AuroraMiddleware/v8
1
0
Fork 0
Code Issues 2 Pull Requests Releases Wiki Activity
6ad6dd1a7a
v8/test/unittests/wasm/trap-handler-x64-unittest.cc

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

472 lines
16 KiB
C++
Raw Normal View History

Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
// Copyright 2018 the V8 project authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
#include "include/v8config.h"
#if V8_OS_LINUX
#include <signal.h>
#include <ucontext.h>
#elif V8_OS_MACOSX
#include <signal.h>
#include <sys/ucontext.h>
#elif V8_OS_WIN
#include <windows.h>
#endif
#include "testing/gtest/include/gtest/gtest.h"
#if V8_OS_POSIX
#include "include/v8-wasm-trap-handler-posix.h"
#elif V8_OS_WIN
#include "include/v8-wasm-trap-handler-win.h"
#endif
#include "src/base/page-allocator.h"
Move code generation related files to src/codegen Bug: v8:9247 TBR=bmeurer@chromium.org,neis@chromium.org NOPRESUBMIT=true Change-Id: Ia1e49d1aac09c4ff9e05d58fab9d08dd71198878 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1621931 Reviewed-by: Yang Guo <yangguo@chromium.org> Reviewed-by: Benedikt Meurer <bmeurer@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Yang Guo <yangguo@chromium.org> Cr-Commit-Position: refs/heads/master@{#61682}
2019-05-21 09:30:15 +00:00
#include "src/codegen/assembler-inl.h"
#include "src/codegen/macro-assembler-inl.h"
Move relevant files to src/execution Bug: v8:9247 Change-Id: I79e0553e8a0d6dac2aa16b94a6c0e05b6ccde4a1 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1621934 Commit-Queue: Yang Guo <yangguo@chromium.org> Reviewed-by: Yang Guo <yangguo@chromium.org> Reviewed-by: Toon Verwaest <verwaest@chromium.org> Reviewed-by: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Benedikt Meurer <bmeurer@chromium.org> Cr-Commit-Position: refs/heads/master@{#61725}
2019-05-22 07:55:37 +00:00
#include "src/execution/simulator.h"
[arraybuffer] Rearchitect backing store ownership This CL completely rearchitects the ownership of array buffer backing stores, consolidating ownership into a {BackingStore} C++ object that is tracked throughout V8 using unique_ptr and shared_ptr where appropriate. Overall, lifetime management is simpler and more explicit. The numerous ways that array buffers were initialized have been streamlined to one Attach() method on JSArrayBuffer. The array buffer tracker in the GC implementation now manages std::shared_ptr<BackingStore> pointers, and the construction and destruction of the BackingStore object itself handles the underlying page or embedder-allocated memory. The embedder API remains unchanged for now. We use the v8::ArrayBuffer::Contents struct to hide an additional shared_ptr to keep the backing store alive properly, even in the case of aliases from live heap objects. Thus the embedder has a lower chance of making a mistake. Long-term, we should move the embedder to a model where they manage backing stores using shared_ptr to an opaque backing store object. R=mlippautz@chromium.org BUG=v8:9380,v8:9221 Change-Id: I48fae5ac85dcf6172a83f252439e77e7c1a16ccd Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1584323 Commit-Queue: Ben Titzer <titzer@chromium.org> Reviewed-by: Ben Titzer <titzer@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Reviewed-by: Yang Guo <yangguo@chromium.org> Reviewed-by: Deepti Gandluri <gdeepti@chromium.org> Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Lippautz <mlippautz@chromium.org> Cr-Commit-Position: refs/heads/master@{#62572}
2019-07-08 12:46:08 +00:00
#include "src/objects/backing-store.h"
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
#include "src/trap-handler/trap-handler.h"
Move utility code to src/utils NOPRESUBMIT=true TBR=mstarzinger@chromium.org Bug: v8:9247 Change-Id: I4cd6b79a1c2cba944f6f23caed59d4f1a4ee358b Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1624217 Commit-Queue: Yang Guo <yangguo@chromium.org> Reviewed-by: Igor Sheludko <ishell@chromium.org> Reviewed-by: Georg Neis <neis@chromium.org> Reviewed-by: Jaroslav Sevcik <jarin@chromium.org> Cr-Commit-Position: refs/heads/master@{#61790}
2019-05-23 13:27:57 +00:00
#include "src/utils/allocation.h"
#include "src/utils/vector.h"
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
#include "src/wasm/wasm-engine.h"
#include "test/common/assembler-tester.h"
#include "test/unittests/test-utils.h"
namespace v8 {
namespace internal {
namespace wasm {
namespace {
constexpr Register scratch = r10;
bool g_test_handler_executed = false;
#if V8_OS_LINUX || V8_OS_MACOSX
struct sigaction g_old_segv_action;
struct sigaction g_old_fpe_action;
struct sigaction g_old_bus_action; // We get SIGBUS on Mac sometimes.
