X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fmesa%2Fstate_tracker%2Fst_draw.c;h=3db5749725e917a93e19d4ef545ac1820ad30e48;hb=f3b62d4c74414c002d25e65decb9d0bda41ab5b6;hp=954f15a3437569948497cb21dc6058690bd8968f;hpb=efa1fac2158c9146b87f0d4340a864661721de21;p=mesa.git diff --git a/src/mesa/state_tracker/st_draw.c b/src/mesa/state_tracker/st_draw.c index 954f15a3437..3db5749725e 100644 --- a/src/mesa/state_tracker/st_draw.c +++ b/src/mesa/state_tracker/st_draw.c @@ -1,8 +1,8 @@ /************************************************************************** - * - * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. + * + * Copyright 2007 VMware, Inc. * All Rights Reserved. - * + * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the * "Software"), to deal in the Software without restriction, including @@ -10,19 +10,19 @@ * distribute, sub license, and/or sell copies of the Software, and to * permit persons to whom the Software is furnished to do so, subject to * the following conditions: - * + * * The above copyright notice and this permission notice (including the * next paragraph) shall be included in all copies or substantial portions * of the Software. - * + * * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. - * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR + * IN NO EVENT SHALL VMWARE AND/OR ITS SUPPLIERS BE LIABLE FOR * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. - * + * **************************************************************************/ /* @@ -31,12 +31,8 @@ * this function whether the user called glBegin/End, glDrawArrays, * glDrawElements, glEvalMesh, or glCalList, etc. * - * We basically convert the VBO's vertex attribute/array information into - * Gallium vertex state, bind the vertex buffer objects and call - * pipe->draw_vbo(). - * * Authors: - * Keith Whitwell + * Keith Whitwell */ @@ -44,14 +40,18 @@ #include "main/image.h" #include "main/bufferobj.h" #include "main/macros.h" -#include "main/mfeatures.h" +#include "main/varray.h" + +#include "compiler/glsl/ir_uniform.h" #include "vbo/vbo.h" #include "st_context.h" #include "st_atom.h" +#include "st_cb_bitmap.h" #include "st_cb_bufferobjects.h" #include "st_cb_xformfb.h" +#include "st_debug.h" #include "st_draw.h" #include "st_program.h" @@ -60,213 +60,11 @@ #include "util/u_inlines.h" #include "util/u_format.h" #include "util/u_prim.h" -#include "util/u_draw_quad.h" +#include "util/u_draw.h" +#include "util/u_upload_mgr.h" #include "draw/draw_context.h" #include "cso_cache/cso_context.h" -#include "../glsl/ir_uniform.h" - - -static GLuint double_types[4] = { - PIPE_FORMAT_R64_FLOAT, - PIPE_FORMAT_R64G64_FLOAT, - PIPE_FORMAT_R64G64B64_FLOAT, - PIPE_FORMAT_R64G64B64A64_FLOAT -}; - -static GLuint float_types[4] = { - PIPE_FORMAT_R32_FLOAT, - PIPE_FORMAT_R32G32_FLOAT, - PIPE_FORMAT_R32G32B32_FLOAT, - PIPE_FORMAT_R32G32B32A32_FLOAT -}; - -static GLuint half_float_types[4] = { - PIPE_FORMAT_R16_FLOAT, - PIPE_FORMAT_R16G16_FLOAT, - PIPE_FORMAT_R16G16B16_FLOAT, - PIPE_FORMAT_R16G16B16A16_FLOAT -}; - -static GLuint uint_types_norm[4] = { - PIPE_FORMAT_R32_UNORM, - PIPE_FORMAT_R32G32_UNORM, - PIPE_FORMAT_R32G32B32_UNORM, - PIPE_FORMAT_R32G32B32A32_UNORM -}; - -static GLuint uint_types_scale[4] = { - PIPE_FORMAT_R32_USCALED, - PIPE_FORMAT_R32G32_USCALED, - PIPE_FORMAT_R32G32B32_USCALED, - PIPE_FORMAT_R32G32B32A32_USCALED -}; - -static GLuint int_types_norm[4] = { - PIPE_FORMAT_R32_SNORM, - PIPE_FORMAT_R32G32_SNORM, - PIPE_FORMAT_R32G32B32_SNORM, - PIPE_FORMAT_R32G32B32A32_SNORM -}; - -static GLuint int_types_scale[4] = { - PIPE_FORMAT_R32_SSCALED, - PIPE_FORMAT_R32G32_SSCALED, - PIPE_FORMAT_R32G32B32_SSCALED, - PIPE_FORMAT_R32G32B32A32_SSCALED -}; - -static GLuint ushort_types_norm[4] = { - PIPE_FORMAT_R16_UNORM, - PIPE_FORMAT_R16G16_UNORM, - PIPE_FORMAT_R16G16B16_UNORM, - PIPE_FORMAT_R16G16B16A16_UNORM -}; - -static GLuint ushort_types_scale[4] = { - PIPE_FORMAT_R16_USCALED, - PIPE_FORMAT_R16G16_USCALED, - PIPE_FORMAT_R16G16B16_USCALED, - PIPE_FORMAT_R16G16B16A16_USCALED -}; - -static GLuint short_types_norm[4] = { - PIPE_FORMAT_R16_SNORM, - PIPE_FORMAT_R16G16_SNORM, - PIPE_FORMAT_R16G16B16_SNORM, - PIPE_FORMAT_R16G16B16A16_SNORM -}; - -static GLuint short_types_scale[4] = { - PIPE_FORMAT_R16_SSCALED, - PIPE_FORMAT_R16G16_SSCALED, - PIPE_FORMAT_R16G16B16_SSCALED, - PIPE_FORMAT_R16G16B16A16_SSCALED -}; - -static GLuint ubyte_types_norm[4] = { - PIPE_FORMAT_R8_UNORM, - PIPE_FORMAT_R8G8_UNORM, - PIPE_FORMAT_R8G8B8_UNORM, - PIPE_FORMAT_R8G8B8A8_UNORM -}; - -static GLuint ubyte_types_scale[4] = { - PIPE_FORMAT_R8_USCALED, - PIPE_FORMAT_R8G8_USCALED, - PIPE_FORMAT_R8G8B8_USCALED, - PIPE_FORMAT_R8G8B8A8_USCALED -}; - -static GLuint byte_types_norm[4] = { - PIPE_FORMAT_R8_SNORM, - PIPE_FORMAT_R8G8_SNORM, - PIPE_FORMAT_R8G8B8_SNORM, - PIPE_FORMAT_R8G8B8A8_SNORM -}; - -static GLuint byte_types_scale[4] = { - PIPE_FORMAT_R8_SSCALED, - PIPE_FORMAT_R8G8_SSCALED, - PIPE_FORMAT_R8G8B8_SSCALED, - PIPE_FORMAT_R8G8B8A8_SSCALED -}; - -static GLuint fixed_types[4] = { - PIPE_FORMAT_R32_FIXED, - PIPE_FORMAT_R32G32_FIXED, - PIPE_FORMAT_R32G32B32_FIXED, - PIPE_FORMAT_R32G32B32A32_FIXED -}; - - - -/** - * Return a PIPE_FORMAT_x for the given GL datatype and size. - */ -enum pipe_format -st_pipe_vertex_format(GLenum type, GLuint size, GLenum format, - GLboolean normalized) -{ - assert((type >= GL_BYTE && type <= GL_DOUBLE) || - type == GL_FIXED || type == GL_HALF_FLOAT || - type == GL_INT_2_10_10_10_REV || - type == GL_UNSIGNED_INT_2_10_10_10_REV); - assert(size >= 1); - assert(size <= 4); - assert(format == GL_RGBA || format == GL_BGRA); - - if (type == GL_INT_2_10_10_10_REV || - type == GL_UNSIGNED_INT_2_10_10_10_REV) { - assert(size == 4); - - if (format == GL_BGRA) { - if (type == GL_INT_2_10_10_10_REV) { - if (normalized) - return PIPE_FORMAT_B10G10R10A2_SNORM; - else - return PIPE_FORMAT_B10G10R10A2_SSCALED; - } else { - if (normalized) - return PIPE_FORMAT_B10G10R10A2_UNORM; - else - return PIPE_FORMAT_B10G10R10A2_USCALED; - } - } else { - if (type == GL_INT_2_10_10_10_REV) { - if (normalized) - return PIPE_FORMAT_R10G10B10A2_SNORM; - else - return PIPE_FORMAT_R10G10B10A2_SSCALED; - } else { - if (normalized) - return PIPE_FORMAT_R10G10B10A2_UNORM; - else - return PIPE_FORMAT_R10G10B10A2_USCALED; - } - } - } - - if (format == GL_BGRA) { - /* this is an odd-ball case */ - assert(type == GL_UNSIGNED_BYTE); - assert(normalized); - return PIPE_FORMAT_B8G8R8A8_UNORM; - } - - if (normalized) { - switch (type) { - case GL_DOUBLE: return double_types[size-1]; - case GL_FLOAT: return float_types[size-1]; - case GL_HALF_FLOAT: return half_float_types[size-1]; - case GL_INT: return int_types_norm[size-1]; - case GL_SHORT: return short_types_norm[size-1]; - case GL_BYTE: return byte_types_norm[size-1]; - case GL_UNSIGNED_INT: return uint_types_norm[size-1]; - case GL_UNSIGNED_SHORT: return ushort_types_norm[size-1]; - case GL_UNSIGNED_BYTE: return ubyte_types_norm[size-1]; - case GL_FIXED: return fixed_types[size-1]; - default: assert(0); return 0; - } - } - else { - switch (type) { - case GL_DOUBLE: return double_types[size-1]; - case GL_FLOAT: return float_types[size-1]; - case GL_HALF_FLOAT: return half_float_types[size-1]; - case GL_INT: return int_types_scale[size-1]; - case GL_SHORT: return short_types_scale[size-1]; - case GL_BYTE: return byte_types_scale[size-1]; - case GL_UNSIGNED_INT: return uint_types_scale[size-1]; - case GL_UNSIGNED_SHORT: return ushort_types_scale[size-1]; - case GL_UNSIGNED_BYTE: return ubyte_types_scale[size-1]; - case GL_FIXED: return fixed_types[size-1]; - default: assert(0); return 0; - } - } - return PIPE_FORMAT_NONE; /* silence compiler warning */ -} - /** * This is very similar to vbo_all_varyings_in_vbos() but we are @@ -276,7 +74,7 @@ static GLboolean all_varyings_in_vbos(const struct gl_client_array *arrays[]) { GLuint i; - + for (i = 0; i < VERT_ATTRIB_MAX; i++) if (arrays[i]->StrideB && !arrays[i]->InstanceDivisor && @@ -288,519 +86,43 @@ all_varyings_in_vbos(const struct gl_client_array *arrays[]) /** - * Examine the active arrays to determine if we have interleaved - * vertex arrays all living in one VBO, or all living in user space. - */ -static GLboolean -is_interleaved_arrays(const struct st_vertex_program *vp, - const struct st_vp_variant *vpv, - const struct gl_client_array **arrays) -{ - GLuint attr; - const struct gl_buffer_object *firstBufObj = NULL; - GLint firstStride = -1; - const GLubyte *firstPtr = NULL; - GLboolean userSpaceBuffer = GL_FALSE; - - for (attr = 0; attr < vpv->num_inputs; attr++) { - const GLuint mesaAttr = vp->index_to_input[attr]; - const struct gl_client_array *array = arrays[mesaAttr]; - const struct gl_buffer_object *bufObj = array->BufferObj; - const GLsizei stride = array->StrideB; /* in bytes */ - - if (attr == 0) { - /* save info about the first array */ - firstStride = stride; - firstPtr = array->Ptr; - firstBufObj = bufObj; - userSpaceBuffer = !bufObj || !bufObj->Name; - } - else { - /* check if other arrays interleave with the first, in same buffer */ - if (stride != firstStride) - return GL_FALSE; /* strides don't match */ - - if (bufObj != firstBufObj) - return GL_FALSE; /* arrays in different VBOs */ - - if (abs(array->Ptr - firstPtr) > firstStride) - return GL_FALSE; /* arrays start too far apart */ - - if ((!