Skip to content

scsi_debug: improve failure rates

Luis Chamberlain fstests lkml 1 version In review

Get series with b4
b4 am [email protected]

Versions newest first

  1. v1 Latest4 patches 2 repliesIn review

Patches in v1 No applies recorded yet

#Subject and filesSizeRepliesApplied
1/4common/config: disable udevadm settle if CONFIG_NET is disabledcommon/config+8 -10 replies
2/4common/scsi_debug: use udevadm settle instead of sleepingcommon/scsi_debug+2 -20 replies
3/4common/module: add a patient module rmmodcommon/module+48 -02 replies
4/4common/scsi_debug: use the patient module removercommon/scsi_debug+4 -20 replies

Reviews trailers per patch

No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.

Read the v1 thread

Cover letter [PATCH 0/4] scsi_debug: improve failure rates

When using scsi_debug to create virtual devices we can often run into failures which are really just false positives, and the failure was the inability to remove the module. Addressing this is not easy. While we can learn from what blktests folks do and use udevadm settle, that's not sufficient by any means. This addresses the other pieces needed.

More work is also needed on the scsi_debug driver front.

Makes me wonder if a patient kmod removal option is then desirable upstream on kmod (just the one which waits to refcnt 0), given this sort of test case. Lucas?

Luis Chamberlain (4):
common/config: disable udevadm settle if CONFIG_NET is disabled common/scsi_debug: use udevadm settle instead of sleeping common/module: add a patient module rmmod common/scsi_debug: use the patient module remover

 common/config     |  9 ++++++++-
 common/module     | 48 +++++++++++++++++++++++++++++++++++++++++++++++
 common/scsi_debug | 10 ++++++----
 3 files changed, 62 insertions(+), 5 deletions(-)
Signature · 1 line
2.29.2