Re: [NetBSD 5.99.23] kern_lwp cannot allocate memory

[BERTRAND Joel <joel.bertrand%systella.fr@localhost>]

BERTRAND Joel a écrit :
> Martin Husemann a écrit :
> > On Sun, Jan 24, 2010 at 07:08:37PM +0100, BERTRAND Joel wrote:
> > > tc-end: lwpctl, passed
> >
> > Ok, this is how it should be, no obvious bug here. Something strange is
> > going on - could you give me a quick receipe how to reproduce the problem
> > locally? I'll update my dual SS20 and try to debug it...
>
> It is very strange. My userland was a 5.99 (bug I don't remember exact
> version, I have installed current from tarfile and build it from a
> NetBSD 4.0.1 some time ago...) and this SS20 randomly hanged on thread
> allocation (in perl) and somewhere in fork() (or execve). I have rebuilt
> an recent userland with some difficulties due to these bugs and install
> it. For yesterday, same workstation is building xsane with all
> dependencies and another huge application without any trouble. I suppose
> there was a bug somewhere in my old userland.
>
> For yesterday, this SS20 runs with GENERIC.MP kernel like a charm. I do
> not try to add more processors because I do not see any significatif
> change for my last experience (panic with dual SM71, reset with dual
> RT626).

	Bad news. Same bug occurs. Kernel is unable to do some fork and 
sometimes lwpctl cannot allocate memory. But if I reboot this station, 
after reboot, I cannot reproduce this allocation failure. Here is a top 
(I cannot obtain more information. Kernel doesn't panic, it only refuse 
to fork process...).

load averages:  2.07,  2.02,  2.01;       up 0+22:06:01        17:00:26
48 processes: 2 runnable, 43 sleeping, 2 zombie, 1 on CPU
CPU states: 0.0% user, 0.0% nice, 0.0% system, 0.0% interrupt, 100% idle
Memory:170M Act,10M Inact,5212K Wired, 9612K Exec, 153M File, 157M Free
Swap: 769M Total, 769M Free

  PID USERNAME PRI NICE   SIZE   RES STATE      TIME   WCPU    CPU COMMAND
    0 root  126    0     0K   11M pgdaemon   9:13  0.00%  0.00% [system]
16908 root  85    0  8208K 3744K vm_map     1:49  0.00%  0.00% cvs
  717 root  78    0  8208K 3744K biolock    1:27  0.00%  0.00% cvs
  184 root  85    0  6172K 1776K select     0:43  0.00%  0.00% amd
  498 bertrand  85    0    11M 3068K select     0:29  0.00%  0.00% sshd
  154 root  85    0  3496K 1032K select     0:20  0.00%  0.00% ypbind

Both CVS are in Z state. I shall rebuild a kernel with all debug 
information.

        Regards,

        JKB