Re: mount: can't read superblock on - corrupt leaf - read time tree block corruption detected

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

在 2024/10/9 06:02, Johannes Hirte 写道:
> On 2024 Okt 03, Qu Wenruo wrote:
>>
>> 在 2024/10/3 17:02, cwalou@gmail.com 写道:
>>> Hello.
>>>
>>> A 4TB drive taken out of a synology NAS. When I try to mount it, it
>>> won't. This is what I did :
>>
>> Synology has out-of-tree features that upstream kernel doesn't support.
>>
>
> Hello,
>
> I've noticed similar errors on my server too. It's a gentoo box with
> vanilla kernel. It happend first with linux-6.10.9 and happens with 6.11
> too:

This is a different one from the original report.

>
> [   67.310316] BTRFS: device label bigraid1 devid 1 transid 98985 /dev/sdd1 (8:49) scanned by mount (4069)
> [   67.313109] BTRFS info (device sdd1): first mount of filesystem 83a02224-c807-4e70-a2cd-8b7156bfbbd2
> [   67.313137] BTRFS info (device sdd1): using crc32c (crc32c-generic) checksum algorithm
> [   67.313156] BTRFS info (device sdd1): using free-space-tree
> [   67.527088] page: refcount:4 mapcount:0 mapping:000000004bf8c7c2 index:0x2ffb1be8 pfn:0x11b55c
> [   67.527114] memcg:ffff8881174ab800
> [   67.527120] aops:0xffffffff8206c960 ino:1
> [   67.527129] flags: 0x2000000000004000(private|node=0|zone=2)
> [   67.527144] raw: 2000000000004000 0000000000000000 dead000000000122 ffff8881022c9678
> [   67.527152] raw: 000000002ffb1be8 ffff88811b5304b0 00000004ffffffff ffff8881174ab800
> [   67.527157] page dumped because: eb page dump
> [   67.527162] BTRFS critical (device sdd1): corrupt leaf: block=3297221967872 slot=47 extent bytenr=2176663552 len=36864 invalid data ref objectid value 259

This is caused by the long deprecated inode_cache feature.

Newer btrfs-check can report it as an error.

You can fix it by "btrfs rescue clear-ino-cache <device>" on the
unmounted fs, using btrfs-progs v6.11 (the command is there for a long
time, but has some small bugs).

