subv3 _run_btrfs_util_prog quota enable $SCRATCH_MNT _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT -# if we don't support noinode_cache mount option, then we should double check -# whether inode cache is enabled before executing the real test payload. -if [ $SUPPORT_NOINODE_CACHE == "no" ]; then - EMPTY_SIZE=`$BTRFS_UTIL_PROG qgroup show $units $SCRATCH_MNT | \ - $SED_PROG -n '/[0-9]/p' | $AWK_PROG '{print $2}' | head -n1` - if [ $EMPTY_SIZE != $NODESIZE ]; then - _notrun "Kernel doesn't support to disable inode cache" - fi -fi - $XFS_IO_PROG -f -c "pwrite 0 256K" $SCRATCH_MNT/subv1/file1 | _filter_xfs_io cp --reflink $SCRATCH_MNT/subv1/file1 $SCRATCH_MNT/subv2/file1 cp --reflink $SCRATCH_MNT/subv1/file1 $SCRATCH_MNT/subv3/file1
Signature · 1 line
2.20.1
Re: [PATCH 2/2] btrfs/091: remove noinode_cache option
Quoted 3 lines · Ma Xinjian
inode cache feature has been removed
From above link, the inode cache feathre has been removed since v5.11 kernel, which is relatively a recent kernel, but old kernels may still need this noinode_cache mount option. So we'd better to keep this check.
We could add check for inode_cache option as well, and only check for noinode_cache when inode cache is supported.
Quoted 17 lines
Reported-by kernel test robot <[email protected]>Signed-off-by Ma Xinjian <[email protected]>--- tests/btrfs/091 | 21 +-------------------- 1 file changed, 1 insertion(+), 20 deletions(-)tests/btrfs/091 +0
-1@@ -23,21 +23,12 @@ _require_cp_reflink # use largest node/leaf size (64K) to allow the test to be run on arch with # page size > 4k. NODESIZE=65536 -SUPPORT_NOINODE_CACHE="yes" run_check _scratch_mkfs "--nodesize $NODESIZE"
tests/btrfs/091 _require_cp_reflink · line 38
Quoted 10 lines
tests/btrfs/091 +0
-7-# inode cache will also take space in fs tree, disable them to get consistent -# result. # discard error output since we will check return value manually. # also disable all compression, or output will mismatch with golden output -_try_scratch_mount "-o noinode_cache,compress=no,compress-force=no" 2> /dev/null - -# Check for old kernel which doesn't support 'noinode_cache' mount option -if [ $? -ne 0 ]; then - support_noinode_cache="no"
This seems like a type in the original code, it should be in upper case.
Thanks,
Eryu
Quoted 25 lines
tests/btrfs/091 +1
-12- _scratch_mount -fi +_try_scratch_mount "-o compress=no,compress-force=no" 2> /dev/null _run_btrfs_util_prog subvolume create $SCRATCH_MNT/subv1 _run_btrfs_util_prog subvolume create $SCRATCH_MNT/subv2@@ -46,16 +37,6 @@ _run_btrfs_util_prog subvolume create $SCRATCH_MNT/subv3 _run_btrfs_util_prog quota enable $SCRATCH_MNT _run_btrfs_util_prog quota rescan -w $SCRATCH_MNT -# if we don't support noinode_cache mount option, then we should double check -# whether inode cache is enabled before executing the real test payload. -if [ $SUPPORT_NOINODE_CACHE == "no" ]; then - EMPTY_SIZE=`$BTRFS_UTIL_PROG qgroup show $units $SCRATCH_MNT | \ - $SED_PROG -n '/[0-9]/p' | $AWK_PROG '{print $2}' | head -n1` - if [ $EMPTY_SIZE != $NODESIZE ]; then - _notrun "Kernel doesn't support to disable inode cache" - fi -fi - $XFS_IO_PROG -f -c "pwrite 0 256K" $SCRATCH_MNT/subv1/file1 | _filter_xfs_io cp --reflink $SCRATCH_MNT/subv1/file1 $SCRATCH_MNT/subv2/file1 cp --reflink $SCRATCH_MNT/subv1/file1 $SCRATCH_MNT/subv3/file1Signature · 1 line
2.20.1
Re: [PATCH 2/2] btrfs/091: remove noinode_cache option
Quoted 5 lines · Eryu Guan
Quoted 3 lines · Ma Xinjian
inode cache feature has been removed
Quoted 7 lines
Quoted 1 line
From above link, the inode cache feathre has been removed since v5.11
kernel, which is relatively a recent kernel, but old kernels may still need this
noinode_cachemount option. So we'd better to keep this check.We could add check for
inode_cacheoption as well, and only check fornoinode_cachewhen inode cache is supported.
