Re: [PATCH v6 4/5] dm-thin: Add REQ_OP_PROVISION support

[Mike Snitzer <snitzer@kernel.org>]

On Sat, May 06 2023 at  2:29P -0400,
Sarthak Kukreti <sarthakkukreti@chromium.org> wrote:

> dm-thinpool uses the provision request to provision
> blocks for a dm-thin device. dm-thinpool currently does not
> pass through REQ_OP_PROVISION to underlying devices.
> 
> For shared blocks, provision requests will break sharing and copy the
> contents of the entire block. Additionally, if 'skip_block_zeroing'
> is not set, dm-thin will opt to zero out the entire range as a part
> of provisioning.
> 
> Signed-off-by: Sarthak Kukreti <sarthakkukreti@chromium.org>
> ---
>  drivers/md/dm-thin.c | 70 +++++++++++++++++++++++++++++++++++++++++---
>  1 file changed, 66 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
> index 2b13c949bd72..3f94f53ac956 100644
> --- a/drivers/md/dm-thin.c
> +++ b/drivers/md/dm-thin.c
...
> @@ -4114,6 +4171,8 @@ static void pool_io_hints(struct dm_target *ti, struct queue_limits *limits)
>  	 * The pool uses the same discard limits as the underlying data
>  	 * device.  DM core has already set this up.
>  	 */
> +
> +	limits->max_provision_sectors = pool->sectors_per_block;

Just noticed that setting limits->max_provision_sectors needs to move
above pool_io_hints code that sets up discards -- otherwise the early
return from if (!pt->adjusted_pf.discard_enabled) will cause setting
max_provision_sectors to be skipped.

Here is a roll up of the fixes that need to be folded into this patch:

diff --git a/drivers/md/dm-thin.c b/drivers/md/dm-thin.c
index 3f94f53ac956..90c8e36cb327 100644
--- a/drivers/md/dm-thin.c
+++ b/drivers/md/dm-thin.c
@@ -4151,6 +4151,8 @@ static void pool_io_hints(struct dm_target *ti, struct queue_limits *limits)
 		blk_limits_io_opt(limits, pool->sectors_per_block << SECTOR_SHIFT);
 	}
 
+	limits->max_provision_sectors = pool->sectors_per_block;
+
 	/*
 	 * pt->adjusted_pf is a staging area for the actual features to use.
 	 * They get transferred to the live pool in bind_control_target()
@@ -4171,8 +4173,6 @@ static void pool_io_hints(struct dm_target *ti, struct queue_limits *limits)
 	 * The pool uses the same discard limits as the underlying data
 	 * device.  DM core has already set this up.
 	 */
-
-	limits->max_provision_sectors = pool->sectors_per_block;
 }
 
 static struct target_type pool_target = {
@@ -4349,6 +4349,7 @@ static int thin_ctr(struct dm_target *ti, unsigned int argc, char **argv)
 
 	ti->num_provision_bios = 1;
 	ti->provision_supported = true;
+	ti->max_provision_granularity = true;
 
 	mutex_unlock(&dm_thin_pool_table.mutex);