2012-07-25 15:26:16 +00:00
|
|
|
// Copyright 2012 the V8 project authors. All rights reserved.
|
2008-08-22 13:33:59 +00:00
|
|
|
// 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.
|
|
|
|
|
|
2021-04-29 16:54:13 +00:00
|
|
|
#include <iostream>
|
2014-10-14 05:03:04 +00:00
|
|
|
|
2021-06-24 13:32:01 +00:00
|
|
|
#include "src/base/numbers/double.h"
|
2015-04-21 10:21:50 +00:00
|
|
|
#include "src/base/utils/random-number-generator.h"
|
2019-05-21 09:30:15 +00:00
|
|
|
#include "src/codegen/assembler-inl.h"
|
|
|
|
|
#include "src/codegen/macro-assembler.h"
|
2019-05-22 07:55:37 +00:00
|
|
|
#include "src/execution/simulator.h"
|
2018-04-09 19:11:22 +00:00
|
|
|
#include "src/heap/factory.h"
|
2019-05-23 13:27:57 +00:00
|
|
|
#include "src/utils/ostreams.h"
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
#include "test/cctest/assembler-helper-arm.h"
|
2017-02-28 12:27:32 +00:00
|
|
|
#include "test/cctest/cctest.h"
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
#include "test/cctest/compiler/value-helper.h"
|
2008-08-22 13:33:59 +00:00
|
|
|
|
2017-08-31 12:34:55 +00:00
|
|
|
namespace v8 {
|
|
|
|
|
namespace internal {
|
2017-10-02 21:19:21 +00:00
|
|
|
namespace test_assembler_arm {
|
2008-08-22 13:33:59 +00:00
|
|
|
|
2017-08-31 12:34:55 +00:00
|
|
|
using base::RandomNumberGenerator;
|
2008-08-22 13:33:59 +00:00
|
|
|
|
|
|
|
|
#define __ assm.
|
|
|
|
|
|
|
|
|
|
TEST(0) {
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2008-08-22 13:33:59 +00:00
|
|
|
|
|
|
|
|
__ add(r0, r0, Operand(r1));
|
|
|
|
|
__ mov(pc, Operand(lr));
|
|
|
|
|
|
|
|
|
|
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();
|
2008-08-22 13:33:59 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2008-08-22 13:33:59 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
int res = reinterpret_cast<int>(f.Call(3, 4, 0, 0, 0));
|
2008-08-22 13:33:59 +00:00
|
|
|
::printf("f() = %d\n", res);
|
|
|
|
|
CHECK_EQ(7, res);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(1) {
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2008-08-22 13:33:59 +00:00
|
|
|
Label L, C;
|
|
|
|
|
|
|
|
|
|
__ mov(r1, Operand(r0));
|
2013-01-07 09:43:12 +00:00
|
|
|
__ mov(r0, Operand::Zero());
|
2008-08-22 13:33:59 +00:00
|
|
|
__ b(&C);
|
|
|
|
|
|
|
|
|
|
__ bind(&L);
|
|
|
|
|
__ add(r0, r0, Operand(r1));
|
|
|
|
|
__ sub(r1, r1, Operand(1));
|
|
|
|
|
|
|
|
|
|
__ bind(&C);
|
2013-01-07 09:43:12 +00:00
|
|
|
__ teq(r1, Operand::Zero());
|
2008-08-22 13:33:59 +00:00
|
|
|
__ b(ne, &L);
|
|
|
|
|
__ mov(pc, Operand(lr));
|
|
|
|
|
|
|
|
|
|
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();
|
2008-08-22 13:33:59 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2008-08-22 13:33:59 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
int res = reinterpret_cast<int>(f.Call(100, 0, 0, 0, 0));
|
2008-08-22 13:33:59 +00:00
|
|
|
::printf("f() = %d\n", res);
|
|
|
|
|
CHECK_EQ(5050, res);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(2) {
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2008-08-22 13:33:59 +00:00
|
|
|
Label L, C;
|
|
|
|
|
|
|
|
|
|
__ mov(r1, Operand(r0));
|
|
|
|
|
__ mov(r0, Operand(1));
|
|
|
|
|
__ b(&C);
|
|
|
|
|
|
|
|
|
|
__ bind(&L);
|
|
|
|
|
__ mul(r0, r1, r0);
|
|
|
|
|
__ sub(r1, r1, Operand(1));
|
|
|
|
|
|
|
|
|
|
__ bind(&C);
|
2013-01-07 09:43:12 +00:00
|
|
|
__ teq(r1, Operand::Zero());
|
2008-08-22 13:33:59 +00:00
|
|
|
__ b(ne, &L);
|
|
|
|
|
__ mov(pc, Operand(lr));
|
|
|
|
|
|
|
|
|
|
// some relocated stuff here, not executed
|
|
|
|
|
__ RecordComment("dead code, just testing relocations");
|
2013-06-04 10:30:05 +00:00
|
|
|
__ mov(r0, Operand(isolate->factory()->true_value()));
|
2008-08-22 13:33:59 +00:00
|
|
|
__ RecordComment("dead code, just testing immediate operands");
|
|
|
|
|
__ mov(r0, Operand(-1));
|
|
|
|
|
__ mov(r0, Operand(0xFF000000));
|
|
|
|
|
__ mov(r0, Operand(0xF0F0F0F0));
|
|
|
|
|
__ mov(r0, Operand(0xFFF0FFFF));
|
|
|
|
|
|
|
|
|
|
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();
|
2008-08-22 13:33:59 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2008-08-22 13:33:59 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
int res = reinterpret_cast<int>(f.Call(10, 0, 0, 0, 0));
|
2008-08-22 13:33:59 +00:00
|
|
|
::printf("f() = %d\n", res);
|
|
|
|
|
CHECK_EQ(3628800, res);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(3) {
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
2008-08-22 13:33:59 +00:00
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2008-08-22 13:33:59 +00:00
|
|
|
int i;
|
|
|
|
|
char c;
|
|
|
|
|
int16_t s;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2008-08-22 13:33:59 +00:00
|
|
|
T t;
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2008-08-22 13:33:59 +00:00
|
|
|
|
|
|
|
|
__ mov(ip, Operand(sp));
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, fp, lr});
|
2008-08-22 13:33:59 +00:00
|
|
|
__ sub(fp, ip, Operand(4));
|
|
|
|
|
__ mov(r4, Operand(r0));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ ldr(r0, MemOperand(r4, offsetof(T, i)));
|
2008-08-22 13:33:59 +00:00
|
|
|
__ mov(r2, Operand(r0, ASR, 1));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r2, MemOperand(r4, offsetof(T, i)));
|
|
|
|
|
__ ldrsb(r2, MemOperand(r4, offsetof(T, c)));
|
2008-08-22 13:33:59 +00:00
|
|
|
__ add(r0, r2, Operand(r0));
|
|
|
|
|
__ mov(r2, Operand(r2, LSL, 2));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ strb(r2, MemOperand(r4, offsetof(T, c)));
|
|
|
|
|
__ ldrsh(r2, MemOperand(r4, offsetof(T, s)));
|
2008-08-22 13:33:59 +00:00
|
|
|
__ add(r0, r2, Operand(r0));
|
|
|
|
|
__ mov(r2, Operand(r2, ASR, 3));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ strh(r2, MemOperand(r4, offsetof(T, s)));
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, fp, pc});
|
2008-08-22 13:33:59 +00:00
|
|
|
|
|
|
|
|
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();
|
2008-08-22 13:33:59 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2008-08-22 13:33:59 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2008-08-22 13:33:59 +00:00
|
|
|
t.i = 100000;
|
|
|
|
|
t.c = 10;
|
|
|
|
|
t.s = 1000;
|
2018-01-09 09:50:34 +00:00
|
|
|
int res = reinterpret_cast<int>(f.Call(&t, 0, 0, 0, 0));
|
2008-08-22 13:33:59 +00:00
|
|
|
::printf("f() = %d\n", res);
|
|
|
|
|
CHECK_EQ(101010, res);
|
|
|
|
|
CHECK_EQ(100000/2, t.i);
|
|
|
|
|
CHECK_EQ(10*4, t.c);
|
|
|
|
|
CHECK_EQ(1000/8, t.s);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2010-01-25 11:54:10 +00:00
|
|
|
TEST(4) {
|
|
|
|
|
// Test the VFP floating point instructions.
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
2010-01-25 11:54:10 +00:00
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2010-01-25 11:54:10 +00:00
|
|
|
double a;
|
|
|
|
|
double b;
|
|
|
|
|
double c;
|
2010-08-26 08:53:00 +00:00
|
|
|
double d;
|
|
|
|
|
double e;
|
|
|
|
|
double f;
|
2011-01-31 10:16:28 +00:00
|
|
|
double g;
|
|
|
|
|
double h;
|
2010-08-26 08:53:00 +00:00
|
|
|
int i;
|
2013-05-17 15:38:14 +00:00
|
|
|
double j;
|
2011-03-15 11:19:13 +00:00
|
|
|
double m;
|
|
|
|
|
double n;
|
2016-03-14 15:30:13 +00:00
|
|
|
float o;
|
|
|
|
|
float p;
|
2010-08-26 08:53:00 +00:00
|
|
|
float x;
|
|
|
|
|
float y;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2010-01-25 11:54:10 +00:00
|
|
|
T t;
|
|
|
|
|
|
|
|
|
|
// Create a function that accepts &t, and loads, manipulates, and stores
|
2010-08-26 08:53:00 +00:00
|
|
|
// the doubles and floats.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2010-01-25 11:54:10 +00:00
|
|
|
|
2016-09-23 15:29:11 +00:00
|
|
|
if (CpuFeatures::IsSupported(VFPv3)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, VFPv3);
|
2010-01-25 11:54:10 +00:00
|
|
|
|
|
|
|
|
__ mov(ip, Operand(sp));
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, fp, lr});
|
2010-01-25 11:54:10 +00:00
|
|
|
__ sub(fp, ip, Operand(4));
|
|
|
|
|
|
|
|
|
|
__ mov(r4, Operand(r0));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d6, r4, offsetof(T, a));
|
|
|
|
|
__ vldr(d7, r4, offsetof(T, b));
|
2010-01-25 11:54:10 +00:00
|
|
|
__ vadd(d5, d6, d7);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d5, r4, offsetof(T, c));
|
2010-01-25 11:54:10 +00:00
|
|
|
|
2012-11-14 11:01:18 +00:00
|
|
|
__ vmla(d5, d6, d7);
|
2013-02-27 10:24:40 +00:00
|
|
|
__ vmls(d5, d5, d6);
|
2012-11-14 11:01:18 +00:00
|
|
|
|
2010-01-25 11:54:10 +00:00
|
|
|
__ vmov(r2, r3, d5);
|
|
|
|
|
__ vmov(d4, r2, r3);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d4, r4, offsetof(T, b));
|
2010-01-25 11:54:10 +00:00
|
|
|
|
2010-08-26 08:53:00 +00:00
|
|
|
// Load t.x and t.y, switch values, and store back to the struct.
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(s0, r4, offsetof(T, x));
|
2016-09-08 11:51:01 +00:00
|
|
|
__ vldr(s1, r4, offsetof(T, y));
|
|
|
|
|
__ vmov(s2, s0);
|
|
|
|
|
__ vmov(s0, s1);
|
|
|
|
|
__ vmov(s1, s2);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(s0, r4, offsetof(T, x));
|
2016-09-08 11:51:01 +00:00
|
|
|
__ vstr(s1, r4, offsetof(T, y));
|
2010-08-26 08:53:00 +00:00
|
|
|
|
|
|
|
|
// Move a literal into a register that can be encoded in the instruction.
|
2021-06-24 13:32:01 +00:00
|
|
|
__ vmov(d4, base::Double(1.0));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d4, r4, offsetof(T, e));
|
2010-08-26 08:53:00 +00:00
|
|
|
|
|
|
|
|
// Move a literal into a register that requires 64 bits to encode.
|
2017-12-02 00:30:37 +00:00
|
|
|
// 0x3FF0000010000000 = 1.000000059604644775390625
|
2021-06-24 13:32:01 +00:00
|
|
|
__ vmov(d4, base::Double(1.000000059604644775390625));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d4, r4, offsetof(T, d));
|
2010-08-26 08:53:00 +00:00
|
|
|
|
|
|
|
|
// Convert from floating point to integer.
|
2021-06-24 13:32:01 +00:00
|
|
|
__ vmov(d4, base::Double(2.0));
|
2016-09-08 11:51:01 +00:00
|
|
|
__ vcvt_s32_f64(s1, d4);
|
|
|
|
|
__ vstr(s1, r4, offsetof(T, i));
|
2010-08-26 08:53:00 +00:00
|
|
|
|
|
|
|
|
// Convert from integer to floating point.
|
|
|
|
|
__ mov(lr, Operand(42));
|
2016-09-08 11:51:01 +00:00
|
|
|
__ vmov(s1, lr);
|
|
|
|
|
__ vcvt_f64_s32(d4, s1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d4, r4, offsetof(T, f));
|
2011-01-31 10:16:28 +00:00
|
|
|
|
2013-05-17 15:38:14 +00:00
|
|
|
// Convert from fixed point to floating point.
|
2014-04-04 11:12:40 +00:00
|
|
|
__ mov(lr, Operand(2468));
|
2013-05-17 15:38:14 +00:00
|
|
|
__ vmov(s8, lr);
|
2014-04-04 11:12:40 +00:00
|
|
|
__ vcvt_f64_s32(d4, 2);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d4, r4, offsetof(T, j));
|
2013-05-17 15:38:14 +00:00
|
|
|
|
2011-01-31 10:16:28 +00:00
|
|
|
// Test vabs.
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d1, r4, offsetof(T, g));
|
2011-01-31 10:16:28 +00:00
|
|
|
__ vabs(d0, d1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d0, r4, offsetof(T, g));
|
|
|
|
|
__ vldr(d2, r4, offsetof(T, h));
|
2011-01-31 10:16:28 +00:00
|
|
|
__ vabs(d0, d2);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d0, r4, offsetof(T, h));
|
2011-01-31 10:16:28 +00:00
|
|
|
|
2011-03-15 11:19:13 +00:00
|
|
|
// Test vneg.
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d1, r4, offsetof(T, m));
|
2011-03-15 11:19:13 +00:00
|
|
|
__ vneg(d0, d1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d0, r4, offsetof(T, m));
|
|
|
|
|
__ vldr(d1, r4, offsetof(T, n));
|
2011-03-15 11:19:13 +00:00
|
|
|
__ vneg(d0, d1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d0, r4, offsetof(T, n));
|
2011-03-15 11:19:13 +00:00
|
|
|
|
2016-03-14 15:30:13 +00:00
|
|
|
// Test vmov for single-precision immediates.
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s0, Float32(0.25f));
|
2016-03-14 15:30:13 +00:00
|
|
|
__ vstr(s0, r4, offsetof(T, o));
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s0, Float32(-16.0f));
|
2016-03-14 15:30:13 +00:00
|
|
|
__ vstr(s0, r4, offsetof(T, p));
|
|
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, fp, pc});
|
2010-01-25 15:17:58 +00:00
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2010-01-25 11:54:10 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2010-01-25 11:54:10 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2010-01-25 15:17:58 +00:00
|
|
|
t.a = 1.5;
|
|
|
|
|
t.b = 2.75;
|
|
|
|
|
t.c = 17.17;
|
2010-08-26 08:53:00 +00:00
|
|
|
t.d = 0.0;
|
|
|
|
|
t.e = 0.0;
|
|
|
|
|
t.f = 0.0;
|
2011-01-31 10:16:28 +00:00
|
|
|
t.g = -2718.2818;
|
|
|
|
|
t.h = 31415926.5;
|
2010-08-26 08:53:00 +00:00
|
|
|
t.i = 0;
|
2013-05-17 15:38:14 +00:00
|
|
|
t.j = 0;
|
2011-03-15 11:19:13 +00:00
|
|
|
t.m = -2718.2818;
|
|
|
|
|
t.n = 123.456;
|
2010-08-26 08:53:00 +00:00
|
|
|
t.x = 4.5;
|
|
|
|
|
t.y = 9.0;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, 0, 0, 0, 0);
|
2016-03-14 15:30:13 +00:00
|
|
|
CHECK_EQ(-16.0f, t.p);
|
|
|
|
|
CHECK_EQ(0.25f, t.o);
|
2011-03-15 11:19:13 +00:00
|
|
|
CHECK_EQ(-123.456, t.n);
|
|
|
|
|
CHECK_EQ(2718.2818, t.m);
|
2010-08-26 08:53:00 +00:00
|
|
|
CHECK_EQ(2, t.i);
|
2011-01-31 10:16:28 +00:00
|
|
|
CHECK_EQ(2718.2818, t.g);
|
|
|
|
|
CHECK_EQ(31415926.5, t.h);
|
2013-05-17 15:38:14 +00:00
|
|
|
CHECK_EQ(617.0, t.j);
|
2010-08-26 08:53:00 +00:00
|
|
|
CHECK_EQ(42.0, t.f);
|
|
|
|
|
CHECK_EQ(1.0, t.e);
|
|
|
|
|
CHECK_EQ(1.000000059604644775390625, t.d);
|
2010-01-25 15:17:58 +00:00
|
|
|
CHECK_EQ(4.25, t.c);
|
2013-02-27 10:24:40 +00:00
|
|
|
CHECK_EQ(-4.1875, t.b);
|
2010-01-25 15:17:58 +00:00
|
|
|
CHECK_EQ(1.5, t.a);
|
2015-10-08 19:02:29 +00:00
|
|
|
CHECK_EQ(4.5f, t.y);
|
|
|
|
|
CHECK_EQ(9.0f, t.x);
|
2010-01-25 15:17:58 +00:00
|
|
|
}
|
2010-01-25 11:54:10 +00:00
|
|
|
}
|
|
|
|
|
|
2010-05-27 13:46:18 +00:00
|
|
|
|
|
|
|
|
TEST(5) {
|
|
|
|
|
// Test the ARMv7 bitfield instructions.
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2010-05-27 13:46:18 +00:00
|
|
|
|
2011-03-31 16:17:37 +00:00
|
|
|
if (CpuFeatures::IsSupported(ARMv7)) {
|
2013-03-05 10:48:16 +00:00
|
|
|
CpuFeatureScope scope(&assm, ARMv7);
|
2010-05-27 13:46:18 +00:00
|
|
|
// On entry, r0 = 0xAAAAAAAA = 0b10..10101010.
|
|
|
|
|
__ ubfx(r0, r0, 1, 12); // 0b00..010101010101 = 0x555
|
|
|
|
|
__ sbfx(r0, r0, 0, 5); // 0b11..111111110101 = -11
|
|
|
|
|
__ bfc(r0, 1, 3); // 0b11..111111110001 = -15
|
|
|
|
|
__ mov(r1, Operand(7));
|
|
|
|
|
__ bfi(r0, r1, 3, 3); // 0b11..111111111001 = -7
|
|
|
|
|
__ mov(pc, Operand(lr));
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2010-05-27 13:46:18 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2010-05-27 13:46:18 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
int res = reinterpret_cast<int>(f.Call(0xAAAAAAAA, 0, 0, 0, 0));
|
2010-05-27 13:46:18 +00:00
|
|
|
::printf("f() = %d\n", res);
|
|
|
|
|
CHECK_EQ(-7, res);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2010-07-21 07:42:51 +00:00
|
|
|
|
|
|
|
|
TEST(6) {
|
|
|
|
|
// Test saturating instructions.
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2010-07-21 07:42:51 +00:00
|
|
|
|
2016-05-12 18:18:23 +00:00
|
|
|
__ usat(r1, 8, Operand(r0)); // Sat 0xFFFF to 0-255 = 0xFF.
|
|
|
|
|
__ usat(r2, 12, Operand(r0, ASR, 9)); // Sat (0xFFFF>>9) to 0-4095 = 0x7F.
|
|
|
|
|
__ usat(r3, 1, Operand(r0, LSL, 16)); // Sat (0xFFFF<<16) to 0-1 = 0x0.
|
|
|
|
|
__ add(r0, r1, Operand(r2));
|
|
|
|
|
__ add(r0, r0, Operand(r3));
|
|
|
|
|
__ mov(pc, Operand(lr));
|
2010-07-21 07:42:51 +00:00
|
|
|
|
2016-05-12 18:18:23 +00:00
|
|
|
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();
|
2010-07-21 07:42:51 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2010-07-21 07:42:51 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
int res = reinterpret_cast<int>(f.Call(0xFFFF, 0, 0, 0, 0));
|
2016-05-12 18:18:23 +00:00
|
|
|
::printf("f() = %d\n", res);
|
|
|
|
|
CHECK_EQ(382, res);
|
2010-07-21 07:42:51 +00:00
|
|
|
}
|
|
|
|
|
|
2010-11-09 08:26:02 +00:00
|
|
|
|
2011-02-04 07:08:50 +00:00
|
|
|
enum VCVTTypes {
|
|
|
|
|
s32_f64,
|
|
|
|
|
u32_f64
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
static void TestRoundingMode(VCVTTypes types,
|
|
|
|
|
VFPRoundingMode mode,
|
|
|
|
|
double value,
|
|
|
|
|
int expected,
|
|
|
|
|
bool expected_exception = false) {
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2010-11-09 08:26:02 +00:00
|
|
|
|
2016-09-23 15:29:11 +00:00
|
|
|
Label wrong_exception;
|
2010-11-09 08:26:02 +00:00
|
|
|
|
2016-09-23 15:29:11 +00:00
|
|
|
__ vmrs(r1);
|
|
|
|
|
// Set custom FPSCR.
|
|
|
|
|
__ bic(r2, r1, Operand(kVFPRoundingModeMask | kVFPExceptionMask));
|
|
|
|
|
__ orr(r2, r2, Operand(mode));
|
|
|
|
|
__ vmsr(r2);
|
2010-11-09 08:26:02 +00:00
|
|
|
|
2016-09-23 15:29:11 +00:00
|
|
|
// Load value, convert, and move back result to r0 if everything went well.
|
2021-06-24 13:32:01 +00:00
|
|
|
__ vmov(d1, base::Double(value));
|
2016-09-23 15:29:11 +00:00
|
|
|
switch (types) {
|
|
|
|
|
case s32_f64:
|
|
|
|
|
__ vcvt_s32_f64(s0, d1, kFPSCRRounding);
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case u32_f64:
|
|
|
|
|
__ vcvt_u32_f64(s0, d1, kFPSCRRounding);
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
default:
|
|
|
|
|
UNREACHABLE();
|
|
|
|
|
}
|
|
|
|
|
// Check for vfp exceptions
|
|
|
|
|
__ vmrs(r2);
|
|
|
|
|
__ tst(r2, Operand(kVFPExceptionMask));
|
|
|
|
|
// Check that we behaved as expected.
|
|
|
|
|
__ b(&wrong_exception, expected_exception ? eq : ne);
|
|
|
|
|
// There was no exception. Retrieve the result and return.
|
|
|
|
|
__ vmov(r0, s0);
|
|
|
|
|
__ mov(pc, Operand(lr));
|
|
|
|
|
|
|
|
|
|
// The exception behaviour is not what we expected.
|
|
|
|
|
// Load a special value and return.
|
|
|
|
|
__ bind(&wrong_exception);
|
|
|
|
|
__ mov(r0, Operand(11223344));
|
|
|
|
|
__ mov(pc, Operand(lr));
|
|
|
|
|
|
|
|
|
|
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();
|
2010-11-09 08:26:02 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2010-11-09 08:26:02 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
int res = reinterpret_cast<int>(f.Call(0, 0, 0, 0, 0));
|
2016-09-23 15:29:11 +00:00
|
|
|
::printf("res = %d\n", res);
|
|
|
|
|
CHECK_EQ(expected, res);
|
2010-11-09 08:26:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(7) {
|
2013-09-19 13:30:47 +00:00
|
|
|
CcTest::InitializeVM();
|
2010-11-09 08:26:02 +00:00
|
|
|
// Test vfp rounding modes.
|
|
|
|
|
|
2011-02-04 07:08:50 +00:00
|
|
|
// s32_f64 (double to integer).
|
|
|
|
|
|
|
|
|
|
TestRoundingMode(s32_f64, RN, 0, 0);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, 0.5, 0);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, -0.5, 0);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, 1.5, 2);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, -1.5, -2);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, 123.7, 124);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, -123.7, -124);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, 123456.2, 123456);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, -123456.2, -123456);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, static_cast<double>(kMaxInt), kMaxInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, (kMaxInt + 0.49), kMaxInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, (kMaxInt + 1.0), kMaxInt, true);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, (kMaxInt + 0.5), kMaxInt, true);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, static_cast<double>(kMinInt), kMinInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, (kMinInt - 0.5), kMinInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, (kMinInt - 1.0), kMinInt, true);
|
|
|
|
|
TestRoundingMode(s32_f64, RN, (kMinInt - 0.51), kMinInt, true);
|
|
|
|
|
|
|
|
|
|
TestRoundingMode(s32_f64, RM, 0, 0);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, 0.5, 0);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, -0.5, -1);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, 123.7, 123);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, -123.7, -124);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, 123456.2, 123456);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, -123456.2, -123457);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, static_cast<double>(kMaxInt), kMaxInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, (kMaxInt + 0.5), kMaxInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, (kMaxInt + 1.0), kMaxInt, true);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, static_cast<double>(kMinInt), kMinInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, (kMinInt - 0.5), kMinInt, true);
|
|
|
|
|
TestRoundingMode(s32_f64, RM, (kMinInt + 0.5), kMinInt);
|
|
|
|
|
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, 0, 0);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, 0.5, 0);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, -0.5, 0);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, 123.7, 123);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, -123.7, -123);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, 123456.2, 123456);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, -123456.2, -123456);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, static_cast<double>(kMaxInt), kMaxInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, (kMaxInt + 0.5), kMaxInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, (kMaxInt + 1.0), kMaxInt, true);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, static_cast<double>(kMinInt), kMinInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, (kMinInt - 0.5), kMinInt);
|
|
|
|
|
TestRoundingMode(s32_f64, RZ, (kMinInt - 1.0), kMinInt, true);
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// u32_f64 (double to integer).
|
|
|
|
|
|
|
|
|
|
// Negative values.
|
|
|
|
|
TestRoundingMode(u32_f64, RN, -0.5, 0);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, -123456.7, 0, true);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, static_cast<double>(kMinInt), 0, true);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, kMinInt - 1.0, 0, true);
|
|
|
|
|
|
|
|
|
|
TestRoundingMode(u32_f64, RM, -0.5, 0, true);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, -123456.7, 0, true);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, static_cast<double>(kMinInt), 0, true);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, kMinInt - 1.0, 0, true);
|
|
|
|
|
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, -0.5, 0);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, -123456.7, 0, true);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, static_cast<double>(kMinInt), 0, true);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, kMinInt - 1.0, 0, true);
|
|
|
|
|
|
|
|
|
|
// Positive values.
|
2017-12-02 00:30:37 +00:00
|
|
|
// kMaxInt is the maximum *signed* integer: 0x7FFFFFFF.
|
|
|
|
|
static const uint32_t kMaxUInt = 0xFFFFFFFFu;
|
2011-02-04 07:08:50 +00:00
|
|
|
TestRoundingMode(u32_f64, RZ, 0, 0);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, 0.5, 0);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, 123.7, 123);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, 123456.2, 123456);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, static_cast<double>(kMaxInt), kMaxInt);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, (kMaxInt + 0.5), kMaxInt);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, (kMaxInt + 1.0),
|
|
|
|
|
static_cast<uint32_t>(kMaxInt) + 1);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, (kMaxUInt + 0.5), kMaxUInt);
|
|
|
|
|
TestRoundingMode(u32_f64, RZ, (kMaxUInt + 1.0), kMaxUInt, true);
|
|
|
|
|
|
|
|
|
|
TestRoundingMode(u32_f64, RM, 0, 0);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, 0.5, 0);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, 123.7, 123);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, 123456.2, 123456);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, static_cast<double>(kMaxInt), kMaxInt);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, (kMaxInt + 0.5), kMaxInt);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, (kMaxInt + 1.0),
|
|
|
|
|
static_cast<uint32_t>(kMaxInt) + 1);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, (kMaxUInt + 0.5), kMaxUInt);
|
|
|
|
|
TestRoundingMode(u32_f64, RM, (kMaxUInt + 1.0), kMaxUInt, true);
|
|
|
|
|
|
|
|
|
|
TestRoundingMode(u32_f64, RN, 0, 0);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, 0.5, 0);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, 1.5, 2);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, 123.7, 124);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, 123456.2, 123456);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, static_cast<double>(kMaxInt), kMaxInt);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, (kMaxInt + 0.49), kMaxInt);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, (kMaxInt + 0.5),
|
|
|
|
|
static_cast<uint32_t>(kMaxInt) + 1);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, (kMaxUInt + 0.49), kMaxUInt);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, (kMaxUInt + 0.5), kMaxUInt, true);
|
|
|
|
|
TestRoundingMode(u32_f64, RN, (kMaxUInt + 1.0), kMaxUInt, true);
|
2010-11-09 08:26:02 +00:00
|
|
|
}
|
|
|
|
|
|
2013-07-05 09:52:11 +00:00
|
|
|
|
2011-04-06 09:06:23 +00:00
|
|
|
TEST(8) {
|
|
|
|
|
// Test VFP multi load/store with ia_w.
