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
|
/*-
* Copyright (c) 2002 Poul-Henning Kamp
* Copyright (c) 2002 Networks Associates Technology, Inc.
* All rights reserved.
*
* This software was developed for the FreeBSD Project by Poul-Henning Kamp
* and NAI Labs, the Security Research Division of Network Associates, Inc.
* under DARPA/SPAWAR contract N66001-01-C-8035 ("CBOSS"), as part of the
* DARPA CHATS research program.
*
* 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.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR 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 AUTHOR 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.
*
* $FreeBSD$
*
* This source file contains the state-engine which makes things happen in the
* right order.
*
* Outline:
* 1) g_bde_start1()
* Break the struct bio into multiple work packets one per zone.
* 2) g_bde_start2()
* Setup the necessary sector buffers and start those read operations
* which we can start at this time and put the item on the work-list.
* 3) g_bde_worker()
* Scan the work-list for items which are ready for crypto processing
* and call the matching crypto function in g_bde_crypt.c and schedule
* any writes needed. Read operations finish here by releasing the
* sector buffers and delivering the original bio request.
* 4) g_bde_write_done()
* Release sector buffers and deliver the original bio request.
*
* Because of the C-scope rules, the functions are almost perfectly in the
* opposite order in this source file.
*
* XXX: A switch to the hardware assisted crypto in src/sys/opencrypto will add
* XXX: additional states to this state-engine. Since no hardware available
* XXX: at this time has AES support, implementing this has been postponed
* XXX: until such time as it would result in a benefit.
*/
#include <sys/param.h>
#include <sys/bio.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/queue.h>
#include <sys/malloc.h>
#include <sys/systm.h>
#include <sys/kernel.h>
#include <sys/sysctl.h>
#include <sys/proc.h>
#include <sys/kthread.h>
#include <crypto/rijndael/rijndael.h>
#include <crypto/sha2/sha2.h>
#include <geom/geom.h>
#include <geom/bde/g_bde.h>
static void g_bde_delete_sector(struct g_bde_softc *wp, struct g_bde_sector *sp);
static struct g_bde_sector * g_bde_new_sector(struct g_bde_work *wp, u_int len);
static void g_bde_release_sector(struct g_bde_work *wp, struct g_bde_sector *sp);
static struct g_bde_sector *g_bde_get_sector(struct g_bde_work *wp, off_t offset);
static int g_bde_start_read(struct g_bde_sector *sp);
static void g_bde_purge_sector(struct g_bde_softc *sc, int fraction);
/*
* Work item allocation.
*
* C++ would call these constructors and destructors.
*/
static u_int g_bde_nwork;
SYSCTL_UINT(_debug, OID_AUTO, gbde_nwork, CTLFLAG_RD, &g_bde_nwork, 0, "");
static struct g_bde_work *
g_bde_new_work(struct g_bde_softc *sc)
{
struct g_bde_work *wp;
wp = g_malloc(sizeof *wp, M_NOWAIT | M_ZERO);
if (wp == NULL)
return (wp);
wp->state = SETUP;
wp->softc = sc;
g_bde_nwork++;
sc->nwork++;
TAILQ_INSERT_TAIL(&sc->worklist, wp, list);
return (wp);
}
static void
g_bde_delete_work(struct g_bde_work *wp)
{
struct g_bde_softc *sc;
sc = wp->softc;
g_bde_nwork--;
sc->nwork--;
TAILQ_REMOVE(&sc->worklist, wp, list);
g_free(wp);
}
/*
* Sector buffer allocation
*
* These two functions allocate and free back variable sized sector buffers
*/
static u_int g_bde_nsect;
SYSCTL_UINT(_debug, OID_AUTO, gbde_nsect, CTLFLAG_RD, &g_bde_nsect, 0, "");
static void
g_bde_delete_sector(struct g_bde_softc *sc, struct g_bde_sector *sp)
{
g_bde_nsect--;
sc->nsect--;
if (sp->malloc)
g_free(sp->data);
g_free(sp);
}
static struct g_bde_sector *
g_bde_new_sector(struct g_bde_work *wp, u_int len)
{
struct g_bde_sector *sp;
sp = g_malloc(sizeof *sp, M_NOWAIT | M_ZERO);
if (sp == NULL)
return (sp);
if (len > 0) {
sp->data = g_malloc(len, M_NOWAIT | M_ZERO);
if (sp->data == NULL) {
g_free(sp);
return (NULL);
}
sp->malloc = 1;
}
g_bde_nsect++;
wp->softc->nsect++;
sp->size = len;
sp->softc = wp->softc;
sp->ref = 1;
sp->owner = wp;
sp->offset = wp->so;
sp->state = JUNK;
return (sp);
}
/*
* Skey sector cache.
