aboutsummaryrefslogtreecommitdiff
diff options
context:
space:
mode:
Diffstat
-rw-r--r--Makefile2
-rw-r--r--src/as_analyzer.c9
-rw-r--r--src/as_analyzer.h1
-rw-r--r--src/as_codegen.c112
-rw-r--r--src/as_codegen.h2
-rw-r--r--src/as_main.c36
-rw-r--r--src/as_op.c19
-rw-r--r--src/as_op.h2
-rw-r--r--src/as_tokenizer.c2
9 files changed, 145 insertions, 40 deletions
diff --git a/Makefile b/Makefile
index fdeaf9f..5c99593 100644
--- a/Makefile
+++ b/Makefile
@@ -11,7 +11,7 @@ obj = $(src:.c=.o)
tests=$(shell find tests/ -name '*.c')
tests_bin=$(tests:.c=.bin)
-all: $(target) # fvm-as
+all: $(target) fvm-as
fvm-as: $(obj) src/as_main.c
$(cc) $(cflags) $(ldflags) $^ -o $@
diff --git a/src/as_analyzer.c b/src/as_analyzer.c
index f48353f..5a75794 100644
--- a/src/as_analyzer.c
+++ b/src/as_analyzer.c
@@ -53,6 +53,15 @@ void sym_table_add(struct allocator * alct, struct sym_table* tbl, const char* n
tbl->size += 1;
}
+int sym_table_lookup(sym_table* tbl, const char* name) {
+ for (int i = 0; i < tbl->size; i++) {
+ if (strcmp(name, tbl->buf[i].name) == 0) {
+ return tbl->buf[i].offset;
+ }
+ }
+ return -1;
+}
+
struct sym_table analyze_prog(struct allocator * alct, struct prog * prog) {
process_section_label(alct, prog);
struct stmt * * stmts = prog->stmts->stmts;
diff --git a/src/as_analyzer.h b/src/as_analyzer.h
index 58a9203..a1f781e 100644
--- a/src/as_analyzer.h
+++ b/src/as_analyzer.h
@@ -19,6 +19,7 @@ typedef struct sym_table sym_table;
sym_table new_sym_table(allocator* alct);
void sym_table_add(allocator* alct, sym_table* tbl, const char* name, int pos);
+int sym_table_lookup(sym_table* tbl, const char* name);
sym_table analyze_prog(allocator* alct, prog* prog);
diff --git a/src/as_codegen.c b/src/as_codegen.c
index 5c1d60e..678733f 100644
--- a/src/as_codegen.c
+++ b/src/as_codegen.c
@@ -1,12 +1,38 @@
#include "as_codegen.h"
+
+#include <string.h>
+
#include "as_analyzer.h"
#include "as_op.h"
#include "fvm.h"
#include "utils.h"
-bytearray *new_bytearray(allocator *alct);
-void bytearray_emit8(bytearray *self, int8_t data);
-void bytearray_emit64(bytearray *self, int64_t data);
+bytearray *new_bytearray(allocator *alct) {
+ bytearray* arr = allocate(alct, sizeof(bytearray));
+ arr->len = 0;
+ arr->cap = 16;
+ arr->buf = allocate(alct, 16);
+ arr->alct = alct;
+ return arr;
+}
+
+void bytearray_emit8(bytearray *arr, int8_t data) {
+ if (arr->len == arr->cap) {
+ void* oldbuf = arr->buf;
+ arr->buf = allocate(arr->alct, 2 * arr->cap);
+ memcpy(arr->buf, oldbuf, arr->cap);
+ arr->cap = arr->cap * 2;
+ }
+ arr->buf[arr->len] = data;
+ arr->len++;
+}
+
+void bytearray_emit64(bytearray *self, int64_t data) {
+ int8_t* ptr = (int8_t*)&data;
+ for (int i = 0; i < 8; i++) {
+ bytearray_emit8(self, ptr[i]);
+ }
+}
int8_t op_bytecode(op op) {
@@ -43,6 +69,12 @@ int8_t op_bytecode(op op) {
if (op == OP_XOR) return (int8_t)FVM_OP_XOR;
if (op == OP_INVERT) return (int8_t)FVM_OP_INVERT;
+ if (op == OP_JNZ) return (int8_t)FVM_OP_JNZ;
+ if (op == OP_JZ) return (int8_t)FVM_OP_JZ;
+ if (op == OP_JMP) return (int8_t)FVM_OP_JMP;
+ if (op == OP_CALL) return (int8_t)FVM_OP_CALL;
+ if (op == OP_SYSCALL) return (int8_t)FVM_OP_SYSCALL;
+
if (op == OP_GT) return (int8_t)FVM_OP_GT;
if (op == OP_LT) return (int8_t)FVM_OP_LT;
if (op == OP_GE) return (int8_t)FVM_OP_GE;
