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 <subdev/secboot.h>
28 
29 #include <nvif/class.h>
30 
31 /*******************************************************************************
32  * PGRAPH engine/subdev functions
33  ******************************************************************************/
34 
35 int
36 gm200_gr_rops(struct gf100_gr *gr)
37 {
38 	return nvkm_rd32(gr->base.engine.subdev.device, 0x12006c);
39 }
40 
41 void
42 gm200_gr_init_ds_hww_esr_2(struct gf100_gr *gr)
43 {
44 	struct nvkm_device *device = gr->base.engine.subdev.device;
45 	nvkm_wr32(device, 0x405848, 0xc0000000);
46 	nvkm_mask(device, 0x40584c, 0x00000001, 0x00000001);
47 }
48 
49 void
50 gm200_gr_init_num_active_ltcs(struct gf100_gr *gr)
51 {
52 	struct nvkm_device *device = gr->base.engine.subdev.device;
53 	nvkm_wr32(device, GPC_BCAST(0x08ac), nvkm_rd32(device, 0x100800));
54 	nvkm_wr32(device, GPC_BCAST(0x033c), nvkm_rd32(device, 0x100804));
55 }
56 
57 void
58 gm200_gr_init_gpc_mmu(struct gf100_gr *gr)
59 {
60 	struct nvkm_device *device = gr->base.engine.subdev.device;
61 
62 	nvkm_wr32(device, 0x418880, nvkm_rd32(device, 0x100c80) & 0xf0001fff);
63 	nvkm_wr32(device, 0x418890, 0x00000000);
64 	nvkm_wr32(device, 0x418894, 0x00000000);
65 
66 	nvkm_wr32(device, 0x4188b4, nvkm_rd32(device, 0x100cc8));
67 	nvkm_wr32(device, 0x4188b8, nvkm_rd32(device, 0x100ccc));
68 	nvkm_wr32(device, 0x4188b0, nvkm_rd32(device, 0x100cc4));
69 }
70 
71 static void
72 gm200_gr_init_rop_active_fbps(struct gf100_gr *gr)
73 {
74 	struct nvkm_device *device = gr->base.engine.subdev.device;
75 	const u32 fbp_count = nvkm_rd32(device, 0x12006c);
76 	nvkm_mask(device, 0x408850, 0x0000000f, fbp_count); /* zrop */
77 	nvkm_mask(device, 0x408958, 0x0000000f, fbp_count); /* crop */
78 }
79 
80 static u8
81 gm200_gr_tile_map_6_24[] = {
82 	0, 1, 2, 3, 4, 5, 3, 4, 5, 0, 1, 2, 0, 1, 2, 3, 4, 5, 3, 4, 5, 0, 1, 2,
83 };
84 
85 static u8
86 gm200_gr_tile_map_4_16[] = {
87 	0, 1, 2, 3, 2, 3, 0, 1, 3, 0, 1, 2, 1, 2, 3, 0,
88 };
89 
90 static u8
91 gm200_gr_tile_map_2_8[] = {
92 	0, 1, 1, 0, 0, 1, 1, 0,
93 };
94 
95 void
96 gm200_gr_oneinit_sm_id(struct gf100_gr *gr)
97 {
98 	/*XXX: There's a different algorithm here I've not yet figured out. */
99 	gf100_gr_oneinit_sm_id(gr);
100 }
101 
102 void
103 gm200_gr_oneinit_tiles(struct gf100_gr *gr)
104 {
105 	/*XXX: Not sure what this is about.  The algorithm from NVGPU
106 	 *     seems to work for all boards I tried from earlier (and
107 	 *     later) GPUs except in these specific configurations.
108 	 *
109 	 *     Let's just hardcode them for now.
110 	 */
111 	if (gr->gpc_nr == 2 && gr->tpc_total == 8) {
112 		memcpy(gr->tile, gm200_gr_tile_map_2_8, gr->tpc_total);
113 		gr->screen_tile_row_offset = 1;
114 	} else
115 	if (gr->gpc_nr == 4 && gr->tpc_total == 16) {
116 		memcpy(gr->tile, gm200_gr_tile_map_4_16, gr->tpc_total);
117 		gr->screen_tile_row_offset = 4;
118 	} else
119 	if (gr->gpc_nr == 6 && gr->tpc_total == 24) {
120 		memcpy(gr->tile, gm200_gr_tile_map_6_24, gr->tpc_total);
121 		gr->screen_tile_row_offset = 5;
122 	} else {
123 		gf100_gr_oneinit_tiles(gr);
124 	}
125 }
126 
127 int
128 gm200_gr_new_(const struct gf100_gr_func *func, struct nvkm_device *device,
129 	      int index, struct nvkm_gr **pgr)
130 {
131 	struct gf100_gr *gr;
132 	int ret;
133 
134 	if (!(gr = kzalloc(sizeof(*gr), GFP_KERNEL)))
135 		return -ENOMEM;
136 	*pgr = &gr->base;
137 
138 	ret = gf100_gr_ctor(func, device, index, gr);
139 	if (ret)
140 		return ret;
141 
142 	/* Load firmwares for non-secure falcons */
143 	if (!nvkm_secboot_is_managed(device->secboot,
144 				     NVKM_SECBOOT_FALCON_FECS)) {
145 		if ((ret = gf100_gr_ctor_fw(gr, "gr/fecs_inst", &gr->fuc409c)) ||
146 		    (ret = gf100_gr_ctor_fw(gr, "gr/fecs_data", &gr->fuc409d)))
147 			return ret;
148 	}
149 	if (!nvkm_secboot_is_managed(device->secboot,
150 				     NVKM_SECBOOT_FALCON_GPCCS)) {
151 		if ((ret = gf100_gr_ctor_fw(gr, "gr/gpccs_inst", &gr->fuc41ac)) ||
152 		    (ret = gf100_gr_ctor_fw(gr, "gr/gpccs_data", &gr->fuc41ad)))
153 			return ret;
154 	}
155 
156 	if ((ret = gk20a_gr_av_to_init(gr, "gr/sw_nonctx", &gr->fuc_sw_nonctx)) ||
157 	    (ret = gk20a_gr_aiv_to_init(gr, "gr/sw_ctx", &gr->fuc_sw_ctx)) ||
158 	    (ret = gk20a_gr_av_to_init(gr, "gr/sw_bundle_init", &gr->fuc_bundle)) ||
159 	    (ret = gk20a_gr_av_to_method(gr, "gr/sw_method_init", &gr->fuc_method)))
160 		return ret;
161 
162 	return 0;
163 }
164 
165 static const struct gf100_gr_func
166 gm200_gr = {
167 	.oneinit_tiles = gm200_gr_oneinit_tiles,
168 	.oneinit_sm_id = gm200_gr_oneinit_sm_id,
169 	.init = gf100_gr_init,
170 	.init_gpc_mmu = gm200_gr_init_gpc_mmu,
171 	.init_bios = gm107_gr_init_bios,
172 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,
173 	.init_zcull = gf117_gr_init_zcull,
174 	.init_num_active_ltcs = gm200_gr_init_num_active_ltcs,
175 	.init_rop_active_fbps = gm200_gr_init_rop_active_fbps,
176 	.init_fecs_exceptions = gf100_gr_init_fecs_exceptions,
177 	.init_ds_hww_esr_2 = gm200_gr_init_ds_hww_esr_2,
178 	.init_sked_hww_esr = gk104_gr_init_sked_hww_esr,
179 	.init_419cc0 = gf100_gr_init_419cc0,
180 	.init_ppc_exceptions = gk104_gr_init_ppc_exceptions,
181 	.init_tex_hww_esr = gf100_gr_init_tex_hww_esr,
182 	.init_504430 = gm107_gr_init_504430,
183 	.init_shader_exceptions = gm107_gr_init_shader_exceptions,
184 	.init_400054 = gm107_gr_init_400054,
185 	.trap_mp = gf100_gr_trap_mp,
186 	.rops = gm200_gr_rops,
187 	.tpc_nr = 4,
188 	.ppc_nr = 2,
189 	.grctx = &gm200_grctx,
190 	.zbc = &gf100_gr_zbc,
191 	.sclass = {
192 		{ -1, -1, FERMI_TWOD_A },
193 		{ -1, -1, KEPLER_INLINE_TO_MEMORY_B },
194 		{ -1, -1, MAXWELL_B, &gf100_fermi },
195 		{ -1, -1, MAXWELL_COMPUTE_B },
196 		{}
197 	}
198 };
199 
200 int
201 gm200_gr_new(struct nvkm_device *device, int index, struct nvkm_gr **pgr)
202 {
203 	return gm200_gr_new_(&gm200_gr, device, index, pgr);
204 }
205