Re: [PATCH 3/8] dm: prevent possible buffer overflow in ioctl interface

[Kevin Corry <corryk@us.ibm.com>]

[Sent this earlier, but it doesn't seem to have shown up yet. My outgoing 
email server seems to flake out at times. Sorry if this is a duplicate, but 
it seemed important enough to resend, since we'd like to avoid the extra 
compile failure.]


On Thursday 27 February 2003 10:25, Roland Dreier wrote:
>    > +         char *name = kmalloc(DM_NAME_LEN + strlen(DM_DIR) + 1);
>    > +         if (!name) {
>    > +                 return -ENOMEM;
>    > +         }
>
> Also, kmalloc() needs a second "GFP_xxx" parameter (I guess GFP_KERNEL
> in this case, although I don't know the context this function is
> called from).
>
>  - Roland


Dammit! I'm not having a good morning. :(

Unfortunately, Linus seems to have committed that patch already. So here is a 
patch to fix just that line.

Thanks for catching that.

-- 
Kevin Corry
corryk@us.ibm.com
http://evms.sourceforge.net/


--- a/drivers/md/dm-ioctl.c	2003/02/27 16:29:58
+++ b/drivers/md/dm-ioctl.c	2003/02/27 17:21:54
@@ -174,7 +174,7 @@
 static int register_with_devfs(struct hash_cell *hc)
 {
 	struct gendisk *disk = dm_disk(hc->md);
-	char *name = kmalloc(DM_NAME_LEN + strlen(DM_DIR) + 1);
+	char *name = kmalloc(DM_NAME_LEN + strlen(DM_DIR) + 1, GFP_KERNEL);
 	if (!name) {
 		return -ENOMEM;
 	}