Rewrite the parser for better structure, Add DPI parsing

This commit is contained in:
sBubshait 2024-06-13 17:24:09 +01:00
parent 38e5cd06fa
commit 995c6d02fa

View File

@ -5,18 +5,159 @@
#include <stdbool.h> #include <stdbool.h>
#include "parser.h" #include "parser.h"
#include "a64instruction/a64instruction.h" #include "a64instruction/a64instruction.h"
#include "tokeniser.c"
//takes input string, read from asm file and returns /** Prototypes */
//input as an a64 instruction void parse_instruction(char asmLine[], a64inst_instruction *instr);
static char *duplicateString(char *str);
void parseSingleTransfer(a64inst_instruction *instr, char *opcode, char *operandList[], int numOperands);
void parseBranch(a64inst_instruction *instr, char* opcode, char *operandList[]);
void calcluateAddressFormat(a64inst_instruction *instr, char *operandList[], int numOperands);
void parseDPImmediate(a64inst_instruction *inst, char *tokens[], int tokensCount);
/** Constants */
static const char *BRANCH_OPCODES[] = {"b", "br", "b.eq", "b.ne", "b.ge", "b.lt", "b.gt", "b.le", "b.al"};
static const char *SINGLE_TRANSFER_OPCODES[] = {"ldr", "str"};
static const char *WIDE_MOV_OPCODES[] = {"movn", "movz", "movz", "movk"};
static const char *ARITHMETIC_OPCODES[] = {"add", "adds", "sub", "subs"};
static const char *MULTIPLY_OPCODES[] = {"mul", "madd", "msub", "mneg"};
a64inst_instruction *parse(char **asmLines, int lineCount) {
a64inst_instruction *instructions = malloc(sizeof(a64inst_instruction) * lineCount);
int i = 0;
while (asmLines[i] != NULL) {
parse_instruction(asmLines[i], &instructions[i]);
i++;
}
return instructions;
}
static char *duplicateString(char *str) {
char *newStr = malloc(strlen(str) + 1);
strcpy(newStr, str);
return newStr;
}
static bool isStringIn(char *str, const char *arr[], int arrSize) {
for (int i = 0; i < arrSize; i++) {
if (strcmp(str, arr[i]) == 0) {
return true;
}
}
return false;
}
// If more than one occurance, return the last index
static int indexStringIn(char *str, const char *arr[], int arrSize) {
for (int i = arrSize - 1; i >= 0; i--) {
if (strcmp(str, arr[i]) == 0) {
return i;
}
}
return -1;
}
int isOperandRegister(char regStartChar) {
return((regStartChar == 'x') || (regStartChar == 'w'));
}
int classifyDPInst(char *operandList[]){
return(isOperandRegister(operandList[1][0]) &&
isOperandRegister(operandList[2][0]) &&
isOperandRegister(operandList[3][0]));
}
void classifyOpcode(char* opcode, a64inst_instruction *instr, char *tokens[], int tokensCount){
if (isStringIn(opcode, BRANCH_OPCODES, 9)) {
instr->type = a64inst_BRANCH;
if (strcmp(opcode, "br") == 0) {
instr->data.BranchData.BranchType = a64inst_REGISTER;
} else if (strcmp(opcode, "b") == 0) {
instr->data.BranchData.BranchType = a64inst_UNCONDITIONAL;
} else {
instr->data.BranchData.BranchType = a64inst_CONDITIONAL;
}
} else if (isStringIn(opcode, SINGLE_TRANSFER_OPCODES, 2)) {
instr->type = a64inst_SINGLETRANSFER;
if (*tokens[2] == '[') {
instr->data.SingleTransferData.SingleTransferOpType = a64inst_SINGLE_TRANSFER_SINGLE_DATA_TRANSFER;
instr->data.SingleTransferData.processOpData.singleDataTransferData.transferType = strcmp(opcode, "ldr") == 0;
} else {
instr->type = a64inst_LOADLITERAL;
}
} else if (classifyDPInst(tokens)) {
instr->type = a64inst_DPREGISTER;
} else {
instr->type = a64inst_DPIMMEDIATE;
}
}
void parse_instruction(char asmLine[], a64inst_instruction *instr) {
if (instr == NULL){
exit(EXIT_FAILURE);
