Re: [PATCH 4.4 038/268] Btrfs: fix scrub to repair raid6 corruption

[Ben Hutchings]

On Mon, 2018-05-28 at 12:00 +0200, Greg Kroah-Hartman wrote:
> 4.4-stable review patch.ÂÂIf anyone has any objections, please let me know.
> 
> ------------------
> 
> From: Liu Bo <bo.li.liu@xxxxxxxxxx>
> 
> [ Upstream commit 762221f095e3932669093466aaf4b85ed9ad2ac1 ]

The diff here is actually from commit 8810f7517a3b ("Btrfs: make raid6
rebuild retry more", mentioned in this commit message).  (Sasha, please
try to work out why commit messages and descriptions are getting mixed
up in your auto-selections.)

Maybe stable branches should get the real commit 762221f095e3 as well?

Ben.

> The raid6 corruption is that,
> suppose that all disks can be read without problems and if the content
> that was read out doesn't match its checksum, currently for raid6
> btrfs at most retries twice,
> 
> - the 1st retry is to rebuild with all other stripes, it'll eventually
> Â be a raid5 xor rebuild,
> - if the 1st fails, the 2nd retry will deliberately fail parity p so
> Â that it will do raid6 style rebuild,
> 
> however, the chances are that another non-parity stripe content also
> has something corrupted, so that the above retries are not able to
> return correct content.
> 
> We've fixed normal reads to rebuild raid6 correctly with more retries
> in Patch "Btrfs: make raid6 rebuild retry more"[1], this is to fix
> scrub to do the exactly same rebuild process.
> 
> [1]: https://patchwork.kernel.org/patch/10091755/
> 
> Signed-off-by: Liu Bo <bo.li.liu@xxxxxxxxxx>
> Signed-off-by: David Sterba <dsterba@xxxxxxxx>
> Signed-off-by: Sasha Levin <alexander.levin@xxxxxxxxxxxxx>
> Signed-off-by: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> ---
> Âfs/btrfs/raid56.cÂÂ|ÂÂÂ18 ++++++++++++++----
> Âfs/btrfs/volumes.c |ÂÂÂÂ9 ++++++++-
> Â2 files changed, 22 insertions(+), 5 deletions(-)
> 
> --- a/fs/btrfs/raid56.c
> +++ b/fs/btrfs/raid56.c
> @@ -2160,11 +2160,21 @@ int raid56_parity_recover(struct btrfs_r
> Â	}
> Â
> Â	/*
> -	Â* reconstruct from the q stripe if they are
> -	Â* asking for mirror 3
> +	Â* Loop retry:
> +	Â* for 'mirror == 2', reconstruct from all other stripes.
> +	Â* for 'mirror_num > 2', select a stripe to fail on every retry.
> Â	Â*/
> -	if (mirror_num == 3)
> -		rbio->failb = rbio->real_stripes - 2;
> +	if (mirror_num > 2) {
> +		/*
> +		Â* 'mirror == 3' is to fail the p stripe and
> +		Â* reconstruct from the q stripe.ÂÂ'mirror > 3' is to
> +		Â* fail a data stripe and reconstruct from p+q stripe.
> +		Â*/
> +		rbio->failb = rbio->real_stripes - (mirror_num - 1);
> +		ASSERT(rbio->failb > 0);
> +		if (rbio->failb <= rbio->faila)
> +			rbio->failb--;
> +	}
> Â
> Â	ret = lock_stripe_add(rbio);
> Â
> --- a/fs/btrfs/volumes.c
> +++ b/fs/btrfs/volumes.c
> @@ -5056,7 +5056,14 @@ int btrfs_num_copies(struct btrfs_fs_inf
> Â	else if (map->type & BTRFS_BLOCK_GROUP_RAID5)
> Â		ret = 2;
> Â	else if (map->type & BTRFS_BLOCK_GROUP_RAID6)
> -		ret = 3;
> +		/*
> +		Â* There could be two corrupted data stripes, we need
> +		Â* to loop retry in order to rebuild the correct data.
> +		Â*
> +		Â* Fail a stripe at a time on every retry except the
> +		Â* stripe under reconstruction.
> +		Â*/
> +		ret = map->num_stripes;
> Â	else
> Â		ret = 1;
> Â	free_extent_map(em);

-- 
Ben Hutchings, Software Developer                Â        Codethink Ltd
https://www.codethink.co.uk/                 Dale House, 35 Dale Street
                                     Manchester, M1 2HF, United Kingdom