extract assembling line

Makefile

@@ -8,7 +8,6 @@ C_FLAGS = \
 	-Wall -Wextra -Wpedantic -Wconversion \
 	-pedantic -pedantic-errors \
 	-Wno-unused-variable \
-	-Wno-unused-parameter \
 	-I. \
 
 L_FLAGS = -pthread

asm/asm.h

@@ -114,7 +114,7 @@ typedef struct {
     uint16_t offset;
 } Line;
 
-uint16_t assemble_to_binary(
+uint16_t assemble_lines_with_labels(
     uint16_t* out, const Line* lines, size_t lines_size);
 
 Line s_label(int label);

asm/ctors.c

@@ -105,6 +105,7 @@ Line s_jnz_l(Reg op1_reg, int op2_label)
 {
     return (Line) {
         .ty = LineTy_Jnz_Label,
+        .op1 = (Ex) { .reg = (uint16_t)op1_reg },
         .op2 = (Ex) { .label = op2_label },
     };
 }

asm/main.c

@@ -37,6 +37,8 @@ typedef enum {
     TT_Int,
     TT_Binary,
     TT_Hex,
+    TT_Char,
+    TT_Str,
     TT_Newline = '\n',
     TT_DoubleLt,
     TT_DoubleGt,
@@ -119,6 +121,18 @@ static inline Tok lexer_tok(const Lexer* lexer, TokTy ty, Loc loc)
     return (Tok) { .ty = ty, .loc = loc, .len = lexer->idx - loc.idx };
 }
 
+static inline int lexer_skip_literal_char(Lexer* lexer)
+{
+    char ch = lexer->ch;
+    lexer_step(lexer);
+    if (ch == '\\') {
+        if (lexer_done(lexer))
+            return -1;
+        lexer_step(lexer);
+    }
+    return 0;
+}
+
 static inline void lexer_report(Lexer* lexer, const char* msg, Loc loc)
 {
     lexer->error_occured = true;
@@ -138,7 +152,7 @@ static inline void lexer_report(Lexer* lexer, const char* msg, Loc loc)
     int line_len = (int)line_end - (int)line_start;
 
     fprintf(stderr,
-        "error: %s\n  --> %s:%d:%d\n     |\n%5d|%.*s\n     |%*c^\n",
+        "error: %s\n    --> %s:%d:%d\n     |\n%5d|%.*s\n     |%*c^\n",
         msg,
         lexer->filename,
         loc.line,
@@ -146,14 +160,14 @@ static inline void lexer_report(Lexer* lexer, const char* msg, Loc loc)
         loc.line,
         line_len,
         line,
-        loc.col - 2,
+        loc.col - 1,
         ' ');
 }
 
 Tok lexer_next(Lexer* lexer)
 {
     const char* ident_chars = "abcdefghijklmnopqrstuvwxyz"
-                              "ABCDEFGHIJKLMNOPQRSTUVWXYZ_";
+                              "ABCDEFGHIJKLMNOPQRSTUVWXYZ_$";
     const char* int_chars = "1234567890";
     const char* hex_chars = "01234567889abcdefABCDEF";
 
@@ -212,6 +226,26 @@ Tok lexer_next(Lexer* lexer)
         } else {
             return lexer_tok(lexer, TT_Int, loc);
         }
+    } else if (lexer->ch == '\'') {
+        lexer_step(lexer);
+        lexer_skip_literal_char(lexer);
+        if (lexer_done(lexer) || lexer->ch != '\'') {
+            lexer_report(lexer, "malformed character literal", loc);
+            return lexer_tok(lexer, TT_Err, loc);
+        }
+        lexer_step(lexer);
+        return lexer_tok(lexer, TT_Char, loc);
+    } else if (lexer->ch == '"') {
+        lexer_step(lexer);
+        while (!lexer_done(lexer) && lexer->ch != '"') {
+            lexer_skip_literal_char(lexer);
+        }
+        if (lexer_done(lexer) || lexer->ch != '"') {
+            lexer_report(lexer, "malformed string literal", loc);
+            return lexer_tok(lexer, TT_Err, loc);
+        }
+        lexer_step(lexer);
+        return lexer_tok(lexer, TT_Str, loc);
     } else if (lexer->ch == '<') {
         lexer_step(lexer);
         if (!lexer_done(lexer) && lexer->ch == '<') {
@@ -235,9 +269,9 @@ Tok lexer_next(Lexer* lexer)
         lexer_step(lexer);
         return lexer_tok(lexer, (TokTy)ch, loc);
     } else {
-        lexer_report(lexer, "illegal character '%c'", loc);
+        lexer_report(lexer, "illegal character", loc);
         lexer_step(lexer);
-        return lexer_next(lexer);
+        return lexer_tok(lexer, TT_Err, loc);
     }
 }
 