}
if(strcmp(asmLine, HALT_ASM_CMD) == 0){
instr->type = a64inst_HALT;
return;
}
char *asmLineCopy = duplicateString(asmLine);
int tokensCount = 0;
char **tokens = tokenise(asmLineCopy, &tokensCount);
char *opcode = tokens[0];
if(strcmp(opcode, ".int") == 0){
// Directive
instr->type = a64inst_DIRECTIVE;
} else if(opcode[strlen(opcode)-1]== ':') {
// Label
instr->type = a64inst_LABEL;
opcode[strlen(opcode) - 1] = '\0'; // Remove the colon
instr->data.LabelData.label = opcode;
} else {
// Instruction
classifyOpcode(opcode, instr, tokens, tokensCount);
switch(instr->type){
case a64inst_BRANCH:
parseBranch(instr, opcode, tokens);
break;
case a64inst_SINGLETRANSFER:
parseSingleTransfer(instr, opcode, tokens, tokensCount);
calcluateAddressFormat(instr, tokens, tokensCount);
break;
case a64inst_LOADLITERAL:
parseSingleTransfer(instr, opcode, tokens, tokensCount);
break;
case a64inst_DPREGISTER:
//generate DP operands;
break;
case a64inst_DPIMMEDIATE:
parseDPImmediate(instr, tokens, tokensCount);
break;
default:
printf("Error: Invalid Instruction\n");
break;
}
}
/* TODO: FREE MEMORY! */
}
//TODO:
// - use string matching to get opcode, and operands (DONE)
// - check operand count (DONE)
// - match opcode to a64 struct types (DONE)
// - count operands and match type/values (DONE)
// - generate final a64inst and return (TODO: DP instrs)
// - ASK ABOUT OFFSET CALCULATION
// - CREATE FUNC TO TIDY UP OPERANDS IN DP
//takes inputted char array and returns the integer of the operand, skipping the first character //takes inputted char array and returns the integer of the operand, skipping the first character
//e.g. for a passed "R32", it skips the 'R' and returns 32 //e.g. for a passed "R32", it skips the 'R' and returns 32
@ -28,64 +169,61 @@ int getOperandNumber(char *operand){
return number; return number;
} }
int isOperandRegister(char regStartChar){
return((regStartChar == 'x') || (regStartChar == 'w'));
}
//calculate offsets from string void calcluateAddressFormat(a64inst_instruction *instr, char *tokens[], int tokenCount) {
void calcluateAddressFormat(a64inst_instruction *instr, char *operandList[], int numOperands){ assert(*tokens[2] == '[');
char baseRegParam[strlen(operandList[1])];
strcpy(baseRegParam, operandList[1]);
char *startptr = &baseRegParam[1];
int base = getOperandNumber(startptr);
instr->data.SingleTransferData.processOpData.singleDataTransferData.base = base;
if(operandList[2][strlen(operandList[2])-2] == '!'){ int operandCount = 0;
char **operands = tokeniseOperands(tokens[2], &operandCount);
int baseRegister = getOperandNumber(operands[0]);
instr->data.SingleTransferData.processOpData.singleDataTransferData.base = baseRegister;
if(operands[1][strlen(operands[1])-1] == '!') {
instr->data.SingleTransferData.processOpData.singleDataTransferData.addressingMode = a64inst_PRE_INDEXED; instr->data.SingleTransferData.processOpData.singleDataTransferData.addressingMode = a64inst_PRE_INDEXED;
instr->data.SingleTransferData.processOpData.singleDataTransferData.a64inst_addressingModeData.indexedOffset = getOperandNumber(operandList[2]); instr->data.SingleTransferData.processOpData.singleDataTransferData.a64inst_addressingModeData.indexedOffset = getOperandNumber(operands[1]);
} else if(operandList[1][strlen(operandList[1])-2] == ']') {
//post-indexed } else if(operands[1][strlen(operands[1])-1] == ']') {
char immOffset[strlen(operandList[2])+1]; // POST_INDEXED
strcpy(immOffset, operandList[2]);
