3 files changed, 30 insertions, 0 deletions
diff --git a/test/Assembler/AutoUpgradeIntrinsics.ll b/test/Assembler/AutoUpgradeIntrinsics.ll
index e4e2d3a..417493f 100644
--- a/test/Assembler/AutoUpgradeIntrinsics.ll
+++ b/test/Assembler/AutoUpgradeIntrinsics.ll
@@ -10,6 +10,7 @@
 ; RUN:   not grep {llvm\\.x86\\.sse2\\.loadu}
 ; RUN: llvm-as < %s | llvm-dis | \
 ; RUN:   grep {llvm\\.x86\\.mmx\\.ps} | grep {x86_mmx} | count 16
+; RUN: llvm-as < %s | llvm-dis | FileCheck %s
 
 declare i32 @llvm.ctpop.i28(i28 %val)
 declare i32 @llvm.cttz.i29(i29 %val)
@@ -91,3 +92,20 @@ define void @test_loadu(i8* %a, double* %b) {
   %v2 = call <2 x double> @llvm.x86.sse2.loadu.pd(double* %b)
   ret void
 }
+
+declare void @llvm.x86.sse.movnt.ps(i8*, <4 x float>) nounwind readnone 
+declare void @llvm.x86.sse2.movnt.dq(i8*, <2 x double>) nounwind readnone 
+declare void @llvm.x86.sse2.movnt.pd(i8*, <2 x double>) nounwind readnone 
+declare void @llvm.x86.sse2.movnt.i(i8*, i32) nounwind readnone 
+
+define void @f(<4 x float> %A, i8* %B, <2 x double> %C, i32 %D) {
+; CHECK: store{{.*}}nontemporal
+  call void @llvm.x86.sse.movnt.ps(i8* %B, <4 x float> %A)
+; CHECK: store{{.*}}nontemporal
+  call void @llvm.x86.sse2.movnt.dq(i8* %B, <2 x double> %C)
+; CHECK: store{{.*}}nontemporal
+  call void @llvm.x86.sse2.movnt.pd(i8* %B, <2 x double> %C)
+; CHECK: store{{.*}}nontemporal
+  call void @llvm.x86.sse2.movnt.i(i8* %B, i32 %D)
+  ret void
+}
diff --git a/test/Assembler/invalid_cast.ll b/test/Assembler/invalid_cast.ll
new file mode 100644
index 0000000..c5b082b
--- /dev/null
+++ b/test/Assembler/invalid_cast.ll
@@ -0,0 +1,6 @@
+; RUN: not llvm-as < %s |& grep {invalid cast opcode}
+
+define <3 x i8> @foo(<4 x i64> %x) {
+  %y = trunc <4 x i64> %x to <3 x i8>
+  ret <3 x i8> %y
+}
diff --git a/test/Assembler/invalid_cast2.ll b/test/Assembler/invalid_cast2.ll
new file mode 100644
index 0000000..f2e7c41
--- /dev/null
+++ b/test/Assembler/invalid_cast2.ll
@@ -0,0 +1,6 @@
+; RUN: not llvm-as < %s |& grep {invalid cast opcode}
+
+define i8 @foo(<4 x i64> %x) {
+  %y = trunc <4 x i64> %x to i8
+  ret i8 %y
+}