v8/test/cctest/test-debug-helper.cc

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

414 lines
18 KiB
C++
Raw Normal View History

Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +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 "src/api/api-inl.h"
#include "src/flags/flags.h"
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
#include "src/heap/spaces.h"
#include "test/cctest/cctest.h"
#include "tools/debug_helper/debug-helper.h"
namespace v8 {
namespace internal {
namespace {
namespace d = v8::debug_helper;
uintptr_t memory_fail_start = 0;
uintptr_t memory_fail_end = 0;
class MemoryFailureRegion {
public:
MemoryFailureRegion(uintptr_t start, uintptr_t end) {
memory_fail_start = start;
memory_fail_end = end;
}
~MemoryFailureRegion() {
memory_fail_start = 0;
memory_fail_end = 0;
}
};
// Implement the memory-reading callback. This one just fetches memory from the
// current process, but a real implementation for a debugging extension would
// fetch memory from the debuggee process or crash dump.
d::MemoryAccessResult ReadMemory(uintptr_t address, uint8_t* destination,
size_t byte_count) {
if (address >= memory_fail_start && address <= memory_fail_end) {
// Simulate failure to read debuggee memory.
return d::MemoryAccessResult::kAddressValidButInaccessible;
}
memcpy(destination, reinterpret_cast<void*>(address), byte_count);
return d::MemoryAccessResult::kOk;
}
void CheckPropBase(const d::PropertyBase& property, const char* expected_type,
const char* expected_name) {
CHECK(property.type == std::string("v8::internal::TaggedValue") ||
property.type == std::string(expected_type));
CHECK(property.decompressed_type == std::string(expected_type));
CHECK(property.name == std::string(expected_name));
}
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
void CheckProp(const d::ObjectProperty& property, const char* expected_type,
const char* expected_name,
d::PropertyKind expected_kind = d::PropertyKind::kSingle,
size_t expected_num_values = 1) {
CheckPropBase(property, expected_type, expected_name);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CHECK_EQ(property.num_values, expected_num_values);
CHECK(property.kind == expected_kind);
}
template <typename TValue>
void CheckProp(const d::ObjectProperty& property, const char* expected_type,
const char* expected_name, TValue expected_value) {
CheckProp(property, expected_type, expected_name);
CHECK(*reinterpret_cast<TValue*>(property.address) == expected_value);
}
Fix brittleness of GetObjectProperties test Part of the GetObjectProperties test case is for verifying the human- readable brief object description string that GetObjectProperties returns. That string might look something like this: "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) GetObjectProperties also tries to detect known immortal objects by recognizing their addresses, which is useful in crash dumps with limited memory. The recognized object name, if it exists, is prepended to the description string. In order to provide this data accurately (in builds without pointer compression), GetObjectProperties relies on the caller to provide the addresses of the first pages in read-only space, map space, and old space. If the caller doesn't provide those addresses, then GetObjectProperties does the best it can with limited information and reports possible matches based on an object's offset within the heap page that contains it. So the result string might look like this, if the object happened to get allocated at a lucky offset within its page: maybe LoadHandler3Map "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) As a result, when testing these descriptions, we should generally check that they contain the interesting data rather than that they start with it, because some incorrect "maybe" match with a known object might be included at the beginning. Bug: v8:10034 Change-Id: I0cf5afd67793a239614aba3665ef57cd2d663a47 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1950233 Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#65432}
2019-12-09 23:54:54 +00:00
bool StartsWith(const std::string& full_string, const std::string& prefix) {
return full_string.substr(0, prefix.size()) == prefix;
}
Fix brittleness of GetObjectProperties test Part of the GetObjectProperties test case is for verifying the human- readable brief object description string that GetObjectProperties returns. That string might look something like this: "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) GetObjectProperties also tries to detect known immortal objects by recognizing their addresses, which is useful in crash dumps with limited memory. The recognized object name, if it exists, is prepended to the description string. In order to provide this data accurately (in builds without pointer compression), GetObjectProperties relies on the caller to provide the addresses of the first pages in read-only space, map space, and old space. If the caller doesn't provide those addresses, then GetObjectProperties does the best it can with limited information and reports possible matches based on an object's offset within the heap page that contains it. So the result string might look like this, if the object happened to get allocated at a lucky offset within its page: maybe LoadHandler3Map "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) As a result, when testing these descriptions, we should generally check that they contain the interesting data rather than that they start with it, because some incorrect "maybe" match with a known object might be included at the beginning. Bug: v8:10034 Change-Id: I0cf5afd67793a239614aba3665ef57cd2d663a47 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1950233 Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#65432}
2019-12-09 23:54:54 +00:00
bool Contains(const std::string& full_string, const std::string& substr) {
return full_string.find(substr) != std::string::npos;
}
void CheckStructProp(const d::StructProperty& property,
const char* expected_type, const char* expected_name,
size_t expected_offset, uint8_t expected_num_bits = 0,
uint8_t expected_shift_bits = 0) {
CheckPropBase(property, expected_type, expected_name);
CHECK_EQ(property.offset, expected_offset);
CHECK_EQ(property.num_bits, expected_num_bits);
CHECK_EQ(property.shift_bits, expected_shift_bits);
}
const d::ObjectProperty& FindProp(const d::ObjectPropertiesResult& props,
std::string name) {
for (size_t i = 0; i < props.num_properties; ++i) {
if (name == props.properties[i]->name) {
return *props.properties[i];
}
}
CHECK_WITH_MSG(false, ("property '" + name + "' not found").c_str());
UNREACHABLE();
}
template <typename TValue>
TValue ReadProp(const d::ObjectPropertiesResult& props, std::string name) {
const d::ObjectProperty& prop = FindProp(props, name);
return *reinterpret_cast<TValue*>(prop.address);
}
// A simple implementation of ExternalStringResource that lets us control the
// result of IsCacheable().
