headers
Blaz Antonic | 2 Nov 2003 09:45
Favicon

Beeping keyboard

Hello,

While experimenting with some stuff few minutes ago i made my LK401 beep
while running NetBSD. Since those beep sequences (and in the end just
one unterminated beep which finally convinced me to turn the box off,
after it wouldn't shut up even when halting NetBSD) are probably not the
normal operating procedure i'm really curious as to why it started in
the first place :-)

Assuming one wanted to use wscons (say on 3100 with monitor attached),
what would be the steps to enable it ? Wscons FAQ is less than useful,
mentioning /dev nodes that don't exist and lines in /etc/tty that don't
exist either. 

Finally, almost unrelated: if one was to grab, say, 1.5 MB of memory for
kernel driver using those neato vax_map_physmem functions for vax and
kernel choked with

panic: vax_map_physmem: kernel map full
Stopped in pid 0 (swapper) at   vax_map_physmem+0x32:   pushl  
r7              

and dropped to debugger prompt, what would be the correct course of
action to take, other than - obviously - not try to allocate so much
memory ? Computer has 56 MB so it shouldn't have problems sparing me
entries for allocation of 1.5 MB of memory space, should it ?

Blaz Antonic
--

-- 
Hi! I'm a signature virus!
(Continue reading)

Permalink | Reply | 1 comment |
headers
ragge | 2 Nov 2003 22:34
Picon
Picon

Re: Beeping keyboard

> 
> Finally, almost unrelated: if one was to grab, say, 1.5 MB of memory for
> kernel driver using those neato vax_map_physmem functions for vax and
> kernel choked with
> 
> panic: vax_map_physmem: kernel map full
> Stopped in pid 0 (swapper) at   vax_map_physmem+0x32:   pushl  
> r7              
> 
The kernel map is sized before the VM system is initialized (because it
needs contiguous physical memory). You must add an entry for it
in arch/vax/vax/pmap.c:calc_kvmsize() to get enough KVA for your needs.

-- Ragge
Permalink | Reply | 0 comments |
headers
ragge | 3 Nov 2003 12:53
Picon
Picon

Re: Sendmail 8.12.10 problems on VAX

> 
> Basically, it runs as a daemon and delivers local mail, but has a problem
> relaying mail. Since the machine is the backup MX for a number of domains,
> it gets a fair bit of email once in a while which ends up sitting in the
> queue. When I try to manually run the queue (sendmail -v -q), I see the
> first of many emails begin to be delivered, but then it segfaults. But I
> don't see a core.
> 
Are you sure that you don't get any core somewhere?  Doesn't sendmail
chroot to some non-obvious place? Is user mail allowed to write a core
file in its directory?  Can you turn on any debug flags?

Just a chance,

-- Ragge
Permalink | Reply | 4 comments |
headers
Johnny Billquist | 3 Nov 2003 22:58
Picon
Picon

Re: Sendmail 8.12.10 problems on VAX

On Mon, 3 Nov 2003 ragge <at> ludd.luth.se wrote:

> > Basically, it runs as a daemon and delivers local mail, but has a problem
> > relaying mail. Since the machine is the backup MX for a number of domains,
> > it gets a fair bit of email once in a while which ends up sitting in the
> > queue. When I try to manually run the queue (sendmail -v -q), I see the
> > first of many emails begin to be delivered, but then it segfaults. But I
> > don't see a core.
> > 
> Are you sure that you don't get any core somewhere?  Doesn't sendmail
> chroot to some non-obvious place? Is user mail allowed to write a core
> file in its directory?  Can you turn on any debug flags?
> 
> Just a chance,

sendmail have not worked on the vax for some time now.
I seem to remember bringing it up this spring, but no solution was
found. I switched to postfix to solve it.
I'd love to get back to sendmail though. postfix is way too slow for my
taste.

I don't think I got a core dump from sendmail when running as a daemon,
but I think I did manage to get a core dump by feeding it a mail to
deliver from the command line.

	Johnny

Johnny Billquist                  || "I'm on a bus
                                  ||  on a psychedelic trip
email: bqt <at> update.uu.se           ||  Reading murder books
(Continue reading)

Permalink | Reply | 3 comments |
headers
J. Buck Caldwell | 4 Nov 2003 05:45

Re: Sendmail 8.12.10 problems on VAX


Johnny Billquist wrote:
> sendmail have not worked on the vax for some time now.
> I seem to remember bringing it up this spring, but no solution was
> found. I switched to postfix to solve it.
> I'd love to get back to sendmail though. postfix is way too slow for my
> taste.

I beg to disagree. I've got pkgsrc-sendmail 8.12.10 (compiled with 
USE_SASL=YES, 1.5.27nb4, compiled just after pkgsrc-sendmail was updated 
to 8.12.10 - dated Sep 24) working on a VS4000/60 running 1.6.1. No 
problems whatsoever. The kernel was last updated to 1_6_release in May - 
with a stock 1.6.1 userland. Works like a charm, no auth (using 
relay_local_from) or cram-md5, with a couple of dnsrbl entries.

Am I alone here?

