BMC/Intel-BMC/linux.git/fs/btrfs/send.c, branch dev-4.17

Btrfs: send, fix missing truncate for inode with prealloc extent past eof

2018-05-02T09:55:29+00:00

An incremental send operation can miss a truncate operation when an inode has an increased size in the send snapshot and a prealloc extent beyond its size. Consider the following scenario where a necessary truncate operation is missing in the incremental send stream: 1) In the parent snapshot an inode has a size of 1282957 bytes and it has no prealloc extents beyond its size; 2) In the the send snapshot it has a size of 5738496 bytes and has a new extent at offsets 1884160 (length of 106496 bytes) and a prealloc extent beyond eof at offset 6729728 (and a length of 339968 bytes); 3) When processing the prealloc extent, at offset 6729728, we end up at send.c:send_write_or_clone() and set the @len variable to a value of 18446744073708560384 because @offset plus the original @len value is larger then the inode's size (6729728 + 339968 > 5738496). We then call send_extent_data(), with that @offset and @len, which in turn calls send_write(), and then the later calls fill_read_buf(). Because the offset passed to fill_read_buf() is greater then inode's i_size, this function returns 0 immediately, which makes send_write() and send_extent_data() do nothing and return immediately as well. When we get back to send.c:send_write_or_clone() we adjust the value of sctx->cur_inode_next_write_offset to @offset plus @len, which corresponds to 6729728 + 18446744073708560384 = 5738496, which is precisely the the size of the inode in the send snapshot; 4) Later when at send.c:finish_inode_if_needed() we determine that we don't need to issue a truncate operation because the value of sctx->cur_inode_next_write_offset corresponds to the inode's new size, 5738496 bytes. This is wrong because the last write operation that was issued started at offset 1884160 with a length of 106496 bytes, so the correct value for sctx->cur_inode_next_write_offset should be 1990656 (1884160 + 106496), so that a truncate operation with a value of 5738496 bytes would have been sent to insert a trailing hole at the destination. So fix the issue by making send.c:send_write_or_clone() not attempt to send write or clone operations for extents that start beyond the inode's size, since such attempts do nothing but waste time by calling helper functions and allocating path structures, and send currently has no fallocate command in order to create prealloc extents at the destination (either beyond a file's eof or not). The issue was found running the test btrfs/007 from fstests using a seed value of 1524346151 for fsstress. Reported-by: Gu, Jinxiang Fixes: ffa7c4296e93 ("Btrfs: send, do not issue unnecessary truncate operations") Signed-off-by: Filipe Manana Signed-off-by: David Sterba

btrfs: replace GPL boilerplate by SPDX -- sources

2018-04-12T14:29:51+00:00

Remove GPL boilerplate text (long, short, one-line) and keep the rest, ie. personal, company or original source copyright statements. Add the SPDX header. Signed-off-by: David Sterba

Btrfs: send: fix typo in TLV_PUT

2018-03-26T13:09:42+00:00

According to tlv_put()'s prototype, data and attrlen needs to be exchanged in the macro, but seems all callers are already aware of this misorder and are therefore not affected. Signed-off-by: Liu Bo Reviewed-by: David Sterba Signed-off-by: David Sterba

Btrfs: send, do not issue unnecessary truncate operations

2018-03-26T13:09:41+00:00

When send finishes processing an inode representing a regular file, it always issues a truncate operation for that file, even if its size did not change or the last write sets the file size correctly. In the most common cases, the issued write operations set the file to correct size (either full or incremental sends) or the file size did not change (for incremental sends), so the only case where a truncate operation is needed is when a file size becomes smaller in the send snapshot when compared to the parent snapshot. By not issuing unnecessary truncate operations we reduce the stream size and save time in the receiver. Currently truncating a file to the same size triggers writeback of its last page (if it's dirty) and waits for it to complete (only if the file size is not aligned with the filesystem's sector size). This is being fixed by another patch and is independent of this change (that patch's title is "Btrfs: skip writeback of last page when truncating file to same size"). The following script was used to measure time spent by a receiver without this change applied, with this change applied, and without this change and with the truncate fix applied (the fix to not make it start and wait for writeback to complete). $ cat test_send.sh #!/bin/bash SRC_DEV=/dev/sdc DST_DEV=/dev/sdd SRC_MNT=/mnt/sdc DST_MNT=/mnt/sdd mkfs.btrfs -f $SRC_DEV >/dev/null mkfs.btrfs -f $DST_DEV >/dev/null mount $SRC_DEV $SRC_MNT mount $DST_DEV $DST_MNT echo "Creating source filesystem" for ((t = 0; t < 10; t++)); do ( for ((i = 1; i <= 20000; i++)); do xfs_io -f -c "pwrite -S 0xab 0 5000" \ $SRC_MNT/file_$i > /dev/null done ) & worker_pids[$t]=$! done wait ${worker_pids[@]} echo "Creating and sending snapshot" btrfs subvolume snapshot -r $SRC_MNT $SRC_MNT/snap1 >/dev/null /usr/bin/time -f "send took %e seconds" \ btrfs send -f $SRC_MNT/send_file $SRC_MNT/snap1 /usr/bin/time -f "receive took %e seconds" \ btrfs receive -f $SRC_MNT/send_file $DST_MNT umount $SRC_MNT umount $DST_MNT The results, which are averages for 5 runs for each case, were the following: * Without this change average receive time was 26.49 seconds standard deviation of 2.53 seconds * Without this change and with the truncate fix average receive time was 12.51 seconds standard deviation of 0.32 seconds * With this change and without the truncate fix average receive time was 10.02 seconds standard deviation of 1.11 seconds Signed-off-by: Filipe Manana Signed-off-by: David Sterba

