#include "intel_batchbuffer.h"
#include "intel_buffer_objects.h"
-static GLuint double_types[5] = {
+static const GLuint double_types[5] = {
0,
BRW_SURFACEFORMAT_R64_FLOAT,
BRW_SURFACEFORMAT_R64G64_FLOAT,
BRW_SURFACEFORMAT_R64G64B64A64_FLOAT
};
-static GLuint float_types[5] = {
+static const GLuint float_types[5] = {
0,
BRW_SURFACEFORMAT_R32_FLOAT,
BRW_SURFACEFORMAT_R32G32_FLOAT,
BRW_SURFACEFORMAT_R32G32B32A32_FLOAT
};
-static GLuint half_float_types[5] = {
+static const GLuint half_float_types[5] = {
0,
BRW_SURFACEFORMAT_R16_FLOAT,
BRW_SURFACEFORMAT_R16G16_FLOAT,
BRW_SURFACEFORMAT_R16G16B16A16_FLOAT
};
-static GLuint fixed_point_types[5] = {
+static const GLuint fixed_point_types[5] = {
0,
BRW_SURFACEFORMAT_R32_SFIXED,
BRW_SURFACEFORMAT_R32G32_SFIXED,
BRW_SURFACEFORMAT_R32G32B32A32_SFIXED,
};
-static GLuint uint_types_direct[5] = {
+static const GLuint uint_types_direct[5] = {
0,
BRW_SURFACEFORMAT_R32_UINT,
BRW_SURFACEFORMAT_R32G32_UINT,
BRW_SURFACEFORMAT_R32G32B32A32_UINT
};
-static GLuint uint_types_norm[5] = {
+static const GLuint uint_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R32_UNORM,
BRW_SURFACEFORMAT_R32G32_UNORM,
BRW_SURFACEFORMAT_R32G32B32A32_UNORM
};
-static GLuint uint_types_scale[5] = {
+static const GLuint uint_types_scale[5] = {
0,
BRW_SURFACEFORMAT_R32_USCALED,
BRW_SURFACEFORMAT_R32G32_USCALED,
BRW_SURFACEFORMAT_R32G32B32A32_USCALED
};
-static GLuint int_types_direct[5] = {
+static const GLuint int_types_direct[5] = {
0,
BRW_SURFACEFORMAT_R32_SINT,
BRW_SURFACEFORMAT_R32G32_SINT,
BRW_SURFACEFORMAT_R32G32B32A32_SINT
};
-static GLuint int_types_norm[5] = {
+static const GLuint int_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R32_SNORM,
BRW_SURFACEFORMAT_R32G32_SNORM,
BRW_SURFACEFORMAT_R32G32B32A32_SNORM
};
-static GLuint int_types_scale[5] = {
+static const GLuint int_types_scale[5] = {
0,
BRW_SURFACEFORMAT_R32_SSCALED,
BRW_SURFACEFORMAT_R32G32_SSCALED,
BRW_SURFACEFORMAT_R32G32B32A32_SSCALED
};
-static GLuint ushort_types_direct[5] = {
+static const GLuint ushort_types_direct[5] = {
0,
BRW_SURFACEFORMAT_R16_UINT,
BRW_SURFACEFORMAT_R16G16_UINT,
BRW_SURFACEFORMAT_R16G16B16A16_UINT
};
-static GLuint ushort_types_norm[5] = {
+static const GLuint ushort_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R16_UNORM,
BRW_SURFACEFORMAT_R16G16_UNORM,
BRW_SURFACEFORMAT_R16G16B16A16_UNORM
};
-static GLuint ushort_types_scale[5] = {
+static const GLuint ushort_types_scale[5] = {
