summaryrefslogtreecommitdiffstats
path: root/sys/geom/geom_ccd.c
blob: 9af6b4a51af3ae6b224e150501b3f9efbdb34654 (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
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
/*
 * Copyright (c) 2003 Poul-Henning Kamp.
 * Copyright (c) 1995 Jason R. Thorpe.
 * Copyright (c) 1990, 1993
 *	The Regents of the University of California.  All rights reserved.
 * All rights reserved.
 * Copyright (c) 1988 University of Utah.
 *
 * This code is derived from software contributed to Berkeley by
 * the Systems Programming Group of the University of Utah Computer
 * Science Department.
 *
 * 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, 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. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed for the NetBSD Project
 *	by Jason R. Thorpe.
 * 4. The names of the authors may not be used to endorse or promote products
 *    derived from this software without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
 * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
 * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
 * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
 * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 * Dynamic configuration and disklabel support by:
 *	Jason R. Thorpe <thorpej@nas.nasa.gov>
 *	Numerical Aerodynamic Simulation Facility
 *	Mail Stop 258-6
 *	NASA Ames Research Center
 *	Moffett Field, CA 94035
 *
 * from: Utah $Hdr: cd.c 1.6 90/11/28$
 *
 *	@(#)cd.c	8.2 (Berkeley) 11/16/93
 *
 *	$NetBSD: ccd.c,v 1.22 1995/12/08 19:13:26 thorpej Exp $ 
 *
 * $FreeBSD$
 */

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/module.h>
#include <sys/proc.h>
#include <sys/bio.h>
#include <sys/malloc.h>
#include <sys/namei.h>
#include <sys/conf.h>
#include <sys/stat.h>
#include <sys/sysctl.h>
#include <sys/disk.h>
#include <sys/fcntl.h>
#include <sys/vnode.h>
#include <geom/geom_disk.h>

#include <sys/ccdvar.h>

MALLOC_DEFINE(M_CCD, "CCD driver", "Concatenated Disk driver");

/*
   This is how mirroring works (only writes are special):

   When initiating a write, ccdbuffer() returns two "struct ccdbuf *"s
   linked together by the cb_mirror field.  "cb_pflags &
   CCDPF_MIRROR_DONE" is set to 0 on both of them.

   When a component returns to ccdiodone(), it checks if "cb_pflags &
   CCDPF_MIRROR_DONE" is set or not.  If not, it sets the partner's
   flag and returns.  If it is, it means its partner has already
   returned, so it will go to the regular cleanup.

 */

struct ccdbuf {
	struct bio	cb_buf;		/* new I/O buf */
	struct bio	*cb_obp;	/* ptr. to original I/O buf */
	struct ccdbuf	*cb_freenext;	/* free list link */
	struct ccd_s	*cb_softc;
	int		cb_comp;	/* target component */
	int		cb_pflags;	/* mirror/parity status flag */
	struct ccdbuf	*cb_mirror;	/* mirror counterpart */
};

/* bits in cb_pflags */
#define CCDPF_MIRROR_DONE 1	/* if set, mirror counterpart is done */

/* convinient macros for often-used statements */
#define IS_ALLOCATED(unit)	(ccdfind(unit) != NULL)
#define IS_INITED(cs)		(((cs)->sc_flags & CCDF_INITED) != 0)

static dev_t	ccdctldev;

static disk_strategy_t ccdstrategy;
static d_ioctl_t ccdctlioctl;

#define NCCDFREEHIWAT	16

#define CDEV_MAJOR 74

static struct cdevsw ccdctl_cdevsw = {
	.d_open =	nullopen,
	.d_close =	nullclose,
	.d_ioctl =	ccdctlioctl,
	.d_name =	"ccdctl",
	.d_maj =	CDEV_MAJOR,
};

static LIST_HEAD(, ccd_s) ccd_softc_list =
	LIST_HEAD_INITIALIZER(&ccd_softc_list);

static struct ccd_s *ccdfind(int);
static struct ccd_s *ccdnew(int);
static int ccddestroy(struct ccd_s *);

/* called during module initialization */
static void ccdattach(void);
static int ccd_modevent(module_t, int, void *);

/* called by biodone() at interrupt time */
static void ccdiodone(struct bio *bp);

static void ccdstart(struct ccd_s *, struct bio *);
static void ccdinterleave(struct ccd_s *, int);
static int ccdinit(struct ccd_s *, char **, struct thread *);
static int ccdlookup(char *, struct thread *p, struct vnode **);
static int ccdbuffer(struct ccdbuf **ret, struct ccd_s *,
		      struct bio *, daddr_t, caddr_t, long);
static int ccdlock(struct ccd_s *);
static void ccdunlock(struct ccd_s *);


/*
 * Number of blocks to untouched in front of a component partition.
 * This is to avoid violating its disklabel area when it starts at the
 * beginning of the slice.
 */
#if !defined(CCD_OFFSET)
#define CCD_OFFSET 16
#endif

static struct ccd_s *
ccdfind(int unit)
{
	struct ccd_s *sc = NULL;

