xref: /titanic_52/usr/src/tools/aw/aw.c (revision 25c28e83beb90e7c80452a7c818c5e6f73a07dc8)
1 /*
2  * CDDL HEADER START
3  *
4  * The contents of this file are subject to the terms of the
5  * Common Development and Distribution License (the "License").
6  * You may not use this file except in compliance with the License.
7  *
8  * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9  * or http://www.opensolaris.org/os/licensing.
10  * See the License for the specific language governing permissions
11  * and limitations under the License.
12  *
13  * When distributing Covered Code, include this CDDL HEADER in each
14  * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15  * If applicable, add the following below this CDDL HEADER, with the
16  * fields enclosed by brackets "[]" replaced with your own identifying
17  * information: Portions Copyright [yyyy] [name of copyright owner]
18  *
19  * CDDL HEADER END
20  */
21 
22 /*
23  * Copyright 2010 Sun Microsystems, Inc.  All rights reserved.
24  * Use is subject to license terms.
25  */
26 
27 /*
28  * Wrapper for the GNU assembler to make it accept the Sun assembler
29  * arguments where possible.
30  *
31  * There are several limitations; the Sun assembler takes multiple
32  * source files, we only take one.
33  *
34  * -b, -s, -xF, -T plain not supported.
35  * -S isn't supported either, because while GNU as does generate
36  * listings with -a, there's no obvious mapping between sub-options.
37  * -K pic, -K PIC not supported either, though it's not clear what
38  * these actually do ..
39  * -Qy (not supported) adds a string to the .comment section
40  * describing the assembler version, while
41  * -Qn (supported) suppresses the string (also the default).
42  *
43  * We also add '-#' support to see invocation lines..
44  * We also add '-xarch=amd64' in case we need to feed the assembler
45  * something different (or in case we need to invoke a different binary
46  * altogether!)
47  */
48 
49 #include <sys/types.h>
50 #include <sys/wait.h>
51 #include <stdio.h>
52 #include <unistd.h>
53 #include <string.h>
54 #include <stdlib.h>
55 #include <sys/param.h>
56 
57 static const char *progname;
58 static int verbose;
59 
60 struct aelist {
61 	int ael_argc;
62 	struct ae {
63 		struct ae *ae_next;
64 		char *ae_arg;
65 	} *ael_head, *ael_tail;
66 };
67 
68 static struct aelist *
69 newael(void)
70 {
71 	return (calloc(sizeof (struct aelist), 1));
72 }
73 
74 static void
75 newae(struct aelist *ael, const char *arg)
76 {
77 	struct ae *ae;
78 
79 	ae = calloc(sizeof (*ae), 1);
80 	ae->ae_arg = strdup(arg);
81 	if (ael->ael_tail == NULL)
82 		ael->ael_head = ae;
83 	else
84 		ael->ael_tail->ae_next = ae;
85 	ael->ael_tail = ae;
86 	ael->ael_argc++;
87 }
88 
89 static void
90 fixae_arg(struct ae *ae, const char *newarg)
91 {
92 	free(ae->ae_arg);
93 	ae->ae_arg = strdup(newarg);
94 }
95 
96 static char **
97 aeltoargv(struct aelist *ael)
98 {
99 	struct ae *ae;
100 	char **argv;
101 	int argc;
102 
103 	argv = calloc(sizeof (*argv), ael->ael_argc + 1);
104 
105 	for (argc = 0, ae = ael->ael_head; ae; ae = ae->ae_next, argc++) {
106 		argv[argc] = ae->ae_arg;
107 		if (ae == ael->ael_tail)
108 			break;
109 	}
110 
111 	return (argv);
112 }
113 
114 static int
115 error(const char *arg)
116 {
117 	(void) fprintf(stderr,
118 	    "%s: as->gas mapping failed at or near arg '%s'\n", progname, arg);
119 	return (2);
120 }
121 
122 static int
123 usage(const char *arg)
124 {
125 	if (arg != NULL)
126 		(void) fprintf(stderr, "error: %s\n", arg);
127 	(void) fprintf(stderr, "Usage: %s [-V] [-#]\n"
128 	    "\t[-xarch=architecture]\n"
129 	    "\t[-o objfile] [-L]\n"
130 	    "\t[-P [[-Ipath] [-Dname] [-Dname=def] [-Uname]]...]\n"
131 	    "\t[-m] [-n] file.s ...\n", progname);
132 	return (3);
133 }
134 
135 static void
136 copyuntil(FILE *in, FILE *out, int termchar)
137 {
138 	int c;
139 
140 	while ((c = fgetc(in)) != EOF) {
141 		if (out && fputc(c, out) == EOF)
142 			exit(1);
143 		if (c == termchar)
144 			break;
145 	}
146 }
147 
148 /*
149  * Variant of copyuntil(), used for copying the path used
150  * for .file directives. This version removes the workspace
151  * from the head of the path, or failing that, attempts to remove
152  * /usr/include. This is a workaround for the way gas handles
153  * these directives. The objects produced by gas contain STT_FILE
154  * symbols for every .file directive. These FILE symbols contain our
155  * workspace paths, leading to wsdiff incorrectly flagging them as
156  * having changed. By clipping off the workspace from these paths,
157  * we eliminate these false positives.
