Merge

.hgtags-top-repo

@@ -351,3 +351,4 @@ be58b02c11f90b88c67e4d0e2cb5e4cf2d9b3c57 jdk-9+105
 54575d8783b3a39a2d710c28cda675d44261f9d9 jdk-9+106
 4d65eba233a8730f913734a6804910b842d2cb54 jdk-9+107
 c7be2a78c31b3b6132f2f5e9e4b3d3bb1c20245c jdk-9+108
+1787bdaabb2b6f4193406e25a50cb0419ea8e8f3 jdk-9+109

common/autoconf/generated-configure.sh

@@ -4005,7 +4005,7 @@ apt_help() {
     devkit)
       PKGHANDLER_COMMAND="sudo apt-get install build-essential" ;;
     openjdk)
-      PKGHANDLER_COMMAND="sudo apt-get install openjdk-7-jdk" ;;
+      PKGHANDLER_COMMAND="sudo apt-get install openjdk-8-jdk" ;;
     alsa)
       PKGHANDLER_COMMAND="sudo apt-get install libasound2-dev" ;;
     cups)
@@ -4026,7 +4026,7 @@ yum_help() {
     devkit)
       PKGHANDLER_COMMAND="sudo yum groupinstall \"Development Tools\"" ;;
     openjdk)
-      PKGHANDLER_COMMAND="sudo yum install java-1.7.0-openjdk" ;;
+      PKGHANDLER_COMMAND="sudo yum install java-1.8.0-openjdk-devel" ;;
     alsa)
       PKGHANDLER_COMMAND="sudo yum install alsa-lib-devel" ;;
     cups)
@@ -4862,7 +4862,7 @@ VS_SDK_PLATFORM_NAME_2013=
 #CUSTOM_AUTOCONF_INCLUDE
 
 # Do not change or remove the following line, it is needed for consistency checks:
-DATE_WHEN_GENERATED=1456136781
+DATE_WHEN_GENERATED=1457684806
 
 ###############################################################################
 #
@@ -15118,6 +15118,10 @@ $as_echo "$COMPILE_TYPE" >&6; }
     REQUIRED_OS_NAME=Darwin
     REQUIRED_OS_VERSION=11.2
   fi
+  if test "x$OPENJDK_TARGET_OS" = "xaix"; then
+    REQUIRED_OS_NAME=AIX
+    REQUIRED_OS_VERSION=7.1
+  fi
 
 
 

common/autoconf/help.m4

@@ -106,7 +106,7 @@ apt_help() {
     devkit)
       PKGHANDLER_COMMAND="sudo apt-get install build-essential" ;;
     openjdk)
-      PKGHANDLER_COMMAND="sudo apt-get install openjdk-7-jdk" ;;
+      PKGHANDLER_COMMAND="sudo apt-get install openjdk-8-jdk" ;;
     alsa)
       PKGHANDLER_COMMAND="sudo apt-get install libasound2-dev" ;;
     cups)
@@ -127,7 +127,7 @@ yum_help() {
     devkit)
       PKGHANDLER_COMMAND="sudo yum groupinstall \"Development Tools\"" ;;
     openjdk)
-      PKGHANDLER_COMMAND="sudo yum install java-1.7.0-openjdk" ;;
+      PKGHANDLER_COMMAND="sudo yum install java-1.8.0-openjdk-devel" ;;
     alsa)
       PKGHANDLER_COMMAND="sudo yum install alsa-lib-devel" ;;
     cups)

common/autoconf/platform.m4

@@ -406,6 +406,10 @@ AC_DEFUN([PLATFORM_SET_RELEASE_FILE_OS_VALUES],
     REQUIRED_OS_NAME=Darwin
     REQUIRED_OS_VERSION=11.2
   fi
+  if test "x$OPENJDK_TARGET_OS" = "xaix"; then
+    REQUIRED_OS_NAME=AIX
+    REQUIRED_OS_VERSION=7.1
+  fi
 
   AC_SUBST(REQUIRED_OS_NAME)
   AC_SUBST(REQUIRED_OS_VERSION)

common/bin/compare_exceptions.sh.incl

@@ -185,7 +185,6 @@ if [ "$OPENJDK_TARGET_OS" = "solaris" ] && [ "$OPENJDK_TARGET_CPU" = "x86_64" ];
       ./lib/amd64/libjava.so
       ./lib/amd64/libjawt.so
       ./lib/amd64/libjdwp.so
-      ./lib/amd64/libjfr.so
       ./lib/amd64/libjpeg.so
       ./lib/amd64/libjsdt.so
       ./lib/amd64/libjsound.so
@@ -321,7 +320,6 @@ if [ "$OPENJDK_TARGET_OS" = "solaris" ] && [ "$OPENJDK_TARGET_CPU" = "sparcv9" ]
       ./lib/sparcv9/libjava.so
       ./lib/sparcv9/libjawt.so
       ./lib/sparcv9/libjdwp.so
-      ./lib/sparcv9/libjfr.so
       ./lib/sparcv9/libjpeg.so
       ./lib/sparcv9/libjsdt.so
       ./lib/sparcv9/libjsound.so

common/bin/unshuffle_list.txt

@@ -1293,12 +1293,8 @@ jdk/src/jdk.crypto.pkcs11/windows/native/libj2pkcs11/j2secmod_md.c : jdk/src/win
 jdk/src/jdk.crypto.pkcs11/windows/native/libj2pkcs11/j2secmod_md.h : jdk/src/windows/native/sun/security/pkcs11/j2secmod_md.h
 jdk/src/jdk.crypto.pkcs11/windows/native/libj2pkcs11/p11_md.c : jdk/src/windows/native/sun/security/pkcs11/wrapper/p11_md.c
 jdk/src/jdk.crypto.pkcs11/windows/native/libj2pkcs11/p11_md.h : jdk/src/windows/native/sun/security/pkcs11/wrapper/p11_md.h
-jdk/src/jdk.deploy.osx/macosx/classes/com/apple/concurrent/package.html : jdk/src/macosx/classes/com/apple/concurrent/package.html
+jdk/src/java.desktop/macosx/native/libosx/CFileManager.m : jdk/src/macosx/native/com/apple/eio/CFileManager.m
-jdk/src/jdk.deploy.osx/macosx/classes/com/apple/concurrent : jdk/src/macosx/classes/com/apple/concurrent
+jdk/src/java.base/macosx/native/libosxsecurity/KeystoreImpl.m : jdk/src/macosx/native/apple/security/KeystoreImpl.m
-jdk/src/jdk.deploy.osx/macosx/native/libosx/CFileManager.m : jdk/src/macosx/native/com/apple/eio/CFileManager.m
-jdk/src/jdk.deploy.osx/macosx/native/libosx/Dispatch.m : jdk/src/macosx/native/com/apple/concurrent/Dispatch.m
-jdk/src/jdk.deploy.osx/macosx/native/libosx/JavaAppLauncher.m : jdk/src/macosx/native/apple/launcher/JavaAppLauncher.m
-jdk/src/jdk.deploy.osx/macosx/native/libosx/KeystoreImpl.m : jdk/src/macosx/native/apple/security/KeystoreImpl.m
 jdk/src/jdk.hprof.agent/share/classes/com/sun/demo/jvmti/hprof : jdk/src/share/classes/com/sun/demo/jvmti/hprof
 jdk/src/jdk.httpserver/share/classes/com/sun/net/httpserver : jdk/src/share/classes/com/sun/net/httpserver
 jdk/src/jdk.httpserver/share/classes/sun/net/httpserver : jdk/src/share/classes/sun/net/httpserver

hotspot/.hgtags

@@ -511,3 +511,4 @@ c5f55130b1b69510d9a6f4a3105b58e21cd7ffe1 jdk-9+103
 7232de4c17c37f60aecec4f3191090bd3d41d334 jdk-9+106
 c5146d4da417f76edfc43097d2e2ced042a65b4e jdk-9+107
 934f6793f5f7dca44f69b4559d525fa64b31840d jdk-9+108
+7e7e50ac4faf19899fc811569e32cfa478759ebb jdk-9+109

hotspot/make/aix/makefiles/trace.make

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2003, 2015, Oracle and/or its affiliates. All rights reserved.
+# Copyright (c) 2003, 2016, Oracle and/or its affiliates. All rights reserved.
 # DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 #
 # This code is free software; you can redistribute it and/or modify it
@@ -57,11 +57,6 @@ ifeq ($(HAS_ALT_SRC), true)
 TraceGeneratedNames +=  \
 	traceRequestables.hpp \
     traceEventControl.hpp
-
-ifneq ($(INCLUDE_TRACE), false)
-TraceGeneratedNames += traceProducer.cpp
-endif
-
 endif
 
 TraceGeneratedFiles = $(TraceGeneratedNames:%=$(TraceOutDir)/%)
@@ -100,9 +95,6 @@ else
 $(TraceOutDir)/traceEventClasses.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceEventClasses.xsl $(XML_DEPS)
 	$(GENERATE_CODE)
 
-$(TraceOutDir)/traceProducer.cpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceProducer.xsl $(XML_DEPS)
-	$(GENERATE_CODE)
-
 $(TraceOutDir)/traceRequestables.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceRequestables.xsl $(XML_DEPS)
 	$(GENERATE_CODE)
 

hotspot/make/bsd/makefiles/trace.make

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2003, 2015, Oracle and/or its affiliates. All rights reserved.
+# Copyright (c) 2003, 2016, Oracle and/or its affiliates. All rights reserved.
 # DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 #
 # This code is free software; you can redistribute it and/or modify it
@@ -57,11 +57,6 @@ ifeq ($(HAS_ALT_SRC), true)
 TraceGeneratedNames +=  \
 	traceRequestables.hpp \
     traceEventControl.hpp
-
-ifneq ($(INCLUDE_TRACE), false)
-TraceGeneratedNames += traceProducer.cpp
-endif
-
 endif
 
 
@@ -101,9 +96,6 @@ else
 $(TraceOutDir)/traceEventClasses.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceEventClasses.xsl $(XML_DEPS)
 	$(GENERATE_CODE)
 
-$(TraceOutDir)/traceProducer.cpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceProducer.xsl $(XML_DEPS)
-	$(GENERATE_CODE)
-
 $(TraceOutDir)/traceRequestables.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceRequestables.xsl $(XML_DEPS)
 	$(GENERATE_CODE)
 

hotspot/make/linux/makefiles/trace.make

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2003, 2015, Oracle and/or its affiliates. All rights reserved.
+# Copyright (c) 2003, 2016, Oracle and/or its affiliates. All rights reserved.
 # DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 #
 # This code is free software; you can redistribute it and/or modify it
@@ -57,11 +57,6 @@ ifeq ($(HAS_ALT_SRC), true)
 TraceGeneratedNames +=  \
 	traceRequestables.hpp \
     traceEventControl.hpp
-
-ifneq ($(INCLUDE_TRACE), false)
-TraceGeneratedNames += traceProducer.cpp
-endif
-
 endif
 
 TraceGeneratedFiles = $(TraceGeneratedNames:%=$(TraceOutDir)/%)
@@ -100,9 +95,6 @@ else
 $(TraceOutDir)/traceEventClasses.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceEventClasses.xsl $(XML_DEPS)
 	$(GENERATE_CODE)
 
-$(TraceOutDir)/traceProducer.cpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceProducer.xsl $(XML_DEPS)
-	$(GENERATE_CODE)
-
 $(TraceOutDir)/traceRequestables.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceRequestables.xsl $(XML_DEPS)
 	$(GENERATE_CODE)
 

hotspot/make/solaris/makefiles/amd64.make

@@ -39,7 +39,7 @@ OPT_CFLAGS/c1_LinearScan.o = -xO2
 # of OPT_CFLAGS. Restore it here.
 ifeq ($(ENABLE_FULL_DEBUG_SYMBOLS),1)
   OPT_CFLAGS/generateOptoStub.o += -g0 -xs
-  OPT_CFLAGS/LinearScan.o += -g0 -xs
+  OPT_CFLAGS/c1_LinearScan.o += -g0 -xs
 endif
 
 else

hotspot/make/solaris/makefiles/trace.make

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2003, 2015, Oracle and/or its affiliates. All rights reserved.
+# Copyright (c) 2003, 2016, Oracle and/or its affiliates. All rights reserved.
 # DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 #
 # This code is free software; you can redistribute it and/or modify it
@@ -56,8 +56,7 @@ TraceGeneratedNames =     \
 ifeq ($(HAS_ALT_SRC), true)
 TraceGeneratedNames +=  \
 	traceRequestables.hpp \
-    traceEventControl.hpp \
+    traceEventControl.hpp
-    traceProducer.cpp
 endif
 
 TraceGeneratedFiles = $(TraceGeneratedNames:%=$(TraceOutDir)/%)
@@ -96,9 +95,6 @@ else
 $(TraceOutDir)/traceEventClasses.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceEventClasses.xsl $(XML_DEPS)
 	$(GENERATE_CODE)
 
-$(TraceOutDir)/traceProducer.cpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceProducer.xsl $(XML_DEPS)
-	$(GENERATE_CODE)
-
 $(TraceOutDir)/traceRequestables.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceRequestables.xsl $(XML_DEPS)
 	$(GENERATE_CODE)
 

hotspot/make/test/JtregNative.gmk

@@ -48,6 +48,7 @@ BUILD_HOTSPOT_JTREG_NATIVE_SRC := \
     $(HOTSPOT_TOPDIR)/test/runtime/SameObject \
     $(HOTSPOT_TOPDIR)/test/compiler/floatingpoint/ \
     $(HOTSPOT_TOPDIR)/test/compiler/calls \
+    $(HOTSPOT_TOPDIR)/test/compiler/native \
     #
 
 # Add conditional directories here when needed.

hotspot/make/windows/makefiles/trace.make

@@ -1,5 +1,5 @@
 #
-# Copyright (c) 2003, 2013, Oracle and/or its affiliates. All rights reserved.
+# Copyright (c) 2003, 2016, Oracle and/or its affiliates. All rights reserved.
 # DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 #
 # This code is free software; you can redistribute it and/or modify it
@@ -43,8 +43,7 @@ TraceGeneratedNames =     \
 !if EXISTS($(TraceAltSrcDir))
 TraceGeneratedNames = $(TraceGeneratedNames) \
     traceRequestables.hpp \
-    traceEventControl.hpp \
+    traceEventControl.hpp
-    traceProducer.cpp
 !endif
 
 
@@ -58,8 +57,7 @@ TraceGeneratedFiles = \
 !if EXISTS($(TraceAltSrcDir))
 TraceGeneratedFiles = $(TraceGeneratedFiles) \
 	$(TraceOutDir)/traceRequestables.hpp \
-    $(TraceOutDir)/traceEventControl.hpp \
+    $(TraceOutDir)/traceEventControl.hpp
-	$(TraceOutDir)/traceProducer.cpp
 !endif
 
 XSLT = $(QUIETLY) $(REMOTE) $(RUN_JAVA) -classpath $(JvmtiOutDir) jvmtiGen
@@ -98,10 +96,6 @@ $(TraceOutDir)/traceEventClasses.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)
 	@echo Generating AltSrc $@
 	@$(XSLT) -IN $(TraceSrcDir)/trace.xml -XSL $(TraceAltSrcDir)/traceEventClasses.xsl -OUT $(TraceOutDir)/traceEventClasses.hpp
 
-$(TraceOutDir)/traceProducer.cpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceProducer.xsl $(XML_DEPS)
-	@echo Generating AltSrc $@
-	@$(XSLT) -IN $(TraceSrcDir)/trace.xml -XSL $(TraceAltSrcDir)/traceProducer.xsl -OUT $(TraceOutDir)/traceProducer.cpp
-
 $(TraceOutDir)/traceRequestables.hpp: $(TraceSrcDir)/trace.xml $(TraceAltSrcDir)/traceRequestables.xsl $(XML_DEPS)
 	@echo Generating AltSrc $@
 	@$(XSLT) -IN $(TraceSrcDir)/trace.xml -XSL $(TraceAltSrcDir)/traceRequestables.xsl -OUT $(TraceOutDir)/traceRequestables.hpp

hotspot/src/cpu/aarch64/vm/jvmciCodeInstaller_aarch64.cpp

@@ -74,7 +74,7 @@ void CodeInstaller::pd_patch_MetaspaceConstant(int pc_offset, Handle constant, T
 void CodeInstaller::pd_patch_DataSectionReference(int pc_offset, int data_offset, TRAPS) {
   address pc = _instructions->start() + pc_offset;
   NativeInstruction* inst = nativeInstruction_at(pc);
-  if (inst->is_adr_aligned()) {
+  if (inst->is_adr_aligned() || inst->is_ldr_literal()) {
     address dest = _constants->start() + data_offset;
     _instructions->relocate(pc, section_word_Relocation::spec((address) dest, CodeBuffer::SECT_CONSTS));
     TRACE_jvmci_3("relocating at " PTR_FORMAT " (+%d) with destination at %d", p2i(pc), pc_offset, data_offset);

hotspot/src/cpu/aarch64/vm/macroAssembler_aarch64.cpp

@@ -4481,225 +4481,126 @@ void MacroAssembler::string_compare(Register str1, Register str2,
   BLOCK_COMMENT("} string_compare");
 }
 
+// Compare Strings or char/byte arrays.
 
-void MacroAssembler::string_equals(Register str1, Register str2,
+// is_string is true iff this is a string comparison.
-                                   Register cnt, Register result,
-                                   Register tmp1) {
-  Label SAME_CHARS, DONE, SHORT_LOOP, SHORT_STRING,
-    NEXT_WORD;
 
-  const Register tmp2 = rscratch1;
+// For Strings we're passed the address of the first characters in a1
-  assert_different_registers(str1, str2, cnt, result, tmp1, tmp2, rscratch2);
+// and a2 and the length in cnt1.
 
-  BLOCK_COMMENT("string_equals {");
+// For byte and char arrays we're passed the arrays themselves and we
+// have to extract length fields and do null checks here.
 
-  // Start by assuming that the strings are not equal.
+// elem_size is the element size in bytes: either 1 or 2.
-  mov(result, zr);
 
-  // A very short string
+// There are two implementations.  For arrays >= 8 bytes, all
-  cmpw(cnt, 4);
+// comparisons (including the final one, which may overlap) are
-  br(Assembler::LT, SHORT_STRING);
+// performed 8 bytes at a time.  For arrays < 8 bytes, we compare a
+// halfword, then a short, and then a byte.
 
-  // Check if the strings start at the same location.
+void MacroAssembler::arrays_equals(Register a1, Register a2,
-  cmp(str1, str2);
+                                   Register result, Register cnt1,
-  br(Assembler::EQ, SAME_CHARS);
+                                   int elem_size, bool is_string)
+{
+  Label SAME, DONE, SHORT, NEXT_WORD, ONE;
+  Register tmp1 = rscratch1;
+  Register tmp2 = rscratch2;
+  Register cnt2 = tmp2;  // cnt2 only used in array length compare
+  int elem_per_word = wordSize/elem_size;
+  int log_elem_size = exact_log2(elem_size);
+  int length_offset = arrayOopDesc::length_offset_in_bytes();
+  int base_offset
+    = arrayOopDesc::base_offset_in_bytes(elem_size == 2 ? T_CHAR : T_BYTE);
 
-  // Compare longwords
+  assert(elem_size == 1 || elem_size == 2, "must be char or byte");
-  {
+  assert_different_registers(a1, a2, result, cnt1, rscratch1, rscratch2);
-    subw(cnt, cnt, 4); // The last longword is a special case
 
-    // Move both string pointers to the last longword of their
+  BLOCK_COMMENT(is_string ? "string_equals {" : "array_equals {");
-    // strings, negate the remaining count, and convert it to bytes.
-    lea(str1, Address(str1, cnt, Address::uxtw(1)));
-    lea(str2, Address(str2, cnt, Address::uxtw(1)));
-    sub(cnt, zr, cnt, LSL, 1);
 
-    // Loop, loading longwords and comparing them into rscratch2.
+  mov(result, false);
-    bind(NEXT_WORD);
-    ldr(tmp1, Address(str1, cnt));
-    ldr(tmp2, Address(str2, cnt));
-    adds(cnt, cnt, wordSize);
-    eor(rscratch2, tmp1, tmp2);
-    cbnz(rscratch2, DONE);
-    br(Assembler::LT, NEXT_WORD);
 
-    // Last longword.  In the case where length == 4 we compare the
+  if (!is_string) {
-    // same longword twice, but that's still faster than another
+    // if (a==a2)
-    // conditional branch.
+    //     return true;
+    eor(rscratch1, a1, a2);
+    cbz(rscratch1, SAME);
+    // if (a==null || a2==null)
+    //     return false;
+    cbz(a1, DONE);
+    cbz(a2, DONE);
+    // if (a1.length != a2.length)
+    //      return false;
+    ldrw(cnt1, Address(a1, length_offset));
+    ldrw(cnt2, Address(a2, length_offset));
+    eorw(tmp1, cnt1, cnt2);
+    cbnzw(tmp1, DONE);
 
-    ldr(tmp1, Address(str1));
+    lea(a1, Address(a1, base_offset));
-    ldr(tmp2, Address(str2));
+    lea(a2, Address(a2, base_offset));
-    eor(rscratch2, tmp1, tmp2);
-    cbz(rscratch2, SAME_CHARS);
-    b(DONE);
   }
 
-  bind(SHORT_STRING);
+  // Check for short strings, i.e. smaller than wordSize.
-  // Is the length zero?
+  subs(cnt1, cnt1, elem_per_word);
-  cbz(cnt, SAME_CHARS);
+  br(Assembler::LT, SHORT);
-
+  // Main 8 byte comparison loop.
-  bind(SHORT_LOOP);
+  bind(NEXT_WORD); {
-  load_unsigned_short(tmp1, Address(post(str1, 2)));
+    ldr(tmp1, Address(post(a1, wordSize)));
-  load_unsigned_short(tmp2, Address(post(str2, 2)));
+    ldr(tmp2, Address(post(a2, wordSize)));
-  subw(tmp1, tmp1, tmp2);
+    subs(cnt1, cnt1, elem_per_word);
+    eor(tmp1, tmp1, tmp2);
+    cbnz(tmp1, DONE);
+  } br(GT, NEXT_WORD);
+  // Last longword.  In the case where length == 4 we compare the
+  // same longword twice, but that's still faster than another
+  // conditional branch.
+  // cnt1 could be 0, -1, -2, -3, -4 for chars; -4 only happens when
+  // length == 4.
+  if (log_elem_size > 0)
+    lsl(cnt1, cnt1, log_elem_size);
+  ldr(tmp1, Address(a1, cnt1));
+  ldr(tmp2, Address(a2, cnt1));
+  eor(tmp1, tmp1, tmp2);
   cbnz(tmp1, DONE);
-  sub(cnt, cnt, 1);
+  b(SAME);
-  cbnz(cnt, SHORT_LOOP);
 
-  // Strings are equal.
+  bind(SHORT);
-  bind(SAME_CHARS);
+  Label TAIL03, TAIL01;
+
+  tbz(cnt1, 2 - log_elem_size, TAIL03); // 0-7 bytes left.
+  {
+    ldrw(tmp1, Address(post(a1, 4)));
+    ldrw(tmp2, Address(post(a2, 4)));
+    eorw(tmp1, tmp1, tmp2);
+    cbnzw(tmp1, DONE);
+  }
+  bind(TAIL03);
+  tbz(cnt1, 1 - log_elem_size, TAIL01); // 0-3 bytes left.
+  {
+    ldrh(tmp1, Address(post(a1, 2)));
+    ldrh(tmp2, Address(post(a2, 2)));
+    eorw(tmp1, tmp1, tmp2);
+    cbnzw(tmp1, DONE);
+  }
+  bind(TAIL01);
+  if (elem_size == 1) { // Only needed when comparing byte arrays.
+    tbz(cnt1, 0, SAME); // 0-1 bytes left.
+    {
+      ldrb(tmp1, a1);
+      ldrb(tmp2, a2);
+      eorw(tmp1, tmp1, tmp2);
+      cbnzw(tmp1, DONE);
+    }
+  }
+  // Arrays are equal.
+  bind(SAME);
   mov(result, true);
 
-  // That's it
+  // That's it.
   bind(DONE);
-
+  BLOCK_COMMENT(is_string ? "} string_equals" : "} array_equals");
-  BLOCK_COMMENT("} string_equals");
 }
 
 
-void MacroAssembler::byte_arrays_equals(Register ary1, Register ary2,
-                                        Register result, Register tmp1)
-{
-  Register cnt1 = rscratch1;
-  Register cnt2 = rscratch2;
-  Register tmp2 = rscratch2;
-
-  Label SAME, DIFFER, NEXT, TAIL07, TAIL03, TAIL01;
-
-  int length_offset  = arrayOopDesc::length_offset_in_bytes();
-  int base_offset    = arrayOopDesc::base_offset_in_bytes(T_BYTE);
-
-  BLOCK_COMMENT("byte_arrays_equals  {");
-
-    // different until proven equal
-    mov(result, false);
-
-    // same array?
-    cmp(ary1, ary2);
-    br(Assembler::EQ, SAME);
-
-    // ne if either null
-    cbz(ary1, DIFFER);
-    cbz(ary2, DIFFER);
-
-    // lengths ne?
-    ldrw(cnt1, Address(ary1, length_offset));
-    ldrw(cnt2, Address(ary2, length_offset));
-    cmp(cnt1, cnt2);
-    br(Assembler::NE, DIFFER);
-
-    lea(ary1, Address(ary1, base_offset));
-    lea(ary2, Address(ary2, base_offset));
-
-    subs(cnt1, cnt1, 8);
-    br(LT, TAIL07);
-
-  BIND(NEXT);
-    ldr(tmp1, Address(post(ary1, 8)));
-    ldr(tmp2, Address(post(ary2, 8)));
-    subs(cnt1, cnt1, 8);
-    eor(tmp1, tmp1, tmp2);
-    cbnz(tmp1, DIFFER);
-    br(GE, NEXT);
-
-  BIND(TAIL07);  // 0-7 bytes left, cnt1 = #bytes left - 4
-    tst(cnt1, 0b100);
-    br(EQ, TAIL03);
-    ldrw(tmp1, Address(post(ary1, 4)));
-    ldrw(tmp2, Address(post(ary2, 4)));
-    cmp(tmp1, tmp2);
-    br(NE, DIFFER);
-
-  BIND(TAIL03);  // 0-3 bytes left, cnt1 = #bytes left - 4
-    tst(cnt1, 0b10);
-    br(EQ, TAIL01);
-    ldrh(tmp1, Address(post(ary1, 2)));
-    ldrh(tmp2, Address(post(ary2, 2)));
-    cmp(tmp1, tmp2);
-    br(NE, DIFFER);
-  BIND(TAIL01);  // 0-1 byte left
-    tst(cnt1, 0b01);
-    br(EQ, SAME);
-    ldrb(tmp1, ary1);
-    ldrb(tmp2, ary2);
-    cmp(tmp1, tmp2);
-    br(NE, DIFFER);
-
-  BIND(SAME);
-    mov(result, true);
-  BIND(DIFFER); // result already set
-
-  BLOCK_COMMENT("} byte_arrays_equals");
-}
-
-// Compare char[] arrays aligned to 4 bytes
-void MacroAssembler::char_arrays_equals(Register ary1, Register ary2,
-                                        Register result, Register tmp1)
-{
-  Register cnt1 = rscratch1;
-  Register cnt2 = rscratch2;
-  Register tmp2 = rscratch2;
-
-  Label SAME, DIFFER, NEXT, TAIL03, TAIL01;
-
-  int length_offset  = arrayOopDesc::length_offset_in_bytes();
-  int base_offset    = arrayOopDesc::base_offset_in_bytes(T_CHAR);
-
-  BLOCK_COMMENT("char_arrays_equals  {");
-
-    // different until proven equal
-    mov(result, false);
-
-    // same array?
-    cmp(ary1, ary2);
-    br(Assembler::EQ, SAME);
-
-    // ne if either null
-    cbz(ary1, DIFFER);
-    cbz(ary2, DIFFER);
-
-    // lengths ne?
-    ldrw(cnt1, Address(ary1, length_offset));
-    ldrw(cnt2, Address(ary2, length_offset));
-    cmp(cnt1, cnt2);
-    br(Assembler::NE, DIFFER);
-
-    lea(ary1, Address(ary1, base_offset));
-    lea(ary2, Address(ary2, base_offset));
-
-    subs(cnt1, cnt1, 4);
-    br(LT, TAIL03);
-
-  BIND(NEXT);
-    ldr(tmp1, Address(post(ary1, 8)));
-    ldr(tmp2, Address(post(ary2, 8)));
-    subs(cnt1, cnt1, 4);
-    eor(tmp1, tmp1, tmp2);
-    cbnz(tmp1, DIFFER);
-    br(GE, NEXT);
-
-  BIND(TAIL03);  // 0-3 chars left, cnt1 = #chars left - 4
-    tst(cnt1, 0b10);
-    br(EQ, TAIL01);
-    ldrw(tmp1, Address(post(ary1, 4)));
-    ldrw(tmp2, Address(post(ary2, 4)));
-    cmp(tmp1, tmp2);
-    br(NE, DIFFER);
-  BIND(TAIL01);  // 0-1 chars left
-    tst(cnt1, 0b01);
-    br(EQ, SAME);
-    ldrh(tmp1, ary1);
-    ldrh(tmp2, ary2);
-    cmp(tmp1, tmp2);
-    br(NE, DIFFER);
-
-  BIND(SAME);
-    mov(result, true);
-  BIND(DIFFER); // result already set
-
-  BLOCK_COMMENT("} char_arrays_equals");
-}
-
 // encode char[] to byte[] in ISO_8859_1
 void MacroAssembler::encode_iso_array(Register src, Register dst,
                       Register len, Register result,

hotspot/src/cpu/aarch64/vm/macroAssembler_aarch64.hpp

@@ -1186,13 +1186,11 @@ public:
   void string_compare(Register str1, Register str2,
                       Register cnt1, Register cnt2, Register result,
                       Register tmp1);
-  void string_equals(Register str1, Register str2,
+
-                     Register cnt, Register result,
+  void arrays_equals(Register a1, Register a2,
-                     Register tmp1);
+                     Register result, Register cnt1,
-  void char_arrays_equals(Register ary1, Register ary2,
+                     int elem_size, bool is_string);
-                          Register result, Register tmp1);
+
-  void byte_arrays_equals(Register ary1, Register ary2,
-                          Register result, Register tmp1);
   void encode_iso_array(Register src, Register dst,
                         Register len, Register result,
                         FloatRegister Vtmp1, FloatRegister Vtmp2,

hotspot/src/cpu/aarch64/vm/nativeInst_aarch64.hpp

@@ -105,13 +105,20 @@ class NativeInstruction VALUE_OBJ_CLASS_SPEC {
   inline friend NativeInstruction* nativeInstruction_at(address address);
 
   static bool is_adrp_at(address instr);
+
   static bool is_ldr_literal_at(address instr);
+
+  bool is_ldr_literal() {
+    return is_ldr_literal_at(addr_at(0));
+  }
+
   static bool is_ldrw_to_zr(address instr);
 
   static bool is_call_at(address instr) {
     const uint32_t insn = (*(uint32_t*)instr);
     return (insn >> 26) == 0b100101;
   }
+
   bool is_call() {
     return is_call_at(addr_at(0));
   }

hotspot/src/cpu/aarch64/vm/stubGenerator_aarch64.cpp

@@ -163,30 +163,20 @@ class StubGenerator: public StubCodeGenerator {
     sp_after_call_off = -26,
 
     d15_off            = -26,
-    d14_off            = -25,
     d13_off            = -24,
-    d12_off            = -23,
     d11_off            = -22,
-    d10_off            = -21,
     d9_off             = -20,
-    d8_off             = -19,
 
     r28_off            = -18,
-    r27_off            = -17,
     r26_off            = -16,
-    r25_off            = -15,
     r24_off            = -14,
-    r23_off            = -13,
     r22_off            = -12,
-    r21_off            = -11,
     r20_off            = -10,
-    r19_off            =  -9,
     call_wrapper_off   =  -8,
     result_off         =  -7,
     result_type_off    =  -6,
     method_off         =  -5,
     entry_point_off    =  -4,
-    parameters_off     =  -3,
     parameter_size_off =  -2,
     thread_off         =  -1,
     fp_f               =   0,
@@ -208,30 +198,20 @@ class StubGenerator: public StubCodeGenerator {
     const Address result_type   (rfp, result_type_off    * wordSize);
     const Address method        (rfp, method_off         * wordSize);
     const Address entry_point   (rfp, entry_point_off    * wordSize);
-    const Address parameters    (rfp, parameters_off     * wordSize);
     const Address parameter_size(rfp, parameter_size_off * wordSize);
 
     const Address thread        (rfp, thread_off         * wordSize);
 
     const Address d15_save      (rfp, d15_off * wordSize);
-    const Address d14_save      (rfp, d14_off * wordSize);
     const Address d13_save      (rfp, d13_off * wordSize);
-    const Address d12_save      (rfp, d12_off * wordSize);
     const Address d11_save      (rfp, d11_off * wordSize);
-    const Address d10_save      (rfp, d10_off * wordSize);
     const Address d9_save       (rfp, d9_off * wordSize);
-    const Address d8_save       (rfp, d8_off * wordSize);
 
     const Address r28_save      (rfp, r28_off * wordSize);
-    const Address r27_save      (rfp, r27_off * wordSize);
     const Address r26_save      (rfp, r26_off * wordSize);
-    const Address r25_save      (rfp, r25_off * wordSize);
     const Address r24_save      (rfp, r24_off * wordSize);
-    const Address r23_save      (rfp, r23_off * wordSize);
     const Address r22_save      (rfp, r22_off * wordSize);
-    const Address r21_save      (rfp, r21_off * wordSize);
     const Address r20_save      (rfp, r20_off * wordSize);
-    const Address r19_save      (rfp, r19_off * wordSize);
 
