static void
intel_viewport(struct gl_context *ctx, GLint x, GLint y, GLsizei w, GLsizei h)
{
struct brw_context *brw = brw_context(ctx);
__DRIcontext *driContext = brw->driContext;
(void) x;
(void) y;
(void) w;
(void) h;
if (_mesa_is_winsys_fbo(ctx->DrawBuffer)) {
dri2InvalidateDrawable(driContext->driDrawablePriv);
dri2InvalidateDrawable(driContext->driReadablePriv);
}
}
/**
* Check if the hardware's cut index support can handle the primitive
* restart case.
*/
static bool
can_cut_index_handle_prims(struct gl_context *ctx,
const struct _mesa_prim *prim,
GLuint nr_prims,
const struct _mesa_index_buffer *ib)
{
struct brw_context *brw = brw_context(ctx);
if (brw->sol.counting_primitives_generated ||
brw->sol.counting_primitives_written) {
/* Counting primitives generated in hardware is not currently
* supported, so take the software path. We need to investigate
* the *_PRIMITIVES_COUNT registers to allow this to be handled
* entirely in hardware.
*/
return false;
}
if (!can_cut_index_handle_restart_index(ctx, ib)) {
/* The primitive restart index can't be handled, so take
* the software path
*/
return false;
}
for ( ; nr_prims > 0; nr_prims--) {
switch(prim->mode) {
case GL_POINTS:
case GL_LINES:
case GL_LINE_STRIP:
case GL_TRIANGLES:
case GL_TRIANGLE_STRIP:
/* Cut index supports these primitive types */
break;
default:
/* Cut index does not support these primitive types */
//case GL_LINE_LOOP:
//case GL_TRIANGLE_FAN:
//case GL_QUADS:
//case GL_QUAD_STRIP:
//case GL_POLYGON:
return false;
}
}
return true;
}
bool
brw_fs_precompile(struct gl_context *ctx,
struct gl_shader_program *shader_prog,
struct gl_program *prog)
{
struct brw_context *brw = brw_context(ctx);
struct brw_wm_prog_key key;
struct gl_fragment_program *fp = (struct gl_fragment_program *) prog;
struct brw_fragment_program *bfp = brw_fragment_program(fp);
memset(&key, 0, sizeof(key));
if (brw->gen < 6) {
if (fp->UsesKill)
key.iz_lookup |= IZ_PS_KILL_ALPHATEST_BIT;
if (fp->Base.OutputsWritten & BITFIELD64_BIT(FRAG_RESULT_DEPTH))
key.iz_lookup |= IZ_PS_COMPUTES_DEPTH_BIT;
/* Just assume depth testing. */
key.iz_lookup |= IZ_DEPTH_TEST_ENABLE_BIT;
key.iz_lookup |= IZ_DEPTH_WRITE_ENABLE_BIT;
}
if (brw->gen < 6 || _mesa_bitcount_64(fp->Base.InputsRead &
BRW_FS_VARYING_INPUT_MASK) > 16)
key.input_slots_valid = fp->Base.InputsRead | VARYING_BIT_POS;
brw_setup_tex_for_precompile(brw, &key.tex, &fp->Base);
key.nr_color_regions = _mesa_bitcount_64(fp->Base.OutputsWritten &
~(BITFIELD64_BIT(FRAG_RESULT_DEPTH) |
BITFIELD64_BIT(FRAG_RESULT_SAMPLE_MASK)));
key.program_string_id = bfp->id;
uint32_t old_prog_offset = brw->wm.base.prog_offset;
struct brw_wm_prog_data *old_prog_data = brw->wm.prog_data;
bool success = brw_codegen_wm_prog(brw, shader_prog, bfp, &key);
brw->wm.base.prog_offset = old_prog_offset;
brw->wm.prog_data = old_prog_data;
return success;
}
static void
intel_image_target_texture_2d(struct gl_context *ctx, GLenum target,
struct gl_texture_object *texObj,
struct gl_texture_image *texImage,
GLeglImageOES image_handle)
{
struct brw_context *brw = brw_context(ctx);
__DRIscreen *screen;
__DRIimage *image;
screen = brw->intelScreen->driScrnPriv;
image = screen->dri2.image->lookupEGLImage(screen, image_handle,
screen->loaderPrivate);
if (image == NULL)
return;
/* We support external textures only for EGLImages created with
* EGL_EXT_image_dma_buf_import. We may lift that restriction in the future.
