stan/jdk-24
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Merge

Browse Source
This commit is contained in:
Alejandro Murillo 2016-09-30 02:52:38 -07:00
commit f0393d5b01
202 changed files with 4987 additions and 2255 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
hotspot/make/gensrc/GensrcJvmti.gmk
View File

@ -67,7 +67,8 @@ define SetupXslTransformBody
$$($1_OUTPUT_DIR)/$1: $$($1_XML_FILE) $$($1_XSL_FILE) $$($1_DEPS) $$(BUILD_JVMTI_TOOLS)
$$(call LogInfo, Generating $$(@F))
$$(call MakeDir, $$(@D))
$$(call ExecuteWithLog, $$@, $$(TOOL_JVMTI_GEN) -IN $$($1_XML_FILE) -XSL $$($1_XSL_FILE) -OUT $$@ $$($1_ARGS))
$$(call ExecuteWithLog, $$@, $$(TOOL_JVMTI_GEN) -IN $$($1_XML_FILE) \
-XSL $$($1_XSL_FILE) -OUT $$@ $$($1_ARGS))
# jvmtiGen does not return error code properly on fail.
# NOTE: We should really fix jvmtiGen.java instead.
test -f $$@
@ -134,8 +135,8 @@ TARGETS += $(JVMTI_OUTPUTDIR)/jvmtiEnvRecommended.cpp
TRACE_OUTPUTDIR := $(JVM_VARIANT_OUTPUTDIR)/gensrc/tracefiles
TRACE_SRCDIR := $(HOTSPOT_TOPDIR)/src/share/vm/trace
# Append directories to search (might have been set by custom extensions)
TRACE_SEARCH_DIRS += $(TRACE_SRCDIR)
# Append list of XSL files to search (might have been set by custom extensions)
TRACE_XSL_FILES += $(wildcard $(TRACE_SRCDIR)/*.xsl)
TRACE_XML ?= $(TRACE_SRCDIR)/trace.xml
@ -155,7 +156,7 @@ TRACE_DEPS += \
define SetupTraceGeneration
$$(eval $$(call SetupXslTransform, $1, \
XML_FILE := $$(TRACE_XML), \
XSL_FILE := $$(firstword $$(wildcard $$(addsuffix /$$(basename $1).xsl, $$(TRACE_SEARCH_DIRS)))), \
XSL_FILE := $$(firstword $$(filter %/$$(basename $1).xsl, $$(TRACE_XSL_FILES))), \
OUTPUT_DIR := $$(TRACE_OUTPUTDIR), \
DEPS := $$(TRACE_DEPS), \
))

1
hotspot/make/lib/JvmOverrideFiles.gmk
View File

@ -31,6 +31,7 @@ $(eval $(call IncludeCustomExtension, hotspot, lib/JvmOverrideFiles.gmk))
ifeq ($(TOOLCHAIN_TYPE), gcc)
BUILD_LIBJVM_vmStructs.cpp_CXXFLAGS := -fno-var-tracking-assignments -O0
BUILD_LIBJVM_jvmciCompilerToVM.cpp_CXXFLAGS := -fno-var-tracking-assignments
endif
ifeq ($(OPENJDK_TARGET_OS), linux)

6
hotspot/make/test/JtregNative.gmk
View File

@ -55,6 +55,9 @@ BUILD_HOTSPOT_JTREG_NATIVE_SRC := \
$(HOTSPOT_TOPDIR)/test/testlibrary/jvmti \
$(HOTSPOT_TOPDIR)/test/compiler/jvmci/jdk.vm.ci.code.test \
$(HOTSPOT_TOPDIR)/test/serviceability/jvmti/GetModulesInfo \
$(HOTSPOT_TOPDIR)/test/serviceability/jvmti/ModuleAwareAgents/ClassFileLoadHook \
$(HOTSPOT_TOPDIR)/test/serviceability/jvmti/ModuleAwareAgents/ClassLoadPrepare \
$(HOTSPOT_TOPDIR)/test/serviceability/jvmti/ModuleAwareAgents/ThreadStart \
#
# Add conditional directories here when needed.
@ -75,6 +78,9 @@ ifeq ($(TOOLCHAIN_TYPE), solstudio)
BUILD_HOTSPOT_JTREG_LIBRARIES_LIBS_liboverflow := -lc
BUILD_HOTSPOT_JTREG_LIBRARIES_LIBS_libSimpleClassFileLoadHook := -lc
BUILD_HOTSPOT_JTREG_LIBRARIES_LIBS_libGetNamedModuleTest := -lc
BUILD_HOTSPOT_JTREG_LIBRARIES_LIBS_libMAAClassFileLoadHook := -lc
BUILD_HOTSPOT_JTREG_LIBRARIES_LIBS_libMAAClassLoadPrepare := -lc
BUILD_HOTSPOT_JTREG_LIBRARIES_LIBS_libMAAThreadStart := -lc
endif
ifeq ($(OPENJDK_TARGET_OS), linux)

31
hotspot/src/cpu/aarch64/vm/c1_LIRGenerator_aarch64.cpp
View File

@ -140,10 +140,11 @@ LIR_Opr LIRGenerator::safepoint_poll_register() {
LIR_Address* LIRGenerator::generate_address(LIR_Opr base, LIR_Opr index,
int shift, int disp, BasicType type) {
assert(base->is_register(), "must be");
intx large_disp = disp;
// accumulate fixed displacements
if (index->is_constant()) {
disp += index->as_constant_ptr()->as_jint() << shift;
large_disp += (intx)(index->as_constant_ptr()->as_jint()) << shift;
index = LIR_OprFact::illegalOpr;
}
@ -154,31 +155,31 @@ LIR_Address* LIRGenerator::generate_address(LIR_Opr base, LIR_Opr index,
__ shift_left(index, shift, tmp);
index = tmp;
}
if (disp != 0) {
if (large_disp != 0) {
LIR_Opr tmp = new_pointer_register();
if (Assembler::operand_valid_for_add_sub_immediate(disp)) {
__ add(tmp, tmp, LIR_OprFact::intptrConst(disp));
if (Assembler::operand_valid_for_add_sub_immediate(large_disp)) {
__ add(tmp, tmp, LIR_OprFact::intptrConst(large_disp));
index = tmp;
} else {
__ move(tmp, LIR_OprFact::intptrConst(disp));
__ move(tmp, LIR_OprFact::intptrConst(large_disp));
__ add(tmp, index, tmp);
index = tmp;
}
disp = 0;
large_disp = 0;
}
} else if (disp != 0 && !Address::offset_ok_for_immed(disp, shift)) {
} else if (large_disp != 0 && !Address::offset_ok_for_immed(large_disp, shift)) {
// index is illegal so replace it with the displacement loaded into a register
index = new_pointer_register();
__ move(LIR_OprFact::intptrConst(disp), index);
disp = 0;
__ move(LIR_OprFact::intptrConst(large_disp), index);
large_disp = 0;
}
// at this point we either have base + index or base + displacement
if (disp == 0) {
if (large_disp == 0) {
return new LIR_Address(base, index, type);
} else {
assert(Address::offset_ok_for_immed(disp, 0), "must be");
return new LIR_Address(base, disp, type);
assert(Address::offset_ok_for_immed(large_disp, 0), "must be");
return new LIR_Address(base, large_disp, type);
}
}
@ -192,7 +193,7 @@ LIR_Address* LIRGenerator::emit_array_address(LIR_Opr array_opr, LIR_Opr index_o
LIR_Address* addr;
if (index_opr->is_constant()) {
addr = new LIR_Address(array_opr,
offset_in_bytes + index_opr->as_jint() * elem_size, type);
offset_in_bytes + (intx)(index_opr->as_jint()) * elem_size, type);
} else {
if (offset_in_bytes) {
LIR_Opr tmp = new_pointer_register();
@ -1032,6 +1033,10 @@ void LIRGenerator::do_update_CRC32C(Intrinsic* x) {
Unimplemented();
}
void LIRGenerator::do_FmaIntrinsic(Intrinsic* x) {
fatal("FMA intrinsic is not implemented on this platform");
}
void LIRGenerator::do_vectorizedMismatch(Intrinsic* x) {
fatal("vectorizedMismatch intrinsic is not implemented on this platform");
}

2
hotspot/src/cpu/aarch64/vm/interp_masm_aarch64.cpp
View File

@ -327,7 +327,7 @@ void InterpreterMacroAssembler::push_i(Register r) {
void InterpreterMacroAssembler::push_l(Register r) {
str(zr, pre(esp, -wordSize));
str(r, pre(esp, -wordsize));
str(r, pre(esp, - wordSize));
}
void InterpreterMacroAssembler::pop_f(FloatRegister r) {

5
hotspot/src/cpu/aarch64/vm/vm_version_aarch64.cpp
View File

@ -262,6 +262,11 @@ void VM_Version::get_processor_features() {
FLAG_SET_DEFAULT(UseCRC32CIntrinsics, false);
}
if (UseFMA) {
warning("FMA instructions are not available on this CPU");
FLAG_SET_DEFAULT(UseFMA, false);
}
if (auxv & (HWCAP_SHA1 | HWCAP_SHA2)) {
if (FLAG_IS_DEFAULT(UseSHA)) {
FLAG_SET_DEFAULT(UseSHA, true);

7
hotspot/src/cpu/ppc/vm/assembler_ppc.hpp
View File

@ -506,6 +506,8 @@ class Assembler : public AbstractAssembler {
// Vector-Scalar (VSX) instruction support.
LXVD2X_OPCODE = (31u << OPCODE_SHIFT | 844u << 1),
STXVD2X_OPCODE = (31u << OPCODE_SHIFT | 972u << 1),
MTVSRD_OPCODE = (31u << OPCODE_SHIFT | 179u << 1),
MFVSRD_OPCODE = (31u << OPCODE_SHIFT | 51u << 1),
// Vector Permute and Formatting
VPKPX_OPCODE = (4u << OPCODE_SHIFT | 782u ),
@ -1573,6 +1575,9 @@ class Assembler : public AbstractAssembler {
inline void stdu( Register d, int si16, Register s1);
inline void stdux(Register s, Register a, Register b);
inline void st_ptr(Register d, int si16, Register s1);
DEBUG_ONLY(inline void st_ptr(Register d, ByteSize b, Register s1);)
// PPC 1, section 3.3.13 Move To/From System Register Instructions
inline void mtlr( Register s1);
inline void mflr( Register d);
@ -2099,6 +2104,8 @@ class Assembler : public AbstractAssembler {
// Vector-Scalar (VSX) instructions.
inline void lxvd2x( VectorSRegister d, Register a, Register b);
inline void stxvd2x( VectorSRegister d, Register a, Register b);
inline void mtvrd( VectorRegister d, Register a);
inline void mfvrd( Register a, VectorRegister d);
// AES (introduced with Power 8)
inline void vcipher( VectorRegister d, VectorRegister a, VectorRegister b);

5
hotspot/src/cpu/ppc/vm/assembler_ppc.inline.hpp
View File

@ -349,6 +349,9 @@ inline void Assembler::stdx( Register d, Register s1, Register s2) { emit_int32(
inline void Assembler::stdu( Register d, int si16, Register s1) { emit_int32(STDU_OPCODE | rs(d) | ds(si16) | rta0mem(s1));}
inline void Assembler::stdux(Register s, Register a, Register b) { emit_int32(STDUX_OPCODE| rs(s) | rta0mem(a) | rb(b));}
inline void Assembler::st_ptr(Register d, int b, Register s1) { std(d, b, s1); }
DEBUG_ONLY(inline void Assembler::st_ptr(Register d, ByteSize b, Register s1) { std(d, in_bytes(b), s1); })
// PPC 1, section 3.3.13 Move To/From System Register Instructions
inline void Assembler::mtlr( Register s1) { emit_int32(MTLR_OPCODE | rs(s1)); }
inline void Assembler::mflr( Register d ) { emit_int32(MFLR_OPCODE | rt(d)); }
@ -733,6 +736,8 @@ inline void Assembler::lvsr( VectorRegister d, Register s1, Register s2) { emit
// Vector-Scalar (VSX) instructions.
inline void Assembler::lxvd2x (VectorSRegister d, Register s1, Register s2) { emit_int32( LXVD2X_OPCODE | vsrt(d) | ra(s1) | rb(s2)); }
inline void Assembler::stxvd2x(VectorSRegister d, Register s1, Register s2) { emit_int32( STXVD2X_OPCODE | vsrt(d) | ra(s1) | rb(s2)); }
inline void Assembler::mtvrd( VectorRegister d, Register a) { emit_int32( MTVSRD_OPCODE | vrt(d) | ra(a) | 1u); } // 1u: d is treated as Vector (VMX/Altivec).
inline void Assembler::mfvrd( Register a, VectorRegister d) { emit_int32( MFVSRD_OPCODE | vrt(d) | ra(a) | 1u); } // 1u: d is treated as Vector (VMX/Altivec).
inline void Assembler::vpkpx( VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VPKPX_OPCODE | vrt(d) | vra(a) | vrb(b)); }
inline void Assembler::vpkshss( VectorRegister d, VectorRegister a, VectorRegister b) { emit_int32( VPKSHSS_OPCODE | vrt(d) | vra(a) | vrb(b)); }

35
hotspot/src/cpu/ppc/vm/c1_LIRGenerator_ppc.cpp
View File

@ -157,10 +157,11 @@ LIR_Opr LIRGenerator::safepoint_poll_register() {
LIR_Address* LIRGenerator::generate_address(LIR_Opr base, LIR_Opr index,
int shift, int disp, BasicType type) {
assert(base->is_register(), "must be");
intx large_disp = disp;
// Accumulate fixed displacements.
if (index->is_constant()) {
disp += index->as_constant_ptr()->as_jint() << shift;
large_disp += (intx)(index->as_constant_ptr()->as_jint()) << shift;
index = LIR_OprFact::illegalOpr;
}
@ -171,31 +172,31 @@ LIR_Address* LIRGenerator::generate_address(LIR_Opr base, LIR_Opr index,
__ shift_left(index, shift, tmp);
index = tmp;
}
if (disp != 0) {
if (large_disp != 0) {
LIR_Opr tmp = new_pointer_register();
if (Assembler::is_simm16(disp)) {
__ add(index, LIR_OprFact::intptrConst(disp), tmp);
if (Assembler::is_simm16(large_disp)) {
__ add(index, LIR_OprFact::intptrConst(large_disp), tmp);
index = tmp;
} else {
__ move(LIR_OprFact::intptrConst(disp), tmp);
__ move(LIR_OprFact::intptrConst(large_disp), tmp);
__ add(tmp, index, tmp);
index = tmp;
}
disp = 0;
large_disp = 0;
}
} else if (!Assembler::is_simm16(disp)) {
} else if (!Assembler::is_simm16(large_disp)) {
// Index is illegal so replace it with the displacement loaded into a register.
index = new_pointer_register();
__ move(LIR_OprFact::intptrConst(disp), index);
disp = 0;
__ move(LIR_OprFact::intptrConst(large_disp), index);
large_disp = 0;
}
// At this point we either have base + index or base + displacement.
if (disp == 0) {
if (large_disp == 0) {
return new LIR_Address(base, index, type);
} else {
assert(Assembler::is_simm16(disp), "must be");
return new LIR_Address(base, disp, type);
assert(Assembler::is_simm16(large_disp), "must be");
return new LIR_Address(base, large_disp, type);
}
}
@ -206,11 +207,11 @@ LIR_Address* LIRGenerator::emit_array_address(LIR_Opr array_opr, LIR_Opr index_o
int shift = exact_log2(elem_size);
LIR_Opr base_opr;
int offset = arrayOopDesc::base_offset_in_bytes(type);
intx offset = arrayOopDesc::base_offset_in_bytes(type);
if (index_opr->is_constant()) {
int i = index_opr->as_constant_ptr()->as_jint();
int array_offset = i * elem_size;
intx i = index_opr->as_constant_ptr()->as_jint();
intx array_offset = i * elem_size;
if (Assembler::is_simm16(array_offset + offset)) {
base_opr = array_opr;
offset = array_offset + offset;
@ -1433,6 +1434,10 @@ void LIRGenerator::do_update_CRC32(Intrinsic* x) {
}
}
void LIRGenerator::do_FmaIntrinsic(Intrinsic* x) {
fatal("FMA intrinsic is not implemented on this platform");
}
void LIRGenerator::do_vectorizedMismatch(Intrinsic* x) {
fatal("vectorizedMismatch intrinsic is not implemented on this platform");
}

3
hotspot/src/cpu/ppc/vm/interp_masm_ppc_64.cpp
View File

@ -261,6 +261,9 @@ void InterpreterMacroAssembler::push_ptr(Register r) {
}
void InterpreterMacroAssembler::push_l(Register r) {
// Clear unused slot.
load_const_optimized(R0, 0L);
std(R0, 0, R15_esp);
std(r, - Interpreter::stackElementSize, R15_esp);
addi(R15_esp, R15_esp, - 2 * Interpreter::stackElementSize );
}

559
hotspot/src/cpu/ppc/vm/macroAssembler_ppc.cpp
View File

@ -4332,6 +4332,565 @@ void MacroAssembler::kernel_crc32_1byte(Register crc, Register buf, Register len
BLOCK_COMMENT("} kernel_crc32_1byte");
}
/**
* @param crc register containing existing CRC (32-bit)
* @param buf register pointing to input byte buffer (byte*)
* @param len register containing number of bytes
* @param table register pointing to CRC table
* @param constants register pointing to CRC table for 128-bit aligned memory
* @param barretConstants register pointing to table for barrett reduction
* @param t0 volatile register
* @param t1 volatile register
* @param t2 volatile register
* @param t3 volatile register
*/
void MacroAssembler::kernel_crc32_1word_vpmsumd(Register crc, Register buf, Register len, Register table,
Register constants, Register barretConstants,
Register t0, Register t1, Register t2, Register t3, Register t4) {
assert_different_registers(crc, buf, len, table);
Label L_alignedHead, L_tail, L_alignTail, L_start, L_end;
Register prealign = t0;
Register postalign = t0;
BLOCK_COMMENT("kernel_crc32_1word_vpmsumb {");
// 1. use kernel_crc32_1word for shorter than 384bit
clrldi(len, len, 32);
cmpdi(CCR0, len, 384);
bge(CCR0, L_start);
Register tc0 = t4;
Register tc1 = constants;
Register tc2 = barretConstants;
kernel_crc32_1word(crc, buf, len, table,t0, t1, t2, t3, tc0, tc1, tc2, table);
b(L_end);
BIND(L_start);
// 2. ~c
nand(crc, crc, crc);
// 3. calculate from 0 to first 128bit-aligned address
clrldi_(prealign, buf, 57);
beq(CCR0, L_alignedHead);
subfic(prealign, prealign, 128);
subf(len, prealign, len);
update_byteLoop_crc32(crc, buf, prealign, table, t2, false, false);
// 4. calculate from first 128bit-aligned address to last 128bit-aligned address
BIND(L_alignedHead);
clrldi(postalign, len, 57);
subf(len, postalign, len);
// len must be more than 256bit
kernel_crc32_1word_aligned(crc, buf, len, constants, barretConstants, t1, t2, t3);
// 5. calculate remaining
cmpdi(CCR0, postalign, 0);
beq(CCR0, L_tail);
update_byteLoop_crc32(crc, buf, postalign, table, t2, false, false);
BIND(L_tail);
// 6. ~c
nand(crc, crc, crc);
BIND(L_end);
BLOCK_COMMENT("} kernel_crc32_1word_vpmsumb");
}
/**
* @param crc register containing existing CRC (32-bit)
* @param buf register pointing to input byte buffer (byte*)
* @param len register containing number of bytes
* @param constants register pointing to CRC table for 128-bit aligned memory
* @param barretConstants register pointing to table for barrett reduction
* @param t0 volatile register
* @param t1 volatile register
* @param t2 volatile register
*/
void MacroAssembler::kernel_crc32_1word_aligned(Register crc, Register buf, Register len,
Register constants, Register barretConstants, Register t0, Register t1, Register t2) {
Label L_mainLoop, L_tail, L_alignTail, L_barrett_reduction, L_end, L_first_warm_up_done, L_first_cool_down, L_second_cool_down, L_XOR, L_test;
Label L_lv0, L_lv1, L_lv2, L_lv3, L_lv4, L_lv5, L_lv6, L_lv7, L_lv8, L_lv9, L_lv10, L_lv11, L_lv12, L_lv13, L_lv14, L_lv15;
Label L_1, L_2, L_3, L_4;
Register rLoaded = t0;
Register rTmp1 = t1;
Register rTmp2 = t2;
Register off16 = R22;
Register off32 = R23;
Register off48 = R24;
Register off64 = R25;
Register off80 = R26;
Register off96 = R27;
Register off112 = R28;
Register rIdx = R29;
Register rMax = R30;
Register constantsPos = R31;
VectorRegister mask_32bit = VR24;
VectorRegister mask_64bit = VR25;
VectorRegister zeroes = VR26;
VectorRegister const1 = VR27;
VectorRegister const2 = VR28;
// Save non-volatile vector registers (frameless).
Register offset = t1; int offsetInt = 0;
offsetInt -= 16; li(offset, -16); stvx(VR20, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); stvx(VR21, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); stvx(VR22, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); stvx(VR23, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); stvx(VR24, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); stvx(VR25, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); stvx(VR26, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); stvx(VR27, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); stvx(VR28, offset, R1_SP);
offsetInt -= 8; std(R22, offsetInt, R1_SP);
offsetInt -= 8; std(R23, offsetInt, R1_SP);
offsetInt -= 8; std(R24, offsetInt, R1_SP);
offsetInt -= 8; std(R25, offsetInt, R1_SP);
offsetInt -= 8; std(R26, offsetInt, R1_SP);
offsetInt -= 8; std(R27, offsetInt, R1_SP);
offsetInt -= 8; std(R28, offsetInt, R1_SP);
offsetInt -= 8; std(R29, offsetInt, R1_SP);
offsetInt -= 8; std(R30, offsetInt, R1_SP);
offsetInt -= 8; std(R31, offsetInt, R1_SP);
// Set constants
li(off16, 16);
li(off32, 32);
li(off48, 48);
li(off64, 64);
li(off80, 80);
li(off96, 96);
li(off112, 112);
clrldi(crc, crc, 32);
vxor(zeroes, zeroes, zeroes);
vspltisw(VR0, -1);
vsldoi(mask_32bit, zeroes, VR0, 4);
vsldoi(mask_64bit, zeroes, VR0, -8);
// Get the initial value into v8
vxor(VR8, VR8, VR8);
mtvrd(VR8, crc);
vsldoi(VR8, zeroes, VR8, -8); // shift into bottom 32 bits
li (rLoaded, 0);
rldicr(rIdx, len, 0, 56);
{
BIND(L_1);
// Checksum in blocks of MAX_SIZE (32768)
lis(rMax, 0);
ori(rMax, rMax, 32768);
mr(rTmp2, rMax);
cmpd(CCR0, rIdx, rMax);
bgt(CCR0, L_2);
mr(rMax, rIdx);
BIND(L_2);
subf(rIdx, rMax, rIdx);
// our main loop does 128 bytes at a time
srdi(rMax, rMax, 7);
/*
* Work out the offset into the constants table to start at. Each
* constant is 16 bytes, and it is used against 128 bytes of input
* data - 128 / 16 = 8
*/
sldi(rTmp1, rMax, 4);
srdi(rTmp2, rTmp2, 3);
subf(rTmp1, rTmp1, rTmp2);
// We reduce our final 128 bytes in a separate step
addi(rMax, rMax, -1);
mtctr(rMax);
// Find the start of our constants
add(constantsPos, constants, rTmp1);
// zero VR0-v7 which will contain our checksums
vxor(VR0, VR0, VR0);
vxor(VR1, VR1, VR1);
vxor(VR2, VR2, VR2);
vxor(VR3, VR3, VR3);
vxor(VR4, VR4, VR4);
vxor(VR5, VR5, VR5);
vxor(VR6, VR6, VR6);
vxor(VR7, VR7, VR7);
lvx(const1, constantsPos);
/*
* If we are looping back to consume more data we use the values
* already in VR16-v23.
*/
cmpdi(CCR0, rLoaded, 1);
beq(CCR0, L_3);
{
// First warm up pass
lvx(VR16, buf);
lvx(VR17, off16, buf);
lvx(VR18, off32, buf);
lvx(VR19, off48, buf);
lvx(VR20, off64, buf);
lvx(VR21, off80, buf);
lvx(VR22, off96, buf);
lvx(VR23, off112, buf);
addi(buf, buf, 8*16);
// xor in initial value
vxor(VR16, VR16, VR8);
}
BIND(L_3);
bdz(L_first_warm_up_done);
addi(constantsPos, constantsPos, 16);
lvx(const2, constantsPos);
// Second warm up pass
vpmsumd(VR8, VR16, const1);
lvx(VR16, buf);
vpmsumd(VR9, VR17, const1);
lvx(VR17, off16, buf);
vpmsumd(VR10, VR18, const1);
lvx(VR18, off32, buf);
vpmsumd(VR11, VR19, const1);
lvx(VR19, off48, buf);
vpmsumd(VR12, VR20, const1);
lvx(VR20, off64, buf);
vpmsumd(VR13, VR21, const1);
lvx(VR21, off80, buf);
vpmsumd(VR14, VR22, const1);
lvx(VR22, off96, buf);
vpmsumd(VR15, VR23, const1);
lvx(VR23, off112, buf);
addi(buf, buf, 8 * 16);
bdz(L_first_cool_down);
/*
* main loop. We modulo schedule it such that it takes three iterations
* to complete - first iteration load, second iteration vpmsum, third
* iteration xor.
*/
{
BIND(L_4);
lvx(const1, constantsPos); addi(constantsPos, constantsPos, 16);
vxor(VR0, VR0, VR8);
vpmsumd(VR8, VR16, const2);
lvx(VR16, buf);
vxor(VR1, VR1, VR9);
vpmsumd(VR9, VR17, const2);
lvx(VR17, off16, buf);
vxor(VR2, VR2, VR10);
vpmsumd(VR10, VR18, const2);
lvx(VR18, off32, buf);
vxor(VR3, VR3, VR11);
vpmsumd(VR11, VR19, const2);
lvx(VR19, off48, buf);
lvx(const2, constantsPos);
vxor(VR4, VR4, VR12);
vpmsumd(VR12, VR20, const1);
lvx(VR20, off64, buf);
vxor(VR5, VR5, VR13);
vpmsumd(VR13, VR21, const1);
lvx(VR21, off80, buf);
vxor(VR6, VR6, VR14);
vpmsumd(VR14, VR22, const1);
lvx(VR22, off96, buf);
vxor(VR7, VR7, VR15);
vpmsumd(VR15, VR23, const1);
lvx(VR23, off112, buf);
addi(buf, buf, 8 * 16);
bdnz(L_4);
}
BIND(L_first_cool_down);
// First cool down pass
lvx(const1, constantsPos);
addi(constantsPos, constantsPos, 16);
vxor(VR0, VR0, VR8);
vpmsumd(VR8, VR16, const1);
vxor(VR1, VR1, VR9);
vpmsumd(VR9, VR17, const1);
vxor(VR2, VR2, VR10);
vpmsumd(VR10, VR18, const1);
vxor(VR3, VR3, VR11);
vpmsumd(VR11, VR19, const1);
vxor(VR4, VR4, VR12);
vpmsumd(VR12, VR20, const1);
vxor(VR5, VR5, VR13);
vpmsumd(VR13, VR21, const1);
vxor(VR6, VR6, VR14);
vpmsumd(VR14, VR22, const1);
vxor(VR7, VR7, VR15);
vpmsumd(VR15, VR23, const1);
BIND(L_second_cool_down);
// Second cool down pass
vxor(VR0, VR0, VR8);
vxor(VR1, VR1, VR9);
vxor(VR2, VR2, VR10);
vxor(VR3, VR3, VR11);
vxor(VR4, VR4, VR12);
vxor(VR5, VR5, VR13);
vxor(VR6, VR6, VR14);
vxor(VR7, VR7, VR15);
/*
* vpmsumd produces a 96 bit result in the least significant bits
* of the register. Since we are bit reflected we have to shift it
* left 32 bits so it occupies the least significant bits in the
* bit reflected domain.
*/
vsldoi(VR0, VR0, zeroes, 4);
vsldoi(VR1, VR1, zeroes, 4);
vsldoi(VR2, VR2, zeroes, 4);
vsldoi(VR3, VR3, zeroes, 4);
vsldoi(VR4, VR4, zeroes, 4);
vsldoi(VR5, VR5, zeroes, 4);
vsldoi(VR6, VR6, zeroes, 4);
vsldoi(VR7, VR7, zeroes, 4);
// xor with last 1024 bits
lvx(VR8, buf);
lvx(VR9, off16, buf);
lvx(VR10, off32, buf);
lvx(VR11, off48, buf);
lvx(VR12, off64, buf);
lvx(VR13, off80, buf);
lvx(VR14, off96, buf);
lvx(VR15, off112, buf);
addi(buf, buf, 8 * 16);
vxor(VR16, VR0, VR8);
vxor(VR17, VR1, VR9);
vxor(VR18, VR2, VR10);
vxor(VR19, VR3, VR11);
vxor(VR20, VR4, VR12);
vxor(VR21, VR5, VR13);
vxor(VR22, VR6, VR14);
vxor(VR23, VR7, VR15);
li(rLoaded, 1);
cmpdi(CCR0, rIdx, 0);
addi(rIdx, rIdx, 128);
bne(CCR0, L_1);
}
// Work out how many bytes we have left
andi_(len, len, 127);
// Calculate where in the constant table we need to start
subfic(rTmp1, len, 128);
add(constantsPos, constantsPos, rTmp1);
// How many 16 byte chunks are in the tail
srdi(rIdx, len, 4);
mtctr(rIdx);
/*
* Reduce the previously calculated 1024 bits to 64 bits, shifting
* 32 bits to include the trailing 32 bits of zeros
*/
lvx(VR0, constantsPos);
lvx(VR1, off16, constantsPos);
lvx(VR2, off32, constantsPos);
lvx(VR3, off48, constantsPos);
lvx(VR4, off64, constantsPos);
lvx(VR5, off80, constantsPos);
lvx(VR6, off96, constantsPos);
lvx(VR7, off112, constantsPos);
addi(constantsPos, constantsPos, 8 * 16);
vpmsumw(VR0, VR16, VR0);
vpmsumw(VR1, VR17, VR1);
vpmsumw(VR2, VR18, VR2);
vpmsumw(VR3, VR19, VR3);
vpmsumw(VR4, VR20, VR4);
vpmsumw(VR5, VR21, VR5);
vpmsumw(VR6, VR22, VR6);
vpmsumw(VR7, VR23, VR7);
// Now reduce the tail (0 - 112 bytes)
cmpdi(CCR0, rIdx, 0);
beq(CCR0, L_XOR);
lvx(VR16, buf); addi(buf, buf, 16);
lvx(VR17, constantsPos);
vpmsumw(VR16, VR16, VR17);
vxor(VR0, VR0, VR16);
beq(CCR0, L_XOR);
lvx(VR16, buf); addi(buf, buf, 16);
lvx(VR17, off16, constantsPos);
vpmsumw(VR16, VR16, VR17);
vxor(VR0, VR0, VR16);
beq(CCR0, L_XOR);
lvx(VR16, buf); addi(buf, buf, 16);
lvx(VR17, off32, constantsPos);
vpmsumw(VR16, VR16, VR17);
vxor(VR0, VR0, VR16);
beq(CCR0, L_XOR);
lvx(VR16, buf); addi(buf, buf, 16);
lvx(VR17, off48,constantsPos);
vpmsumw(VR16, VR16, VR17);
vxor(VR0, VR0, VR16);
beq(CCR0, L_XOR);
lvx(VR16, buf); addi(buf, buf, 16);
lvx(VR17, off64, constantsPos);
vpmsumw(VR16, VR16, VR17);
vxor(VR0, VR0, VR16);
beq(CCR0, L_XOR);
lvx(VR16, buf); addi(buf, buf, 16);
lvx(VR17, off80, constantsPos);
vpmsumw(VR16, VR16, VR17);
vxor(VR0, VR0, VR16);
beq(CCR0, L_XOR);
lvx(VR16, buf); addi(buf, buf, 16);
lvx(VR17, off96, constantsPos);
vpmsumw(VR16, VR16, VR17);
vxor(VR0, VR0, VR16);
// Now xor all the parallel chunks together
BIND(L_XOR);
vxor(VR0, VR0, VR1);
vxor(VR2, VR2, VR3);
vxor(VR4, VR4, VR5);
vxor(VR6, VR6, VR7);
vxor(VR0, VR0, VR2);
vxor(VR4, VR4, VR6);
vxor(VR0, VR0, VR4);
b(L_barrett_reduction);
BIND(L_first_warm_up_done);
lvx(const1, constantsPos);
addi(constantsPos, constantsPos, 16);
vpmsumd(VR8, VR16, const1);
vpmsumd(VR9, VR17, const1);
vpmsumd(VR10, VR18, const1);
vpmsumd(VR11, VR19, const1);
vpmsumd(VR12, VR20, const1);
vpmsumd(VR13, VR21, const1);
vpmsumd(VR14, VR22, const1);
vpmsumd(VR15, VR23, const1);
b(L_second_cool_down);
BIND(L_barrett_reduction);
lvx(const1, barretConstants);
addi(barretConstants, barretConstants, 16);
lvx(const2, barretConstants);
vsldoi(VR1, VR0, VR0, -8);
vxor(VR0, VR0, VR1); // xor two 64 bit results together
// shift left one bit
vspltisb(VR1, 1);
vsl(VR0, VR0, VR1);
vand(VR0, VR0, mask_64bit);
/*
* The reflected version of Barrett reduction. Instead of bit
* reflecting our data (which is expensive to do), we bit reflect our
* constants and our algorithm, which means the intermediate data in
* our vector registers goes from 0-63 instead of 63-0. We can reflect
* the algorithm because we don't carry in mod 2 arithmetic.
*/
vand(VR1, VR0, mask_32bit); // bottom 32 bits of a
vpmsumd(VR1, VR1, const1); // ma
vand(VR1, VR1, mask_32bit); // bottom 32bits of ma
vpmsumd(VR1, VR1, const2); // qn */
vxor(VR0, VR0, VR1); // a - qn, subtraction is xor in GF(2)
/*
* Since we are bit reflected, the result (ie the low 32 bits) is in
* the high 32 bits. We just need to shift it left 4 bytes
* V0 [ 0 1 X 3 ]
* V0 [ 0 X 2 3 ]
*/
vsldoi(VR0, VR0, zeroes, 4); // shift result into top 64 bits of
// Get it into r3
mfvrd(crc, VR0);
BIND(L_end);
offsetInt = 0;
// Restore non-volatile Vector registers (frameless).
offsetInt -= 16; li(offset, -16); lvx(VR20, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); lvx(VR21, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); lvx(VR22, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); lvx(VR23, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); lvx(VR24, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); lvx(VR25, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); lvx(VR26, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); lvx(VR27, offset, R1_SP);
offsetInt -= 16; addi(offset, offset, -16); lvx(VR28, offset, R1_SP);
offsetInt -= 8; ld(R22, offsetInt, R1_SP);
offsetInt -= 8; ld(R23, offsetInt, R1_SP);
offsetInt -= 8; ld(R24, offsetInt, R1_SP);
offsetInt -= 8; ld(R25, offsetInt, R1_SP);
offsetInt -= 8; ld(R26, offsetInt, R1_SP);
offsetInt -= 8; ld(R27, offsetInt, R1_SP);
offsetInt -= 8; ld(R28, offsetInt, R1_SP);
offsetInt -= 8; ld(R29, offsetInt, R1_SP);
offsetInt -= 8; ld(R30, offsetInt, R1_SP);
offsetInt -= 8; ld(R31, offsetInt, R1_SP);
}
void MacroAssembler::kernel_crc32_singleByte(Register crc, Register buf, Register len, Register table, Register tmp) {
assert_different_registers(crc, buf, /* len, not used!! */ table, tmp);