158  */
159 static void
160 copyuntil_path(FILE *in, FILE *out, int termchar,
161     const char *wspace, size_t wspace_len)
162 {
163 #define	PROTO_INC "/proto/root_i386/usr/include/"
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 	if ((cmd = getenv("AW_AS")) != NULL)
507 		strlcpy(as_pgm, cmd, sizeof (as_pgm));
508 	else {
509 		if ((dir = getenv("AW_AS_DIR")) == NULL)
510 			dir = DEFAULT_AS_DIR;	/* /usr/sfw/bin */
511 		(void) snprintf(as_pgm, sizeof (as_pgm), "%s/gas", dir);
512 	}
513 
514 	if ((cmd = getenv("AW_AS64")) != NULL)
515 		strlcpy(as64_pgm, cmd, sizeof (as64_pgm));
516 	else {
517 		if ((dir = getenv("AW_AS64_DIR")) == NULL)
518 			dir = DEFAULT_AS64_DIR;	/* /usr/sfw/bin */
519 		(void) snprintf(as64_pgm, sizeof (as_pgm), "%s/gas", dir);
520 	}
521 
522 	if ((cmd = getenv("AW_M4")) != NULL)
523 		strlcpy(m4_pgm, cmd, sizeof (m4_pgm));
524 	else {
525 		if ((dir = getenv("AW_M4_DIR")) == NULL)
526 			dir = DEFAULT_M4_DIR;	/* /usr/ccs/bin */
527 		(void) snprintf(m4_pgm, sizeof (m4_pgm), "%s/m4", dir);
528 	}
529 
530 	if ((cmd = getenv("AW_M4LIB")) != NULL)
531 		strlcpy(m4_cmdefs, cmd, sizeof (m4_cmdefs));
532 	else {
533 		if ((dir = getenv("AW_M4LIB_DIR")) == NULL)
534 			dir = DEFAULT_M4LIB_DIR;	/* /usr/ccs/lib */
535 		(void) snprintf(m4_cmdefs, sizeof (m4_cmdefs),
536 		    "%s/cm4defs", dir);
537 	}
538 
539 	if ((cmd = getenv("AW_CPP")) != NULL)
540 		strlcpy(cpp_pgm, cmd, sizeof (cpp_pgm));
541 	else {
542 		if ((dir = getenv("AW_CPP_DIR")) == NULL)
543 			dir = DEFAULT_CPP_DIR;	/* /usr/ccs/lib */
544 		(void) snprintf(cpp_pgm, sizeof (cpp_pgm), "%s/cpp", dir);
545 	}
546 
547 	newae(as, as_pgm);
548 	newae(as, "--warn");
549 	newae(as, "--fatal-warnings");
550 	newae(as, "--traditional-format");
551 
552 	/*
553 	 * Walk the argument list, translating as we go ..
554 	 */
555 	while (--argc > 0) {
556 		char *arg;
557 		int arglen;
558 
559 		arg = *++argv;
560 		arglen = strlen(arg);
561 
562 		if (*arg != '-') {
563 			char *filename;
564 
565 			/*
566 			 * filenames ending in '.s' are taken to be
567 			 * assembler files, and provide the default
568 			 * basename of the output file.
569 			 *
570 			 * other files are passed through to the
571 			 * preprocessor, if present, or to gas if not.
572 			 */
573 			filename = arg;
574 			if ((arglen > 2) &&
575 			    ((strcmp(arg + arglen - 2, ".s") == 0) ||
576 			    (strcmp(arg + arglen - 2, ".S") == 0))) {
577 				/*
578 				 * Though 'as' allows multiple assembler
579 				 * files to be processed in one invocation
580 				 * of the assembler, ON only processes one
581 				 * file at a time, which makes things a lot
582 				 * simpler!
583 				 */
584 				if (srcfile == NULL)
585 					srcfile = arg;
586 				else
587 					return (usage(
588 					    "one assembler file at a time"));
589 
590 				/*
591 				 * If we haven't seen a -o option yet,
592 				 * default the output to the basename
593 				 * of the input, substituting a .o on the end
594 				 */
595 				if (outfile == NULL) {
596 					char *argcopy;
597 
598 					argcopy = strdup(arg);
599 					argcopy[arglen - 1] = 'o';
600 
601 					if ((outfile = strrchr(
602 					    argcopy, '/')) == NULL)
603 						outfile = argcopy;
604 					else
605 						outfile++;
606 				}
607 			}
608 			if (cpp)
609 				newae(cpp, filename);
610 			else if (m4)
611 				newae(m4, filename);
612 			else
613 				newae(as, filename);
614 			continue;
615 		} else
616 			arglen--;
617 
618 		switch (arg[1]) {
619 		case 'K':
620 			/*
621 			 * -K pic
622 			 * -K PIC
623 			 */
624 			if (arglen == 1) {
625 				if ((arg = *++argv) == NULL || *arg == '\0')
626 					return (usage("malformed -K"));
627 				argc--;
628 			} else {
629 				arg += 2;
630 			}
631 			if (strcmp(arg, "PIC") != 0 && strcmp(arg, "pic") != 0)
632 				return (usage("malformed -K"));
633 			break;		/* just ignore -Kpic for gcc */
634 		case 'Q':
635 			if (strcmp(arg, "-Qn") == 0)
636 				break;
637 			/*FALLTHROUGH*/
638 		case 'b':
639 		case 's':
640 		case 'T':
641 			/*
642 			 * -b	Extra symbol table for source browser ..
