summaryrefslogtreecommitdiff
path: root/lib/media_fill_gen7.c
blob: 7113fda15a11189e1204c121a9e99f8951a17939 (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
#include <intel_bufmgr.h>
#include <i915_drm.h>

#include "media_fill.h"
#include "gen7_media.h"
#include "intel_reg.h"
#include "drmtest.h"

#include <assert.h>

static const uint32_t media_kernel[][4] = {
	{ 0x00400001, 0x20200231, 0x00000020, 0x00000000 },
	{ 0x00600001, 0x20800021, 0x008d0000, 0x00000000 },
	{ 0x00200001, 0x20800021, 0x00450040, 0x00000000 },
	{ 0x00000001, 0x20880061, 0x00000000, 0x000f000f },
	{ 0x00800001, 0x20a00021, 0x00000020, 0x00000000 },
	{ 0x00800001, 0x20e00021, 0x00000020, 0x00000000 },
	{ 0x00800001, 0x21200021, 0x00000020, 0x00000000 },
	{ 0x00800001, 0x21600021, 0x00000020, 0x00000000 },
	{ 0x05800031, 0x24001ca8, 0x00000080, 0x120a8000 },
	{ 0x00600001, 0x2e000021, 0x008d0000, 0x00000000 },
	{ 0x07800031, 0x20001ca8, 0x00000e00, 0x82000010 },
};

/* shaders/gpgpu/gpgpu_fill.gxa */
static const uint32_t gpgpu_kernel[][4] = {
	{ 0x00400001, 0x20200231, 0x00000020, 0x00000000 },
	{ 0x00000041, 0x20400c21, 0x00000004, 0x00000010 },
	{ 0x00000001, 0x20440021, 0x00000018, 0x00000000 },
	{ 0x00600001, 0x20800021, 0x008d0000, 0x00000000 },
	{ 0x00200001, 0x20800021, 0x00450040, 0x00000000 },
	{ 0x00000001, 0x20880061, 0x00000000, 0x0000000f },
	{ 0x00800001, 0x20a00021, 0x00000020, 0x00000000 },
	{ 0x00800001, 0x20e00021, 0x00000020, 0x00000000 },
	{ 0x00800001, 0x21200021, 0x00000020, 0x00000000 },
	{ 0x00800001, 0x21600021, 0x00000020, 0x00000000 },
	{ 0x05800031, 0x24001ca8, 0x00000080, 0x120a8000 },
	{ 0x00600001, 0x2e000021, 0x008d0000, 0x00000000 },
	{ 0x07800031, 0x20001ca8, 0x00000e00, 0x82000010 },
};

static uint32_t
batch_used(struct intel_batchbuffer *batch)
{
	return batch->ptr - batch->buffer;
}

static uint32_t
batch_align(struct intel_batchbuffer *batch, uint32_t align)
{
	uint32_t offset = batch_used(batch);
	offset = ALIGN(offset, align);
	batch->ptr = batch->buffer + offset;
	return offset;
}

static void *
batch_alloc(struct intel_batchbuffer *batch, uint32_t size, uint32_t align)
{
	uint32_t offset = batch_align(batch, align);
	batch->ptr += size;
	return memset(batch->buffer + offset, 0, size);
}

static uint32_t
batch_offset(struct intel_batchbuffer *batch, void *ptr)
{
	return (uint8_t *)ptr - batch->buffer;
}

static uint32_t
batch_copy(struct intel_batchbuffer *batch, const void *ptr, uint32_t size, uint32_t align)
{
	return batch_offset(batch, memcpy(batch_alloc(batch, size, align), ptr, size));
}

static void
gen7_render_flush(struct intel_batchbuffer *batch, uint32_t batch_end)
{
	int ret;

	ret = drm_intel_bo_subdata(batch->bo, 0, 4096, batch->buffer);
	if (ret == 0)
		ret = drm_intel_bo_mrb_exec(batch->bo, batch_end,
					NULL, 0, 0, 0);
	igt_assert(ret == 0);
}

static uint32_t
gen7_fill_curbe_buffer_data(struct intel_batchbuffer *batch,
			uint8_t color)
{
	uint8_t *curbe_buffer;
	uint32_t offset;

	curbe_buffer = batch_alloc(batch, sizeof(uint32_t) * 8, 64);
	offset = batch_offset(batch, curbe_buffer);
	*curbe_buffer = color;

