2011-12-05 08:58:01 +00:00
|
|
|
// Copyright 2011 the V8 project authors. All rights reserved.
|
|
|
|
|
// Redistribution and use in source and binary forms, with or without
|
|
|
|
|
// modification, are permitted provided that the following conditions are
|
|
|
|
|
// met:
|
|
|
|
|
//
|
|
|
|
|
// * Redistributions of source code must retain the above copyright
|
|
|
|
|
// notice, this list of conditions and the following disclaimer.
|
|
|
|
|
// * Redistributions in binary form must reproduce the above
|
|
|
|
|
// copyright notice, this list of conditions and the following
|
|
|
|
|
// disclaimer in the documentation and/or other materials provided
|
|
|
|
|
// with the distribution.
|
|
|
|
|
// * Neither the name of Google Inc. nor the names of its
|
|
|
|
|
// contributors may be used to endorse or promote products derived
|
|
|
|
|
// from this software without specific prior written permission.
|
|
|
|
|
//
|
|
|
|
|
// THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
|
|
|
|
// "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
|
|
|
|
// LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
|
|
|
|
// A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
|
|
|
|
|
// OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
|
|
|
|
// SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
|
|
|
|
// LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
|
|
|
|
// DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
|
|
|
|
// THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
|
|
|
// (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
|
|
|
|
// OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
|
|
|
|
|
|
|
|
#include <stdlib.h>
|
|
|
|
|
|
2019-05-24 13:51:59 +00:00
|
|
|
#include "src/init/v8.h"
|
2014-06-03 08:12:43 +00:00
|
|
|
|
2019-05-21 09:30:15 +00:00
|
|
|
#include "src/codegen/code-factory.h"
|
|
|
|
|
#include "src/codegen/macro-assembler.h"
|
2015-07-31 11:07:50 +00:00
|
|
|
#include "src/debug/debug.h"
|
2019-05-20 09:15:06 +00:00
|
|
|
#include "src/diagnostics/disasm.h"
|
|
|
|
|
#include "src/diagnostics/disassembler.h"
|
2019-05-22 07:55:37 +00:00
|
|
|
#include "src/execution/frames-inl.h"
|
2019-05-23 13:27:57 +00:00
|
|
|
#include "src/utils/ostreams.h"
|
2019-05-23 08:51:46 +00:00
|
|
|
#include "src/objects/objects-inl.h"
|
2014-06-03 08:12:43 +00:00
|
|
|
#include "test/cctest/cctest.h"
|
2011-12-05 08:58:01 +00:00
|
|
|
|
2017-08-11 11:22:28 +00:00
|
|
|
namespace v8 {
|
|
|
|
|
namespace internal {
|
2011-12-05 08:58:01 +00:00
|
|
|
|
|
|
|
|
#define __ assm.
|
|
|
|
|
|
|
|
|
|
TEST(DisasmX64) {
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2014-03-05 09:28:26 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
2016-09-12 06:24:06 +00:00
|
|
|
v8::internal::byte buffer[8192];
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{},
|
|
|
|
|
ExternalAssemblerBuffer(buffer, sizeof buffer));
|
2011-12-05 08:58:01 +00:00
|
|
|
|
|
|
|
|
// Short immediate instructions
|
|
|
|
|
__ addq(rax, Immediate(12345678));
|
2014-03-26 12:15:35 +00:00
|
|
|
__ orq(rax, Immediate(12345678));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ subq(rax, Immediate(12345678));
|
2014-03-26 12:15:35 +00:00
|
|
|
__ xorq(rax, Immediate(12345678));
|
|
|
|
|
__ andq(rax, Immediate(12345678));
|
2011-12-05 08:58:01 +00:00
|
|
|
|
|
|
|
|
// ---- This one caused crash
|
|
|
|
|
__ movq(rbx, Operand(rsp, rcx, times_2, 0)); // [rsp+rcx*4]
|
|
|
|
|
|
|
|
|
|
// ---- All instructions that I can think of
|
|
|
|
|
__ addq(rdx, rbx);
|
|
|
|
|
__ addq(rdx, Operand(rbx, 0));
|
|
|
|
|
__ addq(rdx, Operand(rbx, 16));
|
|
|
|
|
__ addq(rdx, Operand(rbx, 1999));
|
2014-03-05 09:28:26 +00:00
|
|
|
__ addq(rdx, Operand(rbx, -4));
|
|
|
|
|
__ addq(rdx, Operand(rbx, -1999));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ addq(rdx, Operand(rsp, 0));
|
|
|
|
|
__ addq(rdx, Operand(rsp, 16));
|
|
|
|
|
__ addq(rdx, Operand(rsp, 1999));
|
2014-03-05 09:28:26 +00:00
|
|
|
__ addq(rdx, Operand(rsp, -4));
|
|
|
|
|
__ addq(rdx, Operand(rsp, -1999));
|
|
|
|
|
__ nop();
|
|
|
|
|
__ addq(rsi, Operand(rcx, times_4, 0));
|
|
|
|
|
__ addq(rsi, Operand(rcx, times_4, 24));
|
|
|
|
|
__ addq(rsi, Operand(rcx, times_4, -4));
|
|
|
|
|
__ addq(rsi, Operand(rcx, times_4, -1999));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ nop();
|
|
|
|
|
__ addq(rdi, Operand(rbp, rcx, times_4, 0));
|
|
|
|
|
__ addq(rdi, Operand(rbp, rcx, times_4, 12));
|
2014-03-05 09:28:26 +00:00
|
|
|
__ addq(rdi, Operand(rbp, rcx, times_4, -8));
|
|
|
|
|
__ addq(rdi, Operand(rbp, rcx, times_4, -3999));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ addq(Operand(rbp, rcx, times_4, 12), Immediate(12));
|
|
|
|
|
|
2018-08-14 08:39:05 +00:00
|
|
|
__ bswapl(rax);
|
|
|
|
|
__ bswapq(rdi);
|
2015-03-20 08:37:20 +00:00
|
|
|
__ bsrl(rax, r15);
|
|
|
|
|
__ bsrl(r9, Operand(rcx, times_8, 91919));
|
|
|
|
|
|
2011-12-05 08:58:01 +00:00
|
|
|
__ nop();
|
|
|
|
|
__ addq(rbx, Immediate(12));
|
|
|
|
|
__ nop();
|
|
|
|
|
__ nop();
|
2014-03-26 12:15:35 +00:00
|
|
|
__ andq(rdx, Immediate(3));
|
|
|
|
|
__ andq(rdx, Operand(rsp, 4));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ cmpq(rdx, Immediate(3));
|
|
|
|
|
__ cmpq(rdx, Operand(rsp, 4));
|
|
|
|
|
__ cmpq(Operand(rbp, rcx, times_4, 0), Immediate(1000));
|
|
|
|
|
__ cmpb(rbx, Operand(rbp, rcx, times_2, 0));
|
|
|
|
|
__ cmpb(Operand(rbp, rcx, times_2, 0), rbx);
|
2014-03-26 12:15:35 +00:00
|
|
|
__ orq(rdx, Immediate(3));
|
|
|
|
|
__ xorq(rdx, Immediate(3));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ nop();
|
2013-11-18 15:24:41 +00:00
|
|
|
__ cpuid();
|
2014-04-11 06:32:06 +00:00
|
|
|
__ movsxbl(rdx, Operand(rcx, 0));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ movsxbq(rdx, Operand(rcx, 0));
|
2014-04-11 06:32:06 +00:00
|
|
|
__ movsxwl(rdx, Operand(rcx, 0));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ movsxwq(rdx, Operand(rcx, 0));
|
|
|
|
|
__ movzxbl(rdx, Operand(rcx, 0));
|
|
|
|
|
__ movzxwl(rdx, Operand(rcx, 0));
|
|
|
|
|
__ movzxbq(rdx, Operand(rcx, 0));
|
|
|
|
|
__ movzxwq(rdx, Operand(rcx, 0));
|
|
|
|
|
|
|
|
|
|
__ nop();
|
2014-03-21 02:42:10 +00:00
|
|
|
__ imulq(rdx, rcx);
|
2011-12-05 08:58:01 +00:00
|
|
|
__ shld(rdx, rcx);
|
|
|
|
|
__ shrd(rdx, rcx);
|
2014-10-02 09:08:09 +00:00
|
|
|
__ shlq(Operand(rdi, rax, times_4, 100), Immediate(1));
|
|
|
|
|
__ shlq(Operand(rdi, rax, times_4, 100), Immediate(6));
|
|
|
|
|
__ shlq(Operand(r15, 0), Immediate(1));
|
|
|
|
|
__ shlq(Operand(r15, 0), Immediate(6));
|
|
|
|
|
__ shlq_cl(Operand(r15, 0));
|
|
|
|
|
__ shlq_cl(Operand(r15, 0));
|
|
|
|
|
__ shlq_cl(Operand(rdi, rax, times_4, 100));
|
|
|
|
|
__ shlq_cl(Operand(rdi, rax, times_4, 100));
|
|
|
|
|
__ shlq(rdx, Immediate(1));
|
|
|
|
|
__ shlq(rdx, Immediate(6));
|
|
|
|
|
__ shll(Operand(rdi, rax, times_4, 100), Immediate(1));
|
|
|
|
|
__ shll(Operand(rdi, rax, times_4, 100), Immediate(6));
|
