#define FILE_DEBUG_FLAG DEBUG_MIPTREE
static void *intel_miptree_map_raw(struct brw_context *brw,
- struct intel_mipmap_tree *mt);
+ struct intel_mipmap_tree *mt,
+ GLbitfield mode);
static void intel_miptree_unmap_raw(struct intel_mipmap_tree *mt);
*/
static enum intel_msaa_layout
compute_msaa_layout(struct brw_context *brw, mesa_format format,
- bool disable_aux_buffers)
+ enum intel_aux_disable aux_disable)
{
/* Prior to Gen7, all MSAA surfaces used IMS layout. */
if (brw->gen < 7)
*/
if (brw->gen == 7 && _mesa_get_format_datatype(format) == GL_INT) {
return INTEL_MSAA_LAYOUT_UMS;
- } else if (disable_aux_buffers) {
+ } else if (aux_disable & INTEL_AUX_DISABLE_MCS) {
/* We can't use the CMS layout because it uses an aux buffer, the MCS
* buffer. So fallback to UMS, which is identical to CMS without the
* MCS. */
}
}
-
-/**
- * For single-sampled render targets ("non-MSRT"), the MCS buffer is a
- * scaled-down bitfield representation of the color buffer which is capable of
- * recording when blocks of the color buffer are equal to the clear value.
- * This function returns the block size that will be used by the MCS buffer
- * corresponding to a certain color miptree.
- *
- * From the Ivy Bridge PRM, Vol2 Part1 11.7 "MCS Buffer for Render Target(s)",
- * beneath the "Fast Color Clear" bullet (p327):
- *
- * The following table describes the RT alignment
- *
- * Pixels Lines
- * TiledY RT CL
- * bpp
- * 32 8 4
- * 64 4 4
- * 128 2 4
- * TiledX RT CL
- * bpp
- * 32 16 2
- * 64 8 2
- * 128 4 2
- *
- * This alignment has the following uses:
- *
- * - For figuring out the size of the MCS buffer. Each 4k tile in the MCS
- * buffer contains 128 blocks horizontally and 256 blocks vertically.
- *
- * - For figuring out alignment restrictions for a fast clear operation. Fast
- * clear operations must always clear aligned multiples of 16 blocks
- * horizontally and 32 blocks vertically.
- *
- * - For scaling down the coordinates sent through the render pipeline during
- * a fast clear. X coordinates must be scaled down by 8 times the block
- * width, and Y coordinates by 16 times the block height.
- *
- * - For scaling down the coordinates sent through the render pipeline during
- * a "Render Target Resolve" operation. X coordinates must be scaled down
- * by half the block width, and Y coordinates by half the block height.
- */
-void
-intel_get_non_msrt_mcs_alignment(const struct intel_mipmap_tree *mt,
- unsigned *width_px, unsigned *height)
-{
- switch (mt->tiling) {
- default:
- unreachable("Non-MSRT MCS requires X or Y tiling");
- /* In release builds, fall through */
- case I915_TILING_Y:
- *width_px = 32 / mt->cpp;
- *height = 4;
- break;
- case I915_TILING_X:
- *width_px = 64 / mt->cpp;
- *height = 2;
- }
-}
-
bool
intel_tiling_supports_non_msrt_mcs(const struct brw_context *brw,
unsigned tiling)
if (brw->gen < 7)
return false;
- if (mt->disable_aux_buffers)
+ if (mt->aux_disable & INTEL_AUX_DISABLE_MCS)
return false;
/* This function applies only to non-multisampled render targets. */
if (mt->cpp != 4 && mt->cpp != 8 && mt->cpp != 16)
return false;
- if (mt->first_level != 0 || mt->last_level != 0) {
- if (brw->gen >= 8) {
- perf_debug("Multi-LOD fast clear - giving up (%dx%dx%d).\n",
- mt->logical_width0, mt->logical_height0, mt->last_level);
- }
- return false;
- }
+ const bool mip_mapped = mt->first_level != 0 || mt->last_level != 0;
+ const bool arrayed = mt->physical_depth0 != 1;
- /* Check for layered surfaces. */
- if (mt->physical_depth0 != 1) {
+ if (arrayed) {
/* Multisample surfaces with the CMS layout are not layered surfaces,
* yet still have physical_depth0 > 1. Assert that we don't
* accidentally reject a multisampled surface here. We should have
* rejected it earlier by explicitly checking the sample count.
*/
assert(mt->num_samples <= 1);
+ }
- if (brw->gen >= 8) {
- perf_debug("Layered fast clear - giving up. (%dx%d%d)\n",
- mt->logical_width0, mt->logical_height0,
- mt->physical_depth0);
- }
-
+ /* Handle the hardware restrictions...
+ *
+ * All GENs have the following restriction: "MCS buffer for non-MSRT is
+ * supported only for RT formats 32bpp, 64bpp, and 128bpp."
+ *
+ * From the HSW PRM Volume 7: 3D-Media-GPGPU, page 652: (Color Clear of
+ * Non-MultiSampler Render Target Restrictions) Support is for
+ * non-mip-mapped and non-array surface types only.
+ *
+ * From the BDW PRM Volume 7: 3D-Media-GPGPU, page 649: (Color Clear of
+ * Non-MultiSampler Render Target Restriction). Mip-mapped and arrayed
+ * surfaces are supported with MCS buffer layout with these alignments in
+ * the RT space: Horizontal Alignment = 256 and Vertical Alignment = 128.
+ *
+ * From the SKL PRM Volume 7: 3D-Media-GPGPU, page 632: (Color Clear of
+ * Non-MultiSampler Render Target Restriction). Mip-mapped and arrayed
+ * surfaces are supported with MCS buffer layout with these alignments in
+ * the RT space: Horizontal Alignment = 128 and Vertical Alignment = 64.
+ */
+ if (brw->gen < 8 && (mip_mapped || arrayed))
return false;
- }
/* There's no point in using an MCS buffer if the surface isn't in a
* renderable format.
if (brw->gen >= 9) {
mesa_format linear_format = _mesa_get_srgb_format_linear(mt->format);
- const uint32_t brw_format = brw_format_for_mesa_format(linear_format);
- return isl_format_supports_lossless_compression(brw->intelScreen->devinfo,
- brw_format);
+ const enum isl_format isl_format =
+ brw_isl_format_for_mesa_format(linear_format);
+ return isl_format_supports_ccs_e(&brw->screen->devinfo, isl_format);
} else
return true;
}
return false;
/* Compression always requires auxiliary buffer. */
- if (!mt->mcs_mt)
+ if (!mt->mcs_buf)
return false;
/* Single sample compression is represented re-using msaa compression
mt->logical_width0 = width0;
mt->logical_height0 = height0;
mt->logical_depth0 = depth0;
- mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_NO_MCS;
- mt->disable_aux_buffers = (layout_flags & MIPTREE_LAYOUT_DISABLE_AUX) != 0;
+ mt->aux_disable = (layout_flags & MIPTREE_LAYOUT_DISABLE_AUX) != 0 ?
+ INTEL_AUX_DISABLE_ALL : INTEL_AUX_DISABLE_NONE;
+ mt->aux_disable |= INTEL_AUX_DISABLE_CCS;
mt->is_scanout = (layout_flags & MIPTREE_LAYOUT_FOR_SCANOUT) != 0;
exec_list_make_empty(&mt->hiz_map);
+ exec_list_make_empty(&mt->color_resolve_map);
mt->cpp = _mesa_get_format_bytes(format);
mt->num_samples = num_samples;
mt->compressed = _mesa_is_format_compressed(format);
mt->msaa_layout = INTEL_MSAA_LAYOUT_NONE;
mt->refcount = 1;
+ int depth_multiply = 1;
if (num_samples > 1) {
/* Adjust width/height/depth for MSAA */
- mt->msaa_layout = compute_msaa_layout(brw, format,
- mt->disable_aux_buffers);
+ mt->msaa_layout = compute_msaa_layout(brw, format, mt->aux_disable);
if (mt->msaa_layout == INTEL_MSAA_LAYOUT_IMS) {
/* From the Ivybridge PRM, Volume 1, Part 1, page 108:
* "If the surface is multisampled and it is a depth or stencil
}
} else {
/* Non-interleaved */
- depth0 *= num_samples;
+ depth_multiply = num_samples;
+ depth0 *= depth_multiply;
}
}
}
if (target == GL_TEXTURE_CUBE_MAP)
- assert(depth0 == 6);
+ assert(depth0 == 6 * depth_multiply);
mt->physical_width0 = width0;
mt->physical_height0 = height0;
intel_miptree_wants_hiz_buffer(brw, mt)))) {
uint32_t stencil_flags = MIPTREE_LAYOUT_ACCELERATED_UPLOAD;
if (brw->gen == 6) {
- stencil_flags |= MIPTREE_LAYOUT_FORCE_ALL_SLICE_AT_LOD |
+ stencil_flags |= MIPTREE_LAYOUT_GEN6_HIZ_STENCIL |
MIPTREE_LAYOUT_TILING_ANY;
}
intel_miptree_release(&mt);
return NULL;
}
+ mt->stencil_mt->r8stencil_needs_update = true;
/* Fix up the Z miptree format for how we're splitting out separate
* stencil. Gen7 expects there to be no stencil bits in its depth buffer.
