-/**************************************************************************
- *
+/*
* Copyright 2003 VMware, Inc.
* All Rights Reserved.
*
* 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
+ * distribute, sublicense, 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 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.
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
* 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.
- *
- **************************************************************************/
+ */
#include <sys/errno.h>
-#include "main/glheader.h"
#include "main/context.h"
#include "main/condrender.h"
#include "main/samplerobj.h"
#include "main/enums.h"
#include "main/macros.h"
#include "main/transformfeedback.h"
+#include "main/framebuffer.h"
#include "tnl/tnl.h"
#include "vbo/vbo_context.h"
#include "swrast/swrast.h"
#include "swrast_setup/swrast_setup.h"
#include "drivers/common/meta.h"
+#include "util/bitscan.h"
#include "brw_blorp.h"
#include "brw_draw.h"
#define FILE_DEBUG_FLAG DEBUG_PRIMS
-static const GLuint prim_to_hw_prim[GL_TRIANGLE_STRIP_ADJACENCY+1] = {
- _3DPRIM_POINTLIST,
- _3DPRIM_LINELIST,
- _3DPRIM_LINELOOP,
- _3DPRIM_LINESTRIP,
- _3DPRIM_TRILIST,
- _3DPRIM_TRISTRIP,
- _3DPRIM_TRIFAN,
- _3DPRIM_QUADLIST,
- _3DPRIM_QUADSTRIP,
- _3DPRIM_POLYGON,
- _3DPRIM_LINELIST_ADJ,
- _3DPRIM_LINESTRIP_ADJ,
- _3DPRIM_TRILIST_ADJ,
- _3DPRIM_TRISTRIP_ADJ,
-};
-
static const GLenum reduced_prim[GL_POLYGON+1] = {
- GL_POINTS,
- GL_LINES,
- GL_LINES,
- GL_LINES,
- GL_TRIANGLES,
- GL_TRIANGLES,
- GL_TRIANGLES,
- GL_TRIANGLES,
- GL_TRIANGLES,
- GL_TRIANGLES
+ [GL_POINTS] = GL_POINTS,
+ [GL_LINES] = GL_LINES,
+ [GL_LINE_LOOP] = GL_LINES,
+ [GL_LINE_STRIP] = GL_LINES,
+ [GL_TRIANGLES] = GL_TRIANGLES,
+ [GL_TRIANGLE_STRIP] = GL_TRIANGLES,
+ [GL_TRIANGLE_FAN] = GL_TRIANGLES,
+ [GL_QUADS] = GL_TRIANGLES,
+ [GL_QUAD_STRIP] = GL_TRIANGLES,
+ [GL_POLYGON] = GL_TRIANGLES
};
-uint32_t
-get_hw_prim_for_gl_prim(int mode)
-{
- if (mode >= BRW_PRIM_OFFSET)
- return mode - BRW_PRIM_OFFSET;
- else {
- assert(mode < ARRAY_SIZE(prim_to_hw_prim));
- return prim_to_hw_prim[mode];
- }
-}
-
-
/* When the primitive changes, set a state bit and re-validate. Not
* the nicest and would rather deal with this by having all the
* programs be immune to the active primitive (ie. cope with all
brw->ctx.NewDriverState |= BRW_NEW_PRIMITIVE;
if (reduced_prim[prim->mode] != brw->reduced_primitive) {
- brw->reduced_primitive = reduced_prim[prim->mode];
- brw->ctx.NewDriverState |= BRW_NEW_REDUCED_PRIMITIVE;
+ brw->reduced_primitive = reduced_prim[prim->mode];