|
|
|
|
__ shll(Operand(r15, 0), Immediate(1));
|
|
|
|
|
__ shll(Operand(r15, 0), Immediate(6));
|
|
|
|
|
__ shll_cl(Operand(r15, 0));
|
|
|
|
|
__ shll_cl(Operand(r15, 0));
|
|
|
|
|
__ shll_cl(Operand(rdi, rax, times_4, 100));
|
|
|
|
|
__ shll_cl(Operand(rdi, rax, times_4, 100));
|
|
|
|
|
__ shll(rdx, Immediate(1));
|
|
|
|
|
__ shll(rdx, Immediate(6));
|
2018-09-12 07:29:11 +00:00
|
|
|
__ btq(Operand(rdx, 0), rcx);
|
|
|
|
|
__ btsq(Operand(rdx, 0), rcx);
|
|
|
|
|
__ btsq(Operand(rbx, rcx, times_4, 0), rcx);
|
|
|
|
|
__ btsq(rcx, Immediate(13));
|
|
|
|
|
__ btrq(rcx, Immediate(13));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ nop();
|
2014-03-19 08:59:04 +00:00
|
|
|
__ pushq(Immediate(12));
|
|
|
|
|
__ pushq(Immediate(23456));
|
|
|
|
|
__ pushq(rcx);
|
|
|
|
|
__ pushq(rsi);
|
2019-10-24 14:49:31 +00:00
|
|
|
__ pushq(Operand(rbp, StandardFrameConstants::kFunctionOffset));
|
2014-03-19 08:59:04 +00:00
|
|
|
__ pushq(Operand(rbx, rcx, times_4, 0));
|
|
|
|
|
__ pushq(Operand(rbx, rcx, times_4, 0));
|
|
|
|
|
__ pushq(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ popq(rdx);
|
|
|
|
|
__ popq(rax);
|
|
|
|
|
__ popq(Operand(rbx, rcx, times_4, 0));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ nop();
|
|
|
|
|
|
|
|
|
|
__ addq(rdx, Operand(rsp, 16));
|
|
|
|
|
__ addq(rdx, rcx);
|
|
|
|
|
__ movb(rdx, Operand(rcx, 0));
|
|
|
|
|
__ movb(rcx, Immediate(6));
|
|
|
|
|
__ movb(Operand(rsp, 16), rdx);
|
|
|
|
|
__ movw(Operand(rsp, 16), rdx);
|
|
|
|
|
__ nop();
|
|
|
|
|
__ movsxwq(rdx, Operand(rsp, 12));
|
|
|
|
|
__ movsxbq(rdx, Operand(rsp, 12));
|
|
|
|
|
__ movsxlq(rdx, Operand(rsp, 12));
|
|
|
|
|
__ movzxwq(rdx, Operand(rsp, 12));
|
|
|
|
|
__ movzxbq(rdx, Operand(rsp, 12));
|
|
|
|
|
__ nop();
|
|
|
|
|
__ movq(rdx, Immediate(1234567));
|
|
|
|
|
__ movq(rdx, Operand(rsp, 12));
|
|
|
|
|
__ movq(Operand(rbx, rcx, times_4, 10000), Immediate(12345));
|
|
|
|
|
__ movq(Operand(rbx, rcx, times_4, 10000), rdx);
|
|
|
|
|
__ nop();
|
|
|
|
|
__ decb(rdx);
|
|
|
|
|
__ decb(Operand(rax, 10));
|
|
|
|
|
__ decb(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ decq(rdx);
|
|
|
|
|
__ cdq();
|
|
|
|
|
|
2019-12-04 09:15:26 +00:00
|
|
|
__ repstosl();
|
2019-10-15 08:17:25 +00:00
|
|
|
__ repstosq();
|
|
|
|
|
|
2011-12-05 08:58:01 +00:00
|
|
|
__ nop();
|
|
|
|
|
__ idivq(rdx);
|
2014-11-03 10:04:37 +00:00
|
|
|
__ mull(rdx);
|
|
|
|
|
__ mulq(rdx);
|
2020-01-24 17:08:27 +00:00
|
|
|
|
|
|
|
|
__ negb(rdx);
|
|
|
|
|
__ negb(r10);
|
|
|
|
|
__ negw(rdx);
|
|
|
|
|
__ negl(rdx);
|
2014-03-26 01:13:09 +00:00
|
|
|
__ negq(rdx);
|
2020-01-24 17:08:27 +00:00
|
|
|
__ negb(Operand(rsp, 12));
|
|
|
|
|
__ negw(Operand(rsp, 12));
|
|
|
|
|
__ negl(Operand(rsp, 12));
|
|
|
|
|
__ negb(Operand(rsp, 12));
|
|
|
|
|
|
2014-03-26 12:15:35 +00:00
|
|
|
__ notq(rdx);
|
2011-12-05 08:58:01 +00:00
|
|
|
__ testq(Operand(rbx, rcx, times_4, 10000), rdx);
|
|
|
|
|
|
2014-03-21 02:42:10 +00:00
|
|
|
__ imulq(rdx, rcx, Immediate(12));
|
|
|
|
|
__ imulq(rdx, rcx, Immediate(1000));
|
2014-10-02 12:22:36 +00:00
|
|
|
__ imulq(rdx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ imulq(rdx, Operand(rbx, rcx, times_4, 10000), Immediate(12));
|
2014-09-25 06:37:23 +00:00
|
|
|
__ imulq(rdx, Operand(rbx, rcx, times_4, 10000), Immediate(1000));
|
2014-10-02 12:22:36 +00:00
|
|
|
__ imull(r15, rcx, Immediate(12));
|
|
|
|
|
__ imull(r15, rcx, Immediate(1000));
|
|
|
|
|
__ imull(r15, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ imull(r15, Operand(rbx, rcx, times_4, 10000), Immediate(12));
|
|
|
|
|
__ imull(r15, Operand(rbx, rcx, times_4, 10000), Immediate(1000));
|
2011-12-05 08:58:01 +00:00
|
|
|
|
|
|
|
|
__ incq(rdx);
|
|
|
|
|
__ incq(Operand(rbx, rcx, times_4, 10000));
|
2014-03-19 08:59:04 +00:00
|
|
|
__ pushq(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ popq(Operand(rbx, rcx, times_4, 10000));
|
2019-01-28 13:01:36 +00:00
|
|
|
__ jmp(Operand(rbx, rcx, times_4, 10000));
|
2011-12-05 08:58:01 +00:00
|
|
|
|
2014-03-26 11:17:53 +00:00
|
|
|
__ leaq(rdx, Operand(rbx, rcx, times_4, 10000));
|
2014-03-26 12:15:35 +00:00
|
|
|
__ orq(rdx, Immediate(12345));
|
|
|
|
|
__ orq(rdx, Operand(rbx, rcx, times_4, 10000));
|
2011-12-05 08:58:01 +00:00
|
|
|
|
|
|
|
|
__ nop();
|
|
|
|
|
|
2014-03-28 04:55:00 +00:00
|
|
|
__ rclq(rdx, Immediate(1));
|
|
|
|
|
__ rclq(rdx, Immediate(7));
|
|
|
|
|
__ rcrq(rdx, Immediate(1));
|
|
|
|
|
__ rcrq(rdx, Immediate(7));
|
|
|
|
|
__ sarq(rdx, Immediate(1));
|
|
|
|
|
__ sarq(rdx, Immediate(6));
|
|
|
|
|
__ sarq_cl(rdx);
|
2011-12-05 08:58:01 +00:00
|
|
|
__ sbbq(rdx, rbx);
|
|
|
|
|
__ shld(rdx, rbx);
|
2014-03-28 04:55:00 +00:00
|
|
|
__ shlq(rdx, Immediate(1));
|
|
|
|
|
__ shlq(rdx, Immediate(6));
|
|
|
|
|
__ shlq_cl(rdx);
|
2011-12-05 08:58:01 +00:00
|
|
|
__ shrd(rdx, rbx);
|
2014-03-28 04:55:00 +00:00
|
|
|
__ shrq(rdx, Immediate(1));
|
|
|
|
|
__ shrq(rdx, Immediate(7));
|
|
|
|
|
__ shrq_cl(rdx);
|
2011-12-05 08:58:01 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
// Immediates
|
|
|
|
|
|
|
|
|
|
__ addq(rbx, Immediate(12));
|
|
|
|
|
__ addq(Operand(rdx, rcx, times_4, 10000), Immediate(12));
|
|
|
|
|
|
2014-03-26 12:15:35 +00:00
|
|
|
__ andq(rbx, Immediate(12345));
|
2011-12-05 08:58:01 +00:00
|
|
|
|
|
|
|
|
__ cmpq(rbx, Immediate(12345));
|
|
|
|
|
__ cmpq(rbx, Immediate(12));
|
|
|
|
|
__ cmpq(Operand(rdx, rcx, times_4, 10000), Immediate(12));
|
|
|
|
|
__ cmpb(rax, Immediate(100));
|
|
|
|
|
|
2014-03-26 12:15:35 +00:00
|
|
|
__ orq(rbx, Immediate(12345));
|
2011-12-05 08:58:01 +00:00
|
|
|
|
|
|
|
|
__ subq(rbx, Immediate(12));
|
|
|
|
|
__ subq(Operand(rdx, rcx, times_4, 10000), Immediate(12));
|
|
|
|
|
|
2014-03-26 12:15:35 +00:00
|
|
|
__ xorq(rbx, Immediate(12345));
|
2011-12-05 08:58:01 +00:00
|
|
|
|
2014-03-21 02:42:10 +00:00
|
|
|
__ imulq(rdx, rcx, Immediate(12));
|
|
|
|
|
__ imulq(rdx, rcx, Immediate(1000));
|
2011-12-05 08:58:01 +00:00
|
|
|
|
|
|
|
|
__ cld();
|
|
|
|
|
|
|
|
|
|
__ subq(rdx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ subq(rdx, rbx);
|
|
|
|
|
|
|
|
|
|
__ testq(rdx, Immediate(12345));
|
|
|
|
|
__ testq(Operand(rbx, rcx, times_8, 10000), rdx);
|
|
|
|
|
__ testb(Operand(rcx, rbx, times_2, 1000), rdx);
|
|
|
|
|
__ testb(Operand(rax, -20), Immediate(0x9A));
|
|
|
|
|
__ nop();
|
|
|
|
|
|
2014-03-26 12:15:35 +00:00
|
|
|
__ xorq(rdx, Immediate(12345));
|
|
|
|
|
__ xorq(rdx, Operand(rbx, rcx, times_8, 10000));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ hlt();
|
|
|
|
|
__ int3();
|
|
|
|
|
__ ret(0);
|
|
|
|
|
__ ret(8);
|
|
|
|
|
|
|
|
|
|
// Calls
|
|
|
|
|
|
|
|
|
|
Label L1, L2;
|
|
|
|
|
__ bind(&L1);
|
|
|
|
|
__ nop();
|
|
|
|
|
__ call(&L1);
|
|
|
|
|
__ call(&L2);
|
|
|
|
|
__ nop();
|
|
|
|
|
__ bind(&L2);
|
2019-11-06 13:39:55 +00:00
|
|
|
__ call(rcx);
|
2011-12-05 08:58:01 +00:00
|
|
|
__ nop();
|
2019-11-04 10:01:19 +00:00
|
|
|