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct D {
|
2011-04-06 09:06:23 +00:00
|
|
|
double a;
|
|
|
|
|
double b;
|
|
|
|
|
double c;
|
|
|
|
|
double d;
|
|
|
|
|
double e;
|
|
|
|
|
double f;
|
|
|
|
|
double g;
|
|
|
|
|
double h;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2011-04-06 09:06:23 +00:00
|
|
|
D d;
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct F {
|
2011-04-06 09:06:23 +00:00
|
|
|
float a;
|
|
|
|
|
float b;
|
|
|
|
|
float c;
|
|
|
|
|
float d;
|
|
|
|
|
float e;
|
|
|
|
|
float f;
|
|
|
|
|
float g;
|
|
|
|
|
float h;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2011-04-06 09:06:23 +00:00
|
|
|
F f;
|
|
|
|
|
|
|
|
|
|
// Create a function that uses vldm/vstm to move some double and
|
|
|
|
|
// single precision values around in memory.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2013-04-07 04:34:20 +00:00
|
|
|
__ mov(ip, Operand(sp));
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, fp, lr});
|
2013-04-07 04:34:20 +00:00
|
|
|
__ sub(fp, ip, Operand(4));
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(D, a))));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vldm(ia_w, r4, d0, d3);
|
|
|
|
|
__ vldm(ia_w, r4, d4, d7);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(D, a))));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vstm(ia_w, r4, d6, d7);
|
|
|
|
|
__ vstm(ia_w, r4, d0, d5);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r1, Operand(static_cast<int32_t>(offsetof(F, a))));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vldm(ia_w, r4, s0, s3);
|
|
|
|
|
__ vldm(ia_w, r4, s4, s7);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r1, Operand(static_cast<int32_t>(offsetof(F, a))));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vstm(ia_w, r4, s6, s7);
|
|
|
|
|
__ vstm(ia_w, r4, s0, s5);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, fp, pc});
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2013-04-07 04:34:20 +00:00
|
|
|
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();
|
2011-04-06 09:06:23 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2011-04-06 09:06:23 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto fn = GeneratedCode<F_ppiii>::FromCode(*code);
|
2013-04-07 04:34:20 +00:00
|
|
|
d.a = 1.1;
|
|
|
|
|
d.b = 2.2;
|
|
|
|
|
d.c = 3.3;
|
|
|
|
|
d.d = 4.4;
|
|
|
|
|
d.e = 5.5;
|
|
|
|
|
d.f = 6.6;
|
|
|
|
|
d.g = 7.7;
|
|
|
|
|
d.h = 8.8;
|
|
|
|
|
|
|
|
|
|
f.a = 1.0;
|
|
|
|
|
f.b = 2.0;
|
|
|
|
|
f.c = 3.0;
|
|
|
|
|
f.d = 4.0;
|
|
|
|
|
f.e = 5.0;
|
|
|
|
|
f.f = 6.0;
|
|
|
|
|
f.g = 7.0;
|
|
|
|
|
f.h = 8.0;
|
|
|
|
|
|
2018-01-09 09:50:34 +00:00
|
|
|
fn.Call(&d, &f, 0, 0, 0);
|
2013-04-07 04:34:20 +00:00
|
|
|
|
|
|
|
|
CHECK_EQ(7.7, d.a);
|
|
|
|
|
CHECK_EQ(8.8, d.b);
|
|
|
|
|
CHECK_EQ(1.1, d.c);
|
|
|
|
|
CHECK_EQ(2.2, d.d);
|
|
|
|
|
CHECK_EQ(3.3, d.e);
|
|
|
|
|
CHECK_EQ(4.4, d.f);
|
|
|
|
|
CHECK_EQ(5.5, d.g);
|
|
|
|
|
CHECK_EQ(6.6, d.h);
|
|
|
|
|
|
2015-10-06 14:50:29 +00:00
|
|
|
CHECK_EQ(7.0f, f.a);
|
|
|
|
|
CHECK_EQ(8.0f, f.b);
|
|
|
|
|
CHECK_EQ(1.0f, f.c);
|
|
|
|
|
CHECK_EQ(2.0f, f.d);
|
|
|
|
|
CHECK_EQ(3.0f, f.e);
|
|
|
|
|
CHECK_EQ(4.0f, f.f);
|
|
|
|
|
CHECK_EQ(5.0f, f.g);
|
|
|
|
|
CHECK_EQ(6.0f, f.h);
|
2011-04-06 09:06:23 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(9) {
|
|
|
|
|
// Test VFP multi load/store with ia.
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct D {
|
2011-04-06 09:06:23 +00:00
|
|
|
double a;
|
|
|
|
|
double b;
|
|
|
|
|
double c;
|
|
|
|
|
double d;
|
|
|
|
|
double e;
|
|
|
|
|
double f;
|
|
|
|
|
double g;
|
|
|
|
|
double h;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2011-04-06 09:06:23 +00:00
|
|
|
D d;
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct F {
|
2011-04-06 09:06:23 +00:00
|
|
|
float a;
|
|
|
|
|
float b;
|
|
|
|
|
float c;
|
|
|
|
|
float d;
|
|
|
|
|
float e;
|
|
|
|
|
float f;
|
|
|
|
|
float g;
|
|
|
|
|
float h;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2011-04-06 09:06:23 +00:00
|
|
|
F f;
|
|
|
|
|
|
|
|
|
|
// Create a function that uses vldm/vstm to move some double and
|
|
|
|
|
// single precision values around in memory.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2013-04-07 04:34:20 +00:00
|
|
|
__ mov(ip, Operand(sp));
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, fp, lr});
|
2013-04-07 04:34:20 +00:00
|
|
|
__ sub(fp, ip, Operand(4));
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(D, a))));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vldm(ia, r4, d0, d3);
|
|
|
|
|
__ add(r4, r4, Operand(4 * 8));
|
|
|
|
|
__ vldm(ia, r4, d4, d7);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(D, a))));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vstm(ia, r4, d6, d7);
|
|
|
|
|
__ add(r4, r4, Operand(2 * 8));
|
|
|
|
|
__ vstm(ia, r4, d0, d5);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r1, Operand(static_cast<int32_t>(offsetof(F, a))));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vldm(ia, r4, s0, s3);
|
|
|
|
|
__ add(r4, r4, Operand(4 * 4));
|
|
|
|
|
__ vldm(ia, r4, s4, s7);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r1, Operand(static_cast<int32_t>(offsetof(F, a))));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vstm(ia, r4, s6, s7);
|
|
|
|
|
__ add(r4, r4, Operand(2 * 4));
|
|
|
|
|
__ vstm(ia, r4, s0, s5);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, fp, pc});
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2013-04-07 04:34:20 +00:00
|
|
|
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();
|
2011-04-06 09:06:23 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2011-04-06 09:06:23 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto fn = GeneratedCode<F_ppiii>::FromCode(*code);
|
2013-04-07 04:34:20 +00:00
|
|
|
d.a = 1.1;
|
|
|
|
|
d.b = 2.2;
|
|
|
|
|
d.c = 3.3;
|
|
|
|
|
d.d = 4.4;
|
|
|
|
|
d.e = 5.5;
|
|
|
|
|
d.f = 6.6;
|
|
|
|
|
d.g = 7.7;
|
|
|
|
|
d.h = 8.8;
|
|
|
|
|
|
|
|
|
|
f.a = 1.0;
|
|
|
|
|
f.b = 2.0;
|
|
|
|
|
f.c = 3.0;
|
|
|
|
|
f.d = 4.0;
|
|
|
|
|
f.e = 5.0;
|
|
|
|
|
f.f = 6.0;
|
|
|
|
|
f.g = 7.0;
|
|
|
|
|
f.h = 8.0;
|
|
|
|
|
|
2018-01-09 09:50:34 +00:00
|
|
|
fn.Call(&d, &f, 0, 0, 0);
|
2013-04-07 04:34:20 +00:00
|
|
|
|
|
|
|
|
CHECK_EQ(7.7, d.a);
|
|
|
|
|
CHECK_EQ(8.8, d.b);
|
|
|
|
|
CHECK_EQ(1.1, d.c);
|
|
|
|
|
CHECK_EQ(2.2, d.d);
|
|
|
|
|
CHECK_EQ(3.3, d.e);
|
|
|
|
|
CHECK_EQ(4.4, d.f);
|
|
|
|
|
CHECK_EQ(5.5, d.g);
|
|
|
|
|
CHECK_EQ(6.6, d.h);
|
|
|
|
|
|
2015-10-06 14:50:29 +00:00
|
|
|
CHECK_EQ(7.0f, f.a);
|
|
|
|
|
CHECK_EQ(8.0f, f.b);
|
|
|
|
|
CHECK_EQ(1.0f, f.c);
|
|
|
|
|
CHECK_EQ(2.0f, f.d);
|
|
|
|
|
CHECK_EQ(3.0f, f.e);
|
|
|
|
|
CHECK_EQ(4.0f, f.f);
|
|
|
|
|
CHECK_EQ(5.0f, f.g);
|
|
|
|
|
CHECK_EQ(6.0f, f.h);
|
2011-04-06 09:06:23 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(10) {
|
|
|
|
|
// Test VFP multi load/store with db_w.
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct D {
|
2011-04-06 09:06:23 +00:00
|
|
|
double a;
|
|
|
|
|
double b;
|
|
|
|
|
double c;
|
|
|
|
|
double d;
|
|
|
|
|
double e;
|
|
|
|
|
double f;
|
|
|
|
|
double g;
|
|
|
|
|
double h;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2011-04-06 09:06:23 +00:00
|
|
|
D d;
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct F {
|
2011-04-06 09:06:23 +00:00
|
|
|
float a;
|
|
|
|
|
float b;
|
|
|
|
|
float c;
|
|
|
|
|
float d;
|
|
|
|
|
float e;
|
|
|
|
|
float f;
|
|
|
|
|
float g;
|
|
|
|
|
float h;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2011-04-06 09:06:23 +00:00
|
|
|
F f;
|
|
|
|
|
|
|
|
|
|
// Create a function that uses vldm/vstm to move some double and
|
|
|
|
|
// single precision values around in memory.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2013-04-07 04:34:20 +00:00
|
|
|
__ mov(ip, Operand(sp));
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, fp, lr});
|
2013-04-07 04:34:20 +00:00
|
|
|
__ sub(fp, ip, Operand(4));
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(D, h)) + 8));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vldm(db_w, r4, d4, d7);
|
|
|
|
|
__ vldm(db_w, r4, d0, d3);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(D, h)) + 8));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vstm(db_w, r4, d0, d5);
|
|
|
|
|
__ vstm(db_w, r4, d6, d7);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r1, Operand(static_cast<int32_t>(offsetof(F, h)) + 4));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vldm(db_w, r4, s4, s7);
|
|
|
|
|
__ vldm(db_w, r4, s0, s3);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r1, Operand(static_cast<int32_t>(offsetof(F, h)) + 4));
|
2013-04-07 04:34:20 +00:00
|
|
|
__ vstm(db_w, r4, s0, s5);
|
|
|
|
|
__ vstm(db_w, r4, s6, s7);
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, fp, pc});
|
2011-04-06 09:06:23 +00:00
|
|
|
|
2013-04-07 04:34:20 +00:00
|
|
|
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();
|
2011-04-06 09:06:23 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2011-04-06 09:06:23 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto fn = GeneratedCode<F_ppiii>::FromCode(*code);
|
2013-04-07 04:34:20 +00:00
|
|
|
d.a = 1.1;
|
|
|
|
|
d.b = 2.2;
|
|
|
|
|
d.c = 3.3;
|
|
|
|
|
d.d = 4.4;
|
|
|
|
|
d.e = 5.5;
|
|
|
|
|
d.f = 6.6;
|
|
|
|
|
d.g = 7.7;
|
|
|
|
|
d.h = 8.8;
|
|
|
|
|
|
|
|
|
|
f.a = 1.0;
|
|
|
|
|
f.b = 2.0;
|
|
|
|
|
f.c = 3.0;
|
|
|
|
|
f.d = 4.0;
|
|
|
|
|
f.e = 5.0;
|
|
|
|
|
f.f = 6.0;
|
|
|
|
|
f.g = 7.0;
|
|
|
|
|
f.h = 8.0;
|
|
|
|
|
|
2018-01-09 09:50:34 +00:00
|
|
|
fn.Call(&d, &f, 0, 0, 0);
|
2013-04-07 04:34:20 +00:00
|
|
|
|
|
|
|
|
CHECK_EQ(7.7, d.a);
|
|
|
|
|
CHECK_EQ(8.8, d.b);
|
|
|
|
|
CHECK_EQ(1.1, d.c);
|
|
|
|
|
CHECK_EQ(2.2, d.d);
|
|
|
|
|
CHECK_EQ(3.3, d.e);
|
|
|
|
|
CHECK_EQ(4.4, d.f);
|
|
|
|
|
CHECK_EQ(5.5, d.g);
|
|
|
|
|
CHECK_EQ(6.6, d.h);
|
|
|
|
|
|
2015-10-06 14:50:29 +00:00
|
|
|
CHECK_EQ(7.0f, f.a);
|
|
|
|
|
CHECK_EQ(8.0f, f.b);
|
|
|
|
|
CHECK_EQ(1.0f, f.c);
|
|
|
|
|
CHECK_EQ(2.0f, f.d);
|
|
|
|
|
CHECK_EQ(3.0f, f.e);
|
|
|
|
|
CHECK_EQ(4.0f, f.f);
|
|
|
|
|
CHECK_EQ(5.0f, f.g);
|
|
|
|
|
CHECK_EQ(6.0f, f.h);
|
2011-04-06 09:06:23 +00:00
|
|
|
}
|
|
|
|
|
|
2011-04-18 13:53:11 +00:00
|
|
|
|
|
|
|
|
TEST(11) {
|
|
|
|
|
// Test instructions using the carry flag.
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
2011-04-18 13:53:11 +00:00
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct I {
|
2011-04-18 13:53:11 +00:00
|
|
|
int32_t a;
|
|
|
|
|
int32_t b;
|
|
|
|
|
int32_t c;
|
|
|
|
|
int32_t d;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2011-04-18 13:53:11 +00:00
|
|
|
I i;
|
|
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
i.a = 0xABCD0001;
|
|
|
|
|
i.b = 0xABCD0000;
|
2011-04-18 13:53:11 +00:00
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2011-04-18 13:53:11 +00:00
|
|
|
|
|
|
|
|
// Test HeapObject untagging.
|
2015-06-17 09:06:44 +00:00
|
|
|
__ ldr(r1, MemOperand(r0, offsetof(I, a)));
|
2011-04-18 13:53:11 +00:00
|
|
|
__ mov(r1, Operand(r1, ASR, 1), SetCC);
|
|
|
|
|
__ adc(r1, r1, Operand(r1), LeaveCC, cs);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r1, MemOperand(r0, offsetof(I, a)));
|
2011-04-18 13:53:11 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ ldr(r2, MemOperand(r0, offsetof(I, b)));
|
2011-04-18 13:53:11 +00:00
|
|
|
__ mov(r2, Operand(r2, ASR, 1), SetCC);
|
|
|
|
|
__ adc(r2, r2, Operand(r2), LeaveCC, cs);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r2, MemOperand(r0, offsetof(I, b)));
|
2011-04-18 13:53:11 +00:00
|
|
|
|
|
|
|
|
// Test corner cases.
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r1, Operand(0xFFFFFFFF));
|
2013-01-07 09:43:12 +00:00
|
|
|
__ mov(r2, Operand::Zero());
|
2011-04-18 13:53:11 +00:00
|
|
|
__ mov(r3, Operand(r1, ASR, 1), SetCC); // Set the carry.
|
|
|
|
|
__ adc(r3, r1, Operand(r2));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r3, MemOperand(r0, offsetof(I, c)));
|
2011-04-18 13:53:11 +00:00
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r1, Operand(0xFFFFFFFF));
|
2013-01-07 09:43:12 +00:00
|
|
|
__ mov(r2, Operand::Zero());
|
2011-04-18 13:53:11 +00:00
|
|
|
__ mov(r3, Operand(r2, ASR, 1), SetCC); // Unset the carry.
|
|
|
|
|
__ adc(r3, r1, Operand(r2));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r3, MemOperand(r0, offsetof(I, d)));
|
2011-04-18 13:53:11 +00:00
|
|
|
|
|
|
|
|
__ mov(pc, Operand(lr));
|
|
|
|
|
|
|
|
|
|
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();
|
2011-04-18 13:53:11 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2011-04-18 13:53:11 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
|
|
|
|
f.Call(&i, 0, 0, 0, 0);
|
2011-04-18 13:53:11 +00:00
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(0xABCD0001), i.a);
|
|
|
|
|
CHECK_EQ(static_cast<int32_t>(0xABCD0000) >> 1, i.b);
|
2011-04-18 13:53:11 +00:00
|
|
|
CHECK_EQ(0x00000000, i.c);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(0xFFFFFFFF), i.d);
|
2011-04-18 13:53:11 +00:00
|
|
|
}
|
|
|
|
|
|
2011-08-30 07:36:31 +00:00
|
|
|
|
|
|
|
|
TEST(12) {
|
|
|
|
|
// Test chaining of label usages within instructions (issue 1644).
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
2011-08-30 07:36:31 +00:00
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2011-08-30 07:36:31 +00:00
|
|
|
Label target;
|
|
|
|
|
__ b(eq, &target);
|
|
|
|
|
__ b(ne, &target);
|
|
|
|
|
__ bind(&target);
|
|
|
|
|
__ nop();
|
|
|
|
|
}
|
|
|
|
|
|
2013-01-23 16:29:48 +00:00
|
|
|
|
|
|
|
|
TEST(13) {
|
|
|
|
|
// Test VFP instructions using registers d16-d31.
|
2013-04-10 08:29:39 +00:00
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-03-15 12:06:53 +00:00
|
|
|
HandleScope scope(isolate);
|
2013-01-23 16:29:48 +00:00
|
|
|
|
|
|
|
|
if (!CpuFeatures::IsSupported(VFP32DREGS)) {
|
|
|
|
|
return;
|
|
|
|
|
}
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2013-01-23 16:29:48 +00:00
|
|
|
double a;
|
|
|
|
|
double b;
|
|
|
|
|
double c;
|
|
|
|
|
double x;
|
|
|
|
|
double y;
|
|
|
|
|
double z;
|
|
|
|
|
double i;
|
|
|
|
|
double j;
|
|
|
|
|
double k;
|
2013-07-25 15:04:38 +00:00
|
|
|
uint32_t low;
|
|
|
|
|
uint32_t high;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2013-01-23 16:29:48 +00:00
|
|
|
T t;
|
|
|
|
|
|
|
|
|
|
// Create a function that accepts &t, and loads, manipulates, and stores
|
|
|
|
|
// the doubles and floats.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2013-01-23 16:29:48 +00:00
|
|
|
|
2016-09-23 15:29:11 +00:00
|
|
|
if (CpuFeatures::IsSupported(VFPv3)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, VFPv3);
|
2013-01-23 16:29:48 +00:00
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, lr});
|
2013-01-23 16:29:48 +00:00
|
|
|
|
|
|
|
|
// Load a, b, c into d16, d17, d18.
|
|
|
|
|
__ mov(r4, Operand(r0));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d16, r4, offsetof(T, a));
|
|
|
|
|
__ vldr(d17, r4, offsetof(T, b));
|
|
|
|
|
__ vldr(d18, r4, offsetof(T, c));
|
2013-01-23 16:29:48 +00:00
|
|
|
|
|
|
|
|
__ vneg(d25, d16);
|
|
|
|
|
__ vadd(d25, d25, d17);
|
|
|
|
|
__ vsub(d25, d25, d18);
|
|
|
|
|
__ vmul(d25, d25, d25);
|
|
|
|
|
__ vdiv(d25, d25, d18);
|
|
|
|
|
|
|
|
|
|
__ vmov(d16, d25);
|
|
|
|
|
__ vsqrt(d17, d25);
|
|
|
|
|
__ vneg(d17, d17);
|
|
|
|
|
__ vabs(d17, d17);
|
|
|
|
|
__ vmla(d18, d16, d17);
|
|
|
|
|
|
|
|
|
|
// Store d16, d17, d18 into a, b, c.
|
|
|
|
|
__ mov(r4, Operand(r0));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d16, r4, offsetof(T, a));
|
|
|
|
|
__ vstr(d17, r4, offsetof(T, b));
|
|
|
|
|
__ vstr(d18, r4, offsetof(T, c));
|
2013-01-23 16:29:48 +00:00
|
|
|
|
|
|
|
|
// Load x, y, z into d29-d31.
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, x))));
|
2013-01-23 16:29:48 +00:00
|
|
|
__ vldm(ia_w, r4, d29, d31);
|
|
|
|
|
|
|
|
|
|
// Swap d29 and d30 via r registers.
|
|
|
|
|
__ vmov(r1, r2, d29);
|
|
|
|
|
__ vmov(d29, d30);
|
|
|
|
|
__ vmov(d30, r1, r2);
|
|
|
|
|
|
|
|
|
|
// Convert to and from integer.
|
|
|
|
|
__ vcvt_s32_f64(s1, d31);
|
|
|
|
|
__ vcvt_f64_u32(d31, s1);
|
|
|
|
|
|
|
|
|
|
// Store d29-d31 into x, y, z.
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, x))));
|
2013-01-23 16:29:48 +00:00
|
|
|
__ vstm(ia_w, r4, d29, d31);
|
|
|
|
|
|
|
|
|
|
// Move constants into d20, d21, d22 and store into i, j, k.
|
2021-06-24 13:32:01 +00:00
|
|
|
__ vmov(d20, base::Double(14.7610017472335499));
|
|
|
|
|
__ vmov(d21, base::Double(16.0));
|
2013-01-23 16:29:48 +00:00
|
|
|
__ mov(r1, Operand(372106121));
|
|
|
|
|
__ mov(r2, Operand(1079146608));
|
2018-07-24 16:56:27 +00:00
|
|
|
__ vmov(NeonS32, d22, 0, r1);
|
|
|
|
|
__ vmov(NeonS32, d22, 1, r2);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, i))));
|
2013-01-23 16:29:48 +00:00
|
|
|
__ vstm(ia_w, r4, d20, d22);
|
2013-07-25 15:04:38 +00:00
|
|
|
// Move d22 into low and high.
|
2018-07-24 16:56:27 +00:00
|
|
|
__ vmov(NeonS32, r4, d22, 0);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r4, MemOperand(r0, offsetof(T, low)));
|
2018-07-24 16:56:27 +00:00
|
|
|
__ vmov(NeonS32, r4, d22, 1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r4, MemOperand(r0, offsetof(T, high)));
|
2013-01-23 16:29:48 +00:00
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, pc});
|
2013-01-23 16:29:48 +00:00
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2013-01-23 16:29:48 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2013-01-23 16:29:48 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2013-01-23 16:29:48 +00:00
|
|
|
t.a = 1.5;
|
|
|
|
|
t.b = 2.75;
|
|
|
|
|
t.c = 17.17;
|
|
|
|
|
t.x = 1.5;
|
|
|
|
|
t.y = 2.75;
|
|
|
|
|
t.z = 17.17;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, 0, 0, 0, 0);
|
2013-01-23 16:29:48 +00:00
|
|
|
CHECK_EQ(14.7610017472335499, t.a);
|
|
|
|
|
CHECK_EQ(3.84200491244266251, t.b);
|
|
|
|
|
CHECK_EQ(73.8818412254460241, t.c);
|
|
|
|
|
CHECK_EQ(2.75, t.x);
|
|
|
|
|
CHECK_EQ(1.5, t.y);
|
|
|
|
|
CHECK_EQ(17.0, t.z);
|
|
|
|
|
CHECK_EQ(14.7610017472335499, t.i);
|
|
|
|
|
CHECK_EQ(16.0, t.j);
|
|
|
|
|
CHECK_EQ(73.8818412254460241, t.k);
|
2015-01-30 09:29:25 +00:00
|
|
|
CHECK_EQ(372106121u, t.low);
|
|
|
|
|
CHECK_EQ(1079146608u, t.high);
|
2013-01-23 16:29:48 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2013-04-15 15:55:47 +00:00
|
|
|
|
|
|
|
|
TEST(14) {
|
|
|
|
|
// Test the VFP Canonicalized Nan mode.
|
|
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-04-15 15:55:47 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2013-04-15 15:55:47 +00:00
|
|
|
double left;
|
|
|
|
|
double right;
|
|
|
|
|
double add_result;
|
|
|
|
|
double sub_result;
|
|
|
|
|
double mul_result;
|
|
|
|
|
double div_result;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2013-04-15 15:55:47 +00:00
|
|
|
T t;
|
|
|
|
|
|
|
|
|
|
// Create a function that makes the four basic operations.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2013-04-15 15:55:47 +00:00
|
|
|
|
2018-12-07 08:36:31 +00:00
|
|
|
// Ensure FPSCR state (as JSEntry does).
|
2013-04-15 15:55:47 +00:00
|
|
|
Label fpscr_done;
|
|
|
|
|
__ vmrs(r1);
|
|
|
|
|
__ tst(r1, Operand(kVFPDefaultNaNModeControlBit));
|
|
|
|
|
__ b(ne, &fpscr_done);
|
|
|
|
|
__ orr(r1, r1, Operand(kVFPDefaultNaNModeControlBit));
|
|
|
|
|
__ vmsr(r1);
|
|
|
|
|
__ bind(&fpscr_done);
|
|
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d0, r0, offsetof(T, left));
|
|
|
|
|
__ vldr(d1, r0, offsetof(T, right));
|
2013-04-15 15:55:47 +00:00
|
|
|
__ vadd(d2, d0, d1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d2, r0, offsetof(T, add_result));
|
2013-04-15 15:55:47 +00:00
|
|
|
__ vsub(d2, d0, d1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d2, r0, offsetof(T, sub_result));
|
2013-04-15 15:55:47 +00:00
|
|
|
__ vmul(d2, d0, d1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d2, r0, offsetof(T, mul_result));
|
2013-04-15 15:55:47 +00:00
|
|
|
__ vdiv(d2, d0, d1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d2, r0, offsetof(T, div_result));
|
2013-04-15 15:55:47 +00:00
|
|
|
|
|
|
|
|
__ mov(pc, Operand(lr));
|
|
|
|
|
|
|
|
|
|
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();
|
2013-04-15 15:55:47 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2013-04-15 15:55:47 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2022-05-05 13:37:00 +00:00
|
|
|
t.left = base::bit_cast<double>(kHoleNanInt64);
|
2013-04-15 15:55:47 +00:00
|
|
|
t.right = 1;
|
|
|
|
|
t.add_result = 0;
|
|
|
|
|
t.sub_result = 0;
|
|
|
|
|
t.mul_result = 0;
|
|
|
|
|
t.div_result = 0;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, 0, 0, 0, 0);
|
2017-12-02 00:30:37 +00:00
|
|
|
const uint32_t kArmNanUpper32 = 0x7FF80000;
|
2013-04-15 15:55:47 +00:00
|
|
|
const uint32_t kArmNanLower32 = 0x00000000;
|
|
|
|
|
#ifdef DEBUG
|
|
|
|
|
const uint64_t kArmNanInt64 =
|
|
|
|
|
(static_cast<uint64_t>(kArmNanUpper32) << 32) | kArmNanLower32;
|
2017-10-18 09:06:55 +00:00
|
|
|
CHECK_NE(kArmNanInt64, kHoleNanInt64);
|
2013-04-15 15:55:47 +00:00
|
|
|
#endif
|
|
|
|
|
// With VFP2 the sign of the canonicalized Nan is undefined. So
|
|
|
|
|
// we remove the sign bit for the upper tests.
