* 3DSTATE_URB_DS
* 3DSTATE_URB_GS
*
- * If the 3DSTATE_URB_VS is emitted, than the others must be also. From the
- * BSpec, Volume 2a "3D Pipeline Overview", Section 1.7.1 3DSTATE_URB_VS:
+ * If the 3DSTATE_URB_VS is emitted, than the others must be also.
+ * From the Ivybridge PRM, Volume 2 Part 1, section 1.7.1 3DSTATE_URB_VS:
+ *
* 3DSTATE_URB_HS, 3DSTATE_URB_DS, and 3DSTATE_URB_GS must also be
* programmed in order for the programming of this state to be
* valid.
const brw_blorp_params *params,
uint32_t cc_blend_state_offset)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_BLEND_STATE_POINTERS << 16 | (2 - 2));
OUT_BATCH(cc_blend_state_offset | 1);
const brw_blorp_params *params,
uint32_t cc_state_offset)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_CC_STATE_POINTERS << 16 | (2 - 2));
OUT_BATCH(cc_state_offset | 1);
gen7_blorp_emit_cc_viewport(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
struct brw_cc_viewport *ccv;
uint32_t cc_vp_offset;
const brw_blorp_params *params,
uint32_t depthstencil_offset)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_DEPTH_STENCIL_STATE_POINTERS << 16 | (2 - 2));
OUT_BATCH(depthstencil_offset | 1);
uint32_t read_domains, uint32_t write_domain,
bool is_render_target)
{
- struct intel_context *intel = &brw->intel;
-
uint32_t wm_surf_offset;
uint32_t width = surface->width;
uint32_t height = surface->height;
*/
struct intel_region *region = surface->mt->region;
uint32_t tile_x, tile_y;
+ const uint8_t mocs = GEN7_MOCS_L3;
uint32_t tiling = surface->map_stencil_as_y_tiled
? I915_TILING_Y : region->tiling;
assert(tile_x % 4 == 0);
assert(tile_y % 2 == 0);
surf[5] = SET_FIELD(tile_x / 4, BRW_SURFACE_X_OFFSET) |
- SET_FIELD(tile_y / 2, BRW_SURFACE_Y_OFFSET);
+ SET_FIELD(tile_y / 2, BRW_SURFACE_Y_OFFSET) |
+ SET_FIELD(mocs, GEN7_SURFACE_MOCS);
surf[2] = SET_FIELD(width - 1, GEN7_SURFACE_WIDTH) |
SET_FIELD(height - 1, GEN7_SURFACE_HEIGHT);
is_render_target);
}
- if (intel->is_haswell) {
- surf[7] = SET_FIELD(HSW_SCS_RED, GEN7_SURFACE_SCS_R) |
- SET_FIELD(HSW_SCS_GREEN, GEN7_SURFACE_SCS_G) |
- SET_FIELD(HSW_SCS_BLUE, GEN7_SURFACE_SCS_B) |
- SET_FIELD(HSW_SCS_ALPHA, GEN7_SURFACE_SCS_A);
+ surf[7] = surface->mt->fast_clear_color_value;
+
+ if (brw->is_haswell) {
+ surf[7] |= (SET_FIELD(HSW_SCS_RED, GEN7_SURFACE_SCS_R) |
+ SET_FIELD(HSW_SCS_GREEN, GEN7_SURFACE_SCS_G) |
+ SET_FIELD(HSW_SCS_BLUE, GEN7_SURFACE_SCS_B) |
+ SET_FIELD(HSW_SCS_ALPHA, GEN7_SURFACE_SCS_A));
}
/* Emit relocation to surface contents */
- drm_intel_bo_emit_reloc(intel->batch.bo,
+ drm_intel_bo_emit_reloc(brw->batch.bo,
wm_surf_offset + 4,
region->bo,
surf[1] - region->bo->offset,
gen7_blorp_emit_vs_disable(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(7);
OUT_BATCH(_3DSTATE_CONSTANT_VS << 16 | (7 - 2));
OUT_BATCH(0);
gen7_blorp_emit_hs_disable(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(7);
OUT_BATCH(_3DSTATE_CONSTANT_HS << 16 | (7 - 2));
OUT_BATCH(0);
gen7_blorp_emit_te_disable(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(4);
OUT_BATCH(_3DSTATE_TE << 16 | (4 - 2));
OUT_BATCH(0);
gen7_blorp_emit_ds_disable(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(7);
OUT_BATCH(_3DSTATE_CONSTANT_DS << 16 | (7 - 2));
OUT_BATCH(0);
gen7_blorp_emit_gs_disable(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(7);
OUT_BATCH(_3DSTATE_CONSTANT_GS << 16 | (7 - 2));
OUT_BATCH(0);
gen7_blorp_emit_streamout_disable(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(3);
OUT_BATCH(_3DSTATE_STREAMOUT << 16 | (3 - 2));
OUT_BATCH(0);
gen7_blorp_emit_sf_config(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
/* 3DSTATE_SF
*
* Disable ViewportTransformEnable (dw1.1)
const brw_blorp_params *params,
brw_blorp_prog_data *prog_data)
{
- struct intel_context *intel = &brw->intel;
-
uint32_t dw1 = 0, dw2 = 0;
switch (params->hiz_op) {
*
* Pixel shader dispatch is disabled above in 3DSTATE_WM, dw1.29. Despite
* that, thread dispatch info must still be specified.
