1 /**************************************************************************
2  *
3  * Copyright 2011 Advanced Micro Devices, Inc.
4  * All Rights Reserved.
5  *
6  * Permission is hereby granted, free of charge, to any person obtaining a
7  * copy of this software and associated documentation files (the
8  * "Software"), to deal in the Software without restriction, including
9  * without limitation the rights to use, copy, modify, merge, publish,
10  * distribute, sub license, and/or sell copies of the Software, and to
11  * permit persons to whom the Software is furnished to do so, subject to
12  * the following conditions:
13  *
14  * The above copyright notice and this permission notice (including the
15  * next paragraph) shall be included in all copies or substantial portions
16  * of the Software.
17  *
18  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
19  * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
20  * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.
21  * IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR
22  * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
23  * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
24  * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
25  *
26  **************************************************************************/
27 
28 /*
29  * Authors:
30  *      Christian König <christian.koenig@amd.com>
31  *
32  */
33 
34 #ifndef RADEON_UVD_H
35 #define RADEON_UVD_H
36 
37 #include "radeon/radeon_winsys.h"
38 #include "vl/vl_video_buffer.h"
39 
40 /* UVD uses PM4 packet type 0 and 2 */
41 #define RUVD_PKT_TYPE_S(x)		(((unsigned)(x) & 0x3) << 30)
42 #define RUVD_PKT_TYPE_G(x)		(((x) >> 30) & 0x3)
43 #define RUVD_PKT_TYPE_C			0x3FFFFFFF
44 #define RUVD_PKT_COUNT_S(x)		(((unsigned)(x) & 0x3FFF) << 16)
45 #define RUVD_PKT_COUNT_G(x)		(((x) >> 16) & 0x3FFF)
46 #define RUVD_PKT_COUNT_C		0xC000FFFF
47 #define RUVD_PKT0_BASE_INDEX_S(x)	(((unsigned)(x) & 0xFFFF) << 0)
48 #define RUVD_PKT0_BASE_INDEX_G(x)	(((x) >> 0) & 0xFFFF)
49 #define RUVD_PKT0_BASE_INDEX_C		0xFFFF0000
50 #define RUVD_PKT0(index, count)		(RUVD_PKT_TYPE_S(0) | RUVD_PKT0_BASE_INDEX_S(index) | RUVD_PKT_COUNT_S(count))
51 #define RUVD_PKT2()			(RUVD_PKT_TYPE_S(2))
52 
53 /* registers involved with UVD */
54 #define RUVD_GPCOM_VCPU_CMD		0xEF0C
55 #define RUVD_GPCOM_VCPU_DATA0		0xEF10
56 #define RUVD_GPCOM_VCPU_DATA1		0xEF14
57 #define RUVD_ENGINE_CNTL		0xEF18
58 
59 #define RUVD_GPCOM_VCPU_CMD_SOC15		0x2070c
60 #define RUVD_GPCOM_VCPU_DATA0_SOC15		0x20710
61 #define RUVD_GPCOM_VCPU_DATA1_SOC15		0x20714
62 #define RUVD_ENGINE_CNTL_SOC15			0x20718
63 
64 /* UVD commands to VCPU */
65 #define RUVD_CMD_MSG_BUFFER		0x00000000
66 #define RUVD_CMD_DPB_BUFFER		0x00000001