7
hotspot/src/cpu/ppc/vm/macroAssembler_ppc.hpp
View File

@ -834,6 +834,13 @@ class MacroAssembler: public Assembler {
Register tc0, Register tc1, Register tc2, Register tc3);
void kernel_crc32_1byte(Register crc, Register buf, Register len, Register table,
Register t0, Register t1, Register t2, Register t3);
void kernel_crc32_1word_vpmsumd(Register crc, Register buf, Register len, Register table,
Register constants, Register barretConstants,
Register t0, Register t1, Register t2, Register t3, Register t4);
void kernel_crc32_1word_aligned(Register crc, Register buf, Register len,
Register constants, Register barretConstants,
Register t0, Register t1, Register t2);
void kernel_crc32_singleByte(Register crc, Register buf, Register len, Register table, Register tmp);
//

5
hotspot/src/cpu/ppc/vm/sharedRuntime_ppc.cpp
View File

@ -2489,6 +2489,11 @@ nmethod *SharedRuntime::generate_native_wrapper(MacroAssembler *masm,
__ verify_oop(R3_RET);
}
if (CheckJNICalls) {
// clear_pending_jni_exception_check
__ load_const_optimized(R0, 0L);
__ st_ptr(R0, JavaThread::pending_jni_exception_check_fn_offset(), R16_thread);
}
// Reset handle block.
// --------------------------------------------------------------------------

63
hotspot/src/cpu/ppc/vm/stubGenerator_ppc.cpp
View File

@ -3205,28 +3205,59 @@ class StubGenerator: public StubCodeGenerator {
const Register crc = R3_ARG1; // Current checksum, preset by caller or result from previous call.
const Register data = R4_ARG2; // source byte array
const Register dataLen = R5_ARG3; // #bytes to process
const Register table = R6_ARG4; // crc table address
const Register t0 = R2;
const Register t1 = R7;
const Register t2 = R8;
const Register t3 = R9;
const Register tc0 = R10;
const Register tc1 = R11;
const Register tc2 = R12;
const Register table = R6; // crc table address
BLOCK_COMMENT("Stub body {");
assert_different_registers(crc, data, dataLen, table);
#ifdef VM_LITTLE_ENDIAN
if (VM_Version::has_vpmsumb()) {
const Register constants = R2; // constants address
const Register bconstants = R8; // barret table address
StubRoutines::ppc64::generate_load_crc_table_addr(_masm, table);
const Register t0 = R9;
const Register t1 = R10;
const Register t2 = R11;
const Register t3 = R12;
const Register t4 = R7;
__ kernel_crc32_1word(crc, data, dataLen, table, t0, t1, t2, t3, tc0, tc1, tc2, table);
BLOCK_COMMENT("Stub body {");
assert_different_registers(crc, data, dataLen, table);
BLOCK_COMMENT("return");
__ mr_if_needed(R3_RET, crc); // Updated crc is function result. No copying required (R3_ARG1 == R3_RET).
__ blr();
StubRoutines::ppc64::generate_load_crc_table_addr(_masm, table);
StubRoutines::ppc64::generate_load_crc_constants_addr(_masm, constants);
StubRoutines::ppc64::generate_load_crc_barret_constants_addr(_masm, bconstants);
__ kernel_crc32_1word_vpmsumd(crc, data, dataLen, table, constants, bconstants, t0, t1, t2, t3, t4);
BLOCK_COMMENT("return");
__ mr_if_needed(R3_RET, crc); // Updated crc is function result. No copying required (R3_ARG1 == R3_RET).
__ blr();
BLOCK_COMMENT("} Stub body");
} else
#endif
{
const Register t0 = R2;
const Register t1 = R7;
const Register t2 = R8;
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_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).
__ blr();
BLOCK_COMMENT("} Stub body");
}
BLOCK_COMMENT("} Stub body");
return start;
}

8
hotspot/src/cpu/ppc/vm/stubRoutines_ppc.hpp
View File

@ -45,6 +45,8 @@ enum platform_dependent_constants {
#else
#define CRC32_TABLES 1
#endif
#define CRC32_CONSTANTS_SIZE 1084
#define CRC32_BARRET_CONSTANTS 10
class ppc64 {
friend class StubGenerator;
@ -53,11 +55,17 @@ class ppc64 {
// CRC32 Intrinsics.
static juint _crc_table[CRC32_TABLES][CRC32_COLUMN_SIZE];
static juint* _constants;
static juint* _barret_constants;
public:
// CRC32 Intrinsics.
static void generate_load_crc_table_addr(MacroAssembler* masm, Register table);
static void generate_load_crc_constants_addr(MacroAssembler* masm, Register table);
static void generate_load_crc_barret_constants_addr(MacroAssembler* masm, Register table);
static juint* generate_crc_constants();
static juint* generate_crc_barret_constants();
};

