From: Ben Skeggs <bskeggs@redhat.com>
Date: Tue, 8 May 2018 20:39:47 +1000
Subject: drm/nouveau/gr/gf100-: port zcull tile mapping calculations from NVGPU
Git-commit: d00ffc0c403784c9f88d8da357f9f33f855289a4
Patch-mainline: v4.18-rc1
References: FATE#326289 FATE#326079 FATE#326049 FATE#322398 FATE#326166
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
Acked-by: Petr Tesarik <ptesarik@suse.com>
---
drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c | 32 ++++++++++++-------------
drivers/gpu/drm/nouveau/nvkm/engine/gr/gf117.c | 32 ++++++++++++-------------
2 files changed, 32 insertions(+), 32 deletions(-)
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf100.c
@@ -2042,25 +2042,25 @@ gf100_gr_init_zcull(struct gf100_gr *gr)
{
struct nvkm_device *device = gr->base.engine.subdev.device;
const u32 magicgpc918 = DIV_ROUND_UP(0x00800000, gr->tpc_total);
- u32 data[TPC_MAX / 8] = {};
- u8 tpcnr[GPC_MAX];
- int gpc, tpc;
- int i;
+ const u8 tile_nr = ALIGN(gr->tpc_total, 32);
+ u8 bank[GPC_MAX] = {}, gpc, i, j;
+ u32 data;
- memcpy(tpcnr, gr->tpc_nr, sizeof(gr->tpc_nr));
- for (i = 0, gpc = -1; i < gr->tpc_total; i++) {
- do {
- gpc = (gpc + 1) % gr->gpc_nr;
- } while (!tpcnr[gpc]);
- tpc = gr->tpc_nr[gpc] - tpcnr[gpc]--;
-
- data[i / 8] |= tpc << ((i % 8) * 4);
+ for (i = 0; i < tile_nr; i += 8) {
+ for (data = 0, j = 0; j < 8 && i + j < gr->tpc_total; j++) {
+ data |= bank[gr->tile[i + j]] << (j * 4);
+ bank[gr->tile[i + j]]++;
+ }
+ nvkm_wr32(device, GPC_BCAST(0x0980 + ((i / 8) * 4)), data);
}
- nvkm_wr32(device, GPC_BCAST(0x0980), data[0]);
- nvkm_wr32(device, GPC_BCAST(0x0984), data[1]);
- nvkm_wr32(device, GPC_BCAST(0x0988), data[2]);
- nvkm_wr32(device, GPC_BCAST(0x098c), data[3]);
+ for (gpc = 0; gpc < gr->gpc_nr; gpc++) {
+ nvkm_wr32(device, GPC_UNIT(gpc, 0x0914),
+ gr->screen_tile_row_offset << 8 | gr->tpc_nr[gpc]);
+ nvkm_wr32(device, GPC_UNIT(gpc, 0x0910), 0x00040000 |
+ gr->tpc_total);
+ nvkm_wr32(device, GPC_UNIT(gpc, 0x0918), magicgpc918);
+ }
nvkm_wr32(device, GPC_BCAST(0x1bd4), magicgpc918);
}
--- a/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf117.c
+++ b/drivers/gpu/drm/nouveau/nvkm/engine/gr/gf117.c
@@ -125,25 +125,25 @@ gf117_gr_init_zcull(struct gf100_gr *gr)
{
struct nvkm_device *device = gr->base.engine.subdev.device;
const u32 magicgpc918 = DIV_ROUND_UP(0x00800000, gr->tpc_total);
- u32 data[TPC_MAX / 8] = {};
- u8 tpcnr[GPC_MAX];
- int gpc, tpc;
- int i;
+ const u8 tile_nr = ALIGN(gr->tpc_total, 32);
+ u8 bank[GPC_MAX] = {}, gpc, i, j;
+ u32 data;
- memcpy(tpcnr, gr->tpc_nr, sizeof(gr->tpc_nr));
- for (i = 0, gpc = -1; i < gr->tpc_total; i++) {
- do {
- gpc = (gpc + 1) % gr->gpc_nr;
- } while (!tpcnr[gpc]);
- tpc = gr->tpc_nr[gpc] - tpcnr[gpc]--;
-
- data[i / 8] |= tpc << ((i % 8) * 4);
+ for (i = 0; i < tile_nr; i += 8) {
+ for (data = 0, j = 0; j < 8 && i + j < gr->tpc_total; j++) {
+ data |= bank[gr->tile[i + j]] << (j * 4);
+ bank[gr->tile[i + j]]++;
+ }
+ nvkm_wr32(device, GPC_BCAST(0x0980 + ((i / 8) * 4)), data);
}
- nvkm_wr32(device, GPC_BCAST(0x0980), data[0]);
- nvkm_wr32(device, GPC_BCAST(0x0984), data[1]);
- nvkm_wr32(device, GPC_BCAST(0x0988), data[2]);
- nvkm_wr32(device, GPC_BCAST(0x098c), data[3]);
+ for (gpc = 0; gpc < gr->gpc_nr; gpc++) {
+ nvkm_wr32(device, GPC_UNIT(gpc, 0x0914),
+ gr->screen_tile_row_offset << 8 | gr->tpc_nr[gpc]);
+ nvkm_wr32(device, GPC_UNIT(gpc, 0x0910), 0x00040000 |
+ gr->tpc_total);
+ nvkm_wr32(device, GPC_UNIT(gpc, 0x0918), magicgpc918);
+ }
nvkm_wr32(device, GPC_BCAST(0x3fd4), magicgpc918);
}