Re: [PATCH v2 1/3] tmpfs: zero post-eof uptodate folios on swapout

[Baolin Wang <baolin.wang@linux.alibaba.com>]

On 2025/11/13 00:25, Brian Foster wrote:
> As a first step to facilitate efficient post-eof zeroing in tmpfs,
> zero post-eof uptodate folios at swap out time. This ensures that
> post-eof ranges are zeroed "on disk" (i.e. analogous to traditional
> pagecache writeback) and facilitates zeroing on file size changes by
> allowing it to not have to swap in.
> 
> Note that shmem_writeout() already zeroes !uptodate folios so this
> introduces some duplicate logic. We'll clean this up in the next
> patch.
> 
> Signed-off-by: Brian Foster <bfoster@redhat.com>
> ---
>   mm/shmem.c | 19 +++++++++++++++++--
>   1 file changed, 17 insertions(+), 2 deletions(-)
> 
> diff --git a/mm/shmem.c b/mm/shmem.c
> index 0a25ee095b86..5fb3c911894f 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -1577,6 +1577,8 @@ int shmem_writeout(struct folio *folio, struct swap_iocb **plug,
>   	struct inode *inode = mapping->host;
>   	struct shmem_inode_info *info = SHMEM_I(inode);
>   	struct shmem_sb_info *sbinfo = SHMEM_SB(inode->i_sb);
> +	loff_t i_size = i_size_read(inode);
> +	pgoff_t end_index = DIV_ROUND_UP(i_size, PAGE_SIZE);
>   	pgoff_t index;
>   	int nr_pages;
>   	bool split = false;
> @@ -1596,8 +1598,7 @@ int shmem_writeout(struct folio *folio, struct swap_iocb **plug,
>   	 * (unless fallocate has been used to preallocate beyond EOF).
>   	 */
>   	if (folio_test_large(folio)) {
> -		index = shmem_fallocend(inode,
> -			DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE));
> +		index = shmem_fallocend(inode, end_index);
>   		if ((index > folio->index && index < folio_next_index(folio)) ||
>   		    !IS_ENABLED(CONFIG_THP_SWAP))
>   			split = true;
> @@ -1647,6 +1648,20 @@ int shmem_writeout(struct folio *folio, struct swap_iocb **plug,
>   		folio_mark_uptodate(folio);
>   	}
>   
> +	/*
> +	 * Ranges beyond EOF must be zeroed at writeout time. This mirrors
> +	 * traditional writeback behavior and facilitates zeroing on file size
> +	 * changes without having to swap back in.
> +	 */
> +	if (folio_next_index(folio) >= end_index) {
> +		size_t from = offset_in_folio(folio, i_size);
> +
> +		if (index >= end_index) {
> +			folio_zero_segment(folio, 0, folio_size(folio));
> +		} else if (from)
> +			folio_zero_segment(folio, from, folio_size(folio));
> +	}

As I mentioned before[1], if a large folio is beyond EOF, it will be 
split in shmem_writeout(), and those small folios beyond EOF will be 
dropped and freed in __folio_split(). Of course, there's another special 
case as Hugh mentioned: when there's a 'fallocend' beyond i_size (e.g., 
fallocate()), it will keep the pages allocated beyond EOF after the 
split. However, your 'end_index' here does not consider 'fallocend,' so 
it seems to me that this portion of the code doesn't actually take effect.

[1] 
https://lore.kernel.org/linux-mm/097c0b07-1f43-51c3-3591-aaa2015226c2@google.com/