Subject: Re: Fixing the swapper (was Re: Summarisation on uvII/VS3100 booting)
To: None <port-vax@netbsd.org>
From: Aaron J. Grier <agrier@poofy.goof.com>
List: port-i386
Date: 06/28/1999 14:31:24
On Sun, Jun 27, 1999 at 12:42:35PM -0700, Brian D Chase wrote:

>   ...
>   pagedaemon: deadlock detected!
>   pagedaemon: deadlock detected!
>   pagedaemon: deadlock detected!
>   pagedaemon: deadlock detected!
>   ...

This is not limited to the VAX...  I've seen it on -i386 running 1.4 as
well.

I was compiling 1.4 pkgsrc/benchmarks/benchfft on my 32MB AMD 5x86,
which ran out of memory while compiling "c_source/b512.c" with -O3
flag. [1] This locked up the machine, although I was still able to get
into the debugger from the console.  I tried to kill the cc1 process,
and when I did a continue, I got the very same pagedaemon error.

[1] I also run out of memory on my pmax, which makes me suspect a
    general egcs optimization problem...  I'd be interested in hearing
	if anybody has successfully built the benchfft package, and how much
	memory it takes to compile.
[2] Needless to say I've since added a global ulimit to /etc/profile...  ;)

----
  Aaron J. Grier  | "Not your ordinary poofy goof." | agrier@poofy.goof.com
        "I hear that if you win, you get a free probe by aliens."
		   --  Bill Coldwell, regarding the SETI@home project