headers
Izumi Tsutsui | 8 Sep 2009 17:31
Picon
Gravatar

5.0 mipsel 2009Q1 packages binary

I wrote:

> My RaQ2 has been working on 2009Q1 pkgsrc binary build since
> 5.0 release date and it's still in progress at ~82% (6548/7962),
> but I've uploaded binaries built till last week:
> 
> ftp://ftp.NetBSD.org/pub/pkgsrc/packages/NetBSD/mipsel/5.0_2009Q1/

Now bulkbuild on RaQ2 is complete and I've put binaries again:
ftp://ftp.NetBSD.org/pub/pkgsrc/packages/NetBSD/mipsel/5.0_2009Q1/

Note editors/emacs (emacs-22.3nb1) in pkgsrc-2009Q1 is broken on mips,
so I've also built patched packages separately for popular demand:
ftp://ftp.NetBSD.org/pub/NetBSD/misc/tsutsui/mipsel/emacs-22.3nb1.tgz
ftp://ftp.NetBSD.org/pub/NetBSD/misc/tsutsui/mipsel/emacs-nox11-22.3nb1.tgz

> To use binaries, set or enable PKG_PATH environment,
> which is commented out in /root/.cshrc
> 
>  #setenv PKG_PATH "ftp://ftp.netbsd.org/pub/pkgsrc/packages/NetBSD/`uname -m`/5.0/All"
> 
> or /root/.profile
> 
>  #export PKG_PATH="ftp://ftp.netbsd.org/pub/pkgsrc/packages/NetBSD/$(uname -m)/5.0/All"
> 
> (note there are appropriate symlinks for version and port strings)
> and just type
> # pkg_add tcsh
> # pkg_add perl
> etc.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Sevan / Venture37 | 18 Sep 2009 20:06
Picon
Gravatar

Problems with my Qube 2

Hi everyone
My Qube 2 seems to crash every week or so with no explanation, the 
system is currently sitting at the db prompt, the output of which 
commands would be useful to give an indication of what makes the machine 
fall over?

trace output:
cpu_Debugger+4 (ff00,f9,c6,2000) ra 800f13b4 sz 0
comintr+4d0 (ff00,f9,c6,2000) ra 802c9734 sz 64
cpu_intr+1d0 (ff00,f9,c6,2000) ra 802b3730 sz 64
mips3_KernIntr+90 (0,fe41f,fe3df,0) ra 802ba1c4 sz 128
cpu_idle+28 (0,fe41f,fe3df,0) ra 801f61d8 sz 24
idle_loop+f8 (0,fe41f,fe3df,0) ra 802b3d40 sz 40
mips3_lwp_trampoline+20 (0,fe41f,fe3df,0) ra 0 sz 24
User-level: pid 0.2

Sevan / Venture37
Permalink | Reply | 9 comments |
headers
Izumi Tsutsui | 18 Sep 2009 20:25
Picon
Gravatar

Re: Problems with my Qube 2

> My Qube 2 seems to crash every week or so with no explanation, the 
> system is currently sitting at the db prompt, the output of which 
> commands would be useful to give an indication of what makes the machine 
> fall over?
> 
> trace output:
> cpu_Debugger+4 (ff00,f9,c6,2000) ra 800f13b4 sz 0
> comintr+4d0 (ff00,f9,c6,2000) ra 802c9734 sz 64

It looks your Qube2 got break signal (which activates ddb(4))
via its serial console. It could happen on some terminal
which might pull its data line during reboot etc.
In that case "continue" command may restart your Qube.