|
2014-09-08 09:11:11 +00:00
|
|
|
CHECK_EQ(kArmNanUpper32,
|
2022-05-05 13:37:00 +00:00
|
|
|
(base::bit_cast<int64_t>(t.add_result) >> 32) & 0x7FFFFFFF);
|
|
|
|
|
CHECK_EQ(kArmNanLower32, base::bit_cast<int64_t>(t.add_result) & 0xFFFFFFFFu);
|
2014-09-08 09:11:11 +00:00
|
|
|
CHECK_EQ(kArmNanUpper32,
|
2022-05-05 13:37:00 +00:00
|
|
|
(base::bit_cast<int64_t>(t.sub_result) >> 32) & 0x7FFFFFFF);
|
|
|
|
|
CHECK_EQ(kArmNanLower32, base::bit_cast<int64_t>(t.sub_result) & 0xFFFFFFFFu);
|
2014-09-08 09:11:11 +00:00
|
|
|
CHECK_EQ(kArmNanUpper32,
|
2022-05-05 13:37:00 +00:00
|
|
|
(base::bit_cast<int64_t>(t.mul_result) >> 32) & 0x7FFFFFFF);
|
|
|
|
|
CHECK_EQ(kArmNanLower32, base::bit_cast<int64_t>(t.mul_result) & 0xFFFFFFFFu);
|
2014-09-08 09:11:11 +00:00
|
|
|
CHECK_EQ(kArmNanUpper32,
|
2022-05-05 13:37:00 +00:00
|
|
|
(base::bit_cast<int64_t>(t.div_result) >> 32) & 0x7FFFFFFF);
|
|
|
|
|
CHECK_EQ(kArmNanLower32, base::bit_cast<int64_t>(t.div_result) & 0xFFFFFFFFu);
|
2013-04-15 15:55:47 +00:00
|
|
|
}
|
|
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
#define CHECK_EQ_SPLAT(field, ex) \
|
|
|
|
|
CHECK_EQ(ex, t.field[0]); \
|
|
|
|
|
CHECK_EQ(ex, t.field[1]); \
|
|
|
|
|
CHECK_EQ(ex, t.field[2]); \
|
|
|
|
|
CHECK_EQ(ex, t.field[3]);
|
|
|
|
|
|
2017-03-02 19:50:33 +00:00
|
|
|
#define CHECK_EQ_32X2(field, ex0, ex1) \
|
|
|
|
|
CHECK_EQ(ex0, t.field[0]); \
|
|
|
|
|
CHECK_EQ(ex1, t.field[1]);
|
|
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
#define CHECK_EQ_32X4(field, ex0, ex1, ex2, ex3) \
|
|
|
|
|
CHECK_EQ(ex0, t.field[0]); \
|
|
|
|
|
CHECK_EQ(ex1, t.field[1]); \
|
|
|
|
|
CHECK_EQ(ex2, t.field[2]); \
|
|
|
|
|
CHECK_EQ(ex3, t.field[3]);
|
|
|
|
|
|
2017-01-13 12:03:23 +00:00
|
|
|
#define CHECK_ESTIMATE(expected, tolerance, value) \
|
|
|
|
|
CHECK_LT((expected) - (tolerance), value); \
|
|
|
|
|
CHECK_GT((expected) + (tolerance), value);
|
|
|
|
|
|
|
|
|
|
#define CHECK_ESTIMATE_SPLAT(field, ex, tol) \
|
|
|
|
|
CHECK_ESTIMATE(ex, tol, t.field[0]); \
|
|
|
|
|
CHECK_ESTIMATE(ex, tol, t.field[1]); \
|
|
|
|
|
CHECK_ESTIMATE(ex, tol, t.field[2]); \
|
|
|
|
|
CHECK_ESTIMATE(ex, tol, t.field[3]);
|
|
|
|
|
|
2016-12-15 18:15:23 +00:00
|
|
|
#define INT32_TO_FLOAT(val) \
|
2022-05-05 13:37:00 +00:00
|
|
|
std::round(static_cast<float>(base::bit_cast<int32_t>(val)))
|
2016-12-15 18:15:23 +00:00
|
|
|
#define UINT32_TO_FLOAT(val) \
|
2022-05-05 13:37:00 +00:00
|
|
|
std::round(static_cast<float>(base::bit_cast<uint32_t>(val)))
|
2013-07-10 15:32:39 +00:00
|
|
|
|
|
|
|
|
TEST(15) {
|
|
|
|
|
// Test the Neon instructions.
|
|
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-07-10 15:32:39 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2013-07-10 15:32:39 +00:00
|
|
|
uint32_t src0;
|
|
|
|
|
uint32_t src1;
|
|
|
|
|
uint32_t src2;
|
|
|
|
|
uint32_t src3;
|
|
|
|
|
uint32_t src4;
|
|
|
|
|
uint32_t src5;
|
|
|
|
|
uint32_t src6;
|
|
|
|
|
uint32_t src7;
|
|
|
|
|
uint32_t dst0;
|
|
|
|
|
uint32_t dst1;
|
|
|
|
|
uint32_t dst2;
|
|
|
|
|
uint32_t dst3;
|
|
|
|
|
uint32_t dst4;
|
|
|
|
|
uint32_t dst5;
|
|
|
|
|
uint32_t dst6;
|
|
|
|
|
uint32_t dst7;
|
|
|
|
|
uint32_t srcA0;
|
|
|
|
|
uint32_t srcA1;
|
|
|
|
|
uint32_t dstA0;
|
|
|
|
|
uint32_t dstA1;
|
|
|
|
|
uint32_t dstA2;
|
|
|
|
|
uint32_t dstA3;
|
2016-12-17 21:56:26 +00:00
|
|
|
uint32_t lane_test[4];
|
2016-12-15 18:15:23 +00:00
|
|
|
uint64_t vmov_to_scalar1, vmov_to_scalar2;
|
|
|
|
|
uint32_t vmov_from_scalar_s8, vmov_from_scalar_u8;
|
|
|
|
|
uint32_t vmov_from_scalar_s16, vmov_from_scalar_u16;
|
|
|
|
|
uint32_t vmov_from_scalar_32;
|
2016-12-17 21:56:26 +00:00
|
|
|
uint32_t vmov[4], vmvn[4];
|
2017-03-27 18:23:42 +00:00
|
|
|
uint32_t vmovl_s8[4], vmovl_u16[4], vmovl_s32[4];
|
|
|
|
|
uint32_t vqmovn_s8[2], vqmovn_u16[2], vqmovn_s32[2];
|
2016-12-15 18:15:23 +00:00
|
|
|
int32_t vcvt_s32_f32[4];
|
|
|
|
|
uint32_t vcvt_u32_f32[4];
|
|
|
|
|
float vcvt_f32_s32[4], vcvt_f32_u32[4];
|
2016-12-17 21:56:26 +00:00
|
|
|
uint32_t vdup8[4], vdup16[4], vdup32[4];
|
|
|
|
|
float vabsf[4], vnegf[4];
|
2020-12-10 00:47:56 +00:00
|
|
|
uint32_t vclt0_s8[4], vclt0_s16[4], vclt0_s32[4];
|
2016-12-17 21:56:26 +00:00
|
|
|
uint32_t vabs_s8[4], vabs_s16[4], vabs_s32[4];
|
|
|
|
|
uint32_t vneg_s8[4], vneg_s16[4], vneg_s32[4];
|
2017-01-12 15:36:29 +00:00
|
|
|
uint32_t veor[4], vand[4], vorr[4];
|
2017-04-24 18:53:16 +00:00
|
|
|
float vdupf[4], vaddf[4], vpaddf[2], vsubf[4], vmulf[4];
|
2017-05-16 20:37:16 +00:00
|
|
|
uint32_t vdupf_16[2], vdupf_8[4];
|
2017-01-16 19:36:25 +00:00
|
|
|
uint32_t vmin_s8[4], vmin_u16[4], vmin_s32[4];
|
|
|
|
|
uint32_t vmax_s8[4], vmax_u16[4], vmax_s32[4];
|
2017-04-24 18:53:16 +00:00
|
|
|
uint32_t vpadd_i8[2], vpadd_i16[2], vpadd_i32[2];
|
2021-06-28 13:34:37 +00:00
|
|
|
uint32_t vpadal_s8[4], vpadal_s16[4], vpadal_s32[4];
|
|
|
|
|
uint32_t vpadal_u8[4], vpadal_u16[4], vpadal_u32[4];
|
2020-12-08 06:18:13 +00:00
|
|
|
uint32_t vpaddl_s8[4], vpaddl_s16[4], vpaddl_s32[4];
|
|
|
|
|
uint32_t vpaddl_u8[4], vpaddl_u16[4], vpaddl_u32[4];
|
2017-03-02 19:50:33 +00:00
|
|
|
uint32_t vpmin_s8[2], vpmin_u16[2], vpmin_s32[2];
|
|
|
|
|
uint32_t vpmax_s8[2], vpmax_u16[2], vpmax_s32[2];
|
2016-12-15 18:15:23 +00:00
|
|
|
uint32_t vadd8[4], vadd16[4], vadd32[4];
|
2017-02-01 18:56:28 +00:00
|
|
|
uint32_t vqadd_s8[4], vqadd_u16[4], vqadd_s32[4];
|
2016-12-15 18:15:23 +00:00
|
|
|
uint32_t vsub8[4], vsub16[4], vsub32[4];
|
2017-02-01 18:56:28 +00:00
|
|
|
uint32_t vqsub_u8[4], vqsub_s16[4], vqsub_u32[4];
|
2016-12-17 21:56:26 +00:00
|
|
|
uint32_t vmul8[4], vmul16[4], vmul32[4];
|
2020-11-30 01:05:20 +00:00
|
|
|
uint32_t vshl8[4], vshl16[4], vshl32[4];
|
|
|
|
|
uint32_t vshr_s8[4], vshr_u16[4], vshr_s32[4];
|
|
|
|
|
uint32_t vshr_s8_d[2], vshr_u16_d[2], vshr_s32_d[2];
|
2017-05-16 20:37:16 +00:00
|
|
|
uint32_t vsli_64[2], vsri_64[2], vsli_32[2], vsri_32[2];
|
2020-11-30 01:05:20 +00:00
|
|
|
uint32_t vsra_64[2], vsra_32[2], vsra_16[2];
|
2017-01-12 11:20:08 +00:00
|
|
|
uint32_t vceq[4], vceqf[4], vcgef[4], vcgtf[4];
|
|
|
|
|
uint32_t vcge_s8[4], vcge_u16[4], vcge_s32[4];
|
|
|
|
|
uint32_t vcgt_s8[4], vcgt_u16[4], vcgt_s32[4];
|
2017-01-10 12:36:59 +00:00
|
|
|
float vrecpe[4], vrecps[4], vrsqrte[4], vrsqrts[4];
|
2017-01-16 19:36:25 +00:00
|
|
|
float vminf[4], vmaxf[4];
|
2017-01-12 11:20:08 +00:00
|
|
|
uint32_t vtst[4], vbsl[4];
|
2016-12-17 21:56:26 +00:00
|
|
|
uint32_t vext[4];
|
|
|
|
|
uint32_t vzip8a[4], vzip8b[4], vzip16a[4], vzip16b[4], vzip32a[4],
|
|
|
|
|
vzip32b[4];
|
2017-04-10 21:41:02 +00:00
|
|
|
uint32_t vzipd8a[2], vzipd8b[2], vzipd16a[2], vzipd16b[2];
|
2017-03-14 16:28:45 +00:00
|
|
|
uint32_t vuzp8a[4], vuzp8b[4], vuzp16a[4], vuzp16b[4], vuzp32a[4],
|
|
|
|
|
vuzp32b[4];
|
2017-04-10 21:41:02 +00:00
|
|
|
uint32_t vuzpd8a[2], vuzpd8b[2], vuzpd16a[2], vuzpd16b[2];
|
2016-12-17 21:56:26 +00:00
|
|
|
uint32_t vrev64_32[4], vrev64_16[4], vrev64_8[4];
|
2017-03-14 16:28:45 +00:00
|
|
|
uint32_t vrev32_16[4], vrev32_8[4], vrev16_8[4];
|
|
|
|
|
uint32_t vtrn8a[4], vtrn8b[4], vtrn16a[4], vtrn16b[4], vtrn32a[4],
|
|
|
|
|
vtrn32b[4];
|
2017-04-10 21:41:02 +00:00
|
|
|
uint32_t vtrnd8a[2], vtrnd8b[2], vtrnd16a[2], vtrnd16b[2], vtrnd32a[2],
|
|
|
|
|
vtrnd32b[2];
|
2016-12-17 21:56:26 +00:00
|
|
|
uint32_t vtbl[2], vtbx[2];
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2013-07-10 15:32:39 +00:00
|
|
|
T t;
|
|
|
|
|
|
|
|
|
|
// Create a function that accepts &t, and loads, manipulates, and stores
|
2016-12-17 21:56:26 +00:00
|
|
|
// the doubles, floats, and SIMD values.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2013-07-10 15:32:39 +00:00
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(NEON)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, NEON);
|
|
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, r5, lr});
|
2013-07-10 15:32:39 +00:00
|
|
|
// Move 32 bytes with neon.
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, src0))));
|
2013-07-10 15:32:39 +00:00
|
|
|
__ vld1(Neon8, NeonListOperand(d0, 4), NeonMemOperand(r4));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, dst0))));
|
2013-07-10 15:32:39 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(d0, 4), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
// Expand 8 bytes into 8 words(16 bits).
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, srcA0))));
|
2013-07-10 15:32:39 +00:00
|
|
|
__ vld1(Neon8, NeonListOperand(d0), NeonMemOperand(r4));
|
|
|
|
|
__ vmovl(NeonU8, q0, d0);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, dstA0))));
|
2013-07-10 15:32:39 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(d0, 2), NeonMemOperand(r4));
|
|
|
|
|
|
2014-03-21 13:04:20 +00:00
|
|
|
// The same expansion, but with different source and destination registers.
|
2015-06-17 09:06:44 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, srcA0))));
|
2014-03-21 13:04:20 +00:00
|
|
|
__ vld1(Neon8, NeonListOperand(d1), NeonMemOperand(r4));
|
2017-03-27 18:23:42 +00:00
|
|
|
__ vmovl(NeonS8, q1, d1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmovl_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vmovl(NeonU16, q2, d3);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmovl_u16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
__ vmovl(NeonS32, q3, d4);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmovl_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q3), NeonMemOperand(r4));
|
|
|
|
|
// Narrow what we widened.
|
[wasm-simd] Fix unsigned narrow instructions
These instructions should always treat inputs as signed, and saturate to
unsigned min/max values.
E.g. given -1, it should saturate to 0.
The spec text,
https://github.com/WebAssembly/simd/blob/master/proposals/simd/SIMD.md#integer-to-integer-narrowing,
has been updated to describe this.
The changes here include codegen changes to ia32, x64, arm, and arm64,
changes to arm simulator, assembler, and disassembler to handle the case
of treating input as signed and narrowing to unsigned. The vqmovn
instruction can handle this case, our assembler wasn't allowing callers
to specify this.
The interpreter and scalar lowering are also fixed with this change.
Bug: v8:9729
Change-Id: I6f72baa825f59037f7754485df6a2964af59fe31
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1879423
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Bill Budge <bbudge@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65051}
2019-10-28 18:05:16 +00:00
|
|
|
__ vqmovn(NeonU16, NeonU16, d0, q2);
|
2017-03-27 18:23:42 +00:00
|
|
|
__ vstr(d0, r0, offsetof(T, vqmovn_u16));
|
|
|
|
|
__ vmov(d1, d0);
|
[wasm-simd] Fix unsigned narrow instructions
These instructions should always treat inputs as signed, and saturate to
unsigned min/max values.
E.g. given -1, it should saturate to 0.
The spec text,
https://github.com/WebAssembly/simd/blob/master/proposals/simd/SIMD.md#integer-to-integer-narrowing,
has been updated to describe this.
The changes here include codegen changes to ia32, x64, arm, and arm64,
changes to arm simulator, assembler, and disassembler to handle the case
of treating input as signed and narrowing to unsigned. The vqmovn
instruction can handle this case, our assembler wasn't allowing callers
to specify this.
The interpreter and scalar lowering are also fixed with this change.
Bug: v8:9729
Change-Id: I6f72baa825f59037f7754485df6a2964af59fe31
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1879423
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Bill Budge <bbudge@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65051}
2019-10-28 18:05:16 +00:00
|
|
|
__ vqmovn(NeonS8, NeonS8, d2, q0);
|
2017-03-27 18:23:42 +00:00
|
|
|
__ vstr(d2, r0, offsetof(T, vqmovn_s8));
|
[wasm-simd] Fix unsigned narrow instructions
These instructions should always treat inputs as signed, and saturate to
unsigned min/max values.
E.g. given -1, it should saturate to 0.
The spec text,
https://github.com/WebAssembly/simd/blob/master/proposals/simd/SIMD.md#integer-to-integer-narrowing,
has been updated to describe this.
The changes here include codegen changes to ia32, x64, arm, and arm64,
changes to arm simulator, assembler, and disassembler to handle the case
of treating input as signed and narrowing to unsigned. The vqmovn
instruction can handle this case, our assembler wasn't allowing callers
to specify this.
The interpreter and scalar lowering are also fixed with this change.
Bug: v8:9729
Change-Id: I6f72baa825f59037f7754485df6a2964af59fe31
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1879423
Reviewed-by: Deepti Gandluri <gdeepti@chromium.org>
Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Reviewed-by: Bill Budge <bbudge@chromium.org>
Commit-Queue: Zhi An Ng <zhin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#65051}
2019-10-28 18:05:16 +00:00
|
|
|
__ vqmovn(NeonS32, NeonS32, d4, q3);
|
2017-03-27 18:23:42 +00:00
|
|
|
__ vstr(d4, r0, offsetof(T, vqmovn_s32));
|
2014-03-21 13:04:20 +00:00
|
|
|
|
2016-12-15 18:15:23 +00:00
|
|
|
// ARM core register to scalar.
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFFFFFFF8));
|
2021-06-24 13:32:01 +00:00
|
|
|
__ vmov(d0, base::Double(0.0));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vmov(NeonS8, d0, 1, r4);
|
|
|
|
|
__ vmov(NeonS16, d0, 1, r4);
|
|
|
|
|
__ vmov(NeonS32, d0, 1, r4);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vmov_to_scalar1));
|
2021-06-24 13:32:01 +00:00
|
|
|
__ vmov(d0, base::Double(0.0));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vmov(NeonS8, d0, 3, r4);
|
|
|
|
|
__ vmov(NeonS16, d0, 3, r4);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vmov_to_scalar2));
|
|
|
|
|
|
|
|
|
|
// Scalar to ARM core register.
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFFFFFF00));
|
|
|
|
|
__ mov(r5, Operand(0xFFFFFFFF));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vmov(d0, r4, r5);
|
|
|
|
|
__ vmov(NeonS8, r4, d0, 1);
|
|
|
|
|
__ str(r4, MemOperand(r0, offsetof(T, vmov_from_scalar_s8)));
|
|
|
|
|
__ vmov(NeonU8, r4, d0, 1);
|
|
|
|
|
__ str(r4, MemOperand(r0, offsetof(T, vmov_from_scalar_u8)));
|
|
|
|
|
__ vmov(NeonS16, r4, d0, 1);
|
|
|
|
|
__ str(r4, MemOperand(r0, offsetof(T, vmov_from_scalar_s16)));
|
|
|
|
|
__ vmov(NeonU16, r4, d0, 1);
|
|
|
|
|
__ str(r4, MemOperand(r0, offsetof(T, vmov_from_scalar_u16)));
|
|
|
|
|
__ vmov(NeonS32, r4, d0, 1);
|
|
|
|
|
__ str(r4, MemOperand(r0, offsetof(T, vmov_from_scalar_32)));
|
|
|
|
|
|
|
|
|
|
// vmov for q-registers.
|
2016-12-17 21:56:26 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
2016-11-25 19:36:45 +00:00
|
|
|
__ vmov(q1, q0);
|
2016-12-17 21:56:26 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmov))));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
// vmvn.
|
2016-12-17 21:56:26 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vmvn(q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmvn))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
// vcvt for q-registers.
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s0, Float32(-1.5f));
|
|
|
|
|
__ vmov(s1, Float32(-1.0f));
|
|
|
|
|
__ vmov(s2, Float32(1.0f));
|
|
|
|
|
__ vmov(s3, Float32(1.5f));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vcvt_s32_f32(q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcvt_s32_f32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vcvt_u32_f32(q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcvt_u32_f32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(kMinInt));
|
|
|
|
|
__ mov(r5, Operand(kMaxInt));
|
|
|
|
|
__ vmov(d0, r4, r5);
|
|
|
|
|
__ mov(r4, Operand(kMaxUInt32));
|
|
|
|
|
__ mov(r5, Operand(kMinInt + 1));
|
|
|
|
|
__ vmov(d1, r4, r5); // q0 = [kMinInt, kMaxInt, kMaxUInt32, kMinInt + 1]
|
|
|
|
|
__ vcvt_f32_s32(q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcvt_f32_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vcvt_f32_u32(q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcvt_f32_u32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2017-05-16 20:37:16 +00:00
|
|
|
// vdup (from register).
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xA));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vdup(Neon32, q2, r4);
|
2016-12-17 21:56:26 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vdup8))));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
2016-12-17 21:56:26 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vdup16))));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
2016-12-17 21:56:26 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vdup32))));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
2016-12-17 21:56:26 +00:00
|
|
|
|
2017-05-16 20:37:16 +00:00
|
|
|
// vdup (from scalar).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s0, Float32(-1.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q1, d0, 0);
|
2016-12-17 21:56:26 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vdupf))));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vdup(Neon16, d2, d0, 1);
|
|
|
|
|
__ vstr(d2, r0, offsetof(T, vdupf_16));
|
|
|
|
|
__ vdup(Neon8, q1, d0, 3);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vdupf_8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
2016-12-15 18:15:23 +00:00
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
// vabs (float).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s0, Float32(-1.0f));
|
|
|
|
|
__ vmov(s1, Float32(-0.0f));
|
|
|
|
|
__ vmov(s2, Float32(0.0f));
|
|
|
|
|
__ vmov(s3, Float32(1.0f));
|
2016-12-17 21:56:26 +00:00
|
|
|
__ vabs(q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vabsf))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
// vneg (float).
|
|
|
|
|
__ vneg(q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vnegf))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
// vabs (integer).
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0x7F7F7F7F));
|
2016-12-17 21:56:26 +00:00
|
|
|
__ mov(r5, Operand(0x01010101));
|
|
|
|
|
__ vmov(d0, r4, r5);
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFFFFFFFF));
|
2016-12-17 21:56:26 +00:00
|
|
|
__ mov(r5, Operand(0x80808080));
|
|
|
|
|
__ vmov(d1, r4, r5);
|
|
|
|
|
__ vabs(Neon8, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vabs_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vabs(Neon16, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vabs_s16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vabs(Neon32, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vabs_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
// vneg (integer).
|
|
|
|
|
__ vneg(Neon8, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vneg_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vneg(Neon16, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vneg_s16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vneg(Neon32, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vneg_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2016-12-15 18:15:23 +00:00
|
|
|
// veor.
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xAA));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vdup(Neon16, q0, r4);
|
2017-01-12 15:36:29 +00:00
|
|
|
__ mov(r4, Operand(0x55));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vdup(Neon16, q1, r4);
|
2016-11-25 19:36:45 +00:00
|
|
|
__ veor(q1, q1, q0);
|
2016-12-15 18:15:23 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, veor))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
2017-01-12 15:36:29 +00:00
|
|
|
// vand.
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFF));
|
2017-01-12 15:36:29 +00:00
|
|
|
__ vdup(Neon16, q0, r4);
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFE));
|
2017-01-12 15:36:29 +00:00
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vand(q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vand))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
// vorr.
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xAA));
|
2017-01-12 15:36:29 +00:00
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ mov(r4, Operand(0x55));
|
|
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vorr(q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vorr))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
2016-12-15 18:15:23 +00:00
|
|
|
|
2017-01-16 19:36:25 +00:00
|
|
|
// vmin (float).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(2.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(1.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q1, d2, 0);
|
2017-01-16 19:36:25 +00:00
|
|
|
__ vmin(q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vminf))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
// vmax (float).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(2.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(1.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q1, d2, 0);
|
2017-01-16 19:36:25 +00:00
|
|
|
__ vmax(q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmaxf))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
2016-12-17 21:56:26 +00:00
|
|
|
// vadd (float).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(1.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
|
|
|
|
__ vdup(Neon32, q1, d2, 0);
|
2016-12-17 21:56:26 +00:00
|
|
|
__ vadd(q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vaddf))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
2017-04-24 18:53:16 +00:00
|
|
|
// vpadd (float).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s0, Float32(1.0f));
|
|
|
|
|
__ vmov(s1, Float32(2.0f));
|
|
|
|
|
__ vmov(s2, Float32(3.0f));
|
|
|
|
|
__ vmov(s3, Float32(4.0f));
|
2017-04-24 18:53:16 +00:00
|
|
|
__ vpadd(d2, d0, d1);
|
|
|
|
|
__ vstr(d2, r0, offsetof(T, vpaddf));
|
2016-12-17 21:56:26 +00:00
|
|
|
// vsub (float).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(2.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(1.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q1, d2, 0);
|
2016-12-17 21:56:26 +00:00
|
|
|
__ vsub(q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vsubf))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
// vmul (float).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(2.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
|
|
|
|
__ vdup(Neon32, q1, d2, 0);
|
2016-12-17 21:56:26 +00:00
|
|
|
__ vmul(q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmulf))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
2017-01-10 12:36:59 +00:00
|
|
|
// vrecpe.
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(2.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
2017-01-10 12:36:59 +00:00
|
|
|
__ vrecpe(q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrecpe))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
// vrecps.
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(2.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(1.5f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q1, d2, 0);
|
2017-01-10 12:36:59 +00:00
|
|
|
__ vrecps(q1, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrecps))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
// vrsqrte.
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(4.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
2017-01-10 12:36:59 +00:00
|
|
|
__ vrsqrte(q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrsqrte))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
// vrsqrts.
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(2.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(2.5f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q1, d2, 0);
|
2017-01-10 12:36:59 +00:00
|
|
|
__ vrsqrts(q1, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrsqrts))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
2016-12-20 13:06:58 +00:00
|
|
|
// vceq (float).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s4, Float32(1.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q0, d2, 0);
|
|
|
|
|
__ vdup(Neon32, q1, d2, 0);
|
2016-12-20 13:06:58 +00:00
|
|
|
__ vceq(q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vceqf))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
2017-01-12 11:20:08 +00:00
|
|
|
// vcge (float).
