Re: [PATCH] sched,x86: optimize switch_mm for multi-threaded workloads

[Rik van Riel <riel@redhat.com>]

On 07/31/2013 08:41 PM, Linus Torvalds wrote:
> On Wed, Jul 31, 2013 at 4:14 PM, Paul Turner <pjt@google.com> wrote:
>> We attached the following explanatory comment to our version of the patch:
>>
>> /*
>> * In the common case (two user threads sharing mm
>> * switching) the bit will be set; avoid doing a write
>> * (via atomic test & set) unless we have to.  This is
>> * safe, because no other CPU ever writes to our bit
>> * in the mask, and interrupts are off (so we can't
>> * take a TLB IPI here.)  If we don't do this, then
>> * switching threads will pingpong the cpumask
>> * cacheline.
>> */
>
> So as mentioned, the "interrupts will be off" is actually dubious.
> It's true for the context switch case, but not for the activate_mm().
>
> However, as Rik points out, activate_mm() is different in that we
> shouldn't have any preexisting MMU state anyway. And besides, that
> should never trigger the "prev == next" case.
>
> But it does look a bit messy, and even your comment is a bit
> misleading (it might make somebody think that all of switch_mm() is
> protected from interrupts)
> .
> Anyway, I'm perfectly ok with the patch itself, but I just wanted to
> make sure people had thought about these things.

Would you like me to document the things we found in the comment,
and resend a patch, or is the patch good as-is?

-- 
All rights reversed