On 12/12/20 7:42 AM, Mike Snitzer wrote:

On Sat, Dec 12 2020 at  4:12am -0500,
Song Liu [off-list ref] wrote:

quoted

Hi Jens, 

Please consider pulling the following changes on top of tag
block-5.10-2020-12-05. This is to fix raid10 data corruption [1] in 5.10-rc7. 

Tests run on this change: 

1. md raid10: tested discard on raid10 device works properly (zero mismatch_cnt).
2. dm raid10: tested discard_granularity and discard_max_bytes was set properly. 

Thanks,
Song


[1] https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1907262/


The following changes since commit 7e7986f9d3ba69a7375a41080a1f8c8012cb0923:

  block: use gcd() to fix chunk_sectors limit stacking (2020-12-01 11:02:55 -0700)

are available in the Git repository at:

  ssh://git@gitolite.kernel.org/pub/scm/linux/kernel/git/song/md.git md-fixes

for you to fetch changes up to 0d5c7b890229f8a9bb4b588b34ffe70c62691143:

  Revert "md: add md_submit_discard_bio() for submitting discard bio" (2020-12-12 00:51:41 -0800)

----------------------------------------------------------------
Song Liu (7):
      Revert "dm raid: remove unnecessary discard limits for raid10"
      Revert "dm raid: fix discard limits for raid1 and raid10"
      Revert "md/raid10: improve discard request for far layout"
      Revert "md/raid10: improve raid10 discard request"
      Revert "md/raid10: pull codes that wait for blocked dev into one function"
      Revert "md/raid10: extend r10bio devs to raid disks"
      Revert "md: add md_submit_discard_bio() for submitting discard bio"

 drivers/md/dm-raid.c |   9 +++++
 drivers/md/md.c      |  20 ----------
 drivers/md/md.h      |   2 -
 drivers/md/raid0.c   |  14 ++++++-
 drivers/md/raid10.c  | 423 +++++++++++++++++++++++++++++------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
 drivers/md/raid10.h  |   1 -
 6 files changed, 78 insertions(+), 391 deletions(-)

Jens already pulled 95% of these changes no? why are you sending a pull
that ignores that fact?

And as I stated here:
https://www.redhat.com/archives/dm-devel/2020-December/msg00253.html

I don't agree with reverting commit e0910c8e4f87bb9 ("dm raid: fix
discard limits for raid1 and raid10").  Espeiclaly not like you've done,
given it was marked for stable any follow-on fix/revert needs to be as
well.

Simply changing 'struct mddev' chunk_sectors members from int to
'unsigned int' is the better way forward I think.

I agree, that'd be simpler. Mike, care to send a real patch for that
against block-5.10?

-- 
Jens Axboe