8144534: Refactor templateInterpreter and templateInterpreterGenerator functions

Merged templateInterpreter_x86_32/64 into templateInterpreterGenerator_x86.cpp (some 32/64 functions remain for the hand coded crc functions).

Reviewed-by: goetz, jrose, twisti

hotspot/src/cpu/aarch64/vm/interpreter_aarch64.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2003, 2011, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2003, 2015, Oracle and/or its affiliates. All rights reserved.
  * Copyright (c) 2014, Red Hat Inc. All rights reserved.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
@@ -39,7 +39,6 @@
 #include "prims/jvmtiThreadState.hpp"
 #include "prims/methodHandles.hpp"
 #include "runtime/arguments.hpp"
-#include "runtime/deoptimization.hpp"
 #include "runtime/frame.inline.hpp"
 #include "runtime/sharedRuntime.hpp"
 #include "runtime/stubRoutines.hpp"
@@ -259,20 +258,3 @@ address InterpreterGenerator::generate_abstract_entry(void) {
 
   return entry_point;
 }
-
-
-void Deoptimization::unwind_callee_save_values(frame* f, vframeArray* vframe_array) {
-
-  // This code is sort of the equivalent of C2IAdapter::setup_stack_frame back in
-  // the days we had adapter frames. When we deoptimize a situation where a
-  // compiled caller calls a compiled caller will have registers it expects
-  // to survive the call to the callee. If we deoptimize the callee the only
-  // way we can restore these registers is to have the oldest interpreter
-  // frame that we create restore these values. That is what this routine
-  // will accomplish.
-
-  // At the moment we have modified c2 to not have any callee save registers
-  // so this problem does not exist and this routine is just a place holder.
-
-  assert(f->is_interpreted_frame(), "must be interpreted");
-}

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

@@ -39,7 +39,6 @@
 #include "prims/jvmtiThreadState.hpp"
 #include "prims/methodHandles.hpp"
 #include "runtime/arguments.hpp"
-#include "runtime/deoptimization.hpp"
 #include "runtime/frame.inline.hpp"
 #include "runtime/sharedRuntime.hpp"
 #include "runtime/stubRoutines.hpp"
@@ -61,26 +60,6 @@
 
 #define BIND(label) bind(label); BLOCK_COMMENT(#label ":")
 
-int AbstractInterpreter::BasicType_as_index(BasicType type) {
-  int i = 0;
-  switch (type) {
-    case T_BOOLEAN: i = 0; break;
-    case T_CHAR   : i = 1; break;
-    case T_BYTE   : i = 2; break;
-    case T_SHORT  : i = 3; break;
-    case T_INT    : i = 4; break;
-    case T_LONG   : i = 5; break;
-    case T_VOID   : i = 6; break;
-    case T_FLOAT  : i = 7; break;
-    case T_DOUBLE : i = 8; break;
-    case T_OBJECT : i = 9; break;
-    case T_ARRAY  : i = 9; break;
-    default       : ShouldNotReachHere();
-  }
-  assert(0 <= i && i < AbstractInterpreter::number_of_result_handlers, "index out of bounds");
-  return i;
-}
-
 address AbstractInterpreterGenerator::generate_slow_signature_handler() {
   // Slow_signature handler that respects the PPC C calling conventions.
   //
@@ -579,18 +558,3 @@ address InterpreterGenerator::generate_Reference_get_entry(void) {
 
   return NULL;
 }
-
-void Deoptimization::unwind_callee_save_values(frame* f, vframeArray* vframe_array) {
-  // This code is sort of the equivalent of C2IAdapter::setup_stack_frame back in
-  // the days we had adapter frames. When we deoptimize a situation where a
-  // compiled caller calls a compiled caller will have registers it expects
-  // to survive the call to the callee. If we deoptimize the callee the only
-  // way we can restore these registers is to have the oldest interpreter
-  // frame that we create restore these values. That is what this routine
-  // will accomplish.
-
-  // At the moment we have modified c2 to not have any callee save registers
-  // so this problem does not exist and this routine is just a place holder.
-
-  assert(f->is_interpreted_frame(), "must be interpreted");
-}

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

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2014, 2015, Oracle and/or its affiliates. All rights reserved.
- * Copyright 2013, 2015 SAP AG. All rights reserved.
+ * Copyright (c) 2013, 2015 SAP AG. 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
@@ -33,9 +33,12 @@
   // if too small.
   // Run with +PrintInterpreter to get the VM to print out the size.
   // Max size with JVMTI
-
   const static int InterpreterCodeSize = 230*K;
 
