diff options
author | Eric Dumazet <eric.dumazet@gmail.com> | 2010-09-20 16:16:27 -0400 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2010-09-24 17:41:04 -0400 |
commit | 59104f062435c7816e39ee5ed504a69cb8037f10 (patch) | |
tree | cc77c6202d01b9c152634b836fa06e77b27d14fc /net/ipv4/ip_output.c | |
parent | a02cec2155fbea457eca8881870fd2de1a4c4c76 (diff) |
ip: take care of last fragment in ip_append_data
While investigating a bit, I found ip_fragment() slow path was taken
because ip_append_data() provides following layout for a send(MTU +
N*(MTU - 20)) syscall :
- one skb with 1500 (mtu) bytes
- N fragments of 1480 (mtu-20) bytes (before adding IP header)
last fragment gets 17 bytes of trail data because of following bit:
if (datalen == length + fraggap)
alloclen += rt->dst.trailer_len;
Then esp4 adds 16 bytes of data (while trailer_len is 17... hmm...
another bug ?)
In ip_fragment(), we notice last fragment is too big (1496 + 20) > mtu,
so we take slow path, building another skb chain.
In order to avoid taking slow path, we should correct ip_append_data()
to make sure last fragment has real trail space, under mtu...
Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/ipv4/ip_output.c')
-rw-r--r-- | net/ipv4/ip_output.c | 9 |
1 files changed, 6 insertions, 3 deletions
diff --git a/net/ipv4/ip_output.c b/net/ipv4/ip_output.c index e42762023c27..3551b6dc7419 100644 --- a/net/ipv4/ip_output.c +++ b/net/ipv4/ip_output.c | |||
@@ -926,16 +926,19 @@ alloc_new_skb: | |||
926 | !(rt->dst.dev->features&NETIF_F_SG)) | 926 | !(rt->dst.dev->features&NETIF_F_SG)) |
927 | alloclen = mtu; | 927 | alloclen = mtu; |
928 | else | 928 | else |
929 | alloclen = datalen + fragheaderlen; | 929 | alloclen = fraglen; |
930 | 930 | ||
931 | /* The last fragment gets additional space at tail. | 931 | /* The last fragment gets additional space at tail. |
932 | * Note, with MSG_MORE we overallocate on fragments, | 932 | * Note, with MSG_MORE we overallocate on fragments, |
933 | * because we have no idea what fragment will be | 933 | * because we have no idea what fragment will be |
934 | * the last. | 934 | * the last. |
935 | */ | 935 | */ |
936 | if (datalen == length + fraggap) | 936 | if (datalen == length + fraggap) { |
937 | alloclen += rt->dst.trailer_len; | 937 | alloclen += rt->dst.trailer_len; |
938 | 938 | /* make sure mtu is not reached */ | |
939 | if (datalen > mtu - fragheaderlen - rt->dst.trailer_len) | ||
940 | datalen -= ALIGN(rt->dst.trailer_len, 8); | ||
941 | } | ||
939 | if (transhdrlen) { | 942 | if (transhdrlen) { |
940 | skb = sock_alloc_send_skb(sk, | 943 | skb = sock_alloc_send_skb(sk, |
941 | alloclen + hh_len + 15, | 944 | alloclen + hh_len + 15, |