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
|
/* $NetBSD: subr_disk.c,v 1.58 2004/01/10 14:49:44 yamt Exp $ */
/*-
* Copyright (c) 1996, 1997, 1999, 2000 The NetBSD Foundation, Inc.
* All rights reserved.
*
* This code is derived from software contributed to The NetBSD Foundation
* by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
* NASA Ames Research Center.
*
* 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 by the NetBSD
* Foundation, Inc. and its contributors.
* 4. Neither the name of The NetBSD Foundation nor the names of its
* contributors may be used to endorse or promote products derived
* from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
* ``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 FOUNDATION OR CONTRIBUTORS
* 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.
*/
/*
* Copyright (c) 1982, 1986, 1988, 1993
* The Regents of the University of California. All rights reserved.
* (c) UNIX System Laboratories, Inc.
* All or some portions of this file are derived from material licensed
* to the University of California by American Telephone and Telegraph
* Co. or Unix System Laboratories, Inc. and are reproduced herein with
* the permission of UNIX System Laboratories, Inc.
*
* 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. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``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 REGENTS OR CONTRIBUTORS 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.
*
* @(#)ufs_disksubr.c 8.5 (Berkeley) 1/21/94
*/
#include <sys/cdefs.h>
__KERNEL_RCSID(0, "$NetBSD: subr_disk.c,v 1.58 2004/01/10 14:49:44 yamt Exp $");
#include "opt_compat_netbsd.h"
#include <sys/param.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/buf.h>
#include <sys/syslog.h>
#include <sys/disklabel.h>
#include <sys/disk.h>
#include <sys/sysctl.h>
#include <lib/libkern/libkern.h>
/*
* A global list of all disks attached to the system. May grow or
* shrink over time.
*/
struct disklist_head disklist; /* TAILQ_HEAD */
int disk_count; /* number of drives in global disklist */
struct simplelock disklist_slock = SIMPLELOCK_INITIALIZER;
/*
* Compute checksum for disk label.
*/
u_int
dkcksum(struct disklabel *lp)
{
u_short *start, *end;
u_short sum = 0;
start = (u_short *)lp;
end = (u_short *)&lp->d_partitions[lp->d_npartitions];
while (start < end)
sum ^= *start++;
return (sum);
}
/*
* Disk error is the preface to plaintive error messages
* about failing disk transfers. It prints messages of the form
hp0g: hard error reading fsbn 12345 of 12344-12347 (hp0 bn %d cn %d tn %d sn %d)
* if the offset of the error in the transfer and a disk label
* are both available. blkdone should be -1 if the position of the error
* is unknown; the disklabel pointer may be null from drivers that have not
* been converted to use them. The message is printed with printf
* if pri is LOG_PRINTF, otherwise it uses log at the specified priority.
* The message should be completed (with at least a newline) with printf
* or addlog, respectively. There is no trailing space.
*/
#ifndef PRIdaddr
#define PRIdaddr PRId64
#endif
void
diskerr(const struct buf *bp, const char *dname, const char *what, int pri,
int blkdone, const struct disklabel *lp)
{
int unit = DISKUNIT(bp->b_dev), part = DISKPART(bp->b_dev);
void (*pr)(const char *, ...);
char partname = 'a' + part;
daddr_t sn;
if (/*CONSTCOND*/0)
/* Compiler will error this is the format is wrong... */
printf("%" PRIdaddr, bp->b_blkno);
if (pri != LOG_PRINTF) {
static const char fmt[] = "";
log(pri, fmt);
pr = addlog;
} else
pr = printf;
(*pr)("%s%d%c: %s %sing fsbn ", dname, unit, partname, what,
bp->b_flags & B_READ ? "read" : "writ");
sn = bp->b_blkno;
if (bp->b_bcount <= DEV_BSIZE)
(*pr)("%" PRIdaddr, sn);
else {
if (blkdone >= 0) {
sn += blkdone;
(*pr)("%" PRIdaddr " of ", sn);
}
(*pr)("%" PRIdaddr "-%" PRIdaddr "", bp->b_blkno,
bp->b_blkno + (bp->b_bcount - 1) / DEV_BSIZE);
}
if (lp && (blkdone >= 0 || bp->b_bcount <= lp->d_secsize)) {
sn += lp->d_partitions[part].p_offset;
(*pr)(" (%s%d bn %" PRIdaddr "; cn %" PRIdaddr "",
dname, unit, sn, sn / lp->d_secpercyl);
sn %= lp->d_secpercyl;
(*pr)(" tn %" PRIdaddr " sn %" PRIdaddr ")",
sn / lp->d_nsectors, sn % lp->d_nsectors);
}
}
/*
* Initialize the disklist. Called by main() before autoconfiguration.