*
* Nothing prevents two separate I/O requests from addressing the same zone
* and thereby needing the same skey sector. We therefore need to sequence
* I/O operations to the skey sectors. A certain amount of caching is also
* desirable, although the extent of benefit from this is not at this point
* determined.
*
* XXX: GEOM may be able to grow a generic caching facility at some point
* XXX: to support such needs.
*/
static u_int g_bde_ncache;
SYSCTL_UINT(_debug, OID_AUTO, gbde_ncache, CTLFLAG_RD, &g_bde_ncache, 0, "");
static void
g_bde_purge_one_sector(struct g_bde_softc *sc, struct g_bde_sector *sp)
{
g_trace(G_T_TOPOLOGY, "g_bde_purge_one_sector(%p, %p)", sc, sp);
if (sp->ref != 0)
return;
TAILQ_REMOVE(&sc->freelist, sp, list);
g_bde_ncache--;
sc->ncache--;
bzero(sp->data, sp->size);
g_bde_delete_sector(sc, sp);
}
static struct g_bde_sector *
g_bde_get_sector(struct g_bde_work *wp, off_t offset)
{
struct g_bde_sector *sp;
struct g_bde_softc *sc;
g_trace(G_T_TOPOLOGY, "g_bde_get_sector(%p, %jd)", wp, (intmax_t)offset);
sc = wp->softc;
if (malloc_last_fail() < g_bde_ncache)
g_bde_purge_sector(sc, -1);
sp = TAILQ_FIRST(&sc->freelist);
if (sp != NULL && sp->ref == 0 && sp->used + 300 < time_uptime)
g_bde_purge_one_sector(sc, sp);
TAILQ_FOREACH(sp, &sc->freelist, list) {
if (sp->offset == offset)
break;
}
if (sp != NULL) {
sp->ref++;
KASSERT(sp->offset == offset, ("wrong offset"));
KASSERT(sp->softc == wp->softc, ("wrong softc"));
if (sp->ref == 1)
sp->owner = wp;
} else {
if (malloc_last_fail() < g_bde_ncache) {
TAILQ_FOREACH(sp, &sc->freelist, list)
if (sp->ref == 0)
break;
}
if (sp == NULL && !TAILQ_EMPTY(&sc->freelist))
sp = TAILQ_FIRST(&sc->freelist);
if (sp != NULL && sp->ref > 0)
sp = NULL;
if (sp == NULL) {
g_bde_ncache++;
sc->ncache++;
sp = g_bde_new_sector(wp, sc->sectorsize);
if (sp != NULL) {
TAILQ_INSERT_TAIL(&sc->freelist, sp, list);
sp->malloc = 2;
}
}
if (sp != NULL) {
sp->offset = offset;
sp->softc = wp->softc;
sp->ref = 1;
sp->owner = wp;
sp->state = JUNK;
sp->error = 0;
}
}
if (sp != NULL) {
TAILQ_REMOVE(&sc->freelist, sp, list);
TAILQ_INSERT_TAIL(&sc->freelist, sp, list);
}
wp->ksp = sp;
if (sp == NULL) {
g_bde_purge_sector(sc, -1);
sp = g_bde_get_sector(wp, offset);
}
if (sp != NULL)
sp->used = time_uptime;
KASSERT(sp != NULL, ("get_sector failed"));
return(sp);
}
static void
g_bde_release_sector(struct g_bde_work *wp, struct g_bde_sector *sp)
{
struct g_bde_softc *sc;
struct g_bde_work *wp2;
g_trace(G_T_TOPOLOGY, "g_bde_release_sector(%p)", sp);
KASSERT(sp->malloc == 2, ("Wrong sector released"));
sc = sp->softc;
KASSERT(sc != NULL, ("NULL sp->softc"));
KASSERT(wp == sp->owner, ("Releasing, not owner"));
sp->owner = NULL;
wp->ksp = NULL;
sp->ref--;
if (sp->ref > 0) {
TAILQ_REMOVE(&sc->freelist, sp, list);
