netdev - [PATCH][NET_SCHED] sch_sfq: prevent unnecessary reordering

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Mon, 28 Apr 2008 09:04:53 +0000
From:	Jarek Poplawski <jarkao2@...il.com>
To:	David Miller <davem@...emloft.net>
Cc:	Patrick McHardy <kaber@...sh.net>, netdev@...r.kernel.org
Subject: [PATCH][NET_SCHED] sch_sfq: prevent unnecessary reordering

On 27-04-2008 16:22, Jarek Poplawski wrote:
> [NET_SCHED] sch_sfq: fix queue limiting while enqueuing
> 

David, as Patrick noticed this patch was wrong and I withdraw it.

I still can't see why sfq should have such a low limit, but I think,
there is at least something to improve in the way it's imposed. BTW,
my previous idea to check this before classifying was wrong too. But,
IMHO, this could be done better than now like in this patch.

Regards,
Jarek P.

------------------------->

[NET_SCHED] sch_sfq: prevent unnecessary reordering

Current check of queue limit in sfq_enqueue() isn't optimal: there
is really not much more needed to prevent unnecessary dropping and
possible reordering.


Signed-off-by: Jarek Poplawski <jarkao2@...il.com>

---

 net/sched/sch_sfq.c |   17 ++++++++++++-----
 1 files changed, 12 insertions(+), 5 deletions(-)

diff --git a/net/sched/sch_sfq.c b/net/sched/sch_sfq.c
index a20e2ef..9afef88 100644
--- a/net/sched/sch_sfq.c
+++ b/net/sched/sch_sfq.c
@@ -298,12 +298,19 @@ sfq_enqueue(struct sk_buff *skb, struct Qdisc *sch)
 		q->hash[x] = hash;
 	}
 
-	/* If selected queue has length q->limit, this means that
-	 * all another queues are empty and that we do simple tail drop,
-	 * i.e. drop _this_ packet.
+	/*
+	 * If the queue is full some dropping is necessary. Let's check
+	 * whether this slot is to be selected to avoid reordering at
+	 * least. If max_depth == 1 (very unprobable) another slot could
+	 * be chosen to drop, but there is no reason to do this (and if
+	 * it were this slot yet the reordering would happen.)
 	 */
-	if (q->qs[x].qlen >= q->limit)
-		return qdisc_drop(skb, sch);
+	if (sch->q.qlen >= q->limit) {
+		sfq_index d = q->max_depth;
+
+		if ((d > 1 && x == q->dep[d + SFQ_DEPTH].next) || d == 1)
+			return qdisc_drop(skb, sch);
+	}
 
 	sch->qstats.backlog += skb->len;
 	__skb_queue_tail(&q->qs[x], skb);
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html