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
|
/*
* GP10B Tegra Platform Interface
*
* Copyright (c) 2014-2018, 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.
*/
#include <linux/of_platform.h>
#include <linux/debugfs.h>
#include <linux/dma-buf.h>
#include <linux/nvmap.h>
#include <linux/reset.h>
#include <linux/platform/tegra/emc_bwmgr.h>
#include <uapi/linux/nvgpu.h>
#include <soc/tegra/tegra_bpmp.h>
#include <soc/tegra/tegra_powergate.h>
#include <soc/tegra/tegra-bpmp-dvfs.h>
#include <dt-bindings/memory/tegra-swgroup.h>
#include <nvgpu/kmem.h>
#include <nvgpu/bug.h>
#include <nvgpu/enabled.h>
#include <nvgpu/hashtable.h>
#include <nvgpu/nvhost.h>
#include "os_linux.h"
#include "clk.h"
#include "gk20a/gk20a.h"
#include "platform_gk20a.h"
#include "platform_ecc_sysfs.h"
#include "platform_gk20a_tegra.h"
#include "platform_gp10b.h"
#include "platform_gp10b_tegra.h"
#include "scale.h"
/* Select every GP10B_FREQ_SELECT_STEP'th frequency from h/w table */
#define GP10B_FREQ_SELECT_STEP 8
/* Allow limited set of frequencies to be available */
#define GP10B_NUM_SUPPORTED_FREQS 15
/* Max number of freq supported in h/w */
#define GP10B_MAX_SUPPORTED_FREQS 120
static unsigned long
gp10b_freq_table[GP10B_MAX_SUPPORTED_FREQS / GP10B_FREQ_SELECT_STEP];
#define TEGRA_GP10B_BW_PER_FREQ 64
#define TEGRA_DDR4_BW_PER_FREQ 16
#define EMC_BW_RATIO (TEGRA_GP10B_BW_PER_FREQ / TEGRA_DDR4_BW_PER_FREQ)
#define GPCCLK_INIT_RATE 1000000000
static struct {
char *name;
unsigned long default_rate;
} tegra_gp10b_clocks[] = {
{"gpu", GPCCLK_INIT_RATE},
{"gpu_sys", 204000000} };
/*
* gp10b_tegra_get_clocks()
*
* This function finds clocks in tegra platform and populates
* the clock information to gp10b platform data.
*/
int gp10b_tegra_get_clocks(struct device *dev)
{
struct gk20a_platform *platform = dev_get_drvdata(dev);
unsigned int i;
platform->num_clks = 0;
for (i = 0; i < ARRAY_SIZE(tegra_gp10b_clocks); i++) {
long rate = tegra_gp10b_clocks[i].default_rate;
struct clk *c;
c = clk_get(dev, tegra_gp10b_clocks[i].name);
if (IS_ERR(c)) {
nvgpu_err(platform->g, "cannot get clock %s",
tegra_gp10b_clocks[i].name);
} else {
clk_set_rate(c, rate);
platform->clk[i] = c;
if (i == 0)
platform->cached_rate = rate;
}
}
platform->num_clks = i;
if (platform->clk[0]) {
i = tegra_bpmp_dvfs_get_clk_id(dev->of_node,
tegra_gp10b_clocks[0].name);
if (i > 0)
platform->maxmin_clk_id = i;
}
return 0;
}
void gp10b_tegra_scale_init(struct device *dev)
{
struct gk20a_platform *platform = gk20a_get_platform(dev);
struct gk20a_scale_profile *profile = platform->g->scale_profile;
struct tegra_bwmgr_client *bwmgr_handle;
if (!profile)
return;
if ((struct tegra_bwmgr_client *)profile->private_data)
return;
bwmgr_handle = tegra_bwmgr_register(TEGRA_BWMGR_CLIENT_GPU);
if (!bwmgr_handle)
return;
profile->private_data = (void *)bwmgr_handle;
}
static void gp10b_tegra_scale_exit(struct device *dev)
{
struct gk20a_platform *platform = gk20a_get_platform(dev);
struct gk20a_scale_profile *profile = platform->g->scale_profile;
if (profile && profile->private_data)
tegra_bwmgr_unregister(
(struct tegra_bwmgr_client *)profile->private_data);
}
static int gp10b_tegra_probe(struct device *dev)
{
struct gk20a_platform *platform = dev_get_drvdata(dev);
#ifdef CONFIG_TEGRA_GK20A_NVHOST
int ret;
ret = nvgpu_get_nvhost_dev(platform->g);
if (ret)
return ret;
#endif
ret = gk20a_tegra_init_secure_alloc(platform);
if (ret)
return ret;
platform->disable_bigpage = !device_is_iommuable(dev);
