On Wed, Oct 16, 2024 at 8:38 AM Matthew Wilcox <willy@xxxxxxxxxxxxx> wrote:
>
> On Wed, Oct 16, 2024 at 06:09:30PM +0800, Baolin Wang wrote:
> > @@ -3128,8 +3127,9 @@ static ssize_t shmem_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
> > if (folio) {
> > folio_unlock(folio);
> >
> > - page = folio_file_page(folio, index);
> > - if (PageHWPoison(page)) {
> > + if (folio_test_hwpoison(folio) ||
> > + (folio_test_large(folio) &&
> > + folio_test_has_hwpoisoned(folio))) {
>
> Hm, so if we have hwpoison set on one page in a folio, we now can't read
> bytes from any page in the folio? That seems like we've made a bad
> situation worse.
Yeah, I agree. I think we can fallback to page copy if
folio_test_has_hwpoisoned is true. The PG_hwpoison flag is per page.
The folio_test_has_hwpoisoned is kept set if the folio split is failed
in memory failure handler.
>
>