TAILQ_INSERT_TAIL(&sc->freelist, sp, list);
TAILQ_FOREACH(wp2, &sc->worklist, list) {
if (wp2->ksp == sp) {
KASSERT(wp2 != wp, ("Self-reowning"));
sp->owner = wp2;
wakeup(sp->softc);
break;
}
}
KASSERT(wp2 != NULL, ("Failed to pick up owner for %p\n", sp));
} else if (sp->error != 0) {
sp->offset = ~0;
sp->error = 0;
sp->state = JUNK;
}
TAILQ_REMOVE(&sc->freelist, sp, list);
TAILQ_INSERT_HEAD(&sc->freelist, sp, list);
}
static void
g_bde_purge_sector(struct g_bde_softc *sc, int fraction)
{
struct g_bde_sector *sp;
int n;
g_trace(G_T_TOPOLOGY, "g_bde_purge_sector(%p)", sc);
if (fraction > 0)
n = sc->ncache / fraction + 1;
else
n = g_bde_ncache - malloc_last_fail();
if (n < 0)
return;
if (n > sc->ncache)
n = sc->ncache;
while(n--) {
TAILQ_FOREACH(sp, &sc->freelist, list) {
if (sp->ref != 0)
continue;
TAILQ_REMOVE(&sc->freelist, sp, list);
g_bde_ncache--;
sc->ncache--;
bzero(sp->data, sp->size);
g_bde_delete_sector(sc, sp);
break;
}
}
}
static struct g_bde_sector *
g_bde_read_sector(struct g_bde_softc *sc, struct g_bde_work *wp, off_t offset)
{
struct g_bde_sector *sp;
g_trace(G_T_TOPOLOGY, "g_bde_read_sector(%p)", wp);
sp = g_bde_get_sector(wp, offset);
if (sp == NULL)
return (sp);
if (sp->owner != wp)
return (sp);
if (sp->state == VALID)
return (sp);
if (g_bde_start_read(sp) == 0)
return (sp);
g_bde_release_sector(wp, sp);
return (NULL);
}
/*
* Contribute to the completion of the original bio request.
*
* We have no simple way to tell how many bits the original bio request has
* been segmented into, so the easiest way to determine when we can deliver
* it is to keep track of the number of bytes we have completed. We keep
* track of any errors underway and latch onto the first one.
*
* We always report "nothing done" in case of error, because random bits here
* and there may be completed and returning a number of completed bytes does
* not convey any useful information about which bytes they were. If some
* piece of broken code somewhere interprets this to mean that nothing has
* changed on the underlying media they deserve the lossage headed for them.
*
* A single mutex per g_bde instance is used to prevent contention.
*/
static void
g_bde_contribute(struct bio *bp, off_t bytes, int error)
{
struct g_bde_softc *sc;
g_trace(G_T_TOPOLOGY, "g_bde_contribute bp %p bytes %jd error %d",
bp, (intmax_t)bytes, error);
sc = bp->bio_driver1;
if (bp->bio_error == 0)
bp->bio_error = error;
bp->bio_completed += bytes;
KASSERT(bp->bio_completed <= bp->bio_length, ("Too large contribution"));
if (bp->bio_completed == bp->bio_length) {
if (bp->bio_error != 0)
bp->bio_completed = 0;
g_io_deliver(bp, bp->bio_error);
}
}
/*
* A write operation has finished. When we have all expected cows in the
* barn close the door and call it a day.
