* [PATCH]: fix queue limits in multiple qdiscs
@ 2003-11-13 14:50 Patrick McHardy
2003-11-17 14:22 ` jamal
0 siblings, 1 reply; 5+ messages in thread
From: Patrick McHardy @ 2003-11-13 14:50 UTC (permalink / raw)
To: David S. Miller; +Cc: netdev
[-- Attachment #1: Type: text/plain, Size: 250 bytes --]
This patch fixes multiple qdiscs exceeding their limits:
pfifo/pfifo_fast: by one packet
bfifo/red/gred: by one full-sized packet in bytes
RED and GRED should never reach their limits so this part is
more cosmetic than fix.
Best regards,
Patrick
[-- Attachment #2: 03-queue-limits.diff --]
[-- Type: text/plain, Size: 2967 bytes --]
# This is a BitKeeper generated patch for the following project:
# Project Name: Linux kernel tree
# This patch format is intended for GNU patch command version 2.5 or higher.
# This patch includes the following deltas:
# ChangeSet 1.1431 -> 1.1432
# net/sched/sch_generic.c 1.9 -> 1.10
# net/sched/sch_teql.c 1.8 -> 1.9
# net/sched/sch_fifo.c 1.6 -> 1.7
# net/sched/sch_gred.c 1.12 -> 1.13
# net/sched/sch_red.c 1.8 -> 1.9
#
# The following is the BitKeeper ChangeSet Log
# --------------------------------------------
# 03/11/12 kaber@trash.net 1.1432
# Fix queue limits in multiple qdiscs
# --------------------------------------------
#
diff -Nru a/net/sched/sch_fifo.c b/net/sched/sch_fifo.c
--- a/net/sched/sch_fifo.c Thu Nov 13 15:23:33 2003
+++ b/net/sched/sch_fifo.c Thu Nov 13 15:23:33 2003
@@ -47,7 +47,7 @@
{
struct fifo_sched_data *q = (struct fifo_sched_data *)sch->data;
- if (sch->stats.backlog <= q->limit) {
+ if (sch->stats.backlog + skb->len <= q->limit) {
__skb_queue_tail(&sch->q, skb);
sch->stats.backlog += skb->len;
sch->stats.bytes += skb->len;
@@ -108,7 +108,7 @@
{
struct fifo_sched_data *q = (struct fifo_sched_data *)sch->data;
- if (sch->q.qlen <= q->limit) {
+ if (sch->q.qlen < q->limit) {
__skb_queue_tail(&sch->q, skb);
sch->stats.bytes += skb->len;
sch->stats.packets++;
diff -Nru a/net/sched/sch_generic.c b/net/sched/sch_generic.c
--- a/net/sched/sch_generic.c Thu Nov 13 15:23:33 2003
+++ b/net/sched/sch_generic.c Thu Nov 13 15:23:33 2003
@@ -275,7 +275,7 @@
list = ((struct sk_buff_head*)qdisc->data) +
prio2band[skb->priority&TC_PRIO_MAX];
- if (list->qlen <= qdisc->dev->tx_queue_len) {
+ if (list->qlen < qdisc->dev->tx_queue_len) {
__skb_queue_tail(list, skb);
qdisc->q.qlen++;
qdisc->stats.bytes += skb->len;
diff -Nru a/net/sched/sch_gred.c b/net/sched/sch_gred.c
--- a/net/sched/sch_gred.c Thu Nov 13 15:23:33 2003
+++ b/net/sched/sch_gred.c Thu Nov 13 15:23:33 2003
@@ -110,7 +110,7 @@
unsigned long qave=0;
int i=0;
- if (!t->initd && skb_queue_len(&sch->q) <= sch->dev->tx_queue_len) {
+ if (!t->initd && skb_queue_len(&sch->q) < sch->dev->tx_queue_len) {
D2PRINTK("NO GRED Queues setup yet! Enqueued anyway\n");
goto do_enqueue;
}
@@ -175,7 +175,7 @@
if ((q->qave+qave) < q->qth_min) {
q->qcount = -1;
enqueue:
- if (q->backlog <= q->limit) {
+ if (q->backlog + skb->len <= q->limit) {
q->backlog += skb->len;
do_enqueue:
__skb_queue_tail(&sch->q, skb);
diff -Nru a/net/sched/sch_red.c b/net/sched/sch_red.c
--- a/net/sched/sch_red.c Thu Nov 13 15:23:33 2003
+++ b/net/sched/sch_red.c Thu Nov 13 15:23:33 2003
@@ -257,7 +257,7 @@
if (q->qave < q->qth_min) {
q->qcount = -1;
enqueue:
- if (sch->stats.backlog <= q->limit) {
+ if (sch->stats.backlog + skb->len <= q->limit) {
__skb_queue_tail(&sch->q, skb);
sch->stats.backlog += skb->len;
sch->stats.bytes += skb->len;
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH]: fix queue limits in multiple qdiscs
2003-11-13 14:50 [PATCH]: fix queue limits in multiple qdiscs Patrick McHardy
@ 2003-11-17 14:22 ` jamal
2003-11-17 14:47 ` Patrick McHardy
0 siblings, 1 reply; 5+ messages in thread
From: jamal @ 2003-11-17 14:22 UTC (permalink / raw)
To: Patrick McHardy; +Cc: David S. Miller, netdev
valid but Too cosmetic, IMO; (who cares when you are off by a few
bytes?)