*/
void
disk_init(void)
{
TAILQ_INIT(&disklist);
disk_count = 0;
}
/*
* Searches the disklist for the disk corresponding to the
* name provided.
*/
struct disk *
disk_find(char *name)
{
struct disk *diskp;
if ((name == NULL) || (disk_count <= 0))
return (NULL);
simple_lock(&disklist_slock);
for (diskp = TAILQ_FIRST(&disklist); diskp != NULL;
diskp = TAILQ_NEXT(diskp, dk_link))
if (strcmp(diskp->dk_name, name) == 0) {
simple_unlock(&disklist_slock);
return (diskp);
}
simple_unlock(&disklist_slock);
return (NULL);
}
/*
* Attach a disk.
*/
void
disk_attach(struct disk *diskp)
{
int s;
/*
* Allocate and initialize the disklabel structures. Note that
* it's not safe to sleep here, since we're probably going to be
* called during autoconfiguration.
*/
diskp->dk_label = malloc(sizeof(struct disklabel), M_DEVBUF, M_NOWAIT);
diskp->dk_cpulabel = malloc(sizeof(struct cpu_disklabel), M_DEVBUF,
M_NOWAIT);
if ((diskp->dk_label == NULL) || (diskp->dk_cpulabel == NULL))
panic("disk_attach: can't allocate storage for disklabel");
memset(diskp->dk_label, 0, sizeof(struct disklabel));
memset(diskp->dk_cpulabel, 0, sizeof(struct cpu_disklabel));
/*
* Set the attached timestamp.
*/
s = splclock();
diskp->dk_attachtime = mono_time;
splx(s);
/*
* Link into the disklist.
*/
simple_lock(&disklist_slock);
TAILQ_INSERT_TAIL(&disklist, diskp, dk_link);
simple_unlock(&disklist_slock);
++disk_count;
}
/*
* Detach a disk.
*/
void
disk_detach(struct disk *diskp)
{
/*
* Remove from the disklist.
*/
if (--disk_count < 0)
panic("disk_detach: disk_count < 0");
simple_lock(&disklist_slock);
TAILQ_REMOVE(&disklist, diskp, dk_link);
simple_unlock(&disklist_slock);
/*
* Free the space used by the disklabel structures.
*/
free(diskp->dk_label, M_DEVBUF);
free(diskp->dk_cpulabel, M_DEVBUF);
}
/*
* Increment a disk's busy counter. If the counter is going from
* 0 to 1, set the timestamp.
*/
void
disk_busy(struct disk *diskp)
{
int s;
/*
* XXX We'd like to use something as accurate as microtime(),
* but that doesn't depend on the system TOD clock.
*/
if (diskp->dk_busy++ == 0) {
s = splclock();
diskp->dk_timestamp = mono_time;
splx(s);
}
}
/*
* Decrement a disk's busy counter, increment the byte count, total busy
* time, and reset the timestamp.
