Skip to contentSkip to conversation

virtio-net: fix possible unsigned integer overflow

PATCH Applied Heng Qi 3 participants 8 messages Jan 31, 2023 – Feb 2, 2023

Subsystems (4)
lore
Share & tools
lore

Heng Qi

author
#0 lore raw

PATCH v2 1/1 virtio-net: fix possible unsigned integer overflow

When the single-buffer xdp is loaded and after xdp_linearize_page() is called, *num_buf becomes 0 and (*num_buf - 1) may overflow into a large integer in virtnet_build_xdp_buff_mrg(), resulting in unexpected packet dropping.

Fixes ef75cb51f139 ("virtio-net: build xdp_buff with multi buffers")

---
v1->v2:
- Change the type of num_buf from unsigned int to int. @Michael S . Tsirkin
- Some cleaner codes. @Michael S . Tsirkin

 drivers/net/virtio_net.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)
drivers/net/virtio_net.c +9 -6
index aaa6fe9b214a..8102861785a2 100644
@@ -716,7 +716,7 @@ static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
  * have enough headroom.
  */
 static struct page *xdp_linearize_page(struct receive_queue *rq,
-				       u16 *num_buf,
+				       int *num_buf,
 				       struct page *p,
 				       int offset,
 				       int page_off,
@@ -816,7 +816,7 @@ static struct sk_buff *receive_small(struct net_device *dev,
 		if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
 			int offset = buf - page_address(page) + header_offset;
 			unsigned int tlen = len + vi->hdr_len;
-			u16 num_buf = 1;
+			int num_buf = 1;
 
 			xdp_headroom = virtnet_get_headroom(vi);
 			header_offset = VIRTNET_RX_PAD + xdp_headroom;
@@ -989,7 +989,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 				      void *buf,
 				      unsigned int len,
 				      unsigned int frame_sz,
-				      u16 *num_buf,
+				      int *num_buf,
 				      unsigned int *xdp_frags_truesize,
 				      struct virtnet_rq_stats *stats)
 {
@@ -1007,6 +1007,9 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 	xdp_prepare_buff(xdp, buf - VIRTIO_XDP_HEADROOM,
 			 VIRTIO_XDP_HEADROOM + vi->hdr_len, len - vi->hdr_len, true);
 
+	if (!*num_buf)
+		return 0;
+
 	if (*num_buf > 1) {
 		/* If we want to build multi-buffer xdp, we need
 		 * to specify that the flags of xdp_buff have the
@@ -1020,10 +1023,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 		shinfo->xdp_frags_size = 0;
 	}
 
-	if ((*num_buf - 1) > MAX_SKB_FRAGS)
+	if (*num_buf > MAX_SKB_FRAGS + 1)
 		return -EINVAL;
 
-	while ((--*num_buf) >= 1) {
+	while (--*num_buf > 0) {
 		buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx);
 		if (unlikely(!buf)) {
 			pr_debug("%s: rx error: %d buffers out of %d missing\n",
@@ -1076,7 +1079,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 					 struct virtnet_rq_stats *stats)
 {
 	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
-	u16 num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
+	int num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
 	struct page *page = virt_to_head_page(buf);
 	int offset = buf - page_address(page);
 	struct sk_buff *head_skb, *curr_skb;
Signature · 1 line
2.19.1.6.gb485710b
#1 lore raw
This patch was applied to netdev/net-next.git (master) by Jakub Kicinski <[email protected]>:

Hello:

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

Quoted 11 lines · On Tue, 31 Jan 2023 16:50:04 +0800 you wrote:

When the single-buffer xdp is loaded and after xdp_linearize_page() is called, *num_buf becomes 0 and (*num_buf - 1) may overflow into a large integer in virtnet_build_xdp_buff_mrg(), resulting in unexpected packet dropping.

Fixes ef75cb51f139 ("virtio-net: build xdp_buff with multi buffers")

[...]

