lkml.org 
[lkml]   [2013]   [Dec]   [12]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [PATCH 3/3] x86: mm: Account for the of CPUs that must be flushed during a TLB range flush
On 12/12/2013 07:55 PM, Mel Gorman wrote:
> X86 TLB range flushing uses a balance point to decide if a single global TLB
> flush or multiple single page flushes would perform best. This patch takes into
> account how many CPUs must be flushed.
>
> Signed-off-by: Mel Gorman <mgorman@suse.de>
> ---
> arch/x86/mm/tlb.c | 3 +++
> 1 file changed, 3 insertions(+)
>
> diff --git a/arch/x86/mm/tlb.c b/arch/x86/mm/tlb.c
> index 09b8cb8..0cababa 100644
> --- a/arch/x86/mm/tlb.c
> +++ b/arch/x86/mm/tlb.c
> @@ -217,6 +217,9 @@ void flush_tlb_mm_range(struct mm_struct *mm, unsigned long start,
> act_entries = mm->total_vm > act_entries ? act_entries : mm->total_vm;
> nr_base_pages = (end - start) >> PAGE_SHIFT;
>
> + /* Take the number of CPUs to range flush into account */
> + nr_base_pages *= cpumask_weight(mm_cpumask(mm));
> +

flush range calculation base on per cpu, since no matter how many cpus
in the process, tlb flush and refill time won't change.
> /* tlb_flushall_shift is on balance point, details in commit log */
> if (nr_base_pages > act_entries || has_large_page(mm, start, end)) {
> count_vm_event(NR_TLB_LOCAL_FLUSH_ALL);
>


--
Thanks
Alex


\
 
 \ /
  Last update: 2013-12-12 15:41    [W:0.074 / U:2.308 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site