summaryrefslogtreecommitdiffstats
path: root/drivers/gpu/nvgpu/gk20a/tsg_gk20a.c
blob: f8c8be5e48f2965e2c385060a3f17981636ee945 (plain) (blame)
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
/*
 * Copyright (c) 2014-2017, NVIDIA CORPORATION.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or modify it
 * under the terms and conditions of the GNU General Public License,
 * version 2, as published by the Free Software Foundation.
 *
 * This program is distributed in the hope it will be useful, but WITHOUT
 * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
 * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
 * more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program.  If not, see <http://www.gnu.org/licenses/>.
 */

#include <nvgpu/kmem.h>
#include <nvgpu/log.h>

#include "gk20a.h"
#include "tsg_gk20a.h"

bool gk20a_is_channel_marked_as_tsg(struct channel_gk20a *ch)
{
	return !(ch->tsgid == NVGPU_INVALID_TSG_ID);
}

int gk20a_enable_tsg(struct tsg_gk20a *tsg)
{
	struct gk20a *g = tsg->g;
	struct channel_gk20a *ch;
	bool is_next, is_ctx_reload;

	gk20a_fifo_disable_tsg_sched(g, tsg);

	/*
	 * Due to h/w bug that exists in Maxwell and Pascal,
	 * we first need to enable all channels with NEXT and CTX_RELOAD set,
	 * and then rest of the channels should be enabled
	 */
	down_read(&tsg->ch_list_lock);
	nvgpu_list_for_each_entry(ch, &tsg->ch_list, channel_gk20a, ch_entry) {
		is_next = gk20a_fifo_channel_status_is_next(g, ch->chid);
		is_ctx_reload = gk20a_fifo_channel_status_is_ctx_reload(g, ch->chid);

		if (is_next || is_ctx_reload)
			g->ops.fifo.enable_channel(ch);
	}

	nvgpu_list_for_each_entry(ch, &tsg->ch_list, channel_gk20a, ch_entry) {
		is_next = gk20a_fifo_channel_status_is_next(g, ch->chid);
		is_ctx_reload = gk20a_fifo_channel_status_is_ctx_reload(g, ch->chid);

		if (is_next || is_ctx_reload)
			continue;

		g->ops.fifo.enable_channel(ch);
	}
	up_read(&tsg->ch_list_lock);

	gk20a_fifo_enable_tsg_sched(g, tsg);

	return 0;
}

int gk20a_disable_tsg(struct tsg_gk20a *tsg)
{
	struct gk20a *g = tsg->g;
	struct channel_gk20a *ch;

	down_read(&tsg->ch_list_lock);
	nvgpu_list_for_each_entry(ch, &tsg->ch_list, channel_gk20a, ch_entry) {
		g->ops.fifo.disable_channel(ch);
	}
	up_read(&tsg->ch_list_lock);

	return 0;
}

static bool gk20a_is_channel_active(struct gk20a *g, struct channel_gk20a *ch)
{
	struct fifo_gk20a *f = &g->fifo;
	struct fifo_runlist_info_gk20a *runlist;
	unsigned int i;

	for (i = 0; i < f->max_runlists; ++i) {
		runlist = &f->runlist_info[i];
		if (test_bit(ch->chid, runlist->active_channels))
			return true;
	}

	return false;
}

/*
 * API to mark channel as part of TSG
 *
 * Note that channel is not runnable when we bind it to TSG
 */
int gk20a_tsg_bind_channel(struct tsg_gk20a *tsg,
			struct channel_gk20a *ch)
{
	gk20a_dbg_fn("");

	/* check if channel is already bound to some TSG */
	if (gk20a_is_channel_marked_as_tsg(ch)) {
		return -EINVAL;
	}

	/* channel cannot be bound to TSG if it is already active */
	if (gk20a_is_channel_active(tsg->g, ch)) {
		return -EINVAL;
	}

	ch->tsgid = tsg->tsgid;