+ brw->ctx.NewDriverState |= BRW_NEW_REDUCED_PRIMITIVE;
}
}
}
static void
gen6_set_prim(struct brw_context *brw, const struct _mesa_prim *prim)
{
+ const struct gl_context *ctx = &brw->ctx;
+ uint32_t hw_prim;
+
DBG("PRIM: %s\n", _mesa_enum_to_string(prim->mode));
- const uint32_t hw_prim = get_hw_prim_for_gl_prim(prim->mode);
+ if (prim->mode == GL_PATCHES) {
+ hw_prim = _3DPRIM_PATCHLIST(ctx->TessCtrlProgram.patch_vertices);
+ } else {
+ hw_prim = get_hw_prim_for_gl_prim(prim->mode);
+ }
+
if (hw_prim != brw->primitive) {
brw->primitive = hw_prim;
brw->ctx.NewDriverState |= BRW_NEW_PRIMITIVE;
+ if (prim->mode == GL_PATCHES)
+ brw->ctx.NewDriverState |= BRW_NEW_PATCH_PRIMITIVE;
}
}
static void
brw_emit_prim(struct brw_context *brw,
const struct _mesa_prim *prim,
- uint32_t hw_prim)
+ uint32_t hw_prim,
+ struct brw_transform_feedback_object *xfb_obj,
+ unsigned stream)
{
int verts_per_instance;
int vertex_access_type;
verts_per_instance = prim->count;
/* If nothing to emit, just return. */
- if (verts_per_instance == 0 && !prim->is_indirect)
+ if (verts_per_instance == 0 && !prim->is_indirect && !xfb_obj)
return;
/* If we're set to always flush, do it before and after the primitive emit.
brw_emit_mi_flush(brw);
/* If indirect, emit a bunch of loads from the indirect BO. */
- if (prim->is_indirect) {
+ if (xfb_obj) {
+ indirect_flag = GEN7_3DPRIM_INDIRECT_PARAMETER_ENABLE;
+
+ brw_load_register_mem(brw, GEN7_3DPRIM_VERTEX_COUNT,
+ xfb_obj->prim_count_bo,
+ I915_GEM_DOMAIN_VERTEX, 0,
+ stream * sizeof(uint32_t));
+ BEGIN_BATCH(9);
+ OUT_BATCH(MI_LOAD_REGISTER_IMM | (9 - 2));
+ OUT_BATCH(GEN7_3DPRIM_INSTANCE_COUNT);
+ OUT_BATCH(prim->num_instances);
+ OUT_BATCH(GEN7_3DPRIM_START_VERTEX);
+ OUT_BATCH(0);
+ OUT_BATCH(GEN7_3DPRIM_BASE_VERTEX);
+ OUT_BATCH(0);
+ OUT_BATCH(GEN7_3DPRIM_START_INSTANCE);
+ OUT_BATCH(0);
+ ADVANCE_BATCH();
+ } else if (prim->is_indirect) {
struct gl_buffer_object *indirect_buffer = brw->ctx.DrawIndirectBuffer;
drm_intel_bo *bo = intel_bufferobj_buffer(brw,
intel_buffer_object(indirect_buffer),
static void
brw_merge_inputs(struct brw_context *brw,
- const struct gl_client_array *arrays[])
+ const struct gl_vertex_array *arrays[])
{
const struct gl_context *ctx = &brw->ctx;
GLuint i;
}
if (brw->gen < 8 && !brw->is_haswell) {
- struct gl_program *vp = &ctx->VertexProgram._Current->Base;
+ uint64_t mask = ctx->VertexProgram._Current->info.inputs_read;
/* Prior to Haswell, the hardware can't natively support GL_FIXED or
* 2_10_10_10_REV vertex formats. Set appropriate workaround flags.
