update generator to support multipe isas, expand grammer syntax

2 files changed, 383 insertions, 56 deletions

diff --git a/masm/gen.c b/masm/gen.c
index ba6960b..c4a15a0 100644
--- a/masm/gen.c
+++ b/masm/gen.c
@@ -1,7 +1,9 @@
 #include <stdlib.h>
 #include <merror.h>
 #include <melf.h>
-#include <mips32.h>
+#include <mips.h>
+#include <mips1.h>
+#include <mips32r2.h>
 #include <mips32r6.h>
 
 #include "tab.h"
@@ -263,30 +265,6 @@ static int gen_constant(struct generator *gen, struct expr_const *const expr)
 	return M_ERROR;
 }
 
-static enum grammer_type get_gmr_type(const char *name, size_t *len)
-{
-	#define CHK(part, str) { \
-		if (strncasecmp(str, name, strlen(str)) == 0) { \
-			*len = strlen(str); \
-			return GMR_ ##part; \
-		}} \
-
-	CHK(RD, "rd")
-	CHK(RS, "rs")
-	CHK(RT, "rt")
-	CHK(IMMD, "immd")
-	CHK(OFFSET_BASE, "offset(base)")
-	CHK(OFFSET, "offset")
-	CHK(TARGET, "target")
-	CHK(HI, "hi")
-	CHK(LO, "lo")
-
-	#undef CHK
-
-	ERROR("!!! BUG: this should never hit !!!");
-	exit(1);
-}
-
 static int parse_register(enum mips32_register *reg, struct string *name)
 {
 	int len = name->len;
@@ -403,6 +381,141 @@ static int parse_register(enum mips32_register *reg, struct string *name)
 	return M_ERROR;
 }
 
