On 3/25/25 17:15, Dave Chinner wrote:
> On Wed, Mar 12, 2025 at 03:23:09PM +0800, Chao Yu wrote:
>> This is a regression test to check whether fsck can handle corrupted
>> nlinks correctly, it uses inject.f2fs to inject nlinks w/ wrong value,
>> and expects fsck.f2fs can detect such corruption and do the repair.
>>
>> Cc: Jaegeuk Kim <jaegeuk@xxxxxxxxxx>
>> Signed-off-by: Chao Yu <chao@xxxxxxxxxx>
>> ---
>> v4:
>> - remove unnecessary "_require_command fsck.f2fs"
>> - clean up _fixed_by_git_commit line
>> - fix to clean all tmp files in _cleanup
>> tests/f2fs/009 | 147 +++++++++++++++++++++++++++++++++++++++++++++
>> tests/f2fs/009.out | 2 +
>> 2 files changed, 149 insertions(+)
>> create mode 100755 tests/f2fs/009
>> create mode 100644 tests/f2fs/009.out
>>
>> diff --git a/tests/f2fs/009 b/tests/f2fs/009
>> new file mode 100755
>> index 00000000..9120d8a5
>> --- /dev/null
>> +++ b/tests/f2fs/009
>> @@ -0,0 +1,147 @@
>> +#! /bin/bash
>> +# SPDX-License-Identifier: GPL-2.0
>> +# Copyright (c) 2025 Chao Yu. All Rights Reserved.
>> +#
>> +# FS QA Test No. f2fs/009
>> +#
>> +# This is a regression test to check whether fsck can handle corrupted
>> +# nlinks correctly, it uses inject.f2fs to inject nlinks w/ wrong value,
>> +# and expects fsck.f2fs can detect such corruption and do the repair.
>> +#
>> +. ./common/preamble
>> +_begin_fstest auto quick
>> +
>> +if [ ! -x "$(type -P socket)" ]; then
>> + _notrun "Couldn't find socket"
>> +fi
>> +
>> +_require_scratch
>> +_require_command "$F2FS_INJECT_PROG" inject.f2fs
>> +
>> +_fixed_by_git_commit f2fs-tools 958cd6e \
>> + "fsck.f2fs: support to repair corrupted i_links"
>> +
>> +filename=$SCRATCH_MNT/foo
>> +hardlink=$SCRATCH_MNT/bar
>> +
>> +_cleanup()
>> +{
>> + if [ -n "$pid" ]; then
>> + kill $pid &> /dev/null
>> + wait
>> + fi
>> + cd /
>> + rm -r -f $tmp.*
>> +}
>> +
>> +for ((i=0;i<14;i++)) do
>> + echo "round: " $i >> $seqres.full
>> +
>> + _scratch_mkfs >> $seqres.full
>> + _scratch_mount
>> +
>> + if [ $i == 0 ]; then
>> + touch $filename
>> + ino=`stat -c '%i' $filename`
>> + nlink=0
>> + elif [ $i == 1 ]; then
> .....
>
> Ok, it works, but it's a bit nasty. You want to run the same
> injection and checking on all types of inodes, so factor that out
> like so:
>
> inject_and_check()
> {
> local nlink=$1
> local ino=`stat -c '%i' $filename`
>
> echo "ino:"$ino >> $seqres.full
> echo "nlink:"$nlink >> $seqres.full
>
> _scratch_unmount
>
> $F2FS_INJECT_PROG --node --mb i_links --nid $ino --val $nlink $SCRATCH_DEV \
> >> $seqres.full || _fail "fail to inject"
>
> _check_scratch_fs >> $seqres.full 2>&1 && _fail "can't find corruption"
> _repair_scratch_fs >> $seqres.full
> _check_scratch_fs >> $seqres.full 2>&1 || _fail "fsck can't fix corruption"
>
> _scratch_mount
> _scratch_unmount
>
> _scratch_mkfs >> $seqres.full
> _scratch_mount
> }
>
> And now most of the test cases become this:
>
> _scratch_mkfs >> $seqres.full
> _scratch_mount
>
> touch $filename
> inject_and_check 0
>
> mkdir $filename
> inject_and_check 1
>
> mknod $filename c 9 0
> inject_and_check 0
>
> mknod $filename b 8 0
> inject_and_check 0
>
> .....
>
> The whacky cases with socket and orphan inodes can be handled
> by factoring inject_and_check so there's a second function that
> takes both ino and nlink as parameters.
>
> This gets rid of all the unnecessary if/elif logic and looping,
> leaving only the test code that does the work behind.
Thanks a lot for the suggestion, it will be more clean w/ above
implementation, will update v5.
Thanks,
>
> -Dave.
