[ ghc-Bugs-425965 ] Segfaults with different Heapsize

[noreply@sourceforge.net]

Bugs item #425965, was updated on 2001-05-21 09:33
You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=108032&aid=425965&group_id=8032

Category: Runtime System
Group: 5.00
>Status: Closed
>Resolution: Fixed
Priority: 5
Submitted By: Nobody/Anonymous (nobody)
Assigned to: Nobody/Anonymous (nobody)
Summary: Segfaults with different Heapsize

Initial Comment:
Hi!
My not-so-small programm crashes for some input files.
According to the documentation I tried it with
different heap sizes. On larger heaps it starts not to
segfault anymore. It uses a lot of FFI (especially
functions get dynamically exported) and therefore I
already checked for memory violations. Additionaly I
fired up a debugging ghc and used +RTS -D128 -RTS, but
it kept segfaulting. Sadly I don't know how to shorten
the whole program anymore to pinpoint the problem.


System(s) 
gcc-2.95.2
linux i386,i686,solaris 2.8

A more detailed README is included in the attached file

----------------------------------------------------------------------

Comment By: Axel Krauth (axxl)
Date: 2001-05-30 01:01

Message:
Logged In: YES 
user_id=50935

It looks like the bug has disappeared.
Thanks!

----------------------------------------------------------------------

Comment By: Axel Krauth (axxl)
Date: 2001-05-22 02:20

Message:
Logged In: YES 
user_id=50935

I'm using the source distribution available at the ghc-page.
I try to get the current CVS compiled and see what happens.

----------------------------------------------------------------------

Comment By: Julian Seward (sewardj)
Date: 2001-05-22 01:21

Message:
Logged In: YES 
user_id=50143

Thanks for the bug report and program to reproduce it.  
With -O,
I cannot reproduce the bug:

cam-02-unx:~/Junk/leela2$ ./leela2 Cone.gml +RTS -H1M -RTS 
Everything worked as is should
cam-02-unx:~/Junk/leela2$ ./leela2 Cone.gml +RTS -H2M -RTS 
Everything worked as is should
cam-02-unx:~/Junk/leela2$ ./leela2 Cone.gml +RTS -H500k -
RTS 
Everything worked as is should
cam-02-unx:~/Junk/leela2$ ./leela2 Cone.gml +RTS -G1 -RTS 
Everything worked as is should
cam-02-unx:~/Junk/leela2$ ./leela2 Cone.gml +RTS -G1 -H2m -
RTS 
Everything worked as is should
cam-02-unx:~/Junk/leela2$ ./leela2 Cone.gml +RTS -G3 -H2m -
RTS 
Everything worked as is should
cam-02-unx:~/Junk/leela2$ 

This is using the very latest of the stable branch.  
Exactly where
did you get the GHC you are using for this?  It is possible 
that the bug has already been fixed -- a fix for a 
potentially crashing bug was put in on Friday.

J



----------------------------------------------------------------------

You can respond by visiting: 
http://sourceforge.net/tracker/?func=detail&atid=108032&aid=425965&group_id=8032