btrfs: Fix returned errno codes

From: Luis de Bethencourt <[email protected]>

Hi,

These two patches fix instances where -1 is used to specify a buffer allocation fail, instead of using -ENOMEM.

I could merge the two patches into one if that's more appropriate.

Thanks,
Luis

Luis de Bethencourt (2):
btrfs: check-integrity: Fix returned errno codes
btrfs: reada: Fix returned errno code

 fs/btrfs/check-integrity.c | 4 ++--
 fs/btrfs/reada.c           | 2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

2.5.3

check-integrity is using -1 instead of the -ENOMEM defined macro to specify that a buffer allocation failed. Since the error number is propagated, the caller will get a -EPERM which is the wrong error condition.

Also, the smatch tool complains with the following warnings:
btrfsic_process_superblock() warn: returning -1 instead of -ENOMEM is sloppy btrfsic_read_block() warn: returning -1 instead of -ENOMEM is sloppy

---
 fs/btrfs/check-integrity.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index ce7dec8..a6257e1 100644
@@ -667,7 +667,7 @@ static int btrfsic_process_superblock(struct btrfsic_state *state,
 	selected_super = kzalloc(sizeof(*selected_super), GFP_NOFS);
 	if (NULL == selected_super) {
 		printk(KERN_INFO "btrfsic: error, kmalloc failed!\n");
-		return -1;
+		return -ENOMEM;
 	}
 
 	list_for_each_entry(device, dev_head, dev_list) {

@@ -1660,7 +1660,7 @@ static int btrfsic_read_block(struct btrfsic_state *state,
 					  sizeof(*block_ctx->pagev)) *
 					 num_pages, GFP_NOFS);
 	if (!block_ctx->mem_to_free)
-		return -1;
+		return -ENOMEM;
 	block_ctx->datav = block_ctx->mem_to_free;
 	block_ctx->pagev = (struct page **)(block_ctx->datav + num_pages);
 	for (i = 0; i < num_pages; i++) {

2.5.3

Agreed. btrfsic_process_superblock can be called from the mount path so getting EPERM would be confusing.

reada is using -1 instead of the -ENOMEM defined macro to specify that a buffer allocation failed. Since the error number is propagated, the caller will get a -EPERM which is the wrong error condition.

Smatch tool warning:
reada_add_block() warn: returning -1 instead of -ENOMEM is sloppy

---
 fs/btrfs/reada.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

index 4645cd1..5bfd3cd 100644
@@ -569,7 +569,7 @@ static int reada_add_block(struct reada_control *rc, u64 logical,
 	rec = kzalloc(sizeof(*rec), GFP_NOFS);
 	if (!rec) {
 		reada_extent_put(root->fs_info, re);
-		return -1;
+		return -ENOMEM;
 	}
 
 	rec->rc = rc;

2.5.3

No, it's better to change them separately. If you change the return value, also look at all the callers so you can see the impact of the change and count that into the "size of the patch".