Handle<Code> ic = BUILTIN_CODE(isolate, ArrayFrom);
|
2011-12-05 08:58:01 +00:00
|
|
|
__ call(ic, RelocInfo::CODE_TARGET);
|
|
|
|
|
__ nop();
|
|
|
|
|
|
|
|
|
|
__ jmp(&L1);
|
2019-01-28 13:01:36 +00:00
|
|
|
__ jmp(Operand(rbx, rcx, times_4, 10000));
|
2011-12-05 08:58:01 +00:00
|
|
|
__ jmp(ic, RelocInfo::CODE_TARGET);
|
|
|
|
|
__ nop();
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Label Ljcc;
|
|
|
|
|
__ nop();
|
|
|
|
|
// long jumps
|
|
|
|
|
__ j(overflow, &Ljcc);
|
|
|
|
|
__ j(no_overflow, &Ljcc);
|
|
|
|
|
__ j(below, &Ljcc);
|
|
|
|
|
__ j(above_equal, &Ljcc);
|
|
|
|
|
__ j(equal, &Ljcc);
|
|
|
|
|
__ j(not_equal, &Ljcc);
|
|
|
|
|
__ j(below_equal, &Ljcc);
|
|
|
|
|
__ j(above, &Ljcc);
|
|
|
|
|
__ j(sign, &Ljcc);
|
|
|
|
|
__ j(not_sign, &Ljcc);
|
|
|
|
|
__ j(parity_even, &Ljcc);
|
|
|
|
|
__ j(parity_odd, &Ljcc);
|
|
|
|
|
__ j(less, &Ljcc);
|
|
|
|
|
__ j(greater_equal, &Ljcc);
|
|
|
|
|
__ j(less_equal, &Ljcc);
|
|
|
|
|
__ j(greater, &Ljcc);
|
|
|
|
|
__ nop();
|
|
|
|
|
__ bind(&Ljcc);
|
|
|
|
|
// short jumps
|
|
|
|
|
__ j(overflow, &Ljcc);
|
|
|
|
|
__ j(no_overflow, &Ljcc);
|
|
|
|
|
__ j(below, &Ljcc);
|
|
|
|
|
__ j(above_equal, &Ljcc);
|
|
|
|
|
__ j(equal, &Ljcc);
|
|
|
|
|
__ j(not_equal, &Ljcc);
|
|
|
|
|
__ j(below_equal, &Ljcc);
|
|
|
|
|
__ j(above, &Ljcc);
|
|
|
|
|
__ j(sign, &Ljcc);
|
|
|
|
|
__ j(not_sign, &Ljcc);
|
|
|
|
|
__ j(parity_even, &Ljcc);
|
|
|
|
|
__ j(parity_odd, &Ljcc);
|
|
|
|
|
__ j(less, &Ljcc);
|
|
|
|
|
__ j(greater_equal, &Ljcc);
|
|
|
|
|
__ j(less_equal, &Ljcc);
|
|
|
|
|
__ j(greater, &Ljcc);
|
|
|
|
|
|
|
|
|
|
// 0xD9 instructions
|
|
|
|
|
__ nop();
|
|
|
|
|
|
|
|
|
|
__ fld(1);
|
|
|
|
|
__ fld1();
|
|
|
|
|
__ fldz();
|
|
|
|
|
__ fldpi();
|
|
|
|
|
__ fabs();
|
|
|
|
|
__ fchs();
|
|
|
|
|
__ fprem();
|
|
|
|
|
__ fprem1();
|
|
|
|
|
__ fincstp();
|
|
|
|
|
__ ftst();
|
|
|
|
|
__ fxch(3);
|
|
|
|
|
__ fld_s(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ fstp_s(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ ffree(3);
|
|
|
|
|
__ fld_d(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ fstp_d(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ nop();
|
|
|
|
|
|
|
|
|
|
__ fild_s(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ fistp_s(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ fild_d(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ fistp_d(Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ fnstsw_ax();
|
|
|
|
|
__ nop();
|
|
|
|
|
__ fadd(3);
|
|
|
|
|
__ fsub(3);
|
|
|
|
|
__ fmul(3);
|
|
|
|
|
__ fdiv(3);
|
|
|
|
|
|
|
|
|
|
__ faddp(3);
|
|
|
|
|
__ fsubp(3);
|
|
|
|
|
__ fmulp(3);
|
|
|
|
|
__ fdivp(3);
|
|
|
|
|
__ fcompp();
|
|
|
|
|
__ fwait();
|
2014-02-03 08:13:12 +00:00
|
|
|
__ frndint();
|
|
|
|
|
__ fninit();
|
2011-12-05 08:58:01 +00:00
|
|
|
__ nop();
|
2013-10-28 10:38:40 +00:00
|
|
|
|
|
|
|
|
// SSE instruction
|
2011-12-05 08:58:01 +00:00
|
|
|
{
|
2013-11-18 15:24:41 +00:00
|
|
|
// Move operation
|
2013-10-28 10:38:40 +00:00
|
|
|
__ cvttss2si(rdx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cvttss2si(rdx, xmm1);
|
2020-04-06 16:44:23 +00:00
|
|
|
__ cvtqsi2ss(xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cvtqsi2ss(xmm1, rdx);
|
2018-08-14 19:38:14 +00:00
|
|
|
__ cvttps2dq(xmm0, xmm1);
|
|
|
|
|
__ cvttps2dq(xmm0, Operand(rbx, rcx, times_4, 10000));
|
2013-10-28 10:38:40 +00:00
|
|
|
__ movaps(xmm0, xmm1);
|
2017-12-18 14:32:23 +00:00
|
|
|
__ movdqa(xmm0, Operand(rsp, 12));
|
|
|
|
|
__ movdqa(Operand(rsp, 12), xmm0);
|
|
|
|
|
__ movdqu(xmm0, Operand(rsp, 12));
|
|
|
|
|
__ movdqu(Operand(rsp, 12), xmm0);
|
[x64] Refactor pinsrb family of instructions
The existing macro assembler define Pinsrb, which expects 3 arguments:
- XMMRegister dst
- Register/Operand src
- uint8_t imm
which overwrites dst with src at lane specified by imm.
That means we cannot use the AVX version, which has 4 arguments, and
does not overwrite dst.
This refactoring defines the 4 argument AVX version instead, and if AVX
is not supported, fall back to the SSE version, and ensure that the
value is copied over into dst first.
For convenience, we define an overload with 3 arguments that duplicates
dst, this replicates the SSE behavior, so that not all callers have to
be updated.
Bug: v8:10975, v8:10933
Change-Id: I6f9b9d37fa08d3f5cff4f040ae7d5e1f0cf36455
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2444096
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Reviewed-by: Bill Budge <bbudge@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70392}
2020-10-07 22:45:10 +00:00
|
|
|
__ movdqu(xmm1, xmm0);
|
2020-10-14 16:33:41 +00:00
|
|
|
__ movlps(xmm8, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ movlps(Operand(rbx, rcx, times_4, 10000), xmm9);
|
|
|
|
|
__ movhps(xmm8, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ movhps(Operand(rbx, rcx, times_4, 10000), xmm9);
|
2017-01-26 01:38:17 +00:00
|
|
|
__ shufps(xmm0, xmm9, 0x0);
|
2011-12-05 08:58:01 +00:00
|
|
|
|
2014-11-26 05:31:41 +00:00
|
|
|
__ ucomiss(xmm0, xmm1);
|
|
|
|
|
__ ucomiss(xmm0, Operand(rbx, rcx, times_4, 10000));
|
2019-10-21 22:04:20 +00:00
|
|
|
|
2020-04-06 16:44:23 +00:00
|
|
|
__ movmskps(rdx, xmm9);
|
|
|
|
|
|
2019-10-21 22:04:20 +00:00
|
|
|
#define EMIT_SSE_INSTR(instruction, notUsed1, notUsed2) \
|
|
|
|
|
__ instruction(xmm1, xmm0); \
|
|
|
|
|
__ instruction(xmm1, Operand(rbx, rcx, times_4, 10000));
|
2019-11-25 19:26:01 +00:00
|
|
|
SSE_BINOP_INSTRUCTION_LIST(EMIT_SSE_INSTR)
|
|
|
|
|
SSE_UNOP_INSTRUCTION_LIST(EMIT_SSE_INSTR)
|
2019-10-21 22:04:20 +00:00
|
|
|
#undef EMIT_SSE_INSTR
|
2019-10-29 18:27:55 +00:00
|
|
|
|
|
|
|
|
#define EMIT_SSE_INSTR(instruction, notUsed1, notUsed2, notUse3) \
|
|
|
|
|
__ instruction(xmm1, xmm0); \
|
|
|
|
|
__ instruction(xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
SSE_INSTRUCTION_LIST_SS(EMIT_SSE_INSTR)
|
|
|
|
|
#undef EMIT_SSE_INSTR
|
2013-10-28 10:38:40 +00:00
|
|
|
}
|
2016-09-12 06:24:06 +00:00
|
|
|
|
|
|
|
|
// SSE2 instructions
|
2011-12-05 08:58:01 +00:00
|
|
|
{
|
2013-10-28 10:38:40 +00:00
|
|
|
__ cvttsd2si(rdx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cvttsd2si(rdx, xmm1);
|
|
|
|
|
__ cvttsd2siq(rdx, xmm1);
|
2014-09-19 14:25:13 +00:00
|
|
|
__ cvttsd2siq(rdx, Operand(rbx, rcx, times_4, 10000));
|
2020-04-06 16:44:23 +00:00
|
|
|
__ cvtlsi2sd(xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cvtlsi2sd(xmm1, rdx);
|
2014-09-30 10:24:11 +00:00
|
|
|
__ cvtqsi2sd(xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cvtqsi2sd(xmm1, rdx);
|
2020-04-06 16:44:23 +00:00
|
|
|
__ cvtss2sd(xmm1, xmm9);
|
|
|
|
|
__ cvtss2sd(xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cvtsd2si(rdx, xmm9);
|
|
|
|
|
__ cvtsd2siq(rdx, xmm9);
|
|
|
|
|
|
2013-10-28 10:38:40 +00:00
|
|
|
__ movsd(xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ movsd(Operand(rbx, rcx, times_4, 10000), xmm1);
|
|
|
|
|
// 128 bit move instructions.