}
}
- if (layout_flags & MIPTREE_LAYOUT_FORCE_ALL_SLICE_AT_LOD)
- mt->array_layout = ALL_SLICES_AT_EACH_LOD;
+ if (layout_flags & MIPTREE_LAYOUT_GEN6_HIZ_STENCIL)
+ mt->array_layout = GEN6_HIZ_STENCIL;
/*
* Obey HALIGN_16 constraints for Gen8 and Gen9 buffers which are
(layout_flags & MIPTREE_LAYOUT_FORCE_HALIGN16) == 0);
}
- brw_miptree_layout(brw, mt, layout_flags);
+ if (!brw_miptree_layout(brw, mt, layout_flags)) {
+ intel_miptree_release(&mt);
+ return NULL;
+ }
- if (mt->disable_aux_buffers)
+ if (mt->aux_disable & INTEL_AUX_DISABLE_MCS)
assert(mt->msaa_layout != INTEL_MSAA_LAYOUT_CMS);
return mt;
}
}
-/* This function computes Yf/Ys tiled bo size, alignment and pitch. */
-static unsigned long
-intel_get_yf_ys_bo_size(struct intel_mipmap_tree *mt, unsigned *alignment,
- unsigned long *pitch)
-{
- uint32_t tile_width, tile_height;
- unsigned long stride, size, aligned_y;
-
- assert(mt->tr_mode != INTEL_MIPTREE_TRMODE_NONE);
- intel_get_tile_dims(mt->tiling, mt->tr_mode, mt->cpp,
- &tile_width, &tile_height);
-
- aligned_y = ALIGN(mt->total_height, tile_height);
- stride = mt->total_width * mt->cpp;
- stride = ALIGN(stride, tile_width);
- size = stride * aligned_y;
-
- if (mt->tr_mode == INTEL_MIPTREE_TRMODE_YF) {
- assert(size % 4096 == 0);
- *alignment = 4096;
- } else {
- assert(size % (64 * 1024) == 0);
- *alignment = 64 * 1024;
- }
- *pitch = stride;
- return size;
-}
-
static struct intel_mipmap_tree *
miptree_create(struct brw_context *brw,
GLenum target,
etc_format = (format != tex_format) ? tex_format : MESA_FORMAT_NONE;
- assert((layout_flags & MIPTREE_LAYOUT_DISABLE_AUX) == 0);
assert((layout_flags & MIPTREE_LAYOUT_FOR_BO) == 0);
mt = intel_miptree_create_layout(brw, target, format,
first_level, last_level, width0,
height0, depth0, num_samples,
layout_flags);
- /*
- * pitch == 0 || height == 0 indicates the null texture
- */
- if (!mt || !mt->total_width || !mt->total_height) {
- intel_miptree_release(&mt);
+ if (!mt)
return NULL;
- }
if (mt->tiling == (I915_TILING_Y | I915_TILING_X))
mt->tiling = I915_TILING_Y;
if (layout_flags & MIPTREE_LAYOUT_ACCELERATED_UPLOAD)
alloc_flags |= BO_ALLOC_FOR_RENDER;
- unsigned long pitch;
mt->etc_format = etc_format;
- if (mt->tr_mode != INTEL_MIPTREE_TRMODE_NONE) {
- unsigned alignment = 0;
- unsigned long size;
- size = intel_get_yf_ys_bo_size(mt, &alignment, &pitch);
- assert(size);
- mt->bo = drm_intel_bo_alloc_for_render(brw->bufmgr, "miptree",
- size, alignment);
+ if (format == MESA_FORMAT_S_UINT8) {
+ /* Align to size of W tile, 64x64. */
+ mt->bo = brw_bo_alloc_tiled(brw->bufmgr, "miptree",
+ ALIGN(mt->total_width, 64),
+ ALIGN(mt->total_height, 64),
+ mt->cpp, mt->tiling, &mt->pitch,
+ alloc_flags);
} else {
- if (format == MESA_FORMAT_S_UINT8) {
- /* Align to size of W tile, 64x64. */
- mt->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "miptree",
- ALIGN(mt->total_width, 64),
- ALIGN(mt->total_height, 64),
- mt->cpp, &mt->tiling, &pitch,
- alloc_flags);
- } else {
- mt->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "miptree",
- mt->total_width, mt->total_height,
- mt->cpp, &mt->tiling, &pitch,
- alloc_flags);
- }
+ mt->bo = brw_bo_alloc_tiled(brw->bufmgr, "miptree",
+ mt->total_width, mt->total_height,
+ mt->cpp, mt->tiling, &mt->pitch,
+ alloc_flags);
}
- mt->pitch = pitch;
+ if (layout_flags & MIPTREE_LAYOUT_FOR_SCANOUT)
+ mt->bo->cache_coherent = false;
return mt;
}
*/
if (brw->gen < 6 && mt->bo->size >= brw->max_gtt_map_object_size &&
mt->tiling == I915_TILING_Y) {
- unsigned long pitch = mt->pitch;
const uint32_t alloc_flags =
(layout_flags & MIPTREE_LAYOUT_ACCELERATED_UPLOAD) ?
BO_ALLOC_FOR_RENDER : 0;
mt->total_width, mt->total_height);
mt->tiling = I915_TILING_X;
- drm_intel_bo_unreference(mt->bo);
- mt->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "miptree",
+ brw_bo_unreference(mt->bo);
+ mt->bo = brw_bo_alloc_tiled(brw->bufmgr, "miptree",
mt->total_width, mt->total_height, mt->cpp,
- &mt->tiling, &pitch, alloc_flags);
- mt->pitch = pitch;
+ mt->tiling, &mt->pitch, alloc_flags);
}
mt->offset = 0;
*/
if (intel_tiling_supports_non_msrt_mcs(brw, mt->tiling) &&
intel_miptree_supports_non_msrt_fast_clear(brw, mt)) {
- mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_RESOLVED;
+ mt->aux_disable &= ~INTEL_AUX_DISABLE_CCS;
assert(brw->gen < 8 || mt->halign == 16 || num_samples <= 1);
+
+ /* On Gen9+ clients are not currently capable of consuming compressed
+ * single-sampled buffers. Disabling compression allows us to skip
+ * resolves.
+ */
+ const bool lossless_compression_disabled = INTEL_DEBUG & DEBUG_NO_RBC;
+ const bool is_lossless_compressed =
+ unlikely(!lossless_compression_disabled) &&
+ brw->gen >= 9 && !mt->is_scanout &&
+ intel_miptree_supports_lossless_compressed(brw, mt);
+
+ if (is_lossless_compressed) {
+ intel_miptree_alloc_non_msrt_mcs(brw, mt, is_lossless_compressed);
+ }
}
return mt;
struct intel_mipmap_tree *
intel_miptree_create_for_bo(struct brw_context *brw,
- drm_intel_bo *bo,
+ struct brw_bo *bo,
mesa_format format,
uint32_t offset,
uint32_t width,
uint32_t tiling, swizzle;
GLenum target;
- drm_intel_bo_get_tiling(bo, &tiling, &swizzle);
+ brw_bo_get_tiling(bo, &tiling, &swizzle);
/* Nothing will be able to use this miptree with the BO if the offset isn't
* aligned.
if (!mt)
return NULL;
- drm_intel_bo_reference(bo);
+ brw_bo_reference(bo);
mt->bo = bo;
mt->pitch = pitch;
mt->offset = offset;
void
intel_update_winsys_renderbuffer_miptree(struct brw_context *intel,
struct intel_renderbuffer *irb,
- drm_intel_bo *bo,
+ struct brw_bo *bo,
uint32_t width, uint32_t height,
uint32_t pitch)
{
*/
if (intel_tiling_supports_non_msrt_mcs(intel, singlesample_mt->tiling) &&
intel_miptree_supports_non_msrt_fast_clear(intel, singlesample_mt)) {
- singlesample_mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_RESOLVED;
+ singlesample_mt->aux_disable &= ~INTEL_AUX_DISABLE_CCS;
}
if (num_samples == 0) {
*dst = src;
}
+static void
+intel_miptree_hiz_buffer_free(struct intel_miptree_hiz_buffer *hiz_buf)
+{
+ if (hiz_buf == NULL)
+ return;
+
+ if (hiz_buf->mt)
+ intel_miptree_release(&hiz_buf->mt);
+ else
+ brw_bo_unreference(hiz_buf->aux_base.bo);
+
+ free(hiz_buf);
+}
void
intel_miptree_release(struct intel_mipmap_tree **mt)
DBG("%s deleting %p\n", __func__, *mt);
- drm_intel_bo_unreference((*mt)->bo);
+ brw_bo_unreference((*mt)->bo);
intel_miptree_release(&(*mt)->stencil_mt);
- if ((*mt)->hiz_buf) {
- if ((*mt)->hiz_buf->mt)
- intel_miptree_release(&(*mt)->hiz_buf->mt);
- else
- drm_intel_bo_unreference((*mt)->hiz_buf->bo);
- free((*mt)->hiz_buf);
+ intel_miptree_release(&(*mt)->r8stencil_mt);
+ intel_miptree_hiz_buffer_free((*mt)->hiz_buf);
+ if ((*mt)->mcs_buf) {
+ brw_bo_unreference((*mt)->mcs_buf->bo);
+ free((*mt)->mcs_buf);
}
- intel_miptree_release(&(*mt)->mcs_mt);
intel_resolve_map_clear(&(*mt)->hiz_map);
+ intel_resolve_map_clear(&(*mt)->color_resolve_map);
intel_miptree_release(&(*mt)->plane[0]);
intel_miptree_release(&(*mt)->plane[1]);
* and tile_h is set to 1.
*/
void
-intel_get_tile_dims(uint32_t tiling, uint32_t tr_mode, uint32_t cpp,
+intel_get_tile_dims(uint32_t tiling, uint32_t cpp,
uint32_t *tile_w, uint32_t *tile_h)
{
- if (tr_mode == INTEL_MIPTREE_TRMODE_NONE) {
- switch (tiling) {
- case I915_TILING_X:
- *tile_w = 512;
- *tile_h = 8;
- break;
- case I915_TILING_Y:
- *tile_w = 128;
- *tile_h = 32;
- break;
- case I915_TILING_NONE:
- *tile_w = cpp;
- *tile_h = 1;
- break;
- default:
- unreachable("not reached");
- }
- } else {
- uint32_t aspect_ratio = 1;
- assert(_mesa_is_pow_two(cpp));
-
- switch (cpp) {
- case 1:
- *tile_h = 64;
- break;
- case 2:
- case 4:
- *tile_h = 32;
- break;
- case 8:
- case 16:
- *tile_h = 16;
- break;
- default:
- unreachable("not reached");
- }
-
- if (cpp == 2 || cpp == 8)
- aspect_ratio = 2;
-
- if (tr_mode == INTEL_MIPTREE_TRMODE_YS)
- *tile_h *= 4;
-
- *tile_w = *tile_h * aspect_ratio * cpp;
+ switch (tiling) {
+ case I915_TILING_X:
+ *tile_w = 512;
+ *tile_h = 8;
+ break;
+ case I915_TILING_Y:
+ *tile_w = 128;
+ *tile_h = 32;
+ break;
+ case I915_TILING_NONE:
+ *tile_w = cpp;
+ *tile_h = 1;
+ break;
+ default:
+ unreachable("not reached");
}
}
* untiled, the masks are set to 0.
*/
void
-intel_get_tile_masks(uint32_t tiling, uint32_t tr_mode, uint32_t cpp,
+intel_get_tile_masks(uint32_t tiling, uint32_t cpp,
uint32_t *mask_x, uint32_t *mask_y)
{
uint32_t tile_w_bytes, tile_h;
- intel_get_tile_dims(tiling, tr_mode, cpp, &tile_w_bytes, &tile_h);
+ intel_get_tile_dims(tiling, cpp, &tile_w_bytes, &tile_h);
*mask_x = tile_w_bytes / cpp - 1;
*mask_y = tile_h - 1;
*/
uint32_t
intel_miptree_get_aligned_offset(const struct intel_mipmap_tree *mt,
- uint32_t x, uint32_t y,
- bool map_stencil_as_y_tiled)
+ uint32_t x, uint32_t y)
{
int cpp = mt->cpp;
uint32_t pitch = mt->pitch;
uint32_t tiling = mt->tiling;
- if (map_stencil_as_y_tiled) {
- tiling = I915_TILING_Y;
-
- /* When mapping a W-tiled stencil buffer as Y-tiled, each 64-high W-tile
- * gets transformed into a 32-high Y-tile. Accordingly, the pitch of
- * the resulting surface is twice the pitch of the original miptree,
- * since each row in the Y-tiled view corresponds to two rows in the
- * actual W-tiled surface. So we need to correct the pitch before
- * computing the offsets.