platform->g->gr.ctx_vars.dump_ctxsw_stats_on_channel_close
= false;
platform->g->gr.ctx_vars.dump_ctxsw_stats_on_channel_close
= false;
platform->g->gr.ctx_vars.force_preemption_gfxp = false;
platform->g->gr.ctx_vars.force_preemption_cilp = false;
gp10b_tegra_get_clocks(dev);
nvgpu_linux_init_clk_support(platform->g);
return 0;
}
static int gp10b_tegra_late_probe(struct device *dev)
{
return 0;
}
static int gp10b_tegra_remove(struct device *dev)
{
struct gk20a *g = get_gk20a(dev);
if (g->ops.gr.remove_gr_sysfs)
g->ops.gr.remove_gr_sysfs(g);
/* deinitialise tegra specific scaling quirks */
gp10b_tegra_scale_exit(dev);
#ifdef CONFIG_TEGRA_GK20A_NVHOST
nvgpu_free_nvhost_dev(get_gk20a(dev));
#endif
return 0;
}
static bool gp10b_tegra_is_railgated(struct device *dev)
{
bool ret = false;
if (tegra_bpmp_running())
ret = !tegra_powergate_is_powered(TEGRA186_POWER_DOMAIN_GPU);
return ret;
}
static int gp10b_tegra_railgate(struct device *dev)
{
struct gk20a_platform *platform = gk20a_get_platform(dev);
struct gk20a_scale_profile *profile = platform->g->scale_profile;
/* remove emc frequency floor */
if (profile)
tegra_bwmgr_set_emc(
(struct tegra_bwmgr_client *)profile->private_data,
0, TEGRA_BWMGR_SET_EMC_FLOOR);
if (tegra_bpmp_running() &&
tegra_powergate_is_powered(TEGRA186_POWER_DOMAIN_GPU)) {
int i;
for (i = 0; i < platform->num_clks; i++) {
if (platform->clk[i])
clk_disable_unprepare(platform->clk[i]);
}
tegra_powergate_partition(TEGRA186_POWER_DOMAIN_GPU);
}
return 0;
}
static int gp10b_tegra_unrailgate(struct device *dev)
{
int ret = 0;
struct gk20a_platform *platform = gk20a_get_platform(dev);
struct gk20a_scale_profile *profile = platform->g->scale_profile;
if (tegra_bpmp_running()) {
int i;
ret = tegra_unpowergate_partition(TEGRA186_POWER_DOMAIN_GPU);
for (i = 0; i < platform->num_clks; i++) {
if (platform->clk[i])
clk_prepare_enable(platform->clk[i]);
}
}
/* to start with set emc frequency floor to max rate*/
if (profile)
tegra_bwmgr_set_emc(
(struct tegra_bwmgr_client *)profile->private_data,
tegra_bwmgr_get_max_emc_rate(),
TEGRA_BWMGR_SET_EMC_FLOOR);
return ret;
}
static int gp10b_tegra_suspend(struct device *dev)
{
return 0;
}
int gp10b_tegra_reset_assert(struct device *dev)
{
struct gk20a_platform *platform = gk20a_get_platform(dev);
int ret = 0;
if (!platform->reset_control)
return -EINVAL;
ret = reset_control_assert(platform->reset_control);
return ret;
}
int gp10b_tegra_reset_deassert(struct device *dev)
{
struct gk20a_platform *platform = gk20a_get_platform(dev);
int ret = 0;
if (!platform->reset_control)
return -EINVAL;
ret = reset_control_deassert(platform->reset_control);
return ret;
}
void gp10b_tegra_prescale(struct device *dev)
{
struct gk20a *g = get_gk20a(dev);
u32 avg = 0;
nvgpu_log_fn(g, " ");
nvgpu_pmu_load_norm(g, &avg);
nvgpu_log_fn(g, "done");
}
void gp10b_tegra_postscale(struct device *pdev,
unsigned long freq)
{
struct gk20a_platform *platform = gk20a_get_platform(pdev);
struct gk20a_scale_profile *profile = platform->g->scale_profile;
struct gk20a *g = get_gk20a(pdev);
unsigned long emc_rate;
nvgpu_log_fn(g, " ");
if (profile && profile->private_data &&
!platform->is_railgated(pdev)) {
unsigned long emc_scale;
if (freq <= gp10b_freq_table[0])
emc_scale = 0;
else
emc_scale = g->emc3d_ratio;
emc_rate = (freq * EMC_BW_RATIO * emc_scale) / 1000;
if (emc_rate > tegra_bwmgr_get_max_emc_rate())
emc_rate = tegra_bwmgr_get_max_emc_rate();
tegra_bwmgr_set_emc(
(struct tegra_bwmgr_client *)profile->private_data,
emc_rate, TEGRA_BWMGR_SET_EMC_FLOOR);
}