0,
BRW_SURFACEFORMAT_R16_USCALED,
BRW_SURFACEFORMAT_R16G16_USCALED,
BRW_SURFACEFORMAT_R16G16B16A16_USCALED
};
-static GLuint short_types_direct[5] = {
+static const GLuint short_types_direct[5] = {
0,
BRW_SURFACEFORMAT_R16_SINT,
BRW_SURFACEFORMAT_R16G16_SINT,
BRW_SURFACEFORMAT_R16G16B16A16_SINT
};
-static GLuint short_types_norm[5] = {
+static const GLuint short_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R16_SNORM,
BRW_SURFACEFORMAT_R16G16_SNORM,
BRW_SURFACEFORMAT_R16G16B16A16_SNORM
};
-static GLuint short_types_scale[5] = {
+static const GLuint short_types_scale[5] = {
0,
BRW_SURFACEFORMAT_R16_SSCALED,
BRW_SURFACEFORMAT_R16G16_SSCALED,
BRW_SURFACEFORMAT_R16G16B16A16_SSCALED
};
-static GLuint ubyte_types_direct[5] = {
+static const GLuint ubyte_types_direct[5] = {
0,
BRW_SURFACEFORMAT_R8_UINT,
BRW_SURFACEFORMAT_R8G8_UINT,
BRW_SURFACEFORMAT_R8G8B8A8_UINT
};
-static GLuint ubyte_types_norm[5] = {
+static const GLuint ubyte_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R8_UNORM,
BRW_SURFACEFORMAT_R8G8_UNORM,
BRW_SURFACEFORMAT_R8G8B8A8_UNORM
};
-static GLuint ubyte_types_scale[5] = {
+static const GLuint ubyte_types_scale[5] = {
0,
BRW_SURFACEFORMAT_R8_USCALED,
BRW_SURFACEFORMAT_R8G8_USCALED,
BRW_SURFACEFORMAT_R8G8B8A8_USCALED
};
-static GLuint byte_types_direct[5] = {
+static const GLuint byte_types_direct[5] = {
0,
BRW_SURFACEFORMAT_R8_SINT,
BRW_SURFACEFORMAT_R8G8_SINT,
BRW_SURFACEFORMAT_R8G8B8A8_SINT
};
-static GLuint byte_types_norm[5] = {
+static const GLuint byte_types_norm[5] = {
0,
BRW_SURFACEFORMAT_R8_SNORM,
BRW_SURFACEFORMAT_R8G8_SNORM,
BRW_SURFACEFORMAT_R8G8B8A8_SNORM
};
-static GLuint byte_types_scale[5] = {
+static const GLuint byte_types_scale[5] = {
0,
BRW_SURFACEFORMAT_R8_SSCALED,
BRW_SURFACEFORMAT_R8G8_SSCALED,
if (unlikely(INTEL_DEBUG & DEBUG_VERTS))
fprintf(stderr, "type %s size %d normalized %d\n",
- _mesa_lookup_enum_by_nr(glarray->Type),
+ _mesa_enum_to_string(glarray->Type),
glarray->Size, glarray->Normalized);
if (glarray->Integer) {
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;
+ default: unreachable("not reached");
}
} else if (glarray->Type == GL_UNSIGNED_INT_10F_11F_11F_REV) {
return BRW_SURFACEFORMAT_R11G11B10_FLOAT;
: BRW_SURFACEFORMAT_R10G10B10A2_UNORM;
}
return BRW_SURFACEFORMAT_R10G10B10A2_UINT;
- default: assert(0); return 0;
+ default: unreachable("not reached");
}
}
else {
* INT32_MAX, which will be scaled down by 1/65536 by the VS.
