IcedTea6 on ia64

[Gary Benson]

Gary Benson wrote:
> Andrew John Hughes wrote:
> > We've just had the following build failure from a dgilmore on IRC.
> > This from building the IcedTea6 release on ia64.
> > 
> > http://ia64.koji.fedoraproject.org/koji/getfile?taskID=51511&name=build.log
> > 
> > As it seems to be failing in the second stage of the bootstrap with
> > the just-built zero-based HotSpot, I'm guessing that might be the
> > issue.
> 
> Hmmm, doko had the same failure on Ubuntu a few days ago.  It looks
> like it ran and and compiled a ton of stuff already, so it's actually
> quite promising, but that's me being all glass-half-full ;)
> 
> Since it's in floating point code I'm suspicious it might be the bug
> that icedtea-ia64-bugfix.patch removes the fix for.  I'll ask on one
> of the HotSpot lists.

Ok, could someone try a build with this icedtea-ia64-bugfix.patch
instead of the one that's in the tree?

Cheers,
Gary

-- 
http://gbenson.net/
-------------- next part --------------
diff -r 114e54fc994e openjdk/hotspot/src/share/vm/interpreter/bytecodeInterpreter.cpp
--- openjdk/hotspot/src/share/vm/interpreter/bytecodeInterpreter.cpp	Fri Apr 04 09:33:18 2008 +0100
+++ openjdk/hotspot/src/share/vm/interpreter/bytecodeInterpreter.cpp	Fri Apr 04 09:34:11 2008 +0100
@@ -382,6 +382,13 @@
           CALL_VM_NOCHECK(func);                                  \
           if (THREAD->has_pending_exception()) goto label;        \
         }
+
+/*
+ * IA-64 workaround
+ */
+#ifdef IA64
+double ia64_double_zero = 0.0;
+#endif
 
 /*
  * BytecodeInterpreter::run(interpreterState istate)