>> Since md is freed before the do-while checks it,
>> it's better to change it to while(1).
>
> This actually isn't a bug, since there is no problem checking a
> pointer that has been freed; its only *dereferencing* a pointer which
> is bad. That being said, md is never NULL at the end of the loop,
> since in the middle of the loop is the only break condition:
>
> if (md == NULL)
> break;
>
> So the patch saves a tiny amount of compiled code, but it isn't really
> a fix in any way.
>
> That being said, if we're going to make this sort of change, my
> preference would be to use the more common C idiom:
>
> while (1) {
> ...
> }
>
> as opposed to
>
> do {
> ...
> } while (1);
>
> The former makes it quite clear that any exit from the loop is not
> going to be coming from loop construct itself, but from any embedded
> break statements inside the loop construct
>
Yes. You are right. Revise the patch as you suggested.
Signed-off-by: Shen Feng <shen@xxxxxxxxxxxxxx>
---
fs/ext4/mballoc.c | 4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/fs/ext4/mballoc.c b/fs/ext4/mballoc.c
index c9900aa..bd6cf22 100644
--- a/fs/ext4/mballoc.c
+++ b/fs/ext4/mballoc.c
@@ -2521,7 +2521,7 @@ ext4_mb_free_committed_blocks(struct super_block *sb)
return;
/* there is committed blocks to be freed yet */
- do {
+ while (1) {
/* get next array of blocks */
md = NULL;
spin_lock(&sbi->s_md_lock);
@@ -2561,7 +2561,7 @@ ext4_mb_free_committed_blocks(struct super_block *sb)
kfree(md);
ext4_mb_release_desc(&e4b);
- } while (md);
+ }
mb_debug("freed %u blocks in %u structures\n", count, count2);
}
-- 1.5.4.5
--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