*/
if (target == GL_TEXTURE_EXTERNAL_OES && !image->dma_buf_imported) {
_mesa_error(ctx, GL_INVALID_OPERATION,
"glEGLImageTargetTexture2DOES(external target is enabled only "
"for images created with EGL_EXT_image_dma_buf_import");
return;
}
/* Disallow depth/stencil textures: we don't have a way to pass the
* separate stencil miptree of a GL_DEPTH_STENCIL texture through.
*/
if (image->has_depthstencil) {
_mesa_error(ctx, GL_INVALID_OPERATION, __func__);
return;
}
/* Disable creation of the texture's aux buffers because the driver exposes
* no EGL API to manage them. That is, there is no API for resolving the aux
* buffer's content to the main buffer nor for invalidating the aux buffer's
* content.
*/
intel_set_texture_image_bo(ctx, texImage, image->bo,
target, image->internal_format,
image->format, image->offset,
image->width, image->height,
image->pitch,
image->tile_x, image->tile_y,
MIPTREE_LAYOUT_DISABLE_AUX);
}
static void
intel_image_target_renderbuffer_storage(struct gl_context *ctx,
struct gl_renderbuffer *rb,
void *image_handle)
{
struct brw_context *brw = brw_context(ctx);
struct intel_renderbuffer *irb;
__DRIscreen *screen;
__DRIimage *image;
screen = brw->intelScreen->driScrnPriv;
image = screen->dri2.image->lookupEGLImage(screen, image_handle,
screen->loaderPrivate);
if (image == NULL)
return;
/* __DRIimage is opaque to the core so it has to be checked here */
switch (image->format) {
case MESA_FORMAT_RGBA8888_REV:
_mesa_error(ctx, GL_INVALID_OPERATION,
"glEGLImageTargetRenderbufferStorage(unsupported image format");
return;
break;
default:
break;
}
irb = intel_renderbuffer(rb);
intel_miptree_release(&irb->mt);
irb->mt = intel_miptree_create_for_bo(brw,
image->region->bo,
image->format,
image->offset,
image->region->width,
image->region->height,
image->region->pitch,
image->region->tiling);
if (!irb->mt)
return;
rb->InternalFormat = image->internal_format;
rb->Width = image->region->width;
rb->Height = image->region->height;
rb->Format = image->format;
rb->_BaseFormat = _mesa_base_fbo_format(ctx, image->internal_format);
rb->NeedsFinishRenderTexture = true;
}
static GLboolean
intel_texture_view(struct gl_context *ctx,
struct gl_texture_object *texObj,
struct gl_texture_object *origTexObj)
{
struct brw_context *brw = brw_context(ctx);
struct intel_texture_object *intel_tex = intel_texture_object(texObj);
struct intel_texture_object *intel_orig_tex = intel_texture_object(origTexObj);
assert(intel_orig_tex->mt);
intel_miptree_reference(&intel_tex->mt, intel_orig_tex->mt);
/* Since we can only make views of immutable-format textures,
* we can assume that everything is in origTexObj's miptree.
*
* Mesa core has already made us a copy of all the teximage objects,
* except it hasn't copied our mt pointers, etc.
*/
const int numFaces = _mesa_num_tex_faces(texObj->Target);
const int numLevels = texObj->NumLevels;
int face;
int level;
for (face = 0; face < numFaces; face++) {
for (level = 0; level < numLevels; level++) {
struct gl_texture_image *image = texObj->Image[face][level];
struct intel_texture_image *intel_image = intel_texture_image(image);
intel_miptree_reference(&intel_image->mt, intel_orig_tex->mt);
}
}
/* The miptree is in a validated state, so no need to check later. */
intel_tex->needs_validate = false;
intel_tex->validated_first_level = 0;
intel_tex->validated_last_level = numLevels - 1;
/* Set the validated texture format, with the same adjustments that
* would have been applied to determine the underlying texture's
* mt->format.
