lima/ppir: simplify load uni/temp op lowering and scheduling

The load uniform/temporary operations output only to a pipeline
register, which must be consumed by another op in the same instruction
later.
The current implementation delays the decision of who will consume this
result to until the scheduling step. If the consumer node is not able to
use the pipeline register, a mov node may have to be created, during the
scheduler step.

As part of the ppir scheduler simplification, and now that the ppir
scheduler supports pipeline register dependencies, this can be
simplified by always creating a single mov node outputting to a normal
register that can be used directly by all consumers.

Signed-off-by: Erico Nunes <[email protected]>
Reviewed-by: Vasily Khoruzhick <[email protected]>
Reviewed-by: Qiang Yu <[email protected]>

2 files changed, 33 insertions, 34 deletions

diff --git a/src/gallium/drivers/lima/ir/pp/lower.c b/src/gallium/drivers/lima/ir/pp/lower.c
index 97837f4c198..2244fc7dc31 100644
--- a/src/gallium/drivers/lima/ir/pp/lower.c
+++ b/src/gallium/drivers/lima/ir/pp/lower.c
@@ -87,6 +87,33 @@ static bool ppir_lower_swap_args(ppir_block *block, ppir_node *node)
    return true;
 }
 
+static bool ppir_lower_load(ppir_block *block, ppir_node *node)
+{
+   ppir_node *move = ppir_node_create(block, ppir_op_mov, -1 , 0);
+   if (unlikely(!move))
+      return false;
+
+   ppir_alu_node *alu = ppir_node_to_alu(move);
+
+   ppir_dest *dest = ppir_node_get_dest(node);
+   alu->dest = *dest;
+
+   ppir_node_replace_all_succ(move, node);
+
+   dest->type = ppir_target_pipeline;
+   dest->pipeline = ppir_pipeline_reg_uniform;
+
+   alu->num_src = 1;
+   ppir_node_target_assign(&alu->src[0], dest);
+   for (int i = 0; i < 4; i++)
+      alu->src->swizzle[i] = i;
+
+   ppir_node_add_dep(move, node);
+   list_addtail(&move->list, &node->list);
+
+   return true;
+}
+
 static bool ppir_lower_texture(ppir_block *block, ppir_node *node)
 {
    ppir_load_texture_node *load_tex = ppir_node_to_load_texture(node);
@@ -336,6 +363,8 @@ static bool (*ppir_lower_funcs[ppir_op_num])(ppir_block *, ppir_node *) = {
    [ppir_op_trunc] = ppir_lower_trunc,
    [ppir_op_sat] = ppir_lower_sat,
    [ppir_op_branch] = ppir_lower_branch,
+   [ppir_op_load_uniform] = ppir_lower_load,
+   [ppir_op_load_temp] = ppir_lower_load,
 };
 
 bool ppir_lower_prog(ppir_compiler *comp)
diff --git a/src/gallium/drivers/lima/ir/pp/node_to_instr.c b/src/gallium/drivers/lima/ir/pp/node_to_instr.c
index 711fe2153b4..f4b3114852f 100644
--- a/src/gallium/drivers/lima/ir/pp/node_to_instr.c
+++ b/src/gallium/drivers/lima/ir/pp/node_to_instr.c
@@ -121,15 +121,6 @@ static bool insert_to_each_succ_instr(ppir_block *block, ppir_node *node)
       dup->instr = instr;
       dup->instr_pos = node->instr_pos;
       ppir_node_replace_pred(dep, dup);
-
-      if ((node->op == ppir_op_load_uniform) || (node->op == ppir_op_load_temp)) {
-         ppir_load_node *load = ppir_node_to_load(node);
-         ppir_load_node *dup_load = ppir_node_to_load(dup);
-         dup_load->dest = load->dest;
-         dup_load->index = load->index;
-         dup_load->num_components = load->num_components;
-         instr->slots[node->instr_pos] = dup;
-      }
    }
 
    list_splicetail(&dup_list, &node->list);
@@ -190,31 +181,10 @@ static bool ppir_do_one_node_to_instr(ppir_block *block, ppir_node *node, ppir_n
       break;
    }
    case ppir_node_type_load:
-      if ((node->op == ppir_op_load_uniform) || (node->op == ppir_op_load_temp)) {
-         /* merge pred load_uniform into succ instr can save a reg
-          * by using pipeline reg */
-         if (!insert_to_each_succ_instr(block, node))
-            return false;
-
-         ppir_load_node *load = ppir_node_to_load(node);
-         load->dest.type = ppir_target_pipeline;
-         load->dest.pipeline = ppir_pipeline_reg_uniform;
-      }
-      else if (node->op == ppir_op_load_temp) {
-         /* merge pred load_temp into succ instr can save a reg
-          * by using pipeline reg */
-         if (!insert_to_each_succ_instr(block, node))
-            return false;
-
-         ppir_load_node *load = ppir_node_to_load(node);
-         load->dest.type = ppir_target_pipeline;
-         load->dest.pipeline = ppir_pipeline_reg_uniform;
-      }
-      else if (node->op == ppir_op_load_varying ||
-               node->op == ppir_op_load_fragcoord ||
-               node->op == ppir_op_load_pointcoord ||
-               node->op == ppir_op_load_frontface) {
-         /* delay the load varying dup to scheduler */
+      if (node->op == ppir_op_load_varying ||
+          node->op == ppir_op_load_fragcoord ||
+          node->op == ppir_op_load_pointcoord ||
+          node->op == ppir_op_load_frontface) {
          if (!create_new_instr(block, node))
             return false;
       }