--- /dev/null Thu Jan 01 00:00:00 1970 +0000
+++ b/src/jdk.internal.vm.compiler/share/classes/org.graalvm.compiler.lir.amd64/src/org/graalvm/compiler/lir/amd64/AMD64StringUTF16CompressOp.java Thu Nov 15 09:04:07 2018 -0800
@@ -0,0 +1,342 @@
+/*
+ * Copyright (c) 2018, 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 org.graalvm.compiler.lir.amd64;
+
+import static jdk.vm.ci.amd64.AMD64.k1;
+import static jdk.vm.ci.amd64.AMD64.k2;
+import static jdk.vm.ci.amd64.AMD64.k3;
+import static jdk.vm.ci.amd64.AMD64.rax;
+import static jdk.vm.ci.amd64.AMD64.rdi;
+import static jdk.vm.ci.amd64.AMD64.rdx;
+import static jdk.vm.ci.amd64.AMD64.rsi;
+
+import static jdk.vm.ci.amd64.AMD64.rsp;
+import static jdk.vm.ci.code.ValueUtil.asRegister;
+import static org.graalvm.compiler.lir.LIRInstruction.OperandFlag.REG;
+
+import jdk.vm.ci.amd64.AMD64;
+import org.graalvm.compiler.asm.Label;
+import org.graalvm.compiler.asm.amd64.AMD64Address;
+import org.graalvm.compiler.asm.amd64.AMD64Assembler;
+import org.graalvm.compiler.asm.amd64.AMD64MacroAssembler;
+import org.graalvm.compiler.core.common.LIRKind;
+import org.graalvm.compiler.lir.LIRInstructionClass;
+import org.graalvm.compiler.lir.Opcode;
+import org.graalvm.compiler.lir.asm.CompilationResultBuilder;
+import org.graalvm.compiler.lir.gen.LIRGeneratorTool;
+
+import jdk.vm.ci.amd64.AMD64Kind;
+import jdk.vm.ci.code.Register;
+import jdk.vm.ci.meta.Value;
+
+@Opcode("AMD64_STRING_COMPRESS")
+public final class AMD64StringUTF16CompressOp extends AMD64LIRInstruction {
+ public static final LIRInstructionClass<AMD64StringUTF16CompressOp> TYPE = LIRInstructionClass.create(AMD64StringUTF16CompressOp.class);
+
+ @Def({REG}) private Value rres;
+ @Alive({REG}) private Value rsrc;
+ @Alive({REG}) private Value rdst;
+ @Alive({REG}) private Value rlen;
+
+ @Temp({REG}) private Value vtmp1;
+ @Temp({REG}) private Value vtmp2;
+ @Temp({REG}) private Value vtmp3;
+ @Temp({REG}) private Value vtmp4;
+ @Temp({REG}) private Value rtmp5;
+
+ public AMD64StringUTF16CompressOp(LIRGeneratorTool tool, Value res, Value src, Value dst, Value len) {
+ super(TYPE);
+
+ assert asRegister(src).equals(rsi);
+ assert asRegister(dst).equals(rdi);
+ assert asRegister(len).equals(rdx);
+ assert asRegister(res).equals(rax);
+
+ rres = res;
+ rsrc = src;
+ rdst = dst;
+ rlen = len;
+
+ LIRKind vkind = LIRKind.value(AMD64Kind.V512_BYTE);
+
+ vtmp1 = tool.newVariable(vkind);
+ vtmp2 = tool.newVariable(vkind);
+ vtmp3 = tool.newVariable(vkind);
+ vtmp4 = tool.newVariable(vkind);
+
+ rtmp5 = tool.newVariable(LIRKind.value(AMD64Kind.DWORD));
+ }
+
+ @Override
+ public void emitCode(CompilationResultBuilder crb, AMD64MacroAssembler masm) {
+ Register res = asRegister(rres);
+ Register src = asRegister(rsrc);
+ Register dst = asRegister(rdst);
+ Register len = asRegister(rlen);
+
+ Register tmp1 = asRegister(vtmp1);
+ Register tmp2 = asRegister(vtmp2);
+ Register tmp3 = asRegister(vtmp3);
+ Register tmp4 = asRegister(vtmp4);
+ Register tmp5 = asRegister(rtmp5);
+
+ charArrayCompress(masm, src, dst, len, tmp1, tmp2, tmp3, tmp4, tmp5, res);
+ }
+
+ /**
+ * Compress a UTF16 string which de facto is a Latin1 string into a byte array representation
+ * (buffer).
