static void
gen6_blorp_emit_depth_disable(struct brw_context *brw,
const struct brw_blorp_params *params)
{
brw_emit_depth_stall_flushes(brw);
BEGIN_BATCH(7);
OUT_BATCH(_3DSTATE_DEPTH_BUFFER << 16 | (7 - 2));
OUT_BATCH((BRW_DEPTHFORMAT_D32_FLOAT << 18) |
(BRW_SURFACE_NULL << 29));
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
BEGIN_BATCH(3);
OUT_BATCH(_3DSTATE_HIER_DEPTH_BUFFER << 16 | (3 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
BEGIN_BATCH(3);
OUT_BATCH(_3DSTATE_STENCIL_BUFFER << 16 | (3 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
}
static void
gen8_blorp_emit_depth_disable(struct brw_context *brw)
{
/* Skip repeated NULL depth/stencil emits (think 2D rendering). */
if (brw->no_depth_or_stencil)
return;
brw_emit_depth_stall_flushes(brw);
BEGIN_BATCH(8);
OUT_BATCH(GEN7_3DSTATE_DEPTH_BUFFER << 16 | (8 - 2));
OUT_BATCH((BRW_DEPTHFORMAT_D32_FLOAT << 18) | (BRW_SURFACE_NULL << 29));
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
BEGIN_BATCH(5);
OUT_BATCH(GEN7_3DSTATE_HIER_DEPTH_BUFFER << 16 | (5 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
BEGIN_BATCH(5);
OUT_BATCH(GEN7_3DSTATE_STENCIL_BUFFER << 16 | (5 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
}
static void
gen6_blorp_emit_depth_stencil_config(struct brw_context *brw,
const struct brw_blorp_params *params)
{
uint32_t surfwidth, surfheight;
uint32_t surftype;
unsigned int depth = MAX2(params->depth.mt->logical_depth0, 1);
GLenum gl_target = params->depth.mt->target;
unsigned int lod;
switch (gl_target) {
case GL_TEXTURE_CUBE_MAP_ARRAY:
case GL_TEXTURE_CUBE_MAP:
/* The PRM claims that we should use BRW_SURFACE_CUBE for this
* situation, but experiments show that gl_Layer doesn't work when we do
* this. So we use BRW_SURFACE_2D, since for rendering purposes this is
* equivalent.
*/
surftype = BRW_SURFACE_2D;
depth *= 6;
break;
default:
surftype = translate_tex_target(gl_target);
break;
}
const unsigned min_array_element = params->depth.layer;
lod = params->depth.level - params->depth.mt->first_level;
if (params->hiz_op != GEN6_HIZ_OP_NONE && lod == 0) {
/* HIZ ops for lod 0 may set the width & height a little
* larger to allow the fast depth clear to fit the hardware
* alignment requirements. (8x4)
*/
surfwidth = params->depth.width;
surfheight = params->depth.height;
} else {
surfwidth = params->depth.mt->logical_width0;
surfheight = params->depth.mt->logical_height0;
}
/* 3DSTATE_DEPTH_BUFFER */
{
brw_emit_depth_stall_flushes(brw);
BEGIN_BATCH(7);
/* 3DSTATE_DEPTH_BUFFER dw0 */
OUT_BATCH(_3DSTATE_DEPTH_BUFFER << 16 | (7 - 2));
/* 3DSTATE_DEPTH_BUFFER dw1 */
OUT_BATCH((params->depth.mt->pitch - 1) |
params->depth_format << 18 |
1 << 21 | /* separate stencil enable */
1 << 22 | /* hiz enable */
BRW_TILEWALK_YMAJOR << 26 |
1 << 27 | /* y-tiled */
surftype << 29);
/* 3DSTATE_DEPTH_BUFFER dw2 */
OUT_RELOC(params->depth.mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
0);
/* 3DSTATE_DEPTH_BUFFER dw3 */
OUT_BATCH(BRW_SURFACE_MIPMAPLAYOUT_BELOW << 1 |
(surfwidth - 1) << 6 |
(surfheight - 1) << 19 |
lod << 2);
/* 3DSTATE_DEPTH_BUFFER dw4 */
OUT_BATCH((depth - 1) << 21 |
min_array_element << 10 |
(depth - 1) << 1);
/* 3DSTATE_DEPTH_BUFFER dw5 */
OUT_BATCH(0);
/* 3DSTATE_DEPTH_BUFFER dw6 */
OUT_BATCH(0);
ADVANCE_BATCH();
}
/* 3DSTATE_HIER_DEPTH_BUFFER */
{
struct intel_mipmap_tree *hiz_mt = params->depth.mt->hiz_buf->mt;
uint32_t offset = 0;
if (hiz_mt->array_layout == ALL_SLICES_AT_EACH_LOD) {
offset = intel_miptree_get_aligned_offset(hiz_mt,
hiz_mt->level[lod].level_x,
hiz_mt->level[lod].level_y,
false);
}
BEGIN_BATCH(3);
OUT_BATCH((_3DSTATE_HIER_DEPTH_BUFFER << 16) | (3 - 2));
OUT_BATCH(hiz_mt->pitch - 1);
OUT_RELOC(hiz_mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
offset);
ADVANCE_BATCH();
}
/* 3DSTATE_STENCIL_BUFFER */
{
BEGIN_BATCH(3);
OUT_BATCH((_3DSTATE_STENCIL_BUFFER << 16) | (3 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