|
2016-07-21 11:01:44 +00:00
|
|
|
__ movupd(xmm0, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ movupd(Operand(rbx, rcx, times_4, 10000), xmm0);
|
2013-10-28 10:38:40 +00:00
|
|
|
__ movdqa(xmm0, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ movdqa(Operand(rbx, rcx, times_4, 10000), xmm0);
|
|
|
|
|
|
|
|
|
|
__ ucomisd(xmm0, xmm1);
|
2020-04-06 16:44:23 +00:00
|
|
|
__ ucomisd(xmm8, Operand(rbx, rdx, times_4, 10000));
|
2013-10-28 10:38:40 +00:00
|
|
|
|
2020-04-06 16:44:23 +00:00
|
|
|
__ cmpltsd(xmm3, xmm11);
|
|
|
|
|
|
|
|
|
|
__ movmskpd(rdx, xmm9);
|
2020-04-06 16:42:49 +00:00
|
|
|
__ pmovmskb(rdx, xmm9);
|
|
|
|
|
|
2014-10-17 04:18:38 +00:00
|
|
|
__ pcmpeqd(xmm1, xmm0);
|
2015-03-05 09:22:26 +00:00
|
|
|
|
|
|
|
|
__ punpckldq(xmm1, xmm11);
|
2016-06-17 09:26:58 +00:00
|
|
|
__ punpckldq(xmm5, Operand(rdx, 4));
|
2015-03-05 09:22:26 +00:00
|
|
|
__ punpckhdq(xmm8, xmm15);
|
2016-09-12 06:24:06 +00:00
|
|
|
|
[wasm] Implement wasm x64 I16x8 Ops
- Add I16x8 Splat, ExtractLane, ReplaceLane, shift ops, Some BinOps and compare ops
- Add pshufhw, pshuflw in the assembler, disassembler
- Fix incorrect modrm for pextrw, this bug disregards the register allocated and always makes pextrw use rax.
- Fix pextrw disasm to take the 0 - 7 bits of the immediate instead of 0 - 3.
- Pextrw, pinsrw are in the assembler use 128 bit encodings, pextrw, pinsrw in the disassembler use legacy encodings, fix inconsistencies causing weird code gen when --print-code is used.
Review-Url: https://codereview.chromium.org/2767983002
Cr-Commit-Position: refs/heads/master@{#44664}
2017-04-17 18:47:46 +00:00
|
|
|
__ pshuflw(xmm2, xmm4, 3);
|
|
|
|
|
__ pshufhw(xmm1, xmm9, 6);
|
|
|
|
|
|
2016-09-12 06:24:06 +00:00
|
|
|
#define EMIT_SSE2_INSTR(instruction, notUsed1, notUsed2, notUsed3) \
|
|
|
|
|
__ instruction(xmm5, xmm1); \
|
|
|
|
|
__ instruction(xmm5, Operand(rdx, 4));
|
|
|
|
|
|
|
|
|
|
SSE2_INSTRUCTION_LIST(EMIT_SSE2_INSTR)
|
Fix assembler for sqrtpd
The assembly of sqrtpd when using Sqrtpd macro was wrong, since
Sqrtpd(xmm1, xmm1) will incorrect generated vsqrtpd(xmm1, xmm1, xmm1),
which is nonsensical, since vsqrtpd only takes two operands. The
expected instruction should be vsqrtpd(xmm1, xmm0, xmm1) in terms of the
encoding, which is vsqrtpd(xmm1, xmm1).
So, move sqrtpd and cvtps2dq out into their own macro list, because
they have two operands in their AVX form, unlike the rest of the
instructions in SSE2_INSTRUCTION_LIST.
Also updated disasm and tests to use this new list.
Fixed: v8:10170
Change-Id: Ia9343c9a3ae64596bbc876744556e1dcea2a443b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2032195
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#66088}
2020-01-31 18:28:11 +00:00
|
|
|
SSE2_UNOP_INSTRUCTION_LIST(EMIT_SSE2_INSTR)
|
2019-10-30 19:45:10 +00:00
|
|
|
SSE2_INSTRUCTION_LIST_SD(EMIT_SSE2_INSTR)
|
2016-09-12 06:24:06 +00:00
|
|
|
#undef EMIT_SSE2_INSTR
|
2019-12-04 14:56:48 +00:00
|
|
|
|
|
|
|
|
#define EMIT_SSE2_SHIFT_IMM(instruction, notUsed1, notUsed2, notUsed3, \
|
|
|
|
|
notUsed4) \
|
|
|
|
|
__ instruction(xmm3, 0xA3);
|
|
|
|
|
SSE2_INSTRUCTION_LIST_SHIFT_IMM(EMIT_SSE2_SHIFT_IMM)
|
|
|
|
|
#undef EMIT_SSE2_SHIFT_IMM
|
2011-12-05 08:58:01 +00:00
|
|
|
}
|
|
|
|
|
|
2013-10-28 10:38:40 +00:00
|
|
|
// cmov.
|
2011-12-05 08:58:01 +00:00
|
|
|
{
|
2013-10-28 10:38:40 +00:00
|
|
|
__ cmovq(overflow, rax, Operand(rax, 0));
|
|
|
|
|
__ cmovq(no_overflow, rax, Operand(rax, 1));
|
|
|
|
|
__ cmovq(below, rax, Operand(rax, 2));
|
|
|
|
|
__ cmovq(above_equal, rax, Operand(rax, 3));
|
|
|
|
|
__ cmovq(equal, rax, Operand(rbx, 0));
|
|
|
|
|
__ cmovq(not_equal, rax, Operand(rbx, 1));
|
|
|
|
|
__ cmovq(below_equal, rax, Operand(rbx, 2));
|
|
|
|
|
__ cmovq(above, rax, Operand(rbx, 3));
|
|
|
|
|
__ cmovq(sign, rax, Operand(rcx, 0));
|
|
|
|
|
__ cmovq(not_sign, rax, Operand(rcx, 1));
|
|
|
|
|
__ cmovq(parity_even, rax, Operand(rcx, 2));
|
|
|
|
|
__ cmovq(parity_odd, rax, Operand(rcx, 3));
|
|
|
|
|
__ cmovq(less, rax, Operand(rdx, 0));
|
|
|
|
|
__ cmovq(greater_equal, rax, Operand(rdx, 1));
|
|
|
|
|
__ cmovq(less_equal, rax, Operand(rdx, 2));
|
|
|
|
|
__ cmovq(greater, rax, Operand(rdx, 3));
|
2011-12-05 08:58:01 +00:00
|
|
|
}
|
|
|
|
|
|
2016-09-12 06:24:06 +00:00
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(SSE3)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, SSE3);
|
2018-02-28 06:28:08 +00:00
|
|
|
__ haddps(xmm1, xmm0);
|
|
|
|
|
__ haddps(xmm1, Operand(rbx, rcx, times_4, 10000));
|
2016-09-12 06:24:06 +00:00
|
|
|
__ lddqu(xmm1, Operand(rdx, 4));
|
2019-07-26 00:00:22 +00:00
|
|
|
__ movddup(xmm1, Operand(rax, 5));
|
|
|
|
|
__ movddup(xmm1, xmm2);
|
2016-09-12 06:24:06 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#define EMIT_SSE34_INSTR(instruction, notUsed1, notUsed2, notUsed3, notUsed4) \
|
|
|
|
|
__ instruction(xmm5, xmm1); \
|
|
|
|
|
__ instruction(xmm5, Operand(rdx, 4));
|
|
|
|
|
|
2019-10-30 04:38:36 +00:00
|
|
|
#define EMIT_SSE34_IMM_INSTR(instruction, notUsed1, notUsed2, notUsed3, \
|
|
|
|
|
notUsed4) \
|
|
|
|
|
__ instruction(rbx, xmm15, 0); \
|
|
|
|
|
__ instruction(Operand(rax, 10), xmm0, 1);
|
|
|
|
|
|
2016-09-12 06:24:06 +00:00
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(SSSE3)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, SSSE3);
|
2018-08-14 19:38:14 +00:00
|
|
|
__ palignr(xmm5, xmm1, 5);
|
|
|
|
|
__ palignr(xmm5, Operand(rdx, 4), 5);
|
2016-09-12 06:24:06 +00:00
|
|
|
SSSE3_INSTRUCTION_LIST(EMIT_SSE34_INSTR)
|
2020-02-19 10:26:51 +00:00
|
|
|
SSSE3_UNOP_INSTRUCTION_LIST(EMIT_SSE34_INSTR)
|
2016-09-12 06:24:06 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2013-10-15 12:51:58 +00:00
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(SSE4_1)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, SSE4_1);
|
2016-05-19 19:52:59 +00:00
|
|
|
__ insertps(xmm5, xmm1, 123);
|
[wasm] Implement wasm x64 I16x8 Ops
- Add I16x8 Splat, ExtractLane, ReplaceLane, shift ops, Some BinOps and compare ops
- Add pshufhw, pshuflw in the assembler, disassembler
- Fix incorrect modrm for pextrw, this bug disregards the register allocated and always makes pextrw use rax.
- Fix pextrw disasm to take the 0 - 7 bits of the immediate instead of 0 - 3.
- Pextrw, pinsrw are in the assembler use 128 bit encodings, pextrw, pinsrw in the disassembler use legacy encodings, fix inconsistencies causing weird code gen when --print-code is used.
