kernel.git/fs/btrfs/send.c, branch linux-6.15.y

btrfs: send: use fallocate for hole punching with send stream v2

2025-08-20T16:36:31Z

commit 005b0a0c24e1628313e951516b675109a92cacfe upstream. Currently holes are sent as writes full of zeroes, which results in unnecessarily using disk space at the receiving end and increasing the stream size. In some cases we avoid sending writes of zeroes, like during a full send operation where we just skip writes for holes. But for some cases we fill previous holes with writes of zeroes too, like in this scenario: 1) We have a file with a hole in the range [2M, 3M), we snapshot the subvolume and do a full send. The range [2M, 3M) stays as a hole at the receiver since we skip sending write commands full of zeroes; 2) We punch a hole for the range [3M, 4M) in our file, so that now it has a 2M hole in the range [2M, 4M), and snapshot the subvolume. Now if we do an incremental send, we will send write commands full of zeroes for the range [2M, 4M), removing the hole for [2M, 3M) at the receiver. We could improve cases such as this last one by doing additional comparisons of file extent items (or their absence) between the parent and send snapshots, but that's a lot of code to add plus additional CPU and IO costs. Since the send stream v2 already has a fallocate command and btrfs-progs implements a callback to execute fallocate since the send stream v2 support was added to it, update the kernel to use fallocate for punching holes for V2+ streams. Test coverage is provided by btrfs/284 which is a version of btrfs/007 that exercises send stream v2 instead of v1, using fsstress with random operations and fssum to verify file contents. Link: https://github.com/kdave/btrfs-progs/issues/1001 CC: stable@vger.kernel.org # 6.1+ Reviewed-by: Boris Burkov Signed-off-by: Filipe Manana Reviewed-by: David Sterba Signed-off-by: David Sterba Signed-off-by: Greg Kroah-Hartman

btrfs: make btrfs_iget() return a btrfs inode instead

2025-03-18T19:35:50Z

It's an internal function and most of the time the callers are doing a lot of BTRFS_I() calls on the returned VFS inode to get the btrfs inode, so change the return type to struct btrfs_inode instead. Signed-off-by: Filipe Manana Reviewed-by: David Sterba Signed-off-by: David Sterba

btrfs: send: simplify return logic from send_encoded_extent()

2025-03-18T19:35:50Z

The 'out' label is pointless as we don't have anything to cleanup anymore (we used to have an inode to iput), so remove it and make error paths directly return an error. Reviewed-by: Qu Wenruo Signed-off-by: Filipe Manana Reviewed-by: David Sterba Signed-off-by: David Sterba

btrfs: send: remove unnecessary inode lookup at send_encoded_inline_extent()

2025-03-18T19:35:50Z

We are doing a lookup of the inode but we don't use it at all. So just remove this pointless lookup. Reviewed-by: Qu Wenruo Signed-off-by: Filipe Manana Reviewed-by: David Sterba Signed-off-by: David Sterba

btrfs: pass btrfs_root pointers to send ioctl parameters

2025-03-18T19:35:49Z

The ioctl switch btrfs_ioctl() provides several parameter types for convenience so we don't have to do the conversion in the callbacks. Pass root pointers to the send related functions. Signed-off-by: David Sterba

btrfs: send: simplify return logic from send_set_xattr()

2025-03-18T19:35:46Z

There's no longer any need for the 'out' label as there are no resources to cleanup anymore in case of an error and we can directly return if begin_cmd() fails. Signed-off-by: Filipe Manana Reviewed-by: David Sterba Signed-off-by: David Sterba

btrfs: send: avoid path allocation for the current inode when issuing commands

2025-03-18T19:35:46Z

Whenever we issue a command we allocate a path and then compute it. For the current inode this is not necessary since we have one preallocated and computed in the send context structure, so we can use it instead and avoid allocating and freeing a path. For example if we have 100 extents to send (100 write commands) for a file, we are allocating and freeing paths 100 times. So improve on this by avoiding path allocation and freeing whenever a command is for the current inode by using the current inode's path stored in the send context structure. A test was run before applying this patch and the previous one in the series: "btrfs: send: keep the current inode's path cached" The test script is the following: $ cat test.sh #!/bin/bash DEV=/dev/nullb0 MNT=/mnt/nullb0 mkfs.btrfs -f $DEV > /dev/null mount $DEV $MNT DIR="$MNT/one/two/three/four" FILE="$DIR/foobar" mkdir -p $DIR # Create some empty files to get a deeper btree and therefore make # path computations slower. for ((i = 1; i <= 30000; i++)); do echo -n > "$DIR/filler_$i" done for ((i = 0; i < 10000; i += 2)); do offset=$(( i * 4096 )) xfs_io -f -c "pwrite -S 0xab $offset 4K" $FILE > /dev/null done btrfs subvolume snapshot -r $MNT $MNT/snap start=$(date +%s%N) btrfs send -f /dev/null $MNT/snap end=$(date +%s%N) echo -e "\nsend took $(( (end - start) / 1000000 )) milliseconds" umount $MNT Result before applying the 2 patches: 1121 milliseconds Result after applying the 2 patches: 815 milliseconds (-31.6%) Signed-off-by: Filipe Manana Reviewed-by: David Sterba Signed-off-by: David Sterba

btrfs: send: keep the current inode's path cached

2025-03-18T19:35:46Z

Whenever we need to send a command for the current inode, like sending writes, xattr updates, truncates, utimes, etc, we compute the inode's path each time, which implies doing some memory allocations and traversing the inode hierarchy to extract the name of the inode and each ancestor directory, and that implies doing lookups in the subvolume tree amongst other operations. Most of the time, by far, the current inode's path doesn't change while we are processing it (like if we need to issue 100 write commands, the path remains the same and it's pointless to compute it 100 times). To avoid this keep the current inode's path cached in the send context and invalidate it or update it whenever it's needed (after unlinks or renames). A performance test, and its results, is mentioned in the next patch in the series (subject: "btrfs: send: avoid path allocation for the current inode when issuing commands"). Signed-off-by: Filipe Manana Reviewed-by: David Sterba Signed-off-by: David Sterba

btrfs: send: simplify return logic from send_rmdir()

2025-03-18T19:35:46Z

There is no need to have an 'out' label and jump into it since there are no resource cleanups to perform (release locks, free memory, etc), so make this simpler by removing the label and goto and instead return directly. Signed-off-by: Filipe Manana Reviewed-by: David Sterba Signed-off-by: David Sterba

btrfs: send: simplify return logic from send_unlink()

2025-03-18T19:35:46Z