summaryrefslogtreecommitdiff
path: root/sys/kern/sys_pipe.c
blob: 1823a253d4305fad3bd9201a002cf78b1a12c4ec (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
/*	$OpenBSD: sys_pipe.c,v 1.7 1997/10/06 20:20:02 deraadt Exp $	*/

/*
 * Copyright (c) 1996 John S. Dyson
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice immediately at the beginning of the file, without modification,
 *    this list of conditions, and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Absolutely no warranty of function or purpose is made by the author
 *    John S. Dyson.
 * 4. Modifications may be freely made to this file if the above conditions
 *    are met.
 */

#ifndef OLD_PIPE

/*
 * This file contains a high-performance replacement for the socket-based
 * pipes scheme originally used in FreeBSD/4.4Lite.  It does not support
 * all features of sockets, but does do everything that pipes normally
 * do.
 */

/*
 * This code has two modes of operation, a small write mode and a large
 * write mode.  The small write mode acts like conventional pipes with
 * a kernel buffer.  If the buffer is less than PIPE_MINDIRECT, then the
 * "normal" pipe buffering is done.  If the buffer is between PIPE_MINDIRECT
 * and PIPE_SIZE in size, it is fully mapped and wired into the kernel, and
 * the receiving process can copy it directly from the pages in the sending
 * process.
 *
 * If the sending process receives a signal, it is possible that it will
 * go away, and certainly its address space can change, because control
 * is returned back to the user-mode side.  In that case, the pipe code
 * arranges to copy the buffer supplied by the user process, to a pageable
 * kernel buffer, and the receiving process will grab the data from the
 * pageable kernel buffer.  Since signals don't happen all that often,
 * the copy operation is normally eliminated.
 *
 * The constant PIPE_MINDIRECT is chosen to make sure that buffering will
 * happen for small transfers so that the system will not spend all of
 * its time context switching.  PIPE_SIZE is constrained by the
 * amount of kernel virtual memory.
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/proc.h>
#include <sys/file.h>
#include <sys/protosw.h>
#include <sys/stat.h>
#include <sys/filedesc.h>
#include <sys/malloc.h>
#include <sys/ioctl.h>
#include <sys/stat.h>
#include <sys/select.h>
#include <sys/signalvar.h>
#include <sys/errno.h>
#include <sys/queue.h>
#include <sys/vmmeter.h>
#include <sys/kernel.h>
#if defined(__FreeBSD__)
#include <sys/sysproto.h>
#else /* defined(__NetBSD__) || defined(__OpenBSD__) */
#include <sys/mount.h>
#include <sys/syscallargs.h>
#endif

#include <vm/vm.h>
#include <vm/vm_prot.h>
#include <vm/vm_param.h>
#include <vm/lock.h>
#include <vm/vm_object.h>
#include <vm/vm_kern.h>
#include <vm/vm_extern.h>
#include <vm/pmap.h>
#include <vm/vm_map.h>
#include <vm/vm_page.h>

#include <sys/pipe.h>

/*
 * Use this define if you want to disable *fancy* VM things.  Expect an
 * approx 30% decrease in transfer rate.  This could be useful for
 * NetBSD or OpenBSD.
 */
#if defined(__NetBSD__) || defined(__OpenBSD__)
#define PIPE_NODIRECT
#endif

/*
 * interfaces to the outside world
 */
int	pipe_read __P((struct file *, struct uio *, struct ucred *));
int	pipe_write __P((struct file *, struct uio *, struct ucred *));
int	pipe_close __P((struct file *, struct proc *));
int	pipe_select __P((struct file *, int which, struct proc *));
#if defined(__FreeBSD__)
int	pipe_ioctl __P((struct file *, int, caddr_t, struct proc *));
#else /* defined(__NetBSD__) || defined(__OpenBSD__) */
int	pipe_ioctl __P((struct file *, u_long, caddr_t, struct proc *));
#endif

static struct fileops pipeops =
    { pipe_read, pipe_write, pipe_ioctl, pipe_select, pipe_close };


/*
 * Default pipe buffer size(s), this can be kind-of large now because pipe
 * space is pageable.  The pipe code will try to maintain locality of
 * reference for performance reasons, so small amounts of outstanding I/O
 * will not wipe the cache.
 */
#define MINPIPESIZE (PIPE_SIZE/3)
#define MAXPIPESIZE (2*PIPE_SIZE/3)

/*
 * Maximum amount of kva for pipes -- this is kind-of a soft limit, but
 * is there so that on large systems, we don't exhaust it.
 */
#define MAXPIPEKVA (8*1024*1024)