309
hotspot/src/cpu/ppc/vm/stubRoutines_ppc_64.cpp
View File

@ -37,6 +37,311 @@ void StubRoutines::ppc64::generate_load_crc_table_addr(MacroAssembler* masm, Reg
__ load_const_optimized(table, StubRoutines::_crc_table_adr, R0);
}
void StubRoutines::ppc64::generate_load_crc_constants_addr(MacroAssembler* masm, Register table) {
__ load_const_optimized(table, (address)StubRoutines::ppc64::_constants, R0);
}
void StubRoutines::ppc64::generate_load_crc_barret_constants_addr(MacroAssembler* masm, Register table) {
__ load_const_optimized(table, (address)StubRoutines::ppc64::_barret_constants, R0);
}
juint* StubRoutines::ppc64::generate_crc_constants() {
juint constants[CRC32_CONSTANTS_SIZE] = {
// Reduce 262144 kbits to 1024 bits
0x99ea94a8UL, 0x00000000UL, 0x651797d2UL, 0x00000001UL, // x^261120 mod p(x)` << 1, x^261184 mod p(x)` << 1
0x945a8420UL, 0x00000000UL, 0x21e0d56cUL, 0x00000000UL, // x^260096 mod p(x)` << 1, x^260160 mod p(x)` << 1
0x30762706UL, 0x00000000UL, 0x0f95ecaaUL, 0x00000000UL, // x^259072 mod p(x)` << 1, x^259136 mod p(x)` << 1
0xa52fc582UL, 0x00000001UL, 0xebd224acUL, 0x00000001UL, // x^258048 mod p(x)` << 1, x^258112 mod p(x)` << 1
0xa4a7167aUL, 0x00000001UL, 0x0ccb97caUL, 0x00000000UL, // x^257024 mod p(x)` << 1, x^257088 mod p(x)` << 1
0x0c18249aUL, 0x00000000UL, 0x006ec8a8UL, 0x00000001UL, // x^256000 mod p(x)` << 1, x^256064 mod p(x)` << 1
0xa924ae7cUL, 0x00000000UL, 0x4f58f196UL, 0x00000001UL, // x^254976 mod p(x)` << 1, x^255040 mod p(x)` << 1
0xe12ccc12UL, 0x00000001UL, 0xa7192ca6UL, 0x00000001UL, // x^253952 mod p(x)` << 1, x^254016 mod p(x)` << 1
0xa0b9d4acUL, 0x00000000UL, 0x9a64bab2UL, 0x00000001UL, // x^252928 mod p(x)` << 1, x^252992 mod p(x)` << 1
0x95e8ddfeUL, 0x00000000UL, 0x14f4ed2eUL, 0x00000000UL, // x^251904 mod p(x)` << 1, x^251968 mod p(x)` << 1
0x233fddc4UL, 0x00000000UL, 0x1092b6a2UL, 0x00000001UL, // x^250880 mod p(x)` << 1, x^250944 mod p(x)` << 1
0xb4529b62UL, 0x00000001UL, 0xc8a1629cUL, 0x00000000UL, // x^249856 mod p(x)` << 1, x^249920 mod p(x)` << 1
0xa7fa0e64UL, 0x00000001UL, 0x7bf32e8eUL, 0x00000001UL, // x^248832 mod p(x)` << 1, x^248896 mod p(x)` << 1
0xb5334592UL, 0x00000001UL, 0xf8cc6582UL, 0x00000001UL, // x^247808 mod p(x)` << 1, x^247872 mod p(x)` << 1
0x1f8ee1b4UL, 0x00000001UL, 0x8631ddf0UL, 0x00000000UL, // x^246784 mod p(x)` << 1, x^246848 mod p(x)` << 1
0x6252e632UL, 0x00000000UL, 0x7e5a76d0UL, 0x00000000UL, // x^245760 mod p(x)` << 1, x^245824 mod p(x)` << 1
0xab973e84UL, 0x00000000UL, 0x2b09b31cUL, 0x00000000UL, // x^244736 mod p(x)` << 1, x^244800 mod p(x)` << 1
0x7734f5ecUL, 0x00000000UL, 0xb2df1f84UL, 0x00000001UL, // x^243712 mod p(x)` << 1, x^243776 mod p(x)` << 1
0x7c547798UL, 0x00000000UL, 0xd6f56afcUL, 0x00000001UL, // x^242688 mod p(x)` << 1, x^242752 mod p(x)` << 1
0x7ec40210UL, 0x00000000UL, 0xb9b5e70cUL, 0x00000001UL, // x^241664 mod p(x)` << 1, x^241728 mod p(x)` << 1
0xab1695a8UL, 0x00000001UL, 0x34b626d2UL, 0x00000000UL, // x^240640 mod p(x)` << 1, x^240704 mod p(x)` << 1
0x90494bbaUL, 0x00000000UL, 0x4c53479aUL, 0x00000001UL, // x^239616 mod p(x)` << 1, x^239680 mod p(x)` << 1
0x123fb816UL, 0x00000001UL, 0xa6d179a4UL, 0x00000001UL, // x^238592 mod p(x)` << 1, x^238656 mod p(x)` << 1
0xe188c74cUL, 0x00000001UL, 0x5abd16b4UL, 0x00000001UL, // x^237568 mod p(x)` << 1, x^237632 mod p(x)` << 1
0xc2d3451cUL, 0x00000001UL, 0x018f9852UL, 0x00000000UL, // x^236544 mod p(x)` << 1, x^236608 mod p(x)` << 1
0xf55cf1caUL, 0x00000000UL, 0x1fb3084aUL, 0x00000000UL, // x^235520 mod p(x)` << 1, x^235584 mod p(x)` << 1
0xa0531540UL, 0x00000001UL, 0xc53dfb04UL, 0x00000000UL, // x^234496 mod p(x)` << 1, x^234560 mod p(x)` << 1
0x32cd7ebcUL, 0x00000001UL, 0xe10c9ad6UL, 0x00000000UL, // x^233472 mod p(x)` << 1, x^233536 mod p(x)` << 1
0x73ab7f36UL, 0x00000000UL, 0x25aa994aUL, 0x00000000UL, // x^232448 mod p(x)` << 1, x^232512 mod p(x)` << 1
0x41aed1c2UL, 0x00000000UL, 0xfa3a74c4UL, 0x00000000UL, // x^231424 mod p(x)` << 1, x^231488 mod p(x)` << 1
0x36c53800UL, 0x00000001UL, 0x33eb3f40UL, 0x00000000UL, // x^230400 mod p(x)` << 1, x^230464 mod p(x)` << 1
0x26835a30UL, 0x00000001UL, 0x7193f296UL, 0x00000001UL, // x^229376 mod p(x)` << 1, x^229440 mod p(x)` << 1
0x6241b502UL, 0x00000000UL, 0x43f6c86aUL, 0x00000000UL, // x^228352 mod p(x)` << 1, x^228416 mod p(x)` << 1
0xd5196ad4UL, 0x00000000UL, 0x6b513ec6UL, 0x00000001UL, // x^227328 mod p(x)` << 1, x^227392 mod p(x)` << 1
0x9cfa769aUL, 0x00000000UL, 0xc8f25b4eUL, 0x00000000UL, // x^226304 mod p(x)` << 1, x^226368 mod p(x)` << 1
0x920e5df4UL, 0x00000000UL, 0xa45048ecUL, 0x00000001UL, // x^225280 mod p(x)` << 1, x^225344 mod p(x)` << 1
0x69dc310eUL, 0x00000001UL, 0x0c441004UL, 0x00000000UL, // x^224256 mod p(x)` << 1, x^224320 mod p(x)` << 1
0x09fc331cUL, 0x00000000UL, 0x0e17cad6UL, 0x00000000UL, // x^223232 mod p(x)` << 1, x^223296 mod p(x)` << 1
0x0d94a81eUL, 0x00000001UL, 0x253ae964UL, 0x00000001UL, // x^222208 mod p(x)` << 1, x^222272 mod p(x)` << 1
0x27a20ab2UL, 0x00000000UL, 0xd7c88ebcUL, 0x00000001UL, // x^221184 mod p(x)` << 1, x^221248 mod p(x)` << 1
0x14f87504UL, 0x00000001UL, 0xe7ca913aUL, 0x00000001UL, // x^220160 mod p(x)` << 1, x^220224 mod p(x)` << 1
0x4b076d96UL, 0x00000000UL, 0x33ed078aUL, 0x00000000UL, // x^219136 mod p(x)` << 1, x^219200 mod p(x)` << 1
0xda4d1e74UL, 0x00000000UL, 0xe1839c78UL, 0x00000000UL, // x^218112 mod p(x)` << 1, x^218176 mod p(x)` << 1
0x1b81f672UL, 0x00000000UL, 0x322b267eUL, 0x00000001UL, // x^217088 mod p(x)` << 1, x^217152 mod p(x)` << 1
0x9367c988UL, 0x00000000UL, 0x638231b6UL, 0x00000000UL, // x^216064 mod p(x)` << 1, x^216128 mod p(x)` << 1
0x717214caUL, 0x00000001UL, 0xee7f16f4UL, 0x00000001UL, // x^215040 mod p(x)` << 1, x^215104 mod p(x)` << 1
0x9f47d820UL, 0x00000000UL, 0x17d9924aUL, 0x00000001UL, // x^214016 mod p(x)` << 1, x^214080 mod p(x)` << 1
0x0d9a47d2UL, 0x00000001UL, 0xe1a9e0c4UL, 0x00000000UL, // x^212992 mod p(x)` << 1, x^213056 mod p(x)` << 1
0xa696c58cUL, 0x00000000UL, 0x403731dcUL, 0x00000001UL, // x^211968 mod p(x)` << 1, x^212032 mod p(x)` << 1
0x2aa28ec6UL, 0x00000000UL, 0xa5ea9682UL, 0x00000001UL, // x^210944 mod p(x)` << 1, x^211008 mod p(x)` << 1
0xfe18fd9aUL, 0x00000001UL, 0x01c5c578UL, 0x00000001UL, // x^209920 mod p(x)` << 1, x^209984 mod p(x)` << 1
0x9d4fc1aeUL, 0x00000001UL, 0xdddf6494UL, 0x00000000UL, // x^208896 mod p(x)` << 1, x^208960 mod p(x)` << 1
0xba0e3deaUL, 0x00000001UL, 0xf1c3db28UL, 0x00000000UL, // x^207872 mod p(x)` << 1, x^207936 mod p(x)` << 1
0x74b59a5eUL, 0x00000000UL, 0x3112fb9cUL, 0x00000001UL, // x^206848 mod p(x)` << 1, x^206912 mod p(x)` << 1
0xf2b5ea98UL, 0x00000000UL, 0xb680b906UL, 0x00000000UL, // x^205824 mod p(x)` << 1, x^205888 mod p(x)` << 1
0x87132676UL, 0x00000001UL, 0x1a282932UL, 0x00000000UL, // x^204800 mod p(x)` << 1, x^204864 mod p(x)` << 1
0x0a8c6ad4UL, 0x00000001UL, 0x89406e7eUL, 0x00000000UL, // x^203776 mod p(x)` << 1, x^203840 mod p(x)` << 1
0xe21dfe70UL, 0x00000001UL, 0xdef6be8cUL, 0x00000001UL, // x^202752 mod p(x)` << 1, x^202816 mod p(x)` << 1
0xda0050e4UL, 0x00000001UL, 0x75258728UL, 0x00000000UL, // x^201728 mod p(x)` << 1, x^201792 mod p(x)` << 1
0x772172aeUL, 0x00000000UL, 0x9536090aUL, 0x00000001UL, // x^200704 mod p(x)` << 1, x^200768 mod p(x)` << 1
0xe47724aaUL, 0x00000000UL, 0xf2455bfcUL, 0x00000000UL, // x^199680 mod p(x)` << 1, x^199744 mod p(x)` << 1
0x3cd63ac4UL, 0x00000000UL, 0x8c40baf4UL, 0x00000001UL, // x^198656 mod p(x)` << 1, x^198720 mod p(x)` << 1
0xbf47d352UL, 0x00000001UL, 0x4cd390d4UL, 0x00000000UL, // x^197632 mod p(x)` << 1, x^197696 mod p(x)` << 1
0x8dc1d708UL, 0x00000001UL, 0xe4ece95aUL, 0x00000001UL, // x^196608 mod p(x)` << 1, x^196672 mod p(x)` << 1
0x2d4620a4UL, 0x00000000UL, 0x1a3ee918UL, 0x00000000UL, // x^195584 mod p(x)` << 1, x^195648 mod p(x)` << 1
0x58fd1740UL, 0x00000000UL, 0x7c652fb8UL, 0x00000000UL, // x^194560 mod p(x)` << 1, x^194624 mod p(x)` << 1
0xdadd9bfcUL, 0x00000000UL, 0x1c67842cUL, 0x00000001UL, // x^193536 mod p(x)` << 1, x^193600 mod p(x)` << 1
0xea2140beUL, 0x00000001UL, 0x254f759cUL, 0x00000000UL, // x^192512 mod p(x)` << 1, x^192576 mod p(x)` << 1
0x9de128baUL, 0x00000000UL, 0x7ece94caUL, 0x00000000UL, // x^191488 mod p(x)` << 1, x^191552 mod p(x)` << 1
0x3ac3aa8eUL, 0x00000001UL, 0x38f258c2UL, 0x00000000UL, // x^190464 mod p(x)` << 1, x^190528 mod p(x)` << 1
0x99980562UL, 0x00000000UL, 0xcdf17b00UL, 0x00000001UL, // x^189440 mod p(x)` << 1, x^189504 mod p(x)` << 1
0xc1579c86UL, 0x00000001UL, 0x1f882c16UL, 0x00000001UL, // x^188416 mod p(x)` << 1, x^188480 mod p(x)` << 1
0x68dbbf94UL, 0x00000000UL, 0x00093fc8UL, 0x00000001UL, // x^187392 mod p(x)` << 1, x^187456 mod p(x)` << 1
0x4509fb04UL, 0x00000000UL, 0xcd684f16UL, 0x00000001UL, // x^186368 mod p(x)` << 1, x^186432 mod p(x)` << 1
0x202f6398UL, 0x00000001UL, 0x4bc6a70aUL, 0x00000000UL, // x^185344 mod p(x)` << 1, x^185408 mod p(x)` << 1
0x3aea243eUL, 0x00000001UL, 0x4fc7e8e4UL, 0x00000000UL, // x^184320 mod p(x)` << 1, x^184384 mod p(x)` << 1
0xb4052ae6UL, 0x00000001UL, 0x30103f1cUL, 0x00000001UL, // x^183296 mod p(x)` << 1, x^183360 mod p(x)` << 1
0xcd2a0ae8UL, 0x00000001UL, 0x11b0024cUL, 0x00000001UL, // x^182272 mod p(x)` << 1, x^182336 mod p(x)` << 1
0xfe4aa8b4UL, 0x00000001UL, 0x0b3079daUL, 0x00000001UL, // x^181248 mod p(x)` << 1, x^181312 mod p(x)` << 1
0xd1559a42UL, 0x00000001UL, 0x0192bcc2UL, 0x00000001UL, // x^180224 mod p(x)` << 1, x^180288 mod p(x)` << 1
0xf3e05eccUL, 0x00000001UL, 0x74838d50UL, 0x00000000UL, // x^179200 mod p(x)` << 1, x^179264 mod p(x)` << 1
0x04ddd2ccUL, 0x00000001UL, 0x1b20f520UL, 0x00000000UL, // x^178176 mod p(x)` << 1, x^178240 mod p(x)` << 1
0x5393153cUL, 0x00000001UL, 0x50c3590aUL, 0x00000000UL, // x^177152 mod p(x)` << 1, x^177216 mod p(x)` << 1
0x57e942c6UL, 0x00000000UL, 0xb41cac8eUL, 0x00000000UL, // x^176128 mod p(x)` << 1, x^176192 mod p(x)` << 1
0x2c633850UL, 0x00000001UL, 0x0c72cc78UL, 0x00000000UL, // x^175104 mod p(x)` << 1, x^175168 mod p(x)` << 1
0xebcaae4cUL, 0x00000000UL, 0x30cdb032UL, 0x00000000UL, // x^174080 mod p(x)` << 1, x^174144 mod p(x)` << 1
0x3ee532a6UL, 0x00000001UL, 0x3e09fc32UL, 0x00000001UL, // x^173056 mod p(x)` << 1, x^173120 mod p(x)` << 1
0xbf0cbc7eUL, 0x00000001UL, 0x1ed624d2UL, 0x00000000UL, // x^172032 mod p(x)` << 1, x^172096 mod p(x)` << 1
0xd50b7a5aUL, 0x00000000UL, 0x781aee1aUL, 0x00000000UL, // x^171008 mod p(x)` << 1, x^171072 mod p(x)` << 1
0x02fca6e8UL, 0x00000000UL, 0xc4d8348cUL, 0x00000001UL, // x^169984 mod p(x)` << 1, x^170048 mod p(x)` << 1
0x7af40044UL, 0x00000000UL, 0x57a40336UL, 0x00000000UL, // x^168960 mod p(x)` << 1, x^169024 mod p(x)` << 1
0x16178744UL, 0x00000000UL, 0x85544940UL, 0x00000000UL, // x^167936 mod p(x)` << 1, x^168000 mod p(x)` << 1
0x4c177458UL, 0x00000001UL, 0x9cd21e80UL, 0x00000001UL, // x^166912 mod p(x)` << 1, x^166976 mod p(x)` << 1
0x1b6ddf04UL, 0x00000001UL, 0x3eb95bc0UL, 0x00000001UL, // x^165888 mod p(x)` << 1, x^165952 mod p(x)` << 1
0xf3e29cccUL, 0x00000001UL, 0xdfc9fdfcUL, 0x00000001UL, // x^164864 mod p(x)` << 1, x^164928 mod p(x)` << 1
0x35ae7562UL, 0x00000001UL, 0xcd028bc2UL, 0x00000000UL, // x^163840 mod p(x)` << 1, x^163904 mod p(x)` << 1
0x90ef812cUL, 0x00000001UL, 0x90db8c44UL, 0x00000000UL, // x^162816 mod p(x)` << 1, x^162880 mod p(x)` << 1
0x67a2c786UL, 0x00000000UL, 0x0010a4ceUL, 0x00000001UL, // x^161792 mod p(x)` << 1, x^161856 mod p(x)` << 1
0x48b9496cUL, 0x00000000UL, 0xc8f4c72cUL, 0x00000001UL, // x^160768 mod p(x)` << 1, x^160832 mod p(x)` << 1
0x5a422de6UL, 0x00000001UL, 0x1c26170cUL, 0x00000000UL, // x^159744 mod p(x)` << 1, x^159808 mod p(x)` << 1
0xef0e3640UL, 0x00000001UL, 0xe3fccf68UL, 0x00000000UL, // x^158720 mod p(x)` << 1, x^158784 mod p(x)` << 1
0x006d2d26UL, 0x00000001UL, 0xd513ed24UL, 0x00000000UL, // x^157696 mod p(x)` << 1, x^157760 mod p(x)` << 1
0x170d56d6UL, 0x00000001UL, 0x141beadaUL, 0x00000000UL, // x^156672 mod p(x)` << 1, x^156736 mod p(x)` << 1
0xa5fb613cUL, 0x00000000UL, 0x1071aea0UL, 0x00000001UL, // x^155648 mod p(x)` << 1, x^155712 mod p(x)` << 1
0x40bbf7fcUL, 0x00000000UL, 0x2e19080aUL, 0x00000001UL, // x^154624 mod p(x)` << 1, x^154688 mod p(x)` << 1
0x6ac3a5b2UL, 0x00000001UL, 0x00ecf826UL, 0x00000001UL, // x^153600 mod p(x)` << 1, x^153664 mod p(x)` << 1
0xabf16230UL, 0x00000000UL, 0x69b09412UL, 0x00000000UL, // x^152576 mod p(x)` << 1, x^152640 mod p(x)` << 1
0xebe23facUL, 0x00000001UL, 0x22297bacUL, 0x00000001UL, // x^151552 mod p(x)` << 1, x^151616 mod p(x)` << 1
0x8b6a0894UL, 0x00000000UL, 0xe9e4b068UL, 0x00000000UL, // x^150528 mod p(x)` << 1, x^150592 mod p(x)` << 1
0x288ea478UL, 0x00000001UL, 0x4b38651aUL, 0x00000000UL, // x^149504 mod p(x)` << 1, x^149568 mod p(x)` << 1
0x6619c442UL, 0x00000001UL, 0x468360e2UL, 0x00000001UL, // x^148480 mod p(x)` << 1, x^148544 mod p(x)` << 1
0x86230038UL, 0x00000000UL, 0x121c2408UL, 0x00000000UL, // x^147456 mod p(x)` << 1, x^147520 mod p(x)` << 1
0x7746a756UL, 0x00000001UL, 0xda7e7d08UL, 0x00000000UL, // x^146432 mod p(x)` << 1, x^146496 mod p(x)` << 1
0x91b8f8f8UL, 0x00000001UL, 0x058d7652UL, 0x00000001UL, // x^145408 mod p(x)` << 1, x^145472 mod p(x)` << 1
0x8e167708UL, 0x00000000UL, 0x4a098a90UL, 0x00000001UL, // x^144384 mod p(x)` << 1, x^144448 mod p(x)` << 1
0x48b22d54UL, 0x00000001UL, 0x20dbe72eUL, 0x00000000UL, // x^143360 mod p(x)` << 1, x^143424 mod p(x)` << 1
0x44ba2c3cUL, 0x00000000UL, 0x1e7323e8UL, 0x00000001UL, // x^142336 mod p(x)` << 1, x^142400 mod p(x)` << 1
0xb54d2b52UL, 0x00000000UL, 0xd5d4bf94UL, 0x00000000UL, // x^141312 mod p(x)` << 1, x^141376 mod p(x)` << 1
0x05a4fd8aUL, 0x00000000UL, 0x99d8746cUL, 0x00000001UL, // x^140288 mod p(x)` << 1, x^140352 mod p(x)` << 1
0x39f9fc46UL, 0x00000001UL, 0xce9ca8a0UL, 0x00000000UL, // x^139264 mod p(x)` << 1, x^139328 mod p(x)` << 1
0x5a1fa824UL, 0x00000001UL, 0x136edeceUL, 0x00000000UL, // x^138240 mod p(x)` << 1, x^138304 mod p(x)` << 1
0x0a61ae4cUL, 0x00000000UL, 0x9b92a068UL, 0x00000001UL, // x^137216 mod p(x)` << 1, x^137280 mod p(x)` << 1
0x45e9113eUL, 0x00000001UL, 0x71d62206UL, 0x00000000UL, // x^136192 mod p(x)` << 1, x^136256 mod p(x)` << 1
0x6a348448UL, 0x00000000UL, 0xdfc50158UL, 0x00000000UL, // x^135168 mod p(x)` << 1, x^135232 mod p(x)` << 1
0x4d80a08cUL, 0x00000000UL, 0x517626bcUL, 0x00000001UL, // x^134144 mod p(x)` << 1, x^134208 mod p(x)` << 1
0x4b6837a0UL, 0x00000001UL, 0x48d1e4faUL, 0x00000001UL, // x^133120 mod p(x)` << 1, x^133184 mod p(x)` << 1
0x6896a7fcUL, 0x00000001UL, 0x94d8266eUL, 0x00000000UL, // x^132096 mod p(x)` << 1, x^132160 mod p(x)` << 1
0x4f187140UL, 0x00000001UL, 0x606c5e34UL, 0x00000000UL, // x^131072 mod p(x)` << 1, x^131136 mod p(x)` << 1
0x9581b9daUL, 0x00000001UL, 0x9766beaaUL, 0x00000001UL, // x^130048 mod p(x)` << 1, x^130112 mod p(x)` << 1
0x091bc984UL, 0x00000001UL, 0xd80c506cUL, 0x00000001UL, // x^129024 mod p(x)` << 1, x^129088 mod p(x)` << 1
0x1067223cUL, 0x00000000UL, 0x1e73837cUL, 0x00000000UL, // x^128000 mod p(x)` << 1, x^128064 mod p(x)` << 1
0xab16ea02UL, 0x00000001UL, 0x64d587deUL, 0x00000000UL, // x^126976 mod p(x)` << 1, x^127040 mod p(x)` << 1
0x3c4598a8UL, 0x00000001UL, 0xf4a507b0UL, 0x00000000UL, // x^125952 mod p(x)` << 1, x^126016 mod p(x)` << 1
0xb3735430UL, 0x00000000UL, 0x40e342fcUL, 0x00000000UL, // x^124928 mod p(x)` << 1, x^124992 mod p(x)` << 1
0xbb3fc0c0UL, 0x00000001UL, 0xd5ad9c3aUL, 0x00000001UL, // x^123904 mod p(x)` << 1, x^123968 mod p(x)` << 1
0x570ae19cUL, 0x00000001UL, 0x94a691a4UL, 0x00000000UL, // x^122880 mod p(x)` << 1, x^122944 mod p(x)` << 1
0xea910712UL, 0x00000001UL, 0x271ecdfaUL, 0x00000001UL, // x^121856 mod p(x)` << 1, x^121920 mod p(x)` << 1
0x67127128UL, 0x00000001UL, 0x9e54475aUL, 0x00000000UL, // x^120832 mod p(x)` << 1, x^120896 mod p(x)` << 1
0x19e790a2UL, 0x00000000UL, 0xc9c099eeUL, 0x00000000UL, // x^119808 mod p(x)` << 1, x^119872 mod p(x)` << 1
0x3788f710UL, 0x00000000UL, 0x9a2f736cUL, 0x00000000UL, // x^118784 mod p(x)` << 1, x^118848 mod p(x)` << 1
0x682a160eUL, 0x00000001UL, 0xbb9f4996UL, 0x00000000UL, // x^117760 mod p(x)` << 1, x^117824 mod p(x)` << 1
0x7f0ebd2eUL, 0x00000000UL, 0xdb688050UL, 0x00000001UL, // x^116736 mod p(x)` << 1, x^116800 mod p(x)` << 1
0x2b032080UL, 0x00000000UL, 0xe9b10af4UL, 0x00000000UL, // x^115712 mod p(x)` << 1, x^115776 mod p(x)` << 1
0xcfd1664aUL, 0x00000000UL, 0x2d4545e4UL, 0x00000001UL, // x^114688 mod p(x)` << 1, x^114752 mod p(x)` << 1
0xaa1181c2UL, 0x00000000UL, 0x0361139cUL, 0x00000000UL, // x^113664 mod p(x)` << 1, x^113728 mod p(x)` << 1
0xddd08002UL, 0x00000000UL, 0xa5a1a3a8UL, 0x00000001UL, // x^112640 mod p(x)` << 1, x^112704 mod p(x)` << 1
0xe8dd0446UL, 0x00000000UL, 0x6844e0b0UL, 0x00000000UL, // x^111616 mod p(x)` << 1, x^111680 mod p(x)` << 1
0xbbd94a00UL, 0x00000001UL, 0xc3762f28UL, 0x00000000UL, // x^110592 mod p(x)` << 1, x^110656 mod p(x)` << 1
0xab6cd180UL, 0x00000000UL, 0xd26287a2UL, 0x00000001UL, // x^109568 mod p(x)` << 1, x^109632 mod p(x)` << 1
0x31803ce2UL, 0x00000000UL, 0xf6f0bba8UL, 0x00000001UL, // x^108544 mod p(x)` << 1, x^108608 mod p(x)` << 1
0x24f40b0cUL, 0x00000000UL, 0x2ffabd62UL, 0x00000000UL, // x^107520 mod p(x)` << 1, x^107584 mod p(x)` << 1
0xba1d9834UL, 0x00000001UL, 0xfb4516b8UL, 0x00000000UL, // x^106496 mod p(x)` << 1, x^106560 mod p(x)` << 1
0x04de61aaUL, 0x00000001UL, 0x8cfa961cUL, 0x00000001UL, // x^105472 mod p(x)` << 1, x^105536 mod p(x)` << 1
0x13e40d46UL, 0x00000001UL, 0x9e588d52UL, 0x00000001UL, // x^104448 mod p(x)` << 1, x^104512 mod p(x)` << 1
0x415598a0UL, 0x00000001UL, 0x180f0bbcUL, 0x00000001UL, // x^103424 mod p(x)` << 1, x^103488 mod p(x)` << 1
0xbf6c8c90UL, 0x00000000UL, 0xe1d9177aUL, 0x00000000UL, // x^102400 mod p(x)` << 1, x^102464 mod p(x)` << 1
0x788b0504UL, 0x00000001UL, 0x05abc27cUL, 0x00000001UL, // x^101376 mod p(x)` << 1, x^101440 mod p(x)` << 1
0x38385d02UL, 0x00000000UL, 0x972e4a58UL, 0x00000000UL, // x^100352 mod p(x)` << 1, x^100416 mod p(x)` << 1
0xb6c83844UL, 0x00000001UL, 0x83499a5eUL, 0x00000001UL, // x^99328 mod p(x)` << 1, x^99392 mod p(x)` << 1
0x51061a8aUL, 0x00000000UL, 0xc96a8ccaUL, 0x00000001UL, // x^98304 mod p(x)` << 1, x^98368 mod p(x)` << 1
0x7351388aUL, 0x00000001UL, 0xa1a5b60cUL, 0x00000001UL, // x^97280 mod p(x)` << 1, x^97344 mod p(x)` << 1
0x32928f92UL, 0x00000001UL, 0xe4b6ac9cUL, 0x00000000UL, // x^96256 mod p(x)` << 1, x^96320 mod p(x)` << 1
0xe6b4f48aUL, 0x00000000UL, 0x807e7f5aUL, 0x00000001UL, // x^95232 mod p(x)` << 1, x^95296 mod p(x)` << 1
0x39d15e90UL, 0x00000000UL, 0x7a7e3bc8UL, 0x00000001UL, // x^94208 mod p(x)` << 1, x^94272 mod p(x)` << 1
0x312d6074UL, 0x00000000UL, 0xd73975daUL, 0x00000000UL, // x^93184 mod p(x)` << 1, x^93248 mod p(x)` << 1
0x7bbb2cc4UL, 0x00000001UL, 0x7375d038UL, 0x00000001UL, // x^92160 mod p(x)` << 1, x^92224 mod p(x)` << 1
0x6ded3e18UL, 0x00000001UL, 0x193680bcUL, 0x00000000UL, // x^91136 mod p(x)` << 1, x^91200 mod p(x)` << 1
0xf1638b16UL, 0x00000000UL, 0x999b06f6UL, 0x00000000UL, // x^90112 mod p(x)` << 1, x^90176 mod p(x)` << 1
0xd38b9eccUL, 0x00000001UL, 0xf685d2b8UL, 0x00000001UL, // x^89088 mod p(x)` << 1, x^89152 mod p(x)` << 1
0x8b8d09dcUL, 0x00000001UL, 0xf4ecbed2UL, 0x00000001UL, // x^88064 mod p(x)` << 1, x^88128 mod p(x)` << 1
0xe7bc27d2UL, 0x00000000UL, 0xba16f1a0UL, 0x00000000UL, // x^87040 mod p(x)` << 1, x^87104 mod p(x)` << 1
0x275e1e96UL, 0x00000000UL, 0x15aceac4UL, 0x00000001UL, // x^86016 mod p(x)` << 1, x^86080 mod p(x)` << 1
0xe2e3031eUL, 0x00000000UL, 0xaeff6292UL, 0x00000001UL, // x^84992 mod p(x)` << 1, x^85056 mod p(x)` << 1
0x041c84d8UL, 0x00000001UL, 0x9640124cUL, 0x00000000UL, // x^83968 mod p(x)` << 1, x^84032 mod p(x)` << 1
0x706ce672UL, 0x00000000UL, 0x14f41f02UL, 0x00000001UL, // x^82944 mod p(x)` << 1, x^83008 mod p(x)` << 1
0x5d5070daUL, 0x00000001UL, 0x9c5f3586UL, 0x00000000UL, // x^81920 mod p(x)` << 1, x^81984 mod p(x)` << 1
0x38f9493aUL, 0x00000000UL, 0x878275faUL, 0x00000001UL, // x^80896 mod p(x)` << 1, x^80960 mod p(x)` << 1
0xa3348a76UL, 0x00000000UL, 0xddc42ce8UL, 0x00000000UL, // x^79872 mod p(x)` << 1, x^79936 mod p(x)` << 1
0xad0aab92UL, 0x00000001UL, 0x81d2c73aUL, 0x00000001UL, // x^78848 mod p(x)` << 1, x^78912 mod p(x)` << 1
0x9e85f712UL, 0x00000001UL, 0x41c9320aUL, 0x00000001UL, // x^77824 mod p(x)` << 1, x^77888 mod p(x)` << 1
0x5a871e76UL, 0x00000000UL, 0x5235719aUL, 0x00000001UL, // x^76800 mod p(x)` << 1, x^76864 mod p(x)` << 1
0x7249c662UL, 0x00000001UL, 0xbe27d804UL, 0x00000000UL, // x^75776 mod p(x)` << 1, x^75840 mod p(x)` << 1
0x3a084712UL, 0x00000000UL, 0x6242d45aUL, 0x00000000UL, // x^74752 mod p(x)` << 1, x^74816 mod p(x)` << 1
0xed438478UL, 0x00000000UL, 0x9a53638eUL, 0x00000000UL, // x^73728 mod p(x)` << 1, x^73792 mod p(x)` << 1
0xabac34ccUL, 0x00000000UL, 0x001ecfb6UL, 0x00000001UL, // x^72704 mod p(x)` << 1, x^72768 mod p(x)` << 1
0x5f35ef3eUL, 0x00000000UL, 0x6d7c2d64UL, 0x00000001UL, // x^71680 mod p(x)` << 1, x^71744 mod p(x)` << 1
0x47d6608cUL, 0x00000000UL, 0xd0ce46c0UL, 0x00000001UL, // x^70656 mod p(x)` << 1, x^70720 mod p(x)` << 1
0x2d01470eUL, 0x00000000UL, 0x24c907b4UL, 0x00000001UL, // x^69632 mod p(x)` << 1, x^69696 mod p(x)` << 1
0x58bbc7b0UL, 0x00000001UL, 0x18a555caUL, 0x00000000UL, // x^68608 mod p(x)` << 1, x^68672 mod p(x)` << 1
0xc0a23e8eUL, 0x00000000UL, 0x6b0980bcUL, 0x00000000UL, // x^67584 mod p(x)` << 1, x^67648 mod p(x)` << 1
0xebd85c88UL, 0x00000001UL, 0x8bbba964UL, 0x00000000UL, // x^66560 mod p(x)` << 1, x^66624 mod p(x)` << 1
0x9ee20bb2UL, 0x00000001UL, 0x070a5a1eUL, 0x00000001UL, // x^65536 mod p(x)` << 1, x^65600 mod p(x)` << 1
0xacabf2d6UL, 0x00000001UL, 0x2204322aUL, 0x00000000UL, // x^64512 mod p(x)` << 1, x^64576 mod p(x)` << 1
0xb7963d56UL, 0x00000001UL, 0xa27524d0UL, 0x00000000UL, // x^63488 mod p(x)` << 1, x^63552 mod p(x)` << 1
0x7bffa1feUL, 0x00000001UL, 0x20b1e4baUL, 0x00000000UL, // x^62464 mod p(x)` << 1, x^62528 mod p(x)` << 1
0x1f15333eUL, 0x00000000UL, 0x32cc27fcUL, 0x00000000UL, // x^61440 mod p(x)` << 1, x^61504 mod p(x)` << 1
0x8593129eUL, 0x00000001UL, 0x44dd22b8UL, 0x00000000UL, // x^60416 mod p(x)` << 1, x^60480 mod p(x)` << 1
0x9cb32602UL, 0x00000001UL, 0xdffc9e0aUL, 0x00000000UL, // x^59392 mod p(x)` << 1, x^59456 mod p(x)` << 1
0x42b05cc8UL, 0x00000001UL, 0xb7a0ed14UL, 0x00000001UL, // x^58368 mod p(x)` << 1, x^58432 mod p(x)` << 1
0xbe49e7a4UL, 0x00000001UL, 0xc7842488UL, 0x00000000UL, // x^57344 mod p(x)` << 1, x^57408 mod p(x)` << 1
0x08f69d6cUL, 0x00000001UL, 0xc02a4feeUL, 0x00000001UL, // x^56320 mod p(x)` << 1, x^56384 mod p(x)` << 1
0x6c0971f0UL, 0x00000000UL, 0x3c273778UL, 0x00000000UL, // x^55296 mod p(x)` << 1, x^55360 mod p(x)` << 1
0x5b16467aUL, 0x00000000UL, 0xd63f8894UL, 0x00000001UL, // x^54272 mod p(x)` << 1, x^54336 mod p(x)` << 1
0x551a628eUL, 0x00000001UL, 0x6be557d6UL, 0x00000000UL, // x^53248 mod p(x)` << 1, x^53312 mod p(x)` << 1
0x9e42ea92UL, 0x00000001UL, 0x6a7806eaUL, 0x00000000UL, // x^52224 mod p(x)` << 1, x^52288 mod p(x)` << 1
0x2fa83ff2UL, 0x00000001UL, 0x6155aa0cUL, 0x00000001UL, // x^51200 mod p(x)` << 1, x^51264 mod p(x)` << 1
0x1ca9cde0UL, 0x00000001UL, 0x908650acUL, 0x00000000UL, // x^50176 mod p(x)` << 1, x^50240 mod p(x)` << 1
0xc8e5cd74UL, 0x00000000UL, 0xaa5a8084UL, 0x00000000UL, // x^49152 mod p(x)` << 1, x^49216 mod p(x)` << 1
0x96c27f0cUL, 0x00000000UL, 0x91bb500aUL, 0x00000001UL, // x^48128 mod p(x)` << 1, x^48192 mod p(x)` << 1
0x2baed926UL, 0x00000000UL, 0x64e9bed0UL, 0x00000000UL, // x^47104 mod p(x)` << 1, x^47168 mod p(x)` << 1
0x7c8de8d2UL, 0x00000001UL, 0x9444f302UL, 0x00000000UL, // x^46080 mod p(x)` << 1, x^46144 mod p(x)` << 1
0xd43d6068UL, 0x00000000UL, 0x9db07d3cUL, 0x00000001UL, // x^45056 mod p(x)` << 1, x^45120 mod p(x)` << 1
0xcb2c4b26UL, 0x00000000UL, 0x359e3e6eUL, 0x00000001UL, // x^44032 mod p(x)` << 1, x^44096 mod p(x)` << 1
0x45b8da26UL, 0x00000001UL, 0xe4f10dd2UL, 0x00000001UL, // x^43008 mod p(x)` << 1, x^43072 mod p(x)` << 1
0x8fff4b08UL, 0x00000001UL, 0x24f5735eUL, 0x00000001UL, // x^41984 mod p(x)` << 1, x^42048 mod p(x)` << 1
0x50b58ed0UL, 0x00000001UL, 0x24760a4cUL, 0x00000001UL, // x^40960 mod p(x)` << 1, x^41024 mod p(x)` << 1
0x549f39bcUL, 0x00000001UL, 0x0f1fc186UL, 0x00000000UL, // x^39936 mod p(x)` << 1, x^40000 mod p(x)` << 1
0xef4d2f42UL, 0x00000000UL, 0x150e4cc4UL, 0x00000000UL, // x^38912 mod p(x)` << 1, x^38976 mod p(x)` << 1
0xb1468572UL, 0x00000001UL, 0x2a6204e8UL, 0x00000000UL, // x^37888 mod p(x)` << 1, x^37952 mod p(x)` << 1
0x3d7403b2UL, 0x00000001UL, 0xbeb1d432UL, 0x00000000UL, // x^36864 mod p(x)` << 1, x^36928 mod p(x)` << 1
0xa4681842UL, 0x00000001UL, 0x35f3f1f0UL, 0x00000001UL, // x^35840 mod p(x)` << 1, x^35904 mod p(x)` << 1
0x67714492UL, 0x00000001UL, 0x74fe2232UL, 0x00000000UL, // x^34816 mod p(x)` << 1, x^34880 mod p(x)` << 1
0xe599099aUL, 0x00000001UL, 0x1ac6e2baUL, 0x00000000UL, // x^33792 mod p(x)` << 1, x^33856 mod p(x)` << 1
0xfe128194UL, 0x00000000UL, 0x13fca91eUL, 0x00000000UL, // x^32768 mod p(x)` << 1, x^32832 mod p(x)` << 1
0x77e8b990UL, 0x00000000UL, 0x83f4931eUL, 0x00000001UL, // x^31744 mod p(x)` << 1, x^31808 mod p(x)` << 1
0xa267f63aUL, 0x00000001UL, 0xb6d9b4e4UL, 0x00000000UL, // x^30720 mod p(x)` << 1, x^30784 mod p(x)` << 1
0x945c245aUL, 0x00000001UL, 0xb5188656UL, 0x00000000UL, // x^29696 mod p(x)` << 1, x^29760 mod p(x)` << 1
0x49002e76UL, 0x00000001UL, 0x27a81a84UL, 0x00000000UL, // x^28672 mod p(x)` << 1, x^28736 mod p(x)` << 1
0xbb8310a4UL, 0x00000001UL, 0x25699258UL, 0x00000001UL, // x^27648 mod p(x)` << 1, x^27712 mod p(x)` << 1
0x9ec60bccUL, 0x00000001UL, 0xb23de796UL, 0x00000001UL, // x^26624 mod p(x)` << 1, x^26688 mod p(x)` << 1
0x2d8590aeUL, 0x00000001UL, 0xfe4365dcUL, 0x00000000UL, // x^25600 mod p(x)` << 1, x^25664 mod p(x)` << 1
0x65b00684UL, 0x00000000UL, 0xc68f497aUL, 0x00000000UL, // x^24576 mod p(x)` << 1, x^24640 mod p(x)` << 1
0x5e5aeadcUL, 0x00000001UL, 0xfbf521eeUL, 0x00000000UL, // x^23552 mod p(x)` << 1, x^23616 mod p(x)` << 1
0xb77ff2b0UL, 0x00000000UL, 0x5eac3378UL, 0x00000001UL, // x^22528 mod p(x)` << 1, x^22592 mod p(x)` << 1
0x88da2ff6UL, 0x00000001UL, 0x34914b90UL, 0x00000001UL, // x^21504 mod p(x)` << 1, x^21568 mod p(x)` << 1
0x63da929aUL, 0x00000000UL, 0x16335cfeUL, 0x00000000UL, // x^20480 mod p(x)` << 1, x^20544 mod p(x)` << 1
0x389caa80UL, 0x00000001UL, 0x0372d10cUL, 0x00000001UL, // x^19456 mod p(x)` << 1, x^19520 mod p(x)` << 1
0x3db599d2UL, 0x00000001UL, 0x5097b908UL, 0x00000001UL, // x^18432 mod p(x)` << 1, x^18496 mod p(x)` << 1
0x22505a86UL, 0x00000001UL, 0x227a7572UL, 0x00000001UL, // x^17408 mod p(x)` << 1, x^17472 mod p(x)` << 1
0x6bd72746UL, 0x00000001UL, 0x9a8f75c0UL, 0x00000000UL, // x^16384 mod p(x)` << 1, x^16448 mod p(x)` << 1
0xc3faf1d4UL, 0x00000001UL, 0x682c77a2UL, 0x00000000UL, // x^15360 mod p(x)` << 1, x^15424 mod p(x)` << 1
0x111c826cUL, 0x00000001UL, 0x231f091cUL, 0x00000000UL, // x^14336 mod p(x)` << 1, x^14400 mod p(x)` << 1
0x153e9fb2UL, 0x00000000UL, 0x7d4439f2UL, 0x00000000UL, // x^13312 mod p(x)` << 1, x^13376 mod p(x)` << 1
0x2b1f7b60UL, 0x00000000UL, 0x7e221efcUL, 0x00000001UL, // x^12288 mod p(x)` << 1, x^12352 mod p(x)` << 1
0xb1dba570UL, 0x00000000UL, 0x67457c38UL, 0x00000001UL, // x^11264 mod p(x)` << 1, x^11328 mod p(x)` << 1
0xf6397b76UL, 0x00000001UL, 0xbdf081c4UL, 0x00000000UL, // x^10240 mod p(x)` << 1, x^10304 mod p(x)` << 1
0x56335214UL, 0x00000001UL, 0x6286d6b0UL, 0x00000001UL, // x^9216 mod p(x)` << 1, x^9280 mod p(x)` << 1
0xd70e3986UL, 0x00000001UL, 0xc84f001cUL, 0x00000000UL, // x^8192 mod p(x)` << 1, x^8256 mod p(x)` << 1
0x3701a774UL, 0x00000000UL, 0x64efe7c0UL, 0x00000000UL, // x^7168 mod p(x)` << 1, x^7232 mod p(x)` << 1
0xac81ef72UL, 0x00000000UL, 0x0ac2d904UL, 0x00000000UL, // x^6144 mod p(x)` << 1, x^6208 mod p(x)` << 1
0x33212464UL, 0x00000001UL, 0xfd226d14UL, 0x00000000UL, // x^5120 mod p(x)` << 1, x^5184 mod p(x)` << 1
0xe4e45610UL, 0x00000000UL, 0x1cfd42e0UL, 0x00000001UL, // x^4096 mod p(x)` << 1, x^4160 mod p(x)` << 1
0x0c1bd370UL, 0x00000000UL, 0x6e5a5678UL, 0x00000001UL, // x^3072 mod p(x)` << 1, x^3136 mod p(x)` << 1
0xa7b9e7a6UL, 0x00000001UL, 0xd888fe22UL, 0x00000001UL, // x^2048 mod p(x)` << 1, x^2112 mod p(x)` << 1
0x7d657a10UL, 0x00000000UL, 0xaf77fcd4UL, 0x00000001UL, // x^1024 mod p(x)` << 1, x^1088 mod p(x)` << 1
// Reduce final 1024-2048 bits to 64 bits, shifting 32 bits to include the trailing 32 bits of zeros
0xec447f11UL, 0x99168a18UL, 0x13e8221eUL, 0xed837b26UL, // x^2048 mod p(x)`, x^2016 mod p(x)`, x^1984 mod p(x)`, x^1952 mod p(x)`
0x8fd2cd3cUL, 0xe23e954eUL, 0x47b9ce5aUL, 0xc8acdd81UL, // x^1920 mod p(x)`, x^1888 mod p(x)`, x^1856 mod p(x)`, x^1824 mod p(x)`
0x6b1d2b53UL, 0x92f8befeUL, 0xd4277e25UL, 0xd9ad6d87UL, // x^1792 mod p(x)`, x^1760 mod p(x)`, x^1728 mod p(x)`, x^1696 mod p(x)`
0x291ea462UL, 0xf38a3556UL, 0x33fbca3bUL, 0xc10ec5e0UL, // x^1664 mod p(x)`, x^1632 mod p(x)`, x^1600 mod p(x)`, x^1568 mod p(x)`
0x62b6ca4bUL, 0x974ac562UL, 0x82e02e2fUL, 0xc0b55b0eUL, // x^1536 mod p(x)`, x^1504 mod p(x)`, x^1472 mod p(x)`, x^1440 mod p(x)`
0x784d2a56UL, 0x855712b3UL, 0xe172334dUL, 0x71aa1df0UL, // x^1408 mod p(x)`, x^1376 mod p(x)`, x^1344 mod p(x)`, x^1312 mod p(x)`
0x0eaee722UL, 0xa5abe9f8UL, 0x3969324dUL, 0xfee3053eUL, // x^1280 mod p(x)`, x^1248 mod p(x)`, x^1216 mod p(x)`, x^1184 mod p(x)`
0xdb54814cUL, 0x1fa0943dUL, 0x3eb2bd08UL, 0xf44779b9UL, // x^1152 mod p(x)`, x^1120 mod p(x)`, x^1088 mod p(x)`, x^1056 mod p(x)`
0xd7bbfe6aUL, 0xa53ff440UL, 0x00cc3374UL, 0xf5449b3fUL, // x^1024 mod p(x)`, x^992 mod p(x)`, x^960 mod p(x)`, x^928 mod p(x)`
0x6325605cUL, 0xebe7e356UL, 0xd777606eUL, 0x6f8346e1UL, // x^896 mod p(x)`, x^864 mod p(x)`, x^832 mod p(x)`, x^800 mod p(x)`
0xe5b592b8UL, 0xc65a272cUL, 0xc0b95347UL, 0xe3ab4f2aUL, // x^768 mod p(x)`, x^736 mod p(x)`, x^704 mod p(x)`, x^672 mod p(x)`
0x4721589fUL, 0x5705a9caUL, 0x329ecc11UL, 0xaa2215eaUL, // x^640 mod p(x)`, x^608 mod p(x)`, x^576 mod p(x)`, x^544 mod p(x)`
0x88d14467UL, 0xe3720acbUL, 0xd95efd26UL, 0x1ed8f66eUL, // x^512 mod p(x)`, x^480 mod p(x)`, x^448 mod p(x)`, x^416 mod p(x)`
0x15141c31UL, 0xba1aca03UL, 0xa700e96aUL, 0x78ed02d5UL, // x^384 mod p(x)`, x^352 mod p(x)`, x^320 mod p(x)`, x^288 mod p(x)`
0xed627daeUL, 0xad2a31b3UL, 0x32b39da3UL, 0xba8ccbe8UL, // x^256 mod p(x)`, x^224 mod p(x)`, x^192 mod p(x)`, x^160 mod p(x)`
0xa06a2517UL, 0x6655004fUL, 0xb1e6b092UL, 0xedb88320UL // x^128 mod p(x)`, x^96 mod p(x)`, x^64 mod p(x)`, x^32 mod p(x)`
};
juint* ptr = (juint*) malloc(sizeof(juint) * CRC32_CONSTANTS_SIZE);
guarantee(((intptr_t)ptr & 0xF) == 0, "16-byte alignment needed");
guarantee(ptr != NULL, "allocation error of a crc table");
memcpy((void*)ptr, constants, sizeof(juint) * CRC32_CONSTANTS_SIZE);
return ptr;
}
juint* StubRoutines::ppc64::generate_crc_barret_constants() {
juint barret_constants[CRC32_BARRET_CONSTANTS] = {
0xf7011641UL, 0x00000001UL, 0x00000000UL, 0x00000000UL,
0xdb710641UL, 0x00000001UL, 0x00000000UL, 0x00000000UL
};
juint* ptr = (juint*) malloc(sizeof(juint) * CRC32_CONSTANTS_SIZE);
guarantee(((intptr_t)ptr & 0xF) == 0, "16-byte alignment needed");
guarantee(ptr != NULL, "allocation error of a crc table");
memcpy((void*) ptr, barret_constants, sizeof(juint) * CRC32_BARRET_CONSTANTS);
return ptr;
}
// CRC32 Intrinsics.
/**
* crc_table[] from jdk/src/share/native/java/util/zip/zlib-1.2.8/crc32.h
@ -477,3 +782,7 @@ juint StubRoutines::ppc64::_crc_table[CRC32_TABLES][CRC32_COLUMN_SIZE] = {
#endif
}
};
juint* StubRoutines::ppc64::_constants = StubRoutines::ppc64::generate_crc_constants();
juint* StubRoutines::ppc64::_barret_constants = StubRoutines::ppc64::generate_crc_barret_constants();

6
hotspot/src/cpu/ppc/vm/templateInterpreterGenerator_ppc.cpp
View File

@ -1544,6 +1544,12 @@ address TemplateInterpreterGenerator::generate_native_entry(bool synchronized) {
__ fence();
}
if (CheckJNICalls) {
// clear_pending_jni_exception_check
__ load_const_optimized(R0, 0L);
__ st_ptr(R0, JavaThread::pending_jni_exception_check_fn_offset(), R16_thread);
}
__ reset_last_Java_frame();
// Jvmdi/jvmpi support. Whether we've got an exception pending or

