1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-01 16:33:37 +01:00
llvm-mirror/test/CodeGen/X86/2006-04-04-CrossBlockCrash.ll
Eric Christopher 593cfc9984 Fix PR5315, original patch by Nicolas Capens!
llvm-svn: 86203
2009-11-06 00:11:57 +00:00

51 lines
3.1 KiB
LLVM

; RUN: llc < %s -march=x86 -mcpu=yonah
; END.
target datalayout = "e-p:32:32"
target triple = "i686-apple-darwin8.6.1"
%struct.GLTColor4 = type { float, float, float, float }
%struct.GLTCoord3 = type { float, float, float }
%struct.__GLIContextRec = type { { %struct.anon, { [24 x [16 x float]], [24 x [16 x float]] }, %struct.GLTColor4, { float, float, float, float, %struct.GLTCoord3, float } }, { float, float, float, float, float, float, float, float, [4 x i32], [4 x i32], [4 x i32] } }
%struct.__GLvertex = type { %struct.GLTColor4, %struct.GLTColor4, %struct.GLTColor4, %struct.GLTColor4, %struct.GLTColor4, %struct.GLTCoord3, float, %struct.GLTColor4, float, float, float, i8, i8, i8, i8, [4 x float], [2 x i8*], i32, i32, [16 x %struct.GLTColor4] }
%struct.anon = type { float, float, float, float, float, float, float, float }
declare <4 x float> @llvm.x86.sse.cmp.ps(<4 x float>, <4 x float>, i8)
declare <8 x i16> @llvm.x86.sse2.packssdw.128(<4 x i32>, <4 x i32>)
declare i32 @llvm.x86.sse2.pmovmskb.128(<16 x i8>)
define void @gleLLVMVecInterpolateClip() {
entry:
br i1 false, label %cond_false, label %cond_false183
cond_false: ; preds = %entry
br i1 false, label %cond_false183, label %cond_true69
cond_true69: ; preds = %cond_false
ret void
cond_false183: ; preds = %cond_false, %entry
%vuizmsk.0.1 = phi <4 x i32> [ < i32 -1, i32 -1, i32 -1, i32 0 >, %entry ], [ < i32 -1, i32 0, i32 0, i32 0 >, %cond_false ] ; <<4 x i32>> [#uses=2]
%tmp192 = extractelement <4 x i32> %vuizmsk.0.1, i32 2 ; <i32> [#uses=1]
%tmp193 = extractelement <4 x i32> %vuizmsk.0.1, i32 3 ; <i32> [#uses=2]
%tmp195 = insertelement <4 x i32> zeroinitializer, i32 %tmp192, i32 1 ; <<4 x i32>> [#uses=1]
%tmp196 = insertelement <4 x i32> %tmp195, i32 %tmp193, i32 2 ; <<4 x i32>> [#uses=1]
%tmp197 = insertelement <4 x i32> %tmp196, i32 %tmp193, i32 3 ; <<4 x i32>> [#uses=1]
%tmp336 = and <4 x i32> zeroinitializer, %tmp197 ; <<4 x i32>> [#uses=1]
%tmp337 = bitcast <4 x i32> %tmp336 to <4 x float> ; <<4 x float>> [#uses=1]
%tmp378 = tail call <4 x float> @llvm.x86.sse.cmp.ps( <4 x float> %tmp337, <4 x float> zeroinitializer, i8 1 ) ; <<4 x float>> [#uses=1]
%tmp379 = bitcast <4 x float> %tmp378 to <4 x i32> ; <<4 x i32>> [#uses=1]
%tmp388 = tail call <8 x i16> @llvm.x86.sse2.packssdw.128( <4 x i32> zeroinitializer, <4 x i32> %tmp379 ) ; <<4 x i32>> [#uses=1]
%tmp392 = bitcast <8 x i16> %tmp388 to <8 x i16> ; <<8 x i16>> [#uses=1]
%tmp399 = extractelement <8 x i16> %tmp392, i32 7 ; <i16> [#uses=1]
%tmp423 = insertelement <8 x i16> zeroinitializer, i16 %tmp399, i32 7 ; <<8 x i16>> [#uses=1]
%tmp427 = bitcast <8 x i16> %tmp423 to <16 x i8> ; <<16 x i8>> [#uses=1]
%tmp428 = tail call i32 @llvm.x86.sse2.pmovmskb.128( <16 x i8> %tmp427 ) ; <i32> [#uses=1]
%tmp432 = trunc i32 %tmp428 to i8 ; <i8> [#uses=1]
%tmp = and i8 %tmp432, 42 ; <i8> [#uses=1]
%tmp436 = bitcast i8 %tmp to i8 ; <i8> [#uses=1]
%tmp446 = zext i8 %tmp436 to i32 ; <i32> [#uses=1]
%tmp447 = shl i32 %tmp446, 24 ; <i32> [#uses=1]
%tmp449 = or i32 0, %tmp447 ; <i32> [#uses=1]
store i32 %tmp449, i32* null
ret void
}