Skip to contentSkip to conversation

btrfs: some trivial cleanups related to io trees

PATCH Superseded [email protected] 3 participants 12 messages Apr 1, 2025 Btrfs file system

lore
Share & tools
lore
#0 lore raw

cover btrfs: some trivial cleanups related to io trees

From: Filipe Manana <[email protected]>

Some simple cleanups related to io trees that are very trivial and were initally part of a larger patchset. Details in the change logs.

Filipe Manana (3):
btrfs: use clear_extent_bit() at try_release_extent_state()
btrfs: use clear_extent_bits() at chunk_map_device_clear_bits()
btrfs: simplify last record detection at test_range_bit_exists()

 fs/btrfs/extent-io-tree.c | 8 +++-----
 fs/btrfs/extent_io.c      | 2 +-
 fs/btrfs/volumes.c        | 7 +++----
 3 files changed, 7 insertions(+), 10 deletions(-)
Signature · 1 line
2.45.2
#1 lore raw

PATCH 1/3 btrfs: use clear_extent_bit() at try_release_extent_state()

From: Filipe Manana <[email protected]>

Instead of using __clear_extent_bit() we can use clear_extent_bit() since we pass a NULL value for the changeset argument.

---
 fs/btrfs/extent_io.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/btrfs/extent_io.c +1 -1
index 19f21540475d..50b74531b745 100644
@@ -2651,7 +2651,7 @@ static bool try_release_extent_state(struct extent_io_tree *tree,
 	 * nodatasum, delalloc new and finishing ordered bits. The delalloc new
 	 * bit will be cleared by ordered extent completion.
 	 */
-	ret2 = __clear_extent_bit(tree, start, end, clear_bits, &cached_state, NULL);
+	ret2 = clear_extent_bit(tree, start, end, clear_bits, &cached_state);
 	/*
 	 * If clear_extent_bit failed for enomem reasons, we can't allow the
 	 * release to continue.
Signature · 1 line
2.45.2
#2 lore raw

PATCH 2/3 btrfs: use clear_extent_bits() at chunk_map_device_clear_bits()

In reply to [email protected]

From: Filipe Manana <[email protected]>

Instead of using __clear_extent_bit() we can use clear_extent_bits() since we pass a NULL value for the cached and changeset arguments.

---
 fs/btrfs/volumes.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
fs/btrfs/volumes.c +3 -4
index c8c21c55be53..784d5a15ef29 100644
@@ -5508,10 +5508,9 @@ static void chunk_map_device_clear_bits(struct btrfs_chunk_map *map, unsigned in
 		struct btrfs_io_stripe *stripe = &map->stripes[i];
 		struct btrfs_device *device = stripe->dev;
 
-		__clear_extent_bit(&device->alloc_state, stripe->physical,
-				   stripe->physical + map->stripe_size - 1,
-				   bits | EXTENT_NOWAIT,
-				   NULL, NULL);
+		clear_extent_bits(&device->alloc_state, stripe->physical,
+				  stripe->physical + map->stripe_size - 1,
+				  bits | EXTENT_NOWAIT);
 	}
 }
 
Signature · 1 line
2.45.2
#3 lore raw

PATCH 3/3 btrfs: simplify last record detection at test_range_bit_exists()

In reply to [email protected]

From: Filipe Manana <[email protected]>

Instead of keeping track of the minimum start offset of the next record and detecting overflow everytime we update that offset to be the sum of current record's end offset plus one, we can simply exit when the current record ends at or beyond our end offset and forget about about updating the start offset on every iteration and testing for it at the top of the loop. This makes both the source code and assembly code simpler, more efficient and shorter (reducing the object text size).

Also remove the pointless initializion to NULL of the state variable, as we don't use it before the first assignment to it. This may help avoid some warnings with clang tools such as the one reported/fixed by commit 966de47ff0c9 ("btrfs: remove redundant initialization of variables in log_new_ancestors").

---
 fs/btrfs/extent-io-tree.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)
fs/btrfs/extent-io-tree.c +3 -5
index 7ae24a533404..293cb354259d 100644
@@ -1726,14 +1726,14 @@ u64 count_range_bits(struct extent_io_tree *tree,
  */
 bool test_range_bit_exists(struct extent_io_tree *tree, u64 start, u64 end, u32 bit)
 {
-	struct extent_state *state = NULL;
+	struct extent_state *state;
 	bool bitset = false;
 
 	ASSERT(is_power_of_2(bit));
 
 	spin_lock(&tree->lock);
 	state = tree_search(tree, start);
-	while (state && start <= end) {
+	while (state) {
 		if (state->start > end)
 			break;
 
@@ -1742,9 +1742,7 @@ bool test_range_bit_exists(struct extent_io_tree *tree, u64 start, u64 end, u32
 			break;
 		}
 
-		/* If state->end is (u64)-1, start will overflow to 0 */
-		start = state->end + 1;
-		if (start > end || start == 0)
+		if (state->end >= end)
 			break;
 		state = next_state(state);
 	}
Signature · 1 line
2.45.2
#4 lore raw

v2 cover btrfs: some trivial cleanups related to io trees

In reply to [email protected]

From: Filipe Manana <[email protected]>

Some simple cleanups related to io trees that are very trivial and were initally part of a larger patchset. Details in the change logs.