*/
static void
g_bde_write_done(struct bio *bp)
{
struct g_bde_sector *sp;
struct g_bde_work *wp;
struct g_bde_softc *sc;
sp = bp->bio_caller1;
sc = bp->bio_caller2;
mtx_lock(&sc->worklist_mutex);
KASSERT(sp != NULL, ("NULL sp"));
KASSERT(sc != NULL, ("NULL sc"));
KASSERT(sp->owner != NULL, ("NULL sp->owner"));
g_trace(G_T_TOPOLOGY, "g_bde_write_done(%p)", sp);
sp->error = bp->bio_error;
g_destroy_bio(bp);
wp = sp->owner;
if (wp->error == 0)
wp->error = sp->error;
if (wp->bp->bio_cmd == BIO_DELETE) {
KASSERT(sp == wp->sp, ("trashed delete op"));
g_bde_contribute(wp->bp, wp->length, wp->error);
g_bde_delete_sector(sc, sp);
g_bde_delete_work(wp);
mtx_unlock(&sc->worklist_mutex);
return;
}
KASSERT(wp->bp->bio_cmd == BIO_WRITE, ("Confused in g_bde_write_done()"));
KASSERT(sp == wp->sp || sp == wp->ksp, ("trashed write op"));
if (wp->sp == sp) {
g_bde_delete_sector(sc, wp->sp);
wp->sp = NULL;
} else {
sp->state = VALID;
}
if (wp->sp == NULL && wp->ksp != NULL && wp->ksp->state == VALID) {
g_bde_contribute(wp->bp, wp->length, wp->error);
g_bde_release_sector(wp, wp->ksp);
g_bde_delete_work(wp);
}
mtx_unlock(&sc->worklist_mutex);
return;
}
/*
* Send a write request for the given sector down the pipeline.
*/
static int
g_bde_start_write(struct g_bde_sector *sp)
{
struct bio *bp;
struct g_bde_softc *sc;
g_trace(G_T_TOPOLOGY, "g_bde_start_write(%p)", sp);
sc = sp->softc;
KASSERT(sc != NULL, ("NULL sc in g_bde_start_write"));
KASSERT(sp->owner != NULL, ("NULL sp->owner in g_bde_start_write"));
bp = g_new_bio();
if (bp == NULL)
return (ENOMEM);
bp->bio_cmd = BIO_WRITE;
bp->bio_offset = sp->offset;
bp->bio_data = sp->data;
bp->bio_length = sp->size;
bp->bio_done = g_bde_write_done;
bp->bio_caller1 = sp;
bp->bio_caller2 = sc;
sp->state = IO;
g_io_request(bp, sc->consumer);
return(0);
}
/*
* A read operation has finished. Mark the sector no longer iobusy and
* wake up the worker thread and let it do its thing.
*/
static void
g_bde_read_done(struct bio *bp)
{
struct g_bde_sector *sp;
struct g_bde_softc *sc;
sp = bp->bio_caller1;
g_trace(G_T_TOPOLOGY, "g_bde_read_done(%p)", sp);
sc = bp->bio_caller2;
mtx_lock(&sc->worklist_mutex);
sp->error = bp->bio_error;
sp->state = VALID;
wakeup(sc);
g_destroy_bio(bp);
mtx_unlock(&sc->worklist_mutex);
}
/*
* Send a read request for the given sector down the pipeline.
