xref: /netbsd-src/sbin/dump/tape.c (revision e2325e1f91ff78a0a3b9bef376747a1a1ae02cc6)
1 /*	$NetBSD: tape.c,v 1.57 2021/06/19 13:56:34 christos Exp $	*/
2 
3 /*-
4  * Copyright (c) 1980, 1991, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the University nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 #include <sys/cdefs.h>
33 #ifndef lint
34 #if 0
35 static char sccsid[] = "@(#)tape.c	8.4 (Berkeley) 5/1/95";
36 #else
37 __RCSID("$NetBSD: tape.c,v 1.57 2021/06/19 13:56:34 christos Exp $");
38 #endif
39 #endif /* not lint */
40 
41 #include <sys/param.h>
42 #include <sys/socket.h>
43 #include <sys/time.h>
44 #include <sys/wait.h>
45 #include <sys/ioctl.h>
46 #include <sys/mtio.h>
47 
48 #include <errno.h>
49 #include <fcntl.h>
50 #include <signal.h>
51 #include <stdio.h>
52 #include <stdlib.h>
53 #include <string.h>
54 #include <time.h>
55 #include <unistd.h>
56 
57 #include "dump.h"
58 #include "pathnames.h"
59 
60 int	writesize;		/* size of malloc()ed buffer for tape */
61 int64_t	lastspclrec = -1;	/* tape block number of last written header */
62 int	trecno = 0;		/* next record to write in current block */
63 extern	long blocksperfile;	/* number of blocks per output file */
64 long	blocksthisvol;		/* number of blocks on current output file */
65 extern	int ntrec;		/* blocking factor on tape */
66 extern	int cartridge;
67 extern	const char *host;
68 char	*nexttape;
69 
70 static	ssize_t atomic_read(int, void *, int);
71 static	ssize_t atomic_write(int, const void *, int);
72 static	void doworker(int, int);
73 static	void create_workers(void);
74 static	void flushtape(void);
75 static	void killall(void);
76 static	void proceed(int);
77 static	void rollforward(void);
78 static	void sigpipe(int);
79 static	void tperror(int);
80 
81 /*
82  * Concurrent dump mods (Caltech) - disk block reading and tape writing
83  * are exported to several worker processes.  While one worker writes the
84  * tape, the others read disk blocks; they pass control of the tape in
85  * a ring via signals. The parent process traverses the file system and
86  * sends writeheader()'s and lists of daddr's to the workers via pipes.
87  * The following structure defines the instruction packets sent to workers.
88  */
89 struct req {
90 	daddr_t dblk;
91 	int count;
92 };
93 int reqsiz;
94 
95 #define WORKERS 3		/* 1 worker writing, 1 reading, 1 for slack */
96 struct worker {
97 	int64_t tapea;		/* header number at start of this chunk */
98 	int64_t firstrec;	/* record number of this block */
99 	int count;		/* count to next header (used for TS_TAPE */
100 				/* after EOT) */
101 	int inode;		/* inode that we are currently dealing with */
102 	int fd;			/* FD for this worker */
103 	int pid;		/* PID for this worker */
104 	int sent;		/* 1 == we've sent this worker requests */
105 	char (*tblock)[TP_BSIZE]; /* buffer for data blocks */
106 	struct req *req;	/* buffer for requests */
107 } workers[WORKERS+1];
108 struct worker *wp;
109 
110 char	(*nextblock)[TP_BSIZE];
111 
112 static int64_t tapea_volume;	/* value of spcl.c_tapea at volume start */
113 
114 int master;		/* pid of master, for sending error signals */
115 int tenths;		/* length of tape used per block written */
116 static volatile sig_atomic_t caught;	/* have we caught the signal to proceed? */
117 
118 int
alloctape(void)119 alloctape(void)
120 {
121 	int pgoff = getpagesize() - 1;
122 	char *buf;
123 	int i;
124 
125 	writesize = ntrec * TP_BSIZE;
126 	reqsiz = (ntrec + 1) * sizeof(struct req);
127 	/*
128 	 * CDC 92181's and 92185's make 0.8" gaps in 1600-bpi start/stop mode
129 	 * (see DEC TU80 User's Guide).  The shorter gaps of 6250-bpi require
130 	 * repositioning after stopping, i.e, streaming mode, where the gap is
131 	 * variable, 0.30" to 0.45".  The gap is maximal when the tape stops.
