From goetz.lindenmaier at sap.com Tue Dec 4 13:56:19 2012 From: goetz.lindenmaier at sap.com (goetz.lindenmaier at sap.com) Date: Tue, 04 Dec 2012 21:56:19 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: ppc: compiler fixes & enable OSR Message-ID: <20121204215624.6538D47DBF@hg.openjdk.java.net> Changeset: 2bd6a58395e9 Author: Goetz Date: 2012-12-04 22:25 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/2bd6a58395e9 ppc: compiler fixes & enable OSR ! src/cpu/ppc/vm/assembler_ppc.cpp ! src/cpu/ppc/vm/assembler_ppc.hpp ! src/cpu/ppc/vm/c2_globals_ppc.hpp ! src/cpu/ppc/vm/cppInterpreter_ppc.cpp ! src/cpu/ppc/vm/ppc.ad ! src/cpu/ppc/vm/stubGenerator_ppc.cpp From volker.simonis at gmail.com Thu Dec 6 09:08:01 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Thu, 06 Dec 2012 17:08:01 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: Fix IA64 preprocessor conditionals on AIX!!! Message-ID: <20121206170810.5EEED47F3C@hg.openjdk.java.net> Changeset: 047ed76222c9 Author: simonis Date: 2012-12-06 18:06 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/047ed76222c9 Fix IA64 preprocessor conditionals on AIX!!! On AIX unconditionally defines IA64. At least on AIX 7.1 this is a real problem because 'systemcfg.h' is indirectly included by 'pthread.h' and other common system headers and this breaks all preprocessor conditionals which check for IA64. Also removed some old code which pretended to prevent inlining in some places for GCC platforms other than IA64. ! src/os/aix/vm/os_aix.cpp ! src/share/vm/interpreter/bytecodeInterpreter.cpp ! src/share/vm/opto/generateOptoStub.cpp ! src/share/vm/opto/output.cpp ! src/share/vm/prims/forte.cpp ! src/share/vm/runtime/objectMonitor.cpp ! src/share/vm/runtime/os.cpp ! src/share/vm/runtime/sharedRuntime.cpp ! src/share/vm/runtime/synchronizer.cpp ! src/share/vm/utilities/macros.hpp From volker.simonis at gmail.com Thu Dec 6 10:14:27 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Thu, 06 Dec 2012 18:14:27 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: Check consistency of the PPC-specific flag 'UseSIGTRAP' with 'ImplicitRangeChecks' and 'ImplicitNullChecks' Message-ID: <20121206181432.C2D7447F45@hg.openjdk.java.net> Changeset: 34997231d33a Author: simonis Date: 2012-12-06 19:13 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/34997231d33a Check consistency of the PPC-specific flag 'UseSIGTRAP' with 'ImplicitRangeChecks' and 'ImplicitNullChecks' Because 'UseSIGTRAP' is CPU-specific but ImplicitNullChecks is OS-dependent and ImplicitRangeChecks is CPU-specific a well, we do these consistency checks in 'src/cpu/ppc/vm/vm_version_ppc.cpp' and 'src/os/aix/vm/os_aix.cpp' respectively instead of clobbering the shared 'arguments.cpp'. It would be nice though to have a kind of platform dependent 'do_pd_flag_adjustments()' method but unfortunately this is not the case yet. (See http://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/2012-October/thread.html#4594) ! src/cpu/ppc/vm/vm_version_ppc.cpp ! src/os/aix/vm/os_aix.cpp From volker.simonis at gmail.com Fri Dec 7 05:54:31 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Fri, 07 Dec 2012 13:54:31 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: Fix broken AIX build Message-ID: <20121207135440.77D6447FB1@hg.openjdk.java.net> Changeset: 1d9ec232470d Author: simonis Date: 2012-12-07 14:53 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/1d9ec232470d Fix broken AIX build - We still need to copy some of the Power7 defines from systemcfg.h into os_aix.cpp in order to build on older (i.e. 5.3) AIX releases. - Unfortunately, ImplicitNullChecks is a 'development' flag, so we can only adjust it in non-product builds. In the 'product' build we now just issue a warning if the settigns are incorrect. ! src/os/aix/vm/os_aix.cpp From goetz.lindenmaier at sap.com Fri Dec 7 08:20:10 2012 From: goetz.lindenmaier at sap.com (goetz.lindenmaier at sap.com) Date: Fri, 07 Dec 2012 16:20:10 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: 3 new changesets Message-ID: <20121207162019.2A91C47FBA@hg.openjdk.java.net> Changeset: ca3cac6a53bd Author: Goetz Date: 2012-12-07 16:57 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/ca3cac6a53bd Improve handling of memory ordering in taskqueue.hpp. The current implementation does not work on PPC. - Always access _bottom with ordered operations. - Handle subfields of age directly from TaskQueue. ! src/share/vm/utilities/taskqueue.hpp Changeset: 3cd0e8951ecc Author: Goetz Date: 2012-12-07 17:02 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/3cd0e8951ecc opto: issue more memory barriers as neede on PPC. ! src/share/vm/opto/generateOptoStub.cpp ! src/share/vm/opto/library_call.cpp ! src/share/vm/opto/parse3.cpp Changeset: 07fd72d0ca47 Author: Goetz Date: 2012-12-07 17:05 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/07fd72d0ca47 Memory ordering as required for platforms with weak memory ordering (PPC) ! src/share/vm/code/nmethod.cpp ! src/share/vm/gc_implementation/concurrentMarkSweep/compactibleFreeListSpace.cpp ! src/share/vm/gc_implementation/g1/g1SATBCardTableModRefBS.cpp ! src/share/vm/gc_implementation/g1/g1SATBCardTableModRefBS.hpp ! src/share/vm/interpreter/bytecodeInterpreter.cpp ! src/share/vm/memory/barrierSet.hpp ! src/share/vm/memory/barrierSet.inline.hpp ! src/share/vm/memory/cardTableModRefBS.cpp ! src/share/vm/memory/cardTableModRefBS.hpp ! src/share/vm/memory/modRefBarrierSet.hpp ! src/share/vm/oops/cpCacheOop.cpp ! src/share/vm/oops/cpCacheOop.hpp ! src/share/vm/oops/instanceKlass.cpp ! src/share/vm/oops/methodOop.hpp ! src/share/vm/oops/oop.inline.hpp ! src/share/vm/runtime/sweeper.cpp ! src/share/vm/runtime/synchronizer.cpp ! src/share/vm/runtime/thread.hpp From goetz.lindenmaier at sap.com Tue Dec 11 00:40:44 2012 From: goetz.lindenmaier at sap.com (goetz.lindenmaier at sap.com) Date: Tue, 11 Dec 2012 08:40:44 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: opto: switch off OptimizeFill on PPC as it causes errors. Message-ID: <20121211084048.9AA4647044@hg.openjdk.java.net> Changeset: eaff2cfc5f6a Author: Goetz Date: 2012-12-10 17:58 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/eaff2cfc5f6a opto: switch off OptimizeFill on PPC as it causes errors. E.g., jbb2005 can not properly parse XML. ! src/share/vm/opto/c2_globals.hpp From spoole at linux.vnet.ibm.com Tue Dec 11 02:36:43 2012 From: spoole at linux.vnet.ibm.com (Steve Poole) Date: Tue, 11 Dec 2012 10:36:43 +0000 Subject: jtreg testing Message-ID: <6A9A606F-873B-4605-BDC0-09F3BE143828@linux.vnet.ibm.com> just thought I'd share this with you all. This is the result of a reasonable run of the jtteg hotspot regression tests in PPC-AIX JDK7u repos The matrix has machines across the top and JVM's under test for the rows. The machines are , in order, AIX, Linux x86_64 and Linux X86_32. The IBM and Oracle JVM rows are standard JVM's (which explains the grey button or Dreyfus/oracle as the is no standard Oracle JVM for AIX :-) The other JVMs are Hotspot JVM's built from either the main JDK7u repo or the PPC-AIX repo. I'm not ready to share the results yet as I need to fix some more config issues in running jtreg and I need to create a exclusion list for the tests that will never pass when run against an IBM JVM. I also need do similar for the JDK tests I have running. I'll add the LinuxPPC machine to the matrix once I've sorted out the above - the PPC-AIX JVM is a little slow at the moment and it delays the test runs considerably. Still, hope this is generally encouraging! From spoole at linux.vnet.ibm.com Tue Dec 11 03:07:03 2012 From: spoole at linux.vnet.ibm.com (Steve Poole) Date: Tue, 11 Dec 2012 11:07:03 +0000 Subject: jtreg testing Message-ID: <45DE9C6C-7755-44C8-80F8-B5332E3885CD@linux.vnet.ibm.com> just thought I'd share this with you all. This is the result of a reasonable run of the jtteg hotspot regression tests in PPC-AIX JDK7u repos The matrix has machines across the top and JVM's under test for the rows. The machines are , in order, AIX, Linux x86_64 and Linux X86_32. The IBM and Oracle JVM rows are standard JVM's (which explains the grey button or Dreyfus/oracle as the is no standard Oracle JVM for AIX :-) The other JVMs are Hotspot JVM's built from either the main JDK7u repo or the PPC-AIX repo. I'm not ready to share the results yet as I need to fix some more config issues in running jtreg and I need to create a exclusion list for the tests that will never pass when run against an IBM JVM. I also need do similar for the JDK tests I have running. I'll add the LinuxPPC machine to the matrix once I've sorted out the above - the PPC-AIX JVM is a little slow at the moment and it delays the test runs considerably. Still, hope this is generally encouraging! From spoole at linux.vnet.ibm.com Tue Dec 11 03:55:52 2012 From: spoole at linux.vnet.ibm.com (Steve Poole) Date: Tue, 11 Dec 2012 11:55:52 +0000 Subject: jtreg results v 3 Message-ID: <52CE9683-4027-4904-8527-8468B186D742@linux.vnet.ibm.com> As you may have noticed (or not) there was a problem with the previous note that stripped off the picture I want to share. Here's try 3. Picture is here JenkinsScreenGrab.png just thought I'd share this with you all. This is the result of a reasonable run of the jtteg hotspot regression tests in PPC-AIX JDK7u repos The matrix has machines across the top and JVM's under test for the rows. The machines are , in order, AIX, Linux x86_64 and Linux X86_32. The IBM and Oracle JVM rows are standard JVM's (which explains the grey button or Dreyfus/oracle as the is no standard Oracle JVM for AIX :-) The other JVMs are Hotspot JVM's built from either the main JDK7u repo or the PPC-AIX repo. I'm not ready to share the results yet as I need to fix some more config issues in running jtreg and I need to create a exclusion list for the tests that will never pass when run against an IBM JVM. I also need do similar for the JDK tests I have running. I'll add the LinuxPPC machine to the matrix once I've sorted out the above - the PPC-AIX JVM is a little slow at the moment and it delays the test runs considerably. Still, hope this is generally encouraging! From spoole at linux.vnet.ibm.com Tue Dec 11 05:03:52 2012 From: spoole at linux.vnet.ibm.com (Steve Poole) Date: Tue, 11 Dec 2012 13:03:52 +0000 Subject: jtreg results v 3 In-Reply-To: <52CE9683-4027-4904-8527-8468B186D742@linux.vnet.ibm.com> References: <52CE9683-4027-4904-8527-8468B186D742@linux.vnet.ibm.com> Message-ID: <4FD5E678-3D40-48CE-AAC2-F72829D1E006@linux.vnet.ibm.com> This is becoming embarrassing! here's the link in full http://cr.openjdk.java.net/~spoole/pics/JenkinsScreenGrab.png On 11 Dec 2012, at 11:55, Steve Poole wrote: > As you may have noticed (or not) there was a problem with the previous note that stripped off the picture I want to share. > > Here's try 3. > > Picture is here JenkinsScreenGrab.png > > just thought I'd share this with you all. This is the result of a reasonable run of the jtteg hotspot regression tests in PPC-AIX JDK7u repos > > The matrix has machines across the top and JVM's under test for the rows. > > The machines are , in order, AIX, Linux x86_64 and Linux