Skip to contentSkip to conversation

ext4: extents: Remove unnecessary ‘NULL’ values from ablocks

PATCH In review Li zeming 3 participants 3 messages Apr 2, 2024 – Apr 8, 2024 ext4 file system

lore
Share & tools
lore

Li zeming

author
#0 lore raw

PATCH 1/1 ext4: extents: Remove unnecessary ‘NULL’ values from ablocks

ablocks is assigned first, so it does not need to initialize the assignment.

---
 fs/ext4/extents.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
fs/ext4/extents.c +1 -1
index 4ab96f01a6f31..caace8c3fd3c1 100644
@@ -1061,7 +1061,7 @@ static int ext4_ext_split(handle_t *handle, struct inode *inode,
 	int i = at, k, m, a;
 	ext4_fsblk_t newblock, oldblock;
 	__le32 border;
-	ext4_fsblk_t *ablocks = NULL; /* array of allocated blocks */
+	ext4_fsblk_t *ablocks; /* array of allocated blocks */
 	gfp_t gfp_flags = GFP_NOFS;
 	int err = 0;
 	size_t ext_size = 0;
Signature · 1 line
2.18.2
maintainer
#1 lore raw
Quoted 2 lines · Li zeming

ablocks is assigned first, so it does not need to initialize the assignment.

That's technically true, but the compiler is perfectly capable of optimizing it out. So it's harmless, and removing it does make the code a bit more fragile, since it needs to be set so that the cleanup code doesn't accidentally dereference an uninitialized pointer.

Cheers,

- Ted

#2 lore raw
In reply to Li zeming
Quoted 3 lines · Li zeming

ablocks is assigned first, so it does not need to initialize the assignment.

While it is true that "ablocks" is currently set before use, this is happening a long way away from the variable declaration and also "ablocks" is used after the "cleanup:" label error case:

cleanup:

        if (bh) {
                if (buffer_locked(bh))
                        unlock_buffer(bh);
                brelse(bh);
        }

        if (err) {
                /* free all allocated blocks in error case */
                for (i = 0; i < depth; i++) {
                        if (!ablocks[i])
                                continue;
                        ext4_free_blocks(handle, inode, NULL, ablocks[i], 1,
                                         EXT4_FREE_BLOCKS_METADATA);
                }
        }
        kfree(ablocks);

So there is definitely a risk that a code change in the future would introduce hard-to-debug problems, crashes, or even just spurious static code analysis warnings. My recommendation would be to keep this 1-cycle local variable initialization in place rather than spend hours or days trying to debug and fix a crash here in the future.

Cheers, Andreas

Quoted 20 lines

---
fs/ext4/extents.c | 2 +-

1 file changed, 1 insertion(+), 1 deletion(-)
fs/ext4/extents.c +1 -2
index 4ab96f01a6f31..caace8c3fd3c1 100644
@@ -1061,7 +1061,7 @@ static int ext4_ext_split(handle_t *handle, struct inode *inode,
 	int i = at, k, m, a;
 	ext4_fsblk_t newblock, oldblock;
 	__le32 border;
-	ext4_fsblk_t *ablocks = NULL; /* array of allocated blocks */
+	ext4_fsblk_t *ablocks; /* array of allocated blocks */
 	gfp_t gfp_flags = GFP_NOFS;
 	int err = 0;
 	size_t ext_size = 0;
--
Quoted 2 lines

2.18.2

Cheers, Andreas