summaryrefslogtreecommitdiff
path: root/drivers/gpu/host1x/dev.c
blob: d0ebb70e2fdd6e56750db8a0772b8491f8b7bf54 (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
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
// SPDX-License-Identifier: GPL-2.0-only
/*
 * Tegra host1x driver
 *
 * Copyright (c) 2010-2013, NVIDIA Corporation.
 */

#include <linux/clk.h>
#include <linux/dma-mapping.h>
#include <linux/io.h>
#include <linux/list.h>
#include <linux/module.h>
#include <linux/of_device.h>
#include <linux/of.h>
#include <linux/slab.h>

#define CREATE_TRACE_POINTS
#include <trace/events/host1x.h>
#undef CREATE_TRACE_POINTS

#include "bus.h"
#include "channel.h"
#include "debug.h"
#include "dev.h"
#include "intr.h"

#include "hw/host1x01.h"
#include "hw/host1x02.h"
#include "hw/host1x04.h"
#include "hw/host1x05.h"
#include "hw/host1x06.h"
#include "hw/host1x07.h"

void host1x_hypervisor_writel(struct host1x *host1x, u32 v, u32 r)
{
	writel(v, host1x->hv_regs + r);
}

u32 host1x_hypervisor_readl(struct host1x *host1x, u32 r)
{
	return readl(host1x->hv_regs + r);
}

void host1x_sync_writel(struct host1x *host1x, u32 v, u32 r)
{
	void __iomem *sync_regs = host1x->regs + host1x->info->sync_offset;

	writel(v, sync_regs + r);
}

u32 host1x_sync_readl(struct host1x *host1x, u32 r)
{
	void __iomem *sync_regs = host1x->regs + host1x->info->sync_offset;

	return readl(sync_regs + r);
}

void host1x_ch_writel(struct host1x_channel *ch, u32 v, u32 r)
{
	writel(v, ch->regs + r);
}

u32 host1x_ch_readl(struct host1x_channel *ch, u32 r)
{
	return readl(ch->regs + r);
}

static const struct host1x_info host1x01_info = {
	.nb_channels = 8,
	.nb_pts = 32,
	.nb_mlocks = 16,
	.nb_bases = 8,
	.init = host1x01_init,
	.sync_offset = 0x3000,
	.dma_mask = DMA_BIT_MASK(32),
	.has_wide_gather = false,
	.has_hypervisor = false,
	.num_sid_entries = 0,
	.sid_table = NULL,
};

static const struct host1x_info host1x02_info = {
	.nb_channels = 9,
	.nb_pts = 32,
	.nb_mlocks = 16,
	.nb_bases = 12,
	.init = host1x02_init,
	.sync_offset = 0x3000,
	.dma_mask = DMA_BIT_MASK(32),
	.has_wide_gather = false,
	.has_hypervisor = false,
	.num_sid_entries = 0,
	.sid_table = NULL,
};

static const struct host1x_info host1x04_info = {
	.nb_channels = 12,
	.nb_pts = 192,
	.nb_mlocks = 16,
	.nb_bases = 64,
	.init = host1x04_init,
	.sync_offset = 0x2100,
	.dma_mask = DMA_BIT_MASK(34),
	.has_wide_gather = false,
	.has_hypervisor = false,
	.num_sid_entries = 0,
	.sid_table = NULL,
};

static const struct host1x_info host1x05_info = {
	.nb_channels = 14,
	.nb_pts = 192,
	.nb_mlocks = 16,
	.nb_bases = 64,
	.init = host1x05_init,
	.sync_offset = 0x2100,
	.dma_mask = DMA_BIT_MASK(34),
	.has_wide_gather = false,
	.has_hypervisor = false,
	.num_sid_entries = 0,
	.sid_table = NULL,
};

static const struct host1x_sid_entry tegra186_sid_table[] = {
	{
		/* VIC */
		.base = 0x1af0,
		.offset = 0x30,
		.limit = 0x34
	},
};

static const struct host1x_info host1x06_info = {
	.nb_channels = 63,
	.nb_pts = 576,
	.nb_mlocks = 24,
	.nb_bases = 16,
	.init = host1x06_init,
	.sync_offset = 0x0,
	.dma_mask = DMA_BIT_MASK(40),
	.has_wide_gather = true,
	.has_hypervisor = true,
	.num_sid_entries = ARRAY_SIZE(tegra186_sid_table),
	.sid_table = tegra186_sid_table,
};

