[PATCH 2.6.28] bluetooth: fix leak of uninitialized data to userspace

[PATCH 2.6.28] bluetooth: fix leak of uninitialized data to userspace

[Vegard Nossum]

>From 45be27894a18f87b71b855fcc4afd50f860254b4 Mon Sep 17 00:00:00 2001
From: Vegard Nossum <vegard.nossum@gmail.com>
Date: Sun, 5 Oct 2008 17:25:45 +0200
Subject: [PATCH] bluetooth: fix leak of uninitialized data to userspace

    struct hci_dev_list_req {
            __u16  dev_num;
            struct hci_dev_req dev_req[0];  /* hci_dev_req structures */
    };

sizeof(struct hci_dev_list_req) == 4, so the two bytes immediately
following "dev_num" will never be initialized. When this structure
is copied to userspace, these uninitialized bytes are leaked.

Fix by using kzalloc() instead of kmalloc(). Found using kmemcheck.

Signed-off-by: Vegard Nossum <vegard.nossum@gmail.com>
---
 net/bluetooth/hci_core.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
index 278a3ac..7bb0f1c 100644
--- a/net/bluetooth/hci_core.c
+++ b/net/bluetooth/hci_core.c
@@ -756,7 +756,7 @@ int hci_get_dev_list(void __user *arg)
 
 	size = sizeof(*dl) + dev_num * sizeof(*dr);
 
-	if (!(dl = kmalloc(size, GFP_KERNEL)))
+	if (!(dl = kzalloc(size, GFP_KERNEL)))
 		return -ENOMEM;
 
 	dr = dl->dev_req;
-- 
1.5.5.1

Re: [PATCH 2.6.28] bluetooth: fix leak of uninitialized data to userspace

[Marcel Holtmann]

Hi Vegard,

> From 45be27894a18f87b71b855fcc4afd50f860254b4 Mon Sep 17 00:00:00 2001
> From: Vegard Nossum <vegard.nossum@gmail.com>
> Date: Sun, 5 Oct 2008 17:25:45 +0200
> Subject: [PATCH] bluetooth: fix leak of uninitialized data to userspace
> 
>     struct hci_dev_list_req {
>             __u16  dev_num;
>             struct hci_dev_req dev_req[0];  /* hci_dev_req structures */
>     };
> 
> sizeof(struct hci_dev_list_req) == 4, so the two bytes immediately
> following "dev_num" will never be initialized. When this structure
> is copied to userspace, these uninitialized bytes are leaked.
> 
> Fix by using kzalloc() instead of kmalloc(). Found using kmemcheck.

good catch. I thought we moved everything over to kzalloc, but as it
seems we forgot at least one. Applied your patch to me tree. Thanks.

Regards

Marcel