+ public:
+  // Support abs and sqrt like in compiler.
+  // For others we can use a normal (native) entry.
+  static bool math_entry_available(AbstractInterpreter::MethodKind kind);
 #endif // CPU_PPC_VM_TEMPLATEINTERPRETER_PPC_HPP
 
 

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

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2014, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2015, 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
@@ -38,7 +38,6 @@
 #include "prims/jvmtiThreadState.hpp"
 #include "prims/methodHandles.hpp"
 #include "runtime/arguments.hpp"
-#include "runtime/deoptimization.hpp"
 #include "runtime/frame.inline.hpp"
 #include "runtime/sharedRuntime.hpp"
 #include "runtime/stubRoutines.hpp"
@@ -62,30 +61,6 @@
 
 //----------------------------------------------------------------------------------------------------
 
-
-
-
-int AbstractInterpreter::BasicType_as_index(BasicType type) {
-  int i = 0;
-  switch (type) {
-    case T_BOOLEAN: i = 0; break;
-    case T_CHAR   : i = 1; break;
-    case T_BYTE   : i = 2; break;
-    case T_SHORT  : i = 3; break;
-    case T_INT    : i = 4; break;
-    case T_LONG   : i = 5; break;
-    case T_VOID   : i = 6; break;
-    case T_FLOAT  : i = 7; break;
-    case T_DOUBLE : i = 8; break;
-    case T_OBJECT : i = 9; break;
-    case T_ARRAY  : i = 9; break;
-    default       : ShouldNotReachHere();
-  }
-  assert(0 <= i && i < AbstractInterpreter::number_of_result_handlers, "index out of bounds");
-  return i;
-}
-
-
 #ifndef _LP64
 address AbstractInterpreterGenerator::generate_slow_signature_handler() {
   address entry = __ pc();
@@ -254,28 +229,3 @@ address InterpreterGenerator::generate_abstract_entry(void) {
   return entry;
 
 }
-
-bool AbstractInterpreter::can_be_compiled(methodHandle m) {
-  // No special entry points that preclude compilation
-  return true;
-}
-
-void Deoptimization::unwind_callee_save_values(frame* f, vframeArray* vframe_array) {
-
-  // This code is sort of the equivalent of C2IAdapter::setup_stack_frame back in
-  // the days we had adapter frames. When we deoptimize a situation where a
-  // compiled caller calls a compiled caller will have registers it expects
-  // to survive the call to the callee. If we deoptimize the callee the only
-  // way we can restore these registers is to have the oldest interpreter
-  // frame that we create restore these values. That is what this routine
-  // will accomplish.
-
-  // At the moment we have modified c2 to not have any callee save registers
-  // so this problem does not exist and this routine is just a place holder.
-
-  assert(f->is_interpreted_frame(), "must be interpreted");
-}
-
-
-//----------------------------------------------------------------------------------------------------
-// Exceptions

hotspot/src/cpu/x86/vm/interp_masm_x86.hpp

@@ -175,6 +175,7 @@ class InterpreterMacroAssembler: public MacroAssembler {
     movptr(rsp, Address(rbp, frame::interpreter_frame_monitor_block_top_offset * wordSize));
     // NULL last_sp until next java call
     movptr(Address(rbp, frame::interpreter_frame_last_sp_offset * wordSize), (int32_t)NULL_WORD);
+    NOT_LP64(empty_FPU_stack());
   }
 
   // Helpers for swap and dup

hotspot/src/cpu/x86/vm/interpreter_x86_32.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2014, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2015, 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
@@ -38,7 +38,6 @@
 #include "prims/jvmtiThreadState.hpp"
 #include "prims/methodHandles.hpp"
 #include "runtime/arguments.hpp"
-#include "runtime/deoptimization.hpp"
 #include "runtime/frame.inline.hpp"
 #include "runtime/sharedRuntime.hpp"
 #include "runtime/stubRoutines.hpp"
@@ -184,20 +183,3 @@ address InterpreterGenerator::generate_math_entry(AbstractInterpreter::MethodKin
 
   return entry_point;
 }
-
-
-void Deoptimization::unwind_callee_save_values(frame* f, vframeArray* vframe_array) {
-
-  // This code is sort of the equivalent of C2IAdapter::setup_stack_frame back in
-  // the days we had adapter frames. When we deoptimize a situation where a
-  // compiled caller calls a compiled caller will have registers it expects
-  // to survive the call to the callee. If we deoptimize the callee the only
-  // way we can restore these registers is to have the oldest interpreter
-  // frame that we create restore these values. That is what this routine
-  // will accomplish.
-
-  // At the moment we have modified c2 to not have any callee save registers
-  // so this problem does not exist and this routine is just a place holder.
-
-  assert(f->is_interpreted_frame(), "must be interpreted");
-}

