xref: /illumos-gate/usr/src/cmd/xargs/xargs.c (revision a7cee4e9766ebda975dd156d1f10a70f51c242f0)
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  * Copyright 2014 Garrett D'Amore <garrett@damore.org>
23  * Copyright 2012 DEY Storage Systems, Inc.
24  * Copyright (c) 2018, Joyent, Inc.
25  * Copyright 2026 Oxide Computer Company
26  *
27  * Portions of this file developed by DEY Storage Systems, Inc. are licensed
28  * under the terms of the Common Development and Distribution License (CDDL)
29  * version 1.0 only.  The use of subsequent versions of the License are
30  * is specifically prohibited unless those terms are not in conflict with
31  * version 1.0 of the License.  You can find this license on-line at
32  * http://www.illumos.org/license/CDDL
33  */
34 /*
35  * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
36  * Use is subject to license terms.
37  */
38 
39 /*	Copyright (c) 1984, 1986, 1987, 1988, 1989 AT&T	*/
40 /*	  All Rights Reserved	*/
41 
42 
43 #include <stdio.h>
44 #include <sys/types.h>
45 #include <sys/wait.h>
46 #include <unistd.h>
47 #include <fcntl.h>
48 #include <string.h>
49 #include <stdarg.h>
50 #include <stdlib.h>
51 #include <limits.h>
52 #include <wchar.h>
53 #include <locale.h>
54 #include <langinfo.h>
55 #include <stropts.h>
56 #include <poll.h>
57 #include <errno.h>
58 #include <spawn.h>
59 #include "getresponse.h"
60 
61 extern char **environ;
62 
63 #define	HEAD	0
64 #define	TAIL	1
65 #define	FALSE 0
66 #define	TRUE 1
67 #define	MAXSBUF 255
68 #define	MAXIBUF 512
69 #define	MAXINSERTS 5
70 #define	BUFSIZE LINE_MAX
71 #define	MAXARGS 255
72 #define	INSPAT_STR	"{}"	/* default replstr string for -[Ii]	*/
73 
74 #define	QBUF_STARTLEN 255  /* start size of growable string buffer */
75 #define	QBUF_INC 100	   /* how much to grow a growable string by */
76 
77 /* We use these macros to help make formatting look "consistent" */
78 #define	EMSG(s)		ermsg(gettext(s "\n"))
79 #define	EMSG2(s, a)	ermsg(gettext(s "\n"), a)
80 #define	PERR(s)		perror(gettext("xargs: " s))
81 
82 /* Some common error messages */
83 
84 #define	LIST2LONG	"Argument list too long"
85 #define	ARG2LONG	"A single argument was greater than %d bytes"
86 #define	MALLOCFAIL	"Memory allocation failure"
87 #define	CORRUPTFILE	"Corrupt input file"
88 #define	WAITFAIL	"Wait failure"
89 #define	CHILDSIG	"Child killed with signal %d"
90 #define	CHILDFAIL	"Command could not continue processing data"
91 #define	FORKFAIL	"Could not fork child"
92 #define	EXECFAIL	"Could not exec command"
93 #define	MISSQUOTE	"Missing quote"
94 #define	BADESCAPE	"Incomplete escape"
95 #define	IBUFOVERFLOW	"Insert buffer overflow"
96 #define	NOCHILDSLOT	"No free child slot available"
97 
98 #define	_(x)	gettext(x)
99 
100 static wctype_t	blank;
101 static char	*arglist[MAXARGS+1];
102 static char	argbuf[BUFSIZE * 2 + 1];
103 static char	lastarg[BUFSIZE + 1];
104 static char	**ARGV = arglist;
105 static char	*LEOF = "_";
106 static char	*INSPAT = INSPAT_STR;
107 static char	ins_buf[MAXIBUF];
108 static char	*p_ibuf;
109 
110 static struct inserts {
111 	char	**p_ARGV;	/* where to put newarg ptr in arg list */
112 	char	*p_skel;	/* ptr to arg template */
113 } saveargv[MAXINSERTS];
114 
115 static int	PROMPT = -1;
116 static int	BUFLIM = BUFSIZE;
117 static int	MAXPROCS = 1;
118 static int	N_ARGS = 0;
119 static int	N_args = 0;
120 static int	N_lines = 0;
121 static int	DASHX = FALSE;
122 static int	MORE = TRUE;
123 static int	PER_LINE = FALSE;
124 static int	LINE_CONT = FALSE;
125 static int	EAT_LEAD = FALSE;
126 static int	ERR = FALSE;
127 static int	OK = TRUE;
128 static int	LEGAL = FALSE;
129 static int	TRACE = FALSE;
130 static int	INSERT = FALSE;
131 static int	ZERO = FALSE;
132 static int	linesize = 0;
133 static int	ibufsize = 0;
134 static int	exitstat = 0;	/* our exit status			*/
135 static int	mac;		/* modified argc, after parsing		*/
136 static char	**mav;		/* modified argv, after parsing		*/