*/
return int_types_scale[size];
- default: assert(0); return 0;
+ default: unreachable("not reached");
}
}
}
-unsigned
-brw_get_index_type(GLenum type)
-{
- switch (type) {
- case GL_UNSIGNED_BYTE: return BRW_INDEX_BYTE;
- case GL_UNSIGNED_SHORT: return BRW_INDEX_WORD;
- case GL_UNSIGNED_INT: return BRW_INDEX_DWORD;
- default: assert(0); return 0;
- }
-}
-
static void
copy_array_to_vbo_array(struct brw_context *brw,
struct brw_vertex_element *element,
const unsigned char *src = element->glarray->Ptr + min * src_stride;
int count = max - min + 1;
GLuint size = count * dst_stride;
+ uint8_t *dst = intel_upload_space(brw, size, dst_stride,
+ &buffer->bo, &buffer->offset);
if (dst_stride == src_stride) {
- intel_upload_data(brw, src, size, dst_stride,
- &buffer->bo, &buffer->offset);
+ memcpy(dst, src, size);
} else {
- char * const map = intel_upload_map(brw, size, dst_stride);
- char *dst = map;
-
while (count--) {
memcpy(dst, src, dst_stride);
src += src_stride;
dst += dst_stride;
}
- intel_upload_unmap(brw, map, size, dst_stride,
- &buffer->bo, &buffer->offset);
}
buffer->stride = dst_stride;
}
brw_prepare_vertices(struct brw_context *brw)
{
struct gl_context *ctx = &brw->ctx;
- /* CACHE_NEW_VS_PROG */
+ /* BRW_NEW_VS_PROG_DATA */
GLbitfield64 vs_inputs = brw->vs.prog_data->inputs_read;
const unsigned char *ptr = NULL;
GLuint interleaved = 0;
}
if (0)
- fprintf(stderr, "%s %d..%d\n", __FUNCTION__, min_index, max_index);
+ fprintf(stderr, "%s %d..%d\n", __func__, min_index, max_index);
/* Accumulate the list of enabled arrays. */
brw->vb.nr_enabled = 0;
/* This is a common place to reach if the user mistakenly supplies
* a pointer in place of a VBO offset. If we just let it go through,
* we may end up dereferencing a pointer beyond the bounds of the
- * GTT. We would hope that the VBO's max_index would save us, but
- * Mesa appears to hand us min/max values not clipped to the
- * array object's _MaxElement, and _MaxElement frequently appears
- * to be wrong anyway.
+ * GTT.
*
* The VBO spec allows application termination in this case, and it's
* probably a service to the poor programmer to do so rather than
brw->vb.nr_buffers = j;
}
-static void brw_emit_vertices(struct brw_context *brw)
+void
+brw_prepare_shader_draw_parameters(struct brw_context *brw)
+{
+ /* For non-indirect draws, upload gl_BaseVertex. */
+ if (brw->vs.prog_data->uses_vertexid && brw->draw.draw_params_bo == NULL) {
+ intel_upload_data(brw, &brw->draw.gl_basevertex, 4, 4,
+ &brw->draw.draw_params_bo,
+ &brw->draw.draw_params_offset);
+ }
+}
+
+/**
+ * Emit a VERTEX_BUFFER_STATE entry (part of 3DSTATE_VERTEX_BUFFERS).
+ */
+static uint32_t *
+emit_vertex_buffer_state(struct brw_context *brw,
+ unsigned buffer_nr,
+ drm_intel_bo *bo,
+ unsigned bo_ending_address,
+ unsigned bo_offset,
+ unsigned stride,
+ unsigned step_rate,
+ uint32_t *__map)
{
struct gl_context *ctx = &brw->ctx;
- GLuint i, nr_elements;
+ uint32_t dw0;
+
+ if (brw->gen >= 6) {
+ dw0 = (buffer_nr << GEN6_VB0_INDEX_SHIFT) |
+ (step_rate ? GEN6_VB0_ACCESS_INSTANCEDATA
+ : GEN6_VB0_ACCESS_VERTEXDATA);
+ } else {
+ dw0 = (buffer_nr << BRW_VB0_INDEX_SHIFT) |
+ (step_rate ? BRW_VB0_ACCESS_INSTANCEDATA
+ : BRW_VB0_ACCESS_VERTEXDATA);
+ }
+
+ if (brw->gen >= 7)
+ dw0 |= GEN7_VB0_ADDRESS_MODIFYENABLE;
+
+ if (brw->gen == 7)
+ dw0 |= GEN7_MOCS_L3 << 16;
+
+ WARN_ONCE(stride >= (brw->gen >= 5 ? 2048 : 2047),
+ "VBO stride %d too large, bad rendering may occur\n",
+ stride);
+ OUT_BATCH(dw0 | (stride << BRW_VB0_PITCH_SHIFT));
+ OUT_RELOC(bo, I915_GEM_DOMAIN_VERTEX, 0, bo_offset);
+ if (brw->gen >= 5) {
+ OUT_RELOC(bo, I915_GEM_DOMAIN_VERTEX, 0, bo_ending_address);
+ } else {
+ OUT_BATCH(0);
+ }
+ OUT_BATCH(step_rate);
+
+ return __map;
+}
+#define EMIT_VERTEX_BUFFER_STATE(...) __map = emit_vertex_buffer_state(__VA_ARGS__, __map)
+
+static void brw_emit_vertices(struct brw_context *brw)
+{
+ GLuint i;
brw_prepare_vertices(brw);
+ brw_prepare_shader_draw_parameters(brw);
brw_emit_query_begin(brw);
- nr_elements = brw->vb.nr_enabled + brw->vs.prog_data->uses_vertexid;
+ unsigned nr_elements = brw->vb.nr_enabled;
+ if (brw->vs.prog_data->uses_vertexid || brw->vs.prog_data->uses_instanceid)
+ ++nr_elements;
/* If the VS doesn't read any inputs (calculating vertex position from
* a state variable for some reason, for example), emit a single pad
/* Now emit VB and VEP state packets.