instr->data.SingleTransferData.processOpData.singleDataTransferData.addressingMode = a64inst_POST_INDEXED; instr->data.SingleTransferData.processOpData.singleDataTransferData.addressingMode = a64inst_POST_INDEXED;
instr->data.SingleTransferData.processOpData.singleDataTransferData.a64inst_addressingModeData.indexedOffset = getOperandNumber(immOffset); instr->data.SingleTransferData.processOpData.singleDataTransferData.a64inst_addressingModeData.indexedOffset = getOperandNumber(tokens[3]);
} else if( (isOperandRegister(operandList[1][0]) == 1)
|| (isOperandRegister(operandList[2][0]) == 1)){ } else if( (isOperandRegister(*operands[0]) == 1)
&& (isOperandRegister(*operands[1]) == 1)){
//register //register
instr->data.SingleTransferData.processOpData.singleDataTransferData.addressingMode = a64inst_REGISTER_OFFSET; instr->data.SingleTransferData.processOpData.singleDataTransferData.addressingMode = a64inst_REGISTER_OFFSET;
instr->data.SingleTransferData.processOpData.singleDataTransferData.a64inst_addressingModeData.offsetReg = getOperandNumber(operandList[2]); instr->data.SingleTransferData.processOpData.singleDataTransferData.a64inst_addressingModeData.offsetReg = getOperandNumber(operands[1]);
} else { } else {
instr->data.SingleTransferData.processOpData.singleDataTransferData.addressingMode = a64inst_UNSIGNED_OFFSET; instr->data.SingleTransferData.processOpData.singleDataTransferData.addressingMode = a64inst_UNSIGNED_OFFSET;
if(numOperands==3){ if(operandCount > 1){
int offset = getOperandNumber(operandList[2]); int offset = getOperandNumber(operands[1]);
instr->data.SingleTransferData.processOpData.singleDataTransferData.a64inst_addressingModeData.unsignedOffset = offset/8; instr->data.SingleTransferData.processOpData.singleDataTransferData.a64inst_addressingModeData.unsignedOffset = offset/8;
//NEED TO SCALE IMMEDIATE VALUE BASED ON REGISTER TYPE IN ASSEMBLER //NEED TO SCALE IMMEDIATE VALUE BASED ON REGISTER TYPE IN ASSEMBLER
} }
} }
} }
void generateLoadStoreOperands(a64inst_instruction *instr, char *opcode, char *operandList[], int numOperands){ static int parseRegisterType(char *operand) {
return operand[0] == 'x';
}
void parseSingleTransfer(a64inst_instruction *instr, char *opcode, char *tokens[], int tokensCount) {
switch(instr->type){ switch(instr->type){
case a64inst_SINGLETRANSFER: { case a64inst_SINGLETRANSFER:
if(operandList[0][0] == 'x'){ instr->data.SingleTransferData.regType = parseRegisterType(tokens[1]);
//x-register instr->data.SingleTransferData.target = getOperandNumber(tokens[1]);
instr->data.SingleTransferData.regType = 1;
} else {
instr->data.SingleTransferData.regType = 0;
}
instr->data.SingleTransferData.target = getOperandNumber(operandList[0]);
break; break;
}
case a64inst_LOADLITERAL: case a64inst_LOADLITERAL:
if(operandList[0][0] == 'x') { instr->data.SingleTransferData.regType = parseRegisterType(tokens[1]);
instr->data.SingleTransferData.regType = 1; instr->data.SingleTransferData.target = getOperandNumber(tokens[1]);
} else {
instr->data.SingleTransferData.regType = 0; if(*tokens[2] =='#'){
}
instr->data.SingleTransferData.target = getOperandNumber(operandList[0]);
if(operandList[1][0] =='#'){
//offset is immediate //offset is immediate
int offset = getOperandNumber(operandList[1]); int offset = getOperandNumber(tokens[1]);
instr->data.SingleTransferData.processOpData.loadLiteralData.offset = offset; instr->data.SingleTransferData.processOpData.loadLiteralData.offset = offset;
