-/**************************************************************************\r
- * \r
- * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.\r
- * All Rights Reserved.\r
- * \r
- * Permission is hereby granted, free of charge, to any person obtaining a\r
- * copy of this software and associated documentation files (the\r
- * "Software"), to deal in the Software without restriction, including\r
- * without limitation the rights to use, copy, modify, merge, publish,\r
- * distribute, sub license, and/or sell copies of the Software, and to\r
- * permit persons to whom the Software is furnished to do so, subject to\r
- * the following conditions:\r
- * \r
- * The above copyright notice and this permission notice (including the\r
- * next paragraph) shall be included in all copies or substantial portions\r
- * of the Software.\r
- * \r
- * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS\r
- * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF\r
- * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT.\r
- * IN NO EVENT SHALL TUNGSTEN GRAPHICS AND/OR ITS SUPPLIERS BE LIABLE FOR\r
- * ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,\r
- * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE\r
- * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.\r
- * \r
- **************************************************************************/\r
-\r
-/* Vertices are just an array of floats, with all the attributes\r
- * packed. We currently assume a layout like:\r
- *\r
- * attr[0][0..3] - window position\r
- * attr[1..n][0..3] - remaining attributes.\r
- *\r
- * Attributes are assumed to be 4 floats wide but are packed so that\r
- * all the enabled attributes run contiguously.\r
- */\r
-\r
-#include "pipe/p_util.h"\r
-#include "pipe/p_defines.h"\r
-#include "pipe/p_shader_tokens.h"\r
-\r
-#include "sp_context.h"\r
-#include "sp_state.h"\r
-#include "sp_headers.h"\r
-#include "sp_quad.h"\r
-#include "sp_texture.h"\r
-#include "sp_tex_sample.h"\r
-\r
-\r
-struct quad_shade_stage\r
-{\r
- struct quad_stage stage;\r
- struct tgsi_sampler samplers[PIPE_MAX_SAMPLERS];\r
- struct tgsi_exec_machine machine;\r
- struct tgsi_exec_vector *inputs, *outputs;\r
- int colorOutSlot, depthOutSlot;\r
-};\r
-\r
-\r
-/** cast wrapper */\r
-static INLINE struct quad_shade_stage *\r
-quad_shade_stage(struct quad_stage *qs)\r
-{\r
- return (struct quad_shade_stage *) qs;\r
-}\r
-\r
-\r
-\r
-/**\r
- * Execute fragment shader for the four fragments in the quad.\r
- */\r
-static void\r
-shade_quad(\r
- struct quad_stage *qs,\r
- struct quad_header *quad )\r
-{\r
- struct quad_shade_stage *qss = quad_shade_stage( qs );\r
- struct softpipe_context *softpipe = qs->softpipe;\r
- struct tgsi_exec_machine *machine = &qss->machine;\r
-\r
- /* Consts do not require 16 byte alignment. */\r
- machine->Consts = softpipe->mapped_constants[PIPE_SHADER_FRAGMENT];\r
-\r
- machine->InterpCoefs = quad->coef;\r
-\r
- /* run shader */\r
- quad->mask &= softpipe->fs->run( softpipe->fs, \r
- &qss->machine,\r
- quad );\r
-\r
- /* store result color */\r
- if (qss->colorOutSlot >= 0) {\r
- /* XXX need to handle multiple color outputs someday */\r
- assert(qss->stage.softpipe->fs->info.output_semantic_name[qss->colorOutSlot]\r
- == TGSI_SEMANTIC_COLOR);\r
- memcpy(\r
- quad->outputs.color,\r
- &machine->Outputs[qss->colorOutSlot].xyzw[0].f[0],\r
- sizeof( quad->outputs.color ) );\r
- }\r
-\r
- /*\r
- * XXX the following code for updating quad->outputs.depth\r
- * isn't really needed if we did early z testing.\r
- */\r
-\r
- /* store result Z */\r
- if (qss->depthOutSlot >= 0) {\r
- /* output[slot] is new Z */\r
- uint i;\r
- for (i = 0; i < 4; i++) {\r
