[PATCH 0/2] Page and folio count helper cleanups

[PATCH 0/2] Page and folio count helper cleanups

[David Sterba]

Adding the __pure attribute to num_extent_folios() allows some
optimizations that can reduce stack space usage.

David Sterba (2):
  btrfs: add __pure attribute to eb page and folio counters
  btrfs: use num_extent_folios() in for loop bounds

 fs/btrfs/disk-io.c   |  3 +--
 fs/btrfs/extent_io.c | 48 +++++++++++++++++---------------------------
 fs/btrfs/extent_io.h |  7 +++++--
 3 files changed, 24 insertions(+), 34 deletions(-)

-- 
2.47.1

[PATCH 1/2] btrfs: add __pure attribute to eb page and folio counters

[David Sterba]

The functions qualify for the pure attribute as they always return the
same value for the same argument (in the given scope). This allows to
optimize the calls and cache the value.

Signed-off-by: David Sterba <dsterba@suse.com>
---
 fs/btrfs/extent_io.h | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h
index 6c5328bfabc2..6440ba7cf35e 100644
--- a/fs/btrfs/extent_io.h
+++ b/fs/btrfs/extent_io.h
@@ -276,7 +276,8 @@ void btrfs_readahead_tree_block(struct btrfs_fs_info *fs_info,
 				u64 bytenr, u64 owner_root, u64 gen, int level);
 void btrfs_readahead_node_child(struct extent_buffer *node, int slot);
 
-static inline int num_extent_pages(const struct extent_buffer *eb)
+/* Note: this can be used in for loops without caching the value in a variable. */
+static inline int __pure num_extent_pages(const struct extent_buffer *eb)
 {
 	/*
 	 * For sectorsize == PAGE_SIZE case, since nodesize is always aligned to
@@ -294,8 +295,10 @@ static inline int num_extent_pages(const struct extent_buffer *eb)
  * As we can have either one large folio covering the whole eb
  * (either nodesize <= PAGE_SIZE, or high order folio), or multiple
  * single-paged folios.
+ *
+ * Note: this can be used in for loops without caching the value in a variable.
  */
-static inline int num_extent_folios(const struct extent_buffer *eb)
+static inline int __pure num_extent_folios(const struct extent_buffer *eb)
 {
 	if (folio_order(eb->folios[0]))
 		return 1;
-- 
2.47.1

[PATCH 2/2] btrfs: use num_extent_folios() in for loop bounds

[David Sterba]

As the helper num_extent_folios() is now __pure, we can use it in for
loop without storing its value in a variable explicitly, the compiler
will do this for us.

The effects on btrfs.ko is -200 bytes and there are stack space savings
too:

btrfs_clone_extent_buffer                               -8 (32 -> 24)
btrfs_clear_buffer_dirty                                -8 (48 -> 40)
clear_extent_buffer_uptodate                            -8 (40 -> 32)
set_extent_buffer_dirty                                 -8 (32 -> 24)
write_one_eb                                            -8 (88 -> 80)
set_extent_buffer_uptodate                              -8 (40 -> 32)
read_extent_buffer_pages_nowait                        -16 (64 -> 48)
find_extent_buffer                                      -8 (32 -> 24)

Signed-off-by: David Sterba <dsterba@suse.com>
---
 fs/btrfs/disk-io.c   |  3 +--
 fs/btrfs/extent_io.c | 48 +++++++++++++++++---------------------------
 2 files changed, 19 insertions(+), 32 deletions(-)

diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index a799216aa264..91c7da1ced7d 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -182,13 +182,12 @@ static int btrfs_repair_eb_io_failure(const struct extent_buffer *eb,
 				      int mirror_num)
 {
 	struct btrfs_fs_info *fs_info = eb->fs_info;
-	int num_folios = num_extent_folios(eb);
 	int ret = 0;
 
 	if (sb_rdonly(fs_info->sb))
 		return -EROFS;
 
-	for (int i = 0; i < num_folios; i++) {
+	for (int i = 0; i < num_extent_folios(eb); i++) {
 		struct folio *folio = eb->folios[i];
 		u64 start = max_t(u64, eb->start, folio_pos(folio));
 		u64 end = min_t(u64, eb->start + eb->len,
diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c
index 7b0aa332aedc..7fa4ce5d9ead 100644
--- a/fs/btrfs/extent_io.c
+++ b/fs/btrfs/extent_io.c
@@ -1788,7 +1788,6 @@ static noinline_for_stack void write_one_eb(struct extent_buffer *eb,
 {
 	struct btrfs_fs_info *fs_info = eb->fs_info;
 	struct btrfs_bio *bbio;
-	const int num_folios = num_extent_folios(eb);
 
 	prepare_eb_write(eb);
 