/*
 * Limit for direct transfers, we cannot, of course limit
 * the amount of kva for pipes in general though.
 */
#define LIMITPIPEKVA (16*1024*1024)

/*
 * Limit the number of "big" pipes
 */
#define LIMITBIGPIPES	32
int nbigpipe;

static int amountpipekva;

void	pipeclose __P((struct pipe *));
void	pipeinit __P((struct pipe *));
int	pipe_stat __P((struct pipe *, struct stat *));
static __inline int pipelock __P((struct pipe *, int));
static __inline void pipeunlock __P((struct pipe *));
static __inline void pipeselwakeup __P((struct pipe *));
#ifndef PIPE_NODIRECT
int	pipe_build_write_buffer __P((struct pipe *, struct uio *));
void	pipe_destroy_write_buffer __P((struct pipe *));
int	pipe_direct_write __P((struct pipe *, struct uio *));
void	pipe_clone_write_buffer __P((struct pipe *));
#endif
void	pipespace __P((struct pipe *));

/*
 * The pipe system call for the DTYPE_PIPE type of pipes
 */

/* ARGSUSED */
int
#if defined(__FreeBSD__)
pipe(p, uap, retval)
#else /* (__NetBSD__) || (__OpenBSD__) */
sys_pipe(p, v, retval)
#endif
	struct proc *p;
	void *v;
#if defined(__FreeBSD__)
	int retval[];
#else /* (__NetBSD__) || (__OpenBSD__) */
	register_t *retval;
#endif
{
	register struct filedesc *fdp = p->p_fd;
	struct file *rf, *wf;
	struct pipe *rpipe, *wpipe;
	int fd, error;

	rpipe = malloc( sizeof (*rpipe), M_TEMP, M_WAITOK);
	pipeinit(rpipe);
	rpipe->pipe_state |= PIPE_DIRECTOK;
	wpipe = malloc( sizeof (*wpipe), M_TEMP, M_WAITOK);
	pipeinit(wpipe);
	wpipe->pipe_state |= PIPE_DIRECTOK;

	error = falloc(p, &rf, &fd);
	if (error)
		goto free2;
	retval[0] = fd;
	rf->f_flag = FREAD | FWRITE;
	rf->f_type = DTYPE_PIPE;
	rf->f_ops = &pipeops;
	rf->f_data = (caddr_t)rpipe;
	error = falloc(p, &wf, &fd);
	if (error)
		goto free3;
	wf->f_flag = FREAD | FWRITE;
	wf->f_type = DTYPE_PIPE;
	wf->f_ops = &pipeops;
	wf->f_data = (caddr_t)wpipe;
	retval[1] = fd;

	rpipe->pipe_peer = wpipe;
	wpipe->pipe_peer = rpipe;

	return (0);
free3:
	ffree(rf);
	fdp->fd_ofiles[retval[0]] = 0;
free2:
	(void)pipeclose(wpipe);
	(void)pipeclose(rpipe);
	return (error);
}

/*
 * Allocate kva for pipe circular buffer, the space is pageable
 */
void
pipespace(cpipe)
	struct pipe *cpipe;
{
	int npages, error;

	npages = round_page(cpipe->pipe_buffer.size)/PAGE_SIZE;
	/*
	 * Create an object, I don't like the idea of paging to/from
	 * kernel_object.
	 * XXX -- minor change needed here for NetBSD/OpenBSD VM systems.
	 */
#if defined(__FreeBSD__)
	cpipe->pipe_buffer.object = vm_object_allocate(OBJT_DEFAULT, npages);
#else /* (__NetBSD__) || (__OpenBSD__) */
	cpipe->pipe_buffer.object = vm_object_allocate(npages);
#endif
	cpipe->pipe_buffer.buffer = (caddr_t) vm_map_min(kernel_map);

	/*
	 * Insert the object into the kernel map, and allocate kva for it.
	 * The map entry is, by default, pageable.
	 * XXX -- minor change needed here for NetBSD/OpenBSD VM systems.
	 */
#if defined(__FreeBSD__)
	error = vm_map_find(kernel_map, cpipe->pipe_buffer.object, 0,
		(vm_offset_t *) &cpipe->pipe_buffer.buffer, 
		cpipe->pipe_buffer.size, 1,
		VM_PROT_ALL, VM_PROT_ALL, 0);
#else /* (__NetBSD__) || (__OpenBSD__) */
	error = vm_map_find(kernel_map, cpipe->pipe_buffer.object, 0,
		(vm_offset_t *) &cpipe->pipe_buffer.buffer,
		cpipe->pipe_buffer.size, 1);
#endif

	if (error != KERN_SUCCESS)
		panic("pipeinit: cannot allocate pipe -- out of kvm -- code = %d", error);
	amountpipekva += cpipe->pipe_buffer.size;
}