hotspot/src/cpu/x86/vm/interpreter_x86_64.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2003, 2014, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2003, 2015, 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
@@ -38,7 +38,6 @@
 #include "prims/jvmtiThreadState.hpp"
 #include "prims/methodHandles.hpp"
 #include "runtime/arguments.hpp"
-#include "runtime/deoptimization.hpp"
 #include "runtime/frame.inline.hpp"
 #include "runtime/sharedRuntime.hpp"
 #include "runtime/stubRoutines.hpp"
@@ -298,19 +297,3 @@ address InterpreterGenerator::generate_math_entry(AbstractInterpreter::MethodKin
 
   return entry_point;
 }
-
-void Deoptimization::unwind_callee_save_values(frame* f, vframeArray* vframe_array) {
-
-  // This code is sort of the equivalent of C2IAdapter::setup_stack_frame back in
-  // the days we had adapter frames. When we deoptimize a situation where a
-  // compiled caller calls a compiled caller will have registers it expects
-  // to survive the call to the callee. If we deoptimize the callee the only
-  // way we can restore these registers is to have the oldest interpreter
-  // frame that we create restore these values. That is what this routine
-  // will accomplish.
-
-  // At the moment we have modified c2 to not have any callee save registers
-  // so this problem does not exist and this routine is just a place holder.
-
-  assert(f->is_interpreted_frame(), "must be interpreted");
-}

