1
2 /*
3 * CDDL HEADER START
4 *
5 * The contents of this file are subject to the terms of the
6 * Common Development and Distribution License (the "License").
7 * You may not use this file except in compliance with the License.
8 *
9 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
10 * or http://www.opensolaris.org/os/licensing.
11 * See the License for the specific language governing permissions
12 * and limitations under the License.
13 *
14 * When distributing Covered Code, include this CDDL HEADER in each
15 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
16 * If applicable, add the following below this CDDL HEADER, with the
17 * fields enclosed by brackets "[]" replaced with your own identifying
18 * information: Portions Copyright [yyyy] [name of copyright owner]
19 *
20 * CDDL HEADER END
21 */
22
23 /*
24 * Copyright 2010 Sun Microsystems, Inc. All rights reserved.
25 * Use is subject to license terms.
26 */
27
28 /*
29 * Wrapper for the GNU assembler to make it accept the Sun assembler
30 * arguments where possible.
31 *
32 * There are several limitations; the Sun assembler takes multiple
33 * source files, we only take one.
34 *
35 * -b, -s, -xF, -T plain not supported.
36 * -S isn't supported either, because while GNU as does generate
37 * listings with -a, there's no obvious mapping between sub-options.
38 * -K pic, -K PIC not supported either, though it's not clear what
39 * these actually do ..
40 * -Qy (not supported) adds a string to the .comment section
41 * describing the assembler version, while
42 * -Qn (supported) suppresses the string (also the default).
43 *
44 * We also add '-#' support to see invocation lines..
45 * We also add '-xarch=amd64' in case we need to feed the assembler
46 * something different (or in case we need to invoke a different binary
47 * altogether!)
48 */
49
50 #include <sys/types.h>
51 #include <sys/wait.h>
52 #include <stdio.h>
53 #include <unistd.h>
54 #include <string.h>
55 #include <stdlib.h>
56 #include <sys/param.h>
57
58 static const char *progname;
59 static int verbose;
60
61 struct aelist {
62 int ael_argc;
63 struct ae {
64 struct ae *ae_next;
65 char *ae_arg;
66 } *ael_head, *ael_tail;
67 };
68
69 static struct aelist *
70 newael(void)
71 {
72 return (calloc(sizeof (struct aelist), 1));
73 }
74
75 static void
76 newae(struct aelist *ael, const char *arg)
77 {
78 struct ae *ae;
79
80 ae = calloc(sizeof (*ae), 1);
81 ae->ae_arg = strdup(arg);
82 if (ael->ael_tail == NULL)
83 ael->ael_head = ae;
84 else
85 ael->ael_tail->ae_next = ae;
86 ael->ael_tail = ae;
87 ael->ael_argc++;
88 }
89
90 static void
91 fixae_arg(struct ae *ae, const char *newarg)
92 {
93 free(ae->ae_arg);
94 ae->ae_arg = strdup(newarg);
95 }
96
97 static char **
98 aeltoargv(struct aelist *ael)
99 {
100 struct ae *ae;
101 char **argv;
102 int argc;
103
104 argv = calloc(sizeof (*argv), ael->ael_argc + 1);
105
106 for (argc = 0, ae = ael->ael_head; ae; ae = ae->ae_next, argc++) {
107 argv[argc] = ae->ae_arg;
108 if (ae == ael->ael_tail)
109 break;
110 }
111
112 return (argv);
113 }
114
115 static int
116 error(const char *arg)
117 {
118 (void) fprintf(stderr,
119 "%s: as->gas mapping failed at or near arg '%s'\n", progname, arg);
120 return (2);
121 }
122
123 static int
124 usage(const char *arg)
125 {
126 if (arg != NULL)
127 (void) fprintf(stderr, "error: %s\n", arg);
128 (void) fprintf(stderr, "Usage: %s [-V] [-#]\n"
129 "\t[-xarch=architecture]\n"
130 "\t[-o objfile] [-L]\n"
131 "\t[-P [[-Ipath] [-Dname] [-Dname=def] [-Uname]]...]\n"
132 "\t[-m] [-n] file.s ...\n", progname);
133 return (3);
134 }
135
136 static void
137 copyuntil(FILE *in, FILE *out, int termchar)
138 {
139 int c;
140
141 while ((c = fgetc(in)) != EOF) {
142 if (out && fputc(c, out) == EOF)
143 exit(1);
144 if (c == termchar)
145 break;
146 }
147 }
148
149 /*
150 * Variant of copyuntil(), used for copying the path used
151 * for .file directives. This version removes the workspace
152 * from the head of the path, or failing that, attempts to remove
153 * /usr/include. This is a workaround for the way gas handles
154 * these directives. The objects produced by gas contain STT_FILE
155 * symbols for every .file directive. These FILE symbols contain our
156 * workspace paths, leading to wsdiff incorrectly flagging them as
157 * having changed. By clipping off the workspace from these paths,
158 * we eliminate these false positives.
