Skip to contentSkip to conversation

bnxt: prevent skb UAF after handing over to PTP worker

PATCH Applied Jakub Kicinski 4 participants 4 messages Sep 21, 2022 – Sep 22, 2022

Subsystems (3)
lore
Share & tools
lore

Jakub Kicinski

author maintainer
#0 lore raw

PATCH 1/1 bnxt: prevent skb UAF after handing over to PTP worker

When reading the timestamp is required bnxt_tx_int() hands over the ownership of the completed skb to the PTP worker. The skb should not be used afterwards, as the worker may run before the rest of our code and free the skb, leading to a use-after-free.

Since dev_kfree_skb_any() accepts NULL make the loss of ownership more obvious and set skb to NULL.

Fixes 83bb623c968e ("bnxt_en: Transmit and retrieve packet timestamps")

---
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]

---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +5 -5
index f46eefb5a029..96da0ba3d507 100644
@@ -659,7 +659,6 @@ static void bnxt_tx_int(struct bnxt *bp, struct bnxt_napi *bnapi, int nr_pkts)
 
 	for (i = 0; i < nr_pkts; i++) {
 		struct bnxt_sw_tx_bd *tx_buf;
-		bool compl_deferred = false;
 		struct sk_buff *skb;
 		int j, last;
 
@@ -668,6 +667,8 @@ static void bnxt_tx_int(struct bnxt *bp, struct bnxt_napi *bnapi, int nr_pkts)
 		skb = tx_buf->skb;
 		tx_buf->skb = NULL;
 
+		tx_bytes += skb->len;
+
 		if (tx_buf->is_push) {
 			tx_buf->is_push = 0;
 			goto next_tx_int;
@@ -688,8 +689,9 @@ static void bnxt_tx_int(struct bnxt *bp, struct bnxt_napi *bnapi, int nr_pkts)
 		}
 		if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_IN_PROGRESS)) {
 			if (bp->flags & BNXT_FLAG_CHIP_P5) {
+				/* PTP worker takes ownership of the skb */
 				if (!bnxt_get_tx_ts_p5(bp, skb))
-					compl_deferred = true;
+					skb = NULL;
 				else
 					atomic_inc(&bp->ptp_cfg->tx_avail);
 			}
@@ -698,9 +700,7 @@ static void bnxt_tx_int(struct bnxt *bp, struct bnxt_napi *bnapi, int nr_pkts)
 next_tx_int:
 		cons = NEXT_TX(cons);
 
-		tx_bytes += skb->len;
-		if (!compl_deferred)
-			dev_kfree_skb_any(skb);
+		dev_kfree_skb_any(skb);
 	}
 
 	netdev_tx_completed_queue(txq, nr_pkts, tx_bytes);
Signature · 1 line
2.37.3
Reviewed-by on v1 1/1
#1 lore raw
Quoted 11 lines · Jakub Kicinski

When reading the timestamp is required bnxt_tx_int() hands over the ownership of the completed skb to the PTP worker. The skb should not be used afterwards, as the worker may run before the rest of our code and free the skb, leading to a use-after-free.

Since dev_kfree_skb_any() accepts NULL make the loss of ownership more obvious and set skb to NULL.

Fixes 83bb623c968e ("bnxt_en: Transmit and retrieve packet timestamps")

In general this looks good to me. Let's make sure Pavan and Michael also agree. Thanks for the patch!

Quoted 55 lines

---
CC: [email protected]
CC: [email protected]
CC: [email protected]
CC: [email protected]

---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)
drivers/net/ethernet/broadcom/bnxt/bnxt.c +5 -5
index f46eefb5a029..96da0ba3d507 100644
@@ -659,7 +659,6 @@ static void bnxt_tx_int(struct bnxt *bp, struct bnxt_napi *bnapi, int nr_pkts)
 
 	for (i = 0; i < nr_pkts; i++) {
 		struct bnxt_sw_tx_bd *tx_buf;
-		bool compl_deferred = false;
 		struct sk_buff *skb;
 		int j, last;
 
@@ -668,6 +667,8 @@ static void bnxt_tx_int(struct bnxt *bp, struct bnxt_napi *bnapi, int nr_pkts)
 		skb = tx_buf->skb;
 		tx_buf->skb = NULL;
 
+		tx_bytes += skb->len;
+
 		if (tx_buf->is_push) {
 			tx_buf->is_push = 0;
 			goto next_tx_int;
@@ -688,8 +689,9 @@ static void bnxt_tx_int(struct bnxt *bp, struct bnxt_napi *bnapi, int nr_pkts)
 		}
 		if (unlikely(skb_shinfo(skb)->tx_flags & SKBTX_IN_PROGRESS)) {
 			if (bp->flags & BNXT_FLAG_CHIP_P5) {
+				/* PTP worker takes ownership of the skb */
 				if (!bnxt_get_tx_ts_p5(bp, skb))
-					compl_deferred = true;
+					skb = NULL;
 				else
 					atomic_inc(&bp->ptp_cfg->tx_avail);
 			}
@@ -698,9 +700,7 @@ static void bnxt_tx_int(struct bnxt *bp, struct bnxt_napi *bnapi, int nr_pkts)
 next_tx_int:
 		cons = NEXT_TX(cons);
 
-		tx_bytes += skb->len;
-		if (!compl_deferred)
-			dev_kfree_skb_any(skb);
+		dev_kfree_skb_any(skb);
 	}
 
 	netdev_tx_completed_queue(txq, nr_pkts, tx_bytes);
Signature · 1 line
2.37.3
maintainer Reviewed-by
#2 lore raw
Quoted 11 lines · Andy Gospodarek
Quoted 9 lines · Jakub Kicinski

When reading the timestamp is required bnxt_tx_int() hands over the ownership of the completed skb to the PTP worker. The skb should not be used afterwards, as the worker may run before the rest of our code and free the skb, leading to a use-after-free.

Since dev_kfree_skb_any() accepts NULL make the loss of ownership more obvious and set skb to NULL.

Quoted 7 lines
Quoted 2 lines
Fixes 83bb623c968e ("bnxt_en: Transmit and retrieve packet timestamps")

In general this looks good to me. Let's make sure Pavan and Michael also agree. Thanks for the patch!

Thanks for catching this.

#3 lore raw
In reply to Jakub Kicinski
This patch was applied to netdev/net.git (master) by Jakub Kicinski <[email protected]>:

Hello:

This patch was applied to netdev/net.git (master)
by Jakub Kicinski <[email protected]>:

Quoted 10 lines · On Wed, 21 Sep 2022 13:10:05 -0700 you wrote:

When reading the timestamp is required bnxt_tx_int() hands over the ownership of the completed skb to the PTP worker. The skb should not be used afterwards, as the worker may run before the rest of our code and free the skb, leading to a use-after-free.

Since dev_kfree_skb_any() accepts NULL make the loss of ownership more obvious and set skb to NULL.

[...]

Here is the summary with links:
- [net] bnxt: prevent skb UAF after handing over to PTP worker

    https://git.kernel.org/netdev/net/c/c31f26c8f69f

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html