diff options
author | Jonas Paulsson <paulsson@linux.vnet.ibm.com> | 2017-03-07 09:49:31 +0000 |
---|---|---|
committer | Jonas Paulsson <paulsson@linux.vnet.ibm.com> | 2017-03-07 09:49:31 +0000 |
commit | 1d33cd398880f662cbfbb3a28c2889a617015ee6 (patch) | |
tree | 384d27ca56f02103c27cb27bc60d9e240c87124f | |
parent | 24fb250cad9ecec134287661f024dcdcc4b50d89 (diff) | |
download | bcm5719-llvm-1d33cd398880f662cbfbb3a28c2889a617015ee6.tar.gz bcm5719-llvm-1d33cd398880f662cbfbb3a28c2889a617015ee6.zip |
[SystemZ] Add check VT.isSimple() in canTreateAsByteVector()
Since BB-vectorizer can produce vectors of for example 3 elements,
this check is needed.
Review: Ulrich Weigand
llvm-svn: 297136
-rw-r--r-- | llvm/lib/Target/SystemZ/SystemZISelLowering.cpp | 2 | ||||
-rw-r--r-- | llvm/test/CodeGen/SystemZ/vectorizer-output-3xi32.ll | 10 |
2 files changed, 11 insertions, 1 deletions
diff --git a/llvm/lib/Target/SystemZ/SystemZISelLowering.cpp b/llvm/lib/Target/SystemZ/SystemZISelLowering.cpp index afc9e60bdf2..6322098e1c3 100644 --- a/llvm/lib/Target/SystemZ/SystemZISelLowering.cpp +++ b/llvm/lib/Target/SystemZ/SystemZISelLowering.cpp @@ -4737,7 +4737,7 @@ const char *SystemZTargetLowering::getTargetNodeName(unsigned Opcode) const { // Return true if VT is a vector whose elements are a whole number of bytes // in width. static bool canTreatAsByteVector(EVT VT) { - return VT.isVector() && VT.getScalarSizeInBits() % 8 == 0; + return VT.isVector() && VT.getScalarSizeInBits() % 8 == 0 && VT.isSimple(); } // Try to simplify an EXTRACT_VECTOR_ELT from a vector of type VecVT diff --git a/llvm/test/CodeGen/SystemZ/vectorizer-output-3xi32.ll b/llvm/test/CodeGen/SystemZ/vectorizer-output-3xi32.ll new file mode 100644 index 00000000000..3e7ba609592 --- /dev/null +++ b/llvm/test/CodeGen/SystemZ/vectorizer-output-3xi32.ll @@ -0,0 +1,10 @@ +; RUN: llc < %s -mtriple=s390x-linux-gnu -mcpu=z13 +; +; This tescase origininates from the BB-vectorizer output. + +define void @fun() { + %1 = zext <3 x i1> zeroinitializer to <3 x i32> + %2 = extractelement <3 x i32> %1, i32 2 + store i32 %2, i32* undef, align 8 + unreachable +} |