     // stub code
 
@@ -254,31 +234,20 @@ class StubGenerator: public StubCodeGenerator {
     // rthread because we want to sanity check rthread later
     __ str(c_rarg7,  thread);
     __ strw(c_rarg6, parameter_size);
-    __ str(c_rarg5,  parameters);
+    __ stp(c_rarg4, c_rarg5,  entry_point);
-    __ str(c_rarg4,  entry_point);
+    __ stp(c_rarg2, c_rarg3,  result_type);
-    __ str(c_rarg3,  method);
+    __ stp(c_rarg0, c_rarg1,  call_wrapper);
-    __ str(c_rarg2,  result_type);
-    __ str(c_rarg1,  result);
-    __ str(c_rarg0,  call_wrapper);
-    __ str(r19,      r19_save);
-    __ str(r20,      r20_save);
-    __ str(r21,      r21_save);
-    __ str(r22,      r22_save);
-    __ str(r23,      r23_save);
-    __ str(r24,      r24_save);
-    __ str(r25,      r25_save);
-    __ str(r26,      r26_save);
-    __ str(r27,      r27_save);
-    __ str(r28,      r28_save);
 
-    __ strd(v8,      d8_save);
+    __ stp(r20, r19,   r20_save);
-    __ strd(v9,      d9_save);
+    __ stp(r22, r21,   r22_save);
-    __ strd(v10,     d10_save);
+    __ stp(r24, r23,   r24_save);
-    __ strd(v11,     d11_save);
+    __ stp(r26, r25,   r26_save);
-    __ strd(v12,     d12_save);
+    __ stp(r28, r27,   r28_save);
-    __ strd(v13,     d13_save);
+
-    __ strd(v14,     d14_save);
+    __ stpd(v9,  v8,   d9_save);
-    __ strd(v15,     d15_save);
+    __ stpd(v11, v10,  d11_save);
+    __ stpd(v13, v12,  d13_save);
+    __ stpd(v15, v14,  d15_save);
 
     // install Java thread in global register now we have saved
     // whatever value it held
@@ -385,33 +354,22 @@ class StubGenerator: public StubCodeGenerator {
 #endif
 
     // restore callee-save registers
-    __ ldrd(v15,      d15_save);
+    __ ldpd(v15, v14,  d15_save);
-    __ ldrd(v14,      d14_save);
+    __ ldpd(v13, v12,  d13_save);
-    __ ldrd(v13,      d13_save);
+    __ ldpd(v11, v10,  d11_save);
-    __ ldrd(v12,      d12_save);
+    __ ldpd(v9,  v8,   d9_save);
-    __ ldrd(v11,      d11_save);
-    __ ldrd(v10,      d10_save);
-    __ ldrd(v9,       d9_save);
-    __ ldrd(v8,       d8_save);
 
-    __ ldr(r28,      r28_save);
+    __ ldp(r28, r27,   r28_save);
-    __ ldr(r27,      r27_save);
+    __ ldp(r26, r25,   r26_save);
-    __ ldr(r26,      r26_save);
+    __ ldp(r24, r23,   r24_save);
-    __ ldr(r25,      r25_save);
+    __ ldp(r22, r21,   r22_save);
-    __ ldr(r24,      r24_save);
+    __ ldp(r20, r19,   r20_save);
-    __ ldr(r23,      r23_save);
+
-    __ ldr(r22,      r22_save);
+    __ ldp(c_rarg0, c_rarg1,  call_wrapper);
-    __ ldr(r21,      r21_save);
-    __ ldr(r20,      r20_save);
-    __ ldr(r19,      r19_save);
-    __ ldr(c_rarg0,  call_wrapper);
-    __ ldr(c_rarg1,  result);
     __ ldrw(c_rarg2, result_type);
     __ ldr(c_rarg3,  method);
-    __ ldr(c_rarg4,  entry_point);
+    __ ldp(c_rarg4, c_rarg5,  entry_point);
-    __ ldr(c_rarg5,  parameters);
+    __ ldp(c_rarg6, c_rarg7,  parameter_size);
-    __ ldr(c_rarg6,  parameter_size);
-    __ ldr(c_rarg7,  thread);
 
 #ifndef PRODUCT
     // tell the simulator we are about to end Java execution
@@ -666,7 +624,7 @@ class StubGenerator: public StubCodeGenerator {
   //     count   -  element count
   //     tmp     - scratch register
   //
-  //     Destroy no registers!
+  //     Destroy no registers except rscratch1 and rscratch2
   //
   void  gen_write_ref_array_pre_barrier(Register addr, Register count, bool dest_uninitialized) {
     BarrierSet* bs = Universe::heap()->barrier_set();
@@ -674,12 +632,13 @@ class StubGenerator: public StubCodeGenerator {
     case BarrierSet::G1SATBCTLogging:
       // With G1, don't generate the call if we statically know that the target in uninitialized
       if (!dest_uninitialized) {
-        __ push(RegSet::range(r0, r29), sp);         // integer registers except lr & sp
+        __ push_call_clobbered_registers();
         if (count == c_rarg0) {
           if (addr == c_rarg1) {
             // exactly backwards!!
-            __ stp(c_rarg0, c_rarg1, __ pre(sp, -2 * wordSize));
+            __ mov(rscratch1, c_rarg0);
-            __ ldp(c_rarg1, c_rarg0, __ post(sp, -2 * wordSize));
+            __ mov(c_rarg0, c_rarg1);
+            __ mov(c_rarg1, rscratch1);
           } else {
             __ mov(c_rarg1, count);
             __ mov(c_rarg0, addr);
@@ -689,7 +648,7 @@ class StubGenerator: public StubCodeGenerator {
           __ mov(c_rarg1, count);
         }
         __ call_VM_leaf(CAST_FROM_FN_PTR(address, BarrierSet::static_write_ref_array_pre), 2);
-        __ pop(RegSet::range(r0, r29), sp);         // integer registers except lr & sp        }
+        __ pop_call_clobbered_registers();
         break;
       case BarrierSet::CardTableForRS:
       case BarrierSet::CardTableExtension:
@@ -719,7 +678,7 @@ class StubGenerator: public StubCodeGenerator {
       case BarrierSet::G1SATBCTLogging:
 
         {
-          __ push(RegSet::range(r0, r29), sp);         // integer registers except lr & sp
+          __ push_call_clobbered_registers();
           // must compute element count unless barrier set interface is changed (other platforms supply count)
           assert_different_registers(start, end, scratch);
           __ lea(scratch, Address(end, BytesPerHeapOop));
@@ -728,7 +687,7 @@ class StubGenerator: public StubCodeGenerator {
           __ mov(c_rarg0, start);
           __ mov(c_rarg1, scratch);
           __ call_VM_leaf(CAST_FROM_FN_PTR(address, BarrierSet::static_write_ref_array_post), 2);
-          __ pop(RegSet::range(r0, r29), sp);         // integer registers except lr & sp        }
+          __ pop_call_clobbered_registers();
         }
         break;
       case BarrierSet::CardTableForRS:
@@ -1394,10 +1353,10 @@ class StubGenerator: public StubCodeGenerator {
   //   no-overlap entry point used by generate_conjoint_long_oop_copy().
   //
   address generate_disjoint_oop_copy(bool aligned, address *entry,
-                                     const char *name, bool dest_uninitialized = false) {
+                                     const char *name, bool dest_uninitialized) {
     const bool is_oop = true;
     const size_t size = UseCompressedOops ? sizeof (jint) : sizeof (jlong);
-    return generate_disjoint_copy(size, aligned, is_oop, entry, name);
+    return generate_disjoint_copy(size, aligned, is_oop, entry, name, dest_uninitialized);
   }
 
   // Arguments:
@@ -1412,10 +1371,11 @@ class StubGenerator: public StubCodeGenerator {
   //
   address generate_conjoint_oop_copy(bool aligned,
                                      address nooverlap_target, address *entry,
-                                     const char *name, bool dest_uninitialized = false) {
+                                     const char *name, bool dest_uninitialized) {
     const bool is_oop = true;
     const size_t size = UseCompressedOops ? sizeof (jint) : sizeof (jlong);
-    return generate_conjoint_copy(size, aligned, is_oop, nooverlap_target, entry, name);
+    return generate_conjoint_copy(size, aligned, is_oop, nooverlap_target, entry,
+                                  name, dest_uninitialized);
   }
 
 
@@ -1522,6 +1482,8 @@ class StubGenerator: public StubCodeGenerator {
     }
 #endif //ASSERT
 
+    gen_write_ref_array_pre_barrier(to, count, dest_uninitialized);
+
     // save the original count
     __ mov(count_save, count);
 
@@ -1988,9 +1950,11 @@ class StubGenerator: public StubCodeGenerator {
       bool aligned = !UseCompressedOops;
 
       StubRoutines::_arrayof_oop_disjoint_arraycopy
-        = generate_disjoint_oop_copy(aligned, &entry, "arrayof_oop_disjoint_arraycopy");
+        = generate_disjoint_oop_copy(aligned, &entry, "arrayof_oop_disjoint_arraycopy",
+                                     /*dest_uninitialized*/false);
       StubRoutines::_arrayof_oop_arraycopy
-        = generate_conjoint_oop_copy(aligned, entry, &entry_oop_arraycopy, "arrayof_oop_arraycopy");
+        = generate_conjoint_oop_copy(aligned, entry, &entry_oop_arraycopy, "arrayof_oop_arraycopy",
+                                     /*dest_uninitialized*/false);
       // Aligned versions without pre-barriers
       StubRoutines::_arrayof_oop_disjoint_arraycopy_uninit
         = generate_disjoint_oop_copy(aligned, &entry, "arrayof_oop_disjoint_arraycopy_uninit",

hotspot/src/cpu/ppc/vm/globals_ppc.hpp

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2002, 2016, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2015 SAP SE. All rights reserved.
+ * Copyright (c) 2012, 2016 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -74,8 +74,7 @@ define_pd_global(size_t, CMSYoungGenPerWorker, 16*M);  // Default max size of CM
 
 define_pd_global(uintx, TypeProfileLevel, 111);
 
-// No performance work done here yet.
+define_pd_global(bool, CompactStrings, true);
-define_pd_global(bool, CompactStrings, false);
 
 // Platform dependent flag handling: flags only defined on this platform.
 #define ARCH_FLAGS(develop, product, diagnostic, experimental, notproduct, range, constraint)  \

hotspot/src/cpu/ppc/vm/macroAssembler_ppc.cpp

@@ -45,6 +45,9 @@
 #include "gc/g1/g1SATBCardTableModRefBS.hpp"
 #include "gc/g1/heapRegion.hpp"
 #endif // INCLUDE_ALL_GCS
+#ifdef COMPILER2
+#include "opto/intrinsicnode.hpp"
+#endif
 
 #ifdef PRODUCT
 #define BLOCK_COMMENT(str) // nothing
@@ -3168,6 +3171,553 @@ void MacroAssembler::clear_memory_doubleword(Register base_ptr, Register cnt_dwo
 
 /////////////////////////////////////////// String intrinsics ////////////////////////////////////////////
 
+#ifdef COMPILER2
+// Intrinsics for CompactStrings
+
+// Compress char[] to byte[] by compressing 16 bytes at once.
+void MacroAssembler::string_compress_16(Register src, Register dst, Register cnt,
+                                        Register tmp1, Register tmp2, Register tmp3, Register tmp4, Register tmp5,
+                                        Label& Lfailure) {
+
+  const Register tmp0 = R0;
+  assert_different_registers(src, dst, cnt, tmp0, tmp1, tmp2, tmp3, tmp4, tmp5);
+  Label Lloop, Lslow;
+
+  // Check if cnt >= 8 (= 16 bytes)
+  lis(tmp1, 0xFF);                // tmp1 = 0x00FF00FF00FF00FF
+  srwi_(tmp2, cnt, 3);
+  beq(CCR0, Lslow);
+  ori(tmp1, tmp1, 0xFF);
+  rldimi(tmp1, tmp1, 32, 0);
+  mtctr(tmp2);
+
+  // 2x unrolled loop
+  bind(Lloop);
+  ld(tmp2, 0, src);               // _0_1_2_3 (Big Endian)
+  ld(tmp4, 8, src);               // _4_5_6_7
+
+  orr(tmp0, tmp2, tmp4);
+  rldicl(tmp3, tmp2, 6*8, 64-24); // _____1_2
+  rldimi(tmp2, tmp2, 2*8, 2*8);   // _0_2_3_3
+  rldicl(tmp5, tmp4, 6*8, 64-24); // _____5_6
+  rldimi(tmp4, tmp4, 2*8, 2*8);   // _4_6_7_7
+
+  andc_(tmp0, tmp0, tmp1);
+  bne(CCR0, Lfailure);            // Not latin1.
+  addi(src, src, 16);
+
+  rlwimi(tmp3, tmp2, 0*8, 24, 31);// _____1_3
+  srdi(tmp2, tmp2, 3*8);          // ____0_2_
+  rlwimi(tmp5, tmp4, 0*8, 24, 31);// _____5_7
+  srdi(tmp4, tmp4, 3*8);          // ____4_6_
+
+  orr(tmp2, tmp2, tmp3);          // ____0123
+  orr(tmp4, tmp4, tmp5);          // ____4567
+
+  stw(tmp2, 0, dst);
+  stw(tmp4, 4, dst);
+  addi(dst, dst, 8);
+  bdnz(Lloop);
+
+  bind(Lslow);                    // Fallback to slow version
+}
+
+// Compress char[] to byte[]. cnt must be positive int.
+void MacroAssembler::string_compress(Register src, Register dst, Register cnt, Register tmp, Label& Lfailure) {
+  Label Lloop;
+  mtctr(cnt);
+
+  bind(Lloop);
+  lhz(tmp, 0, src);
+  cmplwi(CCR0, tmp, 0xff);
+  bgt(CCR0, Lfailure);            // Not latin1.
+  addi(src, src, 2);
+  stb(tmp, 0, dst);
+  addi(dst, dst, 1);
+  bdnz(Lloop);
+}
+
+// Inflate byte[] to char[] by inflating 16 bytes at once.
+void MacroAssembler::string_inflate_16(Register src, Register dst, Register cnt,
+                                       Register tmp1, Register tmp2, Register tmp3, Register tmp4, Register tmp5) {
+  const Register tmp0 = R0;
+  assert_different_registers(src, dst, cnt, tmp0, tmp1, tmp2, tmp3, tmp4, tmp5);
+  Label Lloop, Lslow;
+
+  // Check if cnt >= 8
+  srwi_(tmp2, cnt, 3);
+  beq(CCR0, Lslow);
+  lis(tmp1, 0xFF);                // tmp1 = 0x00FF00FF
+  ori(tmp1, tmp1, 0xFF);
+  mtctr(tmp2);
+
+  // 2x unrolled loop
+  bind(Lloop);
+  lwz(tmp2, 0, src);              // ____0123 (Big Endian)
+  lwz(tmp4, 4, src);              // ____4567
+  addi(src, src, 8);
+
+  rldicl(tmp3, tmp2, 7*8, 64-8);  // _______2
+  rlwimi(tmp2, tmp2, 3*8, 16, 23);// ____0113
+  rldicl(tmp5, tmp4, 7*8, 64-8);  // _______6
+  rlwimi(tmp4, tmp4, 3*8, 16, 23);// ____4557
+
+  andc(tmp0, tmp2, tmp1);         // ____0_1_
+  rlwimi(tmp2, tmp3, 2*8, 0, 23); // _____2_3
+  andc(tmp3, tmp4, tmp1);         // ____4_5_
+  rlwimi(tmp4, tmp5, 2*8, 0, 23); // _____6_7
+
+  rldimi(tmp2, tmp0, 3*8, 0*8);   // _0_1_2_3
+  rldimi(tmp4, tmp3, 3*8, 0*8);   // _4_5_6_7
+
+  std(tmp2, 0, dst);
+  std(tmp4, 8, dst);
+  addi(dst, dst, 16);
+  bdnz(Lloop);
+
+  bind(Lslow);                    // Fallback to slow version
+}
+
+// Inflate byte[] to char[]. cnt must be positive int.
+void MacroAssembler::string_inflate(Register src, Register dst, Register cnt, Register tmp) {
+  Label Lloop;
+  mtctr(cnt);
+
+  bind(Lloop);
+  lbz(tmp, 0, src);
+  addi(src, src, 1);
+  sth(tmp, 0, dst);
+  addi(dst, dst, 2);
+  bdnz(Lloop);
+}
+
+void MacroAssembler::string_compare(Register str1, Register str2,
+                                    Register cnt1, Register cnt2,
+                                    Register tmp1, Register result, int ae) {
+  const Register tmp0 = R0,
+                 diff = tmp1;
+
+  assert_different_registers(str1, str2, cnt1, cnt2, tmp0, tmp1, result);
+  Label Ldone, Lslow, Lloop, Lreturn_diff;
+
+  // Note: Making use of the fact that compareTo(a, b) == -compareTo(b, a)
+  // we interchange str1 and str2 in the UL case and negate the result.
+  // Like this, str1 is always latin1 encoded, except for the UU case.
+  // In addition, we need 0 (or sign which is 0) extend.
+
+  if (ae == StrIntrinsicNode::UU) {
+    srwi(cnt1, cnt1, 1);
+  } else {
+    clrldi(cnt1, cnt1, 32);
+  }
+
+  if (ae != StrIntrinsicNode::LL) {
+    srwi(cnt2, cnt2, 1);
+  } else {
+    clrldi(cnt2, cnt2, 32);
+  }
+
+  // See if the lengths are different, and calculate min in cnt1.
+  // Save diff in case we need it for a tie-breaker.
+  subf_(diff, cnt2, cnt1); // diff = cnt1 - cnt2
+  // if (diff > 0) { cnt1 = cnt2; }
+  if (VM_Version::has_isel()) {
+    isel(cnt1, CCR0, Assembler::greater, /*invert*/ false, cnt2);
+  } else {
+    Label Lskip;
+    blt(CCR0, Lskip);
+    mr(cnt1, cnt2);
+    bind(Lskip);
+  }
+
+  // Rename registers
+  Register chr1 = result;
+  Register chr2 = tmp0;
+
+  // Compare multiple characters in fast loop (only implemented for same encoding).
+  int stride1 = 8, stride2 = 8;
+  if (ae == StrIntrinsicNode::LL || ae == StrIntrinsicNode::UU) {
+    int log2_chars_per_iter = (ae == StrIntrinsicNode::LL) ? 3 : 2;
+    Label Lfastloop, Lskipfast;
+
+    srwi_(tmp0, cnt1, log2_chars_per_iter);
+    beq(CCR0, Lskipfast);
+    rldicl(cnt2, cnt1, 0, 64 - log2_chars_per_iter); // Remaining characters.
+    li(cnt1, 1 << log2_chars_per_iter); // Initialize for failure case: Rescan characters from current iteration.
+    mtctr(tmp0);
+
+    bind(Lfastloop);
+    ld(chr1, 0, str1);
+    ld(chr2, 0, str2);
+    cmpd(CCR0, chr1, chr2);
+    bne(CCR0, Lslow);
+    addi(str1, str1, stride1);
+    addi(str2, str2, stride2);
+    bdnz(Lfastloop);
+    mr(cnt1, cnt2); // Remaining characters.
+    bind(Lskipfast);
+  }
+
+  // Loop which searches the first difference character by character.
+  cmpwi(CCR0, cnt1, 0);
+  beq(CCR0, Lreturn_diff);
+  bind(Lslow);
+  mtctr(cnt1);
+
+  switch (ae) {
+    case StrIntrinsicNode::LL: stride1 = 1; stride2 = 1; break;
+    case StrIntrinsicNode::UL: // fallthru (see comment above)
+    case StrIntrinsicNode::LU: stride1 = 1; stride2 = 2; break;
+    case StrIntrinsicNode::UU: stride1 = 2; stride2 = 2; break;
+    default: ShouldNotReachHere(); break;
+  }
+
+  bind(Lloop);
+  if (stride1 == 1) { lbz(chr1, 0, str1); } else { lhz(chr1, 0, str1); }
+  if (stride2 == 1) { lbz(chr2, 0, str2); } else { lhz(chr2, 0, str2); }
+  subf_(result, chr2, chr1); // result = chr1 - chr2
+  bne(CCR0, Ldone);
+  addi(str1, str1, stride1);
+  addi(str2, str2, stride2);
+  bdnz(Lloop);
+
+  // If strings are equal up to min length, return the length difference.
+  bind(Lreturn_diff);
+  mr(result, diff);
+
+  // Otherwise, return the difference between the first mismatched chars.
+  bind(Ldone);
+  if (ae == StrIntrinsicNode::UL) {
+    neg(result, result); // Negate result (see note above).
+  }
+}
+
+void MacroAssembler::array_equals(bool is_array_equ, Register ary1, Register ary2,
+                                  Register limit, Register tmp1, Register result, bool is_byte) {
+  const Register tmp0 = R0;
+  assert_different_registers(ary1, ary2, limit, tmp0, tmp1, result);
+  Label Ldone, Lskiploop, Lloop, Lfastloop, Lskipfast;
+  bool limit_needs_shift = false;
+
+  if (is_array_equ) {
+    const int length_offset = arrayOopDesc::length_offset_in_bytes();
+    const int base_offset   = arrayOopDesc::base_offset_in_bytes(is_byte ? T_BYTE : T_CHAR);
+
+    // Return true if the same array.
+    cmpd(CCR0, ary1, ary2);
+    beq(CCR0, Lskiploop);
+
+    // Return false if one of them is NULL.
+    cmpdi(CCR0, ary1, 0);
+    cmpdi(CCR1, ary2, 0);
+    li(result, 0);
+    cror(CCR0, Assembler::equal, CCR1, Assembler::equal);
+    beq(CCR0, Ldone);
+
+    // Load the lengths of arrays.
+    lwz(limit, length_offset, ary1);
+    lwz(tmp0, length_offset, ary2);
+
+    // Return false if the two arrays are not equal length.
+    cmpw(CCR0, limit, tmp0);
+    bne(CCR0, Ldone);
+
+    // Load array addresses.
+    addi(ary1, ary1, base_offset);
+    addi(ary2, ary2, base_offset);
+  } else {
+    limit_needs_shift = !is_byte;
+    li(result, 0); // Assume not equal.
+  }
+
+  // Rename registers
+  Register chr1 = tmp0;
+  Register chr2 = tmp1;
+
+  // Compare 8 bytes per iteration in fast loop.
+  const int log2_chars_per_iter = is_byte ? 3 : 2;
+
+  srwi_(tmp0, limit, log2_chars_per_iter + (limit_needs_shift ? 1 : 0));
+  beq(CCR0, Lskipfast);
+  mtctr(tmp0);
+
+  bind(Lfastloop);
+  ld(chr1, 0, ary1);
+  ld(chr2, 0, ary2);
+  addi(ary1, ary1, 8);
+  addi(ary2, ary2, 8);
+  cmpd(CCR0, chr1, chr2);
+  bne(CCR0, Ldone);
+  bdnz(Lfastloop);
+
+  bind(Lskipfast);
+  rldicl_(limit, limit, limit_needs_shift ? 64 - 1 : 0, 64 - log2_chars_per_iter); // Remaining characters.
+  beq(CCR0, Lskiploop);
+  mtctr(limit);
+
+  // Character by character.
+  bind(Lloop);
+  if (is_byte) {
+    lbz(chr1, 0, ary1);
+    lbz(chr2, 0, ary2);
+    addi(ary1, ary1, 1);
+    addi(ary2, ary2, 1);
+  } else {
+    lhz(chr1, 0, ary1);
+    lhz(chr2, 0, ary2);
+    addi(ary1, ary1, 2);
+    addi(ary2, ary2, 2);
+  }
+  cmpw(CCR0, chr1, chr2);
+  bne(CCR0, Ldone);
+  bdnz(Lloop);
+
+  bind(Lskiploop);
+  li(result, 1); // All characters are equal.
+  bind(Ldone);
+}
+
+void MacroAssembler::string_indexof(Register result, Register haystack, Register haycnt,
+                                    Register needle, ciTypeArray* needle_values, Register needlecnt, int needlecntval,
+                                    Register tmp1, Register tmp2, Register tmp3, Register tmp4, int ae) {
+
+  // Ensure 0<needlecnt<=haycnt in ideal graph as prerequisite!
+  Label L_TooShort, L_Found, L_NotFound, L_End;
+  Register last_addr = haycnt, // Kill haycnt at the beginning.
+  addr      = tmp1,
+  n_start   = tmp2,
+  ch1       = tmp3,
+  ch2       = R0;
+
+  assert(ae != StrIntrinsicNode::LU, "Invalid encoding");
+  const int h_csize = (ae == StrIntrinsicNode::LL) ? 1 : 2;
+  const int n_csize = (ae == StrIntrinsicNode::UU) ? 2 : 1;
+
+  // **************************************************************************************************
+  // Prepare for main loop: optimized for needle count >=2, bail out otherwise.
+  // **************************************************************************************************
+
+  // Compute last haystack addr to use if no match gets found.
+  clrldi(haycnt, haycnt, 32);         // Ensure positive int is valid as 64 bit value.
+  addi(addr, haystack, -h_csize);     // Accesses use pre-increment.
+  if (needlecntval == 0) { // variable needlecnt
+   cmpwi(CCR6, needlecnt, 2);
+   clrldi(needlecnt, needlecnt, 32);  // Ensure positive int is valid as 64 bit value.
+   blt(CCR6, L_TooShort);             // Variable needlecnt: handle short needle separately.
+  }
+
+  if (n_csize == 2) { lwz(n_start, 0, needle); } else { lhz(n_start, 0, needle); } // Load first 2 characters of needle.
+
+  if (needlecntval == 0) { // variable needlecnt
+   subf(ch1, needlecnt, haycnt);      // Last character index to compare is haycnt-needlecnt.
+   addi(needlecnt, needlecnt, -2);    // Rest of needle.
+  } else { // constant needlecnt
+  guarantee(needlecntval != 1, "IndexOf with single-character needle must be handled separately");
+  assert((needlecntval & 0x7fff) == needlecntval, "wrong immediate");
+   addi(ch1, haycnt, -needlecntval);  // Last character index to compare is haycnt-needlecnt.
+   if (needlecntval > 3) { li(needlecnt, needlecntval - 2); } // Rest of needle.
+  }
+
+  if (h_csize == 2) { slwi(ch1, ch1, 1); } // Scale to number of bytes.
+
+  if (ae ==StrIntrinsicNode::UL) {
+   srwi(tmp4, n_start, 1*8);          // ___0
+   rlwimi(n_start, tmp4, 2*8, 0, 23); // _0_1
+  }
+
+  add(last_addr, haystack, ch1);      // Point to last address to compare (haystack+2*(haycnt-needlecnt)).
+
+  // Main Loop (now we have at least 2 characters).
+  Label L_OuterLoop, L_InnerLoop, L_FinalCheck, L_Comp1, L_Comp2;
+  bind(L_OuterLoop); // Search for 1st 2 characters.
+  Register addr_diff = tmp4;
+   subf(addr_diff, addr, last_addr);  // Difference between already checked address and last address to check.
+   addi(addr, addr, h_csize);         // This is the new address we want to use for comparing.
+   srdi_(ch2, addr_diff, h_csize);
+   beq(CCR0, L_FinalCheck);           // 2 characters left?
+   mtctr(ch2);                        // num of characters / 2
+  bind(L_InnerLoop);                  // Main work horse (2x unrolled search loop)
+   if (h_csize == 2) {                // Load 2 characters of haystack (ignore alignment).
+    lwz(ch1, 0, addr);
+    lwz(ch2, 2, addr);
+   } else {
+    lhz(ch1, 0, addr);
+    lhz(ch2, 1, addr);
+   }
+   cmpw(CCR0, ch1, n_start);          // Compare 2 characters (1 would be sufficient but try to reduce branches to CompLoop).
+   cmpw(CCR1, ch2, n_start);
+   beq(CCR0, L_Comp1);                // Did we find the needle start?
+   beq(CCR1, L_Comp2);
+   addi(addr, addr, 2 * h_csize);
+   bdnz(L_InnerLoop);
+  bind(L_FinalCheck);
+   andi_(addr_diff, addr_diff, h_csize); // Remaining characters not covered by InnerLoop: (num of characters) & 1.
+   beq(CCR0, L_NotFound);
+   if (h_csize == 2) { lwz(ch1, 0, addr); } else { lhz(ch1, 0, addr); } // One position left at which we have to compare.
+   cmpw(CCR1, ch1, n_start);
+   beq(CCR1, L_Comp1);
+  bind(L_NotFound);
+   li(result, -1);                    // not found
+   b(L_End);
+
+   // **************************************************************************************************
+   // Special Case: unfortunately, the variable needle case can be called with needlecnt<2
+   // **************************************************************************************************
+  if (needlecntval == 0) {           // We have to handle these cases separately.
+  Label L_OneCharLoop;
+  bind(L_TooShort);
+   mtctr(haycnt);
+   if (n_csize == 2) { lhz(n_start, 0, needle); } else { lbz(n_start, 0, needle); } // First character of needle
+  bind(L_OneCharLoop);
+   if (h_csize == 2) { lhzu(ch1, 2, addr); } else { lbzu(ch1, 1, addr); }
+   cmpw(CCR1, ch1, n_start);
+   beq(CCR1, L_Found);               // Did we find the one character needle?
+   bdnz(L_OneCharLoop);
+   li(result, -1);                   // Not found.
+   b(L_End);
+  }
+
+  // **************************************************************************************************
+  // Regular Case Part II: compare rest of needle (first 2 characters have been compared already)
+  // **************************************************************************************************
+
+  // Compare the rest
+  bind(L_Comp2);
+   addi(addr, addr, h_csize);        // First comparison has failed, 2nd one hit.
+  bind(L_Comp1);                     // Addr points to possible needle start.
+  if (needlecntval != 2) {           // Const needlecnt==2?
+   if (needlecntval != 3) {
+    if (needlecntval == 0) { beq(CCR6, L_Found); } // Variable needlecnt==2?
+    Register n_ind = tmp4,
+             h_ind = n_ind;
+    li(n_ind, 2 * n_csize);          // First 2 characters are already compared, use index 2.
+    mtctr(needlecnt);                // Decremented by 2, still > 0.
+   Label L_CompLoop;
+   bind(L_CompLoop);
+    if (ae ==StrIntrinsicNode::UL) {
+      h_ind = ch1;
+      sldi(h_ind, n_ind, 1);
+    }
+    if (n_csize == 2) { lhzx(ch2, needle, n_ind); } else { lbzx(ch2, needle, n_ind); }
+    if (h_csize == 2) { lhzx(ch1, addr, h_ind); } else { lbzx(ch1, addr, h_ind); }
+    cmpw(CCR1, ch1, ch2);
+    bne(CCR1, L_OuterLoop);
+    addi(n_ind, n_ind, n_csize);
+    bdnz(L_CompLoop);
+   } else { // No loop required if there's only one needle character left.
+    if (n_csize == 2) { lhz(ch2, 2 * 2, needle); } else { lbz(ch2, 2 * 1, needle); }
+    if (h_csize == 2) { lhz(ch1, 2 * 2, addr); } else { lbz(ch1, 2 * 1, addr); }
+    cmpw(CCR1, ch1, ch2);
+    bne(CCR1, L_OuterLoop);
+   }
+  }
+  // Return index ...
+  bind(L_Found);
+   subf(result, haystack, addr);     // relative to haystack, ...
+   if (h_csize == 2) { srdi(result, result, 1); } // in characters.
+  bind(L_End);
+} // string_indexof
+
+void MacroAssembler::string_indexof_char(Register result, Register haystack, Register haycnt,
+                                         Register needle, jchar needleChar, Register tmp1, Register tmp2, bool is_byte) {
+  assert_different_registers(haystack, haycnt, needle, tmp1, tmp2);
+
+  Label L_InnerLoop, L_FinalCheck, L_Found1, L_Found2, L_NotFound, L_End;
+  Register addr = tmp1,
+           ch1 = tmp2,
+           ch2 = R0;
+
+  const int h_csize = is_byte ? 1 : 2;
+
+//4:
+   srwi_(tmp2, haycnt, 1);   // Shift right by exact_log2(UNROLL_FACTOR).
+   mr(addr, haystack);
+   beq(CCR0, L_FinalCheck);
+   mtctr(tmp2);              // Move to count register.
+//8:
+  bind(L_InnerLoop);         // Main work horse (2x unrolled search loop).
+   if (!is_byte) {
+    lhz(ch1, 0, addr);
+    lhz(ch2, 2, addr);
+   } else {
+    lbz(ch1, 0, addr);
+    lbz(ch2, 1, addr);
+   }
+   (needle != R0) ? cmpw(CCR0, ch1, needle) : cmplwi(CCR0, ch1, (unsigned int)needleChar);
+   (needle != R0) ? cmpw(CCR1, ch2, needle) : cmplwi(CCR1, ch2, (unsigned int)needleChar);
+   beq(CCR0, L_Found1);      // Did we find the needle?
+   beq(CCR1, L_Found2);
+   addi(addr, addr, 2 * h_csize);
+   bdnz(L_InnerLoop);
+//16:
+  bind(L_FinalCheck);
+   andi_(R0, haycnt, 1);
+   beq(CCR0, L_NotFound);
+   if (!is_byte) { lhz(ch1, 0, addr); } else { lbz(ch1, 0, addr); } // One position left at which we have to compare.
+   (needle != R0) ? cmpw(CCR1, ch1, needle) : cmplwi(CCR1, ch1, (unsigned int)needleChar);
+   beq(CCR1, L_Found1);
+//21:
+  bind(L_NotFound);
+   li(result, -1);           // Not found.
+   b(L_End);
+
+  bind(L_Found2);
+   addi(addr, addr, h_csize);
+//24:
+  bind(L_Found1);            // Return index ...
+   subf(result, haystack, addr); // relative to haystack, ...
+   if (!is_byte) { srdi(result, result, 1); } // in characters.
+  bind(L_End);
+} // string_indexof_char
+
+
+void MacroAssembler::has_negatives(Register src, Register cnt, Register result,
+                                   Register tmp1, Register tmp2) {
+  const Register tmp0 = R0;
+  assert_different_registers(src, result, cnt, tmp0, tmp1, tmp2);
+  Label Lfastloop, Lslow, Lloop, Lnoneg, Ldone;
+
+  // Check if cnt >= 8 (= 16 bytes)
+  lis(tmp1, (int)(short)0x8080);  // tmp1 = 0x8080808080808080
+  srwi_(tmp2, cnt, 4);
+  li(result, 1);                  // Assume there's a negative byte.
+  beq(CCR0, Lslow);
+  ori(tmp1, tmp1, 0x8080);
+  rldimi(tmp1, tmp1, 32, 0);
+  mtctr(tmp2);
+
+  // 2x unrolled loop
+  bind(Lfastloop);
+  ld(tmp2, 0, src);
+  ld(tmp0, 8, src);
+
+  orr(tmp0, tmp2, tmp0);
+
+  and_(tmp0, tmp0, tmp1);
+  bne(CCR0, Ldone);               // Found negative byte.
+  addi(src, src, 16);
+
+  bdnz(Lfastloop);
+
+  bind(Lslow);                    // Fallback to slow version
+  rldicl_(tmp0, cnt, 0, 64-4);
+  beq(CCR0, Lnoneg);
+  mtctr(tmp0);
+  bind(Lloop);
+  lbz(tmp0, 0, src);
+  addi(src, src, 1);
+  andi_(tmp0, tmp0, 0x80);
+  bne(CCR0, Ldone);               // Found negative byte.
+  bdnz(Lloop);
+  bind(Lnoneg);
+  li(result, 0);
+
+  bind(Ldone);
+}
+
+
+// Intrinsics for non-CompactStrings
+
 // Search for a single jchar in an jchar[].
 //
 // Assumes that result differs from all other registers.
@@ -3613,6 +4163,8 @@ void MacroAssembler::char_arrays_equalsImm(Register str1_reg, Register str2_reg,
   bind(Ldone_false);
 }
 
