Thread (12 messages) 12 messages, 4 authors, 2013-01-10

[PATCH net-next] net_sched: more precise pkt_len computation

flat view
STALE5038d

From: Eric Dumazet <hidden>
Date: 2012-10-26 09:32:54
Subsystem: networking [general], the rest · Maintainers: "David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, Linus Torvalds

From: Eric Dumazet <redacted>

One long standing problem with TSO/GSO/GRO packets is that skb->len
doesnt represent a precise amount of bytes on wire.

Headers are only accounted for the first segment.
For TCP, thats typically 66 bytes per 1448 bytes segment missing,
an error of 4.5 %

As consequences :

1) TBF/CBQ/HTB/NETEM/... can send more bytes than the assigned limits.
2) Device stats are slightly under estimated as well.

Fix this by taking account of headers in qdisc_skb_cb(skb)->pkt_len
computation.

Packet schedulers should use pkt_len instead of skb->len for their
bandwidth limitations, and TSO enabled devices drivers could use pkt_len
if their statistics are not hardware assisted, and if they dont scratch
skb->cb[] first word.

Signed-off-by: Eric Dumazet <redacted>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Cc: Stephen Hemminger <redacted>
Cc: Paolo Valente <redacted>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Patrick McHardy <redacted>
---
 net/core/dev.c |   11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/net/core/dev.c b/net/core/dev.c
index b4978e2..cf2b5f4 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -2442,7 +2442,6 @@ static inline int __dev_xmit_skb(struct sk_buff *skb, struct Qdisc *q,
 	bool contended;
 	int rc;
 
-	qdisc_skb_cb(skb)->pkt_len = skb->len;
 	qdisc_calculate_pkt_len(skb, q);
 	/*
 	 * Heuristic to force contended enqueues to serialize on a
@@ -2579,6 +2578,16 @@ int dev_queue_xmit(struct sk_buff *skb)
 	skb->tc_verd = SET_TC_AT(skb->tc_verd, AT_EGRESS);
 #endif
 	trace_net_dev_queue(skb);
+	qdisc_skb_cb(skb)->pkt_len = skb->len;
+	if (skb_is_gso(skb)) {
+		unsigned int hdr_len = skb_transport_offset(skb);
+
+		if (skb_shinfo(skb)->gso_type & (SKB_GSO_TCPV4 | SKB_GSO_TCPV6))
+			hdr_len += tcp_hdrlen(skb);
+		else
+			hdr_len += sizeof(struct udphdr);
+		qdisc_skb_cb(skb)->pkt_len += (skb_shinfo(skb)->gso_segs - 1) * hdr_len;
+	}
 	if (q->enqueue) {
 		rc = __dev_xmit_skb(skb, q, dev, txq);
 		goto out;
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help