*/
intel_tex->_Format = intel_depth_format_for_depthstencil_format(
intel_lower_compressed_format(brw, texObj->Image[0][0]->TexFormat));
return GL_TRUE;
}
void brw_draw_prims( struct gl_context *ctx,
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 brw_context *brw = brw_context(ctx);
const struct gl_client_array **arrays = ctx->Array._DrawArrays;
if (!_mesa_check_conditional_render(ctx))
return;
/* Handle primitive restart if needed */
if (brw_handle_primitive_restart(ctx, prims, nr_prims, ib)) {
/* The draw was handled, so we can exit now */
return;
}
/* If we're going to have to upload any of the user's vertex arrays, then
* get the minimum and maximum of their index buffer so we know what range
* to upload.
*/
if (!vbo_all_varyings_in_vbos(arrays) && !index_bounds_valid)
vbo_get_minmax_indices(ctx, prims, ib, &min_index, &max_index, nr_prims);
/* Do GL_SELECT and GL_FEEDBACK rendering using swrast, even though it
* won't support all the extensions we support.
*/
if (ctx->RenderMode != GL_RENDER) {
perf_debug("%s render mode not supported in hardware\n",
_mesa_lookup_enum_by_nr(ctx->RenderMode));
_swsetup_Wakeup(ctx);
_tnl_wakeup(ctx);
_tnl_draw_prims(ctx, arrays, prims, nr_prims, ib, min_index, max_index);
return;
}
/* 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);
}
static void *brw_create_fs_state( struct pipe_context *pipe,
const struct pipe_shader_state *shader )
{
struct brw_context *brw = brw_context(pipe);
struct brw_fragment_shader *fs;
int i;
fs = CALLOC_STRUCT(brw_fragment_shader);
if (fs == NULL)
return NULL;
/* Duplicate tokens, scan shader
*/
fs->id = brw->program_id++;
fs->has_flow_control = has_flow_control(&fs->info);
fs->tokens = tgsi_dup_tokens(shader->tokens);
if (fs->tokens == NULL)
goto fail;
tgsi_scan_shader(fs->tokens, &fs->info);
scan_immediates(fs->tokens, &fs->info, &fs->immediates);
fs->signature.nr_inputs = fs->info.num_inputs;
for (i = 0; i < fs->info.num_inputs; i++) {
fs->signature.input[i].interp = fs->info.input_interpolate[i];
fs->signature.input[i].semantic = fs->info.input_semantic_name[i];
fs->signature.input[i].semantic_index = fs->info.input_semantic_index[i];
}
for (i = 0; i < fs->info.num_inputs; i++)
if (fs->info.input_semantic_name[i] == TGSI_SEMANTIC_POSITION)
fs->uses_depth = 1;
if (fs->info.uses_kill)
fs->iz_lookup |= IZ_PS_KILL_ALPHATEST_BIT;
if (fs->info.writes_z)
fs->iz_lookup |= IZ_PS_COMPUTES_DEPTH_BIT;
return (void *)fs;
fail:
FREE(fs);
return NULL;
}
/**
* Called by ctx->Driver.Clear.
*/
static void
brw_clear(struct gl_context *ctx, GLbitfield mask)
{
struct brw_context *brw = brw_context(ctx);
struct intel_context *intel = &brw->intel;
struct gl_framebuffer *fb = ctx->DrawBuffer;
bool partial_clear = ctx->Scissor.Enabled && !noop_scissor(ctx, fb);
if (!_mesa_check_conditional_render(ctx))
return;
if (mask & (BUFFER_BIT_FRONT_LEFT | BUFFER_BIT_FRONT_RIGHT)) {
intel->front_buffer_dirty = true;
}
intel_prepare_render(intel);
brw_workaround_depthstencil_alignment(brw, partial_clear ? 0 : mask);
if (mask & BUFFER_BIT_DEPTH) {
if (brw_fast_clear_depth(ctx)) {
DBG("fast clear: depth\n");
mask &= ~BUFFER_BIT_DEPTH;
}
}
GLbitfield tri_mask = mask & (BUFFER_BITS_COLOR |
BUFFER_BIT_STENCIL |
BUFFER_BIT_DEPTH);
if (tri_mask) {
debug_mask("tri", tri_mask);
mask &= ~tri_mask;
if (ctx->API == API_OPENGLES) {
_mesa_meta_Clear(&intel->ctx, tri_mask);
} else {
_mesa_meta_glsl_Clear(&intel->ctx, tri_mask);
}
}
/* Any strange buffers get passed off to swrast */
if (mask) {
debug_mask("swrast", mask);
_swrast_Clear(ctx, mask);
}
}
void
intelReadPixels(struct gl_context * ctx,
GLint x, GLint y, GLsizei width, GLsizei height,
GLenum format, GLenum type,
const struct gl_pixelstore_attrib *pack, GLvoid * pixels)
{
bool ok;
struct brw_context *brw = brw_context(ctx);
bool dirty;
DBG("%s\n", __func__);
/* Reading pixels wont dirty the front buffer, so reset the dirty
* flag after calling intel_prepare_render().
