on 2022/07/05 1:57, Zorro Lang wrote:
> On Mon, Jun 27, 2022 at 10:30:03PM +0800, Yang Xu wrote:
>> This case is a regression test for xfsprogs commit
>> f4afdcb0a ("xfs_db: clean up the salvage read callsites in set_cur()").
>>
>> I found this because xfs/270 on older xfsprogs can hit this bug.
>> The code is pasted from xfs/270 but remove mount rw/ro test.
>>
>> Signed-off-by: Yang Xu<xuyang2018.jy@xxxxxxxxxxx>
>> ---
>> tests/xfs/082 | 67 +++++++++++++++++++++++++++++++++++++++++++++++
>> tests/xfs/082.out | 2 ++
>> 2 files changed, 69 insertions(+)
>> create mode 100755 tests/xfs/082
>> create mode 100644 tests/xfs/082.out
>>
>> diff --git a/tests/xfs/082 b/tests/xfs/082
>> new file mode 100755
>> index 00000000..7a5ece8f
>> --- /dev/null
>> +++ b/tests/xfs/082
>> @@ -0,0 +1,67 @@
>> +#! /bin/bash
>> +# SPDX-License-Identifier: GPL-2.0
>> +# Copyright (c) 2022 FUJITSU LIMITED. All rights reserved.
>> +#
>> +# FS QA Test 082
>> +#
>> +# Today xfs_db set_cur whether clean up the LIBXFS_READBUF_SALVAGE
>> +# call sites so that we use the return value directly instead of
>> +# scraping it out later.
>> +#
>> +# It is a regression test for xfsprogs
>> +# f4afdcb0a ("xfs_db: clean up the salvage read callsites in set_cur()").
>> +#
>> +
>> +. ./common/preamble
>> +_begin_fstest auto quick mount
>> +
>> +# Import common functions.
>> +. ./common/filter
>> +
>> +# real QA test starts here
>> +_supported_fs xfs
>> +# skip fs check because superblock contains unknown ro-compat features
>> +_require_scratch_nocheck
>> +# Only V5 XFS disallow rw mount/remount with unknown ro-compat features
>> +_require_scratch_xfs_crc
>> +
>> +_scratch_mkfs_xfs>>$seqres.full 2>&1
>> +
>> +# set the highest bit of features_ro_compat, use it as an unknown
>> +# feature bit. If one day this bit become known feature, please
>> +# change this case.
>> +
>> +ro_compat=$(_scratch_xfs_get_metadata_field "features_ro_compat" "sb 0")
>> +echo $ro_compat | grep -q -E '^0x[[:xdigit:]]$'
>> +if [[ $? != 0 ]]; then
>> + echo "features_ro_compat has an invalid value."
>> +fi
>> +
>> +ro_compat=$(echo $ro_compat | \
>> + awk '/^0x[[:xdigit:]]+/ {
>> + printf("0x%x\n", or(strtonum($1), 0x80000000))
>> + }')
>> +
>> +# write the new ro compat field to the superblock
>> +_scratch_xfs_set_metadata_field "features_ro_compat" "$ro_compat" "sb 0" \
>> + > $seqres.full 2>&1
>> +
>> +# read the newly set ro compat filed for verification
>> +new_ro_compat=$(_scratch_xfs_get_metadata_field "features_ro_compat" "sb 0" \
>> + >$seqres.full 2>&1)
>
> Hi,
>
> Although you hit this issue by xfs/270, but that doesn't mean you need to do the
> same complicated calculation.
>
> You nearly can corrupt many fields to trigger that xfs_db bug, due to
> you don't need to test xfs mount. For example corrupt the features_ro_compat
> simply:
>
> # xfs_db -x -c "sb 0" -c "write -d features_ro_compat 0x80000000" /dev/sda3
> Allowing write of corrupted data with good CRC
> features_ro_compat = 0x80000000
> # xfs_db -c "sb 0" -c "print features_ro_compat" /dev/sda3
> Superblock has unknown read-only compatible features (0x80000000) enabled.
> Attempted to mount read-only compatible filesystem read-write.
> Filesystem can only be safely mounted read only.
> no current type
> cache_purge: shake on cache 0x562ed6463a70 left 1 nodes!?
> cache_purge: shake on cache 0x562ed6463a70 left 1 nodes!?
> cache_zero_check: refcount is 1, not zero (node=0x562ed6478110)
>
> Or even corrupt the sb magicnum (sb 0 need -F) directly:
>
> # xfs_db -x -c "sb 1" -c "write -d magicnum 0" /dev/sda3
> Allowing write of corrupted data and bad CRC
> magicnum = 0
> # xfs_db -c "sb 1" -c "print magicnum" /dev/sda3
> bad magic number
> no current type
> cache_purge: shake on cache 0x5589a86a9a70 left 1 nodes!?
> cache_purge: shake on cache 0x5589a86a9a70 left 1 nodes!?
> cache_zero_check: refcount is 1, not zero (node=0x5589a86be110)
I think corrputing sb 1 magic number is ok. Will use it on v2. Thanks.
>
> Or choose other fields you feel good.
>
>> +
>> +# verify the new ro_compat field is empty
>> +# Without xfsprog patch, xfs_db set_cur function doesn't return directly when
>> +# meeting error by using LIBXFS_READBUF_SALVAGE flag. So cache_purge will complain
>> +# about this.
>> +if [ -z $new_ro_compat ]; then
>> + grep -q "cache_purge: shake on cache" $seqres.full&& \
>> + echo "Hit xfsprogs set_cur bug"
>> + echo "Unable to get new ro compat filed"
>
> Let those "cache_purge ..." things break golden image (.out2 file) directly.
Yes.
Best Regard
Yang Xu
>
>> +fi
>> +
>> +echo "Silence is golden."
>> +
>> +# success, all done
>> +status=0
>> +exit
>> diff --git a/tests/xfs/082.out b/tests/xfs/082.out
>> new file mode 100644
>> index 00000000..6ed56cb1
>> --- /dev/null
>> +++ b/tests/xfs/082.out
>> @@ -0,0 +1,2 @@
>> +QA output created by 082
>> +Silence is golden.
>> --
>> 2.27.0
>>
>