class StringResource : public v8::String::ExternalStringResource {
public:
explicit StringResource(bool cacheable) : cacheable_(cacheable) {}
const uint16_t* data() const override {
return reinterpret_cast<const uint16_t*>(u"abcde");
}
size_t length() const override { return 5; }
bool IsCacheable() const override { return cacheable_; }
private:
bool cacheable_;
};
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
} // namespace
TEST(GetObjectProperties) {
CcTest::InitializeVM();
v8::Isolate* isolate = CcTest::isolate();
Reland "[torque] Support bitfield structs stored within Smis" This reverts commit 4dc1fb4e047cc36cef1308884812c76a66830724. Reason for revert: the regression from the original change was likely due to unlucky factors like code alignment. Original change's description: > Revert "[torque] Support bitfield structs stored within Smis" > > This reverts commit e5e4ea962e48bf138c640b5d6cc1ea9c358cdf86. > > Reason for revert: mysterious performance regression chromium:1052756 > > Original change's description: > > [torque] Support bitfield structs stored within Smis > > > > This change moves the definition of the bits stored in DebugInfo::flags > > to Torque, and updates the only Torque usage of that field to use more > > natural syntax. This is intended as an example of common patterns found > > in various other classes. Several supporting changes are required: > > > > 1. Add a new type representing a bitfield struct stored within a Smi. It > > is currently called SmiTagged, but I'm open to suggestions. > > 2. Add an enum-style output for Torque bitfield structs whose bitfields > > occupy only one bit each. > > 3. Add a new case to MachineOperatorReducer that makes the generated > > code for IncBlockCounter match with what was generated before this > > change. > > 4. Add support for reporting these bitfields in the postmortem debugging > > API. The format matches existing bitfields but with an offset value > > that includes the SMI shift size. > > > > Bug: v8:7793 > > Change-Id: Icaecbe4a162da55d2d9a3a35a8ea85b285b2f1b7 > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2028832 > > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > > Reviewed-by: Tobias Tebbi <tebbi@chromium.org> > > Reviewed-by: Nico Hartmann <nicohartmann@chromium.org> > > Cr-Commit-Position: refs/heads/master@{#66182} > > Bug: chromium:1052756, v8:7793 > Change-Id: I9e2897efbb6321124bf4952cf09de2f179f7310d > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2062569 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Tobias Tebbi <tebbi@chromium.org> > Cr-Commit-Position: refs/heads/master@{#66349} # Not skipping CQ checks because original CL landed > 1 day ago. Bug: chromium:1052756, v8:7793 Change-Id: I6087928aa14c8551ebd294513bd8d6ffa402a0d4 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2070635 Reviewed-by: Jakob Gruber <jgruber@chromium.org> Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#66465}
2020-02-24 18:32:34 +00:00
i::Isolate* i_isolate = reinterpret_cast<i::Isolate*>(isolate);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
v8::HandleScope scope(isolate);
LocalContext context;
// Claim we don't know anything about the heap layout.