*/
void
disk_unbusy(struct disk *diskp, long bcount, int read)
{
int s;
struct timeval dv_time, diff_time;
if (diskp->dk_busy-- == 0) {
printf("%s: dk_busy < 0\n", diskp->dk_name);
panic("disk_unbusy");
}
s = splclock();
dv_time = mono_time;
splx(s);
timersub(&dv_time, &diskp->dk_timestamp, &diff_time);
timeradd(&diskp->dk_time, &diff_time, &diskp->dk_time);
diskp->dk_timestamp = dv_time;
if (bcount > 0) {
if (read) {
diskp->dk_rbytes += bcount;
diskp->dk_rxfer++;
} else {
diskp->dk_wbytes += bcount;
diskp->dk_wxfer++;
}
}
}
/*
* Reset the metrics counters on the given disk. Note that we cannot
* reset the busy counter, as it may case a panic in disk_unbusy().
* We also must avoid playing with the timestamp information, as it
* may skew any pending transfer results.
*/
void
disk_resetstat(struct disk *diskp)
{
int s = splbio(), t;
diskp->dk_rxfer = 0;
diskp->dk_rbytes = 0;
diskp->dk_wxfer = 0;
diskp->dk_wbytes = 0;
t = splclock();
diskp->dk_attachtime = mono_time;
splx(t);
timerclear(&diskp->dk_time);
splx(s);
}
int
sysctl_hw_disknames(SYSCTLFN_ARGS)
{
char buf[DK_DISKNAMELEN + 1];
char *where = oldp;
struct disk *diskp;
size_t needed, left, slen;
int error, first;
if (newp != NULL)
return (EPERM);
if (namelen != 0)
return (EINVAL);
first = 1;
error = 0;
needed = 0;
left = *oldlenp;
simple_lock(&disklist_slock);
for (diskp = TAILQ_FIRST(&disklist); diskp != NULL;
diskp = TAILQ_NEXT(diskp, dk_link)) {
if (where == NULL)
needed += strlen(diskp->dk_name) + 1;
else {
memset(buf, 0, sizeof(buf));
if (first) {
strncpy(buf, diskp->dk_name, sizeof(buf));
first = 0;
} else {
buf[0] = ' ';
strncpy(buf + 1, diskp->dk_name,
sizeof(buf) - 1);
}
buf[DK_DISKNAMELEN] = '\0';
slen = strlen(buf);
if (left < slen + 1)
break;
/* +1 to copy out the trailing NUL byte */
error = copyout(buf, where, slen + 1);
if (error)
break;
where += slen;
needed += slen;
left -= slen;
}
}
simple_unlock(&disklist_slock);
*oldlenp = needed;
return (error);
}
int
sysctl_hw_diskstats(SYSCTLFN_ARGS)
{
struct disk_sysctl sdisk;
struct disk *diskp;
char *where = oldp;
size_t tocopy, left;
int error;
if (newp != NULL)
return (EPERM);
/*
* The original hw.diskstats call was broken and did not require
* the userland to pass in it's size of struct disk_sysctl. This
* was fixed after NetBSD 1.6 was released, and any applications
* that do not pass in the size are given an error only, unless
* we care about 1.6 compatibility.
