const unsigned vec4_size_in_bytes = 4 * sizeof(float);
const unsigned max_num_varyings =
DIV_ROUND_UP(sizeof(params->wm_inputs), vec4_size_in_bytes);
- const unsigned num_varyings = params->wm_prog_data->num_varying_inputs;
+ const unsigned num_varyings =
+ params->wm_prog_data ? params->wm_prog_data->num_varying_inputs : 0;
- *size = num_varyings * vec4_size_in_bytes;
+ *size = 16 + num_varyings * vec4_size_in_bytes;
- const float *const inputs_src = (const float *)¶ms->wm_inputs;
- float *inputs = blorp_alloc_vertex_buffer(batch, *size, addr);
+ const uint32_t *const inputs_src = (const uint32_t *)¶ms->wm_inputs;
+ uint32_t *inputs = blorp_alloc_vertex_buffer(batch, *size, addr);
- /* Walk over the attribute slots, determine if the attribute is used by
- * the program and when necessary copy the values from the input storage to
- * the vertex data buffer.
- */
- for (unsigned i = 0; i < max_num_varyings; i++) {
- const gl_varying_slot attr = VARYING_SLOT_VAR0 + i;
+ /* Zero data for the VUE header */
+ memset(inputs, 0, 4 * sizeof(uint32_t));
+ inputs += 4;
+
+ if (params->wm_prog_data) {
+ /* Walk over the attribute slots, determine if the attribute is used by
+ * the program and when necessary copy the values from the input storage
+ * to the vertex data buffer.
+ */
+ for (unsigned i = 0; i < max_num_varyings; i++) {
+ const gl_varying_slot attr = VARYING_SLOT_VAR0 + i;
- if (!(params->wm_prog_data->inputs_read & (1ull << attr)))
- continue;
+ const int input_index = params->wm_prog_data->urb_setup[attr];
+ if (input_index < 0)
+ continue;
- memcpy(inputs, inputs_src + i * 4, vec4_size_in_bytes);
+ memcpy(inputs, inputs_src + i * 4, vec4_size_in_bytes);
- inputs += 4;
+ inputs += 4;
+ }
}
}
struct GENX(VERTEX_BUFFER_STATE) vb[2];
memset(vb, 0, sizeof(vb));
- unsigned num_buffers = 1;
-
uint32_t size;
blorp_emit_vertex_data(batch, params, &vb[0].BufferStartingAddress, &size);
vb[0].VertexBufferIndex = 0;
vb[0].EndAddress.offset += size - 1;
#endif
- if (params->wm_prog_data && params->wm_prog_data->num_varying_inputs) {
- blorp_emit_input_varying_data(batch, params,
- &vb[1].BufferStartingAddress, &size);
- vb[1].VertexBufferIndex = 1;
- vb[1].BufferPitch = 0;
- vb[1].VertexBufferMOCS = batch->blorp->mocs.vb;
+ blorp_emit_input_varying_data(batch, params,
+ &vb[1].BufferStartingAddress, &size);
+ vb[1].VertexBufferIndex = 1;
+ vb[1].BufferPitch = 0;
+ vb[1].VertexBufferMOCS = batch->blorp->mocs.vb;
#if GEN_GEN >= 7
- vb[1].AddressModifyEnable = true;
+ vb[1].AddressModifyEnable = true;
#endif
#if GEN_GEN >= 8
- vb[1].BufferSize = size;
+ vb[1].BufferSize = size;
#else
- vb[1].BufferAccessType = INSTANCEDATA;
- vb[1].EndAddress = vb[1].BufferStartingAddress;
- vb[1].EndAddress.offset += size - 1;
+ vb[1].BufferAccessType = INSTANCEDATA;
+ vb[1].EndAddress = vb[1].BufferStartingAddress;
+ vb[1].EndAddress.offset += size - 1;
#endif
- num_buffers++;
- }
- const unsigned num_dwords =
- 1 + GENX(VERTEX_BUFFER_STATE_length) * num_buffers;
+ const unsigned num_dwords = 1 + GENX(VERTEX_BUFFER_STATE_length) * 2;
uint32_t *dw = blorp_emitn(batch, GENX(3DSTATE_VERTEX_BUFFERS), num_dwords);
- for (unsigned i = 0; i < num_buffers; i++) {
+ for (unsigned i = 0; i < 2; i++) {
GENX(VERTEX_BUFFER_STATE_pack)(batch, dw, &vb[i]);
dw += GENX(VERTEX_BUFFER_STATE_length);
}
*
* See the vertex element setup below.