hotspot/src/cpu/x86/vm/templateInterpreterGenerator_x86_32.cpp

@@ -0,0 +1,305 @@
+/*
+ * Copyright (c) 1997, 2015, 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
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ *
+ */
+
+#include "precompiled.hpp"
+#include "asm/macroAssembler.hpp"
+#include "interpreter/interpreter.hpp"
+#include "interpreter/interpreterGenerator.hpp"
+#include "runtime/arguments.hpp"
+
+#define __ _masm->
+
+
+#ifndef CC_INTERP
+
+/**
+ * Method entry for static native methods:
+ *   int java.util.zip.CRC32.update(int crc, int b)
+ */
+address InterpreterGenerator::generate_CRC32_update_entry() {
+  if (UseCRC32Intrinsics) {
+    address entry = __ pc();
+
+    // rbx: Method*
+    // rsi: senderSP must preserved for slow path, set SP to it on fast path
+    // rdx: scratch
+    // rdi: scratch
+
+    Label slow_path;
+    // If we need a safepoint check, generate full interpreter entry.
+    ExternalAddress state(SafepointSynchronize::address_of_state());
+    __ cmp32(ExternalAddress(SafepointSynchronize::address_of_state()),
+             SafepointSynchronize::_not_synchronized);
+    __ jcc(Assembler::notEqual, slow_path);
+
+    // We don't generate local frame and don't align stack because
+    // we call stub code and there is no safepoint on this path.
+
+    // Load parameters
+    const Register crc = rax;  // crc
+    const Register val = rdx;  // source java byte value
+    const Register tbl = rdi;  // scratch
+
+    // Arguments are reversed on java expression stack
+    __ movl(val, Address(rsp,   wordSize)); // byte value
+    __ movl(crc, Address(rsp, 2*wordSize)); // Initial CRC
+
+    __ lea(tbl, ExternalAddress(StubRoutines::crc_table_addr()));
+    __ notl(crc); // ~crc
+    __ update_byte_crc32(crc, val, tbl);
+    __ notl(crc); // ~crc
+    // result in rax
+
+    // _areturn
+    __ pop(rdi);                // get return address
+    __ mov(rsp, rsi);           // set sp to sender sp
+    __ jmp(rdi);
+
+    // generate a vanilla native entry as the slow path
+    __ bind(slow_path);
+    __ jump_to_entry(Interpreter::entry_for_kind(Interpreter::native));
+    return entry;
+  }
+  return NULL;
+}
+
+/**
+ * Method entry for static native methods:
+ *   int java.util.zip.CRC32.updateBytes(int crc, byte[] b, int off, int len)
+ *   int java.util.zip.CRC32.updateByteBuffer(int crc, long buf, int off, int len)
+ */
+address InterpreterGenerator::generate_CRC32_updateBytes_entry(AbstractInterpreter::MethodKind kind) {
+  if (UseCRC32Intrinsics) {
+    address entry = __ pc();
+
+    // rbx,: Method*
+    // rsi: senderSP must preserved for slow path, set SP to it on fast path
+    // rdx: scratch
+    // rdi: scratch
+
+    Label slow_path;
+    // If we need a safepoint check, generate full interpreter entry.
+    ExternalAddress state(SafepointSynchronize::address_of_state());
+    __ cmp32(ExternalAddress(SafepointSynchronize::address_of_state()),
+             SafepointSynchronize::_not_synchronized);
+    __ jcc(Assembler::notEqual, slow_path);
+
+    // We don't generate local frame and don't align stack because
+    // we call stub code and there is no safepoint on this path.
+
+    // Load parameters
+    const Register crc = rax;  // crc
+    const Register buf = rdx;  // source java byte array address
+    const Register len = rdi;  // length
+
+    // value              x86_32
+    // interp. arg ptr    ESP + 4
+    // int java.util.zip.CRC32.updateBytes(int crc, byte[] b, int off, int len)
+    //                                         3           2      1        0
+    // int java.util.zip.CRC32.updateByteBuffer(int crc, long buf, int off, int len)
+    //                                              4         2,3      1        0
+
+    // Arguments are reversed on java expression stack
+    __ movl(len,   Address(rsp,   4 + 0)); // Length
+    // Calculate address of start element
+    if (kind == Interpreter::java_util_zip_CRC32_updateByteBuffer) {
+      __ movptr(buf, Address(rsp, 4 + 2 * wordSize)); // long buf
+      __ addptr(buf, Address(rsp, 4 + 1 * wordSize)); // + offset
+      __ movl(crc,   Address(rsp, 4 + 4 * wordSize)); // Initial CRC
+    } else {
+      __ movptr(buf, Address(rsp, 4 + 2 * wordSize)); // byte[] array
+      __ addptr(buf, arrayOopDesc::base_offset_in_bytes(T_BYTE)); // + header size
+      __ addptr(buf, Address(rsp, 4 + 1 * wordSize)); // + offset
+      __ movl(crc,   Address(rsp, 4 + 3 * wordSize)); // Initial CRC
+    }
+
+    __ super_call_VM_leaf(CAST_FROM_FN_PTR(address, StubRoutines::updateBytesCRC32()), crc, buf, len);
+    // result in rax
+
+    // _areturn
+    __ pop(rdi);                // get return address
+    __ mov(rsp, rsi);           // set sp to sender sp
+    __ jmp(rdi);
+
+    // generate a vanilla native entry as the slow path
+    __ bind(slow_path);
+    __ jump_to_entry(Interpreter::entry_for_kind(Interpreter::native));