static const struct host1x_sid_entry tegra194_sid_table[] = {
	{
		/* VIC */
		.base = 0x1af0,
		.offset = 0x30,
		.limit = 0x34
	},
};

static const struct host1x_info host1x07_info = {
	.nb_channels = 63,
	.nb_pts = 704,
	.nb_mlocks = 32,
	.nb_bases = 0,
	.init = host1x07_init,
	.sync_offset = 0x0,
	.dma_mask = DMA_BIT_MASK(40),
	.has_wide_gather = true,
	.has_hypervisor = true,
	.num_sid_entries = ARRAY_SIZE(tegra194_sid_table),
	.sid_table = tegra194_sid_table,
};

static const struct of_device_id host1x_of_match[] = {
	{ .compatible = "nvidia,tegra194-host1x", .data = &host1x07_info, },
	{ .compatible = "nvidia,tegra186-host1x", .data = &host1x06_info, },
	{ .compatible = "nvidia,tegra210-host1x", .data = &host1x05_info, },
	{ .compatible = "nvidia,tegra124-host1x", .data = &host1x04_info, },
	{ .compatible = "nvidia,tegra114-host1x", .data = &host1x02_info, },
	{ .compatible = "nvidia,tegra30-host1x", .data = &host1x01_info, },
	{ .compatible = "nvidia,tegra20-host1x", .data = &host1x01_info, },
	{ },
};
MODULE_DEVICE_TABLE(of, host1x_of_match);

static void host1x_setup_sid_table(struct host1x *host)
{
	const struct host1x_info *info = host->info;
	unsigned int i;

	for (i = 0; i < info->num_sid_entries; i++) {
		const struct host1x_sid_entry *entry = &info->sid_table[i];

		host1x_hypervisor_writel(host, entry->offset, entry->base);
		host1x_hypervisor_writel(host, entry->limit, entry->base + 4);
	}
}

static bool host1x_wants_iommu(struct host1x *host1x)
{
	/*
	 * If we support addressing a maximum of 32 bits of physical memory
	 * and if the host1x firewall is enabled, there's no need to enable
	 * IOMMU support. This can happen for example on Tegra20, Tegra30
	 * and Tegra114.
	 *
	 * Tegra124 and later can address up to 34 bits of physical memory and
	 * many platforms come equipped with more than 2 GiB of system memory,
	 * which requires crossing the 4 GiB boundary. But there's a catch: on
	 * SoCs before Tegra186 (i.e. Tegra124 and Tegra210), the host1x can
	 * only address up to 32 bits of memory in GATHER opcodes, which means
	 * that command buffers need to either be in the first 2 GiB of system
	 * memory (which could quickly lead to memory exhaustion), or command
	 * buffers need to be treated differently from other buffers (which is
	 * not possible with the current ABI).
	 *
	 * A third option is to use the IOMMU in these cases to make sure all
	 * buffers will be mapped into a 32-bit IOVA space that host1x can
	 * address. This allows all of the system memory to be used and works
	 * within the limitations of the host1x on these SoCs.
	 *
	 * In summary, default to enable IOMMU on Tegra124 and later. For any
	 * of the earlier SoCs, only use the IOMMU for additional safety when
	 * the host1x firewall is disabled.
	 */
	if (host1x->info->dma_mask <= DMA_BIT_MASK(32)) {
		if (IS_ENABLED(CONFIG_TEGRA_HOST1X_FIREWALL))
			return false;
	}

	return true;
}

static struct iommu_domain *host1x_iommu_attach(struct host1x *host)
{
	struct iommu_domain *domain = iommu_get_domain_for_dev(host->dev);
	int err;

	/*
	 * We may not always want to enable IOMMU support (for example if the
	 * host1x firewall is already enabled and we don't support addressing
	 * more than 32 bits of physical memory), so check for that first.
	 *
	 * Similarly, if host1x is already attached to an IOMMU (via the DMA
	 * API), don't try to attach again.
	 */
	if (!host1x_wants_iommu(host) || domain)
		return domain;

	host->group = iommu_group_get(host->dev);
	if (host->group) {
		struct iommu_domain_geometry *geometry;
		dma_addr_t start, end;
		unsigned long order;

		err = iova_cache_get();
		if (err < 0)
			goto put_group;

		host->domain = iommu_domain_alloc(&platform_bus_type);
		if (!host->domain) {
			err = -ENOMEM;
			goto put_cache;
		}

