339
|
1 /*
|
|
2 readfiles.c
|
|
3 Copyright © 2009 William Astle
|
|
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 #include <config.h>
|
|
26
|
|
27 #include <argp.h>
|
|
28 #include <errno.h>
|
|
29 #include <stdio.h>
|
|
30 #include <stdlib.h>
|
|
31 #include <string.h>
|
|
32
|
|
33 #include "lwlink.h"
|
|
34 #include "util.h"
|
|
35
|
|
36 void read_lwobj16v0(fileinfo_t *fn);
|
|
37 void read_lwar1v(fileinfo_t *fn);
|
|
38
|
|
39 /*
|
|
40 The logic of reading the entire file into memory is simple. All the symbol
|
|
41 names in the file are NUL terminated strings and can be used directly without
|
|
42 making additional copies.
|
|
43 */
|
|
44 void read_file(fileinfo_t *fn)
|
|
45 {
|
|
46 if (!memcmp(fn -> filedata, "LWOBJ16", 8))
|
|
47 {
|
|
48 // read v0 LWOBJ16 file
|
|
49 read_lwobj16v0(fn);
|
|
50 }
|
|
51 else if (!memcmp(fn -> filedata, "LWAR1V", 6))
|
|
52 {
|
|
53 // archive file
|
|
54 read_lwar1v(fn);
|
|
55 }
|
|
56 else
|
|
57 {
|
|
58 fprintf(stderr, "%s: unknown file format\n", fn -> filename);
|
|
59 exit(1);
|
|
60 }
|
|
61 }
|
|
62
|
|
63 void read_files(void)
|
|
64 {
|
|
65 int i;
|
|
66 long size;
|
|
67 FILE *f;
|
|
68 long bread;
|
|
69 for (i = 0; i < ninputfiles; i++)
|
|
70 {
|
|
71 if (inputfiles[i] -> islib)
|
|
72 {
|
|
73 char *tf;
|
|
74 int s;
|
|
75 int j;
|
|
76
|
|
77 f = NULL;
|
|
78
|
|
79 for (j = 0; j < nlibdirs; j++)
|
|
80 {
|
|
81 s = strlen(libdirs[j]) + 7 + strlen(inputfiles[i] -> filename);
|
|
82 tf = lw_malloc(s + 1);
|
|
83 sprintf(tf, "%s/lib%s.a", libdirs[j], inputfiles[i] -> filename);
|
|
84 f = fopen(tf, "rb");
|
|
85 if (!f)
|
|
86 {
|
|
87 free(tf);
|
|
88 continue;
|
|
89 }
|
|
90 free(tf);
|
|
91 break;
|
|
92 }
|
|
93 if (!f)
|
|
94 {
|
|
95 fprintf(stderr, "Can't open library: -l%s\n", inputfiles[i] -> filename);
|
|
96 exit(1);
|
|
97 }
|
|
98 }
|
|
99 else
|
|
100 {
|
|
101 f = fopen(inputfiles[i] -> filename, "rb");
|
|
102 if (!f)
|
|
103 {
|
|
104 fprintf(stderr, "Can't open file %s:", inputfiles[i] -> filename);
|
|
105 perror("");
|
|
106 exit(1);
|
|
107 }
|
|
108 }
|
|
109 fseek(f, 0, SEEK_END);
|
|
110 size = ftell(f);
|
|
111 rewind(f);
|
|
112
|
|
113 inputfiles[i] -> filedata = lw_malloc(size);
|
|
114 inputfiles[i] -> filesize = size;
|
|
115
|
|
116 bread = fread(inputfiles[i] -> filedata, 1, size, f);
|
|
117 if (bread < size)
|
|
118 {
|
|
119 fprintf(stderr, "Short read on file %s (%ld/%ld):", inputfiles[i] -> filename, bread, size);
|
|
120 perror("");
|
|
121 exit(1);
|
|
122 }
|
|
123
|
|
124 fclose(f);
|
|
125
|
|
126 read_file(inputfiles[i]);
|
|
127 }
|
|
128 }
|
|
129
|
|
130 // this macro is used to bail out if we run off the end of the file data
|
|
131 // while parsing - it keeps the code below cleaner
|
|