/*
 * initialize and allocate VM and memory for pipe
 */
void
pipeinit(cpipe)
	struct pipe *cpipe;
{
	int s;

	cpipe->pipe_buffer.in = 0;
	cpipe->pipe_buffer.out = 0;
	cpipe->pipe_buffer.cnt = 0;
	cpipe->pipe_buffer.size = PIPE_SIZE;

	/* Buffer kva gets dynamically allocated */
	cpipe->pipe_buffer.buffer = NULL;
	/* cpipe->pipe_buffer.object = invalid */

	cpipe->pipe_state = 0;
	cpipe->pipe_peer = NULL;
	cpipe->pipe_busy = 0;
	s = splhigh();
	cpipe->pipe_ctime = time;
	cpipe->pipe_atime = time;
	cpipe->pipe_mtime = time;
	splx(s);
	bzero(&cpipe->pipe_sel, sizeof cpipe->pipe_sel);
	cpipe->pipe_pgid = NO_PID;

#ifndef PIPE_NODIRECT
	/*
	 * pipe data structure initializations to support direct pipe I/O
	 */
	cpipe->pipe_map.cnt = 0;
	cpipe->pipe_map.kva = 0;
	cpipe->pipe_map.pos = 0;
	cpipe->pipe_map.npages = 0;
	/* cpipe->pipe_map.ms[] = invalid */
#endif
}


/*
 * lock a pipe for I/O, blocking other access
 */
static __inline int
pipelock(cpipe, catch)
	struct pipe *cpipe;
	int catch;
{
	int error;
	while (cpipe->pipe_state & PIPE_LOCK) {
		cpipe->pipe_state |= PIPE_LWANT;
		error = tsleep(cpipe, catch ? PRIBIO|PCATCH : PRIBIO,
		    "pipelk", 0);
		if (error)
			return error;
	}
	cpipe->pipe_state |= PIPE_LOCK;
	return 0;
}

/*
 * unlock a pipe I/O lock
 */
static __inline void
pipeunlock(cpipe)
	struct pipe *cpipe;
{
	cpipe->pipe_state &= ~PIPE_LOCK;
	if (cpipe->pipe_state & PIPE_LWANT) {
		cpipe->pipe_state &= ~PIPE_LWANT;
		wakeup(cpipe);
	}
}

static __inline void
pipeselwakeup(cpipe)
	struct pipe *cpipe;
{
	if (cpipe->pipe_state & PIPE_SEL) {
		cpipe->pipe_state &= ~PIPE_SEL;
		selwakeup(&cpipe->pipe_sel);
	}
}

/* ARGSUSED */
int
pipe_read(fp, uio, cred)
	struct file *fp;
	struct uio *uio;
	struct ucred *cred;
{

	struct pipe *rpipe = (struct pipe *) fp->f_data;
	int error = 0;
	int nread = 0;
	int size;

	++rpipe->pipe_busy;
	while (uio->uio_resid) {
		/*
		 * normal pipe buffer receive
		 */
		if (rpipe->pipe_buffer.cnt > 0) {
			size = rpipe->pipe_buffer.size -
			    rpipe->pipe_buffer.out;
			if (size > rpipe->pipe_buffer.cnt)
				size = rpipe->pipe_buffer.cnt;
			if (size > uio->uio_resid)
				size = uio->uio_resid;
			if ((error = pipelock(rpipe,1)) == 0) {
				error = uiomove( &rpipe->pipe_buffer.buffer[rpipe->pipe_buffer.out], 
					size, uio);
				pipeunlock(rpipe);
			}
			if (error) {
				break;
			}
			rpipe->pipe_buffer.out += size;
			if (rpipe->pipe_buffer.out >= rpipe->pipe_buffer.size)
				rpipe->pipe_buffer.out = 0;