+    return entry;
+  }
+  return NULL;
+}
+
+/**
+* Method entry for static native methods:
+*   int java.util.zip.CRC32C.updateBytes(int crc, byte[] b, int off, int end)
+*   int java.util.zip.CRC32C.updateByteBuffer(int crc, long address, int off, int end)
+*/
+address InterpreterGenerator::generate_CRC32C_updateBytes_entry(AbstractInterpreter::MethodKind kind) {
+  if (UseCRC32CIntrinsics) {
+    address entry = __ pc();
+    // Load parameters
+    const Register crc = rax;  // crc
+    const Register buf = rcx;  // source java byte array address
+    const Register len = rdx;  // length
+    const Register end = len;
+
+    // value              x86_32
+    // interp. arg ptr    ESP + 4
+    // int java.util.zip.CRC32.updateBytes(int crc, byte[] b, int off, int end)
+    //                                         3           2      1        0
+    // int java.util.zip.CRC32.updateByteBuffer(int crc, long address, int off, int end)
+    //                                              4         2,3          1        0
+
+    // Arguments are reversed on java expression stack
+    __ movl(end, Address(rsp, 4 + 0)); // end
+    __ subl(len, Address(rsp, 4 + 1 * wordSize));  // end - offset == length
+    // Calculate address of start element
+    if (kind == Interpreter::java_util_zip_CRC32C_updateDirectByteBuffer) {
+      __ movptr(buf, Address(rsp, 4 + 2 * wordSize)); // long address
+      __ addptr(buf, Address(rsp, 4 + 1 * wordSize)); // + offset
+      __ movl(crc, Address(rsp, 4 + 4 * wordSize)); // Initial CRC
+    } else {
+      __ movptr(buf, Address(rsp, 4 + 2 * wordSize)); // byte[] array
+      __ addptr(buf, arrayOopDesc::base_offset_in_bytes(T_BYTE)); // + header size
+      __ addptr(buf, Address(rsp, 4 + 1 * wordSize)); // + offset
+      __ movl(crc, Address(rsp, 4 + 3 * wordSize)); // Initial CRC
+    }
+    __ super_call_VM_leaf(CAST_FROM_FN_PTR(address, StubRoutines::updateBytesCRC32C()), crc, buf, len);
+    // result in rax
+    // _areturn
+    __ pop(rdi);                // get return address
+    __ mov(rsp, rsi);           // set sp to sender sp
+    __ jmp(rdi);
+
+    return entry;
+  }
+  return NULL;
+}
+
+/**
+ * Method entry for static native method:
+ *    java.lang.Float.intBitsToFloat(int bits)
+ */
+address InterpreterGenerator::generate_Float_intBitsToFloat_entry() {
+  if (UseSSE >= 1) {
+    address entry = __ pc();
+
+    // rsi: the sender's SP
+
+    // Skip safepoint check (compiler intrinsic versions of this method
+    // do not perform safepoint checks either).
+
+    // Load 'bits' into xmm0 (interpreter returns results in xmm0)
+    __ movflt(xmm0, Address(rsp, wordSize));
+
+    // Return
+    __ pop(rdi); // get return address
+    __ mov(rsp, rsi); // set rsp to the sender's SP
+    __ jmp(rdi);
+    return entry;
+  }
+
+  return NULL;
+}
+
+/**
+ * Method entry for static native method:
+ *    java.lang.Float.floatToRawIntBits(float value)
+ */
+address InterpreterGenerator::generate_Float_floatToRawIntBits_entry() {
+  if (UseSSE >= 1) {
+    address entry = __ pc();
+
+    // rsi: the sender's SP
+
+    // Skip safepoint check (compiler intrinsic versions of this method
+    // do not perform safepoint checks either).
+
+    // Load the parameter (a floating-point value) into rax.
+    __ movl(rax, Address(rsp, wordSize));
+
+    // Return
+    __ pop(rdi); // get return address
+    __ mov(rsp, rsi); // set rsp to the sender's SP
+    __ jmp(rdi);
+    return entry;
+  }
+
+  return NULL;
+}
+
+
+/**
+ * Method entry for static native method:
+ *    java.lang.Double.longBitsToDouble(long bits)
+ */
+address InterpreterGenerator::generate_Double_longBitsToDouble_entry() {
+   if (UseSSE >= 2) {
+     address entry = __ pc();
+
+     // rsi: the sender's SP
+
+     // Skip safepoint check (compiler intrinsic versions of this method
+     // do not perform safepoint checks either).
+
+     // Load 'bits' into xmm0 (interpreter returns results in xmm0)
+     __ movdbl(xmm0, Address(rsp, wordSize));
+
+     // Return
+     __ pop(rdi); // get return address
+     __ mov(rsp, rsi); // set rsp to the sender's SP
+     __ jmp(rdi);
+     return entry;
+   }
+
+   return NULL;
+}
+
+/**
+ * Method entry for static native method:
+ *    java.lang.Double.doubleToRawLongBits(double value)
+ */
+address InterpreterGenerator::generate_Double_doubleToRawLongBits_entry() {
+  if (UseSSE >= 2) {
+    address entry = __ pc();
+
+    // rsi: the sender's SP
+
+    // Skip safepoint check (compiler intrinsic versions of this method
+    // do not perform safepoint checks either).
+
+    // Load the parameter (a floating-point value) into rax.
+    __ movl(rdx, Address(rsp, 2*wordSize));
+    __ movl(rax, Address(rsp, wordSize));
+
+    // Return
+    __ pop(rdi); // get return address
+    __ mov(rsp, rsi); // set rsp to the sender's SP
+    __ jmp(rdi);
+    return entry;
+  }
+
+  return NULL;
+}
+#endif // CC_INTERP

