Subject: Re: Unpredictable reboots.
To: Peter Seebach <>
From: Andreas Wrede <>
List: port-i386
Date: 03/05/2005 11:18:32
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=US-ASCII; format=flowed

On 5-Mar-05, at 9:24 AM, Peter Seebach wrote:

> In message <d0c1op$oci$>, Matthias Scheler writes:
>> In article <>,
>> (Peter Seebach) writes:
>>> Anyone else seeing anything like this?
>> No:
> Interesting.  I put in a new power supply, just for luck, and went 
> back to a
> DIAGNOSTIC kernel.  This morning, it was wedged at KASSERT(to_ticks >= 
> 0).
> Backtrace said it was executing sendmail, and the backtrace was through
> tcp_output.
> I then saw the exact same panic FIVE more times.  Always within about 
> ten
> seconds of starting sendmail.
> So, I did the obvious thing; made a new kernel which prints to_ticks 
> and sets
> it to zero.
> It's hit that 31 times in 11 minutes.

> I think it's safe to say I can reproduce this.  What this doesn't 
> leave me
> with is any clue how to fix it, or get better debugging info.  But at 
> least I
> have the machine running again in the mean time.  Of course, as you'd 
> expect,
> this is a production server, and I can't make it happen on anything 
> else.
>    5 to_ticks: -150
>    6 to_ticks: -200
>    6 to_ticks: -250
>    9 to_ticks: -300
>    3 to_ticks: -350
> Not sure if the distribution of times means anything.  I suppose next 
> up would
> be adding stack traces and adding debugging code to whatever's calling 
> this.
> And yes, I applied the patches indicated in the PR (29134).

"Me too".  I also applied the diffs from Charles Hannum's Jan 26 commit 
to src/sys/netinet: tcp_input.c tcp_timer.c tcp_var.h.  And I am seeing 
the same to_ticks counts.


content-type: application/pgp-signature; x-mac-type=70674453;
content-description: This is a digitally signed message part
content-disposition: inline; filename=PGP.sig
content-transfer-encoding: 7bit

Version: GnuPG v1.2.4 (Darwin)