KIYOHARA Takashi | 6 Feb 14:39 2011
Picon

panic ibm4xx machines

Hi! all,

My OpenBlockS (evbppc) panics when boot time.

## Transferring control to NetBSD stage-2 loader (at address 00025000) ...
+40000000,
00025000-40000000 -> 00600000-40000000 = 3fa00000
panic: evcnt_attach_static: evcnt non initialized: group=<cpu> name=<tlbmiss>
Stopped in pid 0.1 (system) at  0x4d9e8:        lwz     r0, 0x14(r1)
db> bt
0x005ffe40: at 0x221c88
0x005ffe90: at 0x215328
0x005ffeb0: at 0x1d3538
0x005ffef0: at 0x1d101c
0x005fff10: at 0x250e8
db>

I think this reason is:
http://cvsweb.netbsd.org/bsdweb.cgi/src/sys/kern/subr_evcnt.c.diff?r1=1.7&r2=1.8&only_with_tag=MAIN&f=h

evcnt_attach_static() is called in
arch/powerpc/ibm4xx/pmap.c:pmap_bootstrap().

Please better fix.  ;-<

Thanks,
--
kiyohara

(Continue reading)

Martin Husemann | 6 Feb 19:43 2011
Picon

Re: panic ibm4xx machines

On Sun, Feb 06, 2011 at 10:39:55PM +0900, KIYOHARA Takashi wrote:
> Hi! all,
> 
> 
> My OpenBlockS (evbppc) panics when boot time.
> 
> ## Transferring control to NetBSD stage-2 loader (at address 00025000) ...
> +40000000,
> 00025000-40000000 -> 00600000-40000000 = 3fa00000
> panic: evcnt_attach_static: evcnt non initialized: group=<cpu> name=<tlbmiss>
> Stopped in pid 0.1 (system) at  0x4d9e8:        lwz     r0, 0x14(r1)

See also PR #44502, for a similar losage uncovered by that diagnostic.
Could you do a similar change for ppc and move the evencounter initializiation
to cpu_startup() ?

Martin

Frank Wille | 6 Feb 21:03 2011
Picon

NAS support ready for upcoming releases

Hi!

Support for 824x based consumer NAS systems is now ready for the 6.0
release!

Besides MD kernel sources and drivers the following areas have been
updated now:

- New bootloader altboot (installed as /usr/mdec/altboot.bin).
  Can load a kernel from local disk, TFTP or NFS.

- Install script uses appropriate settings and devices for all supported
  NAS systems.

- New installation notes for NAS. Evaluation board notes were removed.

- The Sandpoint port home page was updated to reflect our new focus on
  NAS: http://www.NetBSD.org/ports/sandpoint/

- There are already detailed preparation instructions for Synology boxes,
  and how to build a TTL-RS232 level shifter to access the serial console:
  http://www.netbsd.org/ports/sandpoint/instSynology.html
  More models will hopefully follow soon.

Tested by the developers themselves and guaranteed to work are:
- KuroBox classic
- Synology DS-101g+

Many others will most likely work too. So please tell us about your
experience with NetBSD/sandpoint on other NAS boxes. Either here on this
(Continue reading)

Erik Fair | 6 Feb 21:30 2011
Picon

Re: NAS support ready for upcoming releases

Frank,

Two questions:

1. Are there any NAS boxes with ECC RAM subsystems?

2. Are the portmasters committed to maintaining a list of specific manufacturers/models that NetBSD is
known to work, and (further) a list of "probable" systems (because they're based on substantially the
same hardware as the port(s) was developed on)?

I believe we'll need those lists so that the fine work gets traction (gets actually used).

	Erik <fair <at> netbsd.org>

Frank Wille | 6 Feb 22:24 2011
Picon

Re: NAS support ready for upcoming releases

Erik Fair wrote:

> 1. Are there any NAS boxes with ECC RAM subsystems?

Don't know. I didn't check yet. The boxes I have seen seem to use non-ECC
RAM.

But I know the 824x's memory interface supports ECC, and it could be useful
for a file server. So it would be strange when such a configuration doesn't
exist.

> 2. Are the portmasters committed to maintaining a list of specific
> manufacturers/models that NetBSD is known to work, and (further) a list
> of "probable" systems (because they're based on substantially the same
> hardware as the port(s) was developed on)?

A list of already supported (but mostly untested) models is on
  http://www.NetBSD.org/ports/sandpoint/

More 8241/8245-based models are listed in:
  src/sys/arch/sandpoint/README.NAS

--

-- 
Frank Wille

KIYOHARA Takashi | 7 Feb 06:37 2011
Picon

Re: panic ibm4xx machines

Hi! Martin,

From: Martin Husemann <martin <at> duskware.de>
Date: Sun, 6 Feb 2011 19:43:10 +0100

> On Sun, Feb 06, 2011 at 10:39:55PM +0900, KIYOHARA Takashi wrote:
> > Hi! all,
> > 
> > 
> > My OpenBlockS (evbppc) panics when boot time.
> > 
> > ## Transferring control to NetBSD stage-2 loader (at address 00025000) ...
> > +40000000,
> > 00025000-40000000 -> 00600000-40000000 = 3fa00000
> > panic: evcnt_attach_static: evcnt non initialized: group=<cpu> name=<tlbmiss>
> > Stopped in pid 0.1 (system) at  0x4d9e8:        lwz     r0, 0x14(r1)
> 
> See also PR #44502, for a similar losage uncovered by that diagnostic.
> Could you do a similar change for ppc and move the evencounter initializiation
> to cpu_startup() ?