*/
if (namelen == 0)
#ifdef COMPAT_16
tocopy = offsetof(struct disk_sysctl, dk_rxfer);
#else
return (EINVAL);
#endif
else
tocopy = name[0];
if (where == NULL) {
*oldlenp = disk_count * tocopy;
return (0);
}
error = 0;
left = *oldlenp;
memset(&sdisk, 0, sizeof(sdisk));
*oldlenp = 0;
simple_lock(&disklist_slock);
TAILQ_FOREACH(diskp, &disklist, dk_link) {
if (left < tocopy)
break;
strncpy(sdisk.dk_name, diskp->dk_name, sizeof(sdisk.dk_name));
sdisk.dk_xfer = diskp->dk_rxfer + diskp->dk_wxfer;
sdisk.dk_rxfer = diskp->dk_rxfer;
sdisk.dk_wxfer = diskp->dk_wxfer;
sdisk.dk_seek = diskp->dk_seek;
sdisk.dk_bytes = diskp->dk_rbytes + diskp->dk_wbytes;
sdisk.dk_rbytes = diskp->dk_rbytes;
sdisk.dk_wbytes = diskp->dk_wbytes;
sdisk.dk_attachtime_sec = diskp->dk_attachtime.tv_sec;
sdisk.dk_attachtime_usec = diskp->dk_attachtime.tv_usec;
sdisk.dk_timestamp_sec = diskp->dk_timestamp.tv_sec;
sdisk.dk_timestamp_usec = diskp->dk_timestamp.tv_usec;
sdisk.dk_time_sec = diskp->dk_time.tv_sec;
sdisk.dk_time_usec = diskp->dk_time.tv_usec;
sdisk.dk_busy = diskp->dk_busy;
error = copyout(&sdisk, where, min(tocopy, sizeof(sdisk)));
if (error)
break;
where += tocopy;
*oldlenp += tocopy;
left -= tocopy;
}
simple_unlock(&disklist_slock);
return (error);
}
struct bufq_fcfs {
TAILQ_HEAD(, buf) bq_head; /* actual list of buffers */
};
struct bufq_disksort {
TAILQ_HEAD(, buf) bq_head; /* actual list of buffers */
};
#define PRIO_READ_BURST 48
#define PRIO_WRITE_REQ 16
struct bufq_prio {
TAILQ_HEAD(, buf) bq_read, bq_write; /* actual list of buffers */
struct buf *bq_write_next; /* next request in bq_write */
struct buf *bq_next; /* current request */
int bq_read_burst; /* # of consecutive reads */
};
/*
* Check if two buf's are in ascending order.
*/
static __inline int
buf_inorder(struct buf *bp, struct buf *bq, int sortby)
{
if (bp == NULL || bq == NULL)
return (bq == NULL);
if (sortby == BUFQ_SORT_CYLINDER) {
if (bp->b_cylinder != bq->b_cylinder)
return bp->b_cylinder < bq->b_cylinder;
else
return bp->b_rawblkno < bq->b_rawblkno;
} else
return bp->b_rawblkno < bq->b_rawblkno;
}
/*
* First-come first-served sort for disks.
*
* Requests are appended to the queue without any reordering.
*/
static void
bufq_fcfs_put(struct bufq_state *bufq, struct buf *bp)
{
struct bufq_fcfs *fcfs = bufq->bq_private;
TAILQ_INSERT_TAIL(&fcfs->bq_head, bp, b_actq);
}
static struct buf *
bufq_fcfs_get(struct bufq_state *bufq, int remove)
{
struct bufq_fcfs *fcfs = bufq->bq_private;
struct buf *bp;
bp = TAILQ_FIRST(&fcfs->bq_head);
if (bp != NULL && remove)
TAILQ_REMOVE(&fcfs->bq_head, bp, b_actq);
return (bp);
}
/*
* Seek sort for disks.
*
* There are actually two queues, sorted in ascendening order. The first
* queue holds those requests which are positioned after the current block;
* the second holds requests which came in after their position was passed.
* Thus we implement a one-way scan, retracting after reaching the end of
* the drive to the first request on the second queue, at which time it
* becomes the first queue.
*
* A one-way scan is natural because of the way UNIX read-ahead blocks are
* allocated.
*/
static void
bufq_disksort_put(struct bufq_state *bufq, struct buf *bp)
{
struct bufq_disksort *disksort = bufq->bq_private;
struct buf *bq, *nbq;
int sortby;
sortby = bufq->bq_flags & BUFQ_SORT_MASK;
bq = TAILQ_FIRST(&disksort->bq_head);
/*
* If the queue is empty it's easy; we just go on the end.
*/
if (bq == NULL) {
TAILQ_INSERT_TAIL(&disksort->bq_head, bp, b_actq);
return;
}
/*
* If we lie before the currently active request, then we
* must locate the second request list and add ourselves to it.
*/
if (buf_inorder(bp, bq, sortby)) {
while ((nbq = TAILQ_NEXT(bq, b_actq)) != NULL) {
/*
* Check for an ``inversion'' in the normally ascending
* block numbers, indicating the start of the second
* request list.