			rpipe->pipe_buffer.cnt -= size;
			nread += size;
#ifndef PIPE_NODIRECT
		/*
		 * Direct copy, bypassing a kernel buffer.
		 */
		} else if ((size = rpipe->pipe_map.cnt) &&
			(rpipe->pipe_state & PIPE_DIRECTW)) {
			caddr_t va;
			if (size > uio->uio_resid)
				size = uio->uio_resid;
			if ((error = pipelock(rpipe,1)) == 0) {
				va = (caddr_t) rpipe->pipe_map.kva + rpipe->pipe_map.pos;
				error = uiomove(va, size, uio);
				pipeunlock(rpipe);
			}
			if (error)
				break;
			nread += size;
			rpipe->pipe_map.pos += size;
			rpipe->pipe_map.cnt -= size;
			if (rpipe->pipe_map.cnt == 0) {
				rpipe->pipe_state &= ~PIPE_DIRECTW;
				wakeup(rpipe);
			}
#endif
		} else {
			/*
			 * detect EOF condition
			 */
			if (rpipe->pipe_state & PIPE_EOF) {
				/* XXX error = ? */
				break;
			}
			/*
			 * If the "write-side" has been blocked, wake it up now.
			 */
			if (rpipe->pipe_state & PIPE_WANTW) {
				rpipe->pipe_state &= ~PIPE_WANTW;
				wakeup(rpipe);
			}
			if (nread > 0)
				break;

			if (fp->f_flag & FNONBLOCK) {
				error = EAGAIN;
				break;
			}

			/*
			 * If there is no more to read in the pipe, reset
			 * its pointers to the beginning.  This improves
			 * cache hit stats.
			 */
		
			if ((error = pipelock(rpipe,1)) == 0) {
				if (rpipe->pipe_buffer.cnt == 0) {
					rpipe->pipe_buffer.in = 0;
					rpipe->pipe_buffer.out = 0;
				}
				pipeunlock(rpipe);
			} else {
				break;
			}

			if (rpipe->pipe_state & PIPE_WANTW) {
				rpipe->pipe_state &= ~PIPE_WANTW;
				wakeup(rpipe);
			}

			rpipe->pipe_state |= PIPE_WANTR;
			error = tsleep(rpipe, PRIBIO|PCATCH, "piperd", 0);
			if (error)
				break;
		}
	}

	if (error == 0) {
		int s = splhigh();
		rpipe->pipe_atime = time;
		splx(s);
	}

	--rpipe->pipe_busy;
	if ((rpipe->pipe_busy == 0) && (rpipe->pipe_state & PIPE_WANT)) {
		rpipe->pipe_state &= ~(PIPE_WANT|PIPE_WANTW);
		wakeup(rpipe);
	} else if (rpipe->pipe_buffer.cnt < MINPIPESIZE) {
		/*
		 * If there is no more to read in the pipe, reset
		 * its pointers to the beginning.  This improves
		 * cache hit stats.
		 */
		if (rpipe->pipe_buffer.cnt == 0) {
			if ((error == 0) && (error = pipelock(rpipe,1)) == 0) {
				rpipe->pipe_buffer.in = 0;
				rpipe->pipe_buffer.out = 0;
				pipeunlock(rpipe);
			}
		}

		/*
		 * If the "write-side" has been blocked, wake it up now.
		 */
		if (rpipe->pipe_state & PIPE_WANTW) {
			rpipe->pipe_state &= ~PIPE_WANTW;
			wakeup(rpipe);
		}
	}

	if ((rpipe->pipe_buffer.size - rpipe->pipe_buffer.cnt) >= PIPE_BUF)
		pipeselwakeup(rpipe);

	return error;
}

#ifndef PIPE_NODIRECT
/*
 * Map the sending processes' buffer into kernel space and wire it.
 * This is similar to a physical write operation.
 */
int
pipe_build_write_buffer(wpipe, uio)
	struct pipe *wpipe;
	struct uio *uio;
{
	int size;
	int i;
	vm_offset_t addr, endaddr, paddr;

	size = uio->uio_iov->iov_len;
	if (size > wpipe->pipe_buffer.size)
		size = wpipe->pipe_buffer.size;

	endaddr = round_page(uio->uio_iov->iov_base + size);
	for(i = 0, addr = trunc_page(uio->uio_iov->iov_base);
		addr < endaddr;
		addr += PAGE_SIZE, i+=1) {

		vm_page_t m;

		vm_fault_quick( (caddr_t) addr, VM_PROT_READ);
		paddr = pmap_kextract(addr);
		if (!paddr) {
			int j;
			for(j=0;j<i;j++)
				vm_page_unwire(wpipe->pipe_map.ms[j]);
			return EFAULT;
		}

		m = PHYS_TO_VM_PAGE(paddr);
		vm_page_wire(m);
		wpipe->pipe_map.ms[i] = m;
	}

/*
 * set up the control block
 */
	wpipe->pipe_map.npages = i;
	wpipe->pipe_map.pos = ((vm_offset_t) uio->uio_iov->iov_base) & PAGE_MASK;
	wpipe->pipe_map.cnt = size;