137 static int	n_inserts;	/* # of insertions.			*/
138 static pid_t	*procs;		/* pids of children			*/
139 static int	n_procs;	/* # of child processes.		*/
140 
141 /* our usage message:							*/
142 #define	USAGEMSG "Usage: xargs: [-t] [-p] [-0] [-e[eofstr]] [-E eofstr] "\
143 	"[-I replstr] [-i[replstr]] [-L #] [-l[#]] [-n # [-x]] [-P maxprocs] "\
144 	"[-s size] [cmd [args ...]]\n"
145 
146 static int	echoargs();
147 static wint_t	getwchr(char *, size_t *);
148 static void	lcall(char *sub, char **subargs);
149 static void	addibuf(struct inserts *p);
150 static void	ermsg(char *messages, ...);
151 static char	*addarg(char *arg);
152 static void	store_str(char **, char *, size_t);
153 static char	*getarg(char *);
154 static char	*insert(char *pattern, char *subst);
155 static void	usage();
156 static void	parseargs();
157 static int	procs_find(pid_t child);
158 static void	procs_store(pid_t child);
159 static boolean_t procs_delete(pid_t child);
160 static pid_t	procs_waitpid(boolean_t blocking, int *stat_loc);
161 static void	procs_wait(boolean_t blocking);
162 
163 int
164 main(int argc, char **argv)
165 {
166 	int	j;
167 	long	l;
168 	struct inserts *psave;
169 	int c;
170 	int	initsize;
171 	char	*cmdname, **initlist;
172 	char	*arg;
173 	char	*next;
174 	char	*eptr;
175 
176 	/* initialization */
177 	blank = wctype("blank");
178 	n_inserts = 0;
179 	psave = saveargv;
180 	(void) setlocale(LC_ALL, "");
181 #if !defined(TEXT_DOMAIN)	/* Should be defined by cc -D		*/
182 #define	TEXT_DOMAIN "SYS_TEST"	/* Use this only if it weren't		*/
183 #endif
184 	(void) textdomain(TEXT_DOMAIN);
185 	if (init_yes() < 0) {
186 		ermsg(_(ERR_MSG_INIT_YES), strerror(errno));
187 		exit(1);
188 	}
189 
190 	parseargs(argc, argv);
191 
192 	/* handling all of xargs arguments:				*/
193 	while ((c = getopt(mac, mav, "0tpe:E:I:i:L:l:n:P:s:x")) != EOF) {
194 		switch (c) {
195 		case '0':
196 			ZERO = TRUE;
197 			break;
198 
199 		case 't':	/* -t: turn trace mode on		*/
200 			TRACE = TRUE;
201 			break;
202 
203 		case 'p':	/* -p: turn on prompt mode.		*/
204 			if ((PROMPT = open("/dev/tty", O_RDONLY)) == -1) {
205 				PERR("can't read from tty for -p");
206 			} else {
207 				TRACE = TRUE;
208 			}
209 			break;
210 
211 		case 'e':
212 			/*
213 			 * -e[eofstr]: set/disable end-of-file.
214 			 * N.B. that an argument *isn't* required here; but
215 			 * parseargs forced an argument if not was given.  The
216 			 * forced argument is the default...
217 			 */
218 			LEOF = optarg; /* can be empty */
219 			break;
220 
221 		case 'E':
222 			/*
223 			 * -E eofstr: change end-of-file string.
224 			 * eofstr *is* required here, but can be empty:
225 			 */
226 			LEOF = optarg;
227 			break;
228 
229 		case 'I':
230 			/* -I replstr: Insert mode. replstr *is* required. */
231 			INSERT = PER_LINE = LEGAL = EAT_LEAD = TRUE;
232 			LINE_CONT = FALSE;
233 			N_ARGS = 0;
234 			INSPAT = optarg;
235 			if (*optarg == '\0') {
236 				ermsg(_("Option requires an argument: -%c\n"),
237 				    c);
238 			}
239 			break;
240 
241 		case 'i':
242 			/*
243 			 * -i [replstr]: insert mode, with *optional* replstr.
244 			 * N.B. that an argument *isn't* required here; if
245 			 * it's not given, then the string INSPAT_STR will
246 			 * be assumed.
247 			 *
248 			 * Since getopt(3C) doesn't handle the case of an
249 			 * optional variable argument at all, we have to
250 			 * parse this by hand:
251 			 */
252 
253 			INSERT = PER_LINE = LEGAL = EAT_LEAD = TRUE;
254 			LINE_CONT = FALSE;
255 			N_ARGS = 0;
256 			if ((optarg != NULL) && (*optarg != '\0')) {
257 				INSPAT = optarg;
258 			} else {
259 				/*
260 				 * here, there is no next argument. so
261 				 * we reset INSPAT to the INSPAT_STR.
262 				 * we *have* to do this, as -i/I may have
263 				 * been given previously, and XCU4 requires
264 				 * that only "the last one specified takes
265 				 * effect".