+#endif // Compiler2
+
 // Helpers for Intrinsic Emitters
 //
 // Revert the byte order of a 32bit value in a register

hotspot/src/cpu/ppc/vm/macroAssembler_ppc.hpp

@@ -1,6 +1,6 @@
 /*
- * Copyright (c) 2002, 2015, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2002, 2016, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2015 SAP SE. All rights reserved.
+ * Copyright (c) 2012, 2016 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -679,6 +679,39 @@ class MacroAssembler: public Assembler {
 
   void clear_memory_doubleword(Register base_ptr, Register cnt_dwords, Register tmp = R0);
 
+#ifdef COMPILER2
+  // Intrinsics for CompactStrings
+  // Compress char[] to byte[] by compressing 16 bytes at once.
+  void string_compress_16(Register src, Register dst, Register cnt,
+                          Register tmp1, Register tmp2, Register tmp3, Register tmp4, Register tmp5,
+                          Label& Lfailure);
+
+  // Compress char[] to byte[]. cnt must be positive int.
+  void string_compress(Register src, Register dst, Register cnt, Register tmp, Label& Lfailure);
+
+  // Inflate byte[] to char[] by inflating 16 bytes at once.
+  void string_inflate_16(Register src, Register dst, Register cnt,
+                         Register tmp1, Register tmp2, Register tmp3, Register tmp4, Register tmp5);
+
+  // Inflate byte[] to char[]. cnt must be positive int.
+  void string_inflate(Register src, Register dst, Register cnt, Register tmp);
+
+  void string_compare(Register str1, Register str2, Register cnt1, Register cnt2,
+                      Register tmp1, Register result, int ae);
+
+  void array_equals(bool is_array_equ, Register ary1, Register ary2,
+                    Register limit, Register tmp1, Register result, bool is_byte);
+
+  void string_indexof(Register result, Register haystack, Register haycnt,
+                      Register needle, ciTypeArray* needle_values, Register needlecnt, int needlecntval,
+                      Register tmp1, Register tmp2, Register tmp3, Register tmp4, int ae);
+
+  void string_indexof_char(Register result, Register haystack, Register haycnt,
+                           Register needle, jchar needleChar, Register tmp1, Register tmp2, bool is_byte);
+
+  void has_negatives(Register src, Register cnt, Register result, Register tmp1, Register tmp2);
+
+  // Intrinsics for non-CompactStrings
   // Needle of length 1.
   void string_indexof_1(Register result, Register haystack, Register haycnt,
                         Register needle, jchar needleChar,
@@ -694,6 +727,7 @@ class MacroAssembler: public Assembler {
                           Register tmp5_reg);
   void char_arrays_equalsImm(Register str1_reg, Register str2_reg, int cntval, Register result_reg,
                              Register tmp1_reg, Register tmp2_reg);
+#endif
 
   // Emitters for BigInteger.multiplyToLen intrinsic.
   inline void multiply64(Register dest_hi, Register dest_lo,

hotspot/src/cpu/ppc/vm/ppc.ad

@@ -1,6 +1,6 @@
 //
 // Copyright (c) 2011, 2016, Oracle and/or its affiliates. All rights reserved.
-// Copyright (c) 2012, 2015 SAP SE. All rights reserved.
+// Copyright (c) 2012, 2016 SAP SE. All rights reserved.
 // DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 //
 // This code is free software; you can redistribute it and/or modify it
@@ -2024,13 +2024,13 @@ const bool Matcher::match_rule_supported(int opcode) {
     return (UsePopCountInstruction && VM_Version::has_popcntw());
 
   case Op_StrComp:
-    return SpecialStringCompareTo && !CompactStrings;
+    return SpecialStringCompareTo;
   case Op_StrEquals:
-    return SpecialStringEquals && !CompactStrings;
+    return SpecialStringEquals;
   case Op_StrIndexOf:
-    return SpecialStringIndexOf && !CompactStrings;
+    return SpecialStringIndexOf;
   case Op_StrIndexOfChar:
-    return SpecialStringIndexOf && !CompactStrings;
+    return SpecialStringIndexOf;
   }
 
   return true;  // Per default match rules are supported.
@@ -11022,6 +11022,584 @@ instruct inlineCallClearArray(rarg1RegL cnt, rarg2RegP base, Universe dummy, reg
   ins_pipe(pipe_class_default);
 %}
 
+instruct string_compareL(rarg1RegP str1, rarg2RegP str2, rarg3RegI cnt1, rarg4RegI cnt2, iRegIdst result,
+                         iRegIdst tmp, regCTR ctr, flagsRegCR0 cr0) %{
+  predicate(((StrCompNode*)n)->encoding() == StrIntrinsicNode::LL);
+  match(Set result (StrComp (Binary str1 cnt1) (Binary str2 cnt2)));
+  effect(TEMP_DEF result, USE_KILL str1, USE_KILL str2, USE_KILL cnt1, USE_KILL cnt2, KILL ctr, KILL cr0, TEMP tmp);
+  ins_cost(300);
+  format %{ "String Compare byte[] $str1,$cnt1,$str2,$cnt2 -> $result \t// KILL $tmp" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ string_compare($str1$$Register, $str2$$Register,
+                      $cnt1$$Register, $cnt2$$Register,
+                      $tmp$$Register,
+                      $result$$Register, StrIntrinsicNode::LL);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+instruct string_compareU(rarg1RegP str1, rarg2RegP str2, rarg3RegI cnt1, rarg4RegI cnt2, iRegIdst result,
+                         iRegIdst tmp, regCTR ctr, flagsRegCR0 cr0) %{
+  predicate(((StrCompNode*)n)->encoding() == StrIntrinsicNode::UU);
+  match(Set result (StrComp (Binary str1 cnt1) (Binary str2 cnt2)));
+  effect(TEMP_DEF result, USE_KILL str1, USE_KILL str2, USE_KILL cnt1, USE_KILL cnt2, KILL ctr, KILL cr0, TEMP tmp);
+  ins_cost(300);
+  format %{ "String Compare char[] $str1,$cnt1,$str2,$cnt2 -> $result \t// KILL $tmp" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ string_compare($str1$$Register, $str2$$Register,
+                      $cnt1$$Register, $cnt2$$Register,
+                      $tmp$$Register,
+                      $result$$Register, StrIntrinsicNode::UU);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+instruct string_compareLU(rarg1RegP str1, rarg2RegP str2, rarg3RegI cnt1, rarg4RegI cnt2, iRegIdst result,
+                          iRegIdst tmp, regCTR ctr, flagsRegCR0 cr0) %{
+  predicate(((StrCompNode*)n)->encoding() == StrIntrinsicNode::LU);
+  match(Set result (StrComp (Binary str1 cnt1) (Binary str2 cnt2)));
+  effect(TEMP_DEF result, USE_KILL str1, USE_KILL str2, USE_KILL cnt1, USE_KILL cnt2, KILL ctr, KILL cr0, TEMP tmp);
+  ins_cost(300);
+  format %{ "String Compare byte[] $str1,$cnt1,$str2,$cnt2 -> $result \t// KILL $tmp" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ string_compare($str1$$Register, $str2$$Register,
+                      $cnt1$$Register, $cnt2$$Register,
+                      $tmp$$Register,
+                      $result$$Register, StrIntrinsicNode::LU);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+instruct string_compareUL(rarg1RegP str1, rarg2RegP str2, rarg3RegI cnt1, rarg4RegI cnt2, iRegIdst result,
+                          iRegIdst tmp, regCTR ctr, flagsRegCR0 cr0) %{
+  predicate(((StrCompNode*)n)->encoding() == StrIntrinsicNode::UL);
+  match(Set result (StrComp (Binary str1 cnt1) (Binary str2 cnt2)));
+  effect(TEMP_DEF result, USE_KILL str1, USE_KILL str2, USE_KILL cnt1, USE_KILL cnt2, KILL ctr, KILL cr0, TEMP tmp);
+  ins_cost(300);
+  format %{ "String Compare byte[] $str1,$cnt1,$str2,$cnt2 -> $result \t// KILL $tmp" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ string_compare($str2$$Register, $str1$$Register,
+                      $cnt2$$Register, $cnt1$$Register,
+                      $tmp$$Register,
+                      $result$$Register, StrIntrinsicNode::UL);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+instruct string_equalsL(rarg1RegP str1, rarg2RegP str2, rarg3RegI cnt, iRegIdst result,
+                        iRegIdst tmp, regCTR ctr, flagsRegCR0 cr0) %{
+  predicate(((StrEqualsNode*)n)->encoding() == StrIntrinsicNode::LL);
+  match(Set result (StrEquals (Binary str1 str2) cnt));
+  effect(TEMP_DEF result, USE_KILL str1, USE_KILL str2, USE_KILL cnt, TEMP tmp, KILL ctr, KILL cr0);
+  ins_cost(300);
+  format %{ "String Equals byte[] $str1,$str2,$cnt -> $result \t// KILL $tmp" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ array_equals(false, $str1$$Register, $str2$$Register,
+                    $cnt$$Register, $tmp$$Register,
+                    $result$$Register, true /* byte */);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+instruct string_equalsU(rarg1RegP str1, rarg2RegP str2, rarg3RegI cnt, iRegIdst result,
+                        iRegIdst tmp, regCTR ctr, flagsRegCR0 cr0) %{
+  predicate(((StrEqualsNode*)n)->encoding() == StrIntrinsicNode::UU);
+  match(Set result (StrEquals (Binary str1 str2) cnt));
+  effect(TEMP_DEF result, USE_KILL str1, USE_KILL str2, USE_KILL cnt, TEMP tmp, KILL ctr, KILL cr0);
+  ins_cost(300);
+  format %{ "String Equals char[]  $str1,$str2,$cnt -> $result \t// KILL $tmp" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ array_equals(false, $str1$$Register, $str2$$Register,
+                    $cnt$$Register, $tmp$$Register,
+                    $result$$Register, false /* byte */);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+instruct array_equalsB(rarg1RegP ary1, rarg2RegP ary2, iRegIdst result,
+                       iRegIdst tmp1, iRegIdst tmp2, regCTR ctr, flagsRegCR0 cr0, flagsRegCR0 cr1) %{
+  predicate(((AryEqNode*)n)->encoding() == StrIntrinsicNode::LL);
+  match(Set result (AryEq ary1 ary2));
+  effect(TEMP_DEF result, USE_KILL ary1, USE_KILL ary2, TEMP tmp1, TEMP tmp2, KILL ctr, KILL cr0, KILL cr1);
+  ins_cost(300);
+  format %{ "Array Equals $ary1,$ary2 -> $result \t// KILL $tmp1,$tmp2" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ array_equals(true, $ary1$$Register, $ary2$$Register,
+                    $tmp1$$Register, $tmp2$$Register,
+                    $result$$Register, true /* byte */);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+instruct array_equalsC(rarg1RegP ary1, rarg2RegP ary2, iRegIdst result,
+                       iRegIdst tmp1, iRegIdst tmp2, regCTR ctr, flagsRegCR0 cr0, flagsRegCR0 cr1) %{
+  predicate(((AryEqNode*)n)->encoding() == StrIntrinsicNode::UU);
+  match(Set result (AryEq ary1 ary2));
+  effect(TEMP_DEF result, USE_KILL ary1, USE_KILL ary2, TEMP tmp1, TEMP tmp2, KILL ctr, KILL cr0, KILL cr1);
+  ins_cost(300);
+  format %{ "Array Equals $ary1,$ary2 -> $result \t// KILL $tmp1,$tmp2" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ array_equals(true, $ary1$$Register, $ary2$$Register,
+                    $tmp1$$Register, $tmp2$$Register,
+                    $result$$Register, false /* byte */);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+instruct indexOf_imm1_char_U(iRegIdst result, iRegPsrc haystack, iRegIsrc haycnt,
+                             immP needleImm, immL offsetImm, immI_1 needlecntImm,
+                             iRegIdst tmp1, iRegIdst tmp2,
+                             flagsRegCR0 cr0, flagsRegCR1 cr1, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary (AddP needleImm offsetImm) needlecntImm)));
+  effect(TEMP tmp1, TEMP tmp2, KILL cr0, KILL cr1, KILL ctr);
+  // Required for EA: check if it is still a type_array.
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::UU);
+  ins_cost(150);
+
+  format %{ "String IndexOf CSCL1 $haystack[0..$haycnt], $needleImm+$offsetImm[0..$needlecntImm]"
+            "-> $result \t// KILL $haycnt, $tmp1, $tmp2, $cr0, $cr1" %}
+
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    immPOper *needleOper = (immPOper *)$needleImm;
+    const TypeOopPtr *t = needleOper->type()->isa_oopptr();
+    ciTypeArray* needle_values = t->const_oop()->as_type_array();  // Pointer to live char *
+    jchar chr;
+#ifdef VM_LITTLE_ENDIAN
+    chr = (((jchar)(unsigned char)needle_values->element_value(1).as_byte()) << 8) |
+           ((jchar)(unsigned char)needle_values->element_value(0).as_byte());
+#else
+    chr = (((jchar)(unsigned char)needle_values->element_value(0).as_byte()) << 8) |
+           ((jchar)(unsigned char)needle_values->element_value(1).as_byte());
+#endif
+    __ string_indexof_char($result$$Register,
+                           $haystack$$Register, $haycnt$$Register,
+                           R0, chr,
+                           $tmp1$$Register, $tmp2$$Register, false /*is_byte*/);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_imm1_char_L(iRegIdst result, iRegPsrc haystack, iRegIsrc haycnt,
+                             immP needleImm, immL offsetImm, immI_1 needlecntImm,
+                             iRegIdst tmp1, iRegIdst tmp2,
+                             flagsRegCR0 cr0, flagsRegCR1 cr1, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary (AddP needleImm offsetImm) needlecntImm)));
+  effect(TEMP tmp1, TEMP tmp2, KILL cr0, KILL cr1, KILL ctr);
+  // Required for EA: check if it is still a type_array.
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::LL);
+  ins_cost(150);
+
+  format %{ "String IndexOf CSCL1 $haystack[0..$haycnt], $needleImm+$offsetImm[0..$needlecntImm]"
+            "-> $result \t// KILL $haycnt, $tmp1, $tmp2, $cr0, $cr1" %}
+
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    immPOper *needleOper = (immPOper *)$needleImm;
+    const TypeOopPtr *t = needleOper->type()->isa_oopptr();
+    ciTypeArray* needle_values = t->const_oop()->as_type_array();  // Pointer to live char *
+    jchar chr = (jchar)needle_values->element_value(0).as_byte();
+    __ string_indexof_char($result$$Register,
+                           $haystack$$Register, $haycnt$$Register,
+                           R0, chr,
+                           $tmp1$$Register, $tmp2$$Register, true /*is_byte*/);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_imm1_char_UL(iRegIdst result, iRegPsrc haystack, iRegIsrc haycnt,
+                              immP needleImm, immL offsetImm, immI_1 needlecntImm,
+                              iRegIdst tmp1, iRegIdst tmp2,
+                              flagsRegCR0 cr0, flagsRegCR1 cr1, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary (AddP needleImm offsetImm) needlecntImm)));
+  effect(TEMP tmp1, TEMP tmp2, KILL cr0, KILL cr1, KILL ctr);
+  // Required for EA: check if it is still a type_array.
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::UL);
+  ins_cost(150);
+
+  format %{ "String IndexOf CSCL1 $haystack[0..$haycnt], $needleImm+$offsetImm[0..$needlecntImm]"
+            "-> $result \t// KILL $haycnt, $tmp1, $tmp2, $cr0, $cr1" %}
+
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    immPOper *needleOper = (immPOper *)$needleImm;
+    const TypeOopPtr *t = needleOper->type()->isa_oopptr();
+    ciTypeArray* needle_values = t->const_oop()->as_type_array();  // Pointer to live char *
+    jchar chr = (jchar)needle_values->element_value(0).as_byte();
+    __ string_indexof_char($result$$Register,
+                           $haystack$$Register, $haycnt$$Register,
+                           R0, chr,
+                           $tmp1$$Register, $tmp2$$Register, false /*is_byte*/);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_imm1_U(iRegIdst result, iRegPsrc haystack, iRegIsrc haycnt,
+                        rscratch2RegP needle, immI_1 needlecntImm,
+                        iRegIdst tmp1, iRegIdst tmp2,
+                        flagsRegCR0 cr0, flagsRegCR1 cr1, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary needle needlecntImm)));
+  effect(USE_KILL needle, TEMP tmp1, TEMP tmp2, KILL cr0, KILL cr1, KILL ctr);
+  // Required for EA: check if it is still a type_array.
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::UU &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop() &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop()->is_type_array());
+  ins_cost(180);
+
+  format %{ "String IndexOf SCL1 $haystack[0..$haycnt], $needle[0..$needlecntImm]"
+            " -> $result \t// KILL $haycnt, $needle, $tmp1, $tmp2, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    Node *ndl = in(operand_index($needle));  // The node that defines needle.
+    ciTypeArray* needle_values = ndl->bottom_type()->is_aryptr()->const_oop()->as_type_array();
+    guarantee(needle_values, "sanity");
+    jchar chr;
+#ifdef VM_LITTLE_ENDIAN
+    chr = (((jchar)(unsigned char)needle_values->element_value(1).as_byte()) << 8) |
+           ((jchar)(unsigned char)needle_values->element_value(0).as_byte());
+#else
+    chr = (((jchar)(unsigned char)needle_values->element_value(0).as_byte()) << 8) |
+           ((jchar)(unsigned char)needle_values->element_value(1).as_byte());
+#endif
+    __ string_indexof_char($result$$Register,
+                           $haystack$$Register, $haycnt$$Register,
+                           R0, chr,
+                           $tmp1$$Register, $tmp2$$Register, false /*is_byte*/);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_imm1_L(iRegIdst result, iRegPsrc haystack, iRegIsrc haycnt,
+                        rscratch2RegP needle, immI_1 needlecntImm,
+                        iRegIdst tmp1, iRegIdst tmp2,
+                        flagsRegCR0 cr0, flagsRegCR1 cr1, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary needle needlecntImm)));
+  effect(USE_KILL needle, TEMP tmp1, TEMP tmp2, KILL cr0, KILL cr1, KILL ctr);
+  // Required for EA: check if it is still a type_array.
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::LL &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop() &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop()->is_type_array());
+  ins_cost(180);
+
+  format %{ "String IndexOf SCL1 $haystack[0..$haycnt], $needle[0..$needlecntImm]"
+            " -> $result \t// KILL $haycnt, $needle, $tmp1, $tmp2, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    Node *ndl = in(operand_index($needle));  // The node that defines needle.
+    ciTypeArray* needle_values = ndl->bottom_type()->is_aryptr()->const_oop()->as_type_array();
+    guarantee(needle_values, "sanity");
+    jchar chr = (jchar)needle_values->element_value(0).as_byte();
+    __ string_indexof_char($result$$Register,
+                           $haystack$$Register, $haycnt$$Register,
+                           R0, chr,
+                           $tmp1$$Register, $tmp2$$Register, true /*is_byte*/);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_imm1_UL(iRegIdst result, iRegPsrc haystack, iRegIsrc haycnt,
+                         rscratch2RegP needle, immI_1 needlecntImm,
+                         iRegIdst tmp1, iRegIdst tmp2,
+                         flagsRegCR0 cr0, flagsRegCR1 cr1, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary needle needlecntImm)));
+  effect(USE_KILL needle, TEMP tmp1, TEMP tmp2, KILL cr0, KILL cr1, KILL ctr);
+  // Required for EA: check if it is still a type_array.
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::UL &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop() &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop()->is_type_array());
+  ins_cost(180);
+
+  format %{ "String IndexOf SCL1 $haystack[0..$haycnt], $needle[0..$needlecntImm]"
+            " -> $result \t// KILL $haycnt, $needle, $tmp1, $tmp2, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    Node *ndl = in(operand_index($needle));  // The node that defines needle.
+    ciTypeArray* needle_values = ndl->bottom_type()->is_aryptr()->const_oop()->as_type_array();
+    guarantee(needle_values, "sanity");
+    jchar chr = (jchar)needle_values->element_value(0).as_byte();
+    __ string_indexof_char($result$$Register,
+                           $haystack$$Register, $haycnt$$Register,
+                           R0, chr,
+                           $tmp1$$Register, $tmp2$$Register, false /*is_byte*/);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOfChar_U(iRegIdst result, iRegPsrc haystack, iRegIsrc haycnt,
+                       iRegIsrc ch, iRegIdst tmp1, iRegIdst tmp2,
+                       flagsRegCR0 cr0, flagsRegCR1 cr1, regCTR ctr) %{
+  match(Set result (StrIndexOfChar (Binary haystack haycnt) ch));
+  effect(TEMP tmp1, TEMP tmp2, KILL cr0, KILL cr1, KILL ctr);
+  predicate(CompactStrings);
+  ins_cost(180);
+
+  format %{ "String IndexOfChar $haystack[0..$haycnt], $ch"
+            " -> $result \t// KILL $haycnt, $tmp1, $tmp2, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ string_indexof_char($result$$Register,
+                           $haystack$$Register, $haycnt$$Register,
+                           $ch$$Register, 0 /* this is not used if the character is already in a register */,
+                           $tmp1$$Register, $tmp2$$Register, false /*is_byte*/);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_imm_U(iRegIdst result, iRegPsrc haystack, rscratch1RegI haycnt,
+                       iRegPsrc needle, uimmI15 needlecntImm,
+                       iRegIdst tmp1, iRegIdst tmp2, iRegIdst tmp3, iRegIdst tmp4, iRegIdst tmp5,
+                       flagsRegCR0 cr0, flagsRegCR1 cr1, flagsRegCR6 cr6, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary needle needlecntImm)));
+  effect(USE_KILL haycnt, /* better: TDEF haycnt, */ TEMP_DEF result,
+         TEMP tmp1, TEMP tmp2, TEMP tmp3, TEMP tmp4, TEMP tmp5, KILL cr0, KILL cr1, KILL cr6, KILL ctr);
+  // Required for EA: check if it is still a type_array.
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::UU &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop() &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop()->is_type_array());
+  ins_cost(250);
+
+  format %{ "String IndexOf SCL $haystack[0..$haycnt], $needle[0..$needlecntImm]"
+            " -> $result \t// KILL $haycnt, $tmp1, $tmp2, $tmp3, $tmp4, $tmp5, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    Node *ndl = in(operand_index($needle));  // The node that defines needle.
+    ciTypeArray* needle_values = ndl->bottom_type()->is_aryptr()->const_oop()->as_type_array();
+
+    __ string_indexof($result$$Register,
+                      $haystack$$Register, $haycnt$$Register,
+                      $needle$$Register, needle_values, $tmp5$$Register, $needlecntImm$$constant,
+                      $tmp1$$Register, $tmp2$$Register, $tmp3$$Register, $tmp4$$Register, StrIntrinsicNode::UU);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_imm_L(iRegIdst result, iRegPsrc haystack, rscratch1RegI haycnt,
+                       iRegPsrc needle, uimmI15 needlecntImm,
+                       iRegIdst tmp1, iRegIdst tmp2, iRegIdst tmp3, iRegIdst tmp4, iRegIdst tmp5,
+                       flagsRegCR0 cr0, flagsRegCR1 cr1, flagsRegCR6 cr6, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary needle needlecntImm)));
+  effect(USE_KILL haycnt, /* better: TDEF haycnt, */ TEMP_DEF result,
+         TEMP tmp1, TEMP tmp2, TEMP tmp3, TEMP tmp4, TEMP tmp5, KILL cr0, KILL cr1, KILL cr6, KILL ctr);
+  // Required for EA: check if it is still a type_array.
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::LL &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop() &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop()->is_type_array());
+  ins_cost(250);
+
+  format %{ "String IndexOf SCL $haystack[0..$haycnt], $needle[0..$needlecntImm]"
+            " -> $result \t// KILL $haycnt, $tmp1, $tmp2, $tmp3, $tmp4, $tmp5, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    Node *ndl = in(operand_index($needle));  // The node that defines needle.
+    ciTypeArray* needle_values = ndl->bottom_type()->is_aryptr()->const_oop()->as_type_array();
+
+    __ string_indexof($result$$Register,
+                      $haystack$$Register, $haycnt$$Register,
+                      $needle$$Register, needle_values, $tmp5$$Register, $needlecntImm$$constant,
+                      $tmp1$$Register, $tmp2$$Register, $tmp3$$Register, $tmp4$$Register, StrIntrinsicNode::LL);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_imm_UL(iRegIdst result, iRegPsrc haystack, rscratch1RegI haycnt,
+                        iRegPsrc needle, uimmI15 needlecntImm,
+                        iRegIdst tmp1, iRegIdst tmp2, iRegIdst tmp3, iRegIdst tmp4, iRegIdst tmp5,
+                        flagsRegCR0 cr0, flagsRegCR1 cr1, flagsRegCR6 cr6, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary needle needlecntImm)));
+  effect(USE_KILL haycnt, /* better: TDEF haycnt, */ TEMP_DEF result,
+         TEMP tmp1, TEMP tmp2, TEMP tmp3, TEMP tmp4, TEMP tmp5, KILL cr0, KILL cr1, KILL cr6, KILL ctr);
+  // Required for EA: check if it is still a type_array.
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::UL &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop() &&
+            n->in(3)->in(1)->bottom_type()->is_aryptr()->const_oop()->is_type_array());
+  ins_cost(250);
+
+  format %{ "String IndexOf SCL $haystack[0..$haycnt], $needle[0..$needlecntImm]"
+            " -> $result \t// KILL $haycnt, $tmp1, $tmp2, $tmp3, $tmp4, $tmp5, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    Node *ndl = in(operand_index($needle));  // The node that defines needle.
+    ciTypeArray* needle_values = ndl->bottom_type()->is_aryptr()->const_oop()->as_type_array();
+
+    __ string_indexof($result$$Register,
+                      $haystack$$Register, $haycnt$$Register,
+                      $needle$$Register, needle_values, $tmp5$$Register, $needlecntImm$$constant,
+                      $tmp1$$Register, $tmp2$$Register, $tmp3$$Register, $tmp4$$Register, StrIntrinsicNode::UL);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_U(iRegIdst result, iRegPsrc haystack, rscratch1RegI haycnt, iRegPsrc needle, rscratch2RegI needlecnt,
+                   iRegLdst tmp1, iRegLdst tmp2, iRegLdst tmp3, iRegLdst tmp4,
+                   flagsRegCR0 cr0, flagsRegCR1 cr1, flagsRegCR6 cr6, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary needle needlecnt)));
+  effect(USE_KILL haycnt, USE_KILL needlecnt, /*better: TDEF haycnt, TDEF needlecnt,*/
+         TEMP_DEF result,
+         TEMP tmp1, TEMP tmp2, TEMP tmp3, TEMP tmp4, KILL cr0, KILL cr1, KILL cr6, KILL ctr);
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::UU);
+  ins_cost(300);
+
+  format %{ "String IndexOf $haystack[0..$haycnt], $needle[0..$needlecnt]"
+             " -> $result \t// KILL $haycnt, $needlecnt, $tmp1, $tmp2, $tmp3, $tmp4, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ string_indexof($result$$Register,
+                      $haystack$$Register, $haycnt$$Register,
+                      $needle$$Register, NULL, $needlecnt$$Register, 0,  // needlecnt not constant.
+                      $tmp1$$Register, $tmp2$$Register, $tmp3$$Register, $tmp4$$Register, StrIntrinsicNode::UU);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_L(iRegIdst result, iRegPsrc haystack, rscratch1RegI haycnt, iRegPsrc needle, rscratch2RegI needlecnt,
+                   iRegLdst tmp1, iRegLdst tmp2, iRegLdst tmp3, iRegLdst tmp4,
+                   flagsRegCR0 cr0, flagsRegCR1 cr1, flagsRegCR6 cr6, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary needle needlecnt)));
+  effect(USE_KILL haycnt, USE_KILL needlecnt, /*better: TDEF haycnt, TDEF needlecnt,*/
+         TEMP_DEF result,
+         TEMP tmp1, TEMP tmp2, TEMP tmp3, TEMP tmp4, KILL cr0, KILL cr1, KILL cr6, KILL ctr);
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::LL);
+  ins_cost(300);
+
+  format %{ "String IndexOf $haystack[0..$haycnt], $needle[0..$needlecnt]"
+             " -> $result \t// KILL $haycnt, $needlecnt, $tmp1, $tmp2, $tmp3, $tmp4, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ string_indexof($result$$Register,
+                      $haystack$$Register, $haycnt$$Register,
+                      $needle$$Register, NULL, $needlecnt$$Register, 0,  // needlecnt not constant.
+                      $tmp1$$Register, $tmp2$$Register, $tmp3$$Register, $tmp4$$Register, StrIntrinsicNode::LL);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+instruct indexOf_UL(iRegIdst result, iRegPsrc haystack, rscratch1RegI haycnt, iRegPsrc needle, rscratch2RegI needlecnt,
+                    iRegLdst tmp1, iRegLdst tmp2, iRegLdst tmp3, iRegLdst tmp4,
+                    flagsRegCR0 cr0, flagsRegCR1 cr1, flagsRegCR6 cr6, regCTR ctr) %{
+  match(Set result (StrIndexOf (Binary haystack haycnt) (Binary needle needlecnt)));
+  effect(USE_KILL haycnt, USE_KILL needlecnt, /*better: TDEF haycnt, TDEF needlecnt,*/
+         TEMP_DEF result,
+         TEMP tmp1, TEMP tmp2, TEMP tmp3, TEMP tmp4, KILL cr0, KILL cr1, KILL cr6, KILL ctr);
+  predicate(((StrIndexOfNode*)n)->encoding() == StrIntrinsicNode::UL);
+  ins_cost(300);
+
+  format %{ "String IndexOf $haystack[0..$haycnt], $needle[0..$needlecnt]"
+             " -> $result \t// KILL $haycnt, $needlecnt, $tmp1, $tmp2, $tmp3, $tmp4, $cr0, $cr1" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ string_indexof($result$$Register,
+                      $haystack$$Register, $haycnt$$Register,
+                      $needle$$Register, NULL, $needlecnt$$Register, 0,  // needlecnt not constant.
+                      $tmp1$$Register, $tmp2$$Register, $tmp3$$Register, $tmp4$$Register, StrIntrinsicNode::UL);
+  %}
+  ins_pipe(pipe_class_compare);
+%}
+
+// char[] to byte[] compression
+instruct string_compress(rarg1RegP src, rarg2RegP dst, iRegIsrc len, iRegIdst result, iRegLdst tmp1,
+                         iRegLdst tmp2, iRegLdst tmp3, iRegLdst tmp4, iRegLdst tmp5, regCTR ctr, flagsRegCR0 cr0) %{
+  match(Set result (StrCompressedCopy src (Binary dst len)));
+  effect(TEMP_DEF result, TEMP tmp1, TEMP tmp2, TEMP tmp3, TEMP tmp4, TEMP tmp5,
+         USE_KILL src, USE_KILL dst, KILL ctr, KILL cr0);
+  ins_cost(300);
+  format %{ "String Compress $src,$dst,$len -> $result \t// KILL $tmp1, $tmp2, $tmp3, $tmp4, $tmp5" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    Label Lskip, Ldone;
+    __ li($result$$Register, 0);
+    __ string_compress_16($src$$Register, $dst$$Register, $len$$Register, $tmp1$$Register,
+                          $tmp2$$Register, $tmp3$$Register, $tmp4$$Register, $tmp5$$Register, Ldone);
+    __ rldicl_($tmp1$$Register, $len$$Register, 0, 64-3); // Remaining characters.
+    __ beq(CCR0, Lskip);
+    __ string_compress($src$$Register, $dst$$Register, $tmp1$$Register, $tmp2$$Register, Ldone);
+    __ bind(Lskip);
+    __ mr($result$$Register, $len$$Register);
+    __ bind(Ldone);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+// byte[] to char[] inflation
+instruct string_inflate(Universe dummy, rarg1RegP src, rarg2RegP dst, iRegIsrc len, iRegLdst tmp1,
+                        iRegLdst tmp2, iRegLdst tmp3, iRegLdst tmp4, iRegLdst tmp5, regCTR ctr, flagsRegCR0 cr0) %{
+  match(Set dummy (StrInflatedCopy src (Binary dst len)));
+  effect(TEMP tmp1, TEMP tmp2, TEMP tmp3, TEMP tmp4, TEMP tmp5, USE_KILL src, USE_KILL dst, KILL ctr, KILL cr0);
+  ins_cost(300);
+  format %{ "String Inflate $src,$dst,$len \t// KILL $tmp1, $tmp2, $tmp3, $tmp4, $tmp5" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    Label Ldone;
+    __ string_inflate_16($src$$Register, $dst$$Register, $len$$Register, $tmp1$$Register,
+                         $tmp2$$Register, $tmp3$$Register, $tmp4$$Register, $tmp5$$Register);
+    __ rldicl_($tmp1$$Register, $len$$Register, 0, 64-3); // Remaining characters.
+    __ beq(CCR0, Ldone);
+    __ string_inflate($src$$Register, $dst$$Register, $tmp1$$Register, $tmp2$$Register);
+    __ bind(Ldone);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+// StringCoding.java intrinsics
+instruct has_negatives(rarg1RegP ary1, iRegIsrc len, iRegIdst result, iRegLdst tmp1, iRegLdst tmp2,
+                       regCTR ctr, flagsRegCR0 cr0)
+%{
+  match(Set result (HasNegatives ary1 len));
+  effect(TEMP_DEF result, USE_KILL ary1, TEMP tmp1, TEMP tmp2, KILL ctr, KILL cr0);
+  ins_cost(300);
+  format %{ "has negatives byte[] $ary1,$len -> $result \t// KILL $tmp1, $tmp2" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ has_negatives($ary1$$Register, $len$$Register, $result$$Register,
+                     $tmp1$$Register, $tmp2$$Register);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+// encode char[] to byte[] in ISO_8859_1
+instruct encode_iso_array(rarg1RegP src, rarg2RegP dst, iRegIsrc len, iRegIdst result, iRegLdst tmp1,
+                          iRegLdst tmp2, iRegLdst tmp3, iRegLdst tmp4, iRegLdst tmp5, regCTR ctr, flagsRegCR0 cr0) %{
+  match(Set result (EncodeISOArray src (Binary dst len)));
+  effect(TEMP_DEF result, TEMP tmp1, TEMP tmp2, TEMP tmp3, TEMP tmp4, TEMP tmp5,
+         USE_KILL src, USE_KILL dst, KILL ctr, KILL cr0);
+  ins_cost(300);
+  format %{ "Encode array $src,$dst,$len -> $result \t// KILL $tmp1, $tmp2, $tmp3, $tmp4, $tmp5" %}
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    Label Lslow, Lfailure1, Lfailure2, Ldone;
+    __ string_compress_16($src$$Register, $dst$$Register, $len$$Register, $tmp1$$Register,
+                          $tmp2$$Register, $tmp3$$Register, $tmp4$$Register, $tmp5$$Register, Lfailure1);
+    __ rldicl_($result$$Register, $len$$Register, 0, 64-3); // Remaining characters.
+    __ beq(CCR0, Ldone);
+    __ bind(Lslow);
+    __ string_compress($src$$Register, $dst$$Register, $result$$Register, $tmp2$$Register, Lfailure2);
+    __ li($result$$Register, 0);
+    __ b(Ldone);
+
+    __ bind(Lfailure1);
+    __ mr($result$$Register, $len$$Register);
+    __ mfctr($tmp1$$Register);
+    __ rldimi_($result$$Register, $tmp1$$Register, 3, 0); // Remaining characters.
+    __ beq(CCR0, Ldone);
+    __ b(Lslow);
+
+    __ bind(Lfailure2);
+    __ mfctr($result$$Register); // Remaining characters.
+
+    __ bind(Ldone);
+    __ subf($result$$Register, $result$$Register, $len$$Register);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
+
 // String_IndexOf for needle of length 1.
 //
 // Match needle into immediate operands: no loadConP node needed. Saves one
@@ -11060,11 +11638,11 @@ instruct string_indexOf_imm1_char(iRegIdst result, iRegPsrc haystack, iRegIsrc h
     if (java_lang_String::has_coder_field()) {
       // New compact strings byte array strings
 #ifdef VM_LITTLE_ENDIAN
-      chr = (((jchar)needle_values->element_value(1).as_byte()) << 8) |
+    chr = (((jchar)(unsigned char)needle_values->element_value(1).as_byte()) << 8) |
-              (jchar)needle_values->element_value(0).as_byte();
+           ((jchar)(unsigned char)needle_values->element_value(0).as_byte());
 #else
-      chr = (((jchar)needle_values->element_value(0).as_byte()) << 8) |
+    chr = (((jchar)(unsigned char)needle_values->element_value(0).as_byte()) << 8) |
-              (jchar)needle_values->element_value(1).as_byte();
+           ((jchar)(unsigned char)needle_values->element_value(1).as_byte());
 #endif
     } else {
       // Old char array strings
@@ -11115,11 +11693,11 @@ instruct string_indexOf_imm1(iRegIdst result, iRegPsrc haystack, iRegIsrc haycnt
     if (java_lang_String::has_coder_field()) {
       // New compact strings byte array strings
 #ifdef VM_LITTLE_ENDIAN
-      chr = (((jchar)needle_values->element_value(1).as_byte()) << 8) |
+    chr = (((jchar)(unsigned char)needle_values->element_value(1).as_byte()) << 8) |
-              (jchar)needle_values->element_value(0).as_byte();
+           ((jchar)(unsigned char)needle_values->element_value(0).as_byte());
 #else
-      chr = (((jchar)needle_values->element_value(0).as_byte()) << 8) |
+    chr = (((jchar)(unsigned char)needle_values->element_value(0).as_byte()) << 8) |
-              (jchar)needle_values->element_value(1).as_byte();
+           ((jchar)(unsigned char)needle_values->element_value(1).as_byte());
 #endif
     } else {
       // Old char array strings
@@ -11321,6 +11899,20 @@ instruct minI_reg_reg_Ex(iRegIdst dst, iRegIsrc src1, iRegIsrc src2) %{
   %}
 %}
 