132 	 */
133 	if (blocksperfile == 0 && !unlimited)
134 		tenths = writesize / density +
135 		    (cartridge ? 16 : density == 625 ? 5 : 8);
136 	/*
137 	 * Allocate tape buffer contiguous with the array of instruction
138 	 * packets, so flushtape() can write them together with one write().
139 	 * Align tape buffer on page boundary to speed up tape write().
140 	 */
141 	for (i = 0; i <= WORKERS; i++) {
142 		buf = (char *)
143 		    xmalloc((unsigned)(reqsiz + writesize + pgoff + TP_BSIZE));
144 		workers[i].tblock = (char (*)[TP_BSIZE])
145 		    (((long)&buf[ntrec + 1] + pgoff) &~ pgoff);
146 		workers[i].req = (struct req *)workers[i].tblock - ntrec - 1;
147 	}
148 	wp = &workers[0];
149 	wp->count = 1;
150 	wp->tapea = 0;
151 	wp->firstrec = 0;
152 	nextblock = wp->tblock;
153 	return(1);
154 }
155 
156 void
writerec(const char * dp,int isspcl)157 writerec(const char *dp, int isspcl)
158 {
159 
160 	wp->req[trecno].dblk = (daddr_t)0;
161 	wp->req[trecno].count = 1;
162 	*(union u_spcl *)(*(nextblock)++) = *(const union u_spcl *)dp;
163 	if (isspcl)
164 		lastspclrec = iswap64(spcl.c_tapea);
165 	trecno++;
166 	spcl.c_tapea = iswap64(iswap64(spcl.c_tapea) +1);
167 	if (trecno >= ntrec)
168 		flushtape();
169 }
170 
171 void
dumpblock(daddr_t blkno,int size)172 dumpblock(daddr_t blkno, int size)
173 {
174 	int avail, tpblks;
175 	daddr_t dblkno;
176 
177 	dblkno = fsatoda(ufsib, blkno);
178 	tpblks = size >> tp_bshift;
179 	while ((avail = MIN(tpblks, ntrec - trecno)) > 0) {
180 		wp->req[trecno].dblk = dblkno;
181 		wp->req[trecno].count = avail;
182 		trecno += avail;
183 		spcl.c_tapea = iswap64(iswap64(spcl.c_tapea) + avail);
184 		if (trecno >= ntrec)
185 			flushtape();
186 		dblkno += avail << (tp_bshift - dev_bshift);
187 		tpblks -= avail;
188 	}
189 }
190 
191 int	nogripe = 0;
192 
193 static void
tperror(int signo __unused)194 tperror(int signo __unused)
195 {
196 
197 	if (pipeout) {
198 		msg("write error on %s\n", tape);
199 		quit("Cannot recover");
200 		/* NOTREACHED */
201 	}
202 	msg("write error %ld blocks into volume %d\n", blocksthisvol, tapeno);
203 	broadcast("DUMP WRITE ERROR!\n");
204 	if (!query("Do you want to restart?"))
