BRW_SURFACEFORMAT_R16G16B16A16_FLOAT
};
+static GLuint uint_types_direct[5] = {
+ 0,
+ BRW_SURFACEFORMAT_R32_UINT,
+ BRW_SURFACEFORMAT_R32G32_UINT,
+ BRW_SURFACEFORMAT_R32G32B32_UINT,
+ BRW_SURFACEFORMAT_R32G32B32A32_UINT
+};
+
static GLuint uint_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R32_UNORM,
BRW_SURFACEFORMAT_R32G32B32A32_USCALED
};
+static GLuint int_types_direct[5] = {
+ 0,
+ BRW_SURFACEFORMAT_R32_SINT,
+ BRW_SURFACEFORMAT_R32G32_SINT,
+ BRW_SURFACEFORMAT_R32G32B32_SINT,
+ BRW_SURFACEFORMAT_R32G32B32A32_SINT
+};
+
static GLuint int_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R32_SNORM,
BRW_SURFACEFORMAT_R32G32B32A32_SSCALED
};
+static GLuint ushort_types_direct[5] = {
+ 0,
+ BRW_SURFACEFORMAT_R16_UINT,
+ BRW_SURFACEFORMAT_R16G16_UINT,
+ BRW_SURFACEFORMAT_R16G16B16A16_UINT,
+ BRW_SURFACEFORMAT_R16G16B16A16_UINT
+};
+
static GLuint ushort_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R16_UNORM,
BRW_SURFACEFORMAT_R16G16B16A16_USCALED
};
+static GLuint short_types_direct[5] = {
+ 0,
+ BRW_SURFACEFORMAT_R16_SINT,
+ BRW_SURFACEFORMAT_R16G16_SINT,
+ BRW_SURFACEFORMAT_R16G16B16A16_SINT,
+ BRW_SURFACEFORMAT_R16G16B16A16_SINT
+};
+
static GLuint short_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R16_SNORM,
BRW_SURFACEFORMAT_R16G16B16A16_SSCALED
};
+static GLuint ubyte_types_direct[5] = {
+ 0,
+ BRW_SURFACEFORMAT_R8_UINT,
+ BRW_SURFACEFORMAT_R8G8_UINT,
+ BRW_SURFACEFORMAT_R8G8B8A8_UINT,
+ BRW_SURFACEFORMAT_R8G8B8A8_UINT
+};
+
static GLuint ubyte_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R8_UNORM,
BRW_SURFACEFORMAT_R8G8B8A8_USCALED
};
+static GLuint byte_types_direct[5] = {
+ 0,
+ BRW_SURFACEFORMAT_R8_SINT,
+ BRW_SURFACEFORMAT_R8G8_SINT,
+ BRW_SURFACEFORMAT_R8G8B8A8_SINT,
+ BRW_SURFACEFORMAT_R8G8B8A8_SINT
+};
+
static GLuint byte_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R8_SNORM,
* Format will be GL_RGBA or possibly GL_BGRA for GLubyte[4] color arrays.
*/
static GLuint get_surface_type( GLenum type, GLuint size,
- GLenum format, bool normalized )
+ GLenum format, bool normalized, bool integer )
{
if (unlikely(INTEL_DEBUG & DEBUG_VERTS))
printf("type %s size %d normalized %d\n",
_mesa_lookup_enum_by_nr(type), size, normalized);
- if (normalized) {
+ if (integer) {
+ assert(format == GL_RGBA); /* sanity check */
+ switch (type) {
+ case GL_INT: return int_types_direct[size];
+ case GL_SHORT: return short_types_direct[size];
+ case GL_BYTE: return byte_types_direct[size];
+ case GL_UNSIGNED_INT: return uint_types_direct[size];
+ case GL_UNSIGNED_SHORT: return ushort_types_direct[size];
+ case GL_UNSIGNED_BYTE: return ubyte_types_direct[size];
+ default: assert(0); return 0;
+ }
+ } else if (normalized) {
switch (type) {
case GL_DOUBLE: return double_types[size];
case GL_FLOAT: return float_types[size];
struct gl_context *ctx = &brw->intel.ctx;
struct intel_context *intel = intel_context(ctx);
/* CACHE_NEW_VS_PROG */
- GLbitfield vs_inputs = brw->vs.prog_data->inputs_read;
+ GLbitfield64 vs_inputs = brw->vs.prog_data->inputs_read;
const unsigned char *ptr = NULL;
GLuint interleaved = 0, total_size = 0;
unsigned int min_index = brw->vb.min_index;
/* Accumulate the list of enabled arrays. */
brw->vb.nr_enabled = 0;
while (vs_inputs) {
- GLuint i = ffs(vs_inputs) - 1;
+ GLuint i = ffsll(vs_inputs) - 1;
struct brw_vertex_element *input = &brw->vb.inputs[i];
- vs_inputs &= ~(1 << i);
+ vs_inputs &= ~BITFIELD64_BIT(i);
if (input->glarray->Size && get_size(input->glarray->Type))
brw->vb.enabled[brw->vb.nr_enabled++] = input;
}
if (brw->vb.nr_buffers)
goto prepare;
- /* XXX: In the rare cases where this happens we fallback all
- * the way to software rasterization, although a tnl fallback
- * would be sufficient. I don't know of *any* real world
- * cases with > 17 vertex attributes enabled, so it probably
- * isn't an issue at this point.