nvgpu_log_fn(g, "done");
}
long gp10b_round_clk_rate(struct device *dev, unsigned long rate)
{
struct gk20a *g = get_gk20a(dev);
struct gk20a_scale_profile *profile = g->scale_profile;
unsigned long *freq_table = profile->devfreq_profile.freq_table;
int max_states = profile->devfreq_profile.max_state;
int i;
for (i = 0; i < max_states; ++i)
if (freq_table[i] >= rate)
return freq_table[i];
return freq_table[max_states - 1];
}
int gp10b_clk_get_freqs(struct device *dev,
unsigned long **freqs, int *num_freqs)
{
struct gk20a_platform *platform = gk20a_get_platform(dev);
struct gk20a *g = platform->g;
unsigned long max_rate;
unsigned long new_rate = 0, prev_rate = 0;
int i, freq_counter = 0;
int sel_freq_cnt;
unsigned long loc_freq_table[GP10B_MAX_SUPPORTED_FREQS];
max_rate = clk_round_rate(platform->clk[0], (UINT_MAX - 1));
/*
* Walk the h/w frequency table and update the local table
*/
for (i = 0; i < GP10B_MAX_SUPPORTED_FREQS; ++i) {
prev_rate = new_rate;
new_rate = clk_round_rate(platform->clk[0],
prev_rate + 1);
loc_freq_table[i] = new_rate;
if (new_rate == max_rate)
break;
}
freq_counter = i + 1;
WARN_ON(freq_counter == GP10B_MAX_SUPPORTED_FREQS);
/*
* If the number of achievable frequencies is less than or
* equal to GP10B_NUM_SUPPORTED_FREQS, select all frequencies
* else, select one out of every 8 frequencies
*/
if (freq_counter <= GP10B_NUM_SUPPORTED_FREQS) {
for (sel_freq_cnt = 0; sel_freq_cnt < freq_counter; ++sel_freq_cnt)
gp10b_freq_table[sel_freq_cnt] =
loc_freq_table[sel_freq_cnt];
} else {
/*
* Walk the h/w frequency table and only select
* GP10B_FREQ_SELECT_STEP'th frequencies and
* add MAX freq to last
*/
sel_freq_cnt = 0;
for (i = 0; i < GP10B_MAX_SUPPORTED_FREQS; ++i) {
new_rate = loc_freq_table[i];
if (i % GP10B_FREQ_SELECT_STEP == 0 ||
new_rate == max_rate) {
gp10b_freq_table[sel_freq_cnt++] =
new_rate;
if (new_rate == max_rate)
break;
}
}
WARN_ON(sel_freq_cnt == GP10B_MAX_SUPPORTED_FREQS);
}
/* Fill freq table */
*freqs = gp10b_freq_table;
*num_freqs = sel_freq_cnt;
nvgpu_log_info(g, "min rate: %ld max rate: %ld num_of_freq %d\n",
gp10b_freq_table[0], max_rate, *num_freqs);
return 0;
}
struct gk20a_platform gp10b_tegra_platform = {
.has_syncpoints = true,
/* power management configuration */
.railgate_delay_init = 500,
/* ldiv slowdown factor */
.ldiv_slowdown_factor_init = SLOWDOWN_FACTOR_FPDIV_BY16,
/* power management configuration */
.can_railgate_init = true,
.enable_elpg = true,
.can_elpg_init = true,
.enable_blcg = true,
.enable_slcg = true,
.enable_elcg = true,
.can_slcg = true,
.can_blcg = true,
.can_elcg = true,
.enable_aelpg = true,
.enable_perfmon = true,
/* ptimer src frequency in hz*/
.ptimer_src_freq = 31250000,
.ch_wdt_timeout_ms = 5000,
.probe = gp10b_tegra_probe,
.late_probe = gp10b_tegra_late_probe,
.remove = gp10b_tegra_remove,
/* power management callbacks */
.suspend = gp10b_tegra_suspend,
.railgate = gp10b_tegra_railgate,
.unrailgate = gp10b_tegra_unrailgate,
.is_railgated = gp10b_tegra_is_railgated,
.busy = gk20a_tegra_busy,
.idle = gk20a_tegra_idle,
.dump_platform_dependencies = gk20a_tegra_debug_dump,
#ifdef CONFIG_NVGPU_SUPPORT_CDE
.has_cde = true,
#endif
.clk_round_rate = gp10b_round_clk_rate,
.get_clk_freqs = gp10b_clk_get_freqs,
/* frequency scaling configuration */
.initscale = gp10b_tegra_scale_init,
.prescale = gp10b_tegra_prescale,
.postscale = gp10b_tegra_postscale,
.devfreq_governor = "nvhost_podgov",
.qos_notify = gk20a_scale_qos_notify,
.reset_assert = gp10b_tegra_reset_assert,