d::HeapAddresses heap_addresses{0, 0, 0, 0};
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
v8::Local<v8::Value> v = CompileRun("42");
Handle<Object> o = v8::Utils::OpenHandle(*v);
d::ObjectPropertiesResultPtr props =
d::GetObjectProperties(o->ptr(), &ReadMemory, heap_addresses);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CHECK(props->type_check_result == d::TypeCheckResult::kSmi);
CHECK(props->brief == std::string("42 (0x2a)"));
CHECK(props->type == std::string("v8::internal::Smi"));
CHECK_EQ(props->num_properties, 0);
v = CompileRun("[\"a\", \"bc\"]");
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
o = v8::Utils::OpenHandle(*v);
props = d::GetObjectProperties(o->ptr(), &ReadMemory, heap_addresses);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CHECK(props->type_check_result == d::TypeCheckResult::kUsedMap);
CHECK(props->type == std::string("v8::internal::JSArray"));
CHECK_EQ(props->num_properties, 4);
CheckProp(*props->properties[0], "v8::internal::Map", "map");
CheckProp(*props->properties[1], "v8::internal::Object",
"properties_or_hash");
CheckProp(*props->properties[2], "v8::internal::FixedArrayBase", "elements");
CheckProp(*props->properties[3], "v8::internal::Object", "length",
static_cast<i::Tagged_t>(IntToSmi(2)));
// We need to supply some valid address for decompression before reading the
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
// elements from the JSArray.
heap_addresses.any_heap_pointer = o->ptr();
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
i::Tagged_t properties_or_hash =
*reinterpret_cast<i::Tagged_t*>(props->properties[1]->address);
i::Tagged_t elements =
*reinterpret_cast<i::Tagged_t*>(props->properties[2]->address);
// The properties_or_hash_code field should be an empty fixed array. Since
// that is at a known offset, we should be able to detect it even without
// any ability to read memory.
{
MemoryFailureRegion failure(0, UINTPTR_MAX);
props =
d::GetObjectProperties(properties_or_hash, &ReadMemory, heap_addresses);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CHECK(props->type_check_result ==
d::TypeCheckResult::kObjectPointerValidButInaccessible);
CHECK(props->type == std::string("v8::internal::HeapObject"));
CHECK_EQ(props->num_properties, 1);
CheckProp(*props->properties[0], "v8::internal::Map", "map");
// "maybe" prefix indicates that GetObjectProperties recognized the offset
// within the page as matching a known object, but didn't know whether the
// object is on the right page. This response can only happen in builds
// without pointer compression, because otherwise heap addresses would be at
// deterministic locations within the heap reservation.
CHECK(COMPRESS_POINTERS_BOOL
? StartsWith(props->brief, "EmptyFixedArray")
Fix brittleness of GetObjectProperties test Part of the GetObjectProperties test case is for verifying the human- readable brief object description string that GetObjectProperties returns. That string might look something like this: "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) GetObjectProperties also tries to detect known immortal objects by recognizing their addresses, which is useful in crash dumps with limited memory. The recognized object name, if it exists, is prepended to the description string. In order to provide this data accurately (in builds without pointer compression), GetObjectProperties relies on the caller to provide the addresses of the first pages in read-only space, map space, and old space. If the caller doesn't provide those addresses, then GetObjectProperties does the best it can with limited information and reports possible matches based on an object's offset within the heap page that contains it. So the result string might look like this, if the object happened to get allocated at a lucky offset within its page: maybe LoadHandler3Map "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) As a result, when testing these descriptions, we should generally check that they contain the interesting data rather than that they start with it, because some incorrect "maybe" match with a known object might be included at the beginning. Bug: v8:10034 Change-Id: I0cf5afd67793a239614aba3665ef57cd2d663a47 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1950233 Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#65432}
2019-12-09 23:54:54 +00:00
: Contains(props->brief, "maybe EmptyFixedArray"));
// Provide a heap first page so the API can be more sure.
Reland "[torque] Support bitfield structs stored within Smis" This reverts commit 4dc1fb4e047cc36cef1308884812c76a66830724. Reason for revert: the regression from the original change was likely due to unlucky factors like code alignment. Original change's description: > Revert "[torque] Support bitfield structs stored within Smis" > > This reverts commit e5e4ea962e48bf138c640b5d6cc1ea9c358cdf86. > > Reason for revert: mysterious performance regression chromium:1052756 > > Original change's description: > > [torque] Support bitfield structs stored within Smis > > > > This change moves the definition of the bits stored in DebugInfo::flags > > to Torque, and updates the only Torque usage of that field to use more > > natural syntax. This is intended as an example of common patterns found > > in various other