@@ -293,6 +327,7 @@ typedef enum {
     PoTy_Imm,
     PoTy_Ident,
     PoTy_SubLabel,
+    PoTy_Str,
     PoTy_MemU8,
     PoTy_MemU16,
     PoTy_Not,
@@ -317,7 +352,7 @@ struct POperand {
     union {
         Reg reg;
         uint16_t imm;
-        char* ident;
+        char* str;
         POperand* operand;
         struct {
             POperand* left;
@@ -338,10 +373,10 @@ POperand* poperand_new_imm(uint16_t imm, Loc loc)
     *operand = (POperand) { .ty = PoTy_Imm, .loc = loc, .imm = imm };
     return operand;
 }
-POperand* poperand_new_ident(POperandTy ty, char* ident, Loc loc)
+POperand* poperand_new_str(POperandTy ty, char* str, Loc loc)
 {
     POperand* operand = malloc(sizeof(POperand));
-    *operand = (POperand) { .ty = ty, .loc = loc, .ident = ident };
+    *operand = (POperand) { .ty = ty, .loc = loc, .str = str };
     return operand;
 }
 POperand* poperand_new_unary(POperandTy ty, POperand* inner, Loc loc)
@@ -366,7 +401,8 @@ void poperand_free(POperand* operand)
             break;
         case PoTy_Ident:
         case PoTy_SubLabel:
-            free(operand->ident);
+        case PoTy_Str:
+            free(operand->str);
             break;
         case PoTy_MemU8:
         case PoTy_MemU16:
@@ -399,40 +435,19 @@ typedef struct {
     POperand* ops[];
 } PLine;
 
-PLine* pline_new_0(char* op, PLabel* labels, Loc loc)
+PLine* pline_new(
+    char* op, PLabel* labels, Loc loc, size_t ops_size, POperand** ops)
 {
-    PLine* line = malloc(sizeof(PLine) + sizeof(POperand*) * 0);
-    *line = (PLine) { .labels = labels, .op = op, .loc = loc, .ops_size = 0 };
-    return line;
-}
-PLine* pline_new_1(char* op, PLabel* labels, Loc loc, POperand* op0)
-{
-    PLine* line = malloc(sizeof(PLine) + sizeof(POperand*) * 1);
-    *line = (PLine) { .labels = labels, .op = op, .loc = loc, .ops_size = 1 };
-    line->ops[0] = op0;
-    return line;
-}
-PLine* pline_new_2(
-    char* op, PLabel* labels, Loc loc, POperand* op0, POperand* op1)
-{
-    PLine* line = malloc(sizeof(PLine) + sizeof(POperand*) * 2);
-    *line = (PLine) { .labels = labels, .op = op, .loc = loc, .ops_size = 2 };
-    line->ops[0] = op0;
-    line->ops[1] = op1;
-    return line;
-}
-PLine* pline_new_3(char* op,
-    PLabel* labels,
-    Loc loc,
-    POperand* op0,
-    POperand* op1,
-    POperand* op2)
-{
-    PLine* line = malloc(sizeof(PLine) + sizeof(POperand*) * 3);
-    *line = (PLine) { .labels = labels, .op = op, .loc = loc, .ops_size = 3 };
-    line->ops[0] = op0;
-    line->ops[1] = op1;
-    line->ops[2] = op2;
+    PLine* line = malloc(sizeof(PLine) + sizeof(POperand*) * ops_size);
+    *line = (PLine) {
+        .labels = labels,
+        .op = op,
+        .loc = loc,
+        .ops_size = ops_size,
+    };
+    for (size_t i = 0; i < ops_size; ++i) {
+        line->ops[i] = ops[i];
+    }
     return line;
 }
 void pline_free(PLine* pline)
@@ -515,7 +530,7 @@ static inline void parser_report(Parser* parser, const char* msg, Loc loc)
     int line_len = (int)line_end - (int)line_start;
 
     fprintf(stderr,
-        "error: %s\n  --> %s:%d:%d\n     |\n%5d|%.*s\n     |%*c^\n",
+        "error: %s\n    --> %s:%d:%d\n     |\n%5d|%.*s\n     |%*c^\n",
         msg,
         parser->lexer.filename,
         loc.line,
@@ -523,7 +538,7 @@ static inline void parser_report(Parser* parser, const char* msg, Loc loc)
         loc.line,
         line_len,
         line,
-        loc.col - 2,
+        loc.col - 1,
         ' ');
 }
 
@@ -572,6 +587,24 @@ static inline PLabel* parser_parse_labels(
     return labels;
 }
 
+static inline char literal_char_val(const char* str)
+{
+    if (str[0] == '\\') {
+        switch (str[1]) {
+            case '0':
+                return 0;
+            case 't':
+                return '\t';
+            case 'n':
+                return '\n';
+            default:
+                return str[1];
+        }
+    } else {
+        return str[0];
+    }
+}
+
 static const int parser_binary_prec = 6;
 static inline POperand* parser_parse_operand_2(Parser* parser, int prec);
 
