1 /*
2  * Copyright 2016 Advanced Micro Devices, Inc.
3  *
4  * Permission is hereby granted, free of charge, to any person obtaining a
5  * copy of this software and associated documentation files (the "Software"),
6  * to deal in the Software without restriction, including without limitation
7  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8  * and/or sell copies of the Software, and to permit persons to whom the
9  * Software is furnished to do so, subject to the following conditions:
10  *
11  * The above copyright notice and this permission notice shall be included in
12  * all copies or substantial portions of the Software.
13  *
14  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
15  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
17  * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
18  * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
19  * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
20  * OTHER DEALINGS IN THE SOFTWARE.
21  *
22  */
23 
24 #ifndef __AMDGPU_VCN_H__
25 #define __AMDGPU_VCN_H__
26 
27 #include "amdgpu_ras.h"
28 
29 #define AMDGPU_VCN_STACK_SIZE		(128*1024)
30 #define AMDGPU_VCN_CONTEXT_SIZE 	(512*1024)
31 
32 #define AMDGPU_VCN_FIRMWARE_OFFSET	256
33 #define AMDGPU_VCN_MAX_ENC_RINGS	3
34 
35 #define AMDGPU_MAX_VCN_INSTANCES	2
36 #define AMDGPU_MAX_VCN_ENC_RINGS  AMDGPU_VCN_MAX_ENC_RINGS * AMDGPU_MAX_VCN_INSTANCES
37 
38 #define AMDGPU_VCN_HARVEST_VCN0 (1 << 0)
39 #define AMDGPU_VCN_HARVEST_VCN1 (1 << 1)
40 
41 #define VCN_DEC_KMD_CMD 		0x80000000
42 #define VCN_DEC_CMD_FENCE		0x00000000
43 #define VCN_DEC_CMD_TRAP		0x00000001
44 #define VCN_DEC_CMD_WRITE_REG		0x00000004
45 #define VCN_DEC_CMD_REG_READ_COND_WAIT	0x00000006
46 #define VCN_DEC_CMD_PACKET_START	0x0000000a
47 #define VCN_DEC_CMD_PACKET_END		0x0000000b
48 
49 #define VCN_DEC_SW_CMD_NO_OP		0x00000000
50 #define VCN_DEC_SW_CMD_END		0x00000001
51 #define VCN_DEC_SW_CMD_IB		0x00000002
52 #define VCN_DEC_SW_CMD_FENCE		0x00000003
53 #define VCN_DEC_SW_CMD_TRAP		0x00000004
54 #define VCN_DEC_SW_CMD_IB_AUTO		0x00000005
55 #define VCN_DEC_SW_CMD_SEMAPHORE	0x00000006
56 #define VCN_DEC_SW_CMD_PREEMPT_FENCE	0x00000009
57 #define VCN_DEC_SW_CMD_REG_WRITE	0x0000000b
58 #define VCN_DEC_SW_CMD_REG_WAIT		0x0000000c
59 
60 #define VCN_ENC_CMD_NO_OP		0x00000000
61 #define VCN_ENC_CMD_END 		0x00000001
62 #define VCN_ENC_CMD_IB			0x00000002
63 #define VCN_ENC_CMD_FENCE		0x00000003
64 #define VCN_ENC_CMD_TRAP		0x00000004
65 #define VCN_ENC_CMD_REG_WRITE		0x0000000b
66 #define VCN_ENC_CMD_REG_WAIT		0x0000000c
67 
68 #define VCN_AON_SOC_ADDRESS_2_0 	0x1f800
69 #define VCN1_AON_SOC_ADDRESS_3_0 	0x48000
70 #define VCN_VID_IP_ADDRESS_2_0		0x0
71 #define VCN_AON_IP_ADDRESS_2_0		0x30000
72 
73 #define mmUVD_RBC_XX_IB_REG_CHECK 					0x026b
74 #define mmUVD_RBC_XX_IB_REG_CHECK_BASE_IDX 				1
75 #define mmUVD_REG_XX_MASK 						0x026c
76 #define mmUVD_REG_XX_MASK_BASE_IDX 					1
77 
78 /* 1 second timeout */
79 #define VCN_IDLE_TIMEOUT	msecs_to_jiffies(1000)
80 
81 #define RREG32_SOC15_DPG_MODE_1_0(ip, inst_idx, reg, mask, sram_sel) 			\
82 	({	WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_MASK, mask); 			\
83 		WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_CTL, 				\
84 			UVD_DPG_LMA_CTL__MASK_EN_MASK | 				\
85 			((adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg) 	\
86 			<< UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT) | 			\
87 			(sram_sel << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT)); 		\
88 		RREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_DATA); 			\
89 	})
90 
91 #define WREG32_SOC15_DPG_MODE_1_0(ip, inst_idx, reg, value, mask, sram_sel) 		\
92 	do { 										\
93 		WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_DATA, value); 			\
94 		WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_MASK, mask); 			\
95 		WREG32_SOC15(ip, inst_idx, mmUVD_DPG_LMA_CTL, 				\
96 			UVD_DPG_LMA_CTL__READ_WRITE_MASK | 				\
97 			((adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg) 	\
98 			<< UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT) | 			\
99 			(sram_sel << UVD_DPG_LMA_CTL__SRAM_SEL__SHIFT)); 		\
100 	} while (0)
101 
102 #define SOC15_DPG_MODE_OFFSET(ip, inst_idx, reg) 						\
103 	({											\
104 		uint32_t internal_reg_offset, addr;						\
105 		bool video_range, video1_range, aon_range, aon1_range;				\
106 												\
107 		addr = (adev->reg_offset[ip##_HWIP][inst_idx][reg##_BASE_IDX] + reg);		\
108 		addr <<= 2; 									\
109 		video_range = ((((0xFFFFF & addr) >= (VCN_VID_SOC_ADDRESS_2_0)) && 		\
110 				((0xFFFFF & addr) < ((VCN_VID_SOC_ADDRESS_2_0 + 0x2600)))));	\
111 		video1_range = ((((0xFFFFF & addr) >= (VCN1_VID_SOC_ADDRESS_3_0)) && 		\
112 				((0xFFFFF & addr) < ((VCN1_VID_SOC_ADDRESS_3_0 + 0x2600)))));	\
113 		aon_range   = ((((0xFFFFF & addr) >= (VCN_AON_SOC_ADDRESS_2_0)) && 		\
114 				((0xFFFFF & addr) < ((VCN_AON_SOC_ADDRESS_2_0 + 0x600)))));	\
115 		aon1_range   = ((((0xFFFFF & addr) >= (VCN1_AON_SOC_ADDRESS_3_0)) && 		\
116 				((0xFFFFF & addr) < ((VCN1_AON_SOC_ADDRESS_3_0 + 0x600)))));	\
117 		if (video_range) 								\
118 			internal_reg_offset = ((0xFFFFF & addr) - (VCN_VID_SOC_ADDRESS_2_0) + 	\
119 				(VCN_VID_IP_ADDRESS_2_0));					\
120 		else if (aon_range)								\
121 			internal_reg_offset = ((0xFFFFF & addr) - (VCN_AON_SOC_ADDRESS_2_0) + 	\
122 				(VCN_AON_IP_ADDRESS_2_0));					\
123 		else if (video1_range) 								\
124 			internal_reg_offset = ((0xFFFFF & addr) - (VCN1_VID_SOC_ADDRESS_3_0) + 	\
125 				(VCN_VID_IP_ADDRESS_2_0));					\
126 		else if (aon1_range)								\
127 			internal_reg_offset = ((0xFFFFF & addr) - (VCN1_AON_SOC_ADDRESS_3_0) + 	\
128 				(VCN_AON_IP_ADDRESS_2_0));					\
129 		else										\
130 			internal_reg_offset = (0xFFFFF & addr);					\
131 												\
132 		internal_reg_offset >>= 2;							\
133 	})
134 
135 #define RREG32_SOC15_DPG_MODE(inst_idx, offset, mask_en) 					\
136 	({											\
137 		WREG32_SOC15(VCN, inst_idx, mmUVD_DPG_LMA_CTL, 					\
138 			(0x0 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT |				\
139 			mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT |				\
140 			offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT));			\
141 		RREG32_SOC15(VCN, inst_idx, mmUVD_DPG_LMA_DATA);				\
142 	})
143 
144 #define WREG32_SOC15_DPG_MODE(inst_idx, offset, value, mask_en, indirect)			\
145 	do {											\
146 		if (!indirect) {								\
147 			WREG32_SOC15(VCN, inst_idx, mmUVD_DPG_LMA_DATA, value);			\
148 			WREG32_SOC15(VCN, inst_idx, mmUVD_DPG_LMA_CTL, 				\
149 				(0x1 << UVD_DPG_LMA_CTL__READ_WRITE__SHIFT |			\
150 				 mask_en << UVD_DPG_LMA_CTL__MASK_EN__SHIFT |			\
151 				 offset << UVD_DPG_LMA_CTL__READ_WRITE_ADDR__SHIFT));		\
152 		} else {									\
153 			*adev->vcn.inst[inst_idx].dpg_sram_curr_addr++ = offset;		\
154 			*adev->vcn.inst[inst_idx].dpg_sram_curr_addr++ = value;			\
155 		}										\
156 	} while (0)
157 
158 #define AMDGPU_FW_SHARED_FLAG_0_UNIFIED_QUEUE (1 << 2)
159 #define AMDGPU_VCN_FW_SHARED_FLAG_0_RB	(1 << 6)
160 #define AMDGPU_VCN_MULTI_QUEUE_FLAG	(1 << 8)
161 #define AMDGPU_VCN_SW_RING_FLAG		(1 << 9)
162 #define AMDGPU_VCN_FW_LOGGING_FLAG	(1 << 10)
163 #define AMDGPU_VCN_SMU_VERSION_INFO_FLAG (1 << 11)
164 #define AMDGPU_VCN_SMU_DPM_INTERFACE_FLAG (1 << 11)
165 #define AMDGPU_VCN_VF_RB_SETUP_FLAG (1 << 14)
166 
167 #define AMDGPU_VCN_IB_FLAG_DECODE_BUFFER	0x00000001
168 #define AMDGPU_VCN_CMD_FLAG_MSG_BUFFER		0x00000001
169 
170 #define VCN_CODEC_DISABLE_MASK_AV1  (1 << 0)
171 #define VCN_CODEC_DISABLE_MASK_VP9  (1 << 1)
172 #define VCN_CODEC_DISABLE_MASK_HEVC (1 << 2)
173 #define VCN_CODEC_DISABLE_MASK_H264 (1 << 3)
174 
175 #define AMDGPU_VCN_SMU_DPM_INTERFACE_DGPU (0)
176 #define AMDGPU_VCN_SMU_DPM_INTERFACE_APU (1)
177 
178 enum fw_queue_mode {
179 	FW_QUEUE_RING_RESET = 1,
180 	FW_QUEUE_DPG_HOLD_OFF = 2,
181 };
182 
183 enum engine_status_constants {
184 	UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON = 0x2AAAA0,
185 	UVD_PGFSM_STATUS__UVDM_UVDU_PWR_ON_2_0 = 0xAAAA0,
186 	UVD_PGFSM_STATUS__UVDM_UVDU_UVDLM_PWR_ON_3_0 = 0x2A2A8AA0,
187 	UVD_PGFSM_CONFIG__UVDM_UVDU_PWR_ON = 0x00000002,
188 	UVD_STATUS__UVD_BUSY = 0x00000004,
189 	GB_ADDR_CONFIG_DEFAULT = 0x26010011,
190 	UVD_STATUS__IDLE = 0x2,
191 	UVD_STATUS__BUSY = 0x5,
192 	UVD_POWER_STATUS__UVD_POWER_STATUS_TILES_OFF = 0x1,
193 	UVD_STATUS__RBC_BUSY = 0x1,
194 	UVD_PGFSM_STATUS_UVDJ_PWR_ON = 0,
195 };
196 
197 enum internal_dpg_state {
198 	VCN_DPG_STATE__UNPAUSE = 0,
199 	VCN_DPG_STATE__PAUSE,
200 };
201 
202 struct dpg_pause_state {
203 	enum internal_dpg_state fw_based;
204 	enum internal_dpg_state jpeg;
205 };
206 
207 struct amdgpu_vcn_reg{
208 	unsigned	data0;
209 	unsigned	data1;
210 	unsigned	cmd;
211 	unsigned	nop;
212 	unsigned	context_id;
213 	unsigned	ib_vmid;
214 	unsigned	ib_bar_low;
215 	unsigned	ib_bar_high;
216 	unsigned	ib_size;
217 	unsigned	gp_scratch8;
218 	unsigned	scratch9;
219 };
220 
221 struct amdgpu_vcn_fw_shared {
222 	void        *cpu_addr;
223 	uint64_t    gpu_addr;
224 	uint32_t    mem_size;
225 	uint32_t    log_offset;
226 };
227 
228 struct amdgpu_vcn_inst {
229 	struct amdgpu_bo	*vcpu_bo;
230 	void			*cpu_addr;
231 	uint64_t		gpu_addr;
232 	void			*saved_bo;
233 	struct amdgpu_ring	ring_dec;
234 	struct amdgpu_ring	ring_enc[AMDGPU_VCN_MAX_ENC_RINGS];
235 	atomic_t		sched_score;
236 	struct amdgpu_irq_src	irq;
237 	struct amdgpu_irq_src	ras_poison_irq;
238 	struct amdgpu_vcn_reg	external;
239 	struct amdgpu_bo	*dpg_sram_bo;
240 	struct dpg_pause_state	pause_state;
241 	void			*dpg_sram_cpu_addr;
242 	uint64_t		dpg_sram_gpu_addr;
243 	uint32_t		*dpg_sram_curr_addr;
244 	atomic_t		dpg_enc_submission_cnt;
245 	struct amdgpu_vcn_fw_shared fw_shared;
246 };
247 
248 struct amdgpu_vcn_ras {
249 	struct amdgpu_ras_block_object ras_block;
250 };
251 
252 struct amdgpu_vcn {
253 	unsigned		fw_version;
254 	struct delayed_work	idle_work;
255 	const struct firmware	*fw;	/* VCN firmware */
256 	unsigned		num_enc_rings;
257 	enum amd_powergating_state cur_state;
258 	bool			indirect_sram;
259 
260 	uint8_t	num_vcn_inst;
261 	struct amdgpu_vcn_inst	 inst[AMDGPU_MAX_VCN_INSTANCES];
262 	uint8_t			 vcn_config[AMDGPU_MAX_VCN_INSTANCES];
263 	uint32_t		 vcn_codec_disable_mask[AMDGPU_MAX_VCN_INSTANCES];
264 	struct amdgpu_vcn_reg	 internal;
265 	struct mutex		 vcn_pg_lock;
266 	struct mutex		vcn1_jpeg1_workaround;
267 	atomic_t		 total_submission_cnt;
268 
269 	unsigned	harvest_config;
270 	int (*pause_dpg_mode)(struct amdgpu_device *adev,
271 		int inst_idx, struct dpg_pause_state *new_state);
272 
273 	struct ras_common_if    *ras_if;
274 	struct amdgpu_vcn_ras   *ras;
275 };
276 
277 struct amdgpu_fw_shared_rb_ptrs_struct {
278 	/* to WA DPG R/W ptr issues.*/
279 	uint32_t  rptr;
280 	uint32_t  wptr;
281 };
282 
283 struct amdgpu_fw_shared_multi_queue {
284 	uint8_t decode_queue_mode;
285 	uint8_t encode_generalpurpose_queue_mode;
286 	uint8_t encode_lowlatency_queue_mode;
287 	uint8_t encode_realtime_queue_mode;
288 	uint8_t padding[4];
289 };
290 
291 struct amdgpu_fw_shared_sw_ring {
292 	uint8_t is_enabled;
293 	uint8_t padding[3];
294 };
295 
296 struct amdgpu_fw_shared_unified_queue_struct {
297 	uint8_t is_enabled;
298 	uint8_t queue_mode;
299 	uint8_t queue_status;
300 	uint8_t padding[5];
301 };
302 
303 struct amdgpu_fw_shared_fw_logging {
304 	uint8_t is_enabled;
305 	uint32_t addr_lo;
306 	uint32_t addr_hi;
307 	uint32_t size;
308 };
309 
310 struct amdgpu_fw_shared_smu_interface_info {
311 	uint8_t smu_interface_type;
312 	uint8_t padding[3];
313 };
314 
315 struct amdgpu_fw_shared {
316 	uint32_t present_flag_0;
317 	uint8_t pad[44];
318 	struct amdgpu_fw_shared_rb_ptrs_struct rb;
319 	uint8_t pad1[1];
320 	struct amdgpu_fw_shared_multi_queue multi_queue;
321 	struct amdgpu_fw_shared_sw_ring sw_ring;
322 	struct amdgpu_fw_shared_fw_logging fw_log;
323 	struct amdgpu_fw_shared_smu_interface_info smu_interface_info;
