summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdkfd/kfd_packet_manager.c
blob: 1092631765cb5b09ef198be0caa2007407db0c13 (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
/*
 * Copyright 2014 Advanced Micro Devices, Inc.
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice shall be included in
 * all copies or substantial portions of the Software.
 *
 * 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 COPYRIGHT HOLDER(S) OR AUTHOR(S) 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.
 *
 */

#include <linux/slab.h>
#include <linux/mutex.h>
#include "kfd_device_queue_manager.h"
#include "kfd_kernel_queue.h"
#include "kfd_priv.h"

static inline void inc_wptr(unsigned int *wptr, unsigned int increment_bytes,
				unsigned int buffer_size_bytes)
{
	unsigned int temp = *wptr + increment_bytes / sizeof(uint32_t);

	WARN((temp * sizeof(uint32_t)) > buffer_size_bytes,
	     "Runlist IB overflow");
	*wptr = temp;
}

static void pm_calc_rlib_size(struct packet_manager *pm,
				unsigned int *rlib_size,
				bool *over_subscription)
{
	unsigned int process_count, queue_count, compute_queue_count;
	unsigned int map_queue_size;
	unsigned int max_proc_per_quantum = 1;
	struct kfd_dev *dev = pm->dqm->dev;

	process_count = pm->dqm->processes_count;
	queue_count = pm->dqm->queue_count;
	compute_queue_count = queue_count - pm->dqm->sdma_queue_count;

	/* check if there is over subscription
	 * Note: the arbitration between the number of VMIDs and
	 * hws_max_conc_proc has been done in
	 * kgd2kfd_device_init().
	 */
	*over_subscription = false;

	if (dev->max_proc_per_quantum > 1)
		max_proc_per_quantum = dev->max_proc_per_quantum;

	if ((process_count > max_proc_per_quantum) ||
	    compute_queue_count > get_queues_num(pm->dqm)) {
		*over_subscription = true;
		pr_debug("Over subscribed runlist\n");
	}

	map_queue_size = pm->pmf->map_queues_size;
	/* calculate run list ib allocation size */
	*rlib_size = process_count * pm->pmf->map_process_size +
		     queue_count * map_queue_size;

	/*
	 * Increase the allocation size in case we need a chained run list
	 * when over subscription
	 */
	if (*over_subscription)
		*rlib_size += pm->pmf->runlist_size;

	pr_debug("runlist ib size %d\n", *rlib_size);
}

static int pm_allocate_runlist_ib(struct packet_manager *pm,
				unsigned int **rl_buffer,
				uint64_t *rl_gpu_buffer,
				unsigned int *rl_buffer_size,
				bool *is_over_subscription)
{
	int retval;

	if (WARN_ON(pm->allocated))
		return -EINVAL;

	pm_calc_rlib_size(pm, rl_buffer_size, is_over_subscription);

	mutex_lock(&pm->lock);

	retval = kfd_gtt_sa_allocate(pm->dqm->dev, *rl_buffer_size,
					&pm->ib_buffer_obj);

	if (retval) {
		pr_err("Failed to allocate runlist IB\n");
		goto out;
	}

	*(void **)rl_buffer = pm->ib_buffer_obj->cpu_ptr;
	*rl_gpu_buffer = pm->ib_buffer_obj->gpu_addr;

	memset(*rl_buffer, 0, *rl_buffer_size);
	pm->allocated = true;

out:
	mutex_unlock(&pm->lock);
	return retval;
}

static int pm_create_runlist_ib(struct packet_manager *pm,
				struct list_head *queues,
				uint64_t *rl_gpu_addr,
				size_t *rl_size_bytes)
{
	unsigned int alloc_size_bytes;
	unsigned int *rl_buffer, rl_wptr, i;
	int retval, proccesses_mapped;
	struct device_process_node *cur;
	struct qcm_process_device *qpd;
	struct queue *q;
	struct kernel_queue *kq;
	bool is_over_subscription;

	rl_wptr = retval = proccesses_mapped = 0;

	retval = pm_allocate_runlist_ib(pm, &rl_buffer, rl_gpu_addr,
				&alloc_size_bytes, &is_over_subscription);
	if (retval)
		return retval;

	*rl_size_bytes = alloc_size_bytes;
	pm->ib_size_bytes = alloc_size_bytes;

	pr_debug("Building runlist ib process count: %d queues count %d\n",
		pm->dqm->processes_count, pm->dqm->queue_count);