- */
- pitch *= 2;
- }
-
switch (tiling) {
default:
unreachable("not reached");
uint32_t x, y;
uint32_t mask_x, mask_y;
- intel_get_tile_masks(mt->tiling, mt->tr_mode, mt->cpp, &mask_x, &mask_y);
+ intel_get_tile_masks(mt->tiling, mt->cpp, &mask_x, &mask_y);
intel_miptree_get_image_offset(mt, level, slice, &x, &y);
*tile_x = x & mask_x;
*tile_y = y & mask_y;
- return intel_miptree_get_aligned_offset(mt, x & ~mask_x, y & ~mask_y, false);
+ return intel_miptree_get_aligned_offset(mt, x & ~mask_x, y & ~mask_y);
}
static void
struct intel_mipmap_tree *mt,
int init_value)
{
+ assert(mt->mcs_buf != NULL);
+
/* From the Ivy Bridge PRM, Vol 2 Part 1 p326:
*
* When MCS buffer is enabled and bound to MSRT, it is required that it
*
* Note: the clear value for MCS buffers is all 1's, so we memset to 0xff.
*/
- void *data = intel_miptree_map_raw(brw, mt->mcs_mt);
- memset(data, init_value, mt->mcs_mt->total_height * mt->mcs_mt->pitch);
- intel_miptree_unmap_raw(mt->mcs_mt);
- mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_CLEAR;
+ void *map = brw_bo_map(brw, mt->mcs_buf->bo, MAP_WRITE);
+ if (unlikely(map == NULL)) {
+ fprintf(stderr, "Failed to map mcs buffer into GTT\n");
+ brw_bo_unreference(mt->mcs_buf->bo);
+ free(mt->mcs_buf);
+ return;
+ }
+ void *data = map;
+ memset(data, init_value, mt->mcs_buf->size);
+ brw_bo_unmap(mt->mcs_buf->bo);
+}
+
+static struct intel_miptree_aux_buffer *
+intel_mcs_miptree_buf_create(struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ mesa_format format,
+ unsigned mcs_width,
+ unsigned mcs_height,
+ uint32_t layout_flags)
+{
+ struct intel_miptree_aux_buffer *buf = calloc(sizeof(*buf), 1);
+ struct intel_mipmap_tree *temp_mt;
+
+ if (!buf)
+ return NULL;
+
+ /* From the Ivy Bridge PRM, Vol4 Part1 p76, "MCS Base Address":
+ *
+ * "The MCS surface must be stored as Tile Y."
+ */
+ layout_flags |= MIPTREE_LAYOUT_TILING_Y;
+ temp_mt = miptree_create(brw,
+ mt->target,
+ format,
+ mt->first_level,
+ mt->last_level,
+ mcs_width,
+ mcs_height,
+ mt->logical_depth0,
+ 0 /* num_samples */,
+ layout_flags);
+ if (!temp_mt) {
+ free(buf);
+ return NULL;
+ }
+
+ buf->bo = temp_mt->bo;
+ buf->offset = temp_mt->offset;
+ buf->size = temp_mt->total_height * temp_mt->pitch;
+ buf->pitch = temp_mt->pitch;
+ buf->qpitch = temp_mt->qpitch;
+
+ /* Just hang on to the BO which backs the AUX buffer; the rest of the miptree
+ * structure should go away. We use miptree create simply as a means to make
+ * sure all the constraints for the buffer are satisfied.
+ */
+ brw_bo_reference(temp_mt->bo);
+ intel_miptree_release(&temp_mt);
+
+ return buf;
}
static bool
GLuint num_samples)
{
assert(brw->gen >= 7); /* MCS only used on Gen7+ */
- assert(mt->mcs_mt == NULL);
- assert(!mt->disable_aux_buffers);
+ assert(mt->mcs_buf == NULL);
+ assert((mt->aux_disable & INTEL_AUX_DISABLE_MCS) == 0);
/* Choose the correct format for the MCS buffer. All that really matters
* is that we allocate the right buffer size, since we'll always be
unreachable("Unrecognized sample count in intel_miptree_alloc_mcs");
};
- /* From the Ivy Bridge PRM, Vol4 Part1 p76, "MCS Base Address":
- *
- * "The MCS surface must be stored as Tile Y."
- */
- const uint32_t mcs_flags = MIPTREE_LAYOUT_ACCELERATED_UPLOAD |
- MIPTREE_LAYOUT_TILING_Y;
- mt->mcs_mt = miptree_create(brw,
- mt->target,
- format,
- mt->first_level,
- mt->last_level,
- mt->logical_width0,
- mt->logical_height0,
- mt->logical_depth0,
- 0 /* num_samples */,
- mcs_flags);
+ mt->mcs_buf =
+ intel_mcs_miptree_buf_create(brw, mt,
+ format,
+ mt->logical_width0,
+ mt->logical_height0,
+ MIPTREE_LAYOUT_ACCELERATED_UPLOAD);
+ if (!mt->mcs_buf)
+ return false;
intel_miptree_init_mcs(brw, mt, 0xFF);
- return mt->mcs_mt;
+ /* Multisampled miptrees are only supported for single level. */
+ assert(mt->first_level == 0);
+ intel_miptree_set_fast_clear_state(brw, mt, mt->first_level, 0,
+ mt->logical_depth0,
+ INTEL_FAST_CLEAR_STATE_CLEAR);
+
+ return true;
}
bool
intel_miptree_alloc_non_msrt_mcs(struct brw_context *brw,
- struct intel_mipmap_tree *mt)
-{
- assert(mt->mcs_mt == NULL);
- assert(!mt->disable_aux_buffers);
-
- /* The format of the MCS buffer is opaque to the driver; all that matters
- * is that we get its size and pitch right. We'll pretend that the format
- * is R32. Since an MCS tile covers 128 blocks horizontally, and a Y-tiled
- * R32 buffer is 32 pixels across, we'll need to scale the width down by
- * the block width and then a further factor of 4. Since an MCS tile
- * covers 256 blocks vertically, and a Y-tiled R32 buffer is 32 rows high,
- * we'll need to scale the height down by the block height and then a
- * further factor of 8.
- */
- const mesa_format format = MESA_FORMAT_R_UINT32;
- unsigned block_width_px;
- unsigned block_height;
- intel_get_non_msrt_mcs_alignment(mt, &block_width_px, &block_height);
- unsigned width_divisor = block_width_px * 4;
- unsigned height_divisor = block_height * 8;
-
- /* The Skylake MCS is twice as tall as the Broadwell MCS.
- *
- * In pre-Skylake, each bit in the MCS contained the state of 2 cachelines
- * in the main surface. In Skylake, it's two bits. The extra bit
- * doubles the MCS height, not width, because in Skylake the MCS is always
- * Y-tiled.
+ struct intel_mipmap_tree *mt,
+ bool is_lossless_compressed)
+{
+ assert(mt->mcs_buf == NULL);
+ assert(!(mt->aux_disable & (INTEL_AUX_DISABLE_MCS | INTEL_AUX_DISABLE_CCS)));
+
+ struct isl_surf temp_main_surf;
+ struct isl_surf temp_ccs_surf;
+
+ /* Create first an ISL presentation for the main color surface and let ISL
+ * calculate equivalent CCS surface against it.
*/
- if (brw->gen >= 9)
- height_divisor /= 2;
+ intel_miptree_get_isl_surf(brw, mt, &temp_main_surf);
+ if (!isl_surf_get_ccs_surf(&brw->isl_dev, &temp_main_surf, &temp_ccs_surf))
+ return false;
- unsigned mcs_width =
- ALIGN(mt->logical_width0, width_divisor) / width_divisor;
- unsigned mcs_height =
- ALIGN(mt->logical_height0, height_divisor) / height_divisor;
- assert(mt->logical_depth0 == 1);
- uint32_t layout_flags = MIPTREE_LAYOUT_TILING_Y;
+ assert(temp_ccs_surf.size &&
+ (temp_ccs_surf.size % temp_ccs_surf.row_pitch == 0));
- if (brw->gen >= 8) {
- layout_flags |= MIPTREE_LAYOUT_FORCE_HALIGN16;
- }
+ struct intel_miptree_aux_buffer *buf = calloc(sizeof(*buf), 1);
+ if (!buf)
+ return false;
- /* On Gen9+ clients are not currently capable of consuming compressed
- * single-sampled buffers. Disabling compression allows us to skip
- * resolves.
- */
- const bool lossless_compression_disabled = INTEL_DEBUG & DEBUG_NO_RBC;
- const bool is_lossless_compressed =
- unlikely(!lossless_compression_disabled) &&
- brw->gen >= 9 && !mt->is_scanout &&
- intel_miptree_supports_lossless_compressed(brw, mt);
+ buf->size = temp_ccs_surf.size;
+ buf->pitch = temp_ccs_surf.row_pitch;
+ buf->qpitch = isl_surf_get_array_pitch_sa_rows(&temp_ccs_surf);
/* In case of compression mcs buffer needs to be initialised requiring the
* buffer to be immediately mapped to cpu space for writing. Therefore do
* not use the gpu access flag which can cause an unnecessary delay if the
* backing pages happened to be just used by the GPU.
*/
- if (!is_lossless_compressed)
- layout_flags |= MIPTREE_LAYOUT_ACCELERATED_UPLOAD;
-
- mt->mcs_mt = miptree_create(brw,
- mt->target,
- format,
- mt->first_level,
- mt->last_level,
- mcs_width,
- mcs_height,
- mt->logical_depth0,
- 0 /* num_samples */,
- layout_flags);
+ const uint32_t alloc_flags =
+ is_lossless_compressed ? 0 : BO_ALLOC_FOR_RENDER;
+
+ /* ISL has stricter set of alignment rules then the drm allocator.
+ * Therefore one can pass the ISL dimensions in terms of bytes instead of
+ * trying to recalculate based on different format block sizes.
+ */
+ buf->bo = brw_bo_alloc_tiled(brw->bufmgr, "ccs-miptree",
+ buf->pitch, buf->size / buf->pitch,
+ 1, I915_TILING_Y, &buf->pitch, alloc_flags);
+ if (!buf->bo) {
+ free(buf);
+ return false;
+ }
+
+ mt->mcs_buf = buf;
/* From Gen9 onwards single-sampled (non-msrt) auxiliary buffers are
* used for lossless compression which requires similar initialisation
* Software needs to initialize MCS with zeros."
*/
intel_miptree_init_mcs(brw, mt, 0);
- mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_RESOLVED;
mt->msaa_layout = INTEL_MSAA_LAYOUT_CMS;
}
- return mt->mcs_mt;
-}
-
-void
-intel_miptree_prepare_mcs(struct brw_context *brw,
- struct intel_mipmap_tree *mt)
-{
- if (mt->mcs_mt)
- return;
-
- if (brw->gen < 9)
- return;
-
- /* Single sample compression is represented re-using msaa compression
- * layout type: "Compressed Multisampled Surfaces".
- */
- if (mt->msaa_layout != INTEL_MSAA_LAYOUT_CMS || mt->num_samples > 1)
- return;
-
- /* Clients are not currently capable of consuming compressed
- * single-sampled buffers.
- */
- if (mt->is_scanout)
- return;
-
- assert(intel_tiling_supports_non_msrt_mcs(brw, mt->tiling) ||
- intel_miptree_supports_lossless_compressed(brw, mt));
-
- /* Consider if lossless compression is supported but the needed
- * auxiliary buffer doesn't exist yet.
- *
- * Failing to allocate the auxiliary buffer means running out of
- * memory. The pointer to the aux miptree is left NULL which should
- * signal non-compressed behavior.
- */
- if (!intel_miptree_alloc_non_msrt_mcs(brw, mt)) {
- _mesa_warning(NULL,
- "Failed to allocated aux buffer for lossless"
- " compressed %p %u:%u %s\n",
- mt, mt->logical_width0, mt->logical_height0,
- _mesa_get_format_name(mt->format));
- }
+ return true;
}
/**
* Helper for intel_miptree_alloc_hiz() that determines the required hiz
* buffer dimensions and allocates a bo for the hiz buffer.
*/
-static struct intel_miptree_aux_buffer *
+static struct intel_miptree_hiz_buffer *
intel_gen7_hiz_buf_create(struct brw_context *brw,
struct intel_mipmap_tree *mt)
{
unsigned z_height = mt->logical_height0;
const unsigned z_depth = MAX2(mt->logical_depth0, 1);
unsigned hz_width, hz_height;
- struct intel_miptree_aux_buffer *buf = calloc(sizeof(*buf), 1);
+ struct intel_miptree_hiz_buffer *buf = calloc(sizeof(*buf), 1);
if (!buf)
return NULL;
hz_height = DIV_ROUND_UP(hz_qpitch * Z0, 2 * 8) * 8;
}
- unsigned long pitch;
- uint32_t tiling = I915_TILING_Y;
- buf->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "hiz",
- hz_width, hz_height, 1,
- &tiling, &pitch,
- BO_ALLOC_FOR_RENDER);
- if (!buf->bo) {
- free(buf);
- return NULL;
- } else if (tiling != I915_TILING_Y) {
- drm_intel_bo_unreference(buf->bo);
+ buf->aux_base.bo = brw_bo_alloc_tiled(brw->bufmgr, "hiz",
+ hz_width, hz_height, 1,
+ I915_TILING_Y, &buf->aux_base.pitch,
+ BO_ALLOC_FOR_RENDER);
+ if (!buf->aux_base.bo) {
free(buf);
return NULL;
}
- buf->pitch = pitch;
+ buf->aux_base.size = hz_width * hz_height;
return buf;
}
* Helper for intel_miptree_alloc_hiz() that determines the required hiz
* buffer dimensions and allocates a bo for the hiz buffer.
*/
-static struct intel_miptree_aux_buffer *
+static struct intel_miptree_hiz_buffer *
intel_gen8_hiz_buf_create(struct brw_context *brw,
struct intel_mipmap_tree *mt)
{
unsigned z_height = mt->logical_height0;
const unsigned z_depth = MAX2(mt->logical_depth0, 1);
unsigned hz_width, hz_height;
- struct intel_miptree_aux_buffer *buf = calloc(sizeof(*buf), 1);
+ struct intel_miptree_hiz_buffer *buf = calloc(sizeof(*buf), 1);
if (!buf)
return NULL;
Z_i = minify(Z_i, 1);
}
/* HZ_QPitch = h0 + max(h1, sum(i=2 to m; h_i)) */
- buf->qpitch = h0 + MAX2(h1, sum_h_i);
+ buf->aux_base.qpitch = h0 + MAX2(h1, sum_h_i);
if (mt->target == GL_TEXTURE_3D) {
/* (1/2) * sum(i=0 to m; h_i * max(1, floor(Z_Depth/2**i))) */
hz_height = DIV_ROUND_UP(hz_height_3d_sum, 2);
} else {
/* HZ_Height (rows) = ceiling( (HZ_QPitch/2)/8) *8 * Z_Depth */
- hz_height = DIV_ROUND_UP(buf->qpitch, 2 * 8) * 8 * Z0;
+ hz_height = DIV_ROUND_UP(buf->aux_base.qpitch, 2 * 8) * 8 * Z0;
}
- unsigned long pitch;
- uint32_t tiling = I915_TILING_Y;
- buf->bo = drm_intel_bo_alloc_tiled(brw->bufmgr, "hiz",
- hz_width, hz_height, 1,
- &tiling, &pitch,
- BO_ALLOC_FOR_RENDER);
- if (!buf->bo) {
- free(buf);
- return NULL;
- } else if (tiling != I915_TILING_Y) {
- drm_intel_bo_unreference(buf->bo);
+ buf->aux_base.bo = brw_bo_alloc_tiled(brw->bufmgr, "hiz",
+ hz_width, hz_height, 1,
+ I915_TILING_Y, &buf->aux_base.pitch,
+ BO_ALLOC_FOR_RENDER);
+ if (!buf->aux_base.bo) {
free(buf);
return NULL;
}
- buf->pitch = pitch;
+ buf->aux_base.size = hz_width * hz_height;
return buf;
}
-static struct intel_miptree_aux_buffer *
+static struct intel_miptree_hiz_buffer *
intel_hiz_miptree_buf_create(struct brw_context *brw,
struct intel_mipmap_tree *mt)
{
- struct intel_miptree_aux_buffer *buf = calloc(sizeof(*buf), 1);
+ struct intel_miptree_hiz_buffer *buf = calloc(sizeof(*buf), 1);
uint32_t layout_flags = MIPTREE_LAYOUT_ACCELERATED_UPLOAD;
if (brw->gen == 6)
- layout_flags |= MIPTREE_LAYOUT_FORCE_ALL_SLICE_AT_LOD;
+ layout_flags |= MIPTREE_LAYOUT_GEN6_HIZ_STENCIL;
if (!buf)
return NULL;
return NULL;
}
- buf->bo = buf->mt->bo;
- buf->pitch = buf->mt->pitch;
- buf->qpitch = buf->mt->qpitch;
+ buf->aux_base.bo = buf->mt->bo;
+ buf->aux_base.size = buf->mt->total_height * buf->mt->pitch;
+ buf->aux_base.pitch = buf->mt->pitch;
+ buf->aux_base.qpitch = buf->mt->qpitch * 2;
return buf;
}
if (mt->hiz_buf != NULL)
return false;
- if (mt->disable_aux_buffers)
+ if (mt->aux_disable & INTEL_AUX_DISABLE_HIZ)
return false;
switch (mt->format) {
struct intel_mipmap_tree *mt)
{
assert(mt->hiz_buf == NULL);
- assert(!mt->disable_aux_buffers);
+ assert((mt->aux_disable & INTEL_AUX_DISABLE_HIZ) == 0);
if (brw->gen == 7) {
mt->hiz_buf = intel_gen7_hiz_buf_create(brw, mt);
exec_node_init(&m->link);
m->level = level;
m->layer = layer;
- m->need = GEN6_HIZ_OP_HIZ_RESOLVE;
+ m->need = BLORP_HIZ_OP_HIZ_RESOLVE;
exec_list_push_tail(&mt->hiz_map, &m->link);
}
return true;
}
+/**
+ * Can the miptree sample using the hiz buffer?
+ */
+bool
+intel_miptree_sample_with_hiz(struct brw_context *brw,
+ struct intel_mipmap_tree *mt)
+{
+ /* It's unclear how well supported sampling from the hiz buffer is on GEN8,
+ * so keep things conservative for now and never enable it unless we're SKL+.
+ */
+ if (brw->gen < 9) {
+ return false;
+ }
+
+ if (!mt->hiz_buf) {
+ return false;
+ }
+
+ /* It seems the hardware won't fallback to the depth buffer if some of the
+ * mipmap levels aren't available in the HiZ buffer. So we need all levels
+ * of the texture to be HiZ enabled.
+ */
+ for (unsigned level = mt->first_level; level <= mt->last_level; ++level) {
+ if (!intel_miptree_level_has_hiz(mt, level))
+ return false;
+ }
+
+ /* If compressed multisampling is enabled, then we use it for the auxiliary
+ * buffer instead.
+ *
+ * From the BDW PRM (Volume 2d: Command Reference: Structures
+ * RENDER_SURFACE_STATE.AuxiliarySurfaceMode):
+ *
+ * "If this field is set to AUX_HIZ, Number of Multisamples must be
+ * MULTISAMPLECOUNT_1, and Surface Type cannot be SURFTYPE_3D.
+ *
+ * There is no such blurb for 1D textures, but there is sufficient evidence
+ * that this is broken on SKL+.
+ */
+ return (mt->num_samples <= 1 &&
+ mt->target != GL_TEXTURE_3D &&
+ mt->target != GL_TEXTURE_1D /* gen9+ restriction */);
+}
+
/**
* Does the miptree slice have hiz enabled?
*/
return;
intel_resolve_map_set(&mt->hiz_map,
- level, layer, GEN6_HIZ_OP_HIZ_RESOLVE);
+ level, layer, BLORP_HIZ_OP_HIZ_RESOLVE);
}
return;
intel_resolve_map_set(&mt->hiz_map,
- level, layer, GEN6_HIZ_OP_DEPTH_RESOLVE);
+ level, layer, BLORP_HIZ_OP_DEPTH_RESOLVE);
}
void
}
static bool
-intel_miptree_slice_resolve(struct brw_context *brw,
- struct intel_mipmap_tree *mt,
- uint32_t level,
- uint32_t layer,
- enum gen6_hiz_op need)
+intel_miptree_depth_hiz_resolve(struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ uint32_t start_level, uint32_t num_levels,
+ uint32_t start_layer, uint32_t num_layers,
+ enum blorp_hiz_op need)
{
- intel_miptree_check_level_layer(mt, level, layer);
+ bool did_resolve = false;
- struct intel_resolve_map *item =
- intel_resolve_map_get(&mt->hiz_map, level, layer);
+ foreach_list_typed_safe(struct intel_resolve_map, map, link, &mt->hiz_map) {
+ if (map->level < start_level ||
+ map->level >= (start_level + num_levels) ||
+ map->layer < start_layer ||
+ map->layer >= (start_layer + num_layers))
+ continue;
- if (!item || item->need != need)
- return false;
+ if (map->need != need)
+ continue;
- intel_hiz_exec(brw, mt, level, layer, need);
- intel_resolve_map_remove(item);
- return true;
+ intel_hiz_exec(brw, mt, map->level, map->layer, 1, need);
+ intel_resolve_map_remove(map);
+ did_resolve = true;
+ }
+
+ return did_resolve;
}
bool
uint32_t level,
uint32_t layer)
{
- return intel_miptree_slice_resolve(brw, mt, level, layer,
- GEN6_HIZ_OP_HIZ_RESOLVE);
+ return intel_miptree_depth_hiz_resolve(brw, mt, level, 1, layer, 1,
+ BLORP_HIZ_OP_HIZ_RESOLVE);
}
bool
uint32_t level,
uint32_t layer)
{
- return intel_miptree_slice_resolve(brw, mt, level, layer,
- GEN6_HIZ_OP_DEPTH_RESOLVE);
-}
-
-static bool
-intel_miptree_all_slices_resolve(struct brw_context *brw,
- struct intel_mipmap_tree *mt,
- enum gen6_hiz_op need)
-{
- bool did_resolve = false;
-
- foreach_list_typed_safe(struct intel_resolve_map, map, link, &mt->hiz_map) {
- if (map->need != need)
- continue;
-
- intel_hiz_exec(brw, mt, map->level, map->layer, need);
- intel_resolve_map_remove(map);
- did_resolve = true;
- }
-
- return did_resolve;
+ return intel_miptree_depth_hiz_resolve(brw, mt, level, 1, layer, 1,
+ BLORP_HIZ_OP_DEPTH_RESOLVE);
}
bool
intel_miptree_all_slices_resolve_hiz(struct brw_context *brw,
struct intel_mipmap_tree *mt)
{
- return intel_miptree_all_slices_resolve(brw, mt,
- GEN6_HIZ_OP_HIZ_RESOLVE);
+ return intel_miptree_depth_hiz_resolve(brw, mt,
+ 0, UINT32_MAX, 0, UINT32_MAX,
+ BLORP_HIZ_OP_HIZ_RESOLVE);
}
bool
intel_miptree_all_slices_resolve_depth(struct brw_context *brw,
struct intel_mipmap_tree *mt)
{
- return intel_miptree_all_slices_resolve(brw, mt,
- GEN6_HIZ_OP_DEPTH_RESOLVE);
+ return intel_miptree_depth_hiz_resolve(brw, mt,
+ 0, UINT32_MAX, 0, UINT32_MAX,
+ BLORP_HIZ_OP_DEPTH_RESOLVE);
}
-
-void
-intel_miptree_resolve_color(struct brw_context *brw,
- struct intel_mipmap_tree *mt,
- int flags)
+enum intel_fast_clear_state
+intel_miptree_get_fast_clear_state(const struct intel_mipmap_tree *mt,
+ unsigned level, unsigned layer)
{
- /* From gen9 onwards there is new compression scheme for single sampled
- * surfaces called "lossless compressed". These don't need to be always
- * resolved.
