drm/nouveau/gr/gf100-: port tile mapping calculations from NVGPU

There's also a couple of hardcoded tables for a couple of very specific
configurations that NVGPU's algorithm didn't work for.
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>

drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf100.c

@@ -1116,27 +1116,14 @@ gf100_grctx_generate_rop_mapping(struct gf100_gr *gr)
 {
 	struct nvkm_device *device = gr->base.engine.subdev.device;
 	u32 data[6] = {}, data2[2] = {};
-	u8  tpcnr[GPC_MAX];
 	u8  shift, ntpcv;
-	int gpc, tpc, i;
-
-	/* calculate first set of magics */
-	memcpy(tpcnr, gr->tpc_nr, sizeof(gr->tpc_nr));
-
-	gpc = -1;
-	for (tpc = 0; tpc < gr->tpc_total; tpc++) {
-		do {
-			gpc = (gpc + 1) % gr->gpc_nr;
-		} while (!tpcnr[gpc]);
-		tpcnr[gpc]--;
-
-		data[tpc / 6] |= gpc << ((tpc % 6) * 5);
-	}
+	int i;
 
-	for (; tpc < 32; tpc++)
-		data[tpc / 6] |= 7 << ((tpc % 6) * 5);
+	/* Pack tile map into register format. */
+	for (i = 0; i < 32; i++)
+		data[i / 6] |= (gr->tile[i] & 0x07) << ((i % 6) * 5);
 
-	/* and the second... */
+	/* Magic. */
 	shift = 0;
 	ntpcv = gr->tpc_total;
 	while (!(ntpcv & (1 << 4))) {

drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgf117.c

@@ -194,27 +194,14 @@ gf117_grctx_generate_rop_mapping(struct gf100_gr *gr)
 {
 	struct nvkm_device *device = gr->base.engine.subdev.device;
 	u32 data[6] = {}, data2[2] = {};
-	u8  tpcnr[GPC_MAX];
 	u8  shift, ntpcv;
-	int gpc, tpc, i;
-
-	/* calculate first set of magics */
-	memcpy(tpcnr, gr->tpc_nr, sizeof(gr->tpc_nr));
-
-	gpc = -1;
-	for (tpc = 0; tpc < gr->tpc_total; tpc++) {
-		do {
-			gpc = (gpc + 1) % gr->gpc_nr;
-		} while (!tpcnr[gpc]);
-		tpcnr[gpc]--;
-
-		data[tpc / 6] |= gpc << ((tpc % 6) * 5);
-	}
+	int i;
 
-	for (; tpc < 32; tpc++)
-		data[tpc / 6] |= 7 << ((tpc % 6) * 5);
+	/* Pack tile map into register format. */
+	for (i = 0; i < 32; i++)
+		data[i / 6] |= (gr->tile[i] & 0x07) << ((i % 6) * 5);
 
-	/* and the second... */
+	/* Magic. */
 	shift = 0;
 	ntpcv = gr->tpc_total;
 	while (!(ntpcv & (1 << 4))) {

drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c

@@ -1652,6 +1652,82 @@ gf100_gr_init_ctxctl(struct gf100_gr *gr)
 	return ret;
 }
 
+void
+gf100_gr_oneinit_tiles(struct gf100_gr *gr)
+{
+	static const u8 primes[] = {
+		3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61
+	};
+	int init_frac[GPC_MAX], init_err[GPC_MAX], run_err[GPC_MAX], i, j;
+	u32 mul_factor, comm_denom;
+	u8  gpc_map[GPC_MAX];
+	bool sorted;
+
+	switch (gr->tpc_total) {
+	case 15: gr->screen_tile_row_offset = 0x06; break;
+	case 14: gr->screen_tile_row_offset = 0x05; break;
+	case 13: gr->screen_tile_row_offset = 0x02; break;
+	case 11: gr->screen_tile_row_offset = 0x07; break;
+	case 10: gr->screen_tile_row_offset = 0x06; break;
+	case  7:
+	case  5: gr->screen_tile_row_offset = 0x01; break;
+	case  3: gr->screen_tile_row_offset = 0x02; break;
+	case  2:
+	case  1: gr->screen_tile_row_offset = 0x01; break;
+	default: gr->screen_tile_row_offset = 0x03;
+		for (i = 0; i < ARRAY_SIZE(primes); i++) {
+			if (gr->tpc_total % primes[i]) {
+				gr->screen_tile_row_offset = primes[i];
+				break;
+			}
+		}
+		break;
+	}
+
+	/* Sort GPCs by TPC count, highest-to-lowest. */
+	for (i = 0; i < gr->gpc_nr; i++)
+		gpc_map[i] = i;
+	sorted = false;
+
+	while (!sorted) {
+		for (sorted = true, i = 0; i < gr->gpc_nr - 1; i++) {
+			if (gr->tpc_nr[gpc_map[i + 1]] >
+			    gr->tpc_nr[gpc_map[i + 0]]) {
+				u8 swap = gpc_map[i];
+				gpc_map[i + 0] = gpc_map[i + 1];
+				gpc_map[i + 1] = swap;
+				sorted = false;
+			}
+		}
+	}
+
+	/* Determine tile->GPC mapping */
+	mul_factor = gr->gpc_nr * gr->tpc_max;
+	if (mul_factor & 1)
+		mul_factor = 2;
+	else
+		mul_factor = 1;
+
+	comm_denom = gr->gpc_nr * gr->tpc_max * mul_factor;
+
+	for (i = 0; i < gr->gpc_nr; i++) {
+		init_frac[i] = gr->tpc_nr[gpc_map[i]] * gr->gpc_nr * mul_factor;
+		 init_err[i] = i * gr->tpc_max * mul_factor - comm_denom/2;
+		  run_err[i] = init_frac[i] + init_err[i];
+	}
+
+	for (i = 0; i < gr->tpc_total;) {
+		for (j = 0; j < gr->gpc_nr; j++) {
+			if ((run_err[j] * 2) >= comm_denom) {
+				gr->tile[i++] = gpc_map[j];
+				run_err[j] += init_frac[j] - comm_denom;
+			} else {
+				run_err[j] += init_frac[j];
+			}
+		}
+	}
+}
+
 static int
 gf100_gr_oneinit(struct nvkm_gr *base)
 {
@@ -1691,45 +1767,8 @@ gf100_gr_oneinit(struct nvkm_gr *base)
 		}
 	}
 
-	/*XXX: these need figuring out... though it might not even matter */
-	switch (device->chipset) {
-	case 0xc0:
-		if (gr->tpc_total == 11) { /* 465, 3/4/4/0, 4 */
-			gr->screen_tile_row_offset = 0x07;
-		} else
-		if (gr->tpc_total == 14) { /* 470, 3/3/4/4, 5 */
-			gr->screen_tile_row_offset = 0x05;
-		} else
-		if (gr->tpc_total == 15) { /* 480, 3/4/4/4, 6 */
-			gr->screen_tile_row_offset = 0x06;
-		}
-		break;
-	case 0xc3: /* 450, 4/0/0/0, 2 */
-		gr->screen_tile_row_offset = 0x03;
-		break;
-	case 0xc4: /* 460, 3/4/0/0, 4 */
-		gr->screen_tile_row_offset = 0x01;
-		break;
-	case 0xc1: /* 2/0/0/0, 1 */
-		gr->screen_tile_row_offset = 0x01;
-		break;
-	case 0xc8: /* 4/4/3/4, 5 */
-		gr->screen_tile_row_offset = 0x06;
-		break;
-	case 0xce: /* 4/4/0/0, 4 */
-		gr->screen_tile_row_offset = 0x03;
-		break;
-	case 0xcf: /* 4/0/0/0, 3 */
-		gr->screen_tile_row_offset = 0x03;
-		break;
-	case 0xd7:
-	case 0xd9: /* 1/0/0/0, 1 */
-	case 0xea: /* gk20a */
-	case 0x12b: /* gm20b */
-		gr->screen_tile_row_offset = 0x01;
-		break;
-	}
-
+	memset(gr->tile, 0xff, sizeof(gr->tile));
+	gr->func->oneinit_tiles(gr);
 	return 0;
 }
 
@@ -2164,6 +2203,7 @@ gf100_gr_gpccs_ucode = {
 
 static const struct gf100_gr_func
 gf100_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gf100_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.h

@@ -107,12 +107,13 @@ struct gf100_gr {
 	u8 ppc_tpc_nr[GPC_MAX][4];
 	u8 ppc_tpc_min;
 
+	u8 screen_tile_row_offset;
+	u8 tile[TPC_MAX];
+
 	struct gf100_gr_data mmio_data[4];
 	struct gf100_gr_mmio mmio_list[4096/8];
 	u32  size;
 	u32 *data;
-
-	u8 screen_tile_row_offset;
 };
 
 int gf100_gr_ctor(const struct gf100_gr_func *, struct nvkm_device *,
@@ -123,6 +124,7 @@ void *gf100_gr_dtor(struct nvkm_gr *);
 
 struct gf100_gr_func {
 	void (*dtor)(struct gf100_gr *);
+	void (*oneinit_tiles)(struct gf100_gr *);
 	int (*init)(struct gf100_gr *);
 	void (*init_gpc_mmu)(struct gf100_gr *);
 	void (*init_r405a14)(struct gf100_gr *);
@@ -164,6 +166,7 @@ struct gf100_gr_func {
 };
 
 int gf100_gr_rops(struct gf100_gr *);
+void gf100_gr_oneinit_tiles(struct gf100_gr *);
 int gf100_gr_init(struct gf100_gr *);
 void gf100_gr_init_vsc_stream_master(struct gf100_gr *);
 void gf100_gr_init_zcull(struct gf100_gr *);
@@ -191,6 +194,7 @@ void gm107_gr_init_400054(struct gf100_gr *);
 
 int gk20a_gr_init(struct gf100_gr *);
 
+void gm200_gr_oneinit_tiles(struct gf100_gr *);
 int gm200_gr_rops(struct gf100_gr *);
 void gm200_gr_init_num_active_ltcs(struct gf100_gr *);
 void gm200_gr_init_ds_hww_esr_2(struct gf100_gr *);

drivers/gpu/drm/nouveau/nvkm/engine/gr/gf104.c

@@ -114,6 +114,7 @@ gf104_gr_pack_mmio[] = {
 
 static const struct gf100_gr_func
 gf104_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gf100_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gf108.c

@@ -111,6 +111,7 @@ gf108_gr_init_r405a14(struct gf100_gr *gr)
 
 static const struct gf100_gr_func
 gf108_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_r405a14 = gf108_gr_init_r405a14,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gf110.c

@@ -86,6 +86,7 @@ gf110_gr_pack_mmio[] = {
 
 static const struct gf100_gr_func
 gf110_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gf100_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gf117.c

@@ -150,6 +150,7 @@ gf117_gr_init_zcull(struct gf100_gr *gr)
 
 static const struct gf100_gr_func
 gf117_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gf100_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gf119.c

@@ -177,6 +177,7 @@ gf119_gr_pack_mmio[] = {
 
 static const struct gf100_gr_func
 gf119_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gf100_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gk104.c

@@ -448,6 +448,7 @@ gk104_gr_gpccs_ucode = {
 
 static const struct gf100_gr_func
 gk104_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110.c

@@ -350,6 +350,7 @@ gk110_gr_init_419eb4(struct gf100_gr *gr)
 
 static const struct gf100_gr_func
 gk110_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gk110b.c

@@ -102,6 +102,7 @@ gk110b_gr_pack_mmio[] = {
 
 static const struct gf100_gr_func
 gk110b_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gk208.c

@@ -161,6 +161,7 @@ gk208_gr_gpccs_ucode = {
 
 static const struct gf100_gr_func
 gk208_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gf100_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gk20a.c

@@ -282,6 +282,7 @@ gk20a_gr_init(struct gf100_gr *gr)
 
 static const struct gf100_gr_func
 gk20a_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gk20a_gr_init,
 	.init_zcull = gf117_gr_init_zcull,
 	.init_rop_active_fbps = gk104_gr_init_rop_active_fbps,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gm107.c

@@ -391,6 +391,7 @@ gm107_gr_gpccs_ucode = {
 
 static const struct gf100_gr_func
 gm107_gr = {
+	.oneinit_tiles = gf100_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gm107_gr_init_gpc_mmu,
 	.init_bios = gm107_gr_init_bios,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gm200.c

@@ -77,6 +77,46 @@ gm200_gr_init_rop_active_fbps(struct gf100_gr *gr)
 	nvkm_mask(device, 0x408958, 0x0000000f, fbp_count); /* crop */
 }
 
+static u8
+gm200_gr_tile_map_6_24[] = {
+	0, 1, 2, 3, 4, 5, 3, 4, 5, 0, 1, 2, 0, 1, 2, 3, 4, 5, 3, 4, 5, 0, 1, 2,
+};
+
+static u8
+gm200_gr_tile_map_4_16[] = {
+	0, 1, 2, 3, 2, 3, 0, 1, 3, 0, 1, 2, 1, 2, 3, 0,
+};
+
+static u8
+gm200_gr_tile_map_2_8[] = {
+	0, 1, 1, 0, 0, 1, 1, 0,
+};
+
+void
+gm200_gr_oneinit_tiles(struct gf100_gr *gr)
+{
+	/*XXX: Not sure what this is about.  The algorithm from NVGPU
+	 *     seems to work for all boards I tried from earlier (and
+	 *     later) GPUs except in these specific configurations.
+	 *
+	 *     Let's just hardcode them for now.
+	 */
+	if (gr->gpc_nr == 2 && gr->tpc_total == 8) {
+		memcpy(gr->tile, gm200_gr_tile_map_2_8, gr->tpc_total);
+		gr->screen_tile_row_offset = 1;
+	} else
+	if (gr->gpc_nr == 4 && gr->tpc_total == 16) {
+		memcpy(gr->tile, gm200_gr_tile_map_4_16, gr->tpc_total);
+		gr->screen_tile_row_offset = 4;
+	} else
+	if (gr->gpc_nr == 6 && gr->tpc_total == 24) {
+		memcpy(gr->tile, gm200_gr_tile_map_6_24, gr->tpc_total);
+		gr->screen_tile_row_offset = 5;
+	} else {
+		gf100_gr_oneinit_tiles(gr);
+	}
+}
+
 int
 gm200_gr_new_(const struct gf100_gr_func *func, struct nvkm_device *device,
 	      int index, struct nvkm_gr **pgr)
@@ -117,6 +157,7 @@ gm200_gr_new_(const struct gf100_gr_func *func, struct nvkm_device *device,
 
 static const struct gf100_gr_func
 gm200_gr = {
+	.oneinit_tiles = gm200_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gm200_gr_init_gpc_mmu,
 	.init_bios = gm107_gr_init_bios,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gm20b.c

@@ -64,6 +64,7 @@ gm20b_gr_set_hww_esr_report_mask(struct gf100_gr *gr)
 
 static const struct gf100_gr_func
 gm20b_gr = {
+	.oneinit_tiles = gm200_gr_oneinit_tiles,
 	.init = gk20a_gr_init,
 	.init_zcull = gf117_gr_init_zcull,
 	.init_gpc_mmu = gm20b_gr_init_gpc_mmu,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gp100.c

@@ -64,6 +64,7 @@ gp100_gr_init_rop_active_fbps(struct gf100_gr *gr)
 
 static const struct gf100_gr_func
 gp100_gr = {
+	.oneinit_tiles = gm200_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gm200_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gp102.c

@@ -42,6 +42,7 @@ gp102_gr_init_swdx_pes_mask(struct gf100_gr *gr)
 
 static const struct gf100_gr_func
 gp102_gr = {
+	.oneinit_tiles = gm200_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gm200_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gp104.c

@@ -26,6 +26,7 @@
 
 static const struct gf100_gr_func
 gp104_gr = {
+	.oneinit_tiles = gm200_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gm200_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gp107.c

@@ -28,6 +28,7 @@
 
 static const struct gf100_gr_func
 gp107_gr = {
+	.oneinit_tiles = gm200_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gm200_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,

drivers/gpu/drm/nouveau/nvkm/engine/gr/gp10b.c

@@ -27,6 +27,7 @@
 
 static const struct gf100_gr_func
 gp10b_gr = {
+	.oneinit_tiles = gm200_gr_oneinit_tiles,
 	.init = gf100_gr_init,
 	.init_gpc_mmu = gm200_gr_init_gpc_mmu,
 	.init_vsc_stream_master = gk104_gr_init_vsc_stream_master,