libbcc

Change-Id: Ieaa3ebd5a38f370752495549f8870b534eeedfc5

1 files changed, 34 insertions, 3 deletions

diff --git a/test/CodeGen/X86/vec_ss_load_fold.ll b/test/CodeGen/X86/vec_ss_load_fold.ll
index b1613fb..c8b2927 100644
--- a/test/CodeGen/X86/vec_ss_load_fold.ll
+++ b/test/CodeGen/X86/vec_ss_load_fold.ll
@@ -1,6 +1,4 @@
-; RUN: llc < %s -march=x86 -mattr=+sse,+sse2 -o %t
-; RUN: grep minss %t | grep CPI | count 2
-; RUN: grep CPI   %t | not grep movss
+; RUN: llc < %s -march=x86 -mattr=+sse,+sse2,+sse41 | FileCheck %s
 
 target datalayout = "e-p:32:32"
 target triple = "i686-apple-darwin8.7.2"
@@ -17,6 +15,10 @@ define i16 @test1(float %f) nounwind {
 	%tmp.upgrd.1 = tail call i32 @llvm.x86.sse.cvttss2si( <4 x float> %tmp59 )		; <i32> [#uses=1]
 	%tmp69 = trunc i32 %tmp.upgrd.1 to i16		; <i16> [#uses=1]
 	ret i16 %tmp69
+; CHECK: test1:
+; CHECK: subss	LCPI1_
+; CHECK: mulss	LCPI1_
+; CHECK: minss	LCPI1_
 }
 
 define i16 @test2(float %f) nounwind {
@@ -28,6 +30,10 @@ define i16 @test2(float %f) nounwind {
 	%tmp = tail call i32 @llvm.x86.sse.cvttss2si( <4 x float> %tmp59 )		; <i32> [#uses=1]
 	%tmp69 = trunc i32 %tmp to i16		; <i16> [#uses=1]
 	ret i16 %tmp69
+; CHECK: test2:
+; CHECK: addss	LCPI2_
+; CHECK: mulss	LCPI2_
+; CHECK: minss	LCPI2_
 }
 
 declare <4 x float> @llvm.x86.sse.sub.ss(<4 x float>, <4 x float>)
@@ -39,3 +45,28 @@ declare <4 x float> @llvm.x86.sse.min.ss(<4 x float>, <4 x float>)
 declare <4 x float> @llvm.x86.sse.max.ss(<4 x float>, <4 x float>)
 
 declare i32 @llvm.x86.sse.cvttss2si(<4 x float>)
+
+
+declare <4 x float> @llvm.x86.sse41.round.ss(<4 x float>, <4 x float>, i32)
+declare <4 x float> @f()
+
+define <4 x float> @test3(<4 x float> %A, float *%b, i32 %C) nounwind {
+  %a = load float *%b
+  %B = insertelement <4 x float> undef, float %a, i32 0
+  %X = call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %A, <4 x float> %B, i32 4)
+  ret <4 x float> %X
+; CHECK: test3:
+; CHECK: roundss	$4, (%eax), %xmm0
+}
+
+define <4 x float> @test4(<4 x float> %A, float *%b, i32 %C) nounwind {
+  %a = load float *%b
+  %B = insertelement <4 x float> undef, float %a, i32 0
+  %q = call <4 x float> @f()
+  %X = call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %q, <4 x float> %B, i32 4)
+  ret <4 x float> %X
+; CHECK: test4:
+; CHECK: movss	(%eax), %xmm
+; CHECK: call
+; CHECK: roundss $4, %xmm{{.*}}, %xmm0
+}