/*
 * and map the buffer
 */
	if (wpipe->pipe_map.kva == 0) {
		/*
		 * We need to allocate space for an extra page because the
		 * address range might (will) span pages at times.
		 */
		wpipe->pipe_map.kva = kmem_alloc_pageable(kernel_map,
			wpipe->pipe_buffer.size + PAGE_SIZE);
		amountpipekva += wpipe->pipe_buffer.size + PAGE_SIZE;
	}
	pmap_qenter(wpipe->pipe_map.kva, wpipe->pipe_map.ms,
		wpipe->pipe_map.npages);

/*
 * and update the uio data
 */

	uio->uio_iov->iov_len -= size;
	uio->uio_iov->iov_base += size;
	if (uio->uio_iov->iov_len == 0)
		uio->uio_iov++;
	uio->uio_resid -= size;
	uio->uio_offset += size;
	return 0;
}

/*
 * unmap and unwire the process buffer
 */
void
pipe_destroy_write_buffer(wpipe)
struct pipe *wpipe;
{
	int i;
	if (wpipe->pipe_map.kva) {
		pmap_qremove(wpipe->pipe_map.kva, wpipe->pipe_map.npages);

		if (amountpipekva > MAXPIPEKVA) {
			vm_offset_t kva = wpipe->pipe_map.kva;
			wpipe->pipe_map.kva = 0;
			kmem_free(kernel_map, kva,
				wpipe->pipe_buffer.size + PAGE_SIZE);
			amountpipekva -= wpipe->pipe_buffer.size + PAGE_SIZE;
		}
	}
	for (i=0;i<wpipe->pipe_map.npages;i++)
		vm_page_unwire(wpipe->pipe_map.ms[i]);
}

/*
 * In the case of a signal, the writing process might go away.  This
 * code copies the data into the circular buffer so that the source
 * pages can be freed without loss of data.
 */
void
pipe_clone_write_buffer(wpipe)
struct pipe *wpipe;
{
	int size;
	int pos;

	size = wpipe->pipe_map.cnt;
	pos = wpipe->pipe_map.pos;
	bcopy((caddr_t) wpipe->pipe_map.kva+pos,
			(caddr_t) wpipe->pipe_buffer.buffer,
			size);

	wpipe->pipe_buffer.in = size;
	wpipe->pipe_buffer.out = 0;
	wpipe->pipe_buffer.cnt = size;
	wpipe->pipe_state &= ~PIPE_DIRECTW;

	pipe_destroy_write_buffer(wpipe);
}

/*
 * This implements the pipe buffer write mechanism.  Note that only
 * a direct write OR a normal pipe write can be pending at any given time.
 * If there are any characters in the pipe buffer, the direct write will
 * be deferred until the receiving process grabs all of the bytes from
 * the pipe buffer.  Then the direct mapping write is set-up.
 */
int
pipe_direct_write(wpipe, uio)
	struct pipe *wpipe;
	struct uio *uio;
{
	int error;
retry:
	while (wpipe->pipe_state & PIPE_DIRECTW) {
		if ( wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
		wpipe->pipe_state |= PIPE_WANTW;
		error = tsleep(wpipe,
				PRIBIO|PCATCH, "pipdww", 0);
		if (error)
			goto error1;
		if (wpipe->pipe_state & PIPE_EOF) {
			error = EPIPE;
			goto error1;
		}
	}
	wpipe->pipe_map.cnt = 0;	/* transfer not ready yet */
	if (wpipe->pipe_buffer.cnt > 0) {
		if ( wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
			
		wpipe->pipe_state |= PIPE_WANTW;
		error = tsleep(wpipe,
				PRIBIO|PCATCH, "pipdwc", 0);
		if (error)
			goto error1;
		if (wpipe->pipe_state & PIPE_EOF) {
			error = EPIPE;
			goto error1;
		}
		goto retry;
	}

	wpipe->pipe_state |= PIPE_DIRECTW;

	error = pipe_build_write_buffer(wpipe, uio);
	if (error) {
		wpipe->pipe_state &= ~PIPE_DIRECTW;
		goto error1;
	}

	error = 0;
	while (!error && (wpipe->pipe_state & PIPE_DIRECTW)) {
		if (wpipe->pipe_state & PIPE_EOF) {
			pipelock(wpipe, 0);
			pipe_destroy_write_buffer(wpipe);
			pipeunlock(wpipe);
			pipeselwakeup(wpipe);
			error = EPIPE;
			goto error1;
		}
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
		pipeselwakeup(wpipe);
		error = tsleep(wpipe, PRIBIO|PCATCH, "pipdwt", 0);
	}