205 		dumpabort(0);
206 	msg("Closing this volume.  Prepare to restart with new media;\n");
207 	msg("this dump volume will be rewritten.\n");
208 	killall();
209 	nogripe = 1;
210 	close_rewind();
211 	Exit(X_REWRITE);
212 }
213 
214 static void
sigpipe(int signo __unused)215 sigpipe(int signo __unused)
216 {
217 
218 	quit("Broken pipe");
219 }
220 
221 /*
222  * do_stats --
223  *	Update xferrate stats
224  */
225 time_t
do_stats(void)226 do_stats(void)
227 {
228 	time_t tnow, ttaken;
229 	int64_t blocks;
230 
231 	(void)time(&tnow);
232 	ttaken = tnow - tstart_volume;
233 	blocks = iswap64(spcl.c_tapea) - tapea_volume;
234 	msg("Volume %d completed at: %s", tapeno, ctime(&tnow));
235 	if (ttaken > 0) {
236 		msg("Volume %d took %d:%02d:%02d\n", tapeno,
237 		    (int) (ttaken / 3600), (int) ((ttaken % 3600) / 60),
238 		    (int) (ttaken % 60));
239 		msg("Volume %d transfer rate: %d KB/s\n", tapeno,
240 		    (int) (blocks / ttaken));
241 		xferrate += blocks / ttaken;
242 	}
243 	return(tnow);
244 }
245 
246 /*
247  * statussig --
248  *	information message upon receipt of SIGINFO
249  *	(derived from optr.c::timeest())
250  */
251 void
statussig(int notused __unused)252 statussig(int notused __unused)
253 {
254 	time_t	tnow, deltat;
255 	char	msgbuf[128];
256 	int	errno_save;
257 
258 	if (blockswritten < 500)
259 		return;
260 	errno_save = errno;
261 	(void) time((time_t *) &tnow);
262 	if (tnow <= tstart_volume)
263 		return;
264 	deltat = tstart_writing - tnow +
265 	    (1.0 * (tnow - tstart_writing)) / blockswritten * tapesize;
266 	(void)snprintf(msgbuf, sizeof(msgbuf),
267 	    "%3.2f%% done at %ld KB/s, finished in %d:%02d\n",
268 	    (blockswritten * 100.0) / tapesize,
269 	    (long)((iswap64(spcl.c_tapea) - tapea_volume) /
270 	      (tnow - tstart_volume)),
271 	    (int)(deltat / 3600), (int)((deltat % 3600) / 60));
272 	write(STDERR_FILENO, msgbuf, strlen(msgbuf));
273 	errno = errno_save;
274 }
275 
276 static void
flushtape(void)277 flushtape(void)
278 {
279 	int i, blks, got;
280 	int64_t lastfirstrec;
281 
282 	int siz = (char *)nextblock - (char *)wp->req;
283 
284 	wp->req[trecno].count = 0;			/* Sentinel */
285 
286 	if (atomic_write(wp->fd, wp->req, siz) != siz)
287 		quite(errno, "error writing command pipe");
288 	wp->sent = 1; /* we sent a request, read the response later */
289 
290 	lastfirstrec = wp->firstrec;
291 
292 	if (++wp >= &workers[WORKERS])
293 		wp = &workers[0];
294 
295 	/* Read results back from next worker */
296 	if (wp->sent) {
297 		if (atomic_read(wp->fd, &got, sizeof got)
298 		    != sizeof got) {
299 			perror("  DUMP: error reading command pipe in master");
300 			dumpabort(0);
301 		}
302 		wp->sent = 0;
303 
304 		/* Check for end of tape */
305 		if (got < writesize) {
306 			msg("End of tape detected\n");
307 
308 			/*
309 			 * Drain the results, don't care what the values were.
310 			 * If we read them here then trewind won't...
311 			 */
312 			for (i = 0; i < WORKERS; i++) {
313 				if (workers[i].sent) {
314 					if (atomic_read(workers[i].fd,
315 					    &got, sizeof got)
316 					    != sizeof got) {
317 						perror("  DUMP: error reading command pipe in master");
318 						dumpabort(0);
319 					}
320 					workers[i].sent = 0;
321 				}
322 			}
323 
324 			close_rewind();
325 			rollforward();
326 			return;
327 		}
328 	}
329 
330 	blks = 0;
331 	if (iswap32(spcl.c_type) == TS_INODE ||
332 	    iswap32(spcl.c_type) == TS_ADDR) {
333 		for (i = 0; i < iswap32(spcl.c_count); i++)
334 			if (spcl.c_addr[i] != 0)
335 				blks++;
336 	}
337 	wp->count = lastspclrec + blks + 1 - iswap64(spcl.c_tapea);
338 	wp->tapea = iswap64(spcl.c_tapea);
339 	wp->firstrec = lastfirstrec + ntrec;
340 	wp->inode = curino;
341 	nextblock = wp->tblock;
342 	trecno = 0;
343 	asize += tenths;
344 	blockswritten += ntrec;
345 	blocksthisvol += ntrec;
346 	if (!pipeout && !unlimited && (blocksperfile ?