}
}
static void
gen7_blorp_emit_depth_stencil_config(struct brw_context *brw,
const struct brw_blorp_params *params)
{
const uint8_t mocs = GEN7_MOCS_L3;
uint32_t surfwidth, surfheight;
uint32_t surftype;
unsigned int depth = MAX2(params->depth.mt->logical_depth0, 1);
unsigned int min_array_element;
GLenum gl_target = params->depth.mt->target;
unsigned int lod;
switch (gl_target) {
case GL_TEXTURE_CUBE_MAP_ARRAY:
case GL_TEXTURE_CUBE_MAP:
/* The PRM claims that we should use BRW_SURFACE_CUBE for this
* situation, but experiments show that gl_Layer doesn't work when we do
* this. So we use BRW_SURFACE_2D, since for rendering purposes this is
* equivalent.
*/
surftype = BRW_SURFACE_2D;
depth *= 6;
break;
default:
surftype = translate_tex_target(gl_target);
break;
}
min_array_element = params->depth.layer;
if (params->depth.mt->num_samples > 1) {
/* Convert physical layer to logical layer. */
min_array_element /= params->depth.mt->num_samples;
}
lod = params->depth.level - params->depth.mt->first_level;
if (params->hiz_op != GEN6_HIZ_OP_NONE && lod == 0) {
/* HIZ ops for lod 0 may set the width & height a little
* larger to allow the fast depth clear to fit the hardware
* alignment requirements. (8x4)
*/
surfwidth = params->depth.width;
surfheight = params->depth.height;
} else {
surfwidth = params->depth.mt->logical_width0;
surfheight = params->depth.mt->logical_height0;
}
/* 3DSTATE_DEPTH_BUFFER */
{
brw_emit_depth_stall_flushes(brw);
BEGIN_BATCH(7);
OUT_BATCH(GEN7_3DSTATE_DEPTH_BUFFER << 16 | (7 - 2));
OUT_BATCH((params->depth.mt->pitch - 1) |
params->depth_format << 18 |
1 << 22 | /* hiz enable */
1 << 28 | /* depth write */
surftype << 29);
OUT_RELOC(params->depth.mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
0);
OUT_BATCH((surfwidth - 1) << 4 |
(surfheight - 1) << 18 |
lod);
OUT_BATCH(((depth - 1) << 21) |
(min_array_element << 10) |
mocs);
OUT_BATCH(0);
OUT_BATCH((depth - 1) << 21);
ADVANCE_BATCH();
}
/* 3DSTATE_HIER_DEPTH_BUFFER */
{
struct intel_miptree_aux_buffer *hiz_buf = params->depth.mt->hiz_buf;
BEGIN_BATCH(3);
OUT_BATCH((GEN7_3DSTATE_HIER_DEPTH_BUFFER << 16) | (3 - 2));
OUT_BATCH((mocs << 25) |
(hiz_buf->pitch - 1));
OUT_RELOC(hiz_buf->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
0);
ADVANCE_BATCH();
}
/* 3DSTATE_STENCIL_BUFFER */
{
BEGIN_BATCH(3);
OUT_BATCH((GEN7_3DSTATE_STENCIL_BUFFER << 16) | (3 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
}
}
void
gen7_emit_depth_stencil_hiz(struct brw_context *brw,
struct intel_mipmap_tree *depth_mt,
uint32_t depth_offset, uint32_t depthbuffer_format,
uint32_t depth_surface_type,
struct intel_mipmap_tree *stencil_mt,
bool hiz, bool separate_stencil,
uint32_t width, uint32_t height,
uint32_t tile_x, uint32_t tile_y)
{
struct gl_context *ctx = &brw->ctx;
const uint8_t mocs = GEN7_MOCS_L3;
struct gl_framebuffer *fb = ctx->DrawBuffer;
uint32_t surftype;
unsigned int depth = 1;
unsigned int min_array_element;
GLenum gl_target = GL_TEXTURE_2D;
unsigned int lod;
const struct intel_mipmap_tree *mt = depth_mt ? depth_mt : stencil_mt;
const struct intel_renderbuffer *irb = NULL;
const struct gl_renderbuffer *rb = NULL;
/* Skip repeated NULL depth/stencil emits (think 2D rendering). */
if (!mt && brw->no_depth_or_stencil) {
assert(brw->hw_ctx);
return;
}
brw_emit_depth_stall_flushes(brw);
irb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
if (!irb)
irb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
rb = (struct gl_renderbuffer*) irb;
if (rb) {
depth = MAX2(irb->layer_count, 1);
if (rb->TexImage)
gl_target = rb->TexImage->TexObject->Target;
}
switch (gl_target) {
case GL_TEXTURE_CUBE_MAP_ARRAY:
case GL_TEXTURE_CUBE_MAP:
/* The PRM claims that we should use BRW_SURFACE_CUBE for this
* situation, but experiments show that gl_Layer doesn't work when we do
* this. So we use BRW_SURFACE_2D, since for rendering purposes this is
* equivalent.