	pipelock(wpipe,0);
	if (wpipe->pipe_state & PIPE_DIRECTW) {
		/*
		 * this bit of trickery substitutes a kernel buffer for
		 * the process that might be going away.
		 */
		pipe_clone_write_buffer(wpipe);
	} else {
		pipe_destroy_write_buffer(wpipe);
	}
	pipeunlock(wpipe);
	return error;

error1:
	wakeup(wpipe);
	return error;
}
#endif
	
int
pipe_write(fp, uio, cred)
	struct file *fp;
	struct uio *uio;
	struct ucred *cred;
{
	int error = 0;
	int orig_resid;

	struct pipe *wpipe, *rpipe;

	rpipe = (struct pipe *) fp->f_data;
	wpipe = rpipe->pipe_peer;

	/*
	 * detect loss of pipe read side, issue SIGPIPE if lost.
	 */
	if ((wpipe == NULL) || (wpipe->pipe_state & PIPE_EOF)) {
		return EPIPE;
	}

	/*
	 * If it is advantageous to resize the pipe buffer, do
	 * so.
	 */
	if ((uio->uio_resid > PIPE_SIZE) &&
		(nbigpipe < LIMITBIGPIPES) &&
		(wpipe->pipe_state & PIPE_DIRECTW) == 0 &&
		(wpipe->pipe_buffer.size <= PIPE_SIZE) &&
		(wpipe->pipe_buffer.cnt == 0)) {

		if (wpipe->pipe_buffer.buffer) {
			amountpipekva -= wpipe->pipe_buffer.size;
			kmem_free(kernel_map,
				(vm_offset_t)wpipe->pipe_buffer.buffer,
				wpipe->pipe_buffer.size);
		}

#ifndef PIPE_NODIRECT
		if (wpipe->pipe_map.kva) {
			amountpipekva -= wpipe->pipe_buffer.size + PAGE_SIZE;
			kmem_free(kernel_map,
				wpipe->pipe_map.kva,
				wpipe->pipe_buffer.size + PAGE_SIZE);
		}
#endif

		wpipe->pipe_buffer.in = 0;
		wpipe->pipe_buffer.out = 0;
		wpipe->pipe_buffer.cnt = 0;
		wpipe->pipe_buffer.size = BIG_PIPE_SIZE;
		wpipe->pipe_buffer.buffer = NULL;
		++nbigpipe;

#ifndef PIPE_NODIRECT
		wpipe->pipe_map.cnt = 0;
		wpipe->pipe_map.kva = 0;
		wpipe->pipe_map.pos = 0;
		wpipe->pipe_map.npages = 0;
#endif

	}
		

	if( wpipe->pipe_buffer.buffer == NULL) {
		if ((error = pipelock(wpipe,1)) == 0) {
			pipespace(wpipe);
			pipeunlock(wpipe);
		} else {
			return error;
		}
	}

	++wpipe->pipe_busy;
	orig_resid = uio->uio_resid;
	while (uio->uio_resid) {
		int space;
#ifndef PIPE_NODIRECT
		/*
		 * If the transfer is large, we can gain performance if
		 * we do process-to-process copies directly.
		 * If the write is non-blocking, we don't use the
		 * direct write mechanism.
		 */
		if ((uio->uio_iov->iov_len >= PIPE_MINDIRECT) &&
		    (fp->f_flag & FNONBLOCK) == 0 &&
			(wpipe->pipe_map.kva || (amountpipekva < LIMITPIPEKVA)) &&
			(uio->uio_iov->iov_len >= PIPE_MINDIRECT)) {
			error = pipe_direct_write( wpipe, uio);
			if (error) {
				break;
			}
			continue;
		}
#endif

		/*
		 * Pipe buffered writes cannot be coincidental with
		 * direct writes.  We wait until the currently executing
		 * direct write is completed before we start filling the
		 * pipe buffer.
		 */
	retrywrite:
		while (wpipe->pipe_state & PIPE_DIRECTW) {
			if (wpipe->pipe_state & PIPE_WANTR) {
				wpipe->pipe_state &= ~PIPE_WANTR;
				wakeup(wpipe);
			}
			error = tsleep(wpipe,
					PRIBIO|PCATCH, "pipbww", 0);
			if (error)
				break;
		}

		space = wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt;

