1
0
mirror of https://github.com/TomHarte/CLK.git synced 2024-11-16 19:08:08 +00:00
CLK/OSBindings/Mac/Clock SignalTests/ARMDecoderTests.mm
2024-02-28 11:27:27 -05:00

338 lines
11 KiB
Plaintext

//
// ARMDecoderTests.m
// Clock Signal
//
// Created by Thomas Harte on 16/02/2024.
// Copyright 2024 Thomas Harte. All rights reserved.
//
#import <XCTest/XCTest.h>
#include "../../../InstructionSets/ARM/OperationMapper.hpp"
#include "../../../InstructionSets/ARM/Registers.hpp"
#include "../../../Numeric/Carry.hpp"
using namespace InstructionSet::ARM;
namespace {
template <ShiftType type, bool set_carry>
void shift(uint32_t &source, uint32_t amount, uint32_t *carry = nullptr) {
switch(type) {
case ShiftType::LogicalLeft:
if(amount > 32) {
if constexpr (set_carry) *carry = 0;
source = 0;
} else if(amount == 32) {
if constexpr (set_carry) *carry = source & 1;
source = 0;
} else if(amount > 0) {
if constexpr (set_carry) *carry = source & (0x8000'0000 >> (amount - 1));
source <<= amount;
}
break;
case ShiftType::LogicalRight:
if(amount > 32) {
if constexpr (set_carry) *carry = 0;
source = 0;
} else if(amount == 32) {
if constexpr (set_carry) *carry = source & 0x8000'0000;
source = 0;
} else if(amount > 0) {
if constexpr (set_carry) *carry = source & (1 << (amount - 1));
source >>= amount;
} else {
// A logical shift right by '0' is treated as a shift by 32;
// assemblers are supposed to map LSR #0 to LSL #0.
if constexpr (set_carry) *carry = source & 0x8000'0000;
source = 0;
}
break;
case ShiftType::ArithmeticRight: {
const uint32_t sign = (source & 0x8000'0000) ? 0xffff'ffff : 0x0000'0000;
if(amount >= 32) {
if constexpr (set_carry) *carry = sign;
source = sign;
} else if(amount > 0) {
if constexpr (set_carry) *carry = source & (1 << (amount - 1));
source = (source >> amount) | (sign << (32 - amount));
} else {
// As per logical right, an arithmetic shift of '0' is
// treated as a shift by 32.
if constexpr (set_carry) *carry = source & 0x8000'0000;
source = sign;
}
} break;
case ShiftType::RotateRight: {
if(amount == 32) {
if constexpr (set_carry) *carry = source & 0x8000'0000;
} else if(amount == 0) {
// Rotate right by 0 is treated as a rotate right by 1 through carry.
const uint32_t high = *carry << 31;
if constexpr (set_carry) *carry = source & 1;
source = (source >> 1) | high;
} else {
amount &= 31;
if constexpr (set_carry) *carry = source & (1 << (amount - 1));
source = (source >> amount) | (source << (32 - amount));
}
} break;
default: break;
}
}
template <bool set_carry>
void shift(ShiftType type, uint32_t &source, uint32_t amount, uint32_t *carry) {
switch(type) {
case ShiftType::LogicalLeft:
shift<ShiftType::LogicalLeft, set_carry>(source, amount, carry);
break;
case ShiftType::LogicalRight:
shift<ShiftType::LogicalRight, set_carry>(source, amount, carry);
break;
case ShiftType::ArithmeticRight:
shift<ShiftType::ArithmeticRight, set_carry>(source, amount, carry);
break;
case ShiftType::RotateRight:
shift<ShiftType::RotateRight, set_carry>(source, amount, carry);
break;
}
}
struct Scheduler {
bool should_schedule(Condition condition) {
return registers_.test(condition);
}
template <Flags f> void perform(DataProcessing fields) {
constexpr DataProcessingFlags flags(f);
const bool shift_by_register = !flags.operand2_is_immediate() && fields.shift_count_is_register();
// Write a raw result into the PC proxy if the target is R15; it'll be stored properly later.
uint32_t pc_proxy;
auto &destination = fields.destination() == 15 ? pc_proxy : registers_.active[fields.destination()];
// "When R15 appears in either of the Rn or Rs positions it will give the value
// of the PC alone, with the PSR bits replaced by zeroes. ...