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s0, Float32(1.0f));
|
|
|
|
|
__ vmov(s1, Float32(-1.0f));
|
|
|
|
|
__ vmov(s2, Float32(-0.0f));
|
|
|
|
|
__ vmov(s3, Float32(0.0f));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q1, d1, 1);
|
2017-01-12 11:20:08 +00:00
|
|
|
__ vcge(q2, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcgef))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
__ vcgt(q2, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcgtf))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
2016-12-17 21:56:26 +00:00
|
|
|
|
2017-01-16 19:36:25 +00:00
|
|
|
// vmin/vmax integer.
|
|
|
|
|
__ mov(r4, Operand(0x03));
|
|
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vmin(NeonS8, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmin_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
__ vmax(NeonS8, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmax_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFF));
|
2017-01-16 19:36:25 +00:00
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vmin(NeonU16, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmin_u16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
__ vmax(NeonU16, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmax_u16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFF));
|
2017-01-16 19:36:25 +00:00
|
|
|
__ vdup(Neon32, q0, r4);
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vmin(NeonS32, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmin_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
__ vmax(NeonS32, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmax_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
2017-04-24 18:53:16 +00:00
|
|
|
// vpadd integer.
|
|
|
|
|
__ mov(r4, Operand(0x03));
|
|
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vpadd(Neon8, d0, d0, d2);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vpadd_i8));
|
|
|
|
|
__ vpadd(Neon16, d0, d0, d2);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vpadd_i16));
|
|
|
|
|
__ vpadd(Neon32, d0, d0, d2);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vpadd_i32));
|
|
|
|
|
|
2021-06-28 13:34:37 +00:00
|
|
|
// vpadal signed.
|
|
|
|
|
__ mov(r4, Operand(0x81));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
|
|
|
|
|
__ mov(r4, Operand(0x01));
|
|
|
|
|
__ vdup(Neon8, q2, r4);
|
|
|
|
|
__ vpadal(NeonS8, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpadal_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ mov(r4, Operand(0x01));
|
|
|
|
|
__ vdup(Neon8, q2, r4);
|
|
|
|
|
__ vpadal(NeonS16, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpadal_s16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ mov(r4, Operand(0x01));
|
|
|
|
|
__ vdup(Neon8, q2, r4);
|
|
|
|
|
__ vpadal(NeonS32, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpadal_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
// vpadal unsigned.
|
|
|
|
|
__ mov(r4, Operand(0x01));
|
|
|
|
|
__ vdup(Neon8, q2, r4);
|
|
|
|
|
__ vpadal(NeonU8, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpadal_u8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ mov(r4, Operand(0x01));
|
|
|
|
|
__ vdup(Neon8, q2, r4);
|
|
|
|
|
__ vpadal(NeonU16, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpadal_u16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ mov(r4, Operand(0x01));
|
|
|
|
|
__ vdup(Neon8, q2, r4);
|
|
|
|
|
__ vpadal(NeonU32, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpadal_u32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
2020-12-08 06:18:13 +00:00
|
|
|
// vpaddl signed.
|
|
|
|
|
__ mov(r4, Operand(0x81));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
|
|
|
|
|
__ vpaddl(NeonS8, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpaddl_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ vpaddl(NeonS16, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpaddl_s16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ vpaddl(NeonS32, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpaddl_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
// vpaddl unsigned.
|
|
|
|
|
__ vpaddl(NeonU8, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpaddl_u8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ vpaddl(NeonU16, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpaddl_u16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ vpaddl(NeonU32, q2, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vpaddl_u32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
2017-03-02 19:50:33 +00:00
|
|
|
// vpmin/vpmax integer.
|
|
|
|
|
__ mov(r4, Operand(0x03));
|
|
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vpmin(NeonS8, d4, d0, d2);
|
|
|
|
|
__ vstr(d4, r0, offsetof(T, vpmin_s8));
|
|
|
|
|
__ vpmax(NeonS8, d4, d0, d2);
|
|
|
|
|
__ vstr(d4, r0, offsetof(T, vpmax_s8));
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFFFF));
|
2017-03-02 19:50:33 +00:00
|
|
|
__ vdup(Neon32, q0, r4);
|
|
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vpmin(NeonU16, d4, d0, d2);
|
|
|
|
|
__ vstr(d4, r0, offsetof(T, vpmin_u16));
|
|
|
|
|
__ vpmax(NeonU16, d4, d0, d2);
|
|
|
|
|
__ vstr(d4, r0, offsetof(T, vpmax_u16));
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFF));
|
2017-03-02 19:50:33 +00:00
|
|
|
__ veor(q0, q0, q0);
|
|
|
|
|
__ vmov(s0, r4);
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vpmin(NeonS32, d4, d0, d2);
|
|
|
|
|
__ vstr(d4, r0, offsetof(T, vpmin_s32));
|
|
|
|
|
__ vpmax(NeonS32, d4, d0, d2);
|
|
|
|
|
__ vstr(d4, r0, offsetof(T, vpmax_s32));
|
|
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
// vadd (integer).
|
2016-12-15 18:15:23 +00:00
|
|
|
__ mov(r4, Operand(0x81));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
__ mov(r4, Operand(0x82));
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vadd(Neon8, q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vadd8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x8001));
|
|
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ mov(r4, Operand(0x8002));
|
|
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vadd(Neon16, q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vadd16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x80000001));
|
|
|
|
|
__ vdup(Neon32, q0, r4);
|
|
|
|
|
__ mov(r4, Operand(0x80000002));
|
|
|
|
|
__ vdup(Neon32, q1, r4);
|
|
|
|
|
__ vadd(Neon32, q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vadd32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2017-02-01 18:56:28 +00:00
|
|
|
// vqadd.
|
|
|
|
|
__ mov(r4, Operand(0x81));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
__ mov(r4, Operand(0x82));
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vqadd(NeonS8, q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vqadd_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x8000));
|
|
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vqadd(NeonU16, q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vqadd_u16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x80000001));
|
|
|
|
|
__ vdup(Neon32, q0, r4);
|
|
|
|
|
__ mov(r4, Operand(0x80000002));
|
|
|
|
|
__ vdup(Neon32, q1, r4);
|
|
|
|
|
__ vqadd(NeonS32, q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vqadd_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
// vsub (integer).
|
2016-12-15 18:15:23 +00:00
|
|
|
__ mov(r4, Operand(0x01));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
2016-12-17 21:56:26 +00:00
|
|
|
__ mov(r4, Operand(0x03));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vsub(Neon8, q1, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vsub8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x0001));
|
|
|
|
|
__ vdup(Neon16, q0, r4);
|
2016-12-17 21:56:26 +00:00
|
|
|
__ mov(r4, Operand(0x0003));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vsub(Neon16, q1, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vsub16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x00000001));
|
|
|
|
|
__ vdup(Neon32, q0, r4);
|
2016-12-17 21:56:26 +00:00
|
|
|
__ mov(r4, Operand(0x00000003));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vdup(Neon32, q1, r4);
|
|
|
|
|
__ vsub(Neon32, q1, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vsub32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2017-02-01 18:56:28 +00:00
|
|
|
// vqsub.
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0x7F));
|
2017-02-01 18:56:28 +00:00
|
|
|
__ vdup(Neon8, q0, r4);
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0x3F));
|
2017-02-01 18:56:28 +00:00
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vqsub(NeonU8, q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vqsub_u8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x8000));
|
|
|
|
|
__ vdup(Neon16, q0, r4);
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0x7FFF));
|
2017-02-01 18:56:28 +00:00
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vqsub(NeonS16, q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vqsub_s16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x80000001));
|
|
|
|
|
__ vdup(Neon32, q0, r4);
|
|
|
|
|
__ mov(r4, Operand(0x80000000));
|
|
|
|
|
__ vdup(Neon32, q1, r4);
|
|
|
|
|
__ vqsub(NeonU32, q1, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vqsub_u32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
// vmul (integer).
|
|
|
|
|
__ mov(r4, Operand(0x02));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
__ vmul(Neon8, q1, q0, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmul8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x0002));
|
|
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ vmul(Neon16, q1, q0, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmul16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ mov(r4, Operand(0x00000002));
|
|
|
|
|
__ vdup(Neon32, q0, r4);
|
|
|
|
|
__ vmul(Neon32, q1, q0, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vmul32))));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2017-01-23 18:24:27 +00:00
|
|
|
// vshl.
|
|
|
|
|
__ mov(r4, Operand(0x55));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
__ vshl(NeonS8, q1, q0, 1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vshl8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vshl(NeonU16, q1, q0, 9);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vshl16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vshl(NeonS32, q1, q0, 17);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vshl32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
// vshr.s, vshr.u.
|
|
|
|
|
__ mov(r4, Operand(0x80));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
__ vshr(NeonS8, q1, q0, 1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vshr_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vshr(NeonU16, q1, q0, 9);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vshr_u16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vshr(NeonS32, q1, q0, 17);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vshr_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2020-11-30 01:05:20 +00:00
|
|
|
// vshr.s, vshr.u with d registers.
|
|
|
|
|
__ mov(r4, Operand(0x80));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
__ vshr(NeonS8, d1, d0, 1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vshr_s8_d))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(d1), NeonMemOperand(r4));
|
|
|
|
|
__ vshr(NeonU16, d2, d0, 9);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vshr_u16_d))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vshr(NeonS32, d2, d0, 17);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vshr_s32_d))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2017-05-16 20:37:16 +00:00
|
|
|
// vsli, vsri.
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFFFFFFFF));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ mov(r5, Operand(0x1));
|
|
|
|
|
__ vmov(d0, r4, r5);
|
|
|
|
|
__ vmov(d1, r5, r5);
|
|
|
|
|
__ vsli(Neon64, d1, d0, 32);
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vsli_64));
|
|
|
|
|
__ vmov(d0, r5, r4);
|
|
|
|
|
__ vmov(d1, r5, r5);
|
|
|
|
|
__ vsri(Neon64, d1, d0, 32);
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vsri_64));
|
|
|
|
|
__ vmov(d0, r4, r5);
|
|
|
|
|
__ vmov(d1, r5, r5);
|
|
|
|
|
__ vsli(Neon32, d1, d0, 16);
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vsli_32));
|
|
|
|
|
__ vmov(d0, r5, r4);
|
|
|
|
|
__ vmov(d1, r5, r5);
|
|
|
|
|
__ vsri(Neon32, d1, d0, 16);
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vsri_32));
|
|
|
|
|
|
2020-11-30 01:05:20 +00:00
|
|
|
// vsra.
|
|
|
|
|
__ vmov(d0, r4, r5);
|
|
|
|
|
// Check same dst and src registers.
|
|
|
|
|
__ vsra(NeonU64, d0, d0, 1);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vsra_64));
|
|
|
|
|
__ vmov(d0, r4, r5);
|
|
|
|
|
__ vmov(d1, r5, r4);
|
|
|
|
|
__ vsra(NeonS32, d1, d0, 16);
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vsra_32));
|
|
|
|
|
__ vmov(d0, r4, r5);
|
|
|
|
|
__ vmov(d1, r5, r4);
|
|
|
|
|
__ vsra(NeonU16, d1, d0, 2);
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vsra_16));
|
|
|
|
|
|
2016-12-15 18:15:23 +00:00
|
|
|
// vceq.
|
|
|
|
|
__ mov(r4, Operand(0x03));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vceq(Neon8, q1, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vceq))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2017-01-16 19:36:25 +00:00
|
|
|
// vcge/vcgt (integer).
|
2017-01-12 11:20:08 +00:00
|
|
|
__ mov(r4, Operand(0x03));
|
|
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vcge(NeonS8, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcge_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
__ vcgt(NeonS8, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcgt_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFF));
|
2017-01-12 11:20:08 +00:00
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vcge(NeonU16, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcge_u16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
__ vcgt(NeonU16, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcgt_u16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0xFF));
|
2017-01-12 11:20:08 +00:00
|
|
|
__ vdup(Neon32, q0, r4);
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ vcge(NeonS32, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcge_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
__ vcgt(NeonS32, q2, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vcgt_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
2020-12-10 00:47:56 +00:00
|
|
|
// vclt #0 (signed integer).
|
|
|
|
|
__ mov(r4, Operand(0x7F80807F));
|
|
|
|
|
__ mov(r5, Operand(0xFF01FF01));
|
|
|
|
|
__ vmov(d0, r4, r5);
|
|
|
|
|
__ mov(r4, Operand(0xFF0000FF));
|
|
|
|
|
__ mov(r5, Operand(0x00800080));
|
|
|
|
|
__ vmov(d1, r4, r5);
|
|
|
|
|
|
|
|
|
|
__ vclt(Neon8, q1, q0, 0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vclt0_s8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ vclt(Neon16, q1, q0, 0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vclt0_s16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
__ vclt(Neon32, q1, q0, 0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vclt0_s32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2016-12-15 18:15:23 +00:00
|
|
|
// vtst.
|
|
|
|
|
__ mov(r4, Operand(0x03));
|
|
|
|
|
__ vdup(Neon8, q0, r4);
|
|
|
|
|
__ mov(r4, Operand(0x02));
|
|
|
|
|
__ vdup(Neon16, q1, r4);
|
|
|
|
|
__ vtst(Neon8, q1, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vtst))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
|
|
|
|
// vbsl.
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(0x00FF));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vdup(Neon16, q0, r4);
|
|
|
|
|
__ mov(r4, Operand(0x01));
|
|
|
|
|
__ vdup(Neon8, q1, r4);
|
|
|
|
|
__ mov(r4, Operand(0x02));
|
|
|
|
|
__ vdup(Neon8, q2, r4);
|
|
|
|
|
__ vbsl(q0, q1, q2);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vbsl))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
// vext.
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vext(q2, q0, q1, 3);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vext))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q2), NeonMemOperand(r4));
|
|
|
|
|
|
2017-04-10 21:41:02 +00:00
|
|
|
// vzip (q-register).
|
2016-12-17 21:56:26 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vzip(Neon8, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vzip8a))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vzip8b))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vzip(Neon16, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vzip16a))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vzip16b))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vzip(Neon32, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vzip32a))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vzip32b))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2017-04-10 21:41:02 +00:00
|
|
|
// vzip (d-register).
|
|
|
|
|
__ vldr(d2, r0, offsetof(T, lane_test));
|
|
|
|
|
__ vmov(d0, d2);
|
|
|
|
|
__ vmov(d1, d2);
|
|
|
|
|
__ vzip(Neon8, d0, d1);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vzipd8a));
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vzipd8b));
|
|
|
|
|
__ vmov(d0, d2);
|
|
|
|
|
__ vmov(d1, d2);
|
|
|
|
|
__ vzip(Neon16, d0, d1);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vzipd16a));
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vzipd16b));
|
|
|
|
|
|
|
|
|
|
// vuzp (q-register).
|
2017-03-14 16:28:45 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vuzp(Neon8, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vuzp8a))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vuzp8b))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vuzp(Neon16, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vuzp16a))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vuzp16b))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vuzp(Neon32, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vuzp32a))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vuzp32b))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2017-04-10 21:41:02 +00:00
|
|
|
// vuzp (d-register).
|
|
|
|
|
__ vldr(d2, r0, offsetof(T, lane_test));
|
|
|
|
|
__ vmov(d0, d2);
|
|
|
|
|
__ vmov(d1, d2);
|
|
|
|
|
__ vuzp(Neon8, d0, d1);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vuzpd8a));
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vuzpd8b));
|
|
|
|
|
__ vmov(d0, d2);
|
|
|
|
|
__ vmov(d1, d2);
|
|
|
|
|
__ vuzp(Neon16, d0, d1);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vuzpd16a));
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vuzpd16b));
|
|
|
|
|
|
|
|
|
|
// vtrn (q-register).
|
2017-03-14 16:28:45 +00:00
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vtrn(Neon8, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vtrn8a))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vtrn8b))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vtrn(Neon16, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vtrn16a))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vtrn16b))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vmov(q1, q0);
|
|
|
|
|
__ vtrn(Neon32, q0, q1);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vtrn32a))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vtrn32b))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2017-04-10 21:41:02 +00:00
|
|
|
// vtrn (d-register).
|
|
|
|
|
__ vldr(d2, r0, offsetof(T, lane_test));
|
|
|
|
|
__ vmov(d0, d2);
|
|
|
|
|
__ vmov(d1, d2);
|
|
|
|
|
__ vtrn(Neon8, d0, d1);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vtrnd8a));
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vtrnd8b));
|
|
|
|
|
__ vmov(d0, d2);
|
|
|
|
|
__ vmov(d1, d2);
|
|
|
|
|
__ vtrn(Neon16, d0, d1);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vtrnd16a));
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vtrnd16b));
|
|
|
|
|
__ vmov(d0, d2);
|
|
|
|
|
__ vmov(d1, d2);
|
|
|
|
|
__ vtrn(Neon32, d0, d1);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vtrnd32a));
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vtrnd32b));
|
|
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
// vrev64/32/16
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, lane_test))));
|
|
|
|
|
__ vld1(Neon8, NeonListOperand(q0), NeonMemOperand(r4));
|
|
|
|
|
__ vrev64(Neon32, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrev64_32))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vrev64(Neon16, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrev64_16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vrev64(Neon8, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrev64_8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vrev32(Neon16, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrev32_16))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vrev32(Neon8, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrev32_8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
__ vrev16(Neon8, q1, q0);
|
|
|
|
|
__ add(r4, r0, Operand(static_cast<int32_t>(offsetof(T, vrev16_8))));
|
|
|
|
|
__ vst1(Neon8, NeonListOperand(q1), NeonMemOperand(r4));
|
|
|
|
|
|
2016-12-15 18:15:23 +00:00
|
|
|
// vtb[l/x].
|
|
|
|
|
__ mov(r4, Operand(0x06040200));
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r5, Operand(0xFF050301));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vmov(d2, r4, r5); // d2 = ff05030106040200
|
|
|
|
|
__ vtbl(d0, NeonListOperand(d2, 1), d2);
|
|
|
|
|
__ vstr(d0, r0, offsetof(T, vtbl));
|
|
|
|
|
__ vtbx(d2, NeonListOperand(d2, 1), d2);
|
|
|
|
|
__ vstr(d2, r0, offsetof(T, vtbx));
|
2016-11-25 19:36:45 +00:00
|
|
|
|
|
|
|
|
// Restore and return.
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, r5, pc});
|
2013-07-10 15:32:39 +00:00
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2013-07-10 15:32:39 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2013-07-10 15:32:39 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2013-07-10 15:32:39 +00:00
|
|
|
t.src0 = 0x01020304;
|
|
|
|
|
t.src1 = 0x11121314;
|
|
|
|
|
t.src2 = 0x21222324;
|
|
|
|
|
t.src3 = 0x31323334;
|
|
|
|
|
t.src4 = 0x41424344;
|
|
|
|
|
t.src5 = 0x51525354;
|
|
|
|
|
t.src6 = 0x61626364;
|
|
|
|
|
t.src7 = 0x71727374;
|
|
|
|
|
t.dst0 = 0;
|
|
|
|
|
t.dst1 = 0;
|
|
|
|
|
t.dst2 = 0;
|
|
|
|
|
t.dst3 = 0;
|
|
|
|
|
t.dst4 = 0;
|
|
|
|
|
t.dst5 = 0;
|
|
|
|
|
t.dst6 = 0;
|
|
|
|
|
t.dst7 = 0;
|
|
|
|
|
t.srcA0 = 0x41424344;
|
|
|
|
|
t.srcA1 = 0x81828384;
|
|
|
|
|
t.dstA0 = 0;
|
|
|
|
|
t.dstA1 = 0;
|
|
|
|
|
t.dstA2 = 0;
|
|
|
|
|
t.dstA3 = 0;
|
2016-12-17 21:56:26 +00:00
|
|
|
t.lane_test[0] = 0x03020100;
|
|
|
|
|
t.lane_test[1] = 0x07060504;
|
2017-12-02 00:30:37 +00:00
|
|
|
t.lane_test[2] = 0x0B0A0908;
|
|
|
|
|
t.lane_test[3] = 0x0F0E0D0C;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, 0, 0, 0, 0);
|
2016-12-15 18:15:23 +00:00
|
|
|
|
2015-01-30 09:29:25 +00:00
|
|
|
CHECK_EQ(0x01020304u, t.dst0);
|
|
|
|
|
CHECK_EQ(0x11121314u, t.dst1);
|
|
|
|
|
CHECK_EQ(0x21222324u, t.dst2);
|
|
|
|
|
CHECK_EQ(0x31323334u, t.dst3);
|
|
|
|
|
CHECK_EQ(0x41424344u, t.dst4);
|
|
|
|
|
CHECK_EQ(0x51525354u, t.dst5);
|
|
|
|
|
CHECK_EQ(0x61626364u, t.dst6);
|
|
|
|
|
CHECK_EQ(0x71727374u, t.dst7);
|
|
|
|
|
CHECK_EQ(0x00430044u, t.dstA0);
|
|
|
|
|
CHECK_EQ(0x00410042u, t.dstA1);
|
|
|
|
|
CHECK_EQ(0x00830084u, t.dstA2);
|
|
|
|
|
CHECK_EQ(0x00810082u, t.dstA3);
|
2017-03-27 18:23:42 +00:00
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vmovl_s8, 0x00430044u, 0x00410042u, 0xFF83FF84u, 0xFF81FF82u);
|
|
|
|
|
CHECK_EQ_32X4(vmovl_u16, 0xFF84u, 0xFF83u, 0xFF82u, 0xFF81u);
|
|
|
|
|
CHECK_EQ_32X4(vmovl_s32, 0xFF84u, 0x0u, 0xFF83u, 0x0u);
|
|
|
|
|
CHECK_EQ_32X2(vqmovn_u16, 0xFF83FF84u, 0xFF81FF82u);
|
2017-03-27 18:23:42 +00:00
|
|
|
CHECK_EQ_32X2(vqmovn_s8, 0x81828384u, 0x81828384u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X2(vqmovn_s32, 0xFF84u, 0xFF83u);
|
2016-12-17 21:56:26 +00:00
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(0xFFFFFFF8FFF8F800u, t.vmov_to_scalar1);
|
|
|
|
|
CHECK_EQ(0xFFF80000F8000000u, t.vmov_to_scalar2);
|
2016-12-15 18:15:23 +00:00
|
|
|
CHECK_EQ(0xFFFFFFFFu, t.vmov_from_scalar_s8);
|
|
|
|
|
CHECK_EQ(0xFFu, t.vmov_from_scalar_u8);
|
|
|
|
|
CHECK_EQ(0xFFFFFFFFu, t.vmov_from_scalar_s16);
|
|
|
|
|
CHECK_EQ(0xFFFFu, t.vmov_from_scalar_u16);
|
|
|
|
|
CHECK_EQ(0xFFFFFFFFu, t.vmov_from_scalar_32);
|
|
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vmov, 0x03020100u, 0x07060504u, 0x0B0A0908u, 0x0F0E0D0Cu);
|
|
|
|
|
CHECK_EQ_32X4(vmvn, 0xFCFDFEFFu, 0xF8F9FAFBu, 0xF4F5F6F7u, 0xF0F1F2F3u);
|
2016-12-17 21:56:26 +00:00
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_SPLAT(vdup8, 0x0A0A0A0Au);
|
|
|
|
|
CHECK_EQ_SPLAT(vdup16, 0x000A000Au);
|
|
|
|
|
CHECK_EQ_SPLAT(vdup32, 0x0000000Au);
|
|
|
|
|
CHECK_EQ_SPLAT(vdupf, -1.0); // bit pattern is 0xBF800000.
|
|
|
|
|
CHECK_EQ_32X2(vdupf_16, 0xBF80BF80u, 0xBF80BF80u);
|
|
|
|
|
CHECK_EQ_SPLAT(vdupf_8, 0xBFBFBFBFu);
|
2016-12-17 21:56:26 +00:00
|
|
|
|
|
|
|
|
// src: [-1, -1, 1, 1]
|
|
|
|
|
CHECK_EQ_32X4(vcvt_s32_f32, -1, -1, 1, 1);
|
|
|
|
|
CHECK_EQ_32X4(vcvt_u32_f32, 0u, 0u, 1u, 1u);
|
2016-12-15 18:15:23 +00:00
|
|
|
// src: [kMinInt, kMaxInt, kMaxUInt32, kMinInt + 1]
|
2016-12-17 21:56:26 +00:00
|
|
|
CHECK_EQ_32X4(vcvt_f32_s32, INT32_TO_FLOAT(kMinInt),
|
|
|
|
|
INT32_TO_FLOAT(kMaxInt), INT32_TO_FLOAT(kMaxUInt32),
|
|
|
|
|
INT32_TO_FLOAT(kMinInt + 1));
|
|
|
|
|
CHECK_EQ_32X4(vcvt_f32_u32, UINT32_TO_FLOAT(kMinInt),
|
|
|
|
|
UINT32_TO_FLOAT(kMaxInt), UINT32_TO_FLOAT(kMaxUInt32),
|
|
|
|
|
UINT32_TO_FLOAT(kMinInt + 1));
|
|
|
|
|
|
2020-12-10 00:47:56 +00:00
|
|
|
CHECK_EQ_32X4(vclt0_s8, 0x00FFFF00u, 0xFF00FF00u, 0xFF0000FFu, 0x00FF00FFu);
|
|
|
|
|
CHECK_EQ_32X4(vclt0_s16, 0x0000FFFF, 0xFFFFFFFFu, 0xFFFF0000u, 0x00000000u);
|
|
|
|
|
CHECK_EQ_32X4(vclt0_s32, 0x00000000u, 0xFFFFFFFFu, 0xFFFFFFFFu,
|
|
|
|
|
0x00000000u);
|
|
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
CHECK_EQ_32X4(vabsf, 1.0, 0.0, 0.0, 1.0);
|
|
|
|
|
CHECK_EQ_32X4(vnegf, 1.0, 0.0, -0.0, -1.0);
|
2017-12-02 00:30:37 +00:00
|
|
|
// src: [0x7F7F7F7F, 0x01010101, 0xFFFFFFFF, 0x80808080]
|
|
|
|
|
CHECK_EQ_32X4(vabs_s8, 0x7F7F7F7Fu, 0x01010101u, 0x01010101u, 0x80808080u);
|
|
|
|
|
CHECK_EQ_32X4(vabs_s16, 0x7F7F7F7Fu, 0x01010101u, 0x00010001u, 0x7F807F80u);
|
|
|
|
|
CHECK_EQ_32X4(vabs_s32, 0x7F7F7F7Fu, 0x01010101u, 0x00000001u, 0x7F7F7F80u);
|
|
|
|
|
CHECK_EQ_32X4(vneg_s8, 0x81818181u, 0xFFFFFFFFu, 0x01010101u, 0x80808080u);
|
|
|
|
|
CHECK_EQ_32X4(vneg_s16, 0x80818081u, 0xFEFFFEFFu, 0x00010001u, 0x7F807F80u);
|
|
|
|
|
CHECK_EQ_32X4(vneg_s32, 0x80808081u, 0xFEFEFEFFu, 0x00000001u, 0x7F7F7F80u);
|
|
|
|
|
|
|
|
|
|
CHECK_EQ_SPLAT(veor, 0x00FF00FFu);
|
|
|
|
|
CHECK_EQ_SPLAT(vand, 0x00FE00FEu);
|
|
|
|
|
CHECK_EQ_SPLAT(vorr, 0x00FF00FFu);
|
2016-12-17 21:56:26 +00:00
|
|
|
CHECK_EQ_SPLAT(vaddf, 2.0);
|
2017-04-24 18:53:16 +00:00
|
|
|
CHECK_EQ_32X2(vpaddf, 3.0, 7.0);
|
2017-01-16 19:36:25 +00:00
|
|
|
CHECK_EQ_SPLAT(vminf, 1.0);
|
|
|
|
|
CHECK_EQ_SPLAT(vmaxf, 2.0);
|
2016-12-17 21:56:26 +00:00
|
|
|
CHECK_EQ_SPLAT(vsubf, -1.0);
|
|
|
|
|
CHECK_EQ_SPLAT(vmulf, 4.0);
|
2017-01-13 12:03:23 +00:00
|
|
|
CHECK_ESTIMATE_SPLAT(vrecpe, 0.5f, 0.1f); // 1 / 2
|
2017-01-10 12:36:59 +00:00
|
|
|
CHECK_EQ_SPLAT(vrecps, -1.0f); // 2 - (2 * 1.5)
|
2017-01-13 12:03:23 +00:00
|
|
|
CHECK_ESTIMATE_SPLAT(vrsqrte, 0.5f, 0.1f); // 1 / sqrt(4)
|
2017-01-10 12:36:59 +00:00
|
|
|
CHECK_EQ_SPLAT(vrsqrts, -1.0f); // (3 - (2 * 2.5)) / 2
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_SPLAT(vceqf, 0xFFFFFFFFu);
|
2017-01-12 11:20:08 +00:00
|
|
|
// [0] >= [-1, 1, -0, 0]
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vcgef, 0u, 0xFFFFFFFFu, 0xFFFFFFFFu, 0xFFFFFFFFu);
|
|
|
|
|
CHECK_EQ_32X4(vcgtf, 0u, 0xFFFFFFFFu, 0u, 0u);
|
2017-01-16 19:36:25 +00:00
|
|
|
// [0, 3, 0, 3, ...] and [3, 3, 3, 3, ...]
|
|
|
|
|
CHECK_EQ_SPLAT(vmin_s8, 0x00030003u);
|
|
|
|
|
CHECK_EQ_SPLAT(vmax_s8, 0x03030303u);
|
2017-12-02 00:30:37 +00:00
|
|
|
// [0x00FF, 0x00FF, ...] and [0xFFFF, 0xFFFF, ...]
|
|
|
|
|
CHECK_EQ_SPLAT(vmin_u16, 0x00FF00FFu);
|
|
|
|
|
CHECK_EQ_SPLAT(vmax_u16, 0xFFFFFFFFu);
|
|
|
|
|
// [0x000000FF, 0x000000FF, ...] and [0xFFFFFFFF, 0xFFFFFFFF, ...]
|
|
|
|
|
CHECK_EQ_SPLAT(vmin_s32, 0xFFFFFFFFu);
|
|
|
|
|
CHECK_EQ_SPLAT(vmax_s32, 0xFFu);
|
2017-03-02 19:50:33 +00:00
|
|
|
// [0, 3, 0, 3, ...] and [3, 3, 3, 3, ...]