14
hotspot/src/cpu/ppc/vm/vm_version_ppc.cpp
View File

@ -230,6 +230,11 @@ void VM_Version::initialize() {
FLAG_SET_DEFAULT(UseGHASHIntrinsics, false);
}
if (UseFMA) {
warning("FMA instructions are not available on this CPU");
FLAG_SET_DEFAULT(UseFMA, false);
}
if (UseSHA) {
warning("SHA instructions are not available on this CPU");
FLAG_SET_DEFAULT(UseSHA, false);
@ -274,7 +279,14 @@ void VM_Version::initialize() {
}
bool os_too_old = true;
#ifdef AIX
if (os::Aix::os_version() >= 0x0701031e) { // at least AIX 7.1.3.30
// Actually, this is supported since AIX 7.1.. Unfortunately, this first
// contained bugs, so that it can only be enabled after AIX 7.1.3.30.
// The Java property os.version, which is used in RTM tests to decide
// whether the feature is available, only knows major and minor versions.
// We don't want to change this property, as user code might depend on it.
// So the tests can not check on subversion 3.30, and we only enable RTM
// with AIX 7.2.
if (os::Aix::os_version() >= 0x07020000) { // At least AIX 7.2.
os_too_old = false;
}
#endif

35
hotspot/src/cpu/sparc/vm/c1_LIRGenerator_sparc.cpp
View File

@ -147,10 +147,11 @@ LIR_Opr LIRGenerator::safepoint_poll_register() {
LIR_Address* LIRGenerator::generate_address(LIR_Opr base, LIR_Opr index,
int shift, int disp, BasicType type) {
assert(base->is_register(), "must be");
intx large_disp = disp;
// accumulate fixed displacements
if (index->is_constant()) {
disp += index->as_constant_ptr()->as_jint() << shift;
large_disp += (intx)(index->as_constant_ptr()->as_jint()) << shift;
index = LIR_OprFact::illegalOpr;
}
@ -161,31 +162,31 @@ LIR_Address* LIRGenerator::generate_address(LIR_Opr base, LIR_Opr index,
__ shift_left(index, shift, tmp);
index = tmp;
}
if (disp != 0) {
if (large_disp != 0) {
LIR_Opr tmp = new_pointer_register();
if (Assembler::is_simm13(disp)) {
__ add(tmp, LIR_OprFact::intptrConst(disp), tmp);
if (Assembler::is_simm13(large_disp)) {
__ add(tmp, LIR_OprFact::intptrConst(large_disp), tmp);
index = tmp;
} else {
__ move(LIR_OprFact::intptrConst(disp), tmp);
__ move(LIR_OprFact::intptrConst(large_disp), tmp);
__ add(tmp, index, tmp);
index = tmp;
}
disp = 0;
large_disp = 0;
}
} else if (disp != 0 && !Assembler::is_simm13(disp)) {
} else if (large_disp != 0 && !Assembler::is_simm13(large_disp)) {
// index is illegal so replace it with the displacement loaded into a register
index = new_pointer_register();
__ move(LIR_OprFact::intptrConst(disp), index);
disp = 0;
__ move(LIR_OprFact::intptrConst(large_disp), index);
large_disp = 0;
}
// at this point we either have base + index or base + displacement
if (disp == 0) {
if (large_disp == 0) {
return new LIR_Address(base, index, type);
} else {
assert(Assembler::is_simm13(disp), "must be");
return new LIR_Address(base, disp, type);
assert(Assembler::is_simm13(large_disp), "must be");
return new LIR_Address(base, large_disp, type);
}
}
@ -196,11 +197,11 @@ LIR_Address* LIRGenerator::emit_array_address(LIR_Opr array_opr, LIR_Opr index_o
int shift = exact_log2(elem_size);
LIR_Opr base_opr;
int offset = arrayOopDesc::base_offset_in_bytes(type);
intx offset = arrayOopDesc::base_offset_in_bytes(type);
if (index_opr->is_constant()) {
int i = index_opr->as_constant_ptr()->as_jint();
int array_offset = i * elem_size;
intx i = index_opr->as_constant_ptr()->as_jint();
intx array_offset = i * elem_size;
if (Assembler::is_simm13(array_offset + offset)) {
base_opr = array_opr;
offset = array_offset + offset;
@ -953,6 +954,10 @@ void LIRGenerator::do_update_CRC32C(Intrinsic* x) {
}
}
void LIRGenerator::do_FmaIntrinsic(Intrinsic* x) {
fatal("FMA intrinsic is not implemented on this platform");
}
void LIRGenerator::do_vectorizedMismatch(Intrinsic* x) {
fatal("vectorizedMismatch intrinsic is not implemented on this platform");
}

18
hotspot/src/cpu/sparc/vm/c1_Runtime1_sparc.cpp
View File

@ -881,20 +881,15 @@ OopMapSet* Runtime1::generate_code_for(StubID id, StubAssembler* sasm) {
__ delayed()->st_ptr(tmp, G2_thread, satb_q_index_byte_offset);
__ bind(refill);
__ save_frame(0);
__ mov(pre_val, L0);
__ mov(tmp, L1);
__ mov(tmp2, L2);
save_live_registers(sasm);
__ call_VM_leaf(L7_thread_cache,
CAST_FROM_FN_PTR(address,
SATBMarkQueueSet::handle_zero_index_for_thread),
G2_thread);
__ mov(L0, pre_val);
__ mov(L1, tmp);
__ mov(L2, tmp2);
restore_live_registers(sasm);
__ br(Assembler::always, /*annul*/false, Assembler::pt, restart);
__ delayed()->restore();
@ -986,20 +981,15 @@ OopMapSet* Runtime1::generate_code_for(StubID id, StubAssembler* sasm) {
__ delayed()->st_ptr(tmp3, G2_thread, dirty_card_q_index_byte_offset);
__ bind(refill);
__ save_frame(0);
__ mov(tmp2, L0);
__ mov(tmp3, L1);
__ mov(tmp4, L2);
save_live_registers(sasm);
__ call_VM_leaf(L7_thread_cache,
CAST_FROM_FN_PTR(address,
DirtyCardQueueSet::handle_zero_index_for_thread),
G2_thread);
__ mov(L0, tmp2);
__ mov(L1, tmp3);
__ mov(L2, tmp4);
restore_live_registers(sasm);
__ br(Assembler::always, /*annul*/false, Assembler::pt, restart);
__ delayed()->restore();

5
hotspot/src/cpu/sparc/vm/vm_version_sparc.cpp
View File

@ -266,6 +266,11 @@ void VM_Version::initialize() {
FLAG_SET_DEFAULT(UseGHASHIntrinsics, false);
}
if (UseFMA) {
warning("FMA instructions are not available on this CPU");
FLAG_SET_DEFAULT(UseFMA, false);
}
// SHA1, SHA256, and SHA512 instructions were added to SPARC T-series at different times
if (has_sha1() || has_sha256() || has_sha512()) {
if (UseVIS > 0) { // SHA intrinsics use VIS1 instructions

4
hotspot/src/cpu/x86/vm/abstractInterpreter_x86.cpp
View File

@ -172,7 +172,9 @@ bool AbstractInterpreter::can_be_compiled(methodHandle m) {
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 :
case Interpreter::java_lang_math_exp : // fall thru
case Interpreter::java_lang_math_fmaD : // fall thru
case Interpreter::java_lang_math_fmaF :
return false;
default:
return true;

16
hotspot/src/cpu/x86/vm/assembler_x86.cpp
View File

@ -4769,6 +4769,22 @@ void Assembler::vdivss(XMMRegister dst, XMMRegister nds, XMMRegister src) {
emit_int8((unsigned char)(0xC0 | encode));
}
void Assembler::vfmadd231sd(XMMRegister dst, XMMRegister src1, XMMRegister src2) {
assert(VM_Version::supports_fma(), "");
InstructionAttr attributes(AVX_128bit, /* vex_w */ true, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
int encode = vex_prefix_and_encode(dst->encoding(), src1->encoding(), src2->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
emit_int8((unsigned char)0xB9);
emit_int8((unsigned char)(0xC0 | encode));
}
void Assembler::vfmadd231ss(XMMRegister dst, XMMRegister src1, XMMRegister src2) {
assert(VM_Version::supports_fma(), "");
InstructionAttr attributes(AVX_128bit, /* vex_w */ false, /* legacy_mode */ false, /* no_mask_reg */ false, /* uses_vl */ false);
int encode = vex_prefix_and_encode(dst->encoding(), src1->encoding(), src2->encoding(), VEX_SIMD_66, VEX_OPCODE_0F_38, &attributes);
emit_int8((unsigned char)0xB9);
emit_int8((unsigned char)(0xC0 | encode));
}
void Assembler::vmulsd(XMMRegister dst, XMMRegister nds, Address src) {
assert(VM_Version::supports_avx(), "");
InstructionMark im(this);

2
hotspot/src/cpu/x86/vm/assembler_x86.hpp
View File

@ -1860,6 +1860,8 @@ private:
void vdivsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
void vdivss(XMMRegister dst, XMMRegister nds, Address src);
void vdivss(XMMRegister dst, XMMRegister nds, XMMRegister src);
void vfmadd231sd(XMMRegister dst, XMMRegister nds, XMMRegister src);
void vfmadd231ss(XMMRegister dst, XMMRegister nds, XMMRegister src);
void vmulsd(XMMRegister dst, XMMRegister nds, Address src);
void vmulsd(XMMRegister dst, XMMRegister nds, XMMRegister src);
void vmulss(XMMRegister dst, XMMRegister nds, Address src);

12
hotspot/src/cpu/x86/vm/c1_LIRAssembler_x86.cpp
View File

@ -1345,6 +1345,18 @@ void LIR_Assembler::emit_op3(LIR_Op3* op) {
op->result_opr(),
op->info());
break;
case lir_fmad:
__ fmad(op->result_opr()->as_xmm_double_reg(),
op->in_opr1()->as_xmm_double_reg(),
op->in_opr2()->as_xmm_double_reg(),
op->in_opr3()->as_xmm_double_reg());
break;
case lir_fmaf:
__ fmaf(op->result_opr()->as_xmm_float_reg(),
op->in_opr1()->as_xmm_float_reg(),
op->in_opr2()->as_xmm_float_reg(),
op->in_opr3()->as_xmm_float_reg());
break;
default: ShouldNotReachHere(); break;
}
}

30
hotspot/src/cpu/x86/vm/c1_LIRGenerator_x86.cpp
View File

@ -152,7 +152,7 @@ LIR_Address* LIRGenerator::generate_address(LIR_Opr base, LIR_Opr index,
assert(base->is_register(), "must be");
if (index->is_constant()) {
return new LIR_Address(base,
(index->as_constant_ptr()->as_jint() << shift) + disp,
((intx)(index->as_constant_ptr()->as_jint()) << shift) + disp,
type);
} else {
return new LIR_Address(base, index, (LIR_Address::Scale)shift, disp, type);
@ -168,7 +168,7 @@ LIR_Address* LIRGenerator::emit_array_address(LIR_Opr array_opr, LIR_Opr index_o
if (index_opr->is_constant()) {
int elem_size = type2aelembytes(type);
addr = new LIR_Address(array_opr,
offset_in_bytes + index_opr->as_jint() * elem_size, type);
offset_in_bytes + (intx)(index_opr->as_jint()) * elem_size, type);
} else {
#ifdef _LP64
if (index_opr->type() == T_INT) {
@ -806,6 +806,32 @@ void LIRGenerator::do_CompareAndSwap(Intrinsic* x, ValueType* type) {
}
}
void LIRGenerator::do_FmaIntrinsic(Intrinsic* x) {
assert(x->number_of_arguments() == 3, "wrong type");
assert(UseFMA, "Needs FMA instructions support.");
LIRItem value(x->argument_at(0), this);
LIRItem value1(x->argument_at(1), this);
LIRItem value2(x->argument_at(2), this);
value2.set_destroys_register();
value.load_item();
value1.load_item();
value2.load_item();
LIR_Opr calc_input = value.result();
LIR_Opr calc_input1 = value1.result();
LIR_Opr calc_input2 = value2.result();
LIR_Opr calc_result = rlock_result(x);
switch (x->id()) {
case vmIntrinsics::_fmaD: __ fmad(calc_input, calc_input1, calc_input2, calc_result); break;
case vmIntrinsics::_fmaF: __ fmaf(calc_input, calc_input1, calc_input2, calc_result); break;
default: ShouldNotReachHere();
}
}
void LIRGenerator::do_MathIntrinsic(Intrinsic* x) {
assert(x->number_of_arguments() == 1 || (x->number_of_arguments() == 2 && x->id() == vmIntrinsics::_dpow), "wrong type");

2
hotspot/src/cpu/x86/vm/c1_LinearScan_x86.hpp
View File

@ -100,9 +100,11 @@ inline void LinearScan::pd_add_temps(LIR_Op* op) {
inline bool LinearScanWalker::pd_init_regs_for_alloc(Interval* cur) {
int last_xmm_reg = pd_last_xmm_reg;
#ifdef _LP64
if (UseAVX < 3) {
last_xmm_reg = pd_first_xmm_reg + (pd_nof_xmm_regs_frame_map / 2) - 1;
}
#endif
if (allocator()->gen()->is_vreg_flag_set(cur->reg_num(), LIRGenerator::byte_reg)) {
assert(cur->type() != T_FLOAT && cur->type() != T_DOUBLE, "cpu regs only");
_first_reg = pd_first_byte_reg;

18
hotspot/src/cpu/x86/vm/macroAssembler_x86.cpp
View File

@ -3147,6 +3147,24 @@ void MacroAssembler::fremr(Register tmp) {
fpop();
}
// dst = c = a * b + c
void MacroAssembler::fmad(XMMRegister dst, XMMRegister a, XMMRegister b, XMMRegister c) {
Assembler::vfmadd231sd(c, a, b);
if (dst != c) {
movdbl(dst, c);
}
}
// dst = c = a * b + c
void MacroAssembler::fmaf(XMMRegister dst, XMMRegister a, XMMRegister b, XMMRegister c) {
Assembler::vfmadd231ss(c, a, b);
if (dst != c) {
movflt(dst, c);
}
}
void MacroAssembler::incrementl(AddressLiteral dst) {
if (reachable(dst)) {

4
hotspot/src/cpu/x86/vm/macroAssembler_x86.hpp
View File

@ -449,6 +449,10 @@ class MacroAssembler: public Assembler {
// tmp is a temporary register, if none is available use noreg
void fremr(Register tmp);
// dst = c = a * b + c
void fmad(XMMRegister dst, XMMRegister a, XMMRegister b, XMMRegister c);
void fmaf(XMMRegister dst, XMMRegister a, XMMRegister b, XMMRegister c);
// same as fcmp2int, but using SSE2
void cmpss2int(XMMRegister opr1, XMMRegister opr2, Register dst, bool unordered_is_less);

8
hotspot/src/cpu/x86/vm/macroAssembler_x86_sha.cpp
View File

@ -720,19 +720,13 @@ const Register& y2 = r15;
enum {
_XFER_SIZE = 2*64*4, // 2 blocks, 64 rounds, 4 bytes/round
#ifndef _WIN64
_XMM_SAVE_SIZE = 0,
#else
_XMM_SAVE_SIZE = 8*16,
#endif
_INP_END_SIZE = 8,
_INP_SIZE = 8,
_CTX_SIZE = 8,
_RSP_SIZE = 8,
_XFER = 0,
_XMM_SAVE = _XFER + _XFER_SIZE,
_INP_END = _XMM_SAVE + _XMM_SAVE_SIZE,
_INP_END = _XFER + _XFER_SIZE,
_INP = _INP_END + _INP_END_SIZE,
_CTX = _INP + _INP_SIZE,
_RSP = _CTX + _CTX_SIZE,

186
hotspot/src/cpu/x86/vm/stubGenerator_x86_64.cpp
View File

@ -3236,11 +3236,6 @@ class StubGenerator: public StubCodeGenerator {
#ifdef _WIN64
// on win64, fill len_reg from stack position
__ movl(len_reg, len_mem);
// save the xmm registers which must be preserved 6-15
__ subptr(rsp, -rsp_after_call_off * wordSize);
for (int i = 6; i <= XMM_REG_NUM_KEY_LAST; i++) {
__ movdqu(xmm_save(i), as_XMMRegister(i));
}
#else
__ push(len_reg); // Save
#endif
@ -3281,10 +3276,6 @@ class StubGenerator: public StubCodeGenerator {
__ movdqu(Address(rvec, 0), xmm_result); // final value of r stored in rvec of CipherBlockChaining object
#ifdef _WIN64
// restore xmm regs belonging to calling function
for (int i = 6; i <= XMM_REG_NUM_KEY_LAST; i++) {
__ movdqu(as_XMMRegister(i), xmm_save(i));
}
__ movl(rax, len_mem);
#else
__ pop(rax); // return length
@ -3446,11 +3437,6 @@ class StubGenerator: public StubCodeGenerator {
#ifdef _WIN64
// on win64, fill len_reg from stack position
__ movl(len_reg, len_mem);
// save the xmm registers which must be preserved 6-15
__ subptr(rsp, -rsp_after_call_off * wordSize);
for (int i = 6; i <= XMM_REG_NUM_KEY_LAST; i++) {
__ movdqu(xmm_save(i), as_XMMRegister(i));
}
#else
__ push(len_reg); // Save
#endif
@ -3644,10 +3630,6 @@ class StubGenerator: public StubCodeGenerator {
__ movdqu(Address(rvec, 0), xmm_prev_block_cipher); // final value of r stored in rvec of CipherBlockChaining object
__ pop(rbx);
#ifdef _WIN64
// restore regs belonging to calling function
for (int i = 6; i <= XMM_REG_NUM_KEY_LAST; i++) {
__ movdqu(as_XMMRegister(i), xmm_save(i));
}
__ movl(rax, len_mem);
#else
__ pop(rax); // return length
@ -3699,25 +3681,12 @@ class StubGenerator: public StubCodeGenerator {
__ enter();
#ifdef _WIN64
// save the xmm registers which must be preserved 6-7
__ subptr(rsp, 4 * wordSize);
__ movdqu(Address(rsp, 0), xmm6);
__ movdqu(Address(rsp, 2 * wordSize), xmm7);
#endif
__ subptr(rsp, 4 * wordSize);
__ fast_sha1(abcd, e0, e1, msg0, msg1, msg2, msg3, shuf_mask,
buf, state, ofs, limit, rsp, multi_block);
__ addptr(rsp, 4 * wordSize);
#ifdef _WIN64
// restore xmm regs belonging to calling function
__ movdqu(xmm6, Address(rsp, 0));
__ movdqu(xmm7, Address(rsp, 2 * wordSize));
__ addptr(rsp, 4 * wordSize);
#endif
__ leave();
__ ret(0);
@ -3775,22 +3744,6 @@ class StubGenerator: public StubCodeGenerator {
const XMMRegister shuf_mask = xmm8;
__ enter();
#ifdef _WIN64
// save the xmm registers which must be preserved 6-7
__ subptr(rsp, 6 * wordSize);
__ movdqu(Address(rsp, 0), xmm6);
__ movdqu(Address(rsp, 2 * wordSize), xmm7);
__ movdqu(Address(rsp, 4 * wordSize), xmm8);
if (!VM_Version::supports_sha() && VM_Version::supports_avx2()) {
__ subptr(rsp, 10 * wordSize);
__ movdqu(Address(rsp, 0), xmm9);
__ movdqu(Address(rsp, 2 * wordSize), xmm10);
__ movdqu(Address(rsp, 4 * wordSize), xmm11);
__ movdqu(Address(rsp, 6 * wordSize), xmm12);
__ movdqu(Address(rsp, 8 * wordSize), xmm13);
}
#endif
__ subptr(rsp, 4 * wordSize);
@ -3802,21 +3755,7 @@ class StubGenerator: public StubCodeGenerator {
buf, state, ofs, limit, rsp, multi_block, shuf_mask);
}
__ addptr(rsp, 4 * wordSize);
#ifdef _WIN64
// restore xmm regs belonging to calling function
if (!VM_Version::supports_sha() && VM_Version::supports_avx2()) {
__ movdqu(xmm9, Address(rsp, 0));
__ movdqu(xmm10, Address(rsp, 2 * wordSize));
__ movdqu(xmm11, Address(rsp, 4 * wordSize));
__ movdqu(xmm12, Address(rsp, 6 * wordSize));
__ movdqu(xmm13, Address(rsp, 8 * wordSize));
__ addptr(rsp, 10 * wordSize);
}
__ movdqu(xmm6, Address(rsp, 0));
__ movdqu(xmm7, Address(rsp, 2 * wordSize));
__ movdqu(xmm8, Address(rsp, 4 * wordSize));
__ addptr(rsp, 6 * wordSize);
#endif
__ leave();
__ ret(0);
return start;
@ -3917,18 +3856,14 @@ class StubGenerator: public StubCodeGenerator {
}
#ifdef _WIN64
// save the xmm registers which must be preserved 6-14
const int XMM_REG_NUM_KEY_LAST = 14;
__ subptr(rsp, -rsp_after_call_off * wordSize);
for (int i = 6; i <= XMM_REG_NUM_KEY_LAST; i++) {
__ movdqu(xmm_save(i), as_XMMRegister(i));
}
const Address r13_save(rbp, rdi_off * wordSize);
const Address r14_save(rbp, rsi_off * wordSize);
__ movptr(r13_save, r13);
__ movptr(r14_save, r14);
// allocate spill slots for r13, r14
enum {
saved_r13_offset,
saved_r14_offset
};
__ subptr(rsp, 2 * wordSize);
__ movptr(Address(rsp, saved_r13_offset * wordSize), r13);
__ movptr(Address(rsp, saved_r14_offset * wordSize), r14);
// on win64, fill len_reg from stack position
__ movl(len_reg, len_mem);
@ -4130,13 +4065,10 @@ class StubGenerator: public StubCodeGenerator {
__ movdqu(Address(counter, 0), xmm_curr_counter); //save counter back
__ pop(rbx); // pop the saved RBX.
#ifdef _WIN64
// restore regs belonging to calling function
for (int i = 6; i <= XMM_REG_NUM_KEY_LAST; i++) {
__ movdqu(as_XMMRegister(i), xmm_save(i));
}
__ movl(rax, len_mem);
__ movptr(r13, r13_save);
__ movptr(r14, r14_save);
__ movptr(r13, Address(rsp, saved_r13_offset * wordSize));
__ movptr(r14, Address(rsp, saved_r14_offset * wordSize));
__ addptr(rsp, 2 * wordSize);
#else
__ pop(rax); // return 'len'
#endif
@ -4177,10 +4109,6 @@ class StubGenerator: public StubCodeGenerator {
const Register data = c_rarg2;
const Register blocks = c_rarg3;
#ifdef _WIN64
const int XMM_REG_LAST = 10;
#endif
const XMMRegister xmm_temp0 = xmm0;
const XMMRegister xmm_temp1 = xmm1;
const XMMRegister xmm_temp2 = xmm2;
@ -4203,14 +4131,6 @@ class StubGenerator: public StubCodeGenerator {
__ kmovql(k1, rax);
}
#ifdef _WIN64
// save the xmm registers which must be preserved 6-10
__ subptr(rsp, -rsp_after_call_off * wordSize);
for (int i = 6; i <= XMM_REG_LAST; i++) {
__ movdqu(xmm_save(i), as_XMMRegister(i));
}
#endif
__ movdqu(xmm_temp10, ExternalAddress(StubRoutines::x86::ghash_long_swap_mask_addr()));
__ movdqu(xmm_temp0, Address(state, 0));
@ -4310,12 +4230,6 @@ class StubGenerator: public StubCodeGenerator {
__ pshufb(xmm_temp6, xmm_temp10); // Byte swap 16-byte result
__ movdqu(Address(state, 0), xmm_temp6); // store the result
#ifdef _WIN64
// restore xmm regs belonging to calling function
for (int i = 6; i <= XMM_REG_LAST; i++) {
__ movdqu(as_XMMRegister(i), xmm_save(i));
}
#endif
__ leave();
__ ret(0);
return start;
@ -4652,21 +4566,8 @@ class StubGenerator: public StubCodeGenerator {
BLOCK_COMMENT("Entry:");
__ enter(); // required for proper stackwalking of RuntimeStub frame
#ifdef _WIN64
// save the xmm registers which must be preserved 6-7
__ subptr(rsp, 4 * wordSize);
__ movdqu(Address(rsp, 0), xmm6);
__ movdqu(Address(rsp, 2 * wordSize), xmm7);
#endif
__ fast_exp(x0, x1, x2, x3, x4, x5, x6, x7, rax, rcx, rdx, tmp);
#ifdef _WIN64
// restore xmm regs belonging to calling function
__ movdqu(xmm6, Address(rsp, 0));
__ movdqu(xmm7, Address(rsp, 2 * wordSize));
__ addptr(rsp, 4 * wordSize);
#endif
__ leave(); // required for proper stackwalking of RuntimeStub frame
__ ret(0);
@ -4693,21 +4594,8 @@ class StubGenerator: public StubCodeGenerator {
BLOCK_COMMENT("Entry:");
__ enter(); // required for proper stackwalking of RuntimeStub frame
#ifdef _WIN64
// save the xmm registers which must be preserved 6-7
__ subptr(rsp, 4 * wordSize);
__ movdqu(Address(rsp, 0), xmm6);
__ movdqu(Address(rsp, 2 * wordSize), xmm7);
#endif
__ fast_log(x0, x1, x2, x3, x4, x5, x6, x7, rax, rcx, rdx, tmp1, tmp2);
#ifdef _WIN64
// restore xmm regs belonging to calling function
__ movdqu(xmm6, Address(rsp, 0));
__ movdqu(xmm7, Address(rsp, 2 * wordSize));
__ addptr(rsp, 4 * wordSize);
#endif
__ leave(); // required for proper stackwalking of RuntimeStub frame
__ ret(0);
@ -4733,21 +4621,8 @@ class StubGenerator: public StubCodeGenerator {
BLOCK_COMMENT("Entry:");
__ enter(); // required for proper stackwalking of RuntimeStub frame
#ifdef _WIN64
// save the xmm registers which must be preserved 6-7
__ subptr(rsp, 4 * wordSize);
__ movdqu(Address(rsp, 0), xmm6);
__ movdqu(Address(rsp, 2 * wordSize), xmm7);
#endif
__ fast_log10(x0, x1, x2, x3, x4, x5, x6, x7, rax, rcx, rdx, tmp);
#ifdef _WIN64
// restore xmm regs belonging to calling function
__ movdqu(xmm6, Address(rsp, 0));
__ movdqu(xmm7, Address(rsp, 2 * wordSize));
__ addptr(rsp, 4 * wordSize);
#endif
__ leave(); // required for proper stackwalking of RuntimeStub frame
__ ret(0);
@ -4776,21 +4651,8 @@ class StubGenerator: public StubCodeGenerator {
BLOCK_COMMENT("Entry:");
__ enter(); // required for proper stackwalking of RuntimeStub frame
#ifdef _WIN64
// save the xmm registers which must be preserved 6-7
__ subptr(rsp, 4 * wordSize);
__ movdqu(Address(rsp, 0), xmm6);
__ movdqu(Address(rsp, 2 * wordSize), xmm7);
#endif
__ fast_pow(x0, x1, x2, x3, x4, x5, x6, x7, rax, rcx, rdx, tmp1, tmp2, tmp3, tmp4);
#ifdef _WIN64
// restore xmm regs belonging to calling function
__ movdqu(xmm6, Address(rsp, 0));
__ movdqu(xmm7, Address(rsp, 2 * wordSize));
__ addptr(rsp, 4 * wordSize);
#endif
__ leave(); // required for proper stackwalking of RuntimeStub frame
__ ret(0);
@ -4822,18 +4684,10 @@ class StubGenerator: public StubCodeGenerator {
#ifdef _WIN64
__ push(rsi);
__ push(rdi);
// save the xmm registers which must be preserved 6-7
__ subptr(rsp, 4 * wordSize);
__ movdqu(Address(rsp, 0), xmm6);
__ movdqu(Address(rsp, 2 * wordSize), xmm7);
#endif
__ fast_sin(x0, x1, x2, x3, x4, x5, x6, x7, rax, rbx, rcx, rdx, tmp1, tmp2, tmp3, tmp4);
#ifdef _WIN64
// restore xmm regs belonging to calling function
__ movdqu(xmm6, Address(rsp, 0));
__ movdqu(xmm7, Address(rsp, 2 * wordSize));
__ addptr(rsp, 4 * wordSize);
__ pop(rdi);
__ pop(rsi);
#endif
@ -4869,18 +4723,10 @@ class StubGenerator: public StubCodeGenerator {
#ifdef _WIN64
__ push(rsi);
__ push(rdi);
// save the xmm registers which must be preserved 6-7
__ subptr(rsp, 4 * wordSize);
__ movdqu(Address(rsp, 0), xmm6);
__ movdqu(Address(rsp, 2 * wordSize), xmm7);
#endif
__ fast_cos(x0, x1, x2, x3, x4, x5, x6, x7, rax, rcx, rdx, tmp1, tmp2, tmp3, tmp4);
#ifdef _WIN64
// restore xmm regs belonging to calling function
__ movdqu(xmm6, Address(rsp, 0));
__ movdqu(xmm7, Address(rsp, 2 * wordSize));
__ addptr(rsp, 4 * wordSize);
__ pop(rdi);
__ pop(rsi);
#endif
@ -4916,18 +4762,10 @@ class StubGenerator: public StubCodeGenerator {
#ifdef _WIN64
__ push(rsi);
__ push(rdi);
// save the xmm registers which must be preserved 6-7
__ subptr(rsp, 4 * wordSize);
__ movdqu(Address(rsp, 0), xmm6);
__ movdqu(Address(rsp, 2 * wordSize), xmm7);
#endif
__ fast_tan(x0, x1, x2, x3, x4, x5, x6, x7, rax, rcx, rdx, tmp1, tmp2, tmp3, tmp4);
#ifdef _WIN64
// restore xmm regs belonging to calling function
__ movdqu(xmm6, Address(rsp, 0));
__ movdqu(xmm7, Address(rsp, 2 * wordSize));
__ addptr(rsp, 4 * wordSize);
__ pop(rdi);
__ pop(rsi);
#endif

2
hotspot/src/cpu/x86/vm/templateInterpreterGenerator_x86.cpp
View File

@ -55,7 +55,7 @@
// Run with +PrintInterpreter to get the VM to print out the size.
// Max size with JVMTI
#ifdef AMD64
int TemplateInterpreter::InterpreterCodeSize = 256 * 1024;
int TemplateInterpreter::InterpreterCodeSize = JVMCI_ONLY(268) NOT_JVMCI(256) * 1024;
#else
int TemplateInterpreter::InterpreterCodeSize = 224 * 1024;
#endif // AMD64