//
// If the shift amount is specified in the instruction, the PC will be 8 bytes ahead.
// If a register is used to specify the shift amount, the PC will be ... 12 bytes ahead
// when used as Rn or Rm."
const uint32_t operand1 =
(fields.operand1() == 15) ?
registers_.pc(shift_by_register ? 12 : 8) :
registers_.active[fields.operand1()];
uint32_t operand2;
uint32_t rotate_carry = registers_.c();
// Populate carry from the shift only if it'll be used.
constexpr bool shift_sets_carry = is_logical(flags.operation()) && flags.set_condition_codes();
// Get operand 2.
if constexpr (flags.operand2_is_immediate()) {
operand2 = fields.immediate();
if(fields.rotate()) {
shift<ShiftType::RotateRight, shift_sets_carry>(operand2, fields.rotate(), &rotate_carry);
}
} else {
uint32_t shift_amount;
if(fields.shift_count_is_register()) {
// "When R15 appears in either of the Rn or Rs positions it will give the value
// of the PC alone, with the PSR bits replaced by zeroes. ...
//
// If a register is used to specify the shift amount, the
// PC will be 8 bytes ahead when used as Rs."
shift_amount =
fields.shift_register() == 15 ?
registers_.pc(8) :
registers_.active[fields.shift_register()];
} else {
shift_amount = fields.shift_amount();
}
// "When R15 appears in the Rm position it will give the value of the PC together
// with the PSR flags to the barrel shifter. ...
//
// If the shift amount is specified in the instruction, the PC will be 8 bytes ahead.
// If a register is used to specify the shift amount, the PC will be ... 12 bytes ahead
// when used as Rn or Rm."
if(fields.operand2() == 15) {
operand2 = registers_.pc_status(shift_by_register ? 12 : 8);
} else {
operand2 = registers_.active[fields.operand2()];
}
shift<shift_sets_carry>(fields.shift_type(), operand2, shift_amount, &rotate_carry);
}
// Perform the data processing operation.
uint32_t conditions = 0;
switch(flags.operation()) {
// Logical operations.
case DataProcessingOperation::AND: conditions = destination = operand1 & operand2; break;
case DataProcessingOperation::EOR: conditions = destination = operand1 ^ operand2; break;
case DataProcessingOperation::ORR: conditions = destination = operand1 | operand2; break;
case DataProcessingOperation::BIC: conditions = destination = operand1 & ~operand2; break;
case DataProcessingOperation::MOV: conditions = destination = operand2; break;
case DataProcessingOperation::MVN: conditions = destination = ~operand2; break;
case DataProcessingOperation::TST: conditions = operand1 & operand2; break;
case DataProcessingOperation::TEQ: conditions = operand1 ^ operand2; break;
case DataProcessingOperation::ADD:
case DataProcessingOperation::ADC:
case DataProcessingOperation::CMN:
conditions = operand1 + operand2;
if constexpr (flags.operation() == DataProcessingOperation::ADC) {
conditions += registers_.c();
}
if constexpr (flags.set_condition_codes()) {
registers_.set_c(Numeric::carried_out<true, 31>(operand1, operand2, conditions));
registers_.set_v(Numeric::overflow<true>(operand1, operand2, conditions));
}
if constexpr (!is_comparison(flags.operation())) {
destination = conditions;
}
break;
case DataProcessingOperation::SUB:
case DataProcessingOperation::SBC:
case DataProcessingOperation::CMP:
conditions = operand1 - operand2;
if constexpr (flags.operation() == DataProcessingOperation::SBC) {
conditions -= registers_.c();
}
if constexpr (flags.set_condition_codes()) {
registers_.set_c(Numeric::carried_out<false, 31>(operand1, operand2, conditions));
registers_.set_v(Numeric::overflow<false>(operand1, operand2, conditions));
}
if constexpr (!is_comparison(flags.operation())) {
destination = conditions;
}
break;
case DataProcessingOperation::RSB:
case DataProcessingOperation::RSC:
conditions = operand2 - operand1;
if constexpr (flags.operation() == DataProcessingOperation::RSC) {
conditions -= registers_.c();
}
if constexpr (flags.set_condition_codes()) {
registers_.set_c(Numeric::carried_out<false, 31>(operand2, operand1, conditions));
registers_.set_v(Numeric::overflow<false>(operand2, operand1, conditions));
}
destination = conditions;
break;
}
if constexpr (flags.set_condition_codes()) {
// "When Rd is a register other than R15, the condition code flags in the PSR may be
// updated from the ALU flags as described above. When Rd is R15 and the S flag in
// the instruction is set, the PSR is overwritten by the corresponding ALU result.