*/
dirty = brw->front_buffer_dirty;
intel_prepare_render(brw);
brw->front_buffer_dirty = dirty;
if (_mesa_is_bufferobj(pack->BufferObj)) {
if (intel_readpixels_blorp(ctx, x, y, width, height,
format, type, pixels, pack))
return;
perf_debug("%s: fallback to CPU mapping in PBO case\n", __func__);
}
ok = intel_readpixels_tiled_memcpy(ctx, x, y, width, height,
format, type, pixels, pack);
if(ok)
return;
/* Update Mesa state before calling _mesa_readpixels().
* XXX this may not be needed since ReadPixels no longer uses the
* span code.
*/
if (ctx->NewState)
_mesa_update_state(ctx);
_mesa_readpixels(ctx, x, y, width, height, format, type, pack, pixels);
/* There's an intel_prepare_render() call in intelSpanRenderStart(). */
brw->front_buffer_dirty = dirty;
}
static void meta_depth_replace( struct intel_context *intel )
{
struct brw_context *brw = brw_context(&intel->ctx);
/* ctx->Driver.Enable( ctx, GL_DEPTH_TEST, GL_TRUE )
* ctx->Driver.DepthMask( ctx, GL_TRUE )
*/
brw->metaops.attribs.Depth->Test = GL_TRUE;
brw->metaops.attribs.Depth->Mask = GL_TRUE;
brw->state.dirty.mesa |= _NEW_DEPTH;
/* ctx->Driver.DepthFunc( ctx, GL_ALWAYS )
*/
brw->metaops.attribs.Depth->Func = GL_ALWAYS;
brw->state.dirty.mesa |= _NEW_DEPTH;
}
static const GLubyte *
intelGetString(struct gl_context * ctx, GLenum name)
{
const struct brw_context *const brw = brw_context(ctx);
switch (name) {
case GL_VENDOR:
return (GLubyte *) brw_vendor_string;
case GL_RENDERER:
return
(GLubyte *) brw_get_renderer_string(brw->intelScreen->deviceID);
default:
return NULL;
}
}
/**
* Check if the hardware's cut index support can handle the primitive
* restart case.
*/
static bool
can_cut_index_handle_prims(struct gl_context *ctx,
const struct _mesa_prim *prim,
GLuint nr_prims,
const struct _mesa_index_buffer *ib)
{
struct brw_context *brw = brw_context(ctx);
/* Otherwise Haswell can do it all. */
if (brw->gen >= 8 || brw->is_haswell)
return true;
if (!can_cut_index_handle_restart_index(ctx, ib)) {
/* The primitive restart index can't be handled, so take
* the software path
*/
return false;
}
for (int i = 0; i < nr_prims; i++) {
switch (prim[i].mode) {
case GL_POINTS:
case GL_LINES:
case GL_LINE_STRIP:
case GL_TRIANGLES:
case GL_TRIANGLE_STRIP:
case GL_LINES_ADJACENCY:
case GL_LINE_STRIP_ADJACENCY:
case GL_TRIANGLES_ADJACENCY:
case GL_TRIANGLE_STRIP_ADJACENCY:
/* Cut index supports these primitive types */
break;
default:
/* Cut index does not support these primitive types */
//case GL_LINE_LOOP:
//case GL_TRIANGLE_FAN:
//case GL_QUADS:
//case GL_QUAD_STRIP:
//case GL_POLYGON:
return false;
}
}
return true;
}
static
void brw_update_texture_surface( GLcontext *ctx,
GLuint unit,
struct brw_surface_state *surf )
{
struct intel_context *intel = intel_context(ctx);
struct brw_context *brw = brw_context(ctx);
struct gl_texture_object *tObj = brw->attribs.Texture->Unit[unit]._Current;
struct intel_texture_object *intelObj = intel_texture_object(tObj);
struct gl_texture_image *firstImage = tObj->Image[0][intelObj->firstLevel];
memset(surf, 0, sizeof(*surf));
surf->ss0.mipmap_layout_mode = BRW_SURFACE_MIPMAPLAYOUT_BELOW;