- quad->outputs.depth[i] = machine->Outputs[0].xyzw[2].f[i];\r
- }\r
- }\r
- else {\r
- /* copy input Z (which was interpolated by the executor) to output Z */\r
- uint i;\r
- for (i = 0; i < 4; i++) {\r
- quad->outputs.depth[i] = machine->Inputs[0].xyzw[2].f[i];\r
- /* XXX not sure the above line is always correct. The following\r
- * might be better:\r
- quad->outputs.depth[i] = machine->QuadPos.xyzw[2].f[i];\r
- */\r
- }\r
- }\r
-\r
- /* shader may cull fragments */\r
- if( quad->mask ) {\r
- qs->next->run( qs->next, quad );\r
- }\r
-}\r
-\r
-/**\r
- * Per-primitive (or per-begin?) setup\r
- */\r
-static void shade_begin(struct quad_stage *qs)\r
-{\r
- struct quad_shade_stage *qss = quad_shade_stage(qs);\r
- struct softpipe_context *softpipe = qs->softpipe;\r
- unsigned i;\r
- unsigned num = MAX2(softpipe->num_textures, softpipe->num_samplers);\r
-\r
- /* set TGSI sampler state that varies */\r
- for (i = 0; i < num; i++) {\r
- qss->samplers[i].state = softpipe->sampler[i];\r
- qss->samplers[i].texture = softpipe->texture[i];\r
- }\r
-\r
- /* find output slots for depth, color */\r
- qss->colorOutSlot = -1;\r
- qss->depthOutSlot = -1;\r
- for (i = 0; i < qss->stage.softpipe->fs->info.num_outputs; i++) {\r
- switch (qss->stage.softpipe->fs->info.output_semantic_name[i]) {\r
- case TGSI_SEMANTIC_POSITION:\r
- qss->depthOutSlot = i;\r
- break;\r
- case TGSI_SEMANTIC_COLOR:\r
- qss->colorOutSlot = i;\r
- break;\r
- }\r
- }\r
- \r
- softpipe->fs->prepare( softpipe->fs, \r
- &qss->machine,\r
- qss->samplers );\r
-\r
- qs->next->begin(qs->next);\r
-}\r
-\r
-\r
-static void shade_destroy(struct quad_stage *qs)\r
-{\r
- struct quad_shade_stage *qss = (struct quad_shade_stage *) qs;\r
-\r
- tgsi_exec_machine_free_data(&qss->machine);\r
- FREE( qss->inputs );\r
- FREE( qss->outputs );\r
- FREE( qs );\r
-}\r
-\r
-\r
-struct quad_stage *sp_quad_shade_stage( struct softpipe_context *softpipe )\r
-{\r
- struct quad_shade_stage *qss = CALLOC_STRUCT(quad_shade_stage);\r
- uint i;\r
-\r
- /* allocate storage for program inputs/outputs, aligned to 16 bytes */\r
- qss->inputs = MALLOC(PIPE_ATTRIB_MAX * sizeof(*qss->inputs) + 16);\r
- qss->outputs = MALLOC(PIPE_ATTRIB_MAX * sizeof(*qss->outputs) + 16);\r
- qss->machine.Inputs = align16(qss->inputs);\r
- qss->machine.Outputs = align16(qss->outputs);\r
-\r
- qss->stage.softpipe = softpipe;\r
- qss->stage.begin = shade_begin;\r
- qss->stage.run = shade_quad;\r
- qss->stage.destroy = shade_destroy;\r
-\r
- /* set TGSI sampler state that's constant */\r
- for (i = 0; i < PIPE_MAX_SAMPLERS; i++) {\r
- assert(softpipe->tex_cache[i]);\r
- qss->samplers[i].get_samples = sp_get_samples;\r
- qss->samplers[i].pipe = &softpipe->pipe;\r
- qss->samplers[i].cache = softpipe->tex_cache[i];\r
- }\r
-\r
- tgsi_exec_machine_init( &qss->machine );\r
-\r
- return &qss->stage;\r
-}\r
+/**************************************************************************
+ *
+ * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas.
+ * All Rights Reserved.
+ * Copyright 2008 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
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * 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
+ * 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.
+ *
+ **************************************************************************/
+
+/* Vertices are just an array of floats, with all the attributes
+ * packed. We currently assume a layout like:
+ *
+ * attr[0][0..3] - window position
+ * attr[1..n][0..3] - remaining attributes.
+ *
+ * Attributes are assumed to be 4 floats wide but are packed so that
+ * all the enabled attributes run contiguously.
