mm/readahead: Do not allow order-1 folio
Subsystems (3)
PATCH 1/1 mm/readahead: Do not allow order-1 folio
The THP machinery does not support order-1 folios because it requires meta data spanning the first 3 struct pages. So order-2 is the smallest large folio that we can safely create.
There was a theoretical bug whereby if ra->size was 2 or 3 pages (due to the device-specific bdi->ra_pages being set that way), we could end up with order = 1. Fix this by unconditionally checking if the preferred order is 1 and if so, set it to 0. Previously this was done in a few specific places, but with this refactoring it is done just once, unconditionally, at the end of the calculation.
This is a theoretical bug found during review of the code; I have no evidence to suggest this manifests in the real world (I expect all device-specific ra_pages values are much bigger than 3).
--- mm/readahead.c | 14 ++++++-------- 1 file changed, 6 insertions(+), 8 deletions(-)
mm/readahead.c +6 -8
@@ -511,16 +511,14 @@ void page_cache_ra_order(struct readahead_control *ractl,unsigned int order = new_order;/* Align with smaller pages if needed */ -if (index & ((1UL << order) -1 )) { +if (index & ((1UL << order) -1 )) order =__ffs (index); -if (order ==1 ) - order =0 ; - }/* Don't allocate pages past EOF */ -while (index + (1UL << order) -1 > limit) { -if (--order ==1 ) - order =0 ; - } +while (index + (1UL << order) -1 > limit) + order--; +/* THP machinery does not support order-1 */ +if (order ==1 ) + order =0 ; err =ra_alloc_folio (ractl, index, mark, order, gfp);if (err)break ;
--
2.25.1
Quoted 11 lines · Ryan Roberts
The THP machinery does not support order-1 folios because it requires meta data spanning the first 3
struct pages. So order-2 is the smallest large folio that we can safely create.There was a theoretical bug whereby if
ra->sizewas 2 or 3 pages (due to the device-specificbdi->ra_pagesbeing set that way), we could end up with order = 1. Fix this by unconditionally checking if the preferred order is 1 and if so, set it to 0. Previously this was done in a few specific places, but with this refactoring it is done just once, unconditionally, at the end of the calculation.
Quoted 5 lines
This is a theoretical bug found during review of the code; I have no evidence to suggest this manifests in the real world (I expect all device-specific
ra_pagesvalues are much bigger than 3).Signed-off-by Ryan Roberts <[email protected]>
It's better code anyway!