	return offset;
}

static uint32_t
gen7_fill_surface_state(struct intel_batchbuffer *batch,
			struct igt_buf *buf,
			uint32_t format,
			int is_dst)
{
	struct gen7_surface_state *ss;
	uint32_t write_domain, read_domain, offset;
	int ret;

	if (is_dst) {
		write_domain = read_domain = I915_GEM_DOMAIN_RENDER;
	} else {
		write_domain = 0;
		read_domain = I915_GEM_DOMAIN_SAMPLER;
	}

	ss = batch_alloc(batch, sizeof(*ss), 64);
	offset = batch_offset(batch, ss);

	ss->ss0.surface_type = GEN7_SURFACE_2D;
	ss->ss0.surface_format = format;
	ss->ss0.render_cache_read_write = 1;

	if (buf->tiling == I915_TILING_X)
		ss->ss0.tiled_mode = 2;
	else if (buf->tiling == I915_TILING_Y)
		ss->ss0.tiled_mode = 3;

	ss->ss1.base_addr = buf->bo->offset;
	ret = drm_intel_bo_emit_reloc(batch->bo,
				batch_offset(batch, ss) + 4,
				buf->bo, 0,
				read_domain, write_domain);
	igt_assert(ret == 0);

	ss->ss2.height = igt_buf_height(buf) - 1;
	ss->ss2.width  = igt_buf_width(buf) - 1;

	ss->ss3.pitch  = buf->stride - 1;

	ss->ss7.shader_chanel_select_r = 4;
	ss->ss7.shader_chanel_select_g = 5;
	ss->ss7.shader_chanel_select_b = 6;
	ss->ss7.shader_chanel_select_a = 7;

	return offset;
}

static uint32_t
gen7_fill_binding_table(struct intel_batchbuffer *batch,
			struct igt_buf *dst)
{
	uint32_t *binding_table, offset;

	binding_table = batch_alloc(batch, 32, 64);
	offset = batch_offset(batch, binding_table);

	binding_table[0] = gen7_fill_surface_state(batch, dst, GEN7_SURFACEFORMAT_R8_UNORM, 1);

	return offset;
}

static uint32_t
gen7_fill_media_kernel(struct intel_batchbuffer *batch,
		const uint32_t kernel[][4],
		size_t size)
{
	uint32_t offset;

	offset = batch_copy(batch, kernel, size, 64);

	return offset;
}

static uint32_t
gen7_fill_interface_descriptor(struct intel_batchbuffer *batch, struct igt_buf *dst,
			       const uint32_t kernel[][4], size_t size)
{
	struct gen7_interface_descriptor_data *idd;
	uint32_t offset;
	uint32_t binding_table_offset, kernel_offset;

	binding_table_offset = gen7_fill_binding_table(batch, dst);
	kernel_offset = gen7_fill_media_kernel(batch, kernel, size);

	idd = batch_alloc(batch, sizeof(*idd), 64);
	offset = batch_offset(batch, idd);

	idd->desc0.kernel_start_pointer = (kernel_offset >> 6);

	idd->desc1.single_program_flow = 1;
	idd->desc1.floating_point_mode = GEN7_FLOATING_POINT_IEEE_754;

	idd->desc2.sampler_count = 0;      /* 0 samplers used */
	idd->desc2.sampler_state_pointer = 0;

	idd->desc3.binding_table_entry_count = 0;
	idd->desc3.binding_table_pointer = (binding_table_offset >> 5);

	idd->desc4.constant_urb_entry_read_offset = 0;
	idd->desc4.constant_urb_entry_read_length = 1; /* grf 1 */

	return offset;
}

static void
gen7_emit_state_base_address(struct intel_batchbuffer *batch)
{
	OUT_BATCH(GEN7_STATE_BASE_ADDRESS | (10 - 2));

	/* general */
	OUT_BATCH(0);

	/* surface */
	OUT_RELOC(batch->bo, I915_GEM_DOMAIN_INSTRUCTION, 0, BASE_ADDRESS_MODIFY);

	/* dynamic */
	OUT_RELOC(batch->bo, I915_GEM_DOMAIN_INSTRUCTION, 0, BASE_ADDRESS_MODIFY);

	/* indirect */
	OUT_BATCH(0);

	/* instruction */
	OUT_RELOC(batch->bo, I915_GEM_DOMAIN_INSTRUCTION, 0, BASE_ADDRESS_MODIFY);

	/* general/dynamic/indirect/instruction access Bound */
	OUT_BATCH(0);
	OUT_BATCH(0 | BASE_ADDRESS_MODIFY);
	OUT_BATCH(0);
	OUT_BATCH(0 | BASE_ADDRESS_MODIFY);
}

static void
gen7_emit_vfe_state(struct intel_batchbuffer *batch)
{
	OUT_BATCH(GEN7_MEDIA_VFE_STATE | (8 - 2));

	/* scratch buffer */
	OUT_BATCH(0);

	/* number of threads & urb entries */
	OUT_BATCH(1 << 16 |
		2 << 8);

	OUT_BATCH(0);

	/* urb entry size & curbe size */
	OUT_BATCH(2 << 16 | 	/* in 256 bits unit */
		2);		/* in 256 bits unit */

	/* scoreboard */
	OUT_BATCH(0);
	OUT_BATCH(0);
	OUT_BATCH(0);
}

static void
gen7_emit_curbe_load(struct intel_batchbuffer *batch, uint32_t curbe_buffer)
{
	OUT_BATCH(GEN7_MEDIA_CURBE_LOAD | (4 - 2));
	OUT_BATCH(0);
	/* curbe total data length */
	OUT_BATCH(64);
	/* curbe data start address, is relative to the dynamics base address */
	OUT_BATCH(curbe_buffer);
}

static void
gen7_emit_interface_descriptor_load(struct intel_batchbuffer *batch, uint32_t interface_descriptor)
{
	OUT_BATCH(GEN7_MEDIA_INTERFACE_DESCRIPTOR_LOAD | (4 - 2));
	OUT_BATCH(0);
	/* interface descriptor data length */
	OUT_BATCH(sizeof(struct gen7_interface_descriptor_data));
	/* interface descriptor address, is relative to the dynamics base address */
	OUT_BATCH(interface_descriptor);
}

static void
gen7_emit_media_objects(struct intel_batchbuffer *batch,
			unsigned x, unsigned y,
			unsigned width, unsigned height)
{
	int i, j;

	for (i = 0; i < width / 16; i++) {
		for (j = 0; j < height / 16; j++) {
			OUT_BATCH(GEN7_MEDIA_OBJECT | (8 - 2));

			/* interface descriptor offset */
			OUT_BATCH(0);

			/* without indirect data */
			OUT_BATCH(0);
			OUT_BATCH(0);

			/* scoreboard */
			OUT_BATCH(0);
			OUT_BATCH(0);

			/* inline data (xoffset, yoffset) */
			OUT_BATCH(x + i * 16);
			OUT_BATCH(y + j * 16);
		}
	}
}

/*
 * This sets up the media pipeline,
 *
 * +---------------+ <---- 4096
 * |       ^       |
 * |       |       |
 * |    various    |
 * |      state    |
 * |       |       |
 * |_______|_______| <---- 2048 + ?
 * |       ^       |
 * |       |       |
 * |   batch       |
 * |    commands   |
 * |       |       |
 * |       |       |
 * +---------------+ <---- 0 + ?
 *
 */

#define BATCH_STATE_SPLIT 2048

void
gen7_media_fillfunc(struct intel_batchbuffer *batch,
		struct igt_buf *dst,
		unsigned x, unsigned y,
		unsigned width, unsigned height,
		uint8_t color)
{
	uint32_t curbe_buffer, interface_descriptor;
	uint32_t batch_end;

	intel_batchbuffer_flush(batch);

	/* setup states */
	batch->ptr = &batch->buffer[BATCH_STATE_SPLIT];

	curbe_buffer = gen7_fill_curbe_buffer_data(batch, color);
	interface_descriptor = gen7_fill_interface_descriptor(batch, dst,
							      media_kernel,
							      sizeof(media_kernel));
	igt_assert(batch->ptr < &batch->buffer[4095]);

	/* media pipeline */
	batch->ptr = batch->buffer;
	OUT_BATCH(GEN7_PIPELINE_SELECT | PIPELINE_SELECT_MEDIA);
	gen7_emit_state_base_address(batch);

	gen7_emit_vfe_state(batch);

	gen7_emit_curbe_load(batch, curbe_buffer);

	gen7_emit_interface_descriptor_load(batch, interface_descriptor);

	gen7_emit_media_objects(batch, x, y, width, height);

	OUT_BATCH(MI_BATCH_BUFFER_END);

	batch_end = batch_align(batch, 8);
	igt_assert(batch_end < BATCH_STATE_SPLIT);

	gen7_render_flush(batch, batch_end);
	intel_batchbuffer_reset(batch);
}

static void
gen7_emit_vfe_state_gpgpu(struct intel_batchbuffer *batch)
{
	OUT_BATCH(GEN7_MEDIA_VFE_STATE | (8 - 2));

	/* scratch buffer */
	OUT_BATCH(0);

	/* number of threads & urb entries */
	OUT_BATCH(1 << 16 | /* max num of threads */
		  0 << 8 | /* num of URB entry */
		  1 << 2); /* GPGPU mode */

	OUT_BATCH(0);

	/* urb entry size & curbe size */
	OUT_BATCH(0 << 16 | 	/* URB entry size in 256 bits unit */
		  1);		/* CURBE entry size in 256 bits unit */

	/* scoreboard */
	OUT_BATCH(0);
	OUT_BATCH(0);
	OUT_BATCH(0);
}

static void
gen7_emit_gpgpu_walk(struct intel_batchbuffer *batch,
		     unsigned x, unsigned y,
		     unsigned width, unsigned height)
{
	uint32_t x_dim, y_dim, tmp, right_mask;

	/*
	 * Simply do SIMD16 based dispatch, so every thread uses
	 * SIMD16 channels.
	 *
	 * Define our own thread group size, e.g 16x1 for every group, then
	 * will have 1 thread each group in SIMD16 dispatch. So thread
	 * width/height/depth are all 1.
	 *
	 * Then thread group X = width / 16 (aligned to 16)
	 * thread group Y = height;
	 */
	x_dim = (width + 15) / 16;
	y_dim = height;

	tmp = width & 15;
	if (tmp == 0)
		right_mask = (1 << 16) - 1;
	else
		right_mask = (1 << tmp) - 1;

	OUT_BATCH(GEN7_GPGPU_WALKER | 9);

	/* interface descriptor offset */
	OUT_BATCH(0);

	/* SIMD size, thread w/h/d */
	OUT_BATCH(1 << 30 | /* SIMD16 */
		  0 << 16 | /* depth:1 */
		  0 << 8 | /* height:1 */
		  0); /* width:1 */

	/* thread group X */
	OUT_BATCH(0);
	OUT_BATCH(x_dim);

	/* thread group Y */
	OUT_BATCH(0);
	OUT_BATCH(y_dim);

	/* thread group Z */
	OUT_BATCH(0);
	OUT_BATCH(1);

	/* right mask */
	OUT_BATCH(right_mask);

	/* bottom mask, height 1, always 0xffffffff */
	OUT_BATCH(0xffffffff);
}

void
gen7_gpgpu_fillfunc(struct intel_batchbuffer *batch,
		    struct igt_buf *dst,
		    unsigned x, unsigned y,
		    unsigned width, unsigned height,
		    uint8_t color)
{
	uint32_t curbe_buffer, interface_descriptor;
	uint32_t batch_end;

	intel_batchbuffer_flush(batch);

	/* setup states */
	batch->ptr = &batch->buffer[BATCH_STATE_SPLIT];

	/*
	 * const buffer needs to fill for every thread, but as we have just 1 thread
	 * per every group, so need only one curbe data.
	 *
	 * For each thread, just use thread group ID for buffer offset.
	 */
	curbe_buffer = gen7_fill_curbe_buffer_data(batch, color);

	interface_descriptor = gen7_fill_interface_descriptor(batch, dst,
							      gpgpu_kernel,
							      sizeof(gpgpu_kernel));
	igt_assert(batch->ptr < &batch->buffer[4095]);

	batch->ptr = batch->buffer;

	/* GPGPU pipeline */
	OUT_BATCH(GEN7_PIPELINE_SELECT | PIPELINE_SELECT_GPGPU);

	gen7_emit_state_base_address(batch);

	gen7_emit_vfe_state_gpgpu(batch);

	gen7_emit_curbe_load(batch, curbe_buffer);

	gen7_emit_interface_descriptor_load(batch, interface_descriptor);

	gen7_emit_gpgpu_walk(batch, x, y, width, height);

	OUT_BATCH(MI_BATCH_BUFFER_END);

	batch_end = batch_align(batch, 8);
	igt_assert(batch_end < BATCH_STATE_SPLIT);

	gen7_render_flush(batch, batch_end);
	intel_batchbuffer_reset(batch);
}