Both ext4 and xfs implement seeking for the next hole or piece of data in unwritten extents by scanning the page cache, and both versions share the same bug when iterating the buffers of a page: the start offset into the page isn't taken into account, so when a page fits more than two filesystem blocks, things will go wrong. For example, on a filesystem with a block size of 1k, the following command will fail: xfs_io -f -c "falloc 0 4k" \ -c "pwrite 1k 1k" \ -c "pwrite 3k 1k" \ -c "seek -a -r 0" foo In this example, neither lseek(fd, 1024, SEEK_HOLE) nor lseek(fd, 2048, SEEK_DATA) will return the correct result. Introduce a generic vfs helper for seeking in the page cache that gets this right. The next commits will replace the filesystem specific implementations.
Showing the changes to include/linux/buffer_head.h (1 of 2 files). Show all
Unified · Split
| @@ -201,6 +201,8 @@ void write_boundary_block(struct block_device *bdev, · 200 unmodified lines | |||
| 201 | sector_t bblock, unsigned blocksize); | 201 | sector_t bblock, unsigned blocksize); |
| 202 | int bh_uptodate_or_lock(struct buffer_head *bh); | 202 | int bh_uptodate_or_lock(struct buffer_head *bh); |
| 203 | int bh_submit_read(struct buffer_head *bh); | 203 | int bh_submit_read(struct buffer_head *bh); |
| 204 | loff_t page_cache_seek_hole_data(struct inode *inode, loff_t offset, | ||
| 205 | loff_t length, int whence); | ||
| 204 | 206 | ||
| 205 | extern int buffer_heads_over_limit; | 207 | extern int buffer_heads_over_limit; |
| 206 | 208 | ||