[Nouveau] [PATCH 3/9] nvkm/fb/ramnv50: Make 0x100da0 per-partition

[Ilia Mirkin]

On Fri, May 22, 2015 at 6:33 PM, Roy Spliet <rspliet at eclipso.eu> wrote:
> Like on GT215
>
> Signed-off-by: Roy Spliet <rspliet at eclipso.eu>
> ---
>  drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv50.c | 20 +++++++++++++-------
>  1 file changed, 13 insertions(+), 7 deletions(-)
>
> diff --git a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv50.c b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv50.c
> index d2c81dd..a96e512 100644
> --- a/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv50.c
> +++ b/drivers/gpu/drm/nouveau/nvkm/subdev/fb/ramnv50.c
> @@ -42,7 +42,7 @@ struct nv50_ramseq {
>         struct hwsq_reg r_0x1002d0;
>         struct hwsq_reg r_0x1002d4;
>         struct hwsq_reg r_0x1002dc;
> -       struct hwsq_reg r_0x100da0[8];
> +       struct hwsq_reg r_0x100da0;
>         struct hwsq_reg r_0x100e20;
>         struct hwsq_reg r_0x100e24;
>         struct hwsq_reg r_0x611200;
> @@ -70,6 +70,7 @@ nv50_ram_calc(struct nvkm_fb *pfb, u32 freq)
>                 u8  size;
>         } ramcfg, timing;
>         u8  ver, hdr, cnt, len, strap;
> +       u32 r100da0;
>         int N1, M1, N2, M2, P;
>         int ret, i;
>
> @@ -109,6 +110,13 @@ nv50_ram_calc(struct nvkm_fb *pfb, u32 freq)
>                 timing.data = 0;
>         }
>
> +       /* XXX: where the fuck does 750MHz come from? */

s/the fuck//

> +       if (freq <= 750000) {
> +               r100da0 = 0x00000010;
> +       } else {
> +               r100da0 = 0x00000000;
> +       }
> +
>         ret = ram_init(hwsq, nv_subdev(pfb));
>         if (ret)
>                 return ret;
> @@ -144,10 +152,9 @@ nv50_ram_calc(struct nvkm_fb *pfb, u32 freq)
>         ram_mask(hwsq, 0x00400c, 0x0000ffff, (N1 << 8) | M1);
>         ram_mask(hwsq, 0x004008, 0x81ff0000, 0x80000000 | (mpll.bias_p << 19) |
>                                              (P << 22) | (P << 16));
> -#if QFX5800NVA0
> -       for (i = 0; i < 8; i++)
> -               ram_mask(hwsq, 0x100da0[i], 0x00000000, 0x00000000); /*XXX*/
> -#endif
> +
> +       if (nv_device(pfb)->chipset == 0xa0)

Don't think this is quite right... probably more like vram == GDDR3. I
definitely remember seeing non-nva0 cards writing 100da0 in the
scripts.

> +               ram_wr32(hwsq, 0x100da0, r100da0); /*XXX: here?*/
>         ram_nsec(hwsq, 96000); /*XXX*/
>         ram_mask(hwsq, 0x004008, 0x00002200, 0x00002000);
>
> @@ -430,8 +437,7 @@ nv50_ram_ctor(struct nvkm_object *parent, struct nvkm_object *engine,
>         ram->hwsq.r_0x1002d0 = hwsq_reg(0x1002d0);
>         ram->hwsq.r_0x1002d4 = hwsq_reg(0x1002d4);
>         ram->hwsq.r_0x1002dc = hwsq_reg(0x1002dc);
> -       for (i = 0; i < 8; i++)
> -               ram->hwsq.r_0x100da0[i] = hwsq_reg(0x100da0 + (i * 0x04));
> +       ram->hwsq.r_0x100da0 = hwsq_stride(0x100da0, 4, ram->base.part_mask);
>         ram->hwsq.r_0x100e20 = hwsq_reg(0x100e20);
>         ram->hwsq.r_0x100e24 = hwsq_reg(0x100e24);
>         ram->hwsq.r_0x611200 = hwsq_reg(0x611200);
> --
> 2.1.0
>
>
>
> _______________________________________________
> Nouveau mailing list
> Nouveau at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/nouveau