@@ -1800,7 +1799,7 @@ static noinline_for_stack void write_one_eb(struct extent_buffer *eb,
 	wbc_init_bio(wbc, &bbio->bio);
 	bbio->inode = BTRFS_I(eb->fs_info->btree_inode);
 	bbio->file_offset = eb->start;
-	for (int i = 0; i < num_folios; i++) {
+	for (int i = 0; i < num_extent_folios(eb); i++) {
 		struct folio *folio = eb->folios[i];
 		u64 range_start = max_t(u64, eb->start, folio_pos(folio));
 		u32 range_len = min_t(u64, folio_pos(folio) + folio_size(folio),
@@ -2677,7 +2676,6 @@ __alloc_extent_buffer(struct btrfs_fs_info *fs_info, u64 start,
 struct extent_buffer *btrfs_clone_extent_buffer(const struct extent_buffer *src)
 {
 	struct extent_buffer *new;
-	int num_folios = num_extent_folios(src);
 	int ret;
 
 	new = __alloc_extent_buffer(src->fs_info, src->start, src->len);
@@ -2697,7 +2695,7 @@ struct extent_buffer *btrfs_clone_extent_buffer(const struct extent_buffer *src)
 		return NULL;
 	}
 
-	for (int i = 0; i < num_folios; i++) {
+	for (int i = 0; i < num_extent_folios(src); i++) {
 		struct folio *folio = new->folios[i];
 
 		ret = attach_extent_buffer_folio(new, folio, NULL);
@@ -2717,7 +2715,6 @@ struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
 						  u64 start, unsigned long len)
 {
 	struct extent_buffer *eb;
-	int num_folios = 0;
 	int ret;
 
 	eb = __alloc_extent_buffer(fs_info, start, len);
@@ -2726,13 +2723,12 @@ struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
 
 	ret = alloc_eb_folio_array(eb, false);
 	if (ret)
-		goto err;
+		goto out;
 
-	num_folios = num_extent_folios(eb);
-	for (int i = 0; i < num_folios; i++) {
+	for (int i = 0; i < num_extent_folios(eb); i++) {
 		ret = attach_extent_buffer_folio(eb, eb->folios[i], NULL);
 		if (ret < 0)
-			goto err;
+			goto out_detach;
 	}
 
 	set_extent_buffer_uptodate(eb);
@@ -2740,13 +2736,15 @@ struct extent_buffer *__alloc_dummy_extent_buffer(struct btrfs_fs_info *fs_info,
 	set_bit(EXTENT_BUFFER_UNMAPPED, &eb->bflags);
 
 	return eb;
-err:
-	for (int i = 0; i < num_folios; i++) {
+
+out_detach:
+	for (int i = 0; i < num_extent_folios(eb); i++) {
 		if (eb->folios[i]) {
 			detach_extent_buffer_folio(eb, eb->folios[i]);
 			folio_put(eb->folios[i]);
 		}
 	}
+out:
 	kmem_cache_free(extent_buffer_cache, eb);
 	return NULL;
 }
@@ -2795,11 +2793,9 @@ static void check_buffer_tree_ref(struct extent_buffer *eb)
 
 static void mark_extent_buffer_accessed(struct extent_buffer *eb)
 {
-	int num_folios= num_extent_folios(eb);
-
 	check_buffer_tree_ref(eb);
 
-	for (int i = 0; i < num_folios; i++)
+	for (int i = 0; i < num_extent_folios(eb); i++)
 		folio_mark_accessed(eb->folios[i]);
 }
 
@@ -3032,7 +3028,6 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
 					  u64 start, u64 owner_root, int level)
 {
 	unsigned long len = fs_info->nodesize;
-	int num_folios;
 	int attached = 0;
 	struct extent_buffer *eb;
 	struct extent_buffer *existing_eb = NULL;
@@ -3096,9 +3091,8 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
 		goto out;
 	}
 
-	num_folios = num_extent_folios(eb);
 	/* Attach all pages to the filemap. */
-	for (int i = 0; i < num_folios; i++) {
+	for (int i = 0; i < num_extent_folios(eb); i++) {
 		struct folio *folio;
 
 		ret = attach_eb_folio_to_filemap(eb, i, prealloc, &existing_eb);
@@ -3192,7 +3186,7 @@ struct extent_buffer *alloc_extent_buffer(struct btrfs_fs_info *fs_info,
 	 * btree_release_folio will correctly detect that a page belongs to a
 	 * live buffer and won't free them prematurely.
 	 */
-	for (int i = 0; i < num_folios; i++)
+	for (int i = 0; i < num_extent_folios(eb); i++)
 		folio_unlock(eb->folios[i]);
 	return eb;
 
