[Bug java/35035] [4.3 Regression] Has any one managed to run the libjava test suite on powerpc-apple-darwin9?

[andreast at gcc dot gnu dot org]

------- Comment #19 from andreast at gcc dot gnu dot org  2008-02-03 21:22 -------
Dominique,

- what is your ppc machine you're running the build on? UP, SMP, which
processor?
- how do you run the test suite, a single make -k check or a parallel one?

I have a G4 PowerBook (1.5GHz), and I have a blocking issue with the OS itself.
If I test a clean tree I fail with the failures originating this PR. Patching
with Richard's second patch attached to this PR I succeed in building but I
send my machine into a deadlock on the deepest level. I can only get it back if
I have the focus on the terminal where I launched the test and hit ctrl-c. If I
do not have the focus I have to reboot!

The problem I have is in this test case:

Index: testsuite/libjava.jvmti/interp/natgetstacktrace.cc
===================================================================
--- testsuite/libjava.jvmti/interp/natgetstacktrace.cc  (revision 132072)
+++ testsuite/libjava.jvmti/interp/natgetstacktrace.cc  (working copy)
@@ -90,7 +90,7 @@
       done = env->GetBooleanField (obj, done_id);
       if (done == false)
         break;
-      usleep (10);
+      usleep (40);
     }
   while (done != false);
 }

>From my point of view this is an OS issue. I worked around in increasing the
usleep value I pass, but the real cause of this I did not find so far. Well, I
do not have the time to track down OS issues atm.

So maybe you hit similar issue and it manifests on your machine in a different
way.

Just my 2 bucks.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35035