openwrt/target/linux/generic/backport-5.4/050-gro-fix-napi_gro_frags-Fast-GRO-breakage-due-to-IP-a.patch
John Audia 0d2873df15 kernel: bump 5.4 to 5.4.114
Manually rebased*
  generic/backport-5.4/700-v5.5-net-core-allow-fast-GRO-for-skbs-with-Ethernet-heade.patch

Added new backport*
  generic/backport-5.4/050-gro-fix-napi_gro_frags-Fast-GRO-breakage-due-to-IP-a.patch

All others updated automatically.

The new backport was included based on this[1] upstream commit that will be
mainlined soon.  This change is needed because Eric Dumazet's check for
NET_IP_ALIGN (landed in 5.4.114) causes huge slowdowns on drivers which use
napi_gro_frags().

Build system: x86_64
Build-tested: ipq806x/R7800
Run-tested: ipq806x/R7800

No dmesg regressions, everything functional

*Credit to Alexander Lobakin
1. https://git.kernel.org/pub/scm/linux/kernel/git/netdev/net.git/commit/?id=7ad18ff6449cbd6beb26b53128ddf56d2685aa93

Signed-off-by: John Audia <graysky@archlinux.us>
2021-04-30 23:48:37 +02:00

76 lines
2.9 KiB
Diff

From 3007b05df4301aad179acc6ca1c3645785576df6 Mon Sep 17 00:00:00 2001
From: Alexander Lobakin <alobakin@pm.me>
Date: Mon, 19 Apr 2021 12:53:06 +0000
Subject: gro: fix napi_gro_frags() Fast GRO breakage due to IP
alignment check
Commit 7ad18ff6449cbd6beb26b53128ddf56d2685aa93 upstream.
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.
From v1 [0]:
- inline tiny skb_gro_reset_offset() to let the code be optimized
more efficively (esp. for the !NET_IP_ALIGN case) (Eric);
- pull in Reviewed-by from Eric.
[0] https://lore.kernel.org/netdev/20210418114200.5839-1-alobakin@pm.me
Fixes: 38ec4944b593 ("gro: ensure frag0 meets IP header alignment")
Reviewed-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Alexander Lobakin <alobakin@pm.me>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
net/core/dev.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -5395,7 +5395,7 @@ static struct list_head *gro_list_prepar
return head;
}
-static void skb_gro_reset_offset(struct sk_buff *skb)
+static inline 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];
@@ -5407,7 +5407,7 @@ static void skb_gro_reset_offset(struct
if (skb_mac_header(skb) == skb_tail_pointer(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),
@@ -5640,7 +5640,7 @@ gro_result_t napi_gro_receive(struct nap
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);
@@ -5733,7 +5733,7 @@ static struct sk_buff *napi_frags_skb(st
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);