+static int parse_fp_register(enum mips32_fp_register *reg, struct string *name)
+{
+	int len = name->len;
+	int c0 = len > 0 ? name->str[0] : '\0',
+	    c1 = len > 1 ? name->str[1] : '\0',
+	    c2 = len > 2 ? name->str[2] : '\0';
+
+	*reg = 0;
+
+	if (c0 != 'f')
+		goto error;
+
+	if (c1 < '0' || c1 > '9')
+		goto error;
+
+	*reg += c1 - '0';
+
+	if (c1 == '\0')
+		return M_SUCCESS;
+
+	if (c2 < '0' || c2 > '9')
+		goto error;
+
+	*reg *= 10;
+	*reg += c2 - '0';
+
+	return M_SUCCESS;
+
+error:
+	ERROR("unknown fp register $%.*s", name->len, name->str);
+	return M_ERROR;
+
+}
+
+static enum grammer_type get_gmr_type(const char *name, size_t *len)
+{
+	#define CHK(part, str) { \
+		if (strncasecmp(str, name, strlen(str)) == 0) { \
+			*len = strlen(str); \
+			return GMR_ ##part; \
+		}} \
+
+	CHK(RD, "rd")
+	CHK(RS, "rs")
+	CHK(RT, "rt")
+	CHK(FS, "fs")
+	CHK(FT, "ft")
+	CHK(FD, "fd")
+	CHK(IMMD, "immd")
+	CHK(CC, "cc")
+	CHK(CODE, "code")
+	CHK(POS, "pos")
+	CHK(SIZE, "size")
+	CHK(HB, "hb")
+	CHK(HINT, "hint")
+	CHK(OFFSET_BASE, "offset(base)")
+	CHK(INDEX_BASE, "index(base)")
+	CHK(OFFSET, "offset")
+	CHK(TARGET, "target")
+	CHK(HI, "hi")
+	CHK(LO, "lo")
+
+	#undef CHK
+
+	BUG("invalid grammer key: %s", name);
+	exit(1);
+}
+
+static int get_gmr_hc(struct gen_ins_state *state,
+		      struct gen_ins_override *over,
+		      char *value)
+{
+	int vlen = 0;
+	int res = M_SUCCESS;
+
+	/* get length of value */
+	for (const char *ptr = value;
+		*ptr != '\0' && *ptr != ',';
+		ptr++, vlen++);
+
+	/* must be at least of length 1 */
+	if (vlen < 1)
+		return M_ERROR;
+
+	#define VAL(v) if (strncmp(v, value, vlen) == 0)
+	/* register to register mapping */
+	VAL("rd")
+		over->reg = state->rd;
+	else VAL("rs")
+		over->reg = state->rs;
+	else VAL("rt")
+	{
+		over->reg = state->rt;
+	}
+	else VAL("fs")
+		over->fpreg = state->fs;
+	else VAL("ft")
+		over->fpreg = state->ft;
+	else VAL("fd")
+		over->fpreg = state->fd;
+	else VAL("immd")
+		over->immd = state->immd;
+	else VAL("-immd")
+		over->immd = -state->immd;
+	/* hardcoded register */
+	else if (*value == '$') {
+		// register
+		if (vlen < 2)
+			return M_ERROR;
+		struct string name;
+		name.str = value + 1;
+		name.len = vlen - 1;
+		if (*name.str == 'f')
+			res = parse_fp_register(&over->fpreg, &name);
+		else
+			res = parse_register(&over->reg, &name);
+	/* hardcoded immediate */
+	} else {
+		// immediate
+		char c;
+		int immd = 0;
+		while (c = *(value++), c != ',' && c != '\0') {
+			if (c < '0' || c > '9')
+				return M_ERROR;
+			immd *= 10;
+			immd += c - '0';
+		}
+		over->immd = immd;
+	}
+
+	#undef KEY
+
+	return res;
+}
+
 static int gen_ins_read_state(struct generator *gen,
 			      struct expr *const expr,
 			      struct gen_ins_state *state,
@@ -459,6 +572,42 @@ static int gen_ins_read_state(struct generator *gen,
 				return M_ERROR;
 			}
 			break;
+		case GMR_FS:
+			// fs
+			if (arg->type != EXPR_INS_ARG_REGISTER) {
+				ERROR("expected a register");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			if (parse_fp_register(&state->fs, &arg->reg)) {
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			break;
+		case GMR_FT:
+			// ft
+			if (arg->type != EXPR_INS_ARG_REGISTER) {
+				ERROR("expected a register");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			if (parse_fp_register(&state->ft, &arg->reg)) {
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			break;
+		case GMR_FD:
+			// fd
+			if (arg->type != EXPR_INS_ARG_REGISTER) {
+				ERROR("expected a register");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			if (parse_fp_register(&state->fd, &arg->reg)) {
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			break;
 		case GMR_IMMD:
 			// immd
 			if (arg->type != EXPR_INS_ARG_IMMEDIATE) {
@@ -468,6 +617,61 @@ static int gen_ins_read_state(struct generator *gen,
 			}
 			state->immd = arg->immd;
 			break;
+		case GMR_CC:
+			// cc
+			if (arg->type != EXPR_INS_ARG_IMMEDIATE) {
+				ERROR("expected an immediate");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			state->cc = arg->immd;
+			break;
+		case GMR_CODE:
+			// code
+			if (arg->type != EXPR_INS_ARG_IMMEDIATE) {
+				ERROR("expected an immediate");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			state->code = arg->immd;
+			break;
+		case GMR_POS:
+			// pos
+			if (arg->type != EXPR_INS_ARG_IMMEDIATE) {
+				ERROR("expected an immediate");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			state->pos = arg->immd;
+			break;
+		case GMR_SIZE:
+			// size
+			if (arg->type != EXPR_INS_ARG_IMMEDIATE) {
+				ERROR("expected an immediate");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			// TODO: check valid size
+			state->size = arg->immd;
+			break;
+		case GMR_HB:
+			// hb
+			if (arg->type != EXPR_INS_ARG_IMMEDIATE) {
+				ERROR("expected an immediate");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			state->hb = !!(arg->immd);
+			break;
+		case GMR_HINT:
+			// hint
+			if (arg->type != EXPR_INS_ARG_IMMEDIATE) {
+				ERROR("expected an immediate");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			state->hint = arg-> immd;
+			break;
 		case GMR_OFFSET:
 			// offset
 			state->offset = 0;
@@ -494,6 +698,19 @@ static int gen_ins_read_state(struct generator *gen,
 				return M_ERROR;
 			}
 			break;
+		case GMR_INDEX_BASE:
+			// index(base)
+			if (arg->type != EXPR_INS_ARG_OFFSET) {
+				ERROR("expected an index($base)");
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			state->index = arg->offset.immd;
+			if (parse_register(&state->base, &arg->offset.reg)) {
+				print_curr_line(gen, expr);
+				return M_ERROR;
+			}
+			break;
 		case GMR_TARGET:
 			// target
 			state->target = 0;
@@ -521,7 +738,7 @@ static int gen_ins_read_state(struct generator *gen,
 		} else if (*ptr == '\0') {
 			break;
 		} else {
-			ERROR("!! BUG3: invalid splitting char %c !!!", *ptr);
+			BUG("invalid grammer '%s'", grammer);
 			exit(1);
 		}
 