*/
- for (i = 0; i < VERT_ATTRIB_MAX; i++) {
- if (!(vp->InputsRead & BITFIELD64_BIT(i)))
- continue;
-
+ while (mask) {
uint8_t wa_flags = 0;
+ i = u_bit_scan64(&mask);
+
switch (brw->vb.inputs[i].glarray->Type) {
case GL_FIXED:
struct intel_renderbuffer *stencil_irb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
struct gl_renderbuffer_attachment *depth_att = &fb->Attachment[BUFFER_DEPTH];
- if (brw_is_front_buffer_drawing(fb))
+ if (_mesa_is_front_buffer_drawing(fb))
front_irb = intel_get_renderbuffer(fb, BUFFER_FRONT_LEFT);
if (front_irb)
front_irb->need_downsample = true;
if (back_irb)
back_irb->need_downsample = true;
- if (depth_irb && ctx->Depth.Mask) {
+ if (depth_irb && brw_depth_writes_enabled(brw)) {
intel_renderbuffer_att_set_needs_depth_resolve(depth_att);
brw_render_cache_set_add_bo(brw, depth_irb->mt->bo);
}
brw_render_cache_set_add_bo(brw, stencil_irb->mt->bo);
}
- for (int i = 0; i < fb->_NumColorDrawBuffers; i++) {
+ for (unsigned i = 0; i < fb->_NumColorDrawBuffers; i++) {
struct intel_renderbuffer *irb =
intel_renderbuffer(fb->_ColorDrawBuffers[i]);
- if (irb)
- brw_render_cache_set_add_bo(brw, irb->mt->bo);
+ if (!irb)
+ continue;
+
+ brw_render_cache_set_add_bo(brw, irb->mt->bo);
+ intel_miptree_used_for_rendering(
+ brw, irb->mt, irb->mt_level, irb->mt_layer, irb->layer_count);
+ }
+}
+
+static void
+brw_predraw_set_aux_buffers(struct brw_context *brw)
+{
+ if (brw->gen < 9)
+ return;
+
+ struct gl_context *ctx = &brw->ctx;
+ struct gl_framebuffer *fb = ctx->DrawBuffer;
+
+ for (unsigned i = 0; i < fb->_NumColorDrawBuffers; i++) {
+ struct intel_renderbuffer *irb =
+ intel_renderbuffer(fb->_ColorDrawBuffers[i]);
+
+ if (!irb) {
+ continue;
+ }
+
+ /* For layered rendering non-compressed fast cleared buffers need to be
+ * resolved. Surface state can carry only one fast color clear value
+ * while each layer may have its own fast clear color value. For
+ * compressed buffers color value is available in the color buffer.
+ */
+ if (irb->layer_count > 1 &&
+ !(irb->mt->aux_disable & INTEL_AUX_DISABLE_CCS) &&
+ !intel_miptree_is_lossless_compressed(brw, irb->mt)) {
+ assert(brw->gen >= 8);
+
+ intel_miptree_resolve_color(brw, irb->mt, irb->mt_level,
+ irb->mt_layer, irb->layer_count, 0);
+ }
}
}
*/
static void
brw_try_draw_prims(struct gl_context *ctx,
- const struct gl_client_array *arrays[],
+ const struct gl_vertex_array *arrays[],
const struct _mesa_prim *prims,
GLuint nr_prims,
const struct _mesa_index_buffer *ib,
+ bool index_bounds_valid,
GLuint min_index,
GLuint max_index,
+ struct brw_transform_feedback_object *xfb_obj,
+ unsigned stream,
struct gl_buffer_object *indirect)
{
struct brw_context *brw = brw_context(ctx);
if (ctx->NewState)
_mesa_update_state(ctx);
- /* Find the highest sampler unit used by each shader program. A bit-count
- * won't work since ARB programs use the texture unit number as the sampler
- * index.
- */
- brw->wm.base.sampler_count =
- _mesa_fls(ctx->FragmentProgram._Current->Base.SamplersUsed);
- brw->gs.base.sampler_count = ctx->GeometryProgram._Current ?
- _mesa_fls(ctx->GeometryProgram._Current->Base.SamplersUsed) : 0;
- brw->vs.base.sampler_count =
- _mesa_fls(ctx->VertexProgram._Current->Base.SamplersUsed);
-
/* We have to validate the textures *before* checking for fallbacks;
* otherwise, the software fallback won't be able to rely on the
* texture state, the firstLevel and lastLevel fields won't be
*/
brw_validate_textures(brw);
+ /* Find the highest sampler unit used by each shader program. A bit-count
+ * won't work since ARB programs use the texture unit number as the sampler
+ * index.
