/* prepend ALU_EXTENDED if we need more than 2 kcache sets */
if (cf->eg_alu_extended) {
bc->bytecode[id++] =
- S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK_INDEX_MODE0(V_SQ_CF_INDEX_NONE) |
- S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK_INDEX_MODE1(V_SQ_CF_INDEX_NONE) |
- S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK_INDEX_MODE2(V_SQ_CF_INDEX_NONE) |
- S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK_INDEX_MODE3(V_SQ_CF_INDEX_NONE) |
+ S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK_INDEX_MODE0(cf->kcache[0].index_mode) |
+ S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK_INDEX_MODE1(cf->kcache[1].index_mode) |
+ S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK_INDEX_MODE2(cf->kcache[2].index_mode) |
+ S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK_INDEX_MODE3(cf->kcache[3].index_mode) |
S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK2(cf->kcache[2].bank) |
S_SQ_CF_ALU_WORD0_EXT_KCACHE_BANK3(cf->kcache[3].bank) |
S_SQ_CF_ALU_WORD0_EXT_KCACHE_MODE2(cf->kcache[2].mode);
bc->bytecode[id++] = S_SQ_CF_ALLOC_EXPORT_WORD0_RW_GPR(cf->output.gpr) |
S_SQ_CF_ALLOC_EXPORT_WORD0_ELEM_SIZE(cf->output.elem_size) |
S_SQ_CF_ALLOC_EXPORT_WORD0_ARRAY_BASE(cf->output.array_base) |
- S_SQ_CF_ALLOC_EXPORT_WORD0_TYPE(cf->output.type);
+ S_SQ_CF_ALLOC_EXPORT_WORD0_TYPE(cf->output.type) |
+ S_SQ_CF_ALLOC_EXPORT_WORD0_INDEX_GPR(cf->output.index_gpr);
bc->bytecode[id] =
S_SQ_CF_ALLOC_EXPORT_WORD1_BURST_COUNT(cf->output.burst_count - 1) |
S_SQ_CF_ALLOC_EXPORT_WORD1_SWIZ_SEL_X(cf->output.swizzle_x) |
S_SQ_CF_ALLOC_EXPORT_WORD1_SWIZ_SEL_Y(cf->output.swizzle_y) |
S_SQ_CF_ALLOC_EXPORT_WORD1_SWIZ_SEL_Z(cf->output.swizzle_z) |
S_SQ_CF_ALLOC_EXPORT_WORD1_SWIZ_SEL_W(cf->output.swizzle_w) |
- S_SQ_CF_ALLOC_EXPORT_WORD1_BARRIER(cf->output.barrier) |
+ S_SQ_CF_ALLOC_EXPORT_WORD1_BARRIER(cf->barrier) |
S_SQ_CF_ALLOC_EXPORT_WORD1_CF_INST(opcode);
if (bc->chip_class == EVERGREEN) /* no EOP on cayman */
- bc->bytecode[id] |= S_SQ_CF_ALLOC_EXPORT_WORD1_END_OF_PROGRAM(cf->output.end_of_program);
+ bc->bytecode[id] |= S_SQ_CF_ALLOC_EXPORT_WORD1_END_OF_PROGRAM(cf->end_of_program);
id++;
- } else if (cfop->flags & CF_STRM) {
- /* MEM_STREAM instructions */
+ } else if (cfop->flags & CF_MEM) {
+ /* MEM_STREAM, MEM_RING instructions */
bc->bytecode[id++] = S_SQ_CF_ALLOC_EXPORT_WORD0_RW_GPR(cf->output.gpr) |
S_SQ_CF_ALLOC_EXPORT_WORD0_ELEM_SIZE(cf->output.elem_size) |