	/* XXX: LOCK(unique unit numbers) */
	LIST_FOREACH(sc, &ccd_softc_list, list) {
		if (sc->sc_unit == unit)
			break;
	}
	/* XXX: UNLOCK(unique unit numbers) */
	return ((sc == NULL) || (sc->sc_unit != unit) ? NULL : sc);
}

static struct ccd_s *
ccdnew(int unit)
{
	struct ccd_s *sc;

	/* XXX: LOCK(unique unit numbers) */
	if (IS_ALLOCATED(unit) || unit > 32)
		return (NULL);

	MALLOC(sc, struct ccd_s *, sizeof(*sc), M_CCD, M_WAITOK | M_ZERO);
	sc->sc_unit = unit;
	LIST_INSERT_HEAD(&ccd_softc_list, sc, list);
	/* XXX: UNLOCK(unique unit numbers) */
	return (sc);
}

static int
ccddestroy(struct ccd_s *sc)
{

	/* XXX: LOCK(unique unit numbers) */
	LIST_REMOVE(sc, list);
	/* XXX: UNLOCK(unique unit numbers) */
	FREE(sc, M_CCD);
	return (0);
}

/*
 * Called by main() during pseudo-device attachment.  All we need
 * to do is to add devsw entries.
 */
static void
ccdattach()
{

	ccdctldev = make_dev(&ccdctl_cdevsw, 0xffff00ff,
		UID_ROOT, GID_OPERATOR, 0640, "ccd.ctl");
	ccdctldev->si_drv1 = ccdctldev;
}

static int
ccd_modevent(module_t mod, int type, void *data)
{
	int error = 0;

	switch (type) {
	case MOD_LOAD:
		ccdattach();
		break;

	case MOD_UNLOAD:
		printf("ccd0: Unload not supported!\n");
		error = EOPNOTSUPP;
		break;

	case MOD_SHUTDOWN:
		break;

	default:
		error = EOPNOTSUPP;
	}
	return (error);
}

DEV_MODULE(ccd, ccd_modevent, NULL);

static int
ccdinit(struct ccd_s *cs, char **cpaths, struct thread *td)
{
	struct ccdcinfo *ci = NULL;	/* XXX */
	size_t size;
	int ix;
	struct vnode *vp;
	size_t minsize;
	int maxsecsize;
	struct ccdgeom *ccg = &cs->sc_geom;
	char *tmppath = NULL;
	int error = 0;
	off_t mediasize;
	u_int sectorsize;


	cs->sc_size = 0;

	/* Allocate space for the component info. */
	cs->sc_cinfo = malloc(cs->sc_nccdisks * sizeof(struct ccdcinfo),
	    M_CCD, M_WAITOK);

	/*
	 * Verify that each component piece exists and record
	 * relevant information about it.
	 */
	maxsecsize = 0;
	minsize = 0;
	tmppath = malloc(MAXPATHLEN, M_CCD, M_WAITOK);
	for (ix = 0; ix < cs->sc_nccdisks; ix++) {
		vp = cs->sc_vpp[ix];
		ci = &cs->sc_cinfo[ix];
		ci->ci_vp = vp;

		/*
		 * Copy in the pathname of the component.
		 */
		if ((error = copyinstr(cpaths[ix], tmppath,
		    MAXPATHLEN, &ci->ci_pathlen)) != 0) {
			goto fail;
		}
		ci->ci_path = malloc(ci->ci_pathlen, M_CCD, M_WAITOK);
		bcopy(tmppath, ci->ci_path, ci->ci_pathlen);

		ci->ci_dev = vn_todev(vp);

		/*
		 * Get partition information for the component.
		 */
		error = VOP_IOCTL(vp, DIOCGMEDIASIZE, (caddr_t)&mediasize,
		    FREAD, td->td_ucred, td);
		if (error != 0) {
			goto fail;
		}
		/*
		 * Get partition information for the component.
		 */
		error = VOP_IOCTL(vp, DIOCGSECTORSIZE, (caddr_t)&sectorsize,
		    FREAD, td->td_ucred, td);
		if (error != 0) {
			goto fail;
		}
		if (sectorsize > maxsecsize)
			maxsecsize = sectorsize;
		size = mediasize / DEV_BSIZE - CCD_OFFSET;

		/*
		 * Calculate the size, truncating to an interleave
		 * boundary if necessary.
		 */

		if (cs->sc_ileave > 1)
			size -= size % cs->sc_ileave;

		if (size == 0) {
			error = ENODEV;
			goto fail;
		}

		if (minsize == 0 || size < minsize)
			minsize = size;
		ci->ci_size = size;
		cs->sc_size += size;
	}

	free(tmppath, M_CCD);
	tmppath = NULL;

	/*
	 * Don't allow the interleave to be smaller than
	 * the biggest component sector.
	 */
	if ((cs->sc_ileave > 0) &&
	    (cs->sc_ileave < (maxsecsize / DEV_BSIZE))) {
		error = EINVAL;
		goto fail;
	}

