Re: tree-checker read time corruption

[telsch <telsch@gmx.de>]

On 2/13/20 1:24 AM, Qu Wenruo wrote:
>
>
> On 2020/2/12 下午10:20, telsch wrote:
>>
>>
>> On 2/12/20 1:41 AM, Qu Wenruo wrote:
>>>
>>>
>>> On 2020/2/11 下午10:17, telsch wrote:
>>>> Dear devs,
>>>>
>>>>
>>>>
>>>> after upgrading from kernel 4.19.101 to 5.5.2 i got read time tree block
>>>> error as
>>>>
>>>> described here:
>>>>
>>>>       https://btrfs.wiki.kernel.org/index.php/Tree-checker#For_end_users
>>>>
>>>>
>>>>
>>>> Working with kernel 4.19.101:
>>>>
>>>>
>>>>
>>>> Linux Arch 4.19.101-1-lts #1 SMP Sat, 01 Feb 2020 16:35:36 +0000 x86_64
>>>> GNU/Linux
>>>>
>>>>
>>>>
>>>> btrfs --version
>>>>
>>>> btrfs-progs v5.4
>>>>
>>>>
>>>>
>>>> btrfs fi show
>>>>
>>>> Label: none  uuid: 56e753f4-1346-49ad-a34f-e93a0235b82a
>>>>
>>>>           Total devices 1 FS bytes used 92.54GiB
>>>>
>>>>           devid    1 size 95.14GiB used 95.14GiB path /dev/mapper/home
>>>>
>>>>
>>>>
>>>> btrfs fi df /home
>>>>
>>>> Data, single: total=94.11GiB, used=91.95GiB
>>>>
>>>> System, single: total=31.00MiB, used=12.00KiB
>>>>
>>>> Metadata, single: total=1.00GiB, used=599.74MiB
>>>>
>>>> GlobalReserve, single: total=199.32MiB, used=0.00B
>>>>
>>>>
>>>>
>>>> After upgrading to kernel 5.5.2:
>>>>
>>>>
>>>>
>>>> [   13.413025] BTRFS: device fsid 56e753f4-1346-49ad-a34f-e93a0235b82a
>>>> devid 1 transid 468295 /dev/dm-1 scanned by systemd-udevd (417)
>>>>
>>>> [   13.589952] BTRFS info (device dm-1): force zstd compression, level 3
>>>>
>>>> [   13.589956] BTRFS info (device dm-1): disk space caching is enabled
>>>>
>>>> [   13.594707] BTRFS info (device dm-1): bdev /dev/mapper/home errs: wr
>>>> 0, rd 47, flush 0, corrupt 0, gen 0
>>>>
>>>> [   13.622912] BTRFS info (device dm-1): enabling ssd optimizations
>>>>
>>>> [   13.624300] BTRFS critical (device dm-1): corrupt leaf: root=5
>>>> block=122395779072 slot=10 ino=265, invalid inode generation: has
>>>> 18446744073709551492 expect [0, 468296]
>>>
>>> An older kernel caused underflow/garbage generation.
>>> Much strict tree checker is detecting it and rejecting the tree block to
>>> prevent further corruption.
>>>
>>> It can be fixed in by btrfs-progs v5.4 and later, by using 'btrfs check
>>> --repair'
>>>
>>> Early btrfs-progs can't detect nor fix it.
>>>
>>> Thanks,
>>> Qu
>>>
>>
>> As you suggest booting to kernel 5.5.3 with btrfs-progs v5.4 and run
>> 'btrfs check --repair'. But didn't fix this error.
>>
>> mount: /home: can't read superblock on /dev/mapper/home.
>> [  325.121475] BTRFS info (device dm-1): force zstd compression, level 3
>> [  325.121482] BTRFS info (device dm-1): disk space caching is enabled
>> [  325.126234] BTRFS info (device dm-1): bdev /dev/mapper/home errs: wr
>> 0, rd 47, flush 0, corrupt 0, gen 0
>> [  325.143521] BTRFS info (device dm-1): enabling ssd optimizations
>> [  325.146138] BTRFS critical (device dm-1): corrupt leaf: root=5
>> block=122395779072 slot=10 ino=265, invalid inode generation: has
>> 18446744073709551492 expect [0, 469820]
>> [  325.148637] BTRFS error (device dm-1): block=122395779072 read time
>> tree block corruption detected
>
> According to the repair log, btrfs-progs doesn't detect it at all.
> Thus I'm not sure if it's a bug in btrfs-progs or it's just not newer
> enough.
>
> Anyway, you can delete inode 265 manually using older kernel.
>
> Thanks,
> Qu

Thanks for support. Deleting inode 265 manually fixed this mount issue
with kernel >= 5.2 for me.

>>
>>>>
>>>> [   13.624381] BTRFS error (device dm-1): block=122395779072 read time
>>>> tree block corruption detected
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> Booting from 4.19 kernel can mount fs again.
>>>
>