+ *
+ * @param masm the assembler
+ * @param src (rsi) the start address of source char[] to be compressed
+ * @param dst (rdi) the start address of destination byte[] vector
+ * @param len (rdx) the length
+ * @param tmp1 (xmm) temporary xmm register
+ * @param tmp2 (xmm) temporary xmm register
+ * @param tmp3 (xmm) temporary xmm register
+ * @param tmp4 (xmm) temporary xmm register
+ * @param tmp (gpr) temporary gpr register
+ * @param res (rax) the result code (length on success, zero otherwise)
+ */
+ private static void charArrayCompress(AMD64MacroAssembler masm, Register src, Register dst, Register len, Register tmp1,
+ Register tmp2, Register tmp3, Register tmp4, Register tmp, Register res) {
+ assert tmp1.getRegisterCategory().equals(AMD64.XMM);
+ assert tmp2.getRegisterCategory().equals(AMD64.XMM);
+ assert tmp3.getRegisterCategory().equals(AMD64.XMM);
+ assert tmp4.getRegisterCategory().equals(AMD64.XMM);
+
+ Label labelReturnLength = new Label();
+ Label labelReturnZero = new Label();
+ Label labelDone = new Label();
+ Label labelBelowThreshold = new Label();
+
+ assert len.number != res.number;
+
+ masm.push(len); // Save length for return.
+
+ if (masm.supports(AMD64.CPUFeature.AVX512BW) &&
+ masm.supports(AMD64.CPUFeature.AVX512VL) &&
+ masm.supports(AMD64.CPUFeature.BMI2)) {
+
+ Label labelRestoreK1ReturnZero = new Label();
+ Label labelAvxPostAlignment = new Label();
+
+ // If the length of the string is less than 32, we chose not to use the
+ // AVX512 instructions.
+ masm.testl(len, -32);
+ masm.jcc(AMD64Assembler.ConditionFlag.Zero, labelBelowThreshold);
+
+ // First check whether a character is compressible (<= 0xff).
+ // Create mask to test for Unicode chars inside (zmm) vector.
+ masm.movl(res, 0x00ff);
+ masm.evpbroadcastw(tmp2, res);
+
+ masm.kmovq(k3, k1); // Save k1
+
+ masm.testl(len, -64);
+ masm.jcc(AMD64Assembler.ConditionFlag.Zero, labelAvxPostAlignment);
+
+ masm.movl(tmp, dst);
+ masm.andl(tmp, (32 - 1));
+ masm.negl(tmp);
+ masm.andl(tmp, (32 - 1));
+
+ // bail out when there is nothing to be done
+ masm.testl(tmp, tmp);
+ masm.jcc(AMD64Assembler.ConditionFlag.Zero, labelAvxPostAlignment);
+
+ // Compute (1 << N) - 1 = ~(~0 << N), where N is the remaining number
+ // of characters to process.
+ masm.movl(res, -1);
+ masm.shlxl(res, res, tmp);
+ masm.notl(res);
+
+ masm.kmovd(k1, res);
+ masm.evmovdqu16(tmp1, k1, new AMD64Address(src));
+ masm.evpcmpuw(k2, k1, tmp1, tmp2, 2 /* le */);
+ masm.ktestd(k2, k1);
+ masm.jcc(AMD64Assembler.ConditionFlag.CarryClear, labelRestoreK1ReturnZero);
+
+ masm.evpmovwb(new AMD64Address(dst), k1, tmp1);
+
+ masm.addq(src, tmp);
+ masm.addq(src, tmp);
+ masm.addq(dst, tmp);
+ masm.subl(len, tmp);
+
+ masm.bind(labelAvxPostAlignment);
+ // end of alignment
+ Label labelAvx512LoopTail = new Label();
+
+ masm.movl(tmp, len);
+ masm.andl(tmp, -32); // The vector count (in chars).