Review-Url: https://codereview.chromium.org/2767983002
Cr-Commit-Position: refs/heads/master@{#44664}
2017-04-17 18:47:46 +00:00
|
|
|
__ pinsrw(xmm2, rcx, 1);
|
2019-07-01 23:58:02 +00:00
|
|
|
__ pextrq(r12, xmm0, 1);
|
2015-03-05 09:22:26 +00:00
|
|
|
__ pinsrd(xmm9, r9, 0);
|
2016-05-19 19:52:59 +00:00
|
|
|
__ pinsrd(xmm5, Operand(rax, 4), 1);
|
2019-07-01 23:58:02 +00:00
|
|
|
__ pinsrq(xmm9, r9, 0);
|
|
|
|
|
__ pinsrq(xmm5, Operand(rax, 4), 1);
|
2018-08-14 19:38:14 +00:00
|
|
|
__ pblendw(xmm5, xmm1, 1);
|
|
|
|
|
__ pblendw(xmm9, Operand(rax, 4), 1);
|
2016-05-19 19:52:59 +00:00
|
|
|
|
|
|
|
|
__ cmpps(xmm5, xmm1, 1);
|
2016-08-03 15:23:57 +00:00
|
|
|
__ cmpps(xmm5, Operand(rbx, rcx, times_4, 10000), 1);
|
2016-05-19 19:52:59 +00:00
|
|
|
__ cmpeqps(xmm5, xmm1);
|
2016-08-03 15:23:57 +00:00
|
|
|
__ cmpeqps(xmm5, Operand(rbx, rcx, times_4, 10000));
|
2016-05-19 19:52:59 +00:00
|
|
|
__ cmpltps(xmm5, xmm1);
|
2016-08-03 15:23:57 +00:00
|
|
|
__ cmpltps(xmm5, Operand(rbx, rcx, times_4, 10000));
|
2016-05-19 19:52:59 +00:00
|
|
|
__ cmpleps(xmm5, xmm1);
|
2016-08-03 15:23:57 +00:00
|
|
|
__ cmpleps(xmm5, Operand(rbx, rcx, times_4, 10000));
|
2016-05-19 19:52:59 +00:00
|
|
|
__ cmpneqps(xmm5, xmm1);
|
2016-08-03 15:23:57 +00:00
|
|
|
__ cmpneqps(xmm5, Operand(rbx, rcx, times_4, 10000));
|
2016-05-19 19:52:59 +00:00
|
|
|
__ cmpnltps(xmm5, xmm1);
|
2016-08-03 15:23:57 +00:00
|
|
|
__ cmpnltps(xmm5, Operand(rbx, rcx, times_4, 10000));
|
2016-05-19 19:52:59 +00:00
|
|
|
__ cmpnleps(xmm5, xmm1);
|
2016-08-03 15:23:57 +00:00
|
|
|
__ cmpnleps(xmm5, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cmppd(xmm5, xmm1, 1);
|
|
|
|
|
__ cmppd(xmm5, Operand(rbx, rcx, times_4, 10000), 1);
|
|
|
|
|
__ cmpeqpd(xmm5, xmm1);
|
|
|
|
|
__ cmpeqpd(xmm5, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cmpltpd(xmm5, xmm1);
|
|
|
|
|
__ cmpltpd(xmm5, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cmplepd(xmm5, xmm1);
|
|
|
|
|
__ cmplepd(xmm5, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cmpneqpd(xmm5, xmm1);
|
|
|
|
|
__ cmpneqpd(xmm5, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cmpnltpd(xmm5, xmm1);
|
|
|
|
|
__ cmpnltpd(xmm5, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ cmpnlepd(xmm5, xmm1);
|
|
|
|
|
__ cmpnlepd(xmm5, Operand(rbx, rcx, times_4, 10000));
|
2016-05-19 19:52:59 +00:00
|
|
|
|
|
|
|
|
__ movups(xmm5, xmm1);
|
|
|
|
|
__ movups(xmm5, Operand(rdx, 4));
|
|
|
|
|
__ movups(Operand(rdx, 4), xmm5);
|
2019-10-16 20:21:00 +00:00
|
|
|
__ movlhps(xmm5, xmm1);
|
2016-05-19 19:52:59 +00:00
|
|
|
__ pmulld(xmm5, xmm1);
|
|
|
|
|
__ pmulld(xmm5, Operand(rdx, 4));
|
2016-09-12 06:24:06 +00:00
|
|
|
__ pmullw(xmm5, xmm1);
|
|
|
|
|
__ pmullw(xmm5, Operand(rdx, 4));
|
2016-05-19 19:52:59 +00:00
|
|
|
__ pmuludq(xmm5, xmm1);
|
|
|
|
|
__ pmuludq(xmm5, Operand(rdx, 4));
|
|
|
|
|
__ psrldq(xmm5, 123);
|
|
|
|
|
__ pshufd(xmm5, xmm1, 3);
|
|
|
|
|
__ cvtps2dq(xmm5, xmm1);
|
|
|
|
|
__ cvtps2dq(xmm5, Operand(rdx, 4));
|
|
|
|
|
__ cvtdq2ps(xmm5, xmm1);
|
|
|
|
|
__ cvtdq2ps(xmm5, Operand(rdx, 4));
|
2020-10-28 02:43:49 +00:00
|
|
|
|
|
|
|
|
__ pblendvb(xmm5, xmm1);
|
|
|
|
|
__ blendvps(xmm5, xmm1);
|
|
|
|
|
__ blendvps(xmm5, Operand(rdx, 4));
|
2020-01-10 17:54:44 +00:00
|
|
|
__ blendvpd(xmm5, xmm1);
|
|
|
|
|
__ blendvpd(xmm5, Operand(rdx, 4));
|
2016-09-12 06:24:06 +00:00
|
|
|
|
2020-06-02 21:11:49 +00:00
|
|
|
__ roundps(xmm8, xmm3, kRoundUp);
|
2020-06-08 16:45:43 +00:00
|
|
|
__ roundpd(xmm8, xmm3, kRoundToNearest);
|
2020-04-06 16:44:23 +00:00
|
|
|
__ roundss(xmm8, xmm3, kRoundDown);
|
|
|
|
|
__ roundsd(xmm8, xmm3, kRoundDown);
|
|
|
|
|
|
2016-09-12 06:24:06 +00:00
|
|
|
SSE4_INSTRUCTION_LIST(EMIT_SSE34_INSTR)
|
2020-02-27 08:45:39 +00:00
|
|
|
SSE4_UNOP_INSTRUCTION_LIST(EMIT_SSE34_INSTR)
|
2019-10-30 04:38:36 +00:00
|
|
|
SSE4_EXTRACT_INSTRUCTION_LIST(EMIT_SSE34_IMM_INSTR)
|
2013-10-15 12:51:58 +00:00
|
|
|
}
|
|
|
|
|
}
|
2019-07-16 08:50:27 +00:00
|
|
|
|
|
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(SSE4_2)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, SSE4_2);
|
|
|
|
|
|
|
|
|
|
SSE4_2_INSTRUCTION_LIST(EMIT_SSE34_INSTR)
|
|
|
|
|
}
|
|
|
|
|
}
|
2016-09-12 06:24:06 +00:00
|
|
|
#undef EMIT_SSE34_INSTR
|
2019-10-30 04:38:36 +00:00
|
|
|
#undef EMIT_SSE34_IMM_INSTR
|
2013-10-15 12:51:58 +00:00
|
|
|
|
2014-12-01 10:45:02 +00:00
|
|
|
// AVX instruction
|
|
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(AVX)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, AVX);
|
2015-10-20 04:58:07 +00:00
|
|
|
__ vmovss(xmm6, xmm14, xmm2);
|
|
|
|
|
__ vmovss(xmm9, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vmovss(Operand(rbx, rcx, times_4, 10000), xmm0);
|
|
|
|
|
|
2015-03-30 07:33:46 +00:00
|
|
|
__ vaddss(xmm0, xmm1, xmm2);
|
|
|
|
|
__ vaddss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vmulss(xmm0, xmm1, xmm2);
|
|
|
|
|
__ vmulss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vsubss(xmm0, xmm1, xmm2);
|
|
|
|
|
__ vsubss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vdivss(xmm0, xmm1, xmm2);
|
|
|
|
|
__ vdivss(xmm0, xmm1, Operand(rbx, rcx, times_2, 10000));
|
|
|
|
|
__ vminss(xmm8, xmm1, xmm2);
|
|
|
|
|
__ vminss(xmm9, xmm1, Operand(rbx, rcx, times_8, 10000));
|
|
|
|
|
__ vmaxss(xmm8, xmm1, xmm2);
|
|
|
|
|
__ vmaxss(xmm9, xmm1, Operand(rbx, rcx, times_1, 10000));
|
2018-03-03 08:53:21 +00:00
|
|
|
__ vsqrtss(xmm8, xmm1, xmm2);
|
|
|
|
|
__ vsqrtss(xmm9, xmm1, Operand(rbx, rcx, times_1, 10000));
|
2015-10-19 20:35:17 +00:00
|
|
|
__ vmovss(xmm9, Operand(r11, rcx, times_8, -10000));
|
|
|
|
|
__ vmovss(Operand(rbx, r9, times_4, 10000), xmm1);
|
2015-03-30 07:33:46 +00:00
|
|
|
__ vucomiss(xmm9, xmm1);
|
|
|
|
|
__ vucomiss(xmm8, Operand(rbx, rdx, times_2, 10981));
|
|
|
|
|
|
2015-10-16 05:20:31 +00:00
|
|
|
__ vmovd(xmm5, rdi);
|
|
|
|
|
__ vmovd(xmm9, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vmovd(r9, xmm6);
|
|
|
|
|
__ vmovq(xmm5, rdi);
|
|
|
|
|
__ vmovq(xmm9, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vmovq(r9, xmm6);
|
|
|
|
|
|
2015-10-20 04:58:07 +00:00
|
|
|
__ vmovsd(xmm6, xmm14, xmm2);
|
2015-10-15 04:14:31 +00:00
|
|
|
__ vmovsd(xmm9, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vmovsd(Operand(rbx, rcx, times_4, 10000), xmm0);
|
|
|