	/*
	 * If uniform interleave is desired set all sizes to that of
	 * the smallest component.  This will guarentee that a single
	 * interleave table is generated.
	 *
	 * Lost space must be taken into account when calculating the
	 * overall size.  Half the space is lost when CCDF_MIRROR is
	 * specified.
	 */
	if (cs->sc_flags & CCDF_UNIFORM) {
		for (ci = cs->sc_cinfo;
		     ci < &cs->sc_cinfo[cs->sc_nccdisks]; ci++) {
			ci->ci_size = minsize;
		}
		if (cs->sc_flags & CCDF_MIRROR) {
			/*
			 * Check to see if an even number of components
			 * have been specified.  The interleave must also
			 * be non-zero in order for us to be able to 
			 * guarentee the topology.
			 */
			if (cs->sc_nccdisks % 2) {
				printf("ccd%d: mirroring requires an even number of disks\n", cs->sc_unit );
				error = EINVAL;
				goto fail;
			}
			if (cs->sc_ileave == 0) {
				printf("ccd%d: an interleave must be specified when mirroring\n", cs->sc_unit);
				error = EINVAL;
				goto fail;
			}
			cs->sc_size = (cs->sc_nccdisks/2) * minsize;
		} else {
			if (cs->sc_ileave == 0) {
				printf("ccd%d: an interleave must be specified when using parity\n", cs->sc_unit);
				error = EINVAL;
				goto fail;
			}
			cs->sc_size = cs->sc_nccdisks * minsize;
		}
	}

	/*
	 * Construct the interleave table.
	 */
	ccdinterleave(cs, cs->sc_unit);

	/*
	 * Create pseudo-geometry based on 1MB cylinders.  It's
	 * pretty close.
	 */
	ccg->ccg_secsize = maxsecsize;
	ccg->ccg_ntracks = 1;
	ccg->ccg_nsectors = 1024 * 1024 / ccg->ccg_secsize;
	ccg->ccg_ncylinders = cs->sc_size / ccg->ccg_nsectors;

	cs->sc_flags |= CCDF_INITED;
	cs->sc_cflags = cs->sc_flags;	/* So we can find out later... */
	return (0);
fail:
	while (ci > cs->sc_cinfo) {
		ci--;
		free(ci->ci_path, M_CCD);
	}
	if (tmppath != NULL)
		free(tmppath, M_CCD);
	free(cs->sc_cinfo, M_CCD);
	ccddestroy(cs);
	return (error);
}

static void
ccdinterleave(struct ccd_s *cs, int unit)
{
	struct ccdcinfo *ci, *smallci;
	struct ccdiinfo *ii;
	daddr_t bn, lbn;
	int ix;
	u_long size;


	/*
	 * Allocate an interleave table.  The worst case occurs when each
	 * of N disks is of a different size, resulting in N interleave
	 * tables.
	 *
	 * Chances are this is too big, but we don't care.
	 */
	size = (cs->sc_nccdisks + 1) * sizeof(struct ccdiinfo);
	cs->sc_itable = (struct ccdiinfo *)malloc(size, M_CCD,
	    M_WAITOK | M_ZERO);

	/*
	 * Trivial case: no interleave (actually interleave of disk size).
	 * Each table entry represents a single component in its entirety.
	 *
	 * An interleave of 0 may not be used with a mirror setup.
	 */
	if (cs->sc_ileave == 0) {
		bn = 0;
		ii = cs->sc_itable;

		for (ix = 0; ix < cs->sc_nccdisks; ix++) {
			/* Allocate space for ii_index. */
			ii->ii_index = malloc(sizeof(int), M_CCD, M_WAITOK);
			ii->ii_ndisk = 1;
			ii->ii_startblk = bn;
			ii->ii_startoff = 0;
			ii->ii_index[0] = ix;
			bn += cs->sc_cinfo[ix].ci_size;
			ii++;
		}
		ii->ii_ndisk = 0;
		return;
	}

	/*
	 * The following isn't fast or pretty; it doesn't have to be.
	 */
	size = 0;
	bn = lbn = 0;
	for (ii = cs->sc_itable; ; ii++) {
		/*
		 * Allocate space for ii_index.  We might allocate more then
		 * we use.
		 */
		ii->ii_index = malloc((sizeof(int) * cs->sc_nccdisks),
		    M_CCD, M_WAITOK);

		/*
		 * Locate the smallest of the remaining components
		 */
		smallci = NULL;
		for (ci = cs->sc_cinfo; ci < &cs->sc_cinfo[cs->sc_nccdisks]; 
		    ci++) {
			if (ci->ci_size > size &&
			    (smallci == NULL ||
			     ci->ci_size < smallci->ci_size)) {
				smallci = ci;
			}
		}

		/*
		 * Nobody left, all done
		 */
		if (smallci == NULL) {
			ii->ii_ndisk = 0;
			free(ii->ii_index, M_CCD);
			break;
		}

		/*
		 * Record starting logical block using an sc_ileave blocksize.
		 */
		ii->ii_startblk = bn / cs->sc_ileave;

		/*
		 * Record starting comopnent block using an sc_ileave 
		 * blocksize.  This value is relative to the beginning of
		 * a component disk.
		 */
		ii->ii_startoff = lbn;