-Buckaroo
Permalink | Reply | 2 comments |
headers
ragge | 4 Nov 2003 09:41
Picon
Picon

Re: Sendmail 8.12.10 problems on VAX

> 
> I beg to disagree. I've got pkgsrc-sendmail 8.12.10 (compiled with 
> USE_SASL=YES, 1.5.27nb4, compiled just after pkgsrc-sendmail was updated 
> to 8.12.10 - dated Sep 24) working on a VS4000/60 running 1.6.1. No 
> problems whatsoever. The kernel was last updated to 1_6_release in May - 
> with a stock 1.6.1 userland. Works like a charm, no auth (using 
> relay_local_from) or cram-md5, with a couple of dnsrbl entries.
> 
I think that this problem has come up in -current. There are many 
programs that do not work anymore, like pic, gawk etc.

-- Ragge
Permalink | Reply | 1 comment |
headers
John Klos | 4 Nov 2003 10:13

Re: Sendmail 8.12.10 problems on VAX

Hi,

> > I beg to disagree. I've got pkgsrc-sendmail 8.12.10 (compiled with
> > USE_SASL=YES, 1.5.27nb4, compiled just after pkgsrc-sendmail was updated
> > to 8.12.10 - dated Sep 24) working on a VS4000/60 running 1.6.1. No
> > problems whatsoever. The kernel was last updated to 1_6_release in May -
> > with a stock 1.6.1 userland. Works like a charm, no auth (using
> > relay_local_from) or cram-md5, with a couple of dnsrbl entries.
> >
> I think that this problem has come up in -current. There are many
> programs that do not work anymore, like pic, gawk etc.

I'm running release, but what I might be doing differently is running a
separate queue runner - I'm not running suid root sendmail. It's when I
try to run the queue that it has problems. And no, unfortunately, it
doesn't drop core anywhere. I've looked everywhere.

In the meanwhile, I might move back to suid sendmail if that is known to
work.

Thanks,
John Klos
Sixgirls Computing Labs
Permalink | Reply | 0 comments |
headers
Blaz Antonic | 15 Nov 2003 18:12
Favicon

wscons & LCG & NetBSD/VAX

Hello,

I now have what should be working LCG driver (for 4000/60), at least
working as much as LCSPX or SMG one (fixed font setting, no palette
changing). My unfamiliarity with wscons system on *BSD and
i386-centricness of FAQs on this subject prevents me from getting it to
do any useful output :-( 

If anybody ever had 3100's SMG working please let mw know what i have to
do to get it working, other than having driver compiled into kernel and
/dev/ttyEcfg & /dev/ttyEX nodes creeated, using part of script stolen
from what appears to be i386 version of NetBSD ... with some node
numbers different than those of NetBSD/VAX. Now i get Operation not
supported or something liek that when wscons tries to initialize near
the end of boot-up process.

On a related note: apparently -current uses different (newer) functions 
for some things, such as getting device major number and so on and i
have 1.6 source tree. I'll have to update driver slightly to fit into
new source tree but i can't do that until i know i have it working for
sure.

Blaz Antonic
--

-- 
Hi! I'm a signature virus!
Copy me into your signature to help me spread!
Permalink | Reply | 1 comment |
headers
Dave McGuire | 17 Nov 2003 05:54

Re: VS3100 - 1.6

On Saturday, September 27, 2003, at 02:23 AM, Kevin D. Ogden wrote:
> I was tempted to load NetBSD/vax 1.6.1 on my VAXstation but I can't 
> seem to find a MOP image that works for some reason.  I've ruled out a 
> configuration issue on boot server, the 1.5.3 MOP image works fine.  
> Is there one hiding on the FTP server somewhere?

   I've not had any luck with the 1.6 image...the latest I've used 
successfully was 1.5.3, though I must admit I didn't have much time to 
hack on getting the 1.6 image working.

> Another question, how bad is moving to 1.6.1 from 1.5.3 going to 
> affect performance on this crate?  I've only got 16 megs of RAM and 
> it's only about 2.8 VUPS, I think the clock rate is like 12mhz not 
> that that really means squat.

   FYI the VS3100 is clocked at 25MHz...just a nit...

        -Dave

--
Dave McGuire                      "My tummy hurts now, but my soul
St. Petersburg, FL                 feels a little better."     -Ed
Permalink | Reply | 6 comments |
headers
Brian Chase | 17 Nov 2003 06:53

Re: VS3100 - 1.6

On Sun, 16 Nov 2003, Dave McGuire wrote:
> On Saturday, September 27, 2003, at 02:23 AM, Kevin D. Ogden wrote:

> > Another question, how bad is moving to 1.6.1 from 1.5.3 going to
> > affect performance on this crate?  I've only got 16 megs of RAM and
> > it's only about 2.8 VUPS, I think the clock rate is like 12mhz not
> > that that really means squat.
>
>    FYI the VS3100 is clocked at 25MHz...just a nit...

Hmmm...

According to the VARM, the cycle time listed for the VS3100/M30,M40 is
90ns or ~11.11 MHz and the VS3100/M38,M48 is listed at 60ns (16.67 MHz).
Other sources indicated the VS3100/M76 had a cycle time of 28ns (35.71
MHz).

-brian.
Permalink | Reply | 5 comments |

Gmane