Re: [PATCH] scsi: Don't wait for completion of in-flight requests

[Bart Van Assche]

On 11/26/24 4:21 AM, John Garry wrote:
> On 26/11/2024 11:50, Qiang Ma wrote:
> > diff --git a/drivers/scsi/scsi_lib.c b/drivers/scsi/scsi_lib.c
> > index adee6f60c966..0a2d5d9327fc 100644
> > --- a/drivers/scsi/scsi_lib.c
> > +++ b/drivers/scsi/scsi_lib.c
> > @@ -2065,7 +2065,7 @@ int scsi_mq_setup_tags(struct Scsi_Host *shost)
> >       tag_set->queue_depth = shost->can_queue;
> >       tag_set->cmd_size = cmd_size;
> >       tag_set->numa_node = dev_to_node(shost->dma_dev);
> > -    tag_set->flags = BLK_MQ_F_SHOULD_MERGE;
> > +    tag_set->flags = BLK_MQ_F_SHOULD_MERGE | BLK_MQ_F_STACKING;
>
> This should not be set for all SCSI hosts. Some SCSI hosts rely on 
> bf0beec0607d.

Are there any SCSI hosts for which it is safe to set this flag? To me
the above change looks like a hack that should not be merged at all. Did
I perhaps overlook something?

Thanks,

Bart.