	/* build the run list ib packet */
	list_for_each_entry(cur, queues, list) {
		qpd = cur->qpd;
		/* build map process packet */
		if (proccesses_mapped >= pm->dqm->processes_count) {
			pr_debug("Not enough space left in runlist IB\n");
			pm_release_ib(pm);
			return -ENOMEM;
		}

		retval = pm->pmf->map_process(pm, &rl_buffer[rl_wptr], qpd);
		if (retval)
			return retval;

		proccesses_mapped++;
		inc_wptr(&rl_wptr, pm->pmf->map_process_size,
				alloc_size_bytes);

		list_for_each_entry(kq, &qpd->priv_queue_list, list) {
			if (!kq->queue->properties.is_active)
				continue;

			pr_debug("static_queue, mapping kernel q %d, is debug status %d\n",
				kq->queue->queue, qpd->is_debug);

			retval = pm->pmf->map_queues(pm,
						&rl_buffer[rl_wptr],
						kq->queue,
						qpd->is_debug);
			if (retval)
				return retval;

			inc_wptr(&rl_wptr,
				pm->pmf->map_queues_size,
				alloc_size_bytes);
		}

		list_for_each_entry(q, &qpd->queues_list, list) {
			if (!q->properties.is_active)
				continue;

			pr_debug("static_queue, mapping user queue %d, is debug status %d\n",
				q->queue, qpd->is_debug);

			retval = pm->pmf->map_queues(pm,
						&rl_buffer[rl_wptr],
						q,
						qpd->is_debug);

			if (retval)
				return retval;

			inc_wptr(&rl_wptr,
				pm->pmf->map_queues_size,
				alloc_size_bytes);
		}
	}

	pr_debug("Finished map process and queues to runlist\n");

	if (is_over_subscription)
		retval = pm->pmf->runlist(pm, &rl_buffer[rl_wptr],
					*rl_gpu_addr,
					alloc_size_bytes / sizeof(uint32_t),
					true);

	for (i = 0; i < alloc_size_bytes / sizeof(uint32_t); i++)
		pr_debug("0x%2X ", rl_buffer[i]);
	pr_debug("\n");

	return retval;
}

int pm_init(struct packet_manager *pm, struct device_queue_manager *dqm)
{
	switch (dqm->dev->device_info->asic_family) {
	case CHIP_KAVERI:
	case CHIP_HAWAII:
		/* PM4 packet structures on CIK are the same as on VI */
	case CHIP_CARRIZO:
	case CHIP_TONGA:
	case CHIP_FIJI:
	case CHIP_POLARIS10:
	case CHIP_POLARIS11:
		pm->pmf = &kfd_vi_pm_funcs;
		break;
	case CHIP_VEGA10:
	case CHIP_RAVEN:
		pm->pmf = &kfd_v9_pm_funcs;
		break;
	default:
		WARN(1, "Unexpected ASIC family %u",
		     dqm->dev->device_info->asic_family);
		return -EINVAL;
	}

	pm->dqm = dqm;
	mutex_init(&pm->lock);
	pm->priv_queue = kernel_queue_init(dqm->dev, KFD_QUEUE_TYPE_HIQ);
	if (!pm->priv_queue) {
		mutex_destroy(&pm->lock);
		return -ENOMEM;
	}
	pm->allocated = false;

	return 0;
}

void pm_uninit(struct packet_manager *pm)
{
	mutex_destroy(&pm->lock);
	kernel_queue_uninit(pm->priv_queue);
}

int pm_send_set_resources(struct packet_manager *pm,
				struct scheduling_resources *res)
{
	uint32_t *buffer, size;
	int retval = 0;

	size = pm->pmf->set_resources_size;
	mutex_lock(&pm->lock);
	pm->priv_queue->ops.acquire_packet_buffer(pm->priv_queue,
					size / sizeof(uint32_t),
					(unsigned int **)&buffer);
	if (!buffer) {
		pr_err("Failed to allocate buffer on kernel queue\n");
		retval = -ENOMEM;
		goto out;
	}

	retval = pm->pmf->set_resources(pm, buffer, res);
	if (!retval)
		pm->priv_queue->ops.submit_packet(pm->priv_queue);
	else
		pm->priv_queue->ops.rollback_packet(pm->priv_queue);

out:
	mutex_unlock(&pm->lock);

	return retval;
}

int pm_send_runlist(struct packet_manager *pm, struct list_head *dqm_queues)
{
	uint64_t rl_gpu_ib_addr;
	uint32_t *rl_buffer;
	size_t rl_ib_size, packet_size_dwords;
	int retval;

	retval = pm_create_runlist_ib(pm, dqm_queues, &rl_gpu_ib_addr,
					&rl_ib_size);
	if (retval)
		goto fail_create_runlist_ib;

	pr_debug("runlist IB address: 0x%llX\n", rl_gpu_ib_addr);

