[PATCH net-next 3/3] ipv6: use skb coalescing in reassembly
flat view
STALE5254d
From: Eric Dumazet <hidden>
Date: 2012-05-19 13:02:38
Subsystem:
networking [general], networking [ipv4/ipv6], the rest · Maintainers:
"David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, David Ahern, Ido Schimmel, Linus Torvalds
From: Eric Dumazet <redacted> ip6_frag_reasm() can use skb_try_coalesce() to build optimized skb, reducing memory used by them (truesize), and reducing number of cache line misses and overhead for the consumer. Signed-off-by: Eric Dumazet <redacted> Cc: Alexander Duyck <redacted> --- net/ipv6/reassembly.c | 26 ++++++++++++++++++++------ 1 file changed, 20 insertions(+), 6 deletions(-)
diff --git a/net/ipv6/reassembly.c b/net/ipv6/reassembly.c
index 5d32dfa..4ff9af6 100644
--- a/net/ipv6/reassembly.c
+++ b/net/ipv6/reassembly.c@@ -415,6 +415,7 @@ static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *prev, struct sk_buff *fp, *head = fq->q.fragments; int payload_len; unsigned int nhoff; + int sum_truesize; fq_kill(fq);
@@ -484,20 +485,33 @@ static int ip6_frag_reasm(struct frag_queue *fq, struct sk_buff *prev, head->mac_header += sizeof(struct frag_hdr); head->network_header += sizeof(struct frag_hdr); - skb_shinfo(head)->frag_list = head->next; skb_reset_transport_header(head); skb_push(head, head->data - skb_network_header(head)); - for (fp=head->next; fp; fp = fp->next) { - head->data_len += fp->len; - head->len += fp->len; + sum_truesize = head->truesize; + for (fp = head->next; fp;) { + bool headstolen; + int delta; + struct sk_buff *next = fp->next; + + sum_truesize += fp->truesize; if (head->ip_summed != fp->ip_summed) head->ip_summed = CHECKSUM_NONE; else if (head->ip_summed == CHECKSUM_COMPLETE) head->csum = csum_add(head->csum, fp->csum); - head->truesize += fp->truesize; + + if (skb_try_coalesce(head, fp, &headstolen, &delta)) { + kfree_skb_partial(fp, headstolen); + } else { + if (!skb_shinfo(head)->frag_list) + skb_shinfo(head)->frag_list = fp; + head->data_len += fp->len; + head->len += fp->len; + head->truesize += fp->truesize; + } + fp = next; } - atomic_sub(head->truesize, &fq->q.net->mem); + atomic_sub(sum_truesize, &fq->q.net->mem); head->next = NULL; head->dev = dev;