|
|
|
2019-10-18 23:58:10 +00:00
|
|
|
__ vmovdqu(xmm9, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vmovdqu(Operand(rbx, rcx, times_4, 10000), xmm0);
|
|
|
|
|
|
2020-10-14 16:33:41 +00:00
|
|
|
__ vmovlps(xmm8, xmm9, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vmovlps(Operand(rbx, rcx, times_4, 10000), xmm9);
|
|
|
|
|
__ vmovhps(xmm8, xmm9, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vmovhps(Operand(rbx, rcx, times_4, 10000), xmm12);
|
|
|
|
|
|
2020-06-02 21:11:49 +00:00
|
|
|
__ vroundps(xmm9, xmm2, kRoundUp);
|
2020-06-08 16:45:43 +00:00
|
|
|
__ vroundpd(xmm9, xmm2, kRoundToNearest);
|
2020-01-10 17:55:04 +00:00
|
|
|
__ vroundss(xmm9, xmm1, xmm2, kRoundDown);
|
2015-10-23 07:48:54 +00:00
|
|
|
__ vroundsd(xmm8, xmm3, xmm0, kRoundDown);
|
2015-03-30 07:33:46 +00:00
|
|
|
__ vucomisd(xmm9, xmm1);
|
|
|
|
|
__ vucomisd(xmm8, Operand(rbx, rdx, times_2, 10981));
|
2015-10-18 15:43:07 +00:00
|
|
|
|
2015-10-18 16:11:11 +00:00
|
|
|
__ vcvtss2sd(xmm4, xmm9, xmm11);
|
|
|
|
|
__ vcvtss2sd(xmm4, xmm9, Operand(rbx, rcx, times_1, 10000));
|
2020-03-03 04:46:45 +00:00
|
|
|
__ vcvttps2dq(xmm4, xmm11);
|
2015-10-15 17:50:03 +00:00
|
|
|
__ vcvtlsi2sd(xmm5, xmm9, rcx);
|
|
|
|
|
__ vcvtlsi2sd(xmm9, xmm3, Operand(rbx, r9, times_4, 10000));
|
2015-10-24 16:49:35 +00:00
|
|
|
__ vcvtqsi2sd(xmm5, xmm9, r11);
|
2015-10-18 15:43:07 +00:00
|
|
|
__ vcvttsd2si(r9, xmm6);
|
|
|
|
|
__ vcvttsd2si(rax, Operand(rbx, r9, times_4, 10000));
|
|
|
|
|
__ vcvttsd2siq(rdi, xmm9);
|
|
|
|
|
__ vcvttsd2siq(r8, Operand(r9, rbx, times_4, 10000));
|
2015-10-24 18:55:27 +00:00
|
|
|
__ vcvtsd2si(rdi, xmm9);
|
2015-04-11 01:02:22 +00:00
|
|
|
|
2015-10-20 04:58:07 +00:00
|
|
|
__ vmovaps(xmm10, xmm11);
|
2015-10-18 16:12:24 +00:00
|
|
|
__ vmovapd(xmm7, xmm0);
|
2016-07-21 11:01:44 +00:00
|
|
|
__ vmovupd(xmm0, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vmovupd(Operand(rbx, rcx, times_4, 10000), xmm0);
|
2015-10-18 16:12:24 +00:00
|
|
|
__ vmovmskpd(r9, xmm4);
|
2020-04-06 16:42:49 +00:00
|
|
|
__ vpmovmskb(r10, xmm9);
|
2015-10-18 16:12:24 +00:00
|
|
|
|
2016-07-11 18:18:17 +00:00
|
|
|
__ vmovups(xmm5, xmm1);
|
|
|
|
|
__ vmovups(xmm5, Operand(rdx, 4));
|
|
|
|
|
__ vmovups(Operand(rdx, 4), xmm5);
|
2020-01-08 19:45:32 +00:00
|
|
|
__ vmovlhps(xmm1, xmm3, xmm5);
|
2016-07-11 18:18:17 +00:00
|
|
|
|
2015-04-11 01:02:22 +00:00
|
|
|
__ vandps(xmm0, xmm9, xmm2);
|
|
|
|
|
__ vandps(xmm9, xmm1, Operand(rbx, rcx, times_4, 10000));
|
2019-03-21 20:30:49 +00:00
|
|
|
__ vandnps(xmm0, xmm9, xmm2);
|
|
|
|
|
__ vandnps(xmm9, xmm1, Operand(rbx, rcx, times_4, 10000));
|
2015-04-11 01:02:22 +00:00
|
|
|
__ vxorps(xmm0, xmm1, xmm9);
|
|
|
|
|
__ vxorps(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
|
2018-02-28 06:28:08 +00:00
|
|
|
__ vhaddps(xmm0, xmm1, xmm9);
|
|
|
|
|
__ vhaddps(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
|
2015-04-11 01:02:22 +00:00
|
|
|
|
2015-10-22 07:00:57 +00:00
|
|
|
__ vpcmpeqd(xmm0, xmm15, xmm5);
|
|
|
|
|
__ vpcmpeqd(xmm15, xmm0, Operand(rbx, rcx, times_4, 10000));
|
2016-08-03 15:23:57 +00:00
|
|
|
|
|
|
|
|
__ vcmpps(xmm5, xmm4, xmm1, 1);
|
|
|
|
|
__ vcmpps(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000), 1);
|
|
|
|
|
__ vcmpeqps(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpeqps(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmpltps(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpltps(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmpleps(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpleps(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmpneqps(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpneqps(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmpnltps(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpnltps(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmpnleps(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpnleps(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmppd(xmm5, xmm4, xmm1, 1);
|
|
|
|
|
__ vcmppd(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000), 1);
|
|
|
|
|
__ vcmpeqpd(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpeqpd(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmpltpd(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpltpd(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmplepd(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmplepd(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmpneqpd(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpneqpd(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmpnltpd(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpnltpd(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vcmpnlepd(xmm5, xmm4, xmm1);
|
|
|
|
|
__ vcmpnlepd(xmm5, xmm4, Operand(rbx, rcx, times_4, 10000));
|
2016-09-12 06:24:06 +00:00
|
|
|
|
2019-11-25 19:26:01 +00:00
|
|
|
#define EMIT_SSE_UNOP_AVXINSTR(instruction, notUsed1, notUsed2) \
|
|
|
|
|
__ v##instruction(xmm10, xmm1); \
|
|
|
|
|
__ v##instruction(xmm10, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
|
|
|
|
|
SSE_UNOP_INSTRUCTION_LIST(EMIT_SSE_UNOP_AVXINSTR)
|
|
|
|
|
#undef EMIT_SSE_UNOP_AVXINSTR
|
|
|
|
|
|
|
|
|
|
#define EMIT_SSE_BINOP_AVXINSTR(instruction, notUsed1, notUsed2) \
|
|
|
|
|
__ v##instruction(xmm10, xmm5, xmm1); \
|
|
|
|
|
__ v##instruction(xmm10, xmm5, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
|
|
|
|
|
SSE_BINOP_INSTRUCTION_LIST(EMIT_SSE_BINOP_AVXINSTR)
|
|
|
|
|
#undef EMIT_SSE_BINOP_AVXINSTR
|
|
|
|
|
|
2016-09-12 06:24:06 +00:00
|
|
|
#define EMIT_SSE2_AVXINSTR(instruction, notUsed1, notUsed2, notUsed3) \
|
|
|
|
|
__ v##instruction(xmm10, xmm5, xmm1); \
|
|
|
|
|
__ v##instruction(xmm10, xmm5, Operand(rdx, 4));
|
|
|
|
|
|
Fix assembler for sqrtpd
The assembly of sqrtpd when using Sqrtpd macro was wrong, since
Sqrtpd(xmm1, xmm1) will incorrect generated vsqrtpd(xmm1, xmm1, xmm1),
which is nonsensical, since vsqrtpd only takes two operands. The
expected instruction should be vsqrtpd(xmm1, xmm0, xmm1) in terms of the
encoding, which is vsqrtpd(xmm1, xmm1).
So, move sqrtpd and cvtps2dq out into their own macro list, because
they have two operands in their AVX form, unlike the rest of the
instructions in SSE2_INSTRUCTION_LIST.
Also updated disasm and tests to use this new list.
