fs/btrfs/file.c

970  * This function locks the extent and properly waits for data=ordered extents
996 		struct btrfs_ordered_extent *ordered;
1013 		ordered = btrfs_lookup_ordered_range(inode, start_pos,
1015 		if (ordered &&
1016 		    ordered->file_offset + ordered->num_bytes > start_pos &&
1017 		    ordered->file_offset <= last_pos) {
1024 			btrfs_start_ordered_extent(ordered);
1025 			btrfs_put_ordered_extent(ordered);
1028 		if (ordered)
1029 			btrfs_put_ordered_extent(ordered);
1053  * This function will flush ordered extents in the range to ensure proper
1529 	 * an ordered extent, and after that it will fault in the pages that the
1532 	 * find that ordered extent and then wait for it to complete (at
1534 	 * obviously the ordered extent can never complete as we didn't submit
1761 	 * update the last_trans of the inode during ordered extent completion,
1777  * It needs to call filemap_fdatawait so that all ordered extent updates are
1804 	 * races between hole detection during logging and completion of ordered
1805 	 * extents outside the range, to missing checksums due to ordered extents
1842 	 * wait for all ordered extents to complete below.
1865 	 * For a full fsync we wait for the ordered extents to complete while
1867 	 * attach the ordered extents to the transaction so that a transaction
1869 	 * the current transaction commits before the ordered extents complete
1873 	 * logical address recorded in the ordered extent may change. We need
1880 		 * Get our ordered extents as soon as possible to avoid doing
1882 		 * checksums attached to the ordered extents.
1903 		 * An ordered extent might have started before and completed
1980 	 * the ordered extents, start it again and commit the transaction.  If
1981 	 * we attempt to wait on the ordered extents here we could deadlock with
1997 		 * making sure the transaction that had the ordered extents is
2228 		 * We can't have ordered extents in the range, nor dirty/writeback
2232 		 * for any ordered extents in the range to complete.
3115 	 * all ordered extents in the range to complete. After this we can lock
3117 	 * can't be more delalloc or ordered extents in the range.
3275 			/* Else trim our search range for ordered extents. */
3285 	 * Now also check if there's any ordered extent in the range.
3290 	 *    an ordered extent for the write. So we might just have been called
3291 	 *    after delalloc is flushed and before the ordered extent completes
3294 	 * 2) We may have an ordered extent created by flushing delalloc for a
3299 	 * being flushed, but using the ordered extents tree is more efficient
3300 	 * because it's usually much smaller as ordered extents are removed from
3309 	/* The ordered extent may span beyond our search range. */
3315 	/* Don't have unflushed delalloc, return the ordered extent range. */
3323 	 * We have both unflushed delalloc (io_tree) and an ordered extent.
3782 	 * our own ordered extent). This is because for direct IO reads,
3868 	 * from there.  We have to do this otherwise we'll miss the ordered