347 	    (blocksthisvol >= blocksperfile) : (asize > tsize))) {
348 		close_rewind();
349 		startnewtape(0);
350 	}
351 	timeest();
352 }
353 
354 void
trewind(int eject)355 trewind(int eject)
356 {
357 	int f;
358 	int got;
359 
360 	for (f = 0; f < WORKERS; f++) {
361 		/*
362 		 * Drain the results, but unlike EOT we DO (or should) care
363 		 * what the return values were, since if we detect EOT after
364 		 * we think we've written the last blocks to the tape anyway,
365 		 * we have to replay those blocks with rollforward.
366 		 *
367 		 * fixme: punt for now.
368 		 */
369 		if (workers[f].sent) {
370 			if (atomic_read(workers[f].fd, &got, sizeof got)
371 			    != sizeof got) {
372 				perror("  DUMP: error reading command pipe in master");
373 				dumpabort(0);
374 			}
375 			workers[f].sent = 0;
376 			if (got != writesize) {
377 				msg("EOT detected in last 2 tape records!\n");
378 				msg("Use a longer tape, decrease the size estimate\n");
379 				quit("or use no size estimate at all");
380 			}
381 		}
382 		(void) close(workers[f].fd);
383 	}
384 	while (wait(NULL) >= 0)	/* wait for any signals from workers */
385 		/* void */;
386 
387 	if (pipeout)
388 		return;
389 
390 	msg("Closing %s\n", tape);
391 
392 #ifdef RDUMP
393 	if (host) {
394 		rmtclose();
395 		while (rmtopen(tape, 0, 0) < 0)
396 			sleep(10);
397 		if (eflag && eject) {
398 			msg("Ejecting %s\n", tape);
399 			(void) rmtioctl(MTOFFL, 0);
400 		}
401 		rmtclose();
402 		return;
403 	}
404 #endif
405 	(void) close(tapefd);
406 	while ((f = open(tape, 0)) < 0)
407 		sleep (10);
408 	if (eflag && eject) {
409 		struct mtop offl;
410 
411 		msg("Ejecting %s\n", tape);
412 		offl.mt_op = MTOFFL;
413 		offl.mt_count = 0;
414 		(void) ioctl(f, MTIOCTOP, &offl);
415 	}
416 	(void) close(f);
417 }
418 
419 void
close_rewind(void)420 close_rewind(void)
421 {
422 	int i, f;
423 
424 	trewind(1);
425 	(void)do_stats();
426 	if (nexttape)
427 		return;
428 	if (!nogripe) {
429 		msg("Change Volumes: Mount volume #%d\n", tapeno+1);
430 		broadcast("CHANGE DUMP VOLUMES!\a\a\n");
431 	}
432 	if (lflag) {
433 		for (i = 0; i < lflag / 10; i++) { /* wait lflag seconds */
434 			if (host) {
435 				if (rmtopen(tape, 0, 0) >= 0) {
436 					rmtclose();
437 					return;
438 				}
439 			} else {
440 				if ((f = open(tape, 0)) >= 0) {
441 					close(f);
442 					return;
443 				}
444 			}
445 			sleep (10);
446 		}
447 	}
448 
449 	while (!query("Is the new volume mounted and ready to go?"))
450 		if (query("Do you want to abort?")) {
451 			dumpabort(0);
452 			/*NOTREACHED*/
453 		}
454 }
455 
456 void
rollforward(void)457 rollforward(void)
458 {
459 	struct req *p, *q, *prev;
460 	struct worker *twp;
461 	int i, size, got;
462 	int64_t savedtapea;
463 	union u_spcl *ntb, *otb;
464 	twp = &workers[WORKERS];
465 	ntb = (union u_spcl *)twp->tblock[1];
466 
467 	/*
468 	 * Each of the N workers should have requests that need to
469 	 * be replayed on the next tape.  Use the extra worker buffers
470 	 * (workers[WORKERS]) to construct request lists to be sent to
471 	 * each worker in turn.
472 	 */
473 	for (i = 0; i < WORKERS; i++) {
474 		q = &twp->req[1];
475 		otb = (union u_spcl *)wp->tblock;
476 
477 		/*
478 		 * For each request in the current worker, copy it to twp.