@@ -3338,7 +3332,6 @@ void btrfs_clear_buffer_dirty(struct btrfs_trans_handle *trans,
 			      struct extent_buffer *eb)
 {
 	struct btrfs_fs_info *fs_info = eb->fs_info;
-	const int num_folios = num_extent_folios(eb);
 
 	btrfs_assert_tree_write_locked(eb);
 
@@ -3365,7 +3358,7 @@ void btrfs_clear_buffer_dirty(struct btrfs_trans_handle *trans,
 	percpu_counter_add_batch(&fs_info->dirty_metadata_bytes, -eb->len,
 				 fs_info->dirty_metadata_batch);
 
-	for (int i = 0; i < num_folios; i++) {
+	for (int i = 0; i < num_extent_folios(eb); i++) {
 		struct folio *folio = eb->folios[i];
 		bool last;
 
@@ -3383,14 +3376,12 @@ void btrfs_clear_buffer_dirty(struct btrfs_trans_handle *trans,
 
 void set_extent_buffer_dirty(struct extent_buffer *eb)
 {
-	int num_folios;
 	bool was_dirty;
 
 	check_buffer_tree_ref(eb);
 
 	was_dirty = test_and_set_bit(EXTENT_BUFFER_DIRTY, &eb->bflags);
 
-	num_folios = num_extent_folios(eb);
 	WARN_ON(atomic_read(&eb->refs) == 0);
 	WARN_ON(!test_bit(EXTENT_BUFFER_TREE_REF, &eb->bflags));
 	WARN_ON(test_bit(EXTENT_BUFFER_ZONED_ZEROOUT, &eb->bflags));
@@ -3411,7 +3402,7 @@ void set_extent_buffer_dirty(struct extent_buffer *eb)
 		 */
 		if (subpage)
 			folio_lock(eb->folios[0]);
-		for (int i = 0; i < num_folios; i++)
+		for (int i = 0; i < num_extent_folios(eb); i++)
 			btrfs_meta_folio_set_dirty(eb->fs_info, eb->folios[i],
 						   eb->start, eb->len);
 		if (subpage)
@@ -3421,7 +3412,7 @@ void set_extent_buffer_dirty(struct extent_buffer *eb)
 					 eb->fs_info->dirty_metadata_batch);
 	}
 #ifdef CONFIG_BTRFS_DEBUG
-	for (int i = 0; i < num_folios; i++)
+	for (int i = 0; i < num_extent_folios(eb); i++)
 		ASSERT(folio_test_dirty(eb->folios[i]));
 #endif
 }
@@ -3429,10 +3420,9 @@ void set_extent_buffer_dirty(struct extent_buffer *eb)
 void clear_extent_buffer_uptodate(struct extent_buffer *eb)
 {
 	struct btrfs_fs_info *fs_info = eb->fs_info;
-	int num_folios = num_extent_folios(eb);
 
 	clear_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
-	for (int i = 0; i < num_folios; i++) {
+	for (int i = 0; i < num_extent_folios(eb); i++) {
 		struct folio *folio = eb->folios[i];
 
 		if (!folio)
@@ -3445,10 +3435,9 @@ void clear_extent_buffer_uptodate(struct extent_buffer *eb)
 void set_extent_buffer_uptodate(struct extent_buffer *eb)
 {
 	struct btrfs_fs_info *fs_info = eb->fs_info;
-	int num_folios = num_extent_folios(eb);
 
 	set_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags);
-	for (int i = 0; i < num_folios; i++) {
+	for (int i = 0; i < num_extent_folios(eb); i++) {
 		struct folio *folio = eb->folios[i];
 
 		btrfs_meta_folio_set_uptodate(fs_info, folio, eb->start, eb->len);
@@ -3496,7 +3485,6 @@ static void end_bbio_meta_read(struct btrfs_bio *bbio)
 int read_extent_buffer_pages_nowait(struct extent_buffer *eb, int mirror_num,
 				    const struct btrfs_tree_parent_check *check)
 {
-	const int num_folios = num_extent_folios(eb);
 	struct btrfs_bio *bbio;
 
 	if (test_bit(EXTENT_BUFFER_UPTODATE, &eb->bflags))
@@ -3537,7 +3525,7 @@ int read_extent_buffer_pages_nowait(struct extent_buffer *eb, int mirror_num,
 	bbio->inode = BTRFS_I(eb->fs_info->btree_inode);
 	bbio->file_offset = eb->start;
 	memcpy(&bbio->parent_check, check, sizeof(*check));
-	for (int i = 0; i < num_folios; i++) {
+	for (int i = 0; i < num_extent_folios(eb); i++) {
 		struct folio *folio = eb->folios[i];
 		u64 range_start = max_t(u64, eb->start, folio_pos(folio));
 		u32 range_len = min_t(u64, folio_pos(folio) + folio_size(folio),
-- 
2.47.1