#elif V8_OS_WIN
void* g_registered_handler = nullptr;
#endif
// The recovery address allows us to recover from an intentional crash.
Address g_recovery_address;
// Flag to indicate if the test handler should call the trap handler as a first
// chance handler.
bool g_use_as_first_chance_handler = false;
} // namespace
#define __ masm.
enum TrapHandlerStyle : int {
// The test uses the default trap handler of V8.
kDefault = 0,
// The test installs the trap handler callback in its own test handler.
kCallback = 1
};
std::string PrintTrapHandlerTestParam(
::testing::TestParamInfo<TrapHandlerStyle> info) {
switch (info.param) {
case kDefault:
return "DefaultTrapHandler";
case kCallback:
return "Callback";
}
UNREACHABLE();
}
class TrapHandlerTest : public TestWithIsolate,
public ::testing::WithParamInterface<TrapHandlerStyle> {
protected:
void SetUp() override {
[arraybuffer] Rearchitect backing store ownership This CL completely rearchitects the ownership of array buffer backing stores, consolidating ownership into a {BackingStore} C++ object that is tracked throughout V8 using unique_ptr and shared_ptr where appropriate. Overall, lifetime management is simpler and more explicit. The numerous ways that array buffers were initialized have been streamlined to one Attach() method on JSArrayBuffer. The array buffer tracker in the GC implementation now manages std::shared_ptr<BackingStore> pointers, and the construction and destruction of the BackingStore object itself handles the underlying page or embedder-allocated memory. The embedder API remains unchanged for now. We use the v8::ArrayBuffer::Contents struct to hide an additional shared_ptr to keep the backing store alive properly, even in the case of aliases from live heap objects. Thus the embedder has a lower chance of making a mistake. Long-term, we should move the embedder to a model where they manage backing stores using shared_ptr to an opaque backing store object. R=mlippautz@chromium.org BUG=v8:9380,v8:9221 Change-Id: I48fae5ac85dcf6172a83f252439e77e7c1a16ccd Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1584323 Commit-Queue: Ben Titzer <titzer@chromium.org> Reviewed-by: Ben Titzer <titzer@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Reviewed-by: Yang Guo <yangguo@chromium.org> Reviewed-by: Deepti Gandluri <gdeepti@chromium.org> Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Lippautz <mlippautz@chromium.org> Cr-Commit-Position: refs/heads/master@{#62572}
2019-07-08 12:46:08 +00:00
backing_store_ = BackingStore::AllocateWasmMemory(i_isolate(), 1, 1,
SharedFlag::kNotShared);
CHECK(backing_store_);
CHECK(backing_store_->has_guard_regions());
// The allocated backing store ends with a guard page.
crash_address_ = reinterpret_cast<Address>(backing_store_->buffer_start()) +
backing_store_->byte_length() + 32;
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
// Allocate a buffer for the generated code.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
buffer_ = AllocateAssemblerBuffer(AssemblerBase::kMinimalBufferSize,
GetRandomMmapAddr());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
InitRecoveryCode();
#if V8_OS_LINUX || V8_OS_MACOSX
// Set up a signal handler to recover from the expected crash.
struct sigaction action;
action.sa_sigaction = SignalHandler;
sigemptyset(&action.sa_mask);
action.sa_flags = SA_SIGINFO;
// SIGSEGV happens for wasm oob memory accesses on Linux.
CHECK_EQ(0, sigaction(SIGSEGV, &action, &g_old_segv_action));
// SIGBUS happens for wasm oob memory accesses on macOS.
CHECK_EQ(0, sigaction(SIGBUS, &action, &g_old_bus_action));
// SIGFPE to simulate crashes which are not handled by the trap handler.
CHECK_EQ(0, sigaction(SIGFPE, &action, &g_old_fpe_action));
#elif V8_OS_WIN
g_registered_handler =
AddVectoredExceptionHandler(/*first=*/0, TestHandler);
#endif
}
void TearDown() override {
// We should always have left wasm code.
