It's not necessary to do rounding for alloca operations when the requested

alignment is equal to the stack alignment.


git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@40004 91177308-0d34-0410-b5e6-96231b3b80d8

1 files changed, 72 insertions, 0 deletions

diff --git a/test/CodeGen/PowerPC/2006-04-19-vmaddfp-crash.ll b/test/CodeGen/PowerPC/2006-04-19-vmaddfp-crash.ll
new file mode 100644
index 0000000..59f7ed4
--- /dev/null
+++ b/test/CodeGen/PowerPC/2006-04-19-vmaddfp-crash.ll
@@ -0,0 +1,72 @@
+; RUN: llvm-upgrade < %s | llvm-as | llc -march=ppc32 -mcpu=g5
+
+void %test(sbyte* %stack) {
+entry:
+	%tmp9 = seteq int 0, 0		; <bool> [#uses=1]
+	%tmp30 = seteq uint 0, 0		; <bool> [#uses=1]
+	br bool %tmp30, label %cond_next54, label %cond_true31
+
+cond_true860:		; preds = %bb855
+	%tmp879 = tail call <4 x float> %llvm.ppc.altivec.vmaddfp( <4 x float> zeroinitializer, <4 x float> zeroinitializer, <4 x float> zeroinitializer )		; <<4 x float>> [#uses=1]
+	%tmp880 = cast <4 x float> %tmp879 to <4 x int>		; <<4 x int>> [#uses=2]
+	%tmp883 = shufflevector <4 x int> %tmp880, <4 x int> undef, <4 x uint> < uint 1, uint 1, uint 1, uint 1 >		; <<4 x int>> [#uses=1]
+	%tmp883 = cast <4 x int> %tmp883 to <4 x float>		; <<4 x float>> [#uses=1]
+	%tmp885 = shufflevector <4 x int> %tmp880, <4 x int> undef, <4 x uint> < uint 2, uint 2, uint 2, uint 2 >		; <<4 x int>> [#uses=1]
+	%tmp885 = cast <4 x int> %tmp885 to <4 x float>		; <<4 x float>> [#uses=1]
+	br label %cond_next905
+
+cond_true31:		; preds = %entry
+	ret void
+
+cond_next54:		; preds = %entry
+	br bool %tmp9, label %cond_false385, label %bb279
+
+bb279:		; preds = %cond_next54
+	ret void
+
+cond_false385:		; preds = %cond_next54
+	%tmp388 = seteq uint 0, 0		; <bool> [#uses=1]
+	br bool %tmp388, label %cond_next463, label %cond_true389
+
+cond_true389:		; preds = %cond_false385
+	ret void
+
+cond_next463:		; preds = %cond_false385
+	%tmp1208107 = setgt sbyte* null, %stack		; <bool> [#uses=1]
+	br bool %tmp1208107, label %cond_true1209.preheader, label %bb1212
+
+cond_true498:		; preds = %cond_true1209.preheader
+	ret void
+
+cond_true519:		; preds = %cond_true1209.preheader
+	%bothcond = or bool false, false		; <bool> [#uses=1]
+	br bool %bothcond, label %bb855, label %bb980
+
+cond_false548:		; preds = %cond_true1209.preheader
+	ret void
+
+bb855:		; preds = %cond_true519
+	%tmp859 = seteq int 0, 0		; <bool> [#uses=1]
+	br bool %tmp859, label %cond_true860, label %cond_next905
+
+cond_next905:		; preds = %bb855, %cond_true860
+	%vfpw2.4 = phi <4 x float> [ %tmp885, %cond_true860 ], [ undef, %bb855 ]		; <<4 x float>> [#uses=0]
+	%vfpw1.4 = phi <4 x float> [ %tmp883, %cond_true860 ], [ undef, %bb855 ]		; <<4 x float>> [#uses=0]
+	%tmp930 = cast <4 x float> zeroinitializer to <4 x int>		; <<4 x int>> [#uses=0]
+	ret void
+
+bb980:		; preds = %cond_true519
+	ret void
+
+cond_true1209.preheader:		; preds = %cond_next463
+	%tmp496 = and uint 0, 12288		; <uint> [#uses=1]
+	switch uint %tmp496, label %cond_false548 [
+		 uint 0, label %cond_true498
+		 uint 4096, label %cond_true519
+	]
+
+bb1212:		; preds = %cond_next463
+	ret void
+}
+
+declare <4 x float> %llvm.ppc.altivec.vmaddfp(<4 x float>, <4 x float>, <4 x float>)