		/*
		 * Determine how many disks take part in this interleave
		 * and record their indices.
		 */
		ix = 0;
		for (ci = cs->sc_cinfo; 
		    ci < &cs->sc_cinfo[cs->sc_nccdisks]; ci++) {
			if (ci->ci_size >= smallci->ci_size) {
				ii->ii_index[ix++] = ci - cs->sc_cinfo;
			}
		}
		ii->ii_ndisk = ix;
		bn += ix * (smallci->ci_size - size);
		lbn = smallci->ci_size / cs->sc_ileave;
		size = smallci->ci_size;
	}
}

static void
ccdstrategy(struct bio *bp)
{
	struct ccd_s *cs;
	int pbn;        /* in sc_secsize chunks */
	long sz;        /* in sc_secsize chunks */

	cs = bp->bio_disk->d_drv1;

	pbn = bp->bio_blkno / (cs->sc_geom.ccg_secsize / DEV_BSIZE);
	sz = howmany(bp->bio_bcount, cs->sc_geom.ccg_secsize);

	/*
	 * If out of bounds return an error. If at the EOF point,
	 * simply read or write less.
	 */

	if (pbn < 0 || pbn >= cs->sc_size) {
		bp->bio_resid = bp->bio_bcount;
		if (pbn != cs->sc_size)
			biofinish(bp, NULL, EINVAL);
		else
			biodone(bp);
		return;
	}

	/*
	 * If the request crosses EOF, truncate the request.
	 */
	if (pbn + sz > cs->sc_size) {
		bp->bio_bcount = (cs->sc_size - pbn) * 
		    cs->sc_geom.ccg_secsize;
	}

	bp->bio_resid = bp->bio_bcount;

	/*
	 * "Start" the unit.
	 */
	ccdstart(cs, bp);
	return;
}

static void
ccdstart(struct ccd_s *cs, struct bio *bp)
{
	long bcount, rcount;
	struct ccdbuf *cbp[2];
	caddr_t addr;
	daddr_t bn;
	int err;
	int sent;

	/*
	 * Translate the partition-relative block number to an absolute.
	 */
	bn = bp->bio_blkno;

	/*
	 * Allocate component buffers and fire off the requests
	 */
	addr = bp->bio_data;
	sent = 0;
	for (bcount = bp->bio_bcount; bcount > 0; bcount -= rcount) {
		err = ccdbuffer(cbp, cs, bp, bn, addr, bcount);
		if (err) {
			printf("ccdbuffer error %d\n", err);
			if (!sent)
				biofinish(bp, NULL, err);
			else {
				/*
				 * XXX: maybe a race where the partners
				 * XXX: we sent already have been in 
				 * XXX: ccdiodone().  Single-threaded g_down
				 * XXX: may protect against this.
				 */
				bp->bio_resid -= bcount;
				bp->bio_error = err;
				bp->bio_flags |= BIO_ERROR;
			}
			return;
		}
		rcount = cbp[0]->cb_buf.bio_bcount;

		if (cs->sc_cflags & CCDF_MIRROR) {
			/*
			 * Mirroring.  Writes go to both disks, reads are
			 * taken from whichever disk seems most appropriate.
			 *
			 * We attempt to localize reads to the disk whos arm
			 * is nearest the read request.  We ignore seeks due
			 * to writes when making this determination and we
			 * also try to avoid hogging.
			 */
			if (cbp[0]->cb_buf.bio_cmd == BIO_WRITE) {
				BIO_STRATEGY(&cbp[0]->cb_buf);
				BIO_STRATEGY(&cbp[1]->cb_buf);
				sent++;
			} else {
				int pick = cs->sc_pick;
				daddr_t range = cs->sc_size / 16;

				if (bn < cs->sc_blk[pick] - range ||
				    bn > cs->sc_blk[pick] + range
				) {
					cs->sc_pick = pick = 1 - pick;
				}
				cs->sc_blk[pick] = bn + btodb(rcount);
				BIO_STRATEGY(&cbp[pick]->cb_buf);
				sent++;
			}
		} else {
			/*
			 * Not mirroring
			 */
			BIO_STRATEGY(&cbp[0]->cb_buf);
			sent++;
		}
		bn += btodb(rcount);
		addr += rcount;
	}
}

/*
 * Build a component buffer header.
 */
static int
ccdbuffer(struct ccdbuf **cb, struct ccd_s *cs, struct bio *bp, daddr_t bn, caddr_t addr, long bcount)
{
	struct ccdcinfo *ci, *ci2 = NULL;	/* XXX */
	struct ccdbuf *cbp;
	daddr_t cbn, cboff;
	off_t cbc;

	/*
	 * Determine which component bn falls in.
	 */
	cbn = bn;
	cboff = 0;

	if (cs->sc_ileave == 0) {
		/*
		 * Serially concatenated and neither a mirror nor a parity
		 * config.  This is a special case.
		 */
		daddr_t sblk;

		sblk = 0;
		for (ci = cs->sc_cinfo; cbn >= sblk + ci->ci_size; ci++)
			sblk += ci->ci_size;
		cbn -= sblk;
	} else {
		struct ccdiinfo *ii;
		int ccdisk, off;