*/
if (buf_inorder(nbq, bq, sortby)) {
/*
* Search the second request list for the first
* request at a larger block number. We go
* after that; if there is no such request, we
* go at the end.
*/
do {
if (buf_inorder(bp, nbq, sortby))
goto insert;
bq = nbq;
} while ((nbq =
TAILQ_NEXT(bq, b_actq)) != NULL);
goto insert; /* after last */
}
bq = nbq;
}
/*
* No inversions... we will go after the last, and
* be the first request in the second request list.
*/
goto insert;
}
/*
* Request is at/after the current request...
* sort in the first request list.
*/
while ((nbq = TAILQ_NEXT(bq, b_actq)) != NULL) {
/*
* We want to go after the current request if there is an
* inversion after it (i.e. it is the end of the first
* request list), or if the next request is a larger cylinder
* than our request.
*/
if (buf_inorder(nbq, bq, sortby) ||
buf_inorder(bp, nbq, sortby))
goto insert;
bq = nbq;
}
/*
* Neither a second list nor a larger request... we go at the end of
* the first list, which is the same as the end of the whole schebang.
*/
insert: TAILQ_INSERT_AFTER(&disksort->bq_head, bq, bp, b_actq);
}
static struct buf *
bufq_disksort_get(struct bufq_state *bufq, int remove)
{
struct bufq_disksort *disksort = bufq->bq_private;
struct buf *bp;
bp = TAILQ_FIRST(&disksort->bq_head);
if (bp != NULL && remove)
TAILQ_REMOVE(&disksort->bq_head, bp, b_actq);
return (bp);
}
/*
* Seek sort for disks.
*
* There are two queues. The first queue holds read requests; the second
* holds write requests. The read queue is first-come first-served; the
* write queue is sorted in ascendening block order.
* The read queue is processed first. After PRIO_READ_BURST consecutive
* read requests with non-empty write queue PRIO_WRITE_REQ requests from
* the write queue will be processed.
*/
static void
bufq_prio_put(struct bufq_state *bufq, struct buf *bp)
{
struct bufq_prio *prio = bufq->bq_private;
struct buf *bq;
int sortby;
sortby = bufq->bq_flags & BUFQ_SORT_MASK;
/*
* If it's a read request append it to the list.
*/
if ((bp->b_flags & B_READ) == B_READ) {
TAILQ_INSERT_TAIL(&prio->bq_read, bp, b_actq);
return;
}
bq = TAILQ_FIRST(&prio->bq_write);
/*
* If the write list is empty, simply append it to the list.
*/
if (bq == NULL) {
TAILQ_INSERT_TAIL(&prio->bq_write, bp, b_actq);
prio->bq_write_next = bp;
return;
}
/*
* If we lie after the next request, insert after this request.
*/
if (buf_inorder(prio->bq_write_next, bp, sortby))
bq = prio->bq_write_next;
/*
* Search for the first request at a larger block number.
* We go before this request if it exists.
*/
while (bq != NULL && buf_inorder(bq, bp, sortby))
bq = TAILQ_NEXT(bq, b_actq);
if (bq != NULL)
TAILQ_INSERT_BEFORE(bq, bp, b_actq);
else
TAILQ_INSERT_TAIL(&prio->bq_write, bp, b_actq);
}
static struct buf *
bufq_prio_get(struct bufq_state *bufq, int remove)
{
struct bufq_prio *prio = bufq->bq_private;
struct buf *bp;
/*
* If no current request, get next from the lists.
*/
if (prio->bq_next == NULL) {
/*
* If at least one list is empty, select the other.
*/
if (TAILQ_FIRST(&prio->bq_read) == NULL) {
prio->bq_next = prio->bq_write_next;
prio->bq_read_burst = 0;
} else if (prio->bq_write_next == NULL) {
prio->bq_next = TAILQ_FIRST(&prio->bq_read);
prio->bq_read_burst = 0;
} else {
/*
* Both list have requests. Select the read list up
* to PRIO_READ_BURST times, then select the write
* list PRIO_WRITE_REQ times.
