Re: [PATCH 10/23] sched,mmc: Convert to sched_set_fifo*()

[Ulf Hansson]

On Thu, 23 Apr 2020 at 10:59, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
>
> On Wed, Apr 22, 2020 at 06:59:35PM +0200, Ulf Hansson wrote:
> > On Wed, 22 Apr 2020 at 13:29, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:
> > >
> > > Because SCHED_FIFO is a broken scheduler model (see previous patches)
> > > take away the priority field, the kernel can't possibly make an
> > > informed decision.
> > >
> > > In this case, use fifo_low, because it only cares about being above
> > > SCHED_NORMAL. Effectively no change in behaviour.
> > >
> > > Cc: ulf.hansson@xxxxxxxxxx
> > > Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
> > > Reviewed-by: Ingo Molnar <mingo@xxxxxxxxxx>
> >
> > Acked-by: Ulf Hansson <ulf.hansson@xxxxxxxxxx>
>
> Thanks!
>
> > FYI: I am slowly moving towards removing the entire kthread for the
> > sdio_irq_thread(). It shouldn't be too far off to be posted, one or
> > two kernel releases or so.
>
> Moving over to regular threaded interrupts? Anyway, cool, if these
> series collide it's easy enough to drop this patch on the floor if it
> turns out obsolete.

In principle, the only reason for the kthread is that we need it for
polling - for hosts that don't support SDIO irqs. So, I am thinking of
replacing the kthread with a workqueue, as we already have one for
hosts that are using non-threaded IRQs.

Kind regards
Uffe