*/
static int
g_bde_start_read(struct g_bde_sector *sp)
{
struct bio *bp;
struct g_bde_softc *sc;
g_trace(G_T_TOPOLOGY, "g_bde_start_read(%p)", sp);
sc = sp->softc;
KASSERT(sc != NULL, ("Null softc in sp %p", sp));
bp = g_new_bio();
if (bp == NULL)
return (ENOMEM);
bp->bio_cmd = BIO_READ;
bp->bio_offset = sp->offset;
bp->bio_data = sp->data;
bp->bio_length = sp->size;
bp->bio_done = g_bde_read_done;
bp->bio_caller1 = sp;
bp->bio_caller2 = sc;
sp->state = IO;
g_io_request(bp, sc->consumer);
return(0);
}
/*
* The worker thread.
*
* The up/down path of GEOM is not allowed to sleep or do any major work
* so we use this thread to do the actual crypto operations and to push
* the state engine onwards.
*
* XXX: if we switch to the src/sys/opencrypt hardware assisted encryption
* XXX: using a thread here is probably not needed.
*/
void
g_bde_worker(void *arg)
{
struct g_bde_softc *sc;
struct g_bde_work *wp;
struct g_geom *gp;
int busy, error;
gp = arg;
sc = gp->softc;
mtx_lock(&sc->worklist_mutex);
for (;;) {
busy = 0;
g_trace(G_T_TOPOLOGY, "g_bde_worker scan");
TAILQ_FOREACH(wp, &sc->worklist, list) {
KASSERT(wp != NULL, ("NULL wp"));
KASSERT(wp->softc != NULL, ("NULL wp->softc"));
if (wp->state != WAIT)
continue; /* Not interesting here */
KASSERT(wp->bp != NULL, ("NULL wp->bp"));
KASSERT(wp->sp != NULL, ("NULL wp->sp"));
if (wp->ksp != NULL) {
if (wp->ksp->owner != wp)
continue;
if (wp->ksp->state == IO)
continue;
KASSERT(wp->ksp->state == VALID,
("Illegal sector state (JUNK ?)"));
}
if (wp->bp->bio_cmd == BIO_READ && wp->sp->state != VALID)
continue;
if (wp->ksp != NULL && wp->ksp->error != 0) {
g_bde_contribute(wp->bp, wp->length,
wp->ksp->error);
g_bde_delete_sector(sc, wp->sp);
g_bde_release_sector(wp, wp->ksp);
g_bde_delete_work(wp);
busy++;
break;
}
switch(wp->bp->bio_cmd) {
case BIO_READ:
if (wp->ksp != NULL && wp->sp->error == 0) {
mtx_unlock(&sc->worklist_mutex);
g_bde_crypt_read(wp);
mtx_lock(&sc->worklist_mutex);
}
g_bde_contribute(wp->bp, wp->length,
wp->sp->error);
g_bde_delete_sector(sc, wp->sp);
if (wp->ksp != NULL)
g_bde_release_sector(wp, wp->ksp);
g_bde_delete_work(wp);
break;
case BIO_WRITE:
wp->state = FINISH;
KASSERT(wp->sp->owner == wp, ("Write not owner sp"));
KASSERT(wp->ksp->owner == wp, ("Write not owner ksp"));
mtx_unlock(&sc->worklist_mutex);
g_bde_crypt_write(wp);
mtx_lock(&sc->worklist_mutex);
g_bde_start_write(wp->sp);
g_bde_start_write(wp->ksp);
break;
case BIO_DELETE:
wp->state = FINISH;
mtx_unlock(&sc->worklist_mutex);
g_bde_crypt_delete(wp);
mtx_lock(&sc->worklist_mutex);
g_bde_start_write(wp->sp);
break;
}
busy++;
break;
}
if (!busy) {
/*
* We don't look for our death-warrant until we are
* idle. Shouldn't make a difference in practice.
*/
if (sc->dead)
break;
g_trace(G_T_TOPOLOGY, "g_bde_worker sleep");
error = msleep(sc, &sc->worklist_mutex,
PRIBIO, "g_bde", hz);
if (error == EWOULDBLOCK) {
/*
* Loose our skey cache in an orderly fashion.