		/*
		 * Calculate cbn, the logical superblock (sc_ileave chunks),
		 * and cboff, a normal block offset (DEV_BSIZE chunks) relative
		 * to cbn.
		 */
		cboff = cbn % cs->sc_ileave;	/* DEV_BSIZE gran */
		cbn = cbn / cs->sc_ileave;	/* DEV_BSIZE * ileave gran */

		/*
		 * Figure out which interleave table to use.
		 */
		for (ii = cs->sc_itable; ii->ii_ndisk; ii++) {
			if (ii->ii_startblk > cbn)
				break;
		}
		ii--;

		/*
		 * off is the logical superblock relative to the beginning 
		 * of this interleave block.  
		 */
		off = cbn - ii->ii_startblk;

		/*
		 * We must calculate which disk component to use (ccdisk),
		 * and recalculate cbn to be the superblock relative to
		 * the beginning of the component.  This is typically done by
		 * adding 'off' and ii->ii_startoff together.  However, 'off'
		 * must typically be divided by the number of components in
		 * this interleave array to be properly convert it from a
		 * CCD-relative logical superblock number to a 
		 * component-relative superblock number.
		 */
		if (ii->ii_ndisk == 1) {
			/*
			 * When we have just one disk, it can't be a mirror
			 * or a parity config.
			 */
			ccdisk = ii->ii_index[0];
			cbn = ii->ii_startoff + off;
		} else {
			if (cs->sc_cflags & CCDF_MIRROR) {
				/*
				 * We have forced a uniform mapping, resulting
				 * in a single interleave array.  We double
				 * up on the first half of the available
				 * components and our mirror is in the second
				 * half.  This only works with a single 
				 * interleave array because doubling up
				 * doubles the number of sectors, so there
				 * cannot be another interleave array because
				 * the next interleave array's calculations
				 * would be off.
				 */
				int ndisk2 = ii->ii_ndisk / 2;
				ccdisk = ii->ii_index[off % ndisk2];
				cbn = ii->ii_startoff + off / ndisk2;
				ci2 = &cs->sc_cinfo[ccdisk + ndisk2];
			} else {
				ccdisk = ii->ii_index[off % ii->ii_ndisk];
				cbn = ii->ii_startoff + off / ii->ii_ndisk;
			}
		}

		ci = &cs->sc_cinfo[ccdisk];

		/*
		 * Convert cbn from a superblock to a normal block so it
		 * can be used to calculate (along with cboff) the normal
		 * block index into this particular disk.
		 */
		cbn *= cs->sc_ileave;
	}

	/*
	 * Fill in the component buf structure.
	 */
	cbp = malloc(sizeof(struct ccdbuf), M_CCD, M_NOWAIT | M_ZERO);
	if (cbp == NULL)
		return (ENOMEM);
	cbp->cb_buf.bio_cmd = bp->bio_cmd;
	cbp->cb_buf.bio_done = ccdiodone;
	cbp->cb_buf.bio_dev = ci->ci_dev;		/* XXX */
	cbp->cb_buf.bio_blkno = cbn + cboff + CCD_OFFSET;
	cbp->cb_buf.bio_offset = dbtob(cbn + cboff + CCD_OFFSET);
	cbp->cb_buf.bio_data = addr;
	cbp->cb_buf.bio_caller2 = cbp;
	if (cs->sc_ileave == 0)
              cbc = dbtob((off_t)(ci->ci_size - cbn));
	else
              cbc = dbtob((off_t)(cs->sc_ileave - cboff));
	cbp->cb_buf.bio_bcount = (cbc < bcount) ? cbc : bcount;
 	cbp->cb_buf.bio_caller1 = (void*)cbp->cb_buf.bio_bcount;

	/*
	 * context for ccdiodone
	 */
	cbp->cb_obp = bp;
	cbp->cb_softc = cs;
	cbp->cb_comp = ci - cs->sc_cinfo;

	cb[0] = cbp;

	/*
	 * Note: both I/O's setup when reading from mirror, but only one
	 * will be executed.
	 */
	if (cs->sc_cflags & CCDF_MIRROR) {
		/* mirror, setup second I/O */
		cbp = malloc(sizeof(struct ccdbuf), M_CCD, M_NOWAIT);
		if (cbp == NULL) {
			free(cb[0], M_CCD);
			cb[0] = NULL;
			return (ENOMEM);
		}
		bcopy(cb[0], cbp, sizeof(struct ccdbuf));
		cbp->cb_buf.bio_caller2 = cbp;
		cbp->cb_buf.bio_dev = ci2->ci_dev;
		cbp->cb_comp = ci2 - cs->sc_cinfo;
		cb[1] = cbp;
		/* link together the ccdbuf's and clear "mirror done" flag */
		cb[0]->cb_mirror = cb[1];
		cb[1]->cb_mirror = cb[0];
		cb[0]->cb_pflags &= ~CCDPF_MIRROR_DONE;
		cb[1]->cb_pflags &= ~CCDPF_MIRROR_DONE;
	}
	return (0);
}

