Reduce size of struct btrfs_path from 144 to 96 bytes and save one
cacheline. This will allow packing of 42 objects into a PAGE_SIZE slab,
(previously 28, measured on x86_64).
Let 'slots' hold at most 65536, which is enough to index items for
reasonably big blocksizes (4k, 64k).
Maximum is 2M: 65536 * 33 (sizeof struct btrfs_key_ptr)
(Making it u8 would allow only 256 * 33 = 8448 ie 8k blocksize)
A 'locks' item holds at most 5 values of lock type.
'lowest_level': maximum level is 8
The 'reada' is set to {-1,0,1,2}, has to be signed.
Signed-off-by: David Sterba <dsterba@xxxxxxx>
---
fs/btrfs/ctree.h | 8 ++++----
1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h
index 04a5dfc..0b5bd2c 100644
--- a/fs/btrfs/ctree.h
+++ b/fs/btrfs/ctree.h
@@ -514,12 +514,12 @@ struct btrfs_node {
*/
struct btrfs_path {
struct extent_buffer *nodes[BTRFS_MAX_LEVEL];
- int slots[BTRFS_MAX_LEVEL];
+ u16 slots[BTRFS_MAX_LEVEL];
/* if there is real range locking, this locks field will change */
- int locks[BTRFS_MAX_LEVEL];
- int reada;
+ u8 locks[BTRFS_MAX_LEVEL];
/* keep some upper locks as we walk down */
- int lowest_level;
+ u8 lowest_level;
+ s8 reada; /* sign represents direction */
/*
* set by btrfs_split_item, tells search_slot to keep all locks
--
1.7.6
--
To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
