1 /* 2 * Copyright 2012 Red Hat 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 * Authors: Ben Skeggs 23 */ 24 #include "gf100.h" 25 #include "ctxgf100.h" 26 #include "fuc/os.h" 27 28 #include <core/client.h> 29 #include <core/firmware.h> 30 #include <core/option.h> 31 #include <subdev/acr.h> 32 #include <subdev/fb.h> 33 #include <subdev/mc.h> 34 #include <subdev/pmu.h> 35 #include <subdev/therm.h> 36 #include <subdev/timer.h> 37 #include <engine/fifo.h> 38 39 #include <nvif/class.h> 40 #include <nvif/cl9097.h> 41 #include <nvif/if900d.h> 42 #include <nvif/unpack.h> 43 44 /******************************************************************************* 45 * Zero Bandwidth Clear 46 ******************************************************************************/ 47 48 static void 49 gf100_gr_zbc_clear_color(struct gf100_gr *gr, int zbc) 50 { 51 struct nvkm_device *device = gr->base.engine.subdev.device; 52 if (gr->zbc_color[zbc].format) { 53 nvkm_wr32(device, 0x405804, gr->zbc_color[zbc].ds[0]); 54 nvkm_wr32(device, 0x405808, gr->zbc_color[zbc].ds[1]); 55 nvkm_wr32(device, 0x40580c, gr->zbc_color[zbc].ds[2]); 56 nvkm_wr32(device, 0x405810, gr->zbc_color[zbc].ds[3]); 57 } 58 nvkm_wr32(device, 0x405814, gr->zbc_color[zbc].format); 59 nvkm_wr32(device, 0x405820, zbc); 60 nvkm_wr32(device, 0x405824, 0x00000004); /* TRIGGER | WRITE | COLOR */ 61 } 62 63 static int 64 gf100_gr_zbc_color_get(struct gf100_gr *gr, int format, 65 const u32 ds[4], const u32 l2[4]) 66 { 67 struct nvkm_ltc *ltc = gr->base.engine.subdev.device->ltc; 68 int zbc = -ENOSPC, i; 69 70 for (i = ltc->zbc_color_min; i <= ltc->zbc_color_max; i++) { 71 if (gr->zbc_color[i].format) { 72 if (gr->zbc_color[i].format != format) 73 continue; 74 if (memcmp(gr->zbc_color[i].ds, ds, sizeof( 75 gr->zbc_color[i].ds))) 76 continue; 77 if (memcmp(gr->zbc_color[i].l2, l2, sizeof( 78 gr->zbc_color[i].l2))) { 79 WARN_ON(1); 80 return -EINVAL; 81 } 82 return i; 83 } else { 84 zbc = (zbc < 0) ? i : zbc; 85 } 86 } 87 88 if (zbc < 0) 89 return zbc; 90 91 memcpy(gr->zbc_color[zbc].ds, ds, sizeof(gr->zbc_color[zbc].ds)); 92 memcpy(gr->zbc_color[zbc].l2, l2, sizeof(gr->zbc_color[zbc].l2)); 93 gr->zbc_color[zbc].format = format; 94 nvkm_ltc_zbc_color_get(ltc, zbc, l2); 95 gr->func->zbc->clear_color(gr, zbc); 96 return zbc; 97 } 98 99 static void 100 gf100_gr_zbc_clear_depth(struct gf100_gr *gr, int zbc) 101 { 102 struct nvkm_device *device = gr->base.engine.subdev.device; 103 if (gr->zbc_depth[zbc].format) 104 nvkm_wr32(device, 0x405818, gr->zbc_depth[zbc].ds); 105 nvkm_wr32(device, 0x40581c, gr->zbc_depth[zbc].format); 106 nvkm_wr32(device, 0x405820, zbc); 107 nvkm_wr32(device, 0x405824, 0x00000005); /* TRIGGER | WRITE | DEPTH */ 108 } 109 110 static int 111 gf100_gr_zbc_depth_get(struct gf100_gr *gr, int format, 112 const u32 ds, const u32 l2) 113 { 114 struct nvkm_ltc *ltc = gr->base.engine.subdev.device->ltc; 115 int zbc = -ENOSPC, i; 116 117 for (i = ltc->zbc_depth_min; i <= ltc->zbc_depth_max; i++) { 118 if (gr->zbc_depth[i].format) { 119 if (gr->zbc_depth[i].format != format) 120 continue; 121 if (gr->zbc_depth[i].ds != ds) 122 continue; 123 if (gr->zbc_depth[i].l2 != l2) { 124 WARN_ON(1); 125 return -EINVAL; 126 } 127 return i; 128 } else { 129 zbc = (zbc < 0) ? i : zbc; 130 } 131 } 132 133 if (zbc < 0) 134 return zbc; 135 136 gr->zbc_depth[zbc].format = format; 137 gr->zbc_depth[zbc].ds = ds; 138 gr->zbc_depth[zbc].l2 = l2; 139 nvkm_ltc_zbc_depth_get(ltc, zbc, l2); 140 gr->func->zbc->clear_depth(gr, zbc); 141 return zbc; 142 } 143 144 const struct gf100_gr_func_zbc 145 gf100_gr_zbc = { 146 .clear_color = gf100_gr_zbc_clear_color, 147 .clear_depth = gf100_gr_zbc_clear_depth, 148 }; 149 150 /******************************************************************************* 151 * Graphics object classes 152 ******************************************************************************/ 153 #define gf100_gr_object(p) container_of((p), struct gf100_gr_object, object) 154 155 struct gf100_gr_object { 156 struct nvkm_object object; 157 struct gf100_gr_chan *chan; 158 }; 159 160 static int 161 gf100_fermi_mthd_zbc_color(struct nvkm_object *object, void *data, u32 size) 162 { 163 struct gf100_gr *gr = gf100_gr(nvkm_gr(object->engine)); 164 union { 165 struct fermi_a_zbc_color_v0 v0; 166 } *args = data; 167 int ret = -ENOSYS; 168 169 if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { 170 switch (args->v0.format) { 171 case FERMI_A_ZBC_COLOR_V0_FMT_ZERO: 172 case FERMI_A_ZBC_COLOR_V0_FMT_UNORM_ONE: 173 case FERMI_A_ZBC_COLOR_V0_FMT_RF32_GF32_BF32_AF32: 174 case FERMI_A_ZBC_COLOR_V0_FMT_R16_G16_B16_A16: 175 case FERMI_A_ZBC_COLOR_V0_FMT_RN16_GN16_BN16_AN16: 176 case FERMI_A_ZBC_COLOR_V0_FMT_RS16_GS16_BS16_AS16: 177 case FERMI_A_ZBC_COLOR_V0_FMT_RU16_GU16_BU16_AU16: 178 case FERMI_A_ZBC_COLOR_V0_FMT_RF16_GF16_BF16_AF16: 179 case FERMI_A_ZBC_COLOR_V0_FMT_A8R8G8B8: 180 case FERMI_A_ZBC_COLOR_V0_FMT_A8RL8GL8BL8: 181 case FERMI_A_ZBC_COLOR_V0_FMT_A2B10G10R10: 182 case FERMI_A_ZBC_COLOR_V0_FMT_AU2BU10GU10RU10: 183 case FERMI_A_ZBC_COLOR_V0_FMT_A8B8G8R8: 184 case FERMI_A_ZBC_COLOR_V0_FMT_A8BL8GL8RL8: 185 case FERMI_A_ZBC_COLOR_V0_FMT_AN8BN8GN8RN8: 186 case FERMI_A_ZBC_COLOR_V0_FMT_AS8BS8GS8RS8: 187 case FERMI_A_ZBC_COLOR_V0_FMT_AU8BU8GU8RU8: 188 case FERMI_A_ZBC_COLOR_V0_FMT_A2R10G10B10: 189 case FERMI_A_ZBC_COLOR_V0_FMT_BF10GF11RF11: 190 ret = gf100_gr_zbc_color_get(gr, args->v0.format, 191 args->v0.ds, 192 args->v0.l2); 193 if (ret >= 0) { 194 args->v0.index = ret; 195 return 0; 196 } 197 break; 198 default: 199 return -EINVAL; 200 } 201 } 202 203 return ret; 204 } 205 206 static int 207 gf100_fermi_mthd_zbc_depth(struct nvkm_object *object, void *data, u32 size) 208 { 209 struct gf100_gr *gr = gf100_gr(nvkm_gr(object->engine)); 210 union { 211 struct fermi_a_zbc_depth_v0 v0; 212 } *args = data; 213 int ret = -ENOSYS; 214 215 if (!(ret = nvif_unpack(ret, &data, &size, args->v0, 0, 0, false))) { 216 switch (args->v0.format) { 217 case FERMI_A_ZBC_DEPTH_V0_FMT_FP32: 218 ret = gf100_gr_zbc_depth_get(gr, args->v0.format, 219 args->v0.ds, 220 args->v0.l2); 221 return (ret >= 0) ? 0 : -ENOSPC; 222 default: 223 return -EINVAL; 224 } 225 } 226 227 return ret; 228 } 229 230 static int 231 gf100_fermi_mthd(struct nvkm_object *object, u32 mthd, void *data, u32 size) 232 { 233 nvif_ioctl(object, "fermi mthd %08x\n", mthd); 234 switch (mthd) { 235 case FERMI_A_ZBC_COLOR: 236 return gf100_fermi_mthd_zbc_color(object, data, size); 237 case FERMI_A_ZBC_DEPTH: 238 return gf100_fermi_mthd_zbc_depth(object, data, size); 239 default: 240 break; 241 } 242 return -EINVAL; 243 } 244 245 const struct nvkm_object_func 246 gf100_fermi = { 247 .mthd = gf100_fermi_mthd, 248 }; 249 250 static void 251 gf100_gr_mthd_set_shader_exceptions(struct nvkm_device *device, u32 data) 252 { 253 nvkm_wr32(device, 0x419e44, data ? 0xffffffff : 0x00000000); 254 nvkm_wr32(device, 0x419e4c, data ? 0xffffffff : 0x00000000); 255 } 256 257 static bool 258 gf100_gr_mthd_sw(struct nvkm_device *device, u16 class, u32 mthd, u32 data) 259 { 260 switch (class & 0x00ff) { 261 case 0x97: 262 case 0xc0: 263 switch (mthd) { 264 case 0x1528: 265 gf100_gr_mthd_set_shader_exceptions(device, data); 266 return true; 267 default: 268 break; 269 } 270 break; 271 default: 272 break; 273 } 274 return false; 275 } 276 277 static const struct nvkm_object_func 278 gf100_gr_object_func = { 279 }; 280 281 static int 282 gf100_gr_object_new(const struct nvkm_oclass *oclass, void *data, u32 size, 283 struct nvkm_object **pobject) 284 { 285 struct gf100_gr_chan *chan = gf100_gr_chan(oclass->parent); 286 struct gf100_gr_object *object; 287 288 if (!(object = kzalloc(sizeof(*object), GFP_KERNEL))) 289 return -ENOMEM; 290 *pobject = &object->object; 291 292 nvkm_object_ctor(oclass->base.func ? oclass->base.func : 293 &gf100_gr_object_func, oclass, &object->object); 294 object->chan = chan; 295 return 0; 296 } 297 298 static int 299 gf100_gr_object_get(struct nvkm_gr *base, int index, struct nvkm_sclass *sclass) 300 { 301 struct gf100_gr *gr = gf100_gr(base); 302 int c = 0; 303 304 while (gr->func->sclass[c].oclass) { 305 if (c++ == index) { 306 *sclass = gr->func->sclass[index]; 307 sclass->ctor = gf100_gr_object_new; 308 return index; 309 } 310 } 311 312 return c; 313 } 314 315 /******************************************************************************* 316 * PGRAPH context 317 ******************************************************************************/ 318 319 static int 320 gf100_gr_chan_bind(struct nvkm_object *object, struct nvkm_gpuobj *parent, 321 int align, struct nvkm_gpuobj **pgpuobj) 322 { 323 struct gf100_gr_chan *chan = gf100_gr_chan(object); 324 struct gf100_gr *gr = chan->gr; 325 int ret, i; 326 327 ret = nvkm_gpuobj_new(gr->base.engine.subdev.device, gr->size, 328 align, false, parent, pgpuobj); 329 if (ret) 330 return ret; 331 332 nvkm_kmap(*pgpuobj); 333 for (i = 0; i < gr->size; i += 4) 334 nvkm_wo32(*pgpuobj, i, gr->data[i / 4]); 335 336 if (!gr->firmware) { 337 nvkm_wo32(*pgpuobj, 0x00, chan->mmio_nr / 2); 338 nvkm_wo32(*pgpuobj, 0x04, chan->mmio_vma->addr >> 8); 339 } else { 340 nvkm_wo32(*pgpuobj, 0xf4, 0); 341 nvkm_wo32(*pgpuobj, 0xf8, 0); 342 nvkm_wo32(*pgpuobj, 0x10, chan->mmio_nr / 2); 343 nvkm_wo32(*pgpuobj, 0x14, lower_32_bits(chan->mmio_vma->addr)); 344 nvkm_wo32(*pgpuobj, 0x18, upper_32_bits(chan->mmio_vma->addr)); 345 nvkm_wo32(*pgpuobj, 0x1c, 1); 346 nvkm_wo32(*pgpuobj, 0x20, 0); 347 nvkm_wo32(*pgpuobj, 0x28, 0); 348 nvkm_wo32(*pgpuobj, 0x2c, 0); 349 } 350 nvkm_done(*pgpuobj); 351 return 0; 352 } 353 354 static void * 355 gf100_gr_chan_dtor(struct nvkm_object *object) 356 { 357 struct gf100_gr_chan *chan = gf100_gr_chan(object); 358 359 nvkm_vmm_put(chan->vmm, &chan->mmio_vma); 360 nvkm_memory_unref(&chan->mmio); 361 362 nvkm_vmm_put(chan->vmm, &chan->attrib_cb); 363 nvkm_vmm_put(chan->vmm, &chan->unknown); 364 nvkm_vmm_put(chan->vmm, &chan->bundle_cb); 365 nvkm_vmm_put(chan->vmm, &chan->pagepool); 366 nvkm_vmm_unref(&chan->vmm); 367 return chan; 368 } 369 370 static const struct nvkm_object_func 371 gf100_gr_chan = { 372 .dtor = gf100_gr_chan_dtor, 373 .bind = gf100_gr_chan_bind, 374 }; 375 376 static int 377 gf100_gr_chan_new(struct nvkm_gr *base, struct nvkm_chan *fifoch, 378 const struct nvkm_oclass *oclass, 379 struct nvkm_object **pobject) 380 { 381 struct gf100_gr *gr = gf100_gr(base); 382 struct gf100_gr_chan *chan; 383 struct gf100_vmm_map_v0 args = { .priv = 1 }; 384 struct nvkm_device *device = gr->base.engine.subdev.device; 385 int ret; 386 387 if (!