*/
surftype = BRW_SURFACE_2D;
depth *= 6;
break;
case GL_TEXTURE_3D:
assert(mt);
depth = MAX2(mt->logical_depth0, 1);
/* fallthrough */
default:
surftype = translate_tex_target(gl_target);
break;
}
min_array_element = irb ? irb->mt_layer : 0;
lod = irb ? irb->mt_level - irb->mt->first_level : 0;
if (mt) {
width = mt->logical_width0;
height = mt->logical_height0;
}
/* _NEW_DEPTH, _NEW_STENCIL, _NEW_BUFFERS */
BEGIN_BATCH(7);
/* 3DSTATE_DEPTH_BUFFER dw0 */
OUT_BATCH(GEN7_3DSTATE_DEPTH_BUFFER << 16 | (7 - 2));
/* 3DSTATE_DEPTH_BUFFER dw1 */
OUT_BATCH((depth_mt ? depth_mt->pitch - 1 : 0) |
(depthbuffer_format << 18) |
((hiz ? 1 : 0) << 22) |
((stencil_mt != NULL && ctx->Stencil._WriteEnabled) << 27) |
((ctx->Depth.Mask != 0) << 28) |
(surftype << 29));
/* 3DSTATE_DEPTH_BUFFER dw2 */
if (depth_mt) {
OUT_RELOC(depth_mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
0);
} else {
OUT_BATCH(0);
}
/* 3DSTATE_DEPTH_BUFFER dw3 */
OUT_BATCH(((width - 1) << 4) |
((height - 1) << 18) |
lod);
/* 3DSTATE_DEPTH_BUFFER dw4 */
OUT_BATCH(((depth - 1) << 21) |
(min_array_element << 10) |
mocs);
/* 3DSTATE_DEPTH_BUFFER dw5 */
OUT_BATCH(0);
/* 3DSTATE_DEPTH_BUFFER dw6 */
OUT_BATCH((depth - 1) << 21);
ADVANCE_BATCH();
if (!hiz) {
BEGIN_BATCH(3);
OUT_BATCH(GEN7_3DSTATE_HIER_DEPTH_BUFFER << 16 | (3 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
} else {
struct intel_miptree_aux_buffer *hiz_buf = depth_mt->hiz_buf;
BEGIN_BATCH(3);
OUT_BATCH(GEN7_3DSTATE_HIER_DEPTH_BUFFER << 16 | (3 - 2));
OUT_BATCH((mocs << 25) |
(hiz_buf->pitch - 1));
OUT_RELOC(hiz_buf->bo,
I915_GEM_DOMAIN_RENDER,
I915_GEM_DOMAIN_RENDER,
0);
ADVANCE_BATCH();
}
if (stencil_mt == NULL) {
BEGIN_BATCH(3);
OUT_BATCH(GEN7_3DSTATE_STENCIL_BUFFER << 16 | (3 - 2));
OUT_BATCH(0);
OUT_BATCH(0);
ADVANCE_BATCH();
} else {
const int enabled = brw->is_haswell ? HSW_STENCIL_ENABLED : 0;
BEGIN_BATCH(3);
OUT_BATCH(GEN7_3DSTATE_STENCIL_BUFFER << 16 | (3 - 2));
/* The stencil buffer has quirky pitch requirements. From the
* Sandybridge PRM, Volume 2 Part 1, page 329 (3DSTATE_STENCIL_BUFFER
* dword 1 bits 16:0 - 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.