479 		 */
480 
481 		prev = NULL;
482 		for (p = wp->req; p->count > 0; p += p->count) {
483 			*q = *p;
484 			if (p->dblk == 0)
485 				*ntb++ = *otb++; /* copy the datablock also */
486 			prev = q;
487 			q += q->count;
488 		}
489 		if (prev == NULL)
490 			quit("rollforward: protocol botch");
491 		if (prev->dblk != 0)
492 			prev->count -= 1;
493 		else
494 			ntb--;
495 		q -= 1;
496 		q->count = 0;
497 		q = &twp->req[0];
498 		if (i == 0) {
499 			q->dblk = 0;
500 			q->count = 1;
501 			trecno = 0;
502 			nextblock = twp->tblock;
503 			savedtapea = iswap64(spcl.c_tapea);
504 			spcl.c_tapea = iswap64(wp->tapea);
505 			startnewtape(0);
506 			spcl.c_tapea = iswap64(savedtapea);
507 			lastspclrec = savedtapea - 1;
508 		}
509 		size = (char *)ntb - (char *)q;
510 		if (atomic_write(wp->fd, q, size) != size) {
511 			perror("  DUMP: error writing command pipe");
512 			dumpabort(0);
513 		}
514 		wp->sent = 1;
515 		if (++wp >= &workers[WORKERS])
516 			wp = &workers[0];
517 
518 		q->count = 1;
519 
520 		if (prev->dblk != 0) {
521 			/*
522 			 * If the last one was a disk block, make the
523 			 * first of this one be the last bit of that disk
524 			 * block...
525 			 */
526 			q->dblk = prev->dblk +
527 				prev->count * (TP_BSIZE / DEV_BSIZE);
528 			ntb = (union u_spcl *)twp->tblock;
529 		} else {
530 			/*
531 			 * It wasn't a disk block.  Copy the data to its
532 			 * new location in the buffer.
533 			 */
534 			q->dblk = 0;
535 			*((union u_spcl *)twp->tblock) = *ntb;
536 			ntb = (union u_spcl *)twp->tblock[1];
537 		}
538 	}
539 	wp->req[0] = *q;
540 	nextblock = wp->tblock;
541 	if (q->dblk == 0)
542 		nextblock++;
543 	trecno = 1;
544 
545 	/*
546 	 * Clear the first workers' response.  One hopes that it
547 	 * worked ok, otherwise the tape is much too short!
548 	 */
549 	if (wp->sent) {
550 		if (atomic_read(wp->fd, &got, sizeof got)
551 		    != sizeof got) {
552 			perror("  DUMP: error reading command pipe in master");
553 			dumpabort(0);
554 		}
555 		wp->sent = 0;
556 
557 		if (got != writesize) {
558 			quit("EOT detected at start of the tape");
559 		}
560 	}
561 }
562 
563 /*
564  * We implement taking and restoring checkpoints on the tape level.
565  * When each tape is opened, a new process is created by forking; this
566  * saves all of the necessary context in the parent.  The child
567  * continues the dump; the parent waits around, saving the context.
568  * If the child returns X_REWRITE, then it had problems writing that tape;
569  * this causes the parent to fork again, duplicating the context, and
570  * everything continues as if nothing had happened.
571  */
572 void
startnewtape(int top)573 startnewtape(int top)
574 {
575 	int	parentpid;
576 	int	childpid;
577 	int	status;
578 	int	waitforpid;
579 	char	*p;
580 	sig_t	interrupt_save;
581 
582 	interrupt_save = signal(SIGINT, SIG_IGN);
583 	parentpid = getpid();
584 	tapea_volume = iswap64(spcl.c_tapea);
585 	(void)time(&tstart_volume);
586 
587 restore_check_point:
588 	(void)signal(SIGINT, interrupt_save);
589 	/*
590 	 *	All signals are inherited...
591 	 */
592 	childpid = fork();
593 	if (childpid < 0) {
594 		msg("Context save fork fails in parent %d\n", parentpid);
595 		Exit(X_ABORT);
596 	}
597 	if (childpid != 0) {
598 		/*
599 		 *	PARENT:
600 		 *	save the context by waiting
601 		 *	until the child doing all of the work returns.
