116
|
1 /*
|
|
2 readfiles.c
|
118
|
3 Copyright © 2009 William Astle
|
116
|
4
|
|
5 This file is part of LWLINK.
|
|
6
|
|
7 LWLINK is free software: you can redistribute it and/or modify it under the
|
|
8 terms of the GNU General Public License as published by the Free Software
|
|
9 Foundation, either version 3 of the License, or (at your option) any later
|
|
10 version.
|
|
11
|
|
12 This program is distributed in the hope that it will be useful, but WITHOUT
|
|
13 ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
|
|
14 FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for
|
|
15 more details.
|
|
16
|
|
17 You should have received a copy of the GNU General Public License along with
|
|
18 this program. If not, see <http://www.gnu.org/licenses/>.
|
|
19
|
|
20
|
|
21 Reads input files
|
|
22
|
|
23 */
|
|
24
|
|
25 #ifdef HAVE_CONFIG_H
|
|
26 #include "config.h"
|
|
27 #endif
|
|
28
|
|
29 #include <argp.h>
|
|
30 #include <errno.h>
|
|
31 #include <stdio.h>
|
|
32 #include <stdlib.h>
|
171
|
33 #include <string.h>
|
116
|
34
|
|
35 #include "lwlink.h"
|
|
36 #include "util.h"
|
|
37
|
|
38 void read_lwobj16v0(fileinfo_t *fn);
|
171
|
39 void read_lwar1v(fileinfo_t *fn);
|
116
|
40
|
|
41 /*
|
|
42 The logic of reading the entire file into memory is simple. All the symbol
|
|
43 names in the file are NUL terminated strings and can be used directly without
|
|
44 making additional copies.
|
|
45 */
|
171
|
46 void read_file(fileinfo_t *fn)
|
|
47 {
|
|
48 if (!memcmp(fn -> filedata, "LWOBJ16", 8))
|
|
49 {
|
|
50 // read v0 LWOBJ16 file
|
|
51 read_lwobj16v0(fn);
|
|
52 }
|
|
53 else if (!memcmp(fn -> filedata, "LWAR1V", 6))
|
|
54 {
|
|
55 // archive file
|
|
56 read_lwar1v(fn);
|
|
57 }
|
|
58 else
|
|
59 {
|
|
60 fprintf(stderr, "%s: unknown file format\n", fn -> filename);
|
|
61 exit(1);
|
|
62 }
|
|
63 }
|
|
64
|
116
|
65 void read_files(void)
|
|
66 {
|
|
67 int i;
|
|
68 long size;
|
|
69 FILE *f;
|
|
70 long bread;
|
|
71 for (i = 0; i < ninputfiles; i++)
|
|
72 {
|
180
|
73 if (inputfiles[i] -> islib)
|
116
|
74 {
|
180
|
75 char *tf;
|
|
76 int s;
|
|
77 int j;
|
|
78
|
|
79 f = NULL;
|
|
80
|
|
81 for (j = 0; j < nlibdirs; j++)
|
|
82 {
|
|
83 s = strlen(libdirs[j]) + 7 + strlen(inputfiles[i] -> filename);
|
|
84 tf = lw_malloc(s + 1);
|
|
85 sprintf(tf, "%s/lib%s.a", libdirs[j], inputfiles[i] -> filename);
|
|
86 f = fopen(tf, "rb");
|
|
87 if (!f)
|
|
88 {
|
|
89 free(tf);
|
|
90 continue;
|
|
91 }
|
|
92 free(tf);
|
|
93 }
|
|
94 if (!f)
|
|
95 {
|
|
96 fprintf(stderr, "Can't open library: -l%s\n", inputfiles[i] -> filename);
|
|
97 exit(1);
|
|
98 }
|
|
99 }
|
|
100 else
|
|
101 {
|
|
102 f = fopen(inputfiles[i] -> filename, "rb");
|
|
103 if (!f)
|
|
104 {
|
|
105 fprintf(stderr, "Can't open file %s:", inputfiles[i] -> filename);
|
|
106 perror("");
|
|
107 exit(1);
|
|
108 }
|
116
|
109 }
|
|
110 fseek(f, 0, SEEK_END);
|
|
111 size = ftell(f);
|
|
112 rewind(f);