Here is the summary with links:
- [net-next,v2] virtio-net: fix possible unsigned integer overflow

    https://git.kernel.org/netdev/net-next/c/981f14d42a7f

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
#2 lore raw
In reply to Heng Qi
Quoted 49 lines · Heng Qi

When the single-buffer xdp is loaded and after xdp_linearize_page() is called, *num_buf becomes 0 and (*num_buf - 1) may overflow into a large integer in virtnet_build_xdp_buff_mrg(), resulting in unexpected packet dropping.

Fixes ef75cb51f139 ("virtio-net: build xdp_buff with multi buffers")

---
v1->v2:
- Change the type of num_buf from unsigned int to int. @Michael S . Tsirkin
- Some cleaner codes. @Michael S . Tsirkin

 drivers/net/virtio_net.c | 15 +++++++++------
 1 file changed, 9 insertions(+), 6 deletions(-)
drivers/net/virtio_net.c +3 -3
index aaa6fe9b214a..8102861785a2 100644
@@ -716,7 +716,7 @@ static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
  * have enough headroom.
  */
 static struct page *xdp_linearize_page(struct receive_queue *rq,
-				       u16 *num_buf,
+				       int *num_buf,
 				       struct page *p,
 				       int offset,
 				       int page_off,
@@ -816,7 +816,7 @@ static struct sk_buff *receive_small(struct net_device *dev,
 		if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
 			int offset = buf - page_address(page) + header_offset;
 			unsigned int tlen = len + vi->hdr_len;
-			u16 num_buf = 1;
+			int num_buf = 1;
 
 			xdp_headroom = virtnet_get_headroom(vi);
 			header_offset = VIRTNET_RX_PAD + xdp_headroom;
@@ -989,7 +989,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 				      void *buf,
 				      unsigned int len,
 				      unsigned int frame_sz,
-				      u16 *num_buf,
+				      int *num_buf,
 				      unsigned int *xdp_frags_truesize,
 				      struct virtnet_rq_stats *stats)
 {

drivers/net/virtio_net.c static int virtnet_build_xdp_buff_mrg(struct net_device *dev, · line 1015

Quoted 9 lines
drivers/net/virtio_net.c +3 -0
@@ -1007,6 +1007,9 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 	xdp_prepare_buff(xdp, buf - VIRTIO_XDP_HEADROOM,
 			 VIRTIO_XDP_HEADROOM + vi->hdr_len, len - vi->hdr_len, true);
 
+	if (!*num_buf)
+		return 0;
+
 	if (*num_buf > 1) {
 		/* If we want to build multi-buffer xdp, we need
 		 * to specify that the flags of xdp_buff have the

Ouch. Why is this here? Merged so pls remove by a follow up patch, the rest of the code handles 0 fine. I'm not sure this introduces a bug by we don't want spaghetti code.

Quoted 24 lines
drivers/net/virtio_net.c +3 -3
@@ -1020,10 +1023,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
 		shinfo->xdp_frags_size = 0;
 	}
 
-	if ((*num_buf - 1) > MAX_SKB_FRAGS)
+	if (*num_buf > MAX_SKB_FRAGS + 1)
 		return -EINVAL;
 
-	while ((--*num_buf) >= 1) {
+	while (--*num_buf > 0) {
 		buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx);
 		if (unlikely(!buf)) {
 			pr_debug("%s: rx error: %d buffers out of %d missing\n",
@@ -1076,7 +1079,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
 					 struct virtnet_rq_stats *stats)
 {
 	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
-	u16 num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
+	int num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
 	struct page *page = virt_to_head_page(buf);
 	int offset = buf - page_address(page);
 	struct sk_buff *head_skb, *curr_skb;
Signature · 1 line
2.19.1.6.gb485710b
author
#3 lore raw
Quoted 53 lines · Michael S. Tsirkin
Quoted 52 lines · Heng Qi

When the single-buffer xdp is loaded and after xdp_linearize_page() is called, *num_buf becomes 0 and (*num_buf - 1) may overflow into a large integer in virtnet_build_xdp_buff_mrg(), resulting in unexpected packet dropping.

