brw_blorp_upload_shader(struct blorp_context *blorp,
const void *key, uint32_t key_size,
const void *kernel, uint32_t kernel_size,
- const void *prog_data, uint32_t prog_data_size,
+ const struct brw_stage_prog_data *prog_data,
+ uint32_t prog_data_size,
uint32_t *kernel_out, void *prog_data_out)
{
struct brw_context *brw = blorp->driver_ctx;
{
blorp_init(&brw->blorp, brw, &brw->isl_dev);
- brw->blorp.compiler = brw->intelScreen->compiler;
+ brw->blorp.compiler = brw->screen->compiler;
switch (brw->gen) {
case 6:
} else {
*offset = intel_miptree_get_aligned_offset(mt,
mt->level[lod].level_x,
- mt->level[lod].level_y,
- false);
+ mt->level[lod].level_y);
}
surf->logical_level0_px.width = minify(surf->logical_level0_px.width, lod);
struct blorp_surf *surf,
struct intel_mipmap_tree *mt,
bool is_render_target,
+ uint32_t safe_aux_usage,
unsigned *level,
+ unsigned start_layer, unsigned num_layers,
struct isl_surf tmp_surfs[2])
{
+ if (mt->msaa_layout == INTEL_MSAA_LAYOUT_UMS ||
+ mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS) {
+ const unsigned num_samples = MAX2(1, mt->num_samples);
+ for (unsigned i = 0; i < num_layers; i++) {
+ for (unsigned s = 0; s < num_samples; s++) {
+ const unsigned phys_layer = (start_layer + i) * num_samples + s;
+ intel_miptree_check_level_layer(mt, *level, phys_layer);
+ }
+ }
+ } else {
+ for (unsigned i = 0; i < num_layers; i++)
+ intel_miptree_check_level_layer(mt, *level, start_layer + i);
+ }
+
intel_miptree_get_isl_surf(brw, mt, &tmp_surfs[0]);
surf->surf = &tmp_surfs[0];
surf->addr = (struct blorp_address) {
struct isl_surf *aux_surf = &tmp_surfs[1];
intel_miptree_get_aux_isl_surf(brw, mt, aux_surf, &surf->aux_usage);
- /* For textures that are in the RESOLVED state, we ignore the MCS */
- if (mt->mcs_mt && !is_render_target &&
- mt->fast_clear_state == INTEL_FAST_CLEAR_STATE_RESOLVED)
- surf->aux_usage = ISL_AUX_USAGE_NONE;
+ if (surf->aux_usage != ISL_AUX_USAGE_NONE) {
+ if (surf->aux_usage == ISL_AUX_USAGE_HIZ) {
+ /* If we're not going to use it as a depth buffer, resolve HiZ */
+ if (!(safe_aux_usage & (1 << ISL_AUX_USAGE_HIZ))) {
+ for (unsigned i = 0; i < num_layers; i++) {
+ intel_miptree_slice_resolve_depth(brw, mt, *level,
+ start_layer + i);
+
+ /* If we're rendering to it then we'll need a HiZ resolve once
+ * we're done before we can use it with HiZ again.
