Re: [PATCH] scsi: pmcraid: shut down command timers before reuse

[Runyu Xiao]

On Fri, Aug 21, 2026 at 02:23:51PM +0800, Greg Kroah-Hartman wrote:
> How was this found and tested?

The issue was found by a static-analysis check and then confirmed by
tracing the command ownership and timer lifetime in pmcraid.c. The v1
patch was generated by an LLM from that finding.

I did not reproduce this with pmcraid hardware. I only ran
git apply --check, checkpatch.pl --strict, and built
drivers/scsi/pmcraid.o against current mainline. I should not have
presented the result as sufficiently tested.

> Why the extra whitespace? Didn't checkpatch complain about this?

I rechecked the exact patch. checkpatch reported 0 errors, 0 warnings,
and 0 checks, but there is an unnecessary blank line in the generated
patch that I missed. I will remove it.

> Why the prototype right before the definition? Did a LLM create this
> patch?

Yes. The initial implementation was generated by an LLM. The prototype
immediately before pmcraid_complete_reset_cmd() is unnecessary and will
be removed.

I will continue reworking the completion path, reset state machine, list
ownership, and teardown synchronization. The current v1 should not be
reused or resent.

Thanks,
Runyu Xiao