(chan = kzalloc(sizeof(*chan), GFP_KERNEL))) 388 return -ENOMEM; 389 nvkm_object_ctor(&gf100_gr_chan, oclass, &chan->object); 390 chan->gr = gr; 391 chan->vmm = nvkm_vmm_ref(fifoch->vmm); 392 *pobject = &chan->object; 393 394 /* Map pagepool. */ 395 ret = nvkm_vmm_get(chan->vmm, 12, nvkm_memory_size(gr->pagepool), &chan->pagepool); 396 if (ret) 397 return ret; 398 399 ret = nvkm_memory_map(gr->pagepool, 0, chan->vmm, chan->pagepool, &args, sizeof(args)); 400 if (ret) 401 return ret; 402 403 /* Map bundle circular buffer. */ 404 ret = nvkm_vmm_get(chan->vmm, 12, nvkm_memory_size(gr->bundle_cb), &chan->bundle_cb); 405 if (ret) 406 return ret; 407 408 ret = nvkm_memory_map(gr->bundle_cb, 0, chan->vmm, chan->bundle_cb, &args, sizeof(args)); 409 if (ret) 410 return ret; 411 412 /* Map attribute circular buffer. */ 413 ret = nvkm_vmm_get(chan->vmm, 12, nvkm_memory_size(gr->attrib_cb), &chan->attrib_cb); 414 if (ret) 415 return ret; 416 417 if (device->card_type < GP100) { 418 ret = nvkm_memory_map(gr->attrib_cb, 0, chan->vmm, chan->attrib_cb, NULL, 0); 419 if (ret) 420 return ret; 421 } else { 422 ret = nvkm_memory_map(gr->attrib_cb, 0, chan->vmm, chan->attrib_cb, 423 &args, sizeof(args)); 424 if (ret) 425 return ret; 426 } 427 428 /* Map some context buffer of unknown purpose. */ 429 if (gr->func->grctx->unknown_size) { 430 ret = nvkm_vmm_get(chan->vmm, 12, nvkm_memory_size(gr->unknown), &chan->unknown); 431 if (ret) 432 return ret; 433 434 ret = nvkm_memory_map(gr->unknown, 0, chan->vmm, chan->unknown, 435 &args, sizeof(args)); 436 if (ret) 437 return ret; 438 } 439 440 /* Generate golden context image. */ 441 mutex_lock(&gr->fecs.mutex); 442 if (gr->data == NULL) { 443 ret = gf100_grctx_generate(gr, chan, fifoch->inst); 444 if (ret) { 445 nvkm_error(&base->engine.subdev, "failed to construct context\n"); 446 mutex_unlock(&gr->fecs.mutex); 447 return ret; 448 } 449 } 450 mutex_unlock(&gr->fecs.mutex); 451 452 /* allocate memory for a "mmio list" buffer that's used by the HUB 453 * fuc to modify some per-context register settings on first load 454 * of the context. 455 */ 456 ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, 0x1000, 0x100, 457 false, &chan->mmio); 458 if (ret) 459 return ret; 460 461 ret = nvkm_vmm_get(fifoch->vmm, 12, 0x1000, &chan->mmio_vma); 462 if (ret) 463 return ret; 464 465 ret = nvkm_memory_map(chan->mmio, 0, fifoch->vmm, 466 chan->mmio_vma, &args, sizeof(args)); 467 if (ret) 468 return ret; 469 470 /* finally, fill in the mmio list and point the context at it */ 471 nvkm_kmap(chan->mmio); 472 gr->func->grctx->pagepool(chan, chan->pagepool->addr); 473 gr->func->grctx->bundle(chan, chan->bundle_cb->addr, gr->func->grctx->bundle_size); 474 gr->func->grctx->attrib_cb(chan, chan->attrib_cb->addr, gr->func->grctx->attrib_cb_size(gr)); 475 gr->func->grctx->attrib(chan); 476 if (gr->func->grctx->patch_ltc) 477 gr->func->grctx->patch_ltc(chan); 478 if (gr->func->grctx->unknown_size) 479 gr->func->grctx->unknown(chan, chan->unknown->addr, gr->func->grctx->unknown_size); 480 nvkm_done(chan->mmio); 481 return 0; 482 } 483 484 /******************************************************************************* 485 * PGRAPH register lists 486 ******************************************************************************/ 487 488 const struct gf100_gr_init 489 gf100_gr_init_main_0[] = { 490 { 0x400080, 1, 0x04, 0x003083c2 }, 491 { 0x400088, 1, 0x04, 0x00006fe7 }, 492 { 0x40008c, 1, 0x04, 0x00000000 }, 493 { 0x400090, 1, 0x04, 0x00000030 }, 494 { 0x40013c, 1, 0x04, 0x013901f7 }, 495 { 0x400140, 1, 0x04, 0x00000100 }, 496 { 0x400144, 1, 0x04, 0x00000000 }, 497 { 0x400148, 1, 0x04, 0x00000110 }, 498 { 0x400138, 1, 0x04, 0x00000000 }, 499 { 0x400130, 2, 0x04, 0x00000000 }, 500 { 0x400124, 1, 0x04, 0x00000002 }, 501 {} 502 }; 503 504 const struct gf100_gr_init 505 gf100_gr_init_fe_0[] = { 506 { 0x40415c, 1, 0x04, 0x00000000 }, 507 { 0x404170, 1, 0x04, 0x00000000 }, 508 {} 509 }; 510 511 const struct gf100_gr_init 512 gf100_gr_init_pri_0[] = { 513 { 0x404488, 2, 0x04, 0x00000000 }, 514 {} 515 }; 516 517 const struct gf100_gr_init 518 gf100_gr_init_rstr2d_0[] = { 519 { 0x407808, 1, 0x04, 0x00000000 }, 520 {} 521 }; 522 523 const struct gf100_gr_init 524 gf100_gr_init_pd_0[] = { 525 { 0x406024, 1, 0x04, 0x00000000 }, 526 {} 527 }; 528 529 const struct gf100_gr_init 530 gf100_gr_init_ds_0[] = { 531 { 0x405844, 1, 0x04, 0x00ffffff }, 532 { 0x405850, 1, 0x04, 0x00000000 }, 533 { 0x405908, 1, 0x04, 0x00000000 }, 534 {} 535 }; 536 537 const struct gf100_gr_init 538 gf100_gr_init_scc_0[] = { 539 { 0x40803c, 1, 0x04, 0x00000000 }, 540 {} 541 }; 542 543 const struct gf100_gr_init 544 gf100_gr_init_prop_0[] = { 545 { 0x4184a0, 1, 0x04, 0x00000000 }, 546 {} 547 }; 548 549 const struct gf100_gr_init 550 gf100_gr_init_gpc_unk_0[] = { 551 { 0x418604, 1, 0x04, 0x00000000 }, 552 { 0x418680, 1, 0x04, 0x00000000 }, 553 { 0x418714, 1, 0x04, 0x80000000 }, 554 { 0x418384, 1, 0x04, 0x00000000 }, 555 {} 556 }; 557 558 const struct gf100_gr_init 559 gf100_gr_init_setup_0[] = { 560 { 0x418814, 3, 0x04, 0x00000000 }, 561 {} 562 }; 563 564 const struct gf100_gr_init 565 gf100_gr_init_crstr_0[] = { 566 { 0x418b04, 1, 0x04, 0x00000000 }, 567 {} 568 }; 569 570 const struct gf100_gr_init 571 gf100_gr_init_setup_1[] = { 572 { 0x4188c8, 1, 0x04, 0x80000000 }, 573 { 0x4188cc, 1, 0x04, 0x00000000 }, 574 { 0x4188d0, 1, 0x04, 0x00010000 }, 575 { 0x4188d4, 1, 0x04, 0x00000001 }, 576 {} 577 }; 578 579 const struct gf100_gr_init 580 gf100_gr_init_zcull_0[] = { 581 { 0x418910, 1, 0x04, 0x00010001 }, 582 { 0x418914, 1, 0x04, 0x00000301 }, 583 { 0x418918, 1, 0x04, 0x00800000 }, 584 { 0x418980, 1, 0x04, 0x77777770 }, 585 { 0x418984, 3, 0x04, 0x77777777 }, 586 {} 587 }; 588 589 const struct gf100_gr_init 590 gf100_gr_init_gpm_0[] = { 591 { 0x418c04, 1, 0x04, 0x00000000 }, 592 { 0x418c88, 1, 0x04, 0x00000000 }, 593 {} 594 }; 595 596 const struct gf100_gr_init 597 gf100_gr_init_gpc_unk_1[] = { 598 { 0x418d00, 1, 0x04, 0x00000000 }, 599 { 0x418f08, 1, 0x04, 0x00000000 }, 600 { 0x418e00, 1, 0x04, 0x00000050 }, 601 { 0x418e08, 1, 0x04, 0x00000000 }, 602 {} 603 }; 604 605 const struct gf100_gr_init 606 gf100_gr_init_gcc_0[] = { 607 { 0x41900c, 1, 0x04, 0x00000000 }, 608 { 0x419018, 1, 0x04, 0x00000000 }, 609 {} 610 }; 611 612 const struct gf100_gr_init 613 gf100_gr_init_tpccs_0[] = { 614 { 0x419d08, 2, 0x04, 0x00000000 }, 615 { 0x419d10, 1, 0x04, 0x00000014 }, 616 {} 617 }; 618 619 const struct gf100_gr_init 620 gf100_gr_init_tex_0[] = { 621 { 0x419ab0, 1, 0x04, 0x00000000 }, 622 { 0x419ab8, 1, 0x04, 0x000000e7 }, 623 { 0x419abc, 2, 0x04, 0x00000000 }, 624 {} 625 }; 626 627 const struct gf100_gr_init 628 gf100_gr_init_pe_0[] = { 629 { 0x41980c, 3, 0x04, 0x00000000 }, 630 { 0x419844, 1, 0x04, 0x00000000 }, 631 { 0x41984c, 1, 0x04, 0x00005bc5 }, 632 { 0x419850, 4, 0x04, 0x00000000 }, 633 {} 634 }; 635 636 const struct gf100_gr_init 637 gf100_gr_init_l1c_0[] = { 638 { 0x419c98, 1, 0x04, 0x00000000 }, 639 { 0x419ca8, 1, 0x04, 0x80000000 }, 640 { 0x419cb4, 1, 0x04, 0x00000000 }, 641 { 0x419cb8, 1, 0x04, 0x00008bf4 }, 642 { 0x419cbc, 1, 0x04, 0x28137606 }, 643 { 0x419cc0, 2, 0x04, 0x00000000 }, 644 {} 645 }; 646 647 const struct gf100_gr_init 648 gf100_gr_init_wwdx_0[] = { 649 { 0x419bd4, 1, 0x04, 0x00800000 }, 650 { 0x419bdc, 1, 0x04, 0x00000000 }, 651 {} 652 }; 653 654 const struct gf100_gr_init 655 gf100_gr_init_tpccs_1[] = { 656 { 0x419d2c, 1, 0x04, 0x00000000 }, 657 {} 658 }; 659 660 const struct gf100_gr_init 661 gf100_gr_init_mpc_0[] = { 662 { 0x419c0c, 1, 0x04, 0x00000000 }, 663 {} 664 }; 665 666 static const struct gf100_gr_init 667 gf100_gr_init_sm_0[] = { 668 { 0x419e00, 1, 0x04, 0x00000000 }, 669 { 0x419ea0, 1, 0x04, 0x00000000 }, 670 { 0x419ea4, 1, 0x04, 0x00000100 }, 671 { 0x419ea8, 1, 0x04, 0x00001100 }, 672 { 0x419eac, 1, 0x04, 0x11100702 }, 673 { 0x419eb0, 1, 0x04, 0x00000003 }, 674 { 0x419eb4, 4, 0x04, 0x00000000 }, 675 { 0x419ec8, 1, 0x04, 0x06060618 }, 676 { 0x419ed0, 1, 0x04, 0x0eff0e38 }, 677 { 0x419ed4, 1, 0x04, 0x011104f1 }, 678 { 0x419edc, 1, 0x04, 0x00000000 }, 679 { 0x419f00, 1, 0x04, 0x00000000 }, 680 { 0x419f2c, 1, 0x04, 0x00000000 }, 681 {} 682 }; 683 684 const struct gf100_gr_init 685 gf100_gr_init_be_0[] = { 686 { 0x40880c, 1, 0x04, 0x00000000 }, 687 { 0x408910, 9, 0x04, 0x00000000 }, 688 { 0x408950, 1, 0x04, 0x00000000 }, 689 { 0x408954, 1, 0x04, 0x0000ffff }, 690 { 0x408984, 1, 0x04, 0x00000000 }, 691 { 0x408988, 1, 0x04, 0x08040201 }, 692 { 0x40898c, 1, 0x04, 0x80402010 }, 693 {} 694 }; 695 696 const struct gf100_gr_init 697 gf100_gr_init_fe_1[] = { 698 { 0x4040f0, 1, 0x04, 0x00000000 }, 699 {} 700 }; 701 702 const struct gf100_gr_init 703 gf100_gr_init_pe_1[] = { 704 { 0x419880, 1, 0x04, 0x00000002 }, 705 {} 706 }; 707 708 static const struct gf100_gr_pack 709 gf100_gr_pack_mmio[] = { 710 { gf100_gr_init_main_0 }, 711 { gf100_gr_init_fe_0 }, 712 { gf100_gr_init_pri_0 }, 713 { gf100_gr_init_rstr2d_0 }, 714 { gf100_gr_init_pd_0 }, 715 { gf100_gr_init_ds_0 }, 716 { gf100_gr_init_scc_0 }, 717 { gf100_gr_init_prop_0 }, 718 { gf100_gr_init_gpc_unk_0 }, 719 { gf100_gr_init_setup_0 }, 720 { gf100_gr_init_crstr_0 }, 721 { gf100_gr_init_setup_1 }, 722 { gf100_gr_init_zcull_0 }, 723 { gf100_gr_init_gpm_0 }, 724 { gf100_gr_init_gpc_unk_1 }, 725 { gf100_gr_init_gcc_0 }, 726 { gf100_gr_init_tpccs_0 }, 727 { gf100_gr_init_tex_0 }, 728 { gf100_gr_init_pe_0 }, 729 { gf100_gr_init_l1c_0 }, 730 { gf100_gr_init_wwdx_0 }, 731 { gf100_gr_init_tpccs_1 }, 732 { gf100_gr_init_mpc_0 }, 733 { gf100_gr_init_sm_0 }, 734 { gf100_gr_init_be_0 }, 735 { gf100_gr_init_fe_1 }, 736 { gf100_gr_init_pe_1 }, 737 {} 738 }; 739 740 /******************************************************************************* 741 * PGRAPH engine/subdev functions 742 ******************************************************************************/ 743 744 static u32 745 gf100_gr_ctxsw_inst(struct nvkm_gr *gr) 746 { 747 return nvkm_rd32(gr->engine.subdev.device, 0x409b00); 748 } 749 750 static int 751 gf100_gr_fecs_ctrl_ctxsw(struct gf100_gr *gr, u32 mthd) 752 { 753 struct nvkm_device *device = gr->base.engine.subdev.device; 754 755 nvkm_wr32(device, 0x409804, 0xffffffff); 756 nvkm_wr32(device, 0x409800, 0x00000000); 757 nvkm_wr32(device, 0x409500, 0xffffffff); 758 nvkm_wr32(device, 0x409504, mthd); 759 nvkm_msec(device, 2000, 760 u32 stat = nvkm_rd32(device, 0x409804); 761 if (stat == 0x00000002) 762 return -EIO; 763 if (stat == 0x00000001) 764 return 0; 765 ); 766 767 return -ETIMEDOUT; 768 } 769 770 static int 771 gf100_gr_fecs_start_ctxsw(struct nvkm_gr *base) 772 { 773 struct gf100_gr *gr = gf100_gr(base); 774 int ret = 0; 775 776 mutex_lock(&gr->fecs.mutex); 777 if (!--gr->fecs.disable) { 778 if (WARN_ON(ret = gf100_gr_fecs_ctrl_ctxsw(gr, 0x39))) 779 gr->fecs.disable++; 780 } 781 mutex_unlock(&gr->fecs.mutex); 782 return ret; 783 } 784 785 static int 786 gf100_gr_fecs_stop_ctxsw(struct nvkm_gr *base) 787 { 788 struct gf100_gr *gr = gf100_gr(base); 789 int ret = 0; 790 791 mutex_lock(&gr->fecs.mutex); 792 if (!gr->fecs.disable++) { 793 if (WARN_ON(ret = gf100_gr_fecs_ctrl_ctxsw(gr, 0x38))) 794 gr->fecs.disable--; 795 } 796 mutex_unlock(&gr->fecs.mutex); 797 return ret; 798 } 799 800 static int 801 gf100_gr_fecs_halt_pipeline(struct gf100_gr *gr) 802 { 803 int ret = 0; 804 805 if (gr->firmware) { 806 mutex_lock(&gr->fecs.mutex); 807 ret = gf100_gr_fecs_ctrl_ctxsw(gr, 0x04); 808 mutex_unlock(&gr->fecs.mutex); 809 } 810 811 return ret; 812 } 813 814 int 815 gf100_gr_fecs_wfi_golden_save(struct gf100_gr *gr, u32 inst) 816 { 817 struct nvkm_device *device = gr->base.engine.subdev.device; 818 819 nvkm_mask(device, 0x409800, 0x00000003, 0x00000000); 820 nvkm_wr32(device, 0x409500, inst); 821 nvkm_wr32(device, 0x409504, 0x00000009); 822 nvkm_msec(device, 2000, 823 u32 stat = nvkm_rd32(device, 0x409800); 824 if (stat & 0x00000002) 825 return -EIO; 826 if (stat & 0x00000001) 827 return 0; 828 ); 829 830 return -ETIMEDOUT; 831 } 832 833 int 834 gf100_gr_fecs_bind_pointer(struct gf100_gr *gr, u32 inst) 835 { 836 struct nvkm_device *device = gr->base.engine.subdev.device; 837 838 nvkm_mask(device, 0x409800, 0x00000030, 0x00000000); 839 nvkm_wr32(device, 0x409500, inst); 840 nvkm_wr32(device, 0x409504, 0x00000003); 841 nvkm_msec(device, 2000, 842 u32 stat = nvkm_rd32(device, 0x409800); 843 if (stat & 0x00000020) 844 return -EIO; 845 if (stat & 0x00000010) 846 return 0; 847 ); 848 849 return -ETIMEDOUT; 850 } 851 852 static int 853 gf100_gr_fecs_set_reglist_virtual_address(struct gf100_gr *gr, u64 addr) 854 { 855 struct nvkm_device *device = gr->base.engine.subdev.device; 856 857 nvkm_wr32(device, 0x409810, addr >> 8); 858 nvkm_wr32(device, 0x409800, 0x00000000); 859 nvkm_wr32(device, 0x409500, 0x00000001); 860 nvkm_wr32(device, 0x409504, 0x00000032); 861 nvkm_msec(device, 2000, 862 if (nvkm_rd32(device, 0x409800) == 0x00000001) 863 return 0; 864 ); 865 866 return -ETIMEDOUT; 867 } 868 869 static int 870 gf100_gr_fecs_set_reglist_bind_instance(struct gf100_gr *gr, u32 inst) 871 { 872 struct nvkm_device *device = gr->base.engine.subdev.device; 873 874 nvkm_wr32(device, 0x409810, inst); 875 nvkm_wr32(device, 0x409800, 0x00000000); 876 nvkm_wr32(device, 0x409500, 0x00000001); 877 nvkm_wr32(device, 0x409504, 0x00000031); 878 nvkm_msec(device, 2000, 879 if (nvkm_rd32(device, 0x409800) == 0x00000001) 880 return 0; 881 ); 882 883 return -ETIMEDOUT; 884 } 885 886 static int 887 gf100_gr_fecs_discover_reglist_image_size(struct gf100_gr *gr, u32 *psize) 888 { 889 struct nvkm_device *device = gr->base.engine.subdev.device; 890 891 nvkm_wr32(device, 0x409800, 0x00000000); 892 nvkm_wr32(device, 0x409500, 0x00000001); 893 nvkm_wr32(device, 0x409504, 0x00000030); 894 nvkm_msec(device, 2000, 895 if ((*psize = nvkm_rd32(device, 0x409800))) 896 return 0; 897 ); 898 899 return -ETIMEDOUT; 900 } 901 902 static int 903 gf100_gr_fecs_elpg_bind(struct gf100_gr *gr) 904 { 905 u32 size; 906 int ret; 907 908 ret = gf100_gr_fecs_discover_reglist_image_size(gr, &size); 909 if (ret) 910 return ret; 911 912 /*XXX: We need to allocate + map the above into PMU's inst block, 913 * which which means we probably need a proper PMU before we 914 * even bother. 915 */ 916 917 ret = gf100_gr_fecs_set_reglist_bind_instance(gr, 0); 918 if (ret) 919 return ret; 920 921 return gf100_gr_fecs_set_reglist_virtual_address(gr, 0); 922 } 923 924 static int 925 gf100_gr_fecs_discover_pm_image_size(struct gf100_gr *gr, u32 *psize) 926 { 927 struct nvkm_device *device = gr->base.engine.subdev.device; 928 929 nvkm_wr32(device, 0x409800, 0x00000000); 930 nvkm_wr32(device, 0x409500, 0x00000000); 931 nvkm_wr32(device, 0x409504, 0x00000025); 932 nvkm_msec(device, 2000, 933 if ((*psize = nvkm_rd32(device, 0x409800))) 934 return 0; 935 ); 936 937 return -ETIMEDOUT; 938 } 939 940 static int 941 gf100_gr_fecs_discover_zcull_image_size(struct gf100_gr *gr, u32 *psize) 942 { 943 struct nvkm_device *device = gr->base.engine.subdev.device; 944 945 nvkm_wr32(device, 0x409800, 0x00000000); 946 nvkm_wr32(device, 0x409500, 0x00000000); 947 nvkm_wr32(device, 0x409504, 0x00000016); 948 nvkm_msec(device, 2000, 949 if ((*psize = nvkm_rd32(device, 0x409800))) 950 return 0; 951 ); 952 953 return -ETIMEDOUT; 954 } 955 956 static int 957 gf100_gr_fecs_discover_image_size(struct gf100_gr *gr, u32 *psize) 958 { 959 struct nvkm_device *device = gr->base.engine.subdev.device; 960 961 nvkm_wr32(device, 0x409800, 0x00000000); 962 nvkm_wr32(device, 0x409500, 0x00000000); 963 nvkm_wr32(device, 0x409504, 0x00000010); 964 nvkm_msec(device, 2000, 965 if ((*psize = nvkm_rd32(device, 0x409800))) 966 return 0; 967 ); 968 969 return -ETIMEDOUT; 970 } 971 972 static void 973 gf100_gr_fecs_set_watchdog_timeout(struct gf100_gr *gr, u32 timeout) 974 { 975 struct nvkm_device *device = gr->base.engine.subdev.device; 976 977 nvkm_wr32(device, 0x409800, 0x00000000); 978 nvkm_wr32(device, 0x409500, timeout); 979 nvkm_wr32(device, 0x409504, 0x00000021); 980 } 981 982 static bool 983 gf100_gr_chsw_load(struct nvkm_gr *base) 984 { 985 struct gf100_gr *gr = gf100_gr(base); 986 if (!gr->firmware) { 987 u32 trace = nvkm_rd32(gr->base.engine.subdev.device, 0x40981c); 988 if (trace & 0x00000040) 989 return true; 990 } else { 991 u32 mthd = nvkm_rd32(gr->base.engine.subdev.device, 0x409808); 992 if (mthd & 0x00080000) 993 return true; 994 } 995 return false; 996 } 997 998 int 999 gf100_gr_rops(struct gf100_gr *gr) 1000 { 1001 struct nvkm_device *device = gr->base.engine.subdev.device; 1002 return (nvkm_rd32(device, 0x409604) & 0x001f0000) >> 16; 1003 } 1004 1005 void 1006 gf100_gr_zbc_init(struct gf100_gr *gr) 1007 { 1008 const u32 zero[] = { 0x00000000, 0x00000000, 0x00000000, 0x00000000, 1009 0x00000000, 0x00000000, 0x00000000, 0x00000000 }; 1010 const u32 one[] = { 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 1011 0xffffffff, 0xffffffff, 0xffffffff, 0xffffffff }; 1012 const u32 f32_0[] = { 0x00000000, 0x00000000, 0x00000000, 0x00000000, 1013 0x00000000, 0x00000000, 0x00000000, 0x00000000 }; 1014 const u32 f32_1[] = { 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000, 1015 0x3f800000, 0x3f800000, 0x3f800000, 0x3f800000 }; 1016 struct nvkm_ltc *ltc = gr->base.engine.subdev.device->ltc; 1017 int index, c = ltc->zbc_color_min, d = ltc->zbc_depth_min, s = ltc->zbc_depth_min; 1018 1019 if (!gr->zbc_color[0].format) { 1020 gf100_gr_zbc_color_get(gr, 1, & zero[0], &zero[4]); c++; 1021 gf100_gr_zbc_color_get(gr, 2, & one[0], &one[4]); c++; 1022 gf100_gr_zbc_color_get(gr, 4, &f32_0[0], &f32_0[4]); c++; 1023 gf100_gr_zbc_color_get(gr, 4, &f32_1[0], &f32_1[4]); c++; 1024 gf100_gr_zbc_depth_get(gr, 1, 0x00000000, 0x00000000); d++; 1025 gf100_gr_zbc_depth_get(gr, 1, 0x3f800000, 0x3f800000); d++; 1026 if (gr->func->zbc->stencil_get) { 1027 gr->func->zbc->stencil_get(gr, 1, 0x00, 0x00); s++; 1028 gr->func->zbc->stencil_get(gr, 1, 0x01, 0x01); s++; 1029 gr->func->zbc->stencil_get(gr, 1, 0xff, 0xff); s++; 1030 } 1031 } 1032 1033 for (index = c; index <= ltc->zbc_color_max; index++) 1034 gr->func->zbc->clear_color(gr, index); 1035 for (index = d; index <= ltc->zbc_depth_max; index++) 1036 gr->func->zbc->clear_depth(gr, index); 1037 1038 if (gr->func->zbc->clear_stencil) { 1039 for (index = s; index <= ltc->zbc_depth_max; index++) 1040 gr->func->zbc->clear_stencil(gr, index); 1041 } 1042 } 1043 1044 /* 1045 * Wait until GR goes idle. GR is considered idle if it is disabled by the 1046 * MC (0x200) register, or GR is not busy and a context switch is not in 1047 * progress. 