Thanks for your reply.
First of all, I will evade DIAGNOSTIC from my kernel though I hope to
fix anyway.

Thanks,
--
kiyohara

(Continue reading)

Matt Thomas | 7 Feb 07:47 2011

Re: -current userland working for others?


On Feb 6, 2011, at 9:53 AM, Jeff Rizzo wrote:

> Since the PPC changes brought in a couple weeks ago, I have not been able to run a -current userland on my G4/400.
> 
> There was a problem where the kernel would not start, which was fixed by matt <at>  after a few days, and I can
successfully boot a GENERIC from the latest autobuilds.  However, when I unpack userland (base.tgz, in
this case), basically all binaries except those in /rescue stop working.  I get these kinds of messages
from the kernel:
> 
> trap: pid 694.1 (ls): user read DSI trap  <at>  0x407eb46c by 0xeff0f2dc (DSISR 0x40000000, err=14)
> trap: pid 541.1 (dmesg): user ISI trap  <at>  0x2000 (SRR1=0x4000f032)
> trap: pid 167.1 (dmesg): user ISI trap  <at>  0x2000 (SRR1=0x4000f032)
> 
> 
> ...once I back out to userland from 20110111, things start working again.
> 
> Does anyone have any clue what's going on here?  This is the box that I was using to run ATF tests on, and it's
basically been idle for three weeks because of this.

The latest commit to powerpc/include/asm.h fixes this.  I didn't realize
that after my changes that binutils/gcc automatically switched to using
secure-plt by default.  The change to asm.h make .S files use secureplt
as well.

One side effect of this is that PPC programs will no longer have 
executable .data/.bss and startup times will be slightly faster.
Jeff Rizzo | 7 Feb 20:29 2011
Picon

Re: -current userland working for others?

On 2/6/11 10:47 PM, Matt Thomas wrote:
> On Feb 6, 2011, at 9:53 AM, Jeff Rizzo wrote:
>
>> Since the PPC changes brought in a couple weeks ago, I have not been able to run a -current userland on my G4/400.
>>
>> There was a problem where the kernel would not start, which was fixed by matt <at>  after a few days, and I can
successfully boot a GENERIC from the latest autobuilds.  However, when I unpack userland (base.tgz, in
this case), basically all binaries except those in /rescue stop working.  I get these kinds of messages
from the kernel:
>>
>> trap: pid 694.1 (ls): user read DSI trap  <at>  0x407eb46c by 0xeff0f2dc (DSISR 0x40000000, err=14)
>> trap: pid 541.1 (dmesg): user ISI trap  <at>  0x2000 (SRR1=0x4000f032)
>> trap: pid 167.1 (dmesg): user ISI trap  <at>  0x2000 (SRR1=0x4000f032)
>>
>>
>> ...once I back out to userland from 20110111, things start working again.
>>
>> Does anyone have any clue what's going on here?  This is the box that I was using to run ATF tests on, and it's
basically been idle for three weeks because of this.
> The latest commit to powerpc/include/asm.h fixes this.  I didn't realize
> that after my changes that binutils/gcc automatically switched to using
> secure-plt by default.  The change to asm.h make .S files use secureplt
> as well.
>
> One side effect of this is that PPC programs will no longer have
> executable .data/.bss and startup times will be slightly faster.

Unfortunately, it doesn't seem to have fixed my problem.  With a system 
compiled from this morning's sources (9AM PDT/1800 UTC), the problem 
persists.
(Continue reading)

Frank Wille | 7 Feb 21:46 2011
Picon

Re: -current userland working for others?

Jeff Rizzo wrote:

>>> trap: pid 694.1 (ls): user read DSI trap  <at>  0x407eb46c by 0xeff0f2dc
>>> (DSISR 0x40000000, err=14) trap: pid 541.1 (dmesg): user ISI trap  <at> 
>>> 0x2000 (SRR1=0x4000f032) trap: pid 167.1 (dmesg): user ISI trap  <at> 
>>> 0x2000 (SRR1=0x4000f032)
>
>>>
>>> ...once I back out to userland from 20110111, things start working
>>> again.
> [...]
>
> Unfortunately, it doesn't seem to have fixed my problem.  With a system 
> compiled from this morning's sources (9AM PDT/1800 UTC), the problem 
> persists.

As an additional information: I just tried today's kernel, including the
asm.h fix, together with a 5.99.43 userland. I'm seeing a crash during
init:

[...]
root file system type: ffs
trap: pid1.1 (init): kernel PGM trap  <at>  0x57b008 (SRR1=0x89032)
panic: trap

The backtrace:

- at panic+0x25c
- at trap+0x100
- kernel PGM trap by trap+0x280: srr1=0x89032
(Continue reading)

Frank Wille | 9 Feb 18:08 2011
Picon

Re: -current userland working for others?

Frank Wille wrote:

> [...]
> root file system type: ffs
> trap: pid1.1 (init): kernel PGM trap  <at>  0x57b008 (SRR1=0x89032)
> panic: trap
> [...]
> At trap+0x280 I found the Altivec instruction "dssall", although I'm
> running a G3. What happened here?

This is fixed now. Thanks.
I will try a current userland in the next days.

--

-- 
Frank Wille


Gmane