lkml.org 
[lkml]   [2020]   [Feb]   [11]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 2/2] arm64: tlb: skip tlbi broadcast for single threaded TLB flushes
Hi Andrea,

On Mon, Feb 10, 2020 at 03:14:11PM -0500, Andrea Arcangeli wrote:
> On Mon, Feb 10, 2020 at 05:51:06PM +0000, Catalin Marinas wrote:
> > It may be better if you used mm_cpumask to mark wherever an mm ever ran
> > than relying on mm_users.
>
> Agreed.
>
> If we can use mm_cpumask to track where the mm ever run, then if I'm
> not mistaken we could optimize also multithreaded processes in the
> same way: if only one thread is running frequently and the others are
> frequently sleeping, we could issue a single tlbi broadcast (modulo
> invalidates of small virtual ranges).

Possibly, though not sure how you'd detect such scenario.

> In the meantime the below should be enough to address the concern you
> raised of the proof of concept RFC patch.
>
> I already experimented with mm_users == 1 earlier and it doesn't
> change the benchmark results for the "best case" below.
>
> (untested)
>
> diff --git a/arch/arm64/include/asm/tlbflush.h b/arch/arm64/include/asm/tlbflush.h
> index 772bbc45b867..a2d53b301f22 100644
> --- a/arch/arm64/include/asm/tlbflush.h
> +++ b/arch/arm64/include/asm/tlbflush.h
[...]
> @@ -212,7 +215,8 @@ static inline void flush_tlb_page(struct vm_area_struct *vma,
> unsigned long addr = __TLBI_VADDR(uaddr, ASID(mm));
>
> /* avoid TLB-i broadcast to remote NUMA nodes if it's a local flush */
> - if (current->mm == mm && atomic_read(&mm->mm_users) <= 1) {
> + if (current->mm == mm && atomic_read(&mm->mm_users) <= 1 &&
> + (system_uses_ttbr0_pan() || atomic_read(&mm->mm_count) == 1)) {
> int cpu = get_cpu();
>
> cpumask_setall(mm_cpumask(mm));

I think there is another race here. IIUC, the assumption you make is
that when mm_users <= 1 && mm_count == 1, the only active user of this
pgd/ASID is on the CPU doing the TLBI. This is not the case for
try_to_unmap() where the above condition may be true but the active
thread on a different CPU won't notice the local TLBI.

> > That's a pretty artificial test and it is indeed improved by this patch.
> > However, it would be nice to have some real-world scenarios where this
> > matters.
[...]
> Still your question if it'll make a difference in practice is a good
> one and I don't have a sure answer yet. I suppose before doing more
> benchmarking it's better to make a new version of this that uses
> mm_cpumask to track where the asid was ever loaded as you suggested,
> so that it will also optimize away tlbi broadcasts from multithreaded
> processes where only one thread is running frequently?

I was actually curious what triggered this patch series, whether you've
seen a real use-case where the TLBI was a bottleneck.

--
Catalin

\
 
 \ /
  Last update: 2020-02-11 15:00    [W:0.105 / U:3.088 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site