On Mon 29-07-19 16:10:37, Minchan Kim wrote:
> In our testing(carmera recording), Miguel and Wei found unmap_page_range
> takes above 6ms with preemption disabled easily. When I see that, the
> reason is it holds page table spinlock during entire 512 page operation
> in a PMD. 6.2ms is never trivial for user experince if RT task couldn't
> run in the time because it could make frame drop or glitch audio problem.
Where is the time spent during the tear down? 512 pages doesn't sound
like a lot to tear down. Is it the TLB flushing?
> This patch adds preemption point like coyp_pte_range.
>
> Reported-by: Miguel de Dios <miguelded...@google.com>
> Reported-by: Wei Wang <w...@google.com>
> Cc: Michal Hocko <mho...@kernel.org>
> Cc: Johannes Weiner <han...@cmpxchg.org>
> Cc: Mel Gorman <mgor...@techsingularity.net>
> Signed-off-by: Minchan Kim <minc...@kernel.org>
> ---
> mm/memory.c | 19 ++++++++++++++++---
> 1 file changed, 16 insertions(+), 3 deletions(-)
>
> diff --git a/mm/memory.c b/mm/memory.c
> index 2e796372927fd..bc3e0c5e4f89b 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -1007,6 +1007,7 @@ static unsigned long zap_pte_range(struct mmu_gather
> *tlb,
> struct zap_details *details)
> {
> struct mm_struct *mm = tlb->mm;
> + int progress = 0;
> int force_flush = 0;
> int rss[NR_MM_COUNTERS];
> spinlock_t *ptl;
> @@ -1022,7 +1023,16 @@ static unsigned long zap_pte_range(struct mmu_gather
> *tlb,
> flush_tlb_batched_pending(mm);
> arch_enter_lazy_mmu_mode();
> do {
> - pte_t ptent = *pte;
> + pte_t ptent;
> +
> + if (progress >= 32) {
> + progress = 0;
> + if (need_resched())
> + break;
> + }
> + progress += 8;
Why 8?
> +
> + ptent = *pte;
> if (pte_none(ptent))
> continue;
>
> @@ -1123,8 +1133,11 @@ static unsigned long zap_pte_range(struct mmu_gather
> *tlb,
> if (force_flush) {
> force_flush = 0;
> tlb_flush_mmu(tlb);
> - if (addr != end)
> - goto again;
> + }
> +
> + if (addr != end) {
> + progress = 0;
> + goto again;
> }
>
> return addr;
> --
> 2.22.0.709.g102302147b-goog
--
Michal Hocko
SUSE Labs