@@ -86,22 +118,66 @@ result codegen(allocator* alct, prog* prog, sym_table tbl) {
int8_t code = op_bytecode(op);
bytearray_emit8(output, code);
offset += 1;
- } else if (op_size(op) == 9) {
- if (op == OP_IMM) {
- if (instr->tag_name != NULL || instr->arg == NULL) {
- return err(safe_sprintf(alct, "line %d: invalid instruction format. (imm)\n", instr->lineno));
- }
- bytearray_emit8(output, FVM_OP_IMM);
- bytearray_emit64(output, instr->arg->ival);
- offset += 9;
- } else if (op == OP_REL) {
- if (instr->tag_name == NULL || instr->arg != NULL) {
- return err(safe_sprintf(alct, "line %d: invalid instruction format. (rel)\n", instr->lineno));
- }
- bytearray_emit8(output, FVM_OP_IMM);
+ } else if (op == OP_IMM) {
+ if (instr->tag_name != NULL || instr->arg == NULL) {
+ return err(safe_sprintf(alct, "line %d: invalid instruction format. (imm)\n", instr->lineno));
}
- } else if (op_size(op) == 10) {
-
+ bytearray_emit8(output, FVM_OP_IMM);
+ bytearray_emit64(output, instr->arg->ival);
+ offset += 9;
+ } else if (op == OP_REL) {
+ if (instr->tag_name == NULL || instr->arg != NULL) {
+ return err(safe_sprintf(alct, "line %d: invalid instruction format. (rel)\n", instr->lineno));
+ }
+ bytearray_emit8(output, FVM_OP_IMM);
+ int target_offset = sym_table_lookup(&tbl, instr->tag_name);
+ if (target_offset == -1) {
+ return err(safe_sprintf(alct, "line %d: unknown tag: %s", instr->lineno, instr->tag_name));
+ }
+ bytearray_emit64(output, target_offset - (offset + 9));
+ bytearray_emit8(output, FVM_OP_PC);
+ bytearray_emit8(output, FVM_OP_ADD);
+ offset += 11;
+ } else if (op == OP_LDARG) {
+ if (instr->tag_name != NULL || instr->arg == NULL) {
+ return err(safe_sprintf(alct, "line %d: invalid instruction format. (ldarg)\n", instr->lineno));
+ }
+ bytearray_emit8(output, FVM_OP_IMM);
+ bytearray_emit64(output, 8 * (instr->arg->ival + 2));
+ bytearray_emit8(output, FVM_OP_BP);
+ bytearray_emit8(output, FVM_OP_ADD);
+ bytearray_emit8(output, FVM_OP_LD);
+ offset += 12;
+ } else if (op == OP_STARG) {
+ if (instr->tag_name != NULL || instr->arg == NULL) {
+ return err(safe_sprintf(alct, "line %d: invalid instruction format. (starg)\n", instr->lineno));
+ }
+ bytearray_emit8(output, FVM_OP_IMM);
+ bytearray_emit64(output, 8 * (instr->arg->ival + 2));
+ bytearray_emit8(output, FVM_OP_BP);
+ bytearray_emit8(output, FVM_OP_ADD);
+ bytearray_emit8(output, FVM_OP_ST);
+ offset += 12;
+ } else if (op == OP_LDVAR) {
+ if (instr->tag_name != NULL || instr->arg == NULL) {
+ return err(safe_sprintf(alct, "line %d: invalid instruction format. (ldvar)\n", instr->lineno));
+ }
+ bytearray_emit8(output, FVM_OP_IMM);
+ bytearray_emit64(output, 8 * (-instr->arg->ival - 1));
+ bytearray_emit8(output, FVM_OP_BP);
+ bytearray_emit8(output, FVM_OP_ADD);
+ bytearray_emit8(output, FVM_OP_LD);
+ offset += 12;
+ } else if (op == OP_STVAR) {
+ if (instr->tag_name != NULL || instr->arg == NULL) {
+ return err(safe_sprintf(alct, "line %d: invalid instruction format. (stvar)\n", instr->lineno));
+ }
+ bytearray_emit8(output, FVM_OP_IMM);
+ bytearray_emit64(output, 8 * (-instr->arg->ival - 2));
+ bytearray_emit8(output, FVM_OP_BP);
+ bytearray_emit8(output, FVM_OP_ADD);
+ bytearray_emit8(output, FVM_OP_ST);
+ offset += 12;
}
}
return ok(output);
diff --git a/src/as_codegen.h b/src/as_codegen.h
