Skip to content

Support ethdebug source locations under EOF #15994

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 8 commits into
base: develop
Choose a base branch
from
Open

Support ethdebug source locations under EOF #15994

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
05ee3a4
Add instruction location info to eof assembly
clonker Apr 9, 2025
c7b8f7a
Refactor legacy assemble to also use RAII instruction location constr…
clonker Apr 10, 2025
906b8c5
Ethdebug requires assembly instance
clonker Apr 12, 2025
d35ccb0
Ethdebug instructions output over multiple code sections
clonker Apr 8, 2025
1bf1d7e
Add ethdebug schema
clonker Apr 11, 2025
7f21b43
Ethdebug uses schema for serialization
clonker Apr 9, 2025
576ce89
Update test expectations with eof-enabled ethdebug
clonker Apr 9, 2025
082d7f9
Test ethdebug output when requested in standard json under EOF
clonker Apr 11, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions Changelog.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,7 @@ Language Features:


Compiler Features:
* Ethdebug: Experimental support for instructions and source locations under EOF.
* NatSpec: Capture Natspec documentation of `enum` values in the AST.


Expand Down
222 changes: 120 additions & 102 deletions libevmasm/Assembly.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -53,6 +53,54 @@ using namespace solidity::evmasm;
using namespace solidity::langutil;
using namespace solidity::util;

namespace
{

/// Produces instruction location info in RAII style. When an assembly instruction is added to the bytecode,
/// this class can be instantiated in that scope. It will record the current bytecode size (before addition)
/// and, at destruction time, record the new bytecode size. This information is then added to an external
/// instruction locations vector.
/// If the instruction decomposes into multiple individual evm instructions, `emit` can be
/// called for all but the last one (which will be emitted by the destructor).
class InstructionLocationEmitter
{
public:
InstructionLocationEmitter(
std::vector<LinkerObject::InstructionLocation>& _instructionLocations,
bytes const& _bytecode,
size_t const _assemblyItemIndex
):
m_instructionLocations(_instructionLocations),
m_bytecode(_bytecode),
m_assemblyItemIndex(_assemblyItemIndex),
m_instructionLocationStart(_bytecode.size())
{}

~InstructionLocationEmitter()
{
emit();
}

void emit()
{
auto const end = m_bytecode.size();
m_instructionLocations.emplace_back(LinkerObject::InstructionLocation{
.start = m_instructionLocationStart,
.end = end,
.assemblyItemIndex = m_assemblyItemIndex
});
m_instructionLocationStart = end;
}

private:
std::vector<LinkerObject::InstructionLocation>& m_instructionLocations;
bytes const& m_bytecode;
size_t const m_assemblyItemIndex;
size_t m_instructionLocationStart;
};

}

std::map<std::string, std::shared_ptr<std::string const>> Assembly::s_sharedSourceNames;

AssemblyItem const& Assembly::append(AssemblyItem _i)
Expand Down Expand Up @@ -1281,163 +1329,119 @@ LinkerObject const& Assembly::assembleLegacy() const
uint8_t dataRefPush = static_cast<uint8_t>(pushInstruction(bytesPerDataRef));

LinkerObject::CodeSectionLocation codeSectionLocation;
codeSectionLocation.instructionLocations.reserve(items.size());
codeSectionLocation.start = 0;
size_t assemblyItemIndex = 0;
auto assembleInstruction = [&](auto&& _addInstruction) {
size_t start = ret.bytecode.size();
_addInstruction();
size_t end = ret.bytecode.size();
codeSectionLocation.instructionLocations.emplace_back(
LinkerObject::InstructionLocation{
.start = start,
.end = end,
.assemblyItemIndex = assemblyItemIndex
}
);
};
for (AssemblyItem const& item: items)
for (auto const& [assemblyItemIndex, item]: items | ranges::views::enumerate)
{
// collect instruction locations via side effects
InstructionLocationEmitter instructionLocationEmitter(codeSectionLocation.instructionLocations, ret.bytecode, assemblyItemIndex);
// store position of the invalid jump destination
if (item.type() != Tag && m_tagPositionsInBytecode[0] == std::numeric_limits<size_t>::max())
m_tagPositionsInBytecode[0] = ret.bytecode.size();