hotspot/src/cpu/x86/vm/templateInterpreterGenerator_x86_64.cpp

@@ -0,0 +1,197 @@
+/*
+ * Copyright (c) 2003, 2015, 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
+ * under the terms of the GNU General Public License version 2 only, as
+ * published by the Free Software Foundation.
+ *
+ * This code is distributed in the hope that it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
+ * version 2 for more details (a copy is included in the LICENSE file that
+ * accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License version
+ * 2 along with this work; if not, write to the Free Software Foundation,
+ * Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.
+ *
+ * Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA
+ * or visit www.oracle.com if you need additional information or have any
+ * questions.
+ *
+ */
+
+#include "precompiled.hpp"
+#include "asm/macroAssembler.hpp"
+#include "interpreter/interpreter.hpp"
+#include "interpreter/interpreterGenerator.hpp"
+#include "runtime/arguments.hpp"
+
+#define __ _masm->
+
+#ifndef CC_INTERP
+
+/**
+ * Method entry for static native methods:
+ *   int java.util.zip.CRC32.update(int crc, int b)
+ */
+address InterpreterGenerator::generate_CRC32_update_entry() {
+  if (UseCRC32Intrinsics) {
+    address entry = __ pc();
+
+    // rbx,: Method*
+    // r13: senderSP must preserved for slow path, set SP to it on fast path
+    // c_rarg0: scratch (rdi on non-Win64, rcx on Win64)
+    // c_rarg1: scratch (rsi on non-Win64, rdx on Win64)
+
+    Label slow_path;
+    // If we need a safepoint check, generate full interpreter entry.
+    ExternalAddress state(SafepointSynchronize::address_of_state());
+    __ cmp32(ExternalAddress(SafepointSynchronize::address_of_state()),
+             SafepointSynchronize::_not_synchronized);
+    __ jcc(Assembler::notEqual, slow_path);
+
+    // We don't generate local frame and don't align stack because
+    // we call stub code and there is no safepoint on this path.
+
+    // Load parameters
+    const Register crc = rax;  // crc
+    const Register val = c_rarg0;  // source java byte value
+    const Register tbl = c_rarg1;  // scratch
+
+    // Arguments are reversed on java expression stack
+    __ movl(val, Address(rsp,   wordSize)); // byte value
+    __ movl(crc, Address(rsp, 2*wordSize)); // Initial CRC
+
+    __ lea(tbl, ExternalAddress(StubRoutines::crc_table_addr()));
+    __ notl(crc); // ~crc
+    __ update_byte_crc32(crc, val, tbl);
+    __ notl(crc); // ~crc
+    // result in rax
+
+    // _areturn
+    __ pop(rdi);                // get return address
+    __ mov(rsp, r13);           // set sp to sender sp
+    __ jmp(rdi);
+
+    // generate a vanilla native entry as the slow path
+    __ bind(slow_path);
+    __ jump_to_entry(Interpreter::entry_for_kind(Interpreter::native));
+    return entry;
+  }
+  return NULL;
+}
+
+/**
+ * Method entry for static native methods:
+ *   int java.util.zip.CRC32.updateBytes(int crc, byte[] b, int off, int len)
+ *   int java.util.zip.CRC32.updateByteBuffer(int crc, long buf, int off, int len)
+ */
+address InterpreterGenerator::generate_CRC32_updateBytes_entry(AbstractInterpreter::MethodKind kind) {
+  if (UseCRC32Intrinsics) {
+    address entry = __ pc();
+
+    // rbx,: Method*
+    // r13: senderSP must preserved for slow path, set SP to it on fast path
+
+    Label slow_path;
+    // If we need a safepoint check, generate full interpreter entry.
+    ExternalAddress state(SafepointSynchronize::address_of_state());
+    __ cmp32(ExternalAddress(SafepointSynchronize::address_of_state()),
+             SafepointSynchronize::_not_synchronized);
+    __ jcc(Assembler::notEqual, slow_path);
+
+    // We don't generate local frame and don't align stack because
+    // we call stub code and there is no safepoint on this path.
+
+    // Load parameters
