Mesa (master): glsl: Don' t emit ir_binop_carry during ir_binop_imul_high lowering

[Ian Romanick]

Module: Mesa
Branch: master
Commit: c01f2bfc6c7abfbbb66b9d91a81b96518285ccf1
URL:    http://cgit.freedesktop.org/mesa/mesa/commit/?id=c01f2bfc6c7abfbbb66b9d91a81b96518285ccf1

Author: Ian Romanick <ian.d.romanick at intel.com>
Date:   Mon Oct  3 15:47:29 2016 -0700

glsl: Don't emit ir_binop_carry during ir_binop_imul_high lowering

st_glsl_to_tgsi only calls lower_instructions once (instead of in a
loop), so the ir_binop_carry generated would not get lowered.  Fixes
assertion failure

state_tracker/st_glsl_to_tgsi.cpp:2265: void glsl_to_tgsi_visitor::visit_expression(ir_expression*, st_src_reg*): Assertion `!"Invalid ir opcode in glsl_to_tgsi_visitor::visit()"' failed.

on softpipe in 16 piglit tests:

    mesa_shader_integer_functions/execution/built-in-functions/fs-imulExtended-nonuniform.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/fs-imulExtended-only-msb-nonuniform.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/fs-imulExtended-only-msb.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/fs-imulExtended.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/fs-umulExtended-nonuniform.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/fs-umulExtended-only-msb-nonuniform.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/fs-umulExtended-only-msb.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/fs-umulExtended.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/vs-imulExtended-nonuniform.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/vs-imulExtended-only-msb-nonuniform.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/vs-imulExtended-only-msb.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/vs-imulExtended.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/vs-umulExtended-nonuniform.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/vs-umulExtended-only-msb-nonuniform.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/vs-umulExtended-only-msb.shader_test
    mesa_shader_integer_functions/execution/built-in-functions/vs-umulExtended.shader_test

Signed-off-by: Ian Romanick <ian.d.romanick at intel.com>
Reviewed-by: Ilia Mirkin <imirkin at alum.mit.edu>

---

 src/compiler/glsl/lower_instructions.cpp | 22 +++++++++++++++++-----
 1 file changed, 17 insertions(+), 5 deletions(-)

diff --git a/src/compiler/glsl/lower_instructions.cpp b/src/compiler/glsl/lower_instructions.cpp
index a9720f2..372ded1 100644
--- a/src/compiler/glsl/lower_instructions.cpp
+++ b/src/compiler/glsl/lower_instructions.cpp
@@ -166,6 +166,8 @@ private:
    void find_lsb_to_float_cast(ir_expression *ir);
    void find_msb_to_float_cast(ir_expression *ir);
    void imul_high_to_mul(ir_expression *ir);
+
+   ir_expression *_carry(operand a, operand b);
 };
 
 } /* anonymous namespace */
@@ -1413,6 +1415,16 @@ lower_instructions_visitor::find_msb_to_float_cast(ir_expression *ir)
    this->progress = true;
 }
 
+ir_expression *
+lower_instructions_visitor::_carry(operand a, operand b)
+{
+   if (lowering(CARRY_TO_ARITH))
+      return i2u(b2i(less(add(a, b),
+                          a.val->clone(ralloc_parent(a.val), NULL))));
+   else
+      return carry(a, b);
+}
+
 void
 lower_instructions_visitor::imul_high_to_mul(ir_expression *ir)
 {
@@ -1518,11 +1530,11 @@ lower_instructions_visitor::imul_high_to_mul(ir_expression *ir)
    i.insert_before(assign(t2, mul(src1h, src2l)));
    i.insert_before(assign(hi, mul(src1h, src2h)));
 
-   i.insert_before(assign(hi, add(hi, carry(lo, lshift(t1, c16->clone(ir, NULL))))));
-   i.insert_before(assign(lo,           add(lo, lshift(t1, c16->clone(ir, NULL)))));
+   i.insert_before(assign(hi, add(hi, _carry(lo, lshift(t1, c16->clone(ir, NULL))))));
+   i.insert_before(assign(lo,            add(lo, lshift(t1, c16->clone(ir, NULL)))));
 
-   i.insert_before(assign(hi, add(hi, carry(lo, lshift(t2, c16->clone(ir, NULL))))));
-   i.insert_before(assign(lo,           add(lo, lshift(t2, c16->clone(ir, NULL)))));
+   i.insert_before(assign(hi, add(hi, _carry(lo, lshift(t2, c16->clone(ir, NULL))))));
+   i.insert_before(assign(lo,            add(lo, lshift(t2, c16->clone(ir, NULL)))));
 
    if (different_signs == NULL) {
       assert(ir->operands[0]->type->base_type == GLSL_TYPE_UINT);
@@ -1547,7 +1559,7 @@ lower_instructions_visitor::imul_high_to_mul(ir_expression *ir)
 
       i.insert_before(neg_hi);
       i.insert_before(assign(neg_hi, add(bit_not(u2i(hi)),
-                                         u2i(carry(bit_not(lo), c1)))));
+                                         u2i(_carry(bit_not(lo), c1)))));
 
       ir->operation = ir_triop_csel;
       ir->operands[0] = new(ir) ir_dereference_variable(different_signs);