This series introduces initial device mapper atomic write support.

Since we already support stacking atomic writes limits, it's quite
straightforward to support.

Personalities dm-linear, dm-stripe, and dm-raid1 are supported here, and
more personalities could be supported in future.

This is still an RFC as I would like to test further.

Based on 3d9a9e9a77c5 (block/for-6.14/block) block: limit disk max
sectors to (LLONG_MAX >> 9)

Changes to v1:
- Generic block layer atomic writes enable flag and dm-table rework
- Add dm-stripe and dm-raid1 support
- Add bio_trim() patch

John Garry (8):
  block: Add common atomic writes enable flag
  block: Don't trim an atomic write
  dm-table: atomic writes support
  dm: Ensure cloned bio is same length for atomic write
  dm-linear: Enable atomic writes
  dm-stripe: Enable atomic writes
  dm-io: Warn on creating multiple atomic write bios for a region
  dm-mirror: Support atomic writes

 block/bio.c                   |  4 ++++
 block/blk-settings.c          |  6 ++++--
 drivers/md/dm-io.c            |  1 +
 drivers/md/dm-linear.c        |  3 ++-
 drivers/md/dm-raid1.c         |  3 ++-
 drivers/md/dm-stripe.c        |  3 ++-
 drivers/md/dm-table.c         | 29 +++++++++++++++++++++++++++++
 drivers/md/dm.c               |  3 +++
 drivers/md/raid0.c            |  2 +-
 drivers/md/raid1.c            |  2 +-
 drivers/md/raid10.c           |  2 +-
 drivers/nvme/host/core.c      |  1 +
 drivers/scsi/sd.c             |  1 +
 include/linux/blkdev.h        |  4 ++--
 include/linux/device-mapper.h |  3 +++
 15 files changed, 57 insertions(+), 10 deletions(-)

-- 
2.31.1

On Thu, Jan 16, 2025 at 05:02:53PM +0000, John Garry wrote:
> This series introduces initial device mapper atomic write support.
> 
> Since we already support stacking atomic writes limits, it's quite
> straightforward to support.
> 
> Personalities dm-linear, dm-stripe, and dm-raid1 are supported here, and
> more personalities could be supported in future.
> 
> This is still an RFC as I would like to test further.
> 
> Based on 3d9a9e9a77c5 (block/for-6.14/block) block: limit disk max
> sectors to (LLONG_MAX >> 9)
> 
> Changes to v1:
> - Generic block layer atomic writes enable flag and dm-table rework
> - Add dm-stripe and dm-raid1 support
> - Add bio_trim() patch

This all looks good.

Mikulas, we need Jens to pick up patches 1 and 2.  I wouldn't be
opposed to him taking the entire set but I did notice the DM core
(ioctl) version and the 3 DM targets that have had atomic support
added need their version numbers bumped.  Given that, likely best for
you (Mikulas) to pick up patches 3-8 after rebasing on Jens' latest
for-6.14/block branch (once Jens picks up patches 1 and 2).

Jens, you cool with picking up patches 1+2 for 6.14?  Or too late and
we circle back to this for 6.15?

Either way, for the series:

Reviewed-by: Mike Snitzer <snitzer@kernel.org>

On Fri, 17 Jan 2025, Mike Snitzer wrote:

> On Thu, Jan 16, 2025 at 05:02:53PM +0000, John Garry wrote:
> > This series introduces initial device mapper atomic write support.
> > 
> > Since we already support stacking atomic writes limits, it's quite
> > straightforward to support.
> > 
> > Personalities dm-linear, dm-stripe, and dm-raid1 are supported here, and
> > more personalities could be supported in future.
> > 
> > This is still an RFC as I would like to test further.
> > 
> > Based on 3d9a9e9a77c5 (block/for-6.14/block) block: limit disk max
> > sectors to (LLONG_MAX >> 9)
> > 
> > Changes to v1:
> > - Generic block layer atomic writes enable flag and dm-table rework
> > - Add dm-stripe and dm-raid1 support
> > - Add bio_trim() patch
> 
> This all looks good.
> 
> Mikulas, we need Jens to pick up patches 1 and 2.  I wouldn't be
> opposed to him taking the entire set but I did notice the DM core
> (ioctl) version and the 3 DM targets that have had atomic support
> added need their version numbers bumped.  Given that, likely best for
> you (Mikulas) to pick up patches 3-8 after rebasing on Jens' latest
> for-6.14/block branch (once Jens picks up patches 1 and 2).
> 
> Jens, you cool with picking up patches 1+2 for 6.14?  Or too late and
> we circle back to this for 6.15?
> 
> Either way, for the series:
> 
> Reviewed-by: Mike Snitzer <snitzer@kernel.org>

Hi

I rebased on Jens' block tree, applied the patches 3-8, increased 
DM_VERSION_MINOR, DM_VERSION_EXTRA, increased version numbers in 
dm-linear, dm-stripe, dm-raid1 and uploaded it to git.kernel.org.