} else { } else {
//offset is literal, use symbol table and calculate difference //offset is literal, use symbol table and calculate difference
@ -97,7 +235,7 @@ void generateLoadStoreOperands(a64inst_instruction *instr, char *opcode, char *o
} }
} }
void generateBranchOperands(a64inst_instruction *instr, char* opcode, char *operandList[]){ void parseBranch(a64inst_instruction *instr, char* opcode, char *operandList[]) {
switch(instr->data.BranchData.BranchType){ switch(instr->data.BranchData.BranchType){
case a64inst_UNCONDITIONAL: case a64inst_UNCONDITIONAL:
//define and sign extend immediate offset //define and sign extend immediate offset
@ -132,155 +270,62 @@ void generateBranchOperands(a64inst_instruction *instr, char* opcode, char *oper
} }
} }
int classifyDPInst(char *operandList[]){ void parseDPImmediate(a64inst_instruction *inst, char *tokens[], int tokensCount) {
return(isOperandRegister(operandList[0][0]) && a64inst_DPImmediateData *data = &inst->data.DPImmediateData;
isOperandRegister(operandList[1][0]) && data->dest = getOperandNumber(tokens[1]);
isOperandRegister(operandList[2][0])); data->regType = parseRegisterType(tokens[1]);
}
void classifyOpcode(char* opcode, a64inst_instruction *instr, char *operandList[], int numOperands){ if (isStringIn(tokens[0], WIDE_MOV_OPCODES, 3)) {
int isUnconditional = strcmp(opcode, "b"); data->DPIOpType = a64inst_DPI_WIDEMOV;
int isRegister = strcmp(opcode, "br"); data->processOp = indexStringIn(tokens[0], WIDE_MOV_OPCODES, 3);
int isLoad = strcmp(opcode, "ldr"); data->processOpData.wideMovData.immediate = getOperandNumber(tokens[2]);
int isStore = strcmp(opcode, "str"); if (tokensCount >= 3) {
data->processOpData.wideMovData.shiftScalar = getOperandNumber(tokens[3]);
if(isUnconditional == 0 ||
isRegister == 0 ||
strncmp(opcode, "b.", 2) == 0){
instr->type = a64inst_BRANCH;
if(isRegister == 0){
instr->data.BranchData.BranchType = a64inst_REGISTER;
} else if (isUnconditional == 0){
instr->data.BranchData.BranchType = a64inst_UNCONDITIONAL;
} else {
instr->data.BranchData.BranchType = a64inst_CONDITIONAL;
}
} else if(isLoad == 0 || isStore == 0){
//loading/storing instruction; classify operands
if( operandList[1][0] == '['){
//type is register
instr->type = a64inst_SINGLETRANSFER;
instr->data.SingleTransferData.SingleTransferOpType = a64inst_SINGLE_TRANSFER_SINGLE_DATA_TRANSFER;
if(isLoad == 0){
instr->data.SingleTransferData.processOpData.singleDataTransferData.transferType = a64inst_LOAD;
} else {
instr->data.SingleTransferData.processOpData.singleDataTransferData.transferType = a64inst_STORE;
}
} else {
instr->type = a64inst_LOADLITERAL;
} }
} else { } else {
if(classifyDPInst(operandList)){ data->DPIOpType = a64inst_DPI_ARITHM;
instr->type = a64inst_DPREGISTER; data->processOp = indexStringIn(tokens[0], ARITHMETIC_OPCODES, 4);
} else { data->processOpData.arithmData.src = getOperandNumber(tokens[2]);
instr->type = a64inst_DPIMMEDIATE; data->processOpData.arithmData.immediate = getOperandNumber(tokens[3]);
if (tokensCount >= 5) {
int shiftAmount = getOperandNumber(tokens[4]);
if (shiftAmount > 0) {
data->processOpData.arithmData.shiftImmediate = true;
}
} }
} }
} }
void tokeniseOperands(char* str, int *operandCount, char *operands[], int *numOperands){ void parseDPRegister(a64inst_instruction *inst, char *tokens[], int tokensCount) {
assert(str != NULL); a64inst_DPRegisterData *data = &inst->data.DPRegisterData;