* The exact rate can be tuned to be less
* aggressive if this is desirable. 10% per
* second means that the cache is gone in a
* few minutes.
*/
g_bde_purge_sector(sc, 10);
}
}
}
g_trace(G_T_TOPOLOGY, "g_bde_worker die");
g_bde_purge_sector(sc, 1);
KASSERT(sc->nwork == 0, ("Dead but %d work remaining", sc->nwork));
KASSERT(sc->ncache == 0, ("Dead but %d cache remaining", sc->ncache));
KASSERT(sc->nsect == 0, ("Dead but %d sect remaining", sc->nsect));
mtx_unlock(&sc->worklist_mutex);
sc->dead = 2;
wakeup(sc);
mtx_lock(&Giant);
kthread_exit(0);
}
/*
* g_bde_start1 has chopped the incoming request up so all the requests
* we see here are inside a single zone. Map the data and key locations
* grab the buffers we need and fire off the first volley of read requests.
*/
static void
g_bde_start2(struct g_bde_work *wp)
{
struct g_bde_softc *sc;
KASSERT(wp != NULL, ("NULL wp in g_bde_start2"));
KASSERT(wp->softc != NULL, ("NULL wp->softc"));
g_trace(G_T_TOPOLOGY, "g_bde_start2(%p)", wp);
sc = wp->softc;
if (wp->bp->bio_cmd == BIO_READ) {
wp->sp = g_bde_new_sector(wp, 0);
if (wp->sp == NULL) {
g_bde_contribute(wp->bp, wp->length, ENOMEM);
g_bde_delete_work(wp);
return;
}
wp->sp->size = wp->length;
wp->sp->data = wp->data;
if (g_bde_start_read(wp->sp) != 0) {
g_bde_contribute(wp->bp, wp->length, ENOMEM);
g_bde_delete_sector(sc, wp->sp);
g_bde_delete_work(wp);
return;
}
g_bde_read_sector(sc, wp, wp->kso);
if (wp->ksp == NULL)
wp->error = ENOMEM;
} else if (wp->bp->bio_cmd == BIO_DELETE) {
wp->sp = g_bde_new_sector(wp, wp->length);
if (wp->sp == NULL) {
g_bde_contribute(wp->bp, wp->length, ENOMEM);
g_bde_delete_work(wp);
return;
}
} else if (wp->bp->bio_cmd == BIO_WRITE) {
wp->sp = g_bde_new_sector(wp, wp->length);
if (wp->sp == NULL) {
g_bde_contribute(wp->bp, wp->length, ENOMEM);
g_bde_delete_work(wp);
return;
}
g_bde_read_sector(sc, wp, wp->kso);
if (wp->ksp == NULL) {
g_bde_contribute(wp->bp, wp->length, ENOMEM);
g_bde_delete_sector(sc, wp->sp);
g_bde_delete_work(wp);
return;
}
} else {
KASSERT(0 == 1,
("Wrong bio_cmd %d in g_bde_start2", wp->bp->bio_cmd));
}
wp->state = WAIT;
wakeup(sc);
}
/*
* Create a sequence of work structures, and have g_bde_map_sector() determine
* how long they each can be. Feed them to g_bde_start2().
*/
void
g_bde_start1(struct bio *bp)
{
struct g_bde_softc *sc;
struct g_bde_work *wp;
off_t done;
sc = bp->bio_to->geom->softc;
bp->bio_driver1 = sc;
mtx_lock(&sc->worklist_mutex);
for(done = 0; done < bp->bio_length; ) {
wp = g_bde_new_work(sc);
if (wp == NULL) {
g_io_deliver(bp, ENOMEM);
mtx_unlock(&sc->worklist_mutex);
return;
}
wp->bp = bp;
wp->offset = bp->bio_offset + done;
wp->data = bp->bio_data + done;
wp->length = bp->bio_length - done;
g_bde_map_sector(wp);
done += wp->length;
g_bde_start2(wp);
}
mtx_unlock(&sc->worklist_mutex);
return;
}
|