Re: [PATCH net v2 0/2] virtio-net: fix for build_skb()
From: "Michael S. Tsirkin" <mst@redhat.com>
Date: 2021-08-02 16:58:29
Also in:
bpf, virtualization
On Fri, Jul 30, 2021 at 05:16:43AM -0700, Jakub Kicinski wrote:
On Tue, 1 Jun 2021 07:06:43 -0400 Michael S. Tsirkin wrote:quoted
On Tue, Jun 01, 2021 at 02:39:58PM +0800, Xuan Zhuo wrote:quoted
#1 Fixed a serious error. #2 Fixed a logical error, but this error did not cause any serious consequences. The logic of this piece is really messy. Fortunately, my refactored patch can be completed with a small amount of testing.Looks good, thanks! Also needed for stable I think. Acked-by: Michael S. Tsirkin <mst@redhat.com>Just a heads up folks, looks like we ended up merging both this and the net-next version of the patch set: 8fb7da9e9907 virtio_net: get build_skb() buf by data ptr 5c37711d9f27 virtio-net: fix for unable to handle page fault for address and 7bf64460e3b2 virtio-net: get build_skb() buf by data ptr 6c66c147b9a4 virtio-net: fix for unable to handle page fault for address Are you okay with the code as is or should we commit something like:
Yes: Acked-by: Michael S. Tsirkin <mst@redhat.com> Will you post this one then?
quoted hunk ↗ jump to hunk
---diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c index 56c3f8519093..74482a52f076 100644 --- a/drivers/net/virtio_net.c +++ b/drivers/net/virtio_net.c@@ -380,7 +380,7 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi, struct page *page, unsigned int offset, unsigned int len, unsigned int truesize, bool hdr_valid, unsigned int metasize, - bool whole_page) + unsigned int headroom) { struct sk_buff *skb; struct virtio_net_hdr_mrg_rxbuf *hdr;@@ -398,28 +398,16 @@ static struct sk_buff *page_to_skb(struct virtnet_info *vi, else hdr_padded_len = sizeof(struct padded_vnet_hdr); - /* If whole_page, there is an offset between the beginning of the + /* If headroom is not 0, there is an offset between the beginning of the * data and the allocated space, otherwise the data and the allocated * space are aligned. * * Buffers with headroom use PAGE_SIZE as alloc size, see * add_recvbuf_mergeable() + get_mergeable_buf_len() */ - if (whole_page) { - /* Buffers with whole_page use PAGE_SIZE as alloc size, - * see add_recvbuf_mergeable() + get_mergeable_buf_len() - */ - truesize = PAGE_SIZE; - - /* page maybe head page, so we should get the buf by p, not the - * page - */ - tailroom = truesize - len - offset_in_page(p); - buf = (char *)((unsigned long)p & PAGE_MASK); - } else { - tailroom = truesize - len; - buf = p; - } + truesize = headroom ? PAGE_SIZE : truesize; + tailroom = truesize - len - headroom; + buf = p - headroom; len -= hdr_len; offset += hdr_padded_len;@@ -978,7 +966,8 @@ static struct sk_buff *receive_mergeable(struct net_device *dev, put_page(page); head_skb = page_to_skb(vi, rq, xdp_page, offset, len, PAGE_SIZE, false, - metasize, true); + metasize, + VIRTIO_XDP_HEADROOM); return head_skb; } break;@@ -1029,7 +1018,7 @@ static struct sk_buff *receive_mergeable(struct net_device *dev, rcu_read_unlock(); head_skb = page_to_skb(vi, rq, page, offset, len, truesize, !xdp_prog, - metasize, !!headroom); + metasize, headroom); curr_skb = head_skb; if (unlikely(!curr_skb))