Re: [PATCH v3 3/4] mm/memory: split non-tlb flushing part from zap_page_range_single()

[Lorenzo Stoakes]

On Wed, Apr 09, 2025 at 05:00:21PM -0700, SeongJae Park wrote:
> Some of zap_page_range_single() callers such as [process_]madvise() with
> MADV_DONTNEED[_LOCKED] cannot batch tlb flushes because
> zap_page_range_single() flushes tlb for each invocation.  Split out the
> body of zap_page_range_single() except mmu_gather object initialization
> and gathered tlb entries flushing for such batched tlb flushing usage.
>
> To avoid hugetlb pages allocation failures from concurrent page faults,
> the tlb flush should be done before hugetlb faults unlocking, though.
> Do the flush and the unlock inside the split out function in the order
> for hugetlb vma case.  Refer to commit 2820b0f09be9 ("hugetlbfs: close
> race between MADV_DONTNEED and page fault") for more details about the
> concurrent faults' page allocation failure problem.
>
> Signed-off-by: SeongJae Park <sj@xxxxxxxxxx>

LGTM, thanks!

Reviewed-by: Lorenzo Stoakes <lorenzo.stoakes@xxxxxxxxxx>

> ---
>  mm/memory.c | 49 +++++++++++++++++++++++++++++++++++++++----------
>  1 file changed, 39 insertions(+), 10 deletions(-)
>
> diff --git a/mm/memory.c b/mm/memory.c
> index fda6d6429a27..690695643dfb 100644
> --- a/mm/memory.c
> +++ b/mm/memory.c
> @@ -1998,36 +1998,65 @@ void unmap_vmas(struct mmu_gather *tlb, struct ma_state *mas,
>  	mmu_notifier_invalidate_range_end(&range);
>  }
>
> -/**
> - * zap_page_range_single - remove user pages in a given range
> +/*
> + * zap_page_range_single_batched - remove user pages in a given range
> + * @tlb: pointer to the caller's struct mmu_gather
>   * @vma: vm_area_struct holding the applicable pages
> - * @address: starting address of pages to zap
> - * @size: number of bytes to zap
> + * @address: starting address of pages to remove
> + * @size: number of bytes to remove
>   * @details: details of shared cache invalidation
>   *
> - * The range must fit into one VMA.
> + * @tlb shouldn't be NULL.  The range must fit into one VMA.  If @vma is for
> + * hugetlb, @tlb is flushed and re-initialized by this function.
>   */
> -void zap_page_range_single(struct vm_area_struct *vma, unsigned long address,
> +static void zap_page_range_single_batched(struct mmu_gather *tlb,
> +		struct vm_area_struct *vma, unsigned long address,
>  		unsigned long size, struct zap_details *details)
>  {
>  	const unsigned long end = address + size;
>  	struct mmu_notifier_range range;
> -	struct mmu_gather tlb;
> +
> +	VM_WARN_ON_ONCE(!tlb || tlb->mm != vma->vm_mm);
>
>  	mmu_notifier_range_init(&range, MMU_NOTIFY_CLEAR, 0, vma->vm_mm,
>  				address, end);
>  	hugetlb_zap_begin(vma, &range.start, &range.end);
> -	tlb_gather_mmu(&tlb, vma->vm_mm);
>  	update_hiwater_rss(vma->vm_mm);
>  	mmu_notifier_invalidate_range_start(&range);
>  	/*
>  	 * unmap 'address-end' not 'range.start-range.end' as range
>  	 * could have been expanded for hugetlb pmd sharing.
>  	 */
> -	unmap_single_vma(&tlb, vma, address, end, details, false);
> +	unmap_single_vma(tlb, vma, address, end, details, false);
>  	mmu_notifier_invalidate_range_end(&range);
> +	if (is_vm_hugetlb_page(vma)) {
> +		/*
> +		 * flush tlb and free resources before hugetlb_zap_end(), to
> +		 * avoid concurrent page faults' allocation failure.
> +		 */
> +		tlb_finish_mmu(tlb);
> +		hugetlb_zap_end(vma, details);
> +		tlb_gather_mmu(tlb, vma->vm_mm);
> +	}
> +}
> +
> +/**
> + * zap_page_range_single - remove user pages in a given range
> + * @vma: vm_area_struct holding the applicable pages
> + * @address: starting address of pages to zap
> + * @size: number of bytes to zap
> + * @details: details of shared cache invalidation
> + *
> + * The range must fit into one VMA.
> + */
> +void zap_page_range_single(struct vm_area_struct *vma, unsigned long address,
> +		unsigned long size, struct zap_details *details)
> +{
> +	struct mmu_gather tlb;
> +
> +	tlb_gather_mmu(&tlb, vma->vm_mm);
> +	zap_page_range_single_batched(&tlb, vma, address, size, details);
>  	tlb_finish_mmu(&tlb);
> -	hugetlb_zap_end(vma, details);
>  }
>
>  /**
> --
> 2.39.5