*/
- ve[0].VertexBufferIndex = 0;
+ ve[0].VertexBufferIndex = 1;
ve[0].Valid = true;
ve[0].SourceElementFormat = ISL_FORMAT_R32G32B32A32_FLOAT;
ve[0].SourceElementOffset = 0;
ve[i + 2].VertexBufferIndex = 1;
ve[i + 2].Valid = true;
ve[i + 2].SourceElementFormat = ISL_FORMAT_R32G32B32A32_FLOAT;
- ve[i + 2].SourceElementOffset = i * 4 * sizeof(float);
+ ve[i + 2].SourceElementOffset = 16 + i * 4 * sizeof(float);
ve[i + 2].Component0Control = VFCOMP_STORE_SRC;
ve[i + 2].Component1Control = VFCOMP_STORE_SRC;
ve[i + 2].Component2Control = VFCOMP_STORE_SRC;
blorp_emit_sf_config(struct blorp_batch *batch,
const struct blorp_params *params)
{
- const struct brw_blorp_prog_data *prog_data = params->wm_prog_data;
+ const struct brw_wm_prog_data *prog_data = params->wm_prog_data;
/* 3DSTATE_SF
*
sf.FrontFaceFillMode = FILL_MODE_SOLID;
sf.BackFaceFillMode = FILL_MODE_SOLID;
- sf.MultisampleRasterizationMode = params->dst.surf.samples > 1 ?
+ sf.MultisampleRasterizationMode = params->num_samples > 1 ?
MSRASTMODE_ON_PATTERN : MSRASTMODE_OFF_PIXEL;
#if GEN_GEN == 7
sf.FrontFaceFillMode = FILL_MODE_SOLID;
sf.BackFaceFillMode = FILL_MODE_SOLID;
- sf.MultisampleRasterizationMode = params->dst.surf.samples > 1 ?
+ sf.MultisampleRasterizationMode = params->num_samples > 1 ?
MSRASTMODE_ON_PATTERN : MSRASTMODE_OFF_PIXEL;
sf.VertexURBEntryReadOffset = 1;
blorp_emit_ps_config(struct blorp_batch *batch,
const struct blorp_params *params)
{
- const struct brw_blorp_prog_data *prog_data = params->wm_prog_data;
+ const struct brw_wm_prog_data *prog_data = params->wm_prog_data;
/* Even when thread dispatch is disabled, max threads (dw5.25:31) must be
* nonzero to prevent the GPU from hanging. While the documentation doesn't
if (prog_data) {
ps.DispatchGRFStartRegisterForConstantSetupData0 =
- prog_data->first_curbe_grf_0;
+ prog_data->base.dispatch_grf_start_reg;
ps.DispatchGRFStartRegisterForConstantSetupData2 =
- prog_data->first_curbe_grf_2;
+ prog_data->dispatch_grf_start_reg_2;
ps._8PixelDispatchEnable = prog_data->dispatch_8;
ps._16PixelDispatchEnable = prog_data->dispatch_16;
ps.KernelStartPointer0 = params->wm_prog_kernel;
ps.KernelStartPointer2 =
- params->wm_prog_kernel + prog_data->ksp_offset_2;
+ params->wm_prog_kernel + prog_data->prog_offset_2;
}
/* 3DSTATE_PS expects the number of threads per PSD, which is always 64;
if (prog_data) {
psx.PixelShaderValid = true;
psx.AttributeEnable = prog_data->num_varying_inputs > 0;
- psx.PixelShaderIsPerSample = prog_data->persample_msaa_dispatch;
+ psx.PixelShaderIsPerSample = prog_data->persample_dispatch;
}
if (params->src.enabled)
wm.ThreadDispatchEnable = true;
if (params->src.enabled)
- wm.PixelShaderKillPixel = true;
+ wm.PixelShaderKillsPixel = true;
- if (params->dst.surf.samples > 1) {
+ if (params->num_samples > 1) {
wm.MultisampleRasterizationMode = MSRASTMODE_ON_PATTERN;
wm.MultisampleDispatchMode =
- (prog_data && prog_data->persample_msaa_dispatch) ?