+    const Register crc = c_rarg0;  // crc
+    const Register buf = c_rarg1;  // source java byte array address
+    const Register len = c_rarg2;  // length
+    const Register off = len;      // offset (never overlaps with 'len')
+
+    // Arguments are reversed on java expression stack
+    // Calculate address of start element
+    if (kind == Interpreter::java_util_zip_CRC32_updateByteBuffer) {
+      __ movptr(buf, Address(rsp, 3*wordSize)); // long buf
+      __ movl2ptr(off, Address(rsp, 2*wordSize)); // offset
+      __ addq(buf, off); // + offset
+      __ movl(crc,   Address(rsp, 5*wordSize)); // Initial CRC
+    } else {
+      __ movptr(buf, Address(rsp, 3*wordSize)); // byte[] array
+      __ addptr(buf, arrayOopDesc::base_offset_in_bytes(T_BYTE)); // + header size
+      __ movl2ptr(off, Address(rsp, 2*wordSize)); // offset
+      __ addq(buf, off); // + offset
+      __ movl(crc,   Address(rsp, 4*wordSize)); // Initial CRC
+    }
+    // Can now load 'len' since we're finished with 'off'
+    __ movl(len, Address(rsp, wordSize)); // Length
+
+    __ super_call_VM_leaf(CAST_FROM_FN_PTR(address, StubRoutines::updateBytesCRC32()), crc, buf, len);
+    // result in rax
+
+    // _areturn
+    __ pop(rdi);                // get return address
+    __ mov(rsp, r13);           // set sp to sender sp
+    __ jmp(rdi);
+
+    // generate a vanilla native entry as the slow path
+    __ bind(slow_path);
+    __ jump_to_entry(Interpreter::entry_for_kind(Interpreter::native));
+    return entry;
+  }
+  return NULL;
+}
+
+/**
+* Method entry for static native methods:
+*   int java.util.zip.CRC32C.updateBytes(int crc, byte[] b, int off, int end)
+*   int java.util.zip.CRC32C.updateByteBuffer(int crc, long address, int off, int end)
+*/
+address InterpreterGenerator::generate_CRC32C_updateBytes_entry(AbstractInterpreter::MethodKind kind) {
+  if (UseCRC32CIntrinsics) {
+    address entry = __ pc();
+    // Load parameters
+    const Register crc = c_rarg0;  // crc
+    const Register buf = c_rarg1;  // source java byte array address
+    const Register len = c_rarg2;
+    const Register off = c_rarg3;  // offset
+    const Register end = len;
+
+    // Arguments are reversed on java expression stack
+    // Calculate address of start element
+    if (kind == Interpreter::java_util_zip_CRC32C_updateDirectByteBuffer) {
+      __ movptr(buf, Address(rsp, 3 * wordSize)); // long buf
+      __ movl2ptr(off, Address(rsp, 2 * wordSize)); // offset
+      __ addq(buf, off); // + offset
+      __ movl(crc, Address(rsp, 5 * wordSize)); // Initial CRC
+      // Note on 5 * wordSize vs. 4 * wordSize:
+      // *   int java.util.zip.CRC32C.updateByteBuffer(int crc, long address, int off, int end)
+      //                                                   4         2,3          1        0
+      // end starts at SP + 8
+      // The Java(R) Virtual Machine Specification Java SE 7 Edition
+      // 4.10.2.3. Values of Types long and double
+      //    "When calculating operand stack length, values of type long and double have length two."
+    } else {
+      __ movptr(buf, Address(rsp, 3 * wordSize)); // byte[] array
+      __ addptr(buf, arrayOopDesc::base_offset_in_bytes(T_BYTE)); // + header size
+      __ movl2ptr(off, Address(rsp, 2 * wordSize)); // offset
+      __ addq(buf, off); // + offset
+      __ movl(crc, Address(rsp, 4 * wordSize)); // Initial CRC
+    }
+    __ movl(end, Address(rsp, wordSize)); // end
+    __ subl(end, off); // end - off
+    __ super_call_VM_leaf(CAST_FROM_FN_PTR(address, StubRoutines::updateBytesCRC32C()), crc, buf, len);
+    // result in rax
+    // _areturn
+    __ pop(rdi);                // get return address
+    __ mov(rsp, r13);           // set sp to sender sp
+    __ jmp(rdi);
+
+    return entry;
+  }
+
+  return NULL;
+}
+#endif // ! CC_INTERP