classes. Several supporting changes are required: > > > > 1. Add a new type representing a bitfield struct stored within a Smi. It > > is currently called SmiTagged, but I'm open to suggestions. > > 2. Add an enum-style output for Torque bitfield structs whose bitfields > > occupy only one bit each. > > 3. Add a new case to MachineOperatorReducer that makes the generated > > code for IncBlockCounter match with what was generated before this > > change. > > 4. Add support for reporting these bitfields in the postmortem debugging > > API. The format matches existing bitfields but with an offset value > > that includes the SMI shift size. > > > > Bug: v8:7793 > > Change-Id: Icaecbe4a162da55d2d9a3a35a8ea85b285b2f1b7 > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2028832 > > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > > Reviewed-by: Tobias Tebbi <tebbi@chromium.org> > > Reviewed-by: Nico Hartmann <nicohartmann@chromium.org> > > Cr-Commit-Position: refs/heads/master@{#66182} > > Bug: chromium:1052756, v8:7793 > Change-Id: I9e2897efbb6321124bf4952cf09de2f179f7310d > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2062569 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Tobias Tebbi <tebbi@chromium.org> > Cr-Commit-Position: refs/heads/master@{#66349} # Not skipping CQ checks because original CL landed > 1 day ago. Bug: chromium:1052756, v8:7793 Change-Id: I6087928aa14c8551ebd294513bd8d6ffa402a0d4 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2070635 Reviewed-by: Jakob Gruber <jgruber@chromium.org> Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#66465}
2020-02-24 18:32:34 +00:00
heap_addresses.read_only_space_first_page = reinterpret_cast<uintptr_t>(
i_isolate->heap()->read_only_space()->first_page());
props =
d::GetObjectProperties(properties_or_hash, &ReadMemory, heap_addresses);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CHECK(props->type_check_result ==
d::TypeCheckResult::kObjectPointerValidButInaccessible);
CHECK(props->type == std::string("v8::internal::HeapObject"));
CHECK_EQ(props->num_properties, 1);
CheckProp(*props->properties[0], "v8::internal::Map", "map");
CHECK(StartsWith(props->brief, "EmptyFixedArray"));
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
}
props = d::GetObjectProperties(elements, &ReadMemory, heap_addresses);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CHECK(props->type_check_result == d::TypeCheckResult::kUsedMap);
CHECK(props->type == std::string("v8::internal::FixedArray"));
CHECK_EQ(props->num_properties, 3);
CheckProp(*props->properties[0], "v8::internal::Map", "map");
CheckProp(*props->properties[1], "v8::internal::Object", "length",
static_cast<i::Tagged_t>(IntToSmi(2)));
CheckProp(*props->properties[2], "v8::internal::Object", "objects",
d::PropertyKind::kArrayOfKnownSize, 2);
// Get the second string value from the FixedArray.
i::Tagged_t second_string_address =
reinterpret_cast<i::Tagged_t*>(props->properties[2]->address)[1];
props = d::GetObjectProperties(second_string_address, &ReadMemory,
heap_addresses);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CHECK(props->type_check_result == d::TypeCheckResult::kUsedMap);
CHECK(props->type == std::string("v8::internal::SeqOneByteString"));
CHECK_EQ(props->num_properties, 4);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CheckProp(*props->properties[0], "v8::internal::Map", "map");
CheckProp(*props->properties[1], "uint32_t", "hash_field");
CheckProp(*props->properties[2], "int32_t", "length", 2);
CheckProp(*props->properties[3], "char", "chars",
d::PropertyKind::kArrayOfKnownSize, 2);
CHECK_EQ(
strncmp("bc",
reinterpret_cast<const char*>(props->properties[3]->address), 2),
0);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
// Read the second string again, using a type hint instead of the map. All of
// its properties should match what we read last time.
d::ObjectPropertiesResultPtr props2;
{
heap_addresses.read_only_space_first_page = 0;
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
uintptr_t map_address =
d::GetObjectProperties(
*reinterpret_cast<i::Tagged_t*>(props->properties[0]->address),
&ReadMemory, heap_addresses)
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
->properties[0]
->address;
MemoryFailureRegion failure(map_address, map_address + i::Map::kSize);
props2 = d::GetObjectProperties(second_string_address, &ReadMemory,
heap_addresses, "v8::internal::String");
if (COMPRESS_POINTERS_BOOL) {
// The first page of each heap space can be automatically detected when
// pointer compression is active, so we expect to use known maps instead
// of the type hint.
CHECK_EQ(props2->type_check_result, d::TypeCheckResult::kKnownMapPointer);
CHECK(props2->type == std::string("v8::internal::SeqOneByteString"));
CHECK_EQ(props2->num_properties, 4);
CheckProp(*props2->properties[3], "char", "chars",
d::PropertyKind::kArrayOfKnownSize, 2);
CHECK_EQ(props2->num_guessed_types, 0);
} else {
CHECK_EQ(props2->type_check_result, d::TypeCheckResult::kUsedTypeHint);
CHECK(props2->type == std::string("v8::internal::String"));
CHECK_EQ(props2->num_properties, 3);
// The type hint we provided was the abstract class String, but
// GetObjectProperties should have recognized that the Map pointer looked
// like the right value for a SeqOneByteString.