switch (item.type())
{
case Operation:
assembleInstruction([&](){
ret.bytecode += assembleOperation(item);
});
ret.bytecode += assembleOperation(item);
break;
case Push:
assembleInstruction([&](){
ret.bytecode += assemblePush(item);
});
ret.bytecode += assemblePush(item);
break;
case PushTag:
{
assembleInstruction([&](){
ret.bytecode.push_back(tagPush);
tagRefs[ret.bytecode.size()] = item.splitForeignPushTag();
ret.bytecode.resize(ret.bytecode.size() + bytesPerTag);
});
ret.bytecode.push_back(tagPush);
tagRefs[ret.bytecode.size()] = item.splitForeignPushTag();
ret.bytecode.resize(ret.bytecode.size() + bytesPerTag);
break;
}
case PushData:
assembleInstruction([&]() {
ret.bytecode.push_back(dataRefPush);
dataRefs.insert(std::make_pair(h256(item.data()), ret.bytecode.size()));
ret.bytecode.resize(ret.bytecode.size() + bytesPerDataRef);
});
ret.bytecode.push_back(dataRefPush);
dataRefs.insert(std::make_pair(h256(item.data()), ret.bytecode.size()));
ret.bytecode.resize(ret.bytecode.size() + bytesPerDataRef);
break;
case PushSub:
assembleInstruction([&]() {
assertThrow(item.data() <= std::numeric_limits<size_t>::max(), AssemblyException, "");
ret.bytecode.push_back(dataRefPush);
subRefs.insert(std::make_pair(static_cast<size_t>(item.data()), ret.bytecode.size()));
ret.bytecode.resize(ret.bytecode.size() + bytesPerDataRef);
});
assertThrow(item.data() <= std::numeric_limits<size_t>::max(), AssemblyException, "");
ret.bytecode.push_back(dataRefPush);
subRefs.insert(std::make_pair(static_cast<size_t>(item.data()), ret.bytecode.size()));
ret.bytecode.resize(ret.bytecode.size() + bytesPerDataRef);
break;
case PushSubSize:
{
assembleInstruction([&](){
assertThrow(item.data() <= std::numeric_limits<size_t>::max(), AssemblyException, "");
auto s = subAssemblyById(static_cast<size_t>(item.data()))->assemble().bytecode.size();
item.setPushedValue(u256(s));
unsigned b = std::max<unsigned>(1, numberEncodingSize(s));
ret.bytecode.push_back(static_cast<uint8_t>(pushInstruction(b)));
ret.bytecode.resize(ret.bytecode.size() + b);
bytesRef byr(&ret.bytecode.back() + 1 - b, b);
toBigEndian(s, byr);
});
assertThrow(item.data() <= std::numeric_limits<size_t>::max(), AssemblyException, "");
auto s = subAssemblyById(static_cast<size_t>(item.data()))->assemble().bytecode.size();
item.setPushedValue(u256(s));
unsigned b = std::max<unsigned>(1, numberEncodingSize(s));
ret.bytecode.push_back(static_cast<uint8_t>(pushInstruction(b)));
ret.bytecode.resize(ret.bytecode.size() + b);
bytesRef byr(&ret.bytecode.back() + 1 - b, b);
toBigEndian(s, byr);
break;
}
case PushProgramSize:
{
assembleInstruction([&](){
ret.bytecode.push_back(dataRefPush);
sizeRefs.push_back(static_cast<unsigned>(ret.bytecode.size()));
ret.bytecode.resize(ret.bytecode.size() + bytesPerDataRef);
});
ret.bytecode.push_back(dataRefPush);
sizeRefs.push_back(static_cast<unsigned>(ret.bytecode.size()));
ret.bytecode.resize(ret.bytecode.size() + bytesPerDataRef);
break;
}
case PushLibraryAddress:
{
assembleInstruction([&]() {
auto const [bytecode, linkRef] = assemblePushLibraryAddress(item, ret.bytecode.size());
ret.bytecode += bytecode;
ret.linkReferences.insert(linkRef);
});
auto const [bytecode, linkRef] = assemblePushLibraryAddress(item, ret.bytecode.size());
ret.bytecode += bytecode;
ret.linkReferences.insert(linkRef);
break;
}
case PushImmutable:
assembleInstruction([&]() {
ret.bytecode.push_back(static_cast<uint8_t>(Instruction::PUSH32));
// Maps keccak back to the "identifier" std::string of that immutable.
ret.immutableReferences[item.data()].first = m_immutables.at(item.data());
// Record the bytecode offset of the PUSH32 argument.
ret.immutableReferences[item.data()].second.emplace_back(ret.bytecode.size());
// Advance bytecode by 32 bytes (default initialized).
ret.bytecode.resize(ret.bytecode.size() + 32);
});
ret.bytecode.push_back(static_cast<uint8_t>(Instruction::PUSH32));
// Maps keccak back to the "identifier" std::string of that immutable.
ret.immutableReferences[item.data()].first = m_immutables.at(item.data());
// Record the bytecode offset of the PUSH32 argument.
ret.immutableReferences[item.data()].second.emplace_back(ret.bytecode.size());
// Advance bytecode by 32 bytes (default initialized).
ret.bytecode.resize(ret.bytecode.size() + 32);
break;
case VerbatimBytecode:
ret.bytecode += assembleVerbatimBytecode(item);
break;
case AssignImmutable:
{
// this decomposes into multiple evm instructions, so we manually call emit on `addInstructionLocation`
// Expect 2 elements on stack (source, dest_base)
auto const& offsets = immutableReferencesBySub[item.data()].second;
for (size_t i = 0; i < offsets.size(); ++i)
{
if (i != offsets.size() - 1)
{
assembleInstruction([&]() {
ret.bytecode.push_back(uint8_t(Instruction::DUP2));
});
assembleInstruction([&]() {
ret.bytecode.push_back(uint8_t(Instruction::DUP2));
});
ret.bytecode.push_back(static_cast<uint8_t>(Instruction::DUP2));
instructionLocationEmitter.emit();
ret.bytecode.push_back(static_cast<uint8_t>(Instruction::DUP2));
instructionLocationEmitter.emit();
}
assembleInstruction([&]() {
// TODO: should we make use of the constant optimizer methods for pushing the offsets?
bytes offsetBytes = toCompactBigEndian(u256(offsets[i]));
ret.bytecode.push_back(static_cast<uint8_t>(pushInstruction(static_cast<unsigned>(offsetBytes.size()))));
ret.bytecode += offsetBytes;
});
assembleInstruction([&]() {
ret.bytecode.push_back(uint8_t(Instruction::ADD));
});
assembleInstruction([&]() {
ret.bytecode.push_back(uint8_t(Instruction::MSTORE));
});
// TODO: should we make use of the constant optimizer methods for pushing the offsets?
bytes offsetBytes = toCompactBigEndian(u256(offsets[i]));
ret.bytecode.push_back(static_cast<uint8_t>(pushInstruction(static_cast<unsigned>(offsetBytes.size()))));
ret.bytecode += offsetBytes;
instructionLocationEmitter.emit();
ret.bytecode.push_back(static_cast<uint8_t>(Instruction::ADD));
instructionLocationEmitter.emit();
ret.bytecode.push_back(static_cast<uint8_t>(Instruction::MSTORE));
// no emit needed here, it's taken care of by the destructor of addInstructionLocation
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

