static __DRIimage *
intel_from_planar(__DRIimage *parent, int plane, void *loaderPrivate)
{
int width, height, offset, stride, dri_format, index;
struct intel_image_format *f;
uint32_t mask_x, mask_y;
__DRIimage *image;
if (parent == NULL || parent->planar_format == NULL)
return NULL;
f = parent->planar_format;
if (plane >= f->nplanes)
return NULL;
width = parent->region->width >> f->planes[plane].width_shift;
height = parent->region->height >> f->planes[plane].height_shift;
dri_format = f->planes[plane].dri_format;
index = f->planes[plane].buffer_index;
offset = parent->offsets[index];
stride = parent->strides[index];
image = intel_allocate_image(dri_format, loaderPrivate);
if (image == NULL)
return NULL;
if (offset + height * stride > parent->region->bo->size) {
_mesa_warning(NULL, "intel_create_sub_image: subimage out of bounds");
free(image);
return NULL;
}
image->region = calloc(sizeof(*image->region), 1);
if (image->region == NULL) {
free(image);
return NULL;
}
image->region->cpp = _mesa_get_format_bytes(image->format);
image->region->width = width;
image->region->height = height;
image->region->pitch = stride;
image->region->refcount = 1;
image->region->bo = parent->region->bo;
drm_intel_bo_reference(image->region->bo);
image->region->tiling = parent->region->tiling;
image->offset = offset;
intel_setup_image_from_dimensions(image);
intel_region_get_tile_masks(image->region, &mask_x, &mask_y, false);
if (offset & mask_x)
_mesa_warning(NULL,
"intel_create_sub_image: offset not on tile boundary");
return image;
}
/**
* Rendering to tiled buffers requires that the base address of the
* buffer be aligned to a page boundary. We generally render to
* textures by pointing the surface at the mipmap image level, which
* may not be aligned to a tile boundary.
*
* This function returns an appropriately-aligned base offset
* according to the tiling restrictions, plus any required x/y offset
* from there.
*/
uint32_t
intel_renderbuffer_tile_offsets(struct intel_renderbuffer *irb,
uint32_t *tile_x,
uint32_t *tile_y)
{
struct intel_region *region = irb->mt->region;
uint32_t mask_x, mask_y;
intel_region_get_tile_masks(region, &mask_x, &mask_y, false);
*tile_x = irb->draw_x & mask_x;
*tile_y = irb->draw_y & mask_y;
return intel_region_get_aligned_offset(region, irb->draw_x & ~mask_x,
irb->draw_y & ~mask_y, false);
}
/**
* Split x_offset and y_offset into a base offset (in bytes) and a remaining
* x/y offset (in pixels). Note: we can't do this by calling
* intel_renderbuffer_tile_offsets(), because the offsets may have been
* adjusted to account for Y vs. W tiling differences. So we compute it
* directly from the adjusted offsets.
*/
uint32_t
brw_blorp_surface_info::compute_tile_offsets(uint32_t *tile_x,
uint32_t *tile_y) const
{
struct intel_region *region = mt->region;
uint32_t mask_x, mask_y;
intel_region_get_tile_masks(region, &mask_x, &mask_y,
map_stencil_as_y_tiled);
*tile_x = x_offset & mask_x;
*tile_y = y_offset & mask_y;
return intel_region_get_aligned_offset(region, x_offset & ~mask_x,
y_offset & ~mask_y,
map_stencil_as_y_tiled);
}
/**
* Rendering with tiled buffers requires that the base address of the buffer
* be aligned to a page boundary. For renderbuffers, and sometimes with
* textures, we may want the surface to point at a texture image level that
* isn't at a page boundary.
*
* This function returns an appropriately-aligned base offset
* according to the tiling restrictions, plus any required x/y offset
* from there.
