[MLton] Memory problems with latest CVS?

Matthew Fluet fluet@cs.cornell.edu
Wed, 8 Sep 2004 19:38:08 -0400 (EDT)


> > Are you able to reproduce the problem (with cvs.HEAD)?  I'm noticing that
> > some of the sequences that led to problems aren't doing so now.
>
> Yes.  I can reproduce the space leak with no problem.  I have never
> seen the seg fault that you mentioned.

I can reproduce that:

mlton on PATH == mlton-20040819

[fluet@tiger mlton 36]% make clean
[fluet@tiger mlton 36]% make dirs runtime compiler world-no-check
[fluet@tiger mlton 36]% rm mlton/mlton-compile
[fluet@tiger mlton 36]% make script mlbpathmap targetmap constants
[fluet@tiger mlton 36]% cd mlton/
[fluet@tiger mlton 36]% ../build/bin/mlton @MLton ram-slop 0.7 gc-summary gc-messages -- -default-ann 'sequenceUnit true' -target self -verbose 2 -output mlton-compile mlton-stubs.cm
...
         simplify Xml starting
/tmp/fileXWiHCF.24.S 4533: GC_gc
Starting gc.  Request 512 nursery bytes and 0 old gen bytes.
[Requested 1388M cannot be satisfied, backing off by 0M (min size =
1388M).
Major mark-compact GC.
Major mark-compact GC done.
Finished gc.
time: 9,530 ms
old gen size: 467,650,716 bytes (16.8%)
            xmlShrink1 starting
Segmentation fault


It doesn't always seg fault.  Sometimes it will get to this point and the
process with just hang.  Don't know how else to describe it -- the process
doesn't exit, but top shows that it's not using any CPU (but is still
holding onto memory); it just seems to sit there, not even spinning its
wheels.