*
* While the Ivybridge PRM lacks this comment, the BSpec contains the
* same text, and experiments indicate that this is necessary.
*/
OUT_BATCH(enabled |
mocs << 25 |
(2 * stencil_mt->pitch - 1));
OUT_RELOC(stencil_mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
0);
ADVANCE_BATCH();
}
BEGIN_BATCH(3);
OUT_BATCH(GEN7_3DSTATE_CLEAR_PARAMS << 16 | (3 - 2));
OUT_BATCH(depth_mt ? depth_mt->depth_clear_value : 0);
OUT_BATCH(1);
ADVANCE_BATCH();
brw->no_depth_or_stencil = !mt;
}
void
gen6_emit_depth_stencil_hiz(struct brw_context *brw,
struct intel_mipmap_tree *depth_mt,
uint32_t depth_offset, uint32_t depthbuffer_format,
uint32_t depth_surface_type,
struct intel_mipmap_tree *stencil_mt,
bool hiz, bool separate_stencil,
uint32_t width, uint32_t height,
uint32_t tile_x, uint32_t tile_y)
{
struct gl_context *ctx = &brw->ctx;
struct gl_framebuffer *fb = ctx->DrawBuffer;
uint32_t surftype;
unsigned int depth = 1;
GLenum gl_target = GL_TEXTURE_2D;
unsigned int lod;
const struct intel_mipmap_tree *mt = depth_mt ? depth_mt : stencil_mt;
const struct intel_renderbuffer *irb = NULL;
const struct gl_renderbuffer *rb = NULL;
/* Enable the hiz bit if we're doing separate stencil, 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
*/
bool enable_hiz_ss = hiz || separate_stencil;
brw_emit_depth_stall_flushes(brw);
irb = intel_get_renderbuffer(fb, BUFFER_DEPTH);
if (!irb)
irb = intel_get_renderbuffer(fb, BUFFER_STENCIL);
rb = (struct gl_renderbuffer*) irb;
if (rb) {
depth = MAX2(irb->layer_count, 1);
if (rb->TexImage)
gl_target = rb->TexImage->TexObject->Target;
}
switch (gl_target) {
case GL_TEXTURE_CUBE_MAP_ARRAY:
case GL_TEXTURE_CUBE_MAP:
/* The PRM claims that we should use BRW_SURFACE_CUBE for this
* situation, but experiments show that gl_Layer doesn't work when we do
* this. So we use BRW_SURFACE_2D, since for rendering purposes this is
* equivalent.
*/
surftype = BRW_SURFACE_2D;
depth *= 6;
break;
case GL_TEXTURE_3D:
assert(mt);
depth = MAX2(mt->logical_depth0, 1);
/* fallthrough */
default:
surftype = translate_tex_target(gl_target);
break;
}
const unsigned min_array_element = irb ? irb->mt_layer : 0;
lod = irb ? irb->mt_level - irb->mt->first_level : 0;
if (mt) {
width = mt->logical_width0;
height = mt->logical_height0;
}
BEGIN_BATCH(7);
/* 3DSTATE_DEPTH_BUFFER dw0 */
OUT_BATCH(_3DSTATE_DEPTH_BUFFER << 16 | (7 - 2));
/* 3DSTATE_DEPTH_BUFFER dw1 */
OUT_BATCH((depth_mt ? depth_mt->pitch - 1 : 0) |
(depthbuffer_format << 18) |
((enable_hiz_ss ? 1 : 0) << 21) | /* separate stencil enable */
((enable_hiz_ss ? 1 : 0) << 22) | /* hiz enable */
(BRW_TILEWALK_YMAJOR << 26) |
((depth_mt ? depth_mt->tiling != I915_TILING_NONE : 1)
<< 27) |
(surftype << 29));
/* 3DSTATE_DEPTH_BUFFER dw2 */
if (depth_mt) {
OUT_RELOC(depth_mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
0);
} else {
OUT_BATCH(0);
}
/* 3DSTATE_DEPTH_BUFFER dw3 */
OUT_BATCH(((width - 1) << 6) |
((height - 1) << 19) |
lod << 2);
/* 3DSTATE_DEPTH_BUFFER dw4 */
OUT_BATCH((depth - 1) << 21 |
min_array_element << 10 |
(depth - 1) << 1);
/* 3DSTATE_DEPTH_BUFFER dw5 */
OUT_BATCH(0);
assert(tile_x == 0 && tile_y == 0);
/* 3DSTATE_DEPTH_BUFFER dw6 */
OUT_BATCH(0);
ADVANCE_BATCH();
if (hiz || 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) {
struct intel_mipmap_tree *hiz_mt = depth_mt->hiz_buf->mt;
uint32_t offset = 0;
if (hiz_mt->array_layout == ALL_SLICES_AT_EACH_LOD) {
offset = intel_miptree_get_aligned_offset(
hiz_mt,
hiz_mt->level[lod].level_x,
hiz_mt->level[lod].level_y,
false);
}
BEGIN_BATCH(3);
OUT_BATCH((_3DSTATE_HIER_DEPTH_BUFFER << 16) | (3 - 2));
OUT_BATCH(hiz_mt->pitch - 1);
OUT_RELOC(hiz_mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
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) {
uint32_t offset = 0;
if (stencil_mt->array_layout == ALL_SLICES_AT_EACH_LOD) {
if (stencil_mt->format == MESA_FORMAT_S_UINT8) {
/* Note: we can't compute the stencil offset using
* intel_region_get_aligned_offset(), because stencil_region
* claims that the region is untiled even though it's W tiled.