_mesa_is_bufferobj(bufObj)) != userSpaceBuffer) - return GL_FALSE; /* mix of VBO and user-space arrays */ - } - } - - return GL_TRUE; -} - - -/** - * Set up for drawing interleaved arrays that all live in one VBO - * or all live in user space. - * \param vbuffer returns vertex buffer info - * \param velements returns vertex element info - * \return GL_TRUE for success, GL_FALSE otherwise (probably out of memory) - */ -static GLboolean -setup_interleaved_attribs(struct gl_context *ctx, - const struct st_vertex_program *vp, - const struct st_vp_variant *vpv, - const struct gl_client_array **arrays, - struct pipe_vertex_buffer *vbuffer, - struct pipe_vertex_element velements[], - unsigned max_index, - unsigned num_instances) -{ - struct st_context *st = st_context(ctx); - struct pipe_context *pipe = st->pipe; - GLuint attr; - const GLubyte *low_addr = NULL; - GLboolean usingVBO; /* all arrays in a VBO? */ - struct gl_buffer_object *bufobj; - GLuint user_buffer_size = 0; - GLuint vertex_size = 0; /* bytes per vertex, in bytes */ - GLsizei stride; - - /* Find the lowest address of the arrays we're drawing, - * Init bufobj and stride. - */ - if (vpv->num_inputs) { - const GLuint mesaAttr0 = vp->index_to_input[0]; - const struct gl_client_array *array = arrays[mesaAttr0]; - - /* Since we're doing interleaved arrays, we know there'll be at most - * one buffer object and the stride will be the same for all arrays. - * Grab them now. - */ - bufobj = array->BufferObj; - stride = array->StrideB; - - low_addr = arrays[vp->index_to_input[0]]->Ptr; - - for (attr = 1; attr < vpv->num_inputs; attr++) { - const GLubyte *start = arrays[vp->index_to_input[attr]]->Ptr; - low_addr = MIN2(low_addr, start); - } - } - else { - /* not sure we'll ever have zero inputs, but play it safe */ - bufobj = NULL; - stride = 0; - low_addr = 0; - } - - /* are the arrays in user space? */ - usingVBO = _mesa_is_bufferobj(bufobj); - - for (attr = 0; attr < vpv->num_inputs; attr++) { - const GLuint mesaAttr = vp->index_to_input[attr]; - const struct gl_client_array *array = arrays[mesaAttr]; - unsigned src_offset = (unsigned) (array->Ptr - low_addr); - GLuint element_size = array->_ElementSize; - - assert(element_size == array->Size * _mesa_sizeof_type(array->Type)); - - velements[attr].src_offset = src_offset; - velements[attr].instance_divisor = array->InstanceDivisor; - velements[attr].vertex_buffer_index = 0; - velements[attr].src_format = st_pipe_vertex_format(array->Type, - array->Size, - array->Format, - array->Normalized); - assert(velements[attr].src_format); - - if (!usingVBO) { - /* how many bytes referenced by this attribute array? */ - uint divisor = array->InstanceDivisor; - uint last_index = divisor ? num_instances / divisor : max_index; - uint bytes = src_offset + stride * last_index + element_size; - - user_buffer_size = MAX2(user_buffer_size, bytes); - - /* update vertex size */ - vertex_size = MAX2(vertex_size, src_offset + element_size); - } - } - - /* - * Return the vbuffer info and setup user-space attrib info, if needed. - */ - if (vpv->num_inputs == 0) { - /* just defensive coding here */ - vbuffer->buffer = NULL; - vbuffer->buffer_offset = 0; - vbuffer->stride = 0; - st->num_user_attribs = 0; - } - else if (usingVBO) { - /* all interleaved arrays in a VBO */ - struct st_buffer_object *stobj = st_buffer_object(bufobj); - - if (!stobj || !stobj->buffer) { - /* probably out of memory (or zero-sized buffer) */ - return GL_FALSE; - } - - vbuffer->buffer = NULL; - pipe_resource_reference(&vbuffer->buffer, stobj->buffer); - vbuffer->buffer_offset = pointer_to_offset(low_addr); - vbuffer->stride = stride; - st->num_user_attribs = 0; - } - else { - /* all interleaved arrays in user memory */ - vbuffer->buffer = pipe_user_buffer_create(pipe->screen, - (void*) low_addr, - user_buffer_size, - PIPE_BIND_VERTEX_BUFFER); - vbuffer->buffer_offset = 0; - vbuffer->stride = stride; - - /* Track user vertex buffers. */ - pipe_resource_reference(&st->user_attrib[0].buffer, vbuffer->buffer); - st->user_attrib[0].element_size = vertex_size; - st->user_attrib[0].stride = stride; - st->num_user_attribs = 1; - } - - return GL_TRUE; -} - - -/** - * Set up a separate pipe_vertex_buffer and pipe_vertex_element for each - * vertex attribute. - * \param vbuffer returns vertex buffer info - * \param velements returns vertex element info - * \return GL_TRUE for success, GL_FALSE otherwise (probably out of memory) + * Basically, translate Mesa's index buffer information into + * a pipe_index_buffer object. + * \return TRUE or FALSE for success/failure */ -static GLboolean -setup_non_interleaved_attribs(struct gl_context *ctx, - const struct st_vertex_program *vp, - const