RE: [PATCH] scsi: ufs: mcq: Limit the amount of inflight requests

[Avri Altman <Avri.Altman@wdc.com>]

 
> On 3/30/23 06:11, Avri Altman wrote:
> > diff --git a/drivers/ufs/core/ufshcd.c b/drivers/ufs/core/ufshcd.c
> > index 35a3bd95c5e4..d529c42a682a 100644
> > --- a/drivers/ufs/core/ufshcd.c
> > +++ b/drivers/ufs/core/ufshcd.c
> > @@ -8468,6 +8468,10 @@ static int ufshcd_alloc_mcq(struct ufs_hba *hba)
> >       if (ret)
> >               goto err;
> >
> > +     if (WARN_ONCE(hba->nutrs * hba->nr_hw_queues > (1 << 16) - 1,
> > +                  "there can be at most 1<<16 inflight requests\n"))
> > +             goto err;
> > +
> >       /*
> >        * Previously allocated memory for nutrs may not be enough in MCQ
> mode.
> >        * Number of supported tags in MCQ mode may be larger than SDB
> mode.
> 
> Hi Avri,
> 
> WARN*() should only be used to report kernel bugs. hba->nutrs *
> hba->nr_hw_queues being too large is not a kernel bug but a
> configuration issue.
Will replace with an appropriate dev_info and a clarifying comment.

> 
> Instead of failing MCQ allocation, shouldn't
> ufshcd_mcq_decide_queue_depth() be modified such that it restricts
> hba->nutrs to a value that can be supported?
The driver's capacity is the product of  hba->nutrs * hba->nr_hw_queues,
Where the latter is being established only later in ufshcd_mcq_init.

Thanks,
Avri
> 
> Thanks,
> 
> Bart.