@@ -536,27 +753,31 @@ static int gen_ins_write_state(
 	struct gen_ins_state *state,  // the current read state
 	char *grammer)		      // the gramemr to parse
 {
+	// grammer pointer
 	char *ptr = grammer;
+
+	// new reference type if needed
 	enum reference_type reftype = REF_NONE;
 
+	// hardcoded
+	struct gen_ins_override hstate;
+	bool hc = false;
+
 	// read values into state
 	while (*ptr != '\0') {
-
 		// parse next dsl entry
 		size_t skip;
 		enum grammer_type gmr = get_gmr_type(ptr, &skip);
 
 		// check for dsl hardcoded register argument
-		bool hardcoded = false;
-		enum mips32_register hard_reg;
+		hc = false;
 		if (*(ptr + skip) == '=') {
-			// parse argument
-			char *rptr = ptr + skip + 2;
-			hardcoded = true;
-			struct string regname;
-			string_bss(&regname, rptr);
-			if (parse_register(&hard_reg, &regname)) {
-				ERROR("!!! BUG2: this should never hit !!!");
+			// get value pointer
+			char *value = ptr + skip + 1;
+			hc = true;
+			// parse value
+			if (get_gmr_hc(state, &hstate, value)) {
+				BUG("invalid grammer: %s", grammer);
 				exit(1);
 			}
 		}
@@ -568,16 +789,43 @@ static int gen_ins_write_state(
 
 		switch (gmr) {
 		case GMR_RD:
-			ins.rd = hardcoded ? hard_reg : state->rd;
+			ins.rd = hc ? hstate.reg : state->rd;
 			break;
 		case GMR_RS:
-			ins.rs = hardcoded ? hard_reg : state->rs;
+			ins.rs = hc ? hstate.reg : state->rs;
 			break;
 		case GMR_RT:
-			ins.rt = hardcoded ? hard_reg : state->rt;
+			ins.rt = hc ? hstate.reg : state->rt;
+			break;
+                case GMR_FS:
+			ins.fs = hc ? hstate.fpreg : state->fs;
+			break;
+                case GMR_FT:
+			ins.ft = hc ? hstate.fpreg : state->ft;
+			break;
+                case GMR_FD:
+			ins.fd = hc ? hstate.fpreg : state->fd;
 			break;
 		case GMR_IMMD:
-			ins.immd = state->immd;
+			ins.immd = hc ? hstate.immd : state->immd;
+			break;
+                case GMR_CC:
+			ins.cc = state->cc;
+			break;
+                case GMR_CODE:
+			ins.code = state->code;
+			break;
+		case GMR_POS:
+			ins.shamt = state->pos;
+			break;
+		case GMR_SIZE:
+			ins.rd = state->size ? state->size - 1 : 0;
+			break;
+		case GMR_HB:
+			ins.hb = hc ? hstate.immd : state->hb;
+			break;
+		case GMR_HINT:
+			ins.rt = state->hint;
 			break;
 		case GMR_OFFSET:
 			ins.offset = state->offset;
@@ -588,6 +836,10 @@ static int gen_ins_write_state(
 			ins.rs = state->base;
 			reftype = REF_MIPS_16;
 			break;
+		case GMR_INDEX_BASE:
+			ins.rt = state->index;
+			ins.rs = state->base;
+			break;
 		case GMR_TARGET:
 			ins.target = state->target;
 			reftype = REF_MIPS_26;
@@ -600,8 +852,9 @@ static int gen_ins_write_state(
 			ins.immd = state->target & 0x0000FFFF;
 			reftype = REF_MIPS_LO16;
 			break;
-		}
-	}
+                  break;
+                }
+        }
 
 	// get offset for reference (if needed)
 	uint32_t offset = gen->current->len;
@@ -721,8 +974,8 @@ static int gen_label(struct generator *gen, struct string *const label)
 	return M_SUCCESS;
 }
 
-/* run codegen */
-static int generate(struct generator *gen)
+/* run codegen for next expression */
+static int generate_next(struct generator *gen)
 {
 	struct expr expr;
 	int res = M_SUCCESS;
@@ -771,16 +1024,10 @@ static int generate(struct generator *gen)
 	return res;
 }
 
