Re: SegFault with Bison Grammar

[Philip Herron]

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

Hey

I'm guessing were not going to debug it for you but say, GDB? But some
tips are, I take it your using LEX with this? If so say you have a
rule like identifier something like [a-zA-Z]+ thats bad rule.. but
anyways, make sure you do a string dump of that or your not going to
be able to reference it again, that might be your >5 char problem.

by the way, a better identifier rule would be
ID       [a-zA-Z][a-zA_Z0-9_]*

I think your need to look at how its all going to work a little more,
are you using a %union? That would help as well for your rule returns
of different types or you have to make your lex return only what
YYSTYPE is and your rule returns will be that also. Try to think of
how you want to structure the execution or things before writing the
grammar helps me, because the parser is the link between how you
interpret the rules.

quick tip:
don't do this your going to have a headache :P

malloc( sizeof(struct ext_cmd) );
        if ( $$ == NULL)
            printf("Memory Allocation Error\n");

something like:

void* xmalloc ( size_t size )
{
  register void* mm= malloc( size );
  if ( !mm )
    fatal("Heap size limit reached!\n");

  return mm;
}


Kyle Brandt wrote:
> I'm not quite sure how memory allocation is working with Bison, but
with the
> following Grammar, my code segfaults when my execute_command function is
> called from the redirect rule with an argument more than 5 char in length,
> but it works fine when called from command rule.
Your not going to be able to fault memory corruption or problems on
bison or lex, they are rock solid make sure you know what your doing
with their outcomes though. anyways that my $0.02

- --Phil

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkqzs6YACgkQAhcOgIaQQ2GIhACdHsDtVP4TlCOmm5e3AM8FDXiD
Z24An2MIUGuaF7MLlEiy1PfJYA/s5hPn
=F9Da
-----END PGP SIGNATURE-----