602 		 *	don't catch the interrupt
603 		 */
604 		signal(SIGINT, SIG_IGN);
605 		signal(SIGINFO, SIG_IGN);	/* only want child's stats */
606 #ifdef TDEBUG
607 		msg("Tape: %d; parent process: %d child process %d\n",
608 			tapeno+1, parentpid, childpid);
609 #endif /* TDEBUG */
610 		while ((waitforpid = wait(&status)) != childpid)
611 			msg("Parent %d waiting for child %d has another child %d return\n",
612 				parentpid, childpid, waitforpid);
613 		if (status & 0xFF) {
614 			msg("Child %d returns LOB status %o\n",
615 				childpid, status&0xFF);
616 		}
617 		status = (status >> 8) & 0xFF;
618 #ifdef TDEBUG
619 		switch(status) {
620 			case X_FINOK:
621 				msg("Child %d finishes X_FINOK\n", childpid);
622 				break;
623 			case X_ABORT:
624 				msg("Child %d finishes X_ABORT\n", childpid);
625 				break;
626 			case X_REWRITE:
627 				msg("Child %d finishes X_REWRITE\n", childpid);
628 				break;
629 			default:
630 				msg("Child %d finishes unknown %d\n",
631 					childpid, status);
632 				break;
633 		}
634 #endif /* TDEBUG */
635 		switch(status) {
636 			case X_FINOK:
637 				Exit(X_FINOK);
638 			case X_ABORT:
639 				Exit(X_ABORT);
640 			case X_REWRITE:
641 				goto restore_check_point;
642 			default:
643 				msg("Bad return code from dump: %d\n", status);
644 				Exit(X_ABORT);
645 		}
646 		/*NOTREACHED*/
647 	} else {	/* we are the child; just continue */
648 		signal(SIGINFO, statussig);	/* now want child's stats */
649 #ifdef TDEBUG
650 		sleep(4);	/* allow time for parent's message to get out */
651 		msg("Child on Tape %d has parent %d, my pid = %d\n",
652 			tapeno+1, parentpid, getpid());
653 #endif /* TDEBUG */
654 		/*
655 		 * If we have a name like "/dev/rst0,/dev/rst1",
656 		 * use the name before the comma first, and save
657 		 * the remaining names for subsequent volumes.
658 		 */
659 		tapeno++;			/* current tape sequence */
660 		if (nexttape || strchr(tape, ',')) {
661 			if (nexttape && *nexttape)
662 				tape = nexttape;
663 			if ((p = strchr(tape, ',')) != NULL) {
664 				*p = '\0';
665 				nexttape = p + 1;
666 			} else
667 				nexttape = NULL;
668 			msg("Dumping volume %d on %s\n", tapeno, tape);
669 		}
670 #ifdef RDUMP
671 		while ((tapefd = (host ? rmtopen(tape, 2, 1) :
672 			pipeout ? 1 : open(tape, O_WRONLY|O_CREAT, 0666))) < 0)
673 #else
674 		while ((tapefd = (pipeout ? 1 :
675 				  open(tape, O_WRONLY|O_CREAT, 0666))) < 0)
676 #endif
677 		    {
678 			msg("Cannot open output \"%s\".\n", tape);
679 			if (!query("Do you want to retry the open?"))
680 				dumpabort(0);
681 		}
682 
683 		create_workers();  /* Share open tape file descriptor with workers */
684 
685 		asize = 0;
686 		blocksthisvol = 0;
687 		if (top)
688 			newtape++;		/* new tape signal */
689 		spcl.c_count = iswap32(wp->count);
690 		/*
691 		 * measure firstrec in TP_BSIZE units since restore doesn't
692 		 * know the correct ntrec value...