67 #define RUVD_CMD_DECODING_TARGET_BUFFER	0x00000002
68 #define RUVD_CMD_FEEDBACK_BUFFER	0x00000003
69 #define RUVD_CMD_SESSION_CONTEXT_BUFFER	0x00000005
70 #define RUVD_CMD_BITSTREAM_BUFFER	0x00000100
71 #define RUVD_CMD_ITSCALING_TABLE_BUFFER	0x00000204
72 #define RUVD_CMD_CONTEXT_BUFFER		0x00000206
73 
74 /* UVD message types */
75 #define RUVD_MSG_CREATE		0
76 #define RUVD_MSG_DECODE		1
77 #define RUVD_MSG_DESTROY	2
78 
79 /* UVD stream types */
80 #define RUVD_CODEC_H264		0x00000000
81 #define RUVD_CODEC_VC1		0x00000001
82 #define RUVD_CODEC_MPEG2	0x00000003
83 #define RUVD_CODEC_MPEG4	0x00000004
84 #define RUVD_CODEC_H264_PERF	0x00000007
85 #define RUVD_CODEC_MJPEG	0x00000008
86 #define RUVD_CODEC_H265		0x00000010
87 
88 /* UVD decode target buffer tiling mode */
89 #define RUVD_TILE_LINEAR	0x00000000
90 #define RUVD_TILE_8X4		0x00000001
91 #define RUVD_TILE_8X8		0x00000002
92 #define RUVD_TILE_32AS8		0x00000003
93 
94 /* UVD decode target buffer array mode */
95 #define RUVD_ARRAY_MODE_LINEAR				0x00000000
96 #define RUVD_ARRAY_MODE_MACRO_LINEAR_MICRO_TILED	0x00000001
97 #define RUVD_ARRAY_MODE_1D_THIN				0x00000002
98 #define RUVD_ARRAY_MODE_2D_THIN				0x00000004
99 #define RUVD_ARRAY_MODE_MACRO_TILED_MICRO_LINEAR	0x00000004
100 #define RUVD_ARRAY_MODE_MACRO_TILED_MICRO_TILED		0x00000005
101 
102 /* UVD tile config */
103 #define RUVD_BANK_WIDTH(x)		((x) << 0)
104 #define RUVD_BANK_HEIGHT(x)		((x) << 3)
105 #define RUVD_MACRO_TILE_ASPECT_RATIO(x)	((x) << 6)
106 #define RUVD_NUM_BANKS(x)		((x) << 9)
107 
108 /* H.264 profile definitions */
109 #define RUVD_H264_PROFILE_BASELINE	0x00000000
110 #define RUVD_H264_PROFILE_MAIN		0x00000001
111 #define RUVD_H264_PROFILE_HIGH		0x00000002
112 #define RUVD_H264_PROFILE_STEREO_HIGH	0x00000003
113 #define RUVD_H264_PROFILE_MVC		0x00000004
114 
115 /* VC-1 profile definitions */
116 #define RUVD_VC1_PROFILE_SIMPLE		0x00000000
117 #define RUVD_VC1_PROFILE_MAIN		0x00000001
118 #define RUVD_VC1_PROFILE_ADVANCED	0x00000002
119 
120 struct ruvd_mvc_element {
121 	uint16_t	viewOrderIndex;
122 	uint16_t	viewId;
123 	uint16_t	numOfAnchorRefsInL0;
124 	uint16_t	viewIdOfAnchorRefsInL0[15];
125 	uint16_t	numOfAnchorRefsInL1;
126 	uint16_t	viewIdOfAnchorRefsInL1[15];
127 	uint16_t	numOfNonAnchorRefsInL0;
128 	uint16_t	viewIdOfNonAnchorRefsInL0[15];
129 	uint16_t	numOfNonAnchorRefsInL1;
130 	uint16_t	viewIdOfNonAnchorRefsInL1[15];
131 };
132 
133 struct ruvd_h264 {
134 	uint32_t	profile;
135 	uint32_t	level;
136 
137 	uint32_t	sps_info_flags;
138 	uint32_t	pps_info_flags;
139 	uint8_t		chroma_format;
140 	uint8_t		bit_depth_luma_minus8;
141 	uint8_t		bit_depth_chroma_minus8;
142 	uint8_t		log2_max_frame_num_minus4;
143 
144 	uint8_t		pic_order_cnt_type;
145 	uint8_t		log2_max_pic_order_cnt_lsb_minus4;
146 	uint8_t		num_ref_frames;
147 	uint8_t		reserved_8bit;
148 
149 	int8_t		pic_init_qp_minus26;
150 	int8_t		pic_init_qs_minus26;
151 	int8_t		chroma_qp_index_offset;
152 	int8_t		second_chroma_qp_index_offset;
153 
154 	uint8_t		num_slice_groups_minus1;
155 	uint8_t		slice_group_map_type;
156 	uint8_t		num_ref_idx_l0_active_minus1;
157 	uint8_t		num_ref_idx_l1_active_minus1;
158 
159 	uint16_t	slice_group_change_rate_minus1;
160 	uint16_t	reserved_16bit_1;
161 
162 	uint8_t		scaling_list_4x4[6][16];
163 	uint8_t		scaling_list_8x8[2][64];
164 
165 	uint32_t	frame_num;
166 	uint32_t	frame_num_list[16];
167 	int32_t		curr_field_order_cnt_list[2];
168 	int32_t		field_order_cnt_list[16][2];
169 
170 	uint32_t	decoded_pic_idx;
171 
172 	uint32_t	curr_pic_ref_frame_num;
173 
174 	uint8_t		ref_frame_list[16];
175 
176 	uint32_t	reserved[122];
177 
178 	struct {
179 		uint32_t			numViews;
180 		uint32_t			viewId0;
181 		struct ruvd_mvc_element	mvcElements[1];
182 	} mvc;
183 };
184 
185 struct ruvd_h265 {
186 	uint32_t	sps_info_flags;
187 	uint32_t	pps_info_flags;
188 
189 	uint8_t		chroma_format;
190 	uint8_t		bit_depth_luma_minus8;
191 	uint8_t		bit_depth_chroma_minus8;
192 	uint8_t		log2_max_pic_order_cnt_lsb_minus4;
193 
194 	uint8_t		sps_max_dec_pic_buffering_minus1;
195 	uint8_t		log2_min_luma_coding_block_size_minus3;
196 	uint8_t		log2_diff_max_min_luma_coding_block_size;
197 	uint8_t		log2_min_transform_block_size_minus2;
198 
199 	uint8_t		log2_diff_max_min_transform_block_size;
200 	uint8_t		max_transform_hierarchy_depth_inter;
201 	uint8_t		max_transform_hierarchy_depth_intra;
202 	uint8_t		pcm_sample_bit_depth_luma_minus1;
203 
204 	uint8_t		pcm_sample_bit_depth_chroma_minus1;
205 	uint8_t		log2_min_pcm_luma_coding_block_size_minus3;
206 	uint8_t		log2_diff_max_min_pcm_luma_coding_block_size;
207 	uint8_t		num_extra_slice_header_bits;
208 
209 	uint8_t		num_short_term_ref_pic_sets;
210 	uint8_t		num_long_term_ref_pic_sps;
211 	uint8_t		num_ref_idx_l0_default_active_minus1;
212 	uint8_t		num_ref_idx_l1_default_active_minus1;
213 
214 	int8_t		pps_cb_qp_offset;
215 	int8_t		pps_cr_qp_offset;
216 	int8_t		pps_beta_offset_div2;
217 	int8_t		pps_tc_offset_div2;
218 
219 	uint8_t		diff_cu_qp_delta_depth;
220 	uint8_t		num_tile_columns_minus1;
221 	uint8_t		num_tile_rows_minus1;
222 	uint8_t		log2_parallel_merge_level_minus2;
223 
224 	uint16_t	column_width_minus1[19];
225 	uint16_t	row_height_minus1[21];
226 
227 	int8_t		init_qp_minus26;
228 	uint8_t		num_delta_pocs_ref_rps_idx;
229 	uint8_t		curr_idx;
230 	uint8_t		reserved1;
231 	int32_t		curr_poc;
232 	uint8_t		ref_pic_list[16];
233 	int32_t		poc_list[16];
234 	uint8_t		ref_pic_set_st_curr_before[8];
235 	uint8_t		ref_pic_set_st_curr_after[8];
236 	uint8_t		ref_pic_set_lt_curr[8];
237 
238 	uint8_t		ucScalingListDCCoefSizeID2[6];
239 	uint8_t		ucScalingListDCCoefSizeID3[2];
240 
241 	uint8_t		highestTid;
242 	uint8_t		isNonRef;
243 
244 	uint8_t 	p010_mode;
245 	uint8_t 	msb_mode;
246 	uint8_t 	luma_10to8;
247 	uint8_t 	chroma_10to8;
248 	uint8_t 	sclr_luma10to8;
249 	uint8_t 	sclr_chroma10to8;
250 
251 	uint8_t 	direct_reflist[2][15];
252 };
253 
254 struct ruvd_vc1 {
255 	uint32_t	profile;
256 	uint32_t	level;
257 	uint32_t	sps_info_flags;
258 	uint32_t	pps_info_flags;
259 	uint32_t	pic_structure;
260 	uint32_t	chroma_format;
261 };
262 
263 struct ruvd_mpeg2 {
264 	uint32_t	decoded_pic_idx;
265 	uint32_t	ref_pic_idx[2];
266 
267 	uint8_t		load_intra_quantiser_matrix;
268 	uint8_t		load_nonintra_quantiser_matrix;
269 	uint8_t		reserved_quantiser_alignement[2];
270 	uint8_t		intra_quantiser_matrix[64];
271 	uint8_t		nonintra_quantiser_matrix[64];
272 
273 	uint8_t		profile_and_level_indication;
274 	uint8_t		chroma_format;
275 
276 	uint8_t		picture_coding_type;
277 
278 	uint8_t		reserved_1;
279 
280 	uint8_t		f_code[2][2];
281 	uint8_t		intra_dc_precision;
282 	uint8_t		pic_structure;
283 	uint8_t		top_field_first;
284 	uint8_t		frame_pred_frame_dct;
285 	uint8_t		concealment_motion_vectors;
286 	uint8_t		q_scale_type;
287 	uint8_t		intra_vlc_format;
288 	uint8_t		alternate_scan;
289 };
290 
291 struct ruvd_mpeg4
292 {
293 	uint32_t	decoded_pic_idx;
294 	uint32_t	ref_pic_idx[2];
295 
296 	uint32_t	variant_type;
297 	uint8_t		profile_and_level_indication;
298 
299 	uint8_t		video_object_layer_verid;
300 	uint8_t		video_object_layer_shape;
301 
302 	uint8_t		reserved_1;
303 
304 	uint16_t	video_object_layer_width;
305 	uint16_t	video_object_layer_height;
306 
307 	uint16_t	vop_time_increment_resolution;
308 
309 	uint16_t	reserved_2;
310 
311 	uint32_t	flags;
312 
313 	uint8_t		quant_type;
314 
315 	uint8_t		reserved_3[3];
316 
317 	uint8_t		intra_quant_mat[64];
318 	uint8_t		nonintra_quant_mat[64];
319 
320 	struct {
321 		uint8_t		sprite_enable;
322 
323 		uint8_t		reserved_4[3];
324 
325 		uint16_t	sprite_width;
326 		uint16_t	sprite_height;
327 		int16_t		sprite_left_coordinate;
328 		int16_t		sprite_top_coordinate;
329 
330 		uint8_t		no_of_sprite_warping_points;
331 		uint8_t		sprite_warping_accuracy;
332 		uint8_t		sprite_brightness_change;
333 		uint8_t		low_latency_sprite_enable;
334 	} sprite_config;
335 
336 	struct {
337 		uint32_t	flags;
338 		uint8_t		vol_mode;
339 		uint8_t		reserved_5[3];
340 	} divx_311_config;
341 };
342 
343 /* message between driver and hardware */
344 struct ruvd_msg {
345 
346 	uint32_t	size;
347 	uint32_t	msg_type;