- */
- if ((flags & INTEL_MIPTREE_IGNORE_CCS_E) &&
- intel_miptree_is_lossless_compressed(brw, mt))
- return;
+ intel_miptree_check_level_layer(mt, level, layer);
- switch (mt->fast_clear_state) {
- case INTEL_FAST_CLEAR_STATE_NO_MCS:
- case INTEL_FAST_CLEAR_STATE_RESOLVED:
- /* No resolve needed */
- break;
- case INTEL_FAST_CLEAR_STATE_UNRESOLVED:
- case INTEL_FAST_CLEAR_STATE_CLEAR:
- /* Fast color clear resolves only make sense for non-MSAA buffers. */
- if (mt->msaa_layout == INTEL_MSAA_LAYOUT_NONE ||
- intel_miptree_is_lossless_compressed(brw, mt)) {
- brw_blorp_resolve_color(brw, mt);
- }
- break;
- }
+ const struct intel_resolve_map *item =
+ intel_resolve_map_const_get(&mt->color_resolve_map, level, layer);
+
+ if (!item)
+ return INTEL_FAST_CLEAR_STATE_RESOLVED;
+
+ return item->fast_clear_state;
}
+static void
+intel_miptree_check_color_resolve(const struct brw_context *brw,
+ const struct intel_mipmap_tree *mt,
+ unsigned level, unsigned layer)
+{
-/**
+ if ((mt->aux_disable & INTEL_AUX_DISABLE_CCS) || !mt->mcs_buf)
+ return;
+
+ /* Fast color clear is supported for mipmapped surfaces only on Gen8+. */
+ assert(brw->gen >= 8 ||
+ (level == 0 && mt->first_level == 0 && mt->last_level == 0));
+
+ /* Compression of arrayed msaa surfaces is supported. */
+ if (mt->num_samples > 1)
+ return;
+
+ /* Fast color clear is supported for non-msaa arrays only on Gen8+. */
+ assert(brw->gen >= 8 || (layer == 0 && mt->logical_depth0 == 1));
+
+ (void)level;
+ (void)layer;
+}
+
+void
+intel_miptree_set_fast_clear_state(const struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ unsigned level,
+ unsigned first_layer,
+ unsigned num_layers,
+ enum intel_fast_clear_state new_state)
+{
+ /* Setting the state to resolved means removing the item from the list
+ * altogether.
+ */
+ assert(new_state != INTEL_FAST_CLEAR_STATE_RESOLVED);
+
+ intel_miptree_check_color_resolve(brw, mt, level, first_layer);
+
+ assert(first_layer + num_layers <= mt->physical_depth0);
+
+ for (unsigned i = 0; i < num_layers; i++)
+ intel_resolve_map_set(&mt->color_resolve_map, level,
+ first_layer + i, new_state);
+}
+
+bool
+intel_miptree_has_color_unresolved(const struct intel_mipmap_tree *mt,
+ unsigned start_level, unsigned num_levels,
+ unsigned start_layer, unsigned num_layers)
+{
+ return intel_resolve_map_find_any(&mt->color_resolve_map,
+ start_level, num_levels,
+ start_layer, num_layers) != NULL;
+}
+
+void
+intel_miptree_used_for_rendering(const struct brw_context *brw,
+ struct intel_mipmap_tree *mt, unsigned level,
+ unsigned start_layer, unsigned num_layers)
+{
+ const bool is_lossless_compressed =
+ intel_miptree_is_lossless_compressed(brw, mt);
+
+ for (unsigned i = 0; i < num_layers; ++i) {
+ const enum intel_fast_clear_state fast_clear_state =
+ intel_miptree_get_fast_clear_state(mt, level, start_layer + i);
+
+ /* If the buffer was previously in fast clear state, change it to
+ * unresolved state, since it won't be guaranteed to be clear after
+ * rendering occurs.
+ */
+ if (is_lossless_compressed ||
+ fast_clear_state == INTEL_FAST_CLEAR_STATE_CLEAR) {
+ intel_miptree_set_fast_clear_state(
+ brw, mt, level, start_layer + i, 1,
+ INTEL_FAST_CLEAR_STATE_UNRESOLVED);
+ }
+ }
+}
+
+static bool
+intel_miptree_needs_color_resolve(const struct brw_context *brw,
+ const struct intel_mipmap_tree *mt,
+ int flags)
+{
+ if (mt->aux_disable & INTEL_AUX_DISABLE_CCS)
+ return false;
+
+ const bool is_lossless_compressed =
+ intel_miptree_is_lossless_compressed(brw, mt);
+
+ /* From gen9 onwards there is new compression scheme for single sampled
+ * surfaces called "lossless compressed". These don't need to be always
+ * resolved.
+ */
+ if ((flags & INTEL_MIPTREE_IGNORE_CCS_E) && is_lossless_compressed)
+ return false;
+
+ /* Fast color clear resolves only make sense for non-MSAA buffers. */
+ if (mt->msaa_layout != INTEL_MSAA_LAYOUT_NONE && !is_lossless_compressed)
+ return false;
+
+ return true;
+}
+
+bool
+intel_miptree_resolve_color(struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ uint32_t start_level, uint32_t num_levels,
+ uint32_t start_layer, uint32_t num_layers,
+ int flags)
+{
+ intel_miptree_check_color_resolve(brw, mt, start_level, start_layer);
+
+ if (!intel_miptree_needs_color_resolve(brw, mt, flags))
+ return false;
+
+ enum blorp_fast_clear_op resolve_op;
+ if (brw->gen >= 9) {
+ if (intel_miptree_is_lossless_compressed(brw, mt)) {
+ resolve_op = BLORP_FAST_CLEAR_OP_RESOLVE_FULL;
+ } else {
+ resolve_op = BLORP_FAST_CLEAR_OP_RESOLVE_PARTIAL;
+ }
+ } else {
+ /* Broadwell and earlier do not have a partial resolve */
+ assert(!intel_miptree_is_lossless_compressed(brw, mt));
+ resolve_op = BLORP_FAST_CLEAR_OP_RESOLVE_FULL;
+ }
+
+ bool resolved = false;
+ foreach_list_typed_safe(struct intel_resolve_map, map, link,
+ &mt->color_resolve_map) {
+ if (map->level < start_level ||
+ map->level >= (start_level + num_levels) ||
+ map->layer < start_layer ||
+ map->layer >= (start_layer + num_layers))
+ continue;
+
+ /* Arrayed and mip-mapped fast clear is only supported for gen8+. */
+ assert(brw->gen >= 8 || (map->level == 0 && map->layer == 0));
+
+ intel_miptree_check_level_layer(mt, map->level, map->layer);
+
+ assert(map->fast_clear_state != INTEL_FAST_CLEAR_STATE_RESOLVED);
+
+ brw_blorp_resolve_color(brw, mt, map->level, map->layer, resolve_op);
+ intel_resolve_map_remove(map);
+ resolved = true;
+ }
+
+ return resolved;
+}
+
+void
+intel_miptree_all_slices_resolve_color(struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ int flags)
+{
+
+ intel_miptree_resolve_color(brw, mt, 0, UINT32_MAX, 0, UINT32_MAX, flags);
+}
+
+static inline uint32_t
+miptree_level_range_length(const struct intel_mipmap_tree *mt,
+ uint32_t start_level, uint32_t num_levels)
+{
+ assert(start_level >= mt->first_level);
+ assert(start_level <= mt->last_level);
+
+ if (num_levels == INTEL_REMAINING_LAYERS)
+ num_levels = mt->last_level - start_level + 1;
+ /* Check for overflow */
+ assert(start_level + num_levels >= start_level);
+ assert(start_level + num_levels <= mt->last_level + 1);
+
+ return num_levels;
+}
+
+static inline uint32_t
+miptree_layer_range_length(const struct intel_mipmap_tree *mt, uint32_t level,
+ uint32_t start_layer, uint32_t num_layers)
+{
+ assert(level <= mt->last_level);
+ uint32_t total_num_layers = mt->level[level].depth;
+
+ assert(start_layer < total_num_layers);
+ if (num_layers == INTEL_REMAINING_LAYERS)
+ num_layers = total_num_layers - start_layer;
+ /* Check for overflow */
+ assert(start_layer + num_layers >= start_layer);
+ assert(start_layer + num_layers <= total_num_layers);
+
+ return num_layers;
+}
+
+void
+intel_miptree_prepare_access(struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ uint32_t start_level, uint32_t num_levels,
+ uint32_t start_layer, uint32_t num_layers,
+ bool aux_supported, bool fast_clear_supported)
+{
+ num_levels = miptree_level_range_length(mt, start_level, num_levels);
+
+ if (_mesa_is_format_color_format(mt->format)) {
+ if (!mt->mcs_buf)
+ return;
+
+ if (mt->num_samples > 1) {
+ /* Nothing to do for MSAA */
+ } else {
+ /* TODO: This is fairly terrible. We can do better. */
+ if (!aux_supported || !fast_clear_supported) {
+ intel_miptree_resolve_color(brw, mt, start_level, num_levels,
+ start_layer, num_layers, 0);
+ }
+ }
+ } else if (mt->format == MESA_FORMAT_S_UINT8) {
+ /* Nothing to do for stencil */
+ } else {
+ if (!mt->hiz_buf)
+ return;
+
+ if (aux_supported) {
+ assert(fast_clear_supported);
+ intel_miptree_depth_hiz_resolve(brw, mt, start_level, num_levels,
+ start_layer, num_layers,
+ BLORP_HIZ_OP_HIZ_RESOLVE);
+ } else {
+ assert(!fast_clear_supported);
+ intel_miptree_depth_hiz_resolve(brw, mt, start_level, num_levels,
+ start_layer, num_layers,
+ BLORP_HIZ_OP_DEPTH_RESOLVE);
+ }
+ }
+}
+
+void
+intel_miptree_finish_write(struct brw_context *brw,
+ struct intel_mipmap_tree *mt, uint32_t level,
+ uint32_t start_layer, uint32_t num_layers,
+ bool written_with_aux)
+{
+ num_layers = miptree_layer_range_length(mt, level, start_layer, num_layers);
+
+ if (_mesa_is_format_color_format(mt->format)) {
+ if (mt->num_samples > 1) {
+ /* Nothing to do for MSAA */
+ } else {
+ if (written_with_aux) {
+ intel_miptree_used_for_rendering(brw, mt, level,
+ start_layer, num_layers);