1048 */ 1049 int 1050 gf100_gr_wait_idle(struct gf100_gr *gr) 1051 { 1052 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1053 struct nvkm_device *device = subdev->device; 1054 unsigned long end_jiffies = jiffies + msecs_to_jiffies(2000); 1055 bool gr_enabled, ctxsw_active, gr_busy; 1056 1057 do { 1058 /* 1059 * required to make sure FIFO_ENGINE_STATUS (0x2640) is 1060 * up-to-date 1061 */ 1062 nvkm_rd32(device, 0x400700); 1063 1064 gr_enabled = nvkm_rd32(device, 0x200) & 0x1000; 1065 ctxsw_active = nvkm_fifo_ctxsw_in_progress(&gr->base.engine); 1066 gr_busy = nvkm_rd32(device, 0x40060c) & 0x1; 1067 1068 if (!gr_enabled || (!gr_busy && !ctxsw_active)) 1069 return 0; 1070 } while (time_before(jiffies, end_jiffies)); 1071 1072 nvkm_error(subdev, 1073 "wait for idle timeout (en: %d, ctxsw: %d, busy: %d)\n", 1074 gr_enabled, ctxsw_active, gr_busy); 1075 return -EAGAIN; 1076 } 1077 1078 void 1079 gf100_gr_mmio(struct gf100_gr *gr, const struct gf100_gr_pack *p) 1080 { 1081 struct nvkm_device *device = gr->base.engine.subdev.device; 1082 const struct gf100_gr_pack *pack; 1083 const struct gf100_gr_init *init; 1084 1085 pack_for_each_init(init, pack, p) { 1086 u32 next = init->addr + init->count * init->pitch; 1087 u32 addr = init->addr; 1088 while (addr < next) { 1089 nvkm_wr32(device, addr, init->data); 1090 addr += init->pitch; 1091 } 1092 } 1093 } 1094 1095 void 1096 gf100_gr_icmd(struct gf100_gr *gr, const struct gf100_gr_pack *p) 1097 { 1098 struct nvkm_device *device = gr->base.engine.subdev.device; 1099 const struct gf100_gr_pack *pack; 1100 const struct gf100_gr_init *init; 1101 u64 data = 0; 1102 1103 nvkm_wr32(device, 0x400208, 0x80000000); 1104 1105 pack_for_each_init(init, pack, p) { 1106 u32 next = init->addr + init->count * init->pitch; 1107 u32 addr = init->addr; 1108 1109 if ((pack == p && init == p->init) || data != init->data) { 1110 nvkm_wr32(device, 0x400204, init->data); 1111 if (pack->type == 64) 1112 nvkm_wr32(device, 0x40020c, upper_32_bits(init->data)); 1113 data = init->data; 1114 } 1115 1116 while (addr < next) { 1117 nvkm_wr32(device, 0x400200, addr); 1118 /** 1119 * Wait for GR to go idle after submitting a 1120 * GO_IDLE bundle 1121 */ 1122 if ((addr & 0xffff) == 0xe100) 1123 gf100_gr_wait_idle(gr); 1124 nvkm_msec(device, 2000, 1125 if (!(nvkm_rd32(device, 0x400700) & 0x00000004)) 1126 break; 1127 ); 1128 addr += init->pitch; 1129 } 1130 } 1131 1132 nvkm_wr32(device, 0x400208, 0x00000000); 1133 } 1134 1135 void 1136 gf100_gr_mthd(struct gf100_gr *gr, const struct gf100_gr_pack *p) 1137 { 1138 struct nvkm_device *device = gr->base.engine.subdev.device; 1139 const struct gf100_gr_pack *pack; 1140 const struct gf100_gr_init *init; 1141 u32 data = 0; 1142 1143 pack_for_each_init(init, pack, p) { 1144 u32 ctrl = 0x80000000 | pack->type; 1145 u32 next = init->addr + init->count * init->pitch; 1146 u32 addr = init->addr; 1147 1148 if ((pack == p && init == p->init) || data != init->data) { 1149 nvkm_wr32(device, 0x40448c, init->data); 1150 data = init->data; 1151 } 1152 1153 while (addr < next) { 1154 nvkm_wr32(device, 0x404488, ctrl | (addr << 14)); 1155 addr += init->pitch; 1156 } 1157 } 1158 } 1159 1160 u64 1161 gf100_gr_units(struct nvkm_gr *base) 1162 { 1163 struct gf100_gr *gr = gf100_gr(base); 1164 u64 cfg; 1165 1166 cfg = (u32)gr->gpc_nr; 1167 cfg |= (u32)gr->tpc_total << 8; 1168 cfg |= (u64)gr->rop_nr << 32; 1169 1170 return cfg; 1171 } 1172 1173 static const struct nvkm_bitfield gf100_dispatch_error[] = { 1174 { 0x00000001, "INJECTED_BUNDLE_ERROR" }, 1175 { 0x00000002, "CLASS_SUBCH_MISMATCH" }, 1176 { 0x00000004, "SUBCHSW_DURING_NOTIFY" }, 1177 {} 1178 }; 1179 1180 static const struct nvkm_bitfield gf100_m2mf_error[] = { 1181 { 0x00000001, "PUSH_TOO_MUCH_DATA" }, 1182 { 0x00000002, "PUSH_NOT_ENOUGH_DATA" }, 1183 {} 1184 }; 1185 1186 static const struct nvkm_bitfield gf100_unk6_error[] = { 1187 { 0x00000001, "TEMP_TOO_SMALL" }, 1188 {} 1189 }; 1190 1191 static const struct nvkm_bitfield gf100_ccache_error[] = { 1192 { 0x00000001, "INTR" }, 1193 { 0x00000002, "LDCONST_OOB" }, 1194 {} 1195 }; 1196 1197 static const struct nvkm_bitfield gf100_macro_error[] = { 1198 { 0x00000001, "TOO_FEW_PARAMS" }, 1199 { 0x00000002, "TOO_MANY_PARAMS" }, 1200 { 0x00000004, "ILLEGAL_OPCODE" }, 1201 { 0x00000008, "DOUBLE_BRANCH" }, 1202 { 0x00000010, "WATCHDOG" }, 1203 {} 1204 }; 1205 1206 static const struct nvkm_bitfield gk104_sked_error[] = { 1207 { 0x00000040, "CTA_RESUME" }, 1208 { 0x00000080, "CONSTANT_BUFFER_SIZE" }, 1209 { 0x00000200, "LOCAL_MEMORY_SIZE_POS" }, 1210 { 0x00000400, "LOCAL_MEMORY_SIZE_NEG" }, 1211 { 0x00000800, "WARP_CSTACK_SIZE" }, 1212 { 0x00001000, "TOTAL_TEMP_SIZE" }, 1213 { 0x00002000, "REGISTER_COUNT" }, 1214 { 0x00040000, "TOTAL_THREADS" }, 1215 { 0x00100000, "PROGRAM_OFFSET" }, 1216 { 0x00200000, "SHARED_MEMORY_SIZE" }, 1217 { 0x00800000, "CTA_THREAD_DIMENSION_ZERO" }, 1218 { 0x01000000, "MEMORY_WINDOW_OVERLAP" }, 1219 { 0x02000000, "SHARED_CONFIG_TOO_SMALL" }, 1220 { 0x04000000, "TOTAL_REGISTER_COUNT" }, 1221 {} 1222 }; 1223 1224 static const struct nvkm_bitfield gf100_gpc_rop_error[] = { 1225 { 0x00000002, "RT_PITCH_OVERRUN" }, 1226 { 0x00000010, "RT_WIDTH_OVERRUN" }, 1227 { 0x00000020, "RT_HEIGHT_OVERRUN" }, 1228 { 0x00000080, "ZETA_STORAGE_TYPE_MISMATCH" }, 1229 { 0x00000100, "RT_STORAGE_TYPE_MISMATCH" }, 1230 { 0x00000400, "RT_LINEAR_MISMATCH" }, 1231 {} 1232 }; 1233 1234 static void 1235 gf100_gr_trap_gpc_rop(struct gf100_gr *gr, int gpc) 1236 { 1237 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1238 struct nvkm_device *device = subdev->device; 1239 char error[128]; 1240 u32 trap[4]; 1241 1242 trap[0] = nvkm_rd32(device, GPC_UNIT(gpc, 0x0420)) & 0x3fffffff; 1243 trap[1] = nvkm_rd32(device, GPC_UNIT(gpc, 0x0434)); 1244 trap[2] = nvkm_rd32(device, GPC_UNIT(gpc, 0x0438)); 1245 trap[3] = nvkm_rd32(device, GPC_UNIT(gpc, 0x043c)); 1246 1247 nvkm_snprintbf(error, sizeof(error), gf100_gpc_rop_error, trap[0]); 1248 1249 nvkm_error(subdev, "GPC%d/PROP trap: %08x [%s] x = %u, y = %u, " 1250 "format = %x, storage type = %x\n", 1251 gpc, trap[0], error, trap[1] & 0xffff, trap[1] >> 16, 1252 (trap[2] >> 8) & 0x3f, trap[3] & 0xff); 1253 nvkm_wr32(device, GPC_UNIT(gpc, 0x0420), 0xc0000000); 1254 } 1255 1256 const struct nvkm_enum gf100_mp_warp_error[] = { 1257 { 0x01, "STACK_ERROR" }, 1258 { 0x02, "API_STACK_ERROR" }, 1259 { 0x03, "RET_EMPTY_STACK_ERROR" }, 1260 { 0x04, "PC_WRAP" }, 1261 { 0x05, "MISALIGNED_PC" }, 1262 { 0x06, "PC_OVERFLOW" }, 1263 { 0x07, "MISALIGNED_IMMC_ADDR" }, 1264 { 0x08, "MISALIGNED_REG" }, 1265 { 0x09, "ILLEGAL_INSTR_ENCODING" }, 1266 { 0x0a, "ILLEGAL_SPH_INSTR_COMBO" }, 1267 { 0x0b, "ILLEGAL_INSTR_PARAM" }, 1268 { 0x0c, "INVALID_CONST_ADDR" }, 1269 { 0x0d, "OOR_REG" }, 1270 { 0x0e, "OOR_ADDR" }, 1271 { 0x0f, "MISALIGNED_ADDR" }, 1272 { 0x10, "INVALID_ADDR_SPACE" }, 1273 { 0x11, "ILLEGAL_INSTR_PARAM2" }, 1274 { 0x12, "INVALID_CONST_ADDR_LDC" }, 1275 { 0x13, "GEOMETRY_SM_ERROR" }, 1276 { 0x14, "DIVERGENT" }, 1277 { 0x15, "WARP_EXIT" }, 1278 {} 1279 }; 1280 1281 const struct nvkm_bitfield gf100_mp_global_error[] = { 1282 { 0x00000001, "SM_TO_SM_FAULT" }, 1283 { 0x00000002, "L1_ERROR" }, 1284 { 0x00000004, "MULTIPLE_WARP_ERRORS" }, 1285 { 0x00000008, "PHYSICAL_STACK_OVERFLOW" }, 1286 { 0x00000010, "BPT_INT" }, 1287 { 0x00000020, "BPT_PAUSE" }, 1288 { 0x00000040, "SINGLE_STEP_COMPLETE" }, 1289 { 0x20000000, "ECC_SEC_ERROR" }, 1290 { 0x40000000, "ECC_DED_ERROR" }, 1291 { 0x80000000, "TIMEOUT" }, 1292 {} 1293 }; 1294 1295 void 1296 gf100_gr_trap_mp(struct gf100_gr *gr, int gpc, int tpc) 1297 { 1298 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1299 struct nvkm_device *device = subdev->device; 1300 u32 werr = nvkm_rd32(device, TPC_UNIT(gpc, tpc, 0x648)); 1301 u32 gerr = nvkm_rd32(device, TPC_UNIT(gpc, tpc, 0x650)); 1302 const struct nvkm_enum *warp; 1303 char glob[128]; 1304 1305 nvkm_snprintbf(glob, sizeof(glob), gf100_mp_global_error, gerr); 1306 warp = nvkm_enum_find(gf100_mp_warp_error, werr & 0xffff); 1307 1308 nvkm_error(subdev, "GPC%i/TPC%i/MP trap: " 1309 "global %08x [%s] warp %04x [%s]\n", 1310 gpc, tpc, gerr, glob, werr, warp ? warp->name : ""); 1311 1312 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x648), 0x00000000); 1313 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x650), gerr); 1314 } 1315 1316 static void 1317 gf100_gr_trap_tpc(struct gf100_gr *gr, int gpc, int tpc) 1318 { 1319 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1320 struct nvkm_device *device = subdev->device; 1321 u32 stat = nvkm_rd32(device, TPC_UNIT(gpc, tpc, 0x0508)); 1322 1323 if (stat & 0x00000001) { 1324 u32 trap = nvkm_rd32(device, TPC_UNIT(gpc, tpc, 0x0224)); 1325 nvkm_error(subdev, "GPC%d/TPC%d/TEX: %08x\n", gpc, tpc, trap); 1326 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x0224), 0xc0000000); 1327 stat &= ~0x00000001; 1328 } 1329 1330 if (stat & 0x00000002) { 1331 gr->func->trap_mp(gr, gpc, tpc); 1332 stat &= ~0x00000002; 1333 } 1334 1335 if (stat & 0x00000004) { 1336 u32 trap = nvkm_rd32(device, TPC_UNIT(gpc, tpc, 0x0084)); 1337 nvkm_error(subdev, "GPC%d/TPC%d/POLY: %08x\n", gpc, tpc, trap); 1338 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x0084), 0xc0000000); 1339 stat &= ~0x00000004; 1340 } 1341 1342 if (stat & 0x00000008) { 1343 u32 trap = nvkm_rd32(device, TPC_UNIT(gpc, tpc, 0x048c)); 1344 nvkm_error(subdev, "GPC%d/TPC%d/L1C: %08x\n", gpc, tpc, trap); 1345 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x048c), 0xc0000000); 1346 stat &= ~0x00000008; 1347 } 1348 1349 if (stat & 0x00000010) { 1350 u32 trap = nvkm_rd32(device, TPC_UNIT(gpc, tpc, 0x0430)); 1351 nvkm_error(subdev, "GPC%d/TPC%d/MPC: %08x\n", gpc, tpc, trap); 1352 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x0430), 0xc0000000); 1353 stat &= ~0x00000010; 1354 } 1355 1356 if (stat) { 1357 nvkm_error(subdev, "GPC%d/TPC%d/%08x: unknown\n", gpc, tpc, stat); 1358 } 1359 } 1360 1361 static void 1362 gf100_gr_trap_gpc(struct gf100_gr *gr, int gpc) 1363 { 1364 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1365 struct nvkm_device *device = subdev->device; 1366 u32 stat = nvkm_rd32(device, GPC_UNIT(gpc, 0x2c90)); 1367 int tpc; 1368 1369 if (stat & 0x00000001) { 1370 gf100_gr_trap_gpc_rop(gr, gpc); 1371 stat &= ~0x00000001; 1372 } 1373 1374 if (stat & 0x00000002) { 1375 u32 trap = nvkm_rd32(device, GPC_UNIT(gpc, 0x0900)); 1376 nvkm_error(subdev, "GPC%d/ZCULL: %08x\n", gpc, trap); 1377 nvkm_wr32(device, GPC_UNIT(gpc, 0x0900), 0xc0000000); 1378 stat &= ~0x00000002; 1379 } 1380 1381 if (stat & 0x00000004) { 1382 u32 trap = nvkm_rd32(device, GPC_UNIT(gpc, 0x1028)); 1383 nvkm_error(subdev, "GPC%d/CCACHE: %08x\n", gpc, trap); 1384 nvkm_wr32(device, GPC_UNIT(gpc, 0x1028), 0xc0000000); 1385 stat &= ~0x00000004; 1386 } 1387 1388 if (stat & 0x00000008) { 1389 u32 trap = nvkm_rd32(device, GPC_UNIT(gpc, 0x0824)); 1390 nvkm_error(subdev, "GPC%d/ESETUP: %08x\n", gpc, trap); 1391 nvkm_wr32(device, GPC_UNIT(gpc, 0x0824), 0xc0000000); 1392 stat &= ~0x00000009; 1393 } 1394 1395 for (tpc = 0; tpc < gr->tpc_nr[gpc]; tpc++) { 1396 u32 mask = 0x00010000 << tpc; 1397 if (stat & mask) { 1398 gf100_gr_trap_tpc(gr, gpc, tpc); 1399 nvkm_wr32(device, GPC_UNIT(gpc, 0x2c90), mask); 1400 stat &= ~mask; 1401 } 1402 } 1403 1404 if (stat) { 1405 nvkm_error(subdev, "GPC%d/%08x: unknown\n", gpc, stat); 1406 } 1407 } 1408 1409 static void 1410 gf100_gr_trap_intr(struct gf100_gr *gr) 1411 { 1412 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1413 struct nvkm_device *device = subdev->device; 1414 char error[128]; 1415 u32 trap = nvkm_rd32(device, 0x400108); 1416 int rop, gpc; 1417 1418 if (trap & 0x00000001) { 1419 u32 stat = nvkm_rd32(device, 0x404000); 1420 1421 nvkm_snprintbf(error, sizeof(error), gf100_dispatch_error, 1422 stat & 0x3fffffff); 1423 nvkm_error(subdev, "DISPATCH %08x [%s]\n", stat, error); 1424 nvkm_wr32(device, 0x404000, 0xc0000000); 1425 nvkm_wr32(device, 0x400108, 0x00000001); 1426 trap &= ~0x00000001; 1427 } 1428 1429 if (trap & 0x00000002) { 1430 u32 stat = nvkm_rd32(device, 0x404600); 1431 1432 nvkm_snprintbf(error, sizeof(error), gf100_m2mf_error, 1433 stat & 0x3fffffff); 1434 nvkm_error(subdev, "M2MF %08x [%s]\n", stat, error); 1435 1436 nvkm_wr32(device, 0x404600, 0xc0000000); 1437 nvkm_wr32(device, 0x400108, 0x00000002); 1438 trap &= ~0x00000002; 1439 } 1440 1441 if (trap & 0x00000008) { 1442 u32 stat = nvkm_rd32(device, 0x408030); 1443 1444 nvkm_snprintbf(error, sizeof(error), gf100_ccache_error, 1445 stat & 0x3fffffff); 1446 nvkm_error(subdev, "CCACHE %08x [%s]\n", stat, error); 1447 nvkm_wr32(device, 0x408030, 0xc0000000); 1448 nvkm_wr32(device, 0x400108, 0x00000008); 1449 trap &= ~0x00000008; 1450 } 1451 1452 if (trap & 0x00000010) { 1453 u32 stat = nvkm_rd32(device, 0x405840); 1454 nvkm_error(subdev, "SHADER %08x, sph: 0x%06x, stage: 0x%02x\n", 1455 stat, stat & 0xffffff, (stat >> 24) & 0x3f); 1456 nvkm_wr32(device, 0x405840, 0xc0000000); 1457 nvkm_wr32(device, 0x400108, 0x00000010); 1458 trap &= ~0x00000010; 1459 } 1460 1461 if (trap & 0x00000040) { 1462 u32 stat = nvkm_rd32(device, 0x40601c); 1463 1464 nvkm_snprintbf(error, sizeof(error), gf100_unk6_error, 1465 stat & 0x3fffffff); 1466 nvkm_error(subdev, "UNK6 %08x [%s]\n", stat, error); 1467 1468 nvkm_wr32(device, 0x40601c, 0xc0000000); 1469 nvkm_wr32(device, 0x400108, 0x00000040); 1470 trap &= ~0x00000040; 1471 } 1472 1473 if (trap & 0x00000080) { 1474 u32 stat = nvkm_rd32(device, 0x404490); 1475 u32 pc = nvkm_rd32(device, 0x404494); 1476 u32 op = nvkm_rd32(device, 0x40449c); 1477 1478 nvkm_snprintbf(error, sizeof(error), gf100_macro_error, 1479 stat & 0x1fffffff); 1480 nvkm_error(subdev, "MACRO %08x [%s], pc: 0x%03x%s, op: 0x%08x\n", 1481 stat, error, pc & 0x7ff, 1482 (pc & 0x10000000) ? "" : " (invalid)", 1483 op); 1484 1485 nvkm_wr32(device, 0x404490, 0xc0000000); 1486 nvkm_wr32(device, 0x400108, 0x00000080); 1487 trap &= ~0x00000080; 1488 } 1489 1490 if (trap & 0x00000100) { 1491 u32 stat = nvkm_rd32(device, 0x407020) & 0x3fffffff; 1492 1493 nvkm_snprintbf(error, sizeof(error), gk104_sked_error, stat); 1494 nvkm_error(subdev, "SKED: %08x [%s]\n", stat, error); 1495 1496 if (stat) 1497 nvkm_wr32(device, 0x407020, 0x40000000); 1498 nvkm_wr32(device, 0x400108, 0x00000100); 1499 trap &= ~0x00000100; 1500 } 1501 1502 if (trap & 0x01000000) { 1503 u32 stat = nvkm_rd32(device, 0x400118); 1504 for (gpc = 0; stat && gpc < gr->gpc_nr; gpc++) { 1505 u32 mask = 0x00000001 << gpc; 1506 if (stat & mask) { 1507 gf100_gr_trap_gpc(gr, gpc); 1508 nvkm_wr32(device, 0x400118, mask); 1509 stat &= ~mask; 1510 } 1511 } 1512 nvkm_wr32(device, 0x400108, 0x01000000); 1513 trap &= ~0x01000000; 1514 } 1515 1516 if (trap & 0x02000000) { 1517 for (rop = 0; rop < gr->rop_nr; rop++) { 1518 u32 statz = nvkm_rd32(device, ROP_UNIT(rop, 0x070)); 1519 u32 statc = nvkm_rd32(device, ROP_UNIT(rop, 0x144)); 1520 nvkm_error(subdev, "ROP%d %08x %08x\n", 1521 rop, statz, statc); 1522 nvkm_wr32(device, ROP_UNIT(rop, 0x070), 0xc0000000); 1523 nvkm_wr32(device, ROP_UNIT(rop, 0x144), 0xc0000000); 1524 } 1525 nvkm_wr32(device, 0x400108, 0x02000000); 1526 trap &= ~0x02000000; 1527 } 1528 1529 if (trap) { 1530 nvkm_error(subdev, "TRAP UNHANDLED %08x\n", trap); 1531 nvkm_wr32(device, 0x400108, trap); 1532 } 1533 } 1534 1535 static void 1536 gf100_gr_ctxctl_debug_unit(struct gf100_gr *gr, u32 base) 1537 { 1538 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1539 struct nvkm_device *device = subdev->device; 1540 nvkm_error(subdev, "%06x - done %08x\n", base, 1541 nvkm_rd32(device, base + 0x400)); 1542 nvkm_error(subdev, "%06x - stat %08x %08x %08x %08x\n", base, 1543 nvkm_rd32(device, base + 0x800), 1544 nvkm_rd32(device, base + 0x804), 1545 nvkm_rd32(device, base + 0x808), 1546 nvkm_rd32(device, base + 0x80c)); 1547 nvkm_error(subdev, "%06x - stat %08x %08x %08x %08x\n", base, 1548 nvkm_rd32(device, base + 0x810), 1549 nvkm_rd32(device, base + 0x814), 1550 nvkm_rd32(device, base + 0x818), 1551 nvkm_rd32(device, base + 0x81c)); 1552 } 1553 1554 void 1555 gf100_gr_ctxctl_debug(struct gf100_gr *gr) 1556 { 1557 struct nvkm_device *device = gr->base.engine.subdev.device; 1558 u32 gpcnr = nvkm_rd32(device, 0x409604) & 0xffff; 1559 u32 gpc; 1560 1561 gf100_gr_ctxctl_debug_unit(gr, 0x409000); 1562 for (gpc = 0; gpc < gpcnr; gpc++) 1563 gf100_gr_ctxctl_debug_unit(gr, 0x502000 + (gpc * 0x8000)); 1564 } 1565 1566 static void 1567 gf100_gr_ctxctl_isr(struct gf100_gr *gr) 1568 { 1569 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1570 struct nvkm_device *device = subdev->device; 1571 u32 stat = nvkm_rd32(device, 0x409c18); 1572 1573 if (!gr->firmware && (stat & 0x00000001)) { 1574 u32 code = nvkm_rd32(device, 0x409814); 1575 if (code == E_BAD_FWMTHD) { 1576 u32 class = nvkm_rd32(device, 0x409808); 1577 u32 addr = nvkm_rd32(device, 0x40980c); 1578 u32 subc = (addr & 0x00070000) >> 16; 1579 u32 mthd = (addr & 0x00003ffc); 1580 u32 data = nvkm_rd32(device, 0x409810); 1581 1582 nvkm_error(subdev, "FECS MTHD subc %d class %04x " 1583 "mthd %04x data %08x\n", 1584 subc, class, mthd, data); 1585 } else { 1586 nvkm_error(subdev, "FECS ucode error %d\n", code); 1587 } 1588 nvkm_wr32(device, 0x409c20, 0x00000001); 1589 stat &= ~0x00000001; 1590 } 1591 1592 if (!gr->firmware && (stat & 0x00080000)) { 1593 nvkm_error(subdev, "FECS watchdog timeout\n"); 1594 gf100_gr_ctxctl_debug(gr); 1595 nvkm_wr32(device, 0x409c20, 0x00080000); 1596 stat &= ~0x00080000; 1597 } 1598 1599 if (stat) { 1600 nvkm_error(subdev, "FECS %08x\n", stat); 1601 gf100_gr_ctxctl_debug(gr); 1602 nvkm_wr32(device, 0x409c20, stat); 1603 } 1604 } 1605 1606 static irqreturn_t 1607 gf100_gr_intr(struct nvkm_inth *inth) 1608 { 1609 struct gf100_gr *gr = container_of(inth, typeof(*gr), base.engine.subdev.inth); 1610 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1611 struct nvkm_device *device = subdev->device; 1612 struct nvkm_chan *chan; 1613 unsigned long flags; 1614 u64 inst = nvkm_rd32(device, 0x409b00) & 0x0fffffff; 1615 u32 stat = nvkm_rd32(device, 0x400100); 1616 u32 addr = nvkm_rd32(device, 0x400704); 1617 u32 mthd = (addr & 0x00003ffc); 1618 u32 subc = (addr & 0x00070000) >> 16; 1619 u32 data = nvkm_rd32(device, 0x400708); 1620 u32 code = nvkm_rd32(device, 0x400110); 1621 u32 class; 1622 const char *name = "unknown"; 1623 int chid = -1; 1624 1625 chan = nvkm_chan_get_inst(&gr->base.engine, (u64)inst << 12, &flags); 1626 if (chan) { 1627 name = chan->name; 1628 chid = chan->id; 1629 } 1630 1631 if (device->card_type < NV_E0 || subc < 4) 1632 class = nvkm_rd32(device, 0x404200 + (subc * 4)); 1633 else 1634 class = 0x0000; 1635 1636 if (stat & 0x00000001) { 1637 /* 1638 * notifier interrupt, only needed for cyclestats 1639 * can be safely ignored 1640 */ 1641 nvkm_wr32(device, 0x400100, 0x00000001); 1642 stat &= ~0x00000001; 1643 } 1644 1645 if (stat & 0x00000010) { 1646 if (!gf100_gr_mthd_sw(device, class, mthd, data)) { 1647 nvkm_error(subdev, "ILLEGAL_MTHD ch %d [%010llx %s] " 1648 "subc %d class %04x mthd %04x data %08x\n", 1649 chid, inst << 12, name, subc, 1650 class, mthd, data); 1651 } 1652 nvkm_wr32(device, 0x400100, 0x00000010); 1653 stat &= ~0x00000010; 1654 } 1655 1656 if (stat & 0x00000020) { 1657 nvkm_error(subdev, "ILLEGAL_CLASS ch %d [%010llx %s] " 1658 "subc %d class %04x mthd %04x data %08x\n", 1659 chid, inst << 12, name, subc, class, mthd, data); 1660 nvkm_wr32(device, 0x400100, 0x00000020); 1661 stat &= ~0x00000020; 1662 } 1663 1664 if (stat & 0x00100000) { 1665 const struct nvkm_enum *en = 1666 nvkm_enum_find(nv50_data_error_names, code); 1667 nvkm_error(subdev, "DATA_ERROR %08x [%s] ch %d [%010llx %s] " 1668 "subc %d class %04x mthd %04x data %08x\n", 1669 code, en ? en->name : "", chid, inst << 12, 1670 name, subc, class, mthd, data); 1671 nvkm_wr32(device, 0x400100, 0x00100000); 1672 stat &= ~0x00100000; 1673 } 1674 1675 if (stat & 0x00200000) { 1676 nvkm_error(subdev, "TRAP ch %d [%010llx %s]\n", 1677 chid, inst << 12, name); 1678 gf100_gr_trap_intr(gr); 1679 nvkm_wr32(device, 0x400100, 0x00200000); 1680 stat &= ~0x00200000; 1681 } 1682 1683 if (stat & 0x00080000) { 1684 gf100_gr_ctxctl_isr(gr); 1685 nvkm_wr32(device, 