Re: [PATCH v2] sdio: optimized SDIO IRQ handling for single irq

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Wed, 4 May 2011, Per Forlin wrote:

> 2011/5/4 Michał Mirosław <mirqus@xxxxxxxxx>:
> > 2011/5/4 Per Forlin <per.forlin@xxxxxxxxxx>:
> >> From: Stefan Nilsson XK <stefan.xk.nilsson@xxxxxxxxxxxxxx>
> >>
> >> If there is only 1 function registered it is possible to
> >> improve performance by directly calling the irq handler
> >> and avoiding the overhead of reading the CCCR registers.
> >>
> > [...]
> >> --- a/drivers/mmc/core/sdio_irq.c
> >> +++ b/drivers/mmc/core/sdio_irq.c
> >> @@ -32,6 +32,16 @@ static int process_sdio_pending_irqs(struct mmc_card *card)
> >>        int i, ret, count;
> >>        unsigned char pending;
> >>
> >> +       /*
> >> +        * Optimization, if there is only 1 function registered
> >> +        * call irq handler directly
> >> +        */
> >> +       if (card->sdio_single_irq && card->sdio_single_irq->irq_handler) {
> >> +               struct sdio_func *func = card->sdio_single_irq;
> >> +               func->irq_handler(func);
> >> +               return 1;
> >> +       }
> > [...]
> >
> > The second condition can be avoided:
> >
> > in process_sdio_pending_irqs():
> >
> > if (card->sdio_irq_func)
> >   call handler and return
> >
> I added the second condition as a sanity check. Same check is used in
> the main for loop
> >	ret = -EINVAL;
> >			} else if (func->irq_handler) {
> >				func->irq_handler(func);
> Is the second check necessary here?

Yes because we want to be notified if the hardware returns pending 
interrupt flags for interrupts we didn't claim.

> > in sdio_claim_irq():
> >
> >  card->func->irq_handler = ...
> >  if (host->sdio_irqs == 1)
> >    card->sdio_irq_func = func;
> >  else
> >    card->sdio_irq_func = NULL;
> I wanted to keep it simple and use same function in claim and release.
> Your code looks nice.
> Is if safe to not check the condition "(card->host->caps &
> MMC_CAP_SDIO_IRQ)". What happens if the SDIO is in polling mode?

You cannot avoid checking MMC_CAP_SDIO_IRQ.  If it isn't set the CCCr 
register must be polled in all cases.


Nicolas

[Index of Archives]     [Linux USB Devel]     [Linux Media]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux