glsl/loops: consolidate bounded loop handling into a lowering pass.

Previously, all of the back-ends (ir_to_mesa, st_glsl_to_tgsi, and the
i965 fs and vec4 visitors) had nearly identical logic for handling
bounded loops.  This replaces the duplicate logic with an equivalent
lowering pass that is used by all the back-ends.

Note: on i965, there is a slight increase in instruction count.  For
example, a loop like this:

    for (int i = 0; i < 100; i++) {
      total += i;
    }

would previously compile down to this (vec4) native code:

          mov(8)       g4<1>.xD 0D
          mov(8)       g8<1>.xD 0D
    loop:
          cmp.ge.f0(8) null     g8<4;4,1>.xD 100D
    (+f0) break(8)
          add(8)       g5<1>.xD g5<4;4,1>.xD g4<4;4,1>.xD
          add(8)       g8<1>.xD g8<4;4,1>.xD 1D
          add(8)       g4<1>.xD g4<4;4,1>.xD 1D
          while(8) loop

After this patch, the "(+f0) break(8)" turns into:

    (+f0) if(8)
          break(8)
          endif(8)

because the back-end isn't smart enough to recognize that "if
(condition) break;" can be done using a conditional break instruction.
However, it should be relatively easy for a future peephole
optimization to properly optimize this.

Reviewed-by: Jordan Justen <[email protected]>
Reviewed-by: Ian Romanick <[email protected]>

3 files changed, 6 insertions, 63 deletions

diff --git a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
index a28dc6cdffe..3e32f6e0eba 100644
--- a/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_fs_visitor.cpp
@@ -2181,39 +2181,16 @@ fs_visitor::visit(ir_if *ir)
 void
 fs_visitor::visit(ir_loop *ir)
 {
-   fs_reg counter = reg_undef;
+   /* Any bounded loops should have been lowered by lower_bounded_loops(). */
+   assert(ir->counter == NULL);
 
    if (brw->gen < 6 && dispatch_width == 16) {
       fail("Can't support (non-uniform) control flow on 16-wide\n");
    }
 
-   if (ir->counter) {
-      this->base_ir = ir->counter;
-      ir->counter->accept(this);
-      counter = *(variable_storage(ir->counter));
-
-      if (ir->from) {
-	 this->base_ir = ir->from;
-	 ir->from->accept(this);
-
-	 emit(MOV(counter, this->result));
-      }
-   }
-
    this->base_ir = NULL;
    emit(BRW_OPCODE_DO);
 
-   if (ir->to) {
-      this->base_ir = ir->to;
-      ir->to->accept(this);
-
-      emit(CMP(reg_null_d, counter, this->result,
-               brw_conditional_for_comparison(ir->cmp)));
-
-      fs_inst *inst = emit(BRW_OPCODE_BREAK);
-      inst->predicate = BRW_PREDICATE_NORMAL;
-   }
-
    foreach_list(node, &ir->body_instructions) {
       ir_instruction *ir = (ir_instruction *)node;
 
@@ -2221,12 +2198,6 @@ fs_visitor::visit(ir_loop *ir)
       ir->accept(this);
    }
 
-   if (ir->increment) {
-      this->base_ir = ir->increment;
-      ir->increment->accept(this);
-      emit(ADD(counter, counter, this->result));
-   }
-
    this->base_ir = NULL;
    emit(BRW_OPCODE_WHILE);
 }
diff --git a/src/mesa/drivers/dri/i965/brw_shader.cpp b/src/mesa/drivers/dri/i965/brw_shader.cpp
index 88aa169fc1f..12035c26793 100644
--- a/src/mesa/drivers/dri/i965/brw_shader.cpp
+++ b/src/mesa/drivers/dri/i965/brw_shader.cpp
@@ -213,6 +213,8 @@ brw_link_shader(struct gl_context *ctx, struct gl_shader_program *shProg)
 	   || progress;
       } while (progress);
 
+      lower_bounded_loops(shader->ir);
+
       /* Make a pass over the IR to add state references for any built-in
        * uniforms that are used.  This has to be done now (during linking).
        * Code generation doesn't happen until the first time this shader is
diff --git a/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp b/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
index b599c29d475..cf62e7ed0b0 100644
--- a/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
+++ b/src/mesa/drivers/dri/i965/brw_vec4_visitor.cpp
@@ -1007,48 +1007,18 @@ vec4_visitor::visit(ir_variable *ir)
 void
 vec4_visitor::visit(ir_loop *ir)
 {
-   dst_reg counter;
+   /* Any bounded loops should have been lowered by lower_bounded_loops(). */
+   assert(ir->counter == NULL);
 
    /* We don't want debugging output to print the whole body of the
     * loop as the annotation.
     */
    this->base_ir = NULL;
 
-   if (ir->counter != NULL) {
-      this->base_ir = ir->counter;
-      ir->counter->accept(this);
-      counter = *(variable_storage(ir->counter));
-
-      if (ir->from != NULL) {
-	 this->base_ir = ir->from;
-	 ir->from->accept(this);
-
-	 emit(MOV(counter, this->result));
-      }
-   }
-
    emit(BRW_OPCODE_DO);
 
-   if (ir->to) {
-      this->base_ir = ir->to;
-      ir->to->accept(this);
-
-      emit(CMP(dst_null_d(), src_reg(counter), this->result,
-	       brw_conditional_for_comparison(ir->cmp)));
-
-      vec4_instruction *inst = emit(BRW_OPCODE_BREAK);
-      inst->predicate = BRW_PREDICATE_NORMAL;
-   }
-
    visit_instructions(&ir->body_instructions);
 
-
-   if (ir->increment) {
-      this->base_ir = ir->increment;
-      ir->increment->accept(this);
-      emit(ADD(counter, src_reg(counter), this->result));
-   }
-
    emit(BRW_OPCODE_WHILE);
 }