643 			 *	not relevant to gas, thus should error.
644 			 * -s	Put stabs in .stabs section not stabs.excl
645 			 *	not clear if there's an equivalent
646 			 * -T	4.x migration option
647 			 */
648 		default:
649 			return (error(arg));
650 		case 'x':
651 			/*
652 			 * Accept -xarch special case to invoke alternate
653 			 * assemblers or assembler flags for different
654 			 * architectures.
655 			 */
656 			if (strcmp(arg, "-xarch=amd64") == 0 ||
657 			    strcmp(arg, "-xarch=generic64") == 0) {
658 				as64++;
659 				fixae_arg(as->ael_head, as64_pgm);
660 				break;
661 			}
662 			/*
663 			 * XX64: Is this useful to gas?
664 			 */
665 			if (strcmp(arg, "-xmodel=kernel") == 0)
666 				break;
667 
668 			/*
669 			 * -xF	Generates performance analysis data
670 			 *	no equivalent
671 			 */
672 			return (error(arg));
673 		case 'V':
674 			newae(as, arg);
675 			break;
676 		case '#':
677 			verbose++;
678 			break;
679 		case 'L':
680 			newae(as, "--keep-locals");
681 			break;
682 		case 'n':
683 			newae(as, "--no-warn");
684 			break;
685 		case 'o':
686 			if (arglen != 1)
687 				return (usage("bad -o flag"));
688 			if ((arg = *++argv) == NULL || *arg == '\0')
689 				return (usage("bad -o flag"));
690 			outfile = arg;
691 			argc--;
692 			arglen = strlen(arg + 1);
693 			break;
694 		case 'm':
695 			if (cpp)
696 				return (usage("-m conflicts with -P"));
697 			if (m4 == NULL) {
698 				m4 = newael();
699 				newae(m4, m4_pgm);
700 				newae(m4, m4_cmdefs);
701 			}
702 			break;
703 		case 'P':
704 			if (m4)
705 				return (usage("-P conflicts with -m"));
706 			if (cpp == NULL) {
707 				cpp = newael();
708 				newae(cpp, cpp_pgm);
709 				newae(cpp, "-D__GNUC_AS__");
710 			}
711 			break;
712 		case 'D':
713 		case 'U':
714 			if (cpp)
715 				newae(cpp, arg);
716 			else if (m4)
717 				newae(m4, arg);
718 			else
719 				newae(as, arg);
720 			break;
721 		case 'I':
722 			if (cpp)
723 				newae(cpp, arg);
724 			else
725 				newae(as, arg);
726 			break;
727 		case '-':	/* a gas-specific option */
728 			newae(as, arg);
729 			break;
730 		}
731 	}
732 
733 #if defined(__i386)
734 	if (as64)
735 		newae(as, "--64");
736 	else
737 		newae(as, "--32");
738 #endif
739 
740 	if (srcfile == NULL)
741 		return (usage("no source file(s) specified"));
742 	if (outfile == NULL)
743 		outfile = "a.out";
744 	newae(as, "-o");
745 	newae(as, outfile);
746 
747 	asargv = aeltoargv(as);
748 	if (cpp) {
749 #if defined(__sparc)
750 		newae(cpp, "-Dsparc");
751 		newae(cpp, "-D__sparc");
752 		if (as64)
753 			newae(cpp, "-D__sparcv9");
754 		else
755 			newae(cpp, "-D__sparcv8");
756 #elif defined(__i386) || defined(__x86)
757 		if (as64) {
758 			newae(cpp, "-D__x86_64");
759 			newae(cpp, "-D__amd64");
760 		} else {
761 			newae(cpp, "-Di386");
762 			newae(cpp, "-D__i386");
763 		}
764 #else
765 #error	"need isa-dependent defines"
766 #endif
767 		code = pipeline(aeltoargv(cpp), asargv);
768 	} else if (m4)
769 		code = pipeline(aeltoargv(m4), asargv);
770 	else {
771 		/*
772 		 * XXX	should arrange to fork/exec so that we
773 		 *	can unlink the output file if errors are
774 		 *	detected..
775 		 */
776 		(void) execvp(asargv[0], asargv);
777 		perror("execvp");
778 		(void) fprintf(stderr, "%s: couldn't run %s\n",
779 		    progname, asargv[0]);
780 		code = 7;
781 	}
782 	if (code != 0)
783 		(void) unlink(outfile);
784 	return (code);
785 }
786