On Wed, 4 Mar 2015, Deng-Cheng Zhu wrote:
> Use sb1250 hpt for sched_clock source. This implementation will give high
> resolution cputime accounting.
>
> Signed-off-by: Deng-Cheng Zhu <dengcheng.zhu@xxxxxxxxxx>
> ---
> arch/mips/kernel/csrc-sb1250.c | 8 ++++++++
> 1 file changed, 8 insertions(+)
>
> diff --git a/arch/mips/kernel/csrc-sb1250.c b/arch/mips/kernel/csrc-sb1250.c
> index df84836..6546fff 100644
> --- a/arch/mips/kernel/csrc-sb1250.c
> +++ b/arch/mips/kernel/csrc-sb1250.c
> @@ -12,6 +12,7 @@
> * GNU General Public License for more details.
> */
> #include <linux/clocksource.h>
> +#include <linux/sched_clock.h>
>
> #include <asm/addrspace.h>
> #include <asm/io.h>
> @@ -46,6 +47,11 @@ struct clocksource bcm1250_clocksource = {
> .flags = CLOCK_SOURCE_IS_CONTINUOUS,
> };
>
> +static u64 notrace sb1250_read_sched_clock(void)
> +{
> + return sb1250_hpt_read(NULL);
> +}
> +
I think you're abusing the API of `sb1250_hpt_read' here, by relying on
the implementation not using its `cs' argument.
I think it would make sense to reverse the implementation, by calling
`sb1250_read_sched_clock' from `sb1250_hpt_read' instead. Or perhaps
better yet use a static inline helper for both, so as to avoid the extra
tail call and the associated performance hit.
Maciej