X86_32. The IBM and Oracle JVM rows are standard JVM's (which explains the grey button or Dreyfus/oracle as the is no standard Oracle JVM for AIX :-) The other JVMs are Hotspot JVM's built from either the main JDK7u repo or the PPC-AIX repo. > > > I'm not ready to share the results yet as I need to fix some more config issues in running jtreg and I need to create a exclusion list for the tests that will never pass when run against an IBM JVM. I also need do similar for the JDK tests I have running. > > I'll add the LinuxPPC machine to the matrix once I've sorted out the above - the PPC-AIX JVM is a little slow at the moment and it delays the test runs considerably. Still, hope this is generally encouraging! > > > > From goetz.lindenmaier at sap.com Tue Dec 11 05:24:32 2012 From: goetz.lindenmaier at sap.com (Lindenmaier, Goetz) Date: Tue, 11 Dec 2012 14:24:32 +0100 Subject: jtreg results v 3 In-Reply-To: <52CE9683-4027-4904-8527-8468B186D742@linux.vnet.ibm.com> References: <52CE9683-4027-4904-8527-8468B186D742@linux.vnet.ibm.com> Message-ID: <140FA3E3585AD840A3316D2853F974DC1BF733DB14@DEWDFECCR03.wdf.sap.corp> Hi Steve, very nice! I hope, the 345 bugs come from the configuration issues? Are they all on the aix machine, or also on the x86 machines? Our aix-vm will get faster soon, if the compiler is enabled. This weekend, linuxppc used the compiler in the nightly build, and aix is on it's way. Yesterday I thought jvm2008 would be stable, but today I found another issue ... Cheers, Goetz. -----Original Message----- From: ppc-aix-port-dev-bounces at openjdk.java.net [mailto:ppc-aix-port-dev-bounces at openjdk.java.net] On Behalf Of Steve Poole Sent: Dienstag, 11. Dezember 2012 12:56 To: ppc-aix-port-dev at openjdk.java.net Subject: jtreg results v 3 As you may have noticed (or not) there was a problem with the previous note that stripped off the picture I want to share. Here's try 3. Picture is here JenkinsScreenGrab.png just thought I'd share this with you all. This is the result of a reasonable run of the jtteg hotspot regression tests in PPC-AIX JDK7u repos The matrix has machines across the top and JVM's under test for the rows. The machines are , in order, AIX, Linux x86_64 and Linux X86_32. The IBM and Oracle JVM rows are standard JVM's (which explains the grey button or Dreyfus/oracle as the is no standard Oracle JVM for AIX :-) The other JVMs are Hotspot JVM's built from either the main JDK7u repo or the PPC-AIX repo. I'm not ready to share the results yet as I need to fix some more config issues in running jtreg and I need to create a exclusion list for the tests that will never pass when run against an IBM JVM. I also need do similar for the JDK tests I have running. I'll add the LinuxPPC machine to the matrix once I've sorted out the above - the PPC-AIX JVM is a little slow at the moment and it delays the test runs considerably. Still, hope this is generally encouraging! From spoole at linux.vnet.ibm.com Tue Dec 11 06:23:10 2012 From: spoole at linux.vnet.ibm.com (Steve Poole) Date: Tue, 11 Dec 2012 14:23:10 +0000 Subject: jtreg testing (a test) Message-ID: <0806B02B-864E-49D5-88C9-179BC495385A@linux.vnet.ibm.com> just thought I'd share this with you all. This is the result of a reasonable run of the jtteg hotspot regression tests in PPC-AIX JDK7u repos The matrix has machines across the top and JVM's under test for the rows. The machines are , in order, AIX, Linux x86_64 and Linux X86_32. The IBM and Oracle JVM rows are standard JVM's (which explains the grey button or Dreyfus/oracle as the is no standard Oracle JVM for AIX :-) The other JVMs are Hotspot JVM's built from either the main JDK7u repo or the PPC-AIX repo. I'm not ready to share the results yet as I need to fix some more config issues in running jtreg and I need to create a exclusion list for the tests that will never pass when run against an IBM JVM. I also need do similar for the JDK tests I have running. I'll add the LinuxPPC machine to the matrix once I've sorted out the above - the PPC-AIX JVM is a little slow at the moment and it delays the test runs considerably. Still, hope this is generally encouraging! -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.openjdk.java.net/pipermail/ppc-aix-port-dev/attachments/20121211/9a2ac899/attachment-0001.html -------------- next part -------------- A non-text attachment was scrubbed... Name: jenkinsscreengrab.png Type: image/png Size: 147457 bytes Desc: not available Url : http://mail.openjdk.java.net/pipermail/ppc-aix-port-dev/attachments/20121211/9a2ac899/jenkinsscreengrab-0001.png From volker.simonis at gmail.com Tue Dec 11 10:05:10 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Tue, 11 Dec 2012 18:05:10 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: Implemented os::thread_cpu_time for AIX Message-ID: <20121211180522.9996747070@hg.openjdk.java.net> Changeset: bcf93b000915 Author: simonis Date: 2012-12-11 19:04 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/bcf93b000915 Implemented os::thread_cpu_time for AIX ! src/os/aix/vm/os_aix.cpp From luchsh at linux.vnet.ibm.com Tue Dec 11 22:48:46 2012 From: luchsh at linux.vnet.ibm.com (Jonathan Lu) Date: Wed, 12 Dec 2012 14:48:46 +0800 Subject: Patch for building Java8 JDK repository Message-ID: <50C828CE.3020802@linux.vnet.ibm.com> Hello, I've got a patch to build Java8 JDK repository [1] on AIX 64 and Linux PPC 64, and plan to push it to [1], before that I'd like to get some comments here :) The patch includes two parts, I) Patches ported from PPC Java7u repository [3], which includes various changes