|
2017-04-24 18:53:16 +00:00
|
|
|
CHECK_EQ_32X2(vpadd_i8, 0x03030303u, 0x06060606u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X2(vpadd_i16, 0x0C0C0606u, 0x06060606u);
|
|
|
|
|
CHECK_EQ_32X2(vpadd_i32, 0x12120C0Cu, 0x06060606u);
|
2020-12-08 06:18:13 +00:00
|
|
|
|
2021-06-28 13:34:37 +00:00
|
|
|
CHECK_EQ_32X4(vpadal_s8, 0x30003, 0x30003, 0x30003, 0x30003);
|
|
|
|
|
CHECK_EQ_32X4(vpadal_s16, 0x1000403, 0x1000403, 0x1000403, 0x1000403);
|
|
|
|
|
CHECK_EQ_32X4(vpadal_s32, 0x4040403, 0x1010100, 0x4040403, 0x1010100);
|
|
|
|
|
|
|
|
|
|
CHECK_EQ_32X4(vpadal_u8, 0x2030203, 0x2030203, 0x2030203, 0x2030203);
|
|
|
|
|
CHECK_EQ_32X4(vpadal_u16, 0x1020403, 0x1020403, 0x1020403, 0x1020403);
|
|
|
|
|
CHECK_EQ_32X4(vpadal_u32, 0x4040403, 0x1010102, 0x4040403, 0x1010102);
|
|
|
|
|
|
2020-12-08 06:18:13 +00:00
|
|
|
CHECK_EQ_32X4(vpaddl_s8, 0xFF02FF02, 0xFF02FF02, 0xFF02FF02, 0xFF02FF02);
|
|
|
|
|
CHECK_EQ_32X4(vpaddl_s16, 0xFFFF0302, 0xFFFF0302, 0xFFFF0302, 0xFFFF0302);
|
|
|
|
|
CHECK_EQ_32X4(vpaddl_s32, 0x03030302, 0xFFFFFFFF, 0x03030302, 0xFFFFFFFF);
|
|
|
|
|
|
|
|
|
|
CHECK_EQ_32X4(vpaddl_u8, 0x01020102, 0x01020102, 0x01020102, 0x01020102);
|
|
|
|
|
CHECK_EQ_32X4(vpaddl_u16, 0x00010302, 0x00010302, 0x00010302, 0x00010302);
|
|
|
|
|
CHECK_EQ_32X4(vpaddl_u32, 0x03030302, 0x00000001, 0x03030302, 0x00000001);
|
|
|
|
|
|
2017-03-02 19:50:33 +00:00
|
|
|
CHECK_EQ_32X2(vpmin_s8, 0x00000000u, 0x03030303u);
|
|
|
|
|
CHECK_EQ_32X2(vpmax_s8, 0x03030303u, 0x03030303u);
|
|
|
|
|
// [0, ffff, 0, ffff] and [ffff, ffff]
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X2(vpmin_u16, 0x00000000u, 0xFFFFFFFFu);
|
|
|
|
|
CHECK_EQ_32X2(vpmax_u16, 0xFFFFFFFFu, 0xFFFFFFFFu);
|
|
|
|
|
// [0x000000FF, 0x00000000u] and [0xFFFFFFFF, 0xFFFFFFFF, ...]
|
|
|
|
|
CHECK_EQ_32X2(vpmin_s32, 0x00u, 0xFFFFFFFFu);
|
|
|
|
|
CHECK_EQ_32X2(vpmax_s32, 0xFFu, 0xFFFFFFFFu);
|
2016-12-17 21:56:26 +00:00
|
|
|
CHECK_EQ_SPLAT(vadd8, 0x03030303u);
|
|
|
|
|
CHECK_EQ_SPLAT(vadd16, 0x00030003u);
|
|
|
|
|
CHECK_EQ_SPLAT(vadd32, 0x00000003u);
|
2017-02-01 18:56:28 +00:00
|
|
|
CHECK_EQ_SPLAT(vqadd_s8, 0x80808080u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_SPLAT(vqadd_u16, 0xFFFFFFFFu);
|
2017-02-01 18:56:28 +00:00
|
|
|
CHECK_EQ_SPLAT(vqadd_s32, 0x80000000u);
|
|
|
|
|
CHECK_EQ_SPLAT(vqsub_u8, 0x00000000u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_SPLAT(vqsub_s16, 0x7FFF7FFFu);
|
2017-02-01 18:56:28 +00:00
|
|
|
CHECK_EQ_SPLAT(vqsub_u32, 0x00000000u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_SPLAT(vsub8, 0xFEFEFEFEu);
|
|
|
|
|
CHECK_EQ_SPLAT(vsub16, 0xFFFEFFFEu);
|
|
|
|
|
CHECK_EQ_SPLAT(vsub32, 0xFFFFFFFEu);
|
2016-12-17 21:56:26 +00:00
|
|
|
CHECK_EQ_SPLAT(vmul8, 0x04040404u);
|
|
|
|
|
CHECK_EQ_SPLAT(vmul16, 0x00040004u);
|
|
|
|
|
CHECK_EQ_SPLAT(vmul32, 0x00000004u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_SPLAT(vshl8, 0xAAAAAAAAu);
|
|
|
|
|
CHECK_EQ_SPLAT(vshl16, 0xAA00AA00u);
|
|
|
|
|
CHECK_EQ_SPLAT(vshl32, 0xAAAA0000u);
|
|
|
|
|
CHECK_EQ_SPLAT(vshr_s8, 0xC0C0C0C0u);
|
2017-01-23 18:24:27 +00:00
|
|
|
CHECK_EQ_SPLAT(vshr_u16, 0x00400040u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_SPLAT(vshr_s32, 0xFFFFC040u);
|
2020-11-30 01:05:20 +00:00
|
|
|
CHECK_EQ_32X2(vshr_s8_d, 0xC0C0C0C0u, 0xC0C0C0C0u);
|
|
|
|
|
CHECK_EQ_32X2(vshr_u16_d, 0x00400040u, 0x00400040u);
|
|
|
|
|
CHECK_EQ_32X2(vshr_s32_d, 0xFFFFC040u, 0xFFFFC040u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X2(vsli_64, 0x01u, 0xFFFFFFFFu);
|
|
|
|
|
CHECK_EQ_32X2(vsri_64, 0xFFFFFFFFu, 0x01u);
|
|
|
|
|
CHECK_EQ_32X2(vsli_32, 0xFFFF0001u, 0x00010001u);
|
|
|
|
|
CHECK_EQ_32X2(vsri_32, 0x00000000u, 0x0000FFFFu);
|
2020-11-30 01:05:20 +00:00
|
|
|
CHECK_EQ_32X2(vsra_64, 0xFFFFFFFEu, 0x2);
|
|
|
|
|
CHECK_EQ_32X2(vsra_32, 0x0, 0xFFFFFFFFu);
|
|
|
|
|
CHECK_EQ_32X2(vsra_16, 0x3FFF4000, 0xFFFFFFFFu);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_SPLAT(vceq, 0x00FF00FFu);
|
2017-01-12 11:20:08 +00:00
|
|
|
// [0, 3, 0, 3, ...] >= [3, 3, 3, 3, ...]
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_SPLAT(vcge_s8, 0x00FF00FFu);
|
2017-01-12 11:20:08 +00:00
|
|
|
CHECK_EQ_SPLAT(vcgt_s8, 0u);
|
2017-12-02 00:30:37 +00:00
|
|
|
// [0x00FF, 0x00FF, ...] >= [0xFFFF, 0xFFFF, ...]
|
2017-01-12 11:20:08 +00:00
|
|
|
CHECK_EQ_SPLAT(vcge_u16, 0u);
|
|
|
|
|
CHECK_EQ_SPLAT(vcgt_u16, 0u);
|
2017-12-02 00:30:37 +00:00
|
|
|
// [0x000000FF, 0x000000FF, ...] >= [0xFFFFFFFF, 0xFFFFFFFF, ...]
|
|
|
|
|
CHECK_EQ_SPLAT(vcge_s32, 0xFFFFFFFFu);
|
|
|
|
|
CHECK_EQ_SPLAT(vcgt_s32, 0xFFFFFFFFu);
|
|
|
|
|
CHECK_EQ_SPLAT(vtst, 0x00FF00FFu);
|
2016-12-17 21:56:26 +00:00
|
|
|
CHECK_EQ_SPLAT(vbsl, 0x02010201u);
|
|
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vext, 0x06050403u, 0x0A090807u, 0x0E0D0C0Bu, 0x0201000Fu);
|
2016-12-17 21:56:26 +00:00
|
|
|
|
|
|
|
|
CHECK_EQ_32X4(vzip8a, 0x01010000u, 0x03030202u, 0x05050404u, 0x07070606u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vzip8b, 0x09090808u, 0x0B0B0A0Au, 0x0D0D0C0Cu, 0x0F0F0E0Eu);
|
2016-12-17 21:56:26 +00:00
|
|
|
CHECK_EQ_32X4(vzip16a, 0x01000100u, 0x03020302u, 0x05040504u, 0x07060706u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vzip16b, 0x09080908u, 0x0B0A0B0Au, 0x0D0C0D0Cu, 0x0F0E0F0Eu);
|
2016-12-17 21:56:26 +00:00
|
|
|
CHECK_EQ_32X4(vzip32a, 0x03020100u, 0x03020100u, 0x07060504u, 0x07060504u);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vzip32b, 0x0B0A0908u, 0x0B0A0908u, 0x0F0E0D0Cu, 0x0F0E0D0Cu);
|
2016-12-17 21:56:26 +00:00
|
|
|
|
2017-04-10 21:41:02 +00:00
|
|
|
CHECK_EQ_32X2(vzipd8a, 0x01010000u, 0x03030202u);
|
|
|
|
|
CHECK_EQ_32X2(vzipd8b, 0x05050404u, 0x07070606u);
|
|
|
|
|
CHECK_EQ_32X2(vzipd16a, 0x01000100u, 0x03020302u);
|
|
|
|
|
CHECK_EQ_32X2(vzipd16b, 0x05040504u, 0x07060706u);
|
|
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vuzp8a, 0x06040200u, 0x0E0C0A08u, 0x06040200u, 0x0E0C0A08u);
|
|
|
|
|
CHECK_EQ_32X4(vuzp8b, 0x07050301u, 0x0F0D0B09u, 0x07050301u, 0x0F0D0B09u);
|
|
|
|
|
CHECK_EQ_32X4(vuzp16a, 0x05040100u, 0x0D0C0908u, 0x05040100u, 0x0D0C0908u);
|
|
|
|
|
CHECK_EQ_32X4(vuzp16b, 0x07060302u, 0x0F0E0B0Au, 0x07060302u, 0x0F0E0B0Au);
|
|
|
|
|
CHECK_EQ_32X4(vuzp32a, 0x03020100u, 0x0B0A0908u, 0x03020100u, 0x0B0A0908u);
|
|
|
|
|
CHECK_EQ_32X4(vuzp32b, 0x07060504u, 0x0F0E0D0Cu, 0x07060504u, 0x0F0E0D0Cu);
|
2017-03-14 16:28:45 +00:00
|
|
|
|
2017-04-10 21:41:02 +00:00
|
|
|
CHECK_EQ_32X2(vuzpd8a, 0x06040200u, 0x06040200u);
|
|
|
|
|
CHECK_EQ_32X2(vuzpd8b, 0x07050301u, 0x07050301u);
|
|
|
|
|
CHECK_EQ_32X2(vuzpd16a, 0x05040100u, 0x05040100u);
|
|
|
|
|
CHECK_EQ_32X2(vuzpd16b, 0x07060302u, 0x07060302u);
|
|
|
|
|
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vtrn8a, 0x02020000u, 0x06060404u, 0x0A0A0808u, 0x0E0E0C0Cu);
|
|
|
|
|
CHECK_EQ_32X4(vtrn8b, 0x03030101u, 0x07070505u, 0x0B0B0909u, 0x0F0F0D0Du);
|
|
|
|
|
CHECK_EQ_32X4(vtrn16a, 0x01000100u, 0x05040504u, 0x09080908u, 0x0D0C0D0Cu);
|
|
|
|
|
CHECK_EQ_32X4(vtrn16b, 0x03020302u, 0x07060706u, 0x0B0A0B0Au, 0x0F0E0F0Eu);
|
|
|
|
|
CHECK_EQ_32X4(vtrn32a, 0x03020100u, 0x03020100u, 0x0B0A0908u, 0x0B0A0908u);
|
|
|
|
|
CHECK_EQ_32X4(vtrn32b, 0x07060504u, 0x07060504u, 0x0F0E0D0Cu, 0x0F0E0D0Cu);
|
2017-03-14 16:28:45 +00:00
|
|
|
|
2017-04-10 21:41:02 +00:00
|
|
|
CHECK_EQ_32X2(vtrnd8a, 0x02020000u, 0x06060404u);
|
|
|
|
|
CHECK_EQ_32X2(vtrnd8b, 0x03030101u, 0x07070505u);
|
|
|
|
|
CHECK_EQ_32X2(vtrnd16a, 0x01000100u, 0x05040504u);
|
|
|
|
|
CHECK_EQ_32X2(vtrnd16b, 0x03020302u, 0x07060706u);
|
|
|
|
|
CHECK_EQ_32X2(vtrnd32a, 0x03020100u, 0x03020100u);
|
|
|
|
|
CHECK_EQ_32X2(vtrnd32b, 0x07060504u, 0x07060504u);
|
|
|
|
|
|
2016-12-17 21:56:26 +00:00
|
|
|
// src: 0 1 2 3 4 5 6 7 8 9 a b c d e f (little endian)
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ_32X4(vrev64_32, 0x07060504u, 0x03020100u, 0x0F0E0D0Cu,
|
|
|
|
|
0x0B0A0908u);
|
|
|
|
|
CHECK_EQ_32X4(vrev64_16, 0x05040706u, 0x01000302u, 0x0D0C0F0Eu,
|
|
|
|
|
0x09080B0Au);
|
|
|
|
|
CHECK_EQ_32X4(vrev64_8, 0x04050607u, 0x00010203u, 0x0C0D0E0Fu, 0x08090A0Bu);
|
|
|
|
|
CHECK_EQ_32X4(vrev32_16, 0x01000302u, 0x05040706u, 0x09080B0Au,
|
|
|
|
|
0x0D0C0F0Eu);
|
|
|
|
|
CHECK_EQ_32X4(vrev32_8, 0x00010203u, 0x04050607u, 0x08090A0Bu, 0x0C0D0E0Fu);
|
|
|
|
|
CHECK_EQ_32X4(vrev16_8, 0x02030001u, 0x06070405u, 0x0A0B0809u, 0x0E0F0C0Du);
|
2016-12-17 21:56:26 +00:00
|
|
|
|
2016-12-15 18:15:23 +00:00
|
|
|
CHECK_EQ(0x05010400u, t.vtbl[0]);
|
|
|
|
|
CHECK_EQ(0x00030602u, t.vtbl[1]);
|
|
|
|
|
CHECK_EQ(0x05010400u, t.vtbx[0]);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(0xFF030602u, t.vtbx[1]);
|
2013-07-10 15:32:39 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(16) {
|
|
|
|
|
// Test the pkh, uxtb, uxtab and uxtb16 instructions.
|
|
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-07-10 15:32:39 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2013-07-10 15:32:39 +00:00
|
|
|
uint32_t src0;
|
|
|
|
|
uint32_t src1;
|
|
|
|
|
uint32_t src2;
|
|
|
|
|
uint32_t dst0;
|
|
|
|
|
uint32_t dst1;
|
|
|
|
|
uint32_t dst2;
|
|
|
|
|
uint32_t dst3;
|
|
|
|
|
uint32_t dst4;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2013-07-10 15:32:39 +00:00
|
|
|
T t;
|
|
|
|
|
|
|
|
|
|
// Create a function that accepts &t, and loads, manipulates, and stores
|
|
|
|
|
// the doubles and floats.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2013-07-10 15:32:39 +00:00
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, lr});
|
2013-07-10 15:32:39 +00:00
|
|
|
|
|
|
|
|
__ mov(r4, Operand(r0));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ ldr(r0, MemOperand(r4, offsetof(T, src0)));
|
|
|
|
|
__ ldr(r1, MemOperand(r4, offsetof(T, src1)));
|
2013-07-10 15:32:39 +00:00
|
|
|
|
|
|
|
|
__ pkhbt(r2, r0, Operand(r1, LSL, 8));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r2, MemOperand(r4, offsetof(T, dst0)));
|
2013-07-10 15:32:39 +00:00
|
|
|
|
|
|
|
|
__ pkhtb(r2, r0, Operand(r1, ASR, 8));
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r2, MemOperand(r4, offsetof(T, dst1)));
|
2013-07-10 15:32:39 +00:00
|
|
|
|
2014-11-10 05:48:38 +00:00
|
|
|
__ uxtb16(r2, r0, 8);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r2, MemOperand(r4, offsetof(T, dst2)));
|
2013-07-10 15:32:39 +00:00
|
|
|
|
2014-11-10 05:48:38 +00:00
|
|
|
__ uxtb(r2, r0, 8);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r2, MemOperand(r4, offsetof(T, dst3)));
|
2013-07-10 15:32:39 +00:00
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ ldr(r0, MemOperand(r4, offsetof(T, src2)));
|
2014-11-10 05:48:38 +00:00
|
|
|
__ uxtab(r2, r0, r1, 8);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r2, MemOperand(r4, offsetof(T, dst4)));
|
2013-07-10 15:32:39 +00:00
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, pc});
|
2013-07-10 15:32:39 +00:00
|
|
|
|
|
|
|
|
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();
|
2013-07-10 15:32:39 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2013-07-10 15:32:39 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2013-07-10 15:32:39 +00:00
|
|
|
t.src0 = 0x01020304;
|
|
|
|
|
t.src1 = 0x11121314;
|
|
|
|
|
t.src2 = 0x11121300;
|
|
|
|
|
t.dst0 = 0;
|
|
|
|
|
t.dst1 = 0;
|
|
|
|
|
t.dst2 = 0;
|
|
|
|
|
t.dst3 = 0;
|
|
|
|
|
t.dst4 = 0;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, 0, 0, 0, 0);
|
2015-01-30 09:29:25 +00:00
|
|
|
CHECK_EQ(0x12130304u, t.dst0);
|
|
|
|
|
CHECK_EQ(0x01021213u, t.dst1);
|
|
|
|
|
CHECK_EQ(0x00010003u, t.dst2);
|
|
|
|
|
CHECK_EQ(0x00000003u, t.dst3);
|
|
|
|
|
CHECK_EQ(0x11121313u, t.dst4);
|
2013-07-10 15:32:39 +00:00
|
|
|
}
|
|
|
|
|
|
2013-08-01 08:13:08 +00:00
|
|
|
|
|
|
|
|
TEST(17) {
|
|
|
|
|
// Test generating labels at high addresses.
|
|
|
|
|
// Should not assert.
|
|
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-08-01 08:13:08 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
|
|
|
|
// Generate a code segment that will be longer than 2^24 bytes.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2013-08-01 08:13:08 +00:00
|
|
|
for (size_t i = 0; i < 1 << 23 ; ++i) { // 2^23
|
|
|
|
|
__ nop();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
Label target;
|
|
|
|
|
__ b(eq, &target);
|
|
|
|
|
__ bind(&target);
|
|
|
|
|
__ nop();
|
|
|
|
|
}
|
|
|
|
|
|
2018-01-09 09:50:34 +00:00
|
|
|
#define TEST_SDIV(expected_, dividend_, divisor_) \
|
|
|
|
|
t.dividend = dividend_; \
|
|
|
|
|
t.divisor = divisor_; \
|
|
|
|
|
t.result = 0; \
|
|
|
|
|
f.Call(&t, 0, 0, 0, 0); \
|
2015-11-23 08:09:34 +00:00
|
|
|
CHECK_EQ(expected_, t.result);
|
2014-01-02 16:36:21 +00:00
|
|
|
|
2014-10-26 12:49:56 +00:00
|
|
|
TEST(sdiv) {
|
2014-01-02 16:36:21 +00:00
|
|
|
// Test the sdiv.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-01-02 16:36:21 +00:00
|
|
|
|
2014-10-26 12:49:56 +00:00
|
|
|
struct T {
|
|
|
|
|
int32_t dividend;
|
|
|
|
|
int32_t divisor;
|
|
|
|
|
int32_t result;
|
|
|
|
|
} t;
|
|
|
|
|
|
2014-01-02 16:36:21 +00:00
|
|
|
if (CpuFeatures::IsSupported(SUDIV)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, SUDIV);
|
|
|
|
|
|
|
|
|
|
__ mov(r3, Operand(r0));
|
|
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ ldr(r0, MemOperand(r3, offsetof(T, dividend)));
|
|
|
|
|
__ ldr(r1, MemOperand(r3, offsetof(T, divisor)));
|
2014-01-02 16:36:21 +00:00
|
|
|
|
|
|
|
|
__ sdiv(r2, r0, r1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r2, MemOperand(r3, offsetof(T, result)));
|
2014-01-02 16:36:21 +00:00
|
|
|
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2014-01-02 16:36:21 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2014-01-02 16:36:21 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-10-26 12:49:56 +00:00
|
|
|
TEST_SDIV(0, kMinInt, 0);
|
|
|
|
|
TEST_SDIV(0, 1024, 0);
|
2014-01-02 16:36:21 +00:00
|
|
|
TEST_SDIV(1073741824, kMinInt, -2);
|
|
|
|
|
TEST_SDIV(kMinInt, kMinInt, -1);
|
|
|
|
|
TEST_SDIV(5, 10, 2);
|
|
|
|
|
TEST_SDIV(3, 10, 3);
|
|
|
|
|
TEST_SDIV(-5, 10, -2);
|
|
|
|
|
TEST_SDIV(-3, 10, -3);
|
|
|
|
|
TEST_SDIV(-5, -10, 2);
|
|
|
|
|
TEST_SDIV(-3, -10, 3);
|
|
|
|
|
TEST_SDIV(5, -10, -2);
|
|
|
|
|
TEST_SDIV(3, -10, -3);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
#undef TEST_SDIV
|
|
|
|
|
|
2018-01-09 09:50:34 +00:00
|
|
|
#define TEST_UDIV(expected_, dividend_, divisor_) \
|
|
|
|
|
t.dividend = dividend_; \
|
|
|
|
|
t.divisor = divisor_; \
|
|
|
|
|
t.result = 0; \
|
|
|
|
|
f.Call(&t, 0, 0, 0, 0); \
|
2014-10-26 12:49:56 +00:00
|
|
|
CHECK_EQ(expected_, t.result);
|
|
|
|
|
|
|
|
|
|
TEST(udiv) {
|
|
|
|
|
// Test the udiv.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-10-26 12:49:56 +00:00
|
|
|
|
|
|
|
|
struct T {
|
|
|
|
|
uint32_t dividend;
|
|
|
|
|
uint32_t divisor;
|
|
|
|
|
uint32_t result;
|
|
|
|
|
} t;
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(SUDIV)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, SUDIV);
|
|
|
|
|
|
|
|
|
|
__ mov(r3, Operand(r0));
|
|
|
|
|
|
2015-06-17 09:06:44 +00:00
|
|
|
__ ldr(r0, MemOperand(r3, offsetof(T, dividend)));
|
|
|
|
|
__ ldr(r1, MemOperand(r3, offsetof(T, divisor)));
|
2014-10-26 12:49:56 +00:00
|
|
|
|
|
|
|
|
__ sdiv(r2, r0, r1);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ str(r2, MemOperand(r3, offsetof(T, result)));
|
2014-10-26 12:49:56 +00:00
|
|
|
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2014-10-26 12:49:56 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2014-10-26 12:49:56 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2015-01-30 09:29:25 +00:00
|
|
|
TEST_UDIV(0u, 0, 0);
|
|
|
|
|
TEST_UDIV(0u, 1024, 0);
|
|
|
|
|
TEST_UDIV(5u, 10, 2);
|
|
|
|
|
TEST_UDIV(3u, 10, 3);
|
2014-10-26 12:49:56 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
#undef TEST_UDIV
|
|
|
|
|
|
|
|
|
|
|
2014-10-14 05:03:04 +00:00
|
|
|
TEST(smmla) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-10-14 05:03:04 +00:00
|
|
|
__ smmla(r1, r1, r2, r3);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-10-14 05:03:04 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-10-14 05:03:04 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-10-14 05:03:04 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt(), y = rng->NextInt(), z = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, y, z, 0);
|
2017-08-31 12:34:55 +00:00
|
|
|
CHECK_EQ(base::bits::SignedMulHighAndAdd32(x, y, z), r);
|
2014-10-14 05:03:04 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(smmul) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-10-14 05:03:04 +00:00
|
|
|
__ smmul(r1, r1, r2);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-10-14 05:03:04 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-10-14 05:03:04 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-10-14 05:03:04 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt(), y = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, y, 0, 0);
|
2017-08-31 12:34:55 +00:00
|
|
|
CHECK_EQ(base::bits::SignedMulHigh32(x, y), r);
|
2014-10-14 05:03:04 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2014-11-10 05:48:38 +00:00
|
|
|
TEST(sxtb) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-11-10 05:48:38 +00:00
|
|
|
__ sxtb(r1, r1);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-11-10 05:48:38 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-11-10 05:48:38 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-11-10 05:48:38 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, 0, 0, 0);
|
2014-11-10 05:48:38 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(static_cast<int8_t>(x)), r);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(sxtab) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-11-10 05:48:38 +00:00
|
|
|
__ sxtab(r1, r2, r1);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-11-10 05:48:38 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-11-10 05:48:38 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-11-10 05:48:38 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt(), y = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, y, 0, 0);
|
2014-11-10 05:48:38 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(static_cast<int8_t>(x)) + y, r);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(sxth) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-11-10 05:48:38 +00:00
|
|
|
__ sxth(r1, r1);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-11-10 05:48:38 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-11-10 05:48:38 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-11-10 05:48:38 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, 0, 0, 0);
|
2014-11-10 05:48:38 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(static_cast<int16_t>(x)), r);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(sxtah) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-11-10 05:48:38 +00:00
|
|
|
__ sxtah(r1, r2, r1);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-11-10 05:48:38 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-11-10 05:48:38 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-11-10 05:48:38 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt(), y = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, y, 0, 0);
|
2014-11-10 05:48:38 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(static_cast<int16_t>(x)) + y, r);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(uxtb) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-11-10 05:48:38 +00:00
|
|
|
__ uxtb(r1, r1);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-11-10 05:48:38 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-11-10 05:48:38 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-11-10 05:48:38 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, 0, 0, 0);
|
2014-11-10 05:48:38 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(static_cast<uint8_t>(x)), r);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(uxtab) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-11-10 05:48:38 +00:00
|
|
|
__ uxtab(r1, r2, r1);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-11-10 05:48:38 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-11-10 05:48:38 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-11-10 05:48:38 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt(), y = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, y, 0, 0);
|
2014-11-10 05:48:38 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(static_cast<uint8_t>(x)) + y, r);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(uxth) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-11-10 05:48:38 +00:00
|
|
|
__ uxth(r1, r1);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-11-10 05:48:38 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-11-10 05:48:38 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-11-10 05:48:38 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, 0, 0, 0);
|
2014-11-10 05:48:38 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(static_cast<uint16_t>(x)), r);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(uxtah) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
RandomNumberGenerator* const rng = isolate->random_number_generator();
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-11-10 05:48:38 +00:00
|
|
|
__ uxtah(r1, r2, r1);
|
|
|
|
|
__ str(r1, MemOperand(r0));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
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();
|
2014-11-10 05:48:38 +00:00
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2014-11-10 05:48:38 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-11-10 05:48:38 +00:00
|
|
|
for (size_t i = 0; i < 128; ++i) {
|
|
|
|
|
int32_t r, x = rng->NextInt(), y = rng->NextInt();
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&r, x, y, 0, 0);
|
2014-11-10 05:48:38 +00:00
|
|
|
CHECK_EQ(static_cast<int32_t>(static_cast<uint16_t>(x)) + y, r);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2018-01-09 09:50:34 +00:00
|
|
|
#define TEST_RBIT(expected_, input_) \
|
|
|
|
|
t.input = input_; \
|
|
|
|
|
t.result = 0; \
|
|
|
|
|
f.Call(&t, 0, 0, 0, 0); \
|
2016-11-11 12:00:34 +00:00
|
|
|
CHECK_EQ(static_cast<uint32_t>(expected_), t.result);
|
2016-02-16 11:14:01 +00:00
|
|
|
|
|
|
|
|
TEST(rbit) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* const isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2016-02-16 11:14:01 +00:00
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(ARMv7)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, ARMv7);
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2016-02-16 11:14:01 +00:00
|
|
|
uint32_t input;
|
|
|
|
|
uint32_t result;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2016-02-16 11:14:01 +00:00
|
|
|
T t;
|
|
|
|
|
|
|
|
|
|
__ ldr(r1, MemOperand(r0, offsetof(T, input)));
|
|
|
|
|
__ rbit(r1, r1);
|
|
|
|
|
__ str(r1, MemOperand(r0, offsetof(T, result)));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2016-02-16 11:14:01 +00:00
|
|
|
|
|
|
|
|
#ifdef OBJECT_PRINT
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(std::cout);
|
2016-02-16 11:14:01 +00:00
|
|
|
#endif
|
|
|
|
|
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2017-12-02 00:30:37 +00:00
|
|
|
TEST_RBIT(0xFFFFFFFF, 0xFFFFFFFF);
|
2016-02-16 11:14:01 +00:00
|
|
|
TEST_RBIT(0x00000000, 0x00000000);
|
2017-12-02 00:30:37 +00:00
|
|
|
TEST_RBIT(0xFFFF0000, 0x0000FFFF);
|
|
|
|
|
TEST_RBIT(0xFF00FF00, 0x00FF00FF);
|
|
|
|
|
TEST_RBIT(0xF0F0F0F0, 0x0F0F0F0F);
|
|
|
|
|
TEST_RBIT(0x1E6A2C48, 0x12345678);
|
2016-02-16 11:14:01 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2013-09-12 10:37:42 +00:00
|
|
|
TEST(code_relative_offset) {
|
|
|
|
|
// Test extracting the offset of a label from the beginning of the code
|
|
|
|
|
// in a register.