struct st_vp_variant *vpv, - const struct gl_client_array **arrays, - struct pipe_vertex_buffer vbuffer[], - struct pipe_vertex_element velements[], - unsigned max_index, - unsigned num_instances) -{ - struct st_context *st = st_context(ctx); - struct pipe_context *pipe = st->pipe; - GLuint attr; - - for (attr = 0; attr < vpv->num_inputs; attr++) { - const GLuint mesaAttr = vp->index_to_input[attr]; - const struct gl_client_array *array = arrays[mesaAttr]; - struct gl_buffer_object *bufobj = array->BufferObj; - GLuint element_size = array->_ElementSize; - GLsizei stride = array->StrideB; - - assert(element_size == array->Size * _mesa_sizeof_type(array->Type)); - - if (_mesa_is_bufferobj(bufobj)) { - /* Attribute data is in a VBO. - * Recall that for VBOs, the gl_client_array->Ptr field is - * really an offset from the start of the VBO, not a pointer. - */ - struct st_buffer_object *stobj = st_buffer_object(bufobj); - - if (!stobj || !stobj->buffer) { - /* probably out of memory (or zero-sized buffer) */ - return GL_FALSE; - } - - vbuffer[attr].buffer = NULL; - pipe_resource_reference(&vbuffer[attr].buffer, stobj->buffer); - vbuffer[attr].buffer_offset = pointer_to_offset(array->Ptr); - } - else { - /* wrap user data */ - uint bytes; - void *ptr; - - if (array->Ptr) { - uint divisor = array->InstanceDivisor; - uint last_index = divisor ? num_instances / divisor : max_index; - - bytes = stride * last_index + element_size; - - ptr = (void *) array->Ptr; - } - else { - /* no array, use ctx->Current.Attrib[] value */ - bytes = element_size = sizeof(ctx->Current.Attrib[0]); - ptr = (void *) ctx->Current.Attrib[mesaAttr]; - stride = 0; - } - - assert(ptr); - assert(bytes); - - vbuffer[attr].buffer = - pipe_user_buffer_create(pipe->screen, ptr, bytes, - PIPE_BIND_VERTEX_BUFFER); - - vbuffer[attr].buffer_offset = 0; - - /* Track user vertex buffers. */ - pipe_resource_reference(&st->user_attrib[attr].buffer, vbuffer[attr].buffer); - st->user_attrib[attr].element_size = element_size; - st->user_attrib[attr].stride = stride; - st->num_user_attribs = MAX2(st->num_user_attribs, attr + 1); - - if (!vbuffer[attr].buffer) { - /* probably ran out of memory */ - return GL_FALSE; - } - } - - /* common-case setup */ - vbuffer[attr].stride = stride; /* in bytes */ - - velements[attr].src_offset = 0; - velements[attr].instance_divisor = array->InstanceDivisor; - velements[attr].vertex_buffer_index = attr; - velements[attr].src_format = st_pipe_vertex_format(array->Type, - array->Size, - array->Format, - array->Normalized); - assert(velements[attr].src_format); - } - - return GL_TRUE; -} - - -static void -setup_index_buffer(struct gl_context *ctx, +static boolean +setup_index_buffer(struct st_context *st, const struct _mesa_index_buffer *ib, struct pipe_index_buffer *ibuffer) { - struct st_context *st = st_context(ctx); - struct pipe_context *pipe = st->pipe; - - memset(ibuffer, 0, sizeof(*ibuffer)); - if (ib) { - struct gl_buffer_object *bufobj = ib->obj; + struct gl_buffer_object *bufobj = ib->obj; - ibuffer->index_size = vbo_sizeof_ib_type(ib->type); + ibuffer->index_size = vbo_sizeof_ib_type(ib->type); - /* get/create the index buffer object */ - if (_mesa_is_bufferobj(bufobj)) { - /* elements/indexes are in a real VBO */ - struct st_buffer_object *stobj = st_buffer_object(bufobj); - pipe_resource_reference(&ibuffer->buffer, stobj->buffer); - ibuffer->offset = pointer_to_offset(ib->ptr); - } - else { - /* element/indicies are in user space memory */ - ibuffer->buffer = - pipe_user_buffer_create(pipe->screen, (void *) ib->ptr, - ib->count * ibuffer->index_size, - PIPE_BIND_INDEX_BUFFER); - } + /* get/create the index buffer object */ + if (_mesa_is_bufferobj(bufobj)) { + /* indices are in a real VBO */ + ibuffer->buffer = st_buffer_object(bufobj)->buffer; + ibuffer->offset = pointer_to_offset(ib->ptr); } -} - - -/** - * Prior to drawing, check that any uniforms referenced by the - * current shader have been set. If a uniform has not been set, - * issue a warning. - */ -static void -check_uniforms(struct gl_context *ctx) -{ - struct gl_shader_program *shProg[3] = { - ctx->Shader.CurrentVertexProgram, - ctx->Shader.CurrentGeometryProgram, - ctx->Shader.CurrentFragmentProgram, - }; - unsigned j; - - for (j = 0; j < 3; j++) { - unsigned i; - - if (shProg[j] == NULL || !shProg[j]->LinkStatus) - continue; - - for (i = 0; i < shProg[j]->NumUserUniformStorage; i++) { - const struct gl_uniform_storage *u = &shProg[j]->UniformStorage[i]; - if (!u->initialized) { - _mesa_warning(ctx, - "Using shader with uninitialized uniform: %s", - u->name); - } + else if (st->indexbuf_uploader) { + /* upload indexes from user memory