CHECK(!GetThreadInWasmFlag());
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
buffer_.reset();
recovery_buffer_.reset();
[arraybuffer] Rearchitect backing store ownership This CL completely rearchitects the ownership of array buffer backing stores, consolidating ownership into a {BackingStore} C++ object that is tracked throughout V8 using unique_ptr and shared_ptr where appropriate. Overall, lifetime management is simpler and more explicit. The numerous ways that array buffers were initialized have been streamlined to one Attach() method on JSArrayBuffer. The array buffer tracker in the GC implementation now manages std::shared_ptr<BackingStore> pointers, and the construction and destruction of the BackingStore object itself handles the underlying page or embedder-allocated memory. The embedder API remains unchanged for now. We use the v8::ArrayBuffer::Contents struct to hide an additional shared_ptr to keep the backing store alive properly, even in the case of aliases from live heap objects. Thus the embedder has a lower chance of making a mistake. Long-term, we should move the embedder to a model where they manage backing stores using shared_ptr to an opaque backing store object. R=mlippautz@chromium.org BUG=v8:9380,v8:9221 Change-Id: I48fae5ac85dcf6172a83f252439e77e7c1a16ccd Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1584323 Commit-Queue: Ben Titzer <titzer@chromium.org> Reviewed-by: Ben Titzer <titzer@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Reviewed-by: Yang Guo <yangguo@chromium.org> Reviewed-by: Deepti Gandluri <gdeepti@chromium.org> Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Lippautz <mlippautz@chromium.org> Cr-Commit-Position: refs/heads/master@{#62572}
2019-07-08 12:46:08 +00:00
backing_store_.reset();
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
// Clean up the trap handler
trap_handler::RemoveTrapHandler();
if (!g_test_handler_executed) {
#if V8_OS_LINUX || V8_OS_MACOSX
// The test handler cleans up the signal handler setup in the test. If the
// test handler was not called, we have to do the cleanup ourselves.
CHECK_EQ(0, sigaction(SIGSEGV, &g_old_segv_action, nullptr));
CHECK_EQ(0, sigaction(SIGFPE, &g_old_fpe_action, nullptr));
CHECK_EQ(0, sigaction(SIGBUS, &g_old_bus_action, nullptr));
#elif V8_OS_WIN
RemoveVectoredExceptionHandler(g_registered_handler);
g_registered_handler = nullptr;
#endif
}
}
void InitRecoveryCode() {
// Create a code snippet where we can jump to to recover from a signal or
// exception. The code snippet only consists of a return statement.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
recovery_buffer_ = AllocateAssemblerBuffer(
AssemblerBase::kMinimalBufferSize, GetRandomMmapAddr());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
MacroAssembler masm(nullptr, AssemblerOptions{}, CodeObjectRequired::kNo,
recovery_buffer_->CreateView());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
int recovery_offset = __ pc_offset();
__ Pop(scratch);
__ Ret();
CodeDesc desc;
masm.GetCode(nullptr, &desc);
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
recovery_buffer_->MakeExecutable();
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
g_recovery_address =
reinterpret_cast<Address>(desc.buffer + recovery_offset);
}
#if V8_OS_LINUX || V8_OS_MACOSX
static void SignalHandler(int signal, siginfo_t* info, void* context) {
if (g_use_as_first_chance_handler) {
if (v8::TryHandleWebAssemblyTrapPosix(signal, info, context)) {
return;
}
}
// Reset the signal handler, to avoid that this signal handler is called
// repeatedly.
sigaction(SIGSEGV, &g_old_segv_action, nullptr);
sigaction(SIGFPE, &g_old_fpe_action, nullptr);
sigaction(SIGBUS, &g_old_bus_action, nullptr);
g_test_handler_executed = true;
// Set the $rip to the recovery code.
