[v2,00/12] Submit zoned writes in order

Message

Bart Van Assche April 7, 2023, 11:58 p.m. UTC

Hi Jens,

Tests with a zoned UFS prototype have shown that there are plenty of
opportunities for reordering in the block layer for zoned writes (REQ_OP_WRITE).
The UFS driver is more likely to trigger reordering than other SCSI drivers
because it reports BLK_STS_DEV_RESOURCE more often, e.g. during clock scaling.
This patch series makes sure that zoned writes are submitted in order without
affecting other workloads significantly.

Please consider this patch series for the next merge window.

Thanks,

Bart.

Changes compared to v1:
- Fixed two issues detected by the kernel test robot.

Bart Van Assche (12):
  block: Send zoned writes to the I/O scheduler
  block: Send flush requests to the I/O scheduler
  block: Send requeued requests to the I/O scheduler
  block: Requeue requests if a CPU is unplugged
  block: One requeue list per hctx
  block: Preserve the order of requeued requests
  block: Make it easier to debug zoned write reordering
  block: mq-deadline: Simplify deadline_skip_seq_writes()
  block: mq-deadline: Disable head insertion for zoned writes
  block: mq-deadline: Introduce a local variable
  block: mq-deadline: Fix a race condition related to zoned writes
  block: mq-deadline: Handle requeued requests correctly

 block/blk-flush.c      |   3 +-
 block/blk-mq-debugfs.c |  66 +++++++++++------------
 block/blk-mq.c         | 115 +++++++++++++++++++++++------------------
 block/blk.h            |  19 +++++++
 block/mq-deadline.c    |  65 ++++++++++++++++++-----
 include/linux/blk-mq.h |  40 +++++++-------
 include/linux/blkdev.h |   4 --
 7 files changed, 194 insertions(+), 118 deletions(-)