summaryrefslogtreecommitdiff
path: root/drivers/infiniband/hw/cxgb4/resource.c
blob: 57ed26b3cc21a6b244af622252fa1646eeeb6edd (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
/*
 * Copyright (c) 2009-2010 Chelsio, Inc. All rights reserved.
 *
 * This software is available to you under a choice of one of two
 * licenses.  You may choose to be licensed under the terms of the GNU
 * General Public License (GPL) Version 2, available from the file
 * COPYING in the main directory of this source tree, or the
 * OpenIB.org BSD license below:
 *
 *     Redistribution and use in source and binary forms, with or
 *     without modification, are permitted provided that the following
 *     conditions are met:
 *
 *      - Redistributions of source code must retain the above
 *        copyright notice, this list of conditions and the following
 *        disclaimer.
 *
 *      - 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.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
 * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS
 * BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN
 * ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */
/* Crude resource management */
#include <linux/spinlock.h>
#include <linux/genalloc.h>
#include <linux/ratelimit.h>
#include "iw_cxgb4.h"

static int c4iw_init_qid_table(struct c4iw_rdev *rdev)
{
	u32 i;

	if (c4iw_id_table_alloc(&rdev->resource.qid_table,
				rdev->lldi.vr->qp.start,
				rdev->lldi.vr->qp.size,
				rdev->lldi.vr->qp.size, 0))
		return -ENOMEM;

	for (i = rdev->lldi.vr->qp.start;
		i < rdev->lldi.vr->qp.start + rdev->lldi.vr->qp.size; i++)
		if (!(i & rdev->qpmask))
			c4iw_id_free(&rdev->resource.qid_table, i);
	return 0;
}

/* nr_* must be power of 2 */
int c4iw_init_resource(struct c4iw_rdev *rdev, u32 nr_tpt,
		       u32 nr_pdid, u32 nr_srqt)
{
	int err = 0;
	err = c4iw_id_table_alloc(&rdev->resource.tpt_table, 0, nr_tpt, 1,
					C4IW_ID_TABLE_F_RANDOM);
	if (err)
		goto tpt_err;
	err = c4iw_init_qid_table(rdev);
	if (err)
		goto qid_err;
	err = c4iw_id_table_alloc(&rdev->resource.pdid_table, 0,
					nr_pdid, 1, 0);
	if (err)
		goto pdid_err;
	if (!nr_srqt)
		err = c4iw_id_table_alloc(&rdev->resource.srq_table, 0,
					  1, 1, 0);
	else
		err = c4iw_id_table_alloc(&rdev->resource.srq_table, 0,
					  nr_srqt, 0, 0);
	if (err)
		goto srq_err;
	return 0;
 srq_err:
	c4iw_id_table_free(&rdev->resource.pdid_table);
 pdid_err:
	c4iw_id_table_free(&rdev->resource.qid_table);
 qid_err:
	c4iw_id_table_free(&rdev->resource.tpt_table);
 tpt_err:
	return -ENOMEM;
}

/*
 * returns 0 if no resource available
 */
u32 c4iw_get_resource(struct c4iw_id_table *id_table)
{
	u32 entry;
	entry = c4iw_id_alloc(id_table);
	if (entry == (u32)(-1))
		return 0;
	return entry;
}

void c4iw_put_resource(struct c4iw_id_table *id_table, u32 entry)
{
	pr_debug("entry 0x%x\n", entry);
	c4iw_id_free(id_table, entry);
}

u32 c4iw_get_cqid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
{
	struct c4iw_qid_list *entry;
	u32 qid;
	int i;

	mutex_lock(&uctx->lock);
	if (!list_empty(&uctx->cqids)) {
		entry = list_entry(uctx->cqids.next, struct c4iw_qid_list,
				   entry);
		list_del(&entry->entry);
		qid = entry->qid;
		kfree(entry);
	} else {
		qid = c4iw_get_resource(&rdev->resource.qid_table);
		if (!qid)
			goto out;
		mutex_lock(&rdev->stats.lock);
		rdev->stats.qid.cur += rdev->qpmask + 1;
		mutex_unlock(&rdev->stats.lock);
		for (i = qid+1; i & rdev->qpmask; i++) {
			entry = kmalloc(sizeof *entry, GFP_KERNEL);
			if (!entry)
				goto out;
			entry->qid = i;
			list_add_tail(&entry->entry, &uctx->cqids);
		}