char *operandsDupe = malloc(strlen(str)+1); data->dest = getOperandNumber(tokens[1]);
assert(operandsDupe != NULL); data->regType = parseRegisterType(tokens[1]);
strcpy(operandsDupe, str); data->src1 = getOperandNumber(tokens[2]);
char *operand = strtok(operandsDupe, OPERAND_DELIMITER); data->src2 = getOperandNumber(tokens[3]);
operands[0] = operand;
while (operand != NULL){ if (isStringIn(tokens[0], MULTIPLY_OPCODES, 4)) {
*operandCount = *(operandCount)+1; // Multiply
operand = strtok(NULL, OPERAND_DELIMITER); data->DPROpType = a64inst_DPR_MULTIPLY;
operands[*(operandCount)] = operand; if (tokensCount >= 5) {
} data->processOpData.multiplydata.summand = getOperandNumber(tokens[4]);
*(numOperands) = *(operandCount); data->processOpData.multiplydata.negProd = strcmp(tokens[4], "mneg") == 0;
}
//takes inputted assembly line and returns a
//pointer to an abstract representation of the instruction
void parser_instruction(char asmLine[], a64inst_instruction *instr) {
printf("%s", asmLine);
int numOperands = 0;
if (instr == NULL){
exit(EXIT_FAILURE);
} }
if(strcmp(asmLine, HALT_ASM_CMD) == 0){
instr->type = a64inst_HALT;
return;
}
//"opcode operand1, {operand2}, ..."
//duplicated as strtok modifies the input string
char stringptr[strlen(asmLine) + 1];
strcpy(stringptr, asmLine);
char *token;
token = strtok(stringptr, " ");
char opcode[strlen(token)+1];
strcpy(opcode, token);
token = strtok(NULL, "");
char operands[strlen(token)+1];
strcpy(operands, token);
if(strcmp(opcode, ".int") == 0){
//type is directive
instr->type = a64inst_DIRECTIVE;
} else if(opcode[strlen(opcode)-1]== ':') {
//type is label
//add to symbol table
instr->type = a64inst_LABEL;
char opcodeCpy[strlen(opcode)+1];
strcpy(opcodeCpy, opcode);
char *labelData = strtok(opcodeCpy, ":");
instr->data.LabelData.label = labelData;
} else { } else {
//type is instruction // Arithmetic/Logic
int operandCount = 0; data->DPROpType = a64inst_DPR_ARITHMLOGIC;
char *operandList[5];
//generate list of operands if (isStringIn(tokens[0], ARITHMETIC_OPCODES, 4)) {
tokeniseOperands(operands, &operandCount, operandList, &numOperands); // Arithmetic
//categorise instruction type from opcode and operands data->processOp = indexStringIn(tokens[0], ARITHMETIC_OPCODES, 4);
classifyOpcode(opcode, instr, operandList, operandCount); data->processOpData.arithmLogicData.type = 1;
//define struct values according to operands and type
printf("got to here"); } else {
switch(instr->type){ // Logic
case a64inst_BRANCH:
generateBranchOperands(instr, opcode, operandList);
break;
case a64inst_SINGLETRANSFER:
generateLoadStoreOperands(instr, opcode, operandList, numOperands);
calcluateAddressFormat(instr, operandList, numOperands);
break;
case a64inst_LOADLITERAL:
generateLoadStoreOperands(instr, opcode, operandList, numOperands);
break;
case a64inst_DPREGISTER:
//generate DP operands;
break;
case a64inst_DPIMMEDIATE:
//generate DP operands;
break;
default:
printf("INVALID INSTRUCTION");
break;
}
} }
}
}
// Takes an array of strings, each string representing an assembly instruction.
// Returns an array of a64inst_instruction pointers, each representing an instruction.
a64inst_instruction *parse(char **asmLines, int lineCount) {
a64inst_instruction *instructions = malloc(sizeof(a64inst_instruction) * lineCount);
int i = 0;
while (asmLines[i] != NULL) {
parser_instruction(asmLines[i], &instructions[i]);
i++;
}
return instructions;
} }