index 1d7392a..4265f07 100644
--- a/src/as_codegen.h
+++ b/src/as_codegen.h
@@ -8,12 +8,14 @@ struct bytearray {
size_t cap;
size_t len;
char* buf;
+ allocator *alct;
};
typedef struct bytearray bytearray;
void bytearray_emit8(bytearray *self, int8_t data);
void bytearray_emit64(bytearray *self, int64_t data);
+// return: result<bytearray*>
struct result codegen(struct allocator *alct, struct prog *prog, struct sym_table tbl);
#endif // FVM_AS_CODEGEN_H_
diff --git a/src/as_main.c b/src/as_main.c
index 6c91a5f..cd9251b 100644
--- a/src/as_main.c
+++ b/src/as_main.c
@@ -1,18 +1,42 @@
#include <stdio.h>
#include "as_tokenizer.h"
+#include "as_analyzer.h"
+#include "as_parser.h"
+#include "as_codegen.h"
#include "utils.h"
-int main(int argc, char** argv) {
- if (argc != 2) {
- fprintf(stderr, "usage: fvm-as <inputfile>\n");
- return 1;
+result main_impl(int argc, char** argv) {
+ if (argc != 3) {
+ return err("usage: fvm-as <input_file> <output_file>");
}
struct allocator * alct = new_allocator();
FILE *fp = fopen(argv[1], "r");
- struct token_stream * ts = new_token_stream(alct, fp);
+ token_stream * ts = new_token_stream(alct, fp);
+ prog* prog = unwrap(parse_prog(alct, ts));
+ sym_table symtbl = analyze_prog(alct, prog);
+ bytearray* output = unwrap(codegen(alct, prog, symtbl));
+ fclose(fp);
+
+ fp = fopen(argv[2], "wb");
+ if (fp == NULL) {
+ return err("open output file failed.");
+ }
+ int ret =fwrite(output->buf, 1, output->len, fp);
+ if (ret != output->len) {
+ return err("write output file failed.");
+ }
+ fclose(fp);
delete_allocator(alct);
- return 0;
+ return ok(NULL);
}
+int main(int argc, char** argv) {
+ result result = main_impl(argc, argv);
+ if (result.errmsg != NULL) {
+ fprintf(stderr, "%s\n", result.errmsg);
+ return -1;
+ }
+ return 0;
+}
diff --git a/src/as_op.c b/src/as_op.c
index 71ba0f3..a50812a 100644
--- a/src/as_op.c
+++ b/src/as_op.c
@@ -94,7 +94,7 @@ struct op_table_entry op_table [] = {
};
int op_size(enum op op) {
- if (op == OP_IMM || op == OP_REL) {
+ if (op == OP_IMM) {
return 8 + 1;
}
if (op == OP_CALL
@@ -104,23 +104,16 @@ int op_size(enum op op) {
|| op == OP_SYSCALL) {
return 8 + 1 + 1;
}
- if (op == OP_LDARG || op == OP_STARG) {
- return 9 + 9 + 1 + 9 + 4;
- }
- if (op == OP_LDVAR || op == OP_STVAR) {
- return 9 + 9 + 2 + 9 + 4;
+ if (op == OP_REL) {
+ return 11;
}
- return 1;
-}
-
-int is_pseudo_op(enum op op) {
if (op == OP_LDARG || op == OP_STARG) {
- return 1;
+ return 12;
}
if (op == OP_LDVAR || op == OP_STVAR) {
- return 1;
+ return 12;
}
- return 0;
+ return 1;
}
enum op str2op(const char* str) {
diff --git a/src/as_op.h b/src/as_op.h
index 3dd546e..90080d0 100644
--- a/src/as_op.h
+++ b/src/as_op.h
@@ -23,7 +23,7 @@ enum op {
typedef enum op op;
enum op str2op(const char *str);
-int isOp(const char *str);
+int is_op(const char *str);
int op_size(op op);
int is_pseudo_op(op op);
diff --git a/src/as_tokenizer.c b/src/as_tokenizer.c
index aa22000..1651ccb 100644
--- a/src/as_tokenizer.c
+++ b/src/as_tokenizer.c
@@ -124,7 +124,7 @@ result next_token_impl(allocator* alct, input_stream* s) {
}
}
sval[i] = '\0';
- if (isOp(sval)) {
+ if (is_op(sval)) {
*t = (struct token){.type = TK_OP, .sval = sval, .line = line, .col = col};
return ok(t);
}
generated by cgit v1.0 at 2025-12-29 12:03:46 +0000