1 files changed, 56 insertions, 64 deletions

diff --git a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm200.c b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm200.c
index db209d33f486..013d05a0f0f6 100644
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm200.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/ctxgm200.c
@@ -28,47 +28,34 @@
  ******************************************************************************/
 
 void
-gm200_grctx_generate_tpcid(struct gf100_gr *gr)
+gm200_grctx_generate_r419a3c(struct gf100_gr *gr)
 {
 	struct nvkm_device *device = gr->base.engine.subdev.device;
-	int gpc, tpc, id;
+	nvkm_mask(device, 0x419a3c, 0x00000014, 0x00000000);
+}
 
-	for (tpc = 0, id = 0; tpc < TPC_MAX_PER_GPC; tpc++) {
-		for (gpc = 0; gpc < gr->gpc_nr; gpc++) {
-			if (tpc < gr->tpc_nr[gpc]) {
-				nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x698), id);
-				nvkm_wr32(device, GPC_UNIT(gpc, 0x0c10 + tpc * 4), id);
-				nvkm_wr32(device, TPC_UNIT(gpc, tpc, 0x088), id);
-				id++;
-			}
-		}
-	}
+static void
+gm200_grctx_generate_r418e94(struct gf100_gr *gr)
+{
+	struct nvkm_device *device = gr->base.engine.subdev.device;
+	nvkm_mask(device, 0x418e94, 0xffffffff, 0xc4230000);
+	nvkm_mask(device, 0x418e4c, 0xffffffff, 0x70000000);
 }
 
 void
-gm200_grctx_generate_405b60(struct gf100_gr *gr)
+gm200_grctx_generate_smid_config(struct gf100_gr *gr)
 {
 	struct nvkm_device *device = gr->base.engine.subdev.device;
 	const u32 dist_nr = DIV_ROUND_UP(gr->tpc_total, 4);
 	u32 dist[TPC_MAX / 4] = {};
 	u32 gpcs[GPC_MAX] = {};
-	u8  tpcnr[GPC_MAX];
-	int tpc, gpc, i;
+	u8  sm, i;
 
-	memcpy(tpcnr, gr->tpc_nr, sizeof(gr->tpc_nr));
-
-	/* won't result in the same distribution as the binary driver where
-	 * some of the gpcs have more tpcs than others, but this shall do
-	 * for the moment.  the code for earlier gpus has this issue too.
-	 */
-	for (gpc = -1, i = 0; i < gr->tpc_total; i++) {
-		do {
-			gpc = (gpc + 1) % gr->gpc_nr;
-		} while(!tpcnr[gpc]);
-		tpc = gr->tpc_nr[gpc] - tpcnr[gpc]--;
-
-		dist[i / 4] |= ((gpc << 4) | tpc) << ((i % 4) * 8);
-		gpcs[gpc] |= i << (tpc * 8);
+	for (sm = 0; sm < gr->sm_nr; sm++) {
+		const u8 gpc = gr->sm[sm].gpc;
+		const u8 tpc = gr->sm[sm].tpc;
+		dist[sm / 4] |= ((gpc << 4) | tpc) << ((sm % 4) * 8);
+		gpcs[gpc] |= sm << (tpc * 8);
 	}
 
 	for (i = 0; i < dist_nr; i++)
@@ -77,50 +64,46 @@ gm200_grctx_generate_405b60(struct gf100_gr *gr)
 		nvkm_wr32(device, 0x405ba0 + (i * 4), gpcs[i]);
 }
 
-static void
-gm200_grctx_generate_main(struct gf100_gr *gr, struct gf100_grctx *info)
+void
+gm200_grctx_generate_tpc_mask(struct gf100_gr *gr)
 {
-	struct nvkm_device *device = gr->base.engine.subdev.device;
-	const struct gf100_grctx_func *grctx = gr->func->grctx;
-	u32 idle_timeout, tmp;
-	int i;
-
-	gf100_gr_mmio(gr, gr->fuc_sw_ctx);
-
-	idle_timeout = nvkm_mask(device, 0x404154, 0xffffffff, 0x00000000);
-
-	grctx->bundle(info);
-	grctx->pagepool(info);
-	grctx->attrib(info);
-	grctx->unkn(gr);
-
-	gm200_grctx_generate_tpcid(gr);
-	gf100_grctx_generate_r406028(gr);
-	gk104_grctx_generate_r418bb8(gr);
-
-	for (i = 0; i < 8; i++)
-		nvkm_wr32(device, 0x4064d0 + (i * 0x04), 0x00000000);
-	nvkm_wr32(device, 0x406500, 0x00000000);
-
-	nvkm_wr32(device, 0x405b00, (gr->tpc_total << 8) | gr->gpc_nr);
-
+	u32 tmp, i;
 	for (tmp = 0, i = 0; i < gr->gpc_nr; i++)
-		tmp |= ((1 << gr->tpc_nr[i]) - 1) << (i * 4);
-	nvkm_wr32(device, 0x4041c4, tmp);
+		tmp |= ((1 << gr->tpc_nr[i]) - 1) << (i * gr->func->tpc_nr);
+	nvkm_wr32(gr->base.engine.subdev.device, 0x4041c4, tmp);
+}
 
-	gm200_grctx_generate_405b60(gr);
+void
+gm200_grctx_generate_r406500(struct gf100_gr *gr)
+{
+	nvkm_wr32(gr->base.engine.subdev.device, 0x406500, 0x00000000);
+}
 
-	gf100_gr_icmd(gr, gr->fuc_bundle);
-	nvkm_wr32(device, 0x404154, idle_timeout);
-	gf100_gr_mthd(gr, gr->fuc_method);
+void
+gm200_grctx_generate_dist_skip_table(struct gf100_gr *gr)
+{
+	struct nvkm_device *device = gr->base.engine.subdev.device;
+	u32 data[8] = {};
+	int gpc, ppc, i;
+
+	for (gpc = 0; gpc < gr->gpc_nr; gpc++) {
+		for (ppc = 0; ppc < gr->ppc_nr[gpc]; ppc++) {
+			u8 ppc_tpcs = gr->ppc_tpc_nr[gpc][ppc];
+			u8 ppc_tpcm = gr->ppc_tpc_mask[gpc][ppc];
+			while (ppc_tpcs-- > gr->ppc_tpc_min)
+				ppc_tpcm &= ppc_tpcm - 1;
+			ppc_tpcm ^= gr->ppc_tpc_mask[gpc][ppc];
+			((u8 *)data)[gpc] |= ppc_tpcm;
+		}
+	}
 
-	nvkm_mask(device, 0x418e94, 0xffffffff, 0xc4230000);
-	nvkm_mask(device, 0x418e4c, 0xffffffff, 0x70000000);
+	for (i = 0; i < ARRAY_SIZE(data); i++)
+		nvkm_wr32(device, 0x4064d0 + (i * 0x04), data[i]);
 }
 
 const struct gf100_grctx_func
 gm200_grctx = {
-	.main  = gm200_grctx_generate_main,
+	.main  = gf100_grctx_generate_main,
 	.unkn  = gk104_grctx_generate_unkn,
 	.bundle = gm107_grctx_generate_bundle,
 	.bundle_size = 0x3000,
@@ -133,4 +116,13 @@ gm200_grctx = {
 	.attrib_nr = 0x400,
 	.alpha_nr_max = 0x1800,
 	.alpha_nr = 0x1000,
+	.sm_id = gm107_grctx_generate_sm_id,
+	.rop_mapping = gf117_grctx_generate_rop_mapping,
+	.dist_skip_table = gm200_grctx_generate_dist_skip_table,
+	.r406500 = gm200_grctx_generate_r406500,
+	.gpc_tpc_nr = gk104_grctx_generate_gpc_tpc_nr,
+	.tpc_mask = gm200_grctx_generate_tpc_mask,
+	.smid_config = gm200_grctx_generate_smid_config,
+	.r418e94 = gm200_grctx_generate_r418e94,
+	.r419a3c = gm200_grctx_generate_r419a3c,
 };