Re: [RFC] perf: a different approach to perf_rotate_context()

[Alexey Budankov]

On 03.03.2018 20:48, Peter Zijlstra wrote:
> On Sat, Mar 03, 2018 at 04:43:16PM +0000, Song Liu wrote:
>>> In any case, there's a ton of conflict against the patches here:
>>>
>>> https://git.kernel.org/pub/scm/linux/kernel/git/peterz/queue.git/log/?h=perf/testing
>>>
>>> And with those the idea was to move to a virtual time based scheduler
>>> (basically schedule those flexible events that have the biggest lag --
>>> that also solves 1).
>>
>> Thanks for these information. I will study this approach. Maybe that is 
>> our path to PMU sharing. 
> 
> So I'm really not convinced on that whole PMU sharing.
> 
>> What's is the status of this work? Would it 
>> land in 4.17? 
> 
> These patches might make 4.17, they got held up because of the whole
> meltdown/spectre crap and I need to get back to them.
> 

That work is long desired and would bring performance boost, specifically on 
server systems in per-process profiling mode, accompanied by good speedup on 
context switches. Undoubtedly meltdown/spectre related activity 
substituted it at some point but that improvements would still bring 
significant value and is still awaited.

BR,
Alexey