mkfs: collapse redundant logic in custom_alloc_extent()

RFC PATCH In review Eric Sandeen 2 participants 2 messages Jan 25, 2013 – Jan 29, 2013 Btrfs file system

lore

Share & tools

lore

Eric Sandeen

author

2013-01-25 15:57 UTC #0 lore raw

RFC PATCH 1/1 mkfs: collapse redundant logic in custom_alloc_extent()

It looks to me like the logic in these two if statements are overlapping.

The test for flags & BTRFS_BLOCK_GROUP_SYSTEM in the 2nd case should never get triggered, because it would have triggered on the first case, right?

And since the actions are identical in both cases, this can be collapsed into one.

Signed-off-by Eric Sandeen <[email protected]>

---

p.s.

Having done that, I now look at the nearly identical custom_alloc_extent() copy in convert.c, and wonder if it's intentional that the convert copy does not care about BTRFS_BLOCK_GROUP_METADATA, but mkfs does? I'm not quite sure what's going on there.

Thanks,
-Eric

mkfs.c +2 -6

index ca850d9..5d77428 100644
@@ -635,14 +635,10 @@ static int custom_alloc_extent(struct btrfs_root *root, u64 num_bytes,
 
 		cache = btrfs_lookup_block_group(root->fs_info, start);
 		BUG_ON(!cache);
-		if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM ||
-		    last > cache->key.objectid + cache->key.offset) {
-			last = cache->key.objectid + cache->key.offset;
-			continue;
-		}
 
 		if (cache->flags & (BTRFS_BLOCK_GROUP_SYSTEM |
-			    BTRFS_BLOCK_GROUP_METADATA)) {
+			    BTRFS_BLOCK_GROUP_METADATA) ||
+		    last > cache->key.objectid + cache->key.offset) {
 			last = cache->key.objectid + cache->key.offset;
 			continue;
 		}

Alex Lyakas

2013-01-29 13:12 UTC #1 lore raw

Hi Eric,

Quoted 13 lines · Eric Sandeen

It looks to me like the logic in these two if statements are overlapping.
The test for flags & BTRFS_BLOCK_GROUP_SYSTEM in the 2nd case should never get triggered, because it would have triggered on the first case, right?
And since the actions are identical in both cases, this can be collapsed into one.
Signed-off-by Eric Sandeen <[email protected]>
---

Quoted 7 lines

p.s.
Having done that, I now look at the nearly identical custom_alloc_extent() copy in convert.c, and wonder if it's intentional that the convert copy does not care about BTRFS_BLOCK_GROUP_METADATA, but mkfs does? I'm not quite sure what's going on there.

When I was digging in the conversion code, I saw that btrfs_make_block_groups() uses some heuristics to define some block groups as DATA and some as METADATA. But later, the conversion code, as you noticed, doesn't care about this, and there are data EXTENT_ITEMs that land in METADATA block groups. I am not sure if this is a problem or not, I asked here:
http://www.spinics.net/lists/linux-btrfs/msg19894.html, but got no answers:(
Since then I tried to develop my own version of convert that lays out block groups more properly, but it makes some assumptions on the free space on the block device being converted.

Alex.

Quoted 30 lines

Thanks,

(quoted diff) +0 -1

-Eric

mkfs.c +2 -7

index ca850d9..5d77428 100644
@@ -635,14 +635,10 @@ static int custom_alloc_extent(struct btrfs_root *root, u64 num_bytes,
 
                cache = btrfs_lookup_block_group(root->fs_info, start);
                BUG_ON(!cache);
-               if (cache->flags & BTRFS_BLOCK_GROUP_SYSTEM ||
-                   last > cache->key.objectid + cache->key.offset) {
-                       last = cache->key.objectid + cache->key.offset;
-                       continue;
-               }
 
                if (cache->flags & (BTRFS_BLOCK_GROUP_SYSTEM |
-                           BTRFS_BLOCK_GROUP_METADATA)) {
+                           BTRFS_BLOCK_GROUP_METADATA) ||
+                   last > cache->key.objectid + cache->key.offset) {
                        last = cache->key.objectid + cache->key.offset;
                        continue;
                }
 
--

To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html