		/*
		 * now put the same ids on the qp list since they all
		 * map to the same db/gts page.
		 */
		entry = kmalloc(sizeof *entry, GFP_KERNEL);
		if (!entry)
			goto out;
		entry->qid = qid;
		list_add_tail(&entry->entry, &uctx->qpids);
		for (i = qid+1; i & rdev->qpmask; i++) {
			entry = kmalloc(sizeof *entry, GFP_KERNEL);
			if (!entry)
				goto out;
			entry->qid = i;
			list_add_tail(&entry->entry, &uctx->qpids);
		}
	}
out:
	mutex_unlock(&uctx->lock);
	pr_debug("qid 0x%x\n", qid);
	mutex_lock(&rdev->stats.lock);
	if (rdev->stats.qid.cur > rdev->stats.qid.max)
		rdev->stats.qid.max = rdev->stats.qid.cur;
	mutex_unlock(&rdev->stats.lock);
	return qid;
}

void c4iw_put_cqid(struct c4iw_rdev *rdev, u32 qid,
		   struct c4iw_dev_ucontext *uctx)
{
	struct c4iw_qid_list *entry;

	entry = kmalloc(sizeof *entry, GFP_KERNEL);
	if (!entry)
		return;
	pr_debug("qid 0x%x\n", qid);
	entry->qid = qid;
	mutex_lock(&uctx->lock);
	list_add_tail(&entry->entry, &uctx->cqids);
	mutex_unlock(&uctx->lock);
}

u32 c4iw_get_qpid(struct c4iw_rdev *rdev, struct c4iw_dev_ucontext *uctx)
{
	struct c4iw_qid_list *entry;
	u32 qid;
	int i;

	mutex_lock(&uctx->lock);
	if (!list_empty(&uctx->qpids)) {
		entry = list_entry(uctx->qpids.next, struct c4iw_qid_list,
				   entry);
		list_del(&entry->entry);
		qid = entry->qid;
		kfree(entry);
	} else {
		qid = c4iw_get_resource(&rdev->resource.qid_table);
		if (!qid) {
			mutex_lock(&rdev->stats.lock);
			rdev->stats.qid.fail++;
			mutex_unlock(&rdev->stats.lock);
			goto out;
		}
		mutex_lock(&rdev->stats.lock);
		rdev->stats.qid.cur += rdev->qpmask + 1;
		mutex_unlock(&rdev->stats.lock);
		for (i = qid+1; i & rdev->qpmask; i++) {
			entry = kmalloc(sizeof *entry, GFP_KERNEL);
			if (!entry)
				goto out;
			entry->qid = i;
			list_add_tail(&entry->entry, &uctx->qpids);
		}

		/*
		 * now put the same ids on the cq list since they all
		 * map to the same db/gts page.
		 */
		entry = kmalloc(sizeof *entry, GFP_KERNEL);
		if (!entry)
			goto out;
		entry->qid = qid;
		list_add_tail(&entry->entry, &uctx->cqids);
		for (i = qid; i & rdev->qpmask; i++) {
			entry = kmalloc(sizeof *entry, GFP_KERNEL);
			if (!entry)
				goto out;
			entry->qid = i;
			list_add_tail(&entry->entry, &uctx->cqids);
		}
	}
out:
	mutex_unlock(&uctx->lock);
	pr_debug("qid 0x%x\n", qid);
	mutex_lock(&rdev->stats.lock);
	if (rdev->stats.qid.cur > rdev->stats.qid.max)
		rdev->stats.qid.max = rdev->stats.qid.cur;
	mutex_unlock(&rdev->stats.lock);
	return qid;
}

void c4iw_put_qpid(struct c4iw_rdev *rdev, u32 qid,
		   struct c4iw_dev_ucontext *uctx)
{
	struct c4iw_qid_list *entry;

	entry = kmalloc(sizeof *entry, GFP_KERNEL);
	if (!entry)
		return;
	pr_debug("qid 0x%x\n", qid);
	entry->qid = qid;
	mutex_lock(&uctx->lock);
	list_add_tail(&entry->entry, &uctx->qpids);
	mutex_unlock(&uctx->lock);
}

void c4iw_destroy_resource(struct c4iw_resource *rscp)
{
	c4iw_id_table_free(&rscp->tpt_table);
	c4iw_id_table_free(&rscp->qid_table);
	c4iw_id_table_free(&rscp->pdid_table);
}

/*
 * PBL Memory Manager.  Uses Linux generic allocator.
 */

#define MIN_PBL_SHIFT 8			/* 256B == min PBL size (32 entries) */

u32 c4iw_pblpool_alloc(struct c4iw_rdev *rdev, int size)
{
	unsigned long addr = gen_pool_alloc(rdev->pbl_pool, size);
	pr_debug("addr 0x%x size %d\n", (u32)addr, size);
	mutex_lock(&rdev->stats.lock);
	if (addr) {
		rdev->stats.pbl.cur += roundup(size, 1 << MIN_PBL_SHIFT);
		if (rdev->stats.pbl.cur > rdev->stats.pbl.max)
			rdev->stats.pbl.max = rdev->stats.pbl.cur;
		kref_get(&rdev->pbl_kref);
	} else
		rdev->stats.pbl.fail++;
	mutex_unlock(&rdev->stats.lock);
	return (u32)addr;
}

