i965: Convert GS_STATE to genxml.

Merge the code with gen6+ 3DSTATE_GS, and delete brw_gs_state.c,
together with brw_gs_unit_state.

Signed-off-by: Rafael Antognolli <rafael.antognolli@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This commit is contained in:
Rafael Antognolli
2017-06-07 13:03:43 -07:00
parent 2936388205
commit 9a9c7e452b
5 changed files with 54 additions and 172 deletions

View File

@@ -24,7 +24,6 @@ i965_FILES = \
brw_formatquery.c \
brw_gs.c \
brw_gs.h \
brw_gs_state.c \
brw_gs_surface_state.c \
brw_link.cpp \
brw_meta_util.c \

View File

@@ -1,101 +0,0 @@
/*
Copyright (C) Intel Corp. 2006. All Rights Reserved.
Intel funded Tungsten Graphics to
develop this 3D driver.
Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
"Software"), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to
permit persons to whom the Software is furnished to do so, subject to
the following conditions:
The above copyright notice and this permission notice (including the
next paragraph) shall be included in all copies or substantial
portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
IN NO EVENT SHALL THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE
LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
**********************************************************************/
/*
* Authors:
* Keith Whitwell <keithw@vmware.com>
*/
#include "brw_context.h"
#include "brw_state.h"
#include "brw_defines.h"
#include "intel_batchbuffer.h"
static void
brw_upload_gs_unit(struct brw_context *brw)
{
struct brw_gs_unit_state *gs;
gs = brw_state_batch(brw, sizeof(*gs), 32, &brw->ff_gs.state_offset);
memset(gs, 0, sizeof(*gs));
/* BRW_NEW_PROGRAM_CACHE | BRW_NEW_GS_PROG_DATA */
if (brw->ff_gs.prog_active) {
gs->thread0.grf_reg_count = (ALIGN(brw->ff_gs.prog_data->total_grf, 16) /
16 - 1);
gs->thread0.kernel_start_pointer =
brw_program_reloc(brw,
brw->ff_gs.state_offset +
offsetof(struct brw_gs_unit_state, thread0),
brw->ff_gs.prog_offset +
(gs->thread0.grf_reg_count << 1)) >> 6;
gs->thread1.floating_point_mode = BRW_FLOATING_POINT_NON_IEEE_754;
gs->thread1.single_program_flow = 1;
gs->thread3.dispatch_grf_start_reg = 1;
gs->thread3.const_urb_entry_read_offset = 0;
gs->thread3.const_urb_entry_read_length = 0;
gs->thread3.urb_entry_read_offset = 0;
gs->thread3.urb_entry_read_length =
brw->ff_gs.prog_data->urb_read_length;
/* BRW_NEW_URB_FENCE */
gs->thread4.nr_urb_entries = brw->urb.nr_gs_entries;
gs->thread4.urb_entry_allocation_size = brw->urb.vsize - 1;
if (brw->urb.nr_gs_entries >= 8)
gs->thread4.max_threads = 1;
else
gs->thread4.max_threads = 0;
}
if (brw->gen == 5)
gs->thread4.rendering_enable = 1;
/* BRW_NEW_VIEWPORT_COUNT */
gs->gs6.max_vp_index = brw->clip.viewport_count - 1;
brw->ctx.NewDriverState |= BRW_NEW_GEN4_UNIT_STATE;
}
const struct brw_tracked_state brw_gs_unit = {
.dirty = {
.mesa = 0,
.brw = BRW_NEW_BATCH |
BRW_NEW_BLORP |
BRW_NEW_PUSH_CONSTANT_ALLOCATION |
BRW_NEW_FF_GS_PROG_DATA |
BRW_NEW_PROGRAM_CACHE |
BRW_NEW_URB_FENCE |
BRW_NEW_VIEWPORT_COUNT,
},
.emit = brw_upload_gs_unit,
};