*/
- if (brw->vb.nr_buffers) {
+ unsigned nr_buffers =
+ brw->vb.nr_buffers + brw->vs.prog_data->uses_vertexid;
+
+ if (nr_buffers) {
if (brw->gen >= 6) {
- assert(brw->vb.nr_buffers <= 33);
+ assert(nr_buffers <= 33);
} else {
- assert(brw->vb.nr_buffers <= 17);
+ assert(nr_buffers <= 17);
}
- BEGIN_BATCH(1 + 4*brw->vb.nr_buffers);
- OUT_BATCH((_3DSTATE_VERTEX_BUFFERS << 16) | (4*brw->vb.nr_buffers - 1));
+ BEGIN_BATCH(1 + 4 * nr_buffers);
+ OUT_BATCH((_3DSTATE_VERTEX_BUFFERS << 16) | (4 * nr_buffers - 1));
for (i = 0; i < brw->vb.nr_buffers; i++) {
struct brw_vertex_buffer *buffer = &brw->vb.buffers[i];
- uint32_t dw0;
-
- if (brw->gen >= 6) {
- dw0 = buffer->step_rate
- ? GEN6_VB0_ACCESS_INSTANCEDATA
- : GEN6_VB0_ACCESS_VERTEXDATA;
- dw0 |= i << GEN6_VB0_INDEX_SHIFT;
- } else {
- dw0 = buffer->step_rate
- ? BRW_VB0_ACCESS_INSTANCEDATA
- : BRW_VB0_ACCESS_VERTEXDATA;
- dw0 |= i << BRW_VB0_INDEX_SHIFT;
- }
+ EMIT_VERTEX_BUFFER_STATE(brw, i, buffer->bo, buffer->bo->size - 1,
+ buffer->offset, buffer->stride,
+ buffer->step_rate);
+
+ }
- if (brw->gen >= 7)
- dw0 |= GEN7_VB0_ADDRESS_MODIFYENABLE;
-
- if (brw->gen == 7)
- dw0 |= GEN7_MOCS_L3 << 16;
-
- WARN_ONCE(buffer->stride >= (brw->gen >= 5 ? 2048 : 2047),
- "VBO stride %d too large, bad rendering may occur\n",
- buffer->stride);
- OUT_BATCH(dw0 | (buffer->stride << BRW_VB0_PITCH_SHIFT));
- OUT_RELOC(buffer->bo, I915_GEM_DOMAIN_VERTEX, 0, buffer->offset);
- if (brw->gen >= 5) {
- OUT_RELOC(buffer->bo, I915_GEM_DOMAIN_VERTEX, 0, buffer->bo->size - 1);
- } else
- OUT_BATCH(0);
- OUT_BATCH(buffer->step_rate);
+ if (brw->vs.prog_data->uses_vertexid) {
+ EMIT_VERTEX_BUFFER_STATE(brw, brw->vb.nr_buffers,
+ brw->draw.draw_params_bo,
+ brw->draw.draw_params_bo->size - 1,
+ brw->draw.draw_params_offset,
+ 0, /* stride */
+ 0); /* step rate */
}
ADVANCE_BATCH();
}
uint32_t comp2 = BRW_VE1_COMPONENT_STORE_SRC;
uint32_t comp3 = BRW_VE1_COMPONENT_STORE_SRC;
- /* The gen4 driver expects edgeflag to come in as a float, and passes
- * that float on to the tests in the clipper. Mesa's current vertex
- * attribute value for EdgeFlag is stored as a float, which works out.
