History

filemap: Correct the conditions for marking a folio as accessed

We had an off-by-one error which meant that we never marked the first page in a read as accessed. This was visible as a slowdown when re-reading a file as pages were being evicted from cache too soon. In reviewing this code, we noticed a second bug where a multi-page folio would be marked as accessed multiple times when doing reads that were less than the size of the folio. Abstract the comparison of whether two file positions are in the same folio into a new function, fixing both of these bugs.

Matthew Wilcox (Oracle) committed 4 years ago

Commit 5ccc9441 parent 78ca5581 file · +10 −3Browse files

Unified · Split

mm/filemap.c+10−3
@@ -2629,6 +2629,13 @@ err: · 2628 unmodified lines
2629 return err;2629 return err;
2630}2630}
2631 2631
2632static inline bool pos_same_folio(loff_t pos1, loff_t pos2, struct folio *folio)
2633{
2634 unsigned int shift = folio_shift(folio);
2635
2636 return (pos1 >> shift == pos2 >> shift);
2637}
2638
2632/**2639/**
2633 * filemap_read - Read data from the page cache.2640 * filemap_read - Read data from the page cache.
2634 * @iocb: The iocb to read.2641 * @iocb: The iocb to read.
@@ -2700,11 +2707,11 @@ ssize_t filemap_read(struct kiocb *iocb, struct iov_iter *iter, · 65 unmodified lines
2700 writably_mapped = mapping_writably_mapped(mapping);2707 writably_mapped = mapping_writably_mapped(mapping);
2701 2708
2702 /*2709 /*
2703 * When a sequential read accesses a page several times, only2710 * When a read accesses the same folio several times, only
2704 * mark it as accessed the first time.2711 * mark it as accessed the first time.
2705 */2712 */
2706 if (iocb->ki_pos >> PAGE_SHIFT !=2713 if (!pos_same_folio(iocb->ki_pos, ra->prev_pos - 1,
2707 ra->prev_pos >> PAGE_SHIFT)2714 fbatch.folios[0]))
2708 folio_mark_accessed(fbatch.folios[0]);2715 folio_mark_accessed(fbatch.folios[0]);
2709 2716
2710 for (i = 0; i < folio_batch_count(&fbatch); i++) {2717 for (i = 0; i < folio_batch_count(&fbatch); i++) {