Skip to contentSkip to conversation

RDMA/rxe: Fix possible NULL pointer dereference

PATCH In review Alex Dewar 3 participants 3 messages Oct 12, 2020 – Oct 16, 2020 1 message not in the archive

Subsystems (2)
lore
Share & tools
lore

Alex Dewar

author
#0 lore raw

PATCH 1/1 RDMA/rxe: Fix possible NULL pointer dereference

skb_clone() is called without checking if the returned pointer is NULL before it is dereferenced. Fix by adding an additional error check.

Fixes e7ec96fc7932 ("RDMA/rxe: Fix skb lifetime in rxe_rcv_mcast_pkt()")

Addresses-Coverity-ID: 1497804: Null pointer dereferences (NULL_RETURNS)

---
 drivers/infiniband/sw/rxe/rxe_recv.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/infiniband/sw/rxe/rxe_recv.c +9 -2
index 11f3daf20768..a65936e12f89 100644
@@ -266,10 +266,13 @@ static void rxe_rcv_mcast_pkt(struct rxe_dev *rxe, struct sk_buff *skb)
 		/* for all but the last qp create a new clone of the
 		 * skb and pass to the qp.
 		 */
-		if (mce->qp_list.next != &mcg->qp_list)
+		if (mce->qp_list.next != &mcg->qp_list) {
 			per_qp_skb = skb_clone(skb, GFP_ATOMIC);
-		else
+			if (!per_qp_skb)
+				goto err2;
+		} else {
 			per_qp_skb = skb;
+		}
 
 		per_qp_pkt = SKB_TO_PKT(per_qp_skb);
 		per_qp_pkt->qp = qp;
@@ -283,6 +286,10 @@ static void rxe_rcv_mcast_pkt(struct rxe_dev *rxe, struct sk_buff *skb)
 
 	return;
 
+err2:
+	spin_unlock_bh(&mcg->mcg_lock);
+	rxe_drop_ref(mcg);	/* drop ref from rxe_pool_get_key. */
+
 err1:
 	kfree_skb(skb);
 }
Signature · 1 line
2.28.0

A message that is not in the archive (sent off-list or not yet received): DM6PR12MB3516EC1E0F28EB8F9B970F24D5050@DM6PR12MB3516.namprd12.prod.outlook.com

maintainer
#2 lore raw
Quoted 9 lines · Yanjun Zhu

-----Original Message-----
From: Alex Dewar <[email protected]>
Sent: Tuesday, October 13, 2020 12:53 AM
Cc: Alex Dewar <[email protected]>; Yanjun Zhu <[email protected]>; Doug Ledford <[email protected]>; Jason Gunthorpe <[email protected]>; Bob Pearson <[email protected]>; [email protected]; [email protected]
Subject: [PATCH] RDMA/rxe: Fix possible NULL pointer dereference

Quoted 5 lines

skb_clone() is called without checking if the returned pointer is NULL before it is dereferenced. Fix by adding an additional error check.

Fixes e7ec96fc7932 ("RDMA/rxe: Fix skb lifetime in rxe_rcv_mcast_pkt()")

Addresses-Coverity-ID: 1497804: Null pointer dereferences (NULL_RETURNS)

Thanks a lot.

Zhu Yanjun

Quoted 38 lines
---
 drivers/infiniband/sw/rxe/rxe_recv.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)
drivers/infiniband/sw/rxe/rxe_recv.c +9 -3
index 11f3daf20768..a65936e12f89 100644
@@ -266,10 +266,13 @@ static void rxe_rcv_mcast_pkt(struct rxe_dev *rxe, struct sk_buff *skb)
                /* for all but the last qp create a new clone of the
                 * skb and pass to the qp.
                 */
-               if (mce->qp_list.next != &mcg->qp_list)
+               if (mce->qp_list.next != &mcg->qp_list) {
                        per_qp_skb = skb_clone(skb, GFP_ATOMIC);
-               else
+                       if (!per_qp_skb)
+                               goto err2;
+               } else {
                        per_qp_skb = skb;
+               }
 
                per_qp_pkt = SKB_TO_PKT(per_qp_skb);
                per_qp_pkt->qp = qp;
@@ -283,6 +286,10 @@ static void rxe_rcv_mcast_pkt(struct rxe_dev *rxe, struct sk_buff *skb)
 
        return;
 
+err2:
+       spin_unlock_bh(&mcg->mcg_lock);
+       rxe_drop_ref(mcg);      /* drop ref from rxe_pool_get_key. */
+
 err1:
        kfree_skb(skb);
 }
--

2.28.0

maintainer
#3 lore raw
In reply to Alex Dewar
Quoted 9 lines · Alex Dewar

skb_clone() is called without checking if the returned pointer is NULL before it is dereferenced. Fix by adding an additional error check.

Fixes e7ec96fc7932 ("RDMA/rxe: Fix skb lifetime in rxe_rcv_mcast_pkt()")

Addresses-Coverity-ID: 1497804: Null pointer dereferences (NULL_RETURNS)

---
 drivers/infiniband/sw/rxe/rxe_recv.c | 11 +++++++++--
 1 file changed, 9 insertions(+), 2 deletions(-)

I took Bob's version of this since it is a little bit simpler

Thanks,
Jason