		/* Writes of size <= PIPE_BUF must be atomic. */
		/* XXX perhaps they need to be contiguous to be atomic? */
		if ((space < uio->uio_resid) && (orig_resid <= PIPE_BUF))
			space = 0;

		if (space > 0 && (wpipe->pipe_buffer.cnt < PIPE_SIZE)) {
			/*
			 * This set the maximum transfer as a segment of
			 * the buffer.
			 */
			int size = wpipe->pipe_buffer.size - wpipe->pipe_buffer.in;
			/*
			 * space is the size left in the buffer
			 */
			if (size > space)
				size = space;
			/*
			 * now limit it to the size of the uio transfer
			 */
			if (size > uio->uio_resid)
				size = uio->uio_resid;
			if ((error = pipelock(wpipe,1)) == 0) {
				/*
				 * It is possible for a direct write to
				 * slip in on us... handle it here...
				 */
				if (wpipe->pipe_state & PIPE_DIRECTW) {
					pipeunlock(wpipe);
					goto retrywrite;
				}
				error = uiomove( &wpipe->pipe_buffer.buffer[wpipe->pipe_buffer.in], 
					size, uio);
				pipeunlock(wpipe);
			}
			if (error)
				break;

			wpipe->pipe_buffer.in += size;
			if (wpipe->pipe_buffer.in >= wpipe->pipe_buffer.size)
				wpipe->pipe_buffer.in = 0;

			wpipe->pipe_buffer.cnt += size;
		} else {
			/*
			 * If the "read-side" has been blocked, wake it up now.
			 */
			if (wpipe->pipe_state & PIPE_WANTR) {
				wpipe->pipe_state &= ~PIPE_WANTR;
				wakeup(wpipe);
			}

			/*
			 * don't block on non-blocking I/O
			 */
			if (fp->f_flag & FNONBLOCK) {
				error = EAGAIN;
				break;
			}

			/*
			 * We have no more space and have something to offer,
			 * wake up selects.
			 */
			pipeselwakeup(wpipe);

			wpipe->pipe_state |= PIPE_WANTW;
			error = tsleep(wpipe, (PRIBIO + 1)|PCATCH,
			    "pipewr", 0);
			if (error)
				break;
			/*
			 * If read side wants to go away, we just issue a
			 * signal to ourselves.
			 */
			if (wpipe->pipe_state & PIPE_EOF) {
				error = EPIPE;
				break;
			}	
		}
	}

	--wpipe->pipe_busy;
	if ((wpipe->pipe_busy == 0) &&
		(wpipe->pipe_state & PIPE_WANT)) {
		wpipe->pipe_state &= ~(PIPE_WANT|PIPE_WANTR);
		wakeup(wpipe);
	} else if (wpipe->pipe_buffer.cnt > 0) {
		/*
		 * If we have put any characters in the buffer, we wake up
		 * the reader.
		 */
		if (wpipe->pipe_state & PIPE_WANTR) {
			wpipe->pipe_state &= ~PIPE_WANTR;
			wakeup(wpipe);
		}
	}

	/*
	 * Don't return EPIPE if I/O was successful
	 */
	if ((wpipe->pipe_buffer.cnt == 0) &&
		(uio->uio_resid == 0) &&
		(error == EPIPE))
		error = 0;

	if (error == 0) {
		int s = splhigh();
		wpipe->pipe_mtime = time;
		splx(s);
	}
	/*
	 * We have something to offer,
	 * wake up select.
	 */
	if (wpipe->pipe_buffer.cnt)
		pipeselwakeup(wpipe);

	return error;
}

/*
 * we implement a very minimal set of ioctls for compatibility with sockets.
 */
int
pipe_ioctl(fp, cmd, data, p)
	struct file *fp;
#if defined(__FreeBSD__)
	int cmd;
#else
	u_long cmd;
#endif
	register caddr_t data;
	struct proc *p;
{
	register struct pipe *mpipe = (struct pipe *)fp->f_data;

	switch (cmd) {

	case FIONBIO:
		return (0);

	case FIOASYNC:
		if (*(int *)data) {
			mpipe->pipe_state |= PIPE_ASYNC;
		} else {
			mpipe->pipe_state &= ~PIPE_ASYNC;
		}
		return (0);

	case FIONREAD:
		if (mpipe->pipe_state & PIPE_DIRECTW)
			*(int *)data = mpipe->pipe_map.cnt;
		else
			*(int *)data = mpipe->pipe_buffer.cnt;
		return (0);

	case SIOCSPGRP:
		mpipe->pipe_pgid = *(int *)data;
		return (0);

	case SIOCGPGRP:
		*(int *)data = mpipe->pipe_pgid;
		return (0);