266 				 */
267 				INSPAT = INSPAT_STR;
268 			}
269 			break;
270 
271 		case 'L':
272 			/*
273 			 * -L number: # of times cmd is executed
274 			 * number *is* required here:
275 			 */
276 			PER_LINE = LINE_CONT = TRUE;
277 			N_ARGS = 0;
278 			INSERT = EAT_LEAD = FALSE;
279 			if ((PER_LINE = atoi(optarg)) <= 0) {
280 				ermsg(_("#lines must be positive int: %s\n"),
281 				    optarg);
282 			}
283 			break;
284 
285 		case 'l':
286 			/*
287 			 * -l [number]: # of times cmd is executed
288 			 * N.B. that an argument *isn't* required here; if
289 			 * it's not given, then 1 is assumed.
290 			 *
291 			 * parseargs handles the optional arg processing.
292 			 */
293 
294 			PER_LINE = LINE_CONT = LEGAL = TRUE;
295 			N_ARGS = 0;
296 			INSERT = EAT_LEAD = FALSE;
297 
298 			if ((optarg != NULL) && (*optarg != '\0')) {
299 				if ((PER_LINE = atoi(optarg)) <= 0)
300 					PER_LINE = 1;
301 			}
302 			break;
303 
304 		case 'n':	/* -n number: # stdin args		*/
305 			/*
306 			 * -n number: # stdin args.
307 			 * number *is* required here:
308 			 */
309 			if ((N_ARGS = atoi(optarg)) <= 0) {
310 				ermsg(_("#args must be positive int: %s\n"),
311 				    optarg);
312 			} else {
313 				LEGAL = DASHX || N_ARGS == 1;
314 				INSERT = PER_LINE = LINE_CONT = FALSE;
315 			}
316 			break;
317 
318 		case 'P':	/* -P maxprocs: # of child processses	*/
319 			errno = 0;
320 			l = strtol(optarg, &eptr, 10);
321 			if (*eptr != '\0' || errno != 0) {
322 				ermsg(_("failed to parse maxprocs (-P): %s\n"),
323 				    optarg);
324 				break;
325 			}
326 
327 			if (l < 0) {
328 				ermsg(_("maximum number of processes (-P) "
329 				    "cannot be negative\n"));
330 				break;
331 			}
332 
333 			/*
334 			 * Come up with an upper bound that'll probably fit in
335 			 * memory.
336 			 */
337 			if (l == 0 || l > ((INT_MAX / sizeof (pid_t) >> 1))) {
338 				l = INT_MAX / sizeof (pid_t) >> 1;
339 			}
340 			MAXPROCS = (int)l;
341 			break;
342 
343 		case 's':	/* -s size: set max size of each arg list */
344 			BUFLIM = atoi(optarg);
345 			if (BUFLIM > BUFSIZE || BUFLIM <= 0) {
346 				ermsg(_("0 < max-cmd-line-size <= %d: %s\n"),
347 				    BUFSIZE, optarg);
348 			}
349 			break;
350 
351 		case 'x':	/* -x: terminate if args > size limit	*/
352 			DASHX = LEGAL = TRUE;
353 			break;
354 
355 		default:
356 			/*
357 			 * bad argument. complain and get ready to die.
358 			 */
359 			usage();
360 			exit(2);
361 			break;
362 		}
363 	}
364 
365 	/*
366 	 * if anything called ermsg(), something screwed up, so
367 	 * we exit early.
368 	 */
369 	if (OK == FALSE) {
370 		usage();
371 		exit(2);
372 	}
373 
374 	/*
375 	 * we're finished handling xargs's options, so now pick up
376 	 * the command name (if any), and it's options.
377 	 */
378 
379 
380 	mac -= optind;	/* dec arg count by what we've processed	*/
381 	mav += optind;	/* inc to current mav				*/
382 
383 	procs = calloc(MAXPROCS, sizeof (pid_t));
384 	if (procs == NULL) {
385 		PERR(MALLOCFAIL);
386 		exit(1);
387 	}
388 
389 	if (mac <= 0) {	/* if there're no more args to process,	*/
390 		cmdname = "/usr/bin/echo";	/* our default command	*/
391 		*ARGV++ = addarg(cmdname);	/* use the default cmd.	*/
392 	} else {	/* otherwise keep parsing rest of the string.	*/
393 		/*
394 		 * note that we can't use getopt(3C), and *must* parse
395 		 * this by hand, as we don't know apriori what options the
396 		 * command will take.