|
|
|
|
|
CcTest::InitializeVM();
|
2013-09-19 09:17:13 +00:00
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
2013-09-12 10:37:42 +00:00
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
// Initialize a code object that will contain the code.
|
2018-07-04 09:10:05 +00:00
|
|
|
Handle<HeapObject> code_object(ReadOnlyRoots(isolate).self_reference_marker(),
|
2018-05-02 11:54:25 +00:00
|
|
|
isolate);
|
2013-09-12 10:37:42 +00:00
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2013-09-12 10:37:42 +00:00
|
|
|
|
|
|
|
|
Label start, target_away, target_faraway;
|
|
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, r5, lr});
|
2013-09-12 10:37:42 +00:00
|
|
|
|
|
|
|
|
// r3 is used as the address zero, the test will crash when we load it.
|
|
|
|
|
__ mov(r3, Operand::Zero());
|
|
|
|
|
|
|
|
|
|
// r5 will be a pointer to the start of the code.
|
|
|
|
|
__ mov(r5, Operand(code_object));
|
|
|
|
|
__ mov_label_offset(r4, &start);
|
|
|
|
|
|
|
|
|
|
__ mov_label_offset(r1, &target_faraway);
|
|
|
|
|
__ str(r1, MemOperand(sp, kPointerSize, NegPreIndex));
|
|
|
|
|
|
|
|
|
|
__ mov_label_offset(r1, &target_away);
|
|
|
|
|
|
|
|
|
|
// Jump straight to 'target_away' the first time and use the relative
|
|
|
|
|
// position the second time. This covers the case when extracting the
|
|
|
|
|
// position of a label which is linked.
|
|
|
|
|
__ mov(r2, Operand::Zero());
|
|
|
|
|
__ bind(&start);
|
|
|
|
|
__ cmp(r2, Operand::Zero());
|
|
|
|
|
__ b(eq, &target_away);
|
|
|
|
|
__ add(pc, r5, r1);
|
|
|
|
|
// Emit invalid instructions to push the label between 2^8 and 2^16
|
|
|
|
|
// instructions away. The test will crash if they are reached.
|
|
|
|
|
for (int i = 0; i < (1 << 10); i++) {
|
|
|
|
|
__ ldr(r3, MemOperand(r3));
|
|
|
|
|
}
|
|
|
|
|
__ bind(&target_away);
|
|
|
|
|
// This will be hit twice: r0 = r0 + 5 + 5.
|
|
|
|
|
__ add(r0, r0, Operand(5));
|
|
|
|
|
|
|
|
|
|
__ ldr(r1, MemOperand(sp, kPointerSize, PostIndex), ne);
|
|
|
|
|
__ add(pc, r5, r4, LeaveCC, ne);
|
|
|
|
|
|
|
|
|
|
__ mov(r2, Operand(1));
|
|
|
|
|
__ b(&start);
|
|
|
|
|
// Emit invalid instructions to push the label between 2^16 and 2^24
|
|
|
|
|
// instructions away. The test will crash if they are reached.
|
|
|
|
|
for (int i = 0; i < (1 << 21); i++) {
|
|
|
|
|
__ ldr(r3, MemOperand(r3));
|
|
|
|
|
}
|
|
|
|
|
__ bind(&target_faraway);
|
|
|
|
|
// r0 = r0 + 5 + 5 + 11
|
|
|
|
|
__ add(r0, r0, Operand(11));
|
|
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, r5, pc});
|
2013-09-12 10:37:42 +00:00
|
|
|
|
|
|
|
|
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)
|
2019-04-29 16:15:47 +00:00
|
|
|
.set_self_reference(code_object)
|
|
|
|
|
.Build();
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
int res = reinterpret_cast<int>(f.Call(21, 0, 0, 0, 0));
|
2013-09-12 10:37:42 +00:00
|
|
|
::printf("f() = %d\n", res);
|
|
|
|
|
CHECK_EQ(42, res);
|
|
|
|
|
}
|
|
|
|
|
|
2016-03-10 11:38:17 +00:00
|
|
|
TEST(msr_mrs) {
|
|
|
|
|
// Test msr and mrs.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2016-03-10 11:38:17 +00:00
|
|
|
|
|
|
|
|
// Create a helper function:
|
|
|
|
|
// void TestMsrMrs(uint32_t nzcv,
|
|
|
|
|
// uint32_t * result_conditionals,
|
|
|
|
|
// uint32_t * result_mrs);
|
|
|
|
|
__ msr(CPSR_f, Operand(r0));
|
|
|
|
|
|
|
|
|
|
// Test that the condition flags have taken effect.
|
|
|
|
|
__ mov(r3, Operand(0));
|
|
|
|
|
__ orr(r3, r3, Operand(1 << 31), LeaveCC, mi); // N
|
|
|
|
|
__ orr(r3, r3, Operand(1 << 30), LeaveCC, eq); // Z
|
|
|
|
|
__ orr(r3, r3, Operand(1 << 29), LeaveCC, cs); // C
|
|
|
|
|
__ orr(r3, r3, Operand(1 << 28), LeaveCC, vs); // V
|
|
|
|
|
__ str(r3, MemOperand(r1));
|
|
|
|
|
|
|
|
|
|
// Also check mrs, ignoring everything other than the flags.
|
|
|
|
|
__ mrs(r3, CPSR);
|
|
|
|
|
__ and_(r3, r3, Operand(kSpecialCondition));
|
|
|
|
|
__ str(r3, MemOperand(r2));
|
|
|
|
|
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
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();
|
2016-03-10 11:38:17 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2016-03-10 11:38:17 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_ippii>::FromCode(*code);
|
|
|
|
|
|
|
|
|
|
#define CHECK_MSR_MRS(n, z, c, v) \
|
|
|
|
|
do { \
|
|
|
|
|
uint32_t nzcv = (n << 31) | (z << 30) | (c << 29) | (v << 28); \
|
|
|
|
|
uint32_t result_conditionals = -1; \
|
|
|
|
|
uint32_t result_mrs = -1; \
|
|
|
|
|
f.Call(nzcv, &result_conditionals, &result_mrs, 0, 0); \
|
|
|
|
|
CHECK_EQ(nzcv, result_conditionals); \
|
|
|
|
|
CHECK_EQ(nzcv, result_mrs); \
|
2016-03-10 11:38:17 +00:00
|
|
|
} while (0);
|
|
|
|
|
|
|
|
|
|
// N Z C V
|
|
|
|
|
CHECK_MSR_MRS(0, 0, 0, 0);
|
|
|
|
|
CHECK_MSR_MRS(0, 0, 0, 1);
|
|
|
|
|
CHECK_MSR_MRS(0, 0, 1, 0);
|
|
|
|
|
CHECK_MSR_MRS(0, 0, 1, 1);
|
|
|
|
|
CHECK_MSR_MRS(0, 1, 0, 0);
|
|
|
|
|
CHECK_MSR_MRS(0, 1, 0, 1);
|
|
|
|
|
CHECK_MSR_MRS(0, 1, 1, 0);
|
|
|
|
|
CHECK_MSR_MRS(0, 1, 1, 1);
|
|
|
|
|
CHECK_MSR_MRS(1, 0, 0, 0);
|
|
|
|
|
CHECK_MSR_MRS(1, 0, 0, 1);
|
|
|
|
|
CHECK_MSR_MRS(1, 0, 1, 0);
|
|
|
|
|
CHECK_MSR_MRS(1, 0, 1, 1);
|
|
|
|
|
CHECK_MSR_MRS(1, 1, 0, 0);
|
|
|
|
|
CHECK_MSR_MRS(1, 1, 0, 1);
|
|
|
|
|
CHECK_MSR_MRS(1, 1, 1, 0);
|
|
|
|
|
CHECK_MSR_MRS(1, 1, 1, 1);
|
|
|
|
|
|
|
|
|
|
#undef CHECK_MSR_MRS
|
|
|
|
|
}
|
2014-10-30 11:00:47 +00:00
|
|
|
|
2015-11-25 19:28:42 +00:00
|
|
|
TEST(ARMv8_float32_vrintX) {
|
|
|
|
|
// Test the vrintX floating point instructions.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2015-11-25 19:28:42 +00:00
|
|
|
float input;
|
|
|
|
|
float ar;
|
|
|
|
|
float nr;
|
|
|
|
|
float mr;
|
|
|
|
|
float pr;
|
|
|
|
|
float zr;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2015-11-25 19:28:42 +00:00
|
|
|
T t;
|
|
|
|
|
|
|
|
|
|
// Create a function that accepts &t, and loads, manipulates, and stores
|
|
|
|
|
// the floats.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2015-11-25 19:28:42 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(ARMv8)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, ARMv8);
|
|
|
|
|
|
|
|
|
|
__ mov(ip, Operand(sp));
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, fp, lr});
|
2015-11-25 19:28:42 +00:00
|
|
|
|
|
|
|
|
__ mov(r4, Operand(r0));
|
|
|
|
|
|
|
|
|
|
// Test vrinta
|
|
|
|
|
__ vldr(s6, r4, offsetof(T, input));
|
|
|
|
|
__ vrinta(s5, s6);
|
|
|
|
|
__ vstr(s5, r4, offsetof(T, ar));
|
|
|
|
|
|
|
|
|
|
// Test vrintn
|
|
|
|
|
__ vldr(s6, r4, offsetof(T, input));
|
|
|
|
|
__ vrintn(s5, s6);
|
|
|
|
|
__ vstr(s5, r4, offsetof(T, nr));
|
|
|
|
|
|
|
|
|
|
// Test vrintp
|
|
|
|
|
__ vldr(s6, r4, offsetof(T, input));
|
|
|
|
|
__ vrintp(s5, s6);
|
|
|
|
|
__ vstr(s5, r4, offsetof(T, pr));
|
|
|
|
|
|
|
|
|
|
// Test vrintm
|
|
|
|
|
__ vldr(s6, r4, offsetof(T, input));
|
|
|
|
|
__ vrintm(s5, s6);
|
|
|
|
|
__ vstr(s5, r4, offsetof(T, mr));
|
|
|
|
|
|
|
|
|
|
// Test vrintz
|
|
|
|
|
__ vldr(s6, r4, offsetof(T, input));
|
|
|
|
|
__ vrintz(s5, s6);
|
|
|
|
|
__ vstr(s5, r4, offsetof(T, zr));
|
|
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, fp, pc});
|
2015-11-25 19:28:42 +00:00
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2015-11-25 19:28:42 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2015-11-25 19:28:42 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2015-11-25 19:28:42 +00:00
|
|
|
|
|
|
|
|
#define CHECK_VRINT(input_val, ares, nres, mres, pres, zres) \
|
|
|
|
|
t.input = input_val; \
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, 0, 0, 0, 0); \
|
2015-11-25 19:28:42 +00:00
|
|
|
CHECK_EQ(ares, t.ar); \
|
|
|
|
|
CHECK_EQ(nres, t.nr); \
|
|
|
|
|
CHECK_EQ(mres, t.mr); \
|
|
|
|
|
CHECK_EQ(pres, t.pr); \
|
|
|
|
|
CHECK_EQ(zres, t.zr);
|
|
|
|
|
|
|
|
|
|
CHECK_VRINT(-0.5, -1.0, -0.0, -1.0, -0.0, -0.0)
|
|
|
|
|
CHECK_VRINT(-0.6, -1.0, -1.0, -1.0, -0.0, -0.0)
|
|
|
|
|
CHECK_VRINT(-1.1, -1.0, -1.0, -2.0, -1.0, -1.0)
|
|
|
|
|
CHECK_VRINT(0.5, 1.0, 0.0, 0.0, 1.0, 0.0)
|
|
|
|
|
CHECK_VRINT(0.6, 1.0, 1.0, 0.0, 1.0, 0.0)
|
|
|
|
|
CHECK_VRINT(1.1, 1.0, 1.0, 1.0, 2.0, 1.0)
|
|
|
|
|
float inf = std::numeric_limits<float>::infinity();
|
|
|
|
|
CHECK_VRINT(inf, inf, inf, inf, inf, inf)
|
|
|
|
|
CHECK_VRINT(-inf, -inf, -inf, -inf, -inf, -inf)
|
|
|
|
|
CHECK_VRINT(-0.0, -0.0, -0.0, -0.0, -0.0, -0.0)
|
|
|
|
|
|
|
|
|
|
// Check NaN propagation.
|
|
|
|
|
float nan = std::numeric_limits<float>::quiet_NaN();
|
|
|
|
|
t.input = nan;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, 0, 0, 0, 0);
|
2022-05-05 13:37:00 +00:00
|
|
|
CHECK_EQ(base::bit_cast<int32_t>(nan), base::bit_cast<int32_t>(t.ar));
|
|
|
|
|
CHECK_EQ(base::bit_cast<int32_t>(nan), base::bit_cast<int32_t>(t.nr));
|
|
|
|
|
CHECK_EQ(base::bit_cast<int32_t>(nan), base::bit_cast<int32_t>(t.mr));
|
|
|
|
|
CHECK_EQ(base::bit_cast<int32_t>(nan), base::bit_cast<int32_t>(t.pr));
|
|
|
|
|
CHECK_EQ(base::bit_cast<int32_t>(nan), base::bit_cast<int32_t>(t.zr));
|
2015-11-25 19:28:42 +00:00
|
|
|
|
|
|
|
|
#undef CHECK_VRINT
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2014-10-30 11:00:47 +00:00
|
|
|
TEST(ARMv8_vrintX) {
|
|
|
|
|
// Test the vrintX floating point instructions.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2014-10-30 11:00:47 +00:00
|
|
|
double input;
|
|
|
|
|
double ar;
|
|
|
|
|
double nr;
|
|
|
|
|
double mr;
|
|
|
|
|
double pr;
|
|
|
|
|
double zr;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2014-10-30 11:00:47 +00:00
|
|
|
T t;
|
|
|
|
|
|
|
|
|
|
// Create a function that accepts &t, and loads, manipulates, and stores
|
|
|
|
|
// the doubles and floats.
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2014-10-30 11:00:47 +00:00
|
|
|
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(ARMv8)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, ARMv8);
|
|
|
|
|
|
|
|
|
|
__ mov(ip, Operand(sp));
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, fp, lr});
|
2014-10-30 11:00:47 +00:00
|
|
|
|
|
|
|
|
__ mov(r4, Operand(r0));
|
|
|
|
|
|
|
|
|
|
// Test vrinta
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d6, r4, offsetof(T, input));
|
2014-10-30 11:00:47 +00:00
|
|
|
__ vrinta(d5, d6);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d5, r4, offsetof(T, ar));
|
2014-10-30 11:00:47 +00:00
|
|
|
|
|
|
|
|
// Test vrintn
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d6, r4, offsetof(T, input));
|
2014-10-30 11:00:47 +00:00
|
|
|
__ vrintn(d5, d6);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d5, r4, offsetof(T, nr));
|
2014-10-30 11:00:47 +00:00
|
|
|
|
|
|
|
|
// Test vrintp
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d6, r4, offsetof(T, input));
|
2014-10-30 11:00:47 +00:00
|
|
|
__ vrintp(d5, d6);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d5, r4, offsetof(T, pr));
|
2014-10-30 11:00:47 +00:00
|
|
|
|
|
|
|
|
// Test vrintm
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d6, r4, offsetof(T, input));
|
2014-10-30 11:00:47 +00:00
|
|
|
__ vrintm(d5, d6);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d5, r4, offsetof(T, mr));
|
2014-10-30 11:00:47 +00:00
|
|
|
|
|
|
|
|
// Test vrintz
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vldr(d6, r4, offsetof(T, input));
|
2014-10-30 11:00:47 +00:00
|
|
|
__ vrintz(d5, d6);
|
2015-06-17 09:06:44 +00:00
|
|
|
__ vstr(d5, r4, offsetof(T, zr));
|
2014-10-30 11:00:47 +00:00
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, fp, pc});
|
2014-10-30 11:00:47 +00:00
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2014-10-30 11:00:47 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2014-10-30 11:00:47 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
2014-10-30 11:00:47 +00:00
|
|
|
|
|
|
|
|
#define CHECK_VRINT(input_val, ares, nres, mres, pres, zres) \
|
|
|
|
|
t.input = input_val; \
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, 0, 0, 0, 0); \
|
2014-10-30 11:00:47 +00:00
|
|
|
CHECK_EQ(ares, t.ar); \
|
|
|
|
|
CHECK_EQ(nres, t.nr); \
|
|
|
|
|
CHECK_EQ(mres, t.mr); \
|
|
|
|
|
CHECK_EQ(pres, t.pr); \
|
|
|
|
|
CHECK_EQ(zres, t.zr);
|
|
|
|
|
|
|
|
|
|
CHECK_VRINT(-0.5, -1.0, -0.0, -1.0, -0.0, -0.0)
|
|
|
|
|
CHECK_VRINT(-0.6, -1.0, -1.0, -1.0, -0.0, -0.0)
|
|
|
|
|
CHECK_VRINT(-1.1, -1.0, -1.0, -2.0, -1.0, -1.0)
|
|
|
|
|
CHECK_VRINT(0.5, 1.0, 0.0, 0.0, 1.0, 0.0)
|
|
|
|
|
CHECK_VRINT(0.6, 1.0, 1.0, 0.0, 1.0, 0.0)
|
|
|
|
|
CHECK_VRINT(1.1, 1.0, 1.0, 1.0, 2.0, 1.0)
|
|
|
|
|
double inf = std::numeric_limits<double>::infinity();
|
|
|
|
|
CHECK_VRINT(inf, inf, inf, inf, inf, inf)
|
|
|
|
|
CHECK_VRINT(-inf, -inf, -inf, -inf, -inf, -inf)
|
|
|
|
|
CHECK_VRINT(-0.0, -0.0, -0.0, -0.0, -0.0, -0.0)
|
2015-10-30 14:10:23 +00:00
|
|
|
|
|
|
|
|
// Check NaN propagation.
|
2014-10-30 11:00:47 +00:00
|
|
|
double nan = std::numeric_limits<double>::quiet_NaN();
|
2015-10-30 14:10:23 +00:00
|
|
|
t.input = nan;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, 0, 0, 0, 0);
|
2022-05-05 13:37:00 +00:00
|
|
|
CHECK_EQ(base::bit_cast<int64_t>(nan), base::bit_cast<int64_t>(t.ar));
|
|
|
|
|
CHECK_EQ(base::bit_cast<int64_t>(nan), base::bit_cast<int64_t>(t.nr));
|
|
|
|
|
CHECK_EQ(base::bit_cast<int64_t>(nan), base::bit_cast<int64_t>(t.mr));
|
|
|
|
|
CHECK_EQ(base::bit_cast<int64_t>(nan), base::bit_cast<int64_t>(t.pr));
|
|
|
|
|
CHECK_EQ(base::bit_cast<int64_t>(nan), base::bit_cast<int64_t>(t.zr));
|
2014-10-30 11:00:47 +00:00
|
|
|
|
|
|
|
|
#undef CHECK_VRINT
|
|
|
|
|
}
|
|
|
|
|
}
|
2015-07-09 09:33:17 +00:00
|
|
|
|
2016-04-06 10:17:14 +00:00
|
|
|
TEST(ARMv8_vsel) {
|
|
|
|
|
// Test the vsel floating point instructions.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2016-04-06 10:17:14 +00:00
|
|
|
|
|
|
|
|
// Used to indicate whether a condition passed or failed.
|
|
|
|
|
static constexpr float kResultPass = 1.0f;
|
|
|
|
|
static constexpr float kResultFail = -kResultPass;
|
|
|
|
|
|
|
|
|
|
struct ResultsF32 {
|
|
|
|
|
float vseleq_;
|
|
|
|
|
float vselge_;
|
|
|
|
|
float vselgt_;
|
|
|
|
|
float vselvs_;
|
|
|
|
|
|
|
|
|
|
// The following conditions aren't architecturally supported, but the
|
|
|
|
|
// assembler implements them by swapping the inputs.
|
|
|
|
|
float vselne_;
|
|
|
|
|
float vsellt_;
|
|
|
|
|
float vselle_;
|
|
|
|
|
float vselvc_;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
struct ResultsF64 {
|
|
|
|
|
double vseleq_;
|
|
|
|
|
double vselge_;
|
|
|
|
|
double vselgt_;
|
|
|
|
|
double vselvs_;
|
|
|
|
|
|
|
|
|
|
// The following conditions aren't architecturally supported, but the
|
|
|
|
|
// assembler implements them by swapping the inputs.
|
|
|
|
|
double vselne_;
|
|
|
|
|
double vsellt_;
|
|
|
|
|
double vselle_;
|
|
|
|
|
double vselvc_;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(ARMv8)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, ARMv8);
|
|
|
|
|
|
|
|
|
|
// Create a helper function:
|
|
|
|
|
// void TestVsel(uint32_t nzcv,
|
|
|
|
|
// ResultsF32* results_f32,
|
|
|
|
|
// ResultsF64* results_f64);
|
|
|
|
|
__ msr(CPSR_f, Operand(r0));
|
|
|
|
|
|
2017-07-07 11:32:01 +00:00
|
|
|
__ vmov(s1, Float32(kResultPass));
|
|
|
|
|
__ vmov(s2, Float32(kResultFail));
|
2016-04-06 10:17:14 +00:00
|
|
|
|
|
|
|
|
__ vsel(eq, s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(ResultsF32, vseleq_));
|
|
|
|
|
__ vsel(ge, s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(ResultsF32, vselge_));
|
|
|
|
|
__ vsel(gt, s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(ResultsF32, vselgt_));
|
|
|
|
|
__ vsel(vs, s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(ResultsF32, vselvs_));
|
|
|
|
|
|
|
|
|
|
__ vsel(ne, s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(ResultsF32, vselne_));
|
|
|
|
|
__ vsel(lt, s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(ResultsF32, vsellt_));
|
|
|
|
|
__ vsel(le, s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(ResultsF32, vselle_));
|
|
|
|
|
__ vsel(vc, s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(ResultsF32, vselvc_));
|
|
|
|
|
|
2021-06-24 13:32:01 +00:00
|
|
|
__ vmov(d1, base::Double(kResultPass));
|
|
|
|
|
__ vmov(d2, base::Double(kResultFail));
|
2016-04-06 10:17:14 +00:00
|
|
|
|
|
|
|
|
__ vsel(eq, d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r2, offsetof(ResultsF64, vseleq_));
|
|
|
|
|
__ vsel(ge, d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r2, offsetof(ResultsF64, vselge_));
|
|
|
|
|
__ vsel(gt, d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r2, offsetof(ResultsF64, vselgt_));
|
|
|
|
|
__ vsel(vs, d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r2, offsetof(ResultsF64, vselvs_));
|
|
|
|
|
|
|
|
|
|
__ vsel(ne, d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r2, offsetof(ResultsF64, vselne_));
|
|
|
|
|
__ vsel(lt, d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r2, offsetof(ResultsF64, vsellt_));
|
|
|
|
|
__ vsel(le, d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r2, offsetof(ResultsF64, vselle_));
|
|
|
|
|
__ vsel(vc, d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r2, offsetof(ResultsF64, vselvc_));
|
|
|
|
|
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2016-04-06 10:17:14 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2016-04-06 10:17:14 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_ippii>::FromCode(*code);
|
2016-04-06 10:17:14 +00:00
|
|
|
|
2022-05-13 09:19:09 +00:00
|
|
|
static_assert(kResultPass == -kResultFail);
|
2018-01-09 09:50:34 +00:00
|
|
|
#define CHECK_VSEL(n, z, c, v, vseleq, vselge, vselgt, vselvs) \
|
|
|
|
|
do { \
|
|
|
|
|
ResultsF32 results_f32; \
|
|
|
|
|
ResultsF64 results_f64; \
|
|
|
|
|
uint32_t nzcv = (n << 31) | (z << 30) | (c << 29) | (v << 28); \
|
|
|
|
|
f.Call(nzcv, &results_f32, &results_f64, 0, 0); \
|
|
|
|
|
CHECK_EQ(vseleq, results_f32.vseleq_); \
|
|
|
|
|
CHECK_EQ(vselge, results_f32.vselge_); \
|
|
|
|
|
CHECK_EQ(vselgt, results_f32.vselgt_); \
|
|
|
|
|
CHECK_EQ(vselvs, results_f32.vselvs_); \
|
|
|
|
|
CHECK_EQ(-vseleq, results_f32.vselne_); \
|
|
|
|
|
CHECK_EQ(-vselge, results_f32.vsellt_); \
|
|
|
|
|
CHECK_EQ(-vselgt, results_f32.vselle_); \
|
|
|
|
|
CHECK_EQ(-vselvs, results_f32.vselvc_); \
|
|
|
|
|
CHECK_EQ(vseleq, results_f64.vseleq_); \
|
|
|
|
|
CHECK_EQ(vselge, results_f64.vselge_); \
|
|
|
|
|
CHECK_EQ(vselgt, results_f64.vselgt_); \
|
|
|
|
|
CHECK_EQ(vselvs, results_f64.vselvs_); \
|
|
|
|
|
CHECK_EQ(-vseleq, results_f64.vselne_); \
|
|
|
|
|
CHECK_EQ(-vselge, results_f64.vsellt_); \
|
|
|
|
|
CHECK_EQ(-vselgt, results_f64.vselle_); \
|
|
|
|
|
CHECK_EQ(-vselvs, results_f64.vselvc_); \
|
2016-04-06 10:17:14 +00:00
|
|
|
} while (0);
|
|
|
|
|
|
|
|
|
|
// N Z C V vseleq vselge vselgt vselvs
|
|
|
|
|
CHECK_VSEL(0, 0, 0, 0, kResultFail, kResultPass, kResultPass, kResultFail);
|
|
|
|
|
CHECK_VSEL(0, 0, 0, 1, kResultFail, kResultFail, kResultFail, kResultPass);
|
|
|
|
|
CHECK_VSEL(0, 0, 1, 0, kResultFail, kResultPass, kResultPass, kResultFail);
|
|
|
|
|
CHECK_VSEL(0, 0, 1, 1, kResultFail, kResultFail, kResultFail, kResultPass);
|
|
|
|
|
CHECK_VSEL(0, 1, 0, 0, kResultPass, kResultPass, kResultFail, kResultFail);
|
|
|
|
|
CHECK_VSEL(0, 1, 0, 1, kResultPass, kResultFail, kResultFail, kResultPass);
|
|
|
|
|
CHECK_VSEL(0, 1, 1, 0, kResultPass, kResultPass, kResultFail, kResultFail);
|
|
|
|
|
CHECK_VSEL(0, 1, 1, 1, kResultPass, kResultFail, kResultFail, kResultPass);
|
|
|
|
|
CHECK_VSEL(1, 0, 0, 0, kResultFail, kResultFail, kResultFail, kResultFail);
|
|
|
|
|
CHECK_VSEL(1, 0, 0, 1, kResultFail, kResultPass, kResultPass, kResultPass);
|
|
|
|
|
CHECK_VSEL(1, 0, 1, 0, kResultFail, kResultFail, kResultFail, kResultFail);
|
|
|
|
|
CHECK_VSEL(1, 0, 1, 1, kResultFail, kResultPass, kResultPass, kResultPass);
|
|
|
|
|
CHECK_VSEL(1, 1, 0, 0, kResultPass, kResultFail, kResultFail, kResultFail);
|
|
|
|
|
CHECK_VSEL(1, 1, 0, 1, kResultPass, kResultPass, kResultFail, kResultPass);
|
|
|
|
|
CHECK_VSEL(1, 1, 1, 0, kResultPass, kResultFail, kResultFail, kResultFail);
|
|
|
|
|
CHECK_VSEL(1, 1, 1, 1, kResultPass, kResultPass, kResultFail, kResultPass);
|
|
|
|
|
|
|
|
|
|
#undef CHECK_VSEL
|
|
|
|
|
}
|
|
|
|
|
}
|
2015-07-09 09:33:17 +00:00
|
|
|
|
2016-09-06 08:46:17 +00:00
|
|
|
TEST(ARMv8_vminmax_f64) {
|
2016-09-06 11:27:41 +00:00
|
|
|
// Test the vminnm and vmaxnm floating point instructions.
