[PATCH][resend] Don't leak memory in failure paths of gfs2_acl_get()

[PATCH][resend] Don't leak memory in failure paths of gfs2_acl_get()

[Jesper Juhl]

In fs/gfs2/acl.c::gfs2_acl_get() we may leak memory in failure scenarios.
gfs2_xattr_acl_get() may return <=0 after having dynamically allocated 
memory for its last argument ('data' in the gfs2_acl_get() caller) and in 
that case the caller leaks the memory.
This patch initializes 'data' to NULL and calls kfree() on 'data' in the 
failure paths. This ensures that we always free the memory on failure or 
that we just call kfree(NULL) on failures where no memory has actually 
been allocated yet.

Signed-off-by: Jesper Juhl <jj@chaosbits.net>
---
 acl.c |   10 +++++++---
 1 file changed, 7 insertions(+), 3 deletions(-)

 Patch against Linus' tree as of right now.
 Compile tested only.

diff --git a/fs/gfs2/acl.c b/fs/gfs2/acl.c
index cbc0715..7d231fb 100644
--- a/fs/gfs2/acl.c
+++ b/fs/gfs2/acl.c
@@ -42,7 +42,7 @@ static struct posix_acl *gfs2_acl_get(struct gfs2_inode *ip, int type)
 {
 	struct posix_acl *acl;
 	const char *name;
-	char *data;
+	char *data = NULL;
 	int len;
 
 	if (!ip->i_eattr)
@@ -57,10 +57,14 @@ static struct posix_acl *gfs2_acl_get(struct gfs2_inode *ip, int type)
 		return ERR_PTR(-EINVAL);
 
 	len = gfs2_xattr_acl_get(ip, name, &data);
-	if (len < 0)
+	if (len < 0) {
+		kfree(data);
 		return ERR_PTR(len);
-	if (len == 0)
+	}
+	if (len == 0) {
+		kfree(data);
 		return NULL;
+	}
 
 	acl = posix_acl_from_xattr(data, len);
 	kfree(data);


-- 
Jesper Juhl <jj@chaosbits.net>       http://www.chaosbits.net/
Don't top-post http://www.catb.org/jargon/html/T/top-post.html
Plain text mails only, please.