Re: Problems with CMU CL (new version)

[William.Lott@cs.cmu.edu]

Peter.Shell@WIENER.PRODIGY.CS.CMU.EDU writes:
> I am trying to run the new CMU CL on a Sparc machine (ankara.mt.cs).
> It was compiling a file which is about 120k long, and after a few GC's
> in the middle of another GC, it seemed to have wedged for a long time.
> After a while, it stopped with the message:
> sendsig: bad signal stack pid=240, sig=10
> sigsp = 0xf7b52f38, action = 0x9db8, upc = 0x4628
> Illegal instruction (core dumped)
> 
>   (After waiting for some time I did ^z and ^c in the lisp but it
> didn't respond immediatly).
> 
>   The memory limits were:
>         datasize 65536 kbytes
>         stacksize 8192 kbytes
> everything else was unlimited.  I have a feeling that it may have
> something to do with the memory management because the first time I
> tried CMUCL, there was MORE stacksize (I set it to unlimited which
> gave it 32768kbytes), and the lisp wedged during a GC earlier than it
> did the second time.
> 
>   There seems to be plenty of disk space for swap space.
> 
>    The core file is in /../katmandu/usr/pshell/core
> 
>   Let me know if/when you're done using the core file so I can delete it.
> 
> --Pete
> 
> (it's the version of Jun-25, Python 1.0, target SPARC/Mach 0.0, and 
>  the Sparcstation is running Mach)

Does it flame out in the same way every time you try to compile this
file?  If so, can you supply me with the file so that I can try it on
my machine?  I doubt it has anything to do with the resource limits,
because we explicitly vm_allocate everything we need.  Also, I was
unable to look at the core file, as /../katmandu doens't seem to
exist.

-William Lott
CMU Common Lisp Group

ps: Please don't send bug reports to both cmucl-bugs and gripe, as we
will then get two copies of the bug report.