Fixes ef75cb51f139 ("virtio-net: build xdp_buff with multi buffers")

---
v1->v2:
- Change the type of num_buf from unsigned int to int. @Michael S . Tsirkin
- Some cleaner codes. @Michael S . Tsirkin

  drivers/net/virtio_net.c | 15 +++++++++------
  1 file changed, 9 insertions(+), 6 deletions(-)
drivers/net/virtio_net.c +3 -3
index aaa6fe9b214a..8102861785a2 100644
@@ -716,7 +716,7 @@ static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
   * have enough headroom.
   */
  static struct page *xdp_linearize_page(struct receive_queue *rq,
-				       u16 *num_buf,
+				       int *num_buf,
  				       struct page *p,
  				       int offset,
  				       int page_off,
@@ -816,7 +816,7 @@ static struct sk_buff *receive_small(struct net_device *dev,
  		if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
  			int offset = buf - page_address(page) + header_offset;
  			unsigned int tlen = len + vi->hdr_len;
-			u16 num_buf = 1;
+			int num_buf = 1;
 
  			xdp_headroom = virtnet_get_headroom(vi);
  			header_offset = VIRTNET_RX_PAD + xdp_headroom;
@@ -989,7 +989,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
  				      void *buf,
  				      unsigned int len,
  				      unsigned int frame_sz,
-				      u16 *num_buf,
+				      int *num_buf,
  				      unsigned int *xdp_frags_truesize,
  				      struct virtnet_rq_stats *stats)
  {
@@ -1007,6 +1007,9 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
  	xdp_prepare_buff(xdp, buf - VIRTIO_XDP_HEADROOM,
  			 VIRTIO_XDP_HEADROOM + vi->hdr_len, len - vi->hdr_len, true);
 
Quoted 9 lines

drivers/net/virtio_net.c static int virtnet_build_xdp_buff_mrg(struct net_device *dev, · line 1015

Quoted 6 lines
drivers/net/virtio_net.c +3 -0
+	if (!*num_buf)
+		return 0;
+
  	if (*num_buf > 1) {
  		/* If we want to build multi-buffer xdp, we need
  		 * to specify that the flags of xdp_buff have the

Ouch. Why is this here? Merged so pls remove by a follow up patch, the rest of the code handles 0 fine. I'm not sure this introduces a bug by we don't want spaghetti code.

Yes it would work without this, but I was keeping this because I wanted it to handle 0 early and exit early.

Do you want to remove this?

Thanks.

Quoted 25 lines
Quoted 25 lines
drivers/net/virtio_net.c +3 -3
@@ -1020,10 +1023,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
  		shinfo->xdp_frags_size = 0;
  	}
 
-	if ((*num_buf - 1) > MAX_SKB_FRAGS)
+	if (*num_buf > MAX_SKB_FRAGS + 1)
  		return -EINVAL;
 
-	while ((--*num_buf) >= 1) {
+	while (--*num_buf > 0) {
  		buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx);
  		if (unlikely(!buf)) {
  			pr_debug("%s: rx error: %d buffers out of %d missing\n",
@@ -1076,7 +1079,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
  					 struct virtnet_rq_stats *stats)
  {
  	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
-	u16 num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
+	int num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
  	struct page *page = virt_to_head_page(buf);
  	int offset = buf - page_address(page);
  	struct sk_buff *head_skb, *curr_skb;
Signature · 1 line
2.19.1.6.gb485710b
#4 lore raw
Quoted 64 lines · Heng Qi
Quoted 52 lines · Michael S. Tsirkin
Quoted 51 lines · Heng Qi

When the single-buffer xdp is loaded and after xdp_linearize_page() is called, *num_buf becomes 0 and (*num_buf - 1) may overflow into a large integer in virtnet_build_xdp_buff_mrg(), resulting in unexpected packet dropping.