- * glEdgeFlagPointer, on the other hand, gives us an unnormalized
- * integer ubyte. Just rewrite that to convert to a float.
- */
- if (input->attrib == VERT_ATTRIB_EDGEFLAG) {
+ if (input == &brw->vb.inputs[VERT_ATTRIB_EDGEFLAG]) {
/* Gen6+ passes edgeflag as sideband along with the vertex, instead
* of in the VUE. We have to upload it sideband as the last vertex
* element according to the B-Spec.
gen6_edgeflag_input = input;
continue;
}
-
- if (format == BRW_SURFACEFORMAT_R8_UINT)
- format = BRW_SURFACEFORMAT_R8_SSCALED;
}
switch (input->glarray->Size) {
(BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_3_SHIFT));
}
- if (brw->vs.prog_data->uses_vertexid) {
+ if (brw->vs.prog_data->uses_vertexid || brw->vs.prog_data->uses_instanceid) {
uint32_t dw0 = 0, dw1 = 0;
+ uint32_t comp0 = BRW_VE1_COMPONENT_STORE_0;
+ uint32_t comp1 = BRW_VE1_COMPONENT_STORE_0;
+ uint32_t comp2 = BRW_VE1_COMPONENT_STORE_0;
+ uint32_t comp3 = BRW_VE1_COMPONENT_STORE_0;
+
+ if (brw->vs.prog_data->uses_vertexid) {
+ comp0 = BRW_VE1_COMPONENT_STORE_SRC;
+ comp2 = BRW_VE1_COMPONENT_STORE_VID;
+ }
- 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_0 << BRW_VE1_COMPONENT_2_SHIFT) |
- (BRW_VE1_COMPONENT_STORE_0 << BRW_VE1_COMPONENT_3_SHIFT));
+ if (brw->vs.prog_data->uses_instanceid) {
+ comp3 = BRW_VE1_COMPONENT_STORE_IID;
+ }
+
+ dw1 = (comp0 << BRW_VE1_COMPONENT_0_SHIFT) |
+ (comp1 << BRW_VE1_COMPONENT_1_SHIFT) |
+ (comp2 << BRW_VE1_COMPONENT_2_SHIFT) |
+ (comp3 << BRW_VE1_COMPONENT_3_SHIFT);
if (brw->gen >= 6) {
- dw0 |= GEN6_VE0_VALID;
+ dw0 |= GEN6_VE0_VALID |
+ brw->vb.nr_buffers << GEN6_VE0_INDEX_SHIFT |
+ BRW_SURFACEFORMAT_R32_UINT << BRW_VE0_FORMAT_SHIFT;
} else {
- dw0 |= BRW_VE0_VALID;
+ dw0 |= BRW_VE0_VALID |
+ brw->vb.nr_buffers << BRW_VE0_INDEX_SHIFT |
+ BRW_SURFACEFORMAT_R32_UINT << BRW_VE0_FORMAT_SHIFT;
dw1 |= (i * 4) << BRW_VE1_DST_OFFSET_SHIFT;
}
const struct brw_tracked_state brw_vertices = {
.dirty = {
.mesa = _NEW_POLYGON,
- .brw = BRW_NEW_BATCH | BRW_NEW_VERTICES,
- .cache = CACHE_NEW_VS_PROG,
+ .brw = BRW_NEW_BATCH |
+ BRW_NEW_VERTICES |
+ BRW_NEW_VS_PROG_DATA,
},
.emit = brw_emit_vertices,
};
struct gl_context *ctx = &brw->ctx;
const struct _mesa_index_buffer *index_buffer = brw->ib.ib;
GLuint ib_size;
- drm_intel_bo *bo = NULL;
+ drm_intel_bo *old_bo = brw->ib.bo;
struct gl_buffer_object *bufferobj;
GLuint offset;
GLuint ib_type_size;
/* Get new bufferobj, offset:
*/
intel_upload_data(brw, index_buffer->ptr, ib_size, ib_type_size,
- &bo, &offset);
+ &brw->ib.bo, &offset);
} else {
offset = (GLuint) (unsigned long) index_buffer->ptr;
/* If the index buffer isn't aligned to its element size, we have to
* rebase it into a temporary.