*/
uint32_t
intel_miptree_get_tile_offsets(struct intel_mipmap_tree *mt,
GLuint level, GLuint slice,
uint32_t *tile_x,
uint32_t *tile_y)
{
struct intel_region *region = mt->region;
uint32_t x, y;
uint32_t mask_x, mask_y;
intel_region_get_tile_masks(region, &mask_x, &mask_y, false);
intel_miptree_get_image_offset(mt, level, slice, &x, &y);
*tile_x = x & mask_x;
*tile_y = y & mask_y;
return intel_region_get_aligned_offset(region, x & ~mask_x, y & ~mask_y,
false);
}
static __DRIimage *
intel_create_sub_image(__DRIimage *parent,
int width, int height, int dri_format,
int offset, int pitch, void *loaderPrivate)
{
__DRIimage *image;
int cpp;
uint32_t mask_x, mask_y;
image = intel_allocate_image(dri_format, loaderPrivate);
cpp = _mesa_get_format_bytes(image->format);
if (offset + height * cpp * pitch > parent->region->bo->size) {
_mesa_warning(NULL, "intel_create_sub_image: subimage out of bounds");
FREE(image);
return NULL;
}
image->region = calloc(sizeof(*image->region), 1);
if (image->region == NULL) {
FREE(image);
return NULL;
}
image->region->cpp = _mesa_get_format_bytes(image->format);
image->region->width = width;
image->region->height = height;
image->region->pitch = pitch;
image->region->refcount = 1;
image->region->bo = parent->region->bo;
drm_intel_bo_reference(image->region->bo);
image->region->tiling = parent->region->tiling;
image->region->screen = parent->region->screen;
image->offset = offset;
intel_region_get_tile_masks(image->region, &mask_x, &mask_y);
if (offset & mask_x)
_mesa_warning(NULL,
"intel_create_sub_image: offset not on tile boundary");
return image;
}
static void emit_depthbuffer(struct brw_context *brw)
{
struct intel_context *intel = &brw->intel;
struct gl_context *ctx = &intel->ctx;
struct gl_framebuffer *fb = ctx->DrawBuffer;
/* _NEW_BUFFERS */
struct intel_renderbuffer *depth_irb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
struct intel_renderbuffer *stencil_irb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
struct intel_mipmap_tree *stencil_mt = NULL;
struct intel_region *hiz_region = NULL;
unsigned int len;
bool separate_stencil = false;
/* Amount by which drawing should be offset in order to draw to the
* appropriate miplevel/zoffset/cubeface. We will extract these values
* from depth_irb or stencil_irb once we determine which is present.
*/
uint32_t draw_x = 0, draw_y = 0;
/* Masks used to determine how much of the draw_x and draw_y offsets should
* be performed using the fine adjustment of "depth coordinate offset X/Y"
* (dw5 of 3DSTATE_DEPTH_BUFFER). Any remaining coarse adjustment will be
* performed by changing the base addresses of the buffers.
*
* Since the HiZ, depth, and stencil buffers all use the same "depth
* coordinate offset X/Y" values, we need to make sure that the coarse
* adjustment will be possible to apply to all three buffers. Since coarse
* adjustment can only be applied in multiples of the tile size, we will OR
* together the tile masks of all the buffers to determine which offsets to
* perform as fine adjustments.
*/
uint32_t tile_mask_x = 0, tile_mask_y = 0;
if (depth_irb) {
intel_region_get_tile_masks(depth_irb->mt->region,
&tile_mask_x, &tile_mask_y);
}
if (depth_irb &&
depth_irb->mt &&
depth_irb->mt->hiz_mt) {
hiz_region = depth_irb->mt->hiz_mt->region;
uint32_t hiz_tile_mask_x, hiz_tile_mask_y;
intel_region_get_tile_masks(hiz_region,
&hiz_tile_mask_x, &hiz_tile_mask_y);
/* Each HiZ row represents 2 rows of pixels */
hiz_tile_mask_y = hiz_tile_mask_y << 1 | 1;
tile_mask_x |= hiz_tile_mask_x;
tile_mask_y |= hiz_tile_mask_y;
}
/* 3DSTATE_DEPTH_BUFFER, 3DSTATE_STENCIL_BUFFER are both
* non-pipelined state that will need the PIPE_CONTROL workaround.
*/
if (intel->gen == 6) {
intel_emit_post_sync_nonzero_flush(intel);
intel_emit_depth_stall_flushes(intel);
}
/* Find the real separate stencil mt if present. */
if (stencil_irb) {
stencil_mt = stencil_irb->mt;
if (stencil_mt->stencil_mt)
stencil_mt = stencil_mt->stencil_mt;
if (stencil_mt->format == MESA_FORMAT_S8) {
separate_stencil = true;
/* Separate stencil buffer uses 64x64 tiles. */
tile_mask_x |= 63;
tile_mask_y |= 63;
} else {
uint32_t stencil_tile_mask_x, stencil_tile_mask_y;
intel_region_get_tile_masks(stencil_mt->region,
&stencil_tile_mask_x,
&stencil_tile_mask_y);
tile_mask_x |= stencil_tile_mask_x;
tile_mask_y |= stencil_tile_mask_y;
}
}
/* If there's a packed depth/stencil bound to stencil only, we need to
* emit the packed depth/stencil buffer packet.
