On Tue, Jan 21, 2025 at 08:09:29PM +0000, Roman Gushchin wrote:
> Commit b67fbebd4cf9 ("mmu_gather: Force tlb-flush VM_PFNMAP vmas")
> added a forced tlbflush to tlb_vma_end(), which is required to avoid a
> race between munmap() and unmap_mapping_range(). However it added some
> overhead to other paths where tlb_vma_end() is used, but vmas are not
> removed, e.g. madvise(MADV_DONTNEED).
>
> Fix this by moving the tlb flush out of tlb_end_vma() into
> free_pgtables(), somewhat similar to the stable version of the
> original commit: e.g. stable commit 895428ee124a ("mm: Force TLB flush
> for PFNMAP mappings before unlink_file_vma()").
>
> Note, that if tlb->fullmm is set, no flush is required, as the whole
> mm is about to be destroyed.
>
> Suggested-by: Jann Horn <jannh@xxxxxxxxxx>
> Signed-off-by: Roman Gushchin <roman.gushchin@xxxxxxxxx>
> diff --git a/mm/memory.c b/mm/memory.c
> index 398c031be9ba..2071415f68dd 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -365,6 +365,13 @@ void free_pgtables(struct mmu_gather *tlb, struct ma_state *mas,
> {
> struct unlink_vma_file_batch vb;
>
> + /*
> + * Ensure we have no stale TLB entries by the time this mapping is
> + * removed from the rmap.
This comment would be ever so much better if it would explain *why* this
is important.
> + */
> + if (tlb->vma_pfn && !tlb->fullmm)
> + tlb_flush_mmu(tlb);
I can't say I particularly like accessing vma_pfn outside of the
mmu_gather code, but I'm also not sure its worth it to add a special
helper.
I do worry this makes the whole thing a little more fragile though.
