Re: Seg fault with mcrypt?
| From: | Sascha Schumann | Date: | Sat, 26 Jun 1999 00:48:22 +0000 |
| Subject: | Re: Seg fault with mcrypt? | ||
| References: | 1 2 | Groups: | php.dev |
| Request: | Send a blank email to php-dev+get-7570@lists.php.net to get a copy of this message | ||
On Sat, Jun 26, 1999 at 02:21:57AM +0200, Sascha Schumann wrote:
> On Fri, Jun 25, 1999 at 07:46:18PM -0400, Mitch Vincent wrote:
> > Redhat 6.0, Kernel 2.2.9
> >
> > Apache 1.3.6, PHP 3.0.10
> >
> > It seems anytime I urlencode, then mcrypt a string and try to urldecode and
> > mcrypt_decrypt it, most of the time in a different script after passing the
> > encrypted strings over a URL, I get a segfault. I say most of the time
> > because it has happened in many different situations. I thought at frst it
> > was something I was doing, but finally decided to submit a bug report to see
> > if you guys could see anything..
> >
> > If I don't use mcrypt, I never see any problems.
>
> Recompile libmcrypt with --disable-posix-threads.
>
> IIRC there is a problem with the interaction of some key parts.
> I'll have a detailed look at this later.
Hm, I remembered a bad interaction between dmalloc and pthreads
which can be worked around by disabling pthreads in libmcrypt (or
upping a parameter in dmalloc). But you're probably encountering
another problem.
Try to reproduce the segfault in a CGI environment. Compile PHP
as a CGI with --enable-debug --with-mcrypt and run your script in
a loop like this one:
for i in
seq 10000; do ./php your_script.php3 ; done
If this produces a core dump, send me a backtrace. Otherwise,
some other suggestions
- upgrade to the latest libc (RH6 shipped with a not so reliable
one)
- recompile libmcrypt with --disable-posix-threads
- buy me an island near Hawaii and a OC3 :-)
--
Regards,
Sascha Schumann
Consultant