V2: Added patch 3/4.

Filipe Manana (4):
btrfs: use clear_extent_bit() at try_release_extent_state()
btrfs: use clear_extent_bits() at chunk_map_device_clear_bits()
btrfs: use clear_extent_bits() instead of clear_extent_bit() where possible
btrfs: simplify last record detection at test_range_bit_exists()

 fs/btrfs/extent-io-tree.c    |  8 +++-----
 fs/btrfs/extent_io.c         |  2 +-
 fs/btrfs/inode.c             |  3 +--
 fs/btrfs/reflink.c           |  5 ++---
 fs/btrfs/tests/inode-tests.c | 24 ++++++++++++------------
 fs/btrfs/volumes.c           |  7 +++----
 6 files changed, 22 insertions(+), 27 deletions(-)
Signature · 1 line
2.45.2
#5 lore raw

PATCH v2 1/4 btrfs: use clear_extent_bit() at try_release_extent_state()

From: Filipe Manana <[email protected]>

Instead of using __clear_extent_bit() we can use clear_extent_bit() since we pass a NULL value for the changeset argument.

---
 fs/btrfs/extent_io.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/btrfs/extent_io.c +1 -1
index 19f21540475d..50b74531b745 100644
@@ -2651,7 +2651,7 @@ static bool try_release_extent_state(struct extent_io_tree *tree,
 	 * nodatasum, delalloc new and finishing ordered bits. The delalloc new
 	 * bit will be cleared by ordered extent completion.
 	 */
-	ret2 = __clear_extent_bit(tree, start, end, clear_bits, &cached_state, NULL);
+	ret2 = clear_extent_bit(tree, start, end, clear_bits, &cached_state);
 	/*
 	 * If clear_extent_bit failed for enomem reasons, we can't allow the
 	 * release to continue.
Signature · 1 line
2.45.2
#6 lore raw

PATCH v2 2/4 btrfs: use clear_extent_bits() at chunk_map_device_clear_bits()

In reply to [email protected]

From: Filipe Manana <[email protected]>

Instead of using __clear_extent_bit() we can use clear_extent_bits() since we pass a NULL value for the cached and changeset arguments.

---
 fs/btrfs/volumes.c | 7 +++----
 1 file changed, 3 insertions(+), 4 deletions(-)
fs/btrfs/volumes.c +3 -4
index c8c21c55be53..784d5a15ef29 100644
@@ -5508,10 +5508,9 @@ static void chunk_map_device_clear_bits(struct btrfs_chunk_map *map, unsigned in
 		struct btrfs_io_stripe *stripe = &map->stripes[i];
 		struct btrfs_device *device = stripe->dev;
 
-		__clear_extent_bit(&device->alloc_state, stripe->physical,
-				   stripe->physical + map->stripe_size - 1,
-				   bits | EXTENT_NOWAIT,
-				   NULL, NULL);
+		clear_extent_bits(&device->alloc_state, stripe->physical,
+				  stripe->physical + map->stripe_size - 1,
+				  bits | EXTENT_NOWAIT);
 	}
 }
 
Signature · 1 line
2.45.2
#7 lore raw

PATCH v2 3/4 btrfs: use clear_extent_bits() instead of clear_extent_bit() where possible

In reply to [email protected]

From: Filipe Manana <[email protected]>

Several places are using clear_extent_bit() and passing a NULL value for the 'cached' argument, which is pointless as they can use instead clear_extent_bits().