397 		 */
398 		cmdname = *mav;	/* get the command name	*/
399 
400 
401 		/* pick up the remaining args from the command line:	*/
402 		while ((OK == TRUE) && (mac-- > 0)) {
403 			/*
404 			 * while we haven't crapped out, and there's
405 			 * work to do:
406 			 */
407 			if (INSERT && ! ERR) {
408 				if (strstr(*mav, INSPAT) != NULL) {
409 					if (++n_inserts > MAXINSERTS) {
410 						ermsg(_("too many args "
411 						    "with %s\n"), INSPAT);
412 						ERR = TRUE;
413 					}
414 					psave->p_ARGV = ARGV;
415 					(psave++)->p_skel = *mav;
416 				}
417 			}
418 			*ARGV++ = addarg(*mav++);
419 		}
420 	}
421 
422 	/* pick up args from standard input */
423 
424 	initlist = ARGV;
425 	initsize = linesize;
426 	lastarg[0] = '\0';
427 
428 	while (OK) {
429 		N_args = 0;
430 		N_lines = 0;
431 		ARGV = initlist;
432 		linesize = initsize;
433 		next = argbuf;
434 
435 		while (MORE || (lastarg[0] != '\0')) {
436 			int l;
437 
438 			if (*lastarg != '\0') {
439 				arg = strcpy(next, lastarg);
440 				*lastarg = '\0';
441 			} else if ((arg = getarg(next)) == NULL) {
442 				break;
443 			}
444 
445 			l = strlen(arg) + 1;
446 			linesize += l;
447 			next += l;
448 
449 			/* Inserts are handled specially later. */
450 			if ((n_inserts == 0) && (linesize >= BUFLIM)) {
451 				/*
452 				 * Legal indicates hard fail if the list is
453 				 * truncated due to size.  So fail, or if we
454 				 * cannot create any list because it would be
455 				 * too big.
456 				 */
457 				if (LEGAL || N_args == 0) {
458 					EMSG(LIST2LONG);
459 					procs_wait(B_TRUE);
460 					exit(2);
461 					/* NOTREACHED */
462 				}
463 
464 				/*
465 				 * Otherwise just save argument for later.
466 				 */
467 				(void) strcpy(lastarg, arg);
468 				break;
469 			}
470 
471 			*ARGV++ = arg;
472 
473 			N_args++;
474 
475 			if ((PER_LINE && (N_lines >= PER_LINE)) ||
476 			    (N_ARGS && (N_args >= N_ARGS))) {
477 				break;
478 			}
479 
480 
481 			if ((ARGV - arglist) == MAXARGS) {
482 				break;
483 			}
484 		}
485 
486 		*ARGV = NULL;
487 		if (N_args == 0) {
488 			/* Reached the end with no more work. */
489 			break;
490 		}
491 
492 		/* insert arg if requested */
493 
494 		if (!ERR && INSERT) {
495 
496 			p_ibuf = ins_buf;
497 			ARGV--;
498 			j = ibufsize = 0;
499 			for (psave = saveargv; ++j <= n_inserts; ++psave) {
500 				addibuf(psave);
501 				if (ERR)
502 					break;
503 			}
504 		}
505 		*ARGV = NULL;
506 
507 		if (n_inserts > 0) {
508 			/*
509 			 * if we've done any insertions, re-calculate the
510 			 * linesize. bomb out if we've exceeded our length.
511 			 */
512 			linesize = 0;
513 			for (ARGV = arglist; *ARGV != NULL; ARGV++) {
514 				linesize += strlen(*ARGV) + 1;
515 			}
516 			if (linesize >= BUFLIM) {
517 				EMSG(LIST2LONG);
518 				procs_wait(B_TRUE);
519 				exit(2);
520 				/* NOTREACHED */
521 			}
522 		}
523 
524 		/* exec command */
525 
526 		if (!ERR) {
527 			if (!MORE &&
528 			    (PER_LINE && N_lines == 0 || N_ARGS && N_args == 0))
529 				exit(exitstat);
530 			OK = TRUE;
531 			j = TRACE ? echoargs() : TRUE;
532 			if (j) {
533 				/*
534 				 * for xcu4, all invocations of cmdname must
535 				 * return 0, in order for us to return 0.
536 				 * so if we have a non-zero status here,
537 				 * quit immediately.
538 				 */
539 				(void) lcall(cmdname, arglist);
540 			}
541 		}
542 	}
543 
544 	procs_wait(B_TRUE);
545 
546 	if (OK)
547 		return (exitstat);
548 
549 	/*
550 	 * if exitstat was set, to match XCU4 complience,
551 	 * return that value, otherwise, return 1.
552 	 */
553 	return (exitstat ? exitstat : 1);
554 }
555 
556 static char *
557 addarg(char *arg)
558 {
559 	linesize += (strlen(arg) + 1);
560 	return (arg);
561 }
562 
563 
564 static void
565 store_str(char **buffer, char *str, size_t len)
566 {
567 	(void) memcpy(*buffer, str, len);
568 	(*buffer)[len] = '\0';
569 	*buffer += len;
570 }
571 
572 
573 static char *
574 getarg(char *arg)
575 {
576 	char	*xarg = arg;
577 	wchar_t	c = 0;
578 	char	mbc[MB_LEN_MAX];
579 	size_t	len;
580 	int	escape = 0;
581 	int	inquote = 0;
582 	int	last = 0;
583 
584 	arg[0] = '\0';
585 
586 	while (MORE) {
587 
588 		len = 0;
589 		last = c;
590 		c = getwchr(mbc, &len);
591 
592 		if (((arg - xarg) + len) > BUFLIM) {
593 			EMSG2(ARG2LONG, BUFLIM);
594 			exit(2);
595 			ERR = TRUE;
596 			return (NULL);
597 		}
598 
599 		switch (c) {
600 		case '\n':
601 			if (ZERO) {
602 				store_str(&arg, mbc, len);
603 				continue;
604 			}
605 			/*
606 			 * NB: Some other versions rip off all of the trailing
607 			 * blanks.  The spec only claims that this should
608 			 * be done for a single blank.  We follow the spec.
