nvc0: add conservative rasterization support

Subpixel precision bias, dilation and the post-snap mode are supported on
GM200 and newer. The pre-snap mode is supported for triangle primitives on
GP100.

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

diff --git a/src/gallium/drivers/nouveau/nvc0/mme/com9097.mme b/src/gallium/drivers/nouveau/nvc0/mme/com9097.mme
index 7c5ec8f52b9fcf72e7c2dd26cf25f6c02a67de1d..ecf99606672b5faa6e26d9d973b0c3d097090640 100644 (file)
--- a/src/gallium/drivers/nouveau/nvc0/mme/com9097.mme
+++ b/src/gallium/drivers/nouveau/nvc0/mme/com9097.mme
@@ -550,3 +550,33 @@ qbw_postclamp:
 qbw_done:
    exit send (extrinsrt 0x0 $r4 0x0 0x10 0x10)
    maddrsend 0x44
+
+/* NVC0_3D_MACRO_CONSERVATIVE_RASTER_STATE:
+ *
+ * This sets basically all the conservative rasterization state. It sets
+ * CONSERVATIVE_RASTER to one while doing so.
+ *
+ * arg = biasx | biasy<<4 | (dilation*4)<<8 | mode<<10
+ */
+.section #mme9097_conservative_raster_state
+   /* Mode and dilation */
+   maddr 0x1d00 /* SCRATCH[0] */
+   send 0x0 /* unknown */
+   send (extrinsrt 0x0 $r1 8 3 23) /* value */
+   mov $r2 0x7
+   send (extrinsrt 0x0 $r2 0 3 23) /* write mask */
+   maddr 0x18c4 /* FIRMWARE[4] */
+   mov $r2 0x831
+   send (extrinsrt 0x0 $r2 0 12 11) /* sends 0x418800 */
+   /* Subpixel precision */
+   mov $r2 (extrinsrt 0x0 $r1 0 3 0)
+   mov $r2 (extrinsrt $r2 $r1 4 4 8)
+   maddr 0x8287 /* SUBPIXEL_PRECISION[0] (incrementing by 8 methods) */
+   mov $r3 16 /* loop counter */
+crs_loop:
+   mov $r3 (add $r3 -1)
+   branz $r3 #crs_loop
+   send $r2
+   /* Enable */
+   exit maddr 0x1452 /* CONSERVATIVE_RASTER */
+   send 0x1

diff --git a/src/gallium/drivers/nouveau/nvc0/mme/com9097.mme.h b/src/gallium/drivers/nouveau/nvc0/mme/com9097.mme.h
index 9618da6e28ccf19589624dd64cecaf0cc7b8f632..3eacda9a27a198d71e831a08dca3eef1ec128ed6 100644 (file)
--- a/src/gallium/drivers/nouveau/nvc0/mme/com9097.mme.h
+++ b/src/gallium/drivers/nouveau/nvc0/mme/com9097.mme.h
@@ -373,3 +373,24 @@ uint32_t mme9097_query_buffer_write[] = {
        0x840100c2,
        0x00110071,
 };
+
+uint32_t mme9097_conservative_raster_state[] = {
+       0x07400021,
+       0x00000041,
+       0xb8d04042,
+/* 0x000c: crs_loop */
+       0x0001c211,
+       0xb8c08042,
+       0x06310021,
+       0x020c4211,
+       0x5b008042,
+       0x00c04212,
+       0x41085212,
+       0x20a1c021,
+       0x00040311,
+       0xffffdb11,
+       0xffffd817,
+       0x00001041,
+       0x051480a1,
+       0x00004041,
+};

diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_3d.xml.h b/src/gallium/drivers/nouveau/nvc0/nvc0_3d.xml.h
index d7245fbcae1c50c326f60fa6784d7627a282b13b..c5456e48b586d96c547e67a713bc3214b32b11b0 100644 (file)
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_3d.xml.h
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_3d.xml.h
@@ -447,6 +447,10 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define NVC0_3D_VIEWPORT_TRANSLATE_Z__ESIZE                    0x00000020
 #define NVC0_3D_VIEWPORT_TRANSLATE_Z__LEN                      0x00000010
 
+#define NVC0_3D_SUBPIXEL_PRECISION(i0)                        (0x00000a1c + 0x20*(i0))
+#define NVC0_3D_SUBPIXEL_PRECISION__ESIZE                      0x00000020
+#define NVC0_3D_SUBPIXEL_PRECISION__LEN                                0x00000010
+
 #define NVC0_3D_VIEWPORT_HORIZ(i0)                            (0x00000c00 + 0x10*(i0))
 #define NVC0_3D_VIEWPORT_HORIZ__ESIZE                          0x00000010
 #define NVC0_3D_VIEWPORT_HORIZ__LEN                            0x00000010
@@ -780,6 +784,7 @@ WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
 #define NVC0_3D_UNK1140                                        0x00001140
 
 #define NVC0_3D_UNK1144                                        0x00001144
+#define NVC0_3D_CONSERVATIVE_RASTER                    0x00001148
 
 #define NVC0_3D_VTX_ATTR_DEFINE                                0x0000114c
 #define NVC0_3D_VTX_ATTR_DEFINE_ATTR__MASK                     0x000000ff

diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_macros.h b/src/gallium/drivers/nouveau/nvc0/nvc0_macros.h
index eeacc714f3e32367dd7713260301b755fa3dcdc1..7aa06337950ab73e653c9e842381dca13f7c57c3 100644 (file)
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_macros.h
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_macros.h
@@ -35,6 +35,8 @@
 
 #define NVC0_3D_MACRO_QUERY_BUFFER_WRITE                       0x00003858
 
-#define NVC0_CP_MACRO_LAUNCH_GRID_INDIRECT       0x00003860
+#define NVC0_CP_MACRO_LAUNCH_GRID_INDIRECT                     0x00003860
+
+#define NVC0_3D_MACRO_CONSERVATIVE_RASTER_STATE                        0x00003868
 
 #endif /* __NVC0_MACROS_H__ */

diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
index 37e43f02e36de9140e57a232ed06d81340218439..e1c7b3a2364fbdb5ca1e0df7351b95621f17c5b1 100644 (file)
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_screen.c
@@ -172,6 +172,8 @@ nvc0_screen_get_param(struct pipe_screen *pscreen, enum pipe_cap param)
       return 30;
    case PIPE_CAP_MAX_WINDOW_RECTANGLES:
       return NVC0_MAX_WINDOW_RECTANGLES;
+   case PIPE_CAP_MAX_CONSERVATIVE_RASTER_SUBPIXEL_PRECISION_BIAS:
+      return class_3d >= GM200_3D_CLASS ? 8 : 0;
 
    /* supported caps */
    case PIPE_CAP_TEXTURE_MIRROR_CLAMP:
@@ -264,7 +266,12 @@ nvc0_screen_get_param(struct pipe_screen *pscreen, enum pipe_cap param)
    case PIPE_CAP_TGSI_VS_LAYER_VIEWPORT:
    case PIPE_CAP_TGSI_TES_LAYER_VIEWPORT:
    case PIPE_CAP_POST_DEPTH_COVERAGE:
+   case PIPE_CAP_CONSERVATIVE_RASTER_POST_SNAP_TRIANGLES:
+   case PIPE_CAP_CONSERVATIVE_RASTER_POST_SNAP_POINTS_LINES:
+   case PIPE_CAP_CONSERVATIVE_RASTER_POST_DEPTH_COVERAGE:
       return class_3d >= GM200_3D_CLASS;
+   case PIPE_CAP_CONSERVATIVE_RASTER_PRE_SNAP_TRIANGLES:
+      return class_3d >= GP100_3D_CLASS;
    case PIPE_CAP_SEAMLESS_CUBE_MAP_PER_TEXTURE:
    case PIPE_CAP_TGSI_BALLOT:
    case PIPE_CAP_BINDLESS_TEXTURE:
@@ -309,12 +316,7 @@ nvc0_screen_get_param(struct pipe_screen *pscreen, enum pipe_cap param)
    case PIPE_CAP_FENCE_SIGNAL:
    case PIPE_CAP_CONSTBUF0_FLAGS:
    case PIPE_CAP_PACKED_UNIFORMS:
-   case PIPE_CAP_CONSERVATIVE_RASTER_POST_SNAP_TRIANGLES:
-   case PIPE_CAP_CONSERVATIVE_RASTER_POST_SNAP_POINTS_LINES:
-   case PIPE_CAP_CONSERVATIVE_RASTER_PRE_SNAP_TRIANGLES:
    case PIPE_CAP_CONSERVATIVE_RASTER_PRE_SNAP_POINTS_LINES:
-   case PIPE_CAP_CONSERVATIVE_RASTER_POST_DEPTH_COVERAGE:
-   case PIPE_CAP_MAX_CONSERVATIVE_RASTER_SUBPIXEL_PRECISION_BIAS:
       return 0;
 