*/
offset =
stencil_mt->level[lod].level_y * stencil_mt->pitch +
stencil_mt->level[lod].level_x * 64;
} else {
offset = intel_miptree_get_aligned_offset(
stencil_mt,
stencil_mt->level[lod].level_x,
stencil_mt->level[lod].level_y,
false);
}
}
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 * stencil_mt->pitch - 1);
OUT_RELOC(stencil_mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
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.
*/
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_CLEAR_PARAMS << 16 |
GEN5_DEPTH_CLEAR_VALID |
(2 - 2));
OUT_BATCH(depth_mt ? depth_mt->depth_clear_value : 0);
ADVANCE_BATCH();
}
void
brw_emit_depth_stencil_hiz(struct brw_context *brw,
struct intel_mipmap_tree *depth_mt,
uint32_t depth_offset, uint32_t depthbuffer_format,
uint32_t depth_surface_type,
struct intel_mipmap_tree *stencil_mt,
bool hiz, bool separate_stencil,
uint32_t width, uint32_t height,
uint32_t tile_x, uint32_t tile_y)
{
/* Enable the hiz bit if we're doing separate stencil, 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
*/
bool enable_hiz_ss = hiz || separate_stencil;
/* 3DSTATE_DEPTH_BUFFER, 3DSTATE_STENCIL_BUFFER are both
* non-pipelined state that will need the PIPE_CONTROL workaround.
*/
if (brw->gen == 6) {
brw_emit_depth_stall_flushes(brw);
}
unsigned int len;
if (brw->gen >= 6)
len = 7;
else if (brw->is_g4x || brw->gen == 5)
len = 6;
else
len = 5;
BEGIN_BATCH(len);
OUT_BATCH(_3DSTATE_DEPTH_BUFFER << 16 | (len - 2));
OUT_BATCH((depth_mt ? depth_mt->pitch - 1 : 0) |
(depthbuffer_format << 18) |
((enable_hiz_ss ? 1 : 0) << 21) | /* separate stencil enable */
((enable_hiz_ss ? 1 : 0) << 22) | /* hiz enable */
(BRW_TILEWALK_YMAJOR << 26) |
((depth_mt ? depth_mt->tiling != I915_TILING_NONE : 1)
<< 27) |
(depth_surface_type << 29));
if (depth_mt) {
OUT_RELOC(depth_mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
depth_offset);
} else {
OUT_BATCH(0);
}
OUT_BATCH(((width + tile_x - 1) << 6) |
((height + tile_y - 1) << 19));
OUT_BATCH(0);
if (brw->is_g4x || brw->gen >= 5)
OUT_BATCH(tile_x | (tile_y << 16));
else
assert(tile_x == 0 && tile_y == 0);
if (brw->gen >= 6)
OUT_BATCH(0);
ADVANCE_BATCH();
if (hiz || 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) {
struct intel_mipmap_tree *hiz_mt = depth_mt->hiz_buf->mt;
BEGIN_BATCH(3);
OUT_BATCH((_3DSTATE_HIER_DEPTH_BUFFER << 16) | (3 - 2));
OUT_BATCH(hiz_mt->pitch - 1);
OUT_RELOC(hiz_mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
brw->depthstencil.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) {
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 * stencil_mt->pitch - 1);
OUT_RELOC(stencil_mt->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
brw->depthstencil.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 (brw->gen >= 6 || hiz) {
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_CLEAR_PARAMS << 16 |
GEN5_DEPTH_CLEAR_VALID |
(2 - 2));
OUT_BATCH(depth_mt ? depth_mt->depth_clear_value : 0);
ADVANCE_BATCH();
}
}
