Regarding Bug 4064 on MacOSX Intel

Regarding Bug 4064 on MacOSX Intel

[Jean-Marie Gaillourdet]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

sorry for posting this to the list but I tried to get an account for  
the bug tracking system, but it didn't work.

This an additional note to Bug 4064.

The same bug, i.e. with the same backtrace in gdb, appears when  
compiling fjavac from http://www.cis.upenn.edu/~stse/javac/ 
fjavac-0.4.3.tgz

That package includes ocfgc a parser generator. For instuctions to  
reproduce that behavior look at the end of the mail.
I think this is easier to reproduce and debug than the original  
instructions because ocfgc does not involve lablgtk or other C  
interfacing code except the unix module of ocaml itself.
This crash does not occur with the bytecode-compiled version, ocfgc.

This is happening with MacOS X 10.4.7 on a MacBookPro and cvs  
checkedout version of today:  
CVSROOT=:pserver:anoncvs@camlcvs.inria.fr:/caml cvs -z3 co -r  
ocaml3092 -P  ocaml
I didn't change anything in the ocaml sources.

Best regards,
   Jean-Marie


$ tar xzvf fjvac-0.4.3.tgz
$ cd fjavac-0.4.3
$ make
<snipped some output>
./ocfgcx java.g j
Generating scanner definitions in file jscan.mll...
Generating syntax definitions in file jsyntax.ml...
Generating parser definitions in file jparse.ml...
make: *** [jsyntax.ml] Bus error
make: *** Deleting file `jsyntax.ml'
$ gdb ./ocfgcx
GNU gdb 6.1-20040303 (Apple version gdb-437) (Fri Jan 13 18:45:48 GMT  
2006)
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and  
you are
welcome to change it and/or distribute copies of it under certain  
conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for  
details.
This GDB was configured as "i386-apple-darwin"...Reading symbols for  
shared libraries .. done

(gdb) r java.g j
Starting program: /Users/jmg/software/src/fjavac-0.4.3/ocfgcx java.g j
Reading symbols for shared libraries . done
Generating scanner definitions in file jscan.mll...
Generating syntax definitions in file jsyntax.ml...
Generating parser definitions in file jparse.ml...

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_PROTECTION_FAILURE at address: 0x00000000
0x0003e693 in caml_oldify_local_roots ()
(gdb) bt
#0  0x0003e693 in caml_oldify_local_roots ()
#1  0x000402f3 in caml_empty_minor_heap ()
#2  0x000403e7 in caml_minor_collection ()
#3  0x0003ee7a in caml_garbage_collection ()
#4  0x0004cea0 in caml_call_gc ()
Previous frame inner to this frame (corrupt stack?)
(gdb)




-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (Darwin)

iD8DBQFEwOEHNIUNP/I5YOgRAhEpAJwMz7LJvpnRJtmjp4/jUJtm8VNJFQCfUD8l
0jZzQbIDMDDSF6lpWMllA9c=
=NqZL
-----END PGP SIGNATURE-----

Re: [Caml-list] Regarding Bug 4064 on MacOSX Intel

[Xavier Leroy]

 > This an additional note to Bug 4064.
 > The same bug, i.e. with the same backtrace in gdb, appears when
 > compiling fjavac from http://www.cis.upenn.edu/~stse/javac/
 > fjavac-0.4.3.tgz

Thanks for the repro case.  It's related to PR#4064 in that both
problems stem from stack alignment constraints under MacOX/Intel,
but the cause of the problem is different.

The bug is now fixed in the 3.09 release branch.  However:

 > CVSROOT=:pserver:anoncvs@camlcvs.inria.fr:/caml cvs -z3 co -r  ocaml3092
 > -P  ocaml

You should use "-r release309" to get the latest sources in the 3.09
release branch.  "-r ocaml3092" will get you the exact sources
released as 3.09.2, which do not include the fixes done since then.

- Xavier Leroy