+instruct minI_reg_reg_isel(iRegIdst dst, iRegIsrc src1, iRegIsrc src2, flagsRegCR0 cr0) %{
+  match(Set dst (MinI src1 src2));
+  effect(KILL cr0);
+  predicate(VM_Version::has_isel());
+  ins_cost(DEFAULT_COST*2);
+
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ cmpw(CCR0, $src1$$Register, $src2$$Register);
+    __ isel($dst$$Register, CCR0, Assembler::less, /*invert*/false, $src1$$Register, $src2$$Register);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
 instruct maxI_reg_reg_Ex(iRegIdst dst, iRegIsrc src1, iRegIsrc src2) %{
   match(Set dst (MaxI src1 src2));
   ins_cost(DEFAULT_COST*6);
@@ -11341,6 +11933,20 @@ instruct maxI_reg_reg_Ex(iRegIdst dst, iRegIsrc src1, iRegIsrc src2) %{
   %}
 %}
 
+instruct maxI_reg_reg_isel(iRegIdst dst, iRegIsrc src1, iRegIsrc src2, flagsRegCR0 cr0) %{
+  match(Set dst (MaxI src1 src2));
+  effect(KILL cr0);
+  predicate(VM_Version::has_isel());
+  ins_cost(DEFAULT_COST*2);
+
+  ins_encode %{
+    // TODO: PPC port $archOpcode(ppc64Opcode_compound);
+    __ cmpw(CCR0, $src1$$Register, $src2$$Register);
+    __ isel($dst$$Register, CCR0, Assembler::greater, /*invert*/false, $src1$$Register, $src2$$Register);
+  %}
+  ins_pipe(pipe_class_default);
+%}
+
 //---------- Population Count Instructions ------------------------------------
 
 // Popcnt for Power7.

hotspot/src/cpu/ppc/vm/stubGenerator_ppc.cpp

@@ -2609,9 +2609,7 @@ class StubGenerator: public StubCodeGenerator {
    *   R5_ARG3    - int   length (of buffer)
    *
    * scratch:
-   *   R6_ARG4    - crc table address
+   *   R2, R6-R12
-   *   R7_ARG5    - tmp1
-   *   R8_ARG6    - tmp2
    *
    * Ouput:
    *   R3_RET     - int   crc result
@@ -2623,22 +2621,25 @@ class StubGenerator: public StubCodeGenerator {
     address start = __ function_entry();  // Remember stub start address (is rtn value).
 
     // arguments to kernel_crc32:
-    Register       crc     = R3_ARG1;  // Current checksum, preset by caller or result from previous call.
+    const Register crc     = R3_ARG1;  // Current checksum, preset by caller or result from previous call.
-    Register       data    = R4_ARG2;  // source byte array
+    const Register data    = R4_ARG2;  // source byte array
-    Register       dataLen = R5_ARG3;  // #bytes to process
+    const Register dataLen = R5_ARG3;  // #bytes to process
-    Register       table   = R6_ARG4;  // crc table address
+    const Register table   = R6_ARG4;  // crc table address
 
-    Register       t0      = R9;       // work reg for kernel* emitters
+    const Register t0      = R2;
-    Register       t1      = R10;      // work reg for kernel* emitters
+    const Register t1      = R7;
-    Register       t2      = R11;      // work reg for kernel* emitters
+    const Register t2      = R8;
-    Register       t3      = R12;      // work reg for kernel* emitters
+    const Register t3      = R9;
+    const Register tc0     = R10;
+    const Register tc1     = R11;
+    const Register tc2     = R12;
 
     BLOCK_COMMENT("Stub body {");
     assert_different_registers(crc, data, dataLen, table);
 
     StubRoutines::ppc64::generate_load_crc_table_addr(_masm, table);
 
-    __ kernel_crc32_1byte(crc, data, dataLen, table, t0, t1, t2, t3);
+    __ kernel_crc32_1word(crc, data, dataLen, table, t0, t1, t2, t3, tc0, tc1, tc2, table);
 
     BLOCK_COMMENT("return");
     __ mr_if_needed(R3_RET, crc);      // Updated crc is function result. No copying required (R3_ARG1 == R3_RET).

hotspot/src/cpu/ppc/vm/vm_version_ppc.cpp

@@ -1,6 +1,6 @@
 /*
- * Copyright (c) 1997, 2015, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2016, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2015 SAP SE. All rights reserved.
+ * Copyright (c) 2012, 2016 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -53,7 +53,7 @@ void VM_Version::initialize() {
 
   // If PowerArchitecturePPC64 hasn't been specified explicitly determine from features.
   if (FLAG_IS_DEFAULT(PowerArchitecturePPC64)) {
-    if (VM_Version::has_tcheck() && VM_Version::has_lqarx()) {
+    if (VM_Version::has_lqarx()) {
       FLAG_SET_ERGO(uintx, PowerArchitecturePPC64, 8);
     } else if (VM_Version::has_popcntw()) {
       FLAG_SET_ERGO(uintx, PowerArchitecturePPC64, 7);
@@ -68,8 +68,7 @@ void VM_Version::initialize() {
 
   bool PowerArchitecturePPC64_ok = false;
   switch (PowerArchitecturePPC64) {
-    case 8: if (!VM_Version::has_tcheck() ) break;
+    case 8: if (!VM_Version::has_lqarx()  ) break;
-            if (!VM_Version::has_lqarx()  ) break;
     case 7: if (!VM_Version::has_popcntw()) break;
     case 6: if (!VM_Version::has_cmpb()   ) break;
     case 5: if (!VM_Version::has_popcntb()) break;
@@ -80,7 +79,7 @@ void VM_Version::initialize() {
             UINTX_FORMAT " on this machine", PowerArchitecturePPC64);
 
   // Power 8: Configure Data Stream Control Register.
-  if (PowerArchitecturePPC64 >= 8) {
+  if (has_mfdscr()) {
     config_dscr();
   }
 
@@ -112,7 +111,7 @@ void VM_Version::initialize() {
   // Create and print feature-string.
   char buf[(num_features+1) * 16]; // Max 16 chars per feature.
   jio_snprintf(buf, sizeof(buf),
-               "ppc64%s%s%s%s%s%s%s%s%s%s%s%s",
+               "ppc64%s%s%s%s%s%s%s%s%s%s%s%s%s",
                (has_fsqrt()   ? " fsqrt"   : ""),
                (has_isel()    ? " isel"    : ""),
                (has_lxarxeh() ? " lxarxeh" : ""),
@@ -125,7 +124,8 @@ void VM_Version::initialize() {
                (has_lqarx()   ? " lqarx"   : ""),
                (has_vcipher() ? " vcipher" : ""),
                (has_vpmsumb() ? " vpmsumb" : ""),
-               (has_tcheck()  ? " tcheck"  : "")
+               (has_tcheck()  ? " tcheck"  : ""),
+               (has_mfdscr()  ? " mfdscr"  : "")
                // Make sure number of %s matches num_features!
               );
   _features_string = os::strdup(buf);
@@ -610,6 +610,7 @@ void VM_Version::determine_features() {
   a->vcipher(VR0, VR1, VR2);                   // code[10] -> vcipher
   a->vpmsumb(VR0, VR1, VR2);                   // code[11] -> vpmsumb
   a->tcheck(0);                                // code[12] -> tcheck
+  a->mfdscr(R0);                               // code[13] -> mfdscr
   a->blr();
 
   // Emit function to set one cache line to zero. Emit function descriptor and get pointer to it.
@@ -657,6 +658,7 @@ void VM_Version::determine_features() {
   if (code[feature_cntr++]) features |= vcipher_m;
   if (code[feature_cntr++]) features |= vpmsumb_m;
   if (code[feature_cntr++]) features |= tcheck_m;
+  if (code[feature_cntr++]) features |= mfdscr_m;
 
   // Print the detection code.
   if (PrintAssembly) {
@@ -670,8 +672,6 @@ void VM_Version::determine_features() {
 
 // Power 8: Configure Data Stream Control Register.
 void VM_Version::config_dscr() {
-  assert(has_tcheck(), "Only execute on Power 8 or later!");
-
   // 7 InstWords for each call (function descriptor + blr instruction).
   const int code_size = (2+2*7)*BytesPerInstWord;
 

hotspot/src/cpu/ppc/vm/vm_version_ppc.hpp

@@ -1,6 +1,6 @@
 /*
- * Copyright (c) 1997, 2015, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2016, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2015 SAP SE. All rights reserved.
+ * Copyright (c) 2012, 2016 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -45,6 +45,7 @@ protected:
     vcipher,
     vpmsumb,
     tcheck,
+    mfdscr,
     num_features // last entry to count features
   };
   enum Feature_Flag_Set {
@@ -62,6 +63,7 @@ protected:
     vcipher_m             = (1 << vcipher),
     vpmsumb_m             = (1 << vpmsumb),
     tcheck_m              = (1 << tcheck ),
+    mfdscr_m              = (1 << mfdscr ),
     all_features_m        = (unsigned long)-1
   };
 
@@ -94,6 +96,7 @@ public:
   static bool has_vcipher() { return (_features & vcipher_m) != 0; }
   static bool has_vpmsumb() { return (_features & vpmsumb_m) != 0; }
   static bool has_tcheck()  { return (_features & tcheck_m) != 0; }
+  static bool has_mfdscr()  { return (_features & mfdscr_m) != 0; }
 
   // Assembler testing
   static void allow_all();

hotspot/src/cpu/sparc/vm/sharedRuntime_sparc.cpp

@@ -1349,9 +1349,12 @@ static void move32_64(MacroAssembler* masm, VMRegPair src, VMRegPair dst) {
     }
   } else if (dst.first()->is_stack()) {
     // reg to stack
-    __ st_ptr(src.first()->as_Register(), SP, reg2offset(dst.first()) + STACK_BIAS);
+    // Some compilers (gcc) expect a clean 32 bit value on function entry
+    __ signx(src.first()->as_Register(), L5);
+    __ st_ptr(L5, SP, reg2offset(dst.first()) + STACK_BIAS);
   } else {
-    __ mov(src.first()->as_Register(), dst.first()->as_Register());
+    // Some compilers (gcc) expect a clean 32 bit value on function entry
+    __ signx(src.first()->as_Register(), dst.first()->as_Register());
   }
 }
 

hotspot/src/cpu/sparc/vm/sparc.ad

@@ -948,28 +948,28 @@ void emit_form3_mem_reg(CodeBuffer &cbuf, PhaseRegAlloc* ra, const MachNode* n,
   }
 #endif
 
-  uint instr;
+  uint instr = (Assembler::ldst_op << 30)
-  instr = (Assembler::ldst_op << 30)
+             | (dst_enc        << 25)
-        | (dst_enc        << 25)
+             | (primary        << 19)
-        | (primary        << 19)
+             | (src1_enc       << 14);
-        | (src1_enc       << 14);
 
   uint index = src2_enc;
   int disp = disp32;
 
   if (src1_enc == R_SP_enc || src1_enc == R_FP_enc) {
     disp += STACK_BIAS;
-    // Quick fix for JDK-8029668: check that stack offset fits, bailout if not
+    // Check that stack offset fits, load into O7 if not
     if (!Assembler::is_simm13(disp)) {
-      ra->C->record_method_not_compilable("unable to handle large constant offsets");
+      MacroAssembler _masm(&cbuf);
-      return;
+      __ set(disp, O7);
+      if (index != R_G0_enc) {
+        __ add(O7, reg_to_register_object(index), O7);
+      }
+      index = R_O7_enc;
+      disp = 0;
     }
   }
 
-  // We should have a compiler bailout here rather than a guarantee.
-  // Better yet would be some mechanism to handle variable-size matches correctly.
-  guarantee(Assembler::is_simm13(disp), "Do not match large constant offsets" );
-
   if( disp == 0 ) {
     // use reg-reg form
     // bit 13 is already zero
@@ -983,7 +983,7 @@ void emit_form3_mem_reg(CodeBuffer &cbuf, PhaseRegAlloc* ra, const MachNode* n,
   cbuf.insts()->emit_int32(instr);
 
 #ifdef ASSERT
-  {
+  if (VerifyOops) {
     MacroAssembler _masm(&cbuf);
     if (is_verified_oop_base) {
       __ verify_oop(reg_to_register_object(src1_enc));
@@ -1342,7 +1342,7 @@ int MachEpilogNode::safepoint_offset() const {
 // Figure out which register class each belongs in: rc_int, rc_float, rc_stack
 enum RC { rc_bad, rc_int, rc_float, rc_stack };
 static enum RC rc_class( OptoReg::Name reg ) {
-  if( !OptoReg::is_valid(reg)  ) return rc_bad;
+  if (!OptoReg::is_valid(reg)) return rc_bad;
   if (OptoReg::is_stack(reg)) return rc_stack;
   VMReg r = OptoReg::as_VMReg(reg);
   if (r->is_Register()) return rc_int;
@@ -1350,66 +1350,79 @@ static enum RC rc_class( OptoReg::Name reg ) {
   return rc_float;
 }
 
-static int impl_helper(const MachNode* mach, CodeBuffer* cbuf, PhaseRegAlloc* ra, bool do_size, bool is_load, int offset, int reg, int opcode, const char *op_str, int size, outputStream* st ) {
+#ifndef PRODUCT
+ATTRIBUTE_PRINTF(2, 3)
+static void print_helper(outputStream* st, const char* format, ...) {
+  if (st->position() > 0) {
+    st->cr();
+    st->sp();
+  }
+  va_list ap;
+  va_start(ap, format);
+  st->vprint(format, ap);
+  va_end(ap);
+}
+#endif // !PRODUCT
+
+static void impl_helper(const MachNode* mach, CodeBuffer* cbuf, PhaseRegAlloc* ra, bool is_load, int offset, int reg, int opcode, const char *op_str, outputStream* st) {
   if (cbuf) {
     emit_form3_mem_reg(*cbuf, ra, mach, opcode, -1, R_SP_enc, offset, 0, Matcher::_regEncode[reg]);
   }
 #ifndef PRODUCT
-  else if (!do_size) {
+  else {
-    if (size != 0) st->print("\n\t");
+    if (is_load) {
-    if (is_load) st->print("%s   [R_SP + #%d],R_%s\t! spill",op_str,offset,OptoReg::regname(reg));
+      print_helper(st, "%s   [R_SP + #%d],R_%s\t! spill", op_str, offset, OptoReg::regname(reg));
-    else         st->print("%s   R_%s,[R_SP + #%d]\t! spill",op_str,OptoReg::regname(reg),offset);
+    } else {
+      print_helper(st, "%s   R_%s,[R_SP + #%d]\t! spill", op_str, OptoReg::regname(reg), offset);
+    }
   }
 #endif
-  return size+4;
 }
 
-static int impl_mov_helper( CodeBuffer *cbuf, bool do_size, int src, int dst, int op1, int op2, const char *op_str, int size, outputStream* st ) {
+static void impl_mov_helper(CodeBuffer *cbuf, int src, int dst, int op1, int op2, const char *op_str, outputStream* st) {
-  if( cbuf ) emit3( *cbuf, Assembler::arith_op, Matcher::_regEncode[dst], op1, 0, op2, Matcher::_regEncode[src] );
+  if (cbuf) {
+    emit3(*cbuf, Assembler::arith_op, Matcher::_regEncode[dst], op1, 0, op2, Matcher::_regEncode[src]);
+  }
 #ifndef PRODUCT
-  else if( !do_size ) {
+  else {
-    if( size != 0 ) st->print("\n\t");
+    print_helper(st, "%s  R_%s,R_%s\t! spill", op_str, OptoReg::regname(src), OptoReg::regname(dst));
-    st->print("%s  R_%s,R_%s\t! spill",op_str,OptoReg::regname(src),OptoReg::regname(dst));
   }
 #endif
-  return size+4;
 }
 
-uint MachSpillCopyNode::implementation( CodeBuffer *cbuf,
+static void mach_spill_copy_implementation_helper(const MachNode* mach,
-                                        PhaseRegAlloc *ra_,
+                                                  CodeBuffer *cbuf,
-                                        bool do_size,
+                                                  PhaseRegAlloc *ra_,
-                                        outputStream* st ) const {
+                                                  outputStream* st) {
   // Get registers to move
-  OptoReg::Name src_second = ra_->get_reg_second(in(1));
+  OptoReg::Name src_second = ra_->get_reg_second(mach->in(1));
-  OptoReg::Name src_first = ra_->get_reg_first(in(1));
+  OptoReg::Name src_first  = ra_->get_reg_first(mach->in(1));
-  OptoReg::Name dst_second = ra_->get_reg_second(this );
+  OptoReg::Name dst_second = ra_->get_reg_second(mach);
-  OptoReg::Name dst_first = ra_->get_reg_first(this );
+  OptoReg::Name dst_first  = ra_->get_reg_first(mach);
 
   enum RC src_second_rc = rc_class(src_second);
-  enum RC src_first_rc = rc_class(src_first);
+  enum RC src_first_rc  = rc_class(src_first);
   enum RC dst_second_rc = rc_class(dst_second);
-  enum RC dst_first_rc = rc_class(dst_first);
+  enum RC dst_first_rc  = rc_class(dst_first);
 
-  assert( OptoReg::is_valid(src_first) && OptoReg::is_valid(dst_first), "must move at least 1 register" );
+  assert(OptoReg::is_valid(src_first) && OptoReg::is_valid(dst_first), "must move at least 1 register");
 
-  // Generate spill code!
+  if (src_first == dst_first && src_second == dst_second) {
-  int size = 0;
+    return; // Self copy, no move
-
+  }
-  if( src_first == dst_first && src_second == dst_second )
-    return size;            // Self copy, no move
 
   // --------------------------------------
   // Check for mem-mem move.  Load into unused float registers and fall into
   // the float-store case.
-  if( src_first_rc == rc_stack && dst_first_rc == rc_stack ) {
+  if (src_first_rc == rc_stack && dst_first_rc == rc_stack) {
     int offset = ra_->reg2offset(src_first);
     // Further check for aligned-adjacent pair, so we can use a double load
-    if( (src_first&1)==0 && src_first+1 == src_second ) {
+    if ((src_first&1) == 0 && src_first+1 == src_second) {
       src_second    = OptoReg::Name(R_F31_num);
       src_second_rc = rc_float;
-      size = impl_helper(this,cbuf,ra_,do_size,true,offset,R_F30_num,Assembler::lddf_op3,"LDDF",size, st);
+      impl_helper(mach, cbuf, ra_, true, offset, R_F30_num, Assembler::lddf_op3, "LDDF", st);
     } else {
-      size = impl_helper(this,cbuf,ra_,do_size,true,offset,R_F30_num,Assembler::ldf_op3 ,"LDF ",size, st);
+      impl_helper(mach, cbuf, ra_, true, offset, R_F30_num, Assembler::ldf_op3, "LDF ", st);
     }
     src_first    = OptoReg::Name(R_F30_num);
     src_first_rc = rc_float;
@@ -1417,7 +1430,7 @@ uint MachSpillCopyNode::implementation( CodeBuffer *cbuf,
 
   if( src_second_rc == rc_stack && dst_second_rc == rc_stack ) {
     int offset = ra_->reg2offset(src_second);
-    size = impl_helper(this,cbuf,ra_,do_size,true,offset,R_F31_num,Assembler::ldf_op3,"LDF ",size, st);
+    impl_helper(mach, cbuf, ra_, true, offset, R_F31_num, Assembler::ldf_op3, "LDF ", st);
     src_second    = OptoReg::Name(R_F31_num);
     src_second_rc = rc_float;
   }
@@ -1427,36 +1440,38 @@ uint MachSpillCopyNode::implementation( CodeBuffer *cbuf,
   if (src_first_rc == rc_float && dst_first_rc == rc_int && UseVIS < 3) {
     int offset = frame::register_save_words*wordSize;
     if (cbuf) {
-      emit3_simm13( *cbuf, Assembler::arith_op, R_SP_enc, Assembler::sub_op3, R_SP_enc, 16 );
+      emit3_simm13(*cbuf, Assembler::arith_op, R_SP_enc, Assembler::sub_op3, R_SP_enc, 16);
-      impl_helper(this,cbuf,ra_,do_size,false,offset,src_first,Assembler::stf_op3 ,"STF ",size, st);
+      impl_helper(mach, cbuf, ra_, false, offset, src_first,  Assembler::stf_op3, "STF ", st);
-      impl_helper(this,cbuf,ra_,do_size,true ,offset,dst_first,Assembler::lduw_op3,"LDUW",size, st);
+      impl_helper(mach, cbuf, ra_,  true, offset, dst_first, Assembler::lduw_op3, "LDUW", st);
-      emit3_simm13( *cbuf, Assembler::arith_op, R_SP_enc, Assembler::add_op3, R_SP_enc, 16 );
+      emit3_simm13(*cbuf, Assembler::arith_op, R_SP_enc, Assembler::add_op3, R_SP_enc, 16);
     }
 #ifndef PRODUCT
-    else if (!do_size) {
+    else {
-      if (size != 0) st->print("\n\t");
+      print_helper(st, "SUB    R_SP,16,R_SP");
-      st->print(  "SUB    R_SP,16,R_SP\n");
+      impl_helper(mach, cbuf, ra_, false, offset, src_first,  Assembler::stf_op3, "STF ", st);
-      impl_helper(this,cbuf,ra_,do_size,false,offset,src_first,Assembler::stf_op3 ,"STF ",size, st);
+      impl_helper(mach, cbuf, ra_,  true, offset, dst_first, Assembler::lduw_op3, "LDUW", st);
-      impl_helper(this,cbuf,ra_,do_size,true ,offset,dst_first,Assembler::lduw_op3,"LDUW",size, st);
+      print_helper(st, "ADD    R_SP,16,R_SP");
-      st->print("\tADD    R_SP,16,R_SP\n");
     }
 #endif
-    size += 16;
   }
 
   // Check for float->int copy on T4
   if (src_first_rc == rc_float && dst_first_rc == rc_int && UseVIS >= 3) {
     // Further check for aligned-adjacent pair, so we can use a double move
-    if ((src_first&1)==0 && src_first+1 == src_second && (dst_first&1)==0 && dst_first+1 == dst_second)
+    if ((src_first & 1) == 0 && src_first + 1 == src_second && (dst_first & 1) == 0 && dst_first + 1 == dst_second) {
-      return impl_mov_helper(cbuf,do_size,src_first,dst_first,Assembler::mftoi_op3,Assembler::mdtox_opf,"MOVDTOX",size, st);
+      impl_mov_helper(cbuf, src_first, dst_first, Assembler::mftoi_op3, Assembler::mdtox_opf, "MOVDTOX", st);
-    size  =  impl_mov_helper(cbuf,do_size,src_first,dst_first,Assembler::mftoi_op3,Assembler::mstouw_opf,"MOVSTOUW",size, st);
+      return;
+    }
+    impl_mov_helper(cbuf, src_first, dst_first, Assembler::mftoi_op3, Assembler::mstouw_opf, "MOVSTOUW", st);
   }
   // Check for int->float copy on T4
   if (src_first_rc == rc_int && dst_first_rc == rc_float && UseVIS >= 3) {
     // Further check for aligned-adjacent pair, so we can use a double move
-    if ((src_first&1)==0 && src_first+1 == src_second && (dst_first&1)==0 && dst_first+1 == dst_second)
+    if ((src_first & 1) == 0 && src_first + 1 == src_second && (dst_first & 1) == 0 && dst_first + 1 == dst_second) {
-      return impl_mov_helper(cbuf,do_size,src_first,dst_first,Assembler::mftoi_op3,Assembler::mxtod_opf,"MOVXTOD",size, st);
+      impl_mov_helper(cbuf, src_first, dst_first, Assembler::mftoi_op3, Assembler::mxtod_opf, "MOVXTOD", st);
-    size  =  impl_mov_helper(cbuf,do_size,src_first,dst_first,Assembler::mftoi_op3,Assembler::mwtos_opf,"MOVWTOS",size, st);
+      return;
+    }
+    impl_mov_helper(cbuf, src_first, dst_first, Assembler::mftoi_op3, Assembler::mwtos_opf, "MOVWTOS", st);
   }
 
   // --------------------------------------
@@ -1466,10 +1481,10 @@ uint MachSpillCopyNode::implementation( CodeBuffer *cbuf,
   // there.  Misaligned sources only come from native-long-returns (handled
   // special below).
 #ifndef _LP64
-  if( src_first_rc == rc_int &&     // source is already big-endian
+  if (src_first_rc == rc_int &&     // source is already big-endian
       src_second_rc != rc_bad &&    // 64-bit move
-      ((dst_first&1)!=0 || dst_second != dst_first+1) ) { // misaligned dst
+      ((dst_first & 1) != 0 || dst_second != dst_first + 1)) { // misaligned dst
-    assert( (src_first&1)==0 && src_second == src_first+1, "source must be aligned" );
+    assert((src_first & 1) == 0 && src_second == src_first + 1, "source must be aligned");
     // Do the big-endian flop.
     OptoReg::Name tmp    = dst_first   ; dst_first    = dst_second   ; dst_second    = tmp   ;
     enum RC       tmp_rc = dst_first_rc; dst_first_rc = dst_second_rc; dst_second_rc = tmp_rc;
@@ -1478,30 +1493,28 @@ uint MachSpillCopyNode::implementation( CodeBuffer *cbuf,
 
   // --------------------------------------
   // Check for integer reg-reg copy
-  if( src_first_rc == rc_int && dst_first_rc == rc_int ) {
+  if (src_first_rc == rc_int && dst_first_rc == rc_int) {
 #ifndef _LP64
-    if( src_first == R_O0_num && src_second == R_O1_num ) {  // Check for the evil O0/O1 native long-return case
+    if (src_first == R_O0_num && src_second == R_O1_num) {  // Check for the evil O0/O1 native long-return case
       // Note: The _first and _second suffixes refer to the addresses of the the 2 halves of the 64-bit value
       //       as stored in memory.  On a big-endian machine like SPARC, this means that the _second
       //       operand contains the least significant word of the 64-bit value and vice versa.
       OptoReg::Name tmp = OptoReg::Name(R_O7_num);
-      assert( (dst_first&1)==0 && dst_second == dst_first+1, "return a native O0/O1 long to an aligned-adjacent 64-bit reg" );
+      assert((dst_first & 1) == 0 && dst_second == dst_first + 1, "return a native O0/O1 long to an aligned-adjacent 64-bit reg" );
       // Shift O0 left in-place, zero-extend O1, then OR them into the dst
-      if( cbuf ) {
+      if ( cbuf ) {
-        emit3_simm13( *cbuf, Assembler::arith_op, Matcher::_regEncode[tmp], Assembler::sllx_op3, Matcher::_regEncode[src_first], 0x1020 );
+        emit3_simm13(*cbuf, Assembler::arith_op, Matcher::_regEncode[tmp], Assembler::sllx_op3, Matcher::_regEncode[src_first], 0x1020);
-        emit3_simm13( *cbuf, Assembler::arith_op, Matcher::_regEncode[src_second], Assembler::srl_op3, Matcher::_regEncode[src_second], 0x0000 );
+        emit3_simm13(*cbuf, Assembler::arith_op, Matcher::_regEncode[src_second], Assembler::srl_op3, Matcher::_regEncode[src_second], 0x0000);
-        emit3       ( *cbuf, Assembler::arith_op, Matcher::_regEncode[dst_first], Assembler:: or_op3, Matcher::_regEncode[tmp], 0, Matcher::_regEncode[src_second] );
+        emit3       (*cbuf, Assembler::arith_op, Matcher::_regEncode[dst_first], Assembler:: or_op3, Matcher::_regEncode[tmp], 0, Matcher::_regEncode[src_second]);
 #ifndef PRODUCT
-      } else if( !do_size ) {
+      } else {
-        if( size != 0 ) st->print("\n\t");
+        print_helper(st, "SLLX   R_%s,32,R_%s\t! Move O0-first to O7-high\n\t", OptoReg::regname(src_first), OptoReg::regname(tmp));
-        st->print("SLLX   R_%s,32,R_%s\t! Move O0-first to O7-high\n\t", OptoReg::regname(src_first), OptoReg::regname(tmp));
+        print_helper(st, "SRL    R_%s, 0,R_%s\t! Zero-extend O1\n\t", OptoReg::regname(src_second), OptoReg::regname(src_second));
-        st->print("SRL    R_%s, 0,R_%s\t! Zero-extend O1\n\t", OptoReg::regname(src_second), OptoReg::regname(src_second));
+        print_helper(st, "OR     R_%s,R_%s,R_%s\t! spill",OptoReg::regname(tmp), OptoReg::regname(src_second), OptoReg::regname(dst_first));
-        st->print("OR     R_%s,R_%s,R_%s\t! spill",OptoReg::regname(tmp), OptoReg::regname(src_second), OptoReg::regname(dst_first));
 #endif
       }
-      return size+12;
+      return;
-    }
+    } else if (dst_first == R_I0_num && dst_second == R_I1_num) {
-    else if( dst_first == R_I0_num && dst_second == R_I1_num ) {
       // returning a long value in I0/I1
       // a SpillCopy must be able to target a return instruction's reg_class
       // Note: The _first and _second suffixes refer to the addresses of the the 2 halves of the 64-bit value
@@ -1511,27 +1524,25 @@ uint MachSpillCopyNode::implementation( CodeBuffer *cbuf,
 
       if (src_first == dst_first) {
         tdest = OptoReg::Name(R_O7_num);
-        size += 4;
       }
 