CHECK_EQ(props2->num_guessed_types, 1);
CHECK(std::string(props2->guessed_types[0]) ==
std::string("v8::internal::SeqOneByteString"));
}
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CheckProp(*props2->properties[0], "v8::internal::Map", "map",
*reinterpret_cast<i::Tagged_t*>(props->properties[0]->address));
CheckProp(*props2->properties[1], "uint32_t", "hash_field",
*reinterpret_cast<int32_t*>(props->properties[1]->address));
CheckProp(*props2->properties[2], "int32_t", "length", 2);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
}
// Try a weak reference.
props2 = d::GetObjectProperties(second_string_address | kWeakHeapObjectMask,
&ReadMemory, heap_addresses);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
std::string weak_ref_prefix = "weak ref to ";
CHECK(weak_ref_prefix + props->brief == props2->brief);
CHECK(props2->type_check_result == d::TypeCheckResult::kUsedMap);
CHECK(props2->type == std::string("v8::internal::SeqOneByteString"));
CHECK_EQ(props2->num_properties, 4);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
CheckProp(*props2->properties[0], "v8::internal::Map", "map",
*reinterpret_cast<i::Tagged_t*>(props->properties[0]->address));
CheckProp(*props2->properties[1], "uint32_t", "hash_field",
*reinterpret_cast<i::Tagged_t*>(props->properties[1]->address));
CheckProp(*props2->properties[2], "int32_t", "length", 2);
// Build a complicated string (multi-level cons with slices inside) to test
// string printing.
v = CompileRun(R"(
const alphabet = "abcdefghijklmnopqrstuvwxyz";
alphabet.substr(3,20) + alphabet.toUpperCase().substr(5,15) + "7")");
o = v8::Utils::OpenHandle(*v);
props = d::GetObjectProperties(o->ptr(), &ReadMemory, heap_addresses);
Fix brittleness of GetObjectProperties test Part of the GetObjectProperties test case is for verifying the human- readable brief object description string that GetObjectProperties returns. That string might look something like this: "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) GetObjectProperties also tries to detect known immortal objects by recognizing their addresses, which is useful in crash dumps with limited memory. The recognized object name, if it exists, is prepended to the description string. In order to provide this data accurately (in builds without pointer compression), GetObjectProperties relies on the caller to provide the addresses of the first pages in read-only space, map space, and old space. If the caller doesn't provide those addresses, then GetObjectProperties does the best it can with limited information and reports possible matches based on an object's offset within the heap page that contains it. So the result string might look like this, if the object happened to get allocated at a lucky offset within its page: maybe LoadHandler3Map "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) As a result, when testing these descriptions, we should generally check that they contain the interesting data rather than that they start with it, because some incorrect "maybe" match with a known object might be included at the beginning. Bug: v8:10034 Change-Id: I0cf5afd67793a239614aba3665ef57cd2d663a47 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1950233 Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#65432}
2019-12-09 23:54:54 +00:00
CHECK(Contains(props->brief, "\"defghijklmnopqrstuvwFGHIJKLMNOPQRST7\""));
// Cause a failure when reading the "second" pointer within the top-level
// ConsString.
{
CheckProp(*props->properties[4], "v8::internal::String", "second");
uintptr_t second_address = props->properties[4]->address;
MemoryFailureRegion failure(second_address, second_address + 4);
props = d::GetObjectProperties(o->ptr(), &ReadMemory, heap_addresses);
Fix brittleness of GetObjectProperties test Part of the GetObjectProperties test case is for verifying the human- readable brief object description string that GetObjectProperties returns. That string might look something like this: "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) GetObjectProperties also tries to detect known immortal objects by recognizing their addresses, which is useful in crash dumps with limited memory. The recognized object name, if it exists, is prepended to the description string. In order to provide this data accurately (in builds without pointer compression), GetObjectProperties relies on the caller to provide the addresses of the first pages in read-only space, map space, and old space. If the caller doesn't provide those addresses, then GetObjectProperties does the best it can with limited information and reports possible matches based on an object's offset within the heap page that contains it. So the result string might look like this, if the object happened to get allocated at a lucky offset within its page: maybe LoadHandler3Map "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) As a result, when testing these descriptions, we should generally check that they contain the interesting data rather than that they start with it, because some incorrect "maybe" match with a known object might be included at the beginning. Bug: v8:10034 Change-Id: I0cf5afd67793a239614aba3665ef57cd2d663a47 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1950233 Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#65432}
2019-12-09 23:54:54 +00:00
CHECK(Contains(props->brief, "\"defghijklmnopqrstuvwFGHIJKLMNOPQRST...\""));
}
// Build a very long string.