	}
	return (ENOTTY);
}

int
pipe_select(fp, which, p)
	struct file *fp;
	int which;
	struct proc *p;
{
	register struct pipe *rpipe = (struct pipe *)fp->f_data;
	struct pipe *wpipe;

	wpipe = rpipe->pipe_peer;
	switch (which) {

	case FREAD:
		if ( (rpipe->pipe_state & PIPE_DIRECTW) ||
			(rpipe->pipe_buffer.cnt > 0) ||
			(rpipe->pipe_state & PIPE_EOF)) {
			return (1);
		}
		selrecord(p, &rpipe->pipe_sel);
		rpipe->pipe_state |= PIPE_SEL;
		break;

	case FWRITE:
		if ((wpipe == NULL) ||
			(wpipe->pipe_state & PIPE_EOF) ||
			(((wpipe->pipe_state & PIPE_DIRECTW) == 0) &&
			 (wpipe->pipe_buffer.size - wpipe->pipe_buffer.cnt) >= PIPE_BUF)) {
			return (1);
		}
		selrecord(p, &wpipe->pipe_sel);
		wpipe->pipe_state |= PIPE_SEL;
		break;

	case 0:
		if ((rpipe->pipe_state & PIPE_EOF) ||
			(wpipe == NULL) ||
			(wpipe->pipe_state & PIPE_EOF)) {
			return (1);
		}
			
		selrecord(p, &rpipe->pipe_sel);
		rpipe->pipe_state |= PIPE_SEL;
		break;
	}
	return (0);
}

int
pipe_stat(pipe, ub)
	register struct pipe *pipe;
	register struct stat *ub;
{
	bzero((caddr_t)ub, sizeof (*ub));
	ub->st_mode = S_IFIFO;
	ub->st_blksize = pipe->pipe_buffer.size;
	ub->st_size = pipe->pipe_buffer.cnt;
	ub->st_blocks = (ub->st_size + ub->st_blksize - 1) / ub->st_blksize;
	TIMEVAL_TO_TIMESPEC(&pipe->pipe_atime, &ub->st_atimespec);
	TIMEVAL_TO_TIMESPEC(&pipe->pipe_mtime, &ub->st_mtimespec);
	TIMEVAL_TO_TIMESPEC(&pipe->pipe_ctime, &ub->st_ctimespec);
	/*
	 * Left as 0: st_dev, st_ino, st_nlink, st_uid, st_gid, st_rdev,
	 * st_flags, st_gen.
	 * XXX (st_dev, st_ino) should be unique.
	 */
	return 0;
}

/* ARGSUSED */
int
pipe_close(fp, p)
	struct file *fp;
	struct proc *p;
{
	struct pipe *cpipe = (struct pipe *)fp->f_data;

	pipeclose(cpipe);
	fp->f_data = NULL;
	return 0;
}

/*
 * shutdown the pipe
 */
void
pipeclose(cpipe)
	struct pipe *cpipe;
{
	struct pipe *ppipe;
	if (cpipe) {
		
		pipeselwakeup(cpipe);

		/*
		 * If the other side is blocked, wake it up saying that
		 * we want to close it down.
		 */
		while (cpipe->pipe_busy) {
			wakeup(cpipe);
			cpipe->pipe_state |= PIPE_WANT|PIPE_EOF;
			tsleep(cpipe, PRIBIO, "pipecl", 0);
		}

		/*
		 * Disconnect from peer
		 */
		if ((ppipe = cpipe->pipe_peer) != NULL) {
			pipeselwakeup(ppipe);

			ppipe->pipe_state |= PIPE_EOF;
			wakeup(ppipe);
			ppipe->pipe_peer = NULL;
		}

		/*
		 * free resources
		 */
		if (cpipe->pipe_buffer.buffer) {
			if (cpipe->pipe_buffer.size > PIPE_SIZE)
				--nbigpipe;
			amountpipekva -= cpipe->pipe_buffer.size;
			kmem_free(kernel_map,
				(vm_offset_t)cpipe->pipe_buffer.buffer,
				cpipe->pipe_buffer.size);
		}
#ifndef PIPE_NODIRECT
		if (cpipe->pipe_map.kva) {
			amountpipekva -= cpipe->pipe_buffer.size + PAGE_SIZE;
			kmem_free(kernel_map,
				cpipe->pipe_map.kva,
				cpipe->pipe_buffer.size + PAGE_SIZE);
		}
#endif
		free(cpipe, M_TEMP);
	}
}
#endif