@@ -590,7 +623,7 @@ static inline POperand* parser_parse_operand_0(Parser* parser)
                 return poperand_new_reg(reg_val[i], loc);
             }
         }
-        return poperand_new_ident(PoTy_Ident, ident, loc);
+        return poperand_new_str(PoTy_Ident, ident, loc);
     } else if (parser_eat(parser, TT_Int)) {
         char* str = parser_ident_val(parser, parser->eaten);
         uint16_t imm = (uint16_t)strtoul(str, NULL, 10);
@@ -606,13 +639,28 @@ static inline POperand* parser_parse_operand_0(Parser* parser)
         uint16_t imm = (uint16_t)strtoul(&str[2], NULL, 16);
         free(str);
         return poperand_new_imm(imm, loc);
+    } else if (parser_eat(parser, TT_Char)) {
+        char* str = parser_ident_val(parser, parser->eaten);
+        uint16_t imm = (uint16_t)literal_char_val(&str[1]);
+        free(str);
+        return poperand_new_imm(imm, loc);
+    } else if (parser_eat(parser, TT_Str)) {
+        char* lit = parser_ident_val(parser, parser->eaten);
+        size_t lit_len = strlen(lit);
+        char* str = calloc(lit_len - 1, sizeof(char));
+        size_t str_len = 0;
+        for (size_t i = 1; i < lit_len - 2; ++i) {
+            str[i] = literal_char_val(&lit[i]);
+        }
+        free(lit);
+        return poperand_new_str(PoTy_Str, str, loc);
     } else if (parser_eat(parser, '.')) {
         if (!parser_eat(parser, TT_Ident)) {
             parser_report(parser, "expected identifier", parser->tok.loc);
             return NULL;
         }
         char* ident = parser_ident_val(parser, parser->eaten);
-        return poperand_new_ident(PoTy_SubLabel, ident, loc);
+        return poperand_new_str(PoTy_SubLabel, ident, loc);
     } else if (parser_eat(parser, '(')) {
         POperand* operand = parser_parse_operand_2(parser, parser_binary_prec);
         if (!parser_eat(parser, ')')) {
@@ -697,13 +745,18 @@ static inline POperand* parser_parse_operand_2(Parser* parser, int prec)
 
 static inline POperand* parser_parse_operand_3(Parser* parser)
 {
+    Loc loc = parser->tok.loc;
+    if (parser_eat(parser, TT_LBracket)) {
+        parser_report(parser, "expected 'u8' or 'u16'", loc);
+        return NULL;
+    }
     if (!parser_test(parser, TT_Ident)) {
         return parser_parse_operand_2(parser, parser_binary_prec);
     }
-    Loc loc = parser->tok.loc;
-    char* ident = parser_ident_val(parser, parser->eaten);
+    char* ident = parser_ident_val(parser, parser->tok);
     if (strcmp(ident, "u8") == 0) {
         free(ident);
+        parser_step(parser);
         if (!parser_eat(parser, '[')) {
             parser_report(parser, "expected '['", parser->tok.loc);
             return NULL;
@@ -717,6 +770,7 @@ static inline POperand* parser_parse_operand_3(Parser* parser)
         return poperand_new_unary(PoTy_MemU8, operand, loc);
     } else if (strcmp(ident, "u16") == 0) {
         free(ident);
+        parser_step(parser);
         if (!parser_eat(parser, '[')) {
             parser_report(parser, "expected '['", parser->tok.loc);
             return NULL;
@@ -740,7 +794,9 @@ PLine* parser_next(Parser* parser)
     Loc loc;
     PLabel* labels = parser_parse_labels(parser, &ident, &loc);
 
-    POperand* ops[3];
+    const size_t max_ops_size = 64;
+    // TODO: Move allocation out-of-band.
+    POperand** ops = malloc(sizeof(POperand) * max_ops_size);
     size_t ops_size = 0;
 
     if (!parser_test(parser, TT_Eof) && !parser_test(parser, '\n')) {
@@ -751,6 +807,12 @@ PLine* parser_next(Parser* parser)
         ops[ops_size++] = operand;
         while (!parser_test(parser, TT_Eof) && !parser_test(parser, '\n')
             && ops_size < 3) {
+            if (ops_size >= max_ops_size) {
+                parser_report(parser,
+                    "exceeded maximum number of operands (64)",
+                    parser->tok.loc);
+                goto error_free_ops;
+            }
             if (!parser_eat(parser, ',')) {
                 parser_report(parser, "expected ','", parser->tok.loc);
                 goto error_free_ops;
@@ -768,22 +830,15 @@ PLine* parser_next(Parser* parser)
     }
     parser_skip_newlines(parser);
 
