summaryrefslogtreecommitdiffstats
path: root/drivers/block/aoe/aoedev.c
blob: 784c92e038d1461de3b2048fc263bf55a3e758a6 (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
/* Copyright (c) 2013 Coraid, Inc.  See COPYING for GPL terms. */
/*
 * aoedev.c
 * AoE device utility functions; maintains device list.
 */

#include <linux/hdreg.h>
#include <linux/blkdev.h>
#include <linux/netdevice.h>
#include <linux/delay.h>
#include <linux/slab.h>
#include <linux/bitmap.h>
#include <linux/kdev_t.h>
#include <linux/moduleparam.h>
#include "aoe.h"

static void dummy_timer(ulong);
static void freetgt(struct aoedev *d, struct aoetgt *t);
static void skbpoolfree(struct aoedev *d);

static int aoe_dyndevs = 1;
module_param(aoe_dyndevs, int, 0644);
MODULE_PARM_DESC(aoe_dyndevs, "Use dynamic minor numbers for devices.");

static struct aoedev *devlist;
static DEFINE_SPINLOCK(devlist_lock);

/* Because some systems will have one, many, or no
 *   - partitions,
 *   - slots per shelf,
 *   - or shelves,
 * we need some flexibility in the way the minor numbers
 * are allocated.  So they are dynamic.
 */
#define N_DEVS ((1U<<MINORBITS)/AOE_PARTITIONS)

static DEFINE_SPINLOCK(used_minors_lock);
static DECLARE_BITMAP(used_minors, N_DEVS);

static int
minor_get_dyn(ulong *sysminor)
{
	ulong flags;
	ulong n;
	int error = 0;

	spin_lock_irqsave(&used_minors_lock, flags);
	n = find_first_zero_bit(used_minors, N_DEVS);
	if (n < N_DEVS)
		set_bit(n, used_minors);
	else
		error = -1;
	spin_unlock_irqrestore(&used_minors_lock, flags);

	*sysminor = n * AOE_PARTITIONS;
	return error;
}

static int
minor_get_static(ulong *sysminor, ulong aoemaj, int aoemin)
{
	ulong flags;
	ulong n;
	int error = 0;
	enum {
		/* for backwards compatibility when !aoe_dyndevs,
		 * a static number of supported slots per shelf */
		NPERSHELF = 16,
	};

	if (aoemin >= NPERSHELF) {
		pr_err("aoe: %s %d slots per shelf\n",
			"static minor device numbers support only",
			NPERSHELF);
		error = -1;
		goto out;
	}

	n = aoemaj * NPERSHELF + aoemin;
	if (n >= N_DEVS) {
		pr_err("aoe: %s with e%ld.%d\n",
			"cannot use static minor device numbers",
			aoemaj, aoemin);
		error = -1;
		goto out;
	}

	spin_lock_irqsave(&used_minors_lock, flags);
	if (test_bit(n, used_minors)) {
		pr_err("aoe: %s %lu\n",
			"existing device already has static minor number",
			n);
		error = -1;
	} else
		set_bit(n, used_minors);
	spin_unlock_irqrestore(&used_minors_lock, flags);
	*sysminor = n * AOE_PARTITIONS;
out:
	return error;
}

static int
minor_get(ulong *sysminor, ulong aoemaj, int aoemin)
{
	if (aoe_dyndevs)
		return minor_get_dyn(sysminor);
	else
		return minor_get_static(sysminor, aoemaj, aoemin);
}

static void
minor_free(ulong minor)
{
	ulong flags;

	minor /= AOE_PARTITIONS;
	BUG_ON(minor >= N_DEVS);

	spin_lock_irqsave(&used_minors_lock, flags);
	BUG_ON(!test_bit(minor, used_minors));
	clear_bit(minor, used_minors);
	spin_unlock_irqrestore(&used_minors_lock, flags);
}

/*
 * Users who grab a pointer to the device with aoedev_by_aoeaddr
 * automatically get a reference count and must be responsible
 * for performing a aoedev_put.  With the addition of async
 * kthread processing I'm no longer confident that we can
 * guarantee consistency in the face of device flushes.
 *
 * For the time being, we only bother to add extra references for
 * frames sitting on the iocq.  When the kthreads finish processing
 * these frames, they will aoedev_put the device.
 */

void
aoedev_put(struct aoedev *d)
{
	ulong flags;

	spin_lock_irqsave(&devlist_lock, flags);
	d->ref--;
	spin_unlock_irqrestore(&devlist_lock, flags);
}

static void
dummy_timer(ulong vp)
{
	struct aoedev *d;

	d = (struct aoedev *)vp;
	if (d->flags & DEVFL_TKILL)
		return;
	d->timer.expires = jiffies + HZ;
	add_timer(&d->timer);
}

static void
aoe_failip(struct aoedev *d)
{
	struct request *rq;
	struct bio *bio;
	unsigned long n;

	aoe_failbuf(d, d->ip.buf);