*/
- if ((ib_type_size - 1) & offset) {
- perf_debug("copying index buffer to a temporary to work around "
- "misaligned offset %d\n", offset);
-
- GLubyte *map = ctx->Driver.MapBufferRange(ctx,
- offset,
- ib_size,
- GL_MAP_READ_BIT,
- bufferobj,
- MAP_INTERNAL);
-
- intel_upload_data(brw, map, ib_size, ib_type_size, &bo, &offset);
-
- ctx->Driver.UnmapBuffer(ctx, bufferobj, MAP_INTERNAL);
- } else {
- bo = intel_bufferobj_buffer(brw, intel_buffer_object(bufferobj),
- offset, ib_size);
- drm_intel_bo_reference(bo);
- }
+ if ((ib_type_size - 1) & offset) {
+ perf_debug("copying index buffer to a temporary to work around "
+ "misaligned offset %d\n", offset);
+
+ GLubyte *map = ctx->Driver.MapBufferRange(ctx,
+ offset,
+ ib_size,
+ GL_MAP_READ_BIT,
+ bufferobj,
+ MAP_INTERNAL);
+
+ intel_upload_data(brw, map, ib_size, ib_type_size,
+ &brw->ib.bo, &offset);
+
+ ctx->Driver.UnmapBuffer(ctx, bufferobj, MAP_INTERNAL);
+ } else {
+ drm_intel_bo *bo =
+ intel_bufferobj_buffer(brw, intel_buffer_object(bufferobj),
+ offset, ib_size);
+ if (bo != brw->ib.bo) {
+ drm_intel_bo_unreference(brw->ib.bo);
+ brw->ib.bo = bo;
+ drm_intel_bo_reference(bo);
+ }
+ }
}
/* Use 3DPRIMITIVE's start_vertex_offset to avoid re-uploading
*/
brw->ib.start_vertex_offset = offset / ib_type_size;
- if (brw->ib.bo != bo) {
- drm_intel_bo_unreference(brw->ib.bo);
- brw->ib.bo = bo;
-
- brw->state.dirty.brw |= BRW_NEW_INDEX_BUFFER;
- } else {
- drm_intel_bo_unreference(bo);
- }
+ if (brw->ib.bo != old_bo)
+ brw->ctx.NewDriverState |= BRW_NEW_INDEX_BUFFER;
if (index_buffer->type != brw->ib.type) {
brw->ib.type = index_buffer->type;
- brw->state.dirty.brw |= BRW_NEW_INDEX_BUFFER;
+ brw->ctx.NewDriverState |= BRW_NEW_INDEX_BUFFER;
}
}
.dirty = {
.mesa = 0,
.brw = BRW_NEW_INDICES,
- .cache = 0,
},
.emit = brw_upload_indices,
};
BEGIN_BATCH(3);
OUT_BATCH(CMD_INDEX_BUFFER << 16 |
cut_index_setting |
- brw_get_index_type(index_buffer->type) << 8 |
+ brw_get_index_type(index_buffer->type) |
1);
OUT_RELOC(brw->ib.bo,
I915_GEM_DOMAIN_VERTEX, 0,
const struct brw_tracked_state brw_index_buffer = {
.dirty = {
.mesa = 0,
- .brw = BRW_NEW_BATCH | BRW_NEW_INDEX_BUFFER,
- .cache = 0,
+ .brw = BRW_NEW_BATCH |
+ BRW_NEW_INDEX_BUFFER,
},
.emit = brw_emit_index_buffer,
};