-      if( cbuf ) {
+      if (cbuf) {
-        assert( (src_first&1) == 0 && (src_first+1) == src_second, "return value was in an aligned-adjacent 64-bit reg");
+        assert((src_first & 1) == 0 && (src_first + 1) == src_second, "return value was in an aligned-adjacent 64-bit reg");
         // Shift value in upper 32-bits of src to lower 32-bits of I0; move lower 32-bits to I1
         // ShrL_reg_imm6
-        emit3_simm13( *cbuf, Assembler::arith_op, Matcher::_regEncode[tdest], Assembler::srlx_op3, Matcher::_regEncode[src_second], 32 | 0x1000 );
+        emit3_simm13(*cbuf, Assembler::arith_op, Matcher::_regEncode[tdest], Assembler::srlx_op3, Matcher::_regEncode[src_second], 32 | 0x1000);
         // ShrR_reg_imm6  src, 0, dst
-        emit3_simm13( *cbuf, Assembler::arith_op, Matcher::_regEncode[dst_second], Assembler::srl_op3, Matcher::_regEncode[src_first], 0x0000 );
+        emit3_simm13(*cbuf, Assembler::arith_op, Matcher::_regEncode[dst_second], Assembler::srl_op3, Matcher::_regEncode[src_first], 0x0000);
         if (tdest != dst_first) {
-          emit3     ( *cbuf, Assembler::arith_op, Matcher::_regEncode[dst_first], Assembler::or_op3, 0/*G0*/, 0/*op2*/, Matcher::_regEncode[tdest] );
+          emit3     (*cbuf, Assembler::arith_op, Matcher::_regEncode[dst_first], Assembler::or_op3, 0/*G0*/, 0/*op2*/, Matcher::_regEncode[tdest]);
         }
       }
 #ifndef PRODUCT
-      else if( !do_size ) {
+      else {
-        if( size != 0 ) st->print("\n\t");  // %%%%% !!!!!
+        print_helper(st, "SRLX   R_%s,32,R_%s\t! Extract MSW\n\t",OptoReg::regname(src_second),OptoReg::regname(tdest));
-        st->print("SRLX   R_%s,32,R_%s\t! Extract MSW\n\t",OptoReg::regname(src_second),OptoReg::regname(tdest));
+        print_helper(st, "SRL    R_%s, 0,R_%s\t! Extract LSW\n\t",OptoReg::regname(src_first),OptoReg::regname(dst_second));
-        st->print("SRL    R_%s, 0,R_%s\t! Extract LSW\n\t",OptoReg::regname(src_first),OptoReg::regname(dst_second));
         if (tdest != dst_first) {
-          st->print("MOV    R_%s,R_%s\t! spill\n\t", OptoReg::regname(tdest), OptoReg::regname(dst_first));
+          print_helper(st, "MOV    R_%s,R_%s\t! spill\n\t", OptoReg::regname(tdest), OptoReg::regname(dst_first));
         }
       }
 #endif // PRODUCT
@@ -1539,65 +1550,77 @@ uint MachSpillCopyNode::implementation( CodeBuffer *cbuf,
     }
 #endif // !_LP64
     // Else normal reg-reg copy
-    assert( src_second != dst_first, "smashed second before evacuating it" );
+    assert(src_second != dst_first, "smashed second before evacuating it");
-    size = impl_mov_helper(cbuf,do_size,src_first,dst_first,Assembler::or_op3,0,"MOV  ",size, st);
+    impl_mov_helper(cbuf, src_first, dst_first, Assembler::or_op3, 0, "MOV  ", st);
-    assert( (src_first&1) == 0 && (dst_first&1) == 0, "never move second-halves of int registers" );
+    assert((src_first & 1) == 0 && (dst_first & 1) == 0, "never move second-halves of int registers");
     // This moves an aligned adjacent pair.
     // See if we are done.
-    if( src_first+1 == src_second && dst_first+1 == dst_second )
+    if (src_first + 1 == src_second && dst_first + 1 == dst_second) {
-      return size;
+      return;
+    }
   }
 
   // Check for integer store
-  if( src_first_rc == rc_int && dst_first_rc == rc_stack ) {
+  if (src_first_rc == rc_int && dst_first_rc == rc_stack) {
     int offset = ra_->reg2offset(dst_first);
     // Further check for aligned-adjacent pair, so we can use a double store
-    if( (src_first&1)==0 && src_first+1 == src_second && (dst_first&1)==0 && dst_first+1 == dst_second )
+    if ((src_first & 1) == 0 && src_first + 1 == src_second && (dst_first & 1) == 0 && dst_first + 1 == dst_second) {
-      return impl_helper(this,cbuf,ra_,do_size,false,offset,src_first,Assembler::stx_op3,"STX ",size, st);
+      impl_helper(mach, cbuf, ra_, false, offset, src_first, Assembler::stx_op3, "STX ", st);
-    size  =  impl_helper(this,cbuf,ra_,do_size,false,offset,src_first,Assembler::stw_op3,"STW ",size, st);
+      return;
+    }
+    impl_helper(mach, cbuf, ra_, false, offset, src_first, Assembler::stw_op3, "STW ", st);
   }
 
   // Check for integer load
-  if( dst_first_rc == rc_int && src_first_rc == rc_stack ) {
+  if (dst_first_rc == rc_int && src_first_rc == rc_stack) {
     int offset = ra_->reg2offset(src_first);
     // Further check for aligned-adjacent pair, so we can use a double load
-    if( (src_first&1)==0 && src_first+1 == src_second && (dst_first&1)==0 && dst_first+1 == dst_second )
+    if ((src_first & 1) == 0 && src_first + 1 == src_second && (dst_first & 1) == 0 && dst_first + 1 == dst_second) {
-      return impl_helper(this,cbuf,ra_,do_size,true,offset,dst_first,Assembler::ldx_op3 ,"LDX ",size, st);
+      impl_helper(mach, cbuf, ra_, true, offset, dst_first, Assembler::ldx_op3, "LDX ", st);
-    size  =  impl_helper(this,cbuf,ra_,do_size,true,offset,dst_first,Assembler::lduw_op3,"LDUW",size, st);
+      return;
+    }
+    impl_helper(mach, cbuf, ra_, true, offset, dst_first, Assembler::lduw_op3, "LDUW", st);
   }
 
   // Check for float reg-reg copy
-  if( src_first_rc == rc_float && dst_first_rc == rc_float ) {
+  if (src_first_rc == rc_float && dst_first_rc == rc_float) {
     // Further check for aligned-adjacent pair, so we can use a double move
-    if( (src_first&1)==0 && src_first+1 == src_second && (dst_first&1)==0 && dst_first+1 == dst_second )
+    if ((src_first & 1) == 0 && src_first + 1 == src_second && (dst_first & 1) == 0 && dst_first + 1 == dst_second) {
-      return impl_mov_helper(cbuf,do_size,src_first,dst_first,Assembler::fpop1_op3,Assembler::fmovd_opf,"FMOVD",size, st);
+      impl_mov_helper(cbuf, src_first, dst_first, Assembler::fpop1_op3, Assembler::fmovd_opf, "FMOVD", st);
-    size  =  impl_mov_helper(cbuf,do_size,src_first,dst_first,Assembler::fpop1_op3,Assembler::fmovs_opf,"FMOVS",size, st);
+      return;
+    }
+    impl_mov_helper(cbuf, src_first, dst_first, Assembler::fpop1_op3, Assembler::fmovs_opf, "FMOVS", st);
   }
 
   // Check for float store
-  if( src_first_rc == rc_float && dst_first_rc == rc_stack ) {
+  if (src_first_rc == rc_float && dst_first_rc == rc_stack) {
     int offset = ra_->reg2offset(dst_first);
     // Further check for aligned-adjacent pair, so we can use a double store
-    if( (src_first&1)==0 && src_first+1 == src_second && (dst_first&1)==0 && dst_first+1 == dst_second )
+    if ((src_first & 1) == 0 && src_first + 1 == src_second && (dst_first & 1) == 0 && dst_first + 1 == dst_second) {
-      return impl_helper(this,cbuf,ra_,do_size,false,offset,src_first,Assembler::stdf_op3,"STDF",size, st);
+      impl_helper(mach, cbuf, ra_, false, offset, src_first, Assembler::stdf_op3, "STDF", st);
-    size  =  impl_helper(this,cbuf,ra_,do_size,false,offset,src_first,Assembler::stf_op3 ,"STF ",size, st);
+      return;
+    }
+    impl_helper(mach, cbuf, ra_, false, offset, src_first, Assembler::stf_op3, "STF ", st);
   }
 
   // Check for float load
-  if( dst_first_rc == rc_float && src_first_rc == rc_stack ) {
+  if (dst_first_rc == rc_float && src_first_rc == rc_stack) {
     int offset = ra_->reg2offset(src_first);
     // Further check for aligned-adjacent pair, so we can use a double load
-    if( (src_first&1)==0 && src_first+1 == src_second && (dst_first&1)==0 && dst_first+1 == dst_second )
+    if ((src_first & 1) == 0 && src_first + 1 == src_second && (dst_first & 1) == 0 && dst_first + 1 == dst_second) {
-      return impl_helper(this,cbuf,ra_,do_size,true,offset,dst_first,Assembler::lddf_op3,"LDDF",size, st);
+      impl_helper(mach, cbuf, ra_, true, offset, dst_first, Assembler::lddf_op3, "LDDF", st);
-    size  =  impl_helper(this,cbuf,ra_,do_size,true,offset,dst_first,Assembler::ldf_op3 ,"LDF ",size, st);
+      return;
+    }
+    impl_helper(mach, cbuf, ra_, true, offset, dst_first, Assembler::ldf_op3, "LDF ", st);
   }
 
   // --------------------------------------------------------------------
   // Check for hi bits still needing moving.  Only happens for misaligned
   // arguments to native calls.
-  if( src_second == dst_second )
+  if (src_second == dst_second) {
-    return size;               // Self copy; no move
+    return; // Self copy; no move
-  assert( src_second_rc != rc_bad && dst_second_rc != rc_bad, "src_second & dst_second cannot be Bad" );
+  }
+  assert(src_second_rc != rc_bad && dst_second_rc != rc_bad, "src_second & dst_second cannot be Bad");
 
 #ifndef _LP64
   // In the LP64 build, all registers can be moved as aligned/adjacent
@@ -1609,52 +1632,57 @@ uint MachSpillCopyNode::implementation( CodeBuffer *cbuf,
   // 32-bits of a 64-bit register, but are needed in low bits of another
   // register (else it's a hi-bits-to-hi-bits copy which should have
   // happened already as part of a 64-bit move)
-  if( src_second_rc == rc_int && dst_second_rc == rc_int ) {
+  if (src_second_rc == rc_int && dst_second_rc == rc_int) {
-    assert( (src_second&1)==1, "its the evil O0/O1 native return case" );
+    assert((src_second & 1) == 1, "its the evil O0/O1 native return case");
-    assert( (dst_second&1)==0, "should have moved with 1 64-bit move" );
+    assert((dst_second & 1) == 0, "should have moved with 1 64-bit move");
     // Shift src_second down to dst_second's low bits.
-    if( cbuf ) {
+    if (cbuf) {
-      emit3_simm13( *cbuf, Assembler::arith_op, Matcher::_regEncode[dst_second], Assembler::srlx_op3, Matcher::_regEncode[src_second-1], 0x1020 );
+      emit3_simm13(*cbuf, Assembler::arith_op, Matcher::_regEncode[dst_second], Assembler::srlx_op3, Matcher::_regEncode[src_second-1], 0x1020);
 #ifndef PRODUCT
-    } else if( !do_size ) {
+    } else  {
-      if( size != 0 ) st->print("\n\t");
+      print_helper(st, "SRLX   R_%s,32,R_%s\t! spill: Move high bits down low", OptoReg::regname(src_second - 1), OptoReg::regname(dst_second));
-      st->print("SRLX   R_%s,32,R_%s\t! spill: Move high bits down low",OptoReg::regname(src_second-1),OptoReg::regname(dst_second));
 #endif
     }
-    return size+4;
+    return;
   }
 
   // Check for high word integer store.  Must down-shift the hi bits
   // into a temp register, then fall into the case of storing int bits.
-  if( src_second_rc == rc_int && dst_second_rc == rc_stack && (src_second&1)==1 ) {
+  if (src_second_rc == rc_int && dst_second_rc == rc_stack && (src_second & 1) == 1) {
     // Shift src_second down to dst_second's low bits.
-    if( cbuf ) {
+    if (cbuf) {
-      emit3_simm13( *cbuf, Assembler::arith_op, Matcher::_regEncode[R_O7_num], Assembler::srlx_op3, Matcher::_regEncode[src_second-1], 0x1020 );
+      emit3_simm13(*cbuf, Assembler::arith_op, Matcher::_regEncode[R_O7_num], Assembler::srlx_op3, Matcher::_regEncode[src_second-1], 0x1020);
 #ifndef PRODUCT
-    } else if( !do_size ) {
+    } else {
-      if( size != 0 ) st->print("\n\t");
+      print_helper(st, "SRLX   R_%s,32,R_%s\t! spill: Move high bits down low", OptoReg::regname(src_second-1), OptoReg::regname(R_O7_num));
-      st->print("SRLX   R_%s,32,R_%s\t! spill: Move high bits down low",OptoReg::regname(src_second-1),OptoReg::regname(R_O7_num));
 #endif
     }
-    size+=4;
     src_second = OptoReg::Name(R_O7_num); // Not R_O7H_num!
   }
 
   // Check for high word integer load
-  if( dst_second_rc == rc_int && src_second_rc == rc_stack )
+  if (dst_second_rc == rc_int && src_second_rc == rc_stack)
-    return impl_helper(this,cbuf,ra_,do_size,true ,ra_->reg2offset(src_second),dst_second,Assembler::lduw_op3,"LDUW",size, st);
+    return impl_helper(this, cbuf, ra_, true, ra_->reg2offset(src_second), dst_second, Assembler::lduw_op3, "LDUW", size, st);
 
   // Check for high word integer store
-  if( src_second_rc == rc_int && dst_second_rc == rc_stack )
+  if (src_second_rc == rc_int && dst_second_rc == rc_stack)
-    return impl_helper(this,cbuf,ra_,do_size,false,ra_->reg2offset(dst_second),src_second,Assembler::stw_op3 ,"STW ",size, st);
+    return impl_helper(this, cbuf, ra_, false, ra_->reg2offset(dst_second), src_second, Assembler::stw_op3, "STW ", size, st);
 
   // Check for high word float store
-  if( src_second_rc == rc_float && dst_second_rc == rc_stack )
+  if (src_second_rc == rc_float && dst_second_rc == rc_stack)
-    return impl_helper(this,cbuf,ra_,do_size,false,ra_->reg2offset(dst_second),src_second,Assembler::stf_op3 ,"STF ",size, st);
+    return impl_helper(this, cbuf, ra_, false, ra_->reg2offset(dst_second), src_second, Assembler::stf_op3, "STF ", size, st);
 
 #endif // !_LP64
 
   Unimplemented();
+}
+
+uint MachSpillCopyNode::implementation(CodeBuffer *cbuf,
+                                       PhaseRegAlloc *ra_,
+                                       bool do_size,
+                                       outputStream* st) const {
+  assert(!do_size, "not supported");
+  mach_spill_copy_implementation_helper(this, cbuf, ra_, st);
   return 0;
 }
 
@@ -1669,19 +1697,19 @@ void MachSpillCopyNode::emit(CodeBuffer &cbuf, PhaseRegAlloc *ra_) const {
 }
 
 uint MachSpillCopyNode::size(PhaseRegAlloc *ra_) const {
-  return implementation( NULL, ra_, true, NULL );
+  return MachNode::size(ra_);
 }
 
 //=============================================================================
 #ifndef PRODUCT
-void MachNopNode::format( PhaseRegAlloc *, outputStream *st ) const {
+void MachNopNode::format(PhaseRegAlloc *, outputStream *st) const {
   st->print("NOP \t# %d bytes pad for loops and calls", 4 * _count);
 }
 #endif
 
-void MachNopNode::emit(CodeBuffer &cbuf, PhaseRegAlloc * ) const {
+void MachNopNode::emit(CodeBuffer &cbuf, PhaseRegAlloc *) const {
   MacroAssembler _masm(&cbuf);
-  for(int i = 0; i < _count; i += 1) {
+  for (int i = 0; i < _count; i += 1) {
     __ nop();
   }
 }
@@ -5197,7 +5225,6 @@ instruct stkI_to_regF(regF dst, stackSlotI src) %{
   // No match rule to avoid chain rule match.
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "LDF    $src,$dst\t! stkI to regF" %}
   opcode(Assembler::ldf_op3);
   ins_encode(simple_form3_mem_reg(src, dst));
@@ -5208,7 +5235,6 @@ instruct stkL_to_regD(regD dst, stackSlotL src) %{
   // No match rule to avoid chain rule match.
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "LDDF   $src,$dst\t! stkL to regD" %}
   opcode(Assembler::lddf_op3);
   ins_encode(simple_form3_mem_reg(src, dst));
@@ -5219,7 +5245,6 @@ instruct regF_to_stkI(stackSlotI dst, regF src) %{
   // No match rule to avoid chain rule match.
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "STF    $src,$dst\t! regF to stkI" %}
   opcode(Assembler::stf_op3);
   ins_encode(simple_form3_mem_reg(dst, src));
@@ -5230,7 +5255,6 @@ instruct regD_to_stkL(stackSlotL dst, regD src) %{
   // No match rule to avoid chain rule match.
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "STDF   $src,$dst\t! regD to stkL" %}
   opcode(Assembler::stdf_op3);
   ins_encode(simple_form3_mem_reg(dst, src));
@@ -5240,7 +5264,6 @@ instruct regD_to_stkL(stackSlotL dst, regD src) %{
 instruct regI_to_stkLHi(stackSlotL dst, iRegI src) %{
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST*2);
-  size(8);
   format %{ "STW    $src,$dst.hi\t! long\n\t"
             "STW    R_G0,$dst.lo" %}
   opcode(Assembler::stw_op3);
@@ -5252,7 +5275,6 @@ instruct regL_to_stkD(stackSlotD dst, iRegL src) %{
   // No match rule to avoid chain rule match.
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "STX    $src,$dst\t! regL to stkD" %}
   opcode(Assembler::stx_op3);
   ins_encode(simple_form3_mem_reg( dst, src ) );
@@ -5266,7 +5288,6 @@ instruct stkI_to_regI( iRegI dst, stackSlotI src ) %{
   match(Set dst src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "LDUW   $src,$dst\t!stk" %}
   opcode(Assembler::lduw_op3);
   ins_encode(simple_form3_mem_reg( src, dst ) );
@@ -5278,7 +5299,6 @@ instruct regI_to_stkI( stackSlotI dst, iRegI src ) %{
   match(Set dst src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STW    $src,$dst\t!stk" %}
   opcode(Assembler::stw_op3);
   ins_encode(simple_form3_mem_reg( dst, src ) );
@@ -5290,7 +5310,6 @@ instruct stkL_to_regL( iRegL dst, stackSlotL src ) %{
   match(Set dst src);
 
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "LDX    $src,$dst\t! long" %}
   opcode(Assembler::ldx_op3);
   ins_encode(simple_form3_mem_reg( src, dst ) );
@@ -5302,7 +5321,6 @@ instruct regL_to_stkL(stackSlotL dst, iRegL src) %{
   match(Set dst src);
 
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "STX    $src,$dst\t! long" %}
   opcode(Assembler::stx_op3);
   ins_encode(simple_form3_mem_reg( dst, src ) );
@@ -5314,7 +5332,6 @@ instruct regL_to_stkL(stackSlotL dst, iRegL src) %{
 instruct stkP_to_regP( iRegP dst, stackSlotP src ) %{
   match(Set dst src);
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "LDX    $src,$dst\t!ptr" %}
   opcode(Assembler::ldx_op3);
   ins_encode(simple_form3_mem_reg( src, dst ) );
@@ -5325,7 +5342,6 @@ instruct stkP_to_regP( iRegP dst, stackSlotP src ) %{
 instruct regP_to_stkP(stackSlotP dst, iRegP src) %{
   match(Set dst src);
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "STX    $src,$dst\t!ptr" %}
   opcode(Assembler::stx_op3);
   ins_encode(simple_form3_mem_reg( dst, src ) );
@@ -5771,7 +5787,6 @@ instruct loadL_unaligned(iRegL dst, memory mem, o7RegI tmp) %{
   match(Set dst (LoadL_unaligned mem));
   effect(KILL tmp);
   ins_cost(MEMORY_REF_COST*2+DEFAULT_COST);
-  size(16);
   format %{ "LDUW   $mem+4,R_O7\t! misaligned long\n"
           "\tLDUW   $mem  ,$dst\n"
           "\tSLLX   #32, $dst, $dst\n"
@@ -5786,7 +5801,6 @@ instruct loadRange(iRegI dst, memory mem) %{
   match(Set dst (LoadRange mem));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "LDUW   $mem,$dst\t! range" %}
   opcode(Assembler::lduw_op3);
   ins_encode(simple_form3_mem_reg( mem, dst ) );
@@ -5797,7 +5811,6 @@ instruct loadRange(iRegI dst, memory mem) %{
 instruct loadI_freg(regF dst, memory mem) %{
   match(Set dst (LoadI mem));
   ins_cost(MEMORY_REF_COST);
-  size(4);
 
   format %{ "LDF    $mem,$dst\t! for fitos/fitod" %}
   opcode(Assembler::ldf_op3);
@@ -5876,7 +5889,6 @@ instruct loadD(regD dst, memory mem) %{
   match(Set dst (LoadD mem));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "LDDF   $mem,$dst" %}
   opcode(Assembler::lddf_op3);
   ins_encode(simple_form3_mem_reg( mem, dst ) );
@@ -5887,7 +5899,6 @@ instruct loadD(regD dst, memory mem) %{
 instruct loadD_unaligned(regD_low dst, memory mem ) %{
   match(Set dst (LoadD_unaligned mem));
   ins_cost(MEMORY_REF_COST*2+DEFAULT_COST);
-  size(8);
   format %{ "LDF    $mem  ,$dst.hi\t! misaligned double\n"
           "\tLDF    $mem+4,$dst.lo\t!" %}
   opcode(Assembler::ldf_op3);
@@ -5900,7 +5911,6 @@ instruct loadF(regF dst, memory mem) %{
   match(Set dst (LoadF mem));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "LDF    $mem,$dst" %}
   opcode(Assembler::ldf_op3);
   ins_encode(simple_form3_mem_reg( mem, dst ) );
@@ -6119,7 +6129,6 @@ instruct prefetchAlloc( memory mem ) %{
   predicate(AllocatePrefetchInstr == 0);
   match( PrefetchAllocation mem );
   ins_cost(MEMORY_REF_COST);
-  size(4);
 
   format %{ "PREFETCH $mem,2\t! Prefetch allocation" %}
   opcode(Assembler::prefetch_op3);
@@ -6175,7 +6184,6 @@ instruct storeB(memory mem, iRegI src) %{
   match(Set mem (StoreB mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STB    $src,$mem\t! byte" %}
   opcode(Assembler::stb_op3);
   ins_encode(simple_form3_mem_reg( mem, src ) );
@@ -6186,7 +6194,6 @@ instruct storeB0(memory mem, immI0 src) %{
   match(Set mem (StoreB mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STB    $src,$mem\t! byte" %}
   opcode(Assembler::stb_op3);
   ins_encode(simple_form3_mem_reg( mem, R_G0 ) );
@@ -6197,7 +6204,6 @@ instruct storeCM0(memory mem, immI0 src) %{
   match(Set mem (StoreCM mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STB    $src,$mem\t! CMS card-mark byte 0" %}
   opcode(Assembler::stb_op3);
   ins_encode(simple_form3_mem_reg( mem, R_G0 ) );
@@ -6209,7 +6215,6 @@ instruct storeC(memory mem, iRegI src) %{
   match(Set mem (StoreC mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STH    $src,$mem\t! short" %}
   opcode(Assembler::sth_op3);
   ins_encode(simple_form3_mem_reg( mem, src ) );
@@ -6220,7 +6225,6 @@ instruct storeC0(memory mem, immI0 src) %{
   match(Set mem (StoreC mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STH    $src,$mem\t! short" %}
   opcode(Assembler::sth_op3);
   ins_encode(simple_form3_mem_reg( mem, R_G0 ) );
@@ -6232,7 +6236,6 @@ instruct storeI(memory mem, iRegI src) %{
   match(Set mem (StoreI mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STW    $src,$mem" %}
   opcode(Assembler::stw_op3);
   ins_encode(simple_form3_mem_reg( mem, src ) );
@@ -6243,7 +6246,6 @@ instruct storeI(memory mem, iRegI src) %{
 instruct storeL(memory mem, iRegL src) %{
   match(Set mem (StoreL mem src));
   ins_cost(MEMORY_REF_COST);
-  size(4);
   format %{ "STX    $src,$mem\t! long" %}
   opcode(Assembler::stx_op3);
   ins_encode(simple_form3_mem_reg( mem, src ) );
@@ -6254,7 +6256,6 @@ instruct storeI0(memory mem, immI0 src) %{
   match(Set mem (StoreI mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STW    $src,$mem" %}
   opcode(Assembler::stw_op3);
   ins_encode(simple_form3_mem_reg( mem, R_G0 ) );
@@ -6265,7 +6266,6 @@ instruct storeL0(memory mem, immL0 src) %{
   match(Set mem (StoreL mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STX    $src,$mem" %}
   opcode(Assembler::stx_op3);
   ins_encode(simple_form3_mem_reg( mem, R_G0 ) );
@@ -6277,7 +6277,6 @@ instruct storeI_Freg(memory mem, regF src) %{
   match(Set mem (StoreI mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STF    $src,$mem\t! after fstoi/fdtoi" %}
   opcode(Assembler::stf_op3);
   ins_encode(simple_form3_mem_reg( mem, src ) );
@@ -6288,7 +6287,6 @@ instruct storeI_Freg(memory mem, regF src) %{
 instruct storeP(memory dst, sp_ptr_RegP src) %{
   match(Set dst (StoreP dst src));
   ins_cost(MEMORY_REF_COST);
-  size(4);
 
 #ifndef _LP64
   format %{ "STW    $src,$dst\t! ptr" %}
@@ -6304,7 +6302,6 @@ instruct storeP(memory dst, sp_ptr_RegP src) %{
 instruct storeP0(memory dst, immP0 src) %{
   match(Set dst (StoreP dst src));
   ins_cost(MEMORY_REF_COST);
-  size(4);
 
 #ifndef _LP64
   format %{ "STW    $src,$dst\t! ptr" %}
@@ -6379,7 +6376,6 @@ instruct storeD( memory mem, regD src) %{
   match(Set mem (StoreD mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STDF   $src,$mem" %}
   opcode(Assembler::stdf_op3);
   ins_encode(simple_form3_mem_reg( mem, src ) );
@@ -6390,7 +6386,6 @@ instruct storeD0( memory mem, immD0 src) %{
   match(Set mem (StoreD mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STX    $src,$mem" %}
   opcode(Assembler::stx_op3);
   ins_encode(simple_form3_mem_reg( mem, R_G0 ) );
@@ -6402,7 +6397,6 @@ instruct storeF( memory mem, regF src) %{
   match(Set mem (StoreF mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STF    $src,$mem" %}
   opcode(Assembler::stf_op3);
   ins_encode(simple_form3_mem_reg( mem, src ) );
@@ -6413,7 +6407,6 @@ instruct storeF0( memory mem, immF0 src) %{
   match(Set mem (StoreF mem src));
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STW    $src,$mem\t! storeF0" %}
   opcode(Assembler::stw_op3);
   ins_encode(simple_form3_mem_reg( mem, R_G0 ) );
@@ -7068,7 +7061,6 @@ instruct loadPLocked(iRegP dst, memory mem) %{
   ins_cost(MEMORY_REF_COST);
 
 #ifndef _LP64
-  size(4);
   format %{ "LDUW   $mem,$dst\t! ptr" %}
   opcode(Assembler::lduw_op3, 0, REGP_OP);
 #else
@@ -8138,7 +8130,6 @@ instruct MoveF2I_stack_reg(iRegI dst, stackSlotF src) %{
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "LDUW   $src,$dst\t! MoveF2I" %}
   opcode(Assembler::lduw_op3);
   ins_encode(simple_form3_mem_reg( src, dst ) );
@@ -8150,7 +8141,6 @@ instruct MoveI2F_stack_reg(regF dst, stackSlotI src) %{
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "LDF    $src,$dst\t! MoveI2F" %}
   opcode(Assembler::ldf_op3);
   ins_encode(simple_form3_mem_reg(src, dst));
@@ -8162,7 +8152,6 @@ instruct MoveD2L_stack_reg(iRegL dst, stackSlotD src) %{
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "LDX    $src,$dst\t! MoveD2L" %}
   opcode(Assembler::ldx_op3);
   ins_encode(simple_form3_mem_reg( src, dst ) );
@@ -8174,7 +8163,6 @@ instruct MoveL2D_stack_reg(regD dst, stackSlotL src) %{
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "LDDF   $src,$dst\t! MoveL2D" %}
   opcode(Assembler::lddf_op3);
   ins_encode(simple_form3_mem_reg(src, dst));
@@ -8186,7 +8174,6 @@ instruct MoveF2I_reg_stack(stackSlotI dst, regF src) %{
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STF   $src,$dst\t! MoveF2I" %}
   opcode(Assembler::stf_op3);
   ins_encode(simple_form3_mem_reg(dst, src));
@@ -8198,7 +8185,6 @@ instruct MoveI2F_reg_stack(stackSlotF dst, iRegI src) %{
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STW    $src,$dst\t! MoveI2F" %}
   opcode(Assembler::stw_op3);
   ins_encode(simple_form3_mem_reg( dst, src ) );
@@ -8210,7 +8196,6 @@ instruct MoveD2L_reg_stack(stackSlotL dst, regD src) %{
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STDF   $src,$dst\t! MoveD2L" %}
   opcode(Assembler::stdf_op3);
   ins_encode(simple_form3_mem_reg(dst, src));
@@ -8222,7 +8207,6 @@ instruct MoveL2D_reg_stack(stackSlotD dst, iRegL src) %{
   effect(DEF dst, USE src);
   ins_cost(MEMORY_REF_COST);
 
-  size(4);
   format %{ "STX    $src,$dst\t! MoveL2D" %}
   opcode(Assembler::stx_op3);
   ins_encode(simple_form3_mem_reg( dst, src ) );
@@ -8427,7 +8411,6 @@ instruct convI2D_reg(regD_low dst, iRegI src) %{
 instruct convI2D_mem(regD_low dst, memory mem) %{
   match(Set dst (ConvI2D (LoadI mem)));
   ins_cost(DEFAULT_COST + MEMORY_REF_COST);
-  size(8);
   format %{ "LDF    $mem,$dst\n\t"
             "FITOD  $dst,$dst" %}
   opcode(Assembler::ldf_op3, Assembler::fitod_opf);
@@ -8468,7 +8451,6 @@ instruct convI2F_reg(regF dst, iRegI src) %{
 instruct convI2F_mem( regF dst, memory mem ) %{
   match(Set dst (ConvI2F (LoadI mem)));
   ins_cost(DEFAULT_COST + MEMORY_REF_COST);
-  size(8);
   format %{ "LDF    $mem,$dst\n\t"
             "FITOS  $dst,$dst" %}
   opcode(Assembler::ldf_op3, Assembler::fitos_opf);

hotspot/src/cpu/sparc/vm/vm_version_sparc.cpp

@@ -463,3 +463,37 @@ unsigned int VM_Version::calc_parallel_worker_threads() {
   }
   return result;
 }
+
+
+int VM_Version::parse_features(const char* implementation) {
+  int features = unknown_m;
+  // Convert to UPPER case before compare.
+  char* impl = os::strdup_check_oom(implementation);
+
+  for (int i = 0; impl[i] != 0; i++)
+    impl[i] = (char)toupper((uint)impl[i]);
+
+  if (strstr(impl, "SPARC64") != NULL) {
+    features |= sparc64_family_m;
+  } else if (strstr(impl, "SPARC-M") != NULL) {
+    // M-series SPARC is based on T-series.
+    features |= (M_family_m | T_family_m);
+  } else if (strstr(impl, "SPARC-T") != NULL) {
+    features |= T_family_m;
+    if (strstr(impl, "SPARC-T1") != NULL) {
+      features |= T1_model_m;
+    }
+  } else {
+    if (strstr(impl, "SPARC") == NULL) {
+#ifndef PRODUCT
+      // kstat on Solaris 8 virtual machines (branded zones)
+      // returns "(unsupported)" implementation. Solaris 8 is not
+      // supported anymore, but include this check to be on the
+      // safe side.
+      warning("Can't parse CPU implementation = '%s', assume generic SPARC", impl);
+#endif
+    }
+  }
+  os::free((void*)impl);
+  return features;
+}

hotspot/src/cpu/sparc/vm/vm_version_sparc.hpp

@@ -121,7 +121,7 @@ protected:
   static bool is_T1_model(int features) { return is_T_family(features) && ((features & T1_model_m) != 0); }
 
   static int maximum_niagara1_processor_count() { return 32; }
-
+  static int parse_features(const char* implementation);
 public:
   // Initialization
   static void initialize();

hotspot/src/cpu/x86/vm/templateInterpreterGenerator_x86.cpp

@@ -161,13 +161,7 @@ address TemplateInterpreterGenerator::generate_exception_handler_common(
                                      create_klass_exception),
                rarg, rarg2);
   } else {
-    // kind of lame ExternalAddress can't take NULL because
+    __ lea(rarg2, ExternalAddress((address)message));
-    // external_word_Relocation will assert.
-    if (message != NULL) {
-      __ lea(rarg2, ExternalAddress((address)message));
-    } else {
-      __ movptr(rarg2, NULL_WORD);
-    }
     __ call_VM(rax,
                CAST_FROM_FN_PTR(address, InterpreterRuntime::create_exception),
                rarg, rarg2);

hotspot/src/cpu/x86/vm/x86_32.ad

@@ -7236,6 +7236,7 @@ instruct storeLConditional( memory mem, eADXRegL oldval, eBCXRegL newval, eFlags
 instruct compareAndSwapL( rRegI res, eSIRegP mem_ptr, eADXRegL oldval, eBCXRegL newval, eFlagsReg cr ) %{
   predicate(VM_Version::supports_cx8());
   match(Set res (CompareAndSwapL mem_ptr (Binary oldval newval)));
+  match(Set res (WeakCompareAndSwapL mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
   format %{ "CMPXCHG8 [$mem_ptr],$newval\t# If EDX:EAX==[$mem_ptr] Then store $newval into [$mem_ptr]\n\t"
             "MOV    $res,0\n\t"
@@ -7249,6 +7250,7 @@ instruct compareAndSwapL( rRegI res, eSIRegP mem_ptr, eADXRegL oldval, eBCXRegL
 
 instruct compareAndSwapP( rRegI res,  pRegP mem_ptr, eAXRegP oldval, eCXRegP newval, eFlagsReg cr) %{
   match(Set res (CompareAndSwapP mem_ptr (Binary oldval newval)));
+  match(Set res (WeakCompareAndSwapP mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
   format %{ "CMPXCHG [$mem_ptr],$newval\t# If EAX==[$mem_ptr] Then store $newval into [$mem_ptr]\n\t"
             "MOV    $res,0\n\t"
@@ -7261,6 +7263,7 @@ instruct compareAndSwapP( rRegI res,  pRegP mem_ptr, eAXRegP oldval, eCXRegP new
 
 instruct compareAndSwapI( rRegI res, pRegP mem_ptr, eAXRegI oldval, eCXRegI newval, eFlagsReg cr) %{
   match(Set res (CompareAndSwapI mem_ptr (Binary oldval newval)));
+  match(Set res (WeakCompareAndSwapI mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
   format %{ "CMPXCHG [$mem_ptr],$newval\t# If EAX==[$mem_ptr] Then store $newval into [$mem_ptr]\n\t"
             "MOV    $res,0\n\t"
@@ -7271,6 +7274,31 @@ instruct compareAndSwapI( rRegI res, pRegP mem_ptr, eAXRegI oldval, eCXRegI newv
   ins_pipe( pipe_cmpxchg );
 %}
 