btrfs: add more __cold annotations

2018-03-26T13:09:39+00:00

The __cold functions are placed to a special section, as they're expected to be called rarely. This could help i-cache prefetches or help compiler to decide which branches are more/less likely to be taken without any other annotations needed. Though we can't add more __exit annotations, it's still possible to add __cold (that's also added with __exit). That way the following function categories are tagged: - printf wrappers, error messages - exit helpers Signed-off-by: David Sterba

btrfs: Remove custom crc32c init code

2018-03-26T13:09:39+00:00

The custom crc32 init code was introduced in 14a958e678cd ("Btrfs: fix btrfs boot when compiled as built-in") to enable using btrfs as a built-in. However, later as pointed out by 60efa5eb2e88 ("Btrfs: use late_initcall instead of module_init") this wasn't enough and finally btrfs was switched to late_initcall which comes after the generic crc32c implementation is initiliased. The latter commit superseeded the former. Now that we don't have to maintain our own code let's just remove it and switch to using the generic implementation. Despite touching a lot of files the patch is really simple. Here is the gist of the changes: 1. Select LIBCRC32C rather than the low-level modules. 2. s/btrfs_crc32c/crc32c/g 3. replace hash.h with linux/crc32c.h 4. Move the btrfs namehash funcs to ctree.h and change the tree accordingly. I've tested this with btrfs being both a module and a built-in and xfstest doesn't complain. Does seem to fix the longstanding problem of not automatically selectiong the crc32c module when btrfs is used. Possibly there is a workaround in dracut. The modinfo confirms that now all the module dependencies are there: before: depends: zstd_compress,zstd_decompress,raid6_pq,xor,zlib_deflate after: depends: libcrc32c,zstd_compress,zstd_decompress,raid6_pq,xor,zlib_deflate Signed-off-by: Nikolay Borisov Reviewed-by: David Sterba [ add more info to changelog from mails ] Signed-off-by: David Sterba

Btrfs: send, fix issuing write op when processing hole in no data mode

2018-03-01T15:18:07+00:00

When doing an incremental send of a filesystem with the no-holes feature enabled, we end up issuing a write operation when using the no data mode send flag, instead of issuing an update extent operation. Fix this by issuing the update extent operation instead. Trivial reproducer: $ mkfs.btrfs -f -O no-holes /dev/sdc $ mkfs.btrfs -f /dev/sdd $ mount /dev/sdc /mnt/sdc $ mount /dev/sdd /mnt/sdd $ xfs_io -f -c "pwrite -S 0xab 0 32K" /mnt/sdc/foobar $ btrfs subvolume snapshot -r /mnt/sdc /mnt/sdc/snap1 $ xfs_io -c "fpunch 8K 8K" /mnt/sdc/foobar $ btrfs subvolume snapshot -r /mnt/sdc /mnt/sdc/snap2 $ btrfs send /mnt/sdc/snap1 | btrfs receive /mnt/sdd $ btrfs send --no-data -p /mnt/sdc/snap1 /mnt/sdc/snap2 \ | btrfs receive -vv /mnt/sdd Before this change the output of the second receive command is: receiving snapshot snap2 uuid=f6922049-8c22-e544-9ff9-fc6755918447... utimes write foobar, offset 8192, len 8192 utimes foobar BTRFS_IOC_SET_RECEIVED_SUBVOL uuid=f6922049-8c22-e544-9ff9-... After this change it is: receiving snapshot snap2 uuid=564d36a3-ebc8-7343-aec9-bf6fda278e64... utimes update_extent foobar: offset=8192, len=8192 utimes foobar BTRFS_IOC_SET_RECEIVED_SUBVOL uuid=564d36a3-ebc8-7343-aec9-bf6fda278e64... Signed-off-by: Filipe Manana Signed-off-by: David Sterba

btrfs: Cleanup existing name_len checks

2018-01-22T15:08:12+00:00

Since tree-checker has verified leaf when reading from disk, we don't need the existing verify_dir_item() or btrfs_is_name_len_valid() checks. Signed-off-by: Qu Wenruo Reviewed-by: Nikolay Borisov Reviewed-by: David Sterba Signed-off-by: David Sterba