surf->ss0.surface_type = translate_tex_target(tObj->Target);
surf->ss0.surface_format = translate_tex_format(firstImage->TexFormat->MesaFormat, tObj->DepthMode);
/* This is ok for all textures with channel width 8bit or less:
*/
/* surf->ss0.data_return_format = BRW_SURFACERETURNFORMAT_S1; */
/* BRW_NEW_LOCK */
surf->ss1.base_addr = bmBufferOffset(intel,
intelObj->mt->region->buffer);
surf->ss2.mip_count = intelObj->lastLevel - intelObj->firstLevel;
surf->ss2.width = firstImage->Width - 1;
surf->ss2.height = firstImage->Height - 1;
surf->ss3.tile_walk = BRW_TILEWALK_XMAJOR;
surf->ss3.tiled_surface = intelObj->mt->region->tiled; /* always zero */
surf->ss3.pitch = (intelObj->mt->pitch * intelObj->mt->cpp) - 1;
surf->ss3.depth = firstImage->Depth - 1;
surf->ss4.min_lod = 0;
if (tObj->Target == GL_TEXTURE_CUBE_MAP) {
surf->ss0.cube_pos_x = 1;
surf->ss0.cube_pos_y = 1;
surf->ss0.cube_pos_z = 1;
surf->ss0.cube_neg_x = 1;
surf->ss0.cube_neg_y = 1;
surf->ss0.cube_neg_z = 1;
}
}
void
brw_begin_transform_feedback(struct gl_context *ctx, GLenum mode,
struct gl_transform_feedback_object *obj)
{
struct brw_context *brw = brw_context(ctx);
const struct gl_shader_program *vs_prog =
ctx->Shader.CurrentVertexProgram;
const struct gl_transform_feedback_info *linked_xfb_info =
&vs_prog->LinkedTransformFeedback;
struct gl_transform_feedback_object *xfb_obj =
ctx->TransformFeedback.CurrentObject;
assert(brw->gen == 6);
/* Compute the maximum number of vertices that we can write without
* overflowing any of the buffers currently being used for feedback.
*/
unsigned max_index
= _mesa_compute_max_transform_feedback_vertices(xfb_obj,
linked_xfb_info);
/* 3DSTATE_GS_SVB_INDEX is non-pipelined. */
intel_emit_post_sync_nonzero_flush(brw);
/* Initialize the SVBI 0 register to zero and set the maximum index. */
BEGIN_BATCH(4);
OUT_BATCH(_3DSTATE_GS_SVB_INDEX << 16 | (4 - 2));
OUT_BATCH(0); /* SVBI 0 */
OUT_BATCH(0); /* starting index */
OUT_BATCH(max_index);
ADVANCE_BATCH();
/* Initialize the rest of the unused streams to sane values. Otherwise,
* they may indicate that there is no room to write data and prevent
* anything from happening at all.
*/
for (int i = 1; i < 4; i++) {
BEGIN_BATCH(4);
OUT_BATCH(_3DSTATE_GS_SVB_INDEX << 16 | (4 - 2));
OUT_BATCH(i << SVB_INDEX_SHIFT);
OUT_BATCH(0); /* starting index */
OUT_BATCH(0xffffffff);
ADVANCE_BATCH();
}
}
/**
* Query information about GPU resets observed by this context
*
* Called via \c dd_function_table::GetGraphicsResetStatus.
*/
GLenum
brw_get_graphics_reset_status(struct gl_context *ctx)
{
struct brw_context *brw = brw_context(ctx);
int err;
uint32_t reset_count;
uint32_t active;
uint32_t pending;
/* If hardware contexts are not being used (or
* DRM_IOCTL_I915_GET_RESET_STATS is not supported), this function should
* not be accessible.
*/
assert(brw->hw_ctx != NULL);
#if 0
/* This is waiting until the kernel code can be merged and a new libdrm
* actually released.
*/
err = drm_intel_get_reset_stats(brw->hw_ctx, &reset_count, &active,
&pending);
if (err)
return GL_NO_ERROR;
#else
return GL_NO_ERROR;
#endif
/* A reset was observed while a batch from this context was executing.