    case PIPE_CAP_VENDOR_ID:
@@ -444,6 +446,8 @@ nvc0_screen_get_shader_param(struct pipe_screen *pscreen,
 static float
 nvc0_screen_get_paramf(struct pipe_screen *pscreen, enum pipe_capf param)
 {
+   const uint16_t class_3d = nouveau_screen(pscreen)->class_3d;
+
    switch (param) {
    case PIPE_CAPF_MAX_LINE_WIDTH:
    case PIPE_CAPF_MAX_LINE_WIDTH_AA:
@@ -457,9 +461,11 @@ nvc0_screen_get_paramf(struct pipe_screen *pscreen, enum pipe_capf param)
    case PIPE_CAPF_MAX_TEXTURE_LOD_BIAS:
       return 15.0f;
    case PIPE_CAPF_MIN_CONSERVATIVE_RASTER_DILATE:
+      return 0.0f;
    case PIPE_CAPF_MAX_CONSERVATIVE_RASTER_DILATE:
+      return class_3d >= GM200_3D_CLASS ? 0.75f : 0.0f;
    case PIPE_CAPF_CONSERVATIVE_RASTER_DILATE_GRANULARITY:
-      return 0.0f;
+      return class_3d >= GM200_3D_CLASS ? 0.25f : 0.0f;
    }
 
    NOUVEAU_ERR("unknown PIPE_CAPF %d\n", param);
@@ -1207,6 +1213,7 @@ nvc0_screen_create(struct nouveau_device *dev)
    MK_MACRO(NVC0_3D_MACRO_DRAW_ARRAYS_INDIRECT_COUNT, mme9097_draw_arrays_indirect_count);
    MK_MACRO(NVC0_3D_MACRO_DRAW_ELEMENTS_INDIRECT_COUNT, mme9097_draw_elts_indirect_count);
    MK_MACRO(NVC0_3D_MACRO_QUERY_BUFFER_WRITE, mme9097_query_buffer_write);
+   MK_MACRO(NVC0_3D_MACRO_CONSERVATIVE_RASTER_STATE, mme9097_conservative_raster_state);
    MK_MACRO(NVC0_CP_MACRO_LAUNCH_GRID_INDIRECT, mme90c0_launch_grid_indirect);
 
    BEGIN_NVC0(push, NVC0_3D(RASTERIZE_ENABLE), 1);

diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_state.c b/src/gallium/drivers/nouveau/nvc0/nvc0_state.c
index 3e204f09a4dba0730c0b66b35ac5eec256a93349..37dbbe66c7c0236fe61e0da1d257b1034040d33b 100644 (file)
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_state.c
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_state.c
@@ -327,6 +327,20 @@ nvc0_rasterizer_state_create(struct pipe_context *pipe,
 
     SB_IMMED_3D(so, PIXEL_CENTER_INTEGER, !cso->half_pixel_center);
 
+    if (class_3d >= GM200_3D_CLASS) {
+        if (cso->conservative_raster_mode != PIPE_CONSERVATIVE_RASTER_OFF) {
+            bool post_snap = cso->conservative_raster_mode ==
+                PIPE_CONSERVATIVE_RASTER_POST_SNAP;
+            uint32_t state = cso->subpixel_precision_x;
+            state |= cso->subpixel_precision_y << 4;
+            state |= (uint32_t)(cso->conservative_raster_dilate * 4) << 8;
+            state |= (post_snap || class_3d < GP100_3D_CLASS) ? 1 << 10 : 0;
+            SB_IMMED_3D(so, MACRO_CONSERVATIVE_RASTER_STATE, state);
+        } else {
+            SB_IMMED_3D(so, CONSERVATIVE_RASTER, 0);
+        }
+    }
+
     assert(so->size <= ARRAY_SIZE(so->state));
     return (void *)so;
 }

diff --git a/src/gallium/drivers/nouveau/nvc0/nvc0_stateobj.h b/src/gallium/drivers/nouveau/nvc0/nvc0_stateobj.h
index 3006ed61956a38a197b1c425b5d7cab71f90d470..e20032845cd0c7bbe1958d00502909a9f97fb6b0 100644 (file)
--- a/src/gallium/drivers/nouveau/nvc0/nvc0_stateobj.h
+++ b/src/gallium/drivers/nouveau/nvc0/nvc0_stateobj.h
@@ -23,7 +23,7 @@ struct nvc0_blend_stateobj {
 struct nvc0_rasterizer_stateobj {
    struct pipe_rasterizer_state pipe;
    int size;
-   uint32_t state[43];
+   uint32_t state[44];
 };
 
 struct nvc0_zsa_stateobj {