|
2016-09-06 08:46:17 +00:00
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2016-09-06 08:46:17 +00:00
|
|
|
|
|
|
|
|
struct Inputs {
|
|
|
|
|
double left_;
|
|
|
|
|
double right_;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
struct Results {
|
|
|
|
|
double vminnm_;
|
|
|
|
|
double vmaxnm_;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(ARMv8)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, ARMv8);
|
|
|
|
|
|
|
|
|
|
// Create a helper function:
|
|
|
|
|
// void TestVminmax(const Inputs* inputs,
|
|
|
|
|
// Results* results);
|
|
|
|
|
__ vldr(d1, r0, offsetof(Inputs, left_));
|
|
|
|
|
__ vldr(d2, r0, offsetof(Inputs, right_));
|
|
|
|
|
|
|
|
|
|
__ vminnm(d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r1, offsetof(Results, vminnm_));
|
|
|
|
|
__ vmaxnm(d0, d1, d2);
|
|
|
|
|
__ vstr(d0, r1, offsetof(Results, vmaxnm_));
|
|
|
|
|
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2016-09-06 08:46:17 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2016-09-06 08:46:17 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_ppiii>::FromCode(*code);
|
2016-09-06 08:46:17 +00:00
|
|
|
|
2022-05-05 13:37:00 +00:00
|
|
|
#define CHECK_VMINMAX(left, right, vminnm, vmaxnm) \
|
|
|
|
|
do { \
|
|
|
|
|
Inputs inputs = {left, right}; \
|
|
|
|
|
Results results; \
|
|
|
|
|
f.Call(&inputs, &results, 0, 0, 0); \
|
|
|
|
|
/* Use a base::bit_cast to correctly identify -0.0 and NaNs. */ \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint64_t>(vminnm), \
|
|
|
|
|
base::bit_cast<uint64_t>(results.vminnm_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint64_t>(vmaxnm), \
|
|
|
|
|
base::bit_cast<uint64_t>(results.vmaxnm_)); \
|
2016-09-06 08:46:17 +00:00
|
|
|
} while (0);
|
|
|
|
|
|
2022-05-05 13:37:00 +00:00
|
|
|
double nan_a = base::bit_cast<double>(UINT64_C(0x7FF8000000000001));
|
|
|
|
|
double nan_b = base::bit_cast<double>(UINT64_C(0x7FF8000000000002));
|
2016-09-06 08:46:17 +00:00
|
|
|
|
|
|
|
|
CHECK_VMINMAX(1.0, -1.0, -1.0, 1.0);
|
|
|
|
|
CHECK_VMINMAX(-1.0, 1.0, -1.0, 1.0);
|
|
|
|
|
CHECK_VMINMAX(0.0, -1.0, -1.0, 0.0);
|
|
|
|
|
CHECK_VMINMAX(-1.0, 0.0, -1.0, 0.0);
|
|
|
|
|
CHECK_VMINMAX(-0.0, -1.0, -1.0, -0.0);
|
|
|
|
|
CHECK_VMINMAX(-1.0, -0.0, -1.0, -0.0);
|
|
|
|
|
CHECK_VMINMAX(0.0, 1.0, 0.0, 1.0);
|
|
|
|
|
CHECK_VMINMAX(1.0, 0.0, 0.0, 1.0);
|
|
|
|
|
|
|
|
|
|
CHECK_VMINMAX(0.0, 0.0, 0.0, 0.0);
|
|
|
|
|
CHECK_VMINMAX(-0.0, -0.0, -0.0, -0.0);
|
|
|
|
|
CHECK_VMINMAX(-0.0, 0.0, -0.0, 0.0);
|
|
|
|
|
CHECK_VMINMAX(0.0, -0.0, -0.0, 0.0);
|
|
|
|
|
|
|
|
|
|
CHECK_VMINMAX(0.0, nan_a, 0.0, 0.0);
|
|
|
|
|
CHECK_VMINMAX(nan_a, 0.0, 0.0, 0.0);
|
|
|
|
|
CHECK_VMINMAX(nan_a, nan_b, nan_a, nan_a);
|
|
|
|
|
CHECK_VMINMAX(nan_b, nan_a, nan_b, nan_b);
|
|
|
|
|
|
|
|
|
|
#undef CHECK_VMINMAX
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(ARMv8_vminmax_f32) {
|
2016-09-06 11:27:41 +00:00
|
|
|
// Test the vminnm and vmaxnm floating point instructions.
|
2016-09-06 08:46:17 +00:00
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2016-09-06 08:46:17 +00:00
|
|
|
|
|
|
|
|
struct Inputs {
|
|
|
|
|
float left_;
|
|
|
|
|
float right_;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
struct Results {
|
|
|
|
|
float vminnm_;
|
|
|
|
|
float vmaxnm_;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(ARMv8)) {
|
|
|
|
|
CpuFeatureScope scope(&assm, ARMv8);
|
|
|
|
|
|
|
|
|
|
// Create a helper function:
|
|
|
|
|
// void TestVminmax(const Inputs* inputs,
|
|
|
|
|
// Results* results);
|
|
|
|
|
__ vldr(s1, r0, offsetof(Inputs, left_));
|
|
|
|
|
__ vldr(s2, r0, offsetof(Inputs, right_));
|
|
|
|
|
|
|
|
|
|
__ vminnm(s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(Results, vminnm_));
|
|
|
|
|
__ vmaxnm(s0, s1, s2);
|
|
|
|
|
__ vstr(s0, r1, offsetof(Results, vmaxnm_));
|
|
|
|
|
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
2016-09-06 08:46:17 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2016-09-06 08:46:17 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_ppiii>::FromCode(*code);
|
2016-09-06 08:46:17 +00:00
|
|
|
|
2022-05-05 13:37:00 +00:00
|
|
|
#define CHECK_VMINMAX(left, right, vminnm, vmaxnm) \
|
|
|
|
|
do { \
|
|
|
|
|
Inputs inputs = {left, right}; \
|
|
|
|
|
Results results; \
|
|
|
|
|
f.Call(&inputs, &results, 0, 0, 0); \
|
|
|
|
|
/* Use a base::bit_cast to correctly identify -0.0 and NaNs. */ \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint32_t>(vminnm), \
|
|
|
|
|
base::bit_cast<uint32_t>(results.vminnm_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint32_t>(vmaxnm), \
|
|
|
|
|
base::bit_cast<uint32_t>(results.vmaxnm_)); \
|
2016-09-06 08:46:17 +00:00
|
|
|
} while (0);
|
|
|
|
|
|
2022-05-05 13:37:00 +00:00
|
|
|
float nan_a = base::bit_cast<float>(UINT32_C(0x7FC00001));
|
|
|
|
|
float nan_b = base::bit_cast<float>(UINT32_C(0x7FC00002));
|
2016-09-06 08:46:17 +00:00
|
|
|
|
|
|
|
|
CHECK_VMINMAX(1.0f, -1.0f, -1.0f, 1.0f);
|
|
|
|
|
CHECK_VMINMAX(-1.0f, 1.0f, -1.0f, 1.0f);
|
|
|
|
|
CHECK_VMINMAX(0.0f, -1.0f, -1.0f, 0.0f);
|
|
|
|
|
CHECK_VMINMAX(-1.0f, 0.0f, -1.0f, 0.0f);
|
|
|
|
|
CHECK_VMINMAX(-0.0f, -1.0f, -1.0f, -0.0f);
|
|
|
|
|
CHECK_VMINMAX(-1.0f, -0.0f, -1.0f, -0.0f);
|
|
|
|
|
CHECK_VMINMAX(0.0f, 1.0f, 0.0f, 1.0f);
|
|
|
|
|
CHECK_VMINMAX(1.0f, 0.0f, 0.0f, 1.0f);
|
|
|
|
|
|
|
|
|
|
CHECK_VMINMAX(0.0f, 0.0f, 0.0f, 0.0f);
|
|
|
|
|
CHECK_VMINMAX(-0.0f, -0.0f, -0.0f, -0.0f);
|
|
|
|
|
CHECK_VMINMAX(-0.0f, 0.0f, -0.0f, 0.0f);
|
|
|
|
|
CHECK_VMINMAX(0.0f, -0.0f, -0.0f, 0.0f);
|
|
|
|
|
|
|
|
|
|
CHECK_VMINMAX(0.0f, nan_a, 0.0f, 0.0f);
|
|
|
|
|
CHECK_VMINMAX(nan_a, 0.0f, 0.0f, 0.0f);
|
|
|
|
|
CHECK_VMINMAX(nan_a, nan_b, nan_a, nan_a);
|
|
|
|
|
CHECK_VMINMAX(nan_b, nan_a, nan_b, nan_b);
|
|
|
|
|
|
|
|
|
|
#undef CHECK_VMINMAX
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2016-09-06 11:27:41 +00:00
|
|
|
template <typename T, typename Inputs, typename Results>
|
2019-07-08 08:42:48 +00:00
|
|
|
static GeneratedCode<F_ppiii> GenerateMacroFloatMinMax(
|
2019-09-10 01:19:59 +00:00
|
|
|
MacroAssembler* assm_ptr) {
|
|
|
|
|
MacroAssembler& assm = *assm_ptr;
|
|
|
|
|
|
2016-09-06 11:27:41 +00:00
|
|
|
T a = T::from_code(0); // d0/s0
|
|
|
|
|
T b = T::from_code(1); // d1/s1
|
|
|
|
|
T c = T::from_code(2); // d2/s2
|
|
|
|
|
|
|
|
|
|
// Create a helper function:
|
|
|
|
|
// void TestFloatMinMax(const Inputs* inputs,
|
|
|
|
|
// Results* results);
|
|
|
|
|
Label ool_min_abc, ool_min_aab, ool_min_aba;
|
|
|
|
|
Label ool_max_abc, ool_max_aab, ool_max_aba;
|
|
|
|
|
|
|
|
|
|
Label done_min_abc, done_min_aab, done_min_aba;
|
|
|
|
|
Label done_max_abc, done_max_aab, done_max_aba;
|
|
|
|
|
|
|
|
|
|
// a = min(b, c);
|
|
|
|
|
__ vldr(b, r0, offsetof(Inputs, left_));
|
|
|
|
|
__ vldr(c, r0, offsetof(Inputs, right_));
|
|
|
|
|
__ FloatMin(a, b, c, &ool_min_abc);
|
|
|
|
|
__ bind(&done_min_abc);
|
|
|
|
|
__ vstr(a, r1, offsetof(Results, min_abc_));
|
|
|
|
|
|
|
|
|
|
// a = min(a, b);
|
|
|
|
|
__ vldr(a, r0, offsetof(Inputs, left_));
|
|
|
|
|
__ vldr(b, r0, offsetof(Inputs, right_));
|
|
|
|
|
__ FloatMin(a, a, b, &ool_min_aab);
|
|
|
|
|
__ bind(&done_min_aab);
|
|
|
|
|
__ vstr(a, r1, offsetof(Results, min_aab_));
|
|
|
|
|
|
|
|
|
|
// a = min(b, a);
|
|
|
|
|
__ vldr(b, r0, offsetof(Inputs, left_));
|
|
|
|
|
__ vldr(a, r0, offsetof(Inputs, right_));
|
|
|
|
|
__ FloatMin(a, b, a, &ool_min_aba);
|
|
|
|
|
__ bind(&done_min_aba);
|
|
|
|
|
__ vstr(a, r1, offsetof(Results, min_aba_));
|
|
|
|
|
|
|
|
|
|
// a = max(b, c);
|
|
|
|
|
__ vldr(b, r0, offsetof(Inputs, left_));
|
|
|
|
|
__ vldr(c, r0, offsetof(Inputs, right_));
|
|
|
|
|
__ FloatMax(a, b, c, &ool_max_abc);
|
|
|
|
|
__ bind(&done_max_abc);
|
|
|
|
|
__ vstr(a, r1, offsetof(Results, max_abc_));
|
|
|
|
|
|
|
|
|
|
// a = max(a, b);
|
|
|
|
|
__ vldr(a, r0, offsetof(Inputs, left_));
|
|
|
|
|
__ vldr(b, r0, offsetof(Inputs, right_));
|
|
|
|
|
__ FloatMax(a, a, b, &ool_max_aab);
|
|
|
|
|
__ bind(&done_max_aab);
|
|
|
|
|
__ vstr(a, r1, offsetof(Results, max_aab_));
|
|
|
|
|
|
|
|
|
|
// a = max(b, a);
|
|
|
|
|
__ vldr(b, r0, offsetof(Inputs, left_));
|
|
|
|
|
__ vldr(a, r0, offsetof(Inputs, right_));
|
|
|
|
|
__ FloatMax(a, b, a, &ool_max_aba);
|
|
|
|
|
__ bind(&done_max_aba);
|
|
|
|
|
__ vstr(a, r1, offsetof(Results, max_aba_));
|
|
|
|
|
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
// Generate out-of-line cases.
|
|
|
|
|
__ bind(&ool_min_abc);
|
|
|
|
|
__ FloatMinOutOfLine(a, b, c);
|
|
|
|
|
__ b(&done_min_abc);
|
|
|
|
|
|
|
|
|
|
__ bind(&ool_min_aab);
|
|
|
|
|
__ FloatMinOutOfLine(a, a, b);
|
|
|
|
|
__ b(&done_min_aab);
|
|
|
|
|
|
|
|
|
|
__ bind(&ool_min_aba);
|
|
|
|
|
__ FloatMinOutOfLine(a, b, a);
|
|
|
|
|
__ b(&done_min_aba);
|
|
|
|
|
|
|
|
|
|
__ bind(&ool_max_abc);
|
|
|
|
|
__ FloatMaxOutOfLine(a, b, c);
|
|
|
|
|
__ b(&done_max_abc);
|
|
|
|
|
|
|
|
|
|
__ bind(&ool_max_aab);
|
|
|
|
|
__ FloatMaxOutOfLine(a, a, b);
|
|
|
|
|
__ b(&done_max_aab);
|
|
|
|
|
|
|
|
|
|
__ bind(&ool_max_aba);
|
|
|
|
|
__ FloatMaxOutOfLine(a, b, a);
|
|
|
|
|
__ b(&done_max_aba);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
2017-05-31 14:00:11 +00:00
|
|
|
assm.GetCode(assm.isolate(), &desc);
|
2017-09-29 14:59:24 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(assm.isolate(), desc, CodeKind::FOR_TESTING).Build();
|
2016-09-06 11:27:41 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2016-09-06 11:27:41 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
return GeneratedCode<F_ppiii>::FromCode(*code);
|
2016-09-06 11:27:41 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(macro_float_minmax_f64) {
|
|
|
|
|
// Test the FloatMin and FloatMax macros.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
MacroAssembler assm(isolate, CodeObjectRequired::kYes);
|
2016-09-06 11:27:41 +00:00
|
|
|
|
|
|
|
|
struct Inputs {
|
|
|
|
|
double left_;
|
|
|
|
|
double right_;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
struct Results {
|
|
|
|
|
// Check all register aliasing possibilities in order to exercise all
|
|
|
|
|
// code-paths in the macro assembler.
|
|
|
|
|
double min_abc_;
|
|
|
|
|
double min_aab_;
|
|
|
|
|
double min_aba_;
|
|
|
|
|
double max_abc_;
|
|
|
|
|
double max_aab_;
|
|
|
|
|
double max_aba_;
|
|
|
|
|
};
|
|
|
|
|
|
2019-09-10 01:19:59 +00:00
|
|
|
auto f = GenerateMacroFloatMinMax<DwVfpRegister, Inputs, Results>(&assm);
|
2016-09-06 11:27:41 +00:00
|
|
|
|
2022-05-05 13:37:00 +00:00
|
|
|
#define CHECK_MINMAX(left, right, min, max) \
|
|
|
|
|
do { \
|
|
|
|
|
Inputs inputs = {left, right}; \
|
|
|
|
|
Results results; \
|
|
|
|
|
f.Call(&inputs, &results, 0, 0, 0); \
|
|
|
|
|
/* Use a base::bit_cast to correctly identify -0.0 and NaNs. */ \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint64_t>(min), \
|
|
|
|
|
base::bit_cast<uint64_t>(results.min_abc_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint64_t>(min), \
|
|
|
|
|
base::bit_cast<uint64_t>(results.min_aab_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint64_t>(min), \
|
|
|
|
|
base::bit_cast<uint64_t>(results.min_aba_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint64_t>(max), \
|
|
|
|
|
base::bit_cast<uint64_t>(results.max_abc_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint64_t>(max), \
|
|
|
|
|
base::bit_cast<uint64_t>(results.max_aab_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint64_t>(max), \
|
|
|
|
|
base::bit_cast<uint64_t>(results.max_aba_)); \
|
2016-09-06 11:27:41 +00:00
|
|
|
} while (0)
|
|
|
|
|
|
2022-05-05 13:37:00 +00:00
|
|
|
double nan_a = base::bit_cast<double>(UINT64_C(0x7FF8000000000001));
|
|
|
|
|
double nan_b = base::bit_cast<double>(UINT64_C(0x7FF8000000000002));
|
2016-09-06 11:27:41 +00:00
|
|
|
|
|
|
|
|
CHECK_MINMAX(1.0, -1.0, -1.0, 1.0);
|
|
|
|
|
CHECK_MINMAX(-1.0, 1.0, -1.0, 1.0);
|
|
|
|
|
CHECK_MINMAX(0.0, -1.0, -1.0, 0.0);
|
|
|
|
|
CHECK_MINMAX(-1.0, 0.0, -1.0, 0.0);
|
|
|
|
|
CHECK_MINMAX(-0.0, -1.0, -1.0, -0.0);
|
|
|
|
|
CHECK_MINMAX(-1.0, -0.0, -1.0, -0.0);
|
|
|
|
|
CHECK_MINMAX(0.0, 1.0, 0.0, 1.0);
|
|
|
|
|
CHECK_MINMAX(1.0, 0.0, 0.0, 1.0);
|
|
|
|
|
|
|
|
|
|
CHECK_MINMAX(0.0, 0.0, 0.0, 0.0);
|
|
|
|
|
CHECK_MINMAX(-0.0, -0.0, -0.0, -0.0);
|
|
|
|
|
CHECK_MINMAX(-0.0, 0.0, -0.0, 0.0);
|
|
|
|
|
CHECK_MINMAX(0.0, -0.0, -0.0, 0.0);
|
|
|
|
|
|
|
|
|
|
CHECK_MINMAX(0.0, nan_a, nan_a, nan_a);
|
|
|
|
|
CHECK_MINMAX(nan_a, 0.0, nan_a, nan_a);
|
|
|
|
|
CHECK_MINMAX(nan_a, nan_b, nan_a, nan_a);
|
|
|
|
|
CHECK_MINMAX(nan_b, nan_a, nan_b, nan_b);
|
|
|
|
|
|
|
|
|
|
#undef CHECK_MINMAX
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(macro_float_minmax_f32) {
|
|
|
|
|
// Test the FloatMin and FloatMax macros.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
MacroAssembler assm(isolate, CodeObjectRequired::kYes);
|
2016-09-06 11:27:41 +00:00
|
|
|
|
|
|
|
|
struct Inputs {
|
|
|
|
|
float left_;
|
|
|
|
|
float right_;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
struct Results {
|
|
|
|
|
// Check all register aliasing possibilities in order to exercise all
|
|
|
|
|
// code-paths in the macro assembler.
|
|
|
|
|
float min_abc_;
|
|
|
|
|
float min_aab_;
|
|
|
|
|
float min_aba_;
|
|
|
|
|
float max_abc_;
|
|
|
|
|
float max_aab_;
|
|
|
|
|
float max_aba_;
|
|
|
|
|
};
|
|
|
|
|
|
2019-09-10 01:19:59 +00:00
|
|
|
auto f = GenerateMacroFloatMinMax<SwVfpRegister, Inputs, Results>(&assm);
|
2016-09-06 11:27:41 +00:00
|
|
|
|
2022-05-05 13:37:00 +00:00
|
|
|
#define CHECK_MINMAX(left, right, min, max) \
|
|
|
|
|
do { \
|
|
|
|
|
Inputs inputs = {left, right}; \
|
|
|
|
|
Results results; \
|
|
|
|
|
f.Call(&inputs, &results, 0, 0, 0); \
|
|
|
|
|
/* Use a base::bit_cast to correctly identify -0.0 and NaNs. */ \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint32_t>(min), \
|
|
|
|
|
base::bit_cast<uint32_t>(results.min_abc_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint32_t>(min), \
|
|
|
|
|
base::bit_cast<uint32_t>(results.min_aab_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint32_t>(min), \
|
|
|
|
|
base::bit_cast<uint32_t>(results.min_aba_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint32_t>(max), \
|
|
|
|
|
base::bit_cast<uint32_t>(results.max_abc_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint32_t>(max), \
|
|
|
|
|
base::bit_cast<uint32_t>(results.max_aab_)); \
|
|
|
|
|
CHECK_EQ(base::bit_cast<uint32_t>(max), \
|
|
|
|
|
base::bit_cast<uint32_t>(results.max_aba_)); \
|
2016-09-06 11:27:41 +00:00
|
|
|
} while (0)
|
|
|
|
|
|
2022-05-05 13:37:00 +00:00
|
|
|
float nan_a = base::bit_cast<float>(UINT32_C(0x7FC00001));
|
|
|
|
|
float nan_b = base::bit_cast<float>(UINT32_C(0x7FC00002));
|
2016-09-06 11:27:41 +00:00
|
|
|
|
|
|
|
|
CHECK_MINMAX(1.0f, -1.0f, -1.0f, 1.0f);
|
|
|
|
|
CHECK_MINMAX(-1.0f, 1.0f, -1.0f, 1.0f);
|
|
|
|
|
CHECK_MINMAX(0.0f, -1.0f, -1.0f, 0.0f);
|
|
|
|
|
CHECK_MINMAX(-1.0f, 0.0f, -1.0f, 0.0f);
|
|
|
|
|
CHECK_MINMAX(-0.0f, -1.0f, -1.0f, -0.0f);
|
|
|
|
|
CHECK_MINMAX(-1.0f, -0.0f, -1.0f, -0.0f);
|
|
|
|
|
CHECK_MINMAX(0.0f, 1.0f, 0.0f, 1.0f);
|
|
|
|
|
CHECK_MINMAX(1.0f, 0.0f, 0.0f, 1.0f);
|
|
|
|
|
|
|
|
|
|
CHECK_MINMAX(0.0f, 0.0f, 0.0f, 0.0f);
|
|
|
|
|
CHECK_MINMAX(-0.0f, -0.0f, -0.0f, -0.0f);
|
|
|
|
|
CHECK_MINMAX(-0.0f, 0.0f, -0.0f, 0.0f);
|
|
|
|
|
CHECK_MINMAX(0.0f, -0.0f, -0.0f, 0.0f);
|
|
|
|
|
|
|
|
|
|
CHECK_MINMAX(0.0f, nan_a, nan_a, nan_a);
|
|
|
|
|
CHECK_MINMAX(nan_a, 0.0f, nan_a, nan_a);
|
|
|
|
|
CHECK_MINMAX(nan_a, nan_b, nan_a, nan_a);
|
|
|
|
|
CHECK_MINMAX(nan_b, nan_a, nan_b, nan_b);
|
|
|
|
|
|
|
|
|
|
#undef CHECK_MINMAX
|
|
|
|
|
}
|
|
|
|
|
|
2016-07-27 13:24:28 +00:00
|
|
|
TEST(unaligned_loads) {
|
|
|
|
|
// All supported ARM targets allow unaligned accesses.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2016-07-27 13:24:28 +00:00
|
|
|
uint32_t ldrh;
|
|
|
|
|
uint32_t ldrsh;
|
|
|
|
|
uint32_t ldr;
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2016-07-27 13:24:28 +00:00
|
|
|
T t;
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2016-07-27 13:24:28 +00:00
|
|
|
__ ldrh(ip, MemOperand(r1, r2));
|
|
|
|
|
__ str(ip, MemOperand(r0, offsetof(T, ldrh)));
|
|
|
|
|
__ ldrsh(ip, MemOperand(r1, r2));
|
|
|
|
|
__ str(ip, MemOperand(r0, offsetof(T, ldrsh)));
|
|
|
|
|
__ ldr(ip, MemOperand(r1, r2));
|
|
|
|
|
__ str(ip, MemOperand(r0, offsetof(T, ldr)));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
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();
|
2016-07-27 13:24:28 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2016-07-27 13:24:28 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_ppiii>::FromCode(*code);
|
2016-07-27 13:24:28 +00:00
|
|
|
|
|
|
|
|
#ifndef V8_TARGET_LITTLE_ENDIAN
|
|
|
|
|
#error This test assumes a little-endian layout.
|
|
|
|
|
#endif
|
2017-12-02 00:30:37 +00:00
|
|
|
uint64_t data = UINT64_C(0x84838281807F7E7D);
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, &data, 0, 0, 0);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(0x00007E7Du, t.ldrh);
|
|
|
|
|
CHECK_EQ(0x00007E7Du, t.ldrsh);
|
|
|
|
|
CHECK_EQ(0x807F7E7Du, t.ldr);
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, &data, 1, 0, 0);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(0x00007F7Eu, t.ldrh);
|
|
|
|
|
CHECK_EQ(0x00007F7Eu, t.ldrsh);
|
|
|
|
|
CHECK_EQ(0x81807F7Eu, t.ldr);
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, &data, 2, 0, 0);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(0x0000807Fu, t.ldrh);
|
|
|
|
|
CHECK_EQ(0xFFFF807Fu, t.ldrsh);
|
|
|
|
|
CHECK_EQ(0x8281807Fu, t.ldr);
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&t, &data, 3, 0, 0);
|
2016-11-11 12:00:34 +00:00
|
|
|
CHECK_EQ(0x00008180u, t.ldrh);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(0xFFFF8180u, t.ldrsh);
|
2016-11-11 12:00:34 +00:00
|
|
|
CHECK_EQ(0x83828180u, t.ldr);
|
2016-07-27 13:24:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(unaligned_stores) {
|
|
|
|
|
// All supported ARM targets allow unaligned accesses.