+ }
+ }
+ } else if (mt->format == MESA_FORMAT_S_UINT8) {
+ /* Nothing to do for stencil */
+ } else {
+ if (written_with_aux) {
+ for (unsigned a = 0; a < num_layers; a++) {
+ intel_miptree_check_level_layer(mt, level, start_layer);
+ intel_miptree_slice_set_needs_depth_resolve(mt, level,
+ start_layer + a);
+ }
+ } else {
+ for (unsigned a = 0; a < num_layers; a++) {
+ intel_miptree_check_level_layer(mt, level, start_layer);
+ intel_miptree_slice_set_needs_hiz_resolve(mt, level,
+ start_layer + a);
+ }
+ }
+ }
+}
+
+enum isl_aux_state
+intel_miptree_get_aux_state(const struct intel_mipmap_tree *mt,
+ uint32_t level, uint32_t layer)
+{
+ if (_mesa_is_format_color_format(mt->format)) {
+ assert(mt->mcs_buf != NULL);
+ if (mt->num_samples > 1) {
+ return ISL_AUX_STATE_COMPRESSED_CLEAR;
+ } else {
+ switch (intel_miptree_get_fast_clear_state(mt, level, layer)) {
+ case INTEL_FAST_CLEAR_STATE_RESOLVED:
+ return ISL_AUX_STATE_RESOLVED;
+ case INTEL_FAST_CLEAR_STATE_UNRESOLVED:
+ return ISL_AUX_STATE_COMPRESSED_CLEAR;
+ case INTEL_FAST_CLEAR_STATE_CLEAR:
+ return ISL_AUX_STATE_CLEAR;
+ default:
+ unreachable("Invalid fast clear state");
+ }
+ }
+ } else if (mt->format == MESA_FORMAT_S_UINT8) {
+ unreachable("Cannot get aux state for stencil");
+ } else {
+ assert(mt->hiz_buf != NULL);
+ const struct intel_resolve_map *map =
+ intel_resolve_map_const_get(&mt->hiz_map, level, layer);
+ if (!map)
+ return ISL_AUX_STATE_RESOLVED;
+ switch (map->need) {
+ case BLORP_HIZ_OP_DEPTH_RESOLVE:
+ return ISL_AUX_STATE_COMPRESSED_CLEAR;
+ case BLORP_HIZ_OP_HIZ_RESOLVE:
+ return ISL_AUX_STATE_AUX_INVALID;
+ default:
+ unreachable("Invalid hiz op");
+ }
+ }
+}
+
+void
+intel_miptree_set_aux_state(struct brw_context *brw,
+ struct intel_mipmap_tree *mt, uint32_t level,
+ uint32_t start_layer, uint32_t num_layers,
+ enum isl_aux_state aux_state)
+{
+ num_layers = miptree_layer_range_length(mt, level, start_layer, num_layers);
+
+ /* Right now, this only applies to clears. */
+ assert(aux_state == ISL_AUX_STATE_CLEAR);
+
+ if (_mesa_is_format_color_format(mt->format)) {
+ if (mt->num_samples > 1)
+ assert(mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS);
+
+ assert(level == 0 && start_layer == 0 && num_layers == 1);
+ intel_miptree_set_fast_clear_state(brw, mt, 0, 0, 1,
+ INTEL_FAST_CLEAR_STATE_CLEAR);
+ } else if (mt->format == MESA_FORMAT_S_UINT8) {
+ assert(!"Cannot set aux state for stencil");
+ } else {
+ for (unsigned a = 0; a < num_layers; a++) {
+ intel_miptree_check_level_layer(mt, level, start_layer);
+ intel_miptree_slice_set_needs_depth_resolve(mt, level,
+ start_layer + a);
+ }
+ }
+}
+
+/* On Gen9 color buffers may be compressed by the hardware (lossless
+ * compression). There are, however, format restrictions and care needs to be
+ * taken that the sampler engine is capable for re-interpreting a buffer with
+ * format different the buffer was originally written with.
+ *
+ * For example, SRGB formats are not compressible and the sampler engine isn't
+ * capable of treating RGBA_UNORM as SRGB_ALPHA. In such a case the underlying
+ * color buffer needs to be resolved so that the sampling surface can be
+ * sampled as non-compressed (i.e., without the auxiliary MCS buffer being
+ * set).
+ */
+static bool
+intel_texture_view_requires_resolve(struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ mesa_format format)
+{
+ if (brw->gen < 9 ||
+ !intel_miptree_is_lossless_compressed(brw, mt))
+ return false;
+
+ const enum isl_format isl_format = brw_isl_format_for_mesa_format(format);
+
+ if (isl_format_supports_ccs_e(&brw->screen->devinfo, isl_format))
+ return false;
+
+ perf_debug("Incompatible sampling format (%s) for rbc (%s)\n",
+ _mesa_get_format_name(format),
+ _mesa_get_format_name(mt->format));
+
+ return true;
+}
+
+static void
+intel_miptree_prepare_texture_slices(struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ mesa_format view_format,
+ uint32_t start_level, uint32_t num_levels,
+ uint32_t start_layer, uint32_t num_layers,
+ bool *aux_supported_out)
+{
+ bool aux_supported;
+ if (_mesa_is_format_color_format(mt->format)) {
+ aux_supported = intel_miptree_is_lossless_compressed(brw, mt) &&
+ !intel_texture_view_requires_resolve(brw, mt, view_format);
+ } else if (mt->format == MESA_FORMAT_S_UINT8) {
+ aux_supported = false;
+ } else {
+ aux_supported = intel_miptree_sample_with_hiz(brw, mt);
+ }
+
+ intel_miptree_prepare_access(brw, mt, start_level, num_levels,
+ start_layer, num_layers,
+ aux_supported, aux_supported);
+ if (aux_supported_out)
+ *aux_supported_out = aux_supported;
+}
+
+void
+intel_miptree_prepare_texture(struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ mesa_format view_format,
+ bool *aux_supported_out)
+{
+ intel_miptree_prepare_texture_slices(brw, mt, view_format,
+ 0, INTEL_REMAINING_LEVELS,
+ 0, INTEL_REMAINING_LAYERS,
+ aux_supported_out);
+}
+
+void
+intel_miptree_prepare_image(struct brw_context *brw,
+ struct intel_mipmap_tree *mt)
+{
+ /* The data port doesn't understand any compression */
+ intel_miptree_prepare_access(brw, mt, 0, INTEL_REMAINING_LEVELS,
+ 0, INTEL_REMAINING_LAYERS, false, false);
+}
+
+void
+intel_miptree_prepare_fb_fetch(struct brw_context *brw,
+ struct intel_mipmap_tree *mt, uint32_t level,
+ uint32_t start_layer, uint32_t num_layers)
+{
+ intel_miptree_prepare_texture_slices(brw, mt, mt->format, level, 1,
+ start_layer, num_layers, NULL);
+}
+
+void
+intel_miptree_prepare_render(struct brw_context *brw,
+ struct intel_mipmap_tree *mt, uint32_t level,
+ uint32_t start_layer, uint32_t layer_count,
+ bool srgb_enabled)
+{
+ /* If FRAMEBUFFER_SRGB is used on Gen9+ then we need to resolve any of
+ * the single-sampled color renderbuffers because the CCS buffer isn't
+ * supported for SRGB formats. This only matters if FRAMEBUFFER_SRGB is
+ * enabled because otherwise the surface state will be programmed with
+ * the linear equivalent format anyway.
+ */
+ if (brw->gen >= 9 && srgb_enabled && mt->num_samples <= 1 &&
+ _mesa_get_srgb_format_linear(mt->format) != mt->format) {
+
+ /* Lossless compression is not supported for SRGB formats, it
+ * should be impossible to get here with such surfaces.
+ */
+ assert(!intel_miptree_is_lossless_compressed(brw, mt));
+ intel_miptree_prepare_access(brw, mt, level, 1, start_layer, layer_count,
+ false, false);
+ }
+
+ /* 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 (layer_count > 1 &&
+ !(mt->aux_disable & INTEL_AUX_DISABLE_CCS) &&
+ !intel_miptree_is_lossless_compressed(brw, mt)) {
+ assert(brw->gen >= 8);
+
+ intel_miptree_prepare_access(brw, mt, level, 1, start_layer, layer_count,
+ false, false);
+ }
+}
+
+void
+intel_miptree_finish_render(struct brw_context *brw,
+ struct intel_mipmap_tree *mt, uint32_t level,
+ uint32_t start_layer, uint32_t layer_count)
+{
+ assert(_mesa_is_format_color_format(mt->format));
+ intel_miptree_finish_write(brw, mt, level, start_layer, layer_count,
+ mt->mcs_buf != NULL);
+}
+
+void
+intel_miptree_prepare_depth(struct brw_context *brw,
+ struct intel_mipmap_tree *mt, uint32_t level,
+ uint32_t start_layer, uint32_t layer_count)
+{
+ intel_miptree_prepare_access(brw, mt, level, 1, start_layer, layer_count,
+ mt->hiz_buf != NULL, mt->hiz_buf != NULL);
+}
+
+void
+intel_miptree_finish_depth(struct brw_context *brw,
+ struct intel_mipmap_tree *mt, uint32_t level,
+ uint32_t start_layer, uint32_t layer_count,
+ bool depth_written)
+{
+ if (depth_written) {
+ intel_miptree_finish_write(brw, mt, level, start_layer, layer_count,
+ mt->hiz_buf != NULL);
+ }
+}
+
+/**
* Make it possible to share the BO backing the given miptree with another
* process or another miptree.
*
* Fast color clears are unsafe with shared buffers, so we need to resolve and
- * then discard the MCS buffer, if present. We also set the fast_clear_state
- * to INTEL_FAST_CLEAR_STATE_NO_MCS to ensure that no MCS buffer gets
- * allocated in the future.
+ * then discard the MCS buffer, if present. We also set the no_ccs flag to
+ * ensure that no MCS buffer gets allocated in the future.
+ *
+ * HiZ is similarly unsafe with shared buffers.
*/
void
intel_miptree_make_shareable(struct brw_context *brw,
* pixel data is stored. Fortunately this code path should never be
* reached for multisample buffers.
