Re: [PATCH 1/6] Free up pf flag PF_KSOFTIRQD

[Venkatesh Pallipadi]

On Thu, Oct 21, 2010 at 7:58 AM, Eric Dumazet <eric.dumazet@xxxxxxxxx> wrote:
> Le jeudi 21 octobre 2010 à 07:36 -0700, Venkatesh Pallipadi a écrit :
>
>> Yes. I thought about static inline part. The reason I did not make
>> this static inline was because ksoftirqd was declared static in
>> softirq.c and this function was getting called from sched.c.
>>
>
> I believe you can remove the 'static' for this kind of thing.
>
>> I did not know that this_cpu_read existed though. I guess I should be
>> looking at using that elsewhere in the patchset too.
>>
>
> Sure :)
>
>> Also, part of the overhead you see below I think is coming from
>> DEBUG_PREEMPT. That would be making every smp_processor_id() call more
>> expensive. No?
>>
>
> Right, but the point is the this_cpu_read() version doesnt have this
> overhead, even if DEBUG_PREEMPT is on, at least on x86.
>

Agreed. Will redo this when I refresh the patchset.

Thanks,
Venki

> BTW, I lied somehow, because the way this_cpu_read() is handled,
> following code :
>
>        p == this_cpu_read(ksoftirqd);
>
> generates :
>
> mov     %gs:offset,%rax
> cmp     %rdi,%rax
>
> not :
>
> cmp     %gs:offset,%rdi
>
>
>
>
--
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/