* Assume that this context was at fault.
*/
if (active != 0)
return GL_GUILTY_CONTEXT_RESET_ARB;
/* A reset was observed while a batch from this context was in progress,
* but the batch was not executing. In this case, assume that the context
* was not at fault.
*/
if (pending != 0)
return GL_INNOCENT_CONTEXT_RESET_ARB;
/* FINISHME: Should we report anything if reset_count > brw->reset_count?
*/
return GL_NO_ERROR;
}
static void
intel_texture_barrier(struct gl_context *ctx)
{
struct brw_context *brw = brw_context(ctx);
const struct gen_device_info *devinfo = &brw->screen->devinfo;
if (devinfo->gen >= 6) {
brw_emit_pipe_control_flush(brw,
PIPE_CONTROL_DEPTH_CACHE_FLUSH |
PIPE_CONTROL_RENDER_TARGET_FLUSH |
PIPE_CONTROL_CS_STALL);
brw_emit_pipe_control_flush(brw,
PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE);
} else {
brw_emit_mi_flush(brw);
}
}
static void brw_bind_fs_state( struct pipe_context *pipe, void *prog )
{
struct brw_fragment_shader *fs = (struct brw_fragment_shader *)prog;
struct brw_context *brw = brw_context(pipe);
if (brw->curr.fragment_shader == fs)
return;
if (brw->curr.fragment_shader == NULL ||
fs == NULL ||
memcmp(&brw->curr.fragment_shader->signature, &fs->signature,
brw_fs_signature_size(&fs->signature)) != 0) {
brw->state.dirty.mesa |= PIPE_NEW_FRAGMENT_SIGNATURE;
}
brw->curr.fragment_shader = fs;
brw->state.dirty.mesa |= PIPE_NEW_FRAGMENT_SHADER;
}
static void
brw_gl_fence_sync(struct gl_context *ctx, struct gl_sync_object *_sync,
GLenum condition, GLbitfield flags)
{
struct brw_context *brw = brw_context(ctx);
struct brw_gl_sync *sync = (struct brw_gl_sync *) _sync;
/* brw_fence_insert_locked() assumes it must do a complete flush */
assert(condition == GL_SYNC_GPU_COMMANDS_COMPLETE);
brw_fence_init(brw, &sync->fence, BRW_FENCE_TYPE_BO_WAIT);
if (!brw_fence_insert_locked(brw, &sync->fence)) {
/* FIXME: There exists no way to report a GL error here. If an error
* occurs, continue silently and hope for the best.
*/
}
}
static void
brw_blend_barrier(struct gl_context *ctx)
{
struct brw_context *brw = brw_context(ctx);
if (!ctx->Extensions.MESA_shader_framebuffer_fetch) {
if (brw->gen >= 6) {
brw_emit_pipe_control_flush(brw,
PIPE_CONTROL_RENDER_TARGET_FLUSH |
PIPE_CONTROL_CS_STALL);
brw_emit_pipe_control_flush(brw,
PIPE_CONTROL_TEXTURE_CACHE_INVALIDATE);
} else {
brw_emit_pipe_control_flush(brw,
PIPE_CONTROL_RENDER_TARGET_FLUSH);
}
}
}
static void
intelDRI2Flush(__DRIdrawable *drawable)
{
GET_CURRENT_CONTEXT(ctx);
struct brw_context *brw = brw_context(ctx);
if (brw == NULL)
return;
intel_resolve_for_dri2_flush(brw, drawable);
brw->need_throttle = true;
if (brw->batch.used)
intel_batchbuffer_flush(brw);
if (INTEL_DEBUG & DEBUG_AUB) {
aub_dump_bmp(ctx);
}
}
struct gl_transform_feedback_object *
brw_new_transform_feedback(struct gl_context *ctx, GLuint name)
{
struct brw_context *brw = brw_context(ctx);
struct brw_transform_feedback_object *brw_obj =
CALLOC_STRUCT(brw_transform_feedback_object);
if (!brw_obj)
return NULL;
_mesa_init_transform_feedback_object(&brw_obj->base, name);
brw_obj->offset_bo =
drm_intel_bo_alloc(brw->bufmgr, "transform feedback offsets", 16, 64);
brw_obj->prim_count_bo =
drm_intel_bo_alloc(brw->bufmgr, "xfb primitive counts", 4096, 64);