	packet_size_dwords = pm->pmf->runlist_size / sizeof(uint32_t);
	mutex_lock(&pm->lock);

	retval = pm->priv_queue->ops.acquire_packet_buffer(pm->priv_queue,
					packet_size_dwords, &rl_buffer);
	if (retval)
		goto fail_acquire_packet_buffer;

	retval = pm->pmf->runlist(pm, rl_buffer, rl_gpu_ib_addr,
					rl_ib_size / sizeof(uint32_t), false);
	if (retval)
		goto fail_create_runlist;

	pm->priv_queue->ops.submit_packet(pm->priv_queue);

	mutex_unlock(&pm->lock);

	return retval;

fail_create_runlist:
	pm->priv_queue->ops.rollback_packet(pm->priv_queue);
fail_acquire_packet_buffer:
	mutex_unlock(&pm->lock);
fail_create_runlist_ib:
	pm_release_ib(pm);
	return retval;
}

int pm_send_query_status(struct packet_manager *pm, uint64_t fence_address,
			uint32_t fence_value)
{
	uint32_t *buffer, size;
	int retval = 0;

	if (WARN_ON(!fence_address))
		return -EFAULT;

	size = pm->pmf->query_status_size;
	mutex_lock(&pm->lock);
	pm->priv_queue->ops.acquire_packet_buffer(pm->priv_queue,
			size / sizeof(uint32_t), (unsigned int **)&buffer);
	if (!buffer) {
		pr_err("Failed to allocate buffer on kernel queue\n");
		retval = -ENOMEM;
		goto out;
	}

	retval = pm->pmf->query_status(pm, buffer, fence_address, fence_value);
	if (!retval)
		pm->priv_queue->ops.submit_packet(pm->priv_queue);
	else
		pm->priv_queue->ops.rollback_packet(pm->priv_queue);

out:
	mutex_unlock(&pm->lock);
	return retval;
}

int pm_send_unmap_queue(struct packet_manager *pm, enum kfd_queue_type type,
			enum kfd_unmap_queues_filter filter,
			uint32_t filter_param, bool reset,
			unsigned int sdma_engine)
{
	uint32_t *buffer, size;
	int retval = 0;

	size = pm->pmf->unmap_queues_size;
	mutex_lock(&pm->lock);
	pm->priv_queue->ops.acquire_packet_buffer(pm->priv_queue,
			size / sizeof(uint32_t), (unsigned int **)&buffer);
	if (!buffer) {
		pr_err("Failed to allocate buffer on kernel queue\n");
		retval = -ENOMEM;
		goto out;
	}

	retval = pm->pmf->unmap_queues(pm, buffer, type, filter, filter_param,
				       reset, sdma_engine);
	if (!retval)
		pm->priv_queue->ops.submit_packet(pm->priv_queue);
	else
		pm->priv_queue->ops.rollback_packet(pm->priv_queue);

out:
	mutex_unlock(&pm->lock);
	return retval;
}

void pm_release_ib(struct packet_manager *pm)
{
	mutex_lock(&pm->lock);
	if (pm->allocated) {
		kfd_gtt_sa_free(pm->dqm->dev, pm->ib_buffer_obj);
		pm->allocated = false;
	}
	mutex_unlock(&pm->lock);
}

#if defined(CONFIG_DEBUG_FS)

int pm_debugfs_runlist(struct seq_file *m, void *data)
{
	struct packet_manager *pm = data;

	mutex_lock(&pm->lock);

	if (!pm->allocated) {
		seq_puts(m, "  No active runlist\n");
		goto out;
	}

	seq_hex_dump(m, "  ", DUMP_PREFIX_OFFSET, 32, 4,
		     pm->ib_buffer_obj->cpu_ptr, pm->ib_size_bytes, false);

out:
	mutex_unlock(&pm->lock);
	return 0;
}

int pm_debugfs_hang_hws(struct packet_manager *pm)
{
	uint32_t *buffer, size;
	int r = 0;

	size = pm->pmf->query_status_size;
	mutex_lock(&pm->lock);
	pm->priv_queue->ops.acquire_packet_buffer(pm->priv_queue,
			size / sizeof(uint32_t), (unsigned int **)&buffer);
	if (!buffer) {
		pr_err("Failed to allocate buffer on kernel queue\n");
		r = -ENOMEM;
		goto out;
	}
	memset(buffer, 0x55, size);
	pm->priv_queue->ops.submit_packet(pm->priv_queue);

	pr_info("Submitting %x %x %x %x %x %x %x to HIQ to hang the HWS.",
		buffer[0], buffer[1], buffer[2], buffer[3],
		buffer[4], buffer[5], buffer[6]);
out:
	mutex_unlock(&pm->lock);
	return r;
}


#endif