aco: execute branch instructions in WQM if necessary

It could happen that only the branch condition was computed in WQM
and not the branch instruction.
There is now some rendundancy which should be cleaned up.

Fixes: 3817fa7a4d ('aco: fix WQM handling in nested loops')
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/6260>

src/amd/compiler/aco_insert_exec_mask.cpp

@@ -165,12 +165,6 @@ void mark_block_wqm(wqm_ctx &ctx, unsigned block_idx)
 
    ctx.branch_wqm[block_idx] = true;
    Block& block = ctx.program->blocks[block_idx];
-   aco_ptr<Instruction>& branch = block.instructions.back();
-
-   if (branch->opcode != aco_opcode::p_branch) {
-      assert(!branch->operands.empty() && branch->operands[0].isTemp());
-      set_needs_wqm(ctx, branch->operands[0].getTemp());
-   }
 
    /* TODO: this sets more branch conditions to WQM than it needs to
     * it should be enough to stop at the "exec mask top level" */
@@ -233,6 +227,11 @@ void get_block_needs(wqm_ctx &ctx, exec_ctx &exec_ctx, Block* block)
          }
       }
 
+      if (instr->format == Format::PSEUDO_BRANCH && ctx.branch_wqm[block->index]) {
+         needs = WQM;
+         propagate_wqm = true;
+      }
+
       if (propagate_wqm) {
          for (const Operand& op : instr->operands) {
             if (op.isTemp()) {