324 };
325 
326 struct amdgpu_fw_shared_rb_setup {
327 	uint32_t is_rb_enabled_flags;
328 	uint32_t rb_addr_lo;
329 	uint32_t rb_addr_hi;
330 	uint32_t  rb_size;
331 	uint32_t  rb4_addr_lo;
332 	uint32_t  rb4_addr_hi;
333 	uint32_t  rb4_size;
334 	uint32_t  reserved[6];
335 };
336 
337 struct amdgpu_vcn4_fw_shared {
338 	uint32_t present_flag_0;
339 	uint8_t pad[12];
340 	struct amdgpu_fw_shared_unified_queue_struct sq;
341 	uint8_t pad1[8];
342 	struct amdgpu_fw_shared_fw_logging fw_log;
343 	uint8_t pad2[20];
344 	struct amdgpu_fw_shared_rb_setup rb_setup;
345 	struct amdgpu_fw_shared_smu_interface_info smu_dpm_interface;
346 };
347 
348 struct amdgpu_vcn_fwlog {
349 	uint32_t rptr;
350 	uint32_t wptr;
351 	uint32_t buffer_size;
352 	uint32_t header_size;
353 	uint8_t wrapped;
354 };
355 
356 struct amdgpu_vcn_decode_buffer {
357 	uint32_t valid_buf_flag;
358 	uint32_t msg_buffer_address_hi;
359 	uint32_t msg_buffer_address_lo;
360 	uint32_t pad[30];
361 };
362 
363 #define VCN_BLOCK_ENCODE_DISABLE_MASK 0x80
364 #define VCN_BLOCK_DECODE_DISABLE_MASK 0x40
365 #define VCN_BLOCK_QUEUE_DISABLE_MASK 0xC0
366 
367 enum vcn_ring_type {
368 	VCN_ENCODE_RING,
369 	VCN_DECODE_RING,
370 	VCN_UNIFIED_RING,
371 };
372 
373 int amdgpu_vcn_early_init(struct amdgpu_device *adev);
374 int amdgpu_vcn_sw_init(struct amdgpu_device *adev);
375 int amdgpu_vcn_sw_fini(struct amdgpu_device *adev);
376 int amdgpu_vcn_suspend(struct amdgpu_device *adev);
377 int amdgpu_vcn_resume(struct amdgpu_device *adev);
378 void amdgpu_vcn_ring_begin_use(struct amdgpu_ring *ring);
379 void amdgpu_vcn_ring_end_use(struct amdgpu_ring *ring);
380 
381 bool amdgpu_vcn_is_disabled_vcn(struct amdgpu_device *adev,
382 				enum vcn_ring_type type, uint32_t vcn_instance);
383 
384 int amdgpu_vcn_dec_ring_test_ring(struct amdgpu_ring *ring);
385 int amdgpu_vcn_dec_ring_test_ib(struct amdgpu_ring *ring, long timeout);
386 int amdgpu_vcn_dec_sw_ring_test_ring(struct amdgpu_ring *ring);
387 int amdgpu_vcn_dec_sw_ring_test_ib(struct amdgpu_ring *ring, long timeout);
388 int amdgpu_vcn_unified_ring_test_ib(struct amdgpu_ring *ring, long timeout);
389 
390 int amdgpu_vcn_enc_ring_test_ring(struct amdgpu_ring *ring);
391 int amdgpu_vcn_enc_ring_test_ib(struct amdgpu_ring *ring, long timeout);
392 
393 enum amdgpu_ring_priority_level amdgpu_vcn_get_enc_ring_prio(int ring);
394 
395 void amdgpu_vcn_setup_ucode(struct amdgpu_device *adev);
396 
397 void amdgpu_vcn_fwlog_init(struct amdgpu_vcn_inst *vcn);
398 void amdgpu_debugfs_vcn_fwlog_init(struct amdgpu_device *adev,
399                                    uint8_t i, struct amdgpu_vcn_inst *vcn);
400 
401 int amdgpu_vcn_process_poison_irq(struct amdgpu_device *adev,
402 			struct amdgpu_irq_src *source,
403 			struct amdgpu_iv_entry *entry);
404 int amdgpu_vcn_ras_late_init(struct amdgpu_device *adev,
405 			struct ras_common_if *ras_block);
406 int amdgpu_vcn_ras_sw_init(struct amdgpu_device *adev);
407 
408 #endif
409