This source file includes following definitions.
- amdgpu_ring_set_preempt_cond_exec
- amdgpu_ring_clear_ring
- amdgpu_ring_write
- amdgpu_ring_write_multiple
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24 #ifndef __AMDGPU_RING_H__
25 #define __AMDGPU_RING_H__
26
27 #include <drm/amdgpu_drm.h>
28 #include <drm/gpu_scheduler.h>
29 #include <drm/drm_print.h>
30
31
32 #define AMDGPU_MAX_RINGS 28
33 #define AMDGPU_MAX_GFX_RINGS 2
34 #define AMDGPU_MAX_COMPUTE_RINGS 8
35 #define AMDGPU_MAX_VCE_RINGS 3
36 #define AMDGPU_MAX_UVD_ENC_RINGS 2
37
38
39 #define AMDGPU_FENCE_OWNER_UNDEFINED ((void *)0ul)
40 #define AMDGPU_FENCE_OWNER_VM ((void *)1ul)
41 #define AMDGPU_FENCE_OWNER_KFD ((void *)2ul)
42
43 #define AMDGPU_FENCE_FLAG_64BIT (1 << 0)
44 #define AMDGPU_FENCE_FLAG_INT (1 << 1)
45 #define AMDGPU_FENCE_FLAG_TC_WB_ONLY (1 << 2)
46
47 #define to_amdgpu_ring(s) container_of((s), struct amdgpu_ring, sched)
48
49 enum amdgpu_ring_type {
50 AMDGPU_RING_TYPE_GFX,
51 AMDGPU_RING_TYPE_COMPUTE,
52 AMDGPU_RING_TYPE_SDMA,
53 AMDGPU_RING_TYPE_UVD,
54 AMDGPU_RING_TYPE_VCE,
55 AMDGPU_RING_TYPE_KIQ,
56 AMDGPU_RING_TYPE_UVD_ENC,
57 AMDGPU_RING_TYPE_VCN_DEC,
58 AMDGPU_RING_TYPE_VCN_ENC,
59 AMDGPU_RING_TYPE_VCN_JPEG
60 };
61
62 struct amdgpu_device;
63 struct amdgpu_ring;
64 struct amdgpu_ib;
65 struct amdgpu_cs_parser;
66 struct amdgpu_job;
67
68
69
70
71 struct amdgpu_fence_driver {
72 uint64_t gpu_addr;
73 volatile uint32_t *cpu_addr;
74
75 uint32_t sync_seq;
76 atomic_t last_seq;
77 bool initialized;
78 struct amdgpu_irq_src *irq_src;
79 unsigned irq_type;
80 struct timer_list fallback_timer;
81 unsigned num_fences_mask;
82 spinlock_t lock;
83 struct dma_fence **fences;
84 };
85
86 int amdgpu_fence_driver_init(struct amdgpu_device *adev);
87 void amdgpu_fence_driver_fini(struct amdgpu_device *adev);
88 void amdgpu_fence_driver_force_completion(struct amdgpu_ring *ring);
89
90 int amdgpu_fence_driver_init_ring(struct amdgpu_ring *ring,
91 unsigned num_hw_submission);
92 int amdgpu_fence_driver_start_ring(struct amdgpu_ring *ring,
93 struct amdgpu_irq_src *irq_src,
94 unsigned irq_type);
95 void amdgpu_fence_driver_suspend(struct amdgpu_device *adev);
96 void amdgpu_fence_driver_resume(struct amdgpu_device *adev);
97 int amdgpu_fence_emit(struct amdgpu_ring *ring, struct dma_fence **fence,
98 unsigned flags);
99 int amdgpu_fence_emit_polling(struct amdgpu_ring *ring, uint32_t *s);
100 bool amdgpu_fence_process(struct amdgpu_ring *ring);
101 int amdgpu_fence_wait_empty(struct amdgpu_ring *ring);
102 signed long amdgpu_fence_wait_polling(struct amdgpu_ring *ring,
103 uint32_t wait_seq,
104 signed long timeout);
105 unsigned amdgpu_fence_count_emitted(struct amdgpu_ring *ring);
106
107
108
109
110
111
112 struct amdgpu_ring_funcs {
