Re: [LARTC] HTB Message Storm HTB Delay <large number> > 5 secs

[Martin Devera <devik@cdi.cz>]

Yes of course ;) I have posted small patch to HTB
mailing list (qos-request@lists.cdi.cz) to solve it.
Here is copy:
--- sch_htb.c.old       Fri Dec  7 13:49:07 2001
+++ sch_htb.c   Fri Dec  7 13:48:32 2001
@@ -402,8 +402,12 @@ static enum htb_cmode
 htb_class_mode(struct htb_sched *q, struct htb_class *cl)
 {
     long toks,diff;
     diff = PSCHED_TDIFF_SAFE(q->now, cl->t_c, cl->mbuffer, 0);
+
+    /* BUG: this is only temporary workaround until next release */
+    if (diff < 0) diff = cl->mbuffer;
+
     HTB_DBG(6,3,"htb_cm diff=%ld\n",diff);

     /* check whether we are over ceil */
     if ((toks = (cl->ctokens + diff)) < 0) {
@@ -480,8 +484,11 @@ htb_dequeue_class(struct Qdisc *sch, str
     /* we have got skb, account it to victim and its parents
        and also to all ceil estimators under victim */
     while (cl) {
        diff = PSCHED_TDIFF_SAFE(q->now, cl->t_c, cl->mbuffer, 0);
+
+       /* BUG: this is only temporary workaround until next release */
+       if (diff < 0) diff = cl->mbuffer;

 #define HTB_ACCNT(T,B,R) toks = diff + cl->T; \
        if (toks > cl->B) toks = cl->B; \
            toks -= L2T(cl, cl->R, skb->len); \


On 8 Dec 2001, John Huttley wrote:

> 
> Hello I've set up a simple system. It seems to work for a short while,
> but now I've got batches of  100's of these messages.
> 
> Also I can't connect through that box any more. It's as if forwarding
> died.
> 
> 
> Has anyone any advice?
> 
> Regards
> 
> John
> 
> 
> 
> 
> _______________________________________________
> LARTC mailing list / LARTC@mailman.ds9a.nl
> http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://ds9a.nl/2.4Routing/
> 
> 




_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://ds9a.nl/2.4Routing/