java broken on powerpc64

java broken on powerpc64

[Alan Modra]

This boehm-gc/ change:

2006-01-24  Bryce McKinlay  <mckinlay@redhat.com>

	Import Boehm GC version 6.6.

results in a complete disaster for libjava testsuite results on
powerpc64.  http://gcc.gnu.org/ml/gcc-testresults/2006-02/msg00113.html

I'm committing the following patch under the obvious rule.

	* include/private/gc_locks.h (GC_test_and_set <POWERPC>): Don't
	use broken 64-bit version.

Bootstrapped, regression tested powerpc64-linux.

Index: boehm-gc/include/private/gc_locks.h
===================================================================
--- boehm-gc/include/private/gc_locks.h	(revision 110570)
+++ boehm-gc/include/private/gc_locks.h	(working copy)
@@ -139,7 +139,7 @@
 #      define GC_TEST_AND_SET_DEFINED
 #    endif
 #    if defined(POWERPC)
-#     if CPP_WORDSZ == 64
+#     if 0 /* CPP_WORDSZ == 64  totally broken to use int locks with ldarx */
         inline static int GC_test_and_set(volatile unsigned int *addr) {
           unsigned long oldval;
           unsigned long temp = 1; /* locked value */

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre

Re: java broken on powerpc64

[Alan Modra]

On Sat, Feb 04, 2006 at 10:44:31AM +1030, Alan Modra wrote:
> 	Import Boehm GC version 6.6.
> results in a complete disaster for libjava testsuite results on
> powerpc64.  http://gcc.gnu.org/ml/gcc-testresults/2006-02/msg00113.html

I meant to add that this is known upstream.  See
http://www.hpl.hp.com/hosted/linux/mail-archives/gc/2006-January/001095.html

-- 
Alan Modra
IBM OzLabs - Linux Technology Centre

RE: java broken on powerpc64

[Boehm, Hans]

Thanks.  The code you ifdef'ed out was recently removed from the
upstream version.  I believe there was never a reason for a "64-bit
version".  I should have pointed that out to Bryce.

Hans

> -----Original Message-----
> From: java-patches-owner@gcc.gnu.org 
> [mailto:java-patches-owner@gcc.gnu.org] On Behalf Of Alan Modra
> Sent: Friday, February 03, 2006 4:15 PM
> To: mckinlay@redhat.com; java-patches@gcc.gnu.org; 
> gcc-patches@gcc.gnu.org
> Subject: java broken on powerpc64
> 
> 
> This boehm-gc/ change:
> 
> 2006-01-24  Bryce McKinlay  <mckinlay@redhat.com>
> 
> 	Import Boehm GC version 6.6.
> 
> results in a complete disaster for libjava testsuite results 
> on powerpc64.  
> http://gcc.gnu.org/ml/gcc-testresults/2006-02/msg00113.html
> 
> I'm committing the following patch under the obvious rule.
> 
> 	* include/private/gc_locks.h (GC_test_and_set <POWERPC>): Don't
> 	use broken 64-bit version.
> 
> Bootstrapped, regression tested powerpc64-linux.
> 
> Index: boehm-gc/include/private/gc_locks.h
> ===================================================================
> --- boehm-gc/include/private/gc_locks.h	(revision 110570)
> +++ boehm-gc/include/private/gc_locks.h	(working copy)
> @@ -139,7 +139,7 @@
>  #      define GC_TEST_AND_SET_DEFINED
>  #    endif
>  #    if defined(POWERPC)
> -#     if CPP_WORDSZ == 64
> +#     if 0 /* CPP_WORDSZ == 64  totally broken to use int 
> locks with ldarx */
>          inline static int GC_test_and_set(volatile unsigned 
> int *addr) {
>            unsigned long oldval;
>            unsigned long temp = 1; /* locked value */
> 
> -- 
> Alan Modra
> IBM OzLabs - Linux Technology Centre
>