[LARTC] HTB: suspicious_delay quick patch and explanation

[devik]

Hi,

seems like if I understand suspicious_delay... and
assertion (delay > 0 && delay < 5000000) failed messages
reason:
 It is not really error. It is ok but my assertions in code
 are rather too restrictive. Inner class can delay up to
 mdelay (1minute) but I tested it to 5sec.

Attached patch against htb 3.5 should fix it so that you should
no longer see any warnings. It it will be succesfull (no other
bug reports) it will go into 3.6 and we could advance from beta
to release status.

Use patch -p0 to apply it from linux kernel subdirectory.

regards,
devik

--- net/sched/sch_htb.c Tue May 21 21:34:51 2002
+++ net/sched/sch_htb_.c        Wed May 22 15:28:26 2002
@@ -411,7 +411,7 @@ static void htb_add_to_wait_tree (struct
 #ifdef HTB_DEBUG
        if (cl->pq_node.rb_color != -1) { BUG_TRAP(0); return; }
        HTB_CHCL(cl);
-       if ((delay <= 0 || delay >= 5000000) && net_ratelimit())
+       if ((delay <= 0 || delay > cl->mbuffer) && net_ratelimit())
                printk(KERN_ERR "HTB: suspicious delay in wait_tree d=%ld cl=%X 
h=%d\n",delay,cl->classid,debug_hint);
 #endif
        DEVIK_MSTART(9);