*/
if (!depth_irb && stencil_irb && !separate_stencil)
depth_irb = stencil_irb;
if (intel->gen >= 6)
len = 7;
else if (intel->is_g4x || intel->gen == 5)
len = 6;
else
len = 5;
if (!depth_irb && !separate_stencil) {
BEGIN_BATCH(len);
OUT_BATCH(_3DSTATE_DEPTH_BUFFER << 16 | (len - 2));
OUT_BATCH((BRW_DEPTHFORMAT_D32_FLOAT << 18) |
(BRW_SURFACE_NULL << 29));
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
if (intel->is_g4x || intel->gen >= 5)
OUT_BATCH(0);
if (intel->gen >= 6)
OUT_BATCH(0);
ADVANCE_BATCH();
} else if (!depth_irb && separate_stencil) {
uint32_t tile_x, tile_y;
/*
* There exists a separate stencil buffer but no depth buffer.
*
* The stencil buffer inherits most of its fields from
* 3DSTATE_DEPTH_BUFFER: namely the tile walk, surface type, width, and
* height.
*
* Enable the hiz bit because it and the separate stencil bit must have
* the same value. From Section 2.11.5.6.1.1 3DSTATE_DEPTH_BUFFER, Bit
* 1.21 "Separate Stencil Enable":
* [DevIL]: If this field is enabled, Hierarchical Depth Buffer
* Enable must also be enabled.
*
* [DevGT]: This field must be set to the same value (enabled or
* disabled) as Hierarchical Depth Buffer Enable
*
* The tiled bit must be set. From the Sandybridge PRM, Volume 2, Part 1,
* Section 7.5.5.1.1 3DSTATE_DEPTH_BUFFER, Bit 1.27 Tiled Surface:
* [DevGT+]: This field must be set to TRUE.
*/
assert(intel->has_separate_stencil);
draw_x = stencil_irb->draw_x;
draw_y = stencil_irb->draw_y;
tile_x = draw_x & tile_mask_x;
tile_y = draw_y & tile_mask_y;
/* According to the Sandy Bridge PRM, volume 2 part 1, pp326-327
* (3DSTATE_DEPTH_BUFFER dw5), in the documentation for "Depth
* Coordinate Offset X/Y":
*
* "The 3 LSBs of both offsets must be zero to ensure correct
* alignment"
*
* We have no guarantee that tile_x and tile_y are correctly aligned,
* since they are determined by the mipmap layout, which is only aligned
* to multiples of 4.
*
* So, to avoid hanging the GPU, just smash the low order 3 bits of
* tile_x and tile_y to 0. This is a temporary workaround until we come
* up with a better solution.
*/
tile_x &= ~7;
tile_y &= ~7;
BEGIN_BATCH(len);
OUT_BATCH(_3DSTATE_DEPTH_BUFFER << 16 | (len - 2));
OUT_BATCH((BRW_DEPTHFORMAT_D32_FLOAT << 18) |
(1 << 21) | /* separate stencil enable */
(1 << 22) | /* hiz enable */
(BRW_TILEWALK_YMAJOR << 26) |
(1 << 27) | /* tiled surface */
(BRW_SURFACE_2D << 29));
OUT_BATCH(0);
OUT_BATCH(((stencil_irb->Base.Base.Width + tile_x - 1) << 6) |
(stencil_irb->Base.Base.Height + tile_y - 1) << 19);
OUT_BATCH(0);
if (intel->is_g4x || intel->gen >= 5)
OUT_BATCH(tile_x | (tile_y << 16));
else
assert(tile_x == 0 && tile_y == 0);
if (intel->gen >= 6)
OUT_BATCH(0);
ADVANCE_BATCH();
} else {
struct intel_region *region = depth_irb->mt->region;
uint32_t tile_x, tile_y, offset;
/* If using separate stencil, hiz must be enabled. */
assert(!separate_stencil || hiz_region);
assert(intel->gen < 6 || region->tiling == I915_TILING_Y);
assert(!hiz_region || region->tiling == I915_TILING_Y);
draw_x = depth_irb->draw_x;
draw_y = depth_irb->draw_y;
tile_x = draw_x & tile_mask_x;
tile_y = draw_y & tile_mask_y;
/* According to the Sandy Bridge PRM, volume 2 part 1, pp326-327
* (3DSTATE_DEPTH_BUFFER dw5), in the documentation for "Depth
* Coordinate Offset X/Y":
*
* "The 3 LSBs of both offsets must be zero to ensure correct
* alignment"
*
* We have no guarantee that tile_x and tile_y are correctly aligned,
* since they are determined by the mipmap layout, which is only aligned
* to multiples of 4.
*
* So, to avoid hanging the GPU, just smash the low order 3 bits of
* tile_x and tile_y to 0. This is a temporary workaround until we come
* up with a better solution.