static void destroy_pblpool(struct kref *kref)
{
	struct c4iw_rdev *rdev;

	rdev = container_of(kref, struct c4iw_rdev, pbl_kref);
	gen_pool_destroy(rdev->pbl_pool);
	complete(&rdev->pbl_compl);
}

void c4iw_pblpool_free(struct c4iw_rdev *rdev, u32 addr, int size)
{
	pr_debug("addr 0x%x size %d\n", addr, size);
	mutex_lock(&rdev->stats.lock);
	rdev->stats.pbl.cur -= roundup(size, 1 << MIN_PBL_SHIFT);
	mutex_unlock(&rdev->stats.lock);
	gen_pool_free(rdev->pbl_pool, (unsigned long)addr, size);
	kref_put(&rdev->pbl_kref, destroy_pblpool);
}

int c4iw_pblpool_create(struct c4iw_rdev *rdev)
{
	unsigned pbl_start, pbl_chunk, pbl_top;

	rdev->pbl_pool = gen_pool_create(MIN_PBL_SHIFT, -1);
	if (!rdev->pbl_pool)
		return -ENOMEM;

	pbl_start = rdev->lldi.vr->pbl.start;
	pbl_chunk = rdev->lldi.vr->pbl.size;
	pbl_top = pbl_start + pbl_chunk;

	while (pbl_start < pbl_top) {
		pbl_chunk = min(pbl_top - pbl_start + 1, pbl_chunk);
		if (gen_pool_add(rdev->pbl_pool, pbl_start, pbl_chunk, -1)) {
			pr_debug("failed to add PBL chunk (%x/%x)\n",
				 pbl_start, pbl_chunk);
			if (pbl_chunk <= 1024 << MIN_PBL_SHIFT) {
				pr_warn("Failed to add all PBL chunks (%x/%x)\n",
					pbl_start, pbl_top - pbl_start);
				return 0;
			}
			pbl_chunk >>= 1;
		} else {
			pr_debug("added PBL chunk (%x/%x)\n",
				 pbl_start, pbl_chunk);
			pbl_start += pbl_chunk;
		}
	}

	return 0;
}

void c4iw_pblpool_destroy(struct c4iw_rdev *rdev)
{
	kref_put(&rdev->pbl_kref, destroy_pblpool);
}

/*
 * RQT Memory Manager.  Uses Linux generic allocator.
 */

#define MIN_RQT_SHIFT 10	/* 1KB == min RQT size (16 entries) */

u32 c4iw_rqtpool_alloc(struct c4iw_rdev *rdev, int size)
{
	unsigned long addr = gen_pool_alloc(rdev->rqt_pool, size << 6);
	pr_debug("addr 0x%x size %d\n", (u32)addr, size << 6);
	if (!addr)
		pr_warn_ratelimited("%s: Out of RQT memory\n",
				    pci_name(rdev->lldi.pdev));
	mutex_lock(&rdev->stats.lock);
	if (addr) {
		rdev->stats.rqt.cur += roundup(size << 6, 1 << MIN_RQT_SHIFT);
		if (rdev->stats.rqt.cur > rdev->stats.rqt.max)
			rdev->stats.rqt.max = rdev->stats.rqt.cur;
		kref_get(&rdev->rqt_kref);
	} else
		rdev->stats.rqt.fail++;
	mutex_unlock(&rdev->stats.lock);
	return (u32)addr;
}

static void destroy_rqtpool(struct kref *kref)
{
	struct c4iw_rdev *rdev;

	rdev = container_of(kref, struct c4iw_rdev, rqt_kref);
	gen_pool_destroy(rdev->rqt_pool);
	complete(&rdev->rqt_compl);
}

void c4iw_rqtpool_free(struct c4iw_rdev *rdev, u32 addr, int size)
{
	pr_debug("addr 0x%x size %d\n", addr, size << 6);
	mutex_lock(&rdev->stats.lock);
	rdev->stats.rqt.cur -= roundup(size << 6, 1 << MIN_RQT_SHIFT);
	mutex_unlock(&rdev->stats.lock);
	gen_pool_free(rdev->rqt_pool, (unsigned long)addr, size << 6);
	kref_put(&rdev->rqt_kref, destroy_rqtpool);
}

int c4iw_rqtpool_create(struct c4iw_rdev *rdev)
{
	unsigned rqt_start, rqt_chunk, rqt_top;
	int skip = 0;

	rdev->rqt_pool = gen_pool_create(MIN_RQT_SHIFT, -1);
	if (!rdev->rqt_pool)
		return -ENOMEM;