Fixes ef75cb51f139 ("virtio-net: build xdp_buff with multi buffers")

---
v1->v2:
- Change the type of num_buf from unsigned int to int. @Michael S . Tsirkin
- Some cleaner codes. @Michael S . Tsirkin

  drivers/net/virtio_net.c | 15 +++++++++------
  1 file changed, 9 insertions(+), 6 deletions(-)
drivers/net/virtio_net.c +4 -3
index aaa6fe9b214a..8102861785a2 100644
@@ -716,7 +716,7 @@ static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
   * have enough headroom.
   */
  static struct page *xdp_linearize_page(struct receive_queue *rq,
-				       u16 *num_buf,
+				       int *num_buf,
  				       struct page *p,
  				       int offset,
  				       int page_off,
@@ -816,7 +816,7 @@ static struct sk_buff *receive_small(struct net_device *dev,
  		if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
  			int offset = buf - page_address(page) + header_offset;
  			unsigned int tlen = len + vi->hdr_len;
-			u16 num_buf = 1;
+			int num_buf = 1;
  			xdp_headroom = virtnet_get_headroom(vi);
  			header_offset = VIRTNET_RX_PAD + xdp_headroom;
@@ -989,7 +989,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
  				      void *buf,
  				      unsigned int len,
  				      unsigned int frame_sz,
-				      u16 *num_buf,
+				      int *num_buf,
  				      unsigned int *xdp_frags_truesize,
  				      struct virtnet_rq_stats *stats)
  {
@@ -1007,6 +1007,9 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
  	xdp_prepare_buff(xdp, buf - VIRTIO_XDP_HEADROOM,
  			 VIRTIO_XDP_HEADROOM + vi->hdr_len, len - vi->hdr_len, true);
+	if (!*num_buf)
Quoted 8 lines

drivers/net/virtio_net.c static int virtnet_build_xdp_buff_mrg(struct net_device *dev, · line 1015

Quoted 5 lines
drivers/net/virtio_net.c +2 -0
+		return 0;
+
  	if (*num_buf > 1) {
  		/* If we want to build multi-buffer xdp, we need
  		 * to specify that the flags of xdp_buff have the

Ouch. Why is this here? Merged so pls remove by a follow up patch, the rest of the code handles 0 fine. I'm not sure this introduces a bug by we don't want spaghetti code.

Quoted 6 lines

Yes it would work without this, but I was keeping this because I wanted it to handle 0 early and exit early.

Do you want to remove this?

Thanks.

why do you want to exit early?

Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
drivers/net/virtio_net.c +3 -3
@@ -1020,10 +1023,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
  		shinfo->xdp_frags_size = 0;
  	}
-	if ((*num_buf - 1) > MAX_SKB_FRAGS)
+	if (*num_buf > MAX_SKB_FRAGS + 1)
  		return -EINVAL;
-	while ((--*num_buf) >= 1) {
+	while (--*num_buf > 0) {
  		buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx);
  		if (unlikely(!buf)) {
  			pr_debug("%s: rx error: %d buffers out of %d missing\n",
@@ -1076,7 +1079,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
  					 struct virtnet_rq_stats *stats)
  {
  	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
-	u16 num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
+	int num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
  	struct page *page = virt_to_head_page(buf);
  	int offset = buf - page_address(page);
  	struct sk_buff *head_skb, *curr_skb;
Signature · 1 line
2.19.1.6.gb485710b
author
#5 lore raw
Quoted 63 lines · Michael S. Tsirkin
Quoted 62 lines · Heng Qi
Quoted 60 lines · Michael S. Tsirkin
Quoted 48 lines · Heng Qi

When the single-buffer xdp is loaded and after xdp_linearize_page() is called, *num_buf becomes 0 and (*num_buf - 1) may overflow into a large integer in virtnet_build_xdp_buff_mrg(), resulting in unexpected packet dropping.

