btrfs corruption: invalid mode: has 00 expect valid S_IF* bit(s)

btrfs corruption: invalid mode: has 00 expect valid S_IF* bit(s)

[Jim Geo]

Hello,

my  /home folder is on a single device btrfs partition.
When I ls a directory, I get these messages on ls:
ls: cannot access 'file1' : Input/output error
ls: cannot access 'file2' : Input/output error
ls: cannot access 'file3' : Input/output error

and dmesg says:
[31209.938486] BTRFS critical (device sdd1): corrupt leaf: root=5
block=859701248 slot=93 ino=5830829, invalid mode: has 00 expect valid
S_IF* bit(s)
[31209.938489] BTRFS error (device sdd1): block=859701248 read time
tree block corruption detected
[31209.948749] BTRFS critical (device sdd1): corrupt leaf: root=5
block=859701248 slot=93 ino=5830829, invalid mode: has 00 expect valid
S_IF* bit(s)
[31209.948751] BTRFS error (device sdd1): block=859701248 read time
tree block corruption detected
[31209.949109] BTRFS critical (device sdd1): corrupt leaf: root=5
block=859701248 slot=93 ino=5830829, invalid mode: has 00 expect valid
S_IF* bit(s)
[31209.949110] BTRFS error (device sdd1): block=859701248 read time
tree block corruption detected
[31209.949751] BTRFS critical (device sdd1): corrupt leaf: root=5
block=859701248 slot=93 ino=5830829, invalid mode: has 00 expect valid
S_IF* bit(s)
[31209.949752] BTRFS error (device sdd1): block=859701248 read time
tree block corruption detected
[31209.950163] BTRFS critical (device sdd1): corrupt leaf: root=5
block=859701248 slot=93 ino=5830829, invalid mode: has 00 expect valid
S_IF* bit(s)
[31209.950165] BTRFS error (device sdd1): block=859701248 read time
tree block corruption detected
[31209.950320] BTRFS critical (device sdd1): corrupt leaf: root=5
block=859701248 slot=93 ino=5830829, invalid mode: has 00 expect valid
S_IF* bit(s)
[31209.950321] BTRFS error (device sdd1): block=859701248 read time
tree block corruption detected

I scrubbed the filesystem but no errors were detected/fixed.

btrfs scrub status /home:
scrub status for myuuid
        scrub started at Thu Aug 15 02:24:42 2019 and finished after 03:55:12
        total bytes scrubbed: 2.05TiB with 0 errors

 uname -a:
Linux gentoo 5.2.8-ck #1 SMP PREEMPT Wed Aug 14 20:44:33 EEST 2019
x86_64 Intel(R) Core(TM) i5-4460 CPU @ 3.20GHz GenuineIntel GNU/Linux

btrfs --version:
btrfs-progs v4.19

btrfs fi show:
Label: 'home_btrfs'  uuid:-------
        Total devices 1 FS bytes used 2.04TiB
        devid    1 size 2.13TiB used 2.13TiB path /dev/sdd1

btrfs fi df /home:
Data, single: total=2.12TiB, used=2.04TiB
System, DUP: total=64.00MiB, used=256.00KiB
Metadata, DUP: total=7.00GiB, used=4.48GiB
GlobalReserve, single: total=512.00MiB, used=0.00B

btrfs device stats /home:
[/dev/sdd1].write_io_errs    4
[/dev/sdd1].read_io_errs     3
[/dev/sdd1].flush_io_errs    0
[/dev/sdd1].corruption_errs  0
[/dev/sdd1].generation_errs  0

mount options:
/dev/sdd1 on /home type btrfs
(rw,relatime,space_cache,autodefrag,subvolid=5,subvol=/)

How can I fix this corruption? How can I detect if more
files/directories are affected?