---
 fs/btrfs/inode.c             |  3 +--
 fs/btrfs/reflink.c           |  5 ++---
 fs/btrfs/tests/inode-tests.c | 24 ++++++++++++------------
 3 files changed, 15 insertions(+), 17 deletions(-)
fs/btrfs/inode.c +1 -2
index 95d29b9282ba..836dca07a06f 100644
@@ -1753,8 +1753,7 @@ static int fallback_to_cow(struct btrfs_inode *inode,
 		spin_unlock(&sinfo->lock);
 
 		if (count > 0)
-			clear_extent_bit(io_tree, start, end, EXTENT_NORESERVE,
-					 NULL);
+			clear_extent_bits(io_tree, start, end, EXTENT_NORESERVE);
 	}
 	unlock_extent(io_tree, start, end, &cached_state);
 
fs/btrfs/reflink.c +2 -3
index 15c296cb4dac..09e40a2fdcf0 100644
@@ -95,9 +95,8 @@ static int copy_inline_to_page(struct btrfs_inode *inode,
 	if (ret < 0)
 		goto out_unlock;
 
-	clear_extent_bit(&inode->io_tree, file_offset, range_end,
-			 EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG,
-			 NULL);
+	clear_extent_bits(&inode->io_tree, file_offset, range_end,
+			  EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG);
 	ret = btrfs_set_extent_delalloc(inode, file_offset, range_end, 0, NULL);
 	if (ret)
 		goto out_unlock;
fs/btrfs/tests/inode-tests.c +12 -12
index 8142a84129b6..09ba6e279d24 100644
@@ -949,10 +949,10 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	}
 
 	/* [BTRFS_MAX_EXTENT_SIZE/2][sectorsize HOLE][the rest] */
-	ret = clear_extent_bit(&BTRFS_I(inode)->io_tree,
-			       BTRFS_MAX_EXTENT_SIZE >> 1,
-			       (BTRFS_MAX_EXTENT_SIZE >> 1) + sectorsize - 1,
-			       EXTENT_DELALLOC | EXTENT_DELALLOC_NEW, NULL);
+	ret = clear_extent_bits(&BTRFS_I(inode)->io_tree,
+				BTRFS_MAX_EXTENT_SIZE >> 1,
+				(BTRFS_MAX_EXTENT_SIZE >> 1) + sectorsize - 1,
+				EXTENT_DELALLOC | EXTENT_DELALLOC_NEW);
 	if (ret) {
 		test_err("clear_extent_bit returned %d", ret);
 		goto out;
@@ -1016,10 +1016,10 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	}
 
 	/* [BTRFS_MAX_EXTENT_SIZE+4k][4K HOLE][BTRFS_MAX_EXTENT_SIZE+4k] */
-	ret = clear_extent_bit(&BTRFS_I(inode)->io_tree,
-			       BTRFS_MAX_EXTENT_SIZE + sectorsize,
-			       BTRFS_MAX_EXTENT_SIZE + 2 * sectorsize - 1,
-			       EXTENT_DELALLOC | EXTENT_DELALLOC_NEW, NULL);
+	ret = clear_extent_bits(&BTRFS_I(inode)->io_tree,
+				BTRFS_MAX_EXTENT_SIZE + sectorsize,
+				BTRFS_MAX_EXTENT_SIZE + 2 * sectorsize - 1,
+				EXTENT_DELALLOC | EXTENT_DELALLOC_NEW);
 	if (ret) {
 		test_err("clear_extent_bit returned %d", ret);
 		goto out;
@@ -1050,8 +1050,8 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	}
 
 	/* Empty */
-	ret = clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, (u64)-1,
-			       EXTENT_DELALLOC | EXTENT_DELALLOC_NEW, NULL);
+	ret = clear_extent_bits(&BTRFS_I(inode)->io_tree, 0, (u64)-1,
+				EXTENT_DELALLOC | EXTENT_DELALLOC_NEW);
 	if (ret) {
 		test_err("clear_extent_bit returned %d", ret);
 		goto out;
@@ -1065,8 +1065,8 @@ static int test_extent_accounting(u32 sectorsize, u32 nodesize)
 	ret = 0;
 out:
 	if (ret)
-		clear_extent_bit(&BTRFS_I(inode)->io_tree, 0, (u64)-1,
-				 EXTENT_DELALLOC | EXTENT_DELALLOC_NEW, NULL);
+		clear_extent_bits(&BTRFS_I(inode)->io_tree, 0, (u64)-1,
+				  EXTENT_DELALLOC | EXTENT_DELALLOC_NEW);
 	iput(inode);
 	btrfs_free_dummy_root(root);
 	btrfs_free_dummy_fs_info(fs_info);
Signature · 1 line
2.45.2
#8 lore raw

PATCH v2 4/4 btrfs: simplify last record detection at test_range_bit_exists()

In reply to [email protected]

From: Filipe Manana <[email protected]>

Instead of keeping track of the minimum start offset of the next record and detecting overflow everytime we update that offset to be the sum of current record's end offset plus one, we can simply exit when the current record ends at or beyond our end offset and forget about updating the start offset on every iteration and testing for it at the top of the loop. This makes both the source code and assembly code simpler, more efficient and shorter (reducing the object text size).