+ */
+ brw->wm.base.sampler_count =
+ util_last_bit(ctx->FragmentProgram._Current->SamplersUsed);
+ brw->gs.base.sampler_count = ctx->GeometryProgram._Current ?
+ util_last_bit(ctx->GeometryProgram._Current->SamplersUsed) : 0;
+ brw->tes.base.sampler_count = ctx->TessEvalProgram._Current ?
+ util_last_bit(ctx->TessEvalProgram._Current->SamplersUsed) : 0;
+ brw->tcs.base.sampler_count = ctx->TessCtrlProgram._Current ?
+ util_last_bit(ctx->TessCtrlProgram._Current->SamplersUsed) : 0;
+ brw->vs.base.sampler_count =
+ util_last_bit(ctx->VertexProgram._Current->SamplersUsed);
+
intel_prepare_render(brw);
+ brw_predraw_set_aux_buffers(brw);
/* This workaround has to happen outside of brw_upload_render_state()
* because it may flush the batchbuffer for a blit, affecting the state
brw->ib.ib = ib;
brw->ctx.NewDriverState |= BRW_NEW_INDICES;
+ brw->vb.index_bounds_valid = index_bounds_valid;
brw->vb.min_index = min_index;
brw->vb.max_index = max_index;
brw->ctx.NewDriverState |= BRW_NEW_VERTICES;
intel_batchbuffer_save_state(brw);
if (brw->num_instances != prims[i].num_instances ||
- brw->basevertex != prims[i].basevertex) {
+ brw->basevertex != prims[i].basevertex ||
+ brw->baseinstance != prims[i].base_instance) {
brw->num_instances = prims[i].num_instances;
brw->basevertex = prims[i].basevertex;
+ brw->baseinstance = prims[i].base_instance;
if (i > 0) { /* For i == 0 we just did this before the loop */
brw->ctx.NewDriverState |= BRW_NEW_VERTICES;
brw_merge_inputs(brw, arrays);
}
}
- brw->draw.gl_basevertex =
+ /* Determine if we need to flag BRW_NEW_VERTICES for updating the
+ * gl_BaseVertexARB or gl_BaseInstanceARB values. For indirect draw, we
+ * always flag if the shader uses one of the values. For direct draws,
+ * we only flag if the values change.
+ */
+ const int new_basevertex =
prims[i].indexed ? prims[i].basevertex : prims[i].start;
+ const int new_baseinstance = prims[i].base_instance;
+ const struct brw_vs_prog_data *vs_prog_data =
+ brw_vs_prog_data(brw->vs.base.prog_data);
+ if (i > 0) {
+ const bool uses_draw_parameters =
+ vs_prog_data->uses_basevertex ||
+ vs_prog_data->uses_baseinstance;
+
+ if ((uses_draw_parameters && prims[i].is_indirect) ||
+ (vs_prog_data->uses_basevertex &&
+ brw->draw.params.gl_basevertex != new_basevertex) ||
+ (vs_prog_data->uses_baseinstance &&
+ brw->draw.params.gl_baseinstance != new_baseinstance))
+ brw->ctx.NewDriverState |= BRW_NEW_VERTICES;
+ }
+ brw->draw.params.gl_basevertex = new_basevertex;
+ brw->draw.params.gl_baseinstance = new_baseinstance;
drm_intel_bo_unreference(brw->draw.draw_params_bo);
if (prims[i].is_indirect) {
brw->draw.draw_params_offset = 0;
}
+ /* gl_DrawID always needs its own vertex buffer since it's not part of
+ * the indirect parameter buffer. If the program uses gl_DrawID we need
+ * to flag BRW_NEW_VERTICES. For the first iteration, we don't have
+ * valid vs_prog_data, but we always flag BRW_NEW_VERTICES before
+ * the loop.