cheers,
jamal
On Thu, 2003-11-13 at 09:50, Patrick McHardy wrote:
> This patch fixes multiple qdiscs exceeding their limits:
>
> pfifo/pfifo_fast: by one packet
> bfifo/red/gred: by one full-sized packet in bytes
>
> RED and GRED should never reach their limits so this part is
> more cosmetic than fix.
>
> Best regards,
> Patrick
>
>
> ______________________________________________________________________
>
> # This is a BitKeeper generated patch for the following project:
> # Project Name: Linux kernel tree
> # This patch format is intended for GNU patch command version 2.5 or higher.
> # This patch includes the following deltas:
> # ChangeSet 1.1431 -> 1.1432
> # net/sched/sch_generic.c 1.9 -> 1.10
> # net/sched/sch_teql.c 1.8 -> 1.9
> # net/sched/sch_fifo.c 1.6 -> 1.7
> # net/sched/sch_gred.c 1.12 -> 1.13
> # net/sched/sch_red.c 1.8 -> 1.9
> #
> # The following is the BitKeeper ChangeSet Log
> # --------------------------------------------
> # 03/11/12 kaber@trash.net 1.1432
> # Fix queue limits in multiple qdiscs
> # --------------------------------------------
> #
> diff -Nru a/net/sched/sch_fifo.c b/net/sched/sch_fifo.c
> --- a/net/sched/sch_fifo.c Thu Nov 13 15:23:33 2003
> +++ b/net/sched/sch_fifo.c Thu Nov 13 15:23:33 2003
> @@ -47,7 +47,7 @@
> {
> struct fifo_sched_data *q = (struct fifo_sched_data *)sch->data;
>
> - if (sch->stats.backlog <= q->limit) {
> + if (sch->stats.backlog + skb->len <= q->limit) {
> __skb_queue_tail(&sch->q, skb);
> sch->stats.backlog += skb->len;
> sch->stats.bytes += skb->len;
> @@ -108,7 +108,7 @@
> {
> struct fifo_sched_data *q = (struct fifo_sched_data *)sch->data;
>
> - if (sch->q.qlen <= q->limit) {
> + if (sch->q.qlen < q->limit) {
> __skb_queue_tail(&sch->q, skb);
> sch->stats.bytes += skb->len;
> sch->stats.packets++;
> diff -Nru a/net/sched/sch_generic.c b/net/sched/sch_generic.c
> --- a/net/sched/sch_generic.c Thu Nov 13 15:23:33 2003
> +++ b/net/sched/sch_generic.c Thu Nov 13 15:23:33 2003
> @@ -275,7 +275,7 @@
> list = ((struct sk_buff_head*)qdisc->data) +
> prio2band[skb->priority&TC_PRIO_MAX];
>
> - if (list->qlen <= qdisc->dev->tx_queue_len) {
> + if (list->qlen < qdisc->dev->tx_queue_len) {
> __skb_queue_tail(list, skb);
> qdisc->q.qlen++;
> qdisc->stats.bytes += skb->len;
> diff -Nru a/net/sched/sch_gred.c b/net/sched/sch_gred.c
> --- a/net/sched/sch_gred.c Thu Nov 13 15:23:33 2003
> +++ b/net/sched/sch_gred.c Thu Nov 13 15:23:33 2003
> @@ -110,7 +110,7 @@
> unsigned long qave=0;
> int i=0;
>
> - if (!t->initd && skb_queue_len(&sch->q) <= sch->dev->tx_queue_len) {
> + if (!t->initd && skb_queue_len(&sch->q) < sch->dev->tx_queue_len) {
> D2PRINTK("NO GRED Queues setup yet! Enqueued anyway\n");
> goto do_enqueue;
> }
> @@ -175,7 +175,7 @@
> if ((q->qave+qave) < q->qth_min) {
> q->qcount = -1;
> enqueue:
> - if (q->backlog <= q->limit) {
> + if (q->backlog + skb->len <= q->limit) {
> q->backlog += skb->len;
> do_enqueue:
> __skb_queue_tail(&sch->q, skb);