+ */
+
+#include "util/u_math.h"
+#include "util/u_memory.h"
+#include "pipe/p_defines.h"
+#include "pipe/p_shader_tokens.h"
+
+#include "sp_context.h"
+#include "sp_state.h"
+#include "sp_quad.h"
+#include "sp_quad_pipe.h"
+
+
+struct quad_shade_stage
+{
+ struct quad_stage stage; /**< base class */
+
+ /* no other fields at this time */
+};
+
+
+/** cast wrapper */
+static INLINE struct quad_shade_stage *
+quad_shade_stage(struct quad_stage *qs)
+{
+ return (struct quad_shade_stage *) qs;
+}
+
+
+/**
+ * Execute fragment shader for the four fragments in the quad.
+ * \return TRUE if quad is alive, FALSE if all four pixels are killed
+ */
+static INLINE boolean
+shade_quad(struct quad_stage *qs, struct quad_header *quad)
+{
+ struct softpipe_context *softpipe = qs->softpipe;
+ struct tgsi_exec_machine *machine = softpipe->fs_machine;
+
+ /* run shader */
+ machine->flatshade_color = softpipe->rasterizer->flatshade ? TRUE : FALSE;
+ return softpipe->fs_variant->run( softpipe->fs_variant, machine, quad );
+}
+
+
+
+static void
+coverage_quad(struct quad_stage *qs, struct quad_header *quad)
+{
+ struct softpipe_context *softpipe = qs->softpipe;
+ uint cbuf;
+
+ /* loop over colorbuffer outputs */
+ for (cbuf = 0; cbuf < softpipe->framebuffer.nr_cbufs; cbuf++) {
+ float (*quadColor)[4] = quad->output.color[cbuf];
+ unsigned j;
+ for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+ assert(quad->input.coverage[j] >= 0.0);
+ assert(quad->input.coverage[j] <= 1.0);
+ quadColor[3][j] *= quad->input.coverage[j];
+ }
+ }
+}
+
+
+/**
+ * Shade/write an array of quads
+ * Called via quad_stage::run()
+ */
+static void
+shade_quads(struct quad_stage *qs,
+ struct quad_header *quads[],
+ unsigned nr)
+{
+ struct softpipe_context *softpipe = qs->softpipe;
+ struct tgsi_exec_machine *machine = softpipe->fs_machine;
+ unsigned i, nr_quads = 0;
+
+ tgsi_exec_set_constant_buffers(machine, PIPE_MAX_CONSTANT_BUFFERS,
+ softpipe->mapped_constants[PIPE_SHADER_FRAGMENT],
+ softpipe->const_buffer_size[PIPE_SHADER_FRAGMENT]);
+
+ machine->InterpCoefs = quads[0]->coef;
+
+ for (i = 0; i < nr; i++) {
+ /* Only omit this quad from the output list if all the fragments
+ * are killed _AND_ it's not the first quad in the list.
+ * The first quad is special in the (optimized) depth-testing code:
+ * the quads' Z coordinates are step-wise interpolated with respect
+ * to the first quad in the list.
+ * For multi-pass algorithms we need to produce exactly the same
+ * Z values in each pass. If interpolation starts with different quads
+ * we can get different Z values for the same (x,y).
+ */
+ if (!shade_quad(qs, quads[i]) && i > 0)
+ continue; /* quad totally culled/killed */
+
+ if (/*do_coverage*/ 0)
+ coverage_quad( qs, quads[i] );
+
+ quads[nr_quads++] = quads[i];
+ }
+
+ if (nr_quads)
+ qs->next->run(qs->next, quads, nr_quads);
+}
+
+
+/**
+ * Per-primitive (or per-begin?) setup
+ */
+static void
+shade_begin(struct quad_stage *qs)
+{
+ struct softpipe_context *softpipe = qs->softpipe;
+
+ softpipe->fs_variant->prepare( softpipe->fs_variant,
+ softpipe->fs_machine,
+ (struct tgsi_sampler **)
+ softpipe->tgsi.frag_samplers_list );
+
+ qs->next->begin(qs->next);
+}
+
+
+static void
+shade_destroy(struct quad_stage *qs)
+{
+ FREE( qs );
+}
+
+
+struct quad_stage *
+sp_quad_shade_stage( struct softpipe_context *softpipe )
+{
+ struct quad_shade_stage *qss = CALLOC_STRUCT(quad_shade_stage);
+ if (!qss)
+ goto fail;
+
+ qss->stage.softpipe = softpipe;
+ qss->stage.begin = shade_begin;
+ qss->stage.run = shade_quads;
+ qss->stage.destroy = shade_destroy;
+
+ return &qss->stage;
+
+fail:
+ FREE(qss);
+ return NULL;
+}