+ */
+ if (is_render_target)
+ intel_miptree_slice_set_needs_hiz_resolve(mt, *level,
+ start_layer + i);
+ }
+ surf->aux_usage = ISL_AUX_USAGE_NONE;
+ }
+ } else if (!(safe_aux_usage & (1 << surf->aux_usage))) {
+ uint32_t flags = 0;
+ if (safe_aux_usage & (1 << ISL_AUX_USAGE_CCS_E))
+ flags |= INTEL_MIPTREE_IGNORE_CCS_E;
+
+ intel_miptree_resolve_color(brw, mt,
+ *level, start_layer, num_layers, flags);
+
+ assert(!intel_miptree_has_color_unresolved(mt, *level, 1,
+ start_layer, num_layers));
+ surf->aux_usage = ISL_AUX_USAGE_NONE;
+ }
+ }
+
+ if (is_render_target)
+ intel_miptree_used_for_rendering(brw, mt, *level,
+ start_layer, num_layers);
if (surf->aux_usage != ISL_AUX_USAGE_NONE) {
/* We only really need a clear color if we also have an auxiliary
.write_domain = is_render_target ? I915_GEM_DOMAIN_RENDER : 0,
};
- if (mt->mcs_mt) {
- surf->aux_addr.buffer = mt->mcs_mt->bo;
- surf->aux_addr.offset = mt->mcs_mt->offset;
+ if (mt->mcs_buf) {
+ surf->aux_addr.buffer = mt->mcs_buf->bo;
+ surf->aux_addr.offset = mt->mcs_buf->offset;
} else {
assert(surf->aux_usage == ISL_AUX_USAGE_HIZ);
struct intel_mipmap_tree *hiz_mt = mt->hiz_buf->mt;
}
assert(hiz_mt->pitch == aux_surf->row_pitch);
} else {
- surf->aux_addr.buffer = mt->hiz_buf->bo;
- surf->aux_addr.offset = 0;
+ surf->aux_addr.buffer = mt->hiz_buf->aux_base.bo;
+ surf->aux_addr.offset = mt->hiz_buf->aux_base.offset;
}
}
} else {
return (enum isl_channel_select)((swizzle + 4) & 7);
}
+static unsigned
+physical_to_logical_layer(struct intel_mipmap_tree *mt,
+ unsigned physical_layer)
+{
+ if (mt->num_samples > 1 &&
+ (mt->msaa_layout == INTEL_MSAA_LAYOUT_UMS ||
+ mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS)) {
+ assert(physical_layer % mt->num_samples == 0);
+ return physical_layer / mt->num_samples;
+ } else {
+ return physical_layer;
+ }
+}
+
/**
* Note: if the src (or dst) is a 2D multisample array texture on Gen7+ using
* INTEL_MSAA_LAYOUT_UMS or INTEL_MSAA_LAYOUT_CMS, src_layer (dst_layer) is
GLenum filter, bool mirror_x, bool mirror_y,
bool decode_srgb, bool encode_srgb)
{
- /* Get ready to blit. This includes depth resolving the src and dst
- * buffers if necessary. Note: it's not necessary to do a color resolve on
- * the destination buffer because we use the standard render path to render
- * to destination color buffers, and the standard render path is
- * fast-color-aware.
- */
- intel_miptree_resolve_color(brw, src_mt, INTEL_MIPTREE_IGNORE_CCS_E);
- intel_miptree_slice_resolve_depth(brw, src_mt, src_level, src_layer);
- intel_miptree_slice_resolve_depth(brw, dst_mt, dst_level, dst_layer);
+ /* Blorp operates in logical layers */
+ src_layer = physical_to_logical_layer(src_mt, src_layer);
+ dst_layer = physical_to_logical_layer(dst_mt, dst_layer);
DBG("%s from %dx %s mt %p %d %d (%f,%f) (%f,%f)"
"to %dx %s mt %p %d %d (%f,%f) (%f,%f) (flip %d,%d)\n",
src_format = dst_format = MESA_FORMAT_R_FLOAT32;
}
- intel_miptree_check_level_layer(src_mt, src_level, src_layer);
- intel_miptree_check_level_layer(dst_mt, dst_level, dst_layer);
- intel_miptree_used_for_rendering(dst_mt);
+ uint32_t src_usage_flags = (1 << ISL_AUX_USAGE_MCS);
+ if (src_format == src_mt->format)
+ src_usage_flags |= (1 << ISL_AUX_USAGE_CCS_E);
+
+ uint32_t dst_usage_flags = (1 << ISL_AUX_USAGE_MCS);
+ if (dst_format == dst_mt->format) {
+ dst_usage_flags |= (1 << ISL_AUX_USAGE_CCS_E) |
+ (1 << ISL_AUX_USAGE_CCS_D);
+ }
struct isl_surf tmp_surfs[4];
struct blorp_surf src_surf, dst_surf;
- blorp_surf_for_miptree(brw, &src_surf, src_mt, false,
- &src_level, &tmp_surfs[0]);