|
|
113
|
|
114 inputfiles[i] -> filedata = lw_malloc(size);
|
|
115 inputfiles[i] -> filesize = size;
|
|
116
|
|
117 bread = fread(inputfiles[i] -> filedata, 1, size, f);
|
|
118 if (bread < size)
|
|
119 {
|
|
120 fprintf(stderr, "Short read on file %s (%ld/%ld):", inputfiles[i] -> filename, bread, size);
|
|
121 perror("");
|
|
122 exit(1);
|
|
123 }
|
|
124
|
|
125 fclose(f);
|
|
126
|
171
|
127 read_file(inputfiles[i]);
|
116
|
128 }
|
|
129 }
|
|
130
|
|
131 // this macro is used to bail out if we run off the end of the file data
|
|
132 // while parsing - it keeps the code below cleaner
|
124
|
133 #define NEXTBYTE() do { cc++; if (cc > fn -> filesize) { fprintf(stderr, "%s: invalid file format\n", fn -> filename); exit(1); } } while (0)
|
116
|
134 // this macro is used to refer to the current byte in the stream
|
124
|
135 #define CURBYTE() (fn -> filedata[cc < fn -> filesize ? cc : fn -> filesize - 1])
|
116
|
136 // this one will leave the input pointer past the trailing NUL
|
|
137 #define CURSTR() read_lwobj16v0_str(&cc, fn)
|
|
138 unsigned char *read_lwobj16v0_str(long *cc1, fileinfo_t *fn)
|
|
139 {
|
|
140 int cc = *cc1;
|
|
141 unsigned char *fp;
|
|
142 fp = &CURBYTE();
|
|
143 while (CURBYTE())
|
|
144 NEXTBYTE();
|
|
145 NEXTBYTE();
|
|
146 *cc1 = cc;
|
|
147 return fp;
|
|
148 }
|
|
149 // the function below can be switched to dealing with data coming from a
|
|
150 // source other than an in-memory byte pool by adjusting the input data
|
|
151 // in "fn" and the above two macros
|
171
|
152
|
116
|
153 void read_lwobj16v0(fileinfo_t *fn)
|
|
154 {
|
|
155 unsigned char *fp;
|
|
156 long cc;
|
|
157 section_t *s;
|
|
158 int val;
|
|
159 symtab_t *se;
|
|
160
|
|
161 // start reading *after* the magic number
|
126
|
162 cc = 8;
|
116
|
163
|
|
164 // init data
|
|
165 fn -> sections = NULL;
|
|
166 fn -> nsections = 0;
|
|
167
|
|
168 while (1)
|
|
169 {
|
126
|
170 // NEXTBYTE();
|
116
|
171 // bail out if no more sections
|
|
172 if (!(CURBYTE()))
|
|
173 break;
|
|
174
|
|
175 fp = CURSTR();
|
|
176
|
|
177 // we now have a section name in fp
|
|
178 // create new section entry
|
|
179 fn -> sections = lw_realloc(fn -> sections, sizeof(section_t) * (fn -> nsections + 1));
|
|
180 s = &(fn -> sections[fn -> nsections]);
|
|
181 fn -> nsections += 1;
|
|
182
|
|
183 s -> localsyms = NULL;
|
|
184 s -> flags = 0;
|
|
185 s -> codesize = 0;
|
|
186 s -> name = fp;
|
|
187 s -> loadaddress = 0;
|
|
188 s -> localsyms = NULL;
|
|
189 s -> exportedsyms = NULL;
|
|
190 s -> incompletes = NULL;
|
119
|
191 s -> processed = 0;
|
|
192 s -> file = fn;
|
|
193
|
116
|
194 // read flags
|
|
195 while (CURBYTE())
|
|
196 {
|
|
197 switch (CURBYTE())
|
|
198 {
|
|
199 case 0x01:
|
|
200 s -> flags |= SECTION_BSS;
|
|
201 break;
|
|
202
|
|
203 default:
|
|
204 fprintf(stderr, "%s (%s): unrecognized section flag %02X\n", fn -> filename, s -> name, (int)(CURBYTE()));
|
|
205 exit(1);