v = CompileRun("'a'.repeat(1000)");
o = v8::Utils::OpenHandle(*v);
props = d::GetObjectProperties(o->ptr(), &ReadMemory, heap_addresses);
Fix brittleness of GetObjectProperties test Part of the GetObjectProperties test case is for verifying the human- readable brief object description string that GetObjectProperties returns. That string might look something like this: "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) GetObjectProperties also tries to detect known immortal objects by recognizing their addresses, which is useful in crash dumps with limited memory. The recognized object name, if it exists, is prepended to the description string. In order to provide this data accurately (in builds without pointer compression), GetObjectProperties relies on the caller to provide the addresses of the first pages in read-only space, map space, and old space. If the caller doesn't provide those addresses, then GetObjectProperties does the best it can with limited information and reports possible matches based on an object's offset within the heap page that contains it. So the result string might look like this, if the object happened to get allocated at a lucky offset within its page: maybe LoadHandler3Map "xy" (0x28f038d5 <v8::internal::SeqOneByteString>) As a result, when testing these descriptions, we should generally check that they contain the interesting data rather than that they start with it, because some incorrect "maybe" match with a known object might be included at the beginning. Bug: v8:10034 Change-Id: I0cf5afd67793a239614aba3665ef57cd2d663a47 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1950233 Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#65432}
2019-12-09 23:54:54 +00:00
CHECK(Contains(props->brief, "\"" + std::string(80, 'a') + "...\""));
// GetObjectProperties can read cacheable external strings.
auto external_string =
v8::String::NewExternalTwoByte(isolate, new StringResource(true));
o = v8::Utils::OpenHandle(*external_string.ToLocalChecked());
props = d::GetObjectProperties(o->ptr(), &ReadMemory, heap_addresses);
CHECK(Contains(props->brief, "\"abcde\""));
// GetObjectProperties cannot read uncacheable external strings.
external_string =
v8::String::NewExternalTwoByte(isolate, new StringResource(false));
o = v8::Utils::OpenHandle(*external_string.ToLocalChecked());
props = d::GetObjectProperties(o->ptr(), &ReadMemory, heap_addresses);
CHECK_EQ(std::string(props->brief).find("\""), std::string::npos);
// Build a basic JS object and get its properties.
v = CompileRun("({a: 1, b: 2})");
o = v8::Utils::OpenHandle(*v);
props = d::GetObjectProperties(o->ptr(), &ReadMemory, heap_addresses);
// Objects constructed from literals get their properties placed inline, so
// the GetObjectProperties response should include an array.
const d::ObjectProperty& prop = FindProp(*props, "in-object properties");
CheckProp(prop, "v8::internal::Object", "in-object properties",
d::PropertyKind::kArrayOfKnownSize, 2);
// The second item in that array is the SMI value 2 from the object literal.
props2 =
d::GetObjectProperties(reinterpret_cast<i::Tagged_t*>(prop.address)[1],
&ReadMemory, heap_addresses);
CHECK(props2->brief == std::string("2 (0x2)"));
// Verify the result for a heap object field which is itself a struct: the
// "descriptors" field on a DescriptorArray.
// Start by getting the object's map and the map's descriptor array.
props = d::GetObjectProperties(ReadProp<i::Tagged_t>(*props, "map"),
&ReadMemory, heap_addresses);
props = d::GetObjectProperties(
ReadProp<i::Tagged_t>(*props, "instance_descriptors"), &ReadMemory,
heap_addresses);
// It should have at least two descriptors (possibly plus slack).
CheckProp(*props->properties[1], "uint16_t", "number_of_all_descriptors");
uint16_t number_of_all_descriptors =
*reinterpret_cast<uint16_t*>(props->properties[1]->address);
CHECK_GE(number_of_all_descriptors, 2);
// The "descriptors" property should describe the struct layout for each
// element in the array.
const d::ObjectProperty& descriptors = *props->properties[6];
// No C++ type is reported directly because there may not be an actual C++
// struct with this layout, hence the empty string in this check.
CheckProp(descriptors, /*type=*/"", "descriptors",
d::PropertyKind::kArrayOfKnownSize, number_of_all_descriptors);
CHECK_EQ(descriptors.size, 3 * i::kTaggedSize);
CHECK_EQ(descriptors.num_struct_fields, 3);
CheckStructProp(*descriptors.struct_fields[0],
"v8::internal::PrimitiveHeapObject", "key",
0 * i::kTaggedSize);
CheckStructProp(*descriptors.struct_fields[1], "v8::internal::Object",
"details", 1 * i::kTaggedSize);
CheckStructProp(*descriptors.struct_fields[2], "v8::internal::Object",
"value", 2 * i::kTaggedSize);
// Build a basic JS function and get its properties. This will allow us to
// exercise bitfield functionality.