*/
if (prio->bq_read_burst++ < PRIO_READ_BURST)
prio->bq_next = TAILQ_FIRST(&prio->bq_read);
else if (prio->bq_read_burst <
PRIO_READ_BURST + PRIO_WRITE_REQ)
prio->bq_next = prio->bq_write_next;
else {
prio->bq_next = TAILQ_FIRST(&prio->bq_read);
prio->bq_read_burst = 0;
}
}
}
bp = prio->bq_next;
if (bp != NULL && remove) {
if ((bp->b_flags & B_READ) == B_READ)
TAILQ_REMOVE(&prio->bq_read, bp, b_actq);
else {
/*
* Advance the write pointer before removing
* bp since it is actually prio->bq_write_next.
*/
prio->bq_write_next =
TAILQ_NEXT(prio->bq_write_next, b_actq);
TAILQ_REMOVE(&prio->bq_write, bp, b_actq);
if (prio->bq_write_next == NULL)
prio->bq_write_next =
TAILQ_FIRST(&prio->bq_write);
}
prio->bq_next = NULL;
}
return (bp);
}
/*
* Cyclical scan (CSCAN)
*/
TAILQ_HEAD(bqhead, buf);
struct cscan_queue {
struct bqhead cq_head[2]; /* actual lists of buffers */
int cq_idx; /* current list index */
int cq_lastcylinder; /* b_cylinder of the last request */
daddr_t cq_lastrawblkno; /* b_rawblkno of the last request */
};
static int __inline cscan_empty(const struct cscan_queue *);
static void cscan_put(struct cscan_queue *, struct buf *, int);
static struct buf *cscan_get(struct cscan_queue *, int);
static void cscan_init(struct cscan_queue *);
static __inline int
cscan_empty(const struct cscan_queue *q)
{
return TAILQ_EMPTY(&q->cq_head[0]) && TAILQ_EMPTY(&q->cq_head[1]);
}
static void
cscan_put(struct cscan_queue *q, struct buf *bp, int sortby)
{
struct buf tmp;
struct buf *it;
struct bqhead *bqh;
int idx;
tmp.b_cylinder = q->cq_lastcylinder;
tmp.b_rawblkno = q->cq_lastrawblkno;
if (buf_inorder(bp, &tmp, sortby))
idx = 1 - q->cq_idx;
else
idx = q->cq_idx;
bqh = &q->cq_head[idx];
TAILQ_FOREACH(it, bqh, b_actq)
if (buf_inorder(bp, it, sortby))
break;
if (it != NULL)
TAILQ_INSERT_BEFORE(it, bp, b_actq);
else
TAILQ_INSERT_TAIL(bqh, bp, b_actq);
}
static struct buf *
cscan_get(struct cscan_queue *q, int remove)
{
int idx = q->cq_idx;
struct bqhead *bqh;
struct buf *bp;
bqh = &q->cq_head[idx];
bp = TAILQ_FIRST(bqh);
if (bp == NULL) {
/* switch queue */
idx = 1 - idx;
bqh = &q->cq_head[idx];
bp = TAILQ_FIRST(bqh);
}
KDASSERT((bp != NULL && !cscan_empty(q)) ||
(bp == NULL && cscan_empty(q)));
if (bp != NULL && remove) {
q->cq_idx = idx;
TAILQ_REMOVE(bqh, bp, b_actq);
q->cq_lastcylinder = bp->b_cylinder;
q->cq_lastrawblkno =
bp->b_rawblkno + (bp->b_bcount >> DEV_BSHIFT);
}
return (bp);
}
static void
cscan_init(struct cscan_queue *q)
{
TAILQ_INIT(&q->cq_head[0]);
TAILQ_INIT(&q->cq_head[1]);
}
/*
* Per-prioritiy CSCAN.