into a real buffer */ + u_upload_data(st->indexbuf_uploader, 0, + ib->count * ibuffer->index_size, 4, ib->ptr, + &ibuffer->offset, &ibuffer->buffer); + if (!ibuffer->buffer) { + /* out of memory */ + return FALSE; } - } -} - - -/* - * Notes on primitive restart: - * The code below is used when the gallium driver does not support primitive - * restart itself. We map the index buffer, find the restart indexes, unmap - * the index buffer then draw the sub-primitives delineated by the restarts. - * A couple possible optimizations: - * 1. Save the list of sub-primitive (start, count) values in a list attached - * to the index buffer for re-use in subsequent draws. The list would be - * invalidated when the contents of the buffer changed. - * 2. If drawing triangle strips or quad strips, create a new index buffer - * that uses duplicated vertices to render the disjoint strips as one - * long strip. We'd have to be careful to avoid using too much memory - * for this. - * Finally, some apps might perform better if they don't use primitive restart - * at all rather than this fallback path. Set MESA_EXTENSION_OVERRIDE to - * "-GL_NV_primitive_restart" to test that. - */ - - -struct sub_primitive -{ - unsigned start, count; -}; - - -/** - * Scan the elements array to find restart indexes. Return a list - * of primitive (start,count) pairs to indicate how to draw the sub- - * primitives delineated by the restart index. - */ -static struct sub_primitive * -find_sub_primitives(const void *elements, unsigned element_size, - unsigned start, unsigned end, unsigned restart_index, - unsigned *num_sub_prims) -{ - const unsigned max_prims = end - start; - struct sub_primitive *sub_prims; - unsigned i, cur_start, cur_count, num; - - sub_prims = (struct sub_primitive *) - malloc(max_prims * sizeof(struct sub_primitive)); - - if (!sub_prims) { - *num_sub_prims = 0; - return NULL; - } - - cur_start = start; - cur_count = 0; - num = 0; - -#define SCAN_ELEMENTS(TYPE) \ - for (i = start; i < end; i++) { \ - if (((const TYPE *) elements)[i] == restart_index) { \ - if (cur_count > 0) { \ - assert(num < max_prims); \ - sub_prims[num].start = cur_start; \ - sub_prims[num].count = cur_count; \ - num++; \ - } \ - cur_start = i + 1; \ - cur_count = 0; \ - } \ - else { \ - cur_count++; \ - } \ - } \ - if (cur_count > 0) { \ - assert(num < max_prims); \ - sub_prims[num].start = cur_start; \ - sub_prims[num].count = cur_count; \ - num++; \ - } - - switch (element_size) { - case 1: - SCAN_ELEMENTS(ubyte); - break; - case 2: - SCAN_ELEMENTS(ushort); - break; - case 4: - SCAN_ELEMENTS(uint); - break; - default: - assert(0 && "bad index_size in find_sub_primitives()"); - } - -#undef SCAN_ELEMENTS - - *num_sub_prims = num; - - return sub_prims; -} - - -/** - * For gallium drivers that don't support the primitive restart - * feature, handle it here by breaking up the indexed primitive into - * sub-primitives. - */ -static void -handle_fallback_primitive_restart(struct pipe_context *pipe, - const struct _mesa_index_buffer *ib, - struct pipe_index_buffer *ibuffer, - struct pipe_draw_info *orig_info) -{ - const unsigned start = orig_info->start; - const unsigned count = orig_info->count; - struct pipe_draw_info info = *orig_info; - struct pipe_transfer *transfer = NULL; - unsigned instance, i; - const void *ptr = NULL; - struct sub_primitive *sub_prims; - unsigned num_sub_prims; - - assert(info.indexed); - assert(ibuffer->buffer); - assert(ib); - - if (!ibuffer->buffer || !ib) - return; - - info.primitive_restart = FALSE; - info.instance_count = 1; - - if (_mesa_is_bufferobj(ib->obj)) { - ptr = pipe_buffer_map_range(pipe, ibuffer->buffer, - start * ibuffer->index_size, /* start */ - count * ibuffer->index_size, /* length */ - PIPE_TRANSFER_READ, &transfer); + u_upload_unmap(st->indexbuf_uploader); } else { - ptr = ib->ptr; + /* indices are in user space memory */ + ibuffer->user_buffer = ib->ptr; } - if (!ptr) - return; - - ptr = ADD_POINTERS(ptr, ibuffer->offset); - - sub_prims = find_sub_primitives(ptr, ibuffer->index_size, - 0, count, orig_info->restart_index, - &num_sub_prims); - - if (transfer) - pipe_buffer_unmap(pipe, transfer); - - /* Now draw the sub primitives. - * Need to loop over instances as well to preserve draw order. - */ - for (instance = 0; instance < orig_info->instance_count; instance++) { - info.start_instance = instance + orig_info->start_instance; - for (i = 0; i < num_sub_prims; i++) { - info.start = sub_prims[i].start; - info.count = sub_prims[i].count; - if (u_trim_pipe_prim(info.mode, &info.count)) { - pipe->draw_vbo(pipe, &info); - } - } - } - - if (sub_prims) - free(sub_prims); + cso_set_index_buffer(st->cso_context, ibuffer); + return