159 */
160 static void
161 copyuntil_path(FILE *in, FILE *out, int termchar,
162 const char *wspace, size_t wspace_len)
163 {
164 #define SYS_INC "/usr/include/"
165
166 static const size_t proto_inc_len = sizeof (PROTO_INC) - 1;
167 static const size_t sys_inc_len = sizeof (SYS_INC) - 1;
168
169 /*
170 * Dynamically sized buffer for reading paths. Retained
171 * and reused between calls.
172 */
173 static char *buf = NULL;
174 static size_t bufsize = 0;
175
176 size_t bufcnt = 0;
177 char *bufptr;
178 int c;
179
180 /* Read the path into the buffer */
181 while ((c = fgetc(in)) != EOF) {
182 /*
183 * If we need a buffer, or need a larger buffer,
184 * fix that here.
185 */
186 if (bufcnt >= bufsize) {
187 bufsize = (bufsize == 0) ? MAXPATHLEN : (bufsize * 2);
188 buf = realloc(buf, bufsize + 1); /* + room for NULL */
189 if (buf == NULL) {
190 perror("realloc");
191 exit(1);
192 }
193 }
194
195 buf[bufcnt++] = c;
196 if (c == termchar)
197 break;
198 }
199 if (bufcnt == 0)
200 return;
201
202 /*
203 * We have a non-empty buffer, and thus the opportunity
204 * to do some surgery on it before passing it to the output.
205 */
206 buf[bufcnt] = '\0';
207 bufptr = buf;
208
209 /*
210 * If our workspace is at the start, remove it.
211 * If not, then look for the system /usr/include instead.
212 */
213 if ((wspace_len > 0) && (wspace_len < bufcnt) &&
214 (strncmp(bufptr, wspace, wspace_len) == 0)) {
215 bufptr += wspace_len;
216 bufcnt -= wspace_len;
217
218 /*
219 * Further opportunity: Also clip the prefix
220 * that leads to /usr/include in the proto.
221 */
222 if ((proto_inc_len < bufcnt) &&
223 (strncmp(bufptr, PROTO_INC, proto_inc_len) == 0)) {
224 bufptr += proto_inc_len;
225 bufcnt -= proto_inc_len;
226 }
227 } else if ((sys_inc_len < bufcnt) &&
228 (strncmp(bufptr, SYS_INC, sys_inc_len) == 0)) {
229 bufptr += sys_inc_len;
230 bufcnt -= sys_inc_len;
231 }
232
233 /* Output whatever is left */
234 if (out && (fwrite(bufptr, 1, bufcnt, out) != bufcnt)) {
235 perror("fwrite");
236 exit(1);
237 }
238
239 #undef PROTO_INC
240 #undef SYS_INC
241 }
242
243 /*
244 * The idea here is to take directives like this emitted
245 * by cpp:
246 *
247 * # num
248 *
249 * and convert them to directives like this that are
250 * understood by the GNU assembler:
251 *
252 * .line num
253 *
254 * and similarly:
255 *
256 * # num "string" optional stuff
257 *
258 * is converted to
259 *
260 * .line num
261 * .file "string"
262 *
263 * While this could be done with a sequence of sed
264 * commands, this is simpler and faster..