*
* XXX probably we should have a way to raise
* priority of the on-queue requests.
*/
#define PRIOCSCAN_NQUEUE 3
struct priocscan_queue {
struct cscan_queue q_queue;
int q_burst;
};
struct bufq_priocscan {
struct priocscan_queue bq_queue[PRIOCSCAN_NQUEUE];
#if 0
/*
* XXX using "global" head position can reduce positioning time
* when switching between queues.
* although it might affect against fairness.
*/
daddr_t bq_lastrawblkno;
int bq_lastcylinder;
#endif
};
/*
* how many requests to serve when having pending requests on other queues.
*
* XXX tune
*/
const int priocscan_burst[] = {
64, 16, 4
};
static void bufq_priocscan_put(struct bufq_state *, struct buf *);
static struct buf *bufq_priocscan_get(struct bufq_state *, int);
static void bufq_priocscan_init(struct bufq_state *);
static __inline struct cscan_queue *bufq_priocscan_selectqueue(
struct bufq_priocscan *, const struct buf *);
static __inline struct cscan_queue *
bufq_priocscan_selectqueue(struct bufq_priocscan *q, const struct buf *bp)
{
static const int priocscan_priomap[] = {
[BPRIO_TIMENONCRITICAL] = 2,
[BPRIO_TIMELIMITED] = 1,
[BPRIO_TIMECRITICAL] = 0
};
return &q->bq_queue[priocscan_priomap[BIO_GETPRIO(bp)]].q_queue;
}
static void
bufq_priocscan_put(struct bufq_state *bufq, struct buf *bp)
{
struct bufq_priocscan *q = bufq->bq_private;
struct cscan_queue *cq;
const int sortby = bufq->bq_flags & BUFQ_SORT_MASK;
cq = bufq_priocscan_selectqueue(q, bp);
cscan_put(cq, bp, sortby);
}
static struct buf *
bufq_priocscan_get(struct bufq_state *bufq, int remove)
{
struct bufq_priocscan *q = bufq->bq_private;
struct priocscan_queue *pq, *npq;
struct priocscan_queue *first; /* first non-empty queue */
const struct priocscan_queue *epq;
const struct cscan_queue *cq;
struct buf *bp;
boolean_t single; /* true if there's only one non-empty queue */
pq = &q->bq_queue[0];
epq = pq + PRIOCSCAN_NQUEUE;
for (; pq < epq; pq++) {
cq = &pq->q_queue;
if (!cscan_empty(cq))
break;
}
if (pq == epq) {
/* there's no requests */
return NULL;
}
first = pq;
single = TRUE;
for (npq = first + 1; npq < epq; npq++) {
cq = &npq->q_queue;
if (!cscan_empty(cq)) {
single = FALSE;
if (pq->q_burst > 0)
break;
pq = npq;
}
}
if (single) {
/*
* there's only a non-empty queue. just serve it.
*/
pq = first;
} else if (pq->q_burst > 0) {
/*
* XXX account only by number of requests. is it good enough?
*/
pq->q_burst--;
} else {
/*
* no queue was selected due to burst counts
*/
int i;
#ifdef DEBUG
for (i = 0; i < PRIOCSCAN_NQUEUE; i++) {
pq = &q->bq_queue[i];
cq = &pq->q_queue;
if (!cscan_empty(cq) && pq->q_burst)
panic("%s: inconsist", __func__);
}
#endif /* DEBUG */
/*
* reset burst counts
*/
for (i = 0; i < PRIOCSCAN_NQUEUE; i++) {
pq = &q->bq_queue[i];
pq->q_burst = priocscan_burst[i];
}
/*
* serve first non-empty queue.