609 			 */
610 			if (LINE_CONT && iswctype(last, blank)) {
611 				len = 0;
612 				*arg = 0;
613 				continue;
614 			}
615 			/* FALLTHRU */
616 
617 		case '\0':
618 		case WEOF:	/* Note WEOF == EOF */
619 
620 			if (escape) {
621 				EMSG(BADESCAPE);
622 				ERR = TRUE;
623 				return (NULL);
624 			}
625 			if (inquote) {
626 				EMSG(MISSQUOTE);
627 				ERR = TRUE;
628 				return (NULL);
629 			}
630 
631 			N_lines++;
632 			break;
633 
634 		case '"':
635 			if (ZERO || escape || (inquote == 1)) {
636 				/* treat it literally */
637 				escape = 0;
638 				store_str(&arg, mbc, len);
639 
640 			} else if (inquote == 2) {
641 				/* terminating double quote */
642 				inquote = 0;
643 
644 			} else {
645 				/* starting quoted string */
646 				inquote = 2;
647 			}
648 			continue;
649 
650 		case '\'':
651 			if (ZERO || escape || (inquote == 2)) {
652 				/* treat it literally */
653 				escape = 0;
654 				store_str(&arg, mbc, len);
655 
656 			} else if (inquote == 1) {
657 				/* terminating single quote */
658 				inquote = 0;
659 
660 			} else {
661 				/* starting quoted string */
662 				inquote = 1;
663 			}
664 			continue;
665 
666 		case '\\':
667 			/*
668 			 * Any unquoted character can be escaped by
669 			 * preceding it with a backslash.
670 			 */
671 			if (ZERO || inquote || escape) {
672 				escape = 0;
673 				store_str(&arg, mbc, len);
674 			} else {
675 				escape = 1;
676 			}
677 			continue;
678 
679 		default:
680 			/* most times we will just want to store it */
681 			if (inquote || escape || ZERO || !iswctype(c, blank)) {
682 				escape = 0;
683 				store_str(&arg, mbc, len);
684 				continue;
685 			}
686 			if (EAT_LEAD && last == 0) {
687 				c = 0;		/* Roll it back */
688 				continue;
689 			}
690 			if (PER_LINE) {
691 				store_str(&arg, mbc, len);
692 				continue;
693 			}
694 
695 			/* unquoted blank without special handling */
696 			break;
697 		}
698 
699 		/*
700 		 * At this point we are processing a complete argument.
701 		 */
702 		if (strcmp(xarg, LEOF) == 0 && *LEOF != '\0') {
703 			MORE = FALSE;
704 			return (NULL);
705 		}
706 		if (c == WEOF) {
707 			MORE = FALSE;
708 		}
709 		if (xarg[0] == '\0')
710 			continue;
711 		break;
712 	}
713 
714 	return (xarg[0] == '\0' ? NULL : xarg);
715 }
716 
717 /*
718  * ermsg():	print out an error message, and indicate failure globally.
719  *
720  *	Assumes that message has already been gettext()'d. It would be
721  *	nice if we could just do the gettext() here, but we can't, since
722  *	since xgettext(1) wouldn't be able to pick up our error message.
723  */
724 /* PRINTFLIKE1 */
725 static void
726 ermsg(char *messages, ...)
727 {
728 	va_list	ap;
729 
730 	va_start(ap, messages);
731 
732 	(void) fprintf(stderr, "xargs: ");
733 	(void) vfprintf(stderr, messages, ap);
734 
735 	va_end(ap);
736 	OK = FALSE;
737 }
738 
739 static int
740 echoargs(void)
741 {
742 	char	**anarg;
743 	char	**tanarg;	/* tmp ptr			*/
744 	int	i;
745 	char	reply[LINE_MAX];
746 
747 	tanarg = anarg = arglist-1;
748 
749 	/*
750 	 * write out each argument, separated by a space. the tanarg
751 	 * nonsense is for xcu4 testsuite compliance - so that an
752 	 * extra space isn't echoed after the last argument.
