netdev
[Top] [All Lists]

Re: [PATCH] IPv6: Refine IPv6 Address Validation Timer

To: kuznet@xxxxxxxxxxxxx
Subject: Re: [PATCH] IPv6: Refine IPv6 Address Validation Timer
From: YOSHIFUJI Hideaki / 吉藤英明 <yoshfuji@xxxxxxxxxxxxxx>
Date: Sat, 28 Sep 2002 01:14:39 +0900 (JST)
Cc: davem@xxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, netdev@xxxxxxxxxxx, usagi@xxxxxxxxxxxxxx
In-reply-to: <200209271530.TAA21206@xxxxxxxxxxxxx>
Organization: USAGI Project
References: <20020928.001617.91124319.yoshfuji@xxxxxxxxxxxxxx> <200209271530.TAA21206@xxxxxxxxxxxxx>
Sender: netdev-bounce@xxxxxxxxxxx
In article <200209271530.TAA21206@xxxxxxxxxxxxx> (at Fri, 27 Sep 2002 19:30:14 
+0400 (MSD)), kuznet@xxxxxxxxxxxxx says:

> Let's only delete this:
> 
> > +   if (time_before(now + HZ/2, jiffies)) {
> > +           ADBG((KERN_WARNING 
> > +                 "addrconf_verify(): too slow; jiffies - now = %ld\n",
> > +                 (long)jiffies - (long)now));
> > +   }
> 
> I do not understand how this survived. If you worry about infinite
> spinning in loop you could make this check real, f.e. breaking loop
> when jiffies >= now+2. In this form it is just mud.

hmm, I supposed it finally exited the loop (and then we got above warn).
Code around end of patch (*) prevent continuous run of this function.
If it is (almost) meaningless, just delete it.

(*) Oops, I slipped at (almost) end of the patch when making patch... sorry;
I inteded to refine timing into ~0.5 sec at worst; so, please change
        addr_chk_timer.expires = time_before(next, jiffies + HZ) ? jiffies + HZ 
: next;
to
        addr_chk_timer.expires = time_before(next, jiffies + HZ/2) ? jiffies + 
HZ/2 : next;
.

Thanks.


(do I need to resend complete patch?)

-- 
Hideaki YOSHIFUJI @ USAGI Project <yoshfuji@xxxxxxxxxxxxxx>
GPG FP: 9022 65EB 1ECF 3AD1 0BDF  80D8 4807 F894 E062 0EEA


<Prev in Thread] Current Thread [Next in Thread>