+ masm.jcc(AMD64Assembler.ConditionFlag.Zero, labelAvx512LoopTail);
+ masm.andl(len, 32 - 1); // The tail count (in chars).
+
+ masm.leaq(src, new AMD64Address(src, tmp, AMD64Address.Scale.Times2));
+ masm.leaq(dst, new AMD64Address(dst, tmp, AMD64Address.Scale.Times1));
+ masm.negq(tmp);
+
+ Label labelAvx512Loop = new Label();
+ // Test and compress 32 chars per iteration, reading 512-bit vectors and
+ // writing 256-bit compressed ditto.
+ masm.bind(labelAvx512Loop);
+ masm.evmovdqu16(tmp1, new AMD64Address(src, tmp, AMD64Address.Scale.Times2));
+ masm.evpcmpuw(k2, tmp1, tmp2, 2 /* le */);
+ masm.kortestd(k2, k2);
+ masm.jcc(AMD64Assembler.ConditionFlag.CarryClear, labelRestoreK1ReturnZero);
+
+ // All 32 chars in the current vector (chunk) are valid for compression,
+ // write truncated byte elements to memory.
+ masm.evpmovwb(new AMD64Address(dst, tmp, AMD64Address.Scale.Times1), tmp1);
+ masm.addq(tmp, 32);
+ masm.jcc(AMD64Assembler.ConditionFlag.NotZero, labelAvx512Loop);
+
+ masm.bind(labelAvx512LoopTail);
+ masm.kmovq(k1, k3); // Restore k1
+
+ // All done if the tail count is zero.
+ masm.testl(len, len);
+ masm.jcc(AMD64Assembler.ConditionFlag.Zero, labelReturnLength);
+
+ // Compute (1 << N) - 1 = ~(~0 << N), where N is the remaining number
+ // of characters to process.
+ masm.movl(res, -1);
+ masm.shlxl(res, res, len);
+ masm.notl(res);
+
+ masm.kmovd(k1, res);
+ masm.evmovdqu16(tmp1, k1, new AMD64Address(src));
+ masm.evpcmpuw(k2, k1, tmp1, tmp2, 2 /* le */);
+ masm.ktestd(k2, k1);
+ masm.jcc(AMD64Assembler.ConditionFlag.CarryClear, labelRestoreK1ReturnZero);
+
+ masm.evpmovwb(new AMD64Address(dst), k1, tmp1);
+
+ masm.kmovq(k1, k3); // Restore k1
+ masm.jmp(labelReturnLength);
+
+ masm.bind(labelRestoreK1ReturnZero);
+ masm.kmovq(k1, k3); // Restore k1
+ masm.jmp(labelReturnZero);
+ }
+
+ if (masm.supports(AMD64.CPUFeature.SSE4_2)) {
+
+ Label labelSSETail = new Label();
+
+ masm.bind(labelBelowThreshold);
+
+ masm.movl(tmp, 0xff00ff00); // Create mask to test for Unicode chars in vectors.
+
+ masm.movl(res, len);
+ masm.andl(res, -16);
+ masm.jccb(AMD64Assembler.ConditionFlag.Zero, labelSSETail);
+ masm.andl(len, 16 - 1);
+
+ // Compress 16 chars per iteration.
+ masm.movdl(tmp1, tmp);
+ masm.pshufd(tmp1, tmp1, 0); // Store Unicode mask in 'vtmp1'.
