Doug / smarc-fsl-linux-kernel

Download as
Browse Code ยป

Commit 623e3db9f9b7d6e7b2a99180f9cf0825c936ab7a

Authored by Hugh Dickins
Committed by Linus Torvalds
1 parent 3748b2f15b
Exists in smarc-l5.0.0_1.0.0-ga and in 5 other branches imx_3.10.17_1.0.1_ga, imx_3.10.53_1.1.0_ga, imx_3.14.28_1.0.0_ga, smarc-imx_3.10.53_1.1.0_ga, smarc-imx_3.14.28_1.0.0_ga

mm for fs: add truncate_pagecache_range() 

Holepunching filesystems ext4 and xfs are using truncate_inode_pages_range
but forgetting to unmap pages first (ocfs2 remembers).  This is not really
a bug, since races already require truncate_inode_page() to handle that
case once the page is locked; but it can be very inefficient if the file
being punched happens to be mapped into many vmas.

Provide a drop-in replacement truncate_pagecache_range() which does the
unmapping pass first, handling the awkward mismatch between arguments to
truncate_inode_pages_range() and arguments to unmap_mapping_range().

Note that holepunching does not unmap privately COWed pages in the range:
POSIX requires that we do so when truncating, but it's hard to justify,
difficult to implement without an i_size cutoff, and no filesystem is
attempting to implement it.

Signed-off-by: Hugh Dickins <hughd@google.com>
Cc: "Theodore Ts'o" <tytso@mit.edu>
Cc: Andreas Dilger <adilger.kernel@dilger.ca>
Cc: Mark Fasheh <mfasheh@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Ben Myers <bpm@sgi.com>
Cc: Alex Elder <elder@kernel.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

Showing 2 changed files with 41 additions and 1 deletions Side-by-side Diff

... ... @@ -954,7 +954,7 @@
954 954 extern void truncate_setsize(struct inode *inode, loff_t newsize);
955 955 extern int vmtruncate(struct inode *inode, loff_t offset);
956 956 extern int vmtruncate_range(struct inode *inode, loff_t offset, loff_t end);
957   -
  957 +void truncate_pagecache_range(struct inode *inode, loff_t offset, loff_t end);
958 958 int truncate_inode_page(struct address_space *mapping, struct page *page);
959 959 int generic_error_remove_page(struct address_space *mapping, struct page *page);
960 960  
... ... @@ -626,4 +626,44 @@
626 626  
627 627 return 0;
628 628 }
  629 +
  630 +/**
  631 + * truncate_pagecache_range - unmap and remove pagecache that is hole-punched
  632 + * @inode: inode
  633 + * @lstart: offset of beginning of hole
  634 + * @lend: offset of last byte of hole
  635 + *
  636 + * This function should typically be called before the filesystem
  637 + * releases resources associated with the freed range (eg. deallocates
  638 + * blocks). This way, pagecache will always stay logically coherent
  639 + * with on-disk format, and the filesystem would not have to deal with
  640 + * situations such as writepage being called for a page that has already
  641 + * had its underlying blocks deallocated.
  642 + */
  643 +void truncate_pagecache_range(struct inode *inode, loff_t lstart, loff_t lend)
  644 +{
  645 + struct address_space *mapping = inode->i_mapping;
  646 + loff_t unmap_start = round_up(lstart, PAGE_SIZE);
  647 + loff_t unmap_end = round_down(1 + lend, PAGE_SIZE) - 1;
  648 + /*
  649 + * This rounding is currently just for example: unmap_mapping_range
  650 + * expands its hole outwards, whereas we want it to contract the hole
  651 + * inwards. However, existing callers of truncate_pagecache_range are
  652 + * doing their own page rounding first; and truncate_inode_pages_range
  653 + * currently BUGs if lend is not pagealigned-1 (it handles partial
  654 + * page at start of hole, but not partial page at end of hole). Note
  655 + * unmap_mapping_range allows holelen 0 for all, and we allow lend -1.
  656 + */
  657 +
  658 + /*
  659 + * Unlike in truncate_pagecache, unmap_mapping_range is called only
  660 + * once (before truncating pagecache), and without "even_cows" flag:
  661 + * hole-punching should not remove private COWed pages from the hole.
  662 + */
  663 + if ((u64)unmap_end > (u64)unmap_start)
  664 + unmap_mapping_range(mapping, unmap_start,
  665 + 1 + unmap_end - unmap_start, 0);
  666 + truncate_inode_pages_range(mapping, lstart, lend);
  667 +}
  668 +EXPORT_SYMBOL(truncate_pagecache_range);