btrfs: precalculate checksums per leaf once

btrfs_csum_bytes_to_leaves shows up in system profiles, which makes it a
candidate for optimizations. After the 64bit division has been replaced
by shift, there's still a calculation done each time the function is
called: checksums per leaf.

As this is a constant value for the entire filesystem lifetime, we
can calculate it once at mount time and reuse. This also allows to
reduce the division to 64bit/32bit as we know the constant will always
fit the 32bit type.

Replace the open-coded rounding up with a macro that internally handles
the 64bit division and as it's now a short function, make it static
inline (slight code increase, slight stack usage reduction).

Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: David Sterba <dsterba@suse.com>

3 files changed, 13 insertions, 20 deletions

diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 3d06596b8746..f28960647c01 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -938,6 +938,7 @@ struct btrfs_fs_info {
 	/* ilog2 of sectorsize, use to avoid 64bit division */
 	u32 sectorsize_bits;
 	u32 csum_size;
+	u32 csums_per_leaf;
 	u32 stripesize;
 
 	/* Block groups and devices containing active swapfiles. */
@@ -2525,7 +2526,17 @@ int btrfs_get_extent_inline_ref_type(const struct extent_buffer *eb,
 				     enum btrfs_inline_ref_type is_data);
 u64 hash_extent_data_ref(u64 root_objectid, u64 owner, u64 offset);
 
-u64 btrfs_csum_bytes_to_leaves(struct btrfs_fs_info *fs_info, u64 csum_bytes);
+/*
+ * Take the number of bytes to be checksummmed and figure out how many leaves
+ * it would require to store the csums for that many bytes.
+ */
+static inline u64 btrfs_csum_bytes_to_leaves(
+			const struct btrfs_fs_info *fs_info, u64 csum_bytes)
+{
+	const u64 num_csums = csum_bytes >> fs_info->sectorsize_bits;
+
+	return DIV_ROUND_UP_ULL(num_csums, fs_info->csums_per_leaf);
+}
 
 /*
  * Use this if we would be adding new items, as we could split nodes as we cow
diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c
index 475bd197c482..0e540ddd7d99 100644
--- a/fs/btrfs/disk-io.c
+++ b/fs/btrfs/disk-io.c
@@ -3081,6 +3081,7 @@ int __cold open_ctree(struct super_block *sb, struct btrfs_fs_devices *fs_device
 	fs_info->sectorsize = sectorsize;
 	fs_info->sectorsize_bits = ilog2(sectorsize);
 	fs_info->csum_size = btrfs_super_csum_size(disk_super);
+	fs_info->csums_per_leaf = BTRFS_MAX_ITEM_SIZE(fs_info) / fs_info->csum_size;
 	fs_info->stripesize = stripesize;
 
 	/*
diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c
index 14d5aaa28399..a27caa47aa62 100644
--- a/fs/btrfs/extent-tree.c
+++ b/fs/btrfs/extent-tree.c
@@ -2133,25 +2133,6 @@ static u64 find_middle(struct rb_root *root)
 #endif
 
 /*
- * Takes the number of bytes to be csumm'ed and figures out how many leaves it
- * would require to store the csums for that many bytes.
- */
-u64 btrfs_csum_bytes_to_leaves(struct btrfs_fs_info *fs_info, u64 csum_bytes)
-{
-	u64 csum_size;
-	u64 num_csums_per_leaf;
-	u64 num_csums;
-
-	csum_size = BTRFS_MAX_ITEM_SIZE(fs_info);
-	num_csums_per_leaf = div64_u64(csum_size,
-			(u64)btrfs_super_csum_size(fs_info->super_copy));
-	num_csums = csum_bytes >> fs_info->sectorsize_bits;
-	num_csums += num_csums_per_leaf - 1;
-	num_csums = div64_u64(num_csums, num_csums_per_leaf);
-	return num_csums;
-}
-
-/*
  * this starts processing the delayed reference count updates and
  * extent insertions we have queued up so far.  count can be
  * 0, which means to process everything in the tree at the start