/*
 * Called at interrupt time.
 * Mark the component as done and if all components are done,
 * take a ccd interrupt.
 */
static void
ccdiodone(struct bio *ibp)
{
	struct ccdbuf *cbp;
	struct bio *bp;
	struct ccd_s *cs;
	int count;

	cbp = ibp->bio_caller2;
	cs = cbp->cb_softc;
	bp = cbp->cb_obp;
	/*
	 * If an error occured, report it.  If this is a mirrored 
	 * configuration and the first of two possible reads, do not
	 * set the error in the bp yet because the second read may
	 * succeed.
	 */

	if (cbp->cb_buf.bio_flags & BIO_ERROR) {
		const char *msg = "";

		if ((cs->sc_cflags & CCDF_MIRROR) &&
		    (cbp->cb_buf.bio_cmd == BIO_READ) &&
		    (cbp->cb_pflags & CCDPF_MIRROR_DONE) == 0) {
			/*
			 * We will try our read on the other disk down
			 * below, also reverse the default pick so if we 
			 * are doing a scan we do not keep hitting the
			 * bad disk first.
			 */

			msg = ", trying other disk";
			cs->sc_pick = 1 - cs->sc_pick;
			cs->sc_blk[cs->sc_pick] = bp->bio_blkno;
		} else {
			bp->bio_flags |= BIO_ERROR;
			bp->bio_error = cbp->cb_buf.bio_error ? 
			    cbp->cb_buf.bio_error : EIO;
		}
		printf("ccd%d: error %d on component %d block %jd "
		    "(ccd block %jd)%s\n", cs->sc_unit, bp->bio_error,
		    cbp->cb_comp, 
		    (intmax_t)cbp->cb_buf.bio_blkno, (intmax_t)bp->bio_blkno,
		    msg);
	}

	/*
	 * Process mirror.  If we are writing, I/O has been initiated on both
	 * buffers and we fall through only after both are finished.
	 *
	 * If we are reading only one I/O is initiated at a time.  If an
	 * error occurs we initiate the second I/O and return, otherwise 
	 * we free the second I/O without initiating it.
	 */

	if (cs->sc_cflags & CCDF_MIRROR) {
		if (cbp->cb_buf.bio_cmd == BIO_WRITE) {
			/*
			 * When writing, handshake with the second buffer
			 * to determine when both are done.  If both are not
			 * done, return here.
			 */
			if ((cbp->cb_pflags & CCDPF_MIRROR_DONE) == 0) {
				cbp->cb_mirror->cb_pflags |= CCDPF_MIRROR_DONE;
				free(cbp, M_CCD);
				return;
			}
		} else {
			/*
			 * When reading, either dispose of the second buffer
			 * or initiate I/O on the second buffer if an error 
			 * occured with this one.
			 */
			if ((cbp->cb_pflags & CCDPF_MIRROR_DONE) == 0) {
				if (cbp->cb_buf.bio_flags & BIO_ERROR) {
					cbp->cb_mirror->cb_pflags |= 
					    CCDPF_MIRROR_DONE;
					BIO_STRATEGY(&cbp->cb_mirror->cb_buf);
					free(cbp, M_CCD);
					return;
				} else {
					free(cbp->cb_mirror, M_CCD);
				}
			}
		}
	}

	/*
	 * use bio_caller1 to determine how big the original request was rather
	 * then bio_bcount, because bio_bcount may have been truncated for EOF.
	 *
	 * XXX We check for an error, but we do not test the resid for an
	 * aligned EOF condition.  This may result in character & block
	 * device access not recognizing EOF properly when read or written 
	 * sequentially, but will not effect filesystems.
	 */
	count = (long)cbp->cb_buf.bio_caller1;
	free(cbp, M_CCD);

	/*
	 * If all done, "interrupt".
	 */
	bp->bio_resid -= count;
	if (bp->bio_resid < 0)
		panic("ccdiodone: count");
	if (bp->bio_resid == 0) {
		if (bp->bio_flags & BIO_ERROR)
			bp->bio_resid = bp->bio_bcount;
		biodone(bp);
	}
}

static int ccdioctltoo(int unit, u_long cmd, caddr_t data, int flag, struct thread *td);

static int
ccdctlioctl(dev_t dev, u_long cmd, caddr_t data, int flag, struct thread *td)
{
	struct ccd_ioctl *ccio;
	u_int unit;
	dev_t dev2;
	int error;

	switch (cmd) {
	case CCDIOCSET:
	case CCDIOCCLR:
		ccio = (struct ccd_ioctl *)data;
		unit = ccio->ccio_size;
		return (ccdioctltoo(unit, cmd, data, flag, td));
	case CCDCONFINFO:
		{
		int ninit = 0;
		struct ccdconf *conf = (struct ccdconf *)data;
		struct ccd_s *tmpcs;
		struct ccd_s *ubuf = conf->buffer;