ucontext_t* uc = reinterpret_cast<ucontext_t*>(context);
#if V8_OS_LINUX
uc->uc_mcontext.gregs[REG_RIP] = g_recovery_address;
#else // V8_OS_MACOSX
uc->uc_mcontext->__ss.__rip = g_recovery_address;
#endif
}
#endif
#if V8_OS_WIN
static LONG WINAPI TestHandler(EXCEPTION_POINTERS* exception) {
if (g_use_as_first_chance_handler) {
if (v8::TryHandleWebAssemblyTrapWindows(exception)) {
return EXCEPTION_CONTINUE_EXECUTION;
}
}
RemoveVectoredExceptionHandler(g_registered_handler);
g_registered_handler = nullptr;
g_test_handler_executed = true;
exception->ContextRecord->Rip = g_recovery_address;
return EXCEPTION_CONTINUE_EXECUTION;
}
#endif
public:
void SetupTrapHandler(TrapHandlerStyle style) {
bool use_default_handler = style == kDefault;
g_use_as_first_chance_handler = !use_default_handler;
CHECK(v8::V8::EnableWebAssemblyTrapHandler(use_default_handler));
}
void GenerateSetThreadInWasmFlagCode(MacroAssembler* masm) {
masm->Move(scratch,
[wasm] Load thread-in-wasm flag from the isolate The existing implementation embedded an isolate-specific pointer to the thread-in-wasm flag in the wrapper code. However, when the module code is shared among multiple workers, this can mean that the workers share the same thread-in-wasm flag. With this change we load the pointer to the flag at runtime from the current isolate. Thereby the correct flag is used even when the same code is executed on different workers. Note that we could access the right flag address by going through the root register. However, changing the code generation to use the root register requires some inconvenient steps: * Pass the isolate to the pipeline again, which we don't want. * Change the WasmCallDescriptor to allow the use of the root register for wrappers but not for other code. To avoid these issues, and allow the CL to be easy to merge back, we got for the changes proposed here. R=mstarzinger@chromium.org, ishell@chromium.org Bug: v8:8533 Change-Id: If15565a7ad7cba835cfc1628e7a4d3fdef90a5c0 Reviewed-on: https://chromium-review.googlesource.com/c/1358518 Commit-Queue: Andreas Haas <ahaas@chromium.org> Reviewed-by: Igor Sheludko <ishell@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58044}
2018-12-05 13:52:50 +00:00
i_isolate()->thread_local_top()->thread_in_wasm_flag_address_,
RelocInfo::NONE);
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
masm->movl(MemOperand(scratch, 0), Immediate(1));
}
void GenerateResetThreadInWasmFlagCode(MacroAssembler* masm) {
masm->Move(scratch,
[wasm] Load thread-in-wasm flag from the isolate The existing implementation embedded an isolate-specific pointer to the thread-in-wasm flag in the wrapper code. However, when the module code is shared among multiple workers, this can mean that the workers share the same thread-in-wasm flag. With this change we load the pointer to the flag at runtime from the current isolate. Thereby the correct flag is used even when the same code is executed on different workers. Note that we could access the right flag address by going through the root register. However, changing the code generation to use the root register requires some inconvenient steps: * Pass the isolate to the pipeline again, which we don't want. * Change the WasmCallDescriptor to allow the use of the root register for wrappers but not for other code. To avoid these issues, and allow the CL to be easy to merge back, we got for the changes proposed here. R=mstarzinger@chromium.org, ishell@chromium.org Bug: v8:8533 Change-Id: If15565a7ad7cba835cfc1628e7a4d3fdef90a5c0 Reviewed-on: https://chromium-review.googlesource.com/c/1358518 Commit-Queue: Andreas Haas <ahaas@chromium.org> Reviewed-by: Igor Sheludko <ishell@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58044}
2018-12-05 13:52:50 +00:00
i_isolate()->thread_local_top()->thread_in_wasm_flag_address_,
RelocInfo::NONE);
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
masm->movl(MemOperand(scratch, 0), Immediate(0));
}
bool GetThreadInWasmFlag() {
return *reinterpret_cast<int*>(
trap_handler::GetThreadInWasmThreadLocalAddress());
}
// Execute the code in buffer.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
void ExecuteBuffer() {
buffer_->MakeExecutable();
GeneratedCode<void>::FromAddress(
i_isolate(), reinterpret_cast<Address>(buffer_->start()))
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
.Call();
[wasm] Load thread-in-wasm flag from the isolate The existing implementation embedded an isolate-specific pointer to the thread-in-wasm flag in the wrapper code. However, when the module code is shared among multiple workers, this can mean that the workers share the same thread-in-wasm flag. With this change we load the pointer to the flag at runtime from the current isolate. Thereby the correct flag is used even when the same code is executed on different workers. Note that we could access the right flag address by going through the root register. However, changing the code generation to use the root register requires some inconvenient steps: * Pass the isolate to the pipeline again, which we don't want. * Change the WasmCallDescriptor to allow the use of the root register for wrappers but not for other code. To avoid these issues, and allow the CL to be easy to merge back, we got for the changes proposed here. R=mstarzinger@chromium.org, ishell@chromium.org Bug: v8:8533 Change-Id: If15565a7ad7cba835cfc1628e7a4d3fdef90a5c0 Reviewed-on: https://chromium-review.googlesource.com/c/1358518 Commit-Queue: Andreas Haas <ahaas@chromium.org> Reviewed-by: Igor Sheludko <ishell@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58044}