0x400100, 0x00080000); 1686 stat &= ~0x00080000; 1687 } 1688 1689 if (stat) { 1690 nvkm_error(subdev, "intr %08x\n", stat); 1691 nvkm_wr32(device, 0x400100, stat); 1692 } 1693 1694 nvkm_wr32(device, 0x400500, 0x00010001); 1695 nvkm_chan_put(&chan, flags); 1696 return IRQ_HANDLED; 1697 } 1698 1699 static void 1700 gf100_gr_init_fw(struct nvkm_falcon *falcon, 1701 struct nvkm_blob *code, struct nvkm_blob *data) 1702 { 1703 nvkm_falcon_load_dmem(falcon, data->data, 0x0, data->size, 0); 1704 nvkm_falcon_load_imem(falcon, code->data, 0x0, code->size, 0, 0, false); 1705 } 1706 1707 static void 1708 gf100_gr_init_csdata(struct gf100_gr *gr, 1709 const struct gf100_gr_pack *pack, 1710 u32 falcon, u32 starstar, u32 base) 1711 { 1712 struct nvkm_device *device = gr->base.engine.subdev.device; 1713 const struct gf100_gr_pack *iter; 1714 const struct gf100_gr_init *init; 1715 u32 addr = ~0, prev = ~0, xfer = 0; 1716 u32 star, temp; 1717 1718 nvkm_wr32(device, falcon + 0x01c0, 0x02000000 + starstar); 1719 star = nvkm_rd32(device, falcon + 0x01c4); 1720 temp = nvkm_rd32(device, falcon + 0x01c4); 1721 if (temp > star) 1722 star = temp; 1723 nvkm_wr32(device, falcon + 0x01c0, 0x01000000 + star); 1724 1725 pack_for_each_init(init, iter, pack) { 1726 u32 head = init->addr - base; 1727 u32 tail = head + init->count * init->pitch; 1728 while (head < tail) { 1729 if (head != prev + 4 || xfer >= 32) { 1730 if (xfer) { 1731 u32 data = ((--xfer << 26) | addr); 1732 nvkm_wr32(device, falcon + 0x01c4, data); 1733 star += 4; 1734 } 1735 addr = head; 1736 xfer = 0; 1737 } 1738 prev = head; 1739 xfer = xfer + 1; 1740 head = head + init->pitch; 1741 } 1742 } 1743 1744 nvkm_wr32(device, falcon + 0x01c4, (--xfer << 26) | addr); 1745 nvkm_wr32(device, falcon + 0x01c0, 0x01000004 + starstar); 1746 nvkm_wr32(device, falcon + 0x01c4, star + 4); 1747 } 1748 1749 /* Initialize context from an external (secure or not) firmware */ 1750 static int 1751 gf100_gr_init_ctxctl_ext(struct gf100_gr *gr) 1752 { 1753 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1754 struct nvkm_device *device = subdev->device; 1755 u32 lsf_mask = 0; 1756 int ret; 1757 1758 /* load fuc microcode */ 1759 nvkm_mc_unk260(device, 0); 1760 1761 /* securely-managed falcons must be reset using secure boot */ 1762 1763 if (!nvkm_acr_managed_falcon(device, NVKM_ACR_LSF_FECS)) { 1764 gf100_gr_init_fw(&gr->fecs.falcon, &gr->fecs.inst, 1765 &gr->fecs.data); 1766 } else { 1767 lsf_mask |= BIT(NVKM_ACR_LSF_FECS); 1768 } 1769 1770 if (!nvkm_acr_managed_falcon(device, NVKM_ACR_LSF_GPCCS)) { 1771 gf100_gr_init_fw(&gr->gpccs.falcon, &gr->gpccs.inst, 1772 &gr->gpccs.data); 1773 } else { 1774 lsf_mask |= BIT(NVKM_ACR_LSF_GPCCS); 1775 } 1776 1777 if (lsf_mask) { 1778 ret = nvkm_acr_bootstrap_falcons(device, lsf_mask); 1779 if (ret) 1780 return ret; 1781 } 1782 1783 nvkm_mc_unk260(device, 1); 1784 1785 /* start both of them running */ 1786 nvkm_wr32(device, 0x409800, 0x00000000); 1787 nvkm_wr32(device, 0x41a10c, 0x00000000); 1788 nvkm_wr32(device, 0x40910c, 0x00000000); 1789 1790 nvkm_falcon_start(&gr->gpccs.falcon); 1791 nvkm_falcon_start(&gr->fecs.falcon); 1792 1793 if (nvkm_msec(device, 2000, 1794 if (nvkm_rd32(device, 0x409800) & 0x00000001) 1795 break; 1796 ) < 0) 1797 return -EBUSY; 1798 1799 gf100_gr_fecs_set_watchdog_timeout(gr, 0x7fffffff); 1800 1801 /* Determine how much memory is required to store main context image. */ 1802 ret = gf100_gr_fecs_discover_image_size(gr, &gr->size); 1803 if (ret) 1804 return ret; 1805 1806 /* Determine how much memory is required to store ZCULL image. */ 1807 ret = gf100_gr_fecs_discover_zcull_image_size(gr, &gr->size_zcull); 1808 if (ret) 1809 return ret; 1810 1811 /* Determine how much memory is required to store PerfMon image. */ 1812 ret = gf100_gr_fecs_discover_pm_image_size(gr, &gr->size_pm); 1813 if (ret) 1814 return ret; 1815 1816 /*XXX: We (likely) require PMU support to even bother with this. 1817 * 1818 * Also, it seems like not all GPUs support ELPG. Traces I 1819 * have here show RM enabling it on Kepler/Turing, but none 1820 * of the GPUs between those. NVGPU decides this by PCIID. 1821 */ 1822 if (0) { 1823 ret = gf100_gr_fecs_elpg_bind(gr); 1824 if (ret) 1825 return ret; 1826 } 1827 1828 return 0; 1829 } 1830 1831 static int 1832 gf100_gr_init_ctxctl_int(struct gf100_gr *gr) 1833 { 1834 const struct gf100_grctx_func *grctx = gr->func->grctx; 1835 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1836 struct nvkm_device *device = subdev->device; 1837 1838 if (!gr->func->fecs.ucode) { 1839 return -ENOSYS; 1840 } 1841 1842 /* load HUB microcode */ 1843 nvkm_mc_unk260(device, 0); 1844 nvkm_falcon_load_dmem(&gr->fecs.falcon, 1845 gr->func->fecs.ucode->data.data, 0x0, 1846 gr->func->fecs.ucode->data.size, 0); 1847 nvkm_falcon_load_imem(&gr->fecs.falcon, 1848 gr->func->fecs.ucode->code.data, 0x0, 1849 gr->func->fecs.ucode->code.size, 0, 0, false); 1850 1851 /* load GPC microcode */ 1852 nvkm_falcon_load_dmem(&gr->gpccs.falcon, 1853 gr->func->gpccs.ucode->data.data, 0x0, 1854 gr->func->gpccs.ucode->data.size, 0); 1855 nvkm_falcon_load_imem(&gr->gpccs.falcon, 1856 gr->func->gpccs.ucode->code.data, 0x0, 1857 gr->func->gpccs.ucode->code.size, 0, 0, false); 1858 nvkm_mc_unk260(device, 1); 1859 1860 /* load register lists */ 1861 gf100_gr_init_csdata(gr, grctx->hub, 0x409000, 0x000, 0x000000); 1862 gf100_gr_init_csdata(gr, grctx->gpc_0, 0x41a000, 0x000, 0x418000); 1863 gf100_gr_init_csdata(gr, grctx->gpc_1, 0x41a000, 0x000, 0x418000); 1864 gf100_gr_init_csdata(gr, grctx->tpc, 0x41a000, 0x004, 0x419800); 1865 gf100_gr_init_csdata(gr, grctx->ppc, 0x41a000, 0x008, 0x41be00); 1866 1867 /* start HUB ucode running, it'll init the GPCs */ 1868 nvkm_wr32(device, 0x40910c, 0x00000000); 1869 nvkm_wr32(device, 0x409100, 0x00000002); 1870 if (nvkm_msec(device, 2000, 1871 if (nvkm_rd32(device, 0x409800) & 0x80000000) 1872 break; 1873 ) < 0) { 1874 gf100_gr_ctxctl_debug(gr); 1875 return -EBUSY; 1876 } 1877 1878 gr->size = nvkm_rd32(device, 0x409804); 1879 return 0; 1880 } 1881 1882 int 1883 gf100_gr_init_ctxctl(struct gf100_gr *gr) 1884 { 1885 int ret; 1886 1887 if (gr->firmware) 1888 ret = gf100_gr_init_ctxctl_ext(gr); 1889 else 1890 ret = gf100_gr_init_ctxctl_int(gr); 1891 1892 return ret; 1893 } 1894 1895 int 1896 gf100_gr_oneinit_sm_id(struct gf100_gr *gr) 1897 { 1898 int tpc, gpc; 1899 1900 for (tpc = 0; tpc < gr->tpc_max; tpc++) { 1901 for (gpc = 0; gpc < gr->gpc_nr; gpc++) { 1902 if (tpc < gr->tpc_nr[gpc]) { 1903 gr->sm[gr->sm_nr].gpc = gpc; 1904 gr->sm[gr->sm_nr].tpc = tpc; 1905 gr->sm_nr++; 1906 } 1907 } 1908 } 1909 1910 return 0; 1911 } 1912 1913 void 1914 gf100_gr_oneinit_tiles(struct gf100_gr *gr) 1915 { 1916 static const u8 primes[] = { 1917 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61 1918 }; 1919 int init_frac[GPC_MAX], init_err[GPC_MAX], run_err[GPC_MAX], i, j; 1920 u32 mul_factor, comm_denom; 1921 u8 gpc_map[GPC_MAX]; 1922 bool sorted; 1923 1924 switch (gr->tpc_total) { 1925 case 15: gr->screen_tile_row_offset = 0x06; break; 1926 case 14: gr->screen_tile_row_offset = 0x05; break; 1927 case 13: gr->screen_tile_row_offset = 0x02; break; 1928 case 11: gr->screen_tile_row_offset = 0x07; break; 1929 case 10: gr->screen_tile_row_offset = 0x06; break; 1930 case 7: 1931 case 5: gr->screen_tile_row_offset = 0x01; break; 1932 case 3: gr->screen_tile_row_offset = 0x02; break; 1933 case 2: 1934 case 1: gr->screen_tile_row_offset = 0x01; break; 1935 default: gr->screen_tile_row_offset = 0x03; 1936 for (i = 0; i < ARRAY_SIZE(primes); i++) { 1937 if (gr->tpc_total % primes[i]) { 1938 gr->screen_tile_row_offset = primes[i]; 1939 break; 1940 } 1941 } 1942 break; 1943 } 1944 1945 /* Sort GPCs by TPC count, highest-to-lowest. */ 1946 for (i = 0; i < gr->gpc_nr; i++) 1947 gpc_map[i] = i; 1948 sorted = false; 1949 1950 while (!sorted) { 1951 for (sorted = true, i = 0; i < gr->gpc_nr - 1; i++) { 1952 if (gr->tpc_nr[gpc_map[i + 1]] > 1953 gr->tpc_nr[gpc_map[i + 0]]) { 1954 u8 swap = gpc_map[i]; 1955 gpc_map[i + 0] = gpc_map[i + 1]; 1956 gpc_map[i + 1] = swap; 1957 sorted = false; 1958 } 1959 } 1960 } 1961 1962 /* Determine tile->GPC mapping */ 1963 mul_factor = gr->gpc_nr * gr->tpc_max; 1964 if (mul_factor & 1) 1965 mul_factor = 2; 1966 else 1967 mul_factor = 1; 1968 1969 comm_denom = gr->gpc_nr * gr->tpc_max * mul_factor; 1970 1971 for (i = 0; i < gr->gpc_nr; i++) { 1972 init_frac[i] = gr->tpc_nr[gpc_map[i]] * gr->gpc_nr * mul_factor; 1973 init_err[i] = i * gr->tpc_max * mul_factor - comm_denom/2; 1974 run_err[i] = init_frac[i] + init_err[i]; 1975 } 1976 1977 for (i = 0; i < gr->tpc_total;) { 1978 for (j = 0; j < gr->gpc_nr; j++) { 1979 if ((run_err[j] * 2) >= comm_denom) { 1980 gr->tile[i++] = gpc_map[j]; 1981 run_err[j] += init_frac[j] - comm_denom; 1982 } else { 1983 run_err[j] += init_frac[j]; 1984 } 1985 } 1986 } 1987 } 1988 1989 static int 1990 gf100_gr_oneinit(struct nvkm_gr *base) 1991 { 1992 struct gf100_gr *gr = gf100_gr(base); 1993 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 1994 struct nvkm_device *device = subdev->device; 1995 struct nvkm_intr *intr = &device->mc->intr; 1996 enum nvkm_intr_type intr_type = NVKM_INTR_SUBDEV; 1997 int ret, i, j; 1998 1999 if (gr->func->oneinit_intr) 2000 intr = gr->func->oneinit_intr(gr, &intr_type); 2001 2002 ret = nvkm_inth_add(intr, intr_type, NVKM_INTR_PRIO_NORMAL, &gr->base.engine.subdev, 2003 gf100_gr_intr, &gr->base.engine.subdev.inth); 2004 if (ret) 2005 return ret; 2006 2007 nvkm_pmu_pgob(device->pmu, false); 2008 2009 gr->rop_nr = gr->func->rops(gr); 2010 gr->gpc_nr = nvkm_rd32(device, 0x409604) & 0x0000001f; 2011 for (i = 0; i < gr->gpc_nr; i++) { 2012 gr->tpc_nr[i] = nvkm_rd32(device, GPC_UNIT(i, 0x2608)); 2013 gr->tpc_max = max(gr->tpc_max, gr->tpc_nr[i]); 2014 gr->tpc_total += gr->tpc_nr[i]; 2015 for (j = 0; j < gr->func->ppc_nr; j++) { 2016 gr->ppc_tpc_mask[i][j] = 2017 nvkm_rd32(device, GPC_UNIT(i, 0x0c30 + (j * 4))); 2018 if (gr->ppc_tpc_mask[i][j] == 0) 2019 continue; 2020 2021 gr->ppc_nr[i]++; 2022 2023 gr->ppc_mask[i] |= (1 << j); 2024 gr->ppc_tpc_nr[i][j] = hweight8(gr->ppc_tpc_mask[i][j]); 2025 if (gr->ppc_tpc_min == 0 || 2026 gr->ppc_tpc_min > gr->ppc_tpc_nr[i][j]) 2027 gr->ppc_tpc_min = gr->ppc_tpc_nr[i][j]; 2028 if (gr->ppc_tpc_max < gr->ppc_tpc_nr[i][j]) 2029 gr->ppc_tpc_max = gr->ppc_tpc_nr[i][j]; 2030 } 2031 2032 gr->ppc_total += gr->ppc_nr[i]; 2033 } 2034 2035 /* Allocate global context buffers. */ 2036 ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST_SR_LOST, 2037 gr->func->grctx->pagepool_size, 0x100, false, &gr->pagepool); 2038 if (ret) 2039 return ret; 2040 2041 ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST_SR_LOST, gr->func->grctx->bundle_size, 2042 0x100, false, &gr->bundle_cb); 2043 if (ret) 2044 return ret; 2045 2046 ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST_SR_LOST, 2047 gr->func->grctx->attrib_cb_size(gr), 0x1000, false, &gr->attrib_cb); 2048 if (ret) 2049 return ret; 2050 2051 if (gr->func->grctx->unknown_size) { 2052 ret = nvkm_memory_new(device, NVKM_MEM_TARGET_INST, gr->func->grctx->unknown_size, 2053 0x100, false, &gr->unknown); 2054 if (ret) 2055 return ret; 2056 } 2057 2058 memset(gr->tile, 0xff, sizeof(gr->tile)); 2059 gr->func->oneinit_tiles(gr); 2060 2061 return gr->func->oneinit_sm_id(gr); 2062 } 2063 2064 static int 2065 gf100_gr_init_(struct nvkm_gr *base) 2066 { 2067 struct gf100_gr *gr = gf100_gr(base); 2068 struct nvkm_subdev *subdev = &base->engine.subdev; 2069 struct nvkm_device *device = subdev->device; 2070 bool reset = device->chipset == 0x137 || device->chipset == 0x138; 2071 int ret; 2072 2073 /* On certain GP107/GP108 boards, we trigger a weird issue where 2074 * GR will stop responding to PRI accesses after we've asked the 2075 * SEC2 RTOS to boot the GR falcons. This happens with far more 2076 * frequency when cold-booting a board (ie. returning from D3). 2077 * 2078 * The root cause for this is not known and has proven difficult 2079 * to isolate, with many avenues being dead-ends. 2080 * 2081 * A workaround was discovered by Karol, whereby putting GR into 2082 * reset for an extended period right before initialisation 2083 * prevents the problem from occuring. 2084 * 2085 * XXX: As RM does not require any such workaround, this is more 2086 * of a hack than a true fix. 2087 */ 2088 reset = nvkm_boolopt(device->cfgopt, "NvGrResetWar", reset); 2089 if (reset) { 2090 nvkm_mask(device, 0x000200, 0x00001000, 0x00000000); 2091 nvkm_rd32(device, 0x000200); 2092 msleep(50); 2093 nvkm_mask(device, 0x000200, 0x00001000, 0x00001000); 2094 nvkm_rd32(device, 0x000200); 2095 } 2096 2097 nvkm_pmu_pgob(gr->base.engine.subdev.device->pmu, false); 2098 2099 ret = nvkm_falcon_get(&gr->fecs.falcon, subdev); 2100 if (ret) 2101 return ret; 2102 2103 ret = nvkm_falcon_get(&gr->gpccs.falcon, subdev); 2104 if (ret) 2105 return ret; 2106 2107 ret = gr->func->init(gr); 2108 if (ret) 2109 return ret; 2110 2111 nvkm_inth_allow(&subdev->inth); 2112 return 0; 2113 } 2114 2115 static int 2116 gf100_gr_fini(struct nvkm_gr *base, bool suspend) 2117 { 2118 struct gf100_gr *gr = gf100_gr(base); 2119 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 2120 2121 nvkm_inth_block(&subdev->inth); 2122 2123 nvkm_falcon_put(&gr->gpccs.falcon, subdev); 2124 nvkm_falcon_put(&gr->fecs.falcon, subdev); 2125 return 0; 2126 } 2127 2128 static void * 2129 gf100_gr_dtor(struct nvkm_gr *base) 2130 { 2131 struct gf100_gr *gr = gf100_gr(base); 2132 2133 kfree(gr->data); 2134 2135 nvkm_memory_unref(&gr->unknown); 2136 nvkm_memory_unref(&gr->attrib_cb); 2137 nvkm_memory_unref(&gr->bundle_cb); 2138 nvkm_memory_unref(&gr->pagepool); 2139 2140 nvkm_falcon_dtor(&gr->gpccs.falcon); 2141 nvkm_falcon_dtor(&gr->fecs.falcon); 2142 2143 nvkm_blob_dtor(&gr->fecs.inst); 2144 nvkm_blob_dtor(&gr->fecs.data); 2145 nvkm_blob_dtor(&gr->gpccs.inst); 2146 nvkm_blob_dtor(&gr->gpccs.data); 2147 2148 vfree(gr->bundle64); 2149 vfree(gr->bundle_veid); 2150 vfree(gr->bundle); 2151 vfree(gr->method); 2152 vfree(gr->sw_ctx); 2153 vfree(gr->sw_nonctx); 2154 vfree(gr->sw_nonctx1); 2155 vfree(gr->sw_nonctx2); 2156 vfree(gr->sw_nonctx3); 2157 vfree(gr->sw_nonctx4); 2158 2159 return gr; 2160 } 2161 2162 static const struct nvkm_falcon_func 2163 gf100_gr_flcn = { 2164 .load_imem = nvkm_falcon_v1_load_imem, 2165 .load_dmem = nvkm_falcon_v1_load_dmem, 2166 .start = nvkm_falcon_v1_start, 2167 }; 2168 2169 void 2170 gf100_gr_init_num_tpc_per_gpc(struct gf100_gr *gr, bool pd, bool ds) 2171 { 2172 struct nvkm_device *device = gr->base.engine.subdev.device; 2173 int gpc, i, j; 2174 u32 data; 2175 2176 for (gpc = 0, i = 0; i < 4; i++) { 2177 for (data = 0, j = 0; j < 8 && gpc < gr->gpc_nr; j++, gpc++) 2178 data |= gr->tpc_nr[gpc] << (j * 4); 2179 if (pd) 2180 nvkm_wr32(device, 0x406028 + (i * 4), data); 2181 if (ds) 2182 nvkm_wr32(device, 0x405870 + (i * 4), data); 2183 } 2184 } 2185 2186 void 2187 gf100_gr_init_400054(struct gf100_gr *gr) 2188 { 2189 nvkm_wr32(gr->base.engine.subdev.device, 0x400054, 0x34ce3464); 2190 } 2191 2192 void 2193 gf100_gr_init_exception2(struct gf100_gr *gr) 2194 { 2195 struct nvkm_device *device = gr->base.engine.subdev.device; 2196 2197 nvkm_wr32(device, 0x40011c, 0xffffffff); 2198 nvkm_wr32(device, 0x400134, 0xffffffff); 2199 } 2200 2201 void 2202 gf100_gr_init_rop_exceptions(struct gf100_gr *gr) 2203 { 2204 struct nvkm_device *device = gr->base.engine.subdev.device; 2205 int rop; 2206 2207 for (rop = 0; rop < gr->rop_nr; rop++) { 2208 nvkm_wr32(device, ROP_UNIT(rop, 0x144), 0x40000000); 2209 nvkm_wr32(device, ROP_UNIT(rop, 0x070), 0x40000000); 2210 nvkm_wr32(device, ROP_UNIT(rop, 0x204), 0xffffffff); 2211 nvkm_wr32(device, ROP_UNIT(rop, 0x208), 0xffffffff); 2212 } 2213 } 2214 2215 void 2216 gf100_gr_init_shader_exceptions(struct gf100_gr *gr, int gpc, int tpc) 2217 { 2218 struct nvkm_device *device = gr->base.engine.subdev.device; 2219 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x644), 0x001ffffe); 2220 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x64c), 0x0000000f); 2221 } 2222 2223 void 2224 gf100_gr_init_tex_hww_esr(struct gf100_gr *gr, int gpc, int tpc) 2225 { 2226 struct nvkm_device *device = gr->base.engine.subdev.device; 2227 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x224), 0xc0000000); 2228 } 2229 2230 void 2231 gf100_gr_init_419eb4(struct gf100_gr *gr) 2232 { 2233 struct nvkm_device *device = gr->base.engine.subdev.device; 2234 nvkm_mask(device, 0x419eb4, 0x00001000, 0x00001000); 2235 } 2236 2237 void 2238 gf100_gr_init_419cc0(struct gf100_gr *gr) 2239 { 2240 struct nvkm_device *device = gr->base.engine.subdev.device; 2241 int gpc, tpc; 2242 2243 nvkm_mask(device, 0x419cc0, 0x00000008, 0x00000008); 2244 2245 for (gpc = 0; gpc < gr->gpc_nr; gpc++) { 2246 for (tpc = 0; tpc < gr->tpc_nr[gpc]; tpc++) 2247 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x48c), 0xc0000000); 2248 } 2249 } 2250 2251 void 2252 gf100_gr_init_40601c(struct gf100_gr *gr) 2253 { 2254 nvkm_wr32(gr->base.engine.subdev.device, 0x40601c, 0xc0000000); 2255 } 2256 2257 void 2258 gf100_gr_init_fecs_exceptions(struct gf100_gr *gr) 2259 { 2260 const u32 data = gr->firmware ? 0x000e0000 : 0x000e0001; 2261 nvkm_wr32(gr->base.engine.subdev.device, 0x409c24, data); 2262 } 2263 2264 void 2265 gf100_gr_init_gpc_mmu(struct gf100_gr *gr) 2266 { 2267 struct nvkm_device *device = gr->base.engine.subdev.device; 2268 struct nvkm_fb *fb = device->fb; 2269 2270 nvkm_wr32(device, 0x418880, nvkm_rd32(device, 0x100c80) & 0x00000001); 2271 nvkm_wr32(device, 0x4188a4, 0x03000000); 2272 nvkm_wr32(device, 0x418888, 0x00000000); 2273 nvkm_wr32(device, 0x41888c, 0x00000000); 2274 nvkm_wr32(device, 0x418890, 0x00000000); 2275 nvkm_wr32(device, 0x418894, 0x00000000); 2276 nvkm_wr32(device, 0x4188b4, nvkm_memory_addr(fb->mmu_wr) >> 8); 2277 nvkm_wr32(device, 0x4188b8, nvkm_memory_addr(fb->mmu_rd) >> 8); 2278 } 2279 2280 void 2281 gf100_gr_init_num_active_ltcs(struct gf100_gr *gr) 2282 { 2283 struct nvkm_device *device = gr->base.engine.subdev.device; 2284 nvkm_wr32(device, GPC_BCAST(0x08ac), nvkm_rd32(device, 0x100800)); 2285 } 2286 2287 void 2288 gf100_gr_init_zcull(struct gf100_gr *gr) 2289 { 2290 struct nvkm_device *device = gr->base.engine.subdev.device; 2291 const u32 magicgpc918 = DIV_ROUND_UP(0x00800000, gr->tpc_total); 2292 const u8 tile_nr = ALIGN(gr->tpc_total, 32); 2293 u8 bank[GPC_MAX] = {}, gpc, i, j; 2294 u32 data; 2295 2296 for (i = 0; i < tile_nr; i += 8) { 2297 for (data = 0, j = 0; j < 8 && i + j < gr->tpc_total; j++) { 2298 data |= bank[gr->tile[i + j]] << (j * 4); 2299 bank[gr->tile[i + j]]++; 2300 } 2301 nvkm_wr32(device, GPC_BCAST(0x0980 + ((i / 8) * 4)), data); 2302 } 2303 2304 for (gpc = 0; gpc < gr->gpc_nr; gpc++) { 2305 nvkm_wr32(device, GPC_UNIT(gpc, 0x0914), 2306 gr->screen_tile_row_offset << 8 | gr->tpc_nr[gpc]); 2307 nvkm_wr32(device, GPC_UNIT(gpc, 0x0910), 0x00040000 | 2308 gr->tpc_total); 2309 nvkm_wr32(device, GPC_UNIT(gpc, 0x0918), magicgpc918); 2310 } 2311 2312 nvkm_wr32(device, GPC_BCAST(0x1bd4), magicgpc918); 2313 } 2314 2315 void 2316 gf100_gr_init_vsc_stream_master(struct gf100_gr *gr) 2317 { 2318 struct nvkm_device *device = gr->base.engine.subdev.device; 2319 nvkm_mask(device, TPC_UNIT(0, 0, 0x05c), 0x00000001, 0x00000001); 2320 } 2321 2322 static int 2323 gf100_gr_reset(struct nvkm_gr *base) 2324 { 2325 struct nvkm_subdev *subdev = &base->engine.subdev; 2326 struct nvkm_device *device = subdev->device; 2327 struct gf100_gr *gr = gf100_gr(base); 2328 2329 nvkm_mask(device, 0x400500, 0x00000001, 0x00000000); 2330 2331 WARN_ON(gf100_gr_fecs_halt_pipeline(gr)); 2332 2333 subdev->func->fini(subdev, false); 2334 nvkm_mc_disable(device, subdev->type, subdev->inst); 2335 if (gr->func->gpccs.reset) 2336 gr->func->gpccs.reset(gr); 2337 2338 nvkm_mc_enable(device, subdev->type, subdev->inst); 2339 return subdev->func->init(subdev); 2340 } 2341 2342 int 2343 gf100_gr_init(struct gf100_gr *gr) 2344 { 2345 struct nvkm_device *device = gr->base.engine.subdev.device; 2346 int gpc, tpc; 2347 2348 nvkm_mask(device, 0x400500, 0x00010001, 0x00000000); 2349 2350 gr->func->init_gpc_mmu(gr); 2351 2352 if (gr->sw_nonctx1) { 2353 gf100_gr_mmio(gr, gr->sw_nonctx1); 2354 gf100_gr_mmio(gr, gr->sw_nonctx2); 2355 gf100_gr_mmio(gr, gr->sw_nonctx3); 2356 gf100_gr_mmio(gr, gr->sw_nonctx4); 2357 } else 2358 if (gr->sw_nonctx) { 2359 gf100_gr_mmio(gr, gr->sw_nonctx); 2360 } else { 2361 gf100_gr_mmio(gr, gr->func->mmio); 2362 } 2363 2364 gf100_gr_wait_idle(gr); 2365 2366 if (gr->func->init_r405a14) 2367 gr->func->init_r405a14(gr); 2368 2369 if (gr->func->clkgate_pack) 2370 nvkm_therm_clkgate_init(device->therm, gr->func->clkgate_pack); 2371 2372 if (gr->func->init_bios) 2373 gr->func->init_bios(gr); 2374 2375 gr->func->init_vsc_stream_master(gr); 2376 gr->func->init_zcull(gr); 2377 gr->func->init_num_active_ltcs(gr); 2378 if (gr->func->init_rop_active_fbps) 2379 gr->func->init_rop_active_fbps(gr); 2380 if (gr->func->init_bios_2) 2381 gr->func->init_bios_2(gr); 2382 if (gr->func->init_swdx_pes_mask) 2383 gr->func->init_swdx_pes_mask(gr); 2384 if (gr->func->init_fs) 2385 gr->func->init_fs(gr); 2386 2387 nvkm_wr32(device, 0x400500, 0x00010001); 2388 2389 nvkm_wr32(device, 0x400100, 0xffffffff); 2390 nvkm_wr32(device, 0x40013c, 0xffffffff); 2391 nvkm_wr32(device, 0x400124, 0x00000002); 2392 2393 gr->func->init_fecs_exceptions(gr); 2394 2395 if (gr->func->init_40a790) 2396 gr->func->init_40a790(gr); 2397 2398 if (gr->func->init_ds_hww_esr_2) 2399 gr->func->init_ds_hww_esr_2(gr); 2400 2401 nvkm_wr32(device, 0x404000, 0xc0000000); 2402 nvkm_wr32(device, 0x404600, 0xc0000000); 2403 nvkm_wr32(device, 0x408030, 0xc0000000); 2404 2405 if (gr->func->init_40601c) 2406 gr->func->init_40601c(gr); 2407 2408 nvkm_wr32(device, 0x406018, 0xc0000000); 2409 nvkm_wr32(device, 0x404490, 0xc0000000); 2410 2411 if (gr->func->init_sked_hww_esr) 2412 gr->func->init_sked_hww_esr(gr); 2413 2414 nvkm_wr32(device, 0x405840, 0xc0000000); 2415 nvkm_wr32(device, 0x405844, 0x00ffffff); 2416 2417 if (gr->func->init_419cc0) 2418 gr->func->init_419cc0(gr); 2419 if (gr->func->init_419eb4) 2420 gr->func->init_419eb4(gr); 2421 if (gr->func->init_419c9c) 2422 gr->func->init_419c9c(gr); 2423 2424 if (gr->func->init_ppc_exceptions) 2425 gr->func->init_ppc_exceptions(gr); 2426 2427 for (gpc = 0; gpc < gr->gpc_nr; gpc++) { 2428 nvkm_wr32(device, GPC_UNIT(gpc, 0x0420), 0xc0000000); 2429 nvkm_wr32(device, GPC_UNIT(gpc, 0x0900), 0xc0000000); 2430 nvkm_wr32(device, GPC_UNIT(gpc, 0x1028), 0xc0000000); 2431 nvkm_wr32(device, GPC_UNIT(gpc, 0x0824), 0xc0000000); 2432 for (tpc = 0; tpc < gr->tpc_nr[gpc]; tpc++) { 2433 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x508), 0xffffffff); 2434 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x50c), 0xffffffff); 2435 if (gr->func->init_tex_hww_esr) 2436 gr->func->init_tex_hww_esr(gr, gpc, tpc); 2437 nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x084), 0xc0000000); 2438 if (gr->func->init_504430) 2439 gr->func->init_504430(gr, gpc, tpc); 2440 gr->func->init_shader_exceptions(gr, gpc, tpc); 2441 } 2442 nvkm_wr32(device, GPC_UNIT(gpc, 0x2c90), 0xffffffff); 2443 nvkm_wr32(device, GPC_UNIT(gpc, 0x2c94), 0xffffffff); 2444 } 2445 2446 gr->func->init_rop_exceptions(gr); 2447 2448 nvkm_wr32(device, 0x400108, 0xffffffff); 2449 nvkm_wr32(device, 0x400138, 0xffffffff); 2450 nvkm_wr32(device, 0x400118, 0xffffffff); 2451 nvkm_wr32(device, 0x400130, 0xffffffff); 2452 if (gr->func->init_exception2) 2453 gr->func->init_exception2(gr); 2454 2455 if (gr->func->init_400054) 2456 gr->func->init_400054(gr); 2457 2458 gf100_gr_zbc_init(gr); 2459 2460 if (gr->func->init_4188a4) 2461 gr->func->init_4188a4(gr); 2462 2463 return gf100_gr_init_ctxctl(gr); 2464 } 2465 2466 void 2467 gf100_gr_fecs_reset(struct gf100_gr *gr) 2468 { 2469 struct nvkm_device *device = gr->base.engine.subdev.device; 2470 2471 nvkm_wr32(device, 0x409614, 0x00000070); 2472 nvkm_usec(device, 10, NVKM_DELAY); 2473 nvkm_mask(device, 0x409614, 0x00000700, 0x00000700); 2474 nvkm_usec(device, 10, NVKM_DELAY); 2475 nvkm_rd32(device, 0x409614); 2476 } 2477 2478 #include "fuc/hubgf100.fuc3.h" 2479 2480 struct gf100_gr_ucode 2481 gf100_gr_fecs_ucode = { 2482 .code.data = gf100_grhub_code, 2483 .code.size = sizeof(gf100_grhub_code), 2484 .data.data = gf100_grhub_data, 2485 .data.size = sizeof(gf100_grhub_data), 2486 }; 2487 2488 #include "fuc/gpcgf100.fuc3.h" 2489 2490 struct gf100_gr_ucode 2491 gf100_gr_gpccs_ucode = { 2492 .code.data = gf100_grgpc_code, 2493 .code.size = sizeof(gf100_grgpc_code), 2494 .data.data = gf100_grgpc_data, 2495 .data.size = sizeof(gf100_grgpc_data), 2496 }; 2497 2498 static int 2499 gf100_gr_nonstall(struct nvkm_gr *base) 2500 { 2501 struct gf100_gr *gr = gf100_gr(base); 2502 2503 if (gr->func->nonstall) 2504 return gr->func->nonstall(gr); 2505 2506 return -EINVAL; 2507 } 2508 2509 static const struct nvkm_gr_func 2510 gf100_gr_ = { 2511 .dtor = gf100_gr_dtor, 2512 .oneinit = gf100_gr_oneinit, 2513 .init = gf100_gr_init_, 2514 .fini = gf100_gr_fini, 2515 .nonstall = gf100_gr_nonstall, 2516 .reset = gf100_gr_reset, 2517 .units = gf100_gr_units, 2518 .chan_new = gf100_gr_chan_new, 2519 .object_get = gf100_gr_object_get, 2520 .chsw_load = gf100_gr_chsw_load, 2521 .ctxsw.pause = gf100_gr_fecs_stop_ctxsw, 2522 .ctxsw.resume = gf100_gr_fecs_start_ctxsw, 2523 .ctxsw.inst = gf100_gr_ctxsw_inst, 2524 }; 2525 2526 static const struct gf100_gr_func 2527 gf100_gr = { 2528 .oneinit_tiles = gf100_gr_oneinit_tiles, 2529 .oneinit_sm_id = gf100_gr_oneinit_sm_id, 2530 .init = gf100_gr_init, 2531 .init_gpc_mmu = gf100_gr_init_gpc_mmu, 2532 .init_vsc_stream_master = gf100_gr_init_vsc_stream_master, 2533 .init_zcull = gf100_gr_init_zcull, 2534 .init_num_active_ltcs = gf100_gr_init_num_active_ltcs, 2535 .init_fecs_exceptions = gf100_gr_init_fecs_exceptions, 2536 .init_40601c = gf100_gr_init_40601c, 2537 .init_419cc0 = gf100_gr_init_419cc0, 2538 .init_419eb4 = gf100_gr_init_419eb4, 2539 .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, 2540 .init_shader_exceptions = gf100_gr_init_shader_exceptions, 2541 .init_rop_exceptions = gf100_gr_init_rop_exceptions, 2542 .init_exception2 = gf100_gr_init_exception2, 2543 .init_400054 = gf100_gr_init_400054, 2544 .trap_mp = gf100_gr_trap_mp, 2545 .mmio = gf100_gr_pack_mmio, 2546 .fecs.ucode = &gf100_gr_fecs_ucode, 2547 .fecs.reset = gf100_gr_fecs_reset, 2548 .gpccs.ucode = &gf100_gr_gpccs_ucode, 2549 .rops = gf100_gr_rops, 2550 .grctx = &gf100_grctx, 2551 .zbc = &gf100_gr_zbc, 2552 .sclass = { 2553 { -1, -1, FERMI_TWOD_A }, 2554 { -1, -1, FERMI_MEMORY_TO_MEMORY_FORMAT_A }, 2555 { -1, -1, FERMI_A, &gf100_fermi }, 2556 { -1, -1, FERMI_COMPUTE_A }, 2557 {} 2558 } 2559 }; 2560 2561 int 2562 gf100_gr_nofw(struct gf100_gr *gr, int ver, const struct gf100_gr_fwif *fwif) 2563 { 2564 gr->firmware = false; 2565 return 0; 2566 } 2567 2568 static int 2569 gf100_gr_load_fw(struct gf100_gr *gr, const char *name, 2570 struct nvkm_blob *blob) 2571 { 2572 struct nvkm_subdev *subdev = &gr->base.engine.subdev; 2573 struct nvkm_device *device = subdev->device; 2574 const struct firmware *fw; 2575 char f[32]; 2576 int ret; 2577 2578 snprintf(f, sizeof(f), "nouveau/nv%02x_%s", device->chipset, name); 2579 ret = request_firmware(&fw, f, device->dev); 2580 if (ret) { 2581 snprintf(f, sizeof(f), "nouveau/%s", name); 2582 ret = request_firmware(&fw, f, device->dev); 2583 if (ret) { 2584 nvkm_error(subdev, "failed to load %s\n", name); 2585 return ret; 2586 } 2587 } 2588 2589 blob->size = fw->size; 2590 blob->data = kmemdup(fw->data, blob->size, GFP_KERNEL); 2591 release_firmware(fw); 2592 return (blob->data != NULL) ? 0 : -ENOMEM; 2593 } 2594 2595 int 2596 gf100_gr_load(struct gf100_gr *gr, int ver, const struct gf100_gr_fwif *fwif) 2597 { 2598 struct nvkm_device *device = gr->base.engine.subdev.device; 2599 2600 if (!nvkm_boolopt(device->cfgopt, "NvGrUseFW", false)) 2601 return -EINVAL; 2602 2603 if (gf100_gr_load_fw(gr, "fuc409c", &gr->fecs.inst) || 2604 gf100_gr_load_fw(gr, "fuc409d", &gr->fecs.data) || 2605 gf100_gr_load_fw(gr, "fuc41ac", &gr->gpccs.inst) || 2606 gf100_gr_load_fw(gr, "fuc41ad", &gr->gpccs.data)) 2607 return -ENOENT; 2608 2609 gr->firmware = true; 2610 return 0; 2611 } 2612 2613 static const struct gf100_gr_fwif 2614 gf100_gr_fwif[] = { 2615 { -1, gf100_gr_load, &gf100_gr }, 2616 { -1, gf100_gr_nofw, &gf100_gr }, 2617 {} 2618 }; 2619 2620 int 2621 gf100_gr_new_(const struct gf100_gr_fwif *fwif, struct nvkm_device *device, 2622 enum nvkm_subdev_type type, int inst, struct nvkm_gr **pgr) 2623 { 2624 struct gf100_gr *gr; 2625 int ret; 2626 2627 if (!(gr = kzalloc(sizeof(*gr), GFP_KERNEL))) 2628 return -ENOMEM; 2629 *pgr = &gr->base; 2630 2631 ret = nvkm_gr_ctor(&gf100_gr_, device, type, inst, true, &gr->base); 2632 if (ret) 2633 return ret; 2634 2635 fwif = nvkm_firmware_load(&gr->base.engine.subdev, fwif, "Gr", gr); 2636 if (IS_ERR(fwif)) 2637 return PTR_ERR(fwif); 2638 2639 gr->func = fwif->func; 2640 2641 ret = nvkm_falcon_ctor(&gf100_gr_flcn, &gr->base.engine.subdev, 2642 "fecs", 0x409000, &gr->fecs.falcon); 2643 if (ret) 2644 return ret; 2645 2646 mutex_init(&gr->fecs.mutex); 2647 2648 ret = nvkm_falcon_ctor(&gf100_gr_flcn, &gr->base.engine.subdev, 2649 "gpccs", 0x41a000, &gr->gpccs.falcon); 2650 if (ret) 2651 return ret; 2652 2653 return 0; 2654 } 2655 2656 int 2657 gf100_gr_new(struct nvkm_device *device, enum nvkm_subdev_type type, int inst, struct nvkm_gr **pgr) 2658 { 2659 return gf100_gr_new_(gf100_gr_fwif, device, type, inst, pgr); 2660 } 2661