From 2bf3cc07f09341d5cb995efe8d84010ca3228253 Mon Sep 17 00:00:00 2001
From: Ulrich Weigand <ulrich.weigand@de.ibm.com>
Date: Thu, 20 Dec 2018 13:09:09 +0000
Subject: [PATCH] [SystemZ] Fix wrong codegen caused by typos in vecintrin.h

The following two bugs in SystemZ high-level vector intrinsics are
fixes by this patch:

- The float case of vec_insert_and_zero should generate a VLLEZF
  pattern, but currently erroneously generates VLLEZLF.

- The float and double versions of vec_orc erroneously generate
  and-with-complement instead of or-with-complement.

The patch also fixes a couple of typos in the associated test.



git-svn-id: https://llvm.org/svn/llvm-project/cfe/trunk@349751 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Headers/vecintrin.h                  |  6 +++---
 test/CodeGen/builtins-systemz-zvector2.c | 10 +++++-----
 2 files changed, 8 insertions(+), 8 deletions(-)

diff --git a/lib/Headers/vecintrin.h b/lib/Headers/vecintrin.h
index f7061e8894..e627389838 100644
--- a/lib/Headers/vecintrin.h
+++ b/lib/Headers/vecintrin.h
@@ -381,7 +381,7 @@ vec_insert_and_zero(const unsigned long long *__ptr) {
 static inline __ATTRS_o_ai vector float
 vec_insert_and_zero(const float *__ptr) {
   vector float __vec = (vector float)0;
-  __vec[0] = *__ptr;
+  __vec[1] = *__ptr;
   return __vec;
 }
 #endif
@@ -5942,13 +5942,13 @@ vec_orc(vector unsigned long long __a, vector unsigned long long __b) {
 
 static inline __ATTRS_o_ai vector float
 vec_orc(vector float __a, vector float __b) {
-  return (vector float)((vector unsigned int)__a &
+  return (vector float)((vector unsigned int)__a |
                         ~(vector unsigned int)__b);
 }
 
 static inline __ATTRS_o_ai vector double
 vec_orc(vector double __a, vector double __b) {
-  return (vector double)((vector unsigned long long)__a &
+  return (vector double)((vector unsigned long long)__a |
                          ~(vector unsigned long long)__b);
 }
 #endif
diff --git a/test/CodeGen/builtins-systemz-zvector2.c b/test/CodeGen/builtins-systemz-zvector2.c
index d9607b3a87..acdaebd7a5 100644
--- a/test/CodeGen/builtins-systemz-zvector2.c
+++ b/test/CodeGen/builtins-systemz-zvector2.c
@@ -65,9 +65,9 @@ void test_core(void) {
   d = vec_extract(vd, idx);
   // CHECK: extractelement <2 x double> %{{.*}}, i32 %{{.*}}
 
-  vf = vec_insert(d, vf, idx);
+  vf = vec_insert(f, vf, idx);
   // CHECK: insertelement <4 x float> %{{.*}}, float %{{.*}}, i32 %{{.*}}
-  vd = vec_insert(f, vd, idx);
+  vd = vec_insert(d, vd, idx);
   // CHECK: insertelement <2 x double> %{{.*}}, double %{{.*}}, i32 %{{.*}}
 
   vf = vec_promote(f, idx);
@@ -76,7 +76,7 @@ void test_core(void) {
   // CHECK: insertelement <2 x double> undef, double %{{.*}}, i32 %{{.*}}
 
   vf = vec_insert_and_zero(cptrf);
-  // CHECK: insertelement <4 x float> <float undef, float 0.000000e+00, float 0.000000e+00, float 0.000000e+00>, float %{{.*}}, i32 0
+  // CHECK: insertelement <4 x float> <float 0.000000e+00, float undef, float 0.000000e+00, float 0.000000e+00>, float %{{.*}}, i32 1
   vd = vec_insert_and_zero(cptrd);
   // CHECK: insertelement <2 x double> <double undef, double 0.000000e+00>, double %{{.*}}, i32 0
 
@@ -227,8 +227,8 @@ void test_compare(void) {
   idx = vec_all_lt(vd, vd);
   // CHECK: call { <2 x i64>, i32 } @llvm.s390.vfchdbs(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 
-  idx = vec_all_nge(vd, vd);
-  // CHECK: call { <2 x i64>, i32 } @llvm.s390.vfchedbs(<2 x double> %{{.*}}, <2 x double> %{{.*}})
+  idx = vec_all_nge(vf, vf);
+  // CHECK: call { <4 x i32>, i32 } @llvm.s390.vfchesbs(<4 x float> %{{.*}}, <4 x float> %{{.*}})
   idx = vec_all_nge(vd, vd);
   // CHECK: call { <2 x i64>, i32 } @llvm.s390.vfchedbs(<2 x double> %{{.*}}, <2 x double> %{{.*}})
 
-- 
2.40.0