v = CompileRun("(function () {})");
o = v8::Utils::OpenHandle(*v);
props = d::GetObjectProperties(o->ptr(), &ReadMemory, heap_addresses);
props = d::GetObjectProperties(
ReadProp<i::Tagged_t>(*props, "shared_function_info"), &ReadMemory,
heap_addresses);
const d::ObjectProperty& flags = FindProp(*props, "flags");
Reland "[torque] Support bitfield structs stored within Smis" This reverts commit 4dc1fb4e047cc36cef1308884812c76a66830724. Reason for revert: the regression from the original change was likely due to unlucky factors like code alignment. Original change's description: > Revert "[torque] Support bitfield structs stored within Smis" > > This reverts commit e5e4ea962e48bf138c640b5d6cc1ea9c358cdf86. > > Reason for revert: mysterious performance regression chromium:1052756 > > Original change's description: > > [torque] Support bitfield structs stored within Smis > > > > This change moves the definition of the bits stored in DebugInfo::flags > > to Torque, and updates the only Torque usage of that field to use more > > natural syntax. This is intended as an example of common patterns found > > in various other classes. Several supporting changes are required: > > > > 1. Add a new type representing a bitfield struct stored within a Smi. It > > is currently called SmiTagged, but I'm open to suggestions. > > 2. Add an enum-style output for Torque bitfield structs whose bitfields > > occupy only one bit each. > > 3. Add a new case to MachineOperatorReducer that makes the generated > > code for IncBlockCounter match with what was generated before this > > change. > > 4. Add support for reporting these bitfields in the postmortem debugging > > API. The format matches existing bitfields but with an offset value > > that includes the SMI shift size. > > > > Bug: v8:7793 > > Change-Id: Icaecbe4a162da55d2d9a3a35a8ea85b285b2f1b7 > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2028832 > > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > > Reviewed-by: Tobias Tebbi <tebbi@chromium.org> > > Reviewed-by: Nico Hartmann <nicohartmann@chromium.org> > > Cr-Commit-Position: refs/heads/master@{#66182} > > Bug: chromium:1052756, v8:7793 > Change-Id: I9e2897efbb6321124bf4952cf09de2f179f7310d > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2062569 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Tobias Tebbi <tebbi@chromium.org> > Cr-Commit-Position: refs/heads/master@{#66349} # Not skipping CQ checks because original CL landed > 1 day ago. Bug: chromium:1052756, v8:7793 Change-Id: I6087928aa14c8551ebd294513bd8d6ffa402a0d4 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2070635 Reviewed-by: Jakob Gruber <jgruber@chromium.org> Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#66465}
2020-02-24 18:32:34 +00:00
CHECK_GE(flags.num_struct_fields, 3);
CheckStructProp(*flags.struct_fields[0], "FunctionKind", "function_kind", 0,
5, 0);
CheckStructProp(*flags.struct_fields[1], "bool", "is_native", 0, 1, 5);
CheckStructProp(*flags.struct_fields[2], "bool", "is_strict", 0, 1, 6);
Reland "[torque] Support bitfield structs stored within Smis" This reverts commit 4dc1fb4e047cc36cef1308884812c76a66830724. Reason for revert: the regression from the original change was likely due to unlucky factors like code alignment. Original change's description: > Revert "[torque] Support bitfield structs stored within Smis" > > This reverts commit e5e4ea962e48bf138c640b5d6cc1ea9c358cdf86. > > Reason for revert: mysterious performance regression chromium:1052756 > > Original change's description: > > [torque] Support bitfield structs stored within Smis > > > > This change moves the definition of the bits stored in DebugInfo::flags > > to Torque, and updates the only Torque usage of that field to use more > > natural syntax. This is intended as an example of common patterns found > > in various other classes. Several supporting changes are required: > > > > 1. Add a new type representing a bitfield struct stored within a Smi. It > > is currently called SmiTagged, but I'm open to suggestions. > > 2. Add an enum-style output for Torque bitfield structs whose bitfields > > occupy only one bit each. > > 3. Add a new case to MachineOperatorReducer that makes the generated > > code for IncBlockCounter match with what was generated before this > > change. > > 4. Add support for reporting these bitfields in the postmortem debugging > > API. The format matches existing bitfields but with an offset value > > that includes the SMI shift size. > > > > Bug: v8:7793 > > Change-Id: Icaecbe4a162da55d2d9a3a35a8ea85b285b2f1b7 > > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2028832 > > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > > Reviewed-by: Tobias Tebbi <tebbi@chromium.org> > > Reviewed-by: Nico Hartmann <nicohartmann@chromium.org> > > Cr-Commit-Position: refs/heads/master@{#66182} > > Bug: chromium:1052756, v8:7793 > Change-Id: I9e2897efbb6321124bf4952cf09de2f179f7310d > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2062569 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Tobias Tebbi <tebbi@chromium.org> > Cr-Commit-Position: refs/heads/master@{#66349} # Not skipping CQ checks because original CL landed > 1 day ago. Bug: chromium:1052756, v8:7793 Change-Id: I6087928aa14c8551ebd294513bd8d6ffa402a0d4 Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/2070635 Reviewed-by: Jakob Gruber <jgruber@chromium.org> Reviewed-by: Tobias Tebbi <tebbi@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#66465}
2020-02-24 18:32:34 +00:00
// Get data about a different bitfield struct which is contained within a smi.