*/
- assert(mt->msaa_layout == INTEL_MSAA_LAYOUT_NONE);
+ assert(mt->msaa_layout == INTEL_MSAA_LAYOUT_NONE || mt->num_samples <= 1);
+
+ intel_miptree_prepare_access(brw, mt, 0, INTEL_REMAINING_LEVELS,
+ 0, INTEL_REMAINING_LAYERS, false, false);
- if (mt->mcs_mt) {
- intel_miptree_resolve_color(brw, mt, 0);
- intel_miptree_release(&mt->mcs_mt);
- mt->fast_clear_state = INTEL_FAST_CLEAR_STATE_NO_MCS;
+ if (mt->mcs_buf) {
+ mt->aux_disable |= (INTEL_AUX_DISABLE_CCS | INTEL_AUX_DISABLE_MCS);
+ brw_bo_unreference(mt->mcs_buf->bo);
+ free(mt->mcs_buf);
+ mt->mcs_buf = NULL;
+
+ /* Any pending MCS/CCS operations are no longer needed. Trying to
+ * execute any will likely crash due to the missing aux buffer. So let's
+ * delete all pending ops.
+ */
+ exec_list_make_empty(&mt->color_resolve_map);
+ }
+
+ if (mt->hiz_buf) {
+ mt->aux_disable |= INTEL_AUX_DISABLE_HIZ;
+ intel_miptree_hiz_buffer_free(mt->hiz_buf);
+ mt->hiz_buf = NULL;
+
+ for (uint32_t l = mt->first_level; l <= mt->last_level; ++l) {
+ mt->level[l].has_hiz = false;
+ }
+
+ /* Any pending HiZ operations are no longer needed. Trying to execute
+ * any will likely crash due to the missing aux buffer. So let's delete
+ * all pending ops.
+ */
+ exec_list_make_empty(&mt->hiz_map);
}
}
}
}
-static void *
-intel_miptree_map_raw(struct brw_context *brw, struct intel_mipmap_tree *mt)
+void
+intel_update_r8stencil(struct brw_context *brw,
+ struct intel_mipmap_tree *mt)
{
- /* CPU accesses to color buffers don't understand fast color clears, so
- * resolve any pending fast color clears before we map.
- */
- intel_miptree_resolve_color(brw, mt, 0);
+ assert(brw->gen >= 7);
+ struct intel_mipmap_tree *src =
+ mt->format == MESA_FORMAT_S_UINT8 ? mt : mt->stencil_mt;
+ if (!src || brw->gen >= 8 || !src->r8stencil_needs_update)
+ return;
- drm_intel_bo *bo = mt->bo;
+ if (!mt->r8stencil_mt) {
+ const uint32_t r8stencil_flags =
+ MIPTREE_LAYOUT_ACCELERATED_UPLOAD | MIPTREE_LAYOUT_TILING_Y |
+ MIPTREE_LAYOUT_DISABLE_AUX;
+ assert(brw->gen > 6); /* Handle MIPTREE_LAYOUT_GEN6_HIZ_STENCIL */
+ mt->r8stencil_mt = intel_miptree_create(brw,
+ src->target,
+ MESA_FORMAT_R_UINT8,
+ src->first_level,
+ src->last_level,
+ src->logical_width0,
+ src->logical_height0,
+ src->logical_depth0,
+ src->num_samples,
+ r8stencil_flags);
+ assert(mt->r8stencil_mt);
+ }
+
+ struct intel_mipmap_tree *dst = mt->r8stencil_mt;
+
+ for (int level = src->first_level; level <= src->last_level; level++) {
+ const unsigned depth = src->level[level].depth;
+
+ for (unsigned layer = 0; layer < depth; layer++) {
+ brw_blorp_copy_miptrees(brw,
+ src, level, layer,
+ dst, level, layer,
+ 0, 0, 0, 0,
+ minify(src->logical_width0, level),
+ minify(src->logical_height0, level));
+ }
+ }
- if (drm_intel_bo_references(brw->batch.bo, bo))
- intel_batchbuffer_flush(brw);
+ brw_render_cache_set_check_flush(brw, dst->bo);
+ src->r8stencil_needs_update = false;
+}
- if (mt->tiling != I915_TILING_NONE)
- brw_bo_map_gtt(brw, bo, "miptree");
- else
- brw_bo_map(brw, bo, true, "miptree");
+static void *
+intel_miptree_map_raw(struct brw_context *brw,
+ struct intel_mipmap_tree *mt,
+ GLbitfield mode)
+{
+ struct brw_bo *bo = mt->bo;
+
+ if (brw_batch_references(&brw->batch, bo))
+ intel_batchbuffer_flush(brw);
- return bo->virtual;
+ return brw_bo_map(brw, bo, mode);
}
static void
intel_miptree_unmap_raw(struct intel_mipmap_tree *mt)
{
- drm_intel_bo_unmap(mt->bo);
+ brw_bo_unmap(mt->bo);
}
static void
y /= bh;
x /= bw;
- base = intel_miptree_map_raw(brw, mt) + mt->offset;
+ base = intel_miptree_map_raw(brw, mt, map->mode) + mt->offset;
if (base == NULL)
map->ptr = NULL;
* temporary buffer back out.
*/
if (!(map->mode & GL_MAP_INVALIDATE_RANGE_BIT)) {
- if (!intel_miptree_blit(brw,
- mt, level, slice,
- map->x, map->y, false,
- map->linear_mt, 0, 0,
- 0, 0, false,
- map->w, map->h, GL_COPY)) {
+ if (!intel_miptree_copy(brw,
+ mt, level, slice, map->x, map->y,
+ map->linear_mt, 0, 0, 0, 0,
+ map->w, map->h)) {
fprintf(stderr, "Failed to blit\n");
goto fail;
}
}
- map->ptr = intel_miptree_map_raw(brw, map->linear_mt);
+ map->ptr = intel_miptree_map_raw(brw, map->linear_mt, map->mode);
DBG("%s: %d,%d %dx%d from mt %p (%s) %d,%d = %p/%d\n", __func__,
map->x, map->y, map->w, map->h,
intel_miptree_unmap_raw(map->linear_mt);
if (map->mode & GL_MAP_WRITE_BIT) {
- bool ok = intel_miptree_blit(brw,
- map->linear_mt, 0, 0,
- 0, 0, false,
- mt, level, slice,
- map->x, map->y, false,
- map->w, map->h, GL_COPY);
+ bool ok = intel_miptree_copy(brw,
+ map->linear_mt, 0, 0, 0, 0,
+ mt, level, slice, map->x, map->y,
+ map->w, map->h);
WARN_ONCE(!ok, "Failed to blit from linear temporary mapping");
}
image_x += map->x;
image_y += map->y;
- void *src = intel_miptree_map_raw(brw, mt);
+ void *src = intel_miptree_map_raw(brw, mt, map->mode);
if (!src)
return;
*/
if (!(map->mode & GL_MAP_INVALIDATE_RANGE_BIT)) {
uint8_t *untiled_s8_map = map->ptr;
- uint8_t *tiled_s8_map = intel_miptree_map_raw(brw, mt);
+ uint8_t *tiled_s8_map = intel_miptree_map_raw(brw, mt, GL_MAP_READ_BIT);
unsigned int image_x, image_y;
intel_miptree_get_image_offset(mt, level, slice, &image_x, &image_y);
if (map->mode & GL_MAP_WRITE_BIT) {
unsigned int image_x, image_y;
uint8_t *untiled_s8_map = map->ptr;
- uint8_t *tiled_s8_map = intel_miptree_map_raw(brw, mt);
+ uint8_t *tiled_s8_map = intel_miptree_map_raw(brw, mt, GL_MAP_WRITE_BIT);
intel_miptree_get_image_offset(mt, level, slice, &image_x, &image_y);
image_x += map->x;
image_y += map->y;
- uint8_t *dst = intel_miptree_map_raw(brw, mt)
+ uint8_t *dst = intel_miptree_map_raw(brw, mt, GL_MAP_WRITE_BIT)
+ image_y * mt->pitch
+ image_x * mt->cpp;
*/
if (!(map->mode & GL_MAP_INVALIDATE_RANGE_BIT)) {
uint32_t *packed_map = map->ptr;
- uint8_t *s_map = intel_miptree_map_raw(brw, s_mt);
- uint32_t *z_map = intel_miptree_map_raw(brw, z_mt);
+ uint8_t *s_map = intel_miptree_map_raw(brw, s_mt, GL_MAP_READ_BIT);
+ uint32_t *z_map = intel_miptree_map_raw(brw, z_mt, GL_MAP_READ_BIT);
unsigned int s_image_x, s_image_y;
unsigned int z_image_x, z_image_y;
if (map->mode & GL_MAP_WRITE_BIT) {
uint32_t *packed_map = map->ptr;
- uint8_t *s_map = intel_miptree_map_raw(brw, s_mt);
- uint32_t *z_map = intel_miptree_map_raw(brw, z_mt);
+ uint8_t *s_map = intel_miptree_map_raw(brw, s_mt, GL_MAP_WRITE_BIT);
+ uint32_t *z_map = intel_miptree_map_raw(brw, z_mt, GL_MAP_WRITE_BIT);
unsigned int s_image_x, s_image_y;
unsigned int z_image_x, z_image_y;
can_blit_slice(struct intel_mipmap_tree *mt,
unsigned int level, unsigned int slice)
{
- uint32_t image_x;
- uint32_t image_y;
- intel_miptree_get_image_offset(mt, level, slice, &image_x, &image_y);
- if (image_x >= 32768 || image_y >= 32768)
- return false;
-
/* See intel_miptree_blit() for details on the 32k pitch limit. */
if (mt->pitch >= 32768)
return false;
{
if (brw->has_llc &&
/* It's probably not worth swapping to the blit ring because of
- * all the overhead involved. But, we must use blitter for the
- * surfaces with INTEL_MIPTREE_TRMODE_{YF,YS}.
+ * all the overhead involved.
