NetBSD Problem Report #25684

Received: (qmail 24083 invoked by uid 605); 23 May 2004 07:27:35 -0000
Message-Id: <20040523072733.03FBD11152@narn.netbsd.org>
Date: Sun, 23 May 2004 07:27:33 +0000 (UTC)
From: danielfdickinson@yahoo.ca
Sender: gnats-bugs-owner@NetBSD.org
Reply-To: danielfdickinson@yahoo.ca
To: gnats-bugs@gnats.NetBSD.org
Subject: tlp0 driver for Davicom DM9102 hangs on some transfers
X-Send-Pr-Version: www-1.0

>Number:         25684
>Category:       kern
>Synopsis:       tlp0 driver for Davicom DM9102 hangs on some transfers
>Confidential:   no
>Severity:       serious
>Priority:       high
>Responsible:    kern-bug-people
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Sun May 23 07:28:00 +0000 2004
>Closed-Date:    
>Last-Modified:  Sun Aug 08 02:24:00 +0000 2004
>Originator:     Daniel Dickinson
>Release:        NetBSD 2.0 BETA
>Organization:
>Environment:
NetBSD davor-netbsd 2.0_BETA NetBsd 2.0_BETA (GENERIC) #0: Sun May 16 02:57:38 EDT 2004  root@davor-netbsd:/usr/obj/sys/arch/i386/compiles/GENERIC i386
>Description:
This is for NetBSD 2.0_BETA (downloaded 2004-05-15)

I'm using a PC-Chips M599LMR Motherboard with an onboard
Davicom DM9102 (rh8 agrees with this) with DM9101 PHY (and
in parentheses dmesg reports AMD AM79C873; I'm not sure
about rh8) and attempting to transfer certain groups of
files causes the system to stop responding (no pretty
keyboard lights changing for caps lock et al, no hard drive
activity, no network activity, can't switch virtual
terminals, etc).

At first I thought it was the amount of traffic or nfs
since I was trying to do a cp -r of a large subtree on an
NFS server, however fmirror also hung the system although
at a different spot, and in another transfer an nfs
transfer hung the system on a particular file every time,
but switching to ftp made it work.  Weird.

There are no log messages when this happens to guide me as
to what's wrong.  Should I PR this and/or what other
information would it be helpful for me to send (if I can
find the file that was always failing would that help, or
is it likely site-specific?).  I am more than willing to
help with the debug, but I don't know where to start, never
having dug into the NetBSD code before, and not even sure
what debug tools are available.




>How-To-Repeat:
1: Transfer a large subtree by cp -r (over nfs)
2: Attempt to cp over nfs the file that caused the hang in 1:

or

1: Transfer a large subtree using fmirror

>Fix:

>Release-Note:
>Audit-Trail:

From: Martin Husemann <martin@duskware.de>
To: danielfdickinson@yahoo.ca
Cc: gnats-bugs@gnats.NetBSD.org
Subject: Re: kern/25684: tlp0 driver for Davicom DM9102 hangs on some transfers
Date: Sun, 23 May 2004 13:33:14 +0200

 Could you build a kernel with "options DDB" enabled and try to break into
 ddb when the system hangs? Look at "man ddb"; on i386 you can use Ctrl-Alt-ESC
 to break into ddb, which will say

  db> 

 and you can use "c" to continue. If this still works when the system hangs,
 you could use the "tr" command to get a stack trace - that should help
 us a lot to diagnose the system.

 I fear that Ctrl-Alt-ESC will not work in your locked situation though.

 Martin

From: "Daniel F. Dickinson" <danielfdickinson@yahoo.ca>
To: Martin Husemann <martin@duskware.de>
Cc: gnats-bugs@gnats.NetBSD.org
Subject: Re: kern/25684: tlp0 driver for Davicom DM9102 hangs on some transfers
Date: Fri, 28 May 2004 16:41:30 -0400

 Martin Husemann wrote:

 > Could you build a kernel with "options DDB" enabled and try to break into
 > ddb when the system hangs? Look at "man ddb"; on i386 you can use Ctrl-Alt-ESC
 > to break into ddb, which will say
 > 
 >  db> 
 > 
 > and you can use "c" to continue. If this still works when the system hangs,
 > you could use the "tr" command to get a stack trace - that should help
 > us a lot to diagnose the system.
 > 
 > I fear that Ctrl-Alt-ESC will not work in your locked situation though.
 > 

 Unfortunately you are correct.  It works before the hanging transfer, 
 but the hang appears total.  If I can manage it I'll try with the serial 
 debugger but that may take a while, and I'm not sure it will work any 
 better.

 Thanks,

 Daniel

 -- 
 Daniel F. Dickinson:
 "Time does not die; the circle is not round."
 http://www.geocities.com/danielfdickinson/index.html
 http://www.indigo.ca : bookstore
 http://www.talkorigins.org : for origins of life archives


From: "Daniel F. Dickinson" <danielfdickinson@yahoo.ca>
To: gnats-bugs@NetBSD.org
Cc:  
Subject: kern/25684
Date: Sat, 07 Aug 2004 21:52:40 -0400

 I have tried a serial debug but when the system hangs it doesn't respond 
 to the break signal from the remove debug machine.  What else can I do 
 to help with the debug?  I don't know where to start looking in the code 
 to figure out if it's a coding error or chipset bug, so any advice would 
 be appreciated.

 I had intended to use this system as my main workstation with NetBSD 2.0 
 but it doesn't look like I will be able to because I need a network 
 connection that doesn't hang the machine when I try a large (or many 
 file) transfer.
 <sigh>


 -- 
 Daniel F. Dickinson:
 "Time does not die; the circle is not round."
 http://www.geocities.com/danielfdickinson/index.html
 http://www.indigo.ca : bookstore
 http://www.talkorigins.org : for origins of life archives

>Unformatted:

Home	PR Database Search