from 'corba' and 'jdk' repositories. II) Several small new changes to make the compilation pass. http://cr.openjdk.java.net/~luchsh/ppc-port-bringup-java8/ Since the HotSpot patches have not yet been ported to Java8, so I was using OpenJDK7 builds [2] provided by Volker as bootstrapping JDK and use JDK7 HotSpot as alternative import. So in part II) patch there's a workaround for non-existing VM interface. Many thanks - Jonathan [1] http://hg.openjdk.java.net/ppc-aix-port/jdk8/jdk [2] http://cr.openjdk.java.net/~simonis/ppc-aix-port/bootstrap/ [3] http://hg.openjdk.java.net/ppc-aix-port/jdk7u From goetz.lindenmaier at sap.com Wed Dec 12 07:58:13 2012 From: goetz.lindenmaier at sap.com (goetz.lindenmaier at sap.com) Date: Wed, 12 Dec 2012 15:58:13 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: opto / regalloc: Fix problem with rematerialization Message-ID: <20121212155816.0EF6F470C6@hg.openjdk.java.net> Changeset: 79eae7bb8433 Author: Goetz Date: 2012-12-12 16:56 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/79eae7bb8433 opto / regalloc: Fix problem with rematerialization On PPC, castX2P is rematerialized just where the live range of another rematerialized node using the same def lives. This causes invalid register dependencies (while the code happens to be correct). In the end the VM aborts with assert(!_reg_node[reg_lo] || edge_from_to(_reg_node[reg_lo],def), msg); Seen in jvm2008: -Xcomp -Xbatch -Djava.io.tmpdir=./tmp -jar SPECjvm2008.jar -ikv -wt 30 -it 60 -bt 2 --base xml.validation in sun.awt.X11.XAtom::getAtomListProperty. Fixed by aborting and retrying the register allocation cycle. We had to increase the number of register allocation cycles, as as consequence we got spill split recycle failures. This fix is suboptimal as it increases the runtime of the register allocation (only on PPC), but we saw similar problems in various situations. Therefor excluding castX2P from rematerialization not only worsens register allocation, but is not sufficient, either. ! src/share/vm/opto/chaitin.cpp ! src/share/vm/opto/reg_split.cpp From volker.simonis at gmail.com Wed Dec 12 09:12:29 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Wed, 12 Dec 2012 17:12:29 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: Fixed memory handling/bookkeeping of mmaped memory on AIX Message-ID: <20121212171231.B617B470C9@hg.openjdk.java.net> Changeset: a3bcdfe8674c Author: simonis Date: 2012-12-12 18:11 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/a3bcdfe8674c Fixed memory handling/bookkeeping of mmaped memory on AIX - also improved the logging with -XX:+Verbose by adding missing new-lines - cleaned up make/aix/makefiles/fastdebug.make ! make/aix/makefiles/fastdebug.make ! src/os/aix/vm/os_aix.cpp From volker.simonis at gmail.com Thu Dec 13 00:27:16 2012 From: volker.simonis at gmail.com (Volker Simonis) Date: Thu, 13 Dec 2012 09:27:16 +0100 Subject: Patch for building Java8 JDK repository In-Reply-To: <50C828CE.3020802@linux.vnet.ibm.com> References: <50C828CE.3020802@linux.vnet.ibm.com> Message-ID: Hi Jonathan, great that you started working on JDK8! could you please send the command lines you used to compile on Linux/PPC and AIX. Regards, Volker On Wed, Dec 12, 2012 at 7:48 AM, Jonathan Lu wrote: > Hello, > > I've got a patch to build Java8 JDK repository [1] on AIX 64 and Linux PPC > 64, and plan to push it to [1], before that I'd like to get some comments > here :) > > The patch includes two parts, > I) Patches ported from PPC Java7u repository [3], which includes various > changes from 'corba' and 'jdk' repositories. > II) Several small new changes to make the compilation pass. > http://cr.openjdk.java.net/~luchsh/ppc-port-bringup-java8/ > > Since the HotSpot patches have not yet been ported to Java8, so I was using > OpenJDK7 builds [2] provided by Volker as bootstrapping JDK and use JDK7 > HotSpot as alternative import. So in part II) patch there's a workaround for > non-existing VM interface. > > Many thanks > - Jonathan > > [1] http://hg.openjdk.java.net/ppc-aix-port/jdk8/jdk > [2] http://cr.openjdk.java.net/~simonis/ppc-aix-port/bootstrap/ > [3] http://hg.openjdk.java.net/ppc-aix-port/jdk7u > From luchsh at linux.vnet.ibm.com Thu Dec 13 01:47:34 2012 From: luchsh at linux.vnet.ibm.com (Jonathan Lu) Date: Thu, 13 Dec 2012 17:47:34 +0800 Subject: Patch for building Java8 JDK repository In-Reply-To: References: <50C828CE.3020802@linux.vnet.ibm.com> Message-ID: <50C9A436.2090207@linux.vnet.ibm.com> Hi Volker, Here's the command I was using for compilation, it was taken from my AIX box, but on Linux PPC it should work with some changes to the path strings. make FT_CFLAGS=-m64 LANG=C \ ALT_BOOTDIR=$HOME/depends/ojdk7 \ ALT_HOTSPOT_IMPORT_PATH=$HOME/depends/ojdk7 \ ALT_JDK_IMPORT_PATH=$ALT_BOOTDIR \ ARCH_DATA_MODEL=64 \ HOST_CC=/usr/vacpp/bin/xlc_r \ HOTSPOT_BUILD_JOBS=8 \ PARALLEL_COMPILE_JOBS=8 \ ALT_FREETYPE_LIB_PATH=/opt/freeware/lib \ ALT_FREETYPE_HEADERS_PATH=/opt/freeware/include \ ALT_CUPS_HEADERS_PATH=/opt/freeware/include \ ANT_HOME=$HOME/depends/apache-ant-1.8.2 \ VERBOSE=true \ CC_INTERP=true \ OPENJDK=true \ CORE_BUILD=true \ ALLOW_DOWNLOAD=true \ The is a place holder for "langtools", "corba", "jaxp", "jaxws" and "jdk", (hope HotSpot can be here soon, then this will be unnecessary ). on my machines I wrap them into a short shell script, and call it with module name as argument in a loop. Many thanks and best regards! Jonathan On 12/13/2012 04:27 PM, Volker Simonis wrote: > Hi Jonathan, > > great that you started working on JDK8! > > could you please send the command lines