btrfs: stop copying old file extents when doing a full fsync

When logging an inode in full sync mode, we go over every leaf that was
modified in the current transaction and has items associated to our inode,
and then copy all those items into the log tree. This includes copying
file extent items that were created and added to the inode in past
transactions, which is useless and only makes use more leaf space in the
log tree.

It's common to have a file with many file extent items spanning many
leaves where only a few file extent items are new and need to be logged,
and in such case we log all the file extent items we find in the modified
leaves.

So change the full sync behaviour to skip over file extent items that are
not needed. Those are the ones that match the following criteria:

1) Have a generation older than the current transaction and the inode
   was not a target of a reflink operation, as that can copy file extent
   items from a past generation from some other inode into our inode, so
   we have to log them;

2) Start at an offset within i_size - we must log anything at or beyond
   i_size, otherwise we would lose prealloc extents after log replay.

The following script exercises a scenario where this happens, and it's
somehow close enough to what happened often on a SQL Server workload which
I had to debug sometime ago to fix an issue where a pattern of writes to
prealloc extents and fsync resulted in fsync failing with -EIO (that was
commit ea7036de0d36c4 ("btrfs: fix fsync failure and transaction abort
after writes to prealloc extents")). In that particular case, we had large
files that had random writes and were often truncated, which made the
next fsync be a full sync.

  $ cat test.sh
  #!/bin/bash

  DEV=/dev/sdi
  MNT=/mnt/sdi

  MKFS_OPTIONS="-O no-holes -R free-space-tree"
  MOUNT_OPTIONS="-o ssd"

  FILE_SIZE=$((1 * 1024 * 1024 * 1024)) # 1G
  # FILE_SIZE=$((2 * 1024 * 1024 * 1024)) # 2G
  # FILE_SIZE=$((512 * 1024 * 1024)) # 512M

  mkfs.btrfs -f $MKFS_OPTIONS $DEV
  mount $MOUNT_OPTIONS $DEV $MNT

  # Create a file with many extents. Use direct IO to make it faster
  # to create the file - using buffered IO we would have to fsync
  # after each write (terribly slow).
  echo "Creating file with $((FILE_SIZE / 4096)) extents of 4K each..."
  xfs_io -f -d -c "pwrite -b 4K 0 $FILE_SIZE" $MNT/foobar

  # Commit the transaction, so every extent after this is from an
  # old generation.
  sync

  # Now rewrite only a few extents, which are all far spread apart from
  # each other (e.g. 1G / 32M = 32 extents).
  # After this only a few extents have a new generation, while all other
  # ones have an old generation.
  echo "Rewriting $((FILE_SIZE / (32 * 1024 * 1024))) extents..."
  for ((i = 0; i < $FILE_SIZE; i += $((32 * 1024 * 1024)))); do
      xfs_io -c "pwrite $i 4K" $MNT/foobar >/dev/null
  done

  # Fsync, the inode logged in full sync mode since it was never fsynced
  # before.
  echo "Fsyncing file..."
  xfs_io -c "fsync" $MNT/foobar

  umount $MNT

And the following bpftrace program was running when executing the test
script:

  $ cat bpf-script.sh
  #!/usr/bin/bpftrace

  k:btrfs_log_inode
  {
      @start_log_inode[tid] = nsecs;
  }

  kr:btrfs_log_inode
  /@start_log_inode[tid]/
  {
      @log_inode_dur[tid] = (nsecs - @start_log_inode[tid]) / 1000;
      delete(@start_log_inode[tid]);
  }

  k:btrfs_sync_log
  {
      @start_sync_log[tid] = nsecs;
  }

  kr:btrfs_sync_log
  /@start_sync_log[tid]/
  {
      $sync_log_dur = (nsecs - @start_sync_log[tid]) / 1000;
      printf("btrfs_log_inode() took %llu us\n", @log_inode_dur[tid]);
      printf("btrfs_sync_log()  took %llu us\n", $sync_log_dur);
      delete(@start_sync_log[tid]);
      delete(@log_inode_dur[tid]);
      exit();
  }

With 512M test file, before this patch:

  btrfs_log_inode() took 15218 us
  btrfs_sync_log()  took 1328 us

  Log tree has 17 leaves and 1 node, its total size is 294912 bytes.