+ (prog_data && prog_data->persample_dispatch) ?
MSDISPMODE_PERSAMPLE : MSDISPMODE_PERPIXEL;
} else {
wm.MultisampleRasterizationMode = MSRASTMODE_OFF_PIXEL;
#endif
if (prog_data) {
- ps.DispatchGRFStartRegisterforConstantSetupData0 =
- prog_data->first_curbe_grf_0;
- ps.DispatchGRFStartRegisterforConstantSetupData2 =
- prog_data->first_curbe_grf_2;
+ ps.DispatchGRFStartRegisterForConstantSetupData0 =
+ prog_data->base.dispatch_grf_start_reg;
+ ps.DispatchGRFStartRegisterForConstantSetupData2 =
+ prog_data->dispatch_grf_start_reg_2;
ps.KernelStartPointer0 = params->wm_prog_kernel;
ps.KernelStartPointer2 =
- params->wm_prog_kernel + prog_data->ksp_offset_2;
+ params->wm_prog_kernel + prog_data->prog_offset_2;
ps._8PixelDispatchEnable = prog_data->dispatch_8;
ps._16PixelDispatchEnable = prog_data->dispatch_16;
if (prog_data) {
wm.ThreadDispatchEnable = true;
- wm.DispatchGRFStartRegisterforConstantSetupData0 =
- prog_data->first_curbe_grf_0;
- wm.DispatchGRFStartRegisterforConstantSetupData2 =
- prog_data->first_curbe_grf_2;
+ wm.DispatchGRFStartRegisterForConstantSetupData0 =
+ prog_data->base.dispatch_grf_start_reg;
+ wm.DispatchGRFStartRegisterForConstantSetupData2 =
+ prog_data->dispatch_grf_start_reg_2;
wm.KernelStartPointer0 = params->wm_prog_kernel;
wm.KernelStartPointer2 =
- params->wm_prog_kernel + prog_data->ksp_offset_2;
+ params->wm_prog_kernel + prog_data->prog_offset_2;
wm._8PixelDispatchEnable = prog_data->dispatch_8;
wm._16PixelDispatchEnable = prog_data->dispatch_16;
if (params->src.enabled) {
wm.SamplerCount = 1; /* Up to 4 samplers */
- wm.PixelShaderKillPixel = true; /* TODO: temporarily smash on */
+ wm.PixelShaderKillsPixel = true; /* TODO: temporarily smash on */
}
- if (params->dst.surf.samples > 1) {
+ if (params->num_samples > 1) {
wm.MultisampleRasterizationMode = MSRASTMODE_ON_PATTERN;
wm.MultisampleDispatchMode =
- (prog_data && prog_data->persample_msaa_dispatch) ?
+ (prog_data && prog_data->persample_dispatch) ?
MSDISPMODE_PERSAMPLE : MSDISPMODE_PERPIXEL;
} else {
wm.MultisampleRasterizationMode = MSRASTMODE_OFF_PIXEL;
return offset;
}
-struct surface_state_info {
- unsigned num_dwords;
- unsigned ss_align; /* Required alignment of RENDER_SURFACE_STATE in bytes */
- unsigned reloc_dw;
- unsigned aux_reloc_dw;
-};
-
-static const struct surface_state_info surface_state_infos[] = {
- [6] = {6, 32, 1, 0},
- [7] = {8, 32, 1, 6},
- [8] = {13, 64, 8, 10},
- [9] = {16, 64, 8, 10},
-};
-
static void
blorp_emit_surface_state(struct blorp_batch *batch,
const struct brw_blorp_surface_info *surface,
- uint32_t *state, uint32_t state_offset,
+ void *state, uint32_t state_offset,
bool is_render_target)
{
- const struct surface_state_info ss_info = surface_state_infos[GEN_GEN];
-
+ const struct isl_device *isl_dev = batch->blorp->isl_dev;
struct isl_surf surf = surface->surf;
if (surf.dim == ISL_SURF_DIM_1D &&
.aux_surf = &surface->aux_surf, .aux_usage = aux_usage,
.mocs = mocs, .clear_color = surface->clear_color);
- blorp_surface_reloc(batch, state_offset + ss_info.reloc_dw * 4,
+ blorp_surface_reloc(batch, state_offset + isl_dev->ss.addr_offset,
surface->addr, 0);
if (aux_usage != ISL_AUX_USAGE_NONE) {
* surface buffer addresses are always 4K page alinged.