	/* all the channel part of TSG should need to be same runlist_id */
	if (tsg->runlist_id == FIFO_INVAL_TSG_ID)
		tsg->runlist_id = ch->runlist_id;
	else if (tsg->runlist_id != ch->runlist_id) {
		nvgpu_err(tsg->g,
			"Error: TSG channel should be share same runlist ch[%d] tsg[%d]",
			ch->runlist_id, tsg->runlist_id);
		return -EINVAL;
	}

	down_write(&tsg->ch_list_lock);
	nvgpu_list_add_tail(&ch->ch_entry, &tsg->ch_list);
	up_write(&tsg->ch_list_lock);

	nvgpu_ref_get(&tsg->refcount);

	gk20a_dbg(gpu_dbg_fn, "BIND tsg:%d channel:%d\n",
					tsg->tsgid, ch->chid);

	gk20a_dbg_fn("done");
	return 0;
}

int gk20a_tsg_unbind_channel(struct channel_gk20a *ch)
{
	struct gk20a *g = ch->g;
	struct tsg_gk20a *tsg = &g->fifo.tsg[ch->tsgid];
	int err;

	err = gk20a_fifo_tsg_unbind_channel(ch);
	if (err)
		return err;

	nvgpu_ref_put(&tsg->refcount, gk20a_tsg_release);
	ch->tsgid = NVGPU_INVALID_TSG_ID;

	gk20a_dbg(gpu_dbg_fn, "UNBIND tsg:%d channel:%d\n",
					tsg->tsgid, ch->chid);

	return 0;
}

int gk20a_init_tsg_support(struct gk20a *g, u32 tsgid)
{
	struct tsg_gk20a *tsg = NULL;
	int err;

	if (tsgid >= g->fifo.num_channels)
		return -EINVAL;

	tsg = &g->fifo.tsg[tsgid];

	tsg->in_use = false;
	tsg->tsgid = tsgid;

	nvgpu_init_list_node(&tsg->ch_list);
	init_rwsem(&tsg->ch_list_lock);

	nvgpu_init_list_node(&tsg->event_id_list);
	err = nvgpu_mutex_init(&tsg->event_id_list_lock);
	if (err) {
		tsg->in_use = true; /* make this TSG unusable */
		return err;
	}

	return 0;
}

int gk20a_tsg_set_priority(struct gk20a *g, struct tsg_gk20a *tsg,
				u32 priority)
{
	u32 timeslice_us;

	switch (priority) {
	case NVGPU_PRIORITY_LOW:
		timeslice_us = g->timeslice_low_priority_us;
		break;
	case NVGPU_PRIORITY_MEDIUM:
		timeslice_us = g->timeslice_medium_priority_us;
		break;
	case NVGPU_PRIORITY_HIGH:
		timeslice_us = g->timeslice_high_priority_us;
		break;
	default:
		pr_err("Unsupported priority");
		return -EINVAL;
	}

	return gk20a_tsg_set_timeslice(tsg, timeslice_us);
}

int gk20a_tsg_set_runlist_interleave(struct tsg_gk20a *tsg, u32 level)
{
	struct gk20a *g = tsg->g;
	int ret;

	gk20a_dbg(gpu_dbg_sched, "tsgid=%u interleave=%u", tsg->tsgid, level);

	switch (level) {
	case NVGPU_RUNLIST_INTERLEAVE_LEVEL_LOW:
	case NVGPU_RUNLIST_INTERLEAVE_LEVEL_MEDIUM:
	case NVGPU_RUNLIST_INTERLEAVE_LEVEL_HIGH:
		ret = g->ops.fifo.set_runlist_interleave(g, tsg->tsgid,
							true, 0, level);
		if (!ret)
			tsg->interleave_level = level;
		break;
	default:
		ret = -EINVAL;
		break;
	}

	return ret ? ret : g->ops.fifo.update_runlist(g, tsg->runlist_id, ~0, true, true);
}

int gk20a_tsg_set_timeslice(struct tsg_gk20a *tsg, u32 timeslice)
{
	struct gk20a *g = tsg->g;

	gk20a_dbg(gpu_dbg_sched, "tsgid=%u timeslice=%u us", tsg->tsgid, timeslice);

	return g->ops.fifo.tsg_set_timeslice(tsg, timeslice);
}

u32 gk20a_tsg_get_timeslice(struct tsg_gk20a *tsg)
{
	struct gk20a *g = tsg->g;

	if (!tsg->timeslice_us)
		return g->ops.fifo.default_timeslice_us(g);

