| Message ID | 20241126145208.985-1-jack@suse.cz (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | Revert "readahead: properly shorten readahead when falling back to do_page_cache_ra()" | expand |
On Tue, 2024-11-26 at 15:52 +0100, Jan Kara wrote: > This reverts commit 7c877586da3178974a8a94577b6045a48377ff25. > > Anders and Philippe have reported that recent kernels occasionally hang > when used with NFS in readahead code. The problem has been bisected to > 7c877586da3 ("readahead: properly shorten readahead when falling back to > do_page_cache_ra()"). The cause of the problem is that ra->size can be > shrunk by read_pages() call and subsequently we end up calling > do_page_cache_ra() with negative (read huge positive) number of pages. > Let's revert 7c877586da3 for now until we can find a proper way how the > logic in read_pages() and page_cache_ra_order() can coexist. This can > lead to reduced readahead throughput due to readahead window confusion > but that's better than outright hangs. > > Reported-by: Anders Blomdell <anders.blomdell@gmail.com> > Reported-by: Philippe Troin <phil@fifi.org> > CC: stable@vger.kernel.org > Signed-off-by: Jan Kara <jack@suse.cz> > --- > mm/readahead.c | 5 ++--- > 1 file changed, 2 insertions(+), 3 deletions(-) > > diff --git a/mm/readahead.c b/mm/readahead.c > index 8f1cf599b572..ea650b8b02fb 100644 > --- a/mm/readahead.c > +++ b/mm/readahead.c > @@ -458,8 +458,7 @@ void page_cache_ra_order(struct readahead_control *ractl, > struct file_ra_state *ra, unsigned int new_order) > { > struct address_space *mapping = ractl->mapping; > - pgoff_t start = readahead_index(ractl); > - pgoff_t index = start; > + pgoff_t index = readahead_index(ractl); > unsigned int min_order = mapping_min_folio_order(mapping); > pgoff_t limit = (i_size_read(mapping->host) - 1) >> PAGE_SHIFT; > pgoff_t mark = index + ra->size - ra->async_size; > @@ -522,7 +521,7 @@ void page_cache_ra_order(struct readahead_control *ractl, > if (!err) > return; > fallback: > - do_page_cache_ra(ractl, ra->size - (index - start), ra->async_size); > + do_page_cache_ra(ractl, ra->size, ra->async_size); > } > > static unsigned long ractl_max_pages(struct readahead_control *ractl, You can add a Tested-by: Philippe Troin <phil@fifi.org> tag as I did experiment and validate the fix with that revert on top of 6.11.10. Phil.
diff --git a/mm/readahead.c b/mm/readahead.c index 8f1cf599b572..ea650b8b02fb 100644 --- a/mm/readahead.c +++ b/mm/readahead.c @@ -458,8 +458,7 @@ void page_cache_ra_order(struct readahead_control *ractl, struct file_ra_state *ra, unsigned int new_order) { struct address_space *mapping = ractl->mapping; - pgoff_t start = readahead_index(ractl); - pgoff_t index = start; + pgoff_t index = readahead_index(ractl); unsigned int min_order = mapping_min_folio_order(mapping); pgoff_t limit = (i_size_read(mapping->host) - 1) >> PAGE_SHIFT; pgoff_t mark = index + ra->size - ra->async_size; @@ -522,7 +521,7 @@ void page_cache_ra_order(struct readahead_control *ractl, if (!err) return; fallback: - do_page_cache_ra(ractl, ra->size - (index - start), ra->async_size); + do_page_cache_ra(ractl, ra->size, ra->async_size); } static unsigned long ractl_max_pages(struct readahead_control *ractl,
This reverts commit 7c877586da3178974a8a94577b6045a48377ff25. Anders and Philippe have reported that recent kernels occasionally hang when used with NFS in readahead code. The problem has been bisected to 7c877586da3 ("readahead: properly shorten readahead when falling back to do_page_cache_ra()"). The cause of the problem is that ra->size can be shrunk by read_pages() call and subsequently we end up calling do_page_cache_ra() with negative (read huge positive) number of pages. Let's revert 7c877586da3 for now until we can find a proper way how the logic in read_pages() and page_cache_ra_order() can coexist. This can lead to reduced readahead throughput due to readahead window confusion but that's better than outright hangs. Reported-by: Anders Blomdell <anders.blomdell@gmail.com> Reported-by: Philippe Troin <phil@fifi.org> CC: stable@vger.kernel.org Signed-off-by: Jan Kara <jack@suse.cz> --- mm/readahead.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-)