annotate lwasm/pass1.c @ 344:0215a0fbf61b

Added assembly error system and additional checks for symbol syntax
author lost@starbug
date Thu, 25 Mar 2010 22:06:50 -0600
parents 7b4123dce741
children 7416c3f9c321
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
332
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
1 /*
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
2 pass1.c
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
3
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
4 Copyright © 2010 William Astle
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
5
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
6 This file is part of LWTOOLS.
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
7
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
8 LWTOOLS is free software: you can redistribute it and/or modify it under the
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
9 terms of the GNU General Public License as published by the Free Software
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
10 Foundation, either version 3 of the License, or (at your option) any later
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
11 version.
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
12
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
13 This program is distributed in the hope that it will be useful, but WITHOUT
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
14 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
15 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
16 more details.
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
17
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
18 You should have received a copy of the GNU General Public License along with
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
19 this program. If not, see <http://www.gnu.org/licenses/>.
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
20 */
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
21
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
22 #include <config.h>
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
23
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
24 #include <stdio.h>
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
25
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
26 #include <lw_alloc.h>
342
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
27 #include <lw_string.h>
332
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
28
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
29 #include "lwasm.h"
342
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
30 #include "instab.h"
332
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
31 #include "input.h"
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
32
337
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
33 /*
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
34 pass 1: parse the lines
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
35
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
36 line format:
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
37
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
38 [<symbol>] <opcode> <operand>[ <comment>]
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
39
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
40 If <symbol> is followed by a :, whitespace may precede the symbol
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
41
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
42 A line may optionally start with a number which must not be preceded by
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
43 white space and must be followed by a single whitespace character. After
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
44 that whitespace character, the line is parsed as if it had no line number.
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
45
337
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
46 */
332
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
47 void do_pass1(asmstate_t *as)
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
48 {
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
49 char *line;
337
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
50 line_t *cl;
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
51 char *p1;
342
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
52 int stspace;
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
53 char *tok, *sym;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
54 int opnum;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
55
332
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
56 for (;;)
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
57 {
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
58 sym = NULL;
332
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
59 line = input_readline(as);
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
60 if (!line)
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
61 break;
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
62 printf("%s\n", line);
337
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
63
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
64 cl = lw_alloc(sizeof(line_t));
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
65 cl -> next = NULL;
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
66 cl -> prev = as -> line_tail;
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
67 cl -> len = -1;
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
68 cl -> insn = -1;
344
0215a0fbf61b Added assembly error system and additional checks for symbol syntax
lost@starbug
parents: 342
diff changeset
69 cl -> err = NULL;
337
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
70 if (!as -> line_tail)
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
71 {
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
72 as -> line_head = cl;
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
73 cl -> addr = lw_expr_build(lw_expr_type_int, 0);
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
74 }
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
75 else
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
76 {
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
77 lw_expr_t te;
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
78 as -> line_tail -> next = cl;
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
79 te = lw_expr_build(lw_expr_type_special, lwasm_expr_linelen, cl -> prev);
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
80 cl -> addr = lw_expr_build(lw_expr_type_oper, lw_expr_oper_plus, cl -> prev -> addr, te);
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
81 lw_expr_destroy(te);
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
82 lw_expr_simplify(cl -> addr);
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
83 }
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
84 as -> line_tail = cl;
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
85
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
86 // blank lines don't count for anything
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
87 if (!*line)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
88 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
89 goto nextline;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
90 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
91
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
92 // skip comments
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
93 if (*line == '*' || *line == ';' || *line == '#')
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
94 goto nextline;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
95
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
96 p1 = line;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
97 if (isdigit(*p1))
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
98 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
99 // skip line number
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
100 while (*p1 && isdigit(*p1))
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
101 p1++;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