TRUE; } @@ -812,95 +134,15 @@ static unsigned translate_prim(const struct gl_context *ctx, unsigned prim) { /* GL prims should match Gallium prims, spot-check a few */ - assert(GL_POINTS == PIPE_PRIM_POINTS); - assert(GL_QUADS == PIPE_PRIM_QUADS); - assert(GL_TRIANGLE_STRIP_ADJACENCY == PIPE_PRIM_TRIANGLE_STRIP_ADJACENCY); - - /* Avoid quadstrips if it's easy to do so: - * Note: it's important to do the correct trimming if we change the - * prim type! We do that wherever this function is called. - */ - if (prim == GL_QUAD_STRIP && - ctx->Light.ShadeModel != GL_FLAT && - ctx->Polygon.FrontMode == GL_FILL && - ctx->Polygon.BackMode == GL_FILL) - prim = GL_TRIANGLE_STRIP; + STATIC_ASSERT(GL_POINTS == PIPE_PRIM_POINTS); + STATIC_ASSERT(GL_QUADS == PIPE_PRIM_QUADS); + STATIC_ASSERT(GL_TRIANGLE_STRIP_ADJACENCY == PIPE_PRIM_TRIANGLE_STRIP_ADJACENCY); + STATIC_ASSERT(GL_PATCHES == PIPE_PRIM_PATCHES); return prim; } -/** - * Setup vertex arrays and buffers prior to drawing. - * \return GL_TRUE for success, GL_FALSE otherwise (probably out of memory) - */ -static GLboolean -st_validate_varrays(struct gl_context *ctx, - const struct gl_client_array **arrays, - unsigned max_index, - unsigned num_instances) -{ - struct st_context *st = st_context(ctx); - const struct st_vertex_program *vp; - const struct st_vp_variant *vpv; - struct pipe_vertex_buffer vbuffer[PIPE_MAX_SHADER_INPUTS]; - struct pipe_vertex_element velements[PIPE_MAX_ATTRIBS]; - unsigned num_vbuffers, num_velements; - GLuint attr; - unsigned i; - - /* must get these after state validation! */ - vp = st->vp; - vpv = st->vp_variant; - - memset(velements, 0, sizeof(struct pipe_vertex_element) * vpv->num_inputs); - - /* Unreference any user vertex buffers. */ - for (i = 0; i < st->num_user_attribs; i++) { - pipe_resource_reference(&st->user_attrib[i].buffer, NULL); - } - st->num_user_attribs = 0; - - /* - * Setup the vbuffer[] and velements[] arrays. - */ - if (is_interleaved_arrays(vp, vpv, arrays)) { - if (!setup_interleaved_attribs(ctx, vp, vpv, arrays, vbuffer, velements, - max_index, num_instances)) { - return GL_FALSE; - } - - num_vbuffers = 1; - num_velements = vpv->num_inputs; - if (num_velements == 0) - num_vbuffers = 0; - } - else { - if (!setup_non_interleaved_attribs(ctx, vp, vpv, arrays, - vbuffer, velements, max_index, - num_instances)) { - return GL_FALSE; - } - - num_vbuffers = vpv->num_inputs; - num_velements = vpv->num_inputs; - } - - cso_set_vertex_buffers(st->cso_context, num_vbuffers, vbuffer); - cso_set_vertex_elements(st->cso_context, num_velements, velements); - - /* unreference buffers (frees wrapped user-space buffer objects) - * This is OK, because the pipe driver should reference buffers by itself - * in set_vertex_buffers. */ - for (attr = 0; attr < num_vbuffers; attr++) { - pipe_resource_reference(&vbuffer[attr].buffer, NULL); - assert(!vbuffer[attr].buffer); - } - - return GL_TRUE; -} - - /** * This function gets plugged into the VBO module and is called when * we have something to render. @@ -908,108 +150,52 @@ st_validate_varrays(struct gl_context *ctx, */ void st_draw_vbo(struct gl_context *ctx, - const struct gl_client_array **arrays, const struct _mesa_prim *prims, GLuint nr_prims, const struct _mesa_index_buffer *ib, GLboolean index_bounds_valid, GLuint min_index, GLuint max_index, - struct gl_transform_feedback_object *tfb_vertcount) + struct gl_transform_feedback_object *tfb_vertcount, + unsigned stream, + struct gl_buffer_object *indirect) { struct st_context *st = st_context(ctx); - struct pipe_context *pipe = st->pipe; - struct pipe_index_buffer ibuffer; + struct pipe_index_buffer ibuffer = {0}; struct pipe_draw_info info; - unsigned i, num_instances = 1; - GLboolean new_array = - st->dirty.st && - (st->dirty.mesa & (_NEW_ARRAY | _NEW_PROGRAM | _NEW_BUFFER_OBJECT)) != 0; + const struct gl_client_array **arrays = ctx->Array._DrawArrays; + unsigned i; /* Mesa core state should have been validated already */ assert(ctx->NewState == 0x0); - if (ib) { - /* Gallium probably doesn't want this in some cases. */ - if (!index_bounds_valid) - if (!all_varyings_in_vbos(arrays)) - vbo_get_minmax_index(ctx, prims, ib, &min_index, &max_index); - - for (i = 0; i < nr_prims; i++) { - num_instances = MAX2(num_instances, prims[i].num_instances); - } - } - else { - /* Get min/max index for non-indexed drawing. */ - min_index = ~0; - max_index = 0; - - for (i = 0; i < nr_prims; i++) { - min_index = MIN2(min_index, prims[i].start); - max_index = MAX2(max_index, prims[i].start + prims[i].count - 1); - num_instances = MAX2(num_instances, prims[i].num_instances); - } - } + st_flush_bitmap_cache(st); /* Validate