132 #define NEXTBYTE() do { cc++; if (cc > fn -> filesize) { fprintf(stderr, "%s: invalid file format\n", fn -> filename); exit(1); } } while (0)
|
|
133 // this macro is used to refer to the current byte in the stream
|
|
134 #define CURBYTE() (fn -> filedata[cc < fn -> filesize ? cc : fn -> filesize - 1])
|
|
135 // this one will leave the input pointer past the trailing NUL
|
|
136 #define CURSTR() read_lwobj16v0_str(&cc, fn)
|
|
137 unsigned char *read_lwobj16v0_str(long *cc1, fileinfo_t *fn)
|
|
138 {
|
|
139 int cc = *cc1;
|
|
140 unsigned char *fp;
|
|
141 fp = &CURBYTE();
|
|
142 while (CURBYTE())
|
|
143 NEXTBYTE();
|
|
144 NEXTBYTE();
|
|
145 *cc1 = cc;
|
|
146 return fp;
|
|
147 }
|
|
148 // the function below can be switched to dealing with data coming from a
|
|
149 // source other than an in-memory byte pool by adjusting the input data
|
|
150 // in "fn" and the above two macros
|
|
151
|
|
152 void read_lwobj16v0(fileinfo_t *fn)
|
|
153 {
|
|
154 unsigned char *fp;
|
|
155 long cc;
|
|
156 section_t *s;
|
|
157 int val;
|
|
158 symtab_t *se;
|
|
159
|
|
160 // start reading *after* the magic number
|
|
161 cc = 8;
|
|
162
|
|
163 // init data
|
|
164 fn -> sections = NULL;
|
|
165 fn -> nsections = 0;
|
|
166
|
|
167 while (1)
|
|
168 {
|
|
169 // NEXTBYTE();
|
|
170 // bail out if no more sections
|
|
171 if (!(CURBYTE()))
|
|
172 break;
|
|
173
|
|
174 fp = CURSTR();
|
|
175
|
|
176 // we now have a section name in fp
|
|
177 // create new section entry
|
|
178 fn -> sections = lw_realloc(fn -> sections, sizeof(section_t) * (fn -> nsections + 1));
|
|
179 s = &(fn -> sections[fn -> nsections]);
|
|
180 fn -> nsections += 1;
|
|
181
|
|
182 s -> localsyms = NULL;
|
|
183 s -> flags = 0;
|
|
184 s -> codesize = 0;
|
|
185 s -> name = fp;
|
|
186 s -> loadaddress = 0;
|
|
187 s -> localsyms = NULL;
|
|
188 s -> exportedsyms = NULL;
|
|
189 s -> incompletes = NULL;
|
|
190 s -> processed = 0;
|
|
191 s -> file = fn;
|
|
192
|
|
193 // read flags
|
|
194 while (CURBYTE())
|
|
195 {
|
|
196 switch (CURBYTE())
|
|
197 {
|
|
198 case 0x01:
|
|
199 s -> flags |= SECTION_BSS;
|
|
200 break;
|
|
201
|
|
202 default:
|
|
203 fprintf(stderr, "%s (%s): unrecognized section flag %02X\n", fn -> filename, s -> name, (int)(CURBYTE()));
|
|
204 exit(1);
|
|
205 }
|
|
206 NEXTBYTE();
|
|
207 }
|
|
208 // skip NUL terminating flags
|
|
209 NEXTBYTE();
|
|
210
|
|
211 // now parse the local symbol table
|
|
212 while (CURBYTE())
|
|
213 {
|
|
214 fp = CURSTR();
|
|
215
|
|
216 // fp is the symbol name
|
|
217 val = (CURBYTE()) << 8;
|
|
218 NEXTBYTE();
|
|
219 val |= (CURBYTE());
|
|
220 NEXTBYTE();
|
|
221 // val is now the symbol value
|
|
222
|
|
223 // create symbol table entry
|
|
224 se = lw_malloc(sizeof(symtab_t));
|
|
225 se -> next = s -> localsyms;
|
|