Kind Regards,
Jim

Re: btrfs corruption: invalid mode: has 00 expect valid S_IF* bit(s)

[Qu Wenruo]

[-- Attachment #1.1: Type: text/plain, Size: 2351 bytes --]



On 2019/8/15 下午4:39, Jim Geo wrote:
> Hello,
> 
> my  /home folder is on a single device btrfs partition.
> When I ls a directory, I get these messages on ls:
> ls: cannot access 'file1' : Input/output error
> ls: cannot access 'file2' : Input/output error
> ls: cannot access 'file3' : Input/output error
> 
> and dmesg says:
> [31209.938486] BTRFS critical (device sdd1): corrupt leaf: root=5
> block=859701248 slot=93 ino=5830829, invalid mode: has 00 expect valid
> S_IF* bit(s)

Please provide the dump of that tree blocks.

# btrfs ins dump-tree -b 859701248 /dev/sdd1

It looks like that tree block has something wrong with it.
Thus kernel reject to accept the data.

[...]
> 
> I scrubbed the filesystem but no errors were detected/fixed.

Scrub mostly only verify checksum, doesn't care about the data in the
tree blocks.

> 
> btrfs scrub status /home:
> scrub status for myuuid
>         scrub started at Thu Aug 15 02:24:42 2019 and finished after 03:55:12
>         total bytes scrubbed: 2.05TiB with 0 errors
> 
>  uname -a:
> Linux gentoo 5.2.8-ck #1 SMP PREEMPT Wed Aug 14 20:44:33 EEST 2019
> x86_64 Intel(R) Core(TM) i5-4460 CPU @ 3.20GHz GenuineIntel GNU/Linux
> 
> btrfs --version:
> btrfs-progs v4.19
> 
> btrfs fi show:
> Label: 'home_btrfs'  uuid:-------
>         Total devices 1 FS bytes used 2.04TiB
>         devid    1 size 2.13TiB used 2.13TiB path /dev/sdd1
> 
> btrfs fi df /home:
> Data, single: total=2.12TiB, used=2.04TiB
> System, DUP: total=64.00MiB, used=256.00KiB
> Metadata, DUP: total=7.00GiB, used=4.48GiB
> GlobalReserve, single: total=512.00MiB, used=0.00B
> 
> btrfs device stats /home:
> [/dev/sdd1].write_io_errs    4
> [/dev/sdd1].read_io_errs     3
> [/dev/sdd1].flush_io_errs    0
> [/dev/sdd1].corruption_errs  0
> [/dev/sdd1].generation_errs  0
> 
> mount options:
> /dev/sdd1 on /home type btrfs
> (rw,relatime,space_cache,autodefrag,subvolid=5,subvol=/)
> 
> How can I fix this corruption? How can I detect if more
> files/directories are affected?

Affected inode number is 5830829. But I need to above mentioned dump to
make sure if it's not a false alert.

If it's truely corrupted by whatever the reason, you'd
better check your memory, as it looks like a bit flip.

Thanks,
Qu
> 
> Kind Regards,
> Jim
> 


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: btrfs corruption: invalid mode: has 00 expect valid S_IF* bit(s)

[Qu Wenruo]

[-- Attachment #1.1: Type: text/plain, Size: 4060 bytes --]



On 2019/8/15 下午5:41, Jim Geo wrote:
> Στις Πέμ, 15 Αυγ 2019 στις 12:12 μ.μ., ο/η Qu Wenruo
> <quwenruo.btrfs@gmx.com> έγραψε:
>>
>>
>>
>> On 2019/8/15 下午4:39, Jim Geo wrote:
>>> Hello,
>>>
>>> my  /home folder is on a single device btrfs partition.
>>> When I ls a directory, I get these messages on ls:
>>> ls: cannot access 'file1' : Input/output error
>>> ls: cannot access 'file2' : Input/output error
>>> ls: cannot access 'file3' : Input/output error
>>>
>>> and dmesg says:
>>> [31209.938486] BTRFS critical (device sdd1): corrupt leaf: root=5
>>> block=859701248 slot=93 ino=5830829, invalid mode: has 00 expect valid
>>> S_IF* bit(s)
>>
>> Please provide the dump of that tree blocks.
>>
>> # btrfs ins dump-tree -b 859701248 /dev/sdd1
>>
>> It looks like that tree block has something wrong with it.
>> Thus kernel reject to accept the data.
> Hello,
> 
> Attached you can find the dump.
> I changed some filenames for privacy but the filenames that came from
> the dump are the ones that ls says that are problematic.