- */
- if (brw->vb.nr_enabled >= BRW_VEP_MAX) {
- intel->Fallback = true; /* boolean, not bitfield */
- return;
- }
-
for (i = j = 0; i < brw->vb.nr_enabled; i++) {
struct brw_vertex_element *input = brw->vb.enabled[i];
const struct gl_client_array *glarray = input->glarray;
{
struct gl_context *ctx = &brw->intel.ctx;
struct intel_context *intel = intel_context(ctx);
- GLuint i;
+ GLuint i, nr_elements;
brw_prepare_vertices(brw);
*/
if (brw->vb.nr_buffers) {
+ if (intel->gen >= 6) {
+ assert(brw->vb.nr_buffers <= 33);
+ } else {
+ assert(brw->vb.nr_buffers <= 17);
+ }
+
BEGIN_BATCH(1 + 4*brw->vb.nr_buffers);
OUT_BATCH((_3DSTATE_VERTEX_BUFFERS << 16) | (4*brw->vb.nr_buffers - 1));
for (i = 0; i < brw->vb.nr_buffers; i++) {
ADVANCE_BATCH();
}
- BEGIN_BATCH(1 + brw->vb.nr_enabled * 2);
- OUT_BATCH((_3DSTATE_VERTEX_ELEMENTS << 16) | (2*brw->vb.nr_enabled - 1));
+ nr_elements = brw->vb.nr_enabled + brw->vs.prog_data->uses_vertexid;
+
+ /* The hardware allows one more VERTEX_ELEMENTS than VERTEX_BUFFERS, presumably
+ * for VertexID/InstanceID.
+ */
+ if (intel->gen >= 6) {
+ assert(nr_elements <= 34);
+ } else {
+ assert(nr_elements <= 18);
+ }
+
+ BEGIN_BATCH(1 + nr_elements * 2);
+ OUT_BATCH((_3DSTATE_VERTEX_ELEMENTS << 16) | (2 * nr_elements - 1));
for (i = 0; i < brw->vb.nr_enabled; i++) {
struct brw_vertex_element *input = brw->vb.enabled[i];
uint32_t format = get_surface_type(input->glarray->Type,
input->glarray->Size,
input->glarray->Format,
- input->glarray->Normalized);
+ input->glarray->Normalized,
+ input->glarray->Integer);
uint32_t comp0 = BRW_VE1_COMPONENT_STORE_SRC;
uint32_t comp1 = BRW_VE1_COMPONENT_STORE_SRC;
uint32_t comp2 = BRW_VE1_COMPONENT_STORE_SRC;
case 0: comp0 = BRW_VE1_COMPONENT_STORE_0;
case 1: comp1 = BRW_VE1_COMPONENT_STORE_0;
case 2: comp2 = BRW_VE1_COMPONENT_STORE_0;
- case 3: comp3 = BRW_VE1_COMPONENT_STORE_1_FLT;
+ case 3: comp3 = input->glarray->Integer ? BRW_VE1_COMPONENT_STORE_1_INT
+ : BRW_VE1_COMPONENT_STORE_1_FLT;
break;
}
(comp3 << BRW_VE1_COMPONENT_3_SHIFT) |
((i * 4) << BRW_VE1_DST_OFFSET_SHIFT));
}
+
+ if (brw->vs.prog_data->uses_vertexid) {
+ uint32_t dw0 = 0, dw1 = 0;
+
+ dw1 = ((BRW_VE1_COMPONENT_STORE_VID << BRW_VE1_COMPONENT_0_SHIFT) |
+ (BRW_VE1_COMPONENT_STORE_IID << BRW_VE1_COMPONENT_1_SHIFT) |
+ (BRW_VE1_COMPONENT_STORE_PID << BRW_VE1_COMPONENT_2_SHIFT) |
+ (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_3_SHIFT));
+
+ if (intel->gen >= 6) {
+ dw0 |= GEN6_VE0_VALID;
+ } else {
+ dw0 |= BRW_VE0_VALID;
+ dw1 |= (i * 4) << BRW_VE1_DST_OFFSET_SHIFT;
+ }
+
+ /* Note that for gl_VertexID, gl_InstanceID, and gl_PrimitiveID values,
+ * the format is ignored and the value is always int.
+ */
+
+ OUT_BATCH(dw0);
+ OUT_BATCH(dw1);
+ }
+
CACHED_BATCH();
}