226 s -> localsyms = se;
|
|
227 se -> sym = fp;
|
|
228 se -> offset = val;
|
|
229 }
|
|
230 // skip terminating NUL
|
|
231 NEXTBYTE();
|
|
232
|
|
233 // now parse the exported symbol table
|
|
234 while (CURBYTE())
|
|
235 {
|
|
236 fp = CURSTR();
|
|
237
|
|
238 // fp is the symbol name
|
|
239 val = (CURBYTE()) << 8;
|
|
240 NEXTBYTE();
|
|
241 val |= (CURBYTE());
|
|
242 NEXTBYTE();
|
|
243 // val is now the symbol value
|
|
244
|
|
245 // create symbol table entry
|
|
246 se = lw_malloc(sizeof(symtab_t));
|
|
247 se -> next = s -> exportedsyms;
|
|
248 s -> exportedsyms = se;
|
|
249 se -> sym = fp;
|
|
250 se -> offset = val;
|
|
251 }
|
|
252 // skip terminating NUL
|
|
253 NEXTBYTE();
|
|
254
|
|
255 // now parse the incomplete references and make a list of
|
|
256 // external references that need resolution
|
|
257 while (CURBYTE())
|
|
258 {
|
|
259 reloc_t *rp;
|
|
260 lw_expr_term_t *term;
|
|
261
|
|
262 // we have a reference
|
|
263 rp = lw_malloc(sizeof(reloc_t));
|
|
264 rp -> next = s -> incompletes;
|
|
265 s -> incompletes = rp;
|
|
266 rp -> offset = 0;
|
|
267 rp -> expr = lw_expr_stack_create();
|
|
268 rp -> flags = RELOC_NORM;
|
|
269
|
|
270 // parse the expression
|
|
271 while (CURBYTE())
|
|
272 {
|
|
273 int tt = CURBYTE();
|
|
274 NEXTBYTE();
|
|
275 switch (tt)
|
|
276 {
|
|
277 case 0xFF:
|
|
278 // a flag specifier
|
|
279 tt = CURBYTE();
|
|
280 rp -> flags = tt;
|
|
281 NEXTBYTE();
|
|
282 term = NULL;
|
|
283 break;
|
|
284
|
|
285 case 0x01:
|
|
286 // 16 bit integer
|
|
287 tt = CURBYTE() << 8;
|
|
288 NEXTBYTE();
|
|
289 tt |= CURBYTE();
|
|
290 NEXTBYTE();
|
|
291 // normalize for negatives...
|
|
292 if (tt > 0x7fff)
|
|
293 tt -= 0x10000;
|
|
294 term = lw_expr_term_create_int(tt);
|
|
295 break;
|
|
296
|
|
297 case 0x02:
|
|
298 // external symbol reference
|
|
299 term = lw_expr_term_create_sym(CURSTR(), 0);
|
|
300 break;
|
|
301
|
|
302 case 0x03:
|
|
303 // internal symbol reference
|
|
304 term = lw_expr_term_create_sym(CURSTR(), 1);
|
|
305 break;
|
|
306
|
|
307 case 0x04:
|
|
308 // operator
|
|
309 term = lw_expr_term_create_oper(CURBYTE());
|
|
310 NEXTBYTE();
|
|
311 break;
|
|
312
|
|
313 case 0x05:
|
|
314 // section base reference (NULL internal reference is
|
|
315 // the section base address
|
|
316 term = lw_expr_term_create_sym(NULL, 1);
|
|
317 break;
|
|
318
|
|
319 default:
|
|
320 fprintf(stderr, "%s (%s): bad relocation expression (%02X)\n", fn -> filename, s -> name, tt);
|
|
321 exit(1);
|
|
322 }
|
|
323 if (term)
|
|
324 {
|
|
325 lw_expr_stack_push(rp -> expr, term);
|
|
326 lw_expr_term_free(term);
|
|
327 }
|
|
328 }
|
|
329 // skip the NUL
|
|
330 NEXTBYTE();
|
|
331
|
|
332 // fetch the offset
|
|
333 rp -> offset = CURBYTE() << 8;
|
|
334 NEXTBYTE();
|
|
