[MBP] Fix a really horrible bug in MachineBlockPlacement, but behind

[oota-llvm.git] / test / CodeGen / X86 / widen_arith-5.ll

diff --git a/test/CodeGen/X86/widen_arith-5.ll b/test/CodeGen/X86/widen_arith-5.ll
index 68d87a6c5051aefcf6ea5feecc5b227df743f33a..04c9ec2def2af21967a55eb77ec15799066e7b76 100644 (file)
--- a/test/CodeGen/X86/widen_arith-5.ll
+++ b/test/CodeGen/X86/widen_arith-5.ll
@@ -1,10 +1,9 @@
-; RUN: llvm-as < %s | llc -march=x86 -mattr=+sse42 -disable-mmx -o %t
-; RUN: grep pmulld  %t | count 1
-; RUN: grep psubd  %t | count 1
-; RUN: grep movaps %t | count 1
+; RUN: llc < %s -march=x86-64 -mattr=+sse4.2  | FileCheck %s
+; CHECK: movdqa
+; CHECK: pslld $2
+; CHECK: psubd
 
 ; widen a v3i32 to v4i32 to do a vector multiple and a subtraction
-target datalayout = "e-p:32:32:32-i1:8:8-i8:8:8-i16:16:16-i32:32:32-i64:32:64-f32:32:32-f64:32:64-v64:64:64-v128:128:128-a0:0:64-f80:128:128"
 
 define void @update(<3 x i32>* %dst, <3 x i32>* %src, i32 %n) nounwind {
 entry:
@@ -21,26 +20,26 @@ entry:
        br label %forcond
 
 forcond:               ; preds = %forinc, %entry
-       %tmp = load i32* %i             ; <i32> [#uses=1]
-       %tmp1 = load i32* %n.addr               ; <i32> [#uses=1]
+       %tmp = load i32, i32* %i                ; <i32> [#uses=1]
+       %tmp1 = load i32, i32* %n.addr          ; <i32> [#uses=1]
        %cmp = icmp slt i32 %tmp, %tmp1         ; <i1> [#uses=1]
        br i1 %cmp, label %forbody, label %afterfor
 
 forbody:               ; preds = %forcond
-       %tmp2 = load i32* %i            ; <i32> [#uses=1]
-       %tmp3 = load <3 x i32>** %dst.addr              ; <<3 x i32>*> [#uses=1]
-       %arrayidx = getelementptr <3 x i32>* %tmp3, i32 %tmp2           ; <<3 x i32>*> [#uses=1]
-       %tmp4 = load i32* %i            ; <i32> [#uses=1]
-       %tmp5 = load <3 x i32>** %src.addr              ; <<3 x i32>*> [#uses=1]
-       %arrayidx6 = getelementptr <3 x i32>* %tmp5, i32 %tmp4          ; <<3 x i32>*> [#uses=1]
-       %tmp7 = load <3 x i32>* %arrayidx6              ; <<3 x i32>> [#uses=1]
+       %tmp2 = load i32, i32* %i               ; <i32> [#uses=1]
+       %tmp3 = load <3 x i32>*, <3 x i32>** %dst.addr          ; <<3 x i32>*> [#uses=1]
+       %arrayidx = getelementptr <3 x i32>, <3 x i32>* %tmp3, i32 %tmp2                ; <<3 x i32>*> [#uses=1]
+       %tmp4 = load i32, i32* %i               ; <i32> [#uses=1]
+       %tmp5 = load <3 x i32>*, <3 x i32>** %src.addr          ; <<3 x i32>*> [#uses=1]
+       %arrayidx6 = getelementptr <3 x i32>, <3 x i32>* %tmp5, i32 %tmp4               ; <<3 x i32>*> [#uses=1]
+       %tmp7 = load <3 x i32>, <3 x i32>* %arrayidx6           ; <<3 x i32>> [#uses=1]
        %mul = mul <3 x i32> %tmp7, < i32 4, i32 4, i32 4 >             ; <<3 x i32>> [#uses=1]
        %sub = sub <3 x i32> %mul, < i32 3, i32 3, i32 3 >              ; <<3 x i32>> [#uses=1]
        store <3 x i32> %sub, <3 x i32>* %arrayidx
        br label %forinc
 
 forinc:                ; preds = %forbody
-       %tmp8 = load i32* %i            ; <i32> [#uses=1]
+       %tmp8 = load i32, i32* %i               ; <i32> [#uses=1]
        %inc = add i32 %tmp8, 1         ; <i32> [#uses=1]
        store i32 %inc, i32* %i
        br label %forcond