With 512M test file, after this patch:

  btrfs_log_inode() took 14760 us
  btrfs_sync_log()  took 588 us

  Log tree has a single leaf, its total size is 16K.

With 1G test file, before this patch:

  btrfs_log_inode() took 27301 us
  btrfs_sync_log()  took 1767 us

  Log tree has 33 leaves and 1 node, its total size is 557056 bytes.

With 1G test file, after this patch:

  btrfs_log_inode() took 26166 us
  btrfs_sync_log()  took 593 us

  Log tree has a single leaf, its total size is 16K

With 2G test file, before this patch:

  btrfs_log_inode() took 50892 us
  btrfs_sync_log()  took 3127 us

  Log tree has 65 leaves and 1 node, its total size is 1081344 bytes.

With 2G test file, after this patch:

  btrfs_log_inode() took 50126 us
  btrfs_sync_log()  took 586 us

  Log tree has a single leaf, its total size is 16K.

Signed-off-by: Filipe Manana <fdmanana@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>

2 files changed, 148 insertions, 73 deletions

diff --git a/fs/btrfs/reflink.c b/fs/btrfs/reflink.c
index a3930da4eb3f..c083ded71ef7 100644
--- a/fs/btrfs/reflink.c
+++ b/fs/btrfs/reflink.c
@@ -518,17 +518,22 @@ process_slot:
 		btrfs_release_path(path);
 
 		/*
-		 * If this is a new extent update the last_reflink_trans of both
-		 * inodes. This is used by fsync to make sure it does not log
-		 * multiple checksum items with overlapping ranges. For older
-		 * extents we don't need to do it since inode logging skips the
-		 * checksums for older extents. Also ignore holes and inline
-		 * extents because they don't have checksums in the csum tree.
+		 * Whenever we share an extent we update the last_reflink_trans
+		 * of each inode to the current transaction. This is needed to
+		 * make sure fsync does not log multiple checksum items with
+		 * overlapping ranges (because some extent items might refer
+		 * only to sections of the original extent). For the destination
+		 * inode we do this regardless of the generation of the extents
+		 * or even if they are inline extents or explicit holes, to make
+		 * sure a full fsync does not skip them. For the source inode,
+		 * we only need to update last_reflink_trans in case it's a new
+		 * extent that is not a hole or an inline extent, to deal with
+		 * the checksums problem on fsync.
 		 */
-		if (extent_gen == trans->transid && disko > 0) {
+		if (extent_gen == trans->transid && disko > 0)
 			BTRFS_I(src)->last_reflink_trans = trans->transid;
-			BTRFS_I(inode)->last_reflink_trans = trans->transid;
-		}
+
+		BTRFS_I(inode)->last_reflink_trans = trans->transid;
 
 		last_dest_end = ALIGN(new_key.offset + datal,
 				      fs_info->sectorsize);
diff --git a/fs/btrfs/tree-log.c b/fs/btrfs/tree-log.c
index d5b7b565036d..bb3781fed8da 100644
--- a/fs/btrfs/tree-log.c
+++ b/fs/btrfs/tree-log.c
@@ -4418,21 +4418,19 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 			       int start_slot, int nr, int inode_only,
 			       u64 logged_isize)
 {
-	struct btrfs_fs_info *fs_info = trans->fs_info;
-	unsigned long src_offset;
-	unsigned long dst_offset;
 	struct btrfs_root *log = inode->root->log_root;
 	struct btrfs_file_extent_item *extent;
-	struct btrfs_inode_item *inode_item;
 	struct extent_buffer *src = src_path->nodes[0];
-	int ret;
+	int ret = 0;
 	struct btrfs_key *ins_keys;
 	u32 *ins_sizes;
 	struct btrfs_item_batch batch;
 	char *ins_data;
 	int i;
+	int dst_index;
 	struct list_head ordered_sums;
-	int skip_csum = inode->flags & BTRFS_INODE_NODATASUM;
+	const bool skip_csum = (inode->flags & BTRFS_INODE_NODATASUM);
+	const u64 i_size = i_size_read(&inode->vfs_inode);
 