Fixed: v8:10170
Change-Id: Ia9343c9a3ae64596bbc876744556e1dcea2a443b
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2032195
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#66088}
2020-01-31 18:28:11 +00:00
|
|
|
SSE2_INSTRUCTION_LIST(EMIT_SSE2_AVXINSTR)
|
|
|
|
|
#undef EMIT_SSE2_AVXINSTR
|
|
|
|
|
|
|
|
|
|
#define EMIT_SSE2_UNOP_AVXINSTR(instruction, notUsed1, notUsed2, notUsed3) \
|
|
|
|
|
__ v##instruction(xmm10, xmm1); \
|
|
|
|
|
__ v##instruction(xmm10, Operand(rdx, 4));
|
|
|
|
|
|
|
|
|
|
SSE2_UNOP_INSTRUCTION_LIST(EMIT_SSE2_UNOP_AVXINSTR)
|
|
|
|
|
#undef EMIT_SSE2_AVXINSTR
|
|
|
|
|
|
2020-02-03 22:51:57 +00:00
|
|
|
#define EMIT_SSE2_SD_AVXINSTR(instruction, notUsed1, notUsed2, notUsed3) \
|
|
|
|
|
__ v##instruction(xmm10, xmm5, xmm1); \
|
|
|
|
|
__ v##instruction(xmm10, xmm5, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
SSE2_INSTRUCTION_LIST_SD(EMIT_SSE2_SD_AVXINSTR)
|
|
|
|
|
#undef EMIT_SSE2_SD_AVXINSTR
|
|
|
|
|
|
2016-09-12 06:24:06 +00:00
|
|
|
#define EMIT_SSE34_AVXINSTR(instruction, notUsed1, notUsed2, notUsed3, \
|
|
|
|
|
notUsed4) \
|
|
|
|
|
__ v##instruction(xmm10, xmm5, xmm1); \
|
|
|
|
|
__ v##instruction(xmm10, xmm5, Operand(rdx, 4));
|
|
|
|
|
|
|
|
|
|
SSSE3_INSTRUCTION_LIST(EMIT_SSE34_AVXINSTR)
|
|
|
|
|
SSE4_INSTRUCTION_LIST(EMIT_SSE34_AVXINSTR)
|
2020-01-10 17:54:18 +00:00
|
|
|
SSE4_2_INSTRUCTION_LIST(EMIT_SSE34_AVXINSTR)
|
2016-09-12 06:24:06 +00:00
|
|
|
#undef EMIT_SSE34_AVXINSTR
|
|
|
|
|
|
2020-02-19 10:26:51 +00:00
|
|
|
#define EMIT_SSSE3_UNOP_AVXINSTR(instruction, notUsed1, notUsed2, notUsed3, \
|
|
|
|
|
notUsed4) \
|
|
|
|
|
__ v##instruction(xmm9, xmm3); \
|
|
|
|
|
__ v##instruction(xmm9, Operand(rdx, 5));
|
|
|
|
|
SSSE3_UNOP_INSTRUCTION_LIST(EMIT_SSSE3_UNOP_AVXINSTR)
|
|
|
|
|
#undef EMIT_SSSE3_UNOP_AVXINSTR
|
|
|
|
|
|
2020-01-06 18:51:23 +00:00
|
|
|
#define EMIT_SSE4_PMOV_AVXINSTR(instruction, notUsed1, notUsed2, notUsed3, \
|
|
|
|
|
notUsed4) \
|
|
|
|
|
__ v##instruction(xmm10, xmm1); \
|
|
|
|
|
__ v##instruction(xmm10, Operand(rdx, 4));
|
2020-02-27 08:45:39 +00:00
|
|
|
SSE4_UNOP_INSTRUCTION_LIST(EMIT_SSE4_PMOV_AVXINSTR)
|
2020-01-06 18:51:23 +00:00
|
|
|
#undef EMIT_SSE4_PMOV_AVXINSTR
|
|
|
|
|
|
2019-12-04 14:56:48 +00:00
|
|
|
#define EMIT_SSE2_SHIFT_IMM_AVX(instruction, notUsed1, notUsed2, notUsed3, \
|
|
|
|
|
notUsed4) \
|
|
|
|
|
__ v##instruction(xmm0, xmm15, 21);
|
|
|
|
|
SSE2_INSTRUCTION_LIST_SHIFT_IMM(EMIT_SSE2_SHIFT_IMM_AVX)
|
|
|
|
|
#undef EMIT_SSE2_SHIFT_IMM_AVX
|
|
|
|
|
|
2019-10-23 20:58:25 +00:00
|
|
|
__ vinsertps(xmm1, xmm2, xmm3, 1);
|
|
|
|
|
__ vinsertps(xmm1, xmm2, Operand(rbx, rcx, times_4, 10000), 1);
|
|
|
|
|
__ vextractps(rax, xmm1, 1);
|
|
|
|
|
|
2016-09-12 06:24:06 +00:00
|
|
|
__ vlddqu(xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vpextrb(rax, xmm2, 12);
|
|
|
|
|
__ vpextrb(Operand(rbx, rcx, times_4, 10000), xmm2, 12);
|
|
|
|
|
__ vpextrw(rax, xmm2, 5);
|
|
|
|
|
__ vpextrw(Operand(rbx, rcx, times_4, 10000), xmm2, 5);
|
|
|
|
|
__ vpextrd(rax, xmm2, 2);
|
|
|
|
|
__ vpextrd(Operand(rbx, rcx, times_4, 10000), xmm2, 2);
|
2020-01-07 23:16:33 +00:00
|
|
|
__ vpextrq(rax, xmm2, 2);
|
2016-09-12 06:24:06 +00:00
|
|
|
|
|
|
|
|
__ vpinsrb(xmm1, xmm2, rax, 12);
|
|
|
|
|
__ vpinsrb(xmm1, xmm2, Operand(rbx, rcx, times_4, 10000), 12);
|
|
|
|
|
__ vpinsrw(xmm1, xmm2, rax, 5);
|
|
|
|
|
__ vpinsrw(xmm1, xmm2, Operand(rbx, rcx, times_4, 10000), 5);
|
|
|
|
|
__ vpinsrd(xmm1, xmm2, rax, 2);
|
|
|
|
|
__ vpinsrd(xmm1, xmm2, Operand(rbx, rcx, times_4, 10000), 2);
|
2019-12-04 11:15:11 +00:00
|
|
|
__ vpinsrq(xmm1, xmm2, rax, 9);
|
|
|
|
|
__ vpinsrq(xmm1, xmm2, Operand(rbx, rcx, times_4, 10000), 9);
|
2016-09-12 06:24:06 +00:00
|
|
|
__ vpshufd(xmm1, xmm2, 85);
|
2020-03-20 00:05:30 +00:00
|
|
|
__ vpshufd(xmm1, Operand(rbx, rcx, times_4, 10000), 85);
|
2019-12-02 09:20:32 +00:00
|
|
|
__ vpshuflw(xmm1, xmm2, 85);
|
|
|
|
|
__ vpshuflw(xmm1, Operand(rbx, rcx, times_4, 10000), 85);
|
2020-03-20 00:05:30 +00:00
|
|
|
__ vpshufhw(xmm1, xmm2, 85);
|
|
|
|
|
__ vpshufhw(xmm1, Operand(rbx, rcx, times_4, 10000), 85);
|
2019-10-19 00:44:25 +00:00
|
|
|
__ vshufps(xmm3, xmm2, xmm3, 3);
|
2019-12-09 09:17:33 +00:00
|
|
|
__ vpblendw(xmm1, xmm2, xmm3, 23);
|
|
|
|
|
__ vpblendw(xmm1, xmm2, Operand(rbx, rcx, times_4, 10000), 23);
|
2020-01-06 19:02:52 +00:00
|
|
|
__ vpalignr(xmm1, xmm2, xmm3, 4);
|
2020-08-25 17:15:47 +00:00
|
|
|
__ vpalignr(xmm1, xmm2, Operand(rbx, rcx, times_4, 10000), 4);
|
2019-12-02 09:20:32 +00:00
|
|
|
|
2020-10-28 02:43:49 +00:00
|
|
|
__ vpblendvb(xmm1, xmm2, xmm3, xmm4);
|
|
|
|
|
__ vblendvps(xmm1, xmm2, xmm3, xmm4);
|
2020-01-10 17:54:44 +00:00
|
|
|
__ vblendvpd(xmm1, xmm2, xmm3, xmm4);
|
|
|
|
|
|
2019-12-04 11:15:11 +00:00
|
|
|
__ vmovddup(xmm1, xmm2);
|
2019-12-02 09:20:32 +00:00
|
|
|
__ vmovddup(xmm1, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ vbroadcastss(xmm1, Operand(rbx, rcx, times_4, 10000));
|
2014-12-01 10:45:02 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2014-11-26 05:31:41 +00:00
|
|
|
// FMA3 instruction
|
|
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(FMA3)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, FMA3);
|
2020-01-02 23:24:28 +00:00
|
|
|
#define EMIT_FMA(instr, notUsed1, notUsed2, notUsed3, notUsed4, notUsed5, \
|
|
|
|
|
notUsed6) \
|
|
|
|
|
__ instr(xmm9, xmm10, xmm11); \
|
|
|
|
|
__ instr(xmm9, xmm10, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
FMA_INSTRUCTION_LIST(EMIT_FMA)
|
|
|
|
|
#undef EMIT_FMA
|
2014-11-26 05:31:41 +00:00
|
|
|
}
|
|
|
|
|
}
|
2015-04-08 07:15:51 +00:00
|
|
|
|
|
|
|
|
// BMI1 instructions
|
|
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(BMI1)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, BMI1);
|
|
|
|
|
__ andnq(rax, rbx, rcx);
|
|
|
|
|
__ andnq(rax, rbx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ andnl(rax, rbx, rcx);
|
|
|
|
|
__ andnl(rax, rbx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ bextrq(rax, rbx, rcx);
|
|
|
|
|
__ bextrq(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ bextrl(rax, rbx, rcx);
|
|
|
|
|
__ bextrl(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ blsiq(rax, rbx);
|
|
|
|
|
__ blsiq(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ blsil(rax, rbx);
|
|
|
|
|
__ blsil(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ blsmskq(rax, rbx);
|
|
|
|
|
__ blsmskq(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ blsmskl(rax, rbx);
|
|
|
|
|
__ blsmskl(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ blsrq(rax, rbx);
|
|
|
|
|
__ blsrq(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ blsrl(rax, rbx);
|
|
|
|
|
__ blsrl(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ tzcntq(rax, rbx);
|
|
|
|
|
__ tzcntq(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ tzcntl(rax, rbx);
|
|
|
|
|
__ tzcntl(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// LZCNT instructions
|
|
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(LZCNT)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, LZCNT);
|
|
|
|
|
__ lzcntq(rax, rbx);
|
|
|
|
|
__ lzcntq(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ lzcntl(rax, rbx);
|
|
|
|
|
__ lzcntl(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// POPCNT instructions
|
|
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(POPCNT)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, POPCNT);
|
|
|
|
|
__ popcntq(rax, rbx);
|
|
|
|
|
__ popcntq(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ popcntl(rax, rbx);
|
|
|
|
|
__ popcntl(rax, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// BMI2 instructions
|
|
|
|
|
{
|
|
|
|
|
if (CpuFeatures::IsSupported(BMI2)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, BMI2);
|
|
|
|
|
__ bzhiq(rax, rbx, rcx);
|
|
|
|
|
__ bzhiq(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ bzhil(rax, rbx, rcx);
|
|
|
|
|
__ bzhil(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ mulxq(rax, rbx, rcx);
|
|
|
|
|
__ mulxq(rax, rbx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ mulxl(rax, rbx, rcx);
|
|
|
|
|
__ mulxl(rax, rbx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ pdepq(rax, rbx, rcx);
|
|
|
|
|
__ pdepq(rax, rbx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ pdepl(rax, rbx, rcx);
|
|
|
|
|
__ pdepl(rax, rbx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ pextq(rax, rbx, rcx);
|
|
|
|
|
__ pextq(rax, rbx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ pextl(rax, rbx, rcx);
|
|
|
|
|
__ pextl(rax, rbx, Operand(rbx, rcx, times_4, 10000));
|
|
|
|
|
__ sarxq(rax, rbx, rcx);
|
|
|
|
|
__ sarxq(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ sarxl(rax, rbx, rcx);
|
|
|
|
|
__ sarxl(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ shlxq(rax, rbx, rcx);
|
|
|
|
|
__ shlxq(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ shlxl(rax, rbx, rcx);
|
|
|
|
|
__ shlxl(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ shrxq(rax, rbx, rcx);
|
|
|
|
|
__ shrxq(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ shrxl(rax, rbx, rcx);
|
|
|
|
|
__ shrxl(rax, Operand(rbx, rcx, times_4, 10000), rbx);
|
|
|
|
|
__ rorxq(rax, rbx, 63);
|
|
|
|
|
__ rorxq(rax, Operand(rbx, rcx, times_4, 10000), 63);
|
|
|
|
|
__ rorxl(rax, rbx, 31);
|
|
|
|
|
__ rorxl(rax, Operand(rbx, rcx, times_4, 10000), 31);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2014-07-30 13:54:45 +00:00
|
|
|
// xchg.