2018-12-05 13:52:50 +00:00
CHECK(!g_test_handler_executed);
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
}
// Execute the code in buffer. We expect a crash which we recover from in the
// test handler.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
void ExecuteExpectCrash(TestingAssemblerBuffer* buffer,
bool check_wasm_flag = true) {
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
CHECK(!g_test_handler_executed);
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
buffer->MakeExecutable();
[wasm] Load thread-in-wasm flag from the isolate The existing implementation embedded an isolate-specific pointer to the thread-in-wasm flag in the wrapper code. However, when the module code is shared among multiple workers, this can mean that the workers share the same thread-in-wasm flag. With this change we load the pointer to the flag at runtime from the current isolate. Thereby the correct flag is used even when the same code is executed on different workers. Note that we could access the right flag address by going through the root register. However, changing the code generation to use the root register requires some inconvenient steps: * Pass the isolate to the pipeline again, which we don't want. * Change the WasmCallDescriptor to allow the use of the root register for wrappers but not for other code. To avoid these issues, and allow the CL to be easy to merge back, we got for the changes proposed here. R=mstarzinger@chromium.org, ishell@chromium.org Bug: v8:8533 Change-Id: If15565a7ad7cba835cfc1628e7a4d3fdef90a5c0 Reviewed-on: https://chromium-review.googlesource.com/c/1358518 Commit-Queue: Andreas Haas <ahaas@chromium.org> Reviewed-by: Igor Sheludko <ishell@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58044}
2018-12-05 13:52:50 +00:00
GeneratedCode<void>::FromAddress(i_isolate(),
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
reinterpret_cast<Address>(buffer->start()))
[wasm] Load thread-in-wasm flag from the isolate The existing implementation embedded an isolate-specific pointer to the thread-in-wasm flag in the wrapper code. However, when the module code is shared among multiple workers, this can mean that the workers share the same thread-in-wasm flag. With this change we load the pointer to the flag at runtime from the current isolate. Thereby the correct flag is used even when the same code is executed on different workers. Note that we could access the right flag address by going through the root register. However, changing the code generation to use the root register requires some inconvenient steps: * Pass the isolate to the pipeline again, which we don't want. * Change the WasmCallDescriptor to allow the use of the root register for wrappers but not for other code. To avoid these issues, and allow the CL to be easy to merge back, we got for the changes proposed here. R=mstarzinger@chromium.org, ishell@chromium.org Bug: v8:8533 Change-Id: If15565a7ad7cba835cfc1628e7a4d3fdef90a5c0 Reviewed-on: https://chromium-review.googlesource.com/c/1358518 Commit-Queue: Andreas Haas <ahaas@chromium.org> Reviewed-by: Igor Sheludko <ishell@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58044}
2018-12-05 13:52:50 +00:00
.Call();
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
CHECK(g_test_handler_executed);
g_test_handler_executed = false;
if (check_wasm_flag) CHECK(!GetThreadInWasmFlag());
}
bool test_handler_executed() { return g_test_handler_executed; }
[arraybuffer] Rearchitect backing store ownership This CL completely rearchitects the ownership of array buffer backing stores, consolidating ownership into a {BackingStore} C++ object that is tracked throughout V8 using unique_ptr and shared_ptr where appropriate. Overall, lifetime management is simpler and more explicit. The numerous ways that array buffers were initialized have been streamlined to one Attach() method on JSArrayBuffer. The array buffer tracker in the GC implementation now manages std::shared_ptr<BackingStore> pointers, and the construction and destruction of the BackingStore object itself handles the underlying page or embedder-allocated memory. The embedder API remains unchanged for now. We use the v8::ArrayBuffer::Contents struct to hide an additional shared_ptr to keep the backing store alive properly, even in the case of aliases from live heap objects. Thus the embedder has a lower chance of making a mistake. Long-term, we should move the embedder to a model where they manage backing stores using shared_ptr to an opaque backing store object. R=mlippautz@chromium.org BUG=v8:9380,v8:9221 Change-Id: I48fae5ac85dcf6172a83f252439e77e7c1a16ccd Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1584323 Commit-Queue: Ben Titzer <titzer@chromium.org> Reviewed-by: Ben Titzer <titzer@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Reviewed-by: Yang Guo <yangguo@chromium.org> Reviewed-by: Deepti Gandluri <gdeepti@chromium.org> Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Lippautz <mlippautz@chromium.org> Cr-Commit-Position: refs/heads/master@{#62572}
2019-07-08 12:46:08 +00:00
// The backing store used for testing the trap handler.
std::unique_ptr<BackingStore> backing_store_;
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
// Address within the guard region of the wasm memory. Accessing this memory
// address causes a signal or exception.