113 enum amdgpu_ring_type type;
114 uint32_t align_mask;
115 u32 nop;
116 bool support_64bit_ptrs;
117 bool no_user_fence;
118 unsigned vmhub;
119 unsigned extra_dw;
120
121
122 u64 (*get_rptr)(struct amdgpu_ring *ring);
123 u64 (*get_wptr)(struct amdgpu_ring *ring);
124 void (*set_wptr)(struct amdgpu_ring *ring);
125
126 int (*parse_cs)(struct amdgpu_cs_parser *p, uint32_t ib_idx);
127 int (*patch_cs_in_place)(struct amdgpu_cs_parser *p, uint32_t ib_idx);
128
129 unsigned emit_frame_size;
130 unsigned emit_ib_size;
131
132 void (*emit_ib)(struct amdgpu_ring *ring,
133 struct amdgpu_job *job,
134 struct amdgpu_ib *ib,
135 uint32_t flags);
136 void (*emit_fence)(struct amdgpu_ring *ring, uint64_t addr,
137 uint64_t seq, unsigned flags);
138 void (*emit_pipeline_sync)(struct amdgpu_ring *ring);
139 void (*emit_vm_flush)(struct amdgpu_ring *ring, unsigned vmid,
140 uint64_t pd_addr);
141 void (*emit_hdp_flush)(struct amdgpu_ring *ring);
142 void (*emit_gds_switch)(struct amdgpu_ring *ring, uint32_t vmid,
143 uint32_t gds_base, uint32_t gds_size,
144 uint32_t gws_base, uint32_t gws_size,
145 uint32_t oa_base, uint32_t oa_size);
146
147 int (*test_ring)(struct amdgpu_ring *ring);
148 int (*test_ib)(struct amdgpu_ring *ring, long timeout);
149
150 void (*insert_nop)(struct amdgpu_ring *ring, uint32_t count);
151 void (*insert_start)(struct amdgpu_ring *ring);
152 void (*insert_end)(struct amdgpu_ring *ring);
153
154 void (*pad_ib)(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
155 unsigned (*init_cond_exec)(struct amdgpu_ring *ring);
156 void (*patch_cond_exec)(struct amdgpu_ring *ring, unsigned offset);
157
158 void (*begin_use)(struct amdgpu_ring *ring);
159 void (*end_use)(struct amdgpu_ring *ring);
160 void (*emit_switch_buffer) (struct amdgpu_ring *ring);
161 void (*emit_cntxcntl) (struct amdgpu_ring *ring, uint32_t flags);
162 void (*emit_rreg)(struct amdgpu_ring *ring, uint32_t reg);
163 void (*emit_wreg)(struct amdgpu_ring *ring, uint32_t reg, uint32_t val);
164 void (*emit_reg_wait)(struct amdgpu_ring *ring, uint32_t reg,
165 uint32_t val, uint32_t mask);
166 void (*emit_reg_write_reg_wait)(struct amdgpu_ring *ring,
167 uint32_t reg0, uint32_t reg1,
168 uint32_t ref, uint32_t mask);
169 void (*emit_tmz)(struct amdgpu_ring *ring, bool start);
170
171 void (*set_priority) (struct amdgpu_ring *ring,
172 enum drm_sched_priority priority);
173
174 void (*soft_recovery)(struct amdgpu_ring *ring, unsigned vmid);
175 int (*preempt_ib)(struct amdgpu_ring *ring);
176 };
177
178 struct amdgpu_ring {
179 struct amdgpu_device *adev;
180 const struct amdgpu_ring_funcs *funcs;
181 struct amdgpu_fence_driver fence_drv;
182 struct drm_gpu_scheduler sched;
183
184 struct amdgpu_bo *ring_obj;
185 volatile uint32_t *ring;
186 unsigned rptr_offs;
187 u64 wptr;
188 u64 wptr_old;
189 unsigned ring_size;
190 unsigned max_dw;
191 int count_dw;
192 uint64_t gpu_addr;
193 uint64_t ptr_mask;
194 uint32_t buf_mask;
195 u32 idx;
196 u32 me;
197 u32 pipe;
198 u32 queue;
199 struct amdgpu_bo *mqd_obj;
200 uint64_t mqd_gpu_addr;
201 void *mqd_ptr;
202 uint64_t eop_gpu_addr;
203 u32 doorbell_index;
204 bool use_doorbell;
205 bool use_pollmem;
206 unsigned wptr_offs;
207 unsigned fence_offs;
208 uint64_t current_ctx;
209 char name[16];
210 u32 trail_seq;
211 unsigned trail_fence_offs;
212 u64 trail_fence_gpu_addr;
213 volatile u32 *trail_fence_cpu_addr;
214 unsigned cond_exe_offs;
215 u64 cond_exe_gpu_addr;
216 volatile u32 *cond_exe_cpu_addr;
217 unsigned vm_inv_eng;
218 struct dma_fence *vmid_wait;
219 bool has_compute_vm_bug;
220
221 atomic_t num_jobs[DRM_SCHED_PRIORITY_MAX];
222 struct mutex priority_mutex;
223
224 int priority;
225
226 #if defined(CONFIG_DEBUG_FS)
227 struct dentry *ent;
228 #endif
229 };
230
231 #define amdgpu_ring_parse_cs(r, p, ib) ((r)->funcs->parse_cs((p), (ib)))
232 #define amdgpu_ring_patch_cs_in_place(r, p, ib) ((r)->funcs->patch_cs_in_place((p), (ib)))
233 #define amdgpu_ring_test_ring(r) (r)->funcs->test_ring((r))
234 #define amdgpu_ring_test_ib(r, t) (r)->funcs->test_ib((r), (t))
235 #define amdgpu_ring_get_rptr(r) (r)->funcs->get_rptr((r))
236 #define amdgpu_ring_get_wptr(r) (r)->funcs->get_wptr((r))
237 #define amdgpu_ring_set_wptr(r) (r)->funcs->set_wptr((r))
238 #define amdgpu_ring_emit_ib(r, job, ib, flags) ((r)->funcs->emit_ib((r), (job), (ib), (flags)))
239 #define amdgpu_ring_emit_pipeline_sync(r) (r)->funcs->emit_pipeline_sync((r))
240 #define amdgpu_ring_emit_vm_flush(r, vmid, addr) (r)->funcs->emit_vm_flush((r), (vmid), (addr))
241 #define amdgpu_ring_emit_fence(r, addr, seq, flags) (r)->funcs->emit_fence((r), (addr), (seq), (flags))
242 #define amdgpu_ring_emit_gds_switch(r, v, db, ds, wb, ws, ab, as) (r)->funcs->emit_gds_switch((r), (v), (db), (ds), (wb), (ws), (ab), (as))
243 #define amdgpu_ring_emit_hdp_flush(r) (r)->funcs->emit_hdp_flush((r))
244 #define amdgpu_ring_emit_switch_buffer(r) (r)->funcs->emit_switch_buffer((r))
245 #define amdgpu_ring_emit_cntxcntl(r, d) (r)->funcs->emit_cntxcntl((r), (d))
246 #define amdgpu_ring_emit_rreg(r, d) (r)->funcs->emit_rreg((r), (d))
247 #define amdgpu_ring_emit_wreg(r, d, v) (r)->funcs->emit_wreg((r), (d), (v))
248 #define amdgpu_ring_emit_reg_wait(r, d, v, m) (r)->funcs->emit_reg_wait((r), (d), (v), (m))
249 #define amdgpu_ring_emit_reg_write_reg_wait(r, d0, d1, v, m) (r)->funcs->emit_reg_write_reg_wait((r), (d0), (d1), (v), (m))
250 #define amdgpu_ring_emit_tmz(r, b) (r)->funcs->emit_tmz((r), (b))
251 #define amdgpu_ring_pad_ib(r, ib) ((r)->funcs->pad_ib((r), (ib)))