21
hotspot/src/cpu/x86/vm/templateInterpreterGenerator_x86_32.cpp
View File

@ -341,6 +341,27 @@ address TemplateInterpreterGenerator::generate_math_entry(AbstractInterpreter::M
// [ lo(arg) ]
// [ hi(arg) ]
//
if (kind == Interpreter::java_lang_math_fmaD) {
__ movdbl(xmm2, Address(rsp, 5 * wordSize));
__ movdbl(xmm1, Address(rsp, 3 * wordSize));
__ movdbl(xmm0, Address(rsp, 1 * wordSize));
__ fmad(xmm0, xmm1, xmm2, xmm0);
__ pop(rdi); // get return address
__ mov(rsp, rsi); // set sp to sender sp
__ jmp(rdi);
return entry_point;
} else if (kind == Interpreter::java_lang_math_fmaF) {
__ movflt(xmm2, Address(rsp, 3 * wordSize));
__ movflt(xmm1, Address(rsp, 2 * wordSize));
__ movflt(xmm0, Address(rsp, 1 * wordSize));
__ fmaf(xmm0, xmm1, xmm2, xmm0);
__ pop(rdi); // get return address
__ mov(rsp, rsi); // set sp to sender sp
__ jmp(rdi);
return entry_point;
}
__ fld_d(Address(rsp, 1*wordSize));
switch (kind) {

13
hotspot/src/cpu/x86/vm/templateInterpreterGenerator_x86_64.cpp
View File

@ -369,8 +369,17 @@ address TemplateInterpreterGenerator::generate_math_entry(AbstractInterpreter::M
// [ hi(arg) ]
//
if (kind == Interpreter::java_lang_math_sqrt) {
if (kind == Interpreter::java_lang_math_fmaD) {
__ movdbl(xmm0, Address(rsp, wordSize));
__ movdbl(xmm1, Address(rsp, 3 * wordSize));
__ movdbl(xmm2, Address(rsp, 5 * wordSize));
__ fmad(xmm0, xmm1, xmm2, xmm0);
} else if (kind == Interpreter::java_lang_math_fmaF) {
__ movflt(xmm0, Address(rsp, wordSize));
__ movflt(xmm1, Address(rsp, 2 * wordSize));
__ movflt(xmm2, Address(rsp, 3 * wordSize));
__ fmaf(xmm0, xmm1, xmm2, xmm0);
} else if (kind == Interpreter::java_lang_math_sqrt) {
__ sqrtsd(xmm0, Address(rsp, wordSize));
} else if (kind == Interpreter::java_lang_math_exp) {
__ movdbl(xmm0, Address(rsp, wordSize));

3
hotspot/src/cpu/x86/vm/vmStructs_x86.hpp
View File

@ -73,6 +73,7 @@
#define VM_LONG_CONSTANTS_CPU(declare_constant, declare_preprocessor_constant, declare_c1_constant, declare_c2_constant, declare_c2_preprocessor_constant) \
declare_preprocessor_constant("VM_Version::CPU_AVX512BW", CPU_AVX512BW) \
declare_preprocessor_constant("VM_Version::CPU_AVX512VL", CPU_AVX512VL) \
declare_preprocessor_constant("VM_Version::CPU_SHA", CPU_SHA)
declare_preprocessor_constant("VM_Version::CPU_SHA", CPU_SHA) \
declare_preprocessor_constant("VM_Version::CPU_FMA", CPU_FMA)
#endif // CPU_X86_VM_VMSTRUCTS_X86_HPP

15
hotspot/src/cpu/x86/vm/vm_version_x86.cpp
View File

@ -578,7 +578,7 @@ void VM_Version::get_processor_features() {
}
char buf[256];
jio_snprintf(buf, sizeof(buf), "(%u cores per cpu, %u threads per core) family %d model %d stepping %d%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s",
jio_snprintf(buf, sizeof(buf), "(%u cores per cpu, %u threads per core) family %d model %d stepping %d%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s%s",
cores_per_cpu(), threads_per_core(),
cpu_family(), _model, _stepping,
(supports_cmov() ? ", cmov" : ""),
@ -610,7 +610,8 @@ void VM_Version::get_processor_features() {
(supports_bmi2() ? ", bmi2" : ""),
(supports_adx() ? ", adx" : ""),
(supports_evex() ? ", evex" : ""),
(supports_sha() ? ", sha" : ""));
(supports_sha() ? ", sha" : ""),
(supports_fma() ? ", fma" : ""));
_features_string = os::strdup(buf);
// UseSSE is set to the smaller of what hardware supports and what
@ -732,6 +733,15 @@ void VM_Version::get_processor_features() {
FLAG_SET_DEFAULT(UseGHASHIntrinsics, false);
}
if (supports_fma() && UseSSE >= 2) {
if (FLAG_IS_DEFAULT(UseFMA)) {
UseFMA = true;
}
} else if (UseFMA) {
warning("FMA instructions are not available on this CPU");
FLAG_SET_DEFAULT(UseFMA, false);
}
if (supports_sha() LP64_ONLY(|| supports_avx2() && supports_bmi2())) {
if (FLAG_IS_DEFAULT(UseSHA)) {
UseSHA = true;
@ -773,7 +783,6 @@ void VM_Version::get_processor_features() {
FLAG_SET_DEFAULT(UseAdler32Intrinsics, false);
}
// Adjust RTM (Restricted Transactional Memory) flags
if (!supports_rtm() && UseRTMLocking) {
// Can't continue because UseRTMLocking affects UseBiasedLocking flag
// setting during arguments processing. See use_biased_locking().

7
hotspot/src/cpu/x86/vm/vm_version_x86.hpp
View File

@ -74,7 +74,8 @@ class VM_Version : public Abstract_VM_Version {
: 1,
ssse3 : 1,
cid : 1,
: 2,
: 1,
fma : 1,
cmpxchg16: 1,
: 4,
dca : 1,
@ -289,6 +290,7 @@ protected:
#define CPU_AVX512BW ((uint64_t)UCONST64(0x100000000)) // enums are limited to 31 bit
#define CPU_AVX512VL ((uint64_t)UCONST64(0x200000000)) // EVEX instructions with smaller vector length
#define CPU_SHA ((uint64_t)UCONST64(0x400000000)) // SHA instructions
#define CPU_FMA ((uint64_t)UCONST64(0x800000000)) // FMA instructions
enum Extended_Family {
// AMD
@ -522,6 +524,8 @@ protected:
result |= CPU_SHA;
if(_cpuid_info.ext_cpuid1_ecx.bits.lzcnt_intel != 0)
result |= CPU_LZCNT;
if (_cpuid_info.std_cpuid1_ecx.bits.fma != 0)
result |= CPU_FMA;
// for Intel, ecx.bits.misalignsse bit (bit 8) indicates support for prefetchw
if (_cpuid_info.ext_cpuid1_ecx.bits.misalignsse != 0) {
result |= CPU_3DNOW_PREFETCH;
@ -726,6 +730,7 @@ public:
static bool supports_avx256only() { return (supports_avx2() && !supports_evex()); }
static bool supports_avxonly() { return ((supports_avx2() || supports_avx()) && !supports_evex()); }
static bool supports_sha() { return (_features & CPU_SHA) != 0; }
static bool supports_fma() { return (_features & CPU_FMA) != 0; }
// Intel features
static bool is_intel_family_core() { return is_intel() &&
extended_cpu_family() == CPU_FAMILY_INTEL_CORE; }

471
hotspot/src/cpu/x86/vm/x86.ad
View File

@ -176,451 +176,6 @@ reg_def XMM5n( SOC, SOC, Op_RegF, 5, xmm5->as_VMReg()->next(13));
reg_def XMM5o( SOC, SOC, Op_RegF, 5, xmm5->as_VMReg()->next(14));
reg_def XMM5p( SOC, SOC, Op_RegF, 5, xmm5->as_VMReg()->next(15));
#ifdef _WIN64
reg_def XMM6 ( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg());
reg_def XMM6b( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(1));
reg_def XMM6c( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(2));
reg_def XMM6d( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(3));
reg_def XMM6e( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(4));
reg_def XMM6f( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(5));
reg_def XMM6g( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(6));
reg_def XMM6h( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(7));
reg_def XMM6i( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(8));
reg_def XMM6j( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(9));
reg_def XMM6k( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(10));
reg_def XMM6l( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(11));
reg_def XMM6m( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(12));
reg_def XMM6n( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(13));
reg_def XMM6o( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(14));
reg_def XMM6p( SOC, SOE, Op_RegF, 6, xmm6->as_VMReg()->next(15));
reg_def XMM7 ( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg());
reg_def XMM7b( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(1));
reg_def XMM7c( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(2));
reg_def XMM7d( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(3));
reg_def XMM7e( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(4));
reg_def XMM7f( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(5));
reg_def XMM7g( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(6));
reg_def XMM7h( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(7));
reg_def XMM7i( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(8));
reg_def XMM7j( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(9));
reg_def XMM7k( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(10));
reg_def XMM7l( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(11));
reg_def XMM7m( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(12));
reg_def XMM7n( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(13));
reg_def XMM7o( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(14));
reg_def XMM7p( SOC, SOE, Op_RegF, 7, xmm7->as_VMReg()->next(15));
reg_def XMM8 ( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg());
reg_def XMM8b( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(1));
reg_def XMM8c( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(2));
reg_def XMM8d( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(3));
reg_def XMM8e( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(4));
reg_def XMM8f( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(5));
reg_def XMM8g( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(6));
reg_def XMM8h( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(7));
reg_def XMM8i( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(8));
reg_def XMM8j( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(9));
reg_def XMM8k( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(10));
reg_def XMM8l( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(11));
reg_def XMM8m( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(12));
reg_def XMM8n( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(13));
reg_def XMM8o( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(14));
reg_def XMM8p( SOC, SOE, Op_RegF, 8, xmm8->as_VMReg()->next(15));
reg_def XMM9 ( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg());
reg_def XMM9b( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(1));
reg_def XMM9c( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(2));
reg_def XMM9d( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(3));
reg_def XMM9e( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(4));
reg_def XMM9f( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(5));
reg_def XMM9g( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(6));
reg_def XMM9h( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(7));
reg_def XMM9i( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(8));
reg_def XMM9j( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(9));
reg_def XMM9k( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(10));
reg_def XMM9l( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(11));
reg_def XMM9m( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(12));
reg_def XMM9n( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(13));
reg_def XMM9o( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(14));
reg_def XMM9p( SOC, SOE, Op_RegF, 9, xmm9->as_VMReg()->next(15));
reg_def XMM10 ( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg());
reg_def XMM10b( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(1));
reg_def XMM10c( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(2));
reg_def XMM10d( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(3));
reg_def XMM10e( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(4));
reg_def XMM10f( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(5));
reg_def XMM10g( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(6));
reg_def XMM10h( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(7));
reg_def XMM10i( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(8));
reg_def XMM10j( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(9));
reg_def XMM10k( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(10));
reg_def XMM10l( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(11));
reg_def XMM10m( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(12));
reg_def XMM10n( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(13));
reg_def XMM10o( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(14));
reg_def XMM10p( SOC, SOE, Op_RegF, 10, xmm10->as_VMReg()->next(15));
reg_def XMM11 ( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg());
reg_def XMM11b( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(1));
reg_def XMM11c( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(2));
reg_def XMM11d( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(3));
reg_def XMM11e( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(4));
reg_def XMM11f( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(5));
reg_def XMM11g( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(6));
reg_def XMM11h( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(7));
reg_def XMM11i( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(8));
reg_def XMM11j( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(9));
reg_def XMM11k( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(10));
reg_def XMM11l( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(11));
reg_def XMM11m( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(12));
reg_def XMM11n( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(13));
reg_def XMM11o( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(14));
reg_def XMM11p( SOC, SOE, Op_RegF, 11, xmm11->as_VMReg()->next(15));
reg_def XMM12 ( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg());
reg_def XMM12b( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(1));
reg_def XMM12c( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(2));
reg_def XMM12d( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(3));
reg_def XMM12e( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(4));
reg_def XMM12f( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(5));
reg_def XMM12g( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(6));
reg_def XMM12h( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(7));
reg_def XMM12i( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(8));
reg_def XMM12j( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(9));
reg_def XMM12k( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(10));
reg_def XMM12l( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(11));
reg_def XMM12m( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(12));
reg_def XMM12n( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(13));
reg_def XMM12o( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(14));
reg_def XMM12p( SOC, SOE, Op_RegF, 12, xmm12->as_VMReg()->next(15));
reg_def XMM13 ( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg());
reg_def XMM13b( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(1));
reg_def XMM13c( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(2));
reg_def XMM13d( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(3));
reg_def XMM13e( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(4));
reg_def XMM13f( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(5));
reg_def XMM13g( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(6));
reg_def XMM13h( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(7));
reg_def XMM13i( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(8));
reg_def XMM13j( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(9));
reg_def XMM13k( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(10));
reg_def XMM13l( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(11));
reg_def XMM13m( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(12));
reg_def XMM13n( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(13));
reg_def XMM13o( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(14));
reg_def XMM13p( SOC, SOE, Op_RegF, 13, xmm13->as_VMReg()->next(15));
reg_def XMM14 ( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg());
reg_def XMM14b( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(1));
reg_def XMM14c( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(2));
reg_def XMM14d( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(3));
reg_def XMM14e( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(4));
reg_def XMM14f( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(5));
reg_def XMM14g( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(6));
reg_def XMM14h( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(7));
reg_def XMM14i( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(8));
reg_def XMM14j( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(9));
reg_def XMM14k( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(10));
reg_def XMM14l( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(11));
reg_def XMM14m( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(12));
reg_def XMM14n( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(13));
reg_def XMM14o( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(14));
reg_def XMM14p( SOC, SOE, Op_RegF, 14, xmm14->as_VMReg()->next(15));
reg_def XMM15 ( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg());
reg_def XMM15b( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(1));
reg_def XMM15c( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(2));
reg_def XMM15d( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(3));
reg_def XMM15e( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(4));
reg_def XMM15f( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(5));
reg_def XMM15g( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(6));
reg_def XMM15h( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(7));
reg_def XMM15i( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(8));
reg_def XMM15j( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(9));
reg_def XMM15k( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(10));
reg_def XMM15l( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(11));
reg_def XMM15m( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(12));
reg_def XMM15n( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(13));
reg_def XMM15o( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(14));
reg_def XMM15p( SOC, SOE, Op_RegF, 15, xmm15->as_VMReg()->next(15));
reg_def XMM16 ( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg());
reg_def XMM16b( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(1));
reg_def XMM16c( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(2));
reg_def XMM16d( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(3));
reg_def XMM16e( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(4));
reg_def XMM16f( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(5));
reg_def XMM16g( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(6));
reg_def XMM16h( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(7));
reg_def XMM16i( SOC, SOE, Op_RegF, 16, xmm15->as_VMReg()->next(8));
reg_def XMM16j( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(9));
reg_def XMM16k( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(10));
reg_def XMM16l( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(11));
reg_def XMM16m( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(12));
reg_def XMM16n( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(13));
reg_def XMM16o( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(14));
reg_def XMM16p( SOC, SOE, Op_RegF, 16, xmm16->as_VMReg()->next(15));
reg_def XMM17 ( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg());
reg_def XMM17b( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(1));
reg_def XMM17c( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(2));
reg_def XMM17d( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(3));
reg_def XMM17e( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(4));
reg_def XMM17f( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(5));
reg_def XMM17g( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(6));
reg_def XMM17h( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(7));
reg_def XMM17i( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(8));
reg_def XMM17j( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(9));
reg_def XMM17k( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(10));
reg_def XMM17l( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(11));
reg_def XMM17m( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(12));
reg_def XMM17n( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(13));
reg_def XMM17o( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(14));
reg_def XMM17p( SOC, SOE, Op_RegF, 17, xmm17->as_VMReg()->next(15));
reg_def XMM18 ( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg());
reg_def XMM18b( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(1));
reg_def XMM18c( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(2));
reg_def XMM18d( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(3));
reg_def XMM18e( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(4));
reg_def XMM18f( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(5));
reg_def XMM18g( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(6));
reg_def XMM18h( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(7));
reg_def XMM18i( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(8));
reg_def XMM18j( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(9));
reg_def XMM18k( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(10));
reg_def XMM18l( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(11));
reg_def XMM18m( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(12));
reg_def XMM18n( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(13));
reg_def XMM18o( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(14));
reg_def XMM18p( SOC, SOE, Op_RegF, 18, xmm18->as_VMReg()->next(15));
reg_def XMM19 ( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg());
reg_def XMM19b( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(1));
reg_def XMM19c( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(2));
reg_def XMM19d( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(3));
reg_def XMM19e( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(4));
reg_def XMM19f( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(5));
reg_def XMM19g( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(6));
reg_def XMM19h( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(7));
reg_def XMM19i( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(8));
reg_def XMM19j( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(9));
reg_def XMM19k( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(10));
reg_def XMM19l( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(11));
reg_def XMM19m( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(12));
reg_def XMM19n( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(13));
reg_def XMM19o( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(14));
reg_def XMM19p( SOC, SOE, Op_RegF, 19, xmm19->as_VMReg()->next(15));
reg_def XMM20 ( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg());
reg_def XMM20b( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(1));
reg_def XMM20c( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(2));
reg_def XMM20d( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(3));
reg_def XMM20e( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(4));
reg_def XMM20f( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(5));
reg_def XMM20g( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(6));
reg_def XMM20h( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(7));
reg_def XMM20i( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(8));
reg_def XMM20j( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(9));
reg_def XMM20k( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(10));
reg_def XMM20l( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(11));
reg_def XMM20m( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(12));
reg_def XMM20n( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(13));
reg_def XMM20o( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(14));
reg_def XMM20p( SOC, SOE, Op_RegF, 20, xmm20->as_VMReg()->next(15));
reg_def XMM21 ( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg());
reg_def XMM21b( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(1));
reg_def XMM21c( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(2));
reg_def XMM21d( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(3));
reg_def XMM21e( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(4));
reg_def XMM21f( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(5));
reg_def XMM21g( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(6));
reg_def XMM21h( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(7));
reg_def XMM21i( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(8));
reg_def XMM21j( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(9));
reg_def XMM21k( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(10));
reg_def XMM21l( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(11));
reg_def XMM21m( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(12));
reg_def XMM21n( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(13));
reg_def XMM21o( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(14));
reg_def XMM21p( SOC, SOE, Op_RegF, 21, xmm21->as_VMReg()->next(15));
reg_def XMM22 ( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg());
reg_def XMM22b( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(1));
reg_def XMM22c( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(2));
reg_def XMM22d( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(3));
reg_def XMM22e( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(4));
reg_def XMM22f( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(5));
reg_def XMM22g( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(6));
reg_def XMM22h( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(7));
reg_def XMM22i( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(8));
reg_def XMM22j( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(9));
reg_def XMM22k( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(10));
reg_def XMM22l( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(11));
reg_def XMM22m( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(12));
reg_def XMM22n( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(13));
reg_def XMM22o( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(14));
reg_def XMM22p( SOC, SOE, Op_RegF, 22, xmm22->as_VMReg()->next(15));
reg_def XMM23 ( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg());
reg_def XMM23b( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(1));
reg_def XMM23c( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(2));
reg_def XMM23d( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(3));
reg_def XMM23e( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(4));
reg_def XMM23f( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(5));
reg_def XMM23g( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(6));
reg_def XMM23h( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(7));
reg_def XMM23i( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(8));
reg_def XMM23j( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(9));
reg_def XMM23k( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(10));
reg_def XMM23l( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(11));
reg_def XMM23m( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(12));
reg_def XMM23n( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(13));
reg_def XMM23o( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(14));
reg_def XMM23p( SOC, SOE, Op_RegF, 23, xmm23->as_VMReg()->next(15));
reg_def XMM24 ( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg());
reg_def XMM24b( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(1));
reg_def XMM24c( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(2));
reg_def XMM24d( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(3));
reg_def XMM24e( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(4));
reg_def XMM24f( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(5));
reg_def XMM24g( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(6));
reg_def XMM24h( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(7));
reg_def XMM24i( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(8));
reg_def XMM24j( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(9));
reg_def XMM24k( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(10));
reg_def XMM24l( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(11));
reg_def XMM24m( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(12));
reg_def XMM24n( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(13));
reg_def XMM24o( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(14));
reg_def XMM24p( SOC, SOE, Op_RegF, 24, xmm24->as_VMReg()->next(15));
reg_def XMM25 ( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg());
reg_def XMM25b( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(1));
reg_def XMM25c( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(2));
reg_def XMM25d( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(3));
reg_def XMM25e( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(4));
reg_def XMM25f( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(5));
reg_def XMM25g( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(6));
reg_def XMM25h( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(7));
reg_def XMM25i( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(8));
reg_def XMM25j( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(9));
reg_def XMM25k( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(10));
reg_def XMM25l( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(11));
reg_def XMM25m( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(12));
reg_def XMM25n( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(13));
reg_def XMM25o( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(14));
reg_def XMM25p( SOC, SOE, Op_RegF, 25, xmm25->as_VMReg()->next(15));
reg_def XMM26 ( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg());
reg_def XMM26b( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(1));
reg_def XMM26c( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(2));
reg_def XMM26d( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(3));
reg_def XMM26e( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(4));
reg_def XMM26f( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(5));
reg_def XMM26g( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(6));
reg_def XMM26h( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(7));
reg_def XMM26i( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(8));
reg_def XMM26j( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(9));
reg_def XMM26k( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(10));
reg_def XMM26l( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(11));
reg_def XMM26m( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(12));
reg_def XMM26n( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(13));
reg_def XMM26o( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(14));
reg_def XMM26p( SOC, SOE, Op_RegF, 26, xmm26->as_VMReg()->next(15));
reg_def XMM27g( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(1));
reg_def XMM27c( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(2));
reg_def XMM27d( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(3));
reg_def XMM27e( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(4));
reg_def XMM27f( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(5));
reg_def XMM27g( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(6));
reg_def XMM27h( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(7));
reg_def XMM27i( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(8));
reg_def XMM27j( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(9));
reg_def XMM27k( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(10));
reg_def XMM27l( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(11));
reg_def XMM27m( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(12));
reg_def XMM27n( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(13));
reg_def XMM27o( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(14));
reg_def XMM27p( SOC, SOE, Op_RegF, 27, xmm27->as_VMReg()->next(15));
reg_def XMM28 ( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg());
reg_def XMM28b( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(1));
reg_def XMM28c( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(2));
reg_def XMM28d( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(3));
reg_def XMM28e( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(4));
reg_def XMM28f( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(5));
reg_def XMM28g( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(6));
reg_def XMM28h( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(7));
reg_def XMM28i( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(8));
reg_def XMM28j( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(9));
reg_def XMM28k( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(10));
reg_def XMM28l( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(11));
reg_def XMM28m( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(12));
reg_def XMM28n( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(13));
reg_def XMM28o( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(14));
reg_def XMM28p( SOC, SOE, Op_RegF, 28, xmm28->as_VMReg()->next(15));
reg_def XMM29 ( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg());
reg_def XMM29b( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(1));
reg_def XMM29c( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(2));
reg_def XMM29d( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(3));
reg_def XMM29e( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(4));
reg_def XMM29f( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(5));
reg_def XMM29g( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(6));
reg_def XMM29h( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(7));
reg_def XMM29i( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(8));
reg_def XMM29j( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(9));
reg_def XMM29k( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(10));
reg_def XMM29l( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(11));
reg_def XMM29m( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(12));
reg_def XMM29n( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(13));
reg_def XMM29o( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(14));
reg_def XMM29p( SOC, SOE, Op_RegF, 29, xmm29->as_VMReg()->next(15));
reg_def XMM30 ( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg());
reg_def XMM30b( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(1));
reg_def XMM30c( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(2));
reg_def XMM30d( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(3));
reg_def XMM30e( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(4));
reg_def XMM30f( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(5));
reg_def XMM30g( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(6));
reg_def XMM30h( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(7));
reg_def XMM30i( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(8));
reg_def XMM30j( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(9));
reg_def XMM30k( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(10));
reg_def XMM30l( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(11));
reg_def XMM30m( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(12));
reg_def XMM30n( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(13));
reg_def XMM30o( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(14));
reg_def XMM30p( SOC, SOE, Op_RegF, 30, xmm30->as_VMReg()->next(15));
reg_def XMM31 ( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg());
reg_def XMM31b( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(1));
reg_def XMM31c( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(2));
reg_def XMM31d( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(3));
reg_def XMM31e( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(4));
reg_def XMM31f( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(5));
reg_def XMM31g( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(6));
reg_def XMM31h( SOC, SOE, Op_RegF, 31, xmm31>-as_VMReg()->next(7));
reg_def XMM31i( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(8));
reg_def XMM31j( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(9));
reg_def XMM31k( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(10));
reg_def XMM31l( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(11));
reg_def XMM31m( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(12));
reg_def XMM31n( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(13));
reg_def XMM31o( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(14));
reg_def XMM31p( SOC, SOE, Op_RegF, 31, xmm31->as_VMReg()->next(15));
#else // _WIN64
reg_def XMM6 ( SOC, SOC, Op_RegF, 6, xmm6->as_VMReg());
reg_def XMM6b( SOC, SOC, Op_RegF, 6, xmm6->as_VMReg()->next(1));
reg_def XMM6c( SOC, SOC, Op_RegF, 6, xmm6->as_VMReg()->next(2));
@ -1067,8 +622,6 @@ reg_def XMM31p( SOC, SOC, Op_RegF, 31, xmm31->as_VMReg()->next(15));
#endif // _LP64
#endif // _WIN64
#ifdef _LP64
reg_def RFLAGS(SOC, SOC, 0, 16, VMRegImpl::Bad());
#else
@ -3113,6 +2666,30 @@ instruct onspinwait() %{
ins_pipe(pipe_slow);
%}
// a * b + c
instruct fmaD_reg(regD a, regD b, regD c) %{
predicate(UseFMA);
match(Set c (FmaD c (Binary a b)));
format %{ "fmasd $a,$b,$c\t# $c = $a * $b + $c" %}
ins_cost(150);
ins_encode %{
__ fmad($c$$XMMRegister, $a$$XMMRegister, $b$$XMMRegister, $c$$XMMRegister);
%}
ins_pipe( pipe_slow );
%}
// a * b + c
instruct fmaF_reg(regF a, regF b, regF c) %{
predicate(UseFMA);
match(Set c (FmaF c (Binary a b)));
format %{ "fmass $a,$b,$c\t# $c = $a * $b + $c" %}
ins_cost(150);
ins_encode %{
__ fmaf($c$$XMMRegister, $a$$XMMRegister, $b$$XMMRegister, $c$$XMMRegister);
%}
ins_pipe( pipe_slow );
%}
// ====================VECTOR INSTRUCTIONS=====================================
// Load vectors (4 bytes long)

16
hotspot/src/cpu/x86/vm/x86_32.ad
View File

@ -104,14 +104,14 @@ reg_def FPR7H( SOC, SOC, Op_RegF, 7, as_FloatRegister(6)->as_VMReg()->next());
//
// Empty fill registers, which are never used, but supply alignment to xmm regs
//
reg_def FILL0( SOC, SOC, Op_RegF, 7, as_FloatRegister(6)->as_VMReg()->next(2));
reg_def FILL1( SOC, SOC, Op_RegF, 7, as_FloatRegister(6)->as_VMReg()->next(3));
reg_def FILL2( SOC, SOC, Op_RegF, 7, as_FloatRegister(6)->as_VMReg()->next(4));
reg_def FILL3( SOC, SOC, Op_RegF, 7, as_FloatRegister(6)->as_VMReg()->next(5));
reg_def FILL4( SOC, SOC, Op_RegF, 7, as_FloatRegister(6)->as_VMReg()->next(6));
reg_def FILL5( SOC, SOC, Op_RegF, 7, as_FloatRegister(6)->as_VMReg()->next(7));
reg_def FILL6( SOC, SOC, Op_RegF, 7, as_FloatRegister(6)->as_VMReg()->next(8));
reg_def FILL7( SOC, SOC, Op_RegF, 7, as_FloatRegister(6)->as_VMReg()->next(9));
reg_def FILL0( SOC, SOC, Op_RegF, 8, VMRegImpl::Bad());
reg_def FILL1( SOC, SOC, Op_RegF, 9, VMRegImpl::Bad());
reg_def FILL2( SOC, SOC, Op_RegF, 10, VMRegImpl::Bad());
reg_def FILL3( SOC, SOC, Op_RegF, 11, VMRegImpl::Bad());
reg_def FILL4( SOC, SOC, Op_RegF, 12, VMRegImpl::Bad());
reg_def FILL5( SOC, SOC, Op_RegF, 13, VMRegImpl::Bad());
reg_def FILL6( SOC, SOC, Op_RegF, 14, VMRegImpl::Bad());
reg_def FILL7( SOC, SOC, Op_RegF, 15, VMRegImpl::Bad());
// Specify priority of register selection within phases of register
// allocation. Highest priority is first. A useful heuristic is to

3
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.amd64/src/jdk/vm/ci/amd64/AMD64.java
View File

@ -205,7 +205,8 @@ public class AMD64 extends Architecture {
AVX512CD,
AVX512BW,
AVX512VL,
SHA
SHA,
FMA
}
private final EnumSet<CPUFeature> features;

3
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot.amd64/src/jdk/vm/ci/hotspot/amd64/AMD64HotSpotJVMCIBackendFactory.java
View File

@ -124,6 +124,9 @@ public class AMD64HotSpotJVMCIBackendFactory implements HotSpotJVMCIBackendFacto
if ((config.vmVersionFeatures & config.amd64SHA) != 0) {
features.add(AMD64.CPUFeature.SHA);
}
if ((config.vmVersionFeatures & config.amd64FMA) != 0) {
features.add(AMD64.CPUFeature.FMA);
}
return features;
}

1
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot.amd64/src/jdk/vm/ci/hotspot/amd64/AMD64HotSpotVMConfig.java
View File

@ -78,4 +78,5 @@ class AMD64HotSpotVMConfig extends HotSpotVMConfigAccess {
final long amd64AVX512BW = getConstant("VM_Version::CPU_AVX512BW", Long.class);
final long amd64AVX512VL = getConstant("VM_Version::CPU_AVX512VL", Long.class);
final long amd64SHA = getConstant("VM_Version::CPU_SHA", Long.class);
final long amd64FMA = getConstant("VM_Version::CPU_FMA", Long.class);
}

13
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/CompilerToVM.java
View File