You can avoid it by "sysctl -w ddb.fromconsole=0"
(or put "ddb.fromconsole?=0" in /etc/sysctl.conf) etc.
---
Izumi Tsutsui
Permalink | Reply | 8 comments |
headers
Sevan / Venture37 | 18 Sep 2009 20:36
Picon
Gravatar

Re: Problems with my Qube 2

On 9/18/09 7:25 PM, Izumi Tsutsui wrote:
>> My Qube 2 seems to crash every week or so with no explanation, the
>> system is currently sitting at the db prompt, the output of which
>> commands would be useful to give an indication of what makes the machine
>> fall over?
>>
>> trace output:
>> cpu_Debugger+4 (ff00,f9,c6,2000) ra 800f13b4 sz 0
>> comintr+4d0 (ff00,f9,c6,2000) ra 802c9734 sz 64
>
> It looks your Qube2 got break signal (which activates ddb(4))
> via its serial console. It could happen on some terminal
> which might pull its data line during reboot etc.
> In that case "continue" command may restart your Qube.
>
> You can avoid it by "sysctl -w ddb.fromconsole=0"
> (or put "ddb.fromconsole?=0" in /etc/sysctl.conf) etc.

hmm, that might have been me plugging in the serial lead, I've set 
"ddb.fromconsole?=0" in /etc/sysctl.conf as instructed & "ddb.onpanic?=1"
Will post back with another trace output if/when it hangs again?

Thanks for your help Izumi :)

Sevan
Permalink | Reply | 7 comments |
headers
Andy Ruhl | 18 Sep 2009 21:19
Picon

Re: Problems with my Qube 2

On Fri, Sep 18, 2009 at 11:36 AM, Sevan / Venture37 <venture37 <at> gmail.com> wrote:
> hmm, that might have been me plugging in the serial lead, I've set
> "ddb.fromconsole?=0" in /etc/sysctl.conf as instructed & "ddb.onpanic?=1"
> Will post back with another trace output if/when it hangs again?

Yeap, that was likely what caused the break.

Mine is also hanging up once in a while but I didn't have it set to
drop to the DB prompt. I do now, but I haven't had a hang (crash?)
since I've done that. If I get one, hopefully we can compare notes.

I think it happens on mine during really high load, but probably also
related to disk activity.

Andy
Permalink | Reply | 6 comments |
headers
Sevan / Venture37 | 18 Sep 2009 21:55
Picon
Gravatar

Re: Problems with my Qube 2

On 9/18/09 8:19 PM, Andy Ruhl wrote:
> On Fri, Sep 18, 2009 at 11:36 AM, Sevan / Venture37<venture37 <at> gmail.com>  wrote:
>> hmm, that might have been me plugging in the serial lead, I've set
>> "ddb.fromconsole?=0" in /etc/sysctl.conf as instructed&  "ddb.onpanic?=1"
>> Will post back with another trace output if/when it hangs again?
>
> Yeap, that was likely what caused the break.
>
> Mine is also hanging up once in a while but I didn't have it set to
> drop to the DB prompt. I do now, but I haven't had a hang (crash?)
> since I've done that. If I get one, hopefully we can compare notes.
>
> I think it happens on mine during really high load, but probably also
> related to disk activity.
>
> Andy

I thought it was high load as well, but the box is sitting there idle 
atm as I've moved all the services off it.

Sevan
Permalink | Reply | 5 comments |
headers
Mike Hebel | 1 Oct 2009 14:25

NetBSD 4.0 "opcode not supported on this processor:mips1"...

My primary DNS server drive died sometime between last night and this  
morning so I'm trying to build BIND from scratch on one of my other  
boxes but it's died with the attached error.

I've read some notes that changing things in the CFLAG line might  
resolve this but I'm clueless as to what needs to be changed.  This  
is a RAQ running 4.0 with 65M of RAM.

A little help here for an unemployed flu-riddled geek? ;-)

===========

/var/tmp//cc6JN3Ze.s: Assembler messages:
/var/tmp//cc6JN3Ze.s:24: Error: opcode not supported on this  
processor: mips1 (mips1) `ll $3,0($4)'
/var/tmp//cc6JN3Ze.s:27: Error: opcode not supported on this  
processor: mips1 (mips1) `sc $3,0($4)'
*** Error code 1

Stop.
make: stopped in /root/bind-9.6.1-P1/lib/isc
*** Error code 1

Stop.
make: stopped in /root/bind-9.6.1-P1/lib
*** Error code 1

Stop.
make: stopped in /root/bind-9.6.1-P1
nothing#
(Continue reading)

Permalink | Reply | 11 comments |
headers
..I'd rather be coding ASM! | 1 Oct 2009 14:56
Picon

Re: NetBSD 4.0 "opcode not supported on this processor:mips1"...