- * - Maximum Number of Threads (dw4.24:31) must be nonzero, as the BSpec
- * states that the valid range for this field is [0x3, 0x2f].
+ * - Maximum Number of Threads (dw4.24:31) must be nonzero, as the
+ * valid range for this field is [0x3, 0x2f].
* - A dispatch mode must be given; that is, at least one of the
* "N Pixel Dispatch Enable" (N=8,16,32) fields must be set. This was
* discovered through simulator error messages.
uint32_t prog_offset,
brw_blorp_prog_data *prog_data)
{
- struct intel_context *intel = &brw->intel;
uint32_t dw2, dw4, dw5;
- const int max_threads_shift = brw->intel.is_haswell ?
+ const int max_threads_shift = brw->is_haswell ?
HSW_PS_MAX_THREADS_SHIFT : IVB_PS_MAX_THREADS_SHIFT;
dw2 = dw4 = dw5 = 0;
*/
dw4 |= GEN7_PS_16_DISPATCH_ENABLE;
- if (intel->is_haswell)
+ if (brw->is_haswell)
dw4 |= SET_FIELD(1, HSW_PS_SAMPLE_MASK); /* 1 sample for now */
if (params->use_wm_prog) {
dw2 |= 1 << GEN7_PS_SAMPLER_COUNT_SHIFT; /* Up to 4 samplers */
dw5 |= prog_data->first_curbe_grf << GEN7_PS_DISPATCH_START_GRF_SHIFT_0;
}
+ switch (params->fast_clear_op) {
+ case GEN7_FAST_CLEAR_OP_FAST_CLEAR:
+ dw4 |= GEN7_PS_RENDER_TARGET_FAST_CLEAR_ENABLE;
+ break;
+ case GEN7_FAST_CLEAR_OP_RESOLVE:
+ dw4 |= GEN7_PS_RENDER_TARGET_RESOLVE_ENABLE;
+ break;
+ default:
+ break;
+ }
+
BEGIN_BATCH(8);
OUT_BATCH(_3DSTATE_PS << 16 | (8 - 2));
OUT_BATCH(params->use_wm_prog ? prog_offset : 0);
const brw_blorp_params *params,
uint32_t wm_bind_bo_offset)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_BINDING_TABLE_POINTERS_PS << 16 | (2 - 2));
OUT_BATCH(wm_bind_bo_offset);
const brw_blorp_params *params,
uint32_t sampler_offset)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(2);
OUT_BATCH(_3DSTATE_SAMPLER_STATE_POINTERS_PS << 16 | (2 - 2));
OUT_BATCH(sampler_offset);
const brw_blorp_params *params,
uint32_t wm_push_const_offset)
{
- struct intel_context *intel = &brw->intel;
+ const uint8_t mocs = GEN7_MOCS_L3;
/* Make sure the push constants fill an exact integer number of
* registers.
(7 - 2));
OUT_BATCH(BRW_BLORP_NUM_PUSH_CONST_REGS);
OUT_BATCH(0);
- OUT_BATCH(wm_push_const_offset);
+ OUT_BATCH(wm_push_const_offset | mocs);
OUT_BATCH(0);
OUT_BATCH(0);
OUT_BATCH(0);
gen7_blorp_emit_constant_ps_disable(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(7);
OUT_BATCH(_3DSTATE_CONSTANT_PS << 16 | (7 - 2));
OUT_BATCH(0);
gen7_blorp_emit_depth_stencil_config(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
- struct gl_context *ctx = &intel->ctx;
- uint32_t draw_x = params->depth.x_offset;
- uint32_t draw_y = params->depth.y_offset;
- uint32_t tile_mask_x, tile_mask_y;
-
- brw_get_depthstencil_tile_masks(params->depth.mt,
- params->depth.level,
- params->depth.layer,
- NULL,
- &tile_mask_x, &tile_mask_y);
+ 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;
+ }
- /* 3DSTATE_DEPTH_BUFFER */
- {
- uint32_t tile_x = draw_x & tile_mask_x;
- uint32_t tile_y = draw_y & tile_mask_y;
- uint32_t offset =
- intel_region_get_aligned_offset(params->depth.mt->region,
- draw_x & ~tile_mask_x,
- draw_y & ~tile_mask_y, false);
-
- /* 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.