Address crash_address_;
// Buffer for generated code.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
std::unique_ptr<TestingAssemblerBuffer> buffer_;
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
// Buffer for the code for the landing pad of the test handler.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
std::unique_ptr<TestingAssemblerBuffer> recovery_buffer_;
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
};
TEST_P(TrapHandlerTest, TestTrapHandlerRecovery) {
// Test that the wasm trap handler can recover a memory access violation in
// wasm code (we fake the wasm code and the access violation).
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
MacroAssembler masm(nullptr, AssemblerOptions{}, CodeObjectRequired::kNo,
buffer_->CreateView());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
__ Push(scratch);
GenerateSetThreadInWasmFlagCode(&masm);
__ Move(scratch, crash_address_, RelocInfo::NONE);
int crash_offset = __ pc_offset();
__ testl(MemOperand(scratch, 0), Immediate(1));
int recovery_offset = __ pc_offset();
GenerateResetThreadInWasmFlagCode(&masm);
__ Pop(scratch);
__ Ret();
CodeDesc desc;
masm.GetCode(nullptr, &desc);
SetupTrapHandler(GetParam());
trap_handler::ProtectedInstructionData protected_instruction{crash_offset,
recovery_offset};
trap_handler::RegisterHandlerData(reinterpret_cast<Address>(desc.buffer),
desc.instr_size, 1, &protected_instruction);
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
ExecuteBuffer();
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
}
TEST_P(TrapHandlerTest, TestReleaseHandlerData) {
// Test that after we release handler data in the trap handler, it cannot
// recover from the specific memory access violation anymore.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
MacroAssembler masm(nullptr, AssemblerOptions{}, CodeObjectRequired::kNo,
buffer_->CreateView());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
__ Push(scratch);
GenerateSetThreadInWasmFlagCode(&masm);
__ Move(scratch, crash_address_, RelocInfo::NONE);
int crash_offset = __ pc_offset();
__ testl(MemOperand(scratch, 0), Immediate(1));
int recovery_offset = __ pc_offset();
GenerateResetThreadInWasmFlagCode(&masm);
__ Pop(scratch);
__ Ret();
CodeDesc desc;
masm.GetCode(nullptr, &desc);
trap_handler::ProtectedInstructionData protected_instruction{crash_offset,
recovery_offset};
int handler_id = trap_handler::RegisterHandlerData(
reinterpret_cast<Address>(desc.buffer), desc.instr_size, 1,
&protected_instruction);
SetupTrapHandler(GetParam());
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
ExecuteBuffer();
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
// Deregister from the trap handler. The trap handler should not do the
// recovery now.
trap_handler::ReleaseHandlerData(handler_id);
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
ExecuteExpectCrash(buffer_.get());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
}
TEST_P(TrapHandlerTest, TestNoThreadInWasmFlag) {
// That that if the thread_in_wasm flag is not set, the trap handler does not
// get active.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
MacroAssembler masm(nullptr, AssemblerOptions{}, CodeObjectRequired::kNo,
buffer_->CreateView());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
__ Push(scratch);
__ Move(scratch, crash_address_, RelocInfo::NONE);
int crash_offset = __ pc_offset();
__ testl(MemOperand(scratch, 0), Immediate(1));
int recovery_offset = __ pc_offset();
__ Pop(scratch);
__ Ret();
CodeDesc desc;
masm.GetCode(nullptr, &desc);
trap_handler::ProtectedInstructionData protected_instruction{crash_offset,
recovery_offset};
trap_handler::RegisterHandlerData(reinterpret_cast<Address>(desc.buffer),
desc.instr_size, 1, &protected_instruction);
SetupTrapHandler(GetParam());
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
ExecuteExpectCrash(buffer_.get());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
}
TEST_P(TrapHandlerTest, TestCrashInWasmNoProtectedInstruction) {
// Test that if the crash in wasm happened at an instruction which is not
// protected, then the trap handler does not handle it.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
MacroAssembler masm(nullptr, AssemblerOptions{}, CodeObjectRequired::kNo,
buffer_->CreateView());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
__ Push(scratch);
GenerateSetThreadInWasmFlagCode(&masm);
int no_crash_offset = __ pc_offset();
__ Move(scratch, crash_address_, RelocInfo::NONE);
__ testl(MemOperand(scratch, 0), Immediate(1));
// Offset where the crash is not happening.