+instruct compareAndExchangeL( eSIRegP mem_ptr, eADXRegL oldval, eBCXRegL newval, eFlagsReg cr ) %{
+  predicate(VM_Version::supports_cx8());
+  match(Set oldval (CompareAndExchangeL mem_ptr (Binary oldval newval)));
+  effect(KILL cr);
+  format %{ "CMPXCHG8 [$mem_ptr],$newval\t# If EDX:EAX==[$mem_ptr] Then store $newval into [$mem_ptr]\n\t" %}
+  ins_encode( enc_cmpxchg8(mem_ptr) );
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct compareAndExchangeP( pRegP mem_ptr, eAXRegP oldval, eCXRegP newval, eFlagsReg cr) %{
+  match(Set oldval (CompareAndExchangeP mem_ptr (Binary oldval newval)));
+  effect(KILL cr);
+  format %{ "CMPXCHG [$mem_ptr],$newval\t# If EAX==[$mem_ptr] Then store $newval into [$mem_ptr]\n\t" %}
+  ins_encode( enc_cmpxchg(mem_ptr) );
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct compareAndExchangeI( pRegP mem_ptr, eAXRegI oldval, eCXRegI newval, eFlagsReg cr) %{
+  match(Set oldval (CompareAndExchangeI mem_ptr (Binary oldval newval)));
+  effect(KILL cr);
+  format %{ "CMPXCHG [$mem_ptr],$newval\t# If EAX==[$mem_ptr] Then store $newval into [$mem_ptr]\n\t" %}
+  ins_encode( enc_cmpxchg(mem_ptr) );
+  ins_pipe( pipe_cmpxchg );
+%}
+
 instruct xaddI_no_res( memory mem, Universe dummy, immI add, eFlagsReg cr) %{
   predicate(n->as_LoadStore()->result_not_used());
   match(Set dummy (GetAndAddI mem add));

hotspot/src/cpu/x86/vm/x86_64.ad

@@ -7281,6 +7281,7 @@ instruct compareAndSwapP(rRegI res,
 %{
   predicate(VM_Version::supports_cx8());
   match(Set res (CompareAndSwapP mem_ptr (Binary oldval newval)));
+  match(Set res (WeakCompareAndSwapP mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
 
   format %{ "cmpxchgq $mem_ptr,$newval\t# "
@@ -7305,6 +7306,7 @@ instruct compareAndSwapL(rRegI res,
 %{
   predicate(VM_Version::supports_cx8());
   match(Set res (CompareAndSwapL mem_ptr (Binary oldval newval)));
+  match(Set res (WeakCompareAndSwapL mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
 
   format %{ "cmpxchgq $mem_ptr,$newval\t# "
@@ -7328,6 +7330,7 @@ instruct compareAndSwapI(rRegI res,
                          rFlagsReg cr)
 %{
   match(Set res (CompareAndSwapI mem_ptr (Binary oldval newval)));
+  match(Set res (WeakCompareAndSwapI mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
 
   format %{ "cmpxchgl $mem_ptr,$newval\t# "
@@ -7351,6 +7354,7 @@ instruct compareAndSwapN(rRegI res,
                           rax_RegN oldval, rRegN newval,
                           rFlagsReg cr) %{
   match(Set res (CompareAndSwapN mem_ptr (Binary oldval newval)));
+  match(Set res (WeakCompareAndSwapN mem_ptr (Binary oldval newval)));
   effect(KILL cr, KILL oldval);
 
   format %{ "cmpxchgl $mem_ptr,$newval\t# "
@@ -7368,6 +7372,83 @@ instruct compareAndSwapN(rRegI res,
   ins_pipe( pipe_cmpxchg );
 %}
 
+instruct compareAndExchangeI(
+                         memory mem_ptr,
+                         rax_RegI oldval, rRegI newval,
+                         rFlagsReg cr)
+%{
+  match(Set oldval (CompareAndExchangeI mem_ptr (Binary oldval newval)));
+  effect(KILL cr);
+
+  format %{ "cmpxchgl $mem_ptr,$newval\t# "
+            "If rax == $mem_ptr then store $newval into $mem_ptr\n\t"  %}
+  opcode(0x0F, 0xB1);
+  ins_encode(lock_prefix,
+             REX_reg_mem(newval, mem_ptr),
+             OpcP, OpcS,
+             reg_mem(newval, mem_ptr) // lock cmpxchg
+             );
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct compareAndExchangeL(
+                         memory mem_ptr,
+                         rax_RegL oldval, rRegL newval,
+                         rFlagsReg cr)
+%{
+  predicate(VM_Version::supports_cx8());
+  match(Set oldval (CompareAndExchangeL mem_ptr (Binary oldval newval)));
+  effect(KILL cr);
+
+  format %{ "cmpxchgq $mem_ptr,$newval\t# "
+            "If rax == $mem_ptr then store $newval into $mem_ptr\n\t"  %}
+  opcode(0x0F, 0xB1);
+  ins_encode(lock_prefix,
+             REX_reg_mem_wide(newval, mem_ptr),
+             OpcP, OpcS,
+             reg_mem(newval, mem_ptr)  // lock cmpxchg
+            );
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct compareAndExchangeN(
+                          memory mem_ptr,
+                          rax_RegN oldval, rRegN newval,
+                          rFlagsReg cr) %{
+  match(Set oldval (CompareAndExchangeN mem_ptr (Binary oldval newval)));
+  effect(KILL cr);
+
+  format %{ "cmpxchgl $mem_ptr,$newval\t# "
+            "If rax == $mem_ptr then store $newval into $mem_ptr\n\t" %}
+  opcode(0x0F, 0xB1);
+  ins_encode(lock_prefix,
+             REX_reg_mem(newval, mem_ptr),
+             OpcP, OpcS,
+             reg_mem(newval, mem_ptr)  // lock cmpxchg
+          );
+  ins_pipe( pipe_cmpxchg );
+%}
+
+instruct compareAndExchangeP(
+                         memory mem_ptr,
+                         rax_RegP oldval, rRegP newval,
+                         rFlagsReg cr)
+%{
+  predicate(VM_Version::supports_cx8());
+  match(Set oldval (CompareAndExchangeP mem_ptr (Binary oldval newval)));
+  effect(KILL cr);
+
+  format %{ "cmpxchgq $mem_ptr,$newval\t# "
+            "If rax == $mem_ptr then store $newval into $mem_ptr\n\t" %}
+  opcode(0x0F, 0xB1);
+  ins_encode(lock_prefix,
+             REX_reg_mem_wide(newval, mem_ptr),
+             OpcP, OpcS,
+             reg_mem(newval, mem_ptr)  // lock cmpxchg
+          );
+  ins_pipe( pipe_cmpxchg );
+%}
+
 instruct xaddI_no_res( memory mem, Universe dummy, immI add, rFlagsReg cr) %{
   predicate(n->as_LoadStore()->result_not_used());
   match(Set dummy (GetAndAddI mem add));

hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.amd64/src/jdk/vm/ci/amd64/AMD64.java

@@ -22,6 +22,7 @@
  */
 package jdk.vm.ci.amd64;
 
+import static jdk.vm.ci.code.MemoryBarriers.LOAD_LOAD;
 import static jdk.vm.ci.code.MemoryBarriers.LOAD_STORE;
 import static jdk.vm.ci.code.MemoryBarriers.STORE_STORE;
 import static jdk.vm.ci.code.Register.SPECIAL;
@@ -220,7 +221,7 @@ public class AMD64 extends Architecture {
     private final AMD64Kind largestKind;
 
     public AMD64(EnumSet<CPUFeature> features, EnumSet<Flag> flags) {
-        super("AMD64", AMD64Kind.QWORD, ByteOrder.LITTLE_ENDIAN, true, allRegisters, LOAD_STORE | STORE_STORE, 1, 8);
+        super("AMD64", AMD64Kind.QWORD, ByteOrder.LITTLE_ENDIAN, true, allRegisters, LOAD_LOAD | LOAD_STORE | STORE_STORE, 1, 8);
         this.features = features;
         this.flags = flags;
         assert features.contains(CPUFeature.SSE2) : "minimum config for x64";

hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/HotSpotVMConfig.java

@@ -1141,7 +1141,7 @@ public class HotSpotVMConfig {
 
     @HotSpotVMField(name = "JavaFrameAnchor::_last_Java_sp", type = "intptr_t*", get = HotSpotVMField.Type.OFFSET) @Stable private int javaFrameAnchorLastJavaSpOffset;
     @HotSpotVMField(name = "JavaFrameAnchor::_last_Java_pc", type = "address", get = HotSpotVMField.Type.OFFSET) @Stable private int javaFrameAnchorLastJavaPcOffset;
-    @HotSpotVMField(name = "JavaFrameAnchor::_last_Java_fp", type = "intptr_t*", get = HotSpotVMField.Type.OFFSET, archs = {"amd64"}) @Stable private int javaFrameAnchorLastJavaFpOffset;
+    @HotSpotVMField(name = "JavaFrameAnchor::_last_Java_fp", type = "intptr_t*", get = HotSpotVMField.Type.OFFSET, archs = {"aarch64, amd64"}) @Stable private int javaFrameAnchorLastJavaFpOffset;
     @HotSpotVMField(name = "JavaFrameAnchor::_flags", type = "int", get = HotSpotVMField.Type.OFFSET, archs = {"sparc"}) @Stable private int javaFrameAnchorFlagsOffset;
 
     public int threadLastJavaSpOffset() {
@@ -1152,11 +1152,8 @@ public class HotSpotVMConfig {
         return javaThreadAnchorOffset + javaFrameAnchorLastJavaPcOffset;
     }
 
-    /**
-     * This value is only valid on AMD64.
-     */
     public int threadLastJavaFpOffset() {
-        // TODO add an assert for AMD64
+        assert getHostArchitectureName().equals("aarch64") || getHostArchitectureName().equals("amd64");
         return javaThreadAnchorOffset + javaFrameAnchorLastJavaFpOffset;
     }
 

hotspot/src/os/aix/vm/os_aix.cpp

@@ -1,6 +1,6 @@
 /*
- * Copyright (c) 1999, 2015, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1999, 2016, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2015 SAP SE. All rights reserved.
+ * Copyright (c) 2012, 2016 SAP SE. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -36,6 +36,7 @@
 #include "compiler/compileBroker.hpp"
 #include "interpreter/interpreter.hpp"
 #include "jvm_aix.h"
+#include "logging/log.hpp"
 #include "libo4.hpp"
 #include "libperfstat_aix.hpp"
 #include "libodm_aix.hpp"
@@ -791,13 +792,8 @@ static void *java_start(Thread *thread) {
   const pthread_t pthread_id = ::pthread_self();
   const tid_t kernel_thread_id = ::thread_self();
 
-  trcVerbose("newborn Thread : pthread-id %u, ktid " UINT64_FORMAT
+  log_info(os, thread)("Thread is alive (tid: " UINTX_FORMAT ", kernel thread id: " UINTX_FORMAT ").",
-    ", stack %p ... %p, stacksize 0x%IX (%IB)",
+    os::current_thread_id(), (uintx) kernel_thread_id);
-    pthread_id, kernel_thread_id,
-    thread->stack_end(),
-    thread->stack_base(),
-    thread->stack_size(),
-    thread->stack_size());
 
   // Normally, pthread stacks on AIX live in the data segment (are allocated with malloc()
   // by the pthread library). In rare cases, this may not be the case, e.g. when third-party
@@ -805,7 +801,7 @@ static void *java_start(Thread *thread) {
   // guard pages on those stacks, because the stacks may reside in memory which is not
   // protectable (shmated).
   if (thread->stack_base() > ::sbrk(0)) {
-    trcVerbose("Thread " UINT64_FORMAT ": stack not in data segment.", (uint64_t) pthread_id);
+    log_warning(os, thread)("Thread stack not in data segment.");
   }
 
   // Try to randomize the cache line index of hot stack frames.
@@ -839,8 +835,8 @@ static void *java_start(Thread *thread) {
   // Call one more level start routine.
   thread->run();
 
-  trcVerbose("Thread finished : pthread-id %u, ktid " UINT64_FORMAT ".",
+  log_info(os, thread)("Thread finished (tid: " UINTX_FORMAT ", kernel thread id: " UINTX_FORMAT ").",
-    pthread_id, kernel_thread_id);
+    os::current_thread_id(), (uintx) kernel_thread_id);
 
   return 0;
 }
@@ -908,20 +904,19 @@ bool os::create_thread(Thread* thread, ThreadType thr_type, size_t stack_size) {
   pthread_t tid;
   int ret = pthread_create(&tid, &attr, (void* (*)(void*)) java_start, thread);
 
+
+  char buf[64];
+  if (ret == 0) {
+    log_info(os, thread)("Thread started (pthread id: " UINTX_FORMAT ", attributes: %s). ",
+      (uintx) tid, os::Posix::describe_pthread_attr(buf, sizeof(buf), &attr));
+  } else {
+    log_warning(os, thread)("Failed to start thread - pthread_create failed (%s) for attributes: %s.",
+      strerror(ret), os::Posix::describe_pthread_attr(buf, sizeof(buf), &attr));
+  }
+
   pthread_attr_destroy(&attr);
 
-  if (ret == 0) {
+  if (ret != 0) {
-    trcVerbose("Created New Thread : pthread-id %u", tid);
-  } else {
-    if (os::Aix::on_pase()) {
-      // QIBM_MULTI_THREADED=Y is needed when the launcher is started on iSeries
-      // using QSH. Otherwise pthread_create fails with errno=11.
-      trcVerbose("(Please make sure you set the environment variable "
-              "QIBM_MULTI_THREADED=Y before running this program.)");
-    }
-    if (PrintMiscellaneous && (Verbose || WizardMode)) {
-      perror("pthread_create()");
-    }
     // Need to clean up stuff we've allocated so far
     thread->set_osthread(NULL);
     delete osthread;
@@ -958,13 +953,6 @@ bool os::create_attached_thread(JavaThread* thread) {
   const pthread_t pthread_id = ::pthread_self();
   const tid_t kernel_thread_id = ::thread_self();
 
-  trcVerbose("attaching Thread : pthread-id %u, ktid " UINT64_FORMAT ", stack %p ... %p, stacksize 0x%IX (%IB)",
-    pthread_id, kernel_thread_id,
-    thread->stack_end(),
-    thread->stack_base(),
-    thread->stack_size(),
-    thread->stack_size());
-
   // OSThread::thread_id is the pthread id.
   osthread->set_thread_id(pthread_id);
 
@@ -990,6 +978,9 @@ bool os::create_attached_thread(JavaThread* thread) {
   // and save the caller's signal mask
   os::Aix::hotspot_sigmask(thread);
 
+  log_info(os, thread)("Thread attached (tid: " UINTX_FORMAT ", kernel thread id: " UINTX_FORMAT ").",
+    os::current_thread_id(), (uintx) kernel_thread_id);
+
   return true;
 }
 

hotspot/src/os/bsd/vm/os_bsd.cpp

@@ -32,6 +32,7 @@
 #include "compiler/disassembler.hpp"
 #include "interpreter/interpreter.hpp"
 #include "jvm_bsd.h"
+#include "logging/log.hpp"
 #include "memory/allocation.inline.hpp"
 #include "memory/filemap.hpp"
 #include "mutex_bsd.inline.hpp"
@@ -681,6 +682,9 @@ static void *java_start(Thread *thread) {
 
   osthread->set_thread_id(os::Bsd::gettid());
 
+  log_info(os, thread)("Thread is alive (tid: " UINTX_FORMAT ", pthread id: " UINTX_FORMAT ").",
+    os::current_thread_id(), (uintx) pthread_self());
+
 #ifdef __APPLE__
   uint64_t unique_thread_id = locate_unique_thread_id(osthread->thread_id());
   guarantee(unique_thread_id != 0, "unique thread id was not found");
@@ -716,6 +720,9 @@ static void *java_start(Thread *thread) {
   // call one more level start routine
   thread->run();
 
+  log_info(os, thread)("Thread finished (tid: " UINTX_FORMAT ", pthread id: " UINTX_FORMAT ").",
+    os::current_thread_id(), (uintx) pthread_self());
+
   return 0;
 }
 
@@ -776,12 +783,18 @@ bool os::create_thread(Thread* thread, ThreadType thr_type, size_t stack_size) {
     pthread_t tid;
     int ret = pthread_create(&tid, &attr, (void* (*)(void*)) java_start, thread);
 
+    char buf[64];
+    if (ret == 0) {
+      log_info(os, thread)("Thread started (pthread id: " UINTX_FORMAT ", attributes: %s). ",
+        (uintx) tid, os::Posix::describe_pthread_attr(buf, sizeof(buf), &attr));
+    } else {
+      log_warning(os, thread)("Failed to start thread - pthread_create failed (%s) for attributes: %s.",
+        strerror(ret), os::Posix::describe_pthread_attr(buf, sizeof(buf), &attr));
+    }
+
     pthread_attr_destroy(&attr);
 
     if (ret != 0) {
-      if (PrintMiscellaneous && (Verbose || WizardMode)) {
-        perror("pthread_create()");
-      }
       // Need to clean up stuff we've allocated so far
       thread->set_osthread(NULL);
       delete osthread;
@@ -858,6 +871,9 @@ bool os::create_attached_thread(JavaThread* thread) {
   // and save the caller's signal mask
   os::Bsd::hotspot_sigmask(thread);
 
+  log_info(os, thread)("Thread attached (tid: " UINTX_FORMAT ", pthread id: " UINTX_FORMAT ").",
+    os::current_thread_id(), (uintx) pthread_self());
+
   return true;
 }
 

hotspot/src/os/linux/vm/os_linux.cpp

@@ -662,6 +662,9 @@ static void *java_start(Thread *thread) {
 
   osthread->set_thread_id(os::current_thread_id());
 
+  log_info(os, thread)("Thread is alive (tid: " UINTX_FORMAT ", pthread id: " UINTX_FORMAT ").",
+    os::current_thread_id(), (uintx) pthread_self());
+
   if (UseNUMA) {
     int lgrp_id = os::numa_get_group_id();
     if (lgrp_id != -1) {
@@ -691,6 +694,9 @@ static void *java_start(Thread *thread) {
   // call one more level start routine
   thread->run();
 
+  log_info(os, thread)("Thread finished (tid: " UINTX_FORMAT ", pthread id: " UINTX_FORMAT ").",
+    os::current_thread_id(), (uintx) pthread_self());
+
   return 0;
 }
 
@@ -756,12 +762,18 @@ bool os::create_thread(Thread* thread, ThreadType thr_type,
     pthread_t tid;
     int ret = pthread_create(&tid, &attr, (void* (*)(void*)) java_start, thread);
 
+    char buf[64];
+    if (ret == 0) {
+      log_info(os, thread)("Thread started (pthread id: " UINTX_FORMAT ", attributes: %s). ",
+        (uintx) tid, os::Posix::describe_pthread_attr(buf, sizeof(buf), &attr));
+    } else {
+      log_warning(os, thread)("Failed to start thread - pthread_create failed (%s) for attributes: %s.",
+        strerror(ret), os::Posix::describe_pthread_attr(buf, sizeof(buf), &attr));
+    }
+
     pthread_attr_destroy(&attr);
 
     if (ret != 0) {
-      if (PrintMiscellaneous && (Verbose || WizardMode)) {
-        perror("pthread_create()");
-      }
       // Need to clean up stuff we've allocated so far
       thread->set_osthread(NULL);
       delete osthread;
@@ -858,6 +870,9 @@ bool os::create_attached_thread(JavaThread* thread) {
   // and save the caller's signal mask
   os::Linux::hotspot_sigmask(thread);
 
+  log_info(os, thread)("Thread attached (tid: " UINTX_FORMAT ", pthread id: " UINTX_FORMAT ").",
+    os::current_thread_id(), (uintx) pthread_self());
+
   return true;
 }
 

hotspot/src/os/posix/vm/os_posix.cpp

@@ -1071,6 +1071,19 @@ void os::Posix::ucontext_set_pc(ucontext_t* ctx, address pc) {
 #endif
 }
 
+char* os::Posix::describe_pthread_attr(char* buf, size_t buflen, const pthread_attr_t* attr) {
+  size_t stack_size = 0;
+  size_t guard_size = 0;
+  int detachstate = 0;
+  pthread_attr_getstacksize(attr, &stack_size);
+  pthread_attr_getguardsize(attr, &guard_size);
+  pthread_attr_getdetachstate(attr, &detachstate);
+  jio_snprintf(buf, buflen, "stacksize: " SIZE_FORMAT "k, guardsize: " SIZE_FORMAT "k, %s",
+    stack_size / 1024, guard_size / 1024,
+    (detachstate == PTHREAD_CREATE_DETACHED ? "detached" : "joinable"));
+  return buf;
+}
+
 
 os::WatcherThreadCrashProtection::WatcherThreadCrashProtection() {
   assert(Thread::current()->is_Watcher_thread(), "Must be WatcherThread");

hotspot/src/os/posix/vm/os_posix.hpp

@@ -76,6 +76,11 @@ public:
   static address ucontext_get_pc(const ucontext_t* ctx);
   // Set PC into context. Needed for continuation after signal.
   static void ucontext_set_pc(ucontext_t* ctx, address pc);
+
+  // Helper function; describes pthread attributes as short string. String is written
+  // to buf with len buflen; buf is returned.
+  static char* describe_pthread_attr(char* buf, size_t buflen, const pthread_attr_t* attr);
+
 };
 
 /*

hotspot/src/os/solaris/vm/os_solaris.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2015, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2016, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -32,6 +32,7 @@
 #include "compiler/disassembler.hpp"
 #include "interpreter/interpreter.hpp"
 #include "jvm_solaris.h"
+#include "logging/log.hpp"
 #include "memory/allocation.inline.hpp"
 #include "memory/filemap.hpp"
 #include "mutex_solaris.inline.hpp"
@@ -68,6 +69,7 @@
 #include "utilities/defaultStream.hpp"
 #include "utilities/events.hpp"
 #include "utilities/growableArray.hpp"
+#include "utilities/macros.hpp"
 #include "utilities/vmError.hpp"
 
 // put OS-includes here
@@ -736,6 +738,9 @@ extern "C" void* java_start(void* thread_addr) {
   osthr->set_lwp_id(_lwp_self());  // Store lwp in case we are bound
   thread->_schedctl = (void *) schedctl_init();
 
+  log_info(os, thread)("Thread is alive (tid: " UINTX_FORMAT ").",
+    os::current_thread_id());
+
   if (UseNUMA) {
     int lgrp_id = os::numa_get_group_id();
     if (lgrp_id != -1) {
@@ -781,6 +786,8 @@ extern "C" void* java_start(void* thread_addr) {
     Atomic::dec(&os::Solaris::_os_thread_count);
   }
 
+  log_info(os, thread)("Thread finished (tid: " UINTX_FORMAT ").", os::current_thread_id());
+
   if (UseDetachedThreads) {
     thr_exit(NULL);
     ShouldNotReachHere();
@@ -853,6 +860,9 @@ bool os::create_attached_thread(JavaThread* thread) {
   // and save the caller's signal mask
   os::Solaris::hotspot_sigmask(thread);
 
+  log_info(os, thread)("Thread attached (tid: " UINTX_FORMAT ").",
+    os::current_thread_id());
+
   return true;
 }
 
@@ -879,6 +889,24 @@ bool os::create_main_thread(JavaThread* thread) {
   return true;
 }
 
+// Helper function to trace thread attributes, similar to os::Posix::describe_pthread_attr()
+static char* describe_thr_create_attributes(char* buf, size_t buflen,
+                                            size_t stacksize, long flags) {
+  stringStream ss(buf, buflen);
+  ss.print("stacksize: " SIZE_FORMAT "k, ", stacksize / 1024);
+  ss.print("flags: ");
+  #define PRINT_FLAG(f) if (flags & f) ss.print( #f " ");
+  #define ALL(X) \
+    X(THR_SUSPENDED) \
+    X(THR_DETACHED) \
+    X(THR_BOUND) \
+    X(THR_NEW_LWP) \
+    X(THR_DAEMON)
+  ALL(PRINT_FLAG)
+  #undef ALL
+  #undef PRINT_FLAG
+  return buf;
+}
 
 bool os::create_thread(Thread* thread, ThreadType thr_type,
                        size_t stack_size) {
@@ -974,10 +1002,17 @@ bool os::create_thread(Thread* thread, ThreadType thr_type,
   osthread->set_thread_id(-1);
 
   status = thr_create(NULL, stack_size, java_start, thread, flags, &tid);
+
+  char buf[64];
+  if (status == 0) {
+    log_info(os, thread)("Thread started (tid: " UINTX_FORMAT ", attributes: %s). ",
+      (uintx) tid, describe_thr_create_attributes(buf, sizeof(buf), stack_size, flags));
+  } else {
+    log_warning(os, thread)("Failed to start thread - thr_create failed (%s) for attributes: %s.",
+      strerror(status), describe_thr_create_attributes(buf, sizeof(buf), stack_size, flags));
+  }
+
   if (status != 0) {
-    if (PrintMiscellaneous && (Verbose || WizardMode)) {
-      perror("os::create_thread");
-    }
     thread->set_osthread(NULL);
     // Need to clean up stuff we've allocated so far
     delete osthread;

hotspot/src/os/windows/vm/os_windows.cpp

@@ -35,6 +35,7 @@
 #include "compiler/disassembler.hpp"
 #include "interpreter/interpreter.hpp"
 #include "jvm_windows.h"
+#include "logging/log.hpp"
 #include "memory/allocation.inline.hpp"
 #include "memory/filemap.hpp"
 #include "mutex_windows.inline.hpp"
@@ -71,6 +72,7 @@
 #include "utilities/defaultStream.hpp"
 #include "utilities/events.hpp"
 #include "utilities/growableArray.hpp"
+#include "utilities/macros.hpp"
 #include "utilities/vmError.hpp"
 
 #ifdef _DEBUG
@@ -436,6 +438,8 @@ static unsigned __stdcall java_start(Thread* thread) {
     res = 20115;    // java thread
   }
 
+  log_info(os, thread)("Thread is alive (tid: " UINTX_FORMAT ").", os::current_thread_id());
+
   // Install a win32 structured exception handler around every thread created
   // by VM, so VM can generate error dump when an exception occurred in non-
   // Java thread (e.g. VM thread).
@@ -446,6 +450,8 @@ static unsigned __stdcall java_start(Thread* thread) {
     // Nothing to do.
   }
 
+  log_info(os, thread)("Thread finished (tid: " UINTX_FORMAT ").", os::current_thread_id());
+
   // One less thread is executing
   // When the VMThread gets here, the main thread may have already exited
   // which frees the CodeHeap containing the Atomic::add code
@@ -509,6 +515,10 @@ bool os::create_attached_thread(JavaThread* thread) {
   osthread->set_state(RUNNABLE);
 
   thread->set_osthread(osthread);
+
+  log_info(os, thread)("Thread attached (tid: " UINTX_FORMAT ").",
+    os::current_thread_id());
+
   return true;
 }
 
@@ -530,6 +540,27 @@ bool os::create_main_thread(JavaThread* thread) {
   return true;
 }
 
+// Helper function to trace _beginthreadex attributes,
+//  similar to os::Posix::describe_pthread_attr()
+static char* describe_beginthreadex_attributes(char* buf, size_t buflen,
+                                               size_t stacksize, unsigned initflag) {
+  stringStream ss(buf, buflen);
+  if (stacksize == 0) {
+    ss.print("stacksize: default, ");
+  } else {
+    ss.print("stacksize: " SIZE_FORMAT "k, ", stacksize / 1024);
+  }
+  ss.print("flags: ");
+  #define PRINT_FLAG(f) if (initflag & f) ss.print( #f " ");
+  #define ALL(X) \
+    X(CREATE_SUSPENDED) \
+    X(STACK_SIZE_PARAM_IS_A_RESERVATION)
+  ALL(PRINT_FLAG)
+  #undef ALL
+  #undef PRINT_FLAG
+  return buf;
+}
+
 // Allocate and initialize a new OSThread
 bool os::create_thread(Thread* thread, ThreadType thr_type,
                        size_t stack_size) {
@@ -596,14 +627,24 @@ bool os::create_thread(Thread* thread, ThreadType thr_type,
   // document because JVM uses C runtime library. The good news is that the
   // flag appears to work with _beginthredex() as well.
 
