netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH net] virtio_net: fix header access in big_packets mode
@ 2025-10-16 22:33 Wen Xin
  2025-10-17  0:57 ` Xuan Zhuo
  0 siblings, 1 reply; 4+ messages in thread
From: Wen Xin @ 2025-10-16 22:33 UTC (permalink / raw)
  To: netdev; +Cc: mst, jasowang, xuanzhuo, eperezma, virtualization, Wen Xin

In Linux virtio-net driver's (drivers/net/virtio_net.c) big packets mode (vi->big_packets && vi->mergeable_rx_bufs), the buf
pointer passed to receive_buf() is a struct page pointer, not a buffer
pointer. The current code incorrectly casts this page pointer directly as
a virtio_net_common_hdr, causing it to read flags from the page struct
memory instead of the actual packet data.

Signed-off-by: Wen Xin <w_xin@apple.com>
---
 drivers/net/virtio_net.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 7646ddd9bef7..c10f5585bc88 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -2566,7 +2566,13 @@ static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
 	 * virtnet_xdp_set()), so packets marked as
 	 * VIRTIO_NET_HDR_F_NEEDS_CSUM get dropped during XDP processing.
 	 */
-	flags = ((struct virtio_net_common_hdr *)buf)->hdr.flags;
+	if (vi->big_packets && !vi->mergeable_rx_bufs) {
+		struct virtio_net_common_hdr *hdr = page_address((struct page *)buf);
+
+		flags = hdr->hdr.flags;
+	} else {
+		flags = ((struct virtio_net_common_hdr *)buf)->hdr.flags;
+	}
 
 	if (vi->mergeable_rx_bufs)
 		skb = receive_mergeable(dev, vi, rq, buf, ctx, len, xdp_xmit,
-- 
2.39.5 (Apple Git-154)


^ permalink raw reply related	[flat|nested] 4+ messages in thread

* Re: [PATCH net] virtio_net: fix header access in big_packets mode
  2025-10-16 22:33 [PATCH net] virtio_net: fix header access in big_packets mode Wen Xin
@ 2025-10-17  0:57 ` Xuan Zhuo
  2025-10-17 23:38   ` Jakub Kicinski
  0 siblings, 1 reply; 4+ messages in thread
From: Xuan Zhuo @ 2025-10-17  0:57 UTC (permalink / raw)
  To: Wen Xin; +Cc: mst, jasowang, eperezma, virtualization, Wen Xin, netdev

http://lore.kernel.org/all/20251013020629.73902-1-xuanzhuo@linux.alibaba.com

do the same thing.


On Thu, 16 Oct 2025 15:33:05 -0700, Wen Xin <w_xin@apple.com> wrote:
> In Linux virtio-net driver's (drivers/net/virtio_net.c) big packets mode (vi->big_packets && vi->mergeable_rx_bufs), the buf
> pointer passed to receive_buf() is a struct page pointer, not a buffer
> pointer. The current code incorrectly casts this page pointer directly as
> a virtio_net_common_hdr, causing it to read flags from the page struct
> memory instead of the actual packet data.
>
> Signed-off-by: Wen Xin <w_xin@apple.com>
> ---
>  drivers/net/virtio_net.c | 8 +++++++-
>  1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
> index 7646ddd9bef7..c10f5585bc88 100644
> --- a/drivers/net/virtio_net.c
> +++ b/drivers/net/virtio_net.c
> @@ -2566,7 +2566,13 @@ static void receive_buf(struct virtnet_info *vi, struct receive_queue *rq,
>  	 * virtnet_xdp_set()), so packets marked as
>  	 * VIRTIO_NET_HDR_F_NEEDS_CSUM get dropped during XDP processing.
>  	 */
> -	flags = ((struct virtio_net_common_hdr *)buf)->hdr.flags;
> +	if (vi->big_packets && !vi->mergeable_rx_bufs) {
> +		struct virtio_net_common_hdr *hdr = page_address((struct page *)buf);
> +
> +		flags = hdr->hdr.flags;
> +	} else {
> +		flags = ((struct virtio_net_common_hdr *)buf)->hdr.flags;
> +	}
>
>  	if (vi->mergeable_rx_bufs)
>  		skb = receive_mergeable(dev, vi, rq, buf, ctx, len, xdp_xmit,
> --
> 2.39.5 (Apple Git-154)
>

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH net] virtio_net: fix header access in big_packets mode
  2025-10-17  0:57 ` Xuan Zhuo
@ 2025-10-17 23:38   ` Jakub Kicinski
  2025-10-22  1:42     ` Xuan Zhuo
  0 siblings, 1 reply; 4+ messages in thread
From: Jakub Kicinski @ 2025-10-17 23:38 UTC (permalink / raw)
  To: Xuan Zhuo; +Cc: Wen Xin, mst, jasowang, eperezma, virtualization, netdev

On Fri, 17 Oct 2025 08:57:36 +0800 Xuan Zhuo wrote:
> http://lore.kernel.org/all/20251013020629.73902-1-xuanzhuo@linux.alibaba.com
> 
> do the same thing.

Could you repost that? looks like there were some questions,
but nobody sent review tags even tho they were answered..

^ permalink raw reply	[flat|nested] 4+ messages in thread

* Re: [PATCH net] virtio_net: fix header access in big_packets mode
  2025-10-17 23:38   ` Jakub Kicinski
@ 2025-10-22  1:42     ` Xuan Zhuo
  0 siblings, 0 replies; 4+ messages in thread
From: Xuan Zhuo @ 2025-10-22  1:42 UTC (permalink / raw)
  To: Jakub Kicinski; +Cc: Wen Xin, mst, jasowang, eperezma, virtualization, netdev

On Fri, 17 Oct 2025 16:38:56 -0700, Jakub Kicinski <kuba@kernel.org> wrote:
> On Fri, 17 Oct 2025 08:57:36 +0800 Xuan Zhuo wrote:
> > http://lore.kernel.org/all/20251013020629.73902-1-xuanzhuo@linux.alibaba.com
> >
> > do the same thing.
>
> Could you repost that? looks like there were some questions,
> but nobody sent review tags even tho they were answered..


YES. In my todo list.



^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2025-10-22  1:43 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-10-16 22:33 [PATCH net] virtio_net: fix header access in big_packets mode Wen Xin
2025-10-17  0:57 ` Xuan Zhuo
2025-10-17 23:38   ` Jakub Kicinski
2025-10-22  1:42     ` Xuan Zhuo

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).