Tried changing your CFLAGS to say: -arch=mips2  at worst, or -arch=mips3 ?

Looks like it's got mips1 hardwired in and something for mips2 is comming 
up. Or try CFLAGS="-arch=mips1" and see if it gets that? Sounds fishy.

The r5xxx in the mips cobalt's should be mips3 capable. Perhaps gcc's
only compiled with mips1 numonics?

Have a play. See how you go.

Al.

On Thu, 1 Oct 2009, Mike Hebel wrote:

> My primary DNS server drive died sometime between last night and this morning 
> so I'm trying to build BIND from scratch on one of my other boxes but it's 
> died with the attached error.
>
> I've read some notes that changing things in the CFLAG line might resolve 
> this but I'm clueless as to what needs to be changed.  This is a RAQ running 
> 4.0 with 65M of RAM.
>
> A little help here for an unemployed flu-riddled geek? ;-)
>
> ===========
>
> /var/tmp//cc6JN3Ze.s: Assembler messages:
> /var/tmp//cc6JN3Ze.s:24: Error: opcode not supported on this processor: mips1 
> (mips1) `ll $3,0($4)'
(Continue reading)

Permalink | Reply | 4 comments |
headers
Mike Hebel | 1 Oct 2009 15:02

Re: NetBSD 4.0 "opcode not supported on this processor:mips1"...

I'll try that.  Thanks!

Mike

On Oct 1, 2009, at 7:56 AM 10/1/09, ..I'd rather be coding ASM! wrote:

>
> Tried changing your CFLAGS to say: -arch=mips2  at worst, or - 
> arch=mips3 ?
>
> Looks like it's got mips1 hardwired in and something for mips2 is  
> comming up. Or try CFLAGS="-arch=mips1" and see if it gets that?  
> Sounds fishy.
>
> The r5xxx in the mips cobalt's should be mips3 capable. Perhaps gcc's
> only compiled with mips1 numonics?
>
> Have a play. See how you go.
>
> Al.
>
> On Thu, 1 Oct 2009, Mike Hebel wrote:
>
>> My primary DNS server drive died sometime between last night and  
>> this morning so I'm trying to build BIND from scratch on one of my  
>> other boxes but it's died with the attached error.
>>
>> I've read some notes that changing things in the CFLAG line might  
>> resolve this but I'm clueless as to what needs to be changed.   
>> This is a RAQ running 4.0 with 65M of RAM.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Mike Hebel | 1 Oct 2009 16:54

Re: NetBSD 4.0 "opcode not supported on this processor:mips1"...


On Oct 1, 2009, at 7:56 AM 10/1/09, ..I'd rather be coding ASM! wrote:

>
> Tried changing your CFLAGS to say: -arch=mips2  at worst, or - 
> arch=mips3 ?
>
> Looks like it's got mips1 hardwired in and something for mips2 is  
> comming up. Or try CFLAGS="-arch=mips1" and see if it gets that?  
> Sounds fishy.
>
> The r5xxx in the mips cobalt's should be mips3 capable. Perhaps gcc's
> only compiled with mips1 numonics?
>
> Have a play. See how you go.

No luck.  I set the CFLAGS variable before doing a 'make  
clean;configure;make" with the same results.

nothing# set
CFLAGS  -g -O2 -arch=mips2
argv    ()
cdpath  (/usr/src/sys /usr/src/bin /usr/src/sbin /usr/src/usr.bin / 
usr/src/usr.sbin /usr/src/lib /usr/src/libexec /usr/src/share /usr/ 
src/local /usr/src/games /usr/src/gnu/usr.bin /usr/src/gnu/usr.sbin / 
usr/src/gnu/libexec /usr/src/gnu/lib /usr/src/gnu/games)
cwd     /root/bind-9.6.1-P1
history 1000
home    /root
old     /root
(Continue reading)

Permalink | Reply | 2 comments |

Gmane