+  const unsigned initflag = CREATE_SUSPENDED | STACK_SIZE_PARAM_IS_A_RESERVATION;
   HANDLE thread_handle =
     (HANDLE)_beginthreadex(NULL,
                            (unsigned)stack_size,
                            (unsigned (__stdcall *)(void*)) java_start,
                            thread,
-                           CREATE_SUSPENDED | STACK_SIZE_PARAM_IS_A_RESERVATION,
+                           initflag,
                            &thread_id);
 
+  char buf[64];
+  if (thread_handle != NULL) {
+    log_info(os, thread)("Thread started (tid: %u, attributes: %s)",
+      thread_id, describe_beginthreadex_attributes(buf, sizeof(buf), stack_size, initflag));
+  } else {
+    log_warning(os, thread)("Failed to start thread - _beginthreadex failed (%s) for attributes: %s.",
+      strerror(errno), describe_beginthreadex_attributes(buf, sizeof(buf), stack_size, initflag));
+  }
+
   if (thread_handle == NULL) {
     // Need to clean up stuff we've allocated so far
     CloseHandle(osthread->interrupt_event());
@@ -1668,8 +1709,7 @@ void os::win32::print_windows_version(outputStream* st) {
     if (is_workstation) {
       st->print("10");
     } else {
-      // The server version name of Windows 10 is not known at this time
+      st->print("Server 2016");
-      st->print("%d.%d", major_version, minor_version);
     }
     break;
 

hotspot/src/os_cpu/solaris_sparc/vm/vm_version_solaris_sparc.cpp

@@ -264,6 +264,7 @@ void PICL::close_library() {
 
 // We need to keep these here as long as we have to build on Solaris
 // versions before 10.
+
 #ifndef SI_ARCHITECTURE_32
 #define SI_ARCHITECTURE_32      516     /* basic 32-bit SI_ARCHITECTURE */
 #endif
@@ -272,36 +273,87 @@ void PICL::close_library() {
 #define SI_ARCHITECTURE_64      517     /* basic 64-bit SI_ARCHITECTURE */
 #endif
 
-static void do_sysinfo(int si, const char* string, int* features, int mask) {
+#ifndef SI_CPUBRAND
-  char   tmp;
+#define SI_CPUBRAND             523     /* return cpu brand string */
-  size_t bufsize = sysinfo(si, &tmp, 1);
+#endif
 
-  // All SI defines used below must be supported.
+class Sysinfo {
-  guarantee(bufsize != -1, "must be supported");
+  char* _string;
+public:
+  Sysinfo(int si) : _string(NULL) {
+    char   tmp;
+    size_t bufsize = sysinfo(si, &tmp, 1);
 
-  char* buf = (char*) os::malloc(bufsize, mtInternal);
+    if (bufsize != -1) {
-
+      char* buf = (char*) os::malloc(bufsize, mtInternal);
-  if (buf == NULL)
+      if (buf != NULL) {
-    return;
+        if (sysinfo(si, buf, bufsize) == bufsize) {
-
+          _string = buf;
-  if (sysinfo(si, buf, bufsize) == bufsize) {
+        } else {
-    // Compare the string.
+          os::free(buf);
-    if (strcmp(buf, string) == 0) {
+        }
-      *features |= mask;
+      }
     }
   }
 
-  os::free(buf);
+  ~Sysinfo() {
-}
+    if (_string != NULL) {
+      os::free(_string);
+    }
+  }
+
+  const char* value() const {
+    return _string;
+  }
+
+  bool valid() const {
+    return _string != NULL;
+  }
+
+  bool match(const char* s) const {
+    return valid() ? strcmp(_string, s) == 0 : false;
+  }
+
+  bool match_substring(const char* s) const {
+    return valid() ? strstr(_string, s) != NULL : false;
+  }
+};
+
+class Sysconf {
+  int _value;
+public:
+  Sysconf(int sc) : _value(-1) {
+    _value = sysconf(sc);
+  }
+  bool valid() const {
+    return _value != -1;
+  }
+  int value() const {
+    return _value;
+  }
+};
+
+
+#ifndef _SC_DCACHE_LINESZ
+#define _SC_DCACHE_LINESZ       508     /* Data cache line size */
+#endif
+
+#ifndef _SC_L2CACHE_LINESZ
+#define _SC_L2CACHE_LINESZ      527     /* Size of L2 cache line */
+#endif
 
 int VM_Version::platform_features(int features) {
   assert(os::Solaris::supports_getisax(), "getisax() must be available");
 
   // Check 32-bit architecture.
-  do_sysinfo(SI_ARCHITECTURE_32, "sparc", &features, v8_instructions_m);
+  if (Sysinfo(SI_ARCHITECTURE_32).match("sparc")) {
+    features |= v8_instructions_m;
+  }
 
   // Check 64-bit architecture.
-  do_sysinfo(SI_ARCHITECTURE_64, "sparcv9", &features, generic_v9_m);
+  if (Sysinfo(SI_ARCHITECTURE_64).match("sparcv9")) {
+    features |= generic_v9_m;
+  }
 
   // Extract valid instruction set extensions.
   uint_t avs[2];
@@ -388,67 +440,63 @@ int VM_Version::platform_features(int features) {
   if (av & AV_SPARC_SHA512)       features |= sha512_instruction_m;
 
   // Determine the machine type.
-  do_sysinfo(SI_MACHINE, "sun4v", &features, sun4v_m);
+  if (Sysinfo(SI_MACHINE).match("sun4v")) {
+    features |= sun4v_m;
+  }
 
-  {
+  bool use_solaris_12_api = false;
-    // Using kstat to determine the machine type.
+  Sysinfo impl(SI_CPUBRAND);
+  if (impl.valid()) {
+    // If SI_CPUBRAND works, that means Solaris 12 API to get the cache line sizes
+    // is available to us as well
+    use_solaris_12_api = true;
+    features |= parse_features(impl.value());
+  } else {
+    // Otherwise use kstat to determine the machine type.
     kstat_ctl_t* kc = kstat_open();
     kstat_t* ksp = kstat_lookup(kc, (char*)"cpu_info", -1, NULL);
-    const char* implementation = "UNKNOWN";
+    const char* implementation;
+    bool has_implementation = false;
     if (ksp != NULL) {
       if (kstat_read(kc, ksp, NULL) != -1 && ksp->ks_data != NULL) {
         kstat_named_t* knm = (kstat_named_t *)ksp->ks_data;
         for (int i = 0; i < ksp->ks_ndata; i++) {
           if (strcmp((const char*)&(knm[i].name),"implementation") == 0) {
             implementation = KSTAT_NAMED_STR_PTR(&knm[i]);
+            has_implementation = true;
 #ifndef PRODUCT
             if (PrintMiscellaneous && Verbose) {
               tty->print_cr("cpu_info.implementation: %s", implementation);
             }
 #endif
-            // Convert to UPPER case before compare.
+            features |= parse_features(implementation);
-            char* impl = os::strdup_check_oom(implementation);
-
-            for (int i = 0; impl[i] != 0; i++)
-              impl[i] = (char)toupper((uint)impl[i]);
-
-            if (strstr(impl, "SPARC64") != NULL) {
-              features |= sparc64_family_m;
-            } else if (strstr(impl, "SPARC-M") != NULL) {
-              // M-series SPARC is based on T-series.
-              features |= (M_family_m | T_family_m);
-            } else if (strstr(impl, "SPARC-T") != NULL) {
-              features |= T_family_m;
-              if (strstr(impl, "SPARC-T1") != NULL) {
-                features |= T1_model_m;
-              }
-            } else {
-              if (strstr(impl, "SPARC") == NULL) {
-#ifndef PRODUCT
-                // kstat on Solaris 8 virtual machines (branded zones)
-                // returns "(unsupported)" implementation. Solaris 8 is not
-                // supported anymore, but include this check to be on the
-                // safe side.
-                warning("kstat cpu_info implementation = '%s', assume generic SPARC", impl);
-#endif
-                implementation = "SPARC";
-              }
-            }
-            os::free((void*)impl);
             break;
           }
         } // for(
       }
     }
-    assert(strcmp(implementation, "UNKNOWN") != 0,
+    assert(has_implementation, "unknown cpu info (changed kstat interface?)");
-           "unknown cpu info (changed kstat interface?)");
     kstat_close(kc);
   }
 
-  // Figure out cache line sizes using PICL
+  bool is_sun4v = (features & sun4v_m) != 0;
-  PICL picl((features & sparc64_family_m) != 0, (features & sun4v_m) != 0);
+  if (use_solaris_12_api && is_sun4v) {
-  _L1_data_cache_line_size = picl.L1_data_cache_line_size();
+    // If Solaris 12 API is supported and it's sun4v use sysconf() to get the cache line sizes
-  _L2_data_cache_line_size = picl.L2_data_cache_line_size();
+    Sysconf l1_dcache_line_size(_SC_DCACHE_LINESZ);
+    if (l1_dcache_line_size.valid()) {
+      _L1_data_cache_line_size =  l1_dcache_line_size.value();
+    }
 
+    Sysconf l2_dcache_line_size(_SC_L2CACHE_LINESZ);
+    if (l2_dcache_line_size.valid()) {
+      _L2_data_cache_line_size = l2_dcache_line_size.value();
+    }
+  } else {
+    // Otherwise figure out the cache line sizes using PICL
+    bool is_fujitsu = (features & sparc64_family_m) != 0;
+    PICL picl(is_fujitsu, is_sun4v);
+    _L1_data_cache_line_size = picl.L1_data_cache_line_size();
+    _L2_data_cache_line_size = picl.L2_data_cache_line_size();
+  }
   return features;
 }

hotspot/src/os_cpu/windows_x86/vm/assembler_windows_x86.cpp

@@ -51,11 +51,8 @@ void MacroAssembler::int3() {
 // movl reg, [reg + thread_ptr_offset]     Load thread
 //
 void MacroAssembler::get_thread(Register thread) {
-  // can't use ExternalAddress because it can't take NULL
-  AddressLiteral null(0, relocInfo::none);
-
   prefix(FS_segment);
-  movptr(thread, null);
+  movptr(thread, ExternalAddress(NULL));
   assert(os::win32::get_thread_ptr_offset() != 0,
          "Thread Pointer Offset has not been initialized");
   movl(thread, Address(thread, os::win32::get_thread_ptr_offset()));

hotspot/src/share/vm/adlc/formssel.cpp

@@ -3491,6 +3491,8 @@ int MatchNode::needs_ideal_memory_edge(FormDict &globals) const {
     "LoadPLocked",
     "StorePConditional", "StoreIConditional", "StoreLConditional",
     "CompareAndSwapI", "CompareAndSwapL", "CompareAndSwapP", "CompareAndSwapN",
+    "WeakCompareAndSwapI", "WeakCompareAndSwapL", "WeakCompareAndSwapP", "WeakCompareAndSwapN",
+    "CompareAndExchangeI", "CompareAndExchangeL", "CompareAndExchangeP", "CompareAndExchangeN",
     "StoreCM",
     "ClearArray",
     "GetAndAddI", "GetAndSetI", "GetAndSetP",

hotspot/src/share/vm/c1/c1_CFGPrinter.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2005, 2016, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -34,7 +34,9 @@
 // compilation for later analysis.
 
 class CFGPrinterOutput;
-class IntervalList;
+class Interval;
+
+typedef GrowableArray<Interval*> IntervalList;
 
 class CFGPrinter : public AllStatic {
 private:

hotspot/src/share/vm/c1/c1_Canonicalizer.cpp

@@ -222,27 +222,36 @@ void Canonicalizer::do_StoreField     (StoreField*      x) {
 }
 
 void Canonicalizer::do_ArrayLength    (ArrayLength*     x) {
-  NewArray* array = x->array()->as_NewArray();
+  NewArray*  na;
-  if (array != NULL && array->length() != NULL) {
+  Constant*  ct;
-    Constant* length = array->length()->as_Constant();
+  LoadField* lf;
-    if (length != NULL) {
+
-      // do not use the Constant itself, but create a new Constant
+  if ((na = x->array()->as_NewArray()) != NULL) {
-      // with same value Otherwise a Constant is live over multiple
+    // New arrays might have the known length.
-      // blocks without being registered in a state array.
+    // Do not use the Constant itself, but create a new Constant
+    // with same value Otherwise a Constant is live over multiple
+    // blocks without being registered in a state array.
+    Constant* length;
+    if (na->length() != NULL &&
+        (length = na->length()->as_Constant()) != NULL) {
       assert(length->type()->as_IntConstant() != NULL, "array length must be integer");
       set_constant(length->type()->as_IntConstant()->value());
     }
-  } else {
+
-    LoadField* lf = x->array()->as_LoadField();
+  } else if ((ct = x->array()->as_Constant()) != NULL) {
-    if (lf != NULL) {
+    // Constant arrays have constant lengths.
-      ciField* field = lf->field();
+    ArrayConstant* cnst = ct->type()->as_ArrayConstant();
-      if (field->is_constant() && field->is_static()) {
+    if (cnst != NULL) {
-        // final static field
+      set_constant(cnst->value()->length());
-        ciObject* c = field->constant_value().as_object();
+    }
-        if (c->is_array()) {
+
-          ciArray* array = (ciArray*) c;
+  } else if ((lf = x->array()->as_LoadField()) != NULL) {
-          set_constant(array->length());
+    ciField* field = lf->field();
-        }
+    if (field->is_constant() && field->is_static()) {
+      assert(PatchALot || ScavengeRootsInCode < 2, "Constant field loads are folded during parsing");
+      ciObject* c = field->constant_value().as_object();
+      if (!c->is_null_object()) {
+        set_constant(c->as_array()->length());
       }
     }
   }

hotspot/src/share/vm/c1/c1_Compiler.cpp

@@ -228,8 +228,6 @@ bool Compiler::is_intrinsic_supported(const methodHandle& method) {
   case vmIntrinsics::_getCharStringU:
   case vmIntrinsics::_putCharStringU:
 #ifdef TRACE_HAVE_INTRINSICS
-  case vmIntrinsics::_classID:
-  case vmIntrinsics::_threadID:
   case vmIntrinsics::_counterTime:
 #endif
     break;

hotspot/src/share/vm/c1/c1_LIRGenerator.cpp

@@ -43,6 +43,9 @@
 #if INCLUDE_ALL_GCS
 #include "gc/g1/heapRegion.hpp"
 #endif // INCLUDE_ALL_GCS
+#ifdef TRACE_HAVE_INTRINSICS
+#include "trace/traceMacros.hpp"
+#endif
 
 #ifdef ASSERT
 #define __ gen()->lir(__FILE__, __LINE__)->
@@ -3067,42 +3070,7 @@ void LIRGenerator::do_RuntimeCall(address routine, Intrinsic* x) {
   __ move(reg, result);
 }
 
-#ifdef TRACE_HAVE_INTRINSICS
-void LIRGenerator::do_ThreadIDIntrinsic(Intrinsic* x) {
-    LIR_Opr thread = getThreadPointer();
-    LIR_Opr osthread = new_pointer_register();
-    __ move(new LIR_Address(thread, in_bytes(JavaThread::osthread_offset()), osthread->type()), osthread);
-    size_t thread_id_size = OSThread::thread_id_size();
-    if (thread_id_size == (size_t) BytesPerLong) {
-      LIR_Opr id = new_register(T_LONG);
-      __ move(new LIR_Address(osthread, in_bytes(OSThread::thread_id_offset()), T_LONG), id);
-      __ convert(Bytecodes::_l2i, id, rlock_result(x));
-    } else if (thread_id_size == (size_t) BytesPerInt) {
-      __ move(new LIR_Address(osthread, in_bytes(OSThread::thread_id_offset()), T_INT), rlock_result(x));
-    } else {
-      ShouldNotReachHere();
-    }
-}
 
-void LIRGenerator::do_ClassIDIntrinsic(Intrinsic* x) {
-    CodeEmitInfo* info = state_for(x);
-    CodeEmitInfo* info2 = new CodeEmitInfo(info); // Clone for the second null check
-    BasicType klass_pointer_type = NOT_LP64(T_INT) LP64_ONLY(T_LONG);
-    assert(info != NULL, "must have info");
-    LIRItem arg(x->argument_at(1), this);
-    arg.load_item();
-    LIR_Opr klass = new_pointer_register();
-    __ move(new LIR_Address(arg.result(), java_lang_Class::klass_offset_in_bytes(), klass_pointer_type), klass, info);
-    LIR_Opr id = new_register(T_LONG);
-    ByteSize offset = TRACE_ID_OFFSET;
-    LIR_Address* trace_id_addr = new LIR_Address(klass, in_bytes(offset), T_LONG);
-    __ move(trace_id_addr, id);
-    __ logical_or(id, LIR_OprFact::longConst(0x01l), id);
-    __ store(id, trace_id_addr);
-    __ logical_and(id, LIR_OprFact::longConst(~0x3l), id);
-    __ move(id, rlock_result(x));
-}
-#endif
 
 void LIRGenerator::do_Intrinsic(Intrinsic* x) {
   switch (x->id()) {
@@ -3115,8 +3083,6 @@ void LIRGenerator::do_Intrinsic(Intrinsic* x) {
   }
 
 #ifdef TRACE_HAVE_INTRINSICS
-  case vmIntrinsics::_threadID: do_ThreadIDIntrinsic(x); break;
-  case vmIntrinsics::_classID: do_ClassIDIntrinsic(x); break;
   case vmIntrinsics::_counterTime:
     do_RuntimeCall(CAST_FROM_FN_PTR(address, TRACE_TIME_METHOD), x);
     break;

hotspot/src/share/vm/c1/c1_LIRGenerator.hpp

@@ -440,10 +440,7 @@ class LIRGenerator: public InstructionVisitor, public BlockClosure {
   void do_SwitchRanges(SwitchRangeArray* x, LIR_Opr value, BlockBegin* default_sux);
 
   void do_RuntimeCall(address routine, Intrinsic* x);
-#ifdef TRACE_HAVE_INTRINSICS
+
-  void do_ThreadIDIntrinsic(Intrinsic* x);
-  void do_ClassIDIntrinsic(Intrinsic* x);
-#endif
   ciKlass* profile_type(ciMethodData* md, int md_first_offset, int md_offset, intptr_t profiled_k,
                         Value arg, LIR_Opr& mdp, bool not_null, ciKlass* signature_at_call_k,
                         ciKlass* callee_signature_k);

hotspot/src/share/vm/c1/c1_LinearScan.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2005, 2015, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2005, 2016, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -1434,42 +1434,74 @@ int LinearScan::interval_cmp(Interval** a, Interval** b) {
 }
 
 #ifndef PRODUCT
-bool LinearScan::is_sorted(IntervalArray* intervals) {
+int interval_cmp(Interval* const& l, Interval* const& r) {
-  int from = -1;
+  return l->from() - r->from();
-  int i, j;
+}
-  for (i = 0; i < intervals->length(); i ++) {
+
-    Interval* it = intervals->at(i);
+bool find_interval(Interval* interval, IntervalArray* intervals) {
-    if (it != NULL) {
+  bool found;
-      if (from > it->from()) {
+  int idx = intervals->find_sorted<Interval*, interval_cmp>(interval, found);
-        assert(false, "");
+
-        return false;
+  if (!found) {
-      }
+    return false;
-      from = it->from();
+  }
+
+  int from = interval->from();
+
+  // The index we've found using binary search is pointing to an interval
+  // that is defined in the same place as the interval we were looking for.
+  // So now we have to look around that index and find exact interval.
+  for (int i = idx; i >= 0; i--) {
+    if (intervals->at(i) == interval) {
+      return true;
+    }
+    if (intervals->at(i)->from() != from) {
+      break;
     }
   }
 
-  // check in both directions if sorted list and unsorted list contain same intervals
+  for (int i = idx + 1; i < intervals->length(); i++) {
-  for (i = 0; i < interval_count(); i++) {
+    if (intervals->at(i) == interval) {
-    if (interval_at(i) != NULL) {
+      return true;
-      int num_found = 0;
+    }
-      for (j = 0; j < intervals->length(); j++) {
+    if (intervals->at(i)->from() != from) {
-        if (interval_at(i) == intervals->at(j)) {
+      break;
-          num_found++;
-        }
-      }
-      assert(num_found == 1, "lists do not contain same intervals");
     }
   }
-  for (j = 0; j < intervals->length(); j++) {
+
-    int num_found = 0;
+  return false;
-    for (i = 0; i < interval_count(); i++) {
+}
-      if (interval_at(i) == intervals->at(j)) {
+
-        num_found++;
+bool LinearScan::is_sorted(IntervalArray* intervals) {
-      }
+  int from = -1;
+  int null_count = 0;
+
+  for (int i = 0; i < intervals->length(); i++) {
+    Interval* it = intervals->at(i);
+    if (it != NULL) {
+      assert(from <= it->from(), "Intervals are unordered");
+      from = it->from();
+    } else {
+      null_count++;
     }
-    assert(num_found == 1, "lists do not contain same intervals");
   }
 
+  assert(null_count == 0, "Sorted intervals should not contain nulls");
+
+  null_count = 0;
+
+  for (int i = 0; i < interval_count(); i++) {
+    Interval* interval = interval_at(i);
+    if (interval != NULL) {
+      assert(find_interval(interval, intervals), "Lists do not contain same intervals");
+    } else {
+      null_count++;
+    }
+  }
+
+  assert(interval_count() - null_count == intervals->length(),
+      "Sorted list should contain the same amount of non-NULL intervals as unsorted list");
+
   return true;
 }
 #endif
@@ -1536,7 +1568,7 @@ void LinearScan::sort_intervals_before_allocation() {
       sorted_len++;
     }
   }
-  IntervalArray* sorted_list = new IntervalArray(sorted_len);
+  IntervalArray* sorted_list = new IntervalArray(sorted_len, sorted_len, NULL);
 
   // special sorting algorithm: the original interval-list is almost sorted,
   // only some intervals are swapped. So this is much faster than a complete QuickSort
@@ -1574,8 +1606,8 @@ void LinearScan::sort_intervals_after_allocation() {
     _needs_full_resort = false;
   }
 
-  IntervalArray* old_list      = _sorted_intervals;
+  IntervalArray* old_list = _sorted_intervals;
-  IntervalList*  new_list      = _new_intervals_from_allocation;
+  IntervalList* new_list = _new_intervals_from_allocation;
   int old_len = old_list->length();
   int new_len = new_list->length();
 
@@ -1589,7 +1621,8 @@ void LinearScan::sort_intervals_after_allocation() {
   new_list->sort(interval_cmp);
 
   // merge old and new list (both already sorted) into one combined list
-  IntervalArray* combined_list = new IntervalArray(old_len + new_len);
+  int combined_list_len = old_len + new_len;
+  IntervalArray* combined_list = new IntervalArray(combined_list_len, combined_list_len, NULL);
   int old_idx = 0;
   int new_idx = 0;
 
@@ -3211,6 +3244,10 @@ void LinearScan::verify_intervals() {
       has_error = true;
     }
 
+    // special intervals that are created in MoveResolver
+    // -> ignore them because the range information has no meaning there
+    if (i1->from() == 1 && i1->to() == 2) continue;
+
     if (i1->first() == Range::end()) {
       tty->print_cr("Interval %d has no Range", i1->reg_num()); i1->print(); tty->cr();
       has_error = true;
@@ -3225,18 +3262,13 @@ void LinearScan::verify_intervals() {
 
     for (int j = i + 1; j < len; j++) {
       Interval* i2 = interval_at(j);
-      if (i2 == NULL) continue;
+      if (i2 == NULL || (i2->from() == 1 && i2->to() == 2)) continue;
-
-      // special intervals that are created in MoveResolver
-      // -> ignore them because the range information has no meaning there
-      if (i1->from() == 1 && i1->to() == 2) continue;
-      if (i2->from() == 1 && i2->to() == 2) continue;
 
       int r1 = i1->assigned_reg();
       int r1Hi = i1->assigned_regHi();
       int r2 = i2->assigned_reg();
       int r2Hi = i2->assigned_regHi();
-      if (i1->intersects(i2) && (r1 == r2 || r1 == r2Hi || (r1Hi != any_reg && (r1Hi == r2 || r1Hi == r2Hi)))) {
+      if ((r1 == r2 || r1 == r2Hi || (r1Hi != any_reg && (r1Hi == r2 || r1Hi == r2Hi))) && i1->intersects(i2)) {
         tty->print_cr("Intervals %d and %d overlap and have the same register assigned", i1->reg_num(), i2->reg_num());
         i1->print(); tty->cr();
         i2->print(); tty->cr();
@@ -3429,7 +3461,8 @@ void LinearScan::verify_registers() {
 
 void RegisterVerifier::verify(BlockBegin* start) {
   // setup input registers (method arguments) for first block
-  IntervalList* input_state = new IntervalList(state_size(), NULL);
+  int input_state_len = state_size();
+  IntervalList* input_state = new IntervalList(input_state_len, input_state_len, NULL);
   CallingConvention* args = compilation()->frame_map()->incoming_arguments();
   for (int n = 0; n < args->length(); n++) {
     LIR_Opr opr = args->at(n);
@@ -3543,7 +3576,7 @@ void RegisterVerifier::process_successor(BlockBegin* block, IntervalList* input_
 
 IntervalList* RegisterVerifier::copy(IntervalList* input_state) {
   IntervalList* copy_state = new IntervalList(input_state->length());
-  copy_state->push_all(input_state);
+  copy_state->appendAll(input_state);
   return copy_state;
 }
 
@@ -5506,7 +5539,7 @@ void LinearScanWalker::split_and_spill_intersecting_intervals(int reg, int regHi
     IntervalList* processed = _spill_intervals[reg];
     for (int i = 0; i < _spill_intervals[regHi]->length(); i++) {
       Interval* it = _spill_intervals[regHi]->at(i);
-      if (processed->index_of(it) == -1) {
+      if (processed->find_from_end(it) == -1) {
         remove_from_list(it);
         split_and_spill_interval(it);
       }

hotspot/src/share/vm/c1/c1_LinearScan.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2005, 2012, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2005, 2016, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -42,8 +42,8 @@ class LinearScan;
 class MoveResolver;
 class Range;
 
-define_array(IntervalArray, Interval*)
+typedef GrowableArray<Interval*> IntervalArray;
-define_stack(IntervalList, IntervalArray)
+typedef GrowableArray<Interval*> IntervalList;
 
 define_array(IntervalsArray, IntervalList*)
 define_stack(IntervalsList, IntervalsArray)

hotspot/src/share/vm/ci/ciArray.cpp

@@ -107,8 +107,9 @@ ciConstant ciArray::element_value_by_offset(intptr_t element_offset) {
   intptr_t header = arrayOopDesc::base_offset_in_bytes(elembt);
   intptr_t index = (element_offset - header) >> shift;
   intptr_t offset = header + ((intptr_t)index << shift);
-  if (offset != element_offset || index != (jint)index)
+  if (offset != element_offset || index != (jint)index || index < 0 || index >= length()) {
     return ciConstant();
+  }
   return element_value((jint) index);
 }
 

hotspot/src/share/vm/ci/ciConstantPoolCache.cpp

@@ -41,58 +41,37 @@ ciConstantPoolCache::ciConstantPoolCache(Arena* arena,
   _keys = new (arena) GrowableArray<int>(arena, expected_size, 0, 0);
 }
 
+int ciConstantPoolCache::key_compare(const int& key, const int& elt) {
+  if (key < elt)      return -1;
+  else if (key > elt) return 1;
+  else                  return 0;
+}
+
 // ------------------------------------------------------------------
 // ciConstantPoolCache::get
 //
 // Get the entry at some index
 void* ciConstantPoolCache::get(int index) {
   ASSERT_IN_VM;
-  int pos = find(index);
+  bool found = false;
-  if (pos >= _keys->length() ||
+  int pos = _keys->find_sorted<int, ciConstantPoolCache::key_compare>(index, found);
-      _keys->at(pos) != index) {
+  if (!found) {
     // This element is not present in the cache.
     return NULL;
   }
   return _elements->at(pos);
 }
 
-// ------------------------------------------------------------------
-// ciConstantPoolCache::find
-//
-// Use binary search to find the position of this index in the cache.
-// If there is no entry in the cache corresponding to this oop, return
-// the position at which the index would be inserted.
-int ciConstantPoolCache::find(int key) {
-  int min = 0;
-  int max = _keys->length()-1;
-
-  while (max >= min) {
-    int mid = (max + min) / 2;
-    int value = _keys->at(mid);
-    if (value < key) {
-      min = mid + 1;
-    } else if (value > key) {
-      max = mid - 1;
-    } else {
-      return mid;
-    }
-  }
-  return min;
-}
-
 // ------------------------------------------------------------------
 // ciConstantPoolCache::insert
 //
 // Insert a ciObject into the table at some index.
 void ciConstantPoolCache::insert(int index, void* elem) {
-  int i;
+  bool found = false;
-  int pos = find(index);
+  int pos = _keys->find_sorted<int, ciConstantPoolCache::key_compare>(index, found);
-  for (i = _keys->length()-1; i >= pos; i--) {
+  assert(!found, "duplicate");
-    _keys->at_put_grow(i+1, _keys->at(i));
+  _keys->insert_before(pos, index);
-    _elements->at_put_grow(i+1, _elements->at(i));
+  _elements->insert_before(pos, elem);
-  }
-  _keys->at_put_grow(pos, index);
-  _elements->at_put_grow(pos, elem);
 }
 
 // ------------------------------------------------------------------

hotspot/src/share/vm/ci/ciConstantPoolCache.hpp

@@ -38,7 +38,7 @@ private:
   GrowableArray<int>*   _keys;
   GrowableArray<void*>* _elements;
 
-  int find(int index);
+  static int key_compare(const int& key, const int& elt);
 
 public:
   ciConstantPoolCache(Arena* arena, int expected_size);

hotspot/src/share/vm/ci/ciObjectFactory.cpp

@@ -260,6 +260,13 @@ ciObject* ciObjectFactory::get(oop key) {
   return new_object;
 }
 
+int ciObjectFactory::metadata_compare(Metadata* const& key, ciMetadata* const& elt) {
+  Metadata* value = elt->constant_encoding();
+  if (key < value)      return -1;
+  else if (key > value) return 1;
+  else                  return 0;
+}
+
 // ------------------------------------------------------------------
 // ciObjectFactory::get_metadata
 //
@@ -280,7 +287,8 @@ ciMetadata* ciObjectFactory::get_metadata(Metadata* key) {
   }
 #endif // ASSERT
   int len = _ci_metadata->length();
-  int index = find(key, _ci_metadata);
+  bool found = false;
+  int index = _ci_metadata->find_sorted<Metadata*, ciObjectFactory::metadata_compare>(key, found);
 #ifdef ASSERT
   if (CIObjectFactoryVerify) {
     for (int i=0; i<_ci_metadata->length(); i++) {
@@ -290,7 +298,8 @@ ciMetadata* ciObjectFactory::get_metadata(Metadata* key) {
     }
   }
 #endif
-  if (!is_found_at(index, key, _ci_metadata)) {
+
+  if (!found) {
     // The ciMetadata does not yet exist. Create it and insert it
     // into the cache.
     ciMetadata* new_object = create_new_metadata(key);
@@ -300,10 +309,10 @@ ciMetadata* ciObjectFactory::get_metadata(Metadata* key) {
     if (len != _ci_metadata->length()) {
       // creating the new object has recursively entered new objects
       // into the table.  We need to recompute our index.
-      index = find(key, _ci_metadata);
+      index = _ci_metadata->find_sorted<Metadata*, ciObjectFactory::metadata_compare>(key, found);
     }
-    assert(!is_found_at(index, key, _ci_metadata), "no double insert");
+    assert(!found, "no double insert");
-    insert(index, new_object, _ci_metadata);
+    _ci_metadata->insert_before(index, new_object);
     return new_object;
   }
   return _ci_metadata->at(index)->as_metadata();
@@ -655,60 +664,6 @@ void ciObjectFactory::init_ident_of(ciBaseObject* obj) {
   obj->set_ident(_next_ident++);
 }
 
-// ------------------------------------------------------------------
-// ciObjectFactory::find
-//
-// Use binary search to find the position of this oop in the cache.
-// If there is no entry in the cache corresponding to this oop, return
-// the position at which the oop should be inserted.
-int ciObjectFactory::find(Metadata* key, GrowableArray<ciMetadata*>* objects) {
-  int min = 0;
-  int max = objects->length()-1;
-
-  // print_contents();
-
-  while (max >= min) {
-    int mid = (max + min) / 2;
-    Metadata* value = objects->at(mid)->constant_encoding();
-    if (value < key) {
-      min = mid + 1;
-    } else if (value > key) {
-      max = mid - 1;
-    } else {
-      return mid;
-    }
-  }
-  return min;
-}
-
-// ------------------------------------------------------------------
-// ciObjectFactory::is_found_at
-//
-// Verify that the binary seach found the given key.
-bool ciObjectFactory::is_found_at(int index, Metadata* key, GrowableArray<ciMetadata*>* objects) {
-  return (index < objects->length() &&
-          objects->at(index)->constant_encoding() == key);
-}
-
-
-// ------------------------------------------------------------------
-// ciObjectFactory::insert
-//
-// Insert a ciObject into the table at some index.
-void ciObjectFactory::insert(int index, ciMetadata* obj, GrowableArray<ciMetadata*>* objects) {
-  int len = objects->length();
-  if (len == index) {
-    objects->append(obj);
-  } else {
-    objects->append(objects->at(len-1));
-    int pos;
-    for (pos = len-2; pos >= index; pos--) {
-      objects->at_put(pos+1,objects->at(pos));
-    }
-    objects->at_put(index, obj);
-  }
-}
-
 static ciObjectFactory::NonPermObject* emptyBucket = NULL;
 
 // ------------------------------------------------------------------

hotspot/src/share/vm/ci/ciObjectFactory.hpp

@@ -68,9 +68,7 @@ private:
   NonPermObject* _non_perm_bucket[NON_PERM_BUCKETS];
   int _non_perm_count;
 
-  int find(Metadata* key, GrowableArray<ciMetadata*>* objects);
+  static int metadata_compare(Metadata* const& key, ciMetadata* const& elt);
-  bool is_found_at(int index, Metadata* key, GrowableArray<ciMetadata*>* objects);
-  void insert(int index, ciMetadata* obj, GrowableArray<ciMetadata*>* objects);
 
   ciObject* create_new_object(oop o);
   ciMetadata* create_new_metadata(Metadata* o);

hotspot/src/share/vm/classfile/classFileParser.cpp

@@ -5380,7 +5380,7 @@ void ClassFileParser::fill_instance_klass(InstanceKlass* ik, TRAPS) {
     }
   }
 
-  TRACE_INIT_ID(ik);
+  TRACE_INIT_KLASS_ID(ik);
 
   // If we reach here, all is well.
   // Now remove the InstanceKlass* from the _klass_to_deallocate field

hotspot/src/share/vm/classfile/classLoader.cpp

@@ -37,6 +37,7 @@
 #include "gc/shared/generation.hpp"
 #include "interpreter/bytecodeStream.hpp"
 #include "interpreter/oopMapCache.hpp"
+#include "logging/logTag.hpp"
 #include "memory/allocation.inline.hpp"
 #include "memory/filemap.hpp"
 #include "memory/oopFactory.hpp"
@@ -417,34 +418,30 @@ bool ClassPathImageEntry::is_jrt() {
 #if INCLUDE_CDS
 void ClassLoader::exit_with_path_failure(const char* error, const char* message) {
   assert(DumpSharedSpaces, "only called at dump time");
-  tty->print_cr("Hint: enable -XX:+TraceClassPaths to diagnose the failure");
+  tty->print_cr("Hint: enable -Xlog:classpath=info to diagnose the failure");
   vm_exit_during_initialization(error, message);
 }
 #endif
 
-void ClassLoader::trace_class_path(outputStream* out, const char* msg, const char* name) {
+void ClassLoader::trace_class_path(const char* msg, const char* name) {
-  if (!TraceClassPaths) {
+  if (log_is_enabled(Info, classpath)) {
-    return;
+    ResourceMark rm;
-  }
+    outputStream* out = LogHandle(classpath)::info_stream();
-
+    if (msg) {
-  if (msg) {
+      out->print("%s", msg);
-    out->print("%s", msg);
+    }
-  }
+    if (name) {
-  if (name) {
+      if (strlen(name) < 256) {
-    if (strlen(name) < 256) {
+        out->print("%s", name);
-      out->print("%s", name);
+      } else {
-    } else {
+        // For very long paths, we need to print each character separately,
-      // For very long paths, we need to print each character separately,
+        // as print_cr() has a length limit
-      // as print_cr() has a length limit
+        while (name[0] != '\0') {
-      while (name[0] != '\0') {
+          out->print("%c", name[0]);
-        out->print("%c", name[0]);
+          name++;
-        name++;
+        }
       }
     }
-  }
-  if (msg && msg[0] == '[') {
-    out->print_cr("]");
-  } else {
     out->cr();
   }
 }
@@ -470,11 +467,13 @@ void ClassLoader::check_shared_classpath(const char *path) {
 void ClassLoader::setup_bootstrap_search_path() {
   assert(_first_entry == NULL, "should not setup bootstrap class search path twice");
   const char* sys_class_path = Arguments::get_sysclasspath();
+  const char* java_class_path = Arguments::get_appclasspath();
   if (PrintSharedArchiveAndExit) {
     // Don't print sys_class_path - this is the bootcp of this current VM process, not necessarily
     // the same as the bootcp of the shared archive.
   } else {
-    trace_class_path(tty, "[Bootstrap loader class path=", sys_class_path);
+    trace_class_path("bootstrap loader class path=", sys_class_path);
+    trace_class_path("classpath: ", java_class_path);
   }
 #if INCLUDE_CDS
   if (DumpSharedSpaces) {
@@ -578,9 +577,7 @@ ClassPathEntry* ClassLoader::create_class_path_entry(const char *path, const str
         }
       }
     }
-    if (TraceClassPaths) {
+    log_info(classpath)("opened: %s", path);
-      tty->print_cr("[Opened %s]", path);
-    }
     log_info(classload)("opened: %s", path);
   } else {
     // Directory

hotspot/src/share/vm/classfile/classLoader.hpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2015, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2016, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -331,7 +331,7 @@ class ClassLoader: AllStatic {
   static void  exit_with_path_failure(const char* error, const char* message);
 #endif
 
