Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.18 as permitted sender) client-ip=23.128.96.18;
Date:   Wed, 9 Dec 2020 17:36:15 -0500
From:   Mike Snitzer <snitzer@redhat.com>
To:     Song Liu <songliubraving@fb.com>, axboe@kernel.dk
Cc:     linux-raid@vger.kernel.org, linux-kernel@vger.kernel.org,
        dm-devel@redhat.com,
        Matthew Ruffell <matthew.ruffell@canonical.com>,
        Xiao Ni <xni@redhat.com>
Subject: Re: Revert "dm raid: remove unnecessary discard limits for raid10"
Message-ID: <20201209223615.GA2752@redhat.com>
References: <20201209215814.2623617-1-songliubraving@fb.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20201209215814.2623617-1-songliubraving@fb.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
Precedence: bulk

On Wed, Dec 09 2020 at  4:58pm -0500,
Song Liu <songliubraving@fb.com> wrote:

> This reverts commit f0e90b6c663a7e3b4736cb318c6c7c589f152c28.
> 
> Matthew Ruffell reported data corruption in raid10 due to the changes
> in discard handling [1]. Revert these changes before we find a proper fix.
> 
> [1] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1907262/
> Cc: Matthew Ruffell <matthew.ruffell@canonical.com>
> Cc: Xiao Ni <xni@redhat.com>
> Cc: Mike Snitzer <snitzer@redhat.com>
> Signed-off-by: Song Liu <songliubraving@fb.com>

If you're reverting all the MD code that enabled this DM change, then
obviously the DM change must be reverted too.  But please do _not_
separate the DM revert from the MD reverts.

Please include this in a v2 pull request to Jens.

Mike

> ---
>  drivers/md/dm-raid.c | 11 +++++++++++
>  1 file changed, 11 insertions(+)
> 
> diff --git a/drivers/md/dm-raid.c b/drivers/md/dm-raid.c
> index 9c1f7c4de65b3..dc8568ab96f24 100644
> --- a/drivers/md/dm-raid.c
> +++ b/drivers/md/dm-raid.c
> @@ -3728,6 +3728,17 @@ static void raid_io_hints(struct dm_target *ti, struct queue_limits *limits)
>  
>  	blk_limits_io_min(limits, chunk_size_bytes);
>  	blk_limits_io_opt(limits, chunk_size_bytes * mddev_data_stripes(rs));
> +
> +	/*
> +	 * RAID10 personality requires bio splitting,
> +	 * RAID0/1/4/5/6 don't and process large discard bios properly.
> +	 */
> +	if (rs_is_raid10(rs)) {
> +		limits->discard_granularity = max(chunk_size_bytes,
> +						  limits->discard_granularity);
> +		limits->max_discard_sectors = min_not_zero(rs->md.chunk_sectors,
> +							   limits->max_discard_sectors);
> +	}
>  }
>  
>  static void raid_postsuspend(struct dm_target *ti)
> -- 
> 2.24.1
>