[PATCH net] gro: fix napi_gro_frags() Fast GRO breakage due to IP alignment check

From: Alexander Lobakin <alobakin@pm.me>
To: "David S. Miller" <davem@davemloft.net>,
	Jakub Kicinski <kuba@kernel.org>
Cc: "Alexei Starovoitov" <ast@kernel.org>,
	"Andrii Nakryiko" <andriin@fb.com>,
	"Daniel Borkmann" <daniel@iogearbox.net>,
	"Eric Dumazet" <edumazet@google.com>,
	"Wei Wang" <weiwan@google.com>,
	"Cong Wang" <cong.wang@bytedance.com>,
	"Taehee Yoo" <ap420073@gmail.com>,
	"Björn Töpel" <bjorn@kernel.org>,
	"Michael S. Tsirkin" <mst@redhat.com>,
	"Alexander Lobakin" <alobakin@pm.me>,
	netdev@vger.kernel.org, linux-kernel@vger.kernel.org
Subject: [PATCH net] gro: fix napi_gro_frags() Fast GRO breakage due to IP alignment check
Date: Sun, 18 Apr 2021 11:42:54 +0000	[thread overview]
Message-ID: <20210418114200.5839-1-alobakin@pm.me> (raw)

Commit 38ec4944b593 ("gro: ensure frag0 meets IP header alignment")
did the right thing, but missed the fact that napi_gro_frags() logics
calls for skb_gro_reset_offset() *before* pulling Ethernet header
to the skb linear space.
That said, the introduced check for frag0 address being aligned to 4
always fails for it as Ethernet header is obviously 14 bytes long,
and in case with NET_IP_ALIGN its start is not aligned to 4.

Fix this by adding @nhoff argument to skb_gro_reset_offset() which
tells if an IP header is placed right at the start of frag0 or not.
This restores Fast GRO for napi_gro_frags() that became very slow
after the mentioned commit, and preserves the introduced check to
avoid silent unaligned accesses.

Fixes: 38ec4944b593 ("gro: ensure frag0 meets IP header alignment")
Signed-off-by: Alexander Lobakin <alobakin@pm.me>
---
 net/core/dev.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/net/core/dev.c b/net/core/dev.c
index 1f79b9aa9a3f..965d5f9b6fee 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -5914,7 +5914,7 @@ static struct list_head *gro_list_prepare(struct napi_struct *napi,
 	return head;
 }

-static void skb_gro_reset_offset(struct sk_buff *skb)
+static void skb_gro_reset_offset(struct sk_buff *skb, u32 nhoff)
 {
 	const struct skb_shared_info *pinfo = skb_shinfo(skb);
 	const skb_frag_t *frag0 = &pinfo->frags[0];
@@ -5925,7 +5925,7 @@ static void skb_gro_reset_offset(struct sk_buff *skb)

 	if (!skb_headlen(skb) && pinfo->nr_frags &&
 	    !PageHighMem(skb_frag_page(frag0)) &&
-	    (!NET_IP_ALIGN || !(skb_frag_off(frag0) & 3))) {
+	    (!NET_IP_ALIGN || !((skb_frag_off(frag0) + nhoff) & 3))) {
 		NAPI_GRO_CB(skb)->frag0 = skb_frag_address(frag0);
 		NAPI_GRO_CB(skb)->frag0_len = min_t(unsigned int,
 						    skb_frag_size(frag0),
@@ -6143,7 +6143,7 @@ gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb)
 	skb_mark_napi_id(skb, napi);
 	trace_napi_gro_receive_entry(skb);

-	skb_gro_reset_offset(skb);
+	skb_gro_reset_offset(skb, 0);

 	ret = napi_skb_finish(napi, skb, dev_gro_receive(napi, skb));
 	trace_napi_gro_receive_exit(ret);
@@ -6232,7 +6232,7 @@ static struct sk_buff *napi_frags_skb(struct napi_struct *napi)
 	napi->skb = NULL;

 	skb_reset_mac_header(skb);
-	skb_gro_reset_offset(skb);
+	skb_gro_reset_offset(skb, hlen);

 	if (unlikely(skb_gro_header_hard(skb, hlen))) {
 		eth = skb_gro_header_slow(skb, hlen, 0);
--
2.31.1