-  static void  trace_class_path(outputStream* out, const char* msg, const char* name = NULL);
+  static void  trace_class_path(const char* msg, const char* name = NULL);
 
   // VM monitoring and management support
   static jlong classloader_time_ms();

hotspot/src/share/vm/classfile/dictionary.cpp

@@ -135,8 +135,10 @@ void DictionaryEntry::add_protection_domain(Dictionary* dict, oop protection_dom
     //          via a store to _pd_set.
     OrderAccess::release_store_ptr(&_pd_set, new_head);
   }
-  if (TraceProtectionDomainVerification && WizardMode) {
+  if (log_is_enabled(Trace, protectiondomain)) {
-    print();
+    ResourceMark rm;
+    outputStream* log = LogHandle(protectiondomain)::trace_stream();
+    print_count(log);
   }
 }
 

hotspot/src/share/vm/classfile/dictionary.hpp

@@ -29,6 +29,7 @@
 #include "oops/instanceKlass.hpp"
 #include "oops/oop.hpp"
 #include "utilities/hashtable.hpp"
+#include "utilities/ostream.hpp"
 
 class DictionaryEntry;
 class PSPromotionManager;
@@ -323,14 +324,14 @@ class DictionaryEntry : public HashtableEntry<Klass*, mtClass> {
     return (klass->name() == class_name && _loader_data == loader_data);
   }
 
-  void print() {
+  void print_count(outputStream *st) {
     int count = 0;
     for (ProtectionDomainEntry* current = _pd_set;
                                 current != NULL;
                                 current = current->_next) {
       count++;
     }
-    tty->print_cr("pd set = #%d", count);
+    st->print_cr("pd set count = #%d", count);
   }
 };
 

hotspot/src/share/vm/classfile/klassFactory.cpp

@@ -1,5 +1,5 @@
 /*
-* Copyright (c) 2015, Oracle and/or its affiliates. All rights reserved.
+* Copyright (c) 2015, 2016, Oracle and/or its affiliates. All rights reserved.
 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
 *
 * This code is free software; you can redistribute it and/or modify it
@@ -29,6 +29,7 @@
 #include "classfile/klassFactory.hpp"
 #include "memory/resourceArea.hpp"
 #include "prims/jvmtiEnvBase.hpp"
+#include "trace/traceMacros.hpp"
 
 static ClassFileStream* prologue(ClassFileStream* stream,
                                  Symbol* name,
@@ -136,5 +137,7 @@ instanceKlassHandle KlassFactory::create_from_stream(ClassFileStream* stream,
     result->set_cached_class_file(cached_class_file);
   }
 
+  TRACE_KLASS_CREATION(result, parser, THREAD);
+
   return result;
 }

hotspot/src/share/vm/classfile/sharedPathsMiscInfo.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2014, 2015, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2014, 2016, Oracle and/or its affiliates. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
  * This code is free software; you can redistribute it and/or modify it
@@ -26,15 +26,15 @@
 #include "classfile/classLoader.hpp"
 #include "classfile/classLoaderData.inline.hpp"
 #include "classfile/sharedPathsMiscInfo.hpp"
+#include "logging/log.hpp"
 #include "memory/allocation.inline.hpp"
 #include "memory/metaspaceShared.hpp"
 #include "runtime/arguments.hpp"
+#include "utilities/ostream.hpp"
 
 void SharedPathsMiscInfo::add_path(const char* path, int type) {
-  if (TraceClassPaths) {
+  log_info(classpath)("type=%s ", type_name(type));
-    tty->print("[type=%s] ", type_name(type));
+  ClassLoader::trace_class_path("add misc shared path ", path);
-    trace_class_path("[Add misc shared path ", path);
-  }
   write(path, strlen(path) + 1);
   write_jint(jint(type));
 }
@@ -67,11 +67,29 @@ bool SharedPathsMiscInfo::read(void* ptr, size_t size) {
 }
 
 bool SharedPathsMiscInfo::fail(const char* msg, const char* name) {
-  ClassLoader::trace_class_path(tty, msg, name);
+  ClassLoader::trace_class_path(msg, name);
   MetaspaceShared::set_archive_loading_failed();
   return false;
 }
 
+void SharedPathsMiscInfo::print_path(int type, const char* path) {
+  ResourceMark rm;
+  outputStream* out = LogHandle(classpath)::info_stream();
+  switch (type) {
+  case BOOT:
+    out->print("Expecting -Dsun.boot.class.path=%s", path);
+    break;
+  case NON_EXIST:
+    out->print("Expecting that %s does not exist", path);
+    break;
+  case REQUIRED:
+    out->print("Expecting that file %s must exist and is not altered", path);
+    break;
+  default:
+    ShouldNotReachHere();
+  }
+}
+
 bool SharedPathsMiscInfo::check() {
   // The whole buffer must be 0 terminated so that we can use strlen and strcmp
   // without fear.
@@ -90,17 +108,14 @@ bool SharedPathsMiscInfo::check() {
     if (!read_jint(&type)) {
       return fail("Corrupted archive file header");
     }
-    if (TraceClassPaths) {
+    log_info(classpath)("type=%s ", type_name(type));
-      tty->print("[type=%s ", type_name(type));
+    print_path(type, path);
-      print_path(tty, type, path);
-      tty->print_cr("]");
-    }
     if (!check(type, path)) {
       if (!PrintSharedArchiveAndExit) {
         return false;
       }
     } else {
-      trace_class_path("[ok");
+      ClassLoader::trace_class_path("ok");
     }
   }
 

hotspot/src/share/vm/classfile/sharedPathsMiscInfo.hpp

@@ -64,9 +64,6 @@ protected:
   void write(const void* ptr, size_t size);
   bool read(void* ptr, size_t size);
 
-  static void trace_class_path(const char* msg, const char* name = NULL) {
-    ClassLoader::trace_class_path(tty, msg, name);
-  }
 protected:
   static bool fail(const char* msg, const char* name = NULL);
   virtual bool check(jint type, const char* path);
@@ -144,21 +141,7 @@ public:
     }
   }
 
-  virtual void print_path(outputStream* out, int type, const char* path) {
+  virtual void print_path(int type, const char* path);
-    switch (type) {
-    case BOOT:
-      out->print("Expecting -Dsun.boot.class.path=%s", path);
-      break;
-    case NON_EXIST:
-      out->print("Expecting that %s does not exist", path);
-      break;
-    case REQUIRED:
-      out->print("Expecting that file %s must exist and is not altered", path);
-      break;
-    default:
-      ShouldNotReachHere();
-    }
-  }
 
   bool check();
   bool read_jint(jint *ptr) {

hotspot/src/share/vm/classfile/systemDictionary.cpp

@@ -430,12 +430,15 @@ void SystemDictionary::validate_protection_domain(instanceKlassHandle klass,
 
   // Now we have to call back to java to check if the initating class has access
   JavaValue result(T_VOID);
-  if (TraceProtectionDomainVerification) {
+  if (log_is_enabled(Debug, protectiondomain)) {
+    ResourceMark rm;
     // Print out trace information
-    tty->print_cr("Checking package access");
+    outputStream* log = LogHandle(protectiondomain)::debug_stream();
-    tty->print(" - class loader:      "); class_loader()->print_value_on(tty);      tty->cr();
+    log->print_cr("Checking package access");
-    tty->print(" - protection domain: "); protection_domain()->print_value_on(tty); tty->cr();
+    log->print("class loader: "); class_loader()->print_value_on(log);
-    tty->print(" - loading:           "); klass()->print_value_on(tty);             tty->cr();
+    log->print(" protection domain: "); protection_domain()->print_value_on(log);
+    log->print(" loading: "); klass()->print_value_on(log);
+    log->cr();
   }
 
   KlassHandle system_loader(THREAD, SystemDictionary::ClassLoader_klass());
@@ -448,13 +451,10 @@ void SystemDictionary::validate_protection_domain(instanceKlassHandle klass,
                          protection_domain,
                          THREAD);
 
-  if (TraceProtectionDomainVerification) {
+  if (HAS_PENDING_EXCEPTION) {
-    if (HAS_PENDING_EXCEPTION) {
+    log_debug(protectiondomain)("DENIED !!!!!!!!!!!!!!!!!!!!!");
-      tty->print_cr(" -> DENIED !!!!!!!!!!!!!!!!!!!!!");
+  } else {
-    } else {
+   log_debug(protectiondomain)("granted");
-     tty->print_cr(" -> granted");
-    }
-    tty->cr();
   }
 
   if (HAS_PENDING_EXCEPTION) return;

hotspot/src/share/vm/classfile/vmSymbols.cpp

@@ -328,8 +328,6 @@ bool vmIntrinsics::preserves_state(vmIntrinsics::ID id) {
   assert(id != vmIntrinsics::_none, "must be a VM intrinsic");
   switch(id) {
 #ifdef TRACE_HAVE_INTRINSICS
-  case vmIntrinsics::_classID:
-  case vmIntrinsics::_threadID:
   case vmIntrinsics::_counterTime:
 #endif
   case vmIntrinsics::_currentTimeMillis:
@@ -544,6 +542,42 @@ bool vmIntrinsics::is_disabled_by_flags(const methodHandle& method) {
   case vmIntrinsics::_putLongVolatile:
   case vmIntrinsics::_putFloatVolatile:
   case vmIntrinsics::_putDoubleVolatile:
+  case vmIntrinsics::_getObjectAcquire:
+  case vmIntrinsics::_getBooleanAcquire:
+  case vmIntrinsics::_getByteAcquire:
+  case vmIntrinsics::_getShortAcquire:
+  case vmIntrinsics::_getCharAcquire:
+  case vmIntrinsics::_getIntAcquire:
+  case vmIntrinsics::_getLongAcquire:
+  case vmIntrinsics::_getFloatAcquire:
+  case vmIntrinsics::_getDoubleAcquire:
+  case vmIntrinsics::_putObjectRelease:
+  case vmIntrinsics::_putBooleanRelease:
+  case vmIntrinsics::_putByteRelease:
+  case vmIntrinsics::_putShortRelease:
+  case vmIntrinsics::_putCharRelease:
+  case vmIntrinsics::_putIntRelease:
+  case vmIntrinsics::_putLongRelease:
+  case vmIntrinsics::_putFloatRelease:
+  case vmIntrinsics::_putDoubleRelease:
+  case vmIntrinsics::_getObjectOpaque:
+  case vmIntrinsics::_getBooleanOpaque:
+  case vmIntrinsics::_getByteOpaque:
+  case vmIntrinsics::_getShortOpaque:
+  case vmIntrinsics::_getCharOpaque:
+  case vmIntrinsics::_getIntOpaque:
+  case vmIntrinsics::_getLongOpaque:
+  case vmIntrinsics::_getFloatOpaque:
+  case vmIntrinsics::_getDoubleOpaque:
+  case vmIntrinsics::_putObjectOpaque:
+  case vmIntrinsics::_putBooleanOpaque:
+  case vmIntrinsics::_putByteOpaque:
+  case vmIntrinsics::_putShortOpaque:
+  case vmIntrinsics::_putCharOpaque:
+  case vmIntrinsics::_putIntOpaque:
+  case vmIntrinsics::_putLongOpaque:
+  case vmIntrinsics::_putFloatOpaque:
+  case vmIntrinsics::_putDoubleOpaque:
   case vmIntrinsics::_getByte_raw:
   case vmIntrinsics::_getShort_raw:
   case vmIntrinsics::_getChar_raw:
@@ -569,9 +603,27 @@ bool vmIntrinsics::is_disabled_by_flags(const methodHandle& method) {
   case vmIntrinsics::_loadFence:
   case vmIntrinsics::_storeFence:
   case vmIntrinsics::_fullFence:
-  case vmIntrinsics::_compareAndSwapObject:
   case vmIntrinsics::_compareAndSwapLong:
+  case vmIntrinsics::_weakCompareAndSwapLong:
+  case vmIntrinsics::_weakCompareAndSwapLongAcquire:
+  case vmIntrinsics::_weakCompareAndSwapLongRelease:
   case vmIntrinsics::_compareAndSwapInt:
+  case vmIntrinsics::_weakCompareAndSwapInt:
+  case vmIntrinsics::_weakCompareAndSwapIntAcquire:
+  case vmIntrinsics::_weakCompareAndSwapIntRelease:
+  case vmIntrinsics::_compareAndSwapObject:
+  case vmIntrinsics::_weakCompareAndSwapObject:
+  case vmIntrinsics::_weakCompareAndSwapObjectAcquire:
+  case vmIntrinsics::_weakCompareAndSwapObjectRelease:
+  case vmIntrinsics::_compareAndExchangeIntVolatile:
+  case vmIntrinsics::_compareAndExchangeIntAcquire:
+  case vmIntrinsics::_compareAndExchangeIntRelease:
+  case vmIntrinsics::_compareAndExchangeLongVolatile:
+  case vmIntrinsics::_compareAndExchangeLongAcquire:
+  case vmIntrinsics::_compareAndExchangeLongRelease:
+  case vmIntrinsics::_compareAndExchangeObjectVolatile:
+  case vmIntrinsics::_compareAndExchangeObjectAcquire:
+  case vmIntrinsics::_compareAndExchangeObjectRelease:
     if (!InlineUnsafeOps) return true;
     break;
   case vmIntrinsics::_getShortUnaligned:

hotspot/src/share/vm/classfile/vmSymbols.hpp

@@ -1146,6 +1146,64 @@
   do_intrinsic(_putFloatVolatile,         jdk_internal_misc_Unsafe,     putFloatVolatile_name, putFloat_signature,     F_RN)  \
   do_intrinsic(_putDoubleVolatile,        jdk_internal_misc_Unsafe,     putDoubleVolatile_name, putDouble_signature,   F_RN)  \
                                                                                                                         \
+  do_name(getObjectOpaque_name,"getObjectOpaque")     do_name(putObjectOpaque_name,"putObjectOpaque")                   \
+  do_name(getBooleanOpaque_name,"getBooleanOpaque")   do_name(putBooleanOpaque_name,"putBooleanOpaque")                 \
+  do_name(getByteOpaque_name,"getByteOpaque")         do_name(putByteOpaque_name,"putByteOpaque")                       \
+  do_name(getShortOpaque_name,"getShortOpaque")       do_name(putShortOpaque_name,"putShortOpaque")                     \
+  do_name(getCharOpaque_name,"getCharOpaque")         do_name(putCharOpaque_name,"putCharOpaque")                       \
+  do_name(getIntOpaque_name,"getIntOpaque")           do_name(putIntOpaque_name,"putIntOpaque")                         \
+  do_name(getLongOpaque_name,"getLongOpaque")         do_name(putLongOpaque_name,"putLongOpaque")                       \
+  do_name(getFloatOpaque_name,"getFloatOpaque")       do_name(putFloatOpaque_name,"putFloatOpaque")                     \
+  do_name(getDoubleOpaque_name,"getDoubleOpaque")     do_name(putDoubleOpaque_name,"putDoubleOpaque")                   \
+                                                                                                                        \
+  do_intrinsic(_getObjectOpaque,          jdk_internal_misc_Unsafe,        getObjectOpaque_name, getObject_signature,   F_R)  \
+  do_intrinsic(_getBooleanOpaque,         jdk_internal_misc_Unsafe,        getBooleanOpaque_name, getBoolean_signature, F_R)  \
+  do_intrinsic(_getByteOpaque,            jdk_internal_misc_Unsafe,        getByteOpaque_name, getByte_signature,       F_R)  \
+  do_intrinsic(_getShortOpaque,           jdk_internal_misc_Unsafe,        getShortOpaque_name, getShort_signature,     F_R)  \
+  do_intrinsic(_getCharOpaque,            jdk_internal_misc_Unsafe,        getCharOpaque_name, getChar_signature,       F_R)  \
+  do_intrinsic(_getIntOpaque,             jdk_internal_misc_Unsafe,        getIntOpaque_name, getInt_signature,         F_R)  \
+  do_intrinsic(_getLongOpaque,            jdk_internal_misc_Unsafe,        getLongOpaque_name, getLong_signature,       F_R)  \
+  do_intrinsic(_getFloatOpaque,           jdk_internal_misc_Unsafe,        getFloatOpaque_name, getFloat_signature,     F_R)  \
+  do_intrinsic(_getDoubleOpaque,          jdk_internal_misc_Unsafe,        getDoubleOpaque_name, getDouble_signature,   F_R)  \
+  do_intrinsic(_putObjectOpaque,          jdk_internal_misc_Unsafe,        putObjectOpaque_name, putObject_signature,   F_R)  \
+  do_intrinsic(_putBooleanOpaque,         jdk_internal_misc_Unsafe,        putBooleanOpaque_name, putBoolean_signature, F_R)  \
+  do_intrinsic(_putByteOpaque,            jdk_internal_misc_Unsafe,        putByteOpaque_name, putByte_signature,       F_R)  \
+  do_intrinsic(_putShortOpaque,           jdk_internal_misc_Unsafe,        putShortOpaque_name, putShort_signature,     F_R)  \
+  do_intrinsic(_putCharOpaque,            jdk_internal_misc_Unsafe,        putCharOpaque_name, putChar_signature,       F_R)  \
+  do_intrinsic(_putIntOpaque,             jdk_internal_misc_Unsafe,        putIntOpaque_name, putInt_signature,         F_R)  \
+  do_intrinsic(_putLongOpaque,            jdk_internal_misc_Unsafe,        putLongOpaque_name, putLong_signature,       F_R)  \
+  do_intrinsic(_putFloatOpaque,           jdk_internal_misc_Unsafe,        putFloatOpaque_name, putFloat_signature,     F_R)  \
+  do_intrinsic(_putDoubleOpaque,          jdk_internal_misc_Unsafe,        putDoubleOpaque_name, putDouble_signature,   F_R)  \
+                                                                                                                        \
+  do_name(getObjectAcquire_name,  "getObjectAcquire")    do_name(putObjectRelease_name,  "putObjectRelease")            \
+  do_name(getBooleanAcquire_name, "getBooleanAcquire")   do_name(putBooleanRelease_name, "putBooleanRelease")           \
+  do_name(getByteAcquire_name,    "getByteAcquire")      do_name(putByteRelease_name,    "putByteRelease")              \
+  do_name(getShortAcquire_name,   "getShortAcquire")     do_name(putShortRelease_name,   "putShortRelease")             \
+  do_name(getCharAcquire_name,    "getCharAcquire")      do_name(putCharRelease_name,    "putCharRelease")              \
+  do_name(getIntAcquire_name,     "getIntAcquire")       do_name(putIntRelease_name,     "putIntRelease")               \
+  do_name(getLongAcquire_name,    "getLongAcquire")      do_name(putLongRelease_name,    "putLongRelease")              \
+  do_name(getFloatAcquire_name,   "getFloatAcquire")     do_name(putFloatRelease_name,   "putFloatRelease")             \
+  do_name(getDoubleAcquire_name,  "getDoubleAcquire")    do_name(putDoubleRelease_name,  "putDoubleRelease")            \
+                                                                                                                        \
+  do_intrinsic(_getObjectAcquire,        jdk_internal_misc_Unsafe,        getObjectAcquire_name, getObject_signature,   F_R)  \
+  do_intrinsic(_getBooleanAcquire,       jdk_internal_misc_Unsafe,        getBooleanAcquire_name, getBoolean_signature, F_R)  \
+  do_intrinsic(_getByteAcquire,          jdk_internal_misc_Unsafe,        getByteAcquire_name, getByte_signature,       F_R)  \
+  do_intrinsic(_getShortAcquire,         jdk_internal_misc_Unsafe,        getShortAcquire_name, getShort_signature,     F_R)  \
+  do_intrinsic(_getCharAcquire,          jdk_internal_misc_Unsafe,        getCharAcquire_name, getChar_signature,       F_R)  \
+  do_intrinsic(_getIntAcquire,           jdk_internal_misc_Unsafe,        getIntAcquire_name, getInt_signature,         F_R)  \
+  do_intrinsic(_getLongAcquire,          jdk_internal_misc_Unsafe,        getLongAcquire_name, getLong_signature,       F_R)  \
+  do_intrinsic(_getFloatAcquire,         jdk_internal_misc_Unsafe,        getFloatAcquire_name, getFloat_signature,     F_R)  \
+  do_intrinsic(_getDoubleAcquire,        jdk_internal_misc_Unsafe,        getDoubleAcquire_name, getDouble_signature,   F_R)  \
+  do_intrinsic(_putObjectRelease,        jdk_internal_misc_Unsafe,        putObjectRelease_name, putObject_signature,   F_R)  \
+  do_intrinsic(_putBooleanRelease,       jdk_internal_misc_Unsafe,        putBooleanRelease_name, putBoolean_signature, F_R)  \
+  do_intrinsic(_putByteRelease,          jdk_internal_misc_Unsafe,        putByteRelease_name, putByte_signature,       F_R)  \
+  do_intrinsic(_putShortRelease,         jdk_internal_misc_Unsafe,        putShortRelease_name, putShort_signature,     F_R)  \
+  do_intrinsic(_putCharRelease,          jdk_internal_misc_Unsafe,        putCharRelease_name, putChar_signature,       F_R)  \
+  do_intrinsic(_putIntRelease,           jdk_internal_misc_Unsafe,        putIntRelease_name, putInt_signature,         F_R)  \
+  do_intrinsic(_putLongRelease,          jdk_internal_misc_Unsafe,        putLongRelease_name, putLong_signature,       F_R)  \
+  do_intrinsic(_putFloatRelease,         jdk_internal_misc_Unsafe,        putFloatRelease_name, putFloat_signature,     F_R)  \
+  do_intrinsic(_putDoubleRelease,        jdk_internal_misc_Unsafe,        putDoubleRelease_name, putDouble_signature,   F_R)  \
+                                                                                                                        \
   do_name(getShortUnaligned_name,"getShortUnaligned")     do_name(putShortUnaligned_name,"putShortUnaligned")           \
   do_name(getCharUnaligned_name,"getCharUnaligned")       do_name(putCharUnaligned_name,"putCharUnaligned")             \
   do_name(getIntUnaligned_name,"getIntUnaligned")         do_name(putIntUnaligned_name,"putIntUnaligned")               \
@@ -1197,24 +1255,68 @@
   do_intrinsic(_putDouble_raw,            jdk_internal_misc_Unsafe,     putDouble_name, putDouble_raw_signature,       F_R)  \
   do_intrinsic(_putAddress_raw,           jdk_internal_misc_Unsafe,     putAddress_name, putAddress_raw_signature,     F_R)  \
                                                                                                                         \
-  do_intrinsic(_compareAndSwapObject,     jdk_internal_misc_Unsafe,     compareAndSwapObject_name, compareAndSwapObject_signature, F_R) \
+  do_signature(compareAndSwapObject_signature,     "(Ljava/lang/Object;JLjava/lang/Object;Ljava/lang/Object;)Z")        \
-   do_name(     compareAndSwapObject_name,                              "compareAndSwapObject")                                \
+  do_signature(compareAndExchangeObject_signature, "(Ljava/lang/Object;JLjava/lang/Object;Ljava/lang/Object;)Ljava/lang/Object;") \
-   do_signature(compareAndSwapObject_signature,                         "(Ljava/lang/Object;JLjava/lang/Object;Ljava/lang/Object;)Z")          \
+  do_signature(compareAndSwapLong_signature,       "(Ljava/lang/Object;JJJ)Z")                                          \
-  do_intrinsic(_compareAndSwapLong,       jdk_internal_misc_Unsafe,     compareAndSwapLong_name, compareAndSwapLong_signature, F_R) \
+  do_signature(compareAndExchangeLong_signature,   "(Ljava/lang/Object;JJJ)J")                                          \
-   do_name(     compareAndSwapLong_name,                                "compareAndSwapLong")                                  \
+  do_signature(compareAndSwapInt_signature,        "(Ljava/lang/Object;JII)Z")                                          \
-   do_signature(compareAndSwapLong_signature,                           "(Ljava/lang/Object;JJJ)Z")                            \
+  do_signature(compareAndExchangeInt_signature,    "(Ljava/lang/Object;JII)I")                                          \
-  do_intrinsic(_compareAndSwapInt,        jdk_internal_misc_Unsafe,     compareAndSwapInt_name, compareAndSwapInt_signature, F_R) \
+                                                                                                                        \
-   do_name(     compareAndSwapInt_name,                                 "compareAndSwapInt")                                   \
+  do_name(compareAndSwapObject_name,             "compareAndSwapObject")                                                \
-   do_signature(compareAndSwapInt_signature,                            "(Ljava/lang/Object;JII)Z")                            \
+  do_name(compareAndExchangeObjectVolatile_name, "compareAndExchangeObjectVolatile")                                    \
-  do_intrinsic(_putOrderedObject,         jdk_internal_misc_Unsafe,     putOrderedObject_name, putOrderedObject_signature, F_R) \
+  do_name(compareAndExchangeObjectAcquire_name,  "compareAndExchangeObjectAcquire")                                     \
-   do_name(     putOrderedObject_name,                                  "putOrderedObject")                                    \
+  do_name(compareAndExchangeObjectRelease_name,  "compareAndExchangeObjectRelease")                                     \
-   do_alias(    putOrderedObject_signature,                             /*(LObject;JLObject;)V*/ putObject_signature)           \
+  do_name(compareAndSwapLong_name,               "compareAndSwapLong")                                                  \
-  do_intrinsic(_putOrderedLong,           jdk_internal_misc_Unsafe,     putOrderedLong_name, putOrderedLong_signature, F_R)  \
+  do_name(compareAndExchangeLongVolatile_name,   "compareAndExchangeLongVolatile")                                      \
-   do_name(     putOrderedLong_name,                                    "putOrderedLong")                                      \
+  do_name(compareAndExchangeLongAcquire_name,    "compareAndExchangeLongAcquire")                                       \
-   do_alias(    putOrderedLong_signature,                               /*(Ljava/lang/Object;JJ)V*/ putLong_signature)          \
+  do_name(compareAndExchangeLongRelease_name,    "compareAndExchangeLongRelease")                                       \
-  do_intrinsic(_putOrderedInt,            jdk_internal_misc_Unsafe,     putOrderedInt_name, putOrderedInt_signature,   F_R)  \
+  do_name(compareAndSwapInt_name,                "compareAndSwapInt")                                                   \
-   do_name(     putOrderedInt_name,                                     "putOrderedInt")                                       \
+  do_name(compareAndExchangeIntVolatile_name,    "compareAndExchangeIntVolatile")                                       \
-   do_alias(    putOrderedInt_signature,                                 /*(Ljava/lang/Object;JI)V*/ putInt_signature)           \
+  do_name(compareAndExchangeIntAcquire_name,     "compareAndExchangeIntAcquire")                                        \
+  do_name(compareAndExchangeIntRelease_name,     "compareAndExchangeIntRelease")                                        \
+                                                                                                                        \
+  do_name(weakCompareAndSwapObject_name,         "weakCompareAndSwapObject")                                            \
+  do_name(weakCompareAndSwapObjectAcquire_name,  "weakCompareAndSwapObjectAcquire")                                     \
+  do_name(weakCompareAndSwapObjectRelease_name,  "weakCompareAndSwapObjectRelease")                                     \
+  do_name(weakCompareAndSwapLong_name,           "weakCompareAndSwapLong")                                              \
+  do_name(weakCompareAndSwapLongAcquire_name,    "weakCompareAndSwapLongAcquire")                                       \
+  do_name(weakCompareAndSwapLongRelease_name,    "weakCompareAndSwapLongRelease")                                       \
+  do_name(weakCompareAndSwapInt_name,            "weakCompareAndSwapInt")                                               \
+  do_name(weakCompareAndSwapIntAcquire_name,     "weakCompareAndSwapIntAcquire")                                        \
+  do_name(weakCompareAndSwapIntRelease_name,     "weakCompareAndSwapIntRelease")                                        \
+                                                                                                                        \
+  do_intrinsic(_compareAndSwapObject,             jdk_internal_misc_Unsafe,  compareAndSwapObject_name,             compareAndSwapObject_signature,     F_RN) \
+  do_intrinsic(_compareAndExchangeObjectVolatile, jdk_internal_misc_Unsafe,  compareAndExchangeObjectVolatile_name, compareAndExchangeObject_signature, F_RN) \
+  do_intrinsic(_compareAndExchangeObjectAcquire,  jdk_internal_misc_Unsafe,  compareAndExchangeObjectAcquire_name,  compareAndExchangeObject_signature, F_R)  \
+  do_intrinsic(_compareAndExchangeObjectRelease,  jdk_internal_misc_Unsafe,  compareAndExchangeObjectRelease_name,  compareAndExchangeObject_signature, F_R)  \
+  do_intrinsic(_compareAndSwapLong,               jdk_internal_misc_Unsafe,  compareAndSwapLong_name,               compareAndSwapLong_signature,       F_RN) \
+  do_intrinsic(_compareAndExchangeLongVolatile,   jdk_internal_misc_Unsafe,  compareAndExchangeLongVolatile_name,   compareAndExchangeLong_signature,   F_RN) \
+  do_intrinsic(_compareAndExchangeLongAcquire,    jdk_internal_misc_Unsafe,  compareAndExchangeLongAcquire_name,    compareAndExchangeLong_signature,   F_R)  \
+  do_intrinsic(_compareAndExchangeLongRelease,    jdk_internal_misc_Unsafe,  compareAndExchangeLongRelease_name,    compareAndExchangeLong_signature,   F_R)  \
+  do_intrinsic(_compareAndSwapInt,                jdk_internal_misc_Unsafe,  compareAndSwapInt_name,                compareAndSwapInt_signature,        F_RN) \
+  do_intrinsic(_compareAndExchangeIntVolatile,    jdk_internal_misc_Unsafe,  compareAndExchangeIntVolatile_name,    compareAndExchangeInt_signature,    F_RN) \
+  do_intrinsic(_compareAndExchangeIntAcquire,     jdk_internal_misc_Unsafe,  compareAndExchangeIntAcquire_name,     compareAndExchangeInt_signature,    F_R)  \
+  do_intrinsic(_compareAndExchangeIntRelease,     jdk_internal_misc_Unsafe,  compareAndExchangeIntRelease_name,     compareAndExchangeInt_signature,    F_R)  \
+                                                                                                                                                              \
+  do_intrinsic(_weakCompareAndSwapObject,         jdk_internal_misc_Unsafe,  weakCompareAndSwapObject_name,         compareAndSwapObject_signature,     F_R) \
+  do_intrinsic(_weakCompareAndSwapObjectAcquire,  jdk_internal_misc_Unsafe,  weakCompareAndSwapObjectAcquire_name,  compareAndSwapObject_signature,     F_R) \
+  do_intrinsic(_weakCompareAndSwapObjectRelease,  jdk_internal_misc_Unsafe,  weakCompareAndSwapObjectRelease_name,  compareAndSwapObject_signature,     F_R) \
+  do_intrinsic(_weakCompareAndSwapLong,           jdk_internal_misc_Unsafe,  weakCompareAndSwapLong_name,           compareAndSwapLong_signature,       F_R) \
+  do_intrinsic(_weakCompareAndSwapLongAcquire,    jdk_internal_misc_Unsafe,  weakCompareAndSwapLongAcquire_name,    compareAndSwapLong_signature,       F_R) \
+  do_intrinsic(_weakCompareAndSwapLongRelease,    jdk_internal_misc_Unsafe,  weakCompareAndSwapLongRelease_name,    compareAndSwapLong_signature,       F_R) \
+  do_intrinsic(_weakCompareAndSwapInt,            jdk_internal_misc_Unsafe,  weakCompareAndSwapInt_name,            compareAndSwapInt_signature,        F_R) \
+  do_intrinsic(_weakCompareAndSwapIntAcquire,     jdk_internal_misc_Unsafe,  weakCompareAndSwapIntAcquire_name,     compareAndSwapInt_signature,        F_R) \
+  do_intrinsic(_weakCompareAndSwapIntRelease,     jdk_internal_misc_Unsafe,  weakCompareAndSwapIntRelease_name,     compareAndSwapInt_signature,        F_R) \
+                                                                                                                        \
+  do_intrinsic(_putOrderedObject,         jdk_internal_misc_Unsafe,        putOrderedObject_name, putOrderedObject_signature, F_RN) \
+   do_name(     putOrderedObject_name,                           "putOrderedObject")                                    \
+   do_alias(    putOrderedObject_signature,                     /*(LObject;JLObject;)V*/ putObject_signature)           \
+  do_intrinsic(_putOrderedLong,           jdk_internal_misc_Unsafe,        putOrderedLong_name, putOrderedLong_signature, F_RN)  \
+   do_name(     putOrderedLong_name,                             "putOrderedLong")                                      \
+   do_alias(    putOrderedLong_signature,                       /*(Ljava/lang/Object;JJ)V*/ putLong_signature)          \
+  do_intrinsic(_putOrderedInt,            jdk_internal_misc_Unsafe,        putOrderedInt_name, putOrderedInt_signature,   F_RN)  \
+   do_name(     putOrderedInt_name,                              "putOrderedInt")                                       \
+   do_alias(    putOrderedInt_signature,                        /*(Ljava/lang/Object;JI)V*/ putInt_signature)           \
                                                                                                                         \
   do_intrinsic(_getAndAddInt,             jdk_internal_misc_Unsafe,     getAndAddInt_name, getAndAddInt_signature, F_R)       \
    do_name(     getAndAddInt_name,                                      "getAndAddInt")                                       \