addInstrictionLocation -> InstructionLocationEmitter, other comments also still referring to the old name

}
if (offsets.empty())
{
assembleInstruction([&]() {
ret.bytecode.push_back(uint8_t(Instruction::POP));
});
assembleInstruction([&]() {
ret.bytecode.push_back(uint8_t(Instruction::POP));
});
ret.bytecode.push_back(static_cast<uint8_t>(Instruction::POP));
instructionLocationEmitter.emit();
ret.bytecode.push_back(static_cast<uint8_t>(Instruction::POP));
// no emit needed here, it's taken care of by the destructor of addInstructionLocation
}
immutableReferencesBySub.erase(item.data());
break;
}
case PushDeployTimeAddress:
assembleInstruction([&]() {
ret.bytecode += assemblePushDeployTimeAddress();
});
ret.bytecode += assemblePushDeployTimeAddress();
break;
case Tag:
assembleInstruction([&](){
ret.bytecode += assembleTag(item, ret.bytecode.size(), true);
});
ret.bytecode += assembleTag(item, ret.bytecode.size(), true);
break;
default:
solAssert(false, "Unexpected opcode while assembling.");
}

++assemblyItemIndex;
}

codeSectionLocation.end = ret.bytecode.size();
Expand Down Expand Up @@ -1606,9 +1610,17 @@ LinkerObject const& Assembly::assembleEOF() const
for (auto&& [codeSectionIndex, codeSection]: m_codeSections | ranges::views::enumerate)
{
auto const sectionStart = ret.bytecode.size();

std::vector<LinkerObject::InstructionLocation> instructionLocations;
instructionLocations.reserve(codeSection.items.size());

solAssert(!codeSection.items.empty(), "Empty code section.");
for (AssemblyItem const& item: codeSection.items)

for (auto const& [assemblyItemIndex, item]: codeSection.items | ranges::views::enumerate)
{
// collect instruction locations via side effects
InstructionLocationEmitter instructionLocationEmitter {instructionLocations, ret.bytecode, assemblyItemIndex};

// store position of the invalid jump destination
if (item.type() != Tag && m_tagPositionsInBytecode[0] == std::numeric_limits<size_t>::max())
m_tagPositionsInBytecode[0] = ret.bytecode.size();
Expand Down Expand Up @@ -1724,6 +1736,12 @@ LinkerObject const& Assembly::assembleEOF() const
"Code section too large for EOF."
);
setBigEndianUint16(ret.bytecode, codeSectionSizePositions[codeSectionIndex], ret.bytecode.size() - sectionStart);

ret.codeSectionLocations.emplace_back(LinkerObject::CodeSectionLocation{
.start = sectionStart,
.end = ret.bytecode.size(),
.instructionLocations = std::move(instructionLocations)
});
}

for (auto const& [refPos, tagId]: tagRef)
Expand Down
2 changes: 2 additions & 0 deletions libevmasm/CMakeLists.txt
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,8 @@ set(sources
AssemblyItem.h
Ethdebug.cpp
Ethdebug.h
EthdebugSchema.cpp
EthdebugSchema.h
EVMAssemblyStack.cpp
EVMAssemblyStack.h
BlockDeduplicator.cpp
Expand Down
Loading