return &brw_obj->base;
}
/**
* called from intelDestroyContext()
*/
static void brw_destroy_context( struct intel_context *intel )
{
struct brw_context *brw = brw_context(&intel->ctx);
brw_destroy_state(brw);
brw_draw_destroy( brw );
ralloc_free(brw->wm.compile_data);
dri_bo_release(&brw->curbe.curbe_bo);
dri_bo_release(&brw->vs.const_bo);
dri_bo_release(&brw->wm.const_bo);
free(brw->curbe.last_buf);
free(brw->curbe.next_buf);
drm_intel_gem_context_destroy(intel->hw_ctx);
}
static struct gl_program *brwNewProgram( struct gl_context *ctx,
GLenum target,
GLuint id )
{
struct brw_context *brw = brw_context(ctx);
switch (target) {
case GL_VERTEX_PROGRAM_ARB:
case GL_TESS_CONTROL_PROGRAM_NV:
case GL_TESS_EVALUATION_PROGRAM_NV:
case GL_GEOMETRY_PROGRAM_NV:
case GL_COMPUTE_PROGRAM_NV: {
struct brw_program *prog = rzalloc(NULL, struct brw_program);
if (prog) {
prog->id = get_new_program_id(brw->screen);
return _mesa_init_gl_program(&prog->program, target, id);
}
else
return NULL;
}
case GL_FRAGMENT_PROGRAM_ARB: {
struct brw_program *prog;
if (brw->gen < 6) {
struct gen4_fragment_program *g4_prog =
rzalloc(NULL, struct gen4_fragment_program);
prog = &g4_prog->base;
} else {
prog = rzalloc(NULL, struct brw_program);
}
if (prog) {
prog->id = get_new_program_id(brw->screen);
return _mesa_init_gl_program(&prog->program, target, id);
}
else
return NULL;
}
default:
unreachable("Unsupported target in brwNewProgram()");
}
/**
* called from intelFlushBatchLocked
*/
static void brw_new_batch( struct intel_context *intel )
{
struct brw_context *brw = brw_context(&intel->ctx);
/* If the kernel supports hardware contexts, then most hardware state is
* preserved between batches; we only need to re-emit state that is required
* to be in every batch. Otherwise we need to re-emit all the state that
* would otherwise be stored in the context (which for all intents and
* purposes means everything).
*/
if (intel->hw_ctx == NULL)
brw->state.dirty.brw |= BRW_NEW_CONTEXT;
brw->state.dirty.brw |= BRW_NEW_BATCH;
/* Assume that the last command before the start of our batch was a
* primitive, for safety.
*/
intel->batch.need_workaround_flush = true;
brw->state_batch_count = 0;
/* Gen7 needs to track what the real transform feedback vertex count was at
* the start of the batch, since the kernel will be resetting the offset to
* 0.
*/
brw->sol.offset_0_batch_start = brw->sol.svbi_0_starting_index;
brw->ib.type = -1;
/* Mark that the current program cache BO has been used by the GPU.
* It will be reallocated if we need to put new programs in for the
* next batch.
*/
brw->cache.bo_used_by_gpu = true;
/* We need to periodically reap the shader time results, because rollover
* happens every few seconds. We also want to see results every once in a
* while, because many programs won't cleanly destroy our context, so the
* end-of-run printout may not happen.
*/
if (INTEL_DEBUG & DEBUG_SHADER_TIME)
brw_collect_and_report_shader_time(brw);
}
/* called from intelFlushBatchLocked
*/
static void brw_lost_hardware( struct intel_context *intel )
{
struct brw_context *brw = brw_context(&intel->ctx);
/* Note that we effectively lose the context after this.
*
* Setting this flag provokes a state buffer wrap and also flushes
* the hardware caches.