Btrfs: incremental send, fix wrong unlink path after renaming file

2017-11-28T16:15:30+00:00

Under some circumstances, an incremental send operation can issue wrong paths for unlink commands related to files that have multiple hard links and some (or all) of those links were renamed between the parent and send snapshots. Consider the following example: Parent snapshot . (ino 256) |---- a/ (ino 257) | |---- b/ (ino 259) | | |---- c/ (ino 260) | | |---- f2 (ino 261) | | | |---- f2l1 (ino 261) | |---- d/ (ino 262) |---- f1l1_2 (ino 258) |---- f2l2 (ino 261) |---- f1_2 (ino 258) Send snapshot . (ino 256) |---- a/ (ino 257) | |---- f2l1/ (ino 263) | |---- b2/ (ino 259) | |---- c/ (ino 260) | | |---- d3 (ino 262) | | |---- f1l1_2 (ino 258) | | |---- f2l2_2 (ino 261) | | |---- f1_2 (ino 258) | | | |---- f2 (ino 261) | |---- f1l2 (ino 258) | |---- d (ino 261) When computing the incremental send stream the following steps happen: 1) When processing inode 261, a rename operation is issued that renames inode 262, which currently as a path of "d", to an orphan name of "o262-7-0". This is done because in the send snapshot, inode 261 has of its hard links with a path of "d" as well. 2) Two link operations are issued that create the new hard links for inode 261, whose names are "d" and "f2l2_2", at paths "/" and "o262-7-0/" respectively. 3) Still while processing inode 261, unlink operations are issued to remove the old hard links of inode 261, with names "f2l1" and "f2l2", at paths "a/" and "d/". However path "d/" does not correspond anymore to the directory inode 262 but corresponds instead to a hard link of inode 261 (link command issued in the previous step). This makes the receiver fail with a ENOTDIR error when attempting the unlink operation. The problem happens because before sending the unlink operation, we failed to detect that inode 262 was one of ancestors for inode 261 in the parent snapshot, and therefore we didn't recompute the path for inode 262 before issuing the unlink operation for the link named "f2l2" of inode 262. The detection failed because the function "is_ancestor()" only follows the first hard link it finds for an inode instead of all of its hard links (as it was originally created for being used with directories only, for which only one hard link exists). So fix this by making "is_ancestor()" follow all hard links of the input inode. A test case for fstests follows soon. Signed-off-by: Filipe Manana Signed-off-by: David Sterba

btrfs: add a flag to iterate_inodes_from_logical to find all extent refs for uncompressed extents

2017-11-01T19:45:34+00:00

The LOGICAL_INO ioctl provides a backward mapping from extent bytenr and offset (encoded as a single logical address) to a list of extent refs. LOGICAL_INO complements TREE_SEARCH, which provides the forward mapping (extent ref -> extent bytenr and offset, or logical address). These are useful capabilities for programs that manipulate extents and extent references from userspace (e.g. dedup and defrag utilities). When the extents are uncompressed (and not encrypted and not other), check_extent_in_eb performs filtering of the extent refs to remove any extent refs which do not contain the same extent offset as the 'logical' parameter's extent offset. This prevents LOGICAL_INO from returning references to more than a single block. To find the set of extent references to an uncompressed extent from [a, b), userspace has to run a loop like this pseudocode: for (i = a; i < b; ++i) extent_ref_set += LOGICAL_INO(i); At each iteration of the loop (up to 32768 iterations for a 128M extent), data we are interested in is collected in the kernel, then deleted by the filter in check_extent_in_eb. When the extents are compressed (or encrypted or other), the 'logical' parameter must be an extent bytenr (the 'a' parameter in the loop). No filtering by extent offset is done (or possible?) so the result is the complete set of extent refs for the entire extent. This removes the need for the loop, since we get all the extent refs in one call. Add an 'ignore_offset' argument to iterate_inodes_from_logical, [...several levels of function call graph...], and check_extent_in_eb, so that we can disable the extent offset filtering for uncompressed extents. This flag can be set by an improved version of the LOGICAL_INO ioctl to get either behavior as desired. There is no functional change in this patch. The new flag is always false. Signed-off-by: Zygo Blaxell Reviewed-by: David Sterba [ minor coding style fixes ] Signed-off-by: David Sterba