1 /* 2 * Copyright 2015 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 <bskeggs@redhat.com> 23 */ 24 #include "gf100.h" 25 #include "ctxgf100.h" 26 27 #include <core/firmware.h> 28 #include <subdev/acr.h> 29 #include <subdev/secboot.h> 30 31 #include <nvfw/flcn.h> 32 33 #include <nvif/class.h> 34 35 int 36 gm200_gr_nofw(struct gf100_gr *gr, int ver, const struct gf100_gr_fwif *fwif) 37 { 38 nvkm_warn(&gr->base.engine.subdev, "firmware unavailable\n"); 39 return -ENODEV; 40 } 41 42 /******************************************************************************* 43 * PGRAPH engine/subdev functions 44 ******************************************************************************/ 45 46 static void 47 gm200_gr_acr_bld_patch(struct nvkm_acr *acr, u32 bld, s64 adjust) 48 { 49 struct flcn_bl_dmem_desc_v1 hdr; 50 nvkm_robj(acr->wpr, bld, &hdr, sizeof(hdr)); 51 hdr.code_dma_base = hdr.code_dma_base + adjust; 52 hdr.data_dma_base = hdr.data_dma_base + adjust; 53 nvkm_wobj(acr->wpr, bld, &hdr, sizeof(hdr)); 54 flcn_bl_dmem_desc_v1_dump(&acr->subdev, &hdr); 55 } 56 57 static void 58 gm200_gr_acr_bld_write(struct nvkm_acr *acr, u32 bld, 59 struct nvkm_acr_lsfw *lsfw) 60 { 61 const u64 base = lsfw->offset.img + lsfw->app_start_offset; 62 const u64 code = base + lsfw->app_resident_code_offset; 63 const u64 data = base + lsfw->app_resident_data_offset; 64 const struct flcn_bl_dmem_desc_v1 hdr = { 65 .ctx_dma = FALCON_DMAIDX_UCODE, 66 .code_dma_base = code, 67 .non_sec_code_off = lsfw->app_resident_code_offset, 68 .non_sec_code_size = lsfw->app_resident_code_size, 69 .code_entry_point = lsfw->app_imem_entry, 70 .data_dma_base = data, 71 .data_size = lsfw->app_resident_data_size, 72 }; 73 74 nvkm_wobj(acr->wpr, bld, &hdr, sizeof(hdr)); 75 } 76 77 const struct nvkm_acr_lsf_func 78 gm200_gr_gpccs_acr = { 79 .flags = NVKM_ACR_LSF_FORCE_PRIV_LOAD, 80 .bld_size = sizeof(struct flcn_bl_dmem_desc_v1), 81 .bld_write = gm200_gr_acr_bld_write, 82 .bld_patch = gm200_gr_acr_bld_patch, 83 }; 84 85 const struct nvkm_acr_lsf_func 86 gm200_gr_fecs_acr = { 87 .bld_size = sizeof(struct flcn_bl_dmem_desc_v1), 88 .bld_write = gm200_gr_acr_bld_write, 89 .bld_patch = gm200_gr_acr_bld_patch, 90 }; 91 92 int 93 gm200_gr_rops(struct gf100_gr *gr) 94 { 95 return nvkm_rd32(gr->base.engine.subdev.device, 0x12006c); 96 } 97 98 void 99 gm200_gr_init_ds_hww_esr_2(struct gf100_gr *gr) 100 { 101 struct nvkm_device *device = gr->base.engine.subdev.device; 102 nvkm_wr32(device, 0x405848, 0xc0000000); 103 nvkm_mask(device, 0x40584c, 0x00000001, 0x00000001); 104 } 105 106 void 107 gm200_gr_init_num_active_ltcs(struct gf100_gr *gr) 108 { 109 struct nvkm_device *device = gr->base.engine.subdev.device; 110 nvkm_wr32(device, GPC_BCAST(0x08ac), nvkm_rd32(device, 0x100800)); 111 nvkm_wr32(device, GPC_BCAST(0x033c), nvkm_rd32(device, 0x100804)); 112 } 113 114 void 115 gm200_gr_init_gpc_mmu(struct gf100_gr *gr) 116 { 117 struct nvkm_device *device = gr->base.engine.subdev.device; 118 119 nvkm_wr32(device, 0x418880, nvkm_rd32(device, 0x100c80) & 0xf0001fff); 120 nvkm_wr32(device, 0x418890, 0x00000000); 121 nvkm_wr32(device, 0x418894, 0x00000000); 122 123 nvkm_wr32(device, 0x4188b4, nvkm_rd32(device, 0x100cc8)); 124 nvkm_wr32(device, 0x4188b8, nvkm_rd32(device, 0x100ccc)); 125 nvkm_wr32(device, 0x4188b0, nvkm_rd32(device, 0x100cc4)); 126 } 127 128 static void 129 gm200_gr_init_rop_active_fbps(struct gf100_gr *gr) 130 { 131 struct nvkm_device *device = gr->base.engine.subdev.device; 132 const u32 fbp_count = nvkm_rd32(device, 0x12006c); 133 nvkm_mask(device, 0x408850, 0x0000000f, fbp_count); /* zrop */ 134 nvkm_mask(device, 0x408958, 0x0000000f, fbp_count); /* crop */ 135 } 136 137 static u8 138 gm200_gr_tile_map_6_24[] = { 139 0, 1, 2, 3, 4, 5, 3, 4, 5, 0, 1, 2, 0, 1, 2, 3, 4, 5, 3, 4, 5, 0, 1, 2, 140 }; 141 142 static u8 143 gm200_gr_tile_map_4_16[] = { 144 0, 1, 2, 3, 2, 3, 0, 1, 3, 0, 1, 2, 1, 2, 3, 0, 145 }; 146 147 static u8 148 gm200_gr_tile_map_2_8[] = { 149 0, 1, 1, 0, 0, 1, 1, 0, 150 }; 151 152 void 153 gm200_gr_oneinit_sm_id(struct gf100_gr *gr) 154 { 155 /*XXX: There's a different algorithm here I've not yet figured out. */ 156 gf100_gr_oneinit_sm_id(gr); 157 } 158 159 void 160 gm200_gr_oneinit_tiles(struct gf100_gr *gr) 161 { 162 /*XXX: Not sure what this is about. The algorithm from NVGPU 163 * seems to work for all boards I tried from earlier (and 164 * later) GPUs except in these specific configurations. 165 * 166 * Let's just hardcode them for now. 167 */ 168 if (gr->gpc_nr == 2 && gr->tpc_total == 8) { 169 memcpy(gr->tile, gm200_gr_tile_map_2_8, gr->tpc_total); 170 gr->screen_tile_row_offset = 1; 171 } else 172 if (gr->gpc_nr == 4 && gr->tpc_total == 16) { 173 memcpy(gr->tile, gm200_gr_tile_map_4_16, gr->tpc_total); 174 gr->screen_tile_row_offset = 4; 175 } else 176 if (gr->gpc_nr == 6 && gr->tpc_total == 24) { 177 memcpy(gr->tile, gm200_gr_tile_map_6_24, gr->tpc_total); 178 gr->screen_tile_row_offset = 5; 179 } else { 180 gf100_gr_oneinit_tiles(gr); 181 } 182 } 183 184 static const struct gf100_gr_func 185 gm200_gr = { 186 .oneinit_tiles = gm200_gr_oneinit_tiles, 187 .oneinit_sm_id = gm200_gr_oneinit_sm_id, 188 .init = gf100_gr_init, 189 .init_gpc_mmu = gm200_gr_init_gpc_mmu, 190 .init_bios = gm107_gr_init_bios, 191 .init_vsc_stream_master = gk104_gr_init_vsc_stream_master, 192 .init_zcull = gf117_gr_init_zcull, 193 .init_num_active_ltcs = gm200_gr_init_num_active_ltcs, 194 .init_rop_active_fbps = gm200_gr_init_rop_active_fbps, 195 .init_fecs_exceptions = gf100_gr_init_fecs_exceptions, 196 .init_ds_hww_esr_2 = gm200_gr_init_ds_hww_esr_2, 197 .init_sked_hww_esr = gk104_gr_init_sked_hww_esr, 198 .init_419cc0 = gf100_gr_init_419cc0, 199 .init_ppc_exceptions = gk104_gr_init_ppc_exceptions, 200 .init_tex_hww_esr = gf100_gr_init_tex_hww_esr, 201 .init_504430 = gm107_gr_init_504430, 202 .init_shader_exceptions = gm107_gr_init_shader_exceptions, 203 .init_400054 = gm107_gr_init_400054, 204 .trap_mp = gf100_gr_trap_mp, 205 .rops = gm200_gr_rops, 206 .tpc_nr = 4, 207 .ppc_nr = 2, 208 .grctx = &gm200_grctx, 209 .zbc = &gf100_gr_zbc, 210 .sclass = { 211 { -1, -1, FERMI_TWOD_A }, 212 { -1, -1, KEPLER_INLINE_TO_MEMORY_B }, 213 { -1, -1, MAXWELL_B, &gf100_fermi }, 214 { -1, -1, MAXWELL_COMPUTE_B }, 215 {} 216 } 217 }; 218 219 int 220 gm200_gr_load(struct gf100_gr *gr, int ver, const struct gf100_gr_fwif *fwif) 221 { 222 int ret; 223 224 ret = nvkm_acr_lsfw_load_bl_inst_data_sig(&gr->base.engine.subdev, 225 &gr->fecs.falcon, 226 NVKM_ACR_LSF_FECS, 227 "gr/fecs_", ver, fwif->fecs); 228 if (ret) 229 return ret; 230 231 ret = nvkm_acr_lsfw_load_bl_inst_data_sig(&gr->base.engine.subdev, 232 &gr->gpccs.falcon, 233 NVKM_ACR_LSF_GPCCS, 234 "gr/gpccs_", ver, 235 fwif->gpccs); 236 if (ret) 237 return ret; 238 239 gr->firmware = true; 240 241 return gk20a_gr_load_sw(gr, "gr/", ver); 242 } 243 244 MODULE_FIRMWARE("nvidia/gm200/gr/fecs_bl.bin"); 245 MODULE_FIRMWARE("nvidia/gm200/gr/fecs_inst.bin"); 246 MODULE_FIRMWARE("nvidia/gm200/gr/fecs_data.bin"); 247 MODULE_FIRMWARE("nvidia/gm200/gr/fecs_sig.bin"); 248 MODULE_FIRMWARE("nvidia/gm200/gr/gpccs_bl.bin"); 249 MODULE_FIRMWARE("nvidia/gm200/gr/gpccs_inst.bin"); 250 MODULE_FIRMWARE("nvidia/gm200/gr/gpccs_data.bin"); 251 MODULE_FIRMWARE("nvidia/gm200/gr/gpccs_sig.bin"); 252 MODULE_FIRMWARE("nvidia/gm200/gr/sw_ctx.bin"); 253 MODULE_FIRMWARE("nvidia/gm200/gr/sw_nonctx.bin"); 254 MODULE_FIRMWARE("nvidia/gm200/gr/sw_bundle_init.bin"); 255 MODULE_FIRMWARE("nvidia/gm200/gr/sw_method_init.bin"); 256 257 MODULE_FIRMWARE("nvidia/gm204/gr/fecs_bl.bin"); 258 MODULE_FIRMWARE("nvidia/gm204/gr/fecs_inst.bin"); 259 MODULE_FIRMWARE("nvidia/gm204/gr/fecs_data.bin"); 260 MODULE_FIRMWARE("nvidia/gm204/gr/fecs_sig.bin"); 261 MODULE_FIRMWARE("nvidia/gm204/gr/gpccs_bl.bin"); 262 MODULE_FIRMWARE("nvidia/gm204/gr/gpccs_inst.bin"); 263 MODULE_FIRMWARE("nvidia/gm204/gr/gpccs_data.bin"); 264 MODULE_FIRMWARE("nvidia/gm204/gr/gpccs_sig.bin"); 265 MODULE_FIRMWARE("nvidia/gm204/gr/sw_ctx.bin"); 266 MODULE_FIRMWARE("nvidia/gm204/gr/sw_nonctx.bin"); 267 MODULE_FIRMWARE("nvidia/gm204/gr/sw_bundle_init.bin"); 268 MODULE_FIRMWARE("nvidia/gm204/gr/sw_method_init.bin"); 269 270 MODULE_FIRMWARE("nvidia/gm206/gr/fecs_bl.bin"); 271 MODULE_FIRMWARE("nvidia/gm206/gr/fecs_inst.bin"); 272 MODULE_FIRMWARE("nvidia/gm206/gr/fecs_data.bin"); 273 MODULE_FIRMWARE("nvidia/gm206/gr/fecs_sig.bin"); 274 MODULE_FIRMWARE("nvidia/gm206/gr/gpccs_bl.bin"); 275 MODULE_FIRMWARE("nvidia/gm206/gr/gpccs_inst.bin"); 276 MODULE_FIRMWARE("nvidia/gm206/gr/gpccs_data.bin"); 277 MODULE_FIRMWARE("nvidia/gm206/gr/gpccs_sig.bin"); 278 MODULE_FIRMWARE("nvidia/gm206/gr/sw_ctx.bin"); 279 MODULE_FIRMWARE("nvidia/gm206/gr/sw_nonctx.bin"); 280 MODULE_FIRMWARE("nvidia/gm206/gr/sw_bundle_init.bin"); 281 MODULE_FIRMWARE("nvidia/gm206/gr/sw_method_init.bin"); 282 283 static const struct gf100_gr_fwif 284 gm200_gr_fwif[] = { 285 { 0, gm200_gr_load, &gm200_gr, &gm200_gr_fecs_acr, &gm200_gr_gpccs_acr }, 286 { -1, gm200_gr_nofw }, 287 {} 288 }; 289 290 int 291 gm200_gr_new(struct nvkm_device *device, int index, struct nvkm_gr **pgr) 292 { 293 return gf100_gr_new_(gm200_gr_fwif, device, index, pgr); 294 } 295