335 rp -> offset |= CURBYTE() & 0xff;
|
|
336 NEXTBYTE();
|
|
337 }
|
|
338 // skip the NUL terminating the relocations
|
|
339 NEXTBYTE();
|
|
340
|
|
341 // now set code location and size and verify that the file
|
|
342 // contains data going to the end of the code (if !SECTION_BSS)
|
|
343 s -> codesize = CURBYTE() << 8;
|
|
344 NEXTBYTE();
|
|
345 s -> codesize |= CURBYTE();
|
|
346 NEXTBYTE();
|
|
347
|
|
348 s -> code = &(CURBYTE());
|
|
349
|
|
350 // skip the code if we're not in a BSS section
|
|
351 if (!(s -> flags & SECTION_BSS))
|
|
352 {
|
|
353 int i;
|
|
354 for (i = 0; i < s -> codesize; i++)
|
|
355 NEXTBYTE();
|
|
356 }
|
|
357 }
|
|
358 }
|
|
359
|
|
360 /*
|
|
361 Read an archive file - this will create a "sub" record and farm out the
|
|
362 parsing of the sub files to the regular file parsers
|
|
363
|
|
364 The archive file format consists of the 6 byte magic number followed by a
|
|
365 series of records as follows:
|
|
366
|
|
367 - NUL terminated file name
|
|
368 - 32 bit file length in big endian order
|
|
369 - the file data
|
|
370
|
|
371 An empty file name indicates the end of the file.
|
|
372
|
|
373 */
|
|
374 void read_lwar1v(fileinfo_t *fn)
|
|
375 {
|
|
376 unsigned long cc = 6;
|
|
377 unsigned long flen;
|
|
378 unsigned long l;
|
|
379 for (;;)
|
|
380 {
|
|
381 if (cc >= fn -> filesize || !(fn -> filedata[cc]))
|
|
382 return;
|
|
383
|
|
384 for (l = cc; cc < fn -> filesize && fn -> filedata[cc]; cc++)
|
|
385 /* do nothing */ ;
|
|
386
|
|
387 cc++;
|
|
388
|
|
389 if (cc >= fn -> filesize)
|
|
390 {
|
|
391 fprintf(stderr, "Malformed archive file %s.\n", fn -> filename);
|
|
392 exit(1);
|
|
393 }
|
|
394
|
|
395 if (cc + 4 > fn -> filesize)
|
|
396 return;
|
|
397
|
|
398 flen = (fn -> filedata[cc++] << 24);
|
|
399 flen |= (fn -> filedata[cc++] << 16);
|
|
400 flen |= (fn -> filedata[cc++] << 8);
|
|
401 flen |= (fn -> filedata[cc++]);
|
|
402
|
|
403 if (flen == 0)
|
|
404 return;
|
|
405
|
|
406 if (cc + flen > fn -> filesize)
|
|
407 {
|
|
408 fprintf(stderr, "Malformed archive file %s.\n", fn -> filename);
|
|
409 exit(1);
|
|
410 }
|
|
411
|
|
412 // add the "sub" input file
|
|
413 fn -> subs = lw_realloc(fn -> subs, sizeof(fileinfo_t *) * (fn -> nsubs + 1));
|
|
414 fn -> subs[fn -> nsubs] = lw_malloc(sizeof(fileinfo_t));
|
|
415 memset(fn -> subs[fn -> nsubs], 0, sizeof(fileinfo_t));
|
|
416 fn -> subs[fn -> nsubs] -> filedata = fn -> filedata + cc;
|
|
417 fn -> subs[fn -> nsubs] -> filesize = flen;
|
|
418 fn -> subs[fn -> nsubs] -> filename = lw_strdup(fn -> filedata + l);
|
|
419 fn -> subs[fn -> nsubs] -> parent = fn;
|
|
420 fn -> subs[fn -> nsubs] -> forced = fn -> forced;
|
|
421 read_file(fn -> subs[fn -> nsubs]);
|
|
422 fn -> nsubs++;
|
|
423 cc += flen;
|
|
424 }
|
|
425 }
|