	/*
	 * If SRQs are supported, then never use the first RQE from
	 * the RQT region. This is because HW uses RQT index 0 as NULL.
	 */
	if (rdev->lldi.vr->srq.size)
		skip = T4_RQT_ENTRY_SIZE;

	rqt_start = rdev->lldi.vr->rq.start + skip;
	rqt_chunk = rdev->lldi.vr->rq.size - skip;
	rqt_top = rqt_start + rqt_chunk;

	while (rqt_start < rqt_top) {
		rqt_chunk = min(rqt_top - rqt_start + 1, rqt_chunk);
		if (gen_pool_add(rdev->rqt_pool, rqt_start, rqt_chunk, -1)) {
			pr_debug("failed to add RQT chunk (%x/%x)\n",
				 rqt_start, rqt_chunk);
			if (rqt_chunk <= 1024 << MIN_RQT_SHIFT) {
				pr_warn("Failed to add all RQT chunks (%x/%x)\n",
					rqt_start, rqt_top - rqt_start);
				return 0;
			}
			rqt_chunk >>= 1;
		} else {
			pr_debug("added RQT chunk (%x/%x)\n",
				 rqt_start, rqt_chunk);
			rqt_start += rqt_chunk;
		}
	}
	return 0;
}

void c4iw_rqtpool_destroy(struct c4iw_rdev *rdev)
{
	kref_put(&rdev->rqt_kref, destroy_rqtpool);
}

int c4iw_alloc_srq_idx(struct c4iw_rdev *rdev)
{
	int idx;

	idx = c4iw_id_alloc(&rdev->resource.srq_table);
	mutex_lock(&rdev->stats.lock);
	if (idx == -1) {
		rdev->stats.srqt.fail++;
		mutex_unlock(&rdev->stats.lock);
		return -ENOMEM;
	}
	rdev->stats.srqt.cur++;
	if (rdev->stats.srqt.cur > rdev->stats.srqt.max)
		rdev->stats.srqt.max = rdev->stats.srqt.cur;
	mutex_unlock(&rdev->stats.lock);
	return idx;
}

void c4iw_free_srq_idx(struct c4iw_rdev *rdev, int idx)
{
	c4iw_id_free(&rdev->resource.srq_table, idx);
	mutex_lock(&rdev->stats.lock);
	rdev->stats.srqt.cur--;
	mutex_unlock(&rdev->stats.lock);
}

/*
 * On-Chip QP Memory.
 */
#define MIN_OCQP_SHIFT 12	/* 4KB == min ocqp size */

u32 c4iw_ocqp_pool_alloc(struct c4iw_rdev *rdev, int size)
{
	unsigned long addr = gen_pool_alloc(rdev->ocqp_pool, size);
	pr_debug("addr 0x%x size %d\n", (u32)addr, size);
	if (addr) {
		mutex_lock(&rdev->stats.lock);
		rdev->stats.ocqp.cur += roundup(size, 1 << MIN_OCQP_SHIFT);
		if (rdev->stats.ocqp.cur > rdev->stats.ocqp.max)
			rdev->stats.ocqp.max = rdev->stats.ocqp.cur;
		mutex_unlock(&rdev->stats.lock);
	}
	return (u32)addr;
}

void c4iw_ocqp_pool_free(struct c4iw_rdev *rdev, u32 addr, int size)
{
	pr_debug("addr 0x%x size %d\n", addr, size);
	mutex_lock(&rdev->stats.lock);
	rdev->stats.ocqp.cur -= roundup(size, 1 << MIN_OCQP_SHIFT);
	mutex_unlock(&rdev->stats.lock);
	gen_pool_free(rdev->ocqp_pool, (unsigned long)addr, size);
}

int c4iw_ocqp_pool_create(struct c4iw_rdev *rdev)
{
	unsigned start, chunk, top;

	rdev->ocqp_pool = gen_pool_create(MIN_OCQP_SHIFT, -1);
	if (!rdev->ocqp_pool)
		return -ENOMEM;

	start = rdev->lldi.vr->ocq.start;
	chunk = rdev->lldi.vr->ocq.size;
	top = start + chunk;

	while (start < top) {
		chunk = min(top - start + 1, chunk);
		if (gen_pool_add(rdev->ocqp_pool, start, chunk, -1)) {
			pr_debug("failed to add OCQP chunk (%x/%x)\n",
				 start, chunk);
			if (chunk <= 1024 << MIN_OCQP_SHIFT) {
				pr_warn("Failed to add all OCQP chunks (%x/%x)\n",
					start, top - start);
				return 0;
			}
			chunk >>= 1;
		} else {
			pr_debug("added OCQP chunk (%x/%x)\n",
				 start, chunk);
			start += chunk;
		}
	}
	return 0;
}

void c4iw_ocqp_pool_destroy(struct c4iw_rdev *rdev)
{
	gen_pool_destroy(rdev->ocqp_pool);
}