TestLeak crashes due to GC issues

Ulrich Weigand Ulrich.Weigand@de.ibm.com
Wed Dec 1 13:59:00 GMT 2004

"Boehm, Hans" <hans.boehm@hp.com> wrote on 11/30/2004 11:34:59 PM:

> I assume this is distinct from the test failures associated with Bryce's
> patch?

Hmm, it looks like it is the same problem after all ...

> It's probably worth stepping through thread creation once, and making
> sure that GC_pthread_create is called as expected.

... because pthread_create is no longer intercepted.

This is directly caused by Bryce's patch, specifically the snippet:
      * include/gc.h: Include config.h, not gc_config.h.

While there still is a boehm-gc/include/config.h being generated
(which contains the proper definitions), when gc.h is included
from libjava files during the libjava build process, the line
#include <config.h>
will resolve to libjava/include/config.h, not boehm-gc/include/config.h.

Thus, when building posix-thread.cc, GC_LINUX_THREADS is not defined,
and therefore pthread_create is not intercepted.

I'll update the tree now that Bryce's patch has been reverted
and verify that the problem is fixed now.


More information about the Java mailing list