 	INIT_LIST_HEAD(&ordered_sums);
 
@@ -4446,28 +4444,140 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 	batch.keys = ins_keys;
 	batch.data_sizes = ins_sizes;
 	batch.total_data_size = 0;
-	batch.nr = nr;
+	batch.nr = 0;
 
+	dst_index = 0;
 	for (i = 0; i < nr; i++) {
-		ins_sizes[i] = btrfs_item_size(src, i + start_slot);
-		batch.total_data_size += ins_sizes[i];
-		btrfs_item_key_to_cpu(src, ins_keys + i, i + start_slot);
+		const int src_slot = start_slot + i;
+		struct btrfs_root *csum_root;
+		u64 disk_bytenr;
+		u64 disk_num_bytes;
+		u64 extent_offset;
+		u64 extent_num_bytes;
+		bool is_old_extent;
+
+		btrfs_item_key_to_cpu(src, &ins_keys[dst_index], src_slot);
+
+		if (ins_keys[dst_index].type != BTRFS_EXTENT_DATA_KEY)
+			goto add_to_batch;
+
+		extent = btrfs_item_ptr(src, src_slot,
+					struct btrfs_file_extent_item);
+
+		is_old_extent = (btrfs_file_extent_generation(src, extent) <
+				 trans->transid);
+
+		/*
+		 * Don't copy extents from past generations. That would make us
+		 * log a lot more metadata for common cases like doing only a
+		 * few random writes into a file and then fsync it for the first
+		 * time or after the full sync flag is set on the inode. We can
+		 * get leaves full of extent items, most of which are from past
+		 * generations, so we can skip them - as long as the inode has
+		 * not been the target of a reflink operation in this transaction,
+		 * as in that case it might have had file extent items with old
+		 * generations copied into it. We also must always log prealloc
+		 * extents that start at or beyond eof, otherwise we would lose
+		 * them on log replay.
+		 */
+		if (is_old_extent &&
+		    ins_keys[dst_index].offset < i_size &&
+		    inode->last_reflink_trans < trans->transid)
+			continue;
+
+		if (skip_csum)
+			goto add_to_batch;
+
+		/* Only regular extents have checksums. */
+		if (btrfs_file_extent_type(src, extent) != BTRFS_FILE_EXTENT_REG)
+			goto add_to_batch;
+
+		/*
+		 * If it's an extent created in a past transaction, then its
+		 * checksums are already accessible from the committed csum tree,
+		 * no need to log them.
+		 */
+		if (is_old_extent)
+			goto add_to_batch;
+
+		disk_bytenr = btrfs_file_extent_disk_bytenr(src, extent);
+		/* If it's an explicit hole, there are no checksums. */
+		if (disk_bytenr == 0)
+			goto add_to_batch;
+
+		disk_num_bytes = btrfs_file_extent_disk_num_bytes(src, extent);
+
+		if (btrfs_file_extent_compression(src, extent)) {
+			extent_offset = 0;
+			extent_num_bytes = disk_num_bytes;
+		} else {
+			extent_offset = btrfs_file_extent_offset(src, extent);
+			extent_num_bytes = btrfs_file_extent_num_bytes(src, extent);
+		}
+
+		csum_root = btrfs_csum_root(trans->fs_info, disk_bytenr);
+		disk_bytenr += extent_offset;
+		ret = btrfs_lookup_csums_range(csum_root, disk_bytenr,
+					       disk_bytenr + extent_num_bytes - 1,
+					       &ordered_sums, 0);
+		if (ret)
+			goto out;
+
+add_to_batch:
+		ins_sizes[dst_index] = btrfs_item_size(src, src_slot);
+		batch.total_data_size += ins_sizes[dst_index];
+		batch.nr++;
+		dst_index++;
 	}
+
+	/*
+	 * We have a leaf full of old extent items that don't need to be logged,
+	 * so we don't need to do anything.
+	 */
+	if (batch.nr == 0)
+		goto out;
+
 	ret = btrfs_insert_empty_items(trans, log, dst_path, &batch);
-	if (ret) {
-		kfree(ins_data);
-		return ret;
-	}
+	if (ret)
+		goto out;
 