753 	 */
754 	while (*++anarg) {		/* while there's an argument	*/
755 		++tanarg;		/* follow anarg			*/
756 		(void) write(2, *anarg, strlen(*anarg));
757 
758 		if (*++tanarg) {	/* if there's another argument:	*/
759 			(void) write(2, " ", 1); /* add a space		*/
760 			--tanarg;	/* reset back to anarg		*/
761 		}
762 	}
763 	if (PROMPT == -1) {
764 		(void) write(2, "\n", 1);
765 		return (TRUE);
766 	}
767 
768 	(void) write(2, "?...", 4);	/* ask the user for input	*/
769 
770 	for (i = 0; i < LINE_MAX && read(PROMPT, &reply[i], 1) > 0; i++) {
771 		if (reply[i] == '\n') {
772 			if (i == 0)
773 				return (FALSE);
774 			break;
775 		}
776 	}
777 	if (i < LINE_MAX)
778 		reply[i] = '\0';
779 	else
780 		reply[LINE_MAX - 1] = '\0';
781 
782 	/* flush remainder of line if necessary */
783 	if (i == LINE_MAX) {
784 		char	bitbucket;
785 
786 		while ((read(PROMPT, &bitbucket, 1) > 0) && (bitbucket != '\n'))
787 			;
788 	}
789 
790 	return (yes_check(reply));
791 }
792 
793 
794 static char *
795 insert(char *pattern, char *subst)
796 {
797 	static char	buffer[MAXSBUF+1];
798 	int		len, ipatlen;
799 	char	*pat;
800 	char	*bufend;
801 	char	*pbuf;
802 
803 	len = strlen(subst);
804 	ipatlen = strlen(INSPAT) - 1;
805 	pat = pattern - 1;
806 	pbuf = buffer;
807 	bufend = &buffer[MAXSBUF];
808 
809 	while (*++pat) {
810 		if (strncmp(pat, INSPAT, ipatlen + 1) == 0) {
811 			if (pbuf + len >= bufend) {
812 				break;
813 			} else {
814 				(void) strcpy(pbuf, subst);
815 				pat += ipatlen;
816 				pbuf += len;
817 			}
818 		} else {
819 			*pbuf++ = *pat;
820 			if (pbuf >= bufend)
821 				break;
822 		}
823 	}
824 
825 	if (!*pat) {
826 		*pbuf = '\0';
827 		return (buffer);
828 	} else {
829 		ermsg(gettext("Maximum argument size with insertion via %s's "
830 		    "exceeded\n"), INSPAT);
831 		ERR = TRUE;
832 		return (NULL);
833 	}
834 }
835 
836 
837 static void
838 addibuf(struct inserts	*p)
839 {
840 	char	*newarg, *skel, *sub;
841 	int		l;
842 
843 	skel = p->p_skel;
844 	sub = *ARGV;
845 	newarg = insert(skel, sub);
846 	if (ERR)
847 		return;
848 
849 	l = strlen(newarg) + 1;
850 	if ((ibufsize += l) > MAXIBUF) {
851 		EMSG(IBUFOVERFLOW);
852 		ERR = TRUE;
853 	}
854 	(void) strcpy(p_ibuf, newarg);
855 	*(p->p_ARGV) = p_ibuf;
856 	p_ibuf += l;
857 }
858 
859 
860 /*
861  * getwchr():	get the next wide character.
862  * description:
863  *	we get the next character from stdin.  This returns WEOF if no
864  *	character is present.  If ZERO is set, it gets a single byte instead
865  *	a wide character.
866  */
867 static wint_t
868 getwchr(char *mbc, size_t *sz)
869 {
870 	size_t		i;
871 	int		c;
872 	wchar_t		wch;
873 
874 	i = 0;
875 	while (i < MB_CUR_MAX) {
876 
877 		if ((c = fgetc(stdin)) == EOF) {
878 
879 			if (i == 0) {
880 				/* TRUE EOF has been reached */
881 				return (WEOF);
882 			}
883 
884 			/*
885 			 * We have some characters in our buffer still so it
886 			 * must be an invalid character right before EOF.
887 			 */
888 			break;
889 		}
890 		mbc[i++] = (char)c;
891 
892 		/* If this succeeds then we are done */
893 		if (ZERO) {
894 			*sz = i;
895 			return ((char)c);
896 		}
897 		if (mbtowc(&wch, mbc, i) != -1) {
898 			*sz = i;
899 			return ((wint_t)wch);
900 		}
901 	}
902 
903 	/*
904 	 * We have now encountered an illegal character sequence.
905 	 * There is nothing much we can do at this point but
906 	 * return an error.  If we attempt to recover we may in fact
907 	 * return garbage as arguments, from the customer's point
908 	 * of view.  After all what if they are feeding us a file
909 	 * generated in another locale?
