From 4f41091f88cca9c87a627864ccd6962e7bb44313 Mon Sep 17 00:00:00 2001
From: Benjamin Kramer <kramerb@google.com>
Date: Tue, 21 Aug 2018 12:35:33 -0700
Subject: [XLA] Propagate invalid shape errors through reduce folding and turn
 it on

HloEvaluator should be stable enough for reduce folding, but it shouldn't
crash when it encounters an instruction without a layout. Verify the layout on
every instruction that gets evaluated and return an error on failure.

PiperOrigin-RevId: 209641401
---
 tensorflow/compiler/xla/service/BUILD | 1 +
 1 file changed, 1 insertion(+)

(limited to 'tensorflow/compiler/xla/service/BUILD')

diff --git a/tensorflow/compiler/xla/service/BUILD b/tensorflow/compiler/xla/service/BUILD
index 7fdffe85c0..73964733e8 100644
--- a/tensorflow/compiler/xla/service/BUILD
+++ b/tensorflow/compiler/xla/service/BUILD
@@ -2456,6 +2456,7 @@ tf_cc_test(
         ":hlo",
         ":hlo_constant_folding",
         ":hlo_matchers",
+        ":hlo_parser",
         ":hlo_pass",
         "//tensorflow/compiler/xla:literal",
         "//tensorflow/compiler/xla:shape_util",
-- 
cgit v1.2.3