you used to compile on > Linux/PPC and AIX. > > Regards, > Volker > > > On Wed, Dec 12, 2012 at 7:48 AM, Jonathan Lu wrote: >> Hello, >> >> I've got a patch to build Java8 JDK repository [1] on AIX 64 and Linux PPC >> 64, and plan to push it to [1], before that I'd like to get some comments >> here :) >> >> The patch includes two parts, >> I) Patches ported from PPC Java7u repository [3], which includes various >> changes from 'corba' and 'jdk' repositories. >> II) Several small new changes to make the compilation pass. >> http://cr.openjdk.java.net/~luchsh/ppc-port-bringup-java8/ >> >> Since the HotSpot patches have not yet been ported to Java8, so I was using >> OpenJDK7 builds [2] provided by Volker as bootstrapping JDK and use JDK7 >> HotSpot as alternative import. So in part II) patch there's a workaround for >> non-existing VM interface. >> >> Many thanks >> - Jonathan >> >> [1] http://hg.openjdk.java.net/ppc-aix-port/jdk8/jdk >> [2] http://cr.openjdk.java.net/~simonis/ppc-aix-port/bootstrap/ >> [3] http://hg.openjdk.java.net/ppc-aix-port/jdk7u >> From volker.simonis at gmail.com Thu Dec 13 07:21:37 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Thu, 13 Dec 2012 15:21:37 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: Turned off ImplicitNullChecks on AIX by default because it requires UseSIGTRAP which isn't currently implemented on AIX Message-ID: <20121213152150.E08854710C@hg.openjdk.java.net> Changeset: b69b779a26df Author: simonis Date: 2012-12-13 16:21 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/b69b779a26df Turned off ImplicitNullChecks on AIX by default because it requires UseSIGTRAP which isn't currently implemented on AIX ! src/cpu/ppc/vm/globals_ppc.hpp ! src/os/aix/vm/os_aix.cpp From volker.simonis at gmail.com Mon Dec 17 10:11:36 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Mon, 17 Dec 2012 18:11:36 +0000 Subject: hg: ppc-aix-port/jdk7u: Updated README to reflect the current status of the project (mainly the availability of a mixed-mode VM on Linux and AIX) Message-ID: <20121217181137.061F6471E2@hg.openjdk.java.net> Changeset: 50f2b3cacf77 Author: simonis Date: 2012-12-17 19:09 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/rev/50f2b3cacf77 Updated README to reflect the current status of the project (mainly the availability of a mixed-mode VM on Linux and AIX) Also added an automatically generated TOC which should hopefully make the document more readable. ! README-ppc.html From volker.simonis at gmail.com Mon Dec 17 10:26:50 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Mon, 17 Dec 2012 18:26:50 +0000 Subject: hg: ppc-aix-port/jdk7u: 2 new changesets Message-ID: <20121217182650.2EB3D471E3@hg.openjdk.java.net> Changeset: baa81e95e8e6 Author: simonis Date: 2012-12-17 19:25 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/rev/baa81e95e8e6 Added tag ppc-aix-port-b02 for changeset 06179726206f ! .hgtags Changeset: ef45d0d7216c Author: simonis Date: 2012-12-17 19:26 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/rev/ef45d0d7216c Added tag ppc-aix-port-b03 for changeset 50f2b3cacf77 ! .hgtags From volker.simonis at gmail.com Mon Dec 17 10:29:46 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Mon, 17 Dec 2012 18:29:46 +0000 Subject: hg: ppc-aix-port/jdk7u/jdk: 2 new changesets Message-ID: <20121217183027.E1D7D471E4@hg.openjdk.java.net> Changeset: 00221efd3960 Author: simonis Date: 2012-12-17 19:28 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/jdk/rev/00221efd3960 Added tag ppc-aix-port-b02 for changeset 3097457689ba ! .hgtags Changeset: 67d762607263 Author: simonis Date: 2012-12-17 19:29 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/jdk/rev/67d762607263 Added tag ppc-aix-port-b03 for changeset 3097457689ba ! .hgtags From volker.simonis at gmail.com Mon Dec 17 10:31:49 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Mon, 17 Dec 2012 18:31:49 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: 2 new changesets Message-ID: <20121217183201.96A3D471E5@hg.openjdk.java.net> Changeset: 00aaa21b62b1 Author: simonis Date: 2012-12-17 19:31 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/00aaa21b62b1 Added tag ppc-aix-port-b02 for changeset 1efaab66c81d ! .hgtags Changeset: d576ee4480ab Author: simonis Date: 2012-12-17 19:31 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/d576ee4480ab Added tag ppc-aix-port-b03 for changeset b69b779a26df ! .hgtags From volker.simonis at gmail.com Mon Dec 17 10:33:13 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Mon, 17 Dec 2012 18:33:13 +0000 Subject: hg: ppc-aix-port/jdk7u/corba: 2 new changesets Message-ID: <20121217183316.A411B471E7@hg.openjdk.java.net> Changeset: 983a12c99c81 Author: simonis Date: 2012-12-17 19:32 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/corba/rev/983a12c99c81 Added tag ppc-aix-port-b02 for changeset 325250aef90a ! .hgtags Changeset: 61259bc8c8d6 Author: simonis Date: 2012-12-17 19:32 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/corba/rev/61259bc8c8d6 Added tag ppc-aix-port-b03 for changeset 325250aef90a ! .hgtags From volker.simonis at gmail.com Mon Dec 17 10:34:26 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Mon, 17 Dec 2012 18:34:26 +0000 Subject: hg: ppc-aix-port/jdk7u/langtools: 2 new changesets Message-ID: <20121217183433.40242471E8@hg.openjdk.java.net> Changeset: 7b06d998828a Author: simonis Date: 2012-12-17 19:33 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/langtools/rev/7b06d998828a Added tag ppc-aix-port-b02 for changeset e3eeee75b861 ! .hgtags Changeset: cfd681b67038 Author: simonis