- blorp_surf_for_miptree(brw, &dst_surf, dst_mt, true,
- &dst_level, &tmp_surfs[2]);
+ blorp_surf_for_miptree(brw, &src_surf, src_mt, false, src_usage_flags,
+ &src_level, src_layer, 1, &tmp_surfs[0]);
+ blorp_surf_for_miptree(brw, &dst_surf, dst_mt, true, dst_usage_flags,
+ &dst_level, dst_layer, 1, &tmp_surfs[2]);
struct isl_swizzle src_isl_swizzle = {
.r = swizzle_to_scs(GET_SWZ(src_swizzle, 0)),
};
struct blorp_batch batch;
- blorp_batch_init(&brw->blorp, &batch, brw);
+ blorp_batch_init(&brw->blorp, &batch, brw, 0);
blorp_blit(&batch, &src_surf, src_level, src_layer,
brw_blorp_to_isl_format(brw, src_format, false), src_isl_swizzle,
&dst_surf, dst_level, dst_layer,
brw_blorp_to_isl_format(brw, dst_format, true),
+ ISL_SWIZZLE_IDENTITY,
src_x0, src_y0, src_x1, src_y1,
dst_x0, dst_y0, dst_x1, dst_y1,
filter, mirror_x, mirror_y);
blorp_batch_finish(&batch);
+}
+
+void
+brw_blorp_copy_miptrees(struct brw_context *brw,
+ struct intel_mipmap_tree *src_mt,
+ unsigned src_level, unsigned src_layer,
+ struct intel_mipmap_tree *dst_mt,
+ unsigned dst_level, unsigned dst_layer,
+ unsigned src_x, unsigned src_y,
+ unsigned dst_x, unsigned dst_y,
+ unsigned src_width, unsigned src_height)
+{
+ DBG("%s from %dx %s mt %p %d %d (%d,%d) %dx%d"
+ "to %dx %s mt %p %d %d (%d,%d)\n",
+ __func__,
+ src_mt->num_samples, _mesa_get_format_name(src_mt->format), src_mt,
+ src_level, src_layer, src_x, src_y, src_width, src_height,
+ dst_mt->num_samples, _mesa_get_format_name(dst_mt->format), dst_mt,
+ dst_level, dst_layer, dst_x, dst_y);
- intel_miptree_slice_set_needs_hiz_resolve(dst_mt, dst_level, dst_layer);
+ struct isl_surf tmp_surfs[4];
+ struct blorp_surf src_surf, dst_surf;
+ blorp_surf_for_miptree(brw, &src_surf, src_mt, false,
+ (1 << ISL_AUX_USAGE_MCS) |
+ (1 << ISL_AUX_USAGE_CCS_E),
+ &src_level, src_layer, 1, &tmp_surfs[0]);
+ blorp_surf_for_miptree(brw, &dst_surf, dst_mt, true,
+ (1 << ISL_AUX_USAGE_MCS) |
+ (1 << ISL_AUX_USAGE_CCS_E),
+ &dst_level, dst_layer, 1, &tmp_surfs[2]);
- if (intel_miptree_is_lossless_compressed(brw, dst_mt))
- dst_mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_UNRESOLVED;
+ struct blorp_batch batch;
+ blorp_batch_init(&brw->blorp, &batch, brw, 0);
+ blorp_copy(&batch, &src_surf, src_level, src_layer,
+ &dst_surf, dst_level, dst_layer,
+ src_x, src_y, dst_x, dst_y, src_width, src_height);
+ blorp_batch_finish(&batch);
}
static struct intel_mipmap_tree *
return disables;
}
+static unsigned
+irb_logical_mt_layer(struct intel_renderbuffer *irb)
+{
+ return physical_to_logical_layer(irb->mt, irb->mt_layer);
+}
+
static bool
do_single_blorp_clear(struct brw_context *brw, struct gl_framebuffer *fb,
struct gl_renderbuffer *rb, unsigned buf,
if (set_write_disables(irb, ctx->Color.ColorMask[buf], color_write_disable))
can_fast_clear = false;
- if (irb->mt->fast_clear_state == INTEL_FAST_CLEAR_STATE_NO_MCS ||
+ if (irb->mt->aux_disable & INTEL_AUX_DISABLE_CCS ||
!brw_is_color_fast_clear_compatible(brw, irb->mt, &ctx->Color.ClearColor))
can_fast_clear = false;
- const bool is_lossless_compressed = intel_miptree_is_lossless_compressed(
- brw, irb->mt);
+ const unsigned logical_layer = irb_logical_mt_layer(irb);
+ const enum intel_fast_clear_state fast_clear_state =
+ intel_miptree_get_fast_clear_state(irb->mt, irb->mt_level,
+ logical_layer);
+
+ /* Surface state can only record one fast clear color value. Therefore
+ * unless different levels/layers agree on the color it can be used to
+ * represent only single level/layer. Here it will be reserved for the
+ * first slice (level 0, layer 0).
