Re: [PATCH] drivers/mmc/mmc.c: Replacing yield() with a better alternative

[Nick Piggin]

Arjan van de Ven wrote:
> On Wed, 2006-10-11 at 18:15 +0530, Amol Lad wrote:
>
> > In 2.6, the semantics of calling yield() changed from "sleep for a
> > bit" to "I really don't want to run for a while".  This matches POSIX
> > better, but there's a lot of drivers still using yield() when they mean
> > cond_resched(), schedule() or even schedule_timeout().
> >
> > For this driver cond_resched() seems to be a better
> > alternative
>
>
> are you sure?
>
>
> > Tested compile only
> >
> > Signed-off-by: Amol Lad <amol@xxxxxxxxxxxxxxxxxxx>
> > ---
> > diff -uprN -X linux-2.6.19-rc1-orig/Documentation/dontdiff linux-2.6.19-rc1-orig/drivers/mmc/mmc.c linux-2.6.19-rc1/drivers/mmc/mmc.c
> > --- linux-2.6.19-rc1-orig/drivers/mmc/mmc.c	2006-10-05 14:00:46.000000000 +0530
> > +++ linux-2.6.19-rc1/drivers/mmc/mmc.c	2006-10-11 17:57:02.000000000 +0530
> > @@ -454,7 +454,7 @@ static void mmc_deselect_cards(struct mm
> > static inline void mmc_delay(unsigned int ms)
> > {
> > 	if (ms < HZ / 1000) {
> > -		yield();
> > +		cond_resched();
> > 		mdelay(ms);
>
>
>
> this probably wants msleep(), especially with hrtimers comming up; there
> the sleeps are always exact...

The condition looks broken too. It should be
if (ms < 1000 / HZ) {...}

Shouldn't it?
--
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com 
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/