WARNING: fs/btrfs/extent-tree.c:6974 __btrfs_free_extent.isra.0.cold

WARNING: fs/btrfs/extent-tree.c:6974 __btrfs_free_extent.isra.0.cold

[Chris Murphy]

Filed bug here
https://bugzilla.kernel.org/show_bug.cgi?id=204557

kernel 5.2.8
progs  5.2.1

I'm getting a kernel warning, and tainted kernel following 'btrfs
scrub start' on a file system that was previously corrupted during
failure to clear v1 cache.

It's valid to complain about a problem with the file system that can't
be fixed. But for the scrub to abort and also taint the kernel sounds
like a bug?

-- 
Chris Murphy

Re: WARNING: fs/btrfs/extent-tree.c:6974 __btrfs_free_extent.isra.0.cold

[Qu Wenruo]

[-- Attachment #1.1: Type: text/plain, Size: 1011 bytes --]



On 2019/8/12 下午1:00, Chris Murphy wrote:
> Filed bug here
> https://bugzilla.kernel.org/show_bug.cgi?id=204557

Exactly the WARN_ON()/BUG_ON()s I'm going to kill in this patch:

https://patchwork.kernel.org/patch/11038303/

So hopefully in 5.3/5.4 release, it will be a more developer-friendly
output.
(But still not user-friendly)
> 
> kernel 5.2.8
> progs  5.2.1
> 
> I'm getting a kernel warning, and tainted kernel following 'btrfs
> scrub start' on a file system that was previously corrupted during
> failure to clear v1 cache.
> 
> It's valid to complain about a problem with the file system that can't
> be fixed. But for the scrub to abort and also taint the kernel sounds
> like a bug?
> 
Corruption in extent tree is a serious problem, it could easily screw
btrfs CoW, so we *must* abort transaction when we find anything wrong on
extent tree.
(Ironically, most corruption happens in extent tree).

For the warning part, it is going to be fixed.

Thanks,
Qu


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