Fixes ef75cb51f139 ("virtio-net: build xdp_buff with multi buffers")

---
v1->v2:
- Change the type of num_buf from unsigned int to int. @Michael S . Tsirkin
- Some cleaner codes. @Michael S . Tsirkin

   drivers/net/virtio_net.c | 15 +++++++++------
   1 file changed, 9 insertions(+), 6 deletions(-)
drivers/net/virtio_net.c +3 -3
index aaa6fe9b214a..8102861785a2 100644
@@ -716,7 +716,7 @@ static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
    * have enough headroom.
    */
   static struct page *xdp_linearize_page(struct receive_queue *rq,
-				       u16 *num_buf,
+				       int *num_buf,
   				       struct page *p,
   				       int offset,
   				       int page_off,
@@ -816,7 +816,7 @@ static struct sk_buff *receive_small(struct net_device *dev,
   		if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
   			int offset = buf - page_address(page) + header_offset;
   			unsigned int tlen = len + vi->hdr_len;
-			u16 num_buf = 1;
+			int num_buf = 1;
   			xdp_headroom = virtnet_get_headroom(vi);
   			header_offset = VIRTNET_RX_PAD + xdp_headroom;
@@ -989,7 +989,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
   				      void *buf,
   				      unsigned int len,
   				      unsigned int frame_sz,
-				      u16 *num_buf,
+				      int *num_buf,
   				      unsigned int *xdp_frags_truesize,
   				      struct virtnet_rq_stats *stats)
   {

drivers/net/virtio_net.c static int virtnet_build_xdp_buff_mrg(struct net_device *dev, · line 1014

Quoted 8 lines
drivers/net/virtio_net.c +3 -0
@@ -1007,6 +1007,9 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
   	xdp_prepare_buff(xdp, buf - VIRTIO_XDP_HEADROOM,
   			 VIRTIO_XDP_HEADROOM + vi->hdr_len, len - vi->hdr_len, true);
+	if (!*num_buf)
+		return 0;
+
   	if (*num_buf > 1) {
   		/* If we want to build multi-buffer xdp, we need
   		 * to specify that the flags of xdp_buff have the

Ouch. Why is this here? Merged so pls remove by a follow up patch, the rest of the code handles 0 fine. I'm not sure this introduces a bug by we don't want spaghetti code.

Quoted 7 lines
Quoted 6 lines

Yes it would work without this, but I was keeping this because I wanted it to handle 0 early and exit early.

Do you want to remove this?

Thanks.

why do you want to exit early?

If num_buf is 0, we don't need to judge the subsequent process, because the latter process is used to build multi-buffer xdp, but this fix solves the possible problems of single-buffer xdp.

Thanks.

Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
drivers/net/virtio_net.c +3 -3
@@ -1020,10 +1023,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
   		shinfo->xdp_frags_size = 0;
   	}
-	if ((*num_buf - 1) > MAX_SKB_FRAGS)
+	if (*num_buf > MAX_SKB_FRAGS + 1)
   		return -EINVAL;
-	while ((--*num_buf) >= 1) {
+	while (--*num_buf > 0) {
   		buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx);
   		if (unlikely(!buf)) {
   			pr_debug("%s: rx error: %d buffers out of %d missing\n",
@@ -1076,7 +1079,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
   					 struct virtnet_rq_stats *stats)
   {
   	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
-	u16 num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
+	int num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
   	struct page *page = virt_to_head_page(buf);
   	int offset = buf - page_address(page);
   	struct sk_buff *head_skb, *curr_skb;
Signature · 1 line
2.19.1.6.gb485710b
#6 lore raw
Quoted 72 lines · Heng Qi
Quoted 69 lines · Michael S. Tsirkin
Quoted 68 lines · Heng Qi
Quoted 52 lines · Michael S. Tsirkin
Quoted 51 lines · Heng Qi

When the single-buffer xdp is loaded and after xdp_linearize_page() is called, *num_buf becomes 0 and (*num_buf - 1) may overflow into a large integer in virtnet_build_xdp_buff_mrg(), resulting in unexpected packet dropping.