265 */
266 static pid_t
267 filter(int pipein, int pipeout)
268 {
269 pid_t pid;
270 FILE *in, *out;
271 char *wspace;
272 size_t wspace_len;
273
274 if (verbose)
275 (void) fprintf(stderr, "{#line filter} ");
276
277 switch (pid = fork()) {
278 case 0:
279 if (dup2(pipein, 0) == -1 ||
280 dup2(pipeout, 1) == -1) {
281 perror("dup2");
282 exit(1);
283 }
284 closefrom(3);
285 break;
286 case -1:
287 perror("fork");
288 default:
289 return (pid);
290 }
291
292 in = fdopen(0, "r");
293 out = fdopen(1, "w");
294
295 /*
296 * Key off the CODEMGR_WS environment variable to detect
297 * if we're in an activated workspace, and to get the
298 * path to the workspace.
299 */
300 wspace = getenv("CODEMGR_WS");
301 if (wspace != NULL)
302 wspace_len = strlen(wspace);
303
304 while (!feof(in)) {
305 int c, num;
306
307 switch (c = fgetc(in)) {
308 case '#':
309 switch (fscanf(in, " %d", &num)) {
310 case 0:
311 /*
312 * discard comment lines completely
313 * discard ident strings completely too.
314 * (GNU as politely ignores them..)
315 */
316 copyuntil(in, NULL, '\n');
317 break;
318 default:
319 (void) fprintf(stderr, "fscanf botch?");
320 /*FALLTHROUGH*/
321 case EOF:
322 exit(1);
323 /*NOTREACHED*/
324 case 1:
325 /*
326 * This line has a number at the beginning;
327 * if it has a string after the number, then
328 * it's a filename.
329 *
330 * If this is an activated workspace, use
331 * copyuntil_path() to do path rewriting
332 * that will prevent workspace paths from
333 * being burned into the resulting object.
334 * If not in an activated workspace, then
335 * copy the existing path straight through
336 * without interpretation.
337 */
338 if (fgetc(in) == ' ' && fgetc(in) == '"') {
339 (void) fprintf(out, "\t.file \"");
340 if (wspace != NULL)
341 copyuntil_path(in, out, '"',
342 wspace, wspace_len);
343 else
344 copyuntil(in, out, '"');
345 (void) fputc('\n', out);
346 }
347 (void) fprintf(out, "\t.line %d\n", num - 1);
348 /*
349 * discard the rest of the line
350 */
351 copyuntil(in, NULL, '\n');
352 break;
353 }
354 break;
355 case '\n':
356 /*
357 * preserve newlines
358 */
359 (void) fputc(c, out);
360 break;
361 case EOF:
362 /*
363 * don't write EOF!
364 */
365 break;
366 default:
367 /*
368 * lines that don't begin with '#' are copied
369 */
370 (void) fputc(c, out);
371 copyuntil(in, out, '\n');
372 break;
373 }
374
375 if (ferror(out))
376 exit(1);
377 }
378
379 exit(0);
380 /*NOTREACHED*/
381 }
382
383 static pid_t
384 invoke(char **argv, int pipein, int pipeout)
385 {
386 pid_t pid;
387
388 if (verbose) {
389 char **dargv = argv;
390
391 while (*dargv)
392 (void) fprintf(stderr, "%s ", *dargv++);
393 }
394
395 switch (pid = fork()) {
396 case 0:
397 if (pipein >= 0 && dup2(pipein, 0) == -1) {
398 perror("dup2");
399 exit(1);
400 }
401 if (pipeout >= 0 && dup2(pipeout, 1) == -1) {
402 perror("dup2");
403 exit(1);
404 }
405 closefrom(3);
406 (void) execvp(argv[0], argv);
407 perror("execvp");
408 (void) fprintf(stderr, "%s: couldn't run %s\n",
409 progname, argv[0]);
410 break;
411 case -1:
412 perror("fork");
413 default:
414 return (pid);
415 }
416 exit(2);
417 /*NOTREACHED*/
418 }
419
420 static int
421 pipeline(char **ppargv, char **asargv)
422 {
423 int pipedes[4];
424 int active = 0;
425 int rval = 0;
426 pid_t pid_pp, pid_f, pid_as;
427
428 if (pipe(pipedes) == -1 || pipe(pipedes + 2) == -1) {