693 		 */
694 		spcl.c_firstrec = iswap32(wp->firstrec);
695 		spcl.c_volume = iswap32(iswap32(spcl.c_volume) + 1);
696 		spcl.c_type = iswap32(TS_TAPE);
697 		if (!is_ufs2)
698 			spcl.c_flags = iswap32(iswap32(spcl.c_flags)
699 			    | DR_NEWHEADER);
700 		writeheader((ino_t)wp->inode);
701 		if (!is_ufs2)
702 			spcl.c_flags  = iswap32(iswap32(spcl.c_flags) &
703 			    ~ DR_NEWHEADER);
704 		msg("Volume %d started at: %s", tapeno, ctime(&tstart_volume));
705 		if (tapeno > 1)
706 			msg("Volume %d begins with blocks from inode %d\n",
707 				tapeno, wp->inode);
708 	}
709 }
710 
711 void
dumpabort(int signo __unused)712 dumpabort(int signo __unused)
713 {
714 
715 	if (master != 0 && master != getpid())
716 		/* Signals master to call dumpabort */
717 		(void) kill(master, SIGTERM);
718 	else {
719 #ifdef DUMP_LFS
720 		lfs_wrap_go();
721 #endif
722 		killall();
723 		msg("The ENTIRE dump is aborted.\n");
724 	}
725 #ifdef RDUMP
726 	rmtclose();
727 #endif
728 	Exit(X_ABORT);
729 }
730 
731 void
Exit(int status)732 Exit(int status)
733 {
734 
735 #ifdef TDEBUG
736 	msg("pid = %d exits with status %d\n", getpid(), status);
737 #endif /* TDEBUG */
738 	exit(status);
739 }
740 
741 /*
742  * proceed - handler for SIGUSR2, used to synchronize IO between the workers.
743  */
744 static void
proceed(int signo __unused)745 proceed(int signo __unused)
746 {
747 	caught++;
748 }
749 
750 void
create_workers(void)751 create_workers(void)
752 {
753 	int cmd[2];
754 	int i, j;
755 
756 	master = getpid();
757 
758 	signal(SIGTERM, dumpabort);  /* Slave sends SIGTERM on dumpabort() */
759 	signal(SIGPIPE, sigpipe);
760 	signal(SIGUSR1, tperror);    /* Slave sends SIGUSR1 on tape errors */
761 	signal(SIGUSR2, proceed);    /* Slave sends SIGUSR2 to next worker */
762 
763 	for (i = 0; i < WORKERS; i++) {
764 		if (i == wp - &workers[0]) {
765 			caught = 1;
766 		} else {
767 			caught = 0;
768 		}
769 
770 		if (socketpair(AF_LOCAL, SOCK_STREAM, 0, cmd) < 0 ||
771 		    (workers[i].pid = fork()) < 0)
772 			quite(errno, "too many workers, %d (recompile smaller)",
773 			    i);
774 
775 		workers[i].fd = cmd[1];
776 		workers[i].sent = 0;
777 		if (workers[i].pid == 0) { 	    /* Slave starts up here */
778 			for (j = 0; j <= i; j++)
779 				(void) close(workers[j].fd);
780 			signal(SIGINT, SIG_IGN);    /* Master handles this */
781 			signal(SIGINFO, SIG_IGN);
782 			doworker(cmd[0], i);
783 			Exit(X_FINOK);
784 		}
785 	}
786 
787 	for (i = 0; i < WORKERS; i++)
788 		(void) atomic_write(workers[i].fd,
789 				&workers[(i + 1) % WORKERS].pid,
790 				sizeof workers[0].pid);
791 
792 	master = 0;
793 }
794 
795 void
killall(void)796 killall(void)
797 {
798 	int i;
799 
800 	for (i = 0; i < WORKERS; i++)
801 		if (workers[i].pid > 0) {
802 			(void) kill(workers[i].pid, SIGKILL);
803 			workers[i].sent = 0;
804 		}
805 }
806 
807 /*
808  * Synchronization - each process has a lockfile, and shares file
809  * descriptors to the following process's lockfile.  When our write
810  * completes, we release our lock on the following process's lock-
811  * file, allowing the following process to lock it and proceed. We
812  * get the lock back for the next cycle by swapping descriptors.
813  */
814 static void
doworker(int cmd,int worker_number __unused)815 doworker(int cmd, int worker_number __unused)
816 {
817 	int nread, nextworker, size, wrote, eot_count, werror;
818 	sigset_t nsigset, osigset;
819 
820 	wrote = 0;
821 	/*
822 	 * Need our own seek pointer.
823 	 */
824 	(void) close(diskfd);
825 	if ((diskfd = open(disk_dev, O_RDONLY)) < 0)
826 		quite(errno, "worker couldn't reopen disk");
827 
828 	/*
829 	 * Need the pid of the next worker in the loop...
