static void
set_predicate_for_overflow_query(struct brw_context *brw,
struct brw_query_object *query,
int stream_start, int count)
{
if (!can_do_mi_math_and_lrr(brw->screen)) {
brw->predicate.state = BRW_PREDICATE_STATE_STALL_FOR_QUERY;
return;
}
brw->predicate.state = BRW_PREDICATE_STATE_USE_BIT;
/* Needed to ensure the memory is coherent for the MI_LOAD_REGISTER_MEM
* command when loading the values into the predicate source registers for
* conditional rendering.
*/
brw_emit_pipe_control_flush(brw, PIPE_CONTROL_FLUSH_ENABLE);
hsw_overflow_result_to_gpr0(brw, query, count);
brw_load_register_reg64(brw, MI_PREDICATE_SRC0, HSW_CS_GPR(0));
brw_load_register_imm64(brw, MI_PREDICATE_SRC1, 0ull);
}
/**
* Compute the number of primitives written during our most recent
* transform feedback activity (the current SO_NUM_PRIMS_WRITTEN value
* minus the stashed "start" value), and add it to our running tally.
*
* If \p finalize is true, also compute the number of vertices written
* (by multiplying by the number of vertices per primitive), and store
* that to the "final" location.
*
* Otherwise, just overwrite the old tally with the new one.
*/
static void
tally_prims_written(struct brw_context *brw,
struct brw_transform_feedback_object *obj,
bool finalize)
{
/* Flush any drawing so that the counters have the right values. */
brw_emit_mi_flush(brw);
for (int i = 0; i < BRW_MAX_XFB_STREAMS; i++) {
/* GPR0 = Tally */
brw_load_register_imm32(brw, HSW_CS_GPR(0) + 4, 0);
brw_load_register_mem(brw, HSW_CS_GPR(0), obj->prim_count_bo,
I915_GEM_DOMAIN_INSTRUCTION,
I915_GEM_DOMAIN_INSTRUCTION,
TALLY_OFFSET + i * sizeof(uint32_t));
if (!obj->base.Paused) {
/* GPR1 = Start Snapshot */
brw_load_register_mem64(brw, HSW_CS_GPR(1), obj->prim_count_bo,
I915_GEM_DOMAIN_INSTRUCTION,
I915_GEM_DOMAIN_INSTRUCTION,
START_OFFSET + i * sizeof(uint64_t));
/* GPR2 = Ending Snapshot */
brw_load_register_reg64(brw, GEN7_SO_NUM_PRIMS_WRITTEN(i), HSW_CS_GPR(2));
BEGIN_BATCH(9);
OUT_BATCH(HSW_MI_MATH | (9 - 2));
/* GPR1 = GPR2 (End) - GPR1 (Start) */
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCA, R2));
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCB, R1));
OUT_BATCH(MI_MATH_ALU0(SUB));
OUT_BATCH(MI_MATH_ALU2(STORE, R1, ACCU));
/* GPR0 = GPR0 (Tally) + GPR1 (Diff) */
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCA, R0));
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCB, R1));
OUT_BATCH(MI_MATH_ALU0(ADD));
OUT_BATCH(MI_MATH_ALU2(STORE, R0, ACCU));
ADVANCE_BATCH();
}
if (!finalize) {
/* Write back the new tally */
brw_store_register_mem32(brw, obj->prim_count_bo, HSW_CS_GPR(0),
TALLY_OFFSET + i * sizeof(uint32_t));
} else {
/* Convert the number of primitives to the number of vertices. */
if (obj->primitive_mode == GL_LINES) {
/* Double R0 (R0 = R0 + R0) */
BEGIN_BATCH(5);
OUT_BATCH(HSW_MI_MATH | (5 - 2));
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCA, R0));
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCB, R0));
OUT_BATCH(MI_MATH_ALU0(ADD));
OUT_BATCH(MI_MATH_ALU2(STORE, R0, ACCU));
ADVANCE_BATCH();
} else if (obj->primitive_mode == GL_TRIANGLES) {
/* Triple R0 (R1 = R0 + R0, R0 = R0 + R1) */
BEGIN_BATCH(9);
OUT_BATCH(HSW_MI_MATH | (9 - 2));
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCA, R0));
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCB, R0));
OUT_BATCH(MI_MATH_ALU0(ADD));
OUT_BATCH(MI_MATH_ALU2(STORE, R1, ACCU));
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCA, R0));
OUT_BATCH(MI_MATH_ALU2(LOAD, SRCB, R1));
OUT_BATCH(MI_MATH_ALU0(ADD));
OUT_BATCH(MI_MATH_ALU2(STORE, R0, ACCU));
ADVANCE_BATCH();
}
/* Store it to the final result */
brw_store_register_mem32(brw, obj->prim_count_bo, HSW_CS_GPR(0),
i * sizeof(uint32_t));
}
}
}