state. */ - if (st->dirty.st) { - GLboolean vertDataEdgeFlags; - - /* sanity check for pointer arithmetic below */ - assert(sizeof(arrays[0]->Ptr[0]) == 1); - - vertDataEdgeFlags = arrays[VERT_ATTRIB_EDGEFLAG]->BufferObj && - arrays[VERT_ATTRIB_EDGEFLAG]->BufferObj->Name; - if (vertDataEdgeFlags != st->vertdata_edgeflags) { - st->vertdata_edgeflags = vertDataEdgeFlags; - st->dirty.st |= ST_NEW_EDGEFLAGS_DATA; - } - - st_validate_state(st); - - if (new_array) { - if (!st_validate_varrays(ctx, arrays, max_index, num_instances)) { - /* probably out of memory, no-op the draw call */ - return; - } - } - -#if 0 - if (MESA_VERBOSE & VERBOSE_GLSL) { - check_uniforms(ctx); - } -#else - (void) check_uniforms; -#endif + if (st->dirty.st || st->dirty.mesa || ctx->NewDriverState) { + st_validate_state(st, ST_PIPELINE_RENDER); } - /* Notify the driver that the content of user buffers may have been - * changed. */ - assert(max_index >= min_index); - if (!new_array && st->num_user_attribs) { - for (i = 0; i < st->num_user_attribs; i++) { - if (st->user_attrib[i].buffer) { - unsigned element_size = st->user_attrib[i].element_size; - unsigned stride = st->user_attrib[i].stride; - unsigned min_offset = min_index * stride; - unsigned max_offset = max_index * stride + element_size; - - assert(max_offset > min_offset); - - pipe->redefine_user_buffer(pipe, st->user_attrib[i].buffer, - min_offset, - max_offset - min_offset); - } - } + if (st->vertex_array_out_of_memory) { + return; } - setup_index_buffer(ctx, ib, &ibuffer); - pipe->set_index_buffer(pipe, &ibuffer); - util_draw_init_info(&info); + if (ib) { + /* Get index bounds for user buffers. */ + if (!index_bounds_valid) + if (!all_varyings_in_vbos(arrays)) + vbo_get_minmax_indices(ctx, prims, ib, &min_index, &max_index, + nr_prims); + + if (!setup_index_buffer(st, ib, &ibuffer)) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "glBegin/DrawElements/DrawArray"); + return; + } + info.indexed = TRUE; - if (min_index != ~0 && max_index != ~0) { + if (min_index != ~0U && max_index != ~0U) { info.min_index = min_index; info.max_index = max_index; } @@ -1017,45 +203,136 @@ st_draw_vbo(struct gl_context *ctx, /* The VBO module handles restart for the non-indexed GLDrawArrays * so we only set these fields for indexed drawing: */ - info.primitive_restart = ctx->Array.PrimitiveRestart; - info.restart_index = ctx->Array.RestartIndex; + info.primitive_restart = ctx->Array._PrimitiveRestart; + info.restart_index = _mesa_primitive_restart_index(ctx, ib->type); } - - /* Set info.count_from_stream_output. */ - if (tfb_vertcount) { - st_transform_feedback_draw_init(tfb_vertcount, &info); + else { + /* Transform feedback drawing is always non-indexed. */ + /* Set info.count_from_stream_output. */ + if (tfb_vertcount) { + if (!st_transform_feedback_draw_init(tfb_vertcount, stream, &info)) + return; + } } + assert(!indirect); + /* do actual drawing */ for (i = 0; i < nr_prims; i++) { - info.mode = translate_prim( ctx, prims[i].mode ); + info.mode = translate_prim(ctx, prims[i].mode); info.start = prims[i].start; info.count = prims[i].count; + info.start_instance = prims[i].base_instance; info.instance_count = prims[i].num_instances; + info.vertices_per_patch = ctx->TessCtrlProgram.patch_vertices; info.index_bias = prims[i].basevertex; + info.drawid = prims[i].draw_id; if (!ib) { info.min_index = info.start; info.max_index = info.start + info.count - 1; } + if (ST_DEBUG & DEBUG_DRAW) { + debug_printf("st/draw: mode %s start %u count %u indexed %d\n", + u_prim_name(info.mode), + info.start, + info.count, + info.indexed); + } + if (info.count_from_stream_output) { - pipe->draw_vbo(pipe, &info); + cso_draw_vbo(st->cso_context, &info); } else if (info.primitive_restart) { - if (st->sw_primitive_restart) { - /* Handle primitive restart for drivers that doesn't support it */ - handle_fallback_primitive_restart(pipe, ib, &ibuffer, &info); - } - else { - /* don't trim, restarts might be inside index list */ - pipe->draw_vbo(pipe, &info); - } + /* don't trim, restarts might be inside index list */ + cso_draw_vbo(st->cso_context, &info); + } + else if (u_trim_pipe_prim(prims[i].mode, &info.count)) { + cso_draw_vbo(st->cso_context, &info); } - else if (u_trim_pipe_prim(info.mode, &info.count)) - pipe->draw_vbo(pipe, &info); } - pipe_resource_reference(&ibuffer.buffer, NULL); + if (ib && st->indexbuf_uploader && !