.reset_deassert = gp10b_tegra_reset_deassert,
.force_reset_in_do_idle = false,
.soc_name = "tegra18x",
.unified_memory = true,
.dma_mask = DMA_BIT_MASK(36),
.ltc_streamid = TEGRA_SID_GPUB,
.secure_buffer_size = 401408,
};
void gr_gp10b_create_sysfs(struct gk20a *g)
{
int error = 0;
struct device *dev = dev_from_gk20a(g);
/* This stat creation function is called on GR init. GR can get
initialized multiple times but we only need to create the ECC
stats once. Therefore, add the following check to avoid
creating duplicate stat sysfs nodes. */
if (g->ecc.gr.sm_lrf_single_err_count.counters != NULL)
return;
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"sm_lrf_ecc_single_err_count",
&g->ecc.gr.sm_lrf_single_err_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"sm_lrf_ecc_double_err_count",
&g->ecc.gr.sm_lrf_double_err_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"sm_shm_ecc_sec_count",
&g->ecc.gr.sm_shm_sec_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"sm_shm_ecc_sed_count",
&g->ecc.gr.sm_shm_sed_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"sm_shm_ecc_ded_count",
&g->ecc.gr.sm_shm_ded_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"tex_ecc_total_sec_pipe0_count",
&g->ecc.gr.tex_total_sec_pipe0_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"tex_ecc_total_ded_pipe0_count",
&g->ecc.gr.tex_total_ded_pipe0_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"tex_ecc_unique_sec_pipe0_count",
&g->ecc.gr.tex_unique_sec_pipe0_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"tex_ecc_unique_ded_pipe0_count",
&g->ecc.gr.tex_unique_ded_pipe0_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"tex_ecc_total_sec_pipe1_count",
&g->ecc.gr.tex_total_sec_pipe1_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"tex_ecc_total_ded_pipe1_count",
&g->ecc.gr.tex_total_ded_pipe1_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"tex_ecc_unique_sec_pipe1_count",
&g->ecc.gr.tex_unique_sec_pipe1_count);
error |= nvgpu_gr_ecc_stat_create(dev,
0,
"tex_ecc_unique_ded_pipe1_count",
&g->ecc.gr.tex_unique_ded_pipe1_count);
error |= nvgpu_gr_ecc_stat_create(dev,
2,
"ecc_sec_count",
&g->ecc.ltc.l2_sec_count);
error |= nvgpu_gr_ecc_stat_create(dev,
2,
"ecc_ded_count",
&g->ecc.ltc.l2_ded_count);
if (error)
dev_err(dev, "Failed to create sysfs attributes!\n");
}
void gr_gp10b_remove_sysfs(struct gk20a *g)
{
struct device *dev = dev_from_gk20a(g);
if (!g->ecc.gr.sm_lrf_single_err_count.counters)
return;
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.sm_lrf_single_err_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.sm_lrf_double_err_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.sm_shm_sec_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.sm_shm_sed_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.sm_shm_ded_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.tex_total_sec_pipe0_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.tex_total_ded_pipe0_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.tex_unique_sec_pipe0_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.tex_unique_ded_pipe0_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.tex_total_sec_pipe1_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.tex_total_ded_pipe1_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.tex_unique_sec_pipe1_count);
nvgpu_gr_ecc_stat_remove(dev,
0,
&g->ecc.gr.tex_unique_ded_pipe1_count);
nvgpu_gr_ecc_stat_remove(dev,
2,
&g->ecc.ltc.l2_sec_count);
nvgpu_gr_ecc_stat_remove(dev,
2,
&g->ecc.ltc.l2_ded_count);
}
|