	return tsg->timeslice_us;
}

static void release_used_tsg(struct fifo_gk20a *f, struct tsg_gk20a *tsg)
{
	nvgpu_mutex_acquire(&f->tsg_inuse_mutex);
	f->tsg[tsg->tsgid].in_use = false;
	nvgpu_mutex_release(&f->tsg_inuse_mutex);
}

static struct tsg_gk20a *gk20a_tsg_acquire_unused_tsg(struct fifo_gk20a *f)
{
	struct tsg_gk20a *tsg = NULL;
	unsigned int tsgid;

	nvgpu_mutex_acquire(&f->tsg_inuse_mutex);
	for (tsgid = 0; tsgid < f->num_channels; tsgid++) {
		if (!f->tsg[tsgid].in_use) {
			f->tsg[tsgid].in_use = true;
			tsg = &f->tsg[tsgid];
			break;
		}
	}
	nvgpu_mutex_release(&f->tsg_inuse_mutex);

	return tsg;
}

struct tsg_gk20a *gk20a_tsg_open(struct gk20a *g)
{
	struct tsg_gk20a *tsg;
	int err;

	tsg = gk20a_tsg_acquire_unused_tsg(&g->fifo);
	if (!tsg)
		return NULL;

	tsg->g = g;
	tsg->num_active_channels = 0;
	nvgpu_ref_init(&tsg->refcount);

	tsg->tsg_gr_ctx = NULL;
	tsg->vm = NULL;
	tsg->interleave_level = NVGPU_RUNLIST_INTERLEAVE_LEVEL_LOW;
	tsg->timeslice_us = 0;
	tsg->timeslice_timeout = 0;
	tsg->timeslice_scale = 0;
	tsg->runlist_id = ~0;
	tsg->tgid = current->tgid;

	if (g->ops.fifo.init_eng_method_buffers)
		g->ops.fifo.init_eng_method_buffers(g, tsg);

	if (g->ops.fifo.tsg_open) {
		err = g->ops.fifo.tsg_open(tsg);
		if (err) {
			nvgpu_err(g, "tsg %d fifo open failed %d",
				  tsg->tsgid, err);
			goto clean_up;
		}
	}

	gk20a_dbg(gpu_dbg_fn, "tsg opened %d\n", tsg->tsgid);

	gk20a_sched_ctrl_tsg_added(g, tsg);

	return tsg;

clean_up:
	nvgpu_ref_put(&tsg->refcount, gk20a_tsg_release);
	return NULL;
}

void gk20a_tsg_release(struct nvgpu_ref *ref)
{
	struct tsg_gk20a *tsg = container_of(ref, struct tsg_gk20a, refcount);
	struct gk20a *g = tsg->g;
	struct gk20a_event_id_data *event_id_data, *event_id_data_temp;

	if (tsg->tsg_gr_ctx) {
		gr_gk20a_free_tsg_gr_ctx(tsg);
		tsg->tsg_gr_ctx = NULL;
	}

	if (g->ops.fifo.deinit_eng_method_buffers)
		g->ops.fifo.deinit_eng_method_buffers(g, tsg);

	if (tsg->vm) {
		nvgpu_vm_put(tsg->vm);
		tsg->vm = NULL;
	}

	gk20a_sched_ctrl_tsg_removed(g, tsg);

	/* unhook all events created on this TSG */
	nvgpu_mutex_acquire(&tsg->event_id_list_lock);
	nvgpu_list_for_each_entry_safe(event_id_data, event_id_data_temp,
				&tsg->event_id_list,
				gk20a_event_id_data,
				event_id_node) {
		nvgpu_list_del(&event_id_data->event_id_node);
	}
	nvgpu_mutex_release(&tsg->event_id_list_lock);

	release_used_tsg(&g->fifo, tsg);

	tsg->runlist_id = ~0;

	gk20a_dbg(gpu_dbg_fn, "tsg released %d\n", tsg->tsgid);
	gk20a_put(g);
}

struct tsg_gk20a *tsg_gk20a_from_ch(struct channel_gk20a *ch)
{
	struct tsg_gk20a *tsg = NULL;

	if (gk20a_is_channel_marked_as_tsg(ch)) {
		struct gk20a *g = ch->g;
		struct fifo_gk20a *f = &g->fifo;
		tsg = &f->tsg[ch->tsgid];
	}

	return tsg;
}