Message ID | 1358770172-30125-1-git-send-email-fabio.baltieri@linaro.org (mailing list archive) |
---|---|
State | New, archived |
Headers | show |
On Mon, Jan 21, 2013 at 01:09:31PM +0100, Fabio Baltieri wrote: > From: Rabin Vincent <rabin.vincent@stericsson.com> > > Modify clksrc_dbx500_prcmu_read to replace readl() with readl_relaxed(). > This speeds up calls to the function by about 40%. Hello John, these two patches may go in the existing ux500 clksrc branch together with the nomadik-mtu's patches queued up some time ago. Can I have your Acked-by for that? Thanks, Fabio
On Mon, Jan 21, 2013 at 1:22 PM, Fabio Baltieri <fabio.baltieri@linaro.org> wrote: > On Mon, Jan 21, 2013 at 01:09:31PM +0100, Fabio Baltieri wrote: >> From: Rabin Vincent <rabin.vincent@stericsson.com> >> >> Modify clksrc_dbx500_prcmu_read to replace readl() with readl_relaxed(). >> This speeds up calls to the function by about 40%. > > Hello John, > > these two patches may go in the existing ux500 clksrc branch > together with the nomadik-mtu's patches queued up some time ago. > > Can I have your Acked-by for that? Actually now when I look at it these changes are fully orthogonal to the other stuff, so these two should just go in through John's timer tree as usual. John? Yours, Linus Walleij
On Wed, Jan 23, 2013 at 11:16:19AM +0100, Linus Walleij wrote: > On Mon, Jan 21, 2013 at 1:22 PM, Fabio Baltieri > <fabio.baltieri@linaro.org> wrote: > > On Mon, Jan 21, 2013 at 01:09:31PM +0100, Fabio Baltieri wrote: > >> From: Rabin Vincent <rabin.vincent@stericsson.com> > >> > >> Modify clksrc_dbx500_prcmu_read to replace readl() with readl_relaxed(). > >> This speeds up calls to the function by about 40%. > > > > Hello John, > > > > these two patches may go in the existing ux500 clksrc branch > > together with the nomadik-mtu's patches queued up some time ago. > > > > Can I have your Acked-by for that? > > Actually now when I look at it these changes are fully orthogonal > to the other stuff, so these two should just go in through John's > timer tree as usual. > > John? Hi John, any feedback from this two patches? Thanks, Fabio
diff --git a/drivers/clocksource/clksrc-dbx500-prcmu.c b/drivers/clocksource/clksrc-dbx500-prcmu.c index c26c369..159f54a 100644 --- a/drivers/clocksource/clksrc-dbx500-prcmu.c +++ b/drivers/clocksource/clksrc-dbx500-prcmu.c @@ -33,15 +33,14 @@ static void __iomem *clksrc_dbx500_timer_base; -static cycle_t clksrc_dbx500_prcmu_read(struct clocksource *cs) +static cycle_t notrace clksrc_dbx500_prcmu_read(struct clocksource *cs) { + void __iomem *base = clksrc_dbx500_timer_base; u32 count, count2; do { - count = readl(clksrc_dbx500_timer_base + - PRCMU_TIMER_DOWNCOUNT); - count2 = readl(clksrc_dbx500_timer_base + - PRCMU_TIMER_DOWNCOUNT); + count = readl_relaxed(base + PRCMU_TIMER_DOWNCOUNT); + count2 = readl_relaxed(base + PRCMU_TIMER_DOWNCOUNT); } while (count2 != count); /* Negate because the timer is a decrementing counter */