@ -26,8 +26,7 @@ package jdk.vm.ci.hotspot;
import static jdk.vm.ci.common.InitTimer.timer;
import static jdk.vm.ci.hotspot.HotSpotJVMCIRuntime.runtime;
import java.lang.reflect.Constructor;
import java.lang.reflect.Method;
import java.lang.reflect.Executable;
import jdk.vm.ci.code.BytecodeFrame;
import jdk.vm.ci.code.InstalledCode;
@ -385,10 +384,9 @@ final class CompilerToVM {
native boolean hasFinalizableSubclass(HotSpotResolvedObjectTypeImpl type);
/**
* Gets the method corresponding to {@code holder} and slot number {@code slot} (i.e.
* {@link Method#slot} or {@link Constructor#slot}).
* Gets the method corresponding to {@code executable}.
*/
native HotSpotResolvedJavaMethodImpl getResolvedJavaMethodAtSlot(Class<?> holder, int slot);
native HotSpotResolvedJavaMethodImpl asResolvedJavaMethod(Executable executable);
/**
* Gets the maximum absolute offset of a PC relative call to {@code address} from any position
@ -616,4 +614,9 @@ final class CompilerToVM {
*/
native int interpreterFrameSize(BytecodeFrame frame);
/**
* Invokes non-public method {@code java.lang.invoke.LambdaForm.compileToBytecode()} on
* {@code lambdaForm} (which must be a {@code java.lang.invoke.LambdaForm} instance).
*/
native void compileToBytecode(Object lambdaForm);
}

35
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/HotSpotCodeCacheProvider.java
View File

@ -22,7 +22,7 @@
*/
package jdk.vm.ci.hotspot;
import java.lang.reflect.Field;
import java.util.Map;
import jdk.vm.ci.code.BailoutException;
import jdk.vm.ci.code.BytecodeFrame;
@ -56,16 +56,11 @@ public class HotSpotCodeCacheProvider implements CodeCacheProvider {
@Override
public String getMarkName(Mark mark) {
int markId = (int) mark.id;
Field[] fields = runtime.getConfig().getClass().getDeclaredFields();
for (Field f : fields) {
if (f.getName().startsWith("MARKID_")) {
f.setAccessible(true);
try {
if (f.getInt(runtime.getConfig()) == markId) {
return f.getName();
}
} catch (Exception e) {
}
HotSpotVMConfigStore store = runtime.getConfigStore();
for (Map.Entry<String, Long> e : store.getConstants().entrySet()) {
String name = e.getKey();
if (name.startsWith("MARKID_") && e.getValue() == markId) {
return name;
}
}
return CodeCacheProvider.super.getMarkName(mark);
@ -76,17 +71,13 @@ public class HotSpotCodeCacheProvider implements CodeCacheProvider {
*/
@Override
public String getTargetName(Call call) {
Field[] fields = runtime.getConfig().getClass().getDeclaredFields();
for (Field f : fields) {
if (f.getName().endsWith("Stub")) {
f.setAccessible(true);
Object address;
try {
address = f.get(runtime.getConfig());
if (address.equals(call.target)) {
return f.getName() + ":0x" + Long.toHexString((Long) address);
}
} catch (IllegalArgumentException | IllegalAccessException e) {
if (call.target instanceof HotSpotForeignCallTarget) {
long address = ((HotSpotForeignCallTarget) call.target).address;
HotSpotVMConfigStore store = runtime.getConfigStore();
for (Map.Entry<String, VMField> e : store.getFields().entrySet()) {
VMField field = e.getValue();
if (field.isStatic() && field.value != null && field.value == address) {
return e.getValue() + ":0x" + Long.toHexString(address);
}
}
}

32
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/HotSpotMetaAccessProvider.java
View File

@ -28,11 +28,10 @@ import static jdk.vm.ci.hotspot.HotSpotResolvedObjectTypeImpl.fromObjectClass;
import static jdk.vm.ci.hotspot.UnsafeAccess.UNSAFE;
import java.lang.reflect.Array;
import java.lang.reflect.Constructor;
import java.lang.reflect.Executable;
import java.lang.reflect.Field;
import java.lang.reflect.Method;
import java.lang.reflect.Modifier;
import java.util.Objects;
import jdk.vm.ci.code.CodeUtil;
import jdk.vm.ci.code.TargetDescription;
@ -78,35 +77,8 @@ public class HotSpotMetaAccessProvider implements MetaAccessProvider {
return new HotSpotSignature(runtime, signature);
}
/**
* {@link Field} object of {@link Method#slot}.
*/
private Field reflectionMethodSlot = getReflectionSlotField(Method.class);
/**
* {@link Field} object of {@link Constructor#slot}.
*/
private Field reflectionConstructorSlot = getReflectionSlotField(Constructor.class);
private static Field getReflectionSlotField(Class<?> reflectionClass) {
try {
Field field = reflectionClass.getDeclaredField("slot");
field.setAccessible(true);
return field;
} catch (NoSuchFieldException | SecurityException e) {
throw new JVMCIError(e);
}
}
public ResolvedJavaMethod lookupJavaMethod(Executable reflectionMethod) {
try {
Class<?> holder = reflectionMethod.getDeclaringClass();
Field slotField = reflectionMethod instanceof Constructor ? reflectionConstructorSlot : reflectionMethodSlot;
final int slot = slotField.getInt(reflectionMethod);
return runtime.getCompilerToVM().getResolvedJavaMethodAtSlot(holder, slot);
} catch (IllegalArgumentException | IllegalAccessException e) {
throw new JVMCIError(e);
}
return runtime.getCompilerToVM().asResolvedJavaMethod(Objects.requireNonNull(reflectionMethod));
}
public ResolvedJavaField lookupJavaField(Field reflectionField) {

91
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/HotSpotMethodHandleAccessProvider.java
View File

@ -24,16 +24,17 @@ package jdk.vm.ci.hotspot;
import static jdk.vm.ci.hotspot.CompilerToVM.compilerToVM;
import static jdk.vm.ci.hotspot.HotSpotJVMCIRuntime.runtime;
import static jdk.vm.ci.hotspot.HotSpotResolvedObjectTypeImpl.fromObjectClass;
import java.lang.invoke.MethodHandle;
import java.util.Objects;
import jdk.vm.ci.common.JVMCIError;
import jdk.vm.ci.meta.ConstantReflectionProvider;
import jdk.vm.ci.meta.JavaConstant;
import jdk.vm.ci.meta.JavaKind;
import jdk.vm.ci.meta.MethodHandleAccessProvider;
import jdk.vm.ci.meta.ResolvedJavaField;
import jdk.vm.ci.meta.ResolvedJavaMethod;
import jdk.vm.ci.meta.ResolvedJavaType;
import jdk.vm.ci.meta.Signature;
public class HotSpotMethodHandleAccessProvider implements MethodHandleAccessProvider {
@ -48,88 +49,46 @@ public class HotSpotMethodHandleAccessProvider implements MethodHandleAccessProv
* possible after the {@link HotSpotJVMCIRuntime} is fully initialized.
*/
static class LazyInitialization {
static final ResolvedJavaType lambdaFormType;
static final ResolvedJavaField methodHandleFormField;
static final ResolvedJavaField lambdaFormVmentryField;
static final ResolvedJavaMethod lambdaFormCompileToBytecodeMethod;
static final HotSpotResolvedJavaField memberNameVmtargetField;
static final ResolvedJavaType CLASS = fromObjectClass(LazyInitialization.class);
/**
* Search for an instance field with the given name in a class.
*
* @param className name of the class to search in
* @param declaringType the type declaring the field
* @param fieldName name of the field to be searched
* @param fieldType resolved Java type of the field
* @return resolved Java field
* @throws ClassNotFoundException
* @throws NoSuchFieldError
*/
private static ResolvedJavaField findFieldInClass(String className, String fieldName, ResolvedJavaType fieldType)
throws ClassNotFoundException {
Class<?> clazz = Class.forName(className);
ResolvedJavaType type = runtime().fromClass(clazz);
ResolvedJavaField[] fields = type.getInstanceFields(false);
private static ResolvedJavaField findFieldInClass(ResolvedJavaType declaringType, String fieldName, ResolvedJavaType fieldType) {
ResolvedJavaField[] fields = declaringType.getInstanceFields(false);
for (ResolvedJavaField field : fields) {
if (field.getName().equals(fieldName) && field.getType().equals(fieldType)) {
return field;
}
}
throw new NoSuchFieldError(fieldType.getName() + " " + className + "." + fieldName);
throw new NoSuchFieldError(fieldType.getName() + " " + declaringType + "." + fieldName);
}
private static ResolvedJavaMethod findMethodInClass(String className, String methodName,
ResolvedJavaType resultType, ResolvedJavaType[] parameterTypes) throws ClassNotFoundException {
Class<?> clazz = Class.forName(className);
HotSpotResolvedObjectTypeImpl type = fromObjectClass(clazz);
ResolvedJavaMethod result = null;
for (ResolvedJavaMethod method : type.getDeclaredMethods()) {
if (method.getName().equals(methodName) && signatureMatches(method, resultType, parameterTypes)) {
result = method;
}
}
if (result == null) {
StringBuilder sig = new StringBuilder("(");
for (ResolvedJavaType t : parameterTypes) {
sig.append(t.getName()).append(",");
}
if (sig.length() > 1) {
sig.replace(sig.length() - 1, sig.length(), ")");
} else {
sig.append(')');
}
throw new NoSuchMethodError(resultType.getName() + " " + className + "." + methodName + sig.toString());
}
return result;
private static ResolvedJavaType resolveType(Class<?> c) {
return runtime().fromClass(c);
}
private static boolean signatureMatches(ResolvedJavaMethod m, ResolvedJavaType resultType,
ResolvedJavaType[] parameterTypes) {
Signature s = m.getSignature();
if (!s.getReturnType(CLASS).equals(resultType)) {
return false;
}
if (s.getParameterCount(false) != parameterTypes.length) {
return false;
}
for (int i = 0; i < s.getParameterCount(false); ++i) {
if (!s.getParameterType(i, CLASS).equals(parameterTypes[i])) {
return false;
}
}
return true;
private static ResolvedJavaType resolveType(String className) throws ClassNotFoundException {
return resolveType(Class.forName(className));
}
static {
try {
methodHandleFormField = findFieldInClass("java.lang.invoke.MethodHandle", "form",
fromObjectClass(Class.forName("java.lang.invoke.LambdaForm")));
lambdaFormVmentryField = findFieldInClass("java.lang.invoke.LambdaForm", "vmentry",
fromObjectClass(Class.forName("java.lang.invoke.MemberName")));
lambdaFormCompileToBytecodeMethod = findMethodInClass("java.lang.invoke.LambdaForm", "compileToBytecode",
new HotSpotResolvedPrimitiveType(JavaKind.Void), new ResolvedJavaType[]{});
memberNameVmtargetField = (HotSpotResolvedJavaField) findFieldInClass("java.lang.invoke.MemberName", "vmtarget",
new HotSpotResolvedPrimitiveType(JavaKind.Long));
ResolvedJavaType methodHandleType = resolveType(MethodHandle.class);
ResolvedJavaType memberNameType = resolveType("java.lang.invoke.MemberName");
lambdaFormType = resolveType("java.lang.invoke.LambdaForm");
methodHandleFormField = findFieldInClass(methodHandleType, "form", lambdaFormType);
lambdaFormVmentryField = findFieldInClass(lambdaFormType, "vmentry", memberNameType);
memberNameVmtargetField = (HotSpotResolvedJavaField) findFieldInClass(memberNameType, "vmtarget", resolveType(long.class));
} catch (Throwable ex) {
throw new JVMCIError(ex);
}
@ -173,12 +132,13 @@ public class HotSpotMethodHandleAccessProvider implements MethodHandleAccessProv
return null;
}
if (forceBytecodeGeneration) {
/* Invoke non-public method: MemberName LambdaForm.compileToBytecode() */
LazyInitialization.lambdaFormCompileToBytecodeMethod.invoke(lambdaForm, new JavaConstant[0]);
}
/* Load non-public field: MemberName LambdaForm.vmentry */
JavaConstant memberName = constantReflection.readFieldValue(LazyInitialization.lambdaFormVmentryField, lambdaForm);
if (memberName.isNull() && forceBytecodeGeneration) {
Object lf = ((HotSpotObjectConstant) lambdaForm).asObject(LazyInitialization.lambdaFormType);
compilerToVM().compileToBytecode(Objects.requireNonNull(lf));
memberName = constantReflection.readFieldValue(LazyInitialization.lambdaFormVmentryField, lambdaForm);
assert memberName.isNonNull();
}
return getTargetMethod(memberName);
}
@ -200,4 +160,3 @@ public class HotSpotMethodHandleAccessProvider implements MethodHandleAccessProv
return compilerToVM().getResolvedJavaMethod(object, LazyInitialization.memberNameVmtargetField.offset());
}
}

66
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/HotSpotModifiers.java Normal file
View File

@ -0,0 +1,66 @@
/*
* Copyright (c) 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
* 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.
*/
package jdk.vm.ci.hotspot;
import static java.lang.reflect.Modifier.ABSTRACT;
import static java.lang.reflect.Modifier.FINAL;
import static java.lang.reflect.Modifier.INTERFACE;
import static java.lang.reflect.Modifier.NATIVE;
import static java.lang.reflect.Modifier.PRIVATE;
import static java.lang.reflect.Modifier.PROTECTED;
import static java.lang.reflect.Modifier.PUBLIC;
import static java.lang.reflect.Modifier.STATIC;
import static java.lang.reflect.Modifier.STRICT;
import static java.lang.reflect.Modifier.SYNCHRONIZED;
import static java.lang.reflect.Modifier.TRANSIENT;
import static java.lang.reflect.Modifier.VOLATILE;
import static jdk.vm.ci.hotspot.HotSpotVMConfig.config;
import java.lang.reflect.Modifier;
/**
* The non-public modifiers in {@link Modifier} that need to be retrieved from
* {@link HotSpotVMConfig}.
*/
public class HotSpotModifiers {
// @formatter:off
public static final int ANNOTATION = config().jvmAccAnnotation;
public static final int ENUM = config().jvmAccEnum;
public static final int VARARGS = config().jvmAccVarargs;
public static final int BRIDGE = config().jvmAccBridge;
public static final int SYNTHETIC = config().jvmAccSynthetic;
// @formatter:on
public static int jvmClassModifiers() {
return PUBLIC | FINAL | INTERFACE | ABSTRACT | ANNOTATION | ENUM | SYNTHETIC;
}
public static int jvmMethodModifiers() {
return PUBLIC | PRIVATE | PROTECTED | STATIC | FINAL | SYNCHRONIZED | BRIDGE | VARARGS | NATIVE | ABSTRACT | STRICT | SYNTHETIC;
}
public static int jvmFieldModifiers() {
return PUBLIC | PRIVATE | PROTECTED | STATIC | FINAL | VOLATILE | TRANSIENT | ENUM | SYNTHETIC;
}
}

4
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/HotSpotResolvedJavaFieldImpl.java
View File

@ -22,6 +22,7 @@
*/
package jdk.vm.ci.hotspot;
import static jdk.vm.ci.hotspot.HotSpotModifiers.jvmFieldModifiers;
import static jdk.vm.ci.hotspot.HotSpotVMConfig.config;
import java.lang.annotation.Annotation;
@ -29,7 +30,6 @@ import java.lang.reflect.Field;
import jdk.internal.vm.annotation.Stable;
import jdk.vm.ci.meta.JavaType;
import jdk.vm.ci.meta.ModifiersProvider;
import jdk.vm.ci.meta.ResolvedJavaType;
/**
@ -81,7 +81,7 @@ class HotSpotResolvedJavaFieldImpl implements HotSpotResolvedJavaField {
@Override
public int getModifiers() {
return modifiers & ModifiersProvider.jvmFieldModifiers();
return modifiers & jvmFieldModifiers();
}
@Override

44
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/HotSpotResolvedJavaMethodImpl.java
View File

@ -24,13 +24,15 @@ package jdk.vm.ci.hotspot;
import static jdk.vm.ci.hotspot.CompilerToVM.compilerToVM;
import static jdk.vm.ci.hotspot.HotSpotJVMCIRuntime.runtime;
import static jdk.vm.ci.hotspot.HotSpotModifiers.BRIDGE;
import static jdk.vm.ci.hotspot.HotSpotModifiers.SYNTHETIC;
import static jdk.vm.ci.hotspot.HotSpotModifiers.VARARGS;
import static jdk.vm.ci.hotspot.HotSpotModifiers.jvmMethodModifiers;
import static jdk.vm.ci.hotspot.HotSpotVMConfig.config;
import static jdk.vm.ci.hotspot.UnsafeAccess.UNSAFE;
import java.lang.annotation.Annotation;
import java.lang.reflect.Executable;
import java.lang.reflect.InvocationTargetException;
import java.lang.reflect.Method;
import java.lang.reflect.Modifier;
import java.lang.reflect.Type;
import java.util.HashMap;
@ -42,13 +44,11 @@ import jdk.vm.ci.meta.Constant;
import jdk.vm.ci.meta.ConstantPool;
import jdk.vm.ci.meta.DefaultProfilingInfo;
import jdk.vm.ci.meta.ExceptionHandler;
import jdk.vm.ci.meta.JavaConstant;
import jdk.vm.ci.meta.JavaMethod;
import jdk.vm.ci.meta.JavaType;
import jdk.vm.ci.meta.LineNumberTable;
import jdk.vm.ci.meta.Local;
import jdk.vm.ci.meta.LocalVariableTable;
import jdk.vm.ci.meta.ModifiersProvider;
import jdk.vm.ci.meta.ProfilingInfo;
import jdk.vm.ci.meta.ResolvedJavaMethod;
import jdk.vm.ci.meta.ResolvedJavaType;
@ -210,7 +210,7 @@ final class HotSpotResolvedJavaMethodImpl extends HotSpotMethod implements HotSp
@Override
public int getModifiers() {
return getAllModifiers() & ModifiersProvider.jvmMethodModifiers();
return getAllModifiers() & jvmMethodModifiers();
}
@Override
@ -490,6 +490,19 @@ final class HotSpotResolvedJavaMethodImpl extends HotSpotMethod implements HotSp
return javaMethod == null ? null : javaMethod.getAnnotation(annotationClass);
}
public boolean isBridge() {
return (BRIDGE & getModifiers()) != 0;
}
@Override
public boolean isSynthetic() {
return (SYNTHETIC & getModifiers()) != 0;
}
public boolean isVarArgs() {
return (VARARGS & getModifiers()) != 0;
}
public boolean isDefault() {
if (isConstructor()) {
return false;
@ -697,27 +710,6 @@ final class HotSpotResolvedJavaMethodImpl extends HotSpotMethod implements HotSp
return (getFlags() & config().methodFlagsIntrinsicCandidate) != 0;
}
@Override
public JavaConstant invoke(JavaConstant receiver, JavaConstant[] arguments) {
assert !isConstructor();
Method javaMethod = (Method) toJava();
javaMethod.setAccessible(true);
Object[] objArguments = new Object[arguments.length];
for (int i = 0; i < arguments.length; i++) {
objArguments[i] = HotSpotObjectConstantImpl.asBoxedValue(arguments[i]);
}
Object objReceiver = receiver != null && !receiver.isNull() ? ((HotSpotObjectConstantImpl) receiver).object() : null;
try {
Object objResult = javaMethod.invoke(objReceiver, objArguments);
return javaMethod.getReturnType() == void.class ? null : HotSpotObjectConstantImpl.forBoxedValue(getSignature().getReturnKind(), objResult);
} catch (IllegalAccessException | InvocationTargetException ex) {
throw new IllegalArgumentException(ex);
}
}
/**
* Allocates a compile id for this method by asking the VM for one.
*

6
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/HotSpotResolvedObjectTypeImpl.java
View File

@ -26,6 +26,7 @@ import static java.util.Objects.requireNonNull;
import static jdk.vm.ci.hotspot.CompilerToVM.compilerToVM;
import static jdk.vm.ci.hotspot.HotSpotConstantPool.isSignaturePolymorphicHolder;
import static jdk.vm.ci.hotspot.HotSpotJVMCIRuntime.runtime;
import static jdk.vm.ci.hotspot.HotSpotModifiers.jvmClassModifiers;
import static jdk.vm.ci.hotspot.HotSpotVMConfig.config;
import static jdk.vm.ci.hotspot.UnsafeAccess.UNSAFE;
@ -49,7 +50,6 @@ import jdk.vm.ci.meta.Constant;
import jdk.vm.ci.meta.JavaConstant;
import jdk.vm.ci.meta.JavaKind;
import jdk.vm.ci.meta.JavaType;
import jdk.vm.ci.meta.ModifiersProvider;
import jdk.vm.ci.meta.ResolvedJavaField;
import jdk.vm.ci.meta.ResolvedJavaMethod;
import jdk.vm.ci.meta.ResolvedJavaType;
@ -152,7 +152,7 @@ final class HotSpotResolvedObjectTypeImpl extends HotSpotResolvedJavaType implem
if (isArray()) {
return (getElementalType().getModifiers() & (Modifier.PUBLIC | Modifier.PRIVATE | Modifier.PROTECTED)) | Modifier.FINAL | Modifier.ABSTRACT;
} else {
return getAccessFlags() & ModifiersProvider.jvmClassModifiers();
return getAccessFlags() & jvmClassModifiers();
}
}
@ -507,7 +507,7 @@ final class HotSpotResolvedObjectTypeImpl extends HotSpotResolvedJavaType implem
synchronized HotSpotResolvedJavaField createField(String fieldName, JavaType type, long offset, int rawFlags) {
HotSpotResolvedJavaField result = null;
final int flags = rawFlags & ModifiersProvider.jvmFieldModifiers();
final int flags = rawFlags & HotSpotModifiers.jvmFieldModifiers();
final long id = offset + ((long) flags << 32);

6
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/HotSpotVMConfig.java
View File

@ -117,8 +117,12 @@ class HotSpotVMConfig extends HotSpotVMConfigAccess {
final int jvmAccFieldHasGenericSignature = getConstant("JVM_ACC_FIELD_HAS_GENERIC_SIGNATURE", Integer.class);
final int jvmAccIsCloneableFast = getConstant("JVM_ACC_IS_CLONEABLE_FAST", Integer.class);
// Modifier.SYNTHETIC is not public so we get it via vmStructs.
// These modifiers are not public in Modifier so we get them via vmStructs.
final int jvmAccSynthetic = getConstant("JVM_ACC_SYNTHETIC", Integer.class);
final int jvmAccAnnotation = getConstant("JVM_ACC_ANNOTATION", Integer.class);
final int jvmAccBridge = getConstant("JVM_ACC_BRIDGE", Integer.class);
final int jvmAccVarargs = getConstant("JVM_ACC_VARARGS", Integer.class);
final int jvmAccEnum = getConstant("JVM_ACC_ENUM", Integer.class);
// This is only valid on AMD64.
final int runtimeCallStackSize = getConstant("frame::arg_reg_save_area_bytes", Integer.class, osArch.equals("amd64") ? null : 0);

20
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.hotspot/src/jdk/vm/ci/hotspot/UnsafeAccess.java
View File

@ -22,8 +22,6 @@
*/
package jdk.vm.ci.hotspot;
import java.lang.reflect.Field;
import jdk.internal.misc.Unsafe;
/**
@ -31,21 +29,5 @@ import jdk.internal.misc.Unsafe;
*/
class UnsafeAccess {
static final Unsafe UNSAFE = initUnsafe();
private static Unsafe initUnsafe() {
try {
// Fast path when we are trusted.
return Unsafe.getUnsafe();
} catch (SecurityException se) {
// Slow path when we are not trusted.
try {
Field theUnsafe = Unsafe.class.getDeclaredField("theUnsafe");
theUnsafe.setAccessible(true);
return (Unsafe) theUnsafe.get(Unsafe.class);
} catch (Exception e) {
throw new RuntimeException("exception while trying to get Unsafe", e);
}
}
}
static final Unsafe UNSAFE = Unsafe.getUnsafe();
}

16
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.meta/src/jdk/vm/ci/meta/MetaUtil.java
View File

@ -22,9 +22,6 @@
*/
package jdk.vm.ci.meta;
import java.lang.reflect.Field;
import java.lang.reflect.Modifier;
/**
* Miscellaneous collection of utility methods used by {@code jdk.vm.ci.meta} and its clients.
*/
@ -226,17 +223,4 @@ public class MetaUtil {
}
return obj.getClass().getName() + "@" + System.identityHashCode(obj);
}
/**
* Used to lookup constants from {@link Modifier} that are not public (VARARGS, SYNTHETIC etc.).
*/
static int getNonPublicModifierStaticField(String name) {
try {
Field field = Modifier.class.getDeclaredField(name);
field.setAccessible(true);
return field.getInt(null);
} catch (NoSuchFieldException | SecurityException | IllegalArgumentException | IllegalAccessException e) {
throw new InternalError(e);
}
}
}

32
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.meta/src/jdk/vm/ci/meta/ModifiersProvider.java
View File

@ -22,18 +22,9 @@
*/
package jdk.vm.ci.meta;
import static java.lang.reflect.Modifier.ABSTRACT;
import static java.lang.reflect.Modifier.FINAL;
import static java.lang.reflect.Modifier.INTERFACE;
import static java.lang.reflect.Modifier.NATIVE;
import static java.lang.reflect.Modifier.PRIVATE;
import static java.lang.reflect.Modifier.PROTECTED;
import static java.lang.reflect.Modifier.PUBLIC;
import static java.lang.reflect.Modifier.STATIC;
import static java.lang.reflect.Modifier.STRICT;
import static java.lang.reflect.Modifier.SYNCHRONIZED;
import static java.lang.reflect.Modifier.TRANSIENT;
import static java.lang.reflect.Modifier.VOLATILE;
import java.lang.reflect.Modifier;
@ -42,17 +33,9 @@ import java.lang.reflect.Modifier;
* language {@linkplain #getModifiers() modifiers}.
*/
public interface ModifiersProvider {
int BRIDGE = MetaUtil.getNonPublicModifierStaticField("BRIDGE");
int VARARGS = MetaUtil.getNonPublicModifierStaticField("VARARGS");
int SYNTHETIC = MetaUtil.getNonPublicModifierStaticField("SYNTHETIC");
int ANNOTATION = MetaUtil.getNonPublicModifierStaticField("ANNOTATION");
int ENUM = MetaUtil.getNonPublicModifierStaticField("ENUM");
int MANDATED = MetaUtil.getNonPublicModifierStaticField("MANDATED");
/**
* Returns the Java Virtual Machine modifiers for this element. Note that this can differ from
* standard Java Reflection modifiers. For example at the JVM level, classes (
* {@link ResolvedJavaType}) can not be private or protected.
* Returns the modifiers for this element.
*/
int getModifiers();
@ -161,17 +144,4 @@ public interface ModifiersProvider {
default boolean isConcrete() {
return !isAbstract();
}
static int jvmClassModifiers() {
// no SUPER
return PUBLIC | FINAL | INTERFACE | ABSTRACT | ANNOTATION | ENUM | SYNTHETIC;
}
static int jvmMethodModifiers() {
return PUBLIC | PRIVATE | PROTECTED | STATIC | FINAL | SYNCHRONIZED | BRIDGE | VARARGS | NATIVE | ABSTRACT | STRICT | SYNTHETIC;
}
static int jvmFieldModifiers() {
return PUBLIC | PRIVATE | PROTECTED | STATIC | FINAL | VOLATILE | TRANSIENT | ENUM | SYNTHETIC;
}
}

33
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.meta/src/jdk/vm/ci/meta/ResolvedJavaMethod.java
View File

@ -26,7 +26,6 @@ import java.lang.annotation.Annotation;
import java.lang.reflect.AnnotatedElement;
import java.lang.reflect.Array;
import java.lang.reflect.Method;
import java.lang.reflect.Modifier;
import java.lang.reflect.Type;
/**
@ -72,14 +71,6 @@ public interface ResolvedJavaMethod extends JavaMethod, InvokeTarget, ModifiersP
*/
int getMaxStackSize();
/**
* {@inheritDoc}
* <p>
* Only the {@linkplain Modifier#methodModifiers() method flags} specified in the JVM
* specification will be included in the returned mask.
*/
int getModifiers();
default boolean isFinal() {
return ModifiersProvider.super.isFinalFlagSet();
}
@ -88,9 +79,7 @@ public interface ResolvedJavaMethod extends JavaMethod, InvokeTarget, ModifiersP
* Determines if this method is a synthetic method as defined by the Java Language
* Specification.
*/
default boolean isSynthetic() {
return (SYNTHETIC & getModifiers()) == SYNTHETIC;
}
boolean isSynthetic();
/**
* Checks that the method is a
@ -99,9 +88,7 @@ public interface ResolvedJavaMethod extends JavaMethod, InvokeTarget, ModifiersP
*
* @return whether the method is a varargs method
*/
default boolean isVarArgs() {
return (VARARGS & getModifiers()) == VARARGS;
}
boolean isVarArgs();
/**
* Checks that the method is a
@ -110,9 +97,7 @@ public interface ResolvedJavaMethod extends JavaMethod, InvokeTarget, ModifiersP
*
* @return whether the method is a bridge method
*/
default boolean isBridge() {
return (BRIDGE & getModifiers()) == BRIDGE;
}
boolean isBridge();
/**
* Returns {@code true} if this method is a default method; returns {@code false} otherwise.
@ -227,18 +212,6 @@ public interface ResolvedJavaMethod extends JavaMethod, InvokeTarget, ModifiersP
*/
LocalVariableTable getLocalVariableTable();
/**
* Invokes the underlying method represented by this object, on the specified object with the
* specified parameters. This method is similar to a reflective method invocation by
* {@link Method#invoke}.
*
* @param receiver The receiver for the invocation, or {@code null} if it is a static method.
* @param arguments The arguments for the invocation.
* @return The value returned by the method invocation, or {@code null} if the return type is
* {@code void}.
*/
JavaConstant invoke(JavaConstant receiver, JavaConstant[] arguments);
/**
* Gets the encoding of (that is, a constant representing the value of) this method.
*

9
hotspot/src/jdk.vm.ci/share/classes/jdk.vm.ci.meta/src/jdk/vm/ci/meta/ResolvedJavaType.java
View File

@ -68,15 +68,6 @@ public interface ResolvedJavaType extends JavaType, ModifiersProvider, Annotated
*/
boolean isPrimitive();
/**
* {@inheritDoc}
* <p>
* Only the flags specified in the JVM specification will be included in the returned mask. This
* method is identical to {@link Class#getModifiers()} in terms of the value return for this
* type.
*/
int getModifiers();
/*
* The setting of the final bit for types is a bit confusing since arrays are marked as final.
* This method provides a semantically equivalent test that appropriate for types.