-    switch (ops_size) {
-        case 0:
-            return pline_new_0(ident, labels, loc);
-        case 1:
-            return pline_new_1(ident, labels, loc, ops[0]);
-        case 2:
-            return pline_new_2(ident, labels, loc, ops[0], ops[1]);
-        case 3:
-        default:
-            return pline_new_3(ident, labels, loc, ops[0], ops[1], ops[2]);
-    }
+    PLine* line = pline_new(ident, labels, loc, ops_size, ops);
+    free(ops);
+    return line;
 
 error_free_ops:
     for (size_t i = 0; i < ops_size; ++i)
         if (ops[i])
             poperand_free(ops[i]);
+    free(ops);
     plabel_free(labels);
     free(ident);
     return NULL;

compile_flags.txt

@@ -7,7 +7,7 @@
 -pedantic
 -pedantic-errors
 -Wno-unused-variable
--Wno-unused-parameter
+# -Wno-unused-parameter
 # -Wno-unused-function
 -I.
 

kern/main.asm

@@ -1,7 +1,7 @@
 
 start:
     ; rsp points *at* the top element
-    mov Rbp, 2048
+    mov rbp, 2048
     mov rsp, 2048 - 2
 
     lit interrupt_table
@@ -25,12 +25,22 @@ interrupt_table:
 
 keyboard_interrupt:
     and rfl, rfl, !(1 << Fl_Int)
-    push rbp
+    ; push rbp
+    add rsp, rsp, 2
+    mov u16 [rsp], rbp
     mov rbp, Rsp
-    push r0
-    push r1
-    push r2
-    push r3
+    ; push r0
+    add rsp, rsp, 2
+    mov u16 [rsp], r0
+    ; push r1
+    add rsp, rsp, 2
+    mov u16 [rsp], r1
+    ; push r2
+    add rsp, rsp, 2
+    mov u16 [rsp], r2
+    ; push r3
+    add rsp, rsp, 2
+    mov u16 [rsp], r3
 
     in r0, Device_Keyboard
 
@@ -52,7 +62,7 @@ keyboard_interrupt:
     jmp .L3
 
 .L0:
-    mov R0, 32 ; ' '
+    mov R0, ' '
     call put_char
     jmp .L4
 
@@ -80,26 +90,42 @@ keyboard_interrupt:
     jmp .L4
 
 .L3:
-    ; add r0, r0, 'A' - 4
+    add r0, r0, 'A' - 4
     call put_char
     jmp .L4
 
 .L4:
 
-    pop r3
-    pop r2
-    pop r1
-    pop r0
+    ; pop r3
+    mov r3, u16 [rsp]
+    sub rsp, rsp, 2
+    ; pop r2
+    mov r2, u16 [rsp]
+    sub rsp, rsp, 2
+    ; pop r1
+    mov r1, u16 [rsp]
+    sub rsp, rsp, 2
+    ; pop r0
+    mov r0, u16 [rsp]
+    sub rsp, rsp, 2
     mov rsp, rbp
-    pop rbp
+    ; pop rbp
+    mov rbp, u16 [rsp]
+    sub rsp, rsp, 2
     or rfl, rfl, 1 << Fl_Int
     iret
 
 put_char:
-    push rbp
+    ; push rbp
+    add rsp, rsp, 2
+    mov u16 [rsp], rbp
     mov rbp, rsp
-    push r1
-    push r2
+    ; push r1
+    add rsp, rsp, 2
+    mov u16 [rsp], r1
+    ; push r2
+    add rsp, rsp, 2
+    mov u16 [rsp], r2
 
     mov r2, screen_y
     mul r2, r2, vcd_width_in_ch
@@ -126,10 +152,16 @@ put_char:
     mov screen_x, r1
 
 .L1:
-    pop r1
-    pop r2
+    ; pop r1
+    mov r1, u16 [rsp]
+    sub rsp, rsp, 2
+    ; pop r2
+    mov r2, u16 [rsp]
+    sub rsp, rsp, 2
     mov rsp, rbp
-    pop rbp
+    ; pop rbp
+    mov rbp, u16 [rsp]
+    sub rsp, rsp, 2
     ret
 
 screen_x:

kern/main.c

@@ -203,7 +203,7 @@ void write_program(FILE* fp)
 
     printf("assembling program...\n");
     uint16_t program_size
-        = assemble_to_binary(program, program_asm, program_asm_size);
+        = assemble_lines_with_labels(program, program_asm, program_asm_size);
     printf("done!\n");
     printf("program size = %d\n", program_size);
 

vm/vm.c

@@ -1,5 +1,6 @@
 #include "vm.h"
 #include "common/arch.h"
+#include "common/op_str.h"
 #include <stdbool.h>
 #include <stdint.h>
 #include <stdio.h>