View File

@@ -52,7 +52,6 @@ extern const struct brw_tracked_state brw_cs_pull_constants;
extern const struct brw_tracked_state brw_constant_buffer;
extern const struct brw_tracked_state brw_curbe_offsets;
extern const struct brw_tracked_state brw_invariant_state;
extern const struct brw_tracked_state brw_gs_unit;
extern const struct brw_tracked_state brw_binding_table_pointers;
extern const struct brw_tracked_state brw_depthbuffer;
extern const struct brw_tracked_state brw_recalculate_urb_fence;

View File

@@ -180,50 +180,6 @@ struct brw_clip_unit_state
float viewport_ymax;
};
struct brw_gs_unit_state
{
struct thread0 thread0;
struct thread1 thread1;
struct thread2 thread2;
struct thread3 thread3;
struct
{
unsigned pad0:8;
unsigned rendering_enable:1; /* for Ironlake */
unsigned pad4:1;
unsigned stats_enable:1;
unsigned nr_urb_entries:7;
unsigned pad1:1;
unsigned urb_entry_allocation_size:5;
unsigned pad2:1;
unsigned max_threads:5;
unsigned pad3:2;
} thread4;
struct
{
unsigned sampler_count:3;
unsigned pad0:2;
unsigned sampler_state_pointer:27;
} gs5;
struct
{
unsigned max_vp_index:4;
unsigned pad0:12;
unsigned svbi_post_inc_value:10;
unsigned pad1:1;
unsigned svbi_post_inc_enable:1;
unsigned svbi_payload:1;
unsigned discard_adjaceny:1;
unsigned reorder_enable:1;
unsigned pad2:1;
} gs6;
};
struct brw_wm_unit_state
{
struct thread0 thread0;