*/
pq = first;
}
KDASSERT(!cscan_empty(&pq->q_queue));
bp = cscan_get(&pq->q_queue, remove);
KDASSERT(bp != NULL);
KDASSERT(&pq->q_queue == bufq_priocscan_selectqueue(q, bp));
return bp;
}
static void
bufq_priocscan_init(struct bufq_state *bufq)
{
struct bufq_priocscan *q;
int i;
bufq->bq_get = bufq_priocscan_get;
bufq->bq_put = bufq_priocscan_put;
bufq->bq_private = malloc(sizeof(struct bufq_priocscan),
M_DEVBUF, M_ZERO);
q = bufq->bq_private;
for (i = 0; i < PRIOCSCAN_NQUEUE; i++) {
struct cscan_queue *cq = &q->bq_queue[i].q_queue;
cscan_init(cq);
}
}
/*
* Create a device buffer queue.
*/
void
bufq_alloc(struct bufq_state *bufq, int flags)
{
struct bufq_fcfs *fcfs;
struct bufq_disksort *disksort;
struct bufq_prio *prio;
bufq->bq_flags = flags;
switch (flags & BUFQ_SORT_MASK) {
case BUFQ_SORT_RAWBLOCK:
case BUFQ_SORT_CYLINDER:
break;
case 0:
if ((flags & BUFQ_METHOD_MASK) == BUFQ_FCFS)
break;
/* FALLTHROUGH */
default:
panic("bufq_alloc: sort out of range");
}
switch (flags & BUFQ_METHOD_MASK) {
case BUFQ_FCFS:
bufq->bq_get = bufq_fcfs_get;
bufq->bq_put = bufq_fcfs_put;
MALLOC(bufq->bq_private, struct bufq_fcfs *,
sizeof(struct bufq_fcfs), M_DEVBUF, M_ZERO);
fcfs = (struct bufq_fcfs *)bufq->bq_private;
TAILQ_INIT(&fcfs->bq_head);
break;
case BUFQ_DISKSORT:
bufq->bq_get = bufq_disksort_get;
bufq->bq_put = bufq_disksort_put;
MALLOC(bufq->bq_private, struct bufq_disksort *,
sizeof(struct bufq_disksort), M_DEVBUF, M_ZERO);
disksort = (struct bufq_disksort *)bufq->bq_private;
TAILQ_INIT(&disksort->bq_head);
break;
case BUFQ_READ_PRIO:
bufq->bq_get = bufq_prio_get;
bufq->bq_put = bufq_prio_put;
MALLOC(bufq->bq_private, struct bufq_prio *,
sizeof(struct bufq_prio), M_DEVBUF, M_ZERO);
prio = (struct bufq_prio *)bufq->bq_private;
TAILQ_INIT(&prio->bq_read);
TAILQ_INIT(&prio->bq_write);
break;
case BUFQ_PRIOCSCAN:
bufq_priocscan_init(bufq);
break;
default:
panic("bufq_alloc: method out of range");
}
}
/*
* Destroy a device buffer queue.
*/
void
bufq_free(struct bufq_state *bufq)
{
KASSERT(bufq->bq_private != NULL);
KASSERT(BUFQ_PEEK(bufq) == NULL);
FREE(bufq->bq_private, M_DEVBUF);
bufq->bq_get = NULL;
bufq->bq_put = NULL;
}
/*
* Bounds checking against the media size, used for the raw partition.
* The sector size passed in should currently always be DEV_BSIZE,
* and the media size the size of the device in DEV_BSIZE sectors.
*/
int
bounds_check_with_mediasize(struct buf *bp, int secsize, u_int64_t mediasize)
{
int sz;
sz = howmany(bp->b_bcount, secsize);
if (bp->b_blkno + sz > mediasize) {
sz = mediasize - bp->b_blkno;
if (sz == 0) {
/* If exactly at end of disk, return EOF. */
bp->b_resid = bp->b_bcount;
goto done;
}
if (sz < 0) {
/* If past end of disk, return EINVAL. */
bp->b_error = EINVAL;
goto bad;
}
/* Otherwise, truncate request. */
bp->b_bcount = sz << DEV_BSHIFT;
}
return 1;
bad:
bp->b_flags |= B_ERROR;
done:
return 0;
}
|