hotspot/src/cpu/x86/vm/templateInterpreter_x86.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 1997, 2014, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 1997, 2015, 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
@@ -116,4 +116,87 @@ void AbstractInterpreter::layout_activation(Method* method,
     method->constants()->cache();
 }
 
+#ifndef _LP64
+int AbstractInterpreter::BasicType_as_index(BasicType type) {
+  int i = 0;
+  switch (type) {
+    case T_BOOLEAN: i = 0; break;
+    case T_CHAR   : i = 1; break;
+    case T_BYTE   : i = 2; break;
+    case T_SHORT  : i = 3; break;
+    case T_INT    : // fall through
+    case T_LONG   : // fall through
+    case T_VOID   : i = 4; break;
+    case T_FLOAT  : i = 5; break;  // have to treat float and double separately for SSE
+    case T_DOUBLE : i = 6; break;
+    case T_OBJECT : // fall through
+    case T_ARRAY  : i = 7; break;
+    default       : ShouldNotReachHere();
+  }
+  assert(0 <= i && i < AbstractInterpreter::number_of_result_handlers, "index out of bounds");
+  return i;
+}
+#else
+int AbstractInterpreter::BasicType_as_index(BasicType type) {
+  int i = 0;
+  switch (type) {
+    case T_BOOLEAN: i = 0; break;
+    case T_CHAR   : i = 1; break;
+    case T_BYTE   : i = 2; break;
+    case T_SHORT  : i = 3; break;
+    case T_INT    : i = 4; break;
+    case T_LONG   : i = 5; break;
+    case T_VOID   : i = 6; break;
+    case T_FLOAT  : i = 7; break;
+    case T_DOUBLE : i = 8; break;
+    case T_OBJECT : i = 9; break;
+    case T_ARRAY  : i = 9; break;
+    default       : ShouldNotReachHere();
+  }
+  assert(0 <= i && i < AbstractInterpreter::number_of_result_handlers,
+         "index out of bounds");
+  return i;
+}
+#endif // _LP64
+
+// These should never be compiled since the interpreter will prefer
+// the compiled version to the intrinsic version.
+bool AbstractInterpreter::can_be_compiled(methodHandle m) {
+  switch (method_kind(m)) {
+    case Interpreter::java_lang_math_sin     : // fall thru
+    case Interpreter::java_lang_math_cos     : // fall thru
+    case Interpreter::java_lang_math_tan     : // fall thru
+    case Interpreter::java_lang_math_abs     : // fall thru
+    case Interpreter::java_lang_math_log     : // fall thru
+    case Interpreter::java_lang_math_log10   : // fall thru
+    case Interpreter::java_lang_math_sqrt    : // fall thru
+    case Interpreter::java_lang_math_pow     : // fall thru
+    case Interpreter::java_lang_math_exp     :
+      return false;
+    default:
+      return true;
+  }
+}
+
+// How much stack a method activation needs in words.
+int AbstractInterpreter::size_top_interpreter_activation(Method* method) {
+  const int entry_size = frame::interpreter_frame_monitor_size();
+
+  // total overhead size: entry_size + (saved rbp thru expr stack
+  // bottom).  be sure to change this if you add/subtract anything
+  // to/from the overhead area
+  const int overhead_size =
+    -(frame::interpreter_frame_initial_sp_offset) + entry_size;
+
+#ifndef _LP64
+  const int stub_code = 4;  // see generate_call_stub
+#else
+  const int stub_code = frame::entry_frame_after_call_words;
+#endif
+
+  const int method_stack = (method->max_locals() + method->max_stack()) *
+                           Interpreter::stackElementWords;
+  return (overhead_size + method_stack + stub_code);
+}
+
 #endif // CC_INTERP

hotspot/src/cpu/x86/vm/templateTable_x86.cpp

@@ -43,8 +43,8 @@
 #define __ _masm->
 
 // Global Register Names
-Register rbcp     = LP64_ONLY(r13) NOT_LP64(rsi);
+static const Register rbcp     = LP64_ONLY(r13) NOT_LP64(rsi);
-Register rlocals  = LP64_ONLY(r14) NOT_LP64(rdi);
+static const Register rlocals  = LP64_ONLY(r14) NOT_LP64(rdi);
 
 // Platform-dependent initialization
 void TemplateTable::pd_initialize() {

hotspot/src/cpu/zero/vm/interpreter_zero.cpp

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2003, 2013, Oracle and/or its affiliates. All rights reserved.
+ * Copyright (c) 2003, 2015, Oracle and/or its affiliates. All rights reserved.
  * Copyright 2007, 2008, 2009, 2010, 2011 Red Hat, Inc.
  * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
  *
@@ -38,7 +38,6 @@
 #include "prims/jvmtiThreadState.hpp"
 #include "prims/methodHandles.hpp"
 #include "runtime/arguments.hpp"
-#include "runtime/deoptimization.hpp"
 #include "runtime/frame.inline.hpp"
 #include "runtime/sharedRuntime.hpp"
 #include "runtime/stubRoutines.hpp"
@@ -74,7 +73,3 @@ address InterpreterGenerator::generate_abstract_entry() {
 bool AbstractInterpreter::can_be_compiled(methodHandle m) {
   return true;
 }
-
-void Deoptimization::unwind_callee_save_values(frame* f,
-                                               vframeArray* vframe_array) {
-}

hotspot/src/share/vm/runtime/deoptimization.cpp

@@ -571,6 +571,23 @@ void Deoptimization::cleanup_deopt_info(JavaThread *thread,
   thread->dec_in_deopt_handler();
 }
 
+// Moved from cpu directories because none of the cpus has callee save values.
+// If a cpu implements callee save values, move this to deoptimization_<cpu>.cpp.
+void Deoptimization::unwind_callee_save_values(frame* f, vframeArray* vframe_array) {
+
+  // This code is sort of the equivalent of C2IAdapter::setup_stack_frame back in
+  // the days we had adapter frames. When we deoptimize a situation where a
+  // compiled caller calls a compiled caller will have registers it expects
+  // to survive the call to the callee. If we deoptimize the callee the only
+  // way we can restore these registers is to have the oldest interpreter
+  // frame that we create restore these values. That is what this routine
+  // will accomplish.
+
+  // At the moment we have modified c2 to not have any callee save registers
+  // so this problem does not exist and this routine is just a place holder.
+
+  assert(f->is_interpreted_frame(), "must be interpreted");
+}
 
 // Return BasicType of value being returned
 JRT_LEAF(BasicType, Deoptimization::unpack_frames(JavaThread* thread, int exec_mode))