//
// ... if the instruction is of a type which does not normally produce a result
// (CMP, CMN, TST, TEQ) but Rd is R15 and the S bit is set, the result will be used in
// this case to update those PSR flags which are not protected by virtue of the
// processor mode."
if(fields.destination() == 15) {
if constexpr (is_comparison(flags.operation())) {
registers_.set_status(pc_proxy);
} else {
registers_.set_status(pc_proxy);
registers_.set_pc(pc_proxy);
}
} else {
// Set N and Z in a unified way.
registers_.set_nz(conditions);
// Set C from the barrel shifter if applicable.
if constexpr (shift_sets_carry) {
registers_.set_c(rotate_carry);
}
}
} else {
// "If the S flag is clear when Rd is R15, only the 24 PC bits of R15 will be written."
if(fields.destination() == 15) {
registers_.set_pc(pc_proxy);
}
}
}
template <Flags f> void perform(Multiply fields) {
constexpr MultiplyFlags flags(f);
// R15 rules:
//
// * Rs: no PSR, 8 bytes ahead;
// * Rn: with PSR, 8 bytes ahead;
// * Rm: with PSR, 12 bytes ahead.
const uint32_t multiplicand = fields.multiplicand() == 15 ? registers_.pc(8) : registers_.active[fields.multiplicand()];
const uint32_t multiplier = fields.multiplier() == 15 ? registers_.pc_status(8) : registers_.active[fields.multiplier()];
const uint32_t accumulator =
flags.operation() == MultiplyOperation::MUL ? 0 :
(fields.multiplicand() == 15 ? registers_.pc_status(12) : registers_.active[fields.accumulator()]);
const uint32_t result = multiplicand * multiplier + accumulator;
if constexpr (flags.set_condition_codes()) {
registers_.set_nz(result);
// V is unaffected; C is undefined.
}
if(fields.destination() != 15) {
registers_.active[fields.destination()] = result;
}
}
template <Operation, Flags> void perform(Condition, SingleDataTransfer) {}
template <Operation, Flags> void perform(Condition, BlockDataTransfer) {}
template <Operation op> void perform(Condition condition, Branch branch) {
printf("Branch %sif %d; add %08x\n", op == Operation::BL ? "with link " : "", int(condition), branch.offset());
}
template <Operation, Flags> void perform(Condition, CoprocessorRegisterTransfer) {}
template <Flags> void perform(Condition, CoprocessorDataOperation) {}
template<Operation, Flags> void perform(Condition, CoprocessorDataTransfer) {}
void software_interrupt(Condition) {}
void unknown(uint32_t) {}
private:
Registers registers_;
};
}
@interface ARMDecoderTests : XCTestCase
@end
@implementation ARMDecoderTests
- (void)testXYX {
Scheduler scheduler;
for(int c = 0; c < 65536; c++) {
InstructionSet::ARM::dispatch(c << 16, scheduler);
}
InstructionSet::ARM::dispatch(0xEAE06900, scheduler);
}
@end