redonkable
/
remarkable-linux
1
0
Fork 0
Code Releases Activity

Btrfs: keep track of max_extent_size per space_info 

When we are heavily fragmented we can induce a lot of latency trying to make an
allocation happen that is simply not going to happen.  Thankfully we keep track
of our max_extent_size when going through the allocator, so if we get to the
point where we are exiting find_free_extent with ENOSPC then set our
space_info->max_extent_size so we can keep future allocations from having to pay
this cost.  We reset the max_extent_size whenever we release pinned bytes back
into this space info so we can redo all the work.  Thanks,

Signed-off-by: Josef Bacik <jbacik@fb.com>
Signed-off-by: Chris Mason <clm@fb.com>
steinar/wifi_calib_4_9_kernel
Josef Bacik 2015-09-29 11:40:47 -04:00 committed by Chris Mason
parent 36af4e0737
commit 4f4db2174d
2 changed files with 34 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
fs/btrfs/ctree.h
View File

@ -1154,6 +1154,10 @@ struct btrfs_space_info {
delalloc/allocations */
u64 bytes_readonly; /* total bytes that are read only */
u64 max_extent_size; /* This will hold the maximum extent size of
the space info if we had an ENOSPC in the
allocator. */
unsigned int full:1; /* indicates that we cannot allocate any more
chunks for this space */
unsigned int chunk_alloc:1; /* set if we are allocating a chunk */

31
fs/btrfs/extent-tree.c
View File

@ -3810,6 +3810,7 @@ static int update_space_info(struct btrfs_fs_info *info, u64 flags,
found->bytes_readonly = 0;
found->bytes_may_use = 0;
found->full = 0;
found->max_extent_size = 0;
found->force_alloc = CHUNK_ALLOC_NO_FORCE;
found->chunk_alloc = 0;
found->flush = 0;
@ -6158,6 +6159,7 @@ static int unpin_extent_range(struct btrfs_root *root, u64 start, u64 end,
spin_lock(&cache->lock);
cache->pinned -= len;
space_info->bytes_pinned -= len;
space_info->max_extent_size = 0;
percpu_counter_add(&space_info->total_bytes_pinned, -len);
if (cache->ro) {
space_info->bytes_readonly += len;
@ -6914,6 +6916,29 @@ static noinline int find_free_extent(struct btrfs_root *orig_root,
return -ENOSPC;
}
/*
* If our free space is heavily fragmented we may not be able to make
* big contiguous allocations, so instead of doing the expensive search
* for free space, simply return ENOSPC with our max_extent_size so we
* can go ahead and search for a more manageable chunk.
*
* If our max_extent_size is large enough for our allocation simply
* disable clustering since we will likely not be able to find enough
* space to create a cluster and induce latency trying.
*/
if (unlikely(space_info->max_extent_size)) {
spin_lock(&space_info->lock);
if (space_info->max_extent_size &&
num_bytes > space_info->max_extent_size) {
ins->offset = space_info->max_extent_size;
spin_unlock(&space_info->lock);
return -ENOSPC;
} else if (space_info->max_extent_size) {
use_cluster = false;
}
spin_unlock(&space_info->lock);
}
/*
* If the space info is for both data and metadata it means we have a
* small filesystem and we can't use the clustering stuff.
@ -7287,8 +7312,12 @@ loop:
ret = 0;
}
out:
if (ret == -ENOSPC)
if (ret == -ENOSPC) {
spin_lock(&space_info->lock);
space_info->max_extent_size = max_extent_size;
spin_unlock(&space_info->lock);
ins->offset = max_extent_size;
}
return ret;
}