Also remove the pointless initializion to NULL of the state variable, as we don't use it before the first assignment to it. This may help avoid some warnings with clang tools such as the one reported/fixed by commit 966de47ff0c9 ("btrfs: remove redundant initialization of variables in log_new_ancestors").

---
 fs/btrfs/extent-io-tree.c | 8 +++-----
 1 file changed, 3 insertions(+), 5 deletions(-)
fs/btrfs/extent-io-tree.c +3 -5
index 7ae24a533404..293cb354259d 100644
@@ -1726,14 +1726,14 @@ u64 count_range_bits(struct extent_io_tree *tree,
  */
 bool test_range_bit_exists(struct extent_io_tree *tree, u64 start, u64 end, u32 bit)
 {
-	struct extent_state *state = NULL;
+	struct extent_state *state;
 	bool bitset = false;
 
 	ASSERT(is_power_of_2(bit));
 
 	spin_lock(&tree->lock);
 	state = tree_search(tree, start);
-	while (state && start <= end) {
+	while (state) {
 		if (state->start > end)
 			break;
 
@@ -1742,9 +1742,7 @@ bool test_range_bit_exists(struct extent_io_tree *tree, u64 start, u64 end, u32
 			break;
 		}
 
-		/* If state->end is (u64)-1, start will overflow to 0 */
-		start = state->end + 1;
-		if (start > end || start == 0)
+		if (state->end >= end)
 			break;
 		state = next_state(state);
 	}
Signature · 1 line
2.45.2
Reviewed-by on the v2 series
#9 lore raw
In reply to [email protected]
Quoted 4 lines · On Tue, Apr 01, 2025 at 04:29:50PM +0100, [email protected] wrote:

From: Filipe Manana <[email protected]>

Some simple cleanups related to io trees that are very trivial and were initally part of a larger patchset. Details in the change logs.

Quoted 20 lines

V2: Added patch 3/4.

Filipe Manana (4):
btrfs: use clear_extent_bit() at try_release_extent_state()
btrfs: use clear_extent_bits() at chunk_map_device_clear_bits()
btrfs: use clear_extent_bits() instead of clear_extent_bit() where possible
btrfs: simplify last record detection at test_range_bit_exists()

 fs/btrfs/extent-io-tree.c    |  8 +++-----
 fs/btrfs/extent_io.c         |  2 +-
 fs/btrfs/inode.c             |  3 +--
 fs/btrfs/reflink.c           |  5 ++---
 fs/btrfs/tests/inode-tests.c | 24 ++++++++++++------------
 fs/btrfs/volumes.c           |  7 +++----
 6 files changed, 22 insertions(+), 27 deletions(-)
Signature · 1 line
2.45.2
Reviewed-by on the v2 series
#10 lore raw
In reply to [email protected]
Quoted 4 lines · 在 2025/4/2 01:59, [email protected] 写道:

From: Filipe Manana <[email protected]>

Some simple cleanups related to io trees that are very trivial and were initally part of a larger patchset. Details in the change logs.

My bad, didn't notice the v2 series.

Still looks good.

Thanks,
Qu>

Quoted 16 lines

V2: Added patch 3/4.

Filipe Manana (4):
btrfs: use clear_extent_bit() at try_release_extent_state()
btrfs: use clear_extent_bits() at chunk_map_device_clear_bits()
btrfs: use clear_extent_bits() instead of clear_extent_bit() where possible
btrfs: simplify last record detection at test_range_bit_exists()

  fs/btrfs/extent-io-tree.c    |  8 +++-----
  fs/btrfs/extent_io.c         |  2 +-
  fs/btrfs/inode.c             |  3 +--
  fs/btrfs/reflink.c           |  5 ++---
  fs/btrfs/tests/inode-tests.c | 24 ++++++++++++------------
  fs/btrfs/volumes.c           |  7 +++----
  6 files changed, 22 insertions(+), 27 deletions(-)
Reviewed-by on the v1 series
#11 lore raw
In reply to [email protected]
Quoted 4 lines · 在 2025/4/1 22:35, [email protected] 写道:

From: Filipe Manana <[email protected]>

Some simple cleanups related to io trees that are very trivial and were initally part of a larger patchset. Details in the change logs.

Thanks,
Qu

Quoted 11 lines

Filipe Manana (3):
btrfs: use clear_extent_bit() at try_release_extent_state()
btrfs: use clear_extent_bits() at chunk_map_device_clear_bits()
btrfs: simplify last record detection at test_range_bit_exists()

  fs/btrfs/extent-io-tree.c | 8 +++-----
  fs/btrfs/extent_io.c      | 2 +-
  fs/btrfs/volumes.c        | 7 +++----
  3 files changed, 7 insertions(+), 10 deletions(-)