Date: 2012-12-17 19:34 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/langtools/rev/cfd681b67038 Added tag ppc-aix-port-b03 for changeset e3eeee75b861 ! .hgtags From volker.simonis at gmail.com Mon Dec 17 10:35:31 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Mon, 17 Dec 2012 18:35:31 +0000 Subject: hg: ppc-aix-port/jdk7u/jaxp: 2 new changesets Message-ID: <20121217183536.A72E5471E9@hg.openjdk.java.net> Changeset: e8f597cd4fe2 Author: simonis Date: 2012-12-17 19:35 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/jaxp/rev/e8f597cd4fe2 Added tag ppc-aix-port-b02 for changeset 15b71daf5e69 ! .hgtags Changeset: 8a9a8ee6dd36 Author: simonis Date: 2012-12-17 19:35 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/jaxp/rev/8a9a8ee6dd36 Added tag ppc-aix-port-b03 for changeset 15b71daf5e69 ! .hgtags From volker.simonis at gmail.com Mon Dec 17 10:36:47 2012 From: volker.simonis at gmail.com (volker.simonis at gmail.com) Date: Mon, 17 Dec 2012 18:36:47 +0000 Subject: hg: ppc-aix-port/jdk7u/jaxws: 2 new changesets Message-ID: <20121217183653.A140A471EB@hg.openjdk.java.net> Changeset: 413bd19989c7 Author: simonis Date: 2012-12-17 19:36 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/jaxws/rev/413bd19989c7 Added tag ppc-aix-port-b02 for changeset 4325d1311d55 ! .hgtags Changeset: eed70ca3604d Author: simonis Date: 2012-12-17 19:36 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/jaxws/rev/eed70ca3604d Added tag ppc-aix-port-b03 for changeset 4325d1311d55 ! .hgtags From goetz.lindenmaier at sap.com Wed Dec 19 02:08:44 2012 From: goetz.lindenmaier at sap.com (goetz.lindenmaier at sap.com) Date: Wed, 19 Dec 2012 10:08:44 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: Remove some unnecessary diffs to Oracle coding. Message-ID: <20121219100851.1BE634727B@hg.openjdk.java.net> Changeset: 30f204b78e0d Author: Goetz Date: 2012-12-19 10:58 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/30f204b78e0d Remove some unnecessary diffs to Oracle coding. - Reordering of StarTask was only necessary for xlc8 - formatting in adlc was contributed like this to hsx25 Add some minor fixes to our changes - remove #define _APPLE from taskqueue - fix tracing in bytecodeInterpreter ! make/linux/makefiles/gcc.make ! src/cpu/sparc/vm/compile_sparc.hpp ! src/cpu/x86/vm/compile_x86.hpp ! src/share/vm/adlc/output_c.cpp ! src/share/vm/adlc/output_h.cpp ! src/share/vm/asm/codeBuffer.cpp ! src/share/vm/code/vmreg.hpp ! src/share/vm/interpreter/bytecodeInterpreter.cpp ! src/share/vm/oops/cpCacheOop.hpp ! src/share/vm/opto/graphKit.hpp ! src/share/vm/runtime/deoptimization.cpp ! src/share/vm/runtime/safepoint.cpp ! src/share/vm/runtime/vmStructs.cpp ! src/share/vm/utilities/taskqueue.hpp From spoole at linux.vnet.ibm.com Wed Dec 19 05:33:52 2012 From: spoole at linux.vnet.ibm.com (Steve Poole) Date: Wed, 19 Dec 2012 13:33:52 +0000 Subject: Latest code (b03) building on my linux ppc machine Message-ID: <05469C23-DB7D-4A3A-B10D-ECB070FA9DE0@linux.vnet.ibm.com> Here we go - latest code set compiles on my machine. Now for the tests. -- Build times ---------- Target debug_build Start 2012-12-19 11:51:09 End 2012-12-19 12:59:25 00:10:46 corba 00:04:03 hotspot 00:04:04 jaxp 00:04:41 jaxws 00:40:52 jdk 00:03:50 langtools 01:08:16 TOTAL ------------------------- -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.openjdk.java.net/pipermail/ppc-aix-port-dev/attachments/20121219/03bb2cb7/attachment.html From goetz.lindenmaier at sap.com Thu Dec 20 01:43:39 2012 From: goetz.lindenmaier at sap.com (goetz.lindenmaier at sap.com) Date: Thu, 20 Dec 2012 09:43:39 +0000 Subject: hg: ppc-aix-port/jdk7u/hotspot: 2 new changesets Message-ID: <20121220094348.44FDF472C8@hg.openjdk.java.net> Changeset: 6c726fea31d4 Author: Goetz Date: 2012-12-19 11:39 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/6c726fea31d4 C-interpreter: memory ordering as required for platforms with weak memory ordering (PPC) ! src/share/vm/interpreter/bytecodeInterpreter.cpp Changeset: 2f54febfa437 Author: Goetz Date: 2012-12-20 10:40 +0100 URL: http://hg.openjdk.java.net/ppc-aix-port/jdk7u/hotspot/rev/2f54febfa437 ppc: Implement card mark emitter for concurrentMarkSweep. ! src/cpu/ppc/vm/ppc.ad From zhouyx at linux.vnet.ibm.com Fri Dec 21 01:40:42 2012 From: zhouyx at linux.vnet.ibm.com (Sean Chou) Date: Fri, 21 Dec 2012 17:40:42 +0800 Subject: PPC Linux 64 needs -fsigned-char option for gcc Message-ID: Hello, We found -fsigned-char is added to ppc platform, but not added to ppc64 platform. As they are different platforms, I think it is needed for ppc64 as well. Currently I just added one line modification as follow, but there may be more places to modify. If some one can give some comments, I can make a complete webrev. The buggy scenario we found needs closed code to reproduce, so it is not reproduced with current openjdk build on ppc linux from AIX porting project. I tested with ibmjdk, the patch works. I found CFLAGS_REQUIRED_ppc is from changeset http://hg.openjdk.java.net/jdk8/jdk8/jdk/rev/54d8193f177b . Is it enough to add ppc64 option for places ppc appears in that patch? ///////////////////////// the patch //////////////////////// diff --git a/make/common/Defs-linux.gmk b/make/common/Defs-linux.gmk --- a/make/common/Defs-linux.gmk +++ b/make/common/Defs-linux.gmk @@ -196,6 +196,7 @@ LDFLAGS_COMMON_sparc += -m32 -mcpu=v9 CFLAGS_REQUIRED_arm += -fsigned-char -D_LITTLE_ENDIAN CFLAGS_REQUIRED_ppc += -fsigned-char -D_BIG_ENDIAN +CFLAGS_REQUIRED_ppc64 += -fsigned-char -D_BIG_ENDIAN ifeq ($(ZERO_BUILD), true) CFLAGS_REQUIRED = $(ZERO_ARCHFLAG) ifeq ($(ZERO_ENDIANNESS), little) -- Best Regards, Sean Chou -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.openjdk.java.net/pipermail/ppc-aix-port-dev/attachments/20121221/714b5679/attachment.html From volker.simonis at gmail.com Fri Dec 21 02:48:32 2012 From: volker.simonis at gmail.com (Volker Simonis) Date: Fri, 21 Dec 2012 11:48:32 +0100 Subject: PPC Linux 64 needs -fsigned-char option for gcc In-Reply-To: References: Message-ID: Hi Sean, honestly speaking, I wasn't aware of this problem until now and I just checked that we currently don't use this option, neither internally nor in our port. I found the following nice explanation of the issue: http://www.network-theory.co.uk/docs/gccintro/gccintro_71.html It seems that you only get problems if your programs relies on the fact that 'char' is either unsigned or signed. I suppose that the current OpenJDK doesn't rely on such assumptions (which is good) because we didn't saw any of them until now. If I understand you right, you add some closed code the the JDK which has problems because it makes such assumptions. Is that right? If yes, you should probably first fix that code in the way described in the referenced document. Wouldn't that be possible? Regarding your patch: I suppose you took it against an original JDK and not our port, because in our port we already have the following lines (at least in http://hg.openjdk.java.net/ppc-aix-port/jdk7u//jdk because we haven't started to work on jdk8 until now) CFLAGS_REQUIRED_arm += -fsigned-char -D_LITTLE_ENDIAN CFLAGS_REQUIRED_ppc += -fsigned-char -D_BIG_ENDIAN CFLAGS_REQUIRED_ppc64 += -m64 LDFLAGS_COMMON_ppc64 += -m64 -L/lib64 -Wl,-melf64ppc Notice that we don't set '-D_BIG_ENDIAN' because it is the default. Didn't you observed your problems with jdk7 on Linux/PPC? I think we should patch JDK7 first if this is really necessary. Regards, Volker On Fri, Dec 21, 2012 at 10:40 AM, Sean Chou wrote: > Hello, > > We found -fsigned-char is added to ppc platform, but not added to ppc64 > platform. As they are different platforms, I think it is needed for ppc64 > as well. Currently I just added one line modification as follow, but there > may be more places to modify. If some one can give some comments, I can > make a complete webrev. > > The buggy scenario we found needs closed code to reproduce, so it is not > reproduced with current openjdk build on ppc linux from AIX porting > project. I tested with ibmjdk, the patch works. > > I found CFLAGS_REQUIRED_ppc is from changeset > http://hg.openjdk.java.net/jdk8/jdk8/jdk/rev/54d8193f177b . Is it enough > to add ppc64 option for places ppc appears in that patch? > > ///////////////////////// the patch //////////////////////// > > diff --git a/make/common/Defs-linux.gmk b/make/common/Defs-linux.gmk > --- a/make/common/Defs-linux.gmk > +++ b/make/common/Defs-linux.gmk > @@ -196,6 +196,7 @@ > LDFLAGS_COMMON_sparc += -m32 -mcpu=v9 > CFLAGS_REQUIRED_arm += -fsigned-char -D_LITTLE_ENDIAN > CFLAGS_REQUIRED_ppc += -fsigned-char -D_BIG_ENDIAN > +CFLAGS_REQUIRED_ppc64 += -fsigned-char -D_BIG_ENDIAN > ifeq ($(ZERO_BUILD), true) > CFLAGS_REQUIRED = $(ZERO_ARCHFLAG) > ifeq ($(ZERO_ENDIANNESS), little) > > > -- > > Best Regards, > Sean Chou > -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.openjdk.java.net/pipermail/ppc-aix-port-dev/attachments/20121221/25bfffd8/attachment.html From Alan.Bateman at oracle.com Fri Dec 21 02:18:03 2012 From: Alan.Bateman at oracle.com (Alan Bateman) Date: Fri, 21 Dec 2012 10:18:03 +0000 Subject: PPC Linux 64 needs -fsigned-char option for gcc In-Reply-To: References: Message-ID: <50D4375B.1000207@oracle.com> On 21/12/2012 09:40, Sean Chou wrote: > Hello, > > We found -fsigned-char is added to ppc platform, but not added to ppc64 > platform. As they are different platforms, I think it is needed for ppc64 > as well. Currently I just added one line modification as follow, but there > may be more places to modify. If some one can give some comments, I can > make a complete webrev. > > The buggy scenario we found needs closed code to reproduce, so it is not > reproduced with current openjdk build on ppc linux from AIX porting > project. I tested with ibmjdk, the patch works. > > I found CFLAGS_REQUIRED_ppc is from changeset > http://hg.openjdk.java.net/jdk8/jdk8/jdk/rev/54d8193f177b . Is it enough > to add ppc64 option for places ppc appears in that patch? > > ///////////////////////// the patch //////////////////////// > > diff --git a/make/common/Defs-linux.gmk b/make/common/Defs-linux.gmk > --- a/make/common/Defs-linux.gmk > +++ b/make/common/Defs-linux.gmk > @@ -196,6 +196,7 @@ > LDFLAGS_COMMON_sparc += -m32 -mcpu=v9 > CFLAGS_REQUIRED_arm += -fsigned-char -D_LITTLE_ENDIAN > CFLAGS_REQUIRED_ppc += -fsigned-char -D_BIG_ENDIAN > +CFLAGS_REQUIRED_ppc64 += -fsigned-char -D_BIG_ENDIAN > ifeq ($(ZERO_BUILD), true) > CFLAGS_REQUIRED = $(ZERO_ARCHFLAG) > ifeq ($(ZERO_ENDIANNESS), little) > You should probably bring this to build-dev to discuss. If the linux-ppc64 port is a jdk8 port then you should probably be looking at the new build and we should be cutting over soon. -Alan From david.holmes at oracle.com Fri Dec 21 05:13:42 2012 From: david.holmes at oracle.com (David Holmes) Date: Fri, 21 Dec 2012 23:13:42 +1000 Subject: PPC Linux 64 needs -fsigned-char option for gcc In-Reply-To: References: Message-ID: <50D46086.90503@oracle.com> One thing that would be nice to come out of this porting effort is better organisation of platform specific aspects. I would prefer to see an arch.make file for each arch rather than have all