int recovery_offset = __ pc_offset();
GenerateResetThreadInWasmFlagCode(&masm);
__ Pop(scratch);
__ Ret();
CodeDesc desc;
masm.GetCode(nullptr, &desc);
trap_handler::ProtectedInstructionData protected_instruction{no_crash_offset,
recovery_offset};
trap_handler::RegisterHandlerData(reinterpret_cast<Address>(desc.buffer),
desc.instr_size, 1, &protected_instruction);
SetupTrapHandler(GetParam());
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
ExecuteExpectCrash(buffer_.get());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
}
TEST_P(TrapHandlerTest, TestCrashInWasmWrongCrashType) {
// Test that if the crash reason is not a memory access violation, then the
// wasm trap handler does not handle it.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
MacroAssembler masm(nullptr, AssemblerOptions{}, CodeObjectRequired::kNo,
buffer_->CreateView());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
__ Push(scratch);
GenerateSetThreadInWasmFlagCode(&masm);
__ xorq(scratch, scratch);
int crash_offset = __ pc_offset();
__ divq(scratch);
// Offset where the crash is not happening.
int recovery_offset = __ pc_offset();
GenerateResetThreadInWasmFlagCode(&masm);
__ Pop(scratch);
__ Ret();
CodeDesc desc;
masm.GetCode(nullptr, &desc);
trap_handler::ProtectedInstructionData protected_instruction{crash_offset,
recovery_offset};
trap_handler::RegisterHandlerData(reinterpret_cast<Address>(desc.buffer),
desc.instr_size, 1, &protected_instruction);
SetupTrapHandler(GetParam());
#if V8_OS_POSIX
Fix Wasm trap handler recursion on exceptions raised early Check if storage for thread_local variables has been allocated before attempting to access such variables, as exceptions may be raised in the thread before this initializion is complete, causing an infinite loop. Bug: v8:8966 Change-Id: Ifc6223b74999a55bfd0ed2d6ebf054bbffd7e809 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1507714 Commit-Queue: Ben Titzer <titzer@chromium.org> Reviewed-by: Ben Titzer <titzer@chromium.org> Cr-Commit-Position: refs/heads/master@{#60852}
2019-03-22 19:23:09 +00:00
// On Posix, the V8 default trap handler does not register for SIGFPE,
// therefore the thread-in-wasm flag is never reset in this test. We
// therefore do not check the value of this flag.
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
bool check_wasm_flag = GetParam() != kDefault;
Fix Wasm trap handler recursion on exceptions raised early Check if storage for thread_local variables has been allocated before attempting to access such variables, as exceptions may be raised in the thread before this initializion is complete, causing an infinite loop. Bug: v8:8966 Change-Id: Ifc6223b74999a55bfd0ed2d6ebf054bbffd7e809 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1507714 Commit-Queue: Ben Titzer <titzer@chromium.org> Reviewed-by: Ben Titzer <titzer@chromium.org> Cr-Commit-Position: refs/heads/master@{#60852}
2019-03-22 19:23:09 +00:00
#elif V8_OS_WIN
// On Windows, the trap handler returns immediately if not an exception of
// interest.
bool check_wasm_flag = false;
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
#else
bool check_wasm_flag = true;
#endif
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
ExecuteExpectCrash(buffer_.get(), check_wasm_flag);
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
if (!check_wasm_flag) {
// Reset the thread-in-wasm flag because it was probably not reset in the
// trap handler.
*trap_handler::GetThreadInWasmThreadLocalAddress() = 0;
}
}
class CodeRunner : public v8::base::Thread {
public:
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
CodeRunner(TrapHandlerTest* test, TestingAssemblerBuffer* buffer)
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
: Thread(Options("CodeRunner")), test_(test), buffer_(buffer) {}
void Run() override { test_->ExecuteExpectCrash(buffer_); }
private:
TrapHandlerTest* test_;
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
TestingAssemblerBuffer* buffer_;
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
};
TEST_P(TrapHandlerTest, TestCrashInOtherThread) {
// Test setup:
// The current thread enters wasm land (sets the thread_in_wasm flag)
// A second thread crashes at a protected instruction without having the flag
// set.