	rq = d->ip.rq;
	if (rq == NULL)
		return;
	while ((bio = d->ip.nxbio)) {
		clear_bit(BIO_UPTODATE, &bio->bi_flags);
		d->ip.nxbio = bio->bi_next;
		n = (unsigned long) rq->special;
		rq->special = (void *) --n;
	}
	if ((unsigned long) rq->special == 0)
		aoe_end_request(d, rq, 0);
}

static void
downdev_frame(struct list_head *pos)
{
	struct frame *f;

	f = list_entry(pos, struct frame, head);
	list_del(pos);
	if (f->buf) {
		f->buf->nframesout--;
		aoe_failbuf(f->t->d, f->buf);
	}
	aoe_freetframe(f);
}

void
aoedev_downdev(struct aoedev *d)
{
	struct aoetgt *t, **tt, **te;
	struct list_head *head, *pos, *nx;
	struct request *rq;
	int i;

	d->flags &= ~DEVFL_UP;

	/* clean out active and to-be-retransmitted buffers */
	for (i = 0; i < NFACTIVE; i++) {
		head = &d->factive[i];
		list_for_each_safe(pos, nx, head)
			downdev_frame(pos);
	}
	head = &d->rexmitq;
	list_for_each_safe(pos, nx, head)
		downdev_frame(pos);

	/* reset window dressings */
	tt = d->targets;
	te = tt + d->ntargets;
	for (; tt < te && (t = *tt); tt++) {
		aoecmd_wreset(t);
		t->nout = 0;
	}

	/* clean out the in-process request (if any) */
	aoe_failip(d);

	/* fast fail all pending I/O */
	if (d->blkq) {
		while ((rq = blk_peek_request(d->blkq))) {
			blk_start_request(rq);
			aoe_end_request(d, rq, 1);
		}
	}

	if (d->gd)
		set_capacity(d->gd, 0);
}

/* return whether the user asked for this particular
 * device to be flushed
 */
static int
user_req(char *s, size_t slen, struct aoedev *d)
{
	char *p;
	size_t lim;

	if (!d->gd)
		return 0;
	p = strrchr(d->gd->disk_name, '/');
	if (!p)
		p = d->gd->disk_name;
	else
		p += 1;
	lim = sizeof(d->gd->disk_name);
	lim -= p - d->gd->disk_name;
	if (slen < lim)
		lim = slen;

	return !strncmp(s, p, lim);
}

static void
freedev(struct aoedev *d)
{
	struct aoetgt **t, **e;
	int freeing = 0;
	unsigned long flags;

	spin_lock_irqsave(&d->lock, flags);
	if (d->flags & DEVFL_TKILL
	&& !(d->flags & DEVFL_FREEING)) {
		d->flags |= DEVFL_FREEING;
		freeing = 1;
	}
	spin_unlock_irqrestore(&d->lock, flags);
	if (!freeing)
		return;

	del_timer_sync(&d->timer);
	if (d->gd) {
		aoedisk_rm_sysfs(d);
		del_gendisk(d->gd);
		put_disk(d->gd);
		blk_cleanup_queue(d->blkq);
	}
	t = d->targets;
	e = t + d->ntargets;
	for (; t < e && *t; t++)
		freetgt(d, *t);
	if (d->bufpool)
		mempool_destroy(d->bufpool);
	skbpoolfree(d);
	minor_free(d->sysminor);

	spin_lock_irqsave(&d->lock, flags);
	d->flags |= DEVFL_FREED;
	spin_unlock_irqrestore(&d->lock, flags);
}

enum flush_parms {
	NOT_EXITING = 0,
	EXITING = 1,
};

static int
flush(const char __user *str, size_t cnt, int exiting)
{
	ulong flags;
	struct aoedev *d, **dd;
	char buf[16];
	int all = 0;
	int specified = 0;	/* flush a specific device */
	unsigned int skipflags;

	skipflags = DEVFL_GDALLOC | DEVFL_NEWSIZE | DEVFL_TKILL;

	if (!exiting && cnt >= 3) {
		if (cnt > sizeof buf)
			cnt = sizeof buf;
		if (copy_from_user(buf, str, cnt))
			return -EFAULT;
		all = !strncmp(buf, "all", 3);
		if (!all)
			specified = 1;
	}

	flush_scheduled_work();
	/* pass one: without sleeping, do aoedev_downdev */
	spin_lock_irqsave(&devlist_lock, flags);
	for (d = devlist; d; d = d->next) {
		spin_lock(&d->lock);
		if (exiting) {
			/* unconditionally take each device down */
		} else if (specified) {
			if (!user_req(buf, cnt, d))
				goto cont;
		} else if ((!all && (d->flags & DEVFL_UP))
		|| d->flags & skipflags
		|| d->nopen
		|| d->ref)
			goto cont;

		aoedev_downdev(d);
		d->flags |= DEVFL_TKILL;
cont:
		spin_unlock(&d->lock);
	}
	spin_unlock_irqrestore(&devlist_lock, flags);