You can check it if it's correct.

Mikulas

On Fri, Jan 17, 2025 at 10:27:22PM +0100, Mikulas Patocka wrote:
> 
> 
> On Fri, 17 Jan 2025, Mike Snitzer wrote:
> 
> > On Thu, Jan 16, 2025 at 05:02:53PM +0000, John Garry wrote:
> > > This series introduces initial device mapper atomic write support.
> > > 
> > > Since we already support stacking atomic writes limits, it's quite
> > > straightforward to support.
> > > 
> > > Personalities dm-linear, dm-stripe, and dm-raid1 are supported here, and
> > > more personalities could be supported in future.
> > > 
> > > This is still an RFC as I would like to test further.
> > > 
> > > Based on 3d9a9e9a77c5 (block/for-6.14/block) block: limit disk max
> > > sectors to (LLONG_MAX >> 9)
> > > 
> > > Changes to v1:
> > > - Generic block layer atomic writes enable flag and dm-table rework
> > > - Add dm-stripe and dm-raid1 support
> > > - Add bio_trim() patch
> > 
> > This all looks good.
> > 
> > Mikulas, we need Jens to pick up patches 1 and 2.  I wouldn't be
> > opposed to him taking the entire set but I did notice the DM core
> > (ioctl) version and the 3 DM targets that have had atomic support
> > added need their version numbers bumped.  Given that, likely best for
> > you (Mikulas) to pick up patches 3-8 after rebasing on Jens' latest
> > for-6.14/block branch (once Jens picks up patches 1 and 2).
> > 
> > Jens, you cool with picking up patches 1+2 for 6.14?  Or too late and
> > we circle back to this for 6.15?
> > 
> > Either way, for the series:
> > 
> > Reviewed-by: Mike Snitzer <snitzer@kernel.org>
> 
> Hi
> 
> I rebased on Jens' block tree, applied the patches 3-8, increased 
> DM_VERSION_MINOR, DM_VERSION_EXTRA, increased version numbers in 
> dm-linear, dm-stripe, dm-raid1 and uploaded it to git.kernel.org.
> 
> You can check it if it's correct.

Looks good, thanks!

Mike

On 1/17/25 1:09 PM, Mike Snitzer wrote:
> On Thu, Jan 16, 2025 at 05:02:53PM +0000, John Garry wrote:
>> This series introduces initial device mapper atomic write support.
>>
>> Since we already support stacking atomic writes limits, it's quite
>> straightforward to support.
>>
>> Personalities dm-linear, dm-stripe, and dm-raid1 are supported here, and
>> more personalities could be supported in future.
>>
>> This is still an RFC as I would like to test further.
>>
>> Based on 3d9a9e9a77c5 (block/for-6.14/block) block: limit disk max
>> sectors to (LLONG_MAX >> 9)
>>
>> Changes to v1:
>> - Generic block layer atomic writes enable flag and dm-table rework
>> - Add dm-stripe and dm-raid1 support
>> - Add bio_trim() patch
> 
> This all looks good.
> 
> Mikulas, we need Jens to pick up patches 1 and 2.  I wouldn't be
> opposed to him taking the entire set but I did notice the DM core
> (ioctl) version and the 3 DM targets that have had atomic support
> added need their version numbers bumped.  Given that, likely best for
> you (Mikulas) to pick up patches 3-8 after rebasing on Jens' latest
> for-6.14/block branch (once Jens picks up patches 1 and 2).
> 
> Jens, you cool with picking up patches 1+2 for 6.14?  Or too late and
> we circle back to this for 6.15?

I can do 1+2 for 6.14, they are pretty trivial.

-- 
Jens Axboe

On Thu, 16 Jan 2025 17:02:53 +0000, John Garry wrote:
> This series introduces initial device mapper atomic write support.
> 
> Since we already support stacking atomic writes limits, it's quite
> straightforward to support.
> 
> Personalities dm-linear, dm-stripe, and dm-raid1 are supported here, and
> more personalities could be supported in future.
> 
> [...]

Applied, thanks!

[1/8] block: Add common atomic writes enable flag
      commit: 6a7e17b22062c84a111d7073c67cc677c4190f32
[2/8] block: Don't trim an atomic write
      commit: 554b22864cc79e28cd65e3a6e1d0d1dfa8581c68

Best regards,
-- 
Jens Axboe