2 * Copyright (c) 2014-2015, NVIDIA CORPORATION. All rights reserved.
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:
11 * The above copyright notice and this permission notice shall be included in
12 * all copies or substantial portions of the Software.
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 AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
18 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING
19 * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER
20 * DEALINGS IN THE SOFTWARE.
25 #include <subdev/timer.h>
27 #include <nvif/class.h>
36 gk20a_gr_av_to_init(struct gf100_gr
*gr
, const char *fw_name
,
37 struct gf100_gr_pack
**ppack
)
39 struct gf100_gr_fuc fuc
;
40 struct gf100_gr_init
*init
;
41 struct gf100_gr_pack
*pack
;
46 ret
= gf100_gr_ctor_fw(gr
, fw_name
, &fuc
);
50 nent
= (fuc
.size
/ sizeof(struct gk20a_fw_av
));
51 pack
= vzalloc((sizeof(*pack
) * 2) + (sizeof(*init
) * (nent
+ 1)));
57 init
= (void *)(pack
+ 2);
60 for (i
= 0; i
< nent
; i
++) {
61 struct gf100_gr_init
*ent
= &init
[i
];
62 struct gk20a_fw_av
*av
= &((struct gk20a_fw_av
*)fuc
.data
)[i
];
73 gf100_gr_dtor_fw(&fuc
);
85 gk20a_gr_aiv_to_init(struct gf100_gr
*gr
, const char *fw_name
,
86 struct gf100_gr_pack
**ppack
)
88 struct gf100_gr_fuc fuc
;
89 struct gf100_gr_init
*init
;
90 struct gf100_gr_pack
*pack
;
95 ret
= gf100_gr_ctor_fw(gr
, fw_name
, &fuc
);
99 nent
= (fuc
.size
/ sizeof(struct gk20a_fw_aiv
));
100 pack
= vzalloc((sizeof(*pack
) * 2) + (sizeof(*init
) * (nent
+ 1)));
106 init
= (void *)(pack
+ 2);
109 for (i
= 0; i
< nent
; i
++) {
110 struct gf100_gr_init
*ent
= &init
[i
];
111 struct gk20a_fw_aiv
*av
= &((struct gk20a_fw_aiv
*)fuc
.data
)[i
];
113 ent
->addr
= av
->addr
;
114 ent
->data
= av
->data
;
122 gf100_gr_dtor_fw(&fuc
);
127 gk20a_gr_av_to_method(struct gf100_gr
*gr
, const char *fw_name
,
128 struct gf100_gr_pack
**ppack
)
130 struct gf100_gr_fuc fuc
;
131 struct gf100_gr_init
*init
;
132 struct gf100_gr_pack
*pack
;
133 /* We don't suppose we will initialize more than 16 classes here... */
134 static const unsigned int max_classes
= 16;
135 u32 classidx
= 0, prevclass
= 0;
140 ret
= gf100_gr_ctor_fw(gr
, fw_name
, &fuc
);
144 nent
= (fuc
.size
/ sizeof(struct gk20a_fw_av
));
146 pack
= vzalloc((sizeof(*pack
) * max_classes
) +
147 (sizeof(*init
) * (nent
+ 1)));
153 init
= (void *)(pack
+ max_classes
);
155 for (i
= 0; i
< nent
; i
++) {
156 struct gf100_gr_init
*ent
= &init
[i
];
157 struct gk20a_fw_av
*av
= &((struct gk20a_fw_av
*)fuc
.data
)[i
];
158 u32
class = av
->addr
& 0xffff;
159 u32 addr
= (av
->addr
& 0xffff0000) >> 14;
161 if (prevclass
!= class) {
162 pack
[classidx
].init
= ent
;
163 pack
[classidx
].type
= class;
165 if (++classidx
>= max_classes
) {
173 ent
->data
= av
->data
;
181 gf100_gr_dtor_fw(&fuc
);
186 gk20a_gr_wait_mem_scrubbing(struct gf100_gr
*gr
)
188 struct nvkm_subdev
*subdev
= &gr
->base
.engine
.subdev
;
189 struct nvkm_device
*device
= subdev
->device
;
191 if (nvkm_msec(device
, 2000,
192 if (!(nvkm_rd32(device
, 0x40910c) & 0x00000006))
195 nvkm_error(subdev
, "FECS mem scrubbing timeout\n");
199 if (nvkm_msec(device
, 2000,
200 if (!(nvkm_rd32(device
, 0x41a10c) & 0x00000006))
203 nvkm_error(subdev
, "GPCCS mem scrubbing timeout\n");
211 gk20a_gr_set_hww_esr_report_mask(struct gf100_gr
*gr
)
213 struct nvkm_device
*device
= gr
->base
.engine
.subdev
.device
;
214 nvkm_wr32(device
, 0x419e44, 0x1ffffe);
215 nvkm_wr32(device
, 0x419e4c, 0x7f);
219 gk20a_gr_init(struct gf100_gr
*gr
)
221 struct nvkm_device
*device
= gr
->base
.engine
.subdev
.device
;
222 const u32 magicgpc918
= DIV_ROUND_UP(0x00800000, gr
->tpc_total
);
223 u32 data
[TPC_MAX
/ 8] = {};
229 nvkm_wr32(device
, 0x40802c, 0x1);
231 gf100_gr_mmio(gr
, gr
->fuc_sw_nonctx
);
233 ret
= gk20a_gr_wait_mem_scrubbing(gr
);
237 ret
= gf100_gr_wait_idle(gr
);
241 /* MMU debug buffer */
242 if (gr
->func
->init_gpc_mmu
)
243 gr
->func
->init_gpc_mmu(gr
);
245 /* Set the PE as stream master */
246 nvkm_mask(device
, 0x503018, 0x1, 0x1);
249 memset(data
, 0x00, sizeof(data
));
250 memcpy(tpcnr
, gr
->tpc_nr
, sizeof(gr
->tpc_nr
));
251 for (i
= 0, gpc
= -1; i
< gr
->tpc_total
; i
++) {
253 gpc
= (gpc
+ 1) % gr
->gpc_nr
;
254 } while (!tpcnr
[gpc
]);
255 tpc
= gr
->tpc_nr
[gpc
] - tpcnr
[gpc
]--;
257 data
[i
/ 8] |= tpc
<< ((i
% 8) * 4);
260 nvkm_wr32(device
, GPC_BCAST(0x0980), data
[0]);
261 nvkm_wr32(device
, GPC_BCAST(0x0984), data
[1]);
262 nvkm_wr32(device
, GPC_BCAST(0x0988), data
[2]);
263 nvkm_wr32(device
, GPC_BCAST(0x098c), data
[3]);
265 for (gpc
= 0; gpc
< gr
->gpc_nr
; gpc
++) {
266 nvkm_wr32(device
, GPC_UNIT(gpc
, 0x0914),
267 gr
->screen_tile_row_offset
<< 8 | gr
->tpc_nr
[gpc
]);
268 nvkm_wr32(device
, GPC_UNIT(gpc
, 0x0910), 0x00040000 |
270 nvkm_wr32(device
, GPC_UNIT(gpc
, 0x0918), magicgpc918
);
273 nvkm_wr32(device
, GPC_BCAST(0x3fd4), magicgpc918
);
275 gr
->func
->init_rop_active_fbps(gr
);
277 /* Enable FIFO access */
278 nvkm_wr32(device
, 0x400500, 0x00010001);
280 /* Enable interrupts */
281 nvkm_wr32(device
, 0x400100, 0xffffffff);
282 nvkm_wr32(device
, 0x40013c, 0xffffffff);
284 /* Enable FECS error interrupts */
285 nvkm_wr32(device
, 0x409c24, 0x000f0000);
287 /* Enable hardware warning exceptions */
288 nvkm_wr32(device
, 0x404000, 0xc0000000);
289 nvkm_wr32(device
, 0x404600, 0xc0000000);
291 if (gr
->func
->set_hww_esr_report_mask
)
292 gr
->func
->set_hww_esr_report_mask(gr
);
294 /* Enable TPC exceptions per GPC */
295 nvkm_wr32(device
, 0x419d0c, 0x2);
296 nvkm_wr32(device
, 0x41ac94, (((1 << gr
->tpc_total
) - 1) & 0xff) << 16);
298 /* Reset and enable all exceptions */
299 nvkm_wr32(device
, 0x400108, 0xffffffff);
300 nvkm_wr32(device
, 0x400138, 0xffffffff);
301 nvkm_wr32(device
, 0x400118, 0xffffffff);
302 nvkm_wr32(device
, 0x400130, 0xffffffff);
303 nvkm_wr32(device
, 0x40011c, 0xffffffff);
304 nvkm_wr32(device
, 0x400134, 0xffffffff);
306 gf100_gr_zbc_init(gr
);
308 return gf100_gr_init_ctxctl(gr
);
311 static const struct gf100_gr_func
313 .init
= gk20a_gr_init
,
314 .init_rop_active_fbps
= gk104_gr_init_rop_active_fbps
,
315 .set_hww_esr_report_mask
= gk20a_gr_set_hww_esr_report_mask
,
316 .rops
= gf100_gr_rops
,
318 .grctx
= &gk20a_grctx
,
320 { -1, -1, FERMI_TWOD_A
},
321 { -1, -1, KEPLER_INLINE_TO_MEMORY_A
},
322 { -1, -1, KEPLER_C
, &gf100_fermi
},
323 { -1, -1, KEPLER_COMPUTE_A
},
329 gk20a_gr_new(struct nvkm_device
*device
, int index
, struct nvkm_gr
**pgr
)
334 if (!(gr
= kzalloc(sizeof(*gr
), GFP_KERNEL
)))
338 ret
= gf100_gr_ctor(&gk20a_gr
, device
, index
, gr
);
342 if (gf100_gr_ctor_fw(gr
, "fecs_inst", &gr
->fuc409c
) ||
343 gf100_gr_ctor_fw(gr
, "fecs_data", &gr
->fuc409d
) ||
344 gf100_gr_ctor_fw(gr
, "gpccs_inst", &gr
->fuc41ac
) ||
345 gf100_gr_ctor_fw(gr
, "gpccs_data", &gr
->fuc41ad
))
348 ret
= gk20a_gr_av_to_init(gr
, "sw_nonctx", &gr
->fuc_sw_nonctx
);
352 ret
= gk20a_gr_aiv_to_init(gr
, "sw_ctx", &gr
->fuc_sw_ctx
);
356 ret
= gk20a_gr_av_to_init(gr
, "sw_bundle_init", &gr
->fuc_bundle
);
360 ret
= gk20a_gr_av_to_method(gr
, "sw_method_init", &gr
->fuc_method
);