Thanks,
Qu
> [   67.527179] BTRFS error (device sdd1): read time tree block corruption detected on logical 3297221967872 mirror 1
> [   67.535814] page: refcount:4 mapcount:0 mapping:000000004bf8c7c2 index:0x2ffb1be8 pfn:0x11b55c
> [   67.535872] memcg:ffff8881174ab800
> [   67.535877] aops:0xffffffff8206c960 ino:1
> [   67.535886] flags: 0x2000000000004000(private|node=0|zone=2)
> [   67.535900] raw: 2000000000004000 0000000000000000 dead000000000122 ffff8881022c9678
> [   67.535907] raw: 000000002ffb1be8 ffff88811b5304b0 00000004ffffffff ffff8881174ab800
> [   67.535912] page dumped because: eb page dump
> [   67.535917] BTRFS critical (device sdd1): corrupt leaf: block=3297221967872 slot=47 extent bytenr=2176663552 len=36864 invalid data ref objectid value 259
> [   67.535931] BTRFS error (device sdd1): read time tree block corruption detected on logical 3297221967872 mirror 2
> [   67.535996] BTRFS error (device sdd1): failed to read block groups: -5
> [   67.539666] BTRFS error (device sdd1): open_ctree failed
> [   70.026232] BTRFS info (device sdb1): first mount of filesystem cae06473-5f45-4cd4-a822-733c036e36d9
> [   70.026270] BTRFS info (device sdb1): using crc32c (crc32c-generic) checksum algorithm
> [   70.026300] BTRFS info (device sdb1): disk space caching is enabled
> [   70.026306] BTRFS warning (device sdb1): space cache v1 is being deprecated and will be removed in a future release, please use -o space_cache=v2
> [   70.130069] page: refcount:3 mapcount:0 mapping:00000000a884d48d index:0x5abd618 pfn:0x11b665
> [   70.130094] memcg:ffff8881174ab800
> [   70.130100] aops:0xffffffff8206c960 ino:1
> [   70.130110] flags: 0x2400000000004020(lru|private|node=0|zone=2)
> [   70.130126] raw: 2400000000004020 ffffea00046d9908 ffffea00046d9988 ffff888102262618
> [   70.130134] raw: 0000000005abd618 ffff88811b63a4b0 00000003ffffffff ffff8881174ab800
> [   70.130139] page dumped because: eb page dump
> [   70.130144] BTRFS critical (device sdb1): corrupt leaf: block=389724340224 slot=7 extent bytenr=12652544 len=36864 invalid data ref objectid value 258
> [   70.130160] BTRFS error (device sdb1): read time tree block corruption detected on logical 389724340224 mirror 2
> [   70.136382] page: refcount:3 mapcount:0 mapping:00000000a884d48d index:0x5abd618 pfn:0x11b665
> [   70.136404] memcg:ffff8881174ab800
> [   70.136409] aops:0xffffffff8206c960 ino:1
> [   70.136418] flags: 0x2400000000004020(lru|private|node=0|zone=2)
> [   70.136431] raw: 2400000000004020 ffffea00046d9908 ffffea00046d9988 ffff888102262618
> [   70.136439] raw: 0000000005abd618 ffff88811b63a4b0 00000003ffffffff ffff8881174ab800
> [   70.136443] page dumped because: eb page dump
> [   70.136448] BTRFS critical (device sdb1): corrupt leaf: block=389724340224 slot=7 extent bytenr=12652544 len=36864 invalid data ref objectid value 258
> [   70.136462] BTRFS error (device sdb1): read time tree block corruption detected on logical 389724340224 mirror 1
> [   70.136555] BTRFS error (device sdb1): failed to read block groups: -5
> [   70.138216] BTRFS error (device sdb1): open_ctree failed
> [   71.825004] BTRFS: device label BIGBIGRAID devid 1 transid 4121 /dev/sde1 (8:65) scanned by mount (4102)
> [   71.826496] BTRFS info (device sde1): first mount of filesystem c9fa0269-733e-4e76-88c9-68c8247c81e5
> [   71.826532] BTRFS info (device sde1): using crc32c (crc32c-generic) checksum algorithm
> [   71.826561] BTRFS info (device sde1): using free-space-tree
> [  140.718646] BTRFS info (device sdb1): first mount of filesystem cae06473-5f45-4cd4-a822-733c036e36d9
> [  140.718683] BTRFS info (device sdb1): using crc32c (crc32c-generic) checksum algorithm
> [  140.718711] BTRFS info (device sdb1): disk space caching is enabled
> [  140.718717] BTRFS warning (device sdb1): space cache v1 is being deprecated and will be removed in a future release, please use -o space_cache=v2
> [  140.845547] page: refcount:4 mapcount:0 mapping:0000000049bde2be index:0x5abd618 pfn:0x11d9a3
> [  140.845571] memcg:ffff8881174ab800
> [  140.845578] aops:0xffffffff8206c960 ino:1
> [  140.845587] flags: 0x2000000000004000(private|node=0|zone=2)
> [  140.845602] raw: 2000000000004000 0000000000000000 dead000000000122 ffff888102262de8
> [  140.845611] raw: 0000000005abd618 ffff88811d96f5a0 00000004ffffffff ffff8881174ab800
> [  140.845615] page dumped because: eb page dump
> [  140.845621] BTRFS critical (device sdb1): corrupt leaf: block=389724340224 slot=7 extent bytenr=12652544 len=36864 invalid data ref objectid value 258
> [  140.845636] BTRFS error (device sdb1): read time tree block corruption detected on logical 389724340224 mirror 1
> [  140.846180] page: refcount:4 mapcount:0 mapping:0000000049bde2be index:0x5abd618 pfn:0x11d9a3
> [  140.846202] memcg:ffff8881174ab800
> [  140.846208] aops:0xffffffff8206c960 ino:1
> [  140.846216] flags: 0x2000000000004000(private|node=0|zone=2)
> [  140.846229] raw: 2000000000004000 0000000000000000 dead000000000122 ffff888102262de8
> [  140.846236] raw: 0000000005abd618 ffff88811d96f5a0 00000004ffffffff ffff8881174ab800
> [  140.846240] page dumped because: eb page dump
> [  140.846245] BTRFS critical (device sdb1): corrupt leaf: block=389724340224 slot=7 extent bytenr=12652544 len=36864 invalid data ref objectid value 258
> [  140.846259] BTRFS error (device sdb1): read time tree block corruption detected on logical 389724340224 mirror 2
> [  140.846329] BTRFS error (device sdb1): failed to read block groups: -5
> [  140.848015] BTRFS error (device sdb1): open_ctree failed
>
>
> It seems following commit is responsible:
>
> commit f333a3c7e8323499aa65038e77fe8f3199d4e283
> Author: Qu Wenruo <wqu@suse.com>
> Date:   Mon Jul 15 16:07:07 2024 +0930
>
>      btrfs: tree-checker: validate dref root and objectid
>
>
> After reverting this, I'm able to mount the filesystems again. There are
> three filesystems on this server that are affected, two single disk and
> one with two disk in RAID1/0. I was able to copy all data from all
> filesytems via rsync without noticing any error when running without
> this commit. Also btrfscheck doesn't notice any error. It's just this
> commit that prevents these three filesystem from mounting. As I was able
> to get all data this is not critical to me. But I think this should be
> addressed in some way.
>
> regards,
>    Johannes