While working on a patchset for TP4129 (kato corrections and
clarifications) I run into a bunch small issues:
- nvme-tcp was spamming the kernel messages, thus I think it makes sense
to rate limit those messages. I observed this with my changes, so it
might not happen that often in mainline right now but still I think it
makes sense to rate limit them.
- nvme-fc should use the nvme state accossor to ensure it always sees
the current state.
- blk_mq_tagset_wait_completed_request was hanging on ctrl deletion path
and after a bit of head scratching I figured
blk_mq_tagset_wait_completed_request does not do what it claims. After
this fix, the shutdown path works fine and I think this could go in
without my other pending stuff I am working on. As the only user of this
function is the nvme subsytem, I included in this mini series.
Signed-off-by: Daniel Wagner <wagi@xxxxxxxxxx>
---
Daniel Wagner (3):
nvme-tcp: rate limit error message in send path
nvme-fc: use ctrl state getter
blk-mq: fix wait condition for tagset wait completed check
block/blk-mq-tag.c | 6 +++---
drivers/nvme/host/fc.c | 9 ++++++---
drivers/nvme/host/tcp.c | 4 ++--
3 files changed, 11 insertions(+), 8 deletions(-)
---
base-commit: fd6102e646a888931ad6ab21843c6ee4355e7525
change-id: 20250128-nvme-misc-fixes-07e8dad616cd
Best regards,
--
Daniel Wagner <wagi@xxxxxxxxxx>