I don't think we should keep the option support in the testsuite. There was deprecation period before actual removal from kernel and long before that we discouraged everybody from using it. There were known bugs and I don't think anybody cares for old kernels anyway.
[cc'ed btrfs list]
Quoted 2 lines · Ma Xinjian
inode_cacheis deprecated and will never appear in mount options; remove it entirely
Please also cc btrfs list in future patches for btrfs-specific tests.
And I'd like an ack from btrfs folks.
Thanks,
Eryu
Quoted 102 lines
Reported-by kernel test robot <[email protected]>Signed-off-by Ma Xinjian <[email protected]>--- tests/btrfs/049 | 85 ------------------------------------------------- 1 file changed, 85 deletions(-) delete mode 100755 tests/btrfs/049tests/btrfs/049 deleted +0
-85@@ -1,85 +0,0 @@ -#! /bin/bash -# SPDX-License-Identifier: GPL-2.0 -# Copyright (c) 2014 Fujitsu. All Rights Reserved. -# -# FS QA Test No. btrfs/049 -# -# Regression test for btrfs inode caching vs tree log which was -# addressed by the following kernel patch. -# -# Btrfs: fix inode caching vs tree log -# -. ./common/preamble -_begin_fstest auto quick - -# Override the default cleanup function. -_cleanup() -{ - _cleanup_flakey - rm -rf $tmp -} - -# Import common functions. -. ./common/filter -. ./common/dmflakey - -# real QA test starts here -_supported_fs btrfs -_require_scratch -_require_dm_target flakey -# Zoned btrfs does not support inode cache -_require_non_zoned_device "$SCRATCH_DEV" - -_scratch_mkfs >> $seqres.full 2>&1 - -SAVE_MOUNT_OPTIONS="$MOUNT_OPTIONS" -MOUNT_OPTIONS="$MOUNT_OPTIONS -o inode_cache,commit=100" - -# create a basic flakey device that will never error out -_init_flakey -_mount_flakey - -_get_inode_id() -{ - local inode_id - inode_id=`stat $1 | grep Inode: | $AWK_PROG '{print $4}'` - echo $inode_id -} - -$XFS_IO_PROG -f -c "pwrite 0 10M" -c "fsync" \ - $SCRATCH_MNT/data >& /dev/null - -inode_id=`_get_inode_id "$SCRATCH_MNT/data"` -rm -f $SCRATCH_MNT/data - -for i in `seq 1 5`; -do - mkdir $SCRATCH_MNT/dir_$i - new_inode_id=`_get_inode_id $SCRATCH_MNT/dir_$i` - if [ $new_inode_id -eq $inode_id ] - then - $XFS_IO_PROG -f -c "pwrite 0 1M" -c "fsync" \ - $SCRATCH_MNT/dir_$i/data1 >& /dev/null - _load_flakey_table 1 - _unmount_flakey - need_umount=1 - break - fi - sleep 1 -done - -# restore previous mount options -export MOUNT_OPTIONS="$SAVE_MOUNT_OPTIONS" - -# ok mount so that any recovery that needs to happen is done -if [ $new_inode_id -eq $inode_id ];then - _load_flakey_table $FLAKEY_ALLOW_WRITES - _mount_flakey - _unmount_flakey -fi - -# make sure we got a valid fs after replay -_check_scratch_fs $FLAKEY_DEV - -status=0 -exitSignature · 1 line
2.20.1
David Sterba
Acked-byQuoted 9 lines · Eryu Guan
[cc'ed btrfs list]
Quoted 2 lines · Ma Xinjian
inode_cacheis deprecated and will never appear in mount options; remove it entirelyPlease also cc btrfs list in future patches for btrfs-specific tests.
And I'd like an ack from btrfs folks.