mbox series

[0/6] Convert several functions in page_io.c to use a folio

Message ID 20230717132602.2202147-1-zhangpeng362@huawei.com
Headers
Series Convert several functions in page_io.c to use a folio |

Message

zhangpeng (AS) July 17, 2023, 1:25 p.m. UTC 
  From: ZhangPeng <zhangpeng362@huawei.com>

This patch series converts several functions in page_io.c to use a
folio, which can remove several implicit calls to compound_head().

ZhangPeng (6):
  mm/page_io: use a folio in __end_swap_bio_read()
  mm/page_io: use a folio in sio_read_complete()
  mm/page_io: use a folio in swap_writepage_bdev_sync()
  mm/page_io: use a folio in swap_writepage_bdev_async()
  mm/page_io: convert count_swpout_vm_event() to take in a folio
  mm/page_io: convert bio_associate_blkg_from_page() to take in a folio

 mm/page_io.c | 56 +++++++++++++++++++++++++++-------------------------
 1 file changed, 29 insertions(+), 27 deletions(-)

Comments

Matthew Wilcox July 17, 2023, 1:40 p.m. UTC | #1 
On Mon, Jul 17, 2023 at 09:25:58PM +0800, Peng Zhang wrote:
> +++ b/mm/page_io.c
> @@ -406,19 +406,19 @@ static void sio_read_complete(struct kiocb *iocb, long ret)
>  
>  	if (ret == sio->len) {
>  		for (p = 0; p < sio->pages; p++) {
> -			struct page *page = sio->bvec[p].bv_page;
> +			struct folio *folio = page_folio(sio->bvec[p].bv_page);
>  
> -			SetPageUptodate(page);
> -			unlock_page(page);
> +			folio_mark_uptodate(folio);
> +			folio_unlock(folio);
>  		}

I'm kind of shocked this works today.  Usually bvecs coalesce adjacent
pages into a single entry, so you need to use a real iterator like
bio_for_each_folio_all() to extract individual pages from a bvec.
Maybe the sio bvec is constructed inefficiently.

I think Kent had some bvec folio iterators in progress?

>  		count_vm_events(PSWPIN, sio->pages);
>  	} else {
>  		for (p = 0; p < sio->pages; p++) {
> -			struct page *page = sio->bvec[p].bv_page;
> +			struct folio *folio = page_folio(sio->bvec[p].bv_page);
>  
> -			SetPageError(page);
> -			ClearPageUptodate(page);
> -			unlock_page(page);
> +			folio_set_error(folio);
> +			folio_clear_uptodate(folio);
> +			folio_unlock(folio);

Similar questions to the last patch -- who checks the error flag on this
page/folio, and isn't the folio already !uptodate?

>  		}
>  		pr_alert_ratelimited("Read-error on swap-device\n");
>  	}
> -- 
> 2.25.1
>
Matthew Wilcox July 17, 2023, 1:48 p.m. UTC | #2 
On Mon, Jul 17, 2023 at 09:26:01PM +0800, Peng Zhang wrote:
> -static inline void count_swpout_vm_event(struct page *page)
> +static inline void count_swpout_vm_event(struct folio *folio)
>  {
>  #ifdef CONFIG_TRANSPARENT_HUGEPAGE
> -	if (unlikely(PageTransHuge(page)))
> +	if (unlikely(folio_test_large(folio)))
>  		count_vm_event(THP_SWPOUT);
>  #endif

Since this is a THP_SWPOUT event, we should do this as:

-#ifdef CONFIG_TRANSPARENT_HUGEPAGE
-	if (unlikely(PageTransHuge(page)))
+	if (folio_test_pmd_mappable(folio))
		count_vm_event(THP_SWPOUT);
-#endif
Matthew Wilcox July 17, 2023, 2:01 p.m. UTC | #3 
On Mon, Jul 17, 2023 at 09:26:00PM +0800, Peng Zhang wrote:
> From: ZhangPeng <zhangpeng362@huawei.com>
> 
> Saves one implicit call to compound_head().
> 
> Signed-off-by: ZhangPeng <zhangpeng362@huawei.com>

Reviewed-by: Matthew Wilcox (Oracle) <willy@infradead.org>