429 perror("pipe");
430 return (4);
431 }
432
433 if ((pid_pp = invoke(ppargv, -1, pipedes[0])) > 0)
434 active++;
435
436 if (verbose)
437 (void) fprintf(stderr, "| ");
438
439 if ((pid_f = filter(pipedes[1], pipedes[2])) > 0)
440 active++;
441
442 if (verbose)
443 (void) fprintf(stderr, "| ");
444
445 if ((pid_as = invoke(asargv, pipedes[3], -1)) > 0)
446 active++;
447
448 if (verbose) {
449 (void) fprintf(stderr, "\n");
450 (void) fflush(stderr);
451 }
452
453 closefrom(3);
454
455 if (active != 3)
456 return (5);
457
458 while (active != 0) {
459 pid_t pid;
460 int stat;
461
462 if ((pid = wait(&stat)) == -1) {
463 rval++;
464 break;
465 }
466
467 if (!WIFEXITED(stat))
468 continue;
469
470 if (pid == pid_pp || pid == pid_f || pid == pid_as) {
471 active--;
472 if (WEXITSTATUS(stat) != 0)
473 rval++;
474 }
475 }
476
477 return (rval);
478 }
479
480 int
481 main(int argc, char *argv[])
482 {
483 struct aelist *cpp = NULL;
484 struct aelist *m4 = NULL;
485 struct aelist *as = newael();
486 char **asargv;
487 char *outfile = NULL;
488 char *srcfile = NULL;
489 const char *dir, *cmd;
490 static char as_pgm[MAXPATHLEN];
491 static char as64_pgm[MAXPATHLEN];
492 static char m4_pgm[MAXPATHLEN];
493 static char m4_cmdefs[MAXPATHLEN];
494 static char cpp_pgm[MAXPATHLEN];
495 int as64 = 0;
496 int code;
497
498 if ((progname = strrchr(argv[0], '/')) == NULL)
499 progname = argv[0];
500 else
501 progname++;
502
503 /*
504 * Helpful when debugging, or when changing tool versions..
505 */
506 cmd = dir = NULL;
507 if ((cmd = getenv("AW_" AW_TARGET "_AS")) == NULL)
508 cmd = getenv("AW_AS");
509 if ((dir = getenv("AW_" AW_TARGET "_AS_DIR")) == NULL)
510 dir = getenv("AW_AS_DIR");
511
512 if (cmd != NULL) {
513 (void) strlcpy(as_pgm, cmd, sizeof (as_pgm));
514 } else {
515 if (dir == NULL)
516 dir = DEFAULT_AS_DIR;
517 (void) snprintf(as_pgm, sizeof (as_pgm), "%s/gas", dir);
518 }
519
520 cmd = dir = NULL;
521 if ((cmd = getenv("AW_" AW_TARGET "_AS64")) == NULL)
522 cmd = getenv("AW_AS64");
523 if ((dir = getenv("AW_" AW_TARGET "_AS64_DIR")) == NULL)
524 dir = getenv("AW_AS64_DIR");
525
526 if (cmd != NULL) {
527 (void) strlcpy(as64_pgm, cmd, sizeof (as64_pgm));
528 } else {
529 if (dir == NULL)
530 dir = DEFAULT_AS64_DIR;
531 (void) snprintf(as64_pgm, sizeof (as_pgm), "%s/gas", dir);
532 }
533
534 if ((cmd = getenv("AW_M4")) != NULL)
535 strlcpy(m4_pgm, cmd, sizeof (m4_pgm));
536 else {
537 if ((dir = getenv("AW_M4_DIR")) == NULL)
538 dir = DEFAULT_M4_DIR; /* /usr/ccs/bin */
539 (void) snprintf(m4_pgm, sizeof (m4_pgm), "%s/m4", dir);
540 }
541
542 if ((cmd = getenv("AW_M4LIB")) != NULL)
543 strlcpy(m4_cmdefs, cmd, sizeof (m4_cmdefs));
544 else {
545 if ((dir = getenv("AW_M4LIB_DIR")) == NULL)
546 dir = DEFAULT_M4LIB_DIR; /* /usr/ccs/lib */
547 (void) snprintf(m4_cmdefs, sizeof (m4_cmdefs),
548 "%s/cm4defs", dir);
549 }
550
551 if ((cmd = getenv("AW_CPP")) != NULL)
552 strlcpy(cpp_pgm, cmd, sizeof (cpp_pgm));
553 else {
554 if ((dir = getenv("AW_CPP_DIR")) == NULL)
555 dir = DEFAULT_CPP_DIR; /* /usr/ccs/lib */
556 (void) snprintf(cpp_pgm, sizeof (cpp_pgm), "%s/cpp", dir);
557 }
558
559 newae(as, as_pgm);
560 newae(as, "--warn");
561 newae(as, "--fatal-warnings");
562 newae(as, "--traditional-format");
563
564 /*
565 * Walk the argument list, translating as we go ..