_mesa_is_bufferobj(ib->obj)) { + pipe_resource_reference(&ibuffer.buffer, NULL); + } +} + +static void +st_indirect_draw_vbo(struct gl_context *ctx, + GLuint mode, + struct gl_buffer_object *indirect_data, + GLsizeiptr indirect_offset, + unsigned draw_count, + unsigned stride, + struct gl_buffer_object *indirect_params, + GLsizeiptr indirect_params_offset, + const struct _mesa_index_buffer *ib) +{ + struct st_context *st = st_context(ctx); + struct pipe_index_buffer ibuffer = {0}; + struct pipe_draw_info info; + + /* Mesa core state should have been validated already */ + assert(ctx->NewState == 0x0); + assert(stride); + + /* Validate state. */ + if (st->dirty.st || st->dirty.mesa || ctx->NewDriverState) { + st_validate_state(st, ST_PIPELINE_RENDER); + } + + if (st->vertex_array_out_of_memory) { + return; + } + + util_draw_init_info(&info); + + if (ib) { + if (!setup_index_buffer(st, ib, &ibuffer)) { + _mesa_error(ctx, GL_OUT_OF_MEMORY, "gl%sDrawElementsIndirect%s", + (draw_count > 1) ? "Multi" : "", + indirect_params ? "CountARB" : ""); + return; + } + + info.indexed = TRUE; + } + + info.mode = translate_prim(ctx, mode); + info.vertices_per_patch = ctx->TessCtrlProgram.patch_vertices; + info.indirect = st_buffer_object(indirect_data)->buffer; + info.indirect_offset = indirect_offset; + + /* Primitive restart is not handled by the VBO module in this case. */ + info.primitive_restart = ctx->Array._PrimitiveRestart; + info.restart_index = ctx->Array.RestartIndex; + + if (ST_DEBUG & DEBUG_DRAW) { + debug_printf("st/draw indirect: mode %s drawcount %d indexed %d\n", + u_prim_name(info.mode), + draw_count, + info.indexed); + } + + if (!st->has_multi_draw_indirect) { + int i; + + assert(!indirect_params); + info.indirect_count = 1; + for (i = 0; i < draw_count; i++) { + info.drawid = i; + cso_draw_vbo(st->cso_context, &info); + info.indirect_offset += stride; + } + } else { + info.indirect_count = draw_count; + info.indirect_stride = stride; + if (indirect_params) { + info.indirect_params = st_buffer_object(indirect_params)->buffer; + info.indirect_params_offset = indirect_params_offset; + } + cso_draw_vbo(st->cso_context, &info); + } } @@ -1065,8 +342,8 @@ st_init_draw(struct st_context *st) struct gl_context *ctx = st->ctx; vbo_set_draw_func(ctx, st_draw_vbo); + vbo_set_indirect_draw_func(ctx, st_indirect_draw_vbo); -#if FEATURE_feedback || FEATURE_rastpos st->draw = draw_create(st->pipe); /* for selection/feedback */ /* Disable draw options that might convert points/lines to tris, etc. @@ -1076,14 +353,101 @@ st_init_draw(struct st_context *st) draw_wide_point_threshold(st->draw, 1000.0f); draw_enable_line_stipple(st->draw, FALSE); draw_enable_point_sprites(st->draw, FALSE); -#endif } void st_destroy_draw(struct st_context *st) { -#if FEATURE_feedback || FEATURE_rastpos draw_destroy(st->draw); -#endif +} + + +/** + * Draw a quad with given position, texcoords and color. + */ +bool +st_draw_quad(struct st_context *st, + float x0, float y0, float x1, float y1, float z, + float s0, float t0, float s1, float t1, + const float *color, + unsigned num_instances) +{ + struct pipe_vertex_buffer vb = {0}; + struct st_util_vertex *verts; + + vb.stride = sizeof(struct st_util_vertex); + + u_upload_alloc(st->uploader, 0, 4 * sizeof(struct st_util_vertex), 4, + &vb.buffer_offset, &vb.buffer, (void **) &verts); + if (!vb.buffer) { + return false; + } + + /* lower-left */ + verts[0].x = x0; + verts[0].y = y1; + verts[0].z = z; + verts[0].r = color[0]; + verts[0].g = color[1]; + verts[0].b = color[2]; + verts[0].a = color[3]; + verts[0].s = s0; + verts[0].t = t0; + + /* lower-right */ + verts[1].x = x1; + verts[1].y = y1; + verts[1].z = z; + verts[1].r = color[0]; + verts[1].g = color[1]; + verts[1].b = color[2]; + verts[1].a = color[3]; + verts[1].s = s1; + verts[1].t = t0; + + /* upper-right */ + verts[2].x = x1; + verts[2].y = y0; + verts[2].z = z; + verts[2].r = color[0]; + verts[2].g = color[1]; + verts[2].b = color[2]; + verts[2].a = color[3]; + verts[2].s = s1; + verts[2].t = t1; + + /* upper-left */ + verts[3].x = x0; + verts[3].y = y0; + verts[3].z = z; + verts[3].r = color[0]; + verts[3].g = color[1]; + verts[3].b = color[2]; + verts[3].a = color[3]; + verts[3].s = s0; + verts[3].t = t1; + + u_upload_unmap(st->uploader); + + /* At the time of writing, cso_get_aux_vertex_buffer_slot() always returns + * zero. If that ever changes we need to audit the calls to that function + * and make sure the slot number is used consistently everywhere. + */ + assert(cso_get_aux_vertex_buffer_slot(st->cso_context) == 0); + + cso_set_vertex_buffers(st->cso_context, + cso_get_aux_vertex_buffer_slot(st->cso_context), + 1, &vb); + + if (num_instances > 1) { + cso_draw_arrays_instanced(st->cso_context, PIPE_PRIM_TRIANGLE_FAN, 0, 4, + 0, num_instances); + } else { + cso_draw_arrays(st->cso_context, PIPE_PRIM_TRIANGLE_FAN, 0, 4); + } + + pipe_resource_reference(&vb.buffer, NULL); + + return true; }