Subject: Re: some general questions
To: None <>
From: Havard Eidnes <>
List: tech-smp
Date: 02/25/2001 16:38:09
> 1)  What is the status on SMP in NetBSD?
> I searched the tech-smp archives, and it looks like the MP kernels are
> working, but everyone is sending dmesg reports.  Is the second CPU just
> recognized, or does the OS actually make use of it?

If we're talking about the i386 port on the MP branch, the OS is
actually making use of the second CPU, as witnessed by the "top"
output from my machine (note the two entries with "CPU/n"):

  215 he       104   20    17M   16M CPU/1    158:10 99.02% 99.02% setiathome
  214 he       104   20    17M   16M RUN/0    158:10 99.02% 99.02% setiathome
  177 root      18  -12   644K 3504K pause/0    0:01  0.00%  0.00% ntpd
  295 he        28    0   156K  836K CPU/0      0:00  0.00%  0.00% top
  227 he        18    0   504K 1136K pause/0    0:00  0.00%  0.00% tcsh
  223 he        18    0   404K  996K pause/0    0:00  0.00%  0.00% tcsh
  203 root      18    0    56K  620K pause/1    0:00  0.00%  0.00% inetd=

> 2) When is it most likely, that we will see SMP for i386, in a
> -release version of NetBSD?

Don't know.  As far as I know there are still a few known issues
which prevents the i386/SMP branch merge to the trunk.

> 3) How stable will it be once it's in -release?  (Ok this
> sounds like a stupid question, maybe I should rephrase to "how
> stable is it right now")

It's been fairly stable for me, save for the single crash I had
today.  I don't pound on it heavily, though; mostly just build
the kernels+distribution + the above.  I don't know what's could
cause it, and I wasn't able to get a core dump. :-(

kernel: page fault trap, code=3D0
Stopped in pid 214 (setiathome) at 0xd3ad3bec: adcb %bh,0x9389d3ad
db{1}:> trace
end(dead4004,d4ad4004) at 0xd3ad3bec
preempt(0,0,c01021fc,100000,1b3b6) at preeempt+0x73
trap() at trap+0x347
--- trap (number 3) ---
db{1}> show reg

I've been running this kernel for, oh, about a month.

> 4)  Is there a shortage of hardware or developers? Or both?

I think the availability of people with the combination of
"available time" and "ability to program at this level" is the
problem.  I'm lacking at least in the last category, and have had
too little in the first one lately.


- H=E5vard