		err = iommu_attach_group(host->domain, host->group);
		if (err) {
			if (err == -ENODEV)
				err = 0;

			goto free_domain;
		}

		geometry = &host->domain->geometry;
		start = geometry->aperture_start & host->info->dma_mask;
		end = geometry->aperture_end & host->info->dma_mask;

		order = __ffs(host->domain->pgsize_bitmap);
		init_iova_domain(&host->iova, 1UL << order, start >> order);
		host->iova_end = end;

		domain = host->domain;
	}

	return domain;

free_domain:
	iommu_domain_free(host->domain);
	host->domain = NULL;
put_cache:
	iova_cache_put();
put_group:
	iommu_group_put(host->group);
	host->group = NULL;

	return ERR_PTR(err);
}

static int host1x_iommu_init(struct host1x *host)
{
	u64 mask = host->info->dma_mask;
	struct iommu_domain *domain;
	int err;

	domain = host1x_iommu_attach(host);
	if (IS_ERR(domain)) {
		err = PTR_ERR(domain);
		dev_err(host->dev, "failed to attach to IOMMU: %d\n", err);
		return err;
	}

	/*
	 * If we're not behind an IOMMU make sure we don't get push buffers
	 * that are allocated outside of the range addressable by the GATHER
	 * opcode.
	 *
	 * Newer generations of Tegra (Tegra186 and later) support a wide
	 * variant of the GATHER opcode that allows addressing more bits.
	 */
	if (!domain && !host->info->has_wide_gather)
		mask = DMA_BIT_MASK(32);

	err = dma_coerce_mask_and_coherent(host->dev, mask);
	if (err < 0) {
		dev_err(host->dev, "failed to set DMA mask: %d\n", err);
		return err;
	}

	return 0;
}

static void host1x_iommu_exit(struct host1x *host)
{
	if (host->domain) {
		put_iova_domain(&host->iova);
		iommu_detach_group(host->domain, host->group);

		iommu_domain_free(host->domain);
		host->domain = NULL;

		iova_cache_put();

		iommu_group_put(host->group);
		host->group = NULL;
	}
}

static int host1x_probe(struct platform_device *pdev)
{
	struct host1x *host;
	struct resource *regs, *hv_regs = NULL;
	int syncpt_irq;
	int err;

	host = devm_kzalloc(&pdev->dev, sizeof(*host), GFP_KERNEL);
	if (!host)
		return -ENOMEM;

	host->info = of_device_get_match_data(&pdev->dev);

	if (host->info->has_hypervisor) {
		regs = platform_get_resource_byname(pdev, IORESOURCE_MEM, "vm");
		if (!regs) {
			dev_err(&pdev->dev, "failed to get vm registers\n");
			return -ENXIO;
		}

		hv_regs = platform_get_resource_byname(pdev, IORESOURCE_MEM,
						       "hypervisor");
		if (!hv_regs) {
			dev_err(&pdev->dev,
				"failed to get hypervisor registers\n");
			return -ENXIO;
		}
	} else {
		regs = platform_get_resource(pdev, IORESOURCE_MEM, 0);
		if (!regs) {
			dev_err(&pdev->dev, "failed to get registers\n");
			return -ENXIO;
		}
	}

	syncpt_irq = platform_get_irq(pdev, 0);
	if (syncpt_irq < 0)
		return syncpt_irq;

	mutex_init(&host->devices_lock);
	INIT_LIST_HEAD(&host->devices);
	INIT_LIST_HEAD(&host->list);
	host->dev = &pdev->dev;

	/* set common host1x device data */
	platform_set_drvdata(pdev, host);

	host->regs = devm_ioremap_resource(&pdev->dev, regs);
	if (IS_ERR(host->regs))
		return PTR_ERR(host->regs);

	if (host->info->has_hypervisor) {
		host->hv_regs = devm_ioremap_resource(&pdev->dev, hv_regs);
		if (IS_ERR(host->hv_regs))
			return PTR_ERR(host->hv_regs);
	}

	host->dev->dma_parms = &host->dma_parms;
	dma_set_max_seg_size(host->dev, UINT_MAX);

	if (host->info->init) {
		err = host->info->init(host);
		if (err)
			return err;
	}

	host->clk = devm_clk_get(&pdev->dev, NULL);
	if (IS_ERR(host->clk)) {
		err = PTR_ERR(host->clk);