566 */
567 while (--argc > 0) {
568 char *arg;
569 int arglen;
570
571 arg = *++argv;
572 arglen = strlen(arg);
573
574 if (*arg != '-') {
575 char *filename;
576
577 /*
578 * filenames ending in '.s' are taken to be
579 * assembler files, and provide the default
580 * basename of the output file.
581 *
582 * other files are passed through to the
583 * preprocessor, if present, or to gas if not.
584 */
585 filename = arg;
586 if ((arglen > 2) &&
587 ((strcmp(arg + arglen - 2, ".s") == 0) ||
588 (strcmp(arg + arglen - 2, ".S") == 0))) {
589 /*
590 * Though 'as' allows multiple assembler
591 * files to be processed in one invocation
592 * of the assembler, ON only processes one
593 * file at a time, which makes things a lot
594 * simpler!
595 */
596 if (srcfile == NULL)
597 srcfile = arg;
598 else
599 return (usage(
600 "one assembler file at a time"));
601
602 /*
603 * If we haven't seen a -o option yet,
604 * default the output to the basename
605 * of the input, substituting a .o on the end
606 */
607 if (outfile == NULL) {
608 char *argcopy;
609
610 argcopy = strdup(arg);
611 argcopy[arglen - 1] = 'o';
612
613 if ((outfile = strrchr(
614 argcopy, '/')) == NULL)
615 outfile = argcopy;
616 else
617 outfile++;
618 }
619 }
620 if (cpp)
621 newae(cpp, filename);
622 else if (m4)
623 newae(m4, filename);
624 else
625 newae(as, filename);
626 continue;
627 } else
628 arglen--;
629
630 switch (arg[1]) {
631 case 'K':
632 /*
633 * -K pic
634 * -K PIC
635 */
636 if (arglen == 1) {
637 if ((arg = *++argv) == NULL || *arg == '\0')
638 return (usage("malformed -K"));
639 argc--;
640 } else {
641 arg += 2;
642 }
643 if (strcmp(arg, "PIC") != 0 && strcmp(arg, "pic") != 0)
644 return (usage("malformed -K"));
645 break; /* just ignore -Kpic for gcc */
646 case 'Q':
647 if (strcmp(arg, "-Qn") == 0)
648 break;
649 /*FALLTHROUGH*/
650 case 'b':
651 case 's':
652 case 'T':
653 /*
654 * -b Extra symbol table for source browser ..
655 * not relevant to gas, thus should error.
656 * -s Put stabs in .stabs section not stabs.excl
657 * not clear if there's an equivalent
658 * -T 4.x migration option
659 */
660 default:
661 return (error(arg));
662 case 'x':
663 /*
664 * Accept -xarch special case to invoke alternate
665 * assemblers or assembler flags for different
666 * architectures.
667 */
668 if (strcmp(arg, "-xarch=amd64") == 0 ||
669 strcmp(arg, "-xarch=generic64") == 0) {
670 as64++;
671 fixae_arg(as->ael_head, as64_pgm);
672 break;
673 }
674 /*
675 * XX64: Is this useful to gas?