Forgot to mention that, feel free to censor the filenames.

And it turns out that, it's indeed a corruption.

        item 93 key (5830829 INODE_ITEM 0) itemoff 5613 itemsize 160
                generation 99667 transid 99669 size 0 nbytes 0
                block group 0 mode 0 links 1 uid 0 gid 0 rdev 0
                sequence 2 flags 0x0(none)
                atime 0.0 (1970-01-01 02:00:00)
                ctime 1516386417.601932603 (2018-01-19 20:26:57)
                mtime 0.0 (1970-01-01 02:00:00)
                otime 0.0 (1970-01-01 02:00:00)
        item 94 key (5830829 INODE_REF 376204) itemoff 5584 itemsize 29
                index 61648 namelen 19 name: <censored>

First, it's an empty file without any S_IFMT flag.

I believe it's a regular file, but it's created very recently.

You can try to remove that file with older kernel (before v5.2 should
not trigger such problem).
Then the problem should be solved.

But please consider double check your memory, as it looks like a memory
bitflip or even corruption.

And also, please consider do a btrfs check --readonly to see if there is
any other problem.

Thanks,
Qu

> 
> Thanks a lot!
>>
>> [...]
>>>
>>> I scrubbed the filesystem but no errors were detected/fixed.
>>
>> Scrub mostly only verify checksum, doesn't care about the data in the
>> tree blocks.
>>
>>>
>>> btrfs scrub status /home:
>>> scrub status for myuuid
>>>         scrub started at Thu Aug 15 02:24:42 2019 and finished after 03:55:12
>>>         total bytes scrubbed: 2.05TiB with 0 errors
>>>
>>>  uname -a:
>>> Linux gentoo 5.2.8-ck #1 SMP PREEMPT Wed Aug 14 20:44:33 EEST 2019
>>> x86_64 Intel(R) Core(TM) i5-4460 CPU @ 3.20GHz GenuineIntel GNU/Linux
>>>
>>> btrfs --version:
>>> btrfs-progs v4.19
>>>
>>> btrfs fi show:
>>> Label: 'home_btrfs'  uuid:-------
>>>         Total devices 1 FS bytes used 2.04TiB
>>>         devid    1 size 2.13TiB used 2.13TiB path /dev/sdd1
>>>
>>> btrfs fi df /home:
>>> Data, single: total=2.12TiB, used=2.04TiB
>>> System, DUP: total=64.00MiB, used=256.00KiB
>>> Metadata, DUP: total=7.00GiB, used=4.48GiB
>>> GlobalReserve, single: total=512.00MiB, used=0.00B
>>>
>>> btrfs device stats /home:
>>> [/dev/sdd1].write_io_errs    4
>>> [/dev/sdd1].read_io_errs     3
>>> [/dev/sdd1].flush_io_errs    0
>>> [/dev/sdd1].corruption_errs  0
>>> [/dev/sdd1].generation_errs  0
>>>
>>> mount options:
>>> /dev/sdd1 on /home type btrfs
>>> (rw,relatime,space_cache,autodefrag,subvolid=5,subvol=/)
>>>
>>> How can I fix this corruption? How can I detect if more
>>> files/directories are affected?
>>
>> Affected inode number is 5830829. But I need to above mentioned dump to
>> make sure if it's not a false alert.
>>
>> If it's truely corrupted by whatever the reason, you'd
>> better check your memory, as it looks like a bit flip.
>>
>> Thanks,
>> Qu
>>>
>>> Kind Regards,
>>> Jim
>>>
>>


[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 488 bytes --]