910 	 */
911 	errno = EILSEQ;
912 	PERR(CORRUPTFILE);
913 	exit(1);
914 	/* NOTREACHED */
915 }
916 
917 
918 static void
919 lcall(char *sub, char **subargs)
920 {
921 	int	err;
922 	pid_t	child;
923 	posix_spawnattr_t attr;
924 
925 	if ((err = posix_spawnattr_init(&attr)) == 0)
926 		err = posix_spawnattr_setflags(&attr, POSIX_SPAWN_NOSIGCHLD_NP);
927 	if (err != 0) {
928 		errno = err;
929 		PERR(FORKFAIL);
930 		exit(123);
931 	}
932 
933 	for (;;) {
934 		err = posix_spawnp(&child, sub, NULL, &attr, subargs, environ);
935 		if (err == 0) {
936 			(void) posix_spawnattr_destroy(&attr);
937 			procs_store(child);
938 			/*
939 			 * Note, if we have used up all of our slots, then this
940 			 * call may end up blocking.
941 			 */
942 			procs_wait(B_FALSE);
943 			return;
944 		}
945 		if (err == EAGAIN) {
946 			/* A temporary resource shortage. Wait, then retry. */
947 			(void) sleep(1);
948 			continue;
949 		}
950 		errno = err;
951 		if (err == EPERM) {
952 			/* Creation of the new process was denied. */
953 			PERR(FORKFAIL);
954 			exit(123);
955 		}
956 		/*
957 		 * The utility could not be invoked. POSIX requires that a
958 		 * diagnostic is written and that we exit without processing
959 		 * any remaining input.
960 		 */
961 		PERR(EXECFAIL);
962 		procs_wait(B_TRUE);
963 		exit((err == EACCES) ? 126 : 127);
964 	}
965 }
966 
967 /*
968  * Return the index of child in the procs array.
969  */
970 static int
971 procs_find(pid_t child)
972 {
973 	int	i;
974 
975 	for (i = 0; i < MAXPROCS; i++) {
976 		if (procs[i] == child) {
977 			return (i);
978 		}
979 	}
980 
981 	return (-1);
982 }
983 
984 static void
985 procs_store(pid_t child)
986 {
987 	int	i;
988 
989 	i = procs_find(0);
990 	if (i < 0) {
991 		EMSG(NOCHILDSLOT);
992 		exit(1);
993 	}
994 	procs[i] = child;
995 	n_procs++;
996 }
997 
998 static boolean_t
999 procs_delete(pid_t child)
1000 {
1001 	int	i;
1002 
1003 	i = procs_find(child);
1004 	if (i < 0) {
1005 		return (B_FALSE);
1006 	}
1007 
1008 	procs[i] = (pid_t)0;
1009 	n_procs--;
1010 
1011 	return (B_TRUE);
1012 }
1013 
1014 static pid_t
1015 procs_waitpid(boolean_t blocking, int *stat_loc)
1016 {
1017 	pid_t	child;
1018 	int	options;
1019 
1020 	if (n_procs == 0) {
1021 		errno = ECHILD;
1022 		return (-1);
1023 	}
1024 
1025 	options = 0;
1026 	if (!blocking) {
1027 		options |= WNOHANG;
1028 	}
1029 
1030 	while ((child = waitpid((pid_t)-1, stat_loc, options)) > 0) {
1031 		if (procs_delete(child)) {
1032 			break;
1033 		}
1034 	}
1035 
1036 	return (child);
1037 }
1038 
1039 static void
1040 procs_wait(boolean_t blocking)
1041 {
1042 	pid_t	child;
1043 	int	stat_loc;
1044 
1045 	/*
1046 	 * If we currently have filled all of our slots, then we need to block
1047 	 * further execution.
1048 	 */
1049 	if (n_procs >= MAXPROCS)
1050 		blocking = B_TRUE;
1051 	while ((child = procs_waitpid(blocking, &stat_loc)) > 0) {
1052 		if (WIFSIGNALED(stat_loc)) {
1053 			EMSG2(CHILDSIG, WTERMSIG(stat_loc));
1054 			exit(125);
1055 			/* NOTREACHED */
1056 		} else if ((WEXITSTATUS(stat_loc) & 0377) == 0377) {
1057 			EMSG(CHILDFAIL);
1058 			exit(124);
1059 			/* NOTREACHED */
1060 		} else {
1061 			exitstat |= WEXITSTATUS(stat_loc);
1062 		}
1063 	}
1064 
1065 	if (child == (pid_t)(-1) && errno != ECHILD) {
1066 		EMSG(WAITFAIL);
1067 		exit(122);
1068 		/* NOTREACHED */
1069 	}
1070 }
1071 
1072 static void
1073 usage()
1074 {
1075 	ermsg(_(USAGEMSG));
1076 	OK = FALSE;
1077 }
1078 
1079 
1080 
1081 /*
1082  * parseargs():		modify the args
1083  *	since the -e, -i and -l flags all take optional subarguments,
1084  *	and getopt(3C) is clueless about this nonsense, we change
1085  *	our local argument count and strings to separate this out,
1086  *	and make it easier to handle via getopt(3C).
