[PATCH 3/3] nvme-core: mark internal passthru req REQ_QUIET

Chaitanya Kulkarni chaitanyak at nvidia.com
Mon Apr 11 03:48:14 PDT 2022


On 4/10/22 23:12, Christoph Hellwig wrote:
> On Sun, Apr 10, 2022 at 08:12:49PM -0700, Chaitanya Kulkarni wrote:
>> Mark internal passthru requests quiet in the submission path with
>> RQF_QUIET flag added in the __nvme_submit_sync_cmd(). In the completion
>> path, if nvme request is resulted in the error and request is marked
>> RQF_QUIET then don't log the error with nvme_error_log().
> 
> This misses out on a few other internal submissions like the PCIe
> queue deletion or keep alive.  Probably not much of an issue, but I'll
> ammend the commit log when applying it.  I'll wait a bit for more
> feedback but plan to send this out later this week.
> 

Yes it did miss that, actually I've a WIP series to centralize all 
passthru request submission with cleanup so that we have only one
interface for internal passthru sync-async and also cleanup the code
for __nvme_submit_sync_cmd(), will send that out soon.

This was just a quick fix to mainly fix reports we are getting.

-ck




More information about the Linux-nvme mailing list