*/
- (!(mode & GL_MAP_WRITE_BIT) ||
- mt->tr_mode != INTEL_MIPTREE_TRMODE_NONE) &&
+ !(mode & GL_MAP_WRITE_BIT) &&
!mt->compressed &&
(mt->tiling == I915_TILING_X ||
/* Prior to Sandybridge, the blitter can't handle Y tiling */
return;
}
- intel_miptree_slice_resolve_depth(brw, mt, level, slice);
- if (map->mode & GL_MAP_WRITE_BIT) {
- intel_miptree_slice_set_needs_hiz_resolve(mt, level, slice);
- }
+ intel_miptree_access_raw(brw, mt, level, slice,
+ map->mode & GL_MAP_WRITE_BIT);
if (mt->format == MESA_FORMAT_S_UINT8) {
intel_miptree_map_s8(brw, mt, map, level, slice);
intel_miptree_map_movntdqa(brw, mt, map, level, slice);
#endif
} else {
- /* intel_miptree_map_gtt() doesn't support surfaces with Yf/Ys tiling. */
- assert(mt->tr_mode == INTEL_MIPTREE_TRMODE_NONE);
intel_miptree_map_gtt(brw, mt, map, level, slice);
}
intel_miptree_release_map(mt, level, slice);
}
-void
-intel_miptree_get_isl_surf(struct brw_context *brw,
- const struct intel_mipmap_tree *mt,
- struct isl_surf *surf)
+enum isl_surf_dim
+get_isl_surf_dim(GLenum target)
{
- switch (mt->target) {
+ switch (target) {
case GL_TEXTURE_1D:
- case GL_TEXTURE_1D_ARRAY: {
- surf->dim = ISL_SURF_DIM_1D;
- if (brw->gen >= 9 && mt->tiling == I915_TILING_NONE)
- surf->dim_layout = ISL_DIM_LAYOUT_GEN9_1D;
- else
- surf->dim_layout = ISL_DIM_LAYOUT_GEN4_2D;
- break;
- }
+ case GL_TEXTURE_1D_ARRAY:
+ return ISL_SURF_DIM_1D;
+
case GL_TEXTURE_2D:
case GL_TEXTURE_2D_ARRAY:
case GL_TEXTURE_RECTANGLE:
case GL_TEXTURE_2D_MULTISAMPLE:
case GL_TEXTURE_2D_MULTISAMPLE_ARRAY:
case GL_TEXTURE_EXTERNAL_OES:
- surf->dim = ISL_SURF_DIM_2D;
- surf->dim_layout = ISL_DIM_LAYOUT_GEN4_2D;
- break;
+ return ISL_SURF_DIM_2D;
+
case GL_TEXTURE_3D:
- surf->dim = ISL_SURF_DIM_3D;
- if (brw->gen >= 9)
- surf->dim_layout = ISL_DIM_LAYOUT_GEN4_2D;
- else
- surf->dim_layout = ISL_DIM_LAYOUT_GEN4_3D;
- break;
- default:
- unreachable("Invalid texture target");
+ return ISL_SURF_DIM_3D;
}
+ unreachable("Invalid texture target");
+}
+
+enum isl_dim_layout
+get_isl_dim_layout(const struct gen_device_info *devinfo, uint32_t tiling,
+ GLenum target, enum miptree_array_layout array_layout)
+{
+ if (array_layout == GEN6_HIZ_STENCIL)
+ return ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ;
+
+ switch (target) {
+ case GL_TEXTURE_1D:
+ case GL_TEXTURE_1D_ARRAY:
+ return (devinfo->gen >= 9 && tiling == I915_TILING_NONE ?
+ ISL_DIM_LAYOUT_GEN9_1D : ISL_DIM_LAYOUT_GEN4_2D);
+
+ case GL_TEXTURE_2D:
+ case GL_TEXTURE_2D_ARRAY:
+ case GL_TEXTURE_RECTANGLE:
+ case GL_TEXTURE_2D_MULTISAMPLE:
+ case GL_TEXTURE_2D_MULTISAMPLE_ARRAY:
+ case GL_TEXTURE_EXTERNAL_OES:
+ return ISL_DIM_LAYOUT_GEN4_2D;
+
+ case GL_TEXTURE_CUBE_MAP:
+ case GL_TEXTURE_CUBE_MAP_ARRAY:
+ return (devinfo->gen == 4 ? ISL_DIM_LAYOUT_GEN4_3D :
+ ISL_DIM_LAYOUT_GEN4_2D);
+
+ case GL_TEXTURE_3D:
+ return (devinfo->gen >= 9 ?
+ ISL_DIM_LAYOUT_GEN4_2D : ISL_DIM_LAYOUT_GEN4_3D);
+ }
+
+ unreachable("Invalid texture target");
+}
+
+enum isl_tiling
+intel_miptree_get_isl_tiling(const struct intel_mipmap_tree *mt)
+{
+ if (mt->format == MESA_FORMAT_S_UINT8) {
+ return ISL_TILING_W;
+ } else {
+ switch (mt->tiling) {
+ case I915_TILING_NONE:
+ return ISL_TILING_LINEAR;
+ case I915_TILING_X:
+ return ISL_TILING_X;
+ case I915_TILING_Y:
+ return ISL_TILING_Y0;
+ default:
+ unreachable("Invalid tiling mode");
+ }
+ }
+}
+
+void
+intel_miptree_get_isl_surf(struct brw_context *brw,
+ const struct intel_mipmap_tree *mt,
+ struct isl_surf *surf)
+{
+ surf->dim = get_isl_surf_dim(mt->target);
+ surf->dim_layout = get_isl_dim_layout(&brw->screen->devinfo,
+ mt->tiling, mt->target,
+ mt->array_layout);
+
if (mt->num_samples > 1) {
switch (mt->msaa_layout) {
case INTEL_MSAA_LAYOUT_IMS:
surf->msaa_layout = ISL_MSAA_LAYOUT_NONE;
}
+ surf->tiling = intel_miptree_get_isl_tiling(mt);
+
if (mt->format == MESA_FORMAT_S_UINT8) {
- surf->tiling = ISL_TILING_W;
/* The ISL definition of row_pitch matches the surface state pitch field
* a bit better than intel_mipmap_tree. In particular, ISL incorporates
* the factor of 2 for W-tiling in row_pitch.
*/
surf->row_pitch = 2 * mt->pitch;
} else {
- switch (mt->tiling) {
- case I915_TILING_NONE:
- surf->tiling = ISL_TILING_LINEAR;
- break;
- case I915_TILING_X:
- surf->tiling = ISL_TILING_X;
- break;
- case I915_TILING_Y:
- switch (mt->tr_mode) {
- case INTEL_MIPTREE_TRMODE_NONE:
- surf->tiling = ISL_TILING_Y0;
- break;
- case INTEL_MIPTREE_TRMODE_YF:
- surf->tiling = ISL_TILING_Yf;
- break;
- case INTEL_MIPTREE_TRMODE_YS:
- surf->tiling = ISL_TILING_Ys;
- break;
- }
- break;
- default:
- unreachable("Invalid tiling mode");
- }
-
surf->row_pitch = mt->pitch;
}
surf->phys_level0_sa.array_len = mt->physical_depth0;
}
- surf->levels = mt->last_level + 1;
+ surf->levels = mt->last_level - mt->first_level + 1;
surf->samples = MAX2(mt->num_samples, 1);
surf->size = 0; /* TODO */
switch (surf->dim_layout) {
case ISL_DIM_LAYOUT_GEN4_2D:
case ISL_DIM_LAYOUT_GEN4_3D:
+ case ISL_DIM_LAYOUT_GEN6_STENCIL_HIZ:
if (brw->gen >= 9) {
surf->array_pitch_el_rows = mt->qpitch;
} else {
surf->array_pitch_span = ISL_ARRAY_PITCH_SPAN_FULL;
break;
case ALL_SLICES_AT_EACH_LOD:
+ case GEN6_HIZ_STENCIL:
surf->array_pitch_span = ISL_ARRAY_PITCH_SPAN_COMPACT;
break;
default:
struct isl_surf *surf,
enum isl_aux_usage *usage)
{
- /* Figure out the layout */
- if (_mesa_get_format_base_format(mt->format) == GL_DEPTH_COMPONENT) {
+ uint32_t aux_pitch, aux_qpitch;
+ if (mt->mcs_buf) {
+ aux_pitch = mt->mcs_buf->pitch;
+ aux_qpitch = mt->mcs_buf->qpitch;
+
+ if (mt->num_samples > 1) {
+ assert(mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS);
+ *usage = ISL_AUX_USAGE_MCS;
+ } else if (intel_miptree_is_lossless_compressed(brw, mt)) {
+ assert(brw->gen >= 9);
+ *usage = ISL_AUX_USAGE_CCS_E;
+ } else if ((mt->aux_disable & INTEL_AUX_DISABLE_CCS) == 0) {
+ *usage = ISL_AUX_USAGE_CCS_D;
+ } else {
+ unreachable("Invalid MCS miptree");
+ }
+ } else if (mt->hiz_buf) {
+ aux_pitch = mt->hiz_buf->aux_base.pitch;
+ aux_qpitch = mt->hiz_buf->aux_base.qpitch;
+
*usage = ISL_AUX_USAGE_HIZ;
- } else if (mt->num_samples > 1) {
- assert(mt->msaa_layout == INTEL_MSAA_LAYOUT_CMS);
- *usage = ISL_AUX_USAGE_MCS;
- } else if (intel_miptree_is_lossless_compressed(brw, mt)) {
- assert(brw->gen >= 9);
- *usage = ISL_AUX_USAGE_CCS_E;
- } else if (mt->fast_clear_state != INTEL_FAST_CLEAR_STATE_NO_MCS) {
- *usage = ISL_AUX_USAGE_CCS_D;
} else {
*usage = ISL_AUX_USAGE_NONE;
return;
/* Figure out the format and tiling of the auxiliary surface */
switch (*usage) {
case ISL_AUX_USAGE_NONE:
- unreachable("Invalid MCS miptree");
+ unreachable("Invalid auxiliary usage");
case ISL_AUX_USAGE_HIZ:
isl_surf_get_hiz_surf(&brw->isl_dev, surf, surf);
}
/* We want the pitch of the actual aux buffer. */
- surf->row_pitch = mt->mcs_mt->pitch;
+ surf->row_pitch = aux_pitch;
/* Auxiliary surfaces in ISL have compressed formats and array_pitch_el_rows
* is in elements. This doesn't match intel_mipmap_tree::qpitch which is
* compression block height.
*/
surf->array_pitch_el_rows =
- mt->mcs_mt->qpitch / isl_format_get_layout(surf->format)->bh;
-}
-
-union isl_color_value
-intel_miptree_get_isl_clear_color(struct brw_context *brw,
- const struct intel_mipmap_tree *mt)
-{
- union isl_color_value clear_color;
-
- if (_mesa_get_format_base_format(mt->format) == GL_DEPTH_COMPONENT) {
- clear_color.i32[0] = mt->depth_clear_value;
- clear_color.i32[1] = 0;
- clear_color.i32[2] = 0;
- clear_color.i32[3] = 0;
- } else if (brw->gen >= 9) {
- clear_color.i32[0] = mt->gen9_fast_clear_color.i[0];
- clear_color.i32[1] = mt->gen9_fast_clear_color.i[1];
- clear_color.i32[2] = mt->gen9_fast_clear_color.i[2];
- clear_color.i32[3] = mt->gen9_fast_clear_color.i[3];
- } else if (_mesa_is_format_integer(mt->format)) {
- clear_color.i32[0] = (mt->fast_clear_color_value & (1u << 31)) != 0;
- clear_color.i32[1] = (mt->fast_clear_color_value & (1u << 30)) != 0;
- clear_color.i32[2] = (mt->fast_clear_color_value & (1u << 29)) != 0;
- clear_color.i32[3] = (mt->fast_clear_color_value & (1u << 28)) != 0;
- } else {
- clear_color.f32[0] = (mt->fast_clear_color_value & (1u << 31)) != 0;
- clear_color.f32[1] = (mt->fast_clear_color_value & (1u << 30)) != 0;
- clear_color.f32[2] = (mt->fast_clear_color_value & (1u << 29)) != 0;
- clear_color.f32[3] = (mt->fast_clear_color_value & (1u << 28)) != 0;
- }
-
- return clear_color;
+ aux_qpitch / isl_format_get_layout(surf->format)->bh;
}