102 if (!*p1 && !isspace(*p1))
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
103 p1 = line;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
104 else if (*p1 && isspace(*p1))
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
105 p1++;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
106 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
107
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
108 if (!*p1)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
109 goto nextline;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
110
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
111 if (*p1 == '*' || *p1 == ';' || *p1 == '#')
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
112 goto nextline;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
113
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
114 if (isspace(*p1))
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
115 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
116 for (; *p1 && isspace(*p1); p1++)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
117 /* do nothing */ ;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
118 stspace = 1;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
119 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
120 else
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
121 stspace = 0;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
122
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
123 if (*p1 == '*' || *p1 == ';' || *p1 == '#' || !*p1)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
124 goto nextline;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
125
342
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
126 // find the end of the first token
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
127 for (tok = p1; *p1 && !isspace(*p1) && *p1 != ':'; p1++)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
128 /* do nothing */ ;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
129
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
130 if (*p1 == ':' || stspace == 0)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
131 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
132 // have a symbol here
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
133 sym = lw_strndup(tok, p1 - tok);
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
134 if (*p1 == ':')
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
135 p1++;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
136 for (; *p1 && isspace(*p1); p1++)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
137 /* do nothing */ ;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
138
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
139 for (tok = p1; *p1 && !isspace(*p1); p1++)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
140 /* do nothing */ ;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
141 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
142
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
143 cl -> sym = sym;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
144 cl -> symset = 0;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
145
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
146 // tok points to the opcode for the line or NUL if none
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
147 if (*tok)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
148 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
149 // look up operation code
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
150 sym = lw_strndup(tok, p1 - tok);
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
151 for (; *p1 && isspace(p1); p1++)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
152 /* do nothing */ ;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
153
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
154 for (opnum = 0; instab[opnum].opcode; opnum++)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
155 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
156 if (!strcasecmp(instab[opnum].opcode, sym))
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
157 break;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
158 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
159 lw_free(sym);
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
160
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
161 // p1 points to the start of the operand
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
162
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
163 if (instab[opnum].opcode == NULL)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
164 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
165 cl -> insn = -1;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
166 if (*tok != ';' && *tok != '*')
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
167 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
168 // bad opcode; check for macro here
344
0215a0fbf61b Added assembly error system and additional checks for symbol syntax
lost@starbug
parents: 342
diff changeset
169 lwasm_register_error(as, cl, "Bad opcode");
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
170 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
171 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
172 else
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
173 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
174 cl -> insn = opnum;
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
175 // call parse function
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
176
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
177 if (*p1 && !isspace(*p1))
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
178 {
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
179 // flag bad operand error
344
0215a0fbf61b Added assembly error system and additional checks for symbol syntax
lost@starbug
parents: 342
diff changeset
180 lwasm_register_error(as, cl, "Bad operand (%s)", p1);
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
181 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
182 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
183 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
184
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
185 if (cl -> sym && cl -> symset == 0)
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
186 {
342
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
187 printf("Register symbol %s:", sym);
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
188 lw_expr_print(cl -> addr);
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
189 printf("\n");
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
190
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
191 // register symbol at line address
344
0215a0fbf61b Added assembly error system and additional checks for symbol syntax
lost@starbug
parents: 342
diff changeset
192 if (!register_symbol(as, cl, cl -> sym, cl -> addr, symbol_flag_none))
342
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
193 {
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
194 // symbol error
344
0215a0fbf61b Added assembly error system and additional checks for symbol syntax
lost@starbug
parents: 342
diff changeset
195 lwasm_register_error(as, cl, "Bad symbol '%s'", cl -> sym);
342
7b4123dce741 Added basic symbol registration
lost@starbug
parents: 340
diff changeset
196 }
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
197 }
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
198
337
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
199 lw_expr_print(cl -> addr);
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
200 printf("\n");
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
201 // now parse the line
04c80c51b16a Checkpoint development
lost
parents: 332
diff changeset
202
340
1a6fc6ebb31c Checkpoint
lost
parents: 337
diff changeset
203 nextline:
332
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
204 lw_free(line);
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
205 }
67224d8d1024 Basic input layer works
lost
parents:
diff changeset
206 }