830 	 */
831 	if ((nread = atomic_read(cmd, &nextworker, sizeof nextworker))
832 	    != sizeof nextworker) {
833 		quit("master/worker protocol botched - didn't get pid"
834 		    " of next worker");
835 	}
836 
837 	/*
838 	 * Get list of blocks to dump, read the blocks into tape buffer
839 	 */
840 	while ((nread = atomic_read(cmd, wp->req, reqsiz)) == reqsiz) {
841 		struct req *p = wp->req;
842 
843 		for (trecno = 0; trecno < ntrec;
844 		     trecno += p->count, p += p->count) {
845 			if (p->dblk) {
846 				bread(p->dblk, wp->tblock[trecno],
847 					p->count * TP_BSIZE);
848 			} else {
849 				if (p->count != 1 || atomic_read(cmd,
850 				    wp->tblock[trecno],
851 				    TP_BSIZE) != TP_BSIZE)
852 				       quit("master/worker protocol botched");
853 			}
854 		}
855 
856 		sigemptyset(&nsigset);
857 		sigaddset(&nsigset, SIGUSR2);
858 		sigprocmask(SIG_BLOCK, &nsigset, &osigset);
859 		while (!caught)
860 			sigsuspend(&osigset);
861 		caught = 0;
862 		sigprocmask(SIG_SETMASK, &osigset, NULL);
863 
864 		/* Try to write the data... */
865 		eot_count = 0;
866 		size = 0;
867 		werror = 0;
868 
869 		while (eot_count < 10 && size < writesize) {
870 #ifdef RDUMP
871 			if (host)
872 				wrote = rmtwrite(wp->tblock[0]+size,
873 				    writesize-size);
874 			else
875 #endif
876 				wrote = write(tapefd, wp->tblock[0]+size,
877 				    writesize-size);
878 			werror = errno;
879 #ifdef WRITEDEBUG
880 			fprintf(stderr, "worker %d wrote %d werror %d\n",
881 			    worker_number, wrote, werror);
882 #endif
883 			if (wrote < 0)
884 				break;
885 			if (wrote == 0)
886 				eot_count++;
887 			size += wrote;
888 		}
889 
890 #ifdef WRITEDEBUG
891 		if (size != writesize)
892 			fprintf(stderr,
893 		    "worker %d only wrote %d out of %d bytes and gave up.\n",
894 			    worker_number, size, writesize);
895 #endif
896 
897 		/*
898 		 * Handle ENOSPC as an EOT condition.
899 		 */
900 		if (wrote < 0 && werror == ENOSPC) {
901 			wrote = 0;
902 			eot_count++;
903 		}
904 
905 		if (eot_count > 0)
906 			size = 0;
907 
908 		if (wrote < 0) {
909 			(void) kill(master, SIGUSR1);
910 			sigemptyset(&nsigset);
911 			for (;;)
912 				sigsuspend(&nsigset);
913 		} else {
914 			/*
915 			 * pass size of write back to master
916 			 * (for EOT handling)
917 			 */
918 			(void) atomic_write(cmd, &size, sizeof size);
919 		}
920 
921 		/*
922 		 * If partial write, don't want next worker to go.
923 		 * Also jolts him awake.
924 		 */
925 		(void) kill(nextworker, SIGUSR2);
926 	}
927 	printcachestats();
928 	if (nread != 0)
929 		quite(errno, "error reading command pipe");
930 }
931 
932 /*
933  * Since a read from a pipe may not return all we asked for,
934  * loop until the count is satisfied (or error).
935  */
936 static ssize_t
atomic_read(int fd,void * buf,int count)937 atomic_read(int fd, void *buf, int count)
938 {
939 	ssize_t got, need = count;
940 
941 	while ((got = read(fd, buf, need)) > 0 && (need -= got) > 0)
942 		buf = (char *)buf + got;
943 	return (got < 0 ? got : count - need);
944 }
945 
946 /*
947  * Since a write may not write all we ask if we get a signal,
948  * loop until the count is satisfied (or error).
949  */
950 static ssize_t
atomic_write(int fd,const void * buf,int count)951 atomic_write(int fd, const void *buf, int count)
952 {
953 	ssize_t got, need = count;
954 
955 	while ((got = write(fd, buf, need)) > 0 && (need -= got) > 0)
956 		buf = (const char *)buf + got;
957 	return (got < 0 ? got : count - need);
958 }
959