*/
tile_x &= ~7;
tile_y &= ~7;
offset = intel_region_get_aligned_offset(region,
draw_x & ~tile_mask_x,
draw_y & ~tile_mask_y);
BEGIN_BATCH(len);
OUT_BATCH(_3DSTATE_DEPTH_BUFFER << 16 | (len - 2));
OUT_BATCH(((region->pitch * region->cpp) - 1) |
(brw_depthbuffer_format(brw) << 18) |
((hiz_region ? 1 : 0) << 21) | /* separate stencil enable */
((hiz_region ? 1 : 0) << 22) | /* hiz enable */
(BRW_TILEWALK_YMAJOR << 26) |
((region->tiling != I915_TILING_NONE) << 27) |
(BRW_SURFACE_2D << 29));
OUT_RELOC(region->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
offset);
OUT_BATCH((BRW_SURFACE_MIPMAPLAYOUT_BELOW << 1) |
(((depth_irb->Base.Base.Width + tile_x) - 1) << 6) |
(((depth_irb->Base.Base.Height + tile_y) - 1) << 19));
OUT_BATCH(0);
if (intel->is_g4x || intel->gen >= 5)
OUT_BATCH(tile_x | (tile_y << 16));
else
assert(tile_x == 0 && tile_y == 0);
if (intel->gen >= 6)
OUT_BATCH(0);
ADVANCE_BATCH();
}
if (hiz_region || separate_stencil) {
/*
* In the 3DSTATE_DEPTH_BUFFER batch emitted above, the 'separate
* stencil enable' and 'hiz enable' bits were set. Therefore we must
* emit 3DSTATE_HIER_DEPTH_BUFFER and 3DSTATE_STENCIL_BUFFER. Even if
* there is no stencil buffer, 3DSTATE_STENCIL_BUFFER must be emitted;
* failure to do so causes hangs on gen5 and a stall on gen6.
*/
/* Emit hiz buffer. */
if (hiz_region) {
uint32_t hiz_offset =
intel_region_get_aligned_offset(hiz_region,
draw_x & ~tile_mask_x,
(draw_y & ~tile_mask_y) / 2);
BEGIN_BATCH(3);
OUT_BATCH((_3DSTATE_HIER_DEPTH_BUFFER << 16) | (3 - 2));
OUT_BATCH(hiz_region->pitch * hiz_region->cpp - 1);
OUT_RELOC(hiz_region->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
hiz_offset);
ADVANCE_BATCH();
} else {
BEGIN_BATCH(3);
OUT_BATCH((_3DSTATE_HIER_DEPTH_BUFFER << 16) | (3 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
}
/* Emit stencil buffer. */
if (separate_stencil) {
struct intel_region *region = stencil_mt->region;
/* Note: we can't compute the stencil offset using
* intel_region_get_aligned_offset(), because stencil_region claims
* that the region is untiled; in fact it's W tiled.
*/
uint32_t stencil_offset =
(draw_y & ~tile_mask_y) * region->pitch +
(draw_x & ~tile_mask_x) * 64;
BEGIN_BATCH(3);
OUT_BATCH((_3DSTATE_STENCIL_BUFFER << 16) | (3 - 2));
/* The stencil buffer has quirky pitch requirements. From Vol 2a,
* 11.5.6.2.1 3DSTATE_STENCIL_BUFFER, field "Surface Pitch":
* The pitch must be set to 2x the value computed based on width, as
* the stencil buffer is stored with two rows interleaved.
*/
OUT_BATCH(2 * region->pitch * region->cpp - 1);
OUT_RELOC(region->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
stencil_offset);
ADVANCE_BATCH();
} else {
BEGIN_BATCH(3);
OUT_BATCH((_3DSTATE_STENCIL_BUFFER << 16) | (3 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
}
}
/*
* On Gen >= 6, emit clear params for safety. If using hiz, then clear
* params must be emitted.
*
* From Section 2.11.5.6.4.1 3DSTATE_CLEAR_PARAMS:
* 3DSTATE_CLEAR_PARAMS packet must follow the DEPTH_BUFFER_STATE packet
* when HiZ is enabled and the DEPTH_BUFFER_STATE changes.
*/
if (intel->gen >= 6 || hiz_region) {
if (intel->gen == 6)
intel_emit_post_sync_nonzero_flush(intel);
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_CLEAR_PARAMS << 16 |
GEN5_DEPTH_CLEAR_VALID |
(2 - 2));
OUT_BATCH(depth_irb ? depth_irb->mt->depth_clear_value : 0);
ADVANCE_BATCH();
}
}