> diff -Nru a/net/sched/sch_red.c b/net/sched/sch_red.c
> --- a/net/sched/sch_red.c Thu Nov 13 15:23:33 2003
> +++ b/net/sched/sch_red.c Thu Nov 13 15:23:33 2003
> @@ -257,7 +257,7 @@
> if (q->qave < q->qth_min) {
> q->qcount = -1;
> enqueue:
> - if (sch->stats.backlog <= q->limit) {
> + if (sch->stats.backlog + skb->len <= q->limit) {
> __skb_queue_tail(&sch->q, skb);
> sch->stats.backlog += skb->len;
> sch->stats.bytes += skb->len;
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH]: fix queue limits in multiple qdiscs
2003-11-17 14:22 ` jamal
@ 2003-11-17 14:47 ` Patrick McHardy
2003-11-17 15:01 ` jamal
0 siblings, 1 reply; 5+ messages in thread
From: Patrick McHardy @ 2003-11-17 14:47 UTC (permalink / raw)
To: hadi; +Cc: David S. Miller, netdev
I don't have strong feelings about this, I just kept wondering why tc
displayes "pfifo limit 10 backlog 11".
Best regards,
Patrick
jamal wrote:
>valid but Too cosmetic, IMO; (who cares when you are off by a few
>bytes?)
>
>cheers,
>jamal
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH]: fix queue limits in multiple qdiscs
2003-11-17 14:47 ` Patrick McHardy
@ 2003-11-17 15:01 ` jamal
2003-11-19 1:34 ` David S. Miller
0 siblings, 1 reply; 5+ messages in thread
From: jamal @ 2003-11-17 15:01 UTC (permalink / raw)
To: Patrick McHardy; +Cc: David S. Miller, netdev
Like i said it is valid - but i wouldnt go out of my way to fix it
(especially when 2.6 is frozen).
cheers,
jamal
On Mon, 2003-11-17 at 09:47, Patrick McHardy wrote:
> I don't have strong feelings about this, I just kept wondering why tc
> displayes "pfifo limit 10 backlog 11".
>
> Best regards,
> Patrick
>
> jamal wrote:
>
> >valid but Too cosmetic, IMO; (who cares when you are off by a few
> >bytes?)
> >
> >cheers,
> >jamal
> >
>
>
>
^ permalink raw reply [flat|nested] 5+ messages in thread
* Re: [PATCH]: fix queue limits in multiple qdiscs
2003-11-17 15:01 ` jamal
@ 2003-11-19 1:34 ` David S. Miller
0 siblings, 0 replies; 5+ messages in thread
From: David S. Miller @ 2003-11-19 1:34 UTC (permalink / raw)
To: hadi; +Cc: kaber, netdev
On 17 Nov 2003 10:01:24 -0500
jamal <hadi@cyberus.ca> wrote:
> Like i said it is valid - but i wouldnt go out of my way to fix it
> (especially when 2.6 is frozen).
I agree with this sentiment, but only in part.
Considering:
1) The output from 'tc' is very confusing and quite stupid.
2) Patrick did all the work to fix this bug.
3) His patch is %100 straight forward, simply, and easy to verify.
I am going to apply his patch.
Thanks Patrick.
^ permalink raw reply [flat|nested] 5+ messages in thread
end of thread, other threads:[~2003-11-19 1:34 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2003-11-13 14:50 [PATCH]: fix queue limits in multiple qdiscs Patrick McHardy
2003-11-17 14:22 ` jamal
2003-11-17 14:47 ` Patrick McHardy
2003-11-17 15:01 ` jamal
2003-11-19 1:34 ` David S. Miller
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).