Fixes ef75cb51f139 ("virtio-net: build xdp_buff with multi buffers")

---
v1->v2:
- Change the type of num_buf from unsigned int to int. @Michael S . Tsirkin
- Some cleaner codes. @Michael S . Tsirkin

   drivers/net/virtio_net.c | 15 +++++++++------
   1 file changed, 9 insertions(+), 6 deletions(-)
drivers/net/virtio_net.c +4 -3
index aaa6fe9b214a..8102861785a2 100644
@@ -716,7 +716,7 @@ static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
    * have enough headroom.
    */
   static struct page *xdp_linearize_page(struct receive_queue *rq,
-				       u16 *num_buf,
+				       int *num_buf,
   				       struct page *p,
   				       int offset,
   				       int page_off,
@@ -816,7 +816,7 @@ static struct sk_buff *receive_small(struct net_device *dev,
   		if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
   			int offset = buf - page_address(page) + header_offset;
   			unsigned int tlen = len + vi->hdr_len;
-			u16 num_buf = 1;
+			int num_buf = 1;
   			xdp_headroom = virtnet_get_headroom(vi);
   			header_offset = VIRTNET_RX_PAD + xdp_headroom;
@@ -989,7 +989,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
   				      void *buf,
   				      unsigned int len,
   				      unsigned int frame_sz,
-				      u16 *num_buf,
+				      int *num_buf,
   				      unsigned int *xdp_frags_truesize,
   				      struct virtnet_rq_stats *stats)
   {
@@ -1007,6 +1007,9 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
   	xdp_prepare_buff(xdp, buf - VIRTIO_XDP_HEADROOM,
   			 VIRTIO_XDP_HEADROOM + vi->hdr_len, len - vi->hdr_len, true);
+	if (!*num_buf)
Quoted 8 lines

drivers/net/virtio_net.c static int virtnet_build_xdp_buff_mrg(struct net_device *dev, · line 1015

Quoted 5 lines
drivers/net/virtio_net.c +2 -0
+		return 0;
+
   	if (*num_buf > 1) {
   		/* If we want to build multi-buffer xdp, we need
   		 * to specify that the flags of xdp_buff have the

Ouch. Why is this here? Merged so pls remove by a follow up patch, the rest of the code handles 0 fine. I'm not sure this introduces a bug by we don't want spaghetti code.

Yes it would work without this, but I was keeping this because I wanted it to handle 0 early and exit early.

Do you want to remove this?

Thanks.

Quoted 8 lines
Quoted 1 line

why do you want to exit early?

If num_buf is 0, we don't need to judge the subsequent process, because the latter process
is used to build multi-buffer xdp, but this fix solves the possible problems of single-buffer xdp.

Thanks.

An optimization then? As any optimization I'd like to see some numbers.

Should have been documented as such not as part of a bugfix.

Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
drivers/net/virtio_net.c +3 -3
@@ -1020,10 +1023,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
   		shinfo->xdp_frags_size = 0;
   	}
-	if ((*num_buf - 1) > MAX_SKB_FRAGS)
+	if (*num_buf > MAX_SKB_FRAGS + 1)
   		return -EINVAL;
-	while ((--*num_buf) >= 1) {
+	while (--*num_buf > 0) {
   		buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx);
   		if (unlikely(!buf)) {
   			pr_debug("%s: rx error: %d buffers out of %d missing\n",
@@ -1076,7 +1079,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
   					 struct virtnet_rq_stats *stats)
   {
   	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
-	u16 num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
+	int num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
   	struct page *page = virt_to_head_page(buf);
   	int offset = buf - page_address(page);
   	struct sk_buff *head_skb, *curr_skb;
Signature · 1 line
2.19.1.6.gb485710b
author
#7 lore raw
Quoted 77 lines · Michael S. Tsirkin
Quoted 76 lines · Heng Qi
Quoted 62 lines · Michael S. Tsirkin
Quoted 61 lines · Heng Qi
Quoted 60 lines · Michael S. Tsirkin
Quoted 48 lines · Heng Qi

When the single-buffer xdp is loaded and after xdp_linearize_page() is called, *num_buf becomes 0 and (*num_buf - 1) may overflow into a large integer in virtnet_build_xdp_buff_mrg(), resulting in unexpected packet dropping.