	/* pass two: call freedev, which might sleep,
	 * for aoedevs marked with DEVFL_TKILL
	 */
restart:
	spin_lock_irqsave(&devlist_lock, flags);
	for (d = devlist; d; d = d->next) {
		spin_lock(&d->lock);
		if (d->flags & DEVFL_TKILL
		&& !(d->flags & DEVFL_FREEING)) {
			spin_unlock(&d->lock);
			spin_unlock_irqrestore(&devlist_lock, flags);
			freedev(d);
			goto restart;
		}
		spin_unlock(&d->lock);
	}

	/* pass three: remove aoedevs marked with DEVFL_FREED */
	for (dd = &devlist, d = *dd; d; d = *dd) {
		struct aoedev *doomed = NULL;

		spin_lock(&d->lock);
		if (d->flags & DEVFL_FREED) {
			*dd = d->next;
			doomed = d;
		} else {
			dd = &d->next;
		}
		spin_unlock(&d->lock);
		if (doomed)
			kfree(doomed->targets);
		kfree(doomed);
	}
	spin_unlock_irqrestore(&devlist_lock, flags);

	return 0;
}

int
aoedev_flush(const char __user *str, size_t cnt)
{
	return flush(str, cnt, NOT_EXITING);
}

/* This has been confirmed to occur once with Tms=3*1000 due to the
 * driver changing link and not processing its transmit ring.  The
 * problem is hard enough to solve by returning an error that I'm
 * still punting on "solving" this.
 */
static void
skbfree(struct sk_buff *skb)
{
	enum { Sms = 250, Tms = 30 * 1000};
	int i = Tms / Sms;

	if (skb == NULL)
		return;
	while (atomic_read(&skb_shinfo(skb)->dataref) != 1 && i-- > 0)
		msleep(Sms);
	if (i < 0) {
		printk(KERN_ERR
			"aoe: %s holds ref: %s\n",
			skb->dev ? skb->dev->name : "netif",
			"cannot free skb -- memory leaked.");
		return;
	}
	skb->truesize -= skb->data_len;
	skb_shinfo(skb)->nr_frags = skb->data_len = 0;
	skb_trim(skb, 0);
	dev_kfree_skb(skb);
}

static void
skbpoolfree(struct aoedev *d)
{
	struct sk_buff *skb, *tmp;

	skb_queue_walk_safe(&d->skbpool, skb, tmp)
		skbfree(skb);

	__skb_queue_head_init(&d->skbpool);
}

/* find it or allocate it */
struct aoedev *
aoedev_by_aoeaddr(ulong maj, int min, int do_alloc)
{
	struct aoedev *d;
	int i;
	ulong flags;
	ulong sysminor = 0;

	spin_lock_irqsave(&devlist_lock, flags);

	for (d=devlist; d; d=d->next)
		if (d->aoemajor == maj && d->aoeminor == min) {
			spin_lock(&d->lock);
			if (d->flags & DEVFL_TKILL) {
				spin_unlock(&d->lock);
				d = NULL;
				goto out;
			}
			d->ref++;
			spin_unlock(&d->lock);
			break;
		}
	if (d || !do_alloc || minor_get(&sysminor, maj, min) < 0)
		goto out;
	d = kcalloc(1, sizeof *d, GFP_ATOMIC);
	if (!d)
		goto out;
	d->targets = kcalloc(NTARGETS, sizeof(*d->targets), GFP_ATOMIC);
	if (!d->targets) {
		kfree(d);
		d = NULL;
		goto out;
	}
	d->ntargets = NTARGETS;
	INIT_WORK(&d->work, aoecmd_sleepwork);
	spin_lock_init(&d->lock);
	skb_queue_head_init(&d->skbpool);
	init_timer(&d->timer);
	d->timer.data = (ulong) d;
	d->timer.function = dummy_timer;
	d->timer.expires = jiffies + HZ;
	add_timer(&d->timer);
	d->bufpool = NULL;	/* defer to aoeblk_gdalloc */
	d->tgt = d->targets;
	d->ref = 1;
	for (i = 0; i < NFACTIVE; i++)
		INIT_LIST_HEAD(&d->factive[i]);
	INIT_LIST_HEAD(&d->rexmitq);
	d->sysminor = sysminor;
	d->aoemajor = maj;
	d->aoeminor = min;
	d->rttavg = RTTAVG_INIT;
	d->rttdev = RTTDEV_INIT;
	d->next = devlist;
	devlist = d;
 out:
	spin_unlock_irqrestore(&devlist_lock, flags);
	return d;
}

static void
freetgt(struct aoedev *d, struct aoetgt *t)
{
	struct frame *f;
	struct list_head *pos, *nx, *head;
	struct aoeif *ifp;

	for (ifp = t->ifs; ifp < &t->ifs[NAOEIFS]; ++ifp) {
		if (!ifp->nd)
			break;
		dev_put(ifp->nd);
	}

	head = &t->ffree;
	list_for_each_safe(pos, nx, head) {
		list_del(pos);
		f = list_entry(pos, struct frame, head);
		skbfree(f->skb);
		kfree(f);
	}
	kfree(t);
}

void
aoedev_exit(void)
{
	flush_scheduled_work();
	flush(NULL, 0, EXITING);
}

int __init
aoedev_init(void)
{
	return 0;
}