6
hotspot/src/os/windows/vm/os_windows.cpp
View File

@ -2366,7 +2366,9 @@ bool os::win32::get_frame_at_stack_banging_point(JavaThread* thread,
if (Interpreter::contains(pc)) {
*fr = os::fetch_frame_from_context((void*)exceptionInfo->ContextRecord);
if (!fr->is_first_java_frame()) {
assert(fr->safe_for_sender(thread), "Safety check");
// get_frame_at_stack_banging_point() is only called when we
// have well defined stacks so java_sender() calls do not need
// to assert safe_for_sender() first.
*fr = fr->java_sender();
}
} else {
@ -2383,7 +2385,7 @@ bool os::win32::get_frame_at_stack_banging_point(JavaThread* thread,
// has been pushed on the stack
*fr = frame(fr->sp() + 1, fr->fp(), (address)*(fr->sp()));
if (!fr->is_java_frame()) {
assert(fr->safe_for_sender(thread), "Safety check");
// See java_sender() comment above.
*fr = fr->java_sender();
}
}

6
hotspot/src/os_cpu/bsd_x86/vm/os_bsd_x86.cpp
View File

@ -378,7 +378,9 @@ bool os::Bsd::get_frame_at_stack_banging_point(JavaThread* thread, ucontext_t* u
// method returns the Java sender of the current frame.
*fr = os::fetch_frame_from_ucontext(thread, uc);
if (!fr->is_first_java_frame()) {
assert(fr->safe_for_sender(thread), "Safety check");
// get_frame_at_stack_banging_point() is only called when we
// have well defined stacks so java_sender() calls do not need
// to assert safe_for_sender() first.
*fr = fr->java_sender();
}
} else {
@ -395,7 +397,7 @@ bool os::Bsd::get_frame_at_stack_banging_point(JavaThread* thread, ucontext_t* u
// has been pushed on the stack
*fr = frame(fr->sp() + 1, fr->fp(), (address)*(fr->sp()));
if (!fr->is_java_frame()) {
assert(fr->safe_for_sender(thread), "Safety check");
// See java_sender() comment above.
*fr = fr->java_sender();
}
}

6
hotspot/src/os_cpu/linux_x86/vm/os_linux_x86.cpp
View File

@ -191,7 +191,9 @@ bool os::Linux::get_frame_at_stack_banging_point(JavaThread* thread, ucontext_t*
// method returns the Java sender of the current frame.
*fr = os::fetch_frame_from_ucontext(thread, uc);
if (!fr->is_first_java_frame()) {
assert(fr->safe_for_sender(thread), "Safety check");
// get_frame_at_stack_banging_point() is only called when we
// have well defined stacks so java_sender() calls do not need
// to assert safe_for_sender() first.
*fr = fr->java_sender();
}
} else {
@ -209,8 +211,8 @@ bool os::Linux::get_frame_at_stack_banging_point(JavaThread* thread, ucontext_t*
intptr_t* sp = os::Linux::ucontext_get_sp(uc);
*fr = frame(sp + 1, fp, (address)*sp);
if (!fr->is_java_frame()) {
assert(fr->safe_for_sender(thread), "Safety check");
assert(!fr->is_first_frame(), "Safety check");
// See java_sender() comment above.
*fr = fr->java_sender();
}
}

6
hotspot/src/os_cpu/solaris_x86/vm/os_solaris_x86.cpp
View File

@ -255,7 +255,9 @@ bool os::Solaris::get_frame_at_stack_banging_point(JavaThread* thread, ucontext_
// method returns the Java sender of the current frame.
*fr = os::fetch_frame_from_ucontext(thread, uc);
if (!fr->is_first_java_frame()) {
assert(fr->safe_for_sender(thread), "Safety check");
// get_frame_at_stack_banging_point() is only called when we
// have well defined stacks so java_sender() calls do not need
// to assert safe_for_sender() first.
*fr = fr->java_sender();
}
} else {
@ -273,7 +275,7 @@ bool os::Solaris::get_frame_at_stack_banging_point(JavaThread* thread, ucontext_
intptr_t* sp = os::Solaris::ucontext_get_sp(uc);
*fr = frame(sp + 1, fp, (address)*sp);
if (!fr->is_java_frame()) {
assert(fr->safe_for_sender(thread), "Safety check");
// See java_sender() comment above.
*fr = fr->java_sender();
}
}

3
hotspot/src/share/vm/adlc/formssel.cpp
View File

@ -1245,6 +1245,7 @@ bool InstructForm::check_branch_variant(ArchDesc &AD, InstructForm *short_branch
this != short_branch && // Don't match myself
!is_short_branch() && // Don't match another short branch variant
reduce_result() != NULL &&
strstr(_ident, "restoreMask") == NULL && // Don't match side effects
strcmp(reduce_result(), short_branch->reduce_result()) == 0 &&
_matrule->equivalent(AD.globalNames(), short_branch->_matrule)) {
// The instructions are equivalent.
@ -4037,6 +4038,8 @@ int MatchRule::is_expensive() const {
strcmp(opType,"EncodeP")==0 ||
strcmp(opType,"EncodePKlass")==0 ||
strcmp(opType,"DecodeNKlass")==0 ||
strcmp(opType,"FmaD") == 0 ||
strcmp(opType,"FmaF") == 0 ||
strcmp(opType,"RoundDouble")==0 ||
strcmp(opType,"RoundFloat")==0 ||
strcmp(opType,"ReverseBytesI")==0 ||

2
hotspot/src/share/vm/c1/c1_Compiler.cpp
View File

@ -162,6 +162,8 @@ bool Compiler::is_intrinsic_supported(const methodHandle& method) {
case vmIntrinsics::_dlog10:
case vmIntrinsics::_dexp:
case vmIntrinsics::_dpow:
case vmIntrinsics::_fmaD:
case vmIntrinsics::_fmaF:
case vmIntrinsics::_getObject:
case vmIntrinsics::_getBoolean:
case vmIntrinsics::_getByte:

6
hotspot/src/share/vm/c1/c1_LIR.cpp
View File

@ -666,7 +666,9 @@ void LIR_OpVisitState::visit(LIR_Op* op) {
// LIR_Op3
case lir_idiv:
case lir_irem: {
case lir_irem:
case lir_fmad:
case lir_fmaf: {
assert(op->as_Op3() != NULL, "must be");
LIR_Op3* op3= (LIR_Op3*)op;
@ -1663,6 +1665,8 @@ const char * LIR_Op::name() const {
// LIR_Op3
case lir_idiv: s = "idiv"; break;
case lir_irem: s = "irem"; break;
case lir_fmad: s = "fmad"; break;
case lir_fmaf: s = "fmaf"; break;
// LIR_OpJavaCall
case lir_static_call: s = "static"; break;
case lir_optvirtual_call: s = "optvirtual"; break;

4
hotspot/src/share/vm/c1/c1_LIR.hpp
View File

@ -956,6 +956,8 @@ enum LIR_Code {
, begin_op3
, lir_idiv
, lir_irem
, lir_fmad
, lir_fmaf
, end_op3
, begin_opJavaCall
, lir_static_call
@ -2149,6 +2151,8 @@ class LIR_List: public CompilationResourceObj {
void abs (LIR_Opr from, LIR_Opr to, LIR_Opr tmp) { append(new LIR_Op2(lir_abs , from, tmp, to)); }
void sqrt(LIR_Opr from, LIR_Opr to, LIR_Opr tmp) { append(new LIR_Op2(lir_sqrt, from, tmp, to)); }
void fmad(LIR_Opr from, LIR_Opr from1, LIR_Opr from2, LIR_Opr to) { append(new LIR_Op3(lir_fmad, from, from1, from2, to)); }
void fmaf(LIR_Opr from, LIR_Opr from1, LIR_Opr from2, LIR_Opr to) { append(new LIR_Op3(lir_fmaf, from, from1, from2, to)); }
void log10 (LIR_Opr from, LIR_Opr to, LIR_Opr tmp) { append(new LIR_Op2(lir_log10, from, LIR_OprFact::illegalOpr, to, tmp)); }
void tan (LIR_Opr from, LIR_Opr to, LIR_Opr tmp1, LIR_Opr tmp2) { append(new LIR_Op2(lir_tan , from, tmp1, to, tmp2)); }

7
hotspot/src/share/vm/c1/c1_LIRGenerator.cpp
View File

@ -3181,6 +3181,9 @@ void LIRGenerator::do_Intrinsic(Intrinsic* x) {
case vmIntrinsics::_dpow : do_MathIntrinsic(x); break;
case vmIntrinsics::_arraycopy: do_ArrayCopy(x); break;
case vmIntrinsics::_fmaD: do_FmaIntrinsic(x); break;
case vmIntrinsics::_fmaF: do_FmaIntrinsic(x); break;
// java.nio.Buffer.checkIndex
case vmIntrinsics::_checkIndex: do_NIOCheckIndex(x); break;
@ -3244,14 +3247,14 @@ void LIRGenerator::profile_arguments(ProfileCall* x) {
Bytecodes::Code bc = x->method()->java_code_at_bci(bci);
int start = 0;
int stop = data->is_CallTypeData() ? ((ciCallTypeData*)data)->number_of_arguments() : ((ciVirtualCallTypeData*)data)->number_of_arguments();
if (x->inlined() && x->callee()->is_static() && Bytecodes::has_receiver(bc)) {
if (x->callee()->is_loaded() && x->callee()->is_static() && Bytecodes::has_receiver(bc)) {
// first argument is not profiled at call (method handle invoke)
assert(x->method()->raw_code_at_bci(bci) == Bytecodes::_invokehandle, "invokehandle expected");
start = 1;
}
ciSignature* callee_signature = x->callee()->signature();
// method handle call to virtual method
bool has_receiver = x->inlined() && !x->callee()->is_static() && !Bytecodes::has_receiver(bc);
bool has_receiver = x->callee()->is_loaded() && !x->callee()->is_static() && !Bytecodes::has_receiver(bc);
ciSignatureStream callee_signature_stream(callee_signature, has_receiver ? x->callee()->holder() : NULL);
bool ignored_will_link;

1
hotspot/src/share/vm/c1/c1_LIRGenerator.hpp
View File

@ -245,6 +245,7 @@ class LIRGenerator: public InstructionVisitor, public BlockClosure {
void do_isPrimitive(Intrinsic* x);
void do_getClass(Intrinsic* x);
void do_currentThread(Intrinsic* x);
void do_FmaIntrinsic(Intrinsic* x);
void do_MathIntrinsic(Intrinsic* x);
void do_LibmIntrinsic(Intrinsic* x);
void do_ArrayCopy(Intrinsic* x);

143
hotspot/src/share/vm/classfile/classLoader.cpp
View File

@ -85,6 +85,7 @@ typedef jboolean (JNICALL *ReadMappedEntry_t)(jzfile *zip, jzentry *entry, unsig
typedef jzentry* (JNICALL *GetNextEntry_t)(jzfile *zip, jint n);
typedef jboolean (JNICALL *ZipInflateFully_t)(void *inBuf, jlong inLen, void *outBuf, jlong outLen, char **pmsg);
typedef jint (JNICALL *Crc32_t)(jint crc, const jbyte *buf, jint len);
typedef void (JNICALL *FreeEntry_t)(jzfile *zip, jzentry *entry);
static ZipOpen_t ZipOpen = NULL;
static ZipClose_t ZipClose = NULL;
@ -95,6 +96,7 @@ static GetNextEntry_t GetNextEntry = NULL;
static canonicalize_fn_t CanonicalizeEntry = NULL;
static ZipInflateFully_t ZipInflateFully = NULL;
static Crc32_t Crc32 = NULL;
static FreeEntry_t FreeEntry = NULL;
// Entry points for jimage.dll for loading jimage file entries
@ -150,6 +152,7 @@ int ClassLoader::_num_entries = 0;
GrowableArray<char*>* ClassLoader::_boot_modules_array = NULL;
GrowableArray<char*>* ClassLoader::_platform_modules_array = NULL;
SharedPathsMiscInfo* ClassLoader::_shared_paths_misc_info = NULL;
int ClassLoader::_num_patch_mod_prefixes = 0;
#endif
// helper routines
@ -319,6 +322,20 @@ ClassPathZipEntry::~ClassPathZipEntry() {
FREE_C_HEAP_ARRAY(char, _zip_name);
}
bool ClassPathZipEntry::stream_exists(const char* name) {
// enable call to C land
JavaThread* thread = JavaThread::current();
ThreadToNativeFromVM ttn(thread);
// check whether zip archive contains name
jint name_len, filesize;
jzentry* entry = (*FindEntry)(_zip, name, &filesize, &name_len);
if (entry != NULL) {
(*FreeEntry)(_zip, entry);
return true;
}
return false;
}
u1* ClassPathZipEntry::open_entry(const char* name, jint* filesize, bool nul_terminate, TRAPS) {
// enable call to C land
JavaThread* thread = JavaThread::current();
@ -640,7 +657,7 @@ void ClassLoader::check_shared_classpath(const char *path) {
struct stat st;
if (os::stat(path, &st) == 0) {
if ((st.st_mode & S_IFREG) != S_IFREG) { // is directory
if ((st.st_mode & S_IFMT) != S_IFREG) { // is not a regular file
if (!os::dir_is_empty(path)) {
tty->print_cr("Error: non-empty directory '%s'", path);
exit_with_path_failure("CDS allows only empty directories in archived classpaths", NULL);
@ -693,8 +710,6 @@ void ClassLoader::setup_patch_mod_entries() {
GrowableArray<ModulePatchPath*>* patch_mod_args = Arguments::get_patch_mod_prefix();
int num_of_entries = patch_mod_args->length();
assert(!DumpSharedSpaces, "DumpSharedSpaces not supported with --patch-module");
assert(!UseSharedSpaces, "UseSharedSpaces not supported with --patch-module");
// Set up the boot loader's _patch_mod_entries list
_patch_mod_entries = new (ResourceObj::C_HEAP, mtModule) GrowableArray<ModuleClassPathList*>(num_of_entries, true);
@ -851,7 +866,7 @@ ClassPathEntry* ClassLoader::create_class_path_entry(const char *path, const str
bool is_boot_append, TRAPS) {
JavaThread* thread = JavaThread::current();
ClassPathEntry* new_entry = NULL;
if ((st->st_mode & S_IFREG) == S_IFREG) {
if ((st->st_mode & S_IFMT) == S_IFREG) {
ResourceMark rm(thread);
// Regular file, should be a zip or jimage file
// Canonicalized filename
@ -914,7 +929,7 @@ ClassPathZipEntry* ClassLoader::create_class_path_zip_entry(const char *path, bo
// check for a regular file
struct stat st;
if (os::stat(path, &st) == 0) {
if ((st.st_mode & S_IFREG) == S_IFREG) {
if ((st.st_mode & S_IFMT) == S_IFREG) {
char canonical_path[JVM_MAXPATHLEN];
if (get_canonical_path(path, canonical_path, JVM_MAXPATHLEN)) {
char* error_msg = NULL;
@ -1068,6 +1083,7 @@ void ClassLoader::load_zip_library() {
GetNextEntry = CAST_TO_FN_PTR(GetNextEntry_t, os::dll_lookup(handle, "ZIP_GetNextEntry"));
ZipInflateFully = CAST_TO_FN_PTR(ZipInflateFully_t, os::dll_lookup(handle, "ZIP_InflateFully"));
Crc32 = CAST_TO_FN_PTR(Crc32_t, os::dll_lookup(handle, "ZIP_CRC32"));
FreeEntry = CAST_TO_FN_PTR(FreeEntry_t, os::dll_lookup(handle, "ZIP_FreeEntry"));
// ZIP_Close is not exported on Windows in JDK5.0 so don't abort if ZIP_Close is NULL
if (ZipOpen == NULL || FindEntry == NULL || ReadEntry == NULL ||
@ -1395,6 +1411,57 @@ ClassFileStream* ClassLoader::search_module_entries(const GrowableArray<ModuleCl
return NULL;
}
#if INCLUDE_CDS
// The following function is only used during CDS dump time.
// It checks if a class can be found in the jar entries of the _patch_mod_entries.
// It does not support non-jar entries.
bool ClassLoader::is_in_patch_module(const char* const file_name) {
assert(DumpSharedSpaces, "dump time only");
if (_patch_mod_entries == NULL) {
return false;
}
int num_of_entries = _patch_mod_entries->length();
char* class_module_name = NULL;
ResourceMark rm;
const char *pkg_name = package_from_name(file_name);
// Using the jimage to obtain the class' module name.
// The ModuleEntryTable cannot be used at this point during dump time
// because the module system hasn't been initialized yet.
if (pkg_name != NULL) {
JImageFile *jimage = _jrt_entry->jimage();
class_module_name = (char*)(*JImagePackageToModule)(jimage, pkg_name);
}
if (class_module_name == NULL) {
return false;
}
// Loop through all the patch module entries looking for module
for (int i = 0; i < num_of_entries; i++) {
ModuleClassPathList* module_cpl = _patch_mod_entries->at(i);
Symbol* module_cpl_name = module_cpl->module_name();
if (strcmp(module_cpl_name->as_C_string(), class_module_name) == 0) {
// Class' module has been located, attempt to locate
// the class from the module's ClassPathEntry list.
ClassPathEntry* e = module_cpl->module_first_entry();
while (e != NULL) {
if (e->is_jar_file()) {
if (e->stream_exists(file_name)) {
return true;
} else {
e = e->next();
}
}
}
}
}
return false;
}
#endif // INCLUDE_CDS
instanceKlassHandle ClassLoader::load_class(Symbol* name, bool search_append_only, TRAPS) {
assert(name != NULL, "invariant");
assert(THREAD->is_Java_thread(), "must be a JavaThread");
@ -1420,8 +1487,8 @@ instanceKlassHandle ClassLoader::load_class(Symbol* name, bool search_append_onl
// If DumpSharedSpaces is true boot loader visibility boundaries are set to:
// - [jimage] + [_first_append_entry to _last_append_entry] (all path entries).
// No --patch-module entries or exploded module builds are included since CDS
// is not supported if --patch-module or exploded module builds are used.
// If a class is found in the --patch-module entries, the class will not be included in the
// CDS archive. Also, CDS is not supported if exploded module builds are used.
//
// If search_append_only is true, boot loader visibility boundaries are
// set to be _first_append_entry to the end. This includes:
@ -1444,8 +1511,17 @@ instanceKlassHandle ClassLoader::load_class(Symbol* name, bool search_append_onl
// found within its module specification, the search should continue to Load Attempt #2.
// Note: The --patch-module entries are never searched if the boot loader's
// visibility boundary is limited to only searching the append entries.
if (_patch_mod_entries != NULL && !search_append_only && !DumpSharedSpaces) {
stream = search_module_entries(_patch_mod_entries, class_name, file_name, CHECK_NULL);
if (_patch_mod_entries != NULL && !search_append_only) {
if (!DumpSharedSpaces) {
stream = search_module_entries(_patch_mod_entries, class_name, file_name, CHECK_NULL);
} else {
#if INCLUDE_CDS
if (is_in_patch_module(file_name)) {
tty->print_cr("Preload Warning: Skip archiving class %s found in --patch-module entry", class_name);
return NULL;
}
#endif
}
}
// Load Attempt #2: [jimage | exploded build]
@ -1596,8 +1672,57 @@ void ClassLoader::initialize() {
}
#if INCLUDE_CDS
// Capture all the --patch-module entries specified during CDS dump time.
// It also captures the non-existing path(s) and the required file(s) during inspecting
// the entries.
void ClassLoader::setup_patch_mod_path() {
assert(DumpSharedSpaces, "only used with -Xshare:dump");
ResourceMark rm;
GrowableArray<ModulePatchPath*>* patch_mod_args = Arguments::get_patch_mod_prefix();
if (patch_mod_args != NULL) {
int num_of_entries = patch_mod_args->length();
for (int i = 0; i < num_of_entries; i++) {
const char* module_name = (patch_mod_args->at(i))->module_name();
const char* module_path = (patch_mod_args->at(i))->path_string();
int path_len = (int)strlen(module_path);
int name_len = (int)strlen(module_name);
int buf_len = name_len + path_len + 2; // add 2 for the '=' and NULL terminator
int end = 0;
char* buf = NEW_C_HEAP_ARRAY(char, buf_len, mtInternal);
// Iterate over the module's class path entries
for (int start = 0; start < path_len; start = end) {
while (module_path[end] && module_path[end] != os::path_separator()[0]) {
end++;
}
strncpy(buf, &module_path[start], end - start);
buf[end - start] = '\0';
struct stat st;
if (os::stat(buf, &st) != 0) {
// File not found
_shared_paths_misc_info->add_nonexist_path(buf);
} else {
if ((st.st_mode & S_IFMT) != S_IFREG) { // is not a regular file
vm_exit_during_initialization(
"--patch-module requires a regular file during dumping", buf);
} else {
_shared_paths_misc_info->add_required_file(buf);
}
}
while (module_path[end] == os::path_separator()[0]) {
end++;
}
};
jio_snprintf(buf, buf_len, "%s=%s", module_name, module_path);
_shared_paths_misc_info->add_patch_mod_classpath((const char*)buf);
_num_patch_mod_prefixes++;
FREE_C_HEAP_ARRAY(char, buf);
}
}
}
void ClassLoader::initialize_shared_path() {
if (DumpSharedSpaces) {
setup_patch_mod_path();
ClassLoaderExt::setup_search_paths();
_shared_paths_misc_info->write_jint(0); // see comments in SharedPathsMiscInfo::check()
}

8
hotspot/src/share/vm/classfile/classLoader.hpp
View File

@ -69,6 +69,7 @@ public:
// Attempt to locate file_name through this class path entry.
// Returns a class file parsing stream if successfull.
virtual ClassFileStream* open_stream(const char* name, TRAPS) = 0;
virtual bool stream_exists(const char* name) = 0;
// Debugging
NOT_PRODUCT(virtual void compile_the_world(Handle loader, TRAPS) = 0;)
};
@ -83,6 +84,7 @@ class ClassPathDirEntry: public ClassPathEntry {
JImageFile* jimage() const { return NULL; }
ClassPathDirEntry(const char* dir);
ClassFileStream* open_stream(const char* name, TRAPS);
bool stream_exists(const char* name) { return false; }
// Debugging
NOT_PRODUCT(void compile_the_world(Handle loader, TRAPS);)
};
@ -126,6 +128,7 @@ class ClassPathZipEntry: public ClassPathEntry {
ClassFileStream* open_stream(const char* name, TRAPS);
void contents_do(void f(const char* name, void* context), void* context);
bool is_multiple_versioned(TRAPS) NOT_CDS_RETURN_(false);
bool stream_exists(const char* name);
// Debugging
NOT_PRODUCT(void compile_the_world(Handle loader, TRAPS);)
};
@ -145,6 +148,7 @@ public:
ClassPathImageEntry(JImageFile* jimage, const char* name);
~ClassPathImageEntry();
ClassFileStream* open_stream(const char* name, TRAPS);
bool stream_exists(const char* name) { return false; }
// Debugging
NOT_PRODUCT(void compile_the_world(Handle loader, TRAPS);)
@ -255,6 +259,7 @@ class ClassLoader: AllStatic {
// Info used by CDS
CDS_ONLY(static SharedPathsMiscInfo * _shared_paths_misc_info;)
CDS_ONLY(static int _num_patch_mod_prefixes;)
// Initialization:
// - setup the boot loader's system class path
@ -427,6 +432,9 @@ class ClassLoader: AllStatic {
static void initialize_module_loader_map(JImageFile* jimage);
static s2 classloader_type(Symbol* class_name, ClassPathEntry* e,
int classpath_index, TRAPS);
static bool is_in_patch_module(const char* const file_name);
static void setup_patch_mod_path(); // Only when -Xshare:dump
static int num_patch_mod_prefixes() { return _num_patch_mod_prefixes; }
#endif
static void trace_class_path(const char* msg, const char* name = NULL);

27
hotspot/src/share/vm/classfile/sharedPathsMiscInfo.cpp
View File

@ -86,6 +86,9 @@ void SharedPathsMiscInfo::print_path(int type, const char* path) {
case REQUIRED:
out->print("Expecting that file %s must exist and is not altered", path);
break;
case PATCH_MOD:
out->print("Expecting --patch-module=%s", path);
break;
default:
ShouldNotReachHere();
}
@ -146,6 +149,9 @@ bool SharedPathsMiscInfo::check(jint type, const char* path) {
// But we want it to not exist -> fail
return fail("File must not exist");
}
if ((st.st_mode & S_IFMT) != S_IFREG) {
return fail("Did not get a regular file as expected.");
}
time_t timestamp;
long filesize;
@ -161,7 +167,26 @@ bool SharedPathsMiscInfo::check(jint type, const char* path) {
}
}
break;
case PATCH_MOD:
{
GrowableArray<ModulePatchPath*>* patch_mod_args = Arguments::get_patch_mod_prefix();
if (patch_mod_args != NULL) {
int num_of_entries = patch_mod_args->length();
for (int i = 0; i < num_of_entries; i++) {
const char* module_name = (patch_mod_args->at(i))->module_name();
const char* path_string = (patch_mod_args->at(i))->path_string();
size_t n = strlen(module_name);
// path contains the module name, followed by '=', and one or more entries.
// E.g.: "java.base=foo" or "java.naming=dir1:dir2:dir3"
if ((strncmp(module_name, path, n) != 0) ||
(path[n] != '=') ||
(strcmp(path + n + 1, path_string) != 0)) {
return fail("--patch-module mismatch, path not found in run time: ", path);
}
}
}
}
break;
default:
return fail("Corrupted archive file header");
}

22
hotspot/src/share/vm/classfile/sharedPathsMiscInfo.hpp
View File

@ -104,10 +104,28 @@ public:
add_path(path, NON_EXIST);
}
// The path must exist and have required size and modification time
void add_required_file(const char* path) {
add_path(path, REQUIRED);
struct stat st;
if (os::stat(path, &st) != 0) {
assert(0, "sanity");
#if INCLUDE_CDS
ClassLoader::exit_with_path_failure("failed to os::stat(%s)", path); // should not happen
#endif
}
write_time(st.st_mtime);
write_long(st.st_size);
}
// The path must exist, and must contain exactly <num_entries> files/dirs
void add_boot_classpath(const char* path) {
add_path(path, BOOT);
}
void add_patch_mod_classpath(const char* path) {
add_path(path, PATCH_MOD);
}
int write_jint(jint num) {
write(&num, sizeof(num));
return 0;
@ -129,7 +147,8 @@ public:
enum {
BOOT = 1,
NON_EXIST = 2,
REQUIRED = 3
REQUIRED = 3,
PATCH_MOD = 4
};
virtual const char* type_name(int type) {
@ -137,6 +156,7 @@ public:
case BOOT: return "BOOT";
case NON_EXIST: return "NON_EXIST";
case REQUIRED: return "REQUIRED";
case PATCH_MOD: return "PATCH_MOD";
default: ShouldNotReachHere(); return "?";
}
}

50
hotspot/src/share/vm/classfile/symbolTable.cpp
View File

@ -710,53 +710,3 @@ int SymboltableDCmd::num_arguments() {
return 0;
}
}
#ifndef PRODUCT
// Internal test of TempNewSymbol
void Test_TempNewSymbol() {
// Assert messages assume these symbols are unique, and the refcounts start at
// one, but code does not rely on this.
Thread* THREAD = Thread::current();
Symbol* abc = SymbolTable::new_symbol("abc", CATCH);
int abccount = abc->refcount();
TempNewSymbol ss = abc;
assert(ss->refcount() == abccount, "only one abc");
assert(ss->refcount() == abc->refcount(), "should match TempNewSymbol");
Symbol* efg = SymbolTable::new_symbol("efg", CATCH);
Symbol* hij = SymbolTable::new_symbol("hij", CATCH);
int efgcount = efg->refcount();
int hijcount = hij->refcount();
TempNewSymbol s1 = efg;
TempNewSymbol s2 = hij;
assert(s1->refcount() == efgcount, "one efg");
assert(s2->refcount() == hijcount, "one hij");
// Assignment operator
s1 = s2;
assert(hij->refcount() == hijcount + 1, "should be two hij");
assert(efg->refcount() == efgcount - 1, "should be no efg");
s1 = ss; // s1 is abc
assert(s1->refcount() == abccount + 1, "should be two abc (s1 and ss)");
assert(hij->refcount() == hijcount, "should only have one hij now (s2)");
s1 = s1; // self assignment
assert(s1->refcount() == abccount + 1, "should still be two abc (s1 and ss)");
TempNewSymbol s3;
Symbol* klm = SymbolTable::new_symbol("klm", CATCH);
int klmcount = klm->refcount();
s3 = klm; // assignment
assert(s3->refcount() == klmcount, "only one klm now");
Symbol* xyz = SymbolTable::new_symbol("xyz", CATCH);
int xyzcount = xyz->refcount();
{ // inner scope
TempNewSymbol s_inner = xyz;
}
assert(xyz->refcount() == (xyzcount - 1),
"Should have been decremented by dtor in inner scope");
}
#endif // PRODUCT

8
hotspot/src/share/vm/classfile/vmSymbols.cpp
View File

@ -355,6 +355,8 @@ bool vmIntrinsics::preserves_state(vmIntrinsics::ID id) {
case vmIntrinsics::_updateBytesCRC32:
case vmIntrinsics::_updateByteBufferCRC32:
case vmIntrinsics::_vectorizedMismatch:
case vmIntrinsics::_fmaD:
case vmIntrinsics::_fmaF:
return true;
default:
return false;
@ -388,6 +390,8 @@ bool vmIntrinsics::can_trap(vmIntrinsics::ID id) {
case vmIntrinsics::_updateBytesCRC32:
case vmIntrinsics::_updateByteBufferCRC32:
case vmIntrinsics::_vectorizedMismatch:
case vmIntrinsics::_fmaD:
case vmIntrinsics::_fmaF:
return false;
default:
return true;
@ -536,6 +540,10 @@ bool vmIntrinsics::is_disabled_by_flags(vmIntrinsics::ID id) {
case vmIntrinsics::_doubleToLongBits:
if (!InlineMathNatives) return true;
break;
case vmIntrinsics::_fmaD:
case vmIntrinsics::_fmaF:
if (!InlineMathNatives || !UseFMA) return true;
break;
case vmIntrinsics::_arraycopy:
if (!InlineArrayCopy) return true;
break;

7
hotspot/src/share/vm/classfile/vmSymbols.hpp
View File

@ -755,8 +755,10 @@
do_class(java_lang_Math, "java/lang/Math") \
do_class(java_lang_StrictMath, "java/lang/StrictMath") \
do_signature(double2_double_signature, "(DD)D") \
do_signature(double3_double_signature, "(DDD)D") \
do_signature(float3_float_signature, "(FFF)F") \
do_signature(int2_int_signature, "(II)I") \
do_signature(long2_long_signature, "(JJ)J") \
do_signature(long2_long_signature, "(JJ)J") \
\
/* here are the math names, all together: */ \
do_name(abs_name,"abs") do_name(sin_name,"sin") do_name(cos_name,"cos") \
@ -770,6 +772,7 @@
do_name(multiplyExact_name,"multiplyExact") \
do_name(negateExact_name,"negateExact") \
do_name(subtractExact_name,"subtractExact") \
do_name(fma_name, "fma") \
\
do_intrinsic(_dabs, java_lang_Math, abs_name, double_double_signature, F_S) \
do_intrinsic(_dsin, java_lang_Math, sin_name, double_double_signature, F_S) \
@ -795,6 +798,8 @@
do_intrinsic(_negateExactL, java_lang_Math, negateExact_name, long_long_signature, F_S) \
do_intrinsic(_subtractExactI, java_lang_Math, subtractExact_name, int2_int_signature, F_S) \
do_intrinsic(_subtractExactL, java_lang_Math, subtractExact_name, long2_long_signature, F_S) \
do_intrinsic(_fmaD, java_lang_Math, fma_name, double3_double_signature, F_S) \
do_intrinsic(_fmaF, java_lang_Math, fma_name, float3_float_signature, F_S) \
\
do_intrinsic(_floatToRawIntBits, java_lang_Float, floatToRawIntBits_name, float_int_signature, F_S) \
do_name( floatToRawIntBits_name, "floatToRawIntBits") \

