[NET_SCHED]: Fix prio/ingress classification logic error

[NET_SCHED]: Fix prio/ingress classification logic error

[Patrick McHardy]

[-- Attachment #1: Type: text/plain, Size: 198 bytes --]

This is the final fix for the problem Peter reported.
Turns out most other schedulers get it right, the only
other case is ingress setting skb->tc_index to the
uninitialized value of res.classid.



[-- Attachment #2: x --]
[-- Type: text/plain, Size: 2133 bytes --]

[NET_SCHED]: Fix prio/ingress classification logic error

Fix handling of empty or completely non-matching filter chains. In
that case -1 is returned and tcf_result is uninitialized, the
qdisc should fall back to default classification in that case.

Noticed by PJ Waskiewicz <peter.p.waskiewicz.jr@intel.com>.

Signed-off-by: Patrick McHardy <kaber@trash.net>

---
commit c98436c161ff45850346b08f89cfb444be7c7817
tree 07e3ea0f508243f41e70645b6c358e23244ea0b4
parent e4903fb59590f86190280a549420f6cb85bd7f7e
author Patrick McHardy <kaber@trash.net> Thu, 26 Jul 2007 13:32:26 +0200
committer Patrick McHardy <kaber@trash.net> Thu, 26 Jul 2007 13:32:26 +0200

 net/sched/sch_ingress.c |    3 +--
 net/sched/sch_prio.c    |    7 +++----
 2 files changed, 4 insertions(+), 6 deletions(-)

diff --git a/net/sched/sch_ingress.c b/net/sched/sch_ingress.c
index 51f16b0..2d32fd2 100644
--- a/net/sched/sch_ingress.c
+++ b/net/sched/sch_ingress.c
@@ -158,9 +158,8 @@ static int ingress_enqueue(struct sk_buff *skb,struct Qdisc *sch)
 			break;
 		case TC_ACT_RECLASSIFY:
 		case TC_ACT_OK:
-		case TC_ACT_UNSPEC:
-		default:
 			skb->tc_index = TC_H_MIN(res.classid);
+		default:
 			result = TC_ACT_OK;
 			break;
 	}
diff --git a/net/sched/sch_prio.c b/net/sched/sch_prio.c
index 2d8c084..71bafde 100644
--- a/net/sched/sch_prio.c
+++ b/net/sched/sch_prio.c
@@ -38,9 +38,11 @@ prio_classify(struct sk_buff *skb, struct Qdisc *sch, int *qerr)
 	struct prio_sched_data *q = qdisc_priv(sch);
 	u32 band = skb->priority;
 	struct tcf_result res;
+	int err;
 
 	*qerr = NET_XMIT_BYPASS;
 	if (TC_H_MAJ(skb->priority) != sch->handle) {
+		err = tc_classify(skb, q->filter_list, &res);
 #ifdef CONFIG_NET_CLS_ACT
 		switch (tc_classify(skb, q->filter_list, &res)) {
 		case TC_ACT_STOLEN:
@@ -49,11 +51,8 @@ prio_classify(struct sk_buff *skb, struct Qdisc *sch, int *qerr)
 		case TC_ACT_SHOT:
 			return NULL;
 		}
-
-		if (!q->filter_list ) {
-#else
-		if (!q->filter_list || tc_classify(skb, q->filter_list, &res)) {
 #endif
+		if (!q->filter_list || err < 0) {
 			if (TC_H_MAJ(band))
 				band = 0;
 			band = q->prio2band[band&TC_PRIO_MAX];

RE: [NET_SCHED]: Fix prio/ingress classification logic error

[Waskiewicz Jr, Peter P]

> This is the final fix for the problem Peter reported.
> Turns out most other schedulers get it right, the only other 
> case is ingress setting skb->tc_index to the uninitialized 
> value of res.classid.

I've applied this and tested prio and rr, and everything is happy again.
Thanks for finding this and putting the patch together.

I got lost in looking for the bug in the u32_classify() code, and missed
the easier defect...

Thanks again Patrick,
-PJ Waskiewicz