Skip to content

writeback: Fix bandwidth estimates

Jan Kara linux-mm 4 versions – Applied

Status from this message: [merged] writeback-use-read_once-for-unlocked-reads-of-writeback-stats.patch removed from -mm tree

Get series with b4
b4 am [email protected]

Versions newest first · linked by author and title

  1. v4 Latest5 patches 0 repliesApplied 5/5
  2. v35 patches 1 replySuperseded
  3. v25 patches 1 replySuperseded
Earlier versions · 1
VersionPostedPatchesRepliesStatus
v15 patches8 repliesSuperseded

Changes in v4 from the cover letter

* Now really send updated patches and not previous version :-|

Patches in v4 All 5 applied

#Subject and filesSizeRepliesApplied
1/5writeback: Track number of inodes under writebackfs/fs-writeback.c include/linux/backing-dev-defs.h mm/backing-dev.c +1 more+27 -20 repliesApplied
2/5writeback: Reliably update bandwidth estimationfs/fs-writeback.c include/linux/backing-dev.h include/linux/writeback.h +1 more+46 -160 repliesApplied
3/5writeback: Fix bandwidth estimate for spiky workloadinclude/linux/backing-dev-defs.h include/linux/writeback.h mm/backing-dev.c +1 more+31 -160 repliesApplied
4/5writeback: Rename domain_update_bandwidth()mm/page-writeback.c+4 -40 repliesApplied
5/5writeback: Use READ_ONCE for unlocked reads of writeback statsmm/page-writeback.c+15 -130 repliesApplied

Reviews trailers per patch

None of the 5 patches has a Reviewed-by or Acked-by.

PatchMichael Stapelberg
1/5
2/5
3/5Reported-and-tested-by
4/5
5/5

Trailers given in replies are filled; those carried in the patches from earlier versions are outlined with dashes.

Read the v4 thread

Cover letter [PATCH 0/5 v4] writeback: Fix bandwidth estimates

Hello,

Here is the fourth revision of the patch series to fix estimate of writeback throughput when device is not fully busy doing writeback. Michael Stapelberg has reported that such workload (e.g. generated by linking) tends to push estimated throughput down to 0 and as a result writeback on the device is practically stalled.

The first three patches fix the reported issue, the remaining two patches are unrelated cleanups of problems I've noticed when reading the code.

Andrew, can you please pick up the series? Thanks!

Honza

Changes since v3:
* Now really send updated patches and not previous version :-|

Changes since v2:
* Rebased on top of 5.14-rc1
* Fixed compilation failure introduced by patch 5/5

Changes since v1:
* Added comments to better explain the logic
* Added Tested-by tag
* Added WRITE_ONCE calls matching READ_ONCE in patch 5/5

Previous versions: