On Thu, Dec 26, 2013 at 3:02 AM, Christoph Bumiller <[email protected]> wrote: > This fixes write faults from GPCCLIENT 5 in geometry shader tests. > > For now this is just a guess from a single mmio trace, it looks > like 8 large pages are being used, i.e. one per TPC. > > Can we just ask NV for the correct sizes ? https://android.googlesource.com/kernel/tegra/+/android-tegra-3.10/drivers/video/tegra/host/gk20a/gr_gk20a.c might be of some help here.
The buffer in question is the "attribute buffer" > --- > drivers/gpu/drm/nouveau/core/engine/graph/ctxnve4.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/gpu/drm/nouveau/core/engine/graph/ctxnve4.c > b/drivers/gpu/drm/nouveau/core/engine/graph/ctxnve4.c > index e2de73e..fc1cc32 100644 > --- a/drivers/gpu/drm/nouveau/core/engine/graph/ctxnve4.c > +++ b/drivers/gpu/drm/nouveau/core/engine/graph/ctxnve4.c > @@ -813,7 +813,7 @@ nve4_grctx_generate_mods(struct nvc0_graph_priv *priv, > struct nvc0_grctx *info) > > mmio_data(0x003000, 0x0100, NV_MEM_ACCESS_RW | NV_MEM_ACCESS_SYS); > mmio_data(0x008000, 0x0100, NV_MEM_ACCESS_RW | NV_MEM_ACCESS_SYS); > - mmio_data(0x060000, 0x1000, NV_MEM_ACCESS_RW); > + mmio_data(priv->tpc_total << 17, 0x1000, NV_MEM_ACCESS_RW); > mmio_list(0x40800c, 0x00000000, 8, 1); > mmio_list(0x408010, 0x80000000, 0, 0); > mmio_list(0x419004, 0x00000000, 8, 1); > -- > 1.8.1.5 > > _______________________________________________ > Nouveau mailing list > [email protected] > http://lists.freedesktop.org/mailman/listinfo/nouveau _______________________________________________ Nouveau mailing list [email protected] http://lists.freedesktop.org/mailman/listinfo/nouveau
