On Tue, May 14, 2019 at 02:01:34AM +0000, Nadav Amit wrote:

...

diff --git a/mm/mmu_gather.c b/mm/mmu_gather.c index 99740e1dd273..cc251422d307 100644 --- a/mm/mmu_gather.c +++ b/mm/mmu_gather.c @@ -251,8 +251,9 @@ void tlb_finish_mmu(struct mmu_gather *tlb, * forcefully if we detect parallel PTE batching threads. */ if (mm_tlb_flush_nested(tlb->mm)) {

•         tlb->fullmm = 1;
           __tlb_reset_range(tlb);
  

•         __tlb_adjust_range(tlb, start, end - start);
  

•         tlb->freed_tables = 1;
   }
  
   tlb_flush_mmu(tlb);
  

I think that this should have set need_flush_all and not fullmm.

Difficult, mmu_gather::need_flush_all is arch specific and not everybody implements it.

And while mmu_gather::fullmm isn't strictly correct either; we can (ab)use it here, because at tlb_finish_mmu() time the differences don't matter anymore.