test: LLVM supports now vectors of arbitrary pointers

This allows Polly to vectorize more code. Fix the relevant test cases.

llvm-svn: 167923

3 files changed, 51 insertions, 59 deletions

diff --git a/polly/test/CodeGen/simple_vec_call_2.ll b/polly/test/CodeGen/simple_vec_call_2.ll
index cf26641a1e4..bb2a6806196 100644
--- a/polly/test/CodeGen/simple_vec_call_2.ll
+++ b/polly/test/CodeGen/simple_vec_call_2.ll
@@ -27,9 +27,6 @@ return:
 }
 
 ; CHECK: %p_scevgep = getelementptr [1024 x float**]* @B, i64 0, i64 0
-; CHECK: %p_scevgep1 = getelementptr [1024 x float**]* @B, i64 0, i64 1
-; CHECK: %p_scevgep2 = getelementptr [1024 x float**]* @B, i64 0, i64 2
-; CHECK: %p_scevgep3 = getelementptr [1024 x float**]* @B, i64 0, i64 3
 ; CHECK: %value_p_splat_one = load <1 x float>* bitcast ([1024 x float]* @A to <1 x float>*), align 8
 ; CHECK: %value_p_splat = shufflevector <1 x float> %value_p_splat_one, <1 x float> %value_p_splat_one, <4 x i32> zeroinitializer
 ; CHECK: %0 = extractelement <4 x float> %value_p_splat, i32 0
@@ -40,10 +37,12 @@ return:
 ; CHECK: [[RES2:%[a-zA-Z0-9_]+]] = tail call float** @foo(float %1) nounwind
 ; CHECK: [[RES3:%[a-zA-Z0-9_]+]] = tail call float** @foo(float %2) nounwind
 ; CHECK: [[RES4:%[a-zA-Z0-9_]+]] = tail call float** @foo(float %3) nounwind
-; CHECK: store float** [[RES1]], float*** %p_scevgep, align 4
-; CHECK: store float** [[RES2]], float*** %p_scevgep1, align 4
-; CHECK: store float** [[RES3]], float*** %p_scevgep2, align 4
-; CHECK: store float** [[RES4]], float*** %p_scevgep3, align 4
+; CHECK: %4 = insertelement <4 x float**> undef, float** %p_result, i32 0
+; CHECK: %5 = insertelement <4 x float**> %4, float** %p_result4, i32 1
+; CHECK: %6 = insertelement <4 x float**> %5, float** %p_result5, i32 2
+; CHECK: %7 = insertelement <4 x float**> %6, float** %p_result6, i32 3
+; CHECK: %vector_ptr = bitcast float*** %p_scevgep to <4 x float**>*
+; CHECK: store <4 x float**> %7, <4 x float**>* %vector_ptr, align 8
 