676 */
677 if (strcmp(arg, "-xmodel=kernel") == 0)
678 break;
679
680 /*
681 * -xF Generates performance analysis data
682 * no equivalent
683 */
684 return (error(arg));
685 case 'V':
686 newae(as, arg);
687 break;
688 case '#':
689 verbose++;
690 break;
691 case 'L':
692 newae(as, "--keep-locals");
693 break;
694 case 'n':
695 newae(as, "--no-warn");
696 break;
697 case 'o':
698 if (arglen != 1)
699 return (usage("bad -o flag"));
700 if ((arg = *++argv) == NULL || *arg == '\0')
701 return (usage("bad -o flag"));
702 outfile = arg;
703 argc--;
704 arglen = strlen(arg + 1);
705 break;
706 case 'm':
707 if (cpp)
708 return (usage("-m conflicts with -P"));
709 if (m4 == NULL) {
710 m4 = newael();
711 newae(m4, m4_pgm);
712 newae(m4, m4_cmdefs);
713 }
714 break;
715 case 'P':
716 if (m4)
717 return (usage("-P conflicts with -m"));
718 if (cpp == NULL) {
719 cpp = newael();
720 newae(cpp, cpp_pgm);
721 newae(cpp, "-D__GNUC_AS__");
722 }
723 break;
724 case 'D':
725 case 'U':
726 if (cpp)
727 newae(cpp, arg);
728 else if (m4)
729 newae(m4, arg);
730 else
731 newae(as, arg);
732 break;
733 case 'I':
734 if (cpp)
735 newae(cpp, arg);
736 else
737 newae(as, arg);
738 break;
739 case '-': /* a gas-specific option */
740 newae(as, arg);
741 break;
742 }
743 }
744
745 #if defined(AW_TARGET_i386)
746 if (as64)
747 newae(as, "--64");
748 else
749 newae(as, "--32");
750 #endif
751
752
753 /*
754 * gas for 32-bit arm defaults to a much older version of the arm
755 * architecture than we can really support. Because of that, we instead
756 * opt to make sure that we set the minimum architecture to armv6k, the
757 * minimum of what we actually support.
758 */
759 #if defined(AW_TARGET_arm)
760 if (as64) {
761 return (error("no 64-bit aw target for arm"));
762 } else {
763 newae(as, "-march=armv6k");
764 newae(as, "-mfpu=vfpv2");
765 newae(as, "-mfloat-abi=hard");
766 }
767 #endif
768
769 if (srcfile == NULL)
770 return (usage("no source file(s) specified"));
771 if (outfile == NULL)
772 outfile = "a.out";
773 newae(as, "-o");
774 newae(as, outfile);
775
776 asargv = aeltoargv(as);
777 if (cpp) {
778 #if defined(AW_TARGET_sparc)
779 newae(cpp, "-Dsparc");
780 newae(cpp, "-D__sparc");
781 if (as64)
782 newae(cpp, "-D__sparcv9");
783 else
784 newae(cpp, "-D__sparcv8");
785 #elif defined(AW_TARGET_i386)
786 if (as64) {
787 newae(cpp, "-D__x86_64");
788 newae(cpp, "-D__amd64");
789 } else {
790 newae(cpp, "-Di386");
791 newae(cpp, "-D__i386");
792 }
793 #elif defined(AW_TARGET_arm)
794 newae(cpp, "-Darm");
795 newae(cpp, "-D__arm");
796 #else
797 #error "need isa-dependent defines"
798 #endif
799 code = pipeline(aeltoargv(cpp), asargv);
800 } else if (m4)
801 code = pipeline(aeltoargv(m4), asargv);
802 else {
803 /*
804 * XXX should arrange to fork/exec so that we
805 * can unlink the output file if errors are
806 * detected..
807 */
808 (void) execvp(asargv[0], asargv);
809 perror("execvp");
810 (void) fprintf(stderr, "%s: couldn't run %s\n",
811 progname, asargv[0]);
812 code = 7;
813 }
814 if (code != 0)
815 (void) unlink(outfile);
816 return (code);
817 }