kernel_optimize_test/fs/ext4/truncate.h
Jan Kara ea3d7209ca ext4: fix races between page faults and hole punching
Currently, page faults and hole punching are completely unsynchronized.
This can result in page fault faulting in a page into a range that we
are punching after truncate_pagecache_range() has been called and thus
we can end up with a page mapped to disk blocks that will be shortly
freed. Filesystem corruption will shortly follow. Note that the same
race is avoided for truncate by checking page fault offset against
i_size but there isn't similar mechanism available for punching holes.

Fix the problem by creating new rw semaphore i_mmap_sem in inode and
grab it for writing over truncate, hole punching, and other functions
removing blocks from extent tree and for read over page faults. We
cannot easily use i_data_sem for this since that ranks below transaction
start and we need something ranking above it so that it can be held over
the whole truncate / hole punching operation. Also remove various
workarounds we had in the code to reduce race window when page fault
could have created pages with stale mapping information.

Signed-off-by: Jan Kara <jack@suse.com>
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
2015-12-07 14:28:03 -05:00

46 lines
1.3 KiB
C

/*
* linux/fs/ext4/truncate.h
*
* Common inline functions needed for truncate support
*/
/*
* Truncate blocks that were not used by write. We have to truncate the
* pagecache as well so that corresponding buffers get properly unmapped.
*/
static inline void ext4_truncate_failed_write(struct inode *inode)
{
down_write(&EXT4_I(inode)->i_mmap_sem);
truncate_inode_pages(inode->i_mapping, inode->i_size);
ext4_truncate(inode);
up_write(&EXT4_I(inode)->i_mmap_sem);
}
/*
* Work out how many blocks we need to proceed with the next chunk of a
* truncate transaction.
*/
static inline unsigned long ext4_blocks_for_truncate(struct inode *inode)
{
ext4_lblk_t needed;
needed = inode->i_blocks >> (inode->i_sb->s_blocksize_bits - 9);
/* Give ourselves just enough room to cope with inodes in which
* i_blocks is corrupt: we've seen disk corruptions in the past
* which resulted in random data in an inode which looked enough
* like a regular file for ext4 to try to delete it. Things
* will go a bit crazy if that happens, but at least we should
* try not to panic the whole kernel. */
if (needed < 2)
needed = 2;
/* But we need to bound the transaction so we don't overflow the
* journal. */
if (needed > EXT4_MAX_TRANS_DATA)
needed = EXT4_MAX_TRANS_DATA;
return EXT4_DATA_TRANS_BLOCKS(inode->i_sb) + needed;
}