252 #define amdgpu_ring_init_cond_exec(r) (r)->funcs->init_cond_exec((r))
253 #define amdgpu_ring_patch_cond_exec(r,o) (r)->funcs->patch_cond_exec((r),(o))
254 #define amdgpu_ring_preempt_ib(r) (r)->funcs->preempt_ib(r)
255
256 int amdgpu_ring_alloc(struct amdgpu_ring *ring, unsigned ndw);
257 void amdgpu_ring_insert_nop(struct amdgpu_ring *ring, uint32_t count);
258 void amdgpu_ring_generic_pad_ib(struct amdgpu_ring *ring, struct amdgpu_ib *ib);
259 void amdgpu_ring_commit(struct amdgpu_ring *ring);
260 void amdgpu_ring_undo(struct amdgpu_ring *ring);
261 void amdgpu_ring_priority_get(struct amdgpu_ring *ring,
262 enum drm_sched_priority priority);
263 void amdgpu_ring_priority_put(struct amdgpu_ring *ring,
264 enum drm_sched_priority priority);
265 int amdgpu_ring_init(struct amdgpu_device *adev, struct amdgpu_ring *ring,
266 unsigned ring_size, struct amdgpu_irq_src *irq_src,
267 unsigned irq_type);
268 void amdgpu_ring_fini(struct amdgpu_ring *ring);
269 void amdgpu_ring_emit_reg_write_reg_wait_helper(struct amdgpu_ring *ring,
270 uint32_t reg0, uint32_t val0,
271 uint32_t reg1, uint32_t val1);
272 bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
273 struct dma_fence *fence);
274
275 static inline void amdgpu_ring_set_preempt_cond_exec(struct amdgpu_ring *ring,
276 bool cond_exec)
277 {
278 *ring->cond_exe_cpu_addr = cond_exec;
279 }
280
281 static inline void amdgpu_ring_clear_ring(struct amdgpu_ring *ring)
282 {
283 int i = 0;
284 while (i <= ring->buf_mask)
285 ring->ring[i++] = ring->funcs->nop;
286
287 }
288
289 static inline void amdgpu_ring_write(struct amdgpu_ring *ring, uint32_t v)
290 {
291 if (ring->count_dw <= 0)
292 DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
293 ring->ring[ring->wptr++ & ring->buf_mask] = v;
294 ring->wptr &= ring->ptr_mask;
295 ring->count_dw--;
296 }
297
298 static inline void amdgpu_ring_write_multiple(struct amdgpu_ring *ring,
299 void *src, int count_dw)
300 {
301 unsigned occupied, chunk1, chunk2;
302 void *dst;
303
304 if (unlikely(ring->count_dw < count_dw))
305 DRM_ERROR("amdgpu: writing more dwords to the ring than expected!\n");
306
307 occupied = ring->wptr & ring->buf_mask;
308 dst = (void *)&ring->ring[occupied];
309 chunk1 = ring->buf_mask + 1 - occupied;
310 chunk1 = (chunk1 >= count_dw) ? count_dw: chunk1;
311 chunk2 = count_dw - chunk1;
312 chunk1 <<= 2;
313 chunk2 <<= 2;
314
315 if (chunk1)
316 memcpy(dst, src, chunk1);
317
318 if (chunk2) {
319 src += chunk1;
320 dst = (void *)ring->ring;
321 memcpy(dst, src, chunk2);
322 }
323
324 ring->wptr += count_dw;
325 ring->wptr &= ring->ptr_mask;
326 ring->count_dw -= count_dw;
327 }
328
329 int amdgpu_ring_test_helper(struct amdgpu_ring *ring);
330
331 #endif