 ; CHECK-SCEV: %value_p_splat_one = load <1 x float>* bitcast ([1024 x float]* @A to <1 x float>*), align 8
 ; CHECK-SCEV: %value_p_splat = shufflevector <1 x float> %value_p_splat_one, <1 x float> %value_p_splat_one, <4 x i32> zeroinitializer
@@ -55,7 +54,8 @@ return:
 ; CHECK-SCEV: [[RES2:%[a-zA-Z0-9_]+]] = tail call float** @foo(float %1) nounwind
 ; CHECK-SCEV: [[RES3:%[a-zA-Z0-9_]+]] = tail call float** @foo(float %2) nounwind
 ; CHECK-SCEV: [[RES4:%[a-zA-Z0-9_]+]] = tail call float** @foo(float %3) nounwind
-; CHECK-SCEV: store float** [[RES1]], float*** getelementptr inbounds ([1024 x float**]* @B, i64 0, i64 0), align 4
-; CHECK-SCEV: store float** [[RES2]], float*** getelementptr inbounds ([1024 x float**]* @B, i64 0, i64 1), align 4
-; CHECK-SCEV: store float** [[RES3]], float*** getelementptr inbounds ([1024 x float**]* @B, i64 0, i64 2), align 4
-; CHECK-SCEV: store float** [[RES4]], float*** getelementptr inbounds ([1024 x float**]* @B, i64 0, i64 3), align 4
+; CHECK-SCEV: %4 = insertelement <4 x float**> undef, float** %p_result, i32 0
+; CHECK-SCEV: %5 = insertelement <4 x float**> %4, float** %p_result1, i32 1
+; CHECK-SCEV: %6 = insertelement <4 x float**> %5, float** %p_result2, i32 2
+; CHECK-SCEV: %7 = insertelement <4 x float**> %6, float** %p_result3, i32 3
+; CHECK-SCEV: store <4 x float**> %7, <4 x float**>* bitcast ([1024 x float**]* @B to <4 x float**>*), align 
diff --git a/polly/test/CodeGen/simple_vec_impossible.ll b/polly/test/CodeGen/simple_vec_impossible.ll
deleted file mode 100644
index b63ebf32c65..00000000000
--- a/polly/test/CodeGen/simple_vec_impossible.ll
+++ /dev/null
@@ -1,48 +0,0 @@
-; RUN: opt %loadPolly -basicaa -polly-codegen %vector-opt -S -polly-codegen-scev=false < %s | FileCheck %s
-; RUN: opt %loadPolly -basicaa -polly-codegen %vector-opt -S -polly-codegen-scev=true < %s | FileCheck %s -check-prefix=CHECK-SCEV
-target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
-target triple = "x86_64-unknown-linux-gnu"
-
-@A = common global [1024 x float**] zeroinitializer, align 16
-@B = common global [1024 x float**] zeroinitializer, align 16
-
-declare float @foo(float) readnone
-
-define void @simple_vec_call() nounwind {
-entry:
-  br label %body
-
-body:
-  %indvar = phi i64 [ 0, %entry ], [ %indvar_next, %body ]
-  %scevgep = getelementptr [1024 x float**]* @B, i64 0, i64 %indvar
-  %value = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0), align 16
-  store float** %value, float*** %scevgep, align 4
-  %indvar_next = add i64 %indvar, 1
-  %exitcond = icmp eq i64 %indvar_next, 4
-  br i1 %exitcond, label %return, label %body
-
-return:
-  ret void
-}
-
-; CHECK: %p_scevgep = getelementptr [1024 x float**]* @B, i64 0, i64 0
-; CHECK: %p_scevgep1 = getelementptr [1024 x float**]* @B, i64 0, i64 1
-; CHECK: %p_scevgep2 = getelementptr [1024 x float**]* @B, i64 0, i64 2
-; CHECK: %p_scevgep3 = getelementptr [1024 x float**]* @B, i64 0, i64 3
-; CHECK: %value_p_scalar_ = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0)
-; CHECK: %value_p_scalar_4 = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0)
-; CHECK: %value_p_scalar_5 = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0)
-; CHECK: %value_p_scalar_6 = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0)
-; CHECK: store float** %value_p_scalar_, float*** %p_scevgep, align 4
-; CHECK: store float** %value_p_scalar_4, float*** %p_scevgep1, align 4
-; CHECK: store float** %value_p_scalar_5, float*** %p_scevgep2, align 4
-; CHECK: store float** %value_p_scalar_6, float*** %p_scevgep3, align 4
-
-; CHECK-SCEV: %value_p_scalar_ = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0)
-; CHECK-SCEV: %value_p_scalar_1 = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0)
-; CHECK-SCEV: %value_p_scalar_2 = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0)
-; CHECK-SCEV: %value_p_scalar_3 = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0)
-; CHECK-SCEV: store float** %value_p_scalar_, float*** getelementptr inbounds ([1024 x float**]* @B, i64 0, i64 0), align 4
-; CHECK-SCEV: store float** %value_p_scalar_1, float*** getelementptr inbounds ([1024 x float**]* @B, i64 0, i64 1), align 4
-; CHECK-SCEV: store float** %value_p_scalar_2, float*** getelementptr inbounds ([1024 x float**]* @B, i64 0, i64 2), align 4
-; CHECK-SCEV: store float** %value_p_scalar_3, float*** getelementptr inbounds ([1024 x float**]* @B, i64 0, i64 3), align 4
diff --git a/polly/test/CodeGen/simple_vec_ptr_ptr_ty.ll b/polly/test/CodeGen/simple_vec_ptr_ptr_ty.ll
new file mode 100644
index 00000000000..d295be7aaaf
--- /dev/null
+++ b/polly/test/CodeGen/simple_vec_ptr_ptr_ty.ll
@@ -0,0 +1,40 @@
+; RUN: opt %loadPolly -basicaa -polly-codegen %vector-opt -S -polly-codegen-scev=false < %s | FileCheck %s
+; RUN: opt %loadPolly -basicaa -polly-codegen %vector-opt -S -polly-codegen-scev=true < %s | FileCheck %s -check-prefix=CHECK-SCEV
+target datalayout = "e-p:64:64:64-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:64:64-f32:32:32-f64:64:64-v64:64:64-v128:128:128-a0:0:64-s0:64:64-f80:128:128-n8:16:32:64"
+target triple = "x86_64-unknown-linux-gnu"
+
+@A = common global [1024 x float**] zeroinitializer, align 16
+@B = common global [1024 x float**] zeroinitializer, align 16
+
+declare float @foo(float) readnone
+
+define void @simple_vec_call() nounwind {
+entry:
+  br label %body
+
+body:
+  %indvar = phi i64 [ 0, %entry ], [ %indvar_next, %body ]
+  %scevgep = getelementptr [1024 x float**]* @B, i64 0, i64 %indvar
+  %value = load float*** getelementptr inbounds ([1024 x float**]* @A, i64 0, i64 0), align 16
+  store float** %value, float*** %scevgep, align 4
+  %indvar_next = add i64 %indvar, 1
+  %exitcond = icmp eq i64 %indvar_next, 4
+  br i1 %exitcond, label %return, label %body
+
+return:
+  ret void
+}
+
+; CHECK: %p_scevgep = getelementptr [1024 x float**]* @B, i64 0, i64 0
+; CHECK: %p_scevgep1 = getelementptr [1024 x float**]* @B, i64 0, i64 1
+; CHECK: %p_scevgep2 = getelementptr [1024 x float**]* @B, i64 0, i64 2
+; CHECK: %p_scevgep3 = getelementptr [1024 x float**]* @B, i64 0, i64 3
+; CHECK: %value_p_splat_one = load <1 x float**>* bitcast ([1024 x float**]* @A to <1 x float**>*), align 8
+; CHECK: %value_p_splat = shufflevector <1 x float**> %value_p_splat_one, <1 x float**> %value_p_splat_one, <4 x i32> zeroinitializer
+; CHECK: %vector_ptr = bitcast float*** %p_scevgep to <4 x float**>*
+; CHECK: store <4 x float**> %value_p_splat, <4 x float**>* %vector_ptr
+
+
+; CHECK-SCEV: %value_p_splat_one = load <1 x float**>* bitcast ([1024 x float**]* @A to <1 x float**>*), align 8
+; CHECK-SCEV: %value_p_splat = shufflevector <1 x float**> %value_p_splat_one, <1 x float**> %value_p_splat_one, <4 x i32> zeroinitializer
+; CHECK-SCEV: store <4 x float**> %value_p_splat, <4 x float**>* bitcast ([1024 x float**]* @B to <4 x float**>*), align 8