		/* XXX: LOCK(unique unit numbers) */
		LIST_FOREACH(tmpcs, &ccd_softc_list, list)
			if (IS_INITED(tmpcs))
				ninit++;

		if (conf->size == 0) {
			conf->size = sizeof(struct ccd_s) * ninit;
			return (0);
		} else if ((conf->size / sizeof(struct ccd_s) != ninit) ||
		    (conf->size % sizeof(struct ccd_s) != 0)) {
			/* XXX: UNLOCK(unique unit numbers) */
			return (EINVAL);
		}

		ubuf += ninit;
		LIST_FOREACH(tmpcs, &ccd_softc_list, list) {
			if (!IS_INITED(tmpcs))
				continue;
			error = copyout(tmpcs, --ubuf,
			    sizeof(struct ccd_s));
			if (error != 0)
				/* XXX: UNLOCK(unique unit numbers) */
				return (error);
		}
		/* XXX: UNLOCK(unique unit numbers) */
		return (0);
		}

	case CCDCPPINFO:
		{
		struct ccdcpps *cpps = (struct ccdcpps *)data;
		char *ubuf = cpps->buffer;
		struct ccd_s *cs;

	
		error = copyin(ubuf, &unit, sizeof (unit));
		if (error)
			return (error);

		if (!IS_ALLOCATED(unit))
			return (ENXIO);
		dev2 = makedev(CDEV_MAJOR, unit * 8 + 2);
		cs = ccdfind(unit);
		if (!IS_INITED(cs))
			return (ENXIO);

		{
			int len = 0, i;
			struct ccdcpps *cpps = (struct ccdcpps *)data;
			char *ubuf = cpps->buffer;


			for (i = 0; i < cs->sc_nccdisks; ++i)
				len += cs->sc_cinfo[i].ci_pathlen;

			if (cpps->size < len)
				return (ENOMEM);

			for (i = 0; i < cs->sc_nccdisks; ++i) {
				len = cs->sc_cinfo[i].ci_pathlen;
				error = copyout(cs->sc_cinfo[i].ci_path, ubuf,
				    len);
				if (error != 0)
					return (error);
				ubuf += len;
			}
			return(copyout("", ubuf, 1));
		}
		break;
		}

	default:
		return (ENXIO);
	}
}

static int
ccdioctltoo(int unit, u_long cmd, caddr_t data, int flag, struct thread *td)
{
	int i, j, lookedup = 0, error = 0;
	struct ccd_s *cs;
	struct ccd_ioctl *ccio = (struct ccd_ioctl *)data;
	struct ccdgeom *ccg;
	char **cpp;
	struct vnode **vpp;

	cs = ccdfind(unit);
	switch (cmd) {
	case CCDIOCSET:
		if (cs == NULL)
			cs = ccdnew(unit);
		if (IS_INITED(cs))
			return (EBUSY);

		if ((flag & FWRITE) == 0)
			return (EBADF);

		if ((error = ccdlock(cs)) != 0)
			return (error);

		if (ccio->ccio_ndisks > CCD_MAXNDISKS)
			return (EINVAL);
 
		/* Fill in some important bits. */
		cs->sc_ileave = ccio->ccio_ileave;
		if (cs->sc_ileave == 0 && (ccio->ccio_flags & CCDF_MIRROR)) {
			printf("ccd%d: disabling mirror, interleave is 0\n",
			    unit);
			ccio->ccio_flags &= ~(CCDF_MIRROR);
		}
		if ((ccio->ccio_flags & CCDF_MIRROR) &&
		    !(ccio->ccio_flags & CCDF_UNIFORM)) {
			printf("ccd%d: mirror/parity forces uniform flag\n",
			       unit);
			ccio->ccio_flags |= CCDF_UNIFORM;
		}
		cs->sc_flags = ccio->ccio_flags & CCDF_USERMASK;

		/*
		 * Allocate space for and copy in the array of
		 * componet pathnames and device numbers.
		 */
		cpp = malloc(ccio->ccio_ndisks * sizeof(char *),
		    M_CCD, M_WAITOK);
		vpp = malloc(ccio->ccio_ndisks * sizeof(struct vnode *),
		    M_CCD, M_WAITOK);

		error = copyin((caddr_t)ccio->ccio_disks, (caddr_t)cpp,
		    ccio->ccio_ndisks * sizeof(char **));
		if (error) {
			free(vpp, M_CCD);
			free(cpp, M_CCD);
			ccdunlock(cs);
			return (error);
		}


		for (i = 0; i < ccio->ccio_ndisks; ++i) {
			if ((error = ccdlookup(cpp[i], td, &vpp[i])) != 0) {
				for (j = 0; j < lookedup; ++j)
					(void)vn_close(vpp[j], FREAD|FWRITE,
					    td->td_ucred, td);
				free(vpp, M_CCD);
				free(cpp, M_CCD);
				ccdunlock(cs);
				return (error);
			}
			++lookedup;
		}
		cs->sc_vpp = vpp;
		cs->sc_nccdisks = ccio->ccio_ndisks;

		/*
		 * Initialize the ccd.  Fills in the softc for us.
		 */
		if ((error = ccdinit(cs, cpp, td)) != 0) {
			for (j = 0; j < lookedup; ++j)
				(void)vn_close(vpp[j], FREAD|FWRITE,
				    td->td_ucred, td);
			/*
			 * We can't ccddestroy() cs just yet, because nothing
			 * prevents user-level app to do another ioctl()
			 * without closing the device first, therefore
			 * declare unit null and void and let ccdclose()
			 * destroy it when it is safe to do so.
			 */
			cs->sc_flags &= (CCDF_WANTED | CCDF_LOCKED);
			free(vpp, M_CCD);
			free(cpp, M_CCD);
			ccdunlock(cs);
			return (error);
		}
		free(cpp, M_CCD);