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
MacroAssembler masm(nullptr, AssemblerOptions{}, CodeObjectRequired::kNo,
buffer_->CreateView());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
__ Push(scratch);
__ Move(scratch, crash_address_, RelocInfo::NONE);
int crash_offset = __ pc_offset();
__ testl(MemOperand(scratch, 0), Immediate(1));
int recovery_offset = __ pc_offset();
__ Pop(scratch);
__ Ret();
CodeDesc desc;
masm.GetCode(nullptr, &desc);
trap_handler::ProtectedInstructionData protected_instruction{crash_offset,
recovery_offset};
trap_handler::RegisterHandlerData(reinterpret_cast<Address>(desc.buffer),
desc.instr_size, 1, &protected_instruction);
SetupTrapHandler(GetParam());
[test] Refactor AllocateAssemblerBuffer Refactor the AllocateAssemblerBuffer helper for the new Assembler API. This is the only non-mechanical part, all other callsites that create Assembler instances can be trivially changed to the new API. This will be done in a separate CL. R=mstarzinger@chromium.org Bug: v8:8689, v8:8562 Change-Id: I6c150748eeea778d9b70f41fd66fbb1221035a1b Reviewed-on: https://chromium-review.googlesource.com/c/1415490 Commit-Queue: Clemens Hammacher <clemensh@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Cr-Commit-Position: refs/heads/master@{#58881}
2019-01-17 10:23:16 +00:00
CodeRunner runner(this, buffer_.get());
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
CHECK(!GetThreadInWasmFlag());
// Set the thread-in-wasm flag manually in this thread.
*trap_handler::GetThreadInWasmThreadLocalAddress() = 1;
runner.Start();
runner.Join();
CHECK(GetThreadInWasmFlag());
// Reset the thread-in-wasm flag.
*trap_handler::GetThreadInWasmThreadLocalAddress() = 0;
}
test: Replace _TEST_CASE_ with _TEST_SUITE_. Googletest is (at last) converging with industry-standard terminology [1]. We previously called test suites "test cases", which was rather confusing for folks coming from any other testing framework. Chrome now has a googletest version that supports _TEST_SUITE_ macros instead of _TEST_CASE_, so this CL cleans up some of the outdated usage. [1] https://github.com/google/googletest/blob/master/googletest/docs/primer.md#beware-of-the-nomenclature Bug: chromium:925652 Change-Id: I3cd02b9fa6dbece1594bbfd50a21ad7503c2aab9 Reviewed-on: https://chromium-review.googlesource.com/c/1475654 Reviewed-by: Jaroslav Sevcik <jarin@chromium.org> Commit-Queue: Victor Costan <pwnall@chromium.org> Cr-Commit-Position: refs/heads/master@{#59666}
2019-02-15 16:53:29 +00:00
INSTANTIATE_TEST_SUITE_P(/* no prefix */, TrapHandlerTest,
::testing::Values(kDefault, kCallback),
PrintTrapHandlerTestParam);
Reland: [wasm] Add more unit tests for trap handler The problem were missing V8_EXPORT_PRIVATE and V8_EXPORT. The unittests test if the trap handler only handles those traps it is supposed to handle: * Only handle traps when the thread-in-wasm flag is set. * Only handle traps of the right type, i.e. memory access violations. * Only handle traps at recorded instructions. The tests also test the consistency of the thread-in-wasm flag. I made one change in the trap handler where that consistency could be violated. All tests are executed with the default trap handler provided by V8, and with the trap handler callback installed in a test signal/exception handler. Patchset 1 is the original CL. R=mstarzinger@chromium.org Change-Id: I172d94f24cdba4c3a1f7f344825b059dbb59da79 Reviewed-on: https://chromium-review.googlesource.com/c/1351024 Reviewed-by: Ulan Degenbaev <ulan@chromium.org> Reviewed-by: Michael Starzinger <mstarzinger@chromium.org> Commit-Queue: Andreas Haas <ahaas@chromium.org> Cr-Commit-Position: refs/heads/master@{#57947}
2018-11-27 13:58:18 +00:00
#undef __
} // namespace wasm
} // namespace internal
} // namespace v8
Reference in New Issue Copy Permalink