Handle<i::JSFunction> function = Handle<i::JSFunction>::cast(o);
Handle<i::SharedFunctionInfo> shared(function->shared(), i_isolate);
Handle<i::DebugInfo> debug_info =
i_isolate->debug()->GetOrCreateDebugInfo(shared);
props =
d::GetObjectProperties(debug_info->ptr(), &ReadMemory, heap_addresses);
const d::ObjectProperty& debug_flags = FindProp(*props, "flags");
CHECK_GE(debug_flags.num_struct_fields, 5);
CheckStructProp(*debug_flags.struct_fields[0], "bool", "has_break_info", 0, 1,
i::kSmiTagSize + i::kSmiShiftSize);
CheckStructProp(*debug_flags.struct_fields[4], "bool", "can_break_at_entry",
0, 1, i::kSmiTagSize + i::kSmiShiftSize + 4);
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
}
TEST(ListObjectClasses) {
CcTest::InitializeVM();
// The ListObjectClasses result will change as classes are added, removed, or
// renamed. Just check that a few expected classes are included in the list,
// and that there are no duplicates.
const d::ClassList* class_list = d::ListObjectClasses();
std::unordered_set<std::string> class_set;
for (size_t i = 0; i < class_list->num_class_names; ++i) {
CHECK_WITH_MSG(class_set.insert(class_list->class_names[i]).second,
"there should be no duplicate entries");
}
CHECK_NE(class_set.find("v8::internal::HeapObject"), class_set.end());
CHECK_NE(class_set.find("v8::internal::String"), class_set.end());
CHECK_NE(class_set.find("v8::internal::JSRegExp"), class_set.end());
}
Reland "Add postmortem debugging helper library" This is a reland of 517ab73fd7e3fdb70220b9699bca4c69a32e212e Updates since original: now compressed pointers passed to the function GetObjectProperties are required to be sign-extended. Previously, the function allowed zero-extended values, but that led to ambiguity on pointers like 0x88044919: is it compressed or is the heap range actually centered on 0x100000000? Original change's description: > Add postmortem debugging helper library > > This change begins to implement the functionality described in > https://docs.google.com/document/d/1evHnb1uLlSbvHAAsmOXyc25x3uh1DjgNa8u1RHvwVhk/edit# > for investigating V8 state in crash dumps. > > This change adds a new library, v8_debug_helper, for providing platform- > agnostic assistance with postmortem debugging. This library can be used > by extensions built for debuggers such as WinDbg or lldb. Its public API > is described by debug-helper.h; currently the only method it exposes is > GetObjectProperties, but we'd like to add more functionality over time. > The API surface is restricted to plain C-style structs and pointers, so > that it's easy to link from a debugger extension built with a different > toolchain. > > This change also adds a new cctest file to exercise some basic > interaction with the new library. > > The API function GetObjectProperties takes an object pointer (which > could be compressed, or weak, or a SMI), and returns a string > description of the object and a list of properties the object contains. > For now, the list of properties is entirely based on Torque object > definitions, but we expect to add custom properties in future updates so > that it can be easier to make sense of complex data structures such as > dictionaries. > > GetObjectProperties does several things that are intended to generate > somewhat useful results even in cases where memory may be corrupt or > unavailable: > - The caller may optionally provide a type string which will be used if > the memory for the object's Map is inaccessible. > - All object pointers are compared against the list of known objects > generated by mkgrokdump. The caller may optionally provide the > pointers for the first pages of various heap spaces, to avoid spurious > matches. If those pointers are not provided, then any matches are > prefixed with "maybe" in the resulting description string, such as > "maybe UndefinedValue (0x4288000341 <Oddball>)". > > Bug: v8:9376 > > Change-Id: Iebf3cc2dea3133c7811bcefcdf38d9458b02fded > Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1628012 > Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> > Reviewed-by: Yang Guo <yangguo@chromium.org> > Reviewed-by: Michael Stanton <mvstanton@chromium.org> > Cr-Commit-Position: refs/heads/master@{#62882} Bug: v8:9376 Change-Id: I866a1cc9d4c34bfe10c7b98462451fe69763cf3f Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1717090 Reviewed-by: Michael Achenbach <machenbach@chromium.org> Reviewed-by: Michael Stanton <mvstanton@chromium.org> Commit-Queue: Seth Brenith <seth.brenith@microsoft.com> Cr-Commit-Position: refs/heads/master@{#63008}
2019-07-30 14:38:15 +00:00
} // namespace internal
} // namespace v8