1087  *
1088  *	-e	-> "-e ""
1089  *	-e3	-> "-e "3"
1090  *	-Estr	-> "-E "str"
1091  *	-i	-> "-i "{}"
1092  *	-irep	-> "-i "rep"
1093  *	-l	-> "-l "1"
1094  *	-l10	-> "-l "10"
1095  *
1096  *	since the -e, -i and -l flags all take optional subarguments,
1097  */
1098 static void
1099 parseargs(int ac, char **av)
1100 {
1101 	int i;			/* current argument			*/
1102 	int cflag;		/* 0 = not processing cmd arg		*/
1103 
1104 	if ((mav = malloc((ac * 2 + 1) * sizeof (char *))) == NULL) {
1105 		PERR(MALLOCFAIL);
1106 		exit(1);
1107 	}
1108 
1109 	/* for each argument, see if we need to change things:		*/
1110 	for (i = mac = cflag = 0; (av[i] != NULL) && i < ac; i++, mac++) {
1111 		if ((mav[mac] = strdup(av[i])) == NULL) {
1112 			PERR(MALLOCFAIL);
1113 			exit(1);
1114 		}
1115 
1116 		/* -- has been found or argument list is fully processes */
1117 		if (cflag)
1118 			continue;
1119 
1120 		/*
1121 		 * if we're doing special processing, and we've got a flag
1122 		 */
1123 		else if ((av[i][0] == '-') && (av[i][1] != '\0')) {
1124 			char	*def;
1125 
1126 			switch (av[i][1]) {
1127 			case	'e':
1128 				def = ""; /* -e with no arg turns off eof */
1129 				goto process_special;
1130 			case	'i':
1131 				def = INSPAT_STR;
1132 				goto process_special;
1133 			case	'l':
1134 				def = "1";
1135 process_special:
1136 				/*
1137 				 * if there's no sub-option, we *must* add
1138 				 * a default one. this is because xargs must
1139 				 * be able to distinguish between a valid
1140 				 * suboption, and a command name.
1141 				 */
1142 				if (av[i][2] == '\0') {
1143 					mav[++mac] = strdup(def);
1144 				} else {
1145 					/* clear out our version: */
1146 					mav[mac][2] = '\0';
1147 					mav[++mac] = strdup(&av[i][2]);
1148 				}
1149 				if (mav[mac] == NULL) {
1150 					PERR(MALLOCFAIL);
1151 					exit(1);
1152 				}
1153 				break;
1154 
1155 			/* flags with required subarguments:		*/
1156 
1157 			/*
1158 			 * there are two separate cases here. either the
1159 			 * flag can have the normal XCU4 handling
1160 			 * (of the form: -X subargument); or it can have
1161 			 * the old solaris 2.[0-4] handling (of the
1162 			 * form: -Xsubargument). in order to maintain
1163 			 * backwards compatibility, we must support the
1164 			 * latter case. we handle the latter possibility
1165 			 * first so both the old solaris way of handling
1166 			 * and the new XCU4 way of handling things are allowed.
1167 			 */
1168 			case	'n':	/* FALLTHROUGH			*/
1169 			case	'P':	/* FALLTHROUGH			*/
1170 			case	's':	/* FALLTHROUGH			*/
1171 			case	'E':	/* FALLTHROUGH			*/
1172 			case	'I':	/* FALLTHROUGH			*/
1173 			case	'L':
1174 				/*
1175 				 * if the second character isn't null, then
1176 				 * the user has specified the old syntax.
1177 				 * we move the subargument into our
1178 				 * mod'd argument list.
1179 				 */
1180 				if (av[i][2] != '\0') {
1181 					/* first clean things up:	*/
1182 					mav[mac][2] = '\0';
1183 
1184 					/* now add the separation:	*/
1185 					++mac;	/* inc to next mod'd arg */
1186 					if ((mav[mac] = strdup(&av[i][2])) ==
1187 					    NULL) {
1188 						PERR(MALLOCFAIL);
1189 						exit(1);
1190 					}
1191 					break;
1192 				}
1193 				i++;
1194 				mac++;
1195 
1196 				if (av[i] == NULL) {
1197 					mav[mac] = NULL;
1198 					return;
1199 				}
1200 				if ((mav[mac] = strdup(av[i])) == NULL) {
1201 					PERR(MALLOCFAIL);
1202 					exit(1);
1203 				}
1204 				break;
1205 
1206 			/* flags */
1207 			case 'p' :
1208 			case 't' :
1209 			case 'x' :
1210 			case '0' :
1211 				break;
1212 
1213 			case '-' :
1214 			default:
1215 				/*
1216 				 * here we've hit the cmd argument. so
1217 				 * we'll stop special processing, as the
1218 				 * cmd may have a "-i" etc., argument,
1219 				 * and we don't want to add a "" to it.
1220 				 */
1221 				cflag = 1;
1222 				break;
1223 			}
1224 		} else if (i > 0) {	/* if we're not the 1st arg	*/
1225 			/*
1226 			 * if it's not a flag, then it *must* be the cmd.
1227 			 * set cflag, so we don't mishandle the -[eil] flags.
1228 			 */
1229 			cflag = 1;
1230 		}
1231 	}
1232 
1233 	mav[mac] = NULL;
1234 }
1235