-/* run codegen with the mips32r6 specification */
-int generate_mips32r6(struct generator *gen)
+static int generate(struct generator *gen)
 {
-	gen->instructions_len = __MIPS32R6_INS_LEN;
-	gen->instructions = mips32r6_instructions;
-	gen->grammers_len = __MIPS32R6_GRAMMER_LEN;
-	gen->grammers = mips32r6_grammers;
-
 	int res;
-	while (res = generate(gen), 1) {
+	while (res = generate_next(gen), 1) {
 		if (res == M_ERROR)
 			return M_ERROR;
 		if (res == M_EOF)
@@ -790,6 +1037,36 @@ int generate_mips32r6(struct generator *gen)
 	return M_SUCCESS;
 }
 
+/* run codegen with the mips32r6 specification */
+int generate_mips32r6(struct generator *gen)
+{
+	gen->instructions_len = __MIPS32R6_INS_LEN;
+	gen->instructions = mips32r6_instructions;
+	gen->grammers_len = __MIPS32R6_GRAMMER_LEN;
+	gen->grammers = mips32r6_grammers;
+	return generate(gen);
+}
+
+/* run codegen with the mips32r2 specification */
+int generate_mips32r2(struct generator *gen)
+{
+	gen->instructions_len = __MIPS32R2_INS_LEN;
+	gen->instructions = mips32r2_instructions;
+	gen->grammers_len = __MIPS32R2_GRAMMER_LEN;
+	gen->grammers = mips32r2_grammers;
+	return generate(gen);
+}
+
+/* run codegen with the mips32r6 specification */
+int generate_mips1(struct generator *gen)
+{
+	gen->instructions_len = __MIPS1_INS_LEN;
+	gen->instructions = mips1_instructions;
+	gen->grammers_len = __MIPS1_GRAMMER_LEN;
+	gen->grammers = mips1_grammers;
+	return generate(gen);
+}
+
 int generator_init(const char *file, struct generator *gen)
 {
 	if (parser_init(file, &gen->parser))
diff --git a/masm/gen.h b/masm/gen.h
index 19f575c..42fbf50 100644
--- a/masm/gen.h
+++ b/masm/gen.h
@@ -4,7 +4,7 @@
 #define __GEN_H__
 
 #include <mlimits.h>
-#include <mips32.h>
+#include <mips.h>
 #include <stdint.h>
 
 #include "parse.h"
@@ -44,16 +44,42 @@ void section_free(struct section *section);
 ///
 
 struct gen_ins_state {
-	// rd,rst,rt
+	// rd,rs,rt
 	enum mips32_register rd;
 	enum mips32_register rs;
 	enum mips32_register rt;
 
+	// fs,ft
+	enum mips32_fp_register fs;
+	enum mips32_fp_register ft;
+	enum mips32_fp_register fd;
+
 	// immd
-	uint16_t immd;
+	uint16_t immd;  // 16 bit
+
+	// cc
+	uint16_t cc;    // 3 bit
+
+	// code
+	uint32_t code;  // 5 bit
+
+	// pos
+	uint32_t pos;   // 5 bit
+
+	// size
+	uint32_t size;  // 5 bit - 1
+
+	// hb
+	bool hb; // 1 bit - 1
+
+	// index
+	uint32_t index;
+
+	// hint
+	uint32_t hint;
 
 	// offset(base)
-	uint16_t offset;
+	uint16_t offset; // 16 bit
 	enum mips32_register base;
 
 	// target
@@ -63,15 +89,35 @@ struct gen_ins_state {
 	struct string *label;
 };
 
+struct gen_ins_override {
+	enum mips32_register reg;
+	enum mips32_fp_register fpreg;
+	uint32_t immd;
+};
+
 ///
 /// grammer type
 ///
 
 enum grammer_type {
+	// registers
 	GMR_RD,
 	GMR_RS,
 	GMR_RT,
+	GMR_INDEX_BASE,
+	// fp registers
+	GMR_FS,
+	GMR_FT,
+	GMR_FD,
+	// numeric fields
 	GMR_IMMD,
+	GMR_CC,
+	GMR_CODE,
+	GMR_POS,
+	GMR_SIZE,
+	GMR_HB,
+	GMR_HINT,
+	// addresses
 	GMR_OFFSET,
 	GMR_OFFSET_BASE,
 	GMR_TARGET,
@@ -108,6 +154,10 @@ struct generator {
 
 /* generate the input as mips32r6 */
 int generate_mips32r6(struct generator *gen);
+/* run codegen with the mips32r2 specification */
+int generate_mips32r2(struct generator *gen);
+/* run codegen with the mips32r6 specification */
+int generate_mips1(struct generator *gen);
 
 /* initalize a generator */
 int generator_init(const char *file, struct generator *gen);