Fixes ef75cb51f139 ("virtio-net: build xdp_buff with multi buffers")

---
v1->v2:
- Change the type of num_buf from unsigned int to int. @Michael S . Tsirkin
- Some cleaner codes. @Michael S . Tsirkin

    drivers/net/virtio_net.c | 15 +++++++++------
    1 file changed, 9 insertions(+), 6 deletions(-)
drivers/net/virtio_net.c +3 -3
index aaa6fe9b214a..8102861785a2 100644
@@ -716,7 +716,7 @@ static unsigned int virtnet_get_headroom(struct virtnet_info *vi)
     * have enough headroom.
     */
    static struct page *xdp_linearize_page(struct receive_queue *rq,
-				       u16 *num_buf,
+				       int *num_buf,
    				       struct page *p,
    				       int offset,
    				       int page_off,
@@ -816,7 +816,7 @@ static struct sk_buff *receive_small(struct net_device *dev,
    		if (unlikely(xdp_headroom < virtnet_get_headroom(vi))) {
    			int offset = buf - page_address(page) + header_offset;
    			unsigned int tlen = len + vi->hdr_len;
-			u16 num_buf = 1;
+			int num_buf = 1;
    			xdp_headroom = virtnet_get_headroom(vi);
    			header_offset = VIRTNET_RX_PAD + xdp_headroom;
@@ -989,7 +989,7 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
    				      void *buf,
    				      unsigned int len,
    				      unsigned int frame_sz,
-				      u16 *num_buf,
+				      int *num_buf,
    				      unsigned int *xdp_frags_truesize,
    				      struct virtnet_rq_stats *stats)
    {

drivers/net/virtio_net.c static int virtnet_build_xdp_buff_mrg(struct net_device *dev, · line 1014

Quoted 8 lines
drivers/net/virtio_net.c +3 -0
@@ -1007,6 +1007,9 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
    	xdp_prepare_buff(xdp, buf - VIRTIO_XDP_HEADROOM,
    			 VIRTIO_XDP_HEADROOM + vi->hdr_len, len - vi->hdr_len, true);
+	if (!*num_buf)
+		return 0;
+
    	if (*num_buf > 1) {
    		/* If we want to build multi-buffer xdp, we need
    		 * to specify that the flags of xdp_buff have the

Ouch. Why is this here? Merged so pls remove by a follow up patch, the rest of the code handles 0 fine. I'm not sure this introduces a bug by we don't want spaghetti code.

Quoted 7 lines
Quoted 6 lines

Yes it would work without this, but I was keeping this because I wanted it to handle 0 early and exit early.

Do you want to remove this?

Thanks.

why do you want to exit early?

If num_buf is 0, we don't need to judge the subsequent process, because the latter process
is used to build multi-buffer xdp, but this fix solves the possible problems of single-buffer xdp.

Quoted 4 lines
Quoted 1 line

Thanks.

An optimization then? As any optimization I'd like to see some numbers.

Should have been documented as such not as part of a bugfix.

Sure, I'm going to remove this with a follow-up patch.😁

Thanks.

Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
Quoted 23 lines
drivers/net/virtio_net.c +3 -3
@@ -1020,10 +1023,10 @@ static int virtnet_build_xdp_buff_mrg(struct net_device *dev,
    		shinfo->xdp_frags_size = 0;
    	}
-	if ((*num_buf - 1) > MAX_SKB_FRAGS)
+	if (*num_buf > MAX_SKB_FRAGS + 1)
    		return -EINVAL;
-	while ((--*num_buf) >= 1) {
+	while (--*num_buf > 0) {
    		buf = virtqueue_get_buf_ctx(rq->vq, &len, &ctx);
    		if (unlikely(!buf)) {
    			pr_debug("%s: rx error: %d buffers out of %d missing\n",
@@ -1076,7 +1079,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev,
    					 struct virtnet_rq_stats *stats)
    {
    	struct virtio_net_hdr_mrg_rxbuf *hdr = buf;
-	u16 num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
+	int num_buf = virtio16_to_cpu(vi->vdev, hdr->num_buffers);
    	struct page *page = virt_to_head_page(buf);
    	int offset = buf - page_address(page);
    	struct sk_buff *head_skb, *curr_skb;
Signature · 1 line
2.19.1.6.gb485710b