develooper Front page | perl.perl5.porters | Postings from December 2004

Re: Smoke [5.9.2] 23670 FAIL(m) openbsd 3.6 (i386/1 cpu)

Thread Previous
Ed Allen Smith
December 23, 2004 14:58
Re: Smoke [5.9.2] 23670 FAIL(m) openbsd 3.6 (i386/1 cpu)
Message ID:
In message <> (on 23
December 2004 17:50:12 -0500), (Ed Allen Smith)
>In message <> (on 23 December 2004
>22:19:57 +0000), (Nicholas Clark) wrote:
>>Well, 23666 was an attempt to stop the failures on platforms where mallocwrap
>>doesn't (yet) work. IIRC this included IRIX. So I was hoping that with this
>>Irix would all pass again.
>It should; am running a check now. But I was actually more thinking of it as
>an easy way to simulate a bug that should be wrappable - as in to try to
>determine, for instance, whether non-n32 (-o32 or -64) IRIX, which doesn't
>show false-positive mallocwrap tests, actually _does_ do _useful_ mallocwrap
>testing or simply isn't affected by it (either would coredump even with
>mallocwrap or wouldn't coredump even without mallocwrap), or whether IRIX is
>vulnerable in -n32 mode to the sort of problem that caused mallocwrap to be
>created in the first place. Admittedly, this test may be obvious to someone
>with more understanding than myself of exactly what mallocwrap _does_ -
>malloc is not an area of C that I am strong on (involves too much pointer
>arithmetic, addressing, et al...).

Oh, fascinating. Would perl 5.8.0 be anticipated to coredump or give
an out-of-memory error or something on the following?

perl -e 'print $#a; $#a>>=1; print $#a;'

What it does on a perl 5.8.0 compiled -n32 is give you '-1-1'. That's it. I
will try to find the time to do some test compiles of more recent perl
versions and see what they do with the above.


Allen Smith             
February 1, 2003                               Space Shuttle Columbia
Ad Astra Per Aspera                     To The Stars Through Asperity

Thread Previous Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About