-	for (i = 0; i < nr; i++, dst_path->slots[0]++) {
-		dst_offset = btrfs_item_ptr_offset(dst_path->nodes[0],
-						   dst_path->slots[0]);
+	dst_index = 0;
+	for (i = 0; i < nr; i++) {
+		const int src_slot = start_slot + i;
+		const int dst_slot = dst_path->slots[0] + dst_index;
+		struct btrfs_key key;
+		unsigned long src_offset;
+		unsigned long dst_offset;
 
-		src_offset = btrfs_item_ptr_offset(src, start_slot + i);
+		/*
+		 * We're done, all the remaining items in the source leaf
+		 * correspond to old file extent items.
+		 */
+		if (dst_index >= batch.nr)
+			break;
 
-		if (ins_keys[i].type == BTRFS_INODE_ITEM_KEY) {
-			inode_item = btrfs_item_ptr(dst_path->nodes[0],
-						    dst_path->slots[0],
+		btrfs_item_key_to_cpu(src, &key, src_slot);
+
+		if (key.type != BTRFS_EXTENT_DATA_KEY)
+			goto copy_item;
+
+		extent = btrfs_item_ptr(src, src_slot,
+					struct btrfs_file_extent_item);
+
+		/* See the comment in the previous loop, same logic. */
+		if (btrfs_file_extent_generation(src, extent) < trans->transid &&
+		    key.offset < i_size &&
+		    inode->last_reflink_trans < trans->transid)
+			continue;
+
+copy_item:
+		dst_offset = btrfs_item_ptr_offset(dst_path->nodes[0], dst_slot);
+		src_offset = btrfs_item_ptr_offset(src, src_slot);
+
+		if (key.type == BTRFS_INODE_ITEM_KEY) {
+			struct btrfs_inode_item *inode_item;
+
+			inode_item = btrfs_item_ptr(dst_path->nodes[0], dst_slot,
 						    struct btrfs_inode_item);
 			fill_inode_item(trans, dst_path->nodes[0], inode_item,
 					&inode->vfs_inode,
@@ -4475,55 +4585,15 @@ static noinline int copy_items(struct btrfs_trans_handle *trans,
 					logged_isize);
 		} else {
 			copy_extent_buffer(dst_path->nodes[0], src, dst_offset,
-					   src_offset, ins_sizes[i]);
+					   src_offset, ins_sizes[dst_index]);
 		}
 
-		/* take a reference on file data extents so that truncates
-		 * or deletes of this inode don't have to relog the inode
-		 * again
-		 */
-		if (ins_keys[i].type == BTRFS_EXTENT_DATA_KEY &&
-		    !skip_csum) {
-			int found_type;
-			extent = btrfs_item_ptr(src, start_slot + i,
-						struct btrfs_file_extent_item);
-
-			if (btrfs_file_extent_generation(src, extent) < trans->transid)
-				continue;
-
-			found_type = btrfs_file_extent_type(src, extent);
-			if (found_type == BTRFS_FILE_EXTENT_REG) {
-				struct btrfs_root *csum_root;
-				u64 ds, dl, cs, cl;
-				ds = btrfs_file_extent_disk_bytenr(src,
-								extent);
-				/* ds == 0 is a hole */
-				if (ds == 0)
-					continue;
-
-				dl = btrfs_file_extent_disk_num_bytes(src,
-								extent);
-				cs = btrfs_file_extent_offset(src, extent);
-				cl = btrfs_file_extent_num_bytes(src,
-								extent);
-				if (btrfs_file_extent_compression(src,
-								  extent)) {
-					cs = 0;
-					cl = dl;
-				}
-
-				csum_root = btrfs_csum_root(fs_info, ds);
-				ret = btrfs_lookup_csums_range(csum_root,
-						ds + cs, ds + cs + cl - 1,
-						&ordered_sums, 0);
-				if (ret)
-					break;
-			}
-		}
+		dst_index++;
 	}
 
 	btrfs_mark_buffer_dirty(dst_path->nodes[0]);
 	btrfs_release_path(dst_path);
+out:
 	kfree(ins_data);
 
 	/*