S_SQ_CF_ALLOC_EXPORT_WORD0_ARRAY_BASE(cf->output.array_base) |
- S_SQ_CF_ALLOC_EXPORT_WORD0_TYPE(cf->output.type);
+ S_SQ_CF_ALLOC_EXPORT_WORD0_TYPE(cf->output.type) |
+ S_SQ_CF_ALLOC_EXPORT_WORD0_INDEX_GPR(cf->output.index_gpr);
bc->bytecode[id] = S_SQ_CF_ALLOC_EXPORT_WORD1_BURST_COUNT(cf->output.burst_count - 1) |
- S_SQ_CF_ALLOC_EXPORT_WORD1_BARRIER(cf->output.barrier) |
+ S_SQ_CF_ALLOC_EXPORT_WORD1_BARRIER(cf->barrier) |
S_SQ_CF_ALLOC_EXPORT_WORD1_CF_INST(opcode) |
S_SQ_CF_ALLOC_EXPORT_WORD1_BUF_COMP_MASK(cf->output.comp_mask) |
S_SQ_CF_ALLOC_EXPORT_WORD1_BUF_ARRAY_SIZE(cf->output.array_size);
if (bc->chip_class == EVERGREEN) /* no EOP on cayman */
- bc->bytecode[id] |= S_SQ_CF_ALLOC_EXPORT_WORD1_END_OF_PROGRAM(cf->output.end_of_program);
+ bc->bytecode[id] |= S_SQ_CF_ALLOC_EXPORT_WORD1_END_OF_PROGRAM(cf->end_of_program);
id++;
} else {
- /* branch, loop, call, return instructions */
+ /* other instructions */
bc->bytecode[id++] = S_SQ_CF_WORD0_ADDR(cf->cf_addr >> 1);
bc->bytecode[id++] = S_SQ_CF_WORD1_CF_INST(opcode)|
S_SQ_CF_WORD1_BARRIER(1) |
S_SQ_CF_WORD1_COND(cf->cond) |
- S_SQ_CF_WORD1_POP_COUNT(cf->pop_count);
+ S_SQ_CF_WORD1_POP_COUNT(cf->pop_count) |
+ S_SQ_CF_WORD1_END_OF_PROGRAM(cf->end_of_program);
}
}
return 0;
}
+#if 0
void eg_bytecode_export_read(struct r600_bytecode *bc,
struct r600_bytecode_output *output, uint32_t word0, uint32_t word1)
{
output->array_size = G_SQ_CF_ALLOC_EXPORT_WORD1_BUF_ARRAY_SIZE(word1);
output->comp_mask = G_SQ_CF_ALLOC_EXPORT_WORD1_BUF_COMP_MASK(word1);
}
+#endif
+
+int egcm_load_index_reg(struct r600_bytecode *bc, unsigned id, bool inside_alu_clause)
+{
+ struct r600_bytecode_alu alu;
+ int r;
+ unsigned type;
+
+ assert(id < 2);
+ assert(bc->chip_class >= EVERGREEN);
+
+ if (bc->index_loaded[id])
+ return 0;
+
+ memset(&alu, 0, sizeof(alu));
+ alu.op = ALU_OP1_MOVA_INT;
+ alu.src[0].sel = bc->index_reg[id];
+ alu.src[0].chan = 0;
+ alu.last = 1;
+ r = r600_bytecode_add_alu(bc, &alu);
+ if (r)
+ return r;
+
+ bc->ar_loaded = 0; /* clobbered */
+
+ memset(&alu, 0, sizeof(alu));
+ alu.op = id == 0 ? ALU_OP0_SET_CF_IDX0 : ALU_OP0_SET_CF_IDX1;
+ alu.last = 1;
+ r = r600_bytecode_add_alu(bc, &alu);
+ if (r)
+ return r;
+
+ /* Must split ALU group as index only applies to following group */
+ if (inside_alu_clause) {
+ type = bc->cf_last->op;
+ if ((r = r600_bytecode_add_cf(bc))) {
+ return r;
+ }
+ bc->cf_last->op = type;
+ }
+
+ bc->index_loaded[id] = 1;
+
+ return 0;
+}