+ */
+ if (irb->layer_count > 1 || irb->mt_level || irb->mt_layer)
+ can_fast_clear = false;
if (can_fast_clear) {
+ union gl_color_union override_color =
+ brw_meta_convert_fast_clear_color(brw, irb->mt,
+ &ctx->Color.ClearColor);
+
/* Record the clear color in the miptree so that it will be
* programmed in SURFACE_STATE by later rendering and resolve
* operations.
*/
const bool color_updated = brw_meta_set_fast_clear_color(
- brw, irb->mt, &ctx->Color.ClearColor);
+ brw, &irb->mt->gen9_fast_clear_color,
+ &override_color);
/* If the buffer is already in INTEL_FAST_CLEAR_STATE_CLEAR, the clear
* is redundant and can be skipped.
*/
- if (!color_updated &&
- irb->mt->fast_clear_state == INTEL_FAST_CLEAR_STATE_CLEAR)
+ if (!color_updated && fast_clear_state == INTEL_FAST_CLEAR_STATE_CLEAR)
return true;
/* If the MCS buffer hasn't been allocated yet, we need to allocate
* it now.
*/
- if (!irb->mt->mcs_mt) {
- assert(!is_lossless_compressed);
+ if (!irb->mt->mcs_buf) {
+ assert(!intel_miptree_is_lossless_compressed(brw, irb->mt));
if (!intel_miptree_alloc_non_msrt_mcs(brw, irb->mt, false)) {
/* MCS allocation failed--probably this will only happen in
* out-of-memory conditions. But in any case, try to recover
}
}
- intel_miptree_used_for_rendering(irb->mt);
+ const unsigned num_layers = fb->MaxNumLayers ? irb->layer_count : 1;
/* We can't setup the blorp_surf until we've allocated the MCS above */
struct isl_surf isl_tmp[2];
struct blorp_surf surf;
unsigned level = irb->mt_level;
- blorp_surf_for_miptree(brw, &surf, irb->mt, true, &level, isl_tmp);
- const unsigned num_layers = fb->MaxNumLayers ? irb->layer_count : 1;
+ blorp_surf_for_miptree(brw, &surf, irb->mt, true,
+ (1 << ISL_AUX_USAGE_MCS) |
+ (1 << ISL_AUX_USAGE_CCS_E) |
+ (1 << ISL_AUX_USAGE_CCS_D),
+ &level, logical_layer, num_layers, isl_tmp);
if (can_fast_clear) {
DBG("%s (fast) to mt %p level %d layers %d+%d\n", __FUNCTION__,
irb->mt, irb->mt_level, irb->mt_layer, num_layers);
struct blorp_batch batch;
- blorp_batch_init(&brw->blorp, &batch, brw);
+ blorp_batch_init(&brw->blorp, &batch, brw, 0);
blorp_fast_clear(&batch, &surf,
(enum isl_format)brw->render_target_format[format],
- level, irb->mt_layer, num_layers, x0, y0, x1, y1);
+ level, logical_layer, num_layers,
+ x0, y0, x1, y1);
blorp_batch_finish(&batch);
/* Now that the fast clear has occurred, put the buffer in
* INTEL_FAST_CLEAR_STATE_CLEAR so that we won't waste time doing
* redundant clears.
*/
- irb->mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_CLEAR;
+ intel_miptree_set_fast_clear_state(brw, irb->mt, irb->mt_level,
+ logical_layer, num_layers,
+ INTEL_FAST_CLEAR_STATE_CLEAR);
} else {
DBG("%s (slow) to mt %p level %d layer %d+%d\n", __FUNCTION__,
irb->mt, irb->mt_level, irb->mt_layer, num_layers);
memcpy(clear_color.f32, ctx->Color.ClearColor.f, sizeof(float) * 4);
struct blorp_batch batch;
- blorp_batch_init(&brw->blorp, &batch, brw);
- blorp_clear(&batch, &surf, level, irb->mt_layer, num_layers,
- x0, y0, x1, y1,
+ blorp_batch_init(&brw->blorp, &batch, brw, 0);
+ blorp_clear(&batch, &surf,
(enum isl_format)brw->render_target_format[format],
+ ISL_SWIZZLE_IDENTITY,
+ level, irb_logical_mt_layer(irb), num_layers,
+ x0, y0, x1, y1,
clear_color, color_write_disable);
blorp_batch_finish(&batch);
-
- if (is_lossless_compressed) {
- /* Compressed buffers can be cleared also using normal rep-clear. In
- * such case they behave such as if they were drawn using normal 3D
- * render pipeline, and we simply mark the mcs as dirty.
