Stability problem (maybe stge related, maybe satalink)

Salut,

I have some issues with NetBSD 2.1 on my dual 600MHz ev56 alpha. After
a day or two, it usually just hangs and doesn't react on anything
but the reset button. I don't remember this behavior from tthe time
before I put a satalink controller (Silicon Image) into it, but then
again I had a different machine back then: a dual 400MHz one.

The last thing that happens before the hangups is:

stge0: device timeout
stge0: DMA wait timed out
<machine is hung>

Also, pcictl list on the PCI bus with the SATA controller causes
a reboot.

Some more information about the machine can be found in the dmesg
which I hopefully won't forget to attach.

Any ideas?

				Tonnerre

Unrecognized boot flag '0'.
consinit: not using prom console
Copyright (c) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005
    The NetBSD Foundation, Inc.  All rights reserved.
Copyright (c) 1982, 1986, 1989, 1991, 1993

Re: Stability problem (maybe stge related, maybe satalink)

Tonnerre LOMBARD wrote:
> Salut,
>
> I have some issues with NetBSD 2.1 on my dual 600MHz ev56 alpha. After
> a day or two, it usually just hangs and doesn't react on anything
> but the reset button. I don't remember this behavior from tthe time
> before I put a satalink controller (Silicon Image) into it, but then
> again I had a different machine back then: a dual 400MHz one.
>
> The last thing that happens before the hangups is:
>
> stge0: device timeout
> stge0: DMA wait timed out
> <machine is hung>
>
> Also, pcictl list on the PCI bus with the SATA controller causes
> a reboot.
>
> Some more information about the machine can be found in the dmesg
> which I hopefully won't forget to attach.
>
> Any ideas?
>
> 				Tonnerre
>   
> ------------------------------------------------------------------------
>
>   

Can you try with a non-MP kernel? I saw (infrequent) hangs until I changed

Re: Stability problem (maybe stge related, maybe satalink)

Salut,

On Fri, Jun 02, 2006 at 04:20:03PM +0200, Tobias Nygren wrote:
> Can you try with a non-MP kernel? I saw (infrequent) hangs until I changed
> to uniprocessor. My initial testing indicated that interrupts were lost when
> running MP. This seemed to affect all PCI boards, but was only fatal on the
> satalink ones. I'm not confident enough to try to debug it further.
> AS4100 with two satalink has been up for 2 months without a hang after
> the other cpu boards were pulled.

The problem is that the box is mainly used as a compile box, so disabling
one CPU has quite negativish effects on the performance of this type of
operations. But well.

It might however be worth trying to kick a kgdb in some way... If I build
a cross-gdb, I might also debug it from my non-alpha workstation over the
serial console, I guess.

Let's see...

				Tonnerre

Re: Package binaries for NetBSD/alpha 2.1 / pkgsrc-2006Q1

Hi,

the new files resulting from a new round of bulk builds for
NetBSD/alpha 2.1 has been merged into

   ftp://ftp.NetBSD.org/pub/pkgsrc/packages-2006Q1/NetBSD-2.1/alpha/

This time the volume was 410MB new packages.

The source this was built from was updated May 30, 2006.  A
re-update and rebuild of the packages on the branch has just been
started, and I will send a new announcement once the result of
the new build has been merged into the above.

Regards,

- Håvard

gcc4 on alpha

Running gcc4 userland and kernel on alpha since last night.
Seems to work fine, except that sbin/disklabel/main.c needs a fix:

cc -O2  -Wall -Wstrict-prototypes -Wmissing-prototypes -Wpointer-arith 
-Wno-sign-compare -Wno-traditional -Wreturn-type -Wswitch -Wshadow 
-Wcast-qual -Wwrite-strings  -Werror -mieee      -c    main.c
cc1: warnings being treated as errors
main.c: In function 'write_bootarea':
main.c:924: warning: dereferencing type-punned pointer will break 
strict-aliasing rules
*** Error code 1

I don't know what the best fix for this is, I just compiled disklabel
with  -fno-strict-aliasing, which seems to have worked :)

-Tobias

Re: alpha pkgsrc bulk build, 2006Q1 for 3.0 uploaded

On 07-Jun-2006 Erik E. Fair wrote:
> How long did the bulk build take, and on what configuration of Alpha 
> did you do the build?

I started it on the branch date.. umm.. so just under 2 months.  It's a 600Mhz
164LX, with 256MB of ram.

---
Tim Rightnour <root <at> garbled.net>
NetBSD: Free multi-architecture OS http://www.netbsd.org/
Genecys: Open Source 3D MMORPG: http://www.genecys.org/

Re: alpha pkgsrc bulk build, 2006Q1 for 3.0 uploaded

On Jun 7, 2006, at 2:54 PM, Tim Rightnour wrote:
>> How long did the bulk build take, and on what configuration of Alpha
>> did you do the build?
>
> I started it on the branch date.. umm.. so just under 2 months.  It's 
> a 600Mhz
> 164LX, with 256MB of ram.

   Yowza...RAM-starved I'll bet.

          -Dave

--
Dave McGuire
Cape Coral, FL

Re: alpha pkgsrc bulk build, 2006Q1 for 3.0 uploaded

At 11:54 -0700 6/7/06, Tim Rightnour wrote:
>On 07-Jun-2006 Erik E. Fair wrote:
>>  How long did the bulk build take, and on what configuration of Alpha
>>  did you do the build?
>
>I started it on the branch date.. umm.. so just under 2 months.  It's a 600Mhz
>164LX, with 256MB of ram.

Hmmm, this sounds like a job for a cluster ...

	Erik <fair <at> netbsd.org>

Re: alpha pkgsrc bulk build, 2006Q1 for 3.0 uploaded

On 07-Jun-2006 Erik E. Fair wrote:
> Hmmm, this sounds like a job for a cluster ...

I'm trying again with a distcc to a remote i386.  I dunno if it's any faster
really..  doing pkgsrc with -jN is rather flaky though.. and clustered builds
on pkgsrc are more or less a pipe dream for now.

In reality.. it doesnt bother me.  I'm not presently using the alpha for
anything.. so it can just sit there in the corner grinding it's life away for
all eternity.

---
Tim Rightnour <root <at> garbled.net>
NetBSD: Free multi-architecture OS http://www.netbsd.org/
Genecys: Open Source 3D MMORPG: http://www.genecys.org/

Re: alpha pkgsrc bulk build, 2006Q1 for 3.0 uploaded

On 07-Jun-2006 Dave McGuire wrote:
>    Yowza...RAM-starved I'll bet.

Probably..  after the build finished I actually upgraded that box to 512MB.

---
Tim Rightnour <root <at> garbled.net>
NetBSD: Free multi-architecture OS http://www.netbsd.org/
Genecys: Open Source 3D MMORPG: http://www.genecys.org/