|
|
206 }
|
|
207 NEXTBYTE();
|
|
208 }
|
|
209 // skip NUL terminating flags
|
|
210 NEXTBYTE();
|
|
211
|
|
212 // now parse the local symbol table
|
|
213 while (CURBYTE())
|
|
214 {
|
|
215 fp = CURSTR();
|
|
216
|
|
217 // fp is the symbol name
|
|
218 val = (CURBYTE()) << 8;
|
|
219 NEXTBYTE();
|
|
220 val |= (CURBYTE());
|
|
221 NEXTBYTE();
|
|
222 // val is now the symbol value
|
|
223
|
|
224 // create symbol table entry
|
|
225 se = lw_malloc(sizeof(symtab_t));
|
|
226 se -> next = s -> localsyms;
|
|
227 s -> localsyms = se;
|
|
228 se -> sym = fp;
|
|
229 se -> offset = val;
|
|
230 }
|
124
|
231 // skip terminating NUL
|
|
232 NEXTBYTE();
|
|
233
|
116
|
234 // now parse the exported symbol table
|
|
235 while (CURBYTE())
|
|
236 {
|
|
237 fp = CURSTR();
|
|
238
|
|
239 // fp is the symbol name
|
|
240 val = (CURBYTE()) << 8;
|
|
241 NEXTBYTE();
|
|
242 val |= (CURBYTE());
|
|
243 NEXTBYTE();
|
|
244 // val is now the symbol value
|
|
245
|
|
246 // create symbol table entry
|
|
247 se = lw_malloc(sizeof(symtab_t));
|
|
248 se -> next = s -> exportedsyms;
|
128
|
249 s -> exportedsyms = se;
|
116
|
250 se -> sym = fp;
|
|
251 se -> offset = val;
|
|
252 }
|
124
|
253 // skip terminating NUL
|
|
254 NEXTBYTE();
|
116
|
255
|
|
256 // now parse the incomplete references and make a list of
|
|
257 // external references that need resolution
|
|
258 while (CURBYTE())
|
|
259 {
|
|
260 reloc_t *rp;
|
|
261 lw_expr_term_t *term;
|
|
262
|
|
263 // we have a reference
|
|
264 rp = lw_malloc(sizeof(reloc_t));
|
|
265 rp -> next = s -> incompletes;
|
|
266 s -> incompletes = rp;
|
|
267 rp -> offset = 0;
|
|
268 rp -> expr = lw_expr_stack_create();
|
|
269
|
|
270 // parse the expression
|
|
271 while (CURBYTE())
|
|
272 {
|
|
273 int tt = CURBYTE();
|
|
274 NEXTBYTE();
|
|
275 switch (tt)
|
|
276 {
|
|
277 case 0x01:
|
|
278 // 16 bit integer
|
|
279 tt = CURBYTE() << 8;
|
|
280 NEXTBYTE();
|
|
281 tt |= CURBYTE();
|
|
282 NEXTBYTE();
|
|
283 // normalize for negatives...
|
|
284 if (tt > 0x7fff)
|
|
285 tt -= 0x10000;
|
|
286 term = lw_expr_term_create_int(tt);
|
|
287 break;
|
|
288
|
|
289 case 0x02:
|
|
290 // external symbol reference
|
|
291 term = lw_expr_term_create_sym(CURSTR(), 0);
|
|
292 break;
|
|
293
|
|
294 case 0x03:
|
|
295 // internal symbol reference
|
|
296 term = lw_expr_term_create_sym(CURSTR(), 1);
|
|
297 break;
|
|
298
|
|
299 case 0x04:
|
|
300 // operator
|
|
301 term = lw_expr_term_create_oper(CURBYTE());
|
|
302 NEXTBYTE();
|
|
303 break;
|
|
304
|
|
305 case 0x05:
|
|
306 // section base reference (NULL internal reference is
|
|
307 // the section base address
|
120
|
308 term = lw_expr_term_create_sym(NULL, 1);
|
116
|
309 break;
|
|
310
|
|
311 default:
|
|
312 fprintf(stderr, "%s (%s): bad relocation expression\n", fn -> filename, s -> name);
|
|
313 exit(1);
|
|
314 }
|
|
315 lw_expr_stack_push(rp -> expr, term);
|
|
316 lw_expr_term_free(term);
|
|
317 }