+ masm.pxor(tmp4, tmp4);
+
+ masm.leaq(src, new AMD64Address(src, res, AMD64Address.Scale.Times2));
+ masm.leaq(dst, new AMD64Address(dst, res, AMD64Address.Scale.Times1));
+ masm.negq(res);
+
+ Label lSSELoop = new Label();
+ // Test and compress 16 chars per iteration, reading 128-bit vectors and
+ // writing 64-bit compressed ditto.
+ masm.bind(lSSELoop);
+ masm.movdqu(tmp2, new AMD64Address(src, res, AMD64Address.Scale.Times2)); // load
+ // 1st 8
+ // characters
+ masm.movdqu(tmp3, new AMD64Address(src, res, AMD64Address.Scale.Times2, 16)); // load
+ // next 8
+ // characters
+ masm.por(tmp4, tmp2);
+ masm.por(tmp4, tmp3);
+ masm.ptest(tmp4, tmp1); // Check for Unicode chars in vector.
+ masm.jcc(AMD64Assembler.ConditionFlag.NotZero, labelReturnZero);
+
+ masm.packuswb(tmp2, tmp3); // Only ASCII chars; compress each to a byte.
+ masm.movdqu(new AMD64Address(dst, res, AMD64Address.Scale.Times1), tmp2);
+ masm.addq(res, 16);
+ masm.jcc(AMD64Assembler.ConditionFlag.NotZero, lSSELoop);
+
+ Label labelCopyChars = new Label();
+ // Test and compress another 8 chars before final tail copy.
+ masm.bind(labelSSETail);
+ masm.movl(res, len);
+ masm.andl(res, -8);
+ masm.jccb(AMD64Assembler.ConditionFlag.Zero, labelCopyChars);
+ masm.andl(len, 8 - 1);
+
+ masm.movdl(tmp1, tmp);
+ masm.pshufd(tmp1, tmp1, 0); // Store Unicode mask in 'vtmp1'.
+ masm.pxor(tmp3, tmp3);
+
+ masm.movdqu(tmp2, new AMD64Address(src));
+ masm.ptest(tmp2, tmp1); // Check for Unicode chars in vector.
+ masm.jccb(AMD64Assembler.ConditionFlag.NotZero, labelReturnZero);
+ masm.packuswb(tmp2, tmp3); // Only ASCII chars; compress each to a byte.
+ masm.movq(new AMD64Address(dst), tmp2);
+ masm.addq(src, 16);
+ masm.addq(dst, 8);
+
+ masm.bind(labelCopyChars);
+ }
+
+ // Compress any remaining characters using a vanilla implementation.
+ masm.testl(len, len);
+ masm.jccb(AMD64Assembler.ConditionFlag.Zero, labelReturnLength);
+ masm.leaq(src, new AMD64Address(src, len, AMD64Address.Scale.Times2));
+ masm.leaq(dst, new AMD64Address(dst, len, AMD64Address.Scale.Times1));
+ masm.negq(len);
+
+ Label labelCopyCharsLoop = new Label();
+ // Compress a single character per iteration.
+ masm.bind(labelCopyCharsLoop);
+ masm.movzwl(res, new AMD64Address(src, len, AMD64Address.Scale.Times2));
+ masm.testl(res, 0xff00); // Check if Unicode character.
+ masm.jccb(AMD64Assembler.ConditionFlag.NotZero, labelReturnZero);
+ // An ASCII character; compress to a byte.
+ masm.movb(new AMD64Address(dst, len, AMD64Address.Scale.Times1), res);
+ masm.incrementq(len, 1);
+ masm.jcc(AMD64Assembler.ConditionFlag.NotZero, labelCopyCharsLoop);
+
+ // If compression succeeded, return the length.
+ masm.bind(labelReturnLength);
+ masm.pop(res);
+ masm.jmpb(labelDone);
+
+ // If compression failed, return 0.
+ masm.bind(labelReturnZero);
+ masm.xorl(res, res);
+ masm.addq(rsp, 8 /* wordSize */);
+
+ masm.bind(labelDone);
+ }
+
+}