+ 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)
*/
- WARN_ONCE((tile_x & 7) || (tile_y & 7),
- "Depth/stencil buffer needs alignment to 8-pixel boundaries.\n"
- "Truncating offset, bad rendering may occur.\n");
- tile_x &= ~7;
- tile_y &= ~7;
+ surfwidth = params->depth.width;
+ surfheight = params->depth.height;
+ } else {
+ surfwidth = params->depth.mt->physical_width0;
+ surfheight = params->depth.mt->physical_height0;
+ }
- intel_emit_depth_stall_flushes(intel);
+ /* 3DSTATE_DEPTH_BUFFER */
+ {
+ intel_emit_depth_stall_flushes(brw);
BEGIN_BATCH(7);
OUT_BATCH(GEN7_3DSTATE_DEPTH_BUFFER << 16 | (7 - 2));
params->depth_format << 18 |
1 << 22 | /* hiz enable */
1 << 28 | /* depth write */
- BRW_SURFACE_2D << 29);
+ surftype << 29);
OUT_RELOC(params->depth.mt->region->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
- offset);
- OUT_BATCH((params->depth.width + tile_x - 1) << 4 |
- (params->depth.height + tile_y - 1) << 18);
- OUT_BATCH(0);
- OUT_BATCH(tile_x |
- tile_y << 16);
+ 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_region *hiz_region = params->depth.mt->hiz_mt->region;
- uint32_t hiz_offset =
- intel_region_get_aligned_offset(hiz_region,
- draw_x & ~tile_mask_x,
- (draw_y & ~tile_mask_y) / 2, false);
BEGIN_BATCH(3);
OUT_BATCH((GEN7_3DSTATE_HIER_DEPTH_BUFFER << 16) | (3 - 2));
- OUT_BATCH(hiz_region->pitch - 1);
+ OUT_BATCH((mocs << 25) |
+ (hiz_region->pitch - 1));
OUT_RELOC(hiz_region->bo,
I915_GEM_DOMAIN_RENDER, I915_GEM_DOMAIN_RENDER,
- hiz_offset);
+ 0);
ADVANCE_BATCH();
}
gen7_blorp_emit_depth_disable(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
+ intel_emit_depth_stall_flushes(brw);
BEGIN_BATCH(7);
OUT_BATCH(GEN7_3DSTATE_DEPTH_BUFFER << 16 | (7 - 2));
/* 3DSTATE_CLEAR_PARAMS
*
- * From the BSpec, Volume 2a.11 Windower, Section 1.5.6.3.2
+ * From the Ivybridge PRM, Volume 2 Part 1, Section 11.5.5.4
* 3DSTATE_CLEAR_PARAMS:
- * [DevIVB] 3DSTATE_CLEAR_PARAMS must always be programmed in the along
+ * 3DSTATE_CLEAR_PARAMS must always be programmed in the along
* with the other Depth/Stencil state commands(i.e. 3DSTATE_DEPTH_BUFFER,
* 3DSTATE_STENCIL_BUFFER, or 3DSTATE_HIER_DEPTH_BUFFER).
*/
gen7_blorp_emit_clear_params(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(3);
OUT_BATCH(GEN7_3DSTATE_CLEAR_PARAMS << 16 | (3 - 2));
OUT_BATCH(params->depth.mt ? params->depth.mt->depth_clear_value : 0);
gen7_blorp_emit_primitive(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct intel_context *intel = &brw->intel;
-
BEGIN_BATCH(7);
OUT_BATCH(CMD_3D_PRIM << 16 | (7 - 2));
OUT_BATCH(GEN7_3DPRIM_VERTEXBUFFER_ACCESS_SEQUENTIAL |
* \copydoc gen6_blorp_exec()
*/
void
-gen7_blorp_exec(struct intel_context *intel,
+gen7_blorp_exec(struct brw_context *brw,
const brw_blorp_params *params)
{
- struct gl_context *ctx = &intel->ctx;
- struct brw_context *brw = brw_context(ctx);
brw_blorp_prog_data *prog_data = NULL;
uint32_t cc_blend_state_offset = 0;
uint32_t cc_state_offset = 0;
projects / mesa.git / blobdiff