|
|
|
|
|
{
|
2016-05-19 00:55:08 +00:00
|
|
|
__ xchgb(rax, Operand(rax, 8));
|
|
|
|
|
__ xchgw(rax, Operand(rbx, 8));
|
2014-07-30 13:54:45 +00:00
|
|
|
__ xchgq(rax, rax);
|
|
|
|
|
__ xchgq(rax, rbx);
|
|
|
|
|
__ xchgq(rbx, rbx);
|
|
|
|
|
__ xchgq(rbx, Operand(rsp, 12));
|
|
|
|
|
}
|
|
|
|
|
|
2016-05-19 00:55:08 +00:00
|
|
|
// cmpxchg.
|
|
|
|
|
{
|
|
|
|
|
__ cmpxchgb(Operand(rsp, 12), rax);
|
|
|
|
|
__ cmpxchgw(Operand(rbx, rcx, times_4, 10000), rax);
|
|
|
|
|
__ cmpxchgl(Operand(rbx, rcx, times_4, 10000), rax);
|
|
|
|
|
__ cmpxchgq(Operand(rbx, rcx, times_4, 10000), rax);
|
|
|
|
|
}
|
|
|
|
|
|
2020-01-20 17:33:17 +00:00
|
|
|
// xadd.
|
|
|
|
|
{
|
|
|
|
|
__ xaddb(Operand(rsp, 12), rax);
|
|
|
|
|
__ xaddw(Operand(rsp, 12), rax);
|
|
|
|
|
__ xaddl(Operand(rsp, 12), rax);
|
|
|
|
|
__ xaddq(Operand(rsp, 12), rax);
|
|
|
|
|
__ xaddb(Operand(rbx, rcx, times_4, 10000), rax);
|
|
|
|
|
__ xaddw(Operand(rbx, rcx, times_4, 10000), rax);
|
|
|
|
|
__ xaddl(Operand(rbx, rcx, times_4, 10000), rax);
|
|
|
|
|
__ xaddq(Operand(rbx, rcx, times_4, 10000), rax);
|
|
|
|
|
}
|
|
|
|
|
|
2016-05-19 00:55:08 +00:00
|
|
|
// lock prefix.
|
|
|
|
|
{
|
|
|
|
|
__ lock();
|
|
|
|
|
__ cmpxchgl(Operand(rsp, 12), rbx);
|
|
|
|
|
|
|
|
|
|
__ lock();
|
|
|
|
|
__ xchgw(rax, Operand(rcx, 8));
|
|
|
|
|
}
|
|
|
|
|
|
2011-12-05 08:58:01 +00:00
|
|
|
// Nop instructions
|
|
|
|
|
for (int i = 0; i < 16; i++) {
|
|
|
|
|
__ Nop(i);
|
|
|
|
|
}
|
|
|
|
|
|
2019-07-18 09:10:30 +00:00
|
|
|
__ mfence();
|
|
|
|
|
__ lfence();
|
2018-05-09 15:02:15 +00:00
|
|
|
__ pause();
|
2011-12-05 08:58:01 +00:00
|
|
|
__ ret(0);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
Reland "Reland "[deoptimizer] Change deopt entries into builtins""
This is a reland of fbfa9bf4ec72b1b73a96b70ccb68cd98c321511b
The arm64 was missing proper codegen for CFI, thus sizes were off.
Original change's description:
> Reland "[deoptimizer] Change deopt entries into builtins"
>
> This is a reland of 7f58ced72eb65b6b5530ccabaf2eaebe45bf9d33
>
> It fixes the different exit size emitted on x64/Atom CPUs due to
> performance tuning in TurboAssembler::Call. Additionally, add
> cctests to verify the fixed size exits.
>
> Original change's description:
> > [deoptimizer] Change deopt entries into builtins
> >
> > While the overall goal of this commit is to change deoptimization
> > entries into builtins, there are multiple related things happening:
> >
> > - Deoptimization entries, formerly stubs (i.e. Code objects generated
> > at runtime, guaranteed to be immovable), have been converted into
> > builtins. The major restriction is that we now need to preserve the
> > kRootRegister, which was formerly used on most architectures to pass
> > the deoptimization id. The solution differs based on platform.
> > - Renamed DEOPT_ENTRIES_OR_FOR_TESTING code kind to FOR_TESTING.
> > - Removed heap/ support for immovable Code generation.
> > - Removed the DeserializerData class (no longer needed).
> > - arm64: to preserve 4-byte deopt exits, introduced a new optimization
> > in which the final jump to the deoptimization entry is generated
> > once per Code object, and deopt exits can continue to emit a
> > near-call.
> > - arm,ia32,x64: change to fixed-size deopt exits. This reduces exit
> > sizes by 4/8, 5, and 5 bytes, respectively.
> >
> > On arm the deopt exit size is reduced from 12 (or 16) bytes to 8 bytes
> > by using the same strategy as on arm64 (recalc deopt id from return
> > address). Before:
> >
> > e300a002 movw r10, <id>
> > e59fc024 ldr ip, [pc, <entry offset>]
> > e12fff3c blx ip
> >
> > After:
> >
> > e59acb35 ldr ip, [r10, <entry offset>]
> > e12fff3c blx ip
> >
> > On arm64 the deopt exit size remains 4 bytes (or 8 bytes in same cases
> > with CFI). Additionally, up to 4 builtin jumps are emitted per Code
> > object (max 32 bytes added overhead per Code object). Before:
> >
> > 9401cdae bl <entry offset>
> >
> > After:
> >
> > # eager deoptimization entry jump.
> > f95b1f50 ldr x16, [x26, <eager entry offset>]
> > d61f0200 br x16
> > # lazy deoptimization entry jump.
> > f95b2b50 ldr x16, [x26, <lazy entry offset>]
> > d61f0200 br x16
> > # the deopt exit.
> > 97fffffc bl <eager deoptimization entry jump offset>
> >
> > On ia32 the deopt exit size is reduced from 10 to 5 bytes. Before:
> >
> > bb00000000 mov ebx,<id>
> > e825f5372b call <entry>
> >
> > After:
> >
> > e8ea2256ba call <entry>
> >
> > On x64 the deopt exit size is reduced from 12 to 7 bytes. Before:
> >
> > 49c7c511000000 REX.W movq r13,<id>
> > e8ea2f0700 call <entry>
> >
> > After:
> >
> > 41ff9560360000 call [r13+<entry offset>]
> >
> > Bug: v8:8661,v8:8768
> > Change-Id: I13e30aedc360474dc818fecc528ce87c3bfeed42
> > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2465834
> > Commit-Queue: Jakob Gruber <jgruber@chromium.org>
> > Reviewed-by: Ross McIlroy <rmcilroy@chromium.org>
> > Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> > Reviewed-by: Ulan Degenbaev <ulan@chromium.org>
> > Cr-Commit-Position: refs/heads/master@{#70597}
>
> Tbr: ulan@chromium.org, tebbi@chromium.org, rmcilroy@chromium.org
> Bug: v8:8661,v8:8768,chromium:1140165
> Change-Id: Ibcd5c39c58a70bf2b2ac221aa375fc68d495e144
> Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2485506
> Reviewed-by: Jakob Gruber <jgruber@chromium.org>
> Reviewed-by: Tobias Tebbi <tebbi@chromium.org>
> Commit-Queue: Jakob Gruber <jgruber@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#70655}
Tbr: ulan@chromium.org, tebbi@chromium.org, rmcilroy@chromium.org
Bug: v8:8661
Bug: v8:8768
Bug: chromium:1140165
Change-Id: I471cc94fc085e527dc9bfb5a84b96bd907c2333f
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2488682
Reviewed-by: Jakob Gruber <jgruber@chromium.org>
Commit-Queue: Jakob Gruber <jgruber@chromium.org>
Cr-Commit-Position: refs/heads/master@{#70672}
2020-10-21 05:12:25 +00:00
|
|
|
Handle<Code> code =
|
|
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2014-04-16 11:50:24 +00:00
|
|
|
USE(code);
|
2011-12-05 08:58:01 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2018-04-13 22:28:05 +00:00
|
|
|
Address begin = code->raw_instruction_start();
|
|
|
|
|
Address end = code->raw_instruction_end();
|
|
|
|
|
disasm::Disassembler::Disassemble(stdout, reinterpret_cast<byte*>(begin),
|
|
|
|
|
reinterpret_cast<byte*>(end));
|
2011-12-05 08:58:01 +00:00
|
|
|
#endif
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#undef __
|
2017-08-11 11:22:28 +00:00
|
|
|
|
|
|
|
|
} // namespace internal
|
|
|
|
|
} // namespace v8
|
