1
0
mirror of https://github.com/TomHarte/CLK.git synced 2024-06-17 21:30:14 +00:00

Attempt to support 'exhaustive' disassemblies (i.e. ones that chase every byte).

This commit is contained in:
Thomas Harte 2023-09-04 15:13:06 -04:00
parent a21fe92b7a
commit 251b8e69ad
5 changed files with 64 additions and 19 deletions

View File

@ -316,5 +316,5 @@ Disassembly Analyser::Static::MOS6502::Disassemble(
const std::vector<uint8_t> &memory,
const std::function<std::size_t(uint16_t)> &address_mapper,
std::vector<uint16_t> entry_points) {
return Analyser::Static::Disassembly::Disassemble<Disassembly, uint16_t, MOS6502Disassembler>(memory, address_mapper, entry_points);
return Analyser::Static::Disassembly::Disassemble<Disassembly, uint16_t, MOS6502Disassembler>(memory, address_mapper, entry_points, false);
}

View File

@ -14,30 +14,50 @@ namespace Analyser::Static::Disassembly {
template <typename D, typename S> struct PartialDisassembly {
D disassembly;
std::vector<S> remaining_entry_points;
std::vector<bool> touched;
};
template <typename D, typename S, typename Disassembler> D Disassemble(
const std::vector<uint8_t> &memory,
const std::function<std::size_t(S)> &address_mapper,
std::vector<S> entry_points) {
std::vector<S> entry_points,
bool exhaustive) {
PartialDisassembly<D, S> partial_disassembly;
partial_disassembly.remaining_entry_points = entry_points;
partial_disassembly.touched.resize(memory.size());
while(!partial_disassembly.remaining_entry_points.empty()) {
// pull the next entry point from the back of the vector
S next_entry_point = partial_disassembly.remaining_entry_points.back();
partial_disassembly.remaining_entry_points.pop_back();
while(true) {
// Do a recursive-style disassembly for all current entry points.
while(!partial_disassembly.remaining_entry_points.empty()) {
// Pull the next entry point from the back of the vector.
S next_entry_point = partial_disassembly.remaining_entry_points.back();
partial_disassembly.remaining_entry_points.pop_back();
// if that address has already been visited, forget about it
if( partial_disassembly.disassembly.instructions_by_address.find(next_entry_point)
!= partial_disassembly.disassembly.instructions_by_address.end()) continue;
// If that address has already been visited, forget about it.
if( partial_disassembly.disassembly.instructions_by_address.find(next_entry_point)
!= partial_disassembly.disassembly.instructions_by_address.end()) continue;
// if it's outgoing, log it as such and forget about it; otherwise disassemble
std::size_t mapped_entry_point = address_mapper(next_entry_point);
if(mapped_entry_point >= memory.size())
partial_disassembly.disassembly.outward_calls.insert(next_entry_point);
else
Disassembler::AddToDisassembly(partial_disassembly, memory, address_mapper, next_entry_point);
// If it's outgoing, log it as such and forget about it; otherwise disassemble.
std::size_t mapped_entry_point = address_mapper(next_entry_point);
if(mapped_entry_point >= memory.size())
partial_disassembly.disassembly.outward_calls.insert(next_entry_point);
else
Disassembler::AddToDisassembly(partial_disassembly, memory, address_mapper, next_entry_point);
}
// If this is not an exhaustive disassembly, that's your lot.
if(!exhaustive) {
break;
}
// Otherwise, find the first address that isn't yet disassembled and chuck it onto the list.
auto first_untouched = std::find(partial_disassembly.touched.begin(), partial_disassembly.touched.end(), false);
if(first_untouched == partial_disassembly.touched.end()) {
break;
}
partial_disassembly.remaining_entry_points.push_back(
static_cast<S>(first_untouched - partial_disassembly.touched.begin())
);
}
return partial_disassembly.disassembly;

View File

@ -558,6 +558,12 @@ struct Z80Disassembler {
// Store the instruction away.
disassembly.disassembly.instructions_by_address[instruction.address] = instruction;
// Apply all touches.
std::fill(
disassembly.touched.begin() + instruction.address,
disassembly.touched.begin() + accessor.address(),
true);
// Update access tables.
int access_type =
((instruction.source == Instruction::Location::Operand_Indirect) ? 1 : 0) |
@ -611,9 +617,18 @@ struct Z80Disassembler {
} // end of anonymous namespace
Disassembly Analyser::Static::Z80::Disassemble(
const std::vector<uint8_t> &memory,
const std::function<std::size_t(uint16_t)> &address_mapper,
std::vector<uint16_t> entry_points) {
return Analyser::Static::Disassembly::Disassemble<Disassembly, uint16_t, Z80Disassembler>(memory, address_mapper, entry_points);
std::vector<uint16_t> entry_points,
Approach approach)
{
return Analyser::Static::Disassembly::Disassemble<Disassembly, uint16_t, Z80Disassembler>(
memory,
address_mapper,
entry_points,
approach == Approach::Exhaustive
);
}

View File

@ -76,10 +76,19 @@ struct Disassembly {
std::set<uint16_t> internal_stores, internal_loads, internal_modifies;
};
enum class Approach {
/// Disassemble from the supplied entry points until an indeterminate branch or return only, adding other fully-static
/// entry points as they are observed.
Recursive,
/// Disassemble all supplied bytes, regardless of what nonsense may be encountered by accidental parsing of data areas.
Exhaustive,
};
Disassembly Disassemble(
const std::vector<uint8_t> &memory,
const std::function<std::size_t(uint16_t)> &address_mapper,
std::vector<uint16_t> entry_points);
std::vector<uint16_t> entry_points,
Approach approach);
}

View File

@ -119,7 +119,8 @@ static Analyser::Static::TargetList CartridgeTargetsFrom(
Analyser::Static::Z80::Disassemble(
first_8k,
Analyser::Static::Disassembler::OffsetMapper(start_address),
{ init_address }
{ init_address },
Analyser::Static::Z80::Approach::Exhaustive
);
// // Look for a indirect store followed by an unconditional JP or CALL into another