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2016-07-27 13:24:28 +00:00
|
|
|
__ strh(r3, MemOperand(r0, r2));
|
|
|
|
|
__ str(r3, MemOperand(r1, r2));
|
|
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
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();
|
2016-07-27 13:24:28 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2016-07-27 13:24:28 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_ppiii>::FromCode(*code);
|
2016-07-27 13:24:28 +00:00
|
|
|
|
|
|
|
|
#ifndef V8_TARGET_LITTLE_ENDIAN
|
|
|
|
|
#error This test assumes a little-endian layout.
|
|
|
|
|
#endif
|
|
|
|
|
{
|
|
|
|
|
uint64_t strh = 0;
|
|
|
|
|
uint64_t str = 0;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&strh, &str, 0, 0xFEDCBA98, 0);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(UINT64_C(0x000000000000BA98), strh);
|
|
|
|
|
CHECK_EQ(UINT64_C(0x00000000FEDCBA98), str);
|
2016-07-27 13:24:28 +00:00
|
|
|
}
|
|
|
|
|
{
|
|
|
|
|
uint64_t strh = 0;
|
|
|
|
|
uint64_t str = 0;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&strh, &str, 1, 0xFEDCBA98, 0);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(UINT64_C(0x0000000000BA9800), strh);
|
|
|
|
|
CHECK_EQ(UINT64_C(0x000000FEDCBA9800), str);
|
2016-07-27 13:24:28 +00:00
|
|
|
}
|
|
|
|
|
{
|
|
|
|
|
uint64_t strh = 0;
|
|
|
|
|
uint64_t str = 0;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&strh, &str, 2, 0xFEDCBA98, 0);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(UINT64_C(0x00000000BA980000), strh);
|
|
|
|
|
CHECK_EQ(UINT64_C(0x0000FEDCBA980000), str);
|
2016-07-27 13:24:28 +00:00
|
|
|
}
|
|
|
|
|
{
|
|
|
|
|
uint64_t strh = 0;
|
|
|
|
|
uint64_t str = 0;
|
2018-01-09 09:50:34 +00:00
|
|
|
f.Call(&strh, &str, 3, 0xFEDCBA98, 0);
|
2017-12-02 00:30:37 +00:00
|
|
|
CHECK_EQ(UINT64_C(0x000000BA98000000), strh);
|
|
|
|
|
CHECK_EQ(UINT64_C(0x00FEDCBA98000000), str);
|
2016-07-27 13:24:28 +00:00
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2016-09-08 07:11:32 +00:00
|
|
|
TEST(vswp) {
|
2016-11-25 19:36:45 +00:00
|
|
|
if (!CpuFeatures::IsSupported(NEON)) return;
|
|
|
|
|
|
2016-09-08 07:11:32 +00:00
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2016-09-08 07:11:32 +00:00
|
|
|
|
2019-05-27 13:12:51 +00:00
|
|
|
struct T {
|
2016-11-30 19:44:45 +00:00
|
|
|
uint64_t vswp_d0;
|
|
|
|
|
uint64_t vswp_d1;
|
|
|
|
|
uint64_t vswp_d30;
|
|
|
|
|
uint64_t vswp_d31;
|
|
|
|
|
uint32_t vswp_q4[4];
|
|
|
|
|
uint32_t vswp_q5[4];
|
2019-05-27 13:12:51 +00:00
|
|
|
};
|
2016-09-08 07:11:32 +00:00
|
|
|
T t;
|
|
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ stm(db_w, sp, {r4, r5, r6, r7, lr});
|
2016-11-30 19:44:45 +00:00
|
|
|
|
2022-05-05 13:37:00 +00:00
|
|
|
uint64_t one = base::bit_cast<uint64_t>(1.0);
|
2016-11-30 19:44:45 +00:00
|
|
|
__ mov(r5, Operand(one >> 32));
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r4, Operand(one & 0xFFFFFFFF));
|
2022-05-05 13:37:00 +00:00
|
|
|
uint64_t minus_one = base::bit_cast<uint64_t>(-1.0);
|
2016-11-30 19:44:45 +00:00
|
|
|
__ mov(r7, Operand(minus_one >> 32));
|
2017-12-02 00:30:37 +00:00
|
|
|
__ mov(r6, Operand(minus_one & 0xFFFFFFFF));
|
2016-11-30 19:44:45 +00:00
|
|
|
|
|
|
|
|
__ vmov(d0, r4, r5); // d0 = 1.0
|
|
|
|
|
__ vmov(d1, r6, r7); // d1 = -1.0
|
2016-09-08 07:11:32 +00:00
|
|
|
__ vswp(d0, d1);
|
2016-11-30 19:44:45 +00:00
|
|
|
__ vstr(d0, r0, offsetof(T, vswp_d0));
|
|
|
|
|
__ vstr(d1, r0, offsetof(T, vswp_d1));
|
2016-09-08 07:11:32 +00:00
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(VFP32DREGS)) {
|
2016-11-30 19:44:45 +00:00
|
|
|
__ vmov(d30, r4, r5); // d30 = 1.0
|
|
|
|
|
__ vmov(d31, r6, r7); // d31 = -1.0
|
2016-09-08 07:11:32 +00:00
|
|
|
__ vswp(d30, d31);
|
2016-11-30 19:44:45 +00:00
|
|
|
__ vstr(d30, r0, offsetof(T, vswp_d30));
|
|
|
|
|
__ vstr(d31, r0, offsetof(T, vswp_d31));
|
2016-09-08 07:11:32 +00:00
|
|
|
}
|
|
|
|
|
|
2016-11-25 19:36:45 +00:00
|
|
|
// q-register swap.
|
2016-11-30 19:44:45 +00:00
|
|
|
const uint32_t test_1 = 0x01234567;
|
2017-12-02 00:30:37 +00:00
|
|
|
const uint32_t test_2 = 0x89ABCDEF;
|
2016-11-30 19:44:45 +00:00
|
|
|
__ mov(r4, Operand(test_1));
|
|
|
|
|
__ mov(r5, Operand(test_2));
|
2017-05-16 20:37:16 +00:00
|
|
|
__ vdup(Neon32, q4, r4);
|
|
|
|
|
__ vdup(Neon32, q5, r5);
|
2016-11-25 19:36:45 +00:00
|
|
|
__ vswp(q4, q5);
|
2016-11-30 19:44:45 +00:00
|
|
|
__ add(r6, r0, Operand(static_cast<int32_t>(offsetof(T, vswp_q4))));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(q4), NeonMemOperand(r6));
|
2016-11-30 19:44:45 +00:00
|
|
|
__ add(r6, r0, Operand(static_cast<int32_t>(offsetof(T, vswp_q5))));
|
2016-12-15 18:15:23 +00:00
|
|
|
__ vst1(Neon8, NeonListOperand(q5), NeonMemOperand(r6));
|
2016-11-25 19:36:45 +00:00
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
__ ldm(ia_w, sp, {r4, r5, r6, r7, pc});
|
2016-09-08 07:11:32 +00:00
|
|
|
__ bx(lr);
|
|
|
|
|
|
|
|
|
|
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();
|
2016-09-08 07:11:32 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
2016-09-08 07:11:32 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_piiii>::FromCode(*code);
|
|
|
|
|
f.Call(&t, 0, 0, 0, 0);
|
2016-11-30 19:44:45 +00:00
|
|
|
CHECK_EQ(minus_one, t.vswp_d0);
|
|
|
|
|
CHECK_EQ(one, t.vswp_d1);
|
2016-09-08 07:11:32 +00:00
|
|
|
if (CpuFeatures::IsSupported(VFP32DREGS)) {
|
2016-11-30 19:44:45 +00:00
|
|
|
CHECK_EQ(minus_one, t.vswp_d30);
|
|
|
|
|
CHECK_EQ(one, t.vswp_d31);
|
2016-09-08 07:11:32 +00:00
|
|
|
}
|
2016-11-30 19:44:45 +00:00
|
|
|
CHECK_EQ(t.vswp_q4[0], test_2);
|
|
|
|
|
CHECK_EQ(t.vswp_q4[1], test_2);
|
|
|
|
|
CHECK_EQ(t.vswp_q4[2], test_2);
|
|
|
|
|
CHECK_EQ(t.vswp_q4[3], test_2);
|
|
|
|
|
CHECK_EQ(t.vswp_q5[0], test_1);
|
|
|
|
|
CHECK_EQ(t.vswp_q5[1], test_1);
|
|
|
|
|
CHECK_EQ(t.vswp_q5[2], test_1);
|
|
|
|
|
CHECK_EQ(t.vswp_q5[3], test_1);
|
2016-09-08 07:11:32 +00:00
|
|
|
}
|
|
|
|
|
|
2015-07-09 09:33:17 +00:00
|
|
|
TEST(regress4292_b) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2015-07-09 09:33:17 +00:00
|
|
|
Label end;
|
|
|
|
|
__ mov(r0, Operand(isolate->factory()->infinity_value()));
|
|
|
|
|
for (int i = 0; i < 1020; ++i) {
|
|
|
|
|
__ b(hi, &end);
|
|
|
|
|
}
|
|
|
|
|
__ bind(&end);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(regress4292_bl) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2015-07-09 09:33:17 +00:00
|
|
|
Label end;
|
|
|
|
|
__ mov(r0, Operand(isolate->factory()->infinity_value()));
|
|
|
|
|
for (int i = 0; i < 1020; ++i) {
|
|
|
|
|
__ bl(hi, &end);
|
|
|
|
|
}
|
|
|
|
|
__ bind(&end);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
TEST(regress4292_blx) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2015-07-09 09:33:17 +00:00
|
|
|
Label end;
|
|
|
|
|
__ mov(r0, Operand(isolate->factory()->infinity_value()));
|
|
|
|
|
for (int i = 0; i < 1020; ++i) {
|
|
|
|
|
__ blx(&end);
|
|
|
|
|
}
|
|
|
|
|
__ bind(&end);
|
|
|
|
|
}
|
|
|
|
|
|
2015-07-09 10:32:23 +00:00
|
|
|
|
|
|
|
|
TEST(regress4292_CheckConstPool) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2015-07-09 10:32:23 +00:00
|
|
|
__ mov(r0, Operand(isolate->factory()->infinity_value()));
|
|
|
|
|
__ BlockConstPoolFor(1019);
|
|
|
|
|
for (int i = 0; i < 1019; ++i) __ nop();
|
|
|
|
|
__ vldr(d0, MemOperand(r0, 0));
|
|
|
|
|
}
|
|
|
|
|
|
2017-06-22 10:00:56 +00:00
|
|
|
TEST(use_scratch_register_scope) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2017-06-22 10:00:56 +00:00
|
|
|
|
|
|
|
|
// The assembler should have ip as a scratch by default.
|
2022-03-14 12:14:35 +00:00
|
|
|
CHECK_EQ(*assm.GetScratchRegisterList(), RegList{ip});
|
2017-06-22 10:00:56 +00:00
|
|
|
|
|
|
|
|
{
|
|
|
|
|
UseScratchRegisterScope temps(&assm);
|
2022-03-14 12:14:35 +00:00
|
|
|
CHECK_EQ(*assm.GetScratchRegisterList(), RegList{ip});
|
2017-06-22 10:00:56 +00:00
|
|
|
|
|
|
|
|
Register scratch = temps.Acquire();
|
|
|
|
|
CHECK_EQ(scratch.code(), ip.code());
|
2022-03-14 12:14:35 +00:00
|
|
|
CHECK_EQ(*assm.GetScratchRegisterList(), RegList{});
|
2017-06-22 10:00:56 +00:00
|
|
|
}
|
|
|
|
|
|
2022-03-14 12:14:35 +00:00
|
|
|
CHECK_EQ(*assm.GetScratchRegisterList(), RegList{ip});
|
2017-06-22 10:00:56 +00:00
|
|
|
}
|
|
|
|
|
|
2017-07-19 13:50:20 +00:00
|
|
|
TEST(use_scratch_vfp_register_scope) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
2017-07-19 13:50:20 +00:00
|
|
|
|
|
|
|
|
VfpRegList orig_scratches = *assm.GetScratchVfpRegisterList();
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(VFP32DREGS)) {
|
|
|
|
|
CHECK_EQ(orig_scratches, d14.ToVfpRegList() | d15.ToVfpRegList());
|
|
|
|
|
} else {
|
|
|
|
|
CHECK_EQ(orig_scratches, d14.ToVfpRegList());
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Test each configuration of scratch registers we can have at the same time.
|
|
|
|
|
|
|
|
|
|
{
|
|
|
|
|
UseScratchRegisterScope temps(&assm);
|
|
|
|
|
|
|
|
|
|
SwVfpRegister s1_scratch = temps.AcquireS();
|
|
|
|
|
CHECK_EQ(s1_scratch, s28);
|
|
|
|
|
|
|
|
|
|
SwVfpRegister s2_scratch = temps.AcquireS();
|
|
|
|
|
CHECK_EQ(s2_scratch, s29);
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(VFP32DREGS)) {
|
|
|
|
|
SwVfpRegister s3_scratch = temps.AcquireS();
|
|
|
|
|
CHECK_EQ(s3_scratch, s30);
|
|
|
|
|
|
|
|
|
|
SwVfpRegister s4_scratch = temps.AcquireS();
|
|
|
|
|
CHECK_EQ(s4_scratch, s31);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
CHECK_EQ(*assm.GetScratchVfpRegisterList(), orig_scratches);
|
|
|
|
|
|
|
|
|
|
{
|
|
|
|
|
UseScratchRegisterScope temps(&assm);
|
|
|
|
|
|
|
|
|
|
SwVfpRegister s1_scratch = temps.AcquireS();
|
|
|
|
|
CHECK_EQ(s1_scratch, s28);
|
|
|
|
|
|
|
|
|
|
SwVfpRegister s2_scratch = temps.AcquireS();
|
|
|
|
|
CHECK_EQ(s2_scratch, s29);
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(VFP32DREGS)) {
|
|
|
|
|
DwVfpRegister d_scratch = temps.AcquireD();
|
|
|
|
|
CHECK_EQ(d_scratch, d15);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
CHECK_EQ(*assm.GetScratchVfpRegisterList(), orig_scratches);
|
|
|
|
|
|
|
|
|
|
{
|
|
|
|
|
UseScratchRegisterScope temps(&assm);
|
|
|
|
|
|
|
|
|
|
DwVfpRegister d_scratch = temps.AcquireD();
|
|
|
|
|
CHECK_EQ(d_scratch, d14);
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(VFP32DREGS)) {
|
|
|
|
|
SwVfpRegister s1_scratch = temps.AcquireS();
|
|
|
|
|
CHECK_EQ(s1_scratch, s30);
|
|
|
|
|
|
|
|
|
|
SwVfpRegister s2_scratch = temps.AcquireS();
|
|
|
|
|
CHECK_EQ(s2_scratch, s31);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
CHECK_EQ(*assm.GetScratchVfpRegisterList(), orig_scratches);
|
|
|
|
|
|
|
|
|
|
{
|
|
|
|
|
UseScratchRegisterScope temps(&assm);
|
|
|
|
|
|
|
|
|
|
DwVfpRegister d1_scratch = temps.AcquireD();
|
|
|
|
|
CHECK_EQ(d1_scratch, d14);
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(VFP32DREGS)) {
|
|
|
|
|
DwVfpRegister d2_scratch = temps.AcquireD();
|
|
|
|
|
CHECK_EQ(d2_scratch, d15);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
CHECK_EQ(*assm.GetScratchVfpRegisterList(), orig_scratches);
|
|
|
|
|
|
|
|
|
|
if (CpuFeatures::IsSupported(NEON)) {
|
|
|
|
|
UseScratchRegisterScope temps(&assm);
|
|
|
|
|
|
|
|
|
|
QwNeonRegister q_scratch = temps.AcquireQ();
|
|
|
|
|
CHECK_EQ(q_scratch, q7);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
CHECK_EQ(*assm.GetScratchVfpRegisterList(), orig_scratches);
|
|
|
|
|
}
|
|
|
|
|
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
TEST(split_add_immediate) {
|
|
|
|
|
CcTest::InitializeVM();
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
|
|
|
|
{
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
__ mov(r1, r0);
|
|
|
|
|
// Re-use the destination as a scratch.
|
|
|
|
|
__ add(r0, r1, Operand(0x12345678));
|
|
|
|
|
__ blx(lr);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
|
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
uint32_t res = reinterpret_cast<int>(f.Call(0, 0, 0, 0, 0));
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
::printf("f() = 0x%x\n", res);
|
|
|
|
|
CHECK_EQ(0x12345678, res);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
{
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
// Use ip as a scratch.
|
|
|
|
|
__ add(r0, r0, Operand(0x12345678));
|
|
|
|
|
__ blx(lr);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
|
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
uint32_t res = reinterpret_cast<int>(f.Call(0, 0, 0, 0, 0));
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
::printf("f() = 0x%x\n", res);
|
|
|
|
|
CHECK_EQ(0x12345678, res);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
{
|
2019-01-16 18:29:52 +00:00
|
|
|
Assembler assm(AssemblerOptions{});
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
UseScratchRegisterScope temps(&assm);
|
|
|
|
|
Register reserved = temps.Acquire();
|
|
|
|
|
USE(reserved);
|
|
|
|
|
// If ip is not available, split the operation into multiple additions.
|
|
|
|
|
__ add(r0, r0, Operand(0x12345678));
|
|
|
|
|
__ blx(lr);
|
|
|
|
|
|
|
|
|
|
CodeDesc desc;
|
|
|
|
|
assm.GetCode(isolate, &desc);
|
2020-08-05 11:48:03 +00:00
|
|
|
Handle<Code> code =
|
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
|
|
|
Factory::CodeBuilder(isolate, desc, CodeKind::FOR_TESTING).Build();
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
#ifdef DEBUG
|
2018-06-14 12:46:07 +00:00
|
|
|
StdoutStream os;
|
2018-07-12 10:06:42 +00:00
|
|
|
code->Print(os);
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
#endif
|
2018-01-09 09:50:34 +00:00
|
|
|
auto f = GeneratedCode<F_iiiii>::FromCode(*code);
|
|
|
|
|
uint32_t res = reinterpret_cast<int>(f.Call(0, 0, 0, 0, 0));
|
[arm] Support splitting add with immediate instructions
When an immediate does not fit an add instruction we use a temporary register to
hold the value, using movw/movt to encode it. However, in order to remove a use
of r9 in TurboFan's code generator, we need to cope with no scratch registers
being available. That is to say that the destination and source registers are
the same, and `ip` is not available to use.
In this case, we can split an add instruction into a sequence of additions:
```
UseScratchRegisterScope temps(...);
Register my_scratch = temps.Acquire();
__ add(r0, r0, Operand(0xabcd); // add r0, r0, #0xcd
// add r0, r0, #0xab00
```
As a drive-by fix, make the disassembler test fail if we expected a different
number of instructions generated.
Bug: v8:6553
Change-Id: Ib7fcc765d28bccafe39257f47cd73f922c5873bf
Reviewed-on: https://chromium-review.googlesource.com/685014
Reviewed-by: Benedikt Meurer <bmeurer@chromium.org>
Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
Cr-Commit-Position: refs/heads/master@{#48491}
2017-10-11 12:00:21 +00:00
|
|
|
::printf("f() = 0x%x\n", res);
|
|
|
|
|
CHECK_EQ(0x12345678, res);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
namespace {
|
|
|
|
|
|
|
|
|
|
std::vector<Float32> Float32Inputs() {
|
|
|
|
|
std::vector<Float32> inputs;
|
|
|
|
|
FOR_FLOAT32_INPUTS(f) {
|
2022-05-05 13:37:00 +00:00
|
|
|
inputs.push_back(Float32::FromBits(base::bit_cast<uint32_t>(f)));
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
}
|
2019-02-01 10:37:04 +00:00
|
|
|
FOR_UINT32_INPUTS(bits) { inputs.push_back(Float32::FromBits(bits)); }
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
return inputs;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
std::vector<Float64> Float64Inputs() {
|
|
|
|
|
std::vector<Float64> inputs;
|
|
|
|
|
FOR_FLOAT64_INPUTS(f) {
|
2022-05-05 13:37:00 +00:00
|
|
|
inputs.push_back(Float64::FromBits(base::bit_cast<uint64_t>(f)));
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
}
|
2019-02-01 10:37:04 +00:00
|
|
|
FOR_UINT64_INPUTS(bits) { inputs.push_back(Float64::FromBits(bits)); }
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
return inputs;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
} // namespace
|
|
|
|
|
|
|
|
|
|
TEST(vabs_32) {
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2020-11-18 09:00:45 +00:00
|
|
|
auto f = AssembleCode<F_iiiii>(isolate, [](Assembler& assm) {
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
__ vmov(s0, r0);
|
|
|
|
|
__ vabs(s0, s0);
|
|
|
|
|
__ vmov(r0, s0);
|
2018-01-09 09:50:34 +00:00
|
|
|
});
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
|
|
|
|
|
for (Float32 f32 : Float32Inputs()) {
|
2018-01-09 09:50:34 +00:00
|
|
|
Float32 res = Float32::FromBits(
|
|
|
|
|
reinterpret_cast<uint32_t>(f.Call(f32.get_bits(), 0, 0, 0, 0)));
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
Float32 exp = Float32::FromBits(f32.get_bits() & ~(1 << 31));
|
|
|
|
|
CHECK_EQ(exp.get_bits(), res.get_bits());
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(vabs_64) {
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2020-11-18 09:00:45 +00:00
|
|
|
auto f = AssembleCode<F_iiiii>(isolate, [](Assembler& assm) {
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
__ vmov(d0, r0, r1);
|
|
|
|
|
__ vabs(d0, d0);
|
|
|
|
|
__ vmov(r1, r0, d0);
|
2018-01-09 09:50:34 +00:00
|
|
|
});
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
|
|
|
|
|
for (Float64 f64 : Float64Inputs()) {
|
|
|
|
|
uint32_t p0 = static_cast<uint32_t>(f64.get_bits());
|
|
|
|
|
uint32_t p1 = static_cast<uint32_t>(f64.get_bits() >> 32);
|
2018-01-09 09:50:34 +00:00
|
|
|
uint32_t res = reinterpret_cast<uint32_t>(f.Call(p0, p1, 0, 0, 0));
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
Float64 exp = Float64::FromBits(f64.get_bits() & ~(1ull << 63));
|
|
|
|
|
// We just get back the top word, so only compare that one.
|
|
|
|
|
CHECK_EQ(exp.get_bits() >> 32, res);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(vneg_32) {
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2020-11-18 09:00:45 +00:00
|
|
|
auto f = AssembleCode<F_iiiii>(isolate, [](Assembler& assm) {
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
__ vmov(s0, r0);
|
|
|
|
|
__ vneg(s0, s0);
|
|
|
|
|
__ vmov(r0, s0);
|
2018-01-09 09:50:34 +00:00
|
|
|
});
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
|
|
|
|
|
for (Float32 f32 : Float32Inputs()) {
|
2018-01-09 09:50:34 +00:00
|
|
|
Float32 res = Float32::FromBits(
|
|
|
|
|
reinterpret_cast<uint32_t>(f.Call(f32.get_bits(), 0, 0, 0, 0)));
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
Float32 exp = Float32::FromBits(f32.get_bits() ^ (1 << 31));
|
|
|
|
|
CHECK_EQ(exp.get_bits(), res.get_bits());
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
TEST(vneg_64) {
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2020-11-18 09:00:45 +00:00
|
|
|
auto f = AssembleCode<F_iiiii>(isolate, [](Assembler& assm) {
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
__ vmov(d0, r0, r1);
|
|
|
|
|
__ vneg(d0, d0);
|
|
|
|
|
__ vmov(r1, r0, d0);
|
2018-01-09 09:50:34 +00:00
|
|
|
});
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
|
|
|
|
|
for (Float64 f64 : Float64Inputs()) {
|
|
|
|
|
uint32_t p0 = static_cast<uint32_t>(f64.get_bits());
|
|
|
|
|
uint32_t p1 = static_cast<uint32_t>(f64.get_bits() >> 32);
|
2018-01-09 09:50:34 +00:00
|
|
|
uint32_t res = reinterpret_cast<uint32_t>(f.Call(p0, p1, 0, 0, 0));
|
Revert "Revert "[cctest] Clarify that tests for sync instructions are simulator specific""
This reverts commit 1feadfe81b7cc250df83cbf0b2fdfd17ac348cf8.
Reason for revert: Reland as bot stayed red after revert.
Original change's description:
> Revert "[cctest] Clarify that tests for sync instructions are simulator specific"
>
> This reverts commit 4013518fe3be92f0fbd043b09017f5eb8b8629d8.
>
> Reason for revert:
> https://build.chromium.org/p/client.v8.ports/builders/V8%20Linux%20-%20arm64%20-%20sim%20-%20gc%20stress
>
> Original change's description:
> > [cctest] Clarify that tests for sync instructions are simulator specific
> >
> > Some tests were recently added to test-simulator-arm.cc, however this file is
> > meant for tests that are specific to the simulator and therefore are not written
> > to work on hardware. While this sounds surprising, the reason is that our simulation
> > of synchronisation instructions is more conservative than on hardware.
> >
> > To make this more clear, this patch renames the "test-simulator-arm{,64}.cc"
> > files to "test-sync-primitives-arm{,64}.cc", and moves the vneg and vabs tests
> > into "test-assembler-arm.cc" which is were tests that are garanteed to work in
> > either native or simulated environments live.
> >
> > Finally, take the opportunity to share a little bit of code.
> >
> > Bug: v8:6963
> > Change-Id: Ifb85d3671c823b9bba73d09f419536b089a4e87c
> > Reviewed-on: https://chromium-review.googlesource.com/749387
> > Reviewed-by: Clemens Hammacher <clemensh@chromium.org>
> > Commit-Queue: Pierre Langlois <pierre.langlois@arm.com>
> > Cr-Commit-Position: refs/heads/master@{#49073}
>
> TBR=clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
>
> Change-Id: I1bfb4e9c7c18b716f417a84b18a14cb2e1fa3a7a
> No-Presubmit: true
> No-Tree-Checks: true
> No-Try: true
> Bug: v8:6963
> Reviewed-on: https://chromium-review.googlesource.com/750624
> Reviewed-by: Michael Achenbach <machenbach@chromium.org>
> Commit-Queue: Michael Achenbach <machenbach@chromium.org>
> Cr-Commit-Position: refs/heads/master@{#49074}
TBR=machenbach@chromium.org,clemensh@chromium.org,pierre.langlois@arm.com,bmeurer@chromium.org
Change-Id: I5af7bd3678758130534730a2f6f0b651b64c6956
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: v8:6963
Reviewed-on: https://chromium-review.googlesource.com/750903
Reviewed-by: Michael Achenbach <machenbach@chromium.org>
Commit-Queue: Michael Achenbach <machenbach@chromium.org>
Cr-Commit-Position: refs/heads/master@{#49075}
2017-11-02 13:11:29 +00:00
|
|
|
Float64 exp = Float64::FromBits(f64.get_bits() ^ (1ull << 63));
|
|
|
|
|
// We just get back the top word, so only compare that one.
|
|
|
|
|
CHECK_EQ(exp.get_bits() >> 32, res);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
2019-03-04 15:40:39 +00:00
|
|
|
TEST(move_pair) {
|
|
|
|
|
Isolate* isolate = CcTest::i_isolate();
|
|
|
|
|
HandleScope scope(isolate);
|
|
|
|
|
|
2020-11-18 09:00:45 +00:00
|
|
|
auto f = AssembleCode<F_piiii>(isolate, [](MacroAssembler& assm) {
|
2022-03-14 12:14:35 +00:00
|
|
|
RegList used_callee_saved = {r4, r5, r6, r7, r8};
|
2019-03-04 15:40:39 +00:00
|
|
|
__ stm(db_w, sp, used_callee_saved);
|
|
|
|
|
|
|
|
|
|
// Save output register bank pointer to r8.
|
|
|
|
|
__ mov(r8, r0);
|
|
|
|
|
|
|
|
|
|
__ mov(r0, Operand(0xabababab));
|
|
|
|
|
__ mov(r1, Operand(0xbabababa));
|
|
|
|
|
__ mov(r2, Operand(0x12341234));
|
|
|
|
|
__ mov(r3, Operand(0x43214321));
|
|
|
|
|
|
|
|
|
|
// No overlap:
|
|
|
|
|
// r4 <- r0
|
|
|
|
|
// r5 <- r1
|
|
|
|
|
__ MovePair(r4, r0, r5, r1);
|
|
|
|
|
|
|
|
|
|
// Overlap but we can swap moves:
|
|
|
|
|
// r2 <- r0
|
|
|
|
|
// r6 <- r2
|
|
|
|
|
__ MovePair(r2, r0, r6, r2);
|
|
|
|
|
|
|
|
|
|
// Overlap but can be done:
|
|
|
|
|
// r7 <- r3
|
|
|
|
|
// r3 <- r0
|
|
|
|
|
__ MovePair(r7, r3, r3, r0);
|
|
|
|
|
|
|
|
|
|
// Swap.
|
|
|
|
|
// r0 <- r1
|
|
|
|
|
// r1 <- r0
|
|
|
|
|
__ MovePair(r0, r1, r1, r0);
|
|
|
|
|
|
|
|
|
|
// Fill the fake register bank.
|
|
|
|
|
__ str(r0, MemOperand(r8, 0 * kPointerSize));
|
|
|
|
|
__ str(r1, MemOperand(r8, 1 * kPointerSize));
|
|
|
|
|
__ str(r2, MemOperand(r8, 2 * kPointerSize));
|
|
|
|
|
__ str(r3, MemOperand(r8, 3 * kPointerSize));
|
|
|
|
|
__ str(r4, MemOperand(r8, 4 * kPointerSize));
|
|
|
|
|
__ str(r5, MemOperand(r8, 5 * kPointerSize));
|
|
|
|
|
__ str(r6, MemOperand(r8, 6 * kPointerSize));
|
|
|
|
|
__ str(r7, MemOperand(r8, 7 * kPointerSize));
|
|
|
|
|
|
|
|
|
|
__ ldm(ia_w, sp, used_callee_saved);
|
|
|
|
|
});
|
|
|
|
|
|
|
|
|
|
// Create a fake register bank.
|
|
|
|
|
uint32_t r[] = {0, 0, 0, 0, 0, 0, 0, 0};
|
|
|
|
|
f.Call(r, 0, 0, 0, 0);
|
|
|
|
|
|
|
|
|
|
// r4 <- r0
|
|
|
|
|
// r5 <- r1
|
|
|
|
|
CHECK_EQ(0xabababab, r[4]);
|
|
|
|
|
CHECK_EQ(0xbabababa, r[5]);
|
|
|
|
|
|
|
|
|
|
// r2 <- r0
|
|
|
|
|
// r6 <- r2
|
|
|
|
|
CHECK_EQ(0xabababab, r[2]);
|
|
|
|
|
CHECK_EQ(0x12341234, r[6]);
|
|
|
|
|
|
|
|
|
|
// r7 <- r3
|
|
|
|
|
// r3 <- r0
|
|
|
|
|
CHECK_EQ(0x43214321, r[7]);
|
|
|
|
|
CHECK_EQ(0xabababab, r[3]);
|
|
|
|
|
|
|
|
|
|
// r0 and r1 should be swapped.
|
|
|
|
|
CHECK_EQ(0xbabababa, r[0]);
|
|
|
|
|
CHECK_EQ(0xabababab, r[1]);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2008-08-22 13:33:59 +00:00
|
|
|
#undef __
|
2017-08-31 12:34:55 +00:00
|
|
|
|
2017-10-02 21:19:21 +00:00
|
|
|
} // namespace test_assembler_arm
|
2017-08-31 12:34:55 +00:00
|
|
|
} // namespace internal
|
|
|
|
|
} // namespace v8
|