View File

@@ -2339,18 +2339,18 @@ static const struct brw_tracked_state genX(sf_clip_viewport) = {
/* ---------------------------------------------------------------------- */
#if GEN_GEN >= 6
static void
genX(upload_gs_state)(struct brw_context *brw)
{
const struct gen_device_info *devinfo = &brw->screen->devinfo;
UNUSED struct gl_context *ctx = &brw->ctx;
UNUSED const struct gen_device_info *devinfo = &brw->screen->devinfo;
const struct brw_stage_state *stage_state = &brw->gs.base;
/* BRW_NEW_GEOMETRY_PROGRAM */
bool active = brw->geometry_program;
bool active = GEN_GEN >= 6 && brw->geometry_program;
/* BRW_NEW_GS_PROG_DATA */
struct brw_stage_prog_data *stage_prog_data = stage_state->prog_data;
const struct brw_vue_prog_data *vue_prog_data =
UNUSED const struct brw_vue_prog_data *vue_prog_data =
brw_vue_prog_data(stage_prog_data);
#if GEN_GEN >= 7
const struct brw_gs_prog_data *gs_prog_data =
@@ -2384,7 +2384,14 @@ genX(upload_gs_state)(struct brw_context *brw)
gen7_emit_cs_stall_flush(brw);
#endif
#if GEN_GEN >= 6
brw_batch_emit(brw, GENX(3DSTATE_GS), gs) {
#else
ctx->NewDriverState |= BRW_NEW_GEN4_UNIT_STATE;
brw_state_emit(brw, GENX(GS_STATE), 32, &brw->ff_gs.state_offset, gs) {
#endif
#if GEN_GEN >= 6
if (active) {
INIT_THREAD_DISPATCH_FIELDS(gs, Vertex);
@@ -2435,7 +2442,6 @@ genX(upload_gs_state)(struct brw_context *brw)
#if GEN_GEN < 7
gs.SOStatisticsEnable = true;
gs.RenderingEnabled = 1;
if (brw->geometry_program->info.has_transform_feedback_varyings)
gs.SVBIPayloadEnable = true;
@@ -2469,33 +2475,41 @@ genX(upload_gs_state)(struct brw_context *brw)
gs.VertexURBEntryOutputReadOffset = urb_entry_write_offset;
gs.VertexURBEntryOutputLength = MAX2(urb_entry_output_length, 1);
#endif
#if GEN_GEN < 7
} else if (brw->ff_gs.prog_active) {
}
#endif
#if GEN_GEN <= 6
if (!active && brw->ff_gs.prog_active) {
/* In gen6, transform feedback for the VS stage is done with an
* ad-hoc GS program. This function provides the needed 3DSTATE_GS
* for this.
*/
gs.KernelStartPointer = KSP(brw, brw->ff_gs.prog_offset);
gs.SingleProgramFlow = true;
gs.VectorMaskEnable = true;
gs.DispatchGRFStartRegisterForURBData = 2;
gs.DispatchGRFStartRegisterForURBData = GEN_GEN == 6 ? 2 : 1;
gs.VertexURBEntryReadLength = brw->ff_gs.prog_data->urb_read_length;
gs.MaximumNumberofThreads = devinfo->max_gs_threads - 1;
gs.StatisticsEnable = true;
gs.SOStatisticsEnable = true;
gs.RenderingEnabled = true;
#if GEN_GEN <= 5
gs.GRFRegisterCount =
DIV_ROUND_UP(brw->ff_gs.prog_data->total_grf, 16) - 1;
/* BRW_NEW_URB_FENCE */
gs.NumberofURBEntries = brw->urb.nr_gs_entries;
gs.URBEntryAllocationSize = brw->urb.vsize - 1;
gs.MaximumNumberofThreads = brw->urb.nr_gs_entries >= 8 ? 1 : 0;
gs.FloatingPointMode = FLOATING_POINT_MODE_Alternate;
#else
gs.Enable = true;
gs.VectorMaskEnable = true;
gs.SVBIPayloadEnable = true;
gs.SVBIPostIncrementEnable = true;
gs.SVBIPostIncrementValue =
brw->ff_gs.prog_data->svbi_postincrement_value;
gs.Enable = true;
gs.SOStatisticsEnable = true;
gs.MaximumNumberofThreads = devinfo->max_gs_threads - 1;
#endif
} else {
gs.StatisticsEnable = true;
#if GEN_GEN < 7
gs.RenderingEnabled = true;
}
#endif
if (!active && !brw->ff_gs.prog_active) {
#if GEN_GEN < 8
gs.DispatchGRFStartRegisterForURBData = 1;
#if GEN_GEN >= 7
@@ -2503,6 +2517,16 @@ genX(upload_gs_state)(struct brw_context *brw)
#endif
#endif
}
#if GEN_GEN >= 6
gs.StatisticsEnable = true;
#endif
#if GEN_GEN == 5 || GEN_GEN == 6
gs.RenderingEnabled = true;
#endif
#if GEN_GEN <= 5
gs.MaximumVPIndex = brw->clip.viewport_count - 1;
#endif
}
#if GEN_GEN == 6
@@ -2512,17 +2536,22 @@ genX(upload_gs_state)(struct brw_context *brw)
static const struct brw_tracked_state genX(gs_state) = {
.dirty = {
.mesa = (GEN_GEN < 7 ? _NEW_PROGRAM_CONSTANTS : 0),
.mesa = (GEN_GEN == 6 ? _NEW_PROGRAM_CONSTANTS : 0),
.brw = BRW_NEW_BATCH |
BRW_NEW_BLORP |
BRW_NEW_CONTEXT |
BRW_NEW_GEOMETRY_PROGRAM |
BRW_NEW_GS_PROG_DATA |
(GEN_GEN <= 5 ? BRW_NEW_PUSH_CONSTANT_ALLOCATION |
BRW_NEW_PROGRAM_CACHE |
BRW_NEW_URB_FENCE |
BRW_NEW_VIEWPORT_COUNT
: 0) |
(GEN_GEN >= 6 ? BRW_NEW_CONTEXT |
BRW_NEW_GEOMETRY_PROGRAM |
BRW_NEW_GS_PROG_DATA
: 0) |
(GEN_GEN < 7 ? BRW_NEW_FF_GS_PROG_DATA : 0),
},
.emit = genX(upload_gs_state),
};
#endif
/* ---------------------------------------------------------------------- */
@@ -5037,7 +5066,7 @@ genX(init_atoms)(struct brw_context *brw)
&genX(sf_state),
&genX(vs_state), /* always required, enabled or not */
&brw_clip_unit,
&brw_gs_unit,
&genX(gs_state),
/* Command packets:
*/