		if (err != -EPROBE_DEFER)
			dev_err(&pdev->dev, "failed to get clock: %d\n", err);

		return err;
	}

	host->rst = devm_reset_control_get(&pdev->dev, "host1x");
	if (IS_ERR(host->rst)) {
		err = PTR_ERR(host->rst);
		dev_err(&pdev->dev, "failed to get reset: %d\n", err);
		return err;
	}

	err = host1x_iommu_init(host);
	if (err < 0) {
		dev_err(&pdev->dev, "failed to setup IOMMU: %d\n", err);
		return err;
	}

	err = host1x_channel_list_init(&host->channel_list,
				       host->info->nb_channels);
	if (err) {
		dev_err(&pdev->dev, "failed to initialize channel list\n");
		goto iommu_exit;
	}

	err = clk_prepare_enable(host->clk);
	if (err < 0) {
		dev_err(&pdev->dev, "failed to enable clock\n");
		goto free_channels;
	}

	err = reset_control_deassert(host->rst);
	if (err < 0) {
		dev_err(&pdev->dev, "failed to deassert reset: %d\n", err);
		goto unprepare_disable;
	}

	err = host1x_syncpt_init(host);
	if (err) {
		dev_err(&pdev->dev, "failed to initialize syncpts\n");
		goto reset_assert;
	}

	err = host1x_intr_init(host, syncpt_irq);
	if (err) {
		dev_err(&pdev->dev, "failed to initialize interrupts\n");
		goto deinit_syncpt;
	}

	host1x_debug_init(host);

	if (host->info->has_hypervisor)
		host1x_setup_sid_table(host);

	err = host1x_register(host);
	if (err < 0)
		goto deinit_debugfs;

	err = devm_of_platform_populate(&pdev->dev);
	if (err < 0)
		goto unregister;

	return 0;

unregister:
	host1x_unregister(host);
deinit_debugfs:
	host1x_debug_deinit(host);
	host1x_intr_deinit(host);
deinit_syncpt:
	host1x_syncpt_deinit(host);
reset_assert:
	reset_control_assert(host->rst);
unprepare_disable:
	clk_disable_unprepare(host->clk);
free_channels:
	host1x_channel_list_free(&host->channel_list);
iommu_exit:
	host1x_iommu_exit(host);

	return err;
}

static int host1x_remove(struct platform_device *pdev)
{
	struct host1x *host = platform_get_drvdata(pdev);

	host1x_unregister(host);
	host1x_debug_deinit(host);
	host1x_intr_deinit(host);
	host1x_syncpt_deinit(host);
	reset_control_assert(host->rst);
	clk_disable_unprepare(host->clk);
	host1x_iommu_exit(host);

	return 0;
}

static struct platform_driver tegra_host1x_driver = {
	.driver = {
		.name = "tegra-host1x",
		.of_match_table = host1x_of_match,
	},
	.probe = host1x_probe,
	.remove = host1x_remove,
};

static struct platform_driver * const drivers[] = {
	&tegra_host1x_driver,
	&tegra_mipi_driver,
};

static int __init tegra_host1x_init(void)
{
	int err;

	err = bus_register(&host1x_bus_type);
	if (err < 0)
		return err;

	err = platform_register_drivers(drivers, ARRAY_SIZE(drivers));
	if (err < 0)
		bus_unregister(&host1x_bus_type);

	return err;
}
module_init(tegra_host1x_init);

static void __exit tegra_host1x_exit(void)
{
	platform_unregister_drivers(drivers, ARRAY_SIZE(drivers));
	bus_unregister(&host1x_bus_type);
}
module_exit(tegra_host1x_exit);

/**
 * host1x_get_dma_mask() - query the supported DMA mask for host1x
 * @host1x: host1x instance
 *
 * Note that this returns the supported DMA mask for host1x, which can be
 * different from the applicable DMA mask under certain circumstances.
 */
u64 host1x_get_dma_mask(struct host1x *host1x)
{
	return host1x->info->dma_mask;
}
EXPORT_SYMBOL(host1x_get_dma_mask);

MODULE_AUTHOR("Thierry Reding <thierry.reding@avionic-design.de>");
MODULE_AUTHOR("Terje Bergstrom <tbergstrom@nvidia.com>");
MODULE_DESCRIPTION("Host1x driver for Tegra products");
MODULE_LICENSE("GPL");