- */
- irb->mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_UNRESOLVED;
- }
}
return true;
if (rb == NULL)
continue;
- const unsigned num_layers = fb->MaxNumLayers ? irb->layer_count : 1;
- for (unsigned layer = 0; layer < num_layers; layer++) {
- intel_miptree_check_level_layer(irb->mt, irb->mt_level, layer);
- }
-
if (!do_single_blorp_clear(brw, fb, rb, buf, partial_clear,
encode_srgb)) {
return false;
}
void
-brw_blorp_resolve_color(struct brw_context *brw, struct intel_mipmap_tree *mt)
+brw_blorp_resolve_color(struct brw_context *brw, struct intel_mipmap_tree *mt,
+ unsigned level, unsigned layer)
{
- DBG("%s to mt %p\n", __FUNCTION__, mt);
+ DBG("%s to mt %p level %u layer %u\n", __FUNCTION__, mt, level, layer);
const mesa_format format = _mesa_get_srgb_format_linear(mt->format);
- intel_miptree_check_level_layer(mt, 0 /* level */, 0 /* layer */);
- intel_miptree_used_for_rendering(mt);
-
struct isl_surf isl_tmp[2];
struct blorp_surf surf;
- unsigned level = 0;
- blorp_surf_for_miptree(brw, &surf, mt, true, &level, isl_tmp);
+ blorp_surf_for_miptree(brw, &surf, mt, true,
+ (1 << ISL_AUX_USAGE_CCS_E) |
+ (1 << ISL_AUX_USAGE_CCS_D),
+ &level, layer, 1 /* num_layers */,
+ isl_tmp);
+
+ enum blorp_fast_clear_op resolve_op;
+ if (brw->gen >= 9) {
+ if (surf.aux_usage == ISL_AUX_USAGE_CCS_E)
+ resolve_op = BLORP_FAST_CLEAR_OP_RESOLVE_FULL;
+ else
+ resolve_op = BLORP_FAST_CLEAR_OP_RESOLVE_PARTIAL;
+ } else {
+ assert(surf.aux_usage == ISL_AUX_USAGE_CCS_D);
+ /* Broadwell and earlier do not have a partial resolve */
+ resolve_op = BLORP_FAST_CLEAR_OP_RESOLVE_FULL;
+ }
struct blorp_batch batch;
- blorp_batch_init(&brw->blorp, &batch, brw);
- blorp_ccs_resolve(&batch, &surf,
- brw_blorp_to_isl_format(brw, format, true));
+ blorp_batch_init(&brw->blorp, &batch, brw, 0);
+ blorp_ccs_resolve(&batch, &surf, level, layer,
+ brw_blorp_to_isl_format(brw, format, true),
+ resolve_op);
blorp_batch_finish(&batch);
-
- mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_RESOLVED;
}
static void
gen6_blorp_hiz_exec(struct brw_context *brw, struct intel_mipmap_tree *mt,
unsigned int level, unsigned int layer, enum blorp_hiz_op op)
{
- intel_miptree_check_level_layer(mt, level, layer);
- intel_miptree_used_for_rendering(mt);
-
assert(intel_miptree_level_has_hiz(mt, level));
struct isl_surf isl_tmp[2];
struct blorp_surf surf;
- blorp_surf_for_miptree(brw, &surf, mt, true, &level, isl_tmp);
+ blorp_surf_for_miptree(brw, &surf, mt, true, (1 << ISL_AUX_USAGE_HIZ),
+ &level, layer, 1, isl_tmp);
struct blorp_batch batch;
- blorp_batch_init(&brw->blorp, &batch, brw);
+ blorp_batch_init(&brw->blorp, &batch, brw, 0);
blorp_gen6_hiz_op(&batch, &surf, level, layer, op);
blorp_batch_finish(&batch);
}