lkml.org 
[lkml]   [2018]   [Jan]   [25]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectRe: [RFC PATCH 1/2] x86/ibpb: Skip IBPB when we switch back to same user process
On Thu, Jan 25, 2018 at 12:46 PM, Peter Zijlstra <peterz@infradead.org> wrote:
> On Thu, Jan 25, 2018 at 11:32:46AM -0800, Tim Chen wrote:
>>
>> This patch is not ideal as it comes with the caveats that
>> patch 2 tries to close. I put it out here to see if it can prompt
>> people to come up with a better solution. Keeping active_mm around would
>> have been cleaner but it looks like there are issues that Andy mentioned.
>>
>> The "A -> idle -> A" case would not trigger IBPB if tlb_defer_switch_to_init_mm()
>> is true (non pcid) as we does not change the mm.
>>
>> This patch tries to address the case when we do switch to init_mm and back.
>> Do you still have objections to the approach in this patch
>> to save the last active mm before switching to init_mm?
>
> I still think the existing active_mm is sufficient. Something like:
>
> switch_mm()
> {
> ...
> if (prev && next != prev)
> ibpb();
> ...
> }
>
> should work. Because while the idle crud does leave_mm() and PCID does
> enter_lazy_tlb() and both end up doing: switch_mm(NULL, &init_mm, NULL),
> nothing there affects tsk->active_mm.
>
> So over the "A -> idle -> A" transition, active_mm should actually track
> what you want.
>
>

Can we please not rely on any of the active_mm shit? That thing has
really weird semantics and should just die.

That being said, just stashing last_user_mm without any refcounting
should be fine. After all, the only thing anyone does with it is
comparing to next, and next is always alive. Or we could use
last_user_ctx_id, since we already have a never-reused ctx_id for each
mm on x86.

--Andy

\
 
 \ /
  Last update: 2018-01-25 22:05    [W:0.151 / U:0.160 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site