*/
brw->state.dirty.brw |= BRW_NEW_CONTEXT;
/* Which means there shouldn't be any commands already queued:
*/
assert(intel->batch->ptr == intel->batch->map + intel->batch->offset);
brw->state.dirty.mesa |= ~0;
brw->state.dirty.brw |= ~0;
brw->state.dirty.cache |= ~0;
}
static void leave_meta_state( struct intel_context *intel )
{
GLcontext *ctx = &intel->ctx;
struct brw_context *brw = brw_context(ctx);
restore_attribs(brw);
ctx->DrawBuffer->_ColorDrawBufferMask[0] = brw->metaops.restore_draw_mask;
ctx->FragmentProgram.Current = brw->metaops.restore_fp;
brw->state.draw_region = brw->metaops.saved_draw_region;
brw->state.depth_region = brw->metaops.saved_depth_region;
brw->metaops.saved_draw_region = NULL;
brw->metaops.saved_depth_region = NULL;
brw->metaops.active = 0;
brw->state.dirty.mesa |= _NEW_BUFFERS;
brw->state.dirty.brw |= BRW_NEW_METAOPS;
}
/**
* \see dd_function_table::UnmapRenderbuffer
*/
static void
intel_unmap_renderbuffer(struct gl_context *ctx,
struct gl_renderbuffer *rb)
{
struct brw_context *brw = brw_context(ctx);
struct swrast_renderbuffer *srb = (struct swrast_renderbuffer *)rb;
struct intel_renderbuffer *irb = intel_renderbuffer(rb);
DBG("%s: rb %d (%s)\n", __FUNCTION__,
rb->Name, _mesa_get_format_name(rb->Format));
if (srb->Buffer) {
/* this is a malloc'd renderbuffer (accum buffer) */
/* nothing to do */
return;
}
intel_miptree_unmap(brw, irb->mt, irb->mt_level, irb->mt_layer);
}
static void
intel_get_tex_sub_image(struct gl_context *ctx,
GLint xoffset, GLint yoffset, GLint zoffset,
GLsizei width, GLsizei height, GLint depth,
GLenum format, GLenum type, GLvoid *pixels,
struct gl_texture_image *texImage)
{
struct brw_context *brw = brw_context(ctx);
bool ok;
DBG("%s\n", __func__);
if (_mesa_is_bufferobj(ctx->Pack.BufferObj)) {
if (_mesa_meta_pbo_GetTexSubImage(ctx, 3, texImage,
xoffset, yoffset, zoffset,
width, height, depth, format, type,
pixels, &ctx->Pack)) {
/* Flush to guarantee coherency between the render cache and other
* caches the PBO could potentially be bound to after this point.
* See the related comment in intelReadPixels() for a more detailed
* explanation.
*/
brw_emit_mi_flush(brw);
return;
}
perf_debug("%s: fallback to CPU mapping in PBO case\n", __func__);
}
ok = intel_gettexsubimage_tiled_memcpy(ctx, texImage, xoffset, yoffset,
width, height,
format, type, pixels, &ctx->Pack);
if(ok)
return;
_mesa_meta_GetTexSubImage(ctx, xoffset, yoffset, zoffset,
width, height, depth,
format, type, pixels, texImage);
DBG("%s - DONE\n", __func__);
}
/**
* Called by glFramebufferTexture[123]DEXT() (and other places) to
* prepare for rendering into texture memory. This might be called
* many times to choose different texture levels, cube faces, etc
* before intel_finish_render_texture() is ever called.
*/
static void
intel_render_texture(struct gl_context * ctx,
struct gl_framebuffer *fb,
struct gl_renderbuffer_attachment *att)
{
struct brw_context *brw = brw_context(ctx);
struct gl_renderbuffer *rb = att->Renderbuffer;
struct intel_renderbuffer *irb = intel_renderbuffer(rb);
struct gl_texture_image *image = rb->TexImage;
struct intel_texture_image *intel_image = intel_texture_image(image);
struct intel_mipmap_tree *mt = intel_image->mt;
int layer;
(void) fb;
if (att->CubeMapFace > 0) {
assert(att->Zoffset == 0);
layer = att->CubeMapFace;
} else {
layer = att->Zoffset;
}
if (!intel_image->mt) {
/* Fallback on drawing to a texture that doesn't have a miptree
* (has a border, width/height 0, etc.)
*/
_swrast_render_texture(ctx, fb, att);
return;
}
intel_miptree_check_level_layer(mt, att->TextureLevel, layer);
if (!intel_renderbuffer_update_wrapper(brw, irb, image, layer)) {
_swrast_render_texture(ctx, fb, att);
return;
}
DBG("Begin render %s texture tex=%u w=%d h=%d d=%d refcount=%d\n",
_mesa_get_format_name(image->TexFormat),
att->Texture->Name, image->Width, image->Height, image->Depth,
rb->RefCount);
}