Front page | perl.perl5.porters |
Postings from November 1999
Joerg Schumacher writes:
>
> Hi!
>
> I've had a problem with perl5.005_02 and Term::ReadLine::Gnu dumping
> core on a solaris 5.5.1 box. A little bit of debugging revealed that
> readline-4.0 calls putenv(3) for LINES and COLUMNS during initialization.
> If these vars weren't present in the environment the call to putenv()
> changes the value of char **environ. my_setenv() tests this value to
> decide wether to copy the environment. In my case it didn't copy and
> thus dumped core on Safefree() as result of an assignment to $ENV{PATH}.
>
> Patch appended.
>
> Kind regards,
> Joerg
>
> --
> Gaertner Datensysteme 38114 Braunschweig
> Joerg Schumacher Hamburger Str. 273a
> Tel: 0531-2335555 Fax: 0531-2335556
>
>
>
>
> --- perl5.005_02/util.c.orig Wed Nov 17 00:21:06 1999
> +++ perl5.005_02/util.c Wed Nov 17 00:21:50 1999
> @@ -1413,12 +1413,14 @@
> my_setenv(char *nam, char *val)
> {
> register I32 i=setenv_getix(nam); /* where does it go? */
> + static int copyenv = 1;
>
> - if (environ == PL_origenviron) { /* need we copy environment? */
> + if (copyenv) { /* need we copy environment? */
> I32 j;
> I32 max;
> char **tmpenv;
>
> + copyenv = 0;
> /*SUPPRESS 530*/
> for (max = i; environ[max]; max++) ;
> New(901,tmpenv, max+2, char*);
Without further explanation I hardly think this is a solution. The
real problem is that putenv() of Term::ReadLine::Gnu is mentioning a
wrong malloc() [though I have no idea how this might have happened -
but it does, at least on Solaris 2.7].
Note that malloc.c of 5.005_62 already includes Perl_putenv(). Since
the default now is to have backward compatibility (including
overriding malloc()), we may need to byte a bullet and make Perl by
default override putenv() too (same as it is done for calloc():
#if POLLUTE...
# define Perl_calloc calloc
# define Perl_putenv putenv
#endif
). I think this (and possibly strdup(),
for which Perl_strdup() is already in place) will fix 95% of problems
with mismatched malloc()s.
Ilya