348 	uint32_t	stream_handle;
349 	uint32_t	status_report_feedback_number;
350 
351 	union {
352 		struct {
353 			uint32_t	stream_type;
354 			uint32_t	session_flags;
355 			uint32_t	asic_id;
356 			uint32_t	width_in_samples;
357 			uint32_t	height_in_samples;
358 			uint32_t	dpb_buffer;
359 			uint32_t	dpb_size;
360 			uint32_t	dpb_model;
361 			uint32_t	version_info;
362 		} create;
363 
364 		struct {
365 			uint32_t	stream_type;
366 			uint32_t	decode_flags;
367 			uint32_t	width_in_samples;
368 			uint32_t	height_in_samples;
369 
370 			uint32_t	dpb_buffer;
371 			uint32_t	dpb_size;
372 			uint32_t	dpb_model;
373 			uint32_t	dpb_reserved;
374 
375 			uint32_t	db_offset_alignment;
376 			uint32_t	db_pitch;
377 			uint32_t	db_tiling_mode;
378 			uint32_t	db_array_mode;
379 			uint32_t	db_field_mode;
380 			uint32_t	db_surf_tile_config;
381 			uint32_t	db_aligned_height;
382 			uint32_t	db_reserved;
383 
384 			uint32_t	use_addr_macro;
385 
386 			uint32_t	bsd_buffer;
387 			uint32_t	bsd_size;
388 
389 			uint32_t	pic_param_buffer;
390 			uint32_t	pic_param_size;
391 			uint32_t	mb_cntl_buffer;
392 			uint32_t	mb_cntl_size;
393 
394 			uint32_t	dt_buffer;
395 			uint32_t	dt_pitch;
396 			uint32_t	dt_tiling_mode;
397 			uint32_t	dt_array_mode;
398 			uint32_t	dt_field_mode;
399 			uint32_t	dt_luma_top_offset;
400 			uint32_t	dt_luma_bottom_offset;
401 			uint32_t	dt_chroma_top_offset;
402 			uint32_t	dt_chroma_bottom_offset;
403 			uint32_t	dt_surf_tile_config;
404 			uint32_t	dt_uv_surf_tile_config;
405 			// re-use dt_wa_chroma_top_offset as dt_ext_info for UV pitch in stoney
406 			uint32_t	dt_wa_chroma_top_offset;
407 			uint32_t	dt_wa_chroma_bottom_offset;
408 
409 			uint32_t	reserved[16];
410 
411 			union {
412 				struct ruvd_h264	h264;
413 				struct ruvd_h265	h265;
414 				struct ruvd_vc1		vc1;
415 				struct ruvd_mpeg2	mpeg2;
416 				struct ruvd_mpeg4	mpeg4;
417 
418 				uint32_t info[768];
419 			} codec;
420 
421 			uint8_t		extension_support;
422 			uint8_t		reserved_8bit_1;
423 			uint8_t		reserved_8bit_2;
424 			uint8_t		reserved_8bit_3;
425 			uint32_t	extension_reserved[64];
426 		} decode;
427 	} body;
428 };
429 
430 /* driver dependent callback */
431 typedef struct pb_buffer* (*ruvd_set_dtb)
432 (struct ruvd_msg* msg, struct vl_video_buffer *vb);
433 
434 /* create an UVD decode */
435 struct pipe_video_codec *ruvd_create_decoder(struct pipe_context *context,
436 					     const struct pipe_video_codec *templat,
437 					     ruvd_set_dtb set_dtb);
438 
439 /* fill decoding target field from the luma and chroma surfaces */
440 void ruvd_set_dt_surfaces(struct ruvd_msg *msg, struct radeon_surf *luma,
441 			  struct radeon_surf *chroma);
442 #endif
443