		/*
		 * The ccd has been successfully initialized, so
		 * we can place it into the array and read the disklabel.
		 */
		ccio->ccio_unit = unit;
		ccio->ccio_size = cs->sc_size;
		ccg = &cs->sc_geom;
		cs->sc_disk = malloc(sizeof(struct disk), M_CCD,
		    M_ZERO | M_WAITOK);
		cs->sc_disk->d_strategy = ccdstrategy;
		cs->sc_disk->d_name = "ccd";
		cs->sc_disk->d_sectorsize = ccg->ccg_secsize;
		cs->sc_disk->d_mediasize =
		    cs->sc_size * (off_t)ccg->ccg_secsize;
		cs->sc_disk->d_fwsectors = ccg->ccg_nsectors;
		cs->sc_disk->d_fwheads = ccg->ccg_ntracks;
		cs->sc_disk->d_drv1 = cs;
		cs->sc_disk->d_maxsize = MAXPHYS;
		disk_create(unit, cs->sc_disk, 0, NULL, NULL);

		ccdunlock(cs);

		break;

	case CCDIOCCLR:
		if (cs == NULL)
			return (ENXIO);

		if (!IS_INITED(cs))
			return (ENXIO);

		if ((flag & FWRITE) == 0)
			return (EBADF);

		if ((error = ccdlock(cs)) != 0)
			return (error);

		/* Don't unconfigure if any other partitions are open */
		if (cs->sc_disk->d_flags & DISKFLAG_OPEN) {
			ccdunlock(cs);
			return (EBUSY);
		}

		disk_destroy(cs->sc_disk);
		free(cs->sc_disk, M_CCD);
		cs->sc_disk = NULL;
		/* Declare unit null and void (reset all flags) */
		cs->sc_flags &= (CCDF_WANTED | CCDF_LOCKED);

		/* Close the components and free their pathnames. */
		for (i = 0; i < cs->sc_nccdisks; ++i) {
			/*
			 * XXX: this close could potentially fail and
			 * cause Bad Things.  Maybe we need to force
			 * the close to happen?
			 */
			(void)vn_close(cs->sc_cinfo[i].ci_vp, FREAD|FWRITE,
			    td->td_ucred, td);
			free(cs->sc_cinfo[i].ci_path, M_CCD);
		}

		/* Free interleave index. */
		for (i = 0; cs->sc_itable[i].ii_ndisk; ++i)
			free(cs->sc_itable[i].ii_index, M_CCD);

		/* Free component info and interleave table. */
		free(cs->sc_cinfo, M_CCD);
		free(cs->sc_itable, M_CCD);
		free(cs->sc_vpp, M_CCD);

		/* This must be atomic. */
		ccdunlock(cs);
		ccddestroy(cs);

		break;
	}

	return (0);
}


/*
 * Lookup the provided name in the filesystem.  If the file exists,
 * is a valid block device, and isn't being used by anyone else,
 * set *vpp to the file's vnode.
 */
static int
ccdlookup(char *path, struct thread *td, struct vnode **vpp)
{
	struct nameidata nd;
	struct vnode *vp;
	int error, flags;

	NDINIT(&nd, LOOKUP, FOLLOW, UIO_USERSPACE, path, td);
	flags = FREAD | FWRITE;
	if ((error = vn_open(&nd, &flags, 0)) != 0) {
		return (error);
	}
	vp = nd.ni_vp;

	if (vrefcnt(vp) > 1) {
		error = EBUSY;
		goto bad;
	}

	if (!vn_isdisk(vp, &error)) 
		goto bad;


	VOP_UNLOCK(vp, 0, td);
	NDFREE(&nd, NDF_ONLY_PNBUF);
	*vpp = vp;
	return (0);
bad:
	VOP_UNLOCK(vp, 0, td);
	NDFREE(&nd, NDF_ONLY_PNBUF);
	/* vn_close does vrele() for vp */
	(void)vn_close(vp, FREAD|FWRITE, td->td_ucred, td);
	return (error);
}

/*

 * Wait interruptibly for an exclusive lock.
 *
 * XXX
 * Several drivers do this; it should be abstracted and made MP-safe.
 */
static int
ccdlock(struct ccd_s *cs)
{
	int error;

	while ((cs->sc_flags & CCDF_LOCKED) != 0) {
		cs->sc_flags |= CCDF_WANTED;
		if ((error = tsleep(cs, PRIBIO | PCATCH, "ccdlck", 0)) != 0)
			return (error);
	}
	cs->sc_flags |= CCDF_LOCKED;
	return (0);
}

/*
 * Unlock and wake up any waiters.
 */
static void
ccdunlock(struct ccd_s *cs)
{

	cs->sc_flags &= ~CCDF_LOCKED;
	if ((cs->sc_flags & CCDF_WANTED) != 0) {
		cs->sc_flags &= ~CCDF_WANTED;
		wakeup(cs);
	}
}
OpenPOWER on IntegriCloud