these platform specific sections. (It was fine when you only had a couple of flavours but now the balance has shifted). The JDK ppc/arm references were simply the minimum we needed for our embedded builds to work. They don't really offer proper arm/ppc support. For hotspot we rely on setting ARCH and overriding EXTRA_CFLAGS via the environment or make invocation, so these architectures are not treated as first-class citizens by hotspot. This particular example is particularly bad because the arm/ppc values are completely different to the kinds of values set for the other archs! I don't even think we continue to use these in the new build. David ---- On 21/12/2012 7:40 PM, Sean Chou wrote: > Hello, > > We found -fsigned-char is added to ppc platform, but not added to ppc64 > platform. As they are different platforms, I think it is needed for > ppc64 as well. Currently I just added one line modification as follow, > but there may be more places to modify. If some one can give some > comments, I can make a complete webrev. > > The buggy scenario we found needs closed code to reproduce, so it is not > reproduced with current openjdk build on ppc linux from AIX porting > project. I tested with ibmjdk, the patch works. > > I found CFLAGS_REQUIRED_ppc is from changeset > http://hg.openjdk.java.net/jdk8/jdk8/jdk/rev/54d8193f177b . Is it > enough to add ppc64 option for places ppc appears in that patch? > > ///////////////////////// the patch //////////////////////// > > diff --git a/make/common/Defs-linux.gmk b/make/common/Defs-linux.gmk > --- a/make/common/Defs-linux.gmk > +++ b/make/common/Defs-linux.gmk > @@ -196,6 +196,7 @@ > LDFLAGS_COMMON_sparc += -m32 -mcpu=v9 > CFLAGS_REQUIRED_arm += -fsigned-char -D_LITTLE_ENDIAN > CFLAGS_REQUIRED_ppc += -fsigned-char -D_BIG_ENDIAN > +CFLAGS_REQUIRED_ppc64 += -fsigned-char -D_BIG_ENDIAN > ifeq ($(ZERO_BUILD), true) > CFLAGS_REQUIRED = $(ZERO_ARCHFLAG) > ifeq ($(ZERO_ENDIANNESS), little) > > > -- > > Best Regards, > Sean Chou From zhouyx at linux.vnet.ibm.com Sun Dec 23 19:29:57 2012 From: zhouyx at linux.vnet.ibm.com (Sean Chou) Date: Mon, 24 Dec 2012 11:29:57 +0800 Subject: PPC Linux 64 needs -fsigned-char option for gcc In-Reply-To: References: Message-ID: Yes, we see the problem with jdk7 on linux-ppc64. And we have a patch which addes "signed" to the char field obviously. But I think it is better to have this option rather than just modify that place. On Fri, Dec 21, 2012 at 6:48 PM, Volker Simonis wrote: > Hi Sean, > > honestly speaking, I wasn't aware of this problem until now and I just > checked that we currently don't use this option, neither internally nor in > our port. > I found the following nice explanation of the issue: > http://www.network-theory.co.uk/docs/gccintro/gccintro_71.html > > It seems that you only get problems if your programs relies on the fact > that 'char' is either unsigned or signed. I suppose that the current > OpenJDK doesn't rely on such assumptions (which is good) because we didn't > saw any of them until now. > > If I understand you right, you add some closed code the the JDK which has > problems because it makes such assumptions. Is that right? If yes, you > should probably first fix that code in the way described in the referenced > document. Wouldn't that be possible? > > Regarding your patch: I suppose you took it against an original JDK and > not our port, because in our port we already have the following lines (at > least in http://hg.openjdk.java.net/ppc-aix-port/jdk7u//jdk because we > haven't started to work on jdk8 until now) > > CFLAGS_REQUIRED_arm += -fsigned-char -D_LITTLE_ENDIAN > CFLAGS_REQUIRED_ppc += -fsigned-char -D_BIG_ENDIAN > CFLAGS_REQUIRED_ppc64 += -m64 > LDFLAGS_COMMON_ppc64 += -m64 -L/lib64 -Wl,-melf64ppc > > Notice that we don't set '-D_BIG_ENDIAN' because it is the default. > > Didn't you observed your problems with jdk7 on Linux/PPC? I think we > should patch JDK7 first if this is really necessary. > > Regards, > Volker > > > > On Fri, Dec 21, 2012 at 10:40 AM, Sean Chou wrote: > >> Hello, >> >> We found -fsigned-char is added to ppc platform, but not added to ppc64 >> platform. As they are different platforms, I think it is needed for ppc64 >> as well. Currently I just added one line modification as follow, but there >> may be more places to modify. If some one can give some comments, I can >> make a complete webrev. >> >> The buggy scenario we found needs closed code to reproduce, so it is not >> reproduced with current openjdk build on ppc linux from AIX porting >> project. I tested with ibmjdk, the patch works. >> >> I found CFLAGS_REQUIRED_ppc is from changeset >> http://hg.openjdk.java.net/jdk8/jdk8/jdk/rev/54d8193f177b . Is it >> enough to add ppc64 option for places ppc appears in that patch? >> >> ///////////////////////// the patch //////////////////////// >> >> diff --git a/make/common/Defs-linux.gmk b/make/common/Defs-linux.gmk >> --- a/make/common/Defs-linux.gmk >> +++ b/make/common/Defs-linux.gmk >> @@ -196,6 +196,7 @@ >> LDFLAGS_COMMON_sparc += -m32 -mcpu=v9 >> CFLAGS_REQUIRED_arm += -fsigned-char -D_LITTLE_ENDIAN >> CFLAGS_REQUIRED_ppc += -fsigned-char -D_BIG_ENDIAN >> +CFLAGS_REQUIRED_ppc64 += -fsigned-char -D_BIG_ENDIAN >> ifeq ($(ZERO_BUILD), true) >> CFLAGS_REQUIRED = $(ZERO_ARCHFLAG) >> ifeq ($(ZERO_ENDIANNESS), little) >> >> >> -- >> >> Best Regards, >> Sean Chou >> > > -- Best Regards, Sean Chou -------------- next part -------------- An HTML attachment was scrubbed... URL: http://mail.openjdk.java.net/pipermail/ppc-aix-port-dev/attachments/20121224/dfeebff6/attachment.html