*/
assert((surface->aux_addr.offset & 0xfff) == 0);
- blorp_surface_reloc(batch, state_offset + ss_info.aux_reloc_dw * 4,
- surface->aux_addr, state[ss_info.aux_reloc_dw]);
+ uint32_t *aux_addr = state + isl_dev->ss.aux_addr_offset;
+ blorp_surface_reloc(batch, state_offset + isl_dev->ss.aux_addr_offset,
+ surface->aux_addr, *aux_addr);
}
}
blorp_emit_surface_states(struct blorp_batch *batch,
const struct blorp_params *params)
{
+ const struct isl_device *isl_dev = batch->blorp->isl_dev;
uint32_t bind_offset, surface_offsets[2];
void *surface_maps[2];
- const unsigned ss_size = GENX(RENDER_SURFACE_STATE_length) * 4;
- const unsigned ss_align = GENX(RENDER_SURFACE_STATE_length) > 8 ? 64 : 32;
-
unsigned num_surfaces = 1 + params->src.enabled;
- blorp_alloc_binding_table(batch, num_surfaces, ss_size, ss_align,
+ blorp_alloc_binding_table(batch, num_surfaces,
+ isl_dev->ss.size, isl_dev->ss.align,
&bind_offset, surface_offsets, surface_maps);
if (params->dst.enabled) {
}
#if GEN_GEN >= 7
+ blorp_emit(batch, GENX(3DSTATE_BINDING_TABLE_POINTERS_VS), bt);
+ blorp_emit(batch, GENX(3DSTATE_BINDING_TABLE_POINTERS_HS), bt);
+ blorp_emit(batch, GENX(3DSTATE_BINDING_TABLE_POINTERS_DS), bt);
+ blorp_emit(batch, GENX(3DSTATE_BINDING_TABLE_POINTERS_GS), bt);
+
blorp_emit(batch, GENX(3DSTATE_BINDING_TABLE_POINTERS_PS), bt) {
bt.PointertoPSBindingTable = bind_offset;
}
blorp_emit_3dstate_multisample(struct blorp_batch *batch,
const struct blorp_params *params)
{
- const unsigned samples = params->dst.surf.samples;
-
blorp_emit(batch, GENX(3DSTATE_MULTISAMPLE), ms) {
- ms.NumberofMultisamples = __builtin_ffs(samples) - 1;
+ ms.NumberofMultisamples = __builtin_ffs(params->num_samples) - 1;
#if GEN_GEN >= 8
/* The PRM says that this bit is valid only for DX9:
#elif GEN_GEN >= 7
ms.PixelLocation = PIXLOC_CENTER;
- switch (samples) {
+ switch (params->num_samples) {
case 1:
GEN_SAMPLE_POS_1X(ms.Sample);
break;
blorp_emit_3dstate_multisample(batch, params);
blorp_emit(batch, GENX(3DSTATE_SAMPLE_MASK), mask) {
- mask.SampleMask = (1 << params->dst.surf.samples) - 1;
+ mask.SampleMask = (1 << params->num_samples) - 1;
}
/* From the BSpec, 3D Pipeline > Geometry > Vertex Shader > State,
blorp_emit_viewport_state(batch, params);
- blorp_emit_depth_stencil_config(batch, params);
+ if (!(batch->flags & BLORP_BATCH_NO_EMIT_DEPTH_STENCIL))
+ blorp_emit_depth_stencil_config(batch, params);
blorp_emit(batch, GENX(3DPRIMITIVE), prim) {
prim.VertexAccessType = SEQUENTIAL;
projects / mesa.git / blobdiff