Re: + mm-shmem-fallback-to-page-size-splice-if-large-folio-has-poisoned-pages.patch added to mm-unstable branch

[Baolin Wang <baolin.wang@xxxxxxxxxxxxxxxxx>]

On 2024/10/31 21:16, David Howells wrote:
> Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> > +		/*
> > +		 * Fallback to PAGE_SIZE splice if the large folio has hwpoisoned
> > +		 * pages.
> > +		 */
> > +		if (likely(!fallback_page_splice)) {
> > +			size = len;
> > +		} else {
> > +			size_t offset = *ppos & ~PAGE_MASK;
> > +
> > +			size = min_t(loff_t, PAGE_SIZE - offset, len);
> > +		}
>
> It might look better to write this as:
>
> 		size = len;
> 		if (unlikely(fallback_page_splice)) {
> 			size_t offset = *ppos & ~PAGE_MASK;
>
> 			size = umin(size, PAGE_SIZE - offset);
> 		}

Yes, looks better. Thanks.

Andrew, could you help squash the following changes into this patch? Thanks.

diff --git a/mm/shmem.c b/mm/shmem.c
index 44282a296c33..d910063dac27 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -3335,12 +3335,11 @@ static ssize_t shmem_file_splice_read(struct 
file *in, loff_t *ppos,
                 * Fallback to PAGE_SIZE splice if the large folio has 
hwpoisoned
                 * pages.
                 */
-               if (likely(!fallback_page_splice)) {
-                       size = len;
-               } else {
+               size = len;
+               if (unlikely(fallback_page_splice)) {
                        size_t offset = *ppos & ~PAGE_MASK;

-                       size = min_t(loff_t, PAGE_SIZE - offset, len);
+                       size = umin(size, PAGE_SIZE - offset);
                }
                part = min_t(loff_t, isize - *ppos, size);