21
hotspot/src/share/vm/gc/cms/cmsOopClosures.hpp
View File

@ -258,16 +258,15 @@ class PushOrMarkClosure: public MetadataAwareOopClosure {
// the closure ParMarkFromRootsClosure.
class ParPushOrMarkClosure: public MetadataAwareOopClosure {
private:
CMSCollector* _collector;
MemRegion _whole_span;
MemRegion _span; // local chunk
CMSBitMap* _bit_map;
OopTaskQueue* _work_queue;
CMSMarkStack* _overflow_stack;
HeapWord* const _finger;
HeapWord** const _global_finger_addr;
ParMarkFromRootsClosure* const
_parent;
CMSCollector* _collector;
MemRegion _whole_span;
MemRegion _span; // local chunk
CMSBitMap* _bit_map;
OopTaskQueue* _work_queue;
CMSMarkStack* _overflow_stack;
HeapWord* const _finger;
HeapWord* volatile* const _global_finger_addr;
ParMarkFromRootsClosure* const _parent;
protected:
DO_OOP_WORK_DEFN
public:
@ -277,7 +276,7 @@ class ParPushOrMarkClosure: public MetadataAwareOopClosure {
OopTaskQueue* work_queue,
CMSMarkStack* mark_stack,
HeapWord* finger,
HeapWord** global_finger_addr,
HeapWord* volatile* global_finger_addr,
ParMarkFromRootsClosure* parent);
virtual void do_oop(oop* p);
virtual void do_oop(narrowOop* p);

22
hotspot/src/share/vm/gc/cms/concurrentMarkSweepGeneration.cpp
View File

@ -3025,14 +3025,14 @@ class CMSConcMarkingTerminatorTerminator: public TerminatorTerminator {
// MT Concurrent Marking Task
class CMSConcMarkingTask: public YieldingFlexibleGangTask {
CMSCollector* _collector;
uint _n_workers; // requested/desired # workers
bool _result;
CompactibleFreeListSpace* _cms_space;
char _pad_front[64]; // padding to ...
HeapWord* _global_finger; // ... avoid sharing cache line
char _pad_back[64];
HeapWord* _restart_addr;
CMSCollector* _collector;
uint _n_workers; // requested/desired # workers
bool _result;
CompactibleFreeListSpace* _cms_space;
char _pad_front[64]; // padding to ...
HeapWord* volatile _global_finger; // ... avoid sharing cache line
char _pad_back[64];
HeapWord* _restart_addr;
// Exposed here for yielding support
Mutex* const _bit_map_lock;
@ -3068,7 +3068,7 @@ class CMSConcMarkingTask: public YieldingFlexibleGangTask {
OopTaskQueue* work_queue(int i) { return task_queues()->queue(i); }
HeapWord** global_finger_addr() { return &_global_finger; }
HeapWord* volatile* global_finger_addr() { return &_global_finger; }
CMSConcMarkingTerminator* terminator() { return &_term; }
@ -6554,7 +6554,7 @@ void ParMarkFromRootsClosure::scan_oops_in_oop(HeapWord* ptr) {
// Note: the local finger doesn't advance while we drain
// the stack below, but the global finger sure can and will.
HeapWord** gfa = _task->global_finger_addr();
HeapWord* volatile* gfa = _task->global_finger_addr();
ParPushOrMarkClosure pushOrMarkClosure(_collector,
_span, _bit_map,
_work_queue,
@ -6721,7 +6721,7 @@ ParPushOrMarkClosure::ParPushOrMarkClosure(CMSCollector* collector,
OopTaskQueue* work_queue,
CMSMarkStack* overflow_stack,
HeapWord* finger,
HeapWord** global_finger_addr,
HeapWord* volatile* global_finger_addr,
ParMarkFromRootsClosure* parent) :
MetadataAwareOopClosure(collector->ref_processor()),
_collector(collector),

14
hotspot/src/share/vm/gc/cms/concurrentMarkSweepGeneration.hpp
View File

@ -540,7 +540,7 @@ class CMSCollector: public CHeapObj<mtGC> {
// Overflow list of grey objects, threaded through mark-word
// Manipulated with CAS in the parallel/multi-threaded case.
oop _overflow_list;
oopDesc* volatile _overflow_list;
// The following array-pair keeps track of mark words
// displaced for accommodating overflow list above.
// This code will likely be revisited under RFE#4922830.
@ -724,12 +724,12 @@ class CMSCollector: public CHeapObj<mtGC> {
// Support for parallelizing young gen rescan in CMS remark phase
ParNewGeneration* _young_gen;
HeapWord** _top_addr; // ... Top of Eden
HeapWord** _end_addr; // ... End of Eden
Mutex* _eden_chunk_lock;
HeapWord** _eden_chunk_array; // ... Eden partitioning array
size_t _eden_chunk_index; // ... top (exclusive) of array
size_t _eden_chunk_capacity; // ... max entries in array
HeapWord* volatile* _top_addr; // ... Top of Eden
HeapWord** _end_addr; // ... End of Eden
Mutex* _eden_chunk_lock;
HeapWord** _eden_chunk_array; // ... Eden partitioning array
size_t _eden_chunk_index; // ... top (exclusive) of array
size_t _eden_chunk_capacity; // ... max entries in array
// Support for parallelizing survivor space rescan
HeapWord** _survivor_chunk_array;

4
hotspot/src/share/vm/gc/cms/parNewGeneration.hpp
View File

@ -1,5 +1,5 @@
/*
* Copyright (c) 2001, 2015, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2001, 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
@ -323,7 +323,7 @@ class ParNewGeneration: public DefNewGeneration {
// A list of from-space images of to-be-scanned objects, threaded through
// klass-pointers (klass information already copied to the forwarded
// image.) Manipulated with CAS.
oop _overflow_list;
oopDesc* volatile _overflow_list;
NOT_PRODUCT(ssize_t _num_par_pushes;)
// This closure is used by the reference processor to filter out

12
hotspot/src/share/vm/gc/g1/g1CollectedHeap.cpp
View File

@ -2474,8 +2474,16 @@ size_t G1CollectedHeap::max_capacity() const {
}
jlong G1CollectedHeap::millis_since_last_gc() {
// assert(false, "NYI");
return 0;
// See the notes in GenCollectedHeap::millis_since_last_gc()
// for more information about the implementation.
jlong ret_val = (os::javaTimeNanos() / NANOSECS_PER_MILLISEC) -
_g1_policy->collection_pause_end_millis();
if (ret_val < 0) {
log_warning(gc)("millis_since_last_gc() would return : " JLONG_FORMAT
". returning zero instead.", ret_val);
return 0;
}
return ret_val;
}
void G1CollectedHeap::prepare_for_verify() {

5
hotspot/src/share/vm/gc/g1/g1DefaultPolicy.cpp
View File

@ -66,7 +66,8 @@ G1DefaultPolicy::G1DefaultPolicy() :
_phase_times(new G1GCPhaseTimes(ParallelGCThreads)),
_tenuring_threshold(MaxTenuringThreshold),
_max_survivor_regions(0),
_survivors_age_table(true) { }
_survivors_age_table(true),
_collection_pause_end_millis(os::javaTimeNanos() / NANOSECS_PER_MILLISEC) { }
G1DefaultPolicy::~G1DefaultPolicy() {
delete _ihop_control;
@ -575,6 +576,8 @@ void G1DefaultPolicy::record_collection_pause_end(double pause_time_ms, size_t c
record_pause(young_gc_pause_kind(), end_time_sec - pause_time_ms / 1000.0, end_time_sec);
_collection_pause_end_millis = os::javaTimeNanos() / NANOSECS_PER_MILLISEC;
last_pause_included_initial_mark = collector_state()->during_initial_mark_pause();
if (last_pause_included_initial_mark) {
record_concurrent_mark_init_end(0.0);

4
hotspot/src/share/vm/gc/g1/g1DefaultPolicy.hpp
View File

@ -64,6 +64,8 @@ class G1DefaultPolicy: public G1Policy {
double _full_collection_start_sec;
jlong _collection_pause_end_millis;
uint _young_list_target_length;
uint _young_list_fixed_length;
@ -237,6 +239,8 @@ public:
double reclaimable_bytes_perc(size_t reclaimable_bytes) const;
jlong collection_pause_end_millis() { return _collection_pause_end_millis; }
private:
// Sets up marking if proper conditions are met.
void maybe_start_marking();

117
hotspot/src/share/vm/gc/g1/g1IHOPControl.cpp
View File

@ -81,47 +81,6 @@ G1StaticIHOPControl::G1StaticIHOPControl(double ihop_percent) :
_last_marking_length_s(0.0) {
}
#ifndef PRODUCT
static void test_update(G1IHOPControl* ctrl, double alloc_time, size_t alloc_amount, size_t young_size, double mark_time) {
for (int i = 0; i < 100; i++) {
ctrl->update_allocation_info(alloc_time, alloc_amount, young_size);
ctrl->update_marking_length(mark_time);
}
}
void G1StaticIHOPControl::test() {
size_t const initial_ihop = 45;
G1StaticIHOPControl ctrl(initial_ihop);
ctrl.update_target_occupancy(100);
size_t threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold == initial_ihop,
"Expected IHOP threshold of " SIZE_FORMAT " but is " SIZE_FORMAT, initial_ihop, threshold);
ctrl.update_allocation_info(100.0, 100, 100);
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold == initial_ihop,
"Expected IHOP threshold of " SIZE_FORMAT " but is " SIZE_FORMAT, initial_ihop, threshold);
ctrl.update_marking_length(1000.0);
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold == initial_ihop,
"Expected IHOP threshold of " SIZE_FORMAT " but is " SIZE_FORMAT, initial_ihop, threshold);
// Whatever we pass, the IHOP value must stay the same.
test_update(&ctrl, 2, 10, 10, 3);
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold == initial_ihop,
"Expected IHOP threshold of " SIZE_FORMAT " but is " SIZE_FORMAT, initial_ihop, threshold);
test_update(&ctrl, 12, 10, 10, 3);
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold == initial_ihop,
"Expected IHOP threshold of " SIZE_FORMAT " but is " SIZE_FORMAT, initial_ihop, threshold);
}
#endif
G1AdaptiveIHOPControl::G1AdaptiveIHOPControl(double ihop_percent,
G1Predictions const* predictor,
size_t heap_reserve_percent,
@ -224,79 +183,3 @@ void G1AdaptiveIHOPControl::send_trace_event(G1NewTracer* tracer) {
_predictor->get_new_prediction(&_marking_times_s),
have_enough_data_for_prediction());
}
#ifndef PRODUCT
void G1AdaptiveIHOPControl::test() {
size_t const initial_threshold = 45;
size_t const young_size = 10;
size_t const target_size = 100;
// The final IHOP value is always
// target_size - (young_size + alloc_amount/alloc_time * marking_time)
G1Predictions pred(0.95);
G1AdaptiveIHOPControl ctrl(initial_threshold, &pred, 0, 0);
ctrl.update_target_occupancy(target_size);
// First "load".
size_t const alloc_time1 = 2;
size_t const alloc_amount1 = 10;
size_t const marking_time1 = 2;
size_t const settled_ihop1 = target_size - (young_size + alloc_amount1/alloc_time1 * marking_time1);
size_t threshold;
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold == initial_threshold,
"Expected IHOP threshold of " SIZE_FORMAT " but is " SIZE_FORMAT, initial_threshold, threshold);
for (size_t i = 0; i < G1AdaptiveIHOPNumInitialSamples - 1; i++) {
ctrl.update_allocation_info(alloc_time1, alloc_amount1, young_size);
ctrl.update_marking_length(marking_time1);
// Not enough data yet.
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold == initial_threshold,
"Expected IHOP threshold of " SIZE_FORMAT " but is " SIZE_FORMAT, initial_threshold, threshold);
}
test_update(&ctrl, alloc_time1, alloc_amount1, young_size, marking_time1);
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold == settled_ihop1,
"Expected IHOP threshold to settle at " SIZE_FORMAT " but is " SIZE_FORMAT, settled_ihop1, threshold);
// Second "load". A bit higher allocation rate.
size_t const alloc_time2 = 2;
size_t const alloc_amount2 = 30;
size_t const marking_time2 = 2;
size_t const settled_ihop2 = target_size - (young_size + alloc_amount2/alloc_time2 * marking_time2);
test_update(&ctrl, alloc_time2, alloc_amount2, young_size, marking_time2);
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold < settled_ihop1,
"Expected IHOP threshold to settle at a value lower than " SIZE_FORMAT " but is " SIZE_FORMAT, settled_ihop1, threshold);
// Third "load". Very high (impossible) allocation rate.
size_t const alloc_time3 = 1;
size_t const alloc_amount3 = 50;
size_t const marking_time3 = 2;
size_t const settled_ihop3 = 0;
test_update(&ctrl, alloc_time3, alloc_amount3, young_size, marking_time3);
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold == settled_ihop3,
"Expected IHOP threshold to settle at " SIZE_FORMAT " but is " SIZE_FORMAT, settled_ihop3, threshold);
// And back to some arbitrary value.
test_update(&ctrl, alloc_time2, alloc_amount2, young_size, marking_time2);
threshold = ctrl.get_conc_mark_start_threshold();
assert(threshold > settled_ihop3,
"Expected IHOP threshold to settle at value larger than " SIZE_FORMAT " but is " SIZE_FORMAT, settled_ihop3, threshold);
}
void IHOP_test() {
G1StaticIHOPControl::test();
G1AdaptiveIHOPControl::test();
}
#endif

9
hotspot/src/share/vm/gc/g1/g1IHOPControl.hpp
View File

@ -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
@ -99,10 +99,6 @@ class G1StaticIHOPControl : public G1IHOPControl {
assert(marking_length_s > 0.0, "Marking length must be larger than zero but is %.3f", marking_length_s);
_last_marking_length_s = marking_length_s;
}
#ifndef PRODUCT
static void test();
#endif
};
// This algorithm tries to return a concurrent mark starting occupancy value that
@ -148,9 +144,6 @@ class G1AdaptiveIHOPControl : public G1IHOPControl {
virtual void print();
virtual void send_trace_event(G1NewTracer* tracer);
#ifndef PRODUCT
static void test();
#endif
};
#endif // SHARE_VM_GC_G1_G1IHOPCONTROL_HPP

2
hotspot/src/share/vm/gc/g1/g1Policy.hpp
View File

@ -119,6 +119,8 @@ public:
virtual void record_full_collection_start() = 0;
virtual void record_full_collection_end() = 0;
virtual jlong collection_pause_end_millis() = 0;
// Must currently be called while the world is stopped.
virtual void record_concurrent_mark_init_end(double mark_init_elapsed_time_ms) = 0;

16
hotspot/src/share/vm/gc/g1/g1RemSet.cpp
View File

@ -668,20 +668,18 @@ bool G1RemSet::refine_card(jbyte* card_ptr,
// fail arbitrarily). We tell the iteration code to perform this
// filtering when it has been determined that there has been an actual
// allocation in this region and making it safe to check the young type.
bool filter_young = true;
HeapWord* stop_point =
bool card_processed =
r->oops_on_card_seq_iterate_careful(dirtyRegion,
&filter_then_update_rs_oop_cl,
filter_young,
card_ptr);
// If stop_point is non-null, then we encountered an unallocated region
// (perhaps the unfilled portion of a TLAB.) For now, we'll dirty the
// card and re-enqueue: if we put off the card until a GC pause, then the
// unallocated portion will be filled in. Alternatively, we might try
// the full complexity of the technique used in "regular" precleaning.
if (stop_point != NULL) {
// If unable to process the card then we encountered an unparsable
// part of the heap (e.g. a partially allocated object). Redirty
// and re-enqueue: if we put off the card until a GC pause, then the
// allocation will have completed.
if (!card_processed) {
assert(!_g1->is_gc_active(), "Unparsable heap during GC");
// The card might have gotten re-dirtied and re-enqueued while we
// worked. (In fact, it's pretty likely.)
if (*card_ptr != CardTableModRefBS::dirty_card_val()) {

51
hotspot/src/share/vm/gc/g1/heapRegion.cpp
View File

@ -352,19 +352,10 @@ void HeapRegion::note_self_forwarding_removal_end(bool during_initial_mark,
_prev_marked_bytes = marked_bytes;
}
HeapWord*
HeapRegion::
oops_on_card_seq_iterate_careful(MemRegion mr,
FilterOutOfRegionClosure* cl,
bool filter_young,
jbyte* card_ptr) {
// Currently, we should only have to clean the card if filter_young
// is true and vice versa.
if (filter_young) {
assert(card_ptr != NULL, "pre-condition");
} else {
assert(card_ptr == NULL, "pre-condition");
}
bool HeapRegion::oops_on_card_seq_iterate_careful(MemRegion mr,
FilterOutOfRegionClosure* cl,
jbyte* card_ptr) {
assert(card_ptr != NULL, "pre-condition");
G1CollectedHeap* g1h = G1CollectedHeap::heap();
// If we're within a stop-world GC, then we might look at a card in a
@ -375,7 +366,9 @@ oops_on_card_seq_iterate_careful(MemRegion mr,
} else {
mr = mr.intersection(used_region());
}
if (mr.is_empty()) return NULL;
if (mr.is_empty()) {
return true;
}
// Otherwise, find the obj that extends onto mr.start().
// The intersection of the incoming mr (for the card) and the
@ -384,27 +377,21 @@ oops_on_card_seq_iterate_careful(MemRegion mr,
// G1CollectedHeap.cpp that allocates a new region sets the
// is_young tag on the region before allocating. Thus we
// safely know if this region is young.
if (is_young() && filter_young) {
return NULL;
if (is_young()) {
return true;
}
assert(!is_young(), "check value of filter_young");
// We can only clean the card here, after we make the decision that
// the card is not young. And we only clean the card if we have been
// asked to (i.e., card_ptr != NULL).
if (card_ptr != NULL) {
*card_ptr = CardTableModRefBS::clean_card_val();
// We must complete this write before we do any of the reads below.
OrderAccess::storeload();
}
// the card is not young.
*card_ptr = CardTableModRefBS::clean_card_val();
// We must complete this write before we do any of the reads below.
OrderAccess::storeload();
// Cache the boundaries of the memory region in some const locals
HeapWord* const start = mr.start();
HeapWord* const end = mr.end();
// We used to use "block_start_careful" here. But we're actually happy
// to update the BOT while we do this...
// Update BOT as needed while finding start of (potential) object.
HeapWord* cur = block_start(start);
assert(cur <= start, "Postcondition");
@ -416,7 +403,9 @@ oops_on_card_seq_iterate_careful(MemRegion mr,
obj = oop(cur);
if (obj->klass_or_null() == NULL) {
// Ran into an unparseable point.
return cur;
assert(!g1h->is_gc_active(),
"Unparsable heap during GC at " PTR_FORMAT, p2i(cur));
return false;
}
// Otherwise...
next = cur + block_size(cur);
@ -433,7 +422,9 @@ oops_on_card_seq_iterate_careful(MemRegion mr,
assert((cur + block_size(cur)) > (HeapWord*)obj, "Loop invariant");
if (obj->klass_or_null() == NULL) {
// Ran into an unparseable point.
return cur;
assert(!g1h->is_gc_active(),
"Unparsable heap during GC at " PTR_FORMAT, p2i(cur));
return false;
}
// Advance the current pointer. "obj" still points to the object to iterate.
@ -452,7 +443,7 @@ oops_on_card_seq_iterate_careful(MemRegion mr,
}
} while (cur < end);
return NULL;
return true;
}
// Code roots support

21
hotspot/src/share/vm/gc/g1/heapRegion.hpp
View File

@ -653,16 +653,17 @@ class HeapRegion: public G1ContiguousSpace {
}
}
// filter_young: if true and the region is a young region then we
// skip the iteration.
// card_ptr: if not NULL, and we decide that the card is not young
// and we iterate over it, we'll clean the card before we start the
// iteration.
HeapWord*
oops_on_card_seq_iterate_careful(MemRegion mr,
FilterOutOfRegionClosure* cl,
bool filter_young,
jbyte* card_ptr);
// Iterate over the card in the card designated by card_ptr,
// applying cl to all references in the region.
// mr: the memory region covered by the card.
// card_ptr: if we decide that the card is not young and we iterate
// over it, we'll clean the card before we start the iteration.
// Returns true if card was successfully processed, false if an
// unparsable part of the heap was encountered, which should only
// happen when invoked concurrently with the mutator.
bool oops_on_card_seq_iterate_careful(MemRegion mr,
FilterOutOfRegionClosure* cl,
jbyte* card_ptr);
size_t recorded_rs_length() const { return _recorded_rs_length; }
double predicted_elapsed_time_ms() const { return _predicted_elapsed_time_ms; }

5
hotspot/src/share/vm/gc/g1/heapRegionManager.cpp
View File

@ -482,8 +482,9 @@ void HeapRegionManager::verify_optional() {
HeapRegionClaimer::HeapRegionClaimer(uint n_workers) :
_n_workers(n_workers), _n_regions(G1CollectedHeap::heap()->_hrm._allocated_heapregions_length), _claims(NULL) {
assert(n_workers > 0, "Need at least one worker.");
_claims = NEW_C_HEAP_ARRAY(uint, _n_regions, mtGC);
memset(_claims, Unclaimed, sizeof(*_claims) * _n_regions);
uint* new_claims = NEW_C_HEAP_ARRAY(uint, _n_regions, mtGC);
memset(new_claims, Unclaimed, sizeof(*_claims) * _n_regions);
_claims = new_claims;
}
HeapRegionClaimer::~HeapRegionClaimer() {

7
hotspot/src/share/vm/gc/g1/heapRegionManager.hpp
View File

@ -259,9 +259,9 @@ public:
// The HeapRegionClaimer is used during parallel iteration over heap regions,
// allowing workers to claim heap regions, gaining exclusive rights to these regions.
class HeapRegionClaimer : public StackObj {
uint _n_workers;
uint _n_regions;
uint* _claims;
uint _n_workers;
uint _n_regions;
volatile uint* _claims;
static const uint Unclaimed = 0;
static const uint Claimed = 1;
@ -285,4 +285,3 @@ class HeapRegionClaimer : public StackObj {
bool claim_region(uint region_index);
};
#endif // SHARE_VM_GC_G1_HEAPREGIONMANAGER_HPP

4
hotspot/src/share/vm/gc/g1/heapRegionRemSet.cpp
View File

@ -56,7 +56,7 @@ class PerRegionTable: public CHeapObj<mtGC> {
PerRegionTable * _collision_list_next;
// Global free list of PRTs
static PerRegionTable* _free_list;
static PerRegionTable* volatile _free_list;
protected:
// We need access in order to union things into the base table.
@ -249,7 +249,7 @@ public:
static void test_fl_mem_size();
};
PerRegionTable* PerRegionTable::_free_list = NULL;
PerRegionTable* volatile PerRegionTable::_free_list = NULL;
size_t OtherRegionsTable::_max_fine_entries = 0;
size_t OtherRegionsTable::_mod_max_fine_entries_mask = 0;

2
hotspot/src/share/vm/gc/g1/sparsePRT.cpp
View File

@ -283,7 +283,7 @@ size_t RSHashTable::mem_size() const {
// ----------------------------------------------------------------------
SparsePRT* SparsePRT::_head_expanded_list = NULL;
SparsePRT* volatile SparsePRT::_head_expanded_list = NULL;
void SparsePRT::add_to_expanded_list(SparsePRT* sprt) {
// We could expand multiple times in a pause -- only put on list once.

2
hotspot/src/share/vm/gc/g1/sparsePRT.hpp
View File

@ -250,7 +250,7 @@ class SparsePRT VALUE_OBJ_CLASS_SPEC {
bool should_be_on_expanded_list();
static SparsePRT* _head_expanded_list;
static SparsePRT* volatile _head_expanded_list;
public:
SparsePRT(HeapRegion* hr);

6
hotspot/src/share/vm/gc/parallel/mutableSpace.hpp
View File

@ -1,5 +1,5 @@
/*
* Copyright (c) 2001, 2015, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2001, 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
@ -51,7 +51,7 @@ class MutableSpace: public ImmutableSpace {
MemRegion _last_setup_region;
size_t _alignment;
protected:
HeapWord* _top;
HeapWord* volatile _top;
MutableSpaceMangler* mangler() { return _mangler; }
@ -69,7 +69,7 @@ class MutableSpace: public ImmutableSpace {
HeapWord* top() const { return _top; }
virtual void set_top(HeapWord* value) { _top = value; }
HeapWord** top_addr() { return &_top; }
HeapWord* volatile* top_addr() { return &_top; }
HeapWord** end_addr() { return &_end; }
virtual void set_bottom(HeapWord* value) { _bottom = value; }

4
hotspot/src/share/vm/gc/parallel/parallelScavengeHeap.hpp
View File

@ -1,5 +1,5 @@
/*
* Copyright (c) 2001, 2015, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2001, 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
@ -175,7 +175,7 @@ class ParallelScavengeHeap : public CollectedHeap {
bool supports_inline_contig_alloc() const { return !UseNUMA; }
HeapWord** top_addr() const { return !UseNUMA ? young_gen()->top_addr() : (HeapWord**)-1; }
HeapWord* volatile* top_addr() const { return !UseNUMA ? young_gen()->top_addr() : (HeapWord* volatile*)-1; }
HeapWord** end_addr() const { return !UseNUMA ? young_gen()->end_addr() : (HeapWord**)-1; }
void ensure_parsability(bool retire_tlabs);

4
hotspot/src/share/vm/gc/parallel/psYoungGen.hpp
View File

@ -1,5 +1,5 @@
/*
* Copyright (c) 2001, 2015, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2001, 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
@ -162,7 +162,7 @@ class PSYoungGen : public CHeapObj<mtGC> {
return result;
}
HeapWord** top_addr() const { return eden_space()->top_addr(); }
HeapWord* volatile* top_addr() const { return eden_space()->top_addr(); }
HeapWord** end_addr() const { return eden_space()->end_addr(); }
// Iteration.

7
hotspot/src/share/vm/gc/parallel/vmStructs_parallelgc.hpp
View File

@ -1,5 +1,5 @@
/*
* Copyright (c) 2007, 2015, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2007, 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
@ -26,7 +26,8 @@
#define SHARE_VM_GC_PARALLEL_VMSTRUCTS_PARALLELGC_HPP
#define VM_STRUCTS_PARALLELGC(nonstatic_field, \
static_field) \
volatile_nonstatic_field, \
static_field) \
\
/**********************/ \
/* Parallel GC fields */ \
@ -40,7 +41,7 @@
nonstatic_field(ImmutableSpace, _bottom, HeapWord*) \
nonstatic_field(ImmutableSpace, _end, HeapWord*) \
\
nonstatic_field(MutableSpace, _top, HeapWord*) \
volatile_nonstatic_field(MutableSpace, _top, HeapWord*) \
\
nonstatic_field(PSYoungGen, _reserved, MemRegion) \
nonstatic_field(PSYoungGen, _virtual_space, PSVirtualSpace*) \

2
hotspot/src/share/vm/gc/serial/defNewGeneration.cpp
View File

@ -512,7 +512,7 @@ size_t DefNewGeneration::contiguous_available() const {
}
HeapWord** DefNewGeneration::top_addr() const { return eden()->top_addr(); }
HeapWord* volatile* DefNewGeneration::top_addr() const { return eden()->top_addr(); }
HeapWord** DefNewGeneration::end_addr() const { return eden()->end_addr(); }
void DefNewGeneration::object_iterate(ObjectClosure* blk) {

4
hotspot/src/share/vm/gc/serial/defNewGeneration.hpp
View File

@ -1,5 +1,5 @@
/*
* Copyright (c) 2001, 2015, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2001, 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
@ -225,7 +225,7 @@ protected:
size_t max_survivor_size() const { return _max_survivor_size; }
bool supports_inline_contig_alloc() const { return true; }
HeapWord** top_addr() const;
HeapWord* volatile* top_addr() const;
HeapWord** end_addr() const;
// Thread-local allocation buffers

2
hotspot/src/share/vm/gc/shared/collectedHeap.hpp
View File

@ -350,7 +350,7 @@ class CollectedHeap : public CHeapObj<mtInternal> {
// These functions return the addresses of the fields that define the
// boundaries of the contiguous allocation area. (These fields should be
// physically near to one another.)
virtual HeapWord** top_addr() const {
virtual HeapWord* volatile* top_addr() const {
guarantee(false, "inline contiguous allocation not supported");
return NULL;
}

15
hotspot/src/share/vm/gc/shared/genCollectedHeap.cpp
View File

@ -721,7 +721,7 @@ bool GenCollectedHeap::supports_inline_contig_alloc() const {
return _young_gen->supports_inline_contig_alloc();
}
HeapWord** GenCollectedHeap::top_addr() const {
HeapWord* volatile* GenCollectedHeap::top_addr() const {
return _young_gen->top_addr();
}
@ -1256,21 +1256,20 @@ class GenTimeOfLastGCClosure: public GenCollectedHeap::GenClosure {
};
jlong GenCollectedHeap::millis_since_last_gc() {
// We need a monotonically non-decreasing time in ms but
// os::javaTimeMillis() does not guarantee monotonicity.
// javaTimeNanos() is guaranteed to be monotonically non-decreasing
// provided the underlying platform provides such a time source
// (and it is bug free). So we still have to guard against getting
// back a time later than 'now'.
jlong now = os::javaTimeNanos() / NANOSECS_PER_MILLISEC;
GenTimeOfLastGCClosure tolgc_cl(now);
// iterate over generations getting the oldest
// time that a generation was collected
generation_iterate(&tolgc_cl, false);
// javaTimeNanos() is guaranteed to be monotonically non-decreasing
// provided the underlying platform provides such a time source
// (and it is bug free). So we still have to guard against getting
// back a time later than 'now'.
jlong retVal = now - tolgc_cl.time();
if (retVal < 0) {
NOT_PRODUCT(log_warning(gc)("time warp: " JLONG_FORMAT, retVal);)
log_warning(gc)("millis_since_last_gc() would return : " JLONG_FORMAT
". returning zero instead.", retVal);
return 0;
}
return retVal;

2
hotspot/src/share/vm/gc/shared/genCollectedHeap.hpp
View File

@ -184,7 +184,7 @@ public:
// We may support a shared contiguous allocation area, if the youngest
// generation does.
bool supports_inline_contig_alloc() const;
HeapWord** top_addr() const;
HeapWord* volatile* top_addr() const;
HeapWord** end_addr() const;
// Perform a full collection of the heap; intended for use in implementing

Some files were not shown because too many files have changed in this diff Show More