|
127
|
318 // skip the NUL
|
|
319 NEXTBYTE();
|
|
320
|
|
321 // fetch the offset
|
|
322 rp -> offset = CURBYTE() << 8;
|
|
323 NEXTBYTE();
|
|
324 rp -> offset |= CURBYTE() & 0xff;
|
|
325 NEXTBYTE();
|
116
|
326 }
|
124
|
327 // skip the NUL terminating the relocations
|
116
|
328 NEXTBYTE();
|
|
329
|
|
330 // now set code location and size and verify that the file
|
|
331 // contains data going to the end of the code (if !SECTION_BSS)
|
|
332 s -> codesize = CURBYTE() << 8;
|
|
333 NEXTBYTE();
|
|
334 s -> codesize |= CURBYTE();
|
|
335 NEXTBYTE();
|
|
336
|
|
337 s -> code = &(CURBYTE());
|
|
338
|
|
339 // skip the code if we're not in a BSS section
|
|
340 if (!(s -> flags & SECTION_BSS))
|
|
341 {
|
|
342 int i;
|
|
343 for (i = 0; i < s -> codesize; i++)
|
|
344 NEXTBYTE();
|
|
345 }
|
|
346 }
|
|
347 }
|
171
|
348
|
|
349 /*
|
|
350 Read an archive file - this will create a "sub" record and farm out the
|
|
351 parsing of the sub files to the regular file parsers
|
|
352
|
|
353 The archive file format consists of the 6 byte magic number followed by a
|
|
354 series of records as follows:
|
|
355
|
|
356 - NUL terminated file name
|
|
357 - 32 bit file length in big endian order
|
|
358 - the file data
|
|
359
|
|
360 An empty file name indicates the end of the file.
|
|
361
|
|
362 */
|
|
363 void read_lwar1v(fileinfo_t *fn)
|
|
364 {
|
|
365 unsigned long cc = 6;
|
|
366 unsigned long flen;
|
|
367 unsigned long l;
|
|
368 for (;;)
|
|
369 {
|
|
370 if (cc >= fn -> filesize || !(fn -> filedata[cc]))
|
|
371 return;
|
|
372
|
173
|
373 for (l = cc; cc < fn -> filesize && fn -> filedata[cc]; cc++)
|
171
|
374 /* do nothing */ ;
|
|
375
|
173
|
376 cc++;
|
|
377
|
171
|
378 if (cc >= fn -> filesize)
|
|
379 {
|
|
380 fprintf(stderr, "Malformed archive file %s.\n", fn -> filename);
|
|
381 exit(1);
|
|
382 }
|
|
383
|
|
384 if (cc + 4 > fn -> filesize)
|
|
385 return;
|
|
386
|
173
|
387 flen = (fn -> filedata[cc++] << 24);
|
|
388 flen |= (fn -> filedata[cc++] << 16);
|
|
389 flen |= (fn -> filedata[cc++] << 8);
|
|
390 flen |= (fn -> filedata[cc++]);
|
171
|
391
|
|
392 if (flen == 0)
|
|
393 return;
|
|
394
|
|
395 if (cc + flen > fn -> filesize)
|
|
396 {
|
|
397 fprintf(stderr, "Malformed archive file %s.\n", fn -> filename);
|
|
398 exit(1);
|
|
399 }
|
|
400
|
|
401 // add the "sub" input file
|
|
402 fn -> subs = lw_realloc(fn -> subs, sizeof(fileinfo_t *) * (fn -> nsubs + 1));
|
|
403 fn -> subs[fn -> nsubs] = lw_malloc(sizeof(fileinfo_t));
|
|
404 memset(fn -> subs[fn -> nsubs], 0, sizeof(fileinfo_t));
|
|
405 fn -> subs[fn -> nsubs] -> filedata = fn -> filedata + cc;
|
|
406 fn -> subs[fn -> nsubs] -> filesize = flen;
|
|
407 fn -> subs[fn -> nsubs] -> filename = lw_strdup(fn -> filedata + l);
|
|
408 fn -> subs[fn -> nsubs] -> parent = fn;
|
173
|
409
|
171
|
410 read_file(fn -> subs[fn -> nsubs]);
|
|
411 fn -> nsubs++;
|
173
|
412 cc += flen;
|
171
|
413 }
|
|
414 }
|