+ */
+ brw->draw.gl_drawid = prims[i].draw_id;
+ drm_intel_bo_unreference(brw->draw.draw_id_bo);
+ brw->draw.draw_id_bo = NULL;
+ if (i > 0 && vs_prog_data->uses_drawid)
+ brw->ctx.NewDriverState |= BRW_NEW_VERTICES;
+
if (brw->gen < 6)
- brw_set_prim(brw, &prims[i]);
+ brw_set_prim(brw, &prims[i]);
else
- gen6_set_prim(brw, &prims[i]);
+ gen6_set_prim(brw, &prims[i]);
retry:
* brw->ctx.NewDriverState.
*/
if (brw->ctx.NewDriverState) {
- brw->no_batch_wrap = true;
- brw_upload_render_state(brw);
+ brw->no_batch_wrap = true;
+ brw_upload_render_state(brw);
}
- brw_emit_prim(brw, &prims[i], brw->primitive);
+ brw_emit_prim(brw, &prims[i], brw->primitive, xfb_obj, stream);
brw->no_batch_wrap = false;
if (dri_bufmgr_check_aperture_space(&brw->batch.bo, 1)) {
- if (!fail_next) {
- intel_batchbuffer_reset_to_saved(brw);
- intel_batchbuffer_flush(brw);
- fail_next = true;
- goto retry;
- } else {
+ if (!fail_next) {
+ intel_batchbuffer_reset_to_saved(brw);
+ intel_batchbuffer_flush(brw);
+ fail_next = true;
+ goto retry;
+ } else {
int ret = intel_batchbuffer_flush(brw);
WARN_ONCE(ret == -ENOSPC,
"i965: Single primitive emit exceeded "
"available aperture space\n");
- }
+ }
}
/* Now that we know we haven't run out of aperture space, we can safely
if (brw->always_flush_batch)
intel_batchbuffer_flush(brw);
- brw_state_cache_check_size(brw);
+ brw_program_cache_check_size(brw);
brw_postdraw_set_buffers_need_resolve(brw);
return;
GLboolean index_bounds_valid,
GLuint min_index,
GLuint max_index,
- struct gl_transform_feedback_object *unused_tfb_object,
+ struct gl_transform_feedback_object *gl_xfb_obj,
+ unsigned stream,
struct gl_buffer_object *indirect)
{
struct brw_context *brw = brw_context(ctx);
- const struct gl_client_array **arrays = ctx->Array._DrawArrays;
-
- assert(unused_tfb_object == NULL);
+ const struct gl_vertex_array **arrays = ctx->Array._DrawArrays;
+ struct brw_transform_feedback_object *xfb_obj =
+ (struct brw_transform_feedback_object *) gl_xfb_obj;
if (!brw_check_conditional_render(brw))
return;
_swsetup_Wakeup(ctx);
_tnl_wakeup(ctx);
_tnl_draw_prims(ctx, prims, nr_prims, ib,
- index_bounds_valid, min_index, max_index, NULL, NULL);
+ index_bounds_valid, min_index, max_index, NULL, 0, NULL);
return;
}
perf_debug("Scanning index buffer to compute index buffer bounds. "
"Use glDrawRangeElements() to avoid this.\n");
vbo_get_minmax_indices(ctx, prims, ib, &min_index, &max_index, nr_prims);
+ index_bounds_valid = true;
}
/* Try drawing with the hardware, but don't do anything else if we can't
* manage it. swrast doesn't support our featureset, so we can't fall back
* to it.
*/
- brw_try_draw_prims(ctx, arrays, prims, nr_prims, ib, min_index, max_index,
- indirect);
+ brw_try_draw_prims(ctx, arrays, prims, nr_prims, ib, index_bounds_valid,
+ min_index, max_index, xfb_obj, stream, indirect);
}
void
void
brw_draw_destroy(struct brw_context *brw)
{
- int i;
+ unsigned i;
for (i = 0; i < brw->vb.nr_buffers; i++) {
drm_intel_bo_unreference(brw->vb.buffers[i].bo);