Improve SME check, add runline to test for it

Author: huntergr-arm

llvm/lib/Target/AArch64/AArch64TargetTransformInfo.cpp

@@ -5600,7 +5600,8 @@ AArch64TTIImpl::getShuffleCost(TTI::ShuffleKind Kind, VectorType *DstTy,
   }
 
   // Segmented shuffle matching.
-  if ((ST->hasSVE2p1() || ST->hasSME2p1()) &&
+  if ((ST->hasSVE2p1() ||
+       (ST->hasSME2p1() && ST->isSVEorStreamingSVEAvailable())) &&
       Kind == TTI::SK_PermuteSingleSrc && isa<FixedVectorType>(Tp) &&
       Tp->getPrimitiveSizeInBits().isKnownMultipleOf(
           AArch64::SVEBitsPerBlock)) {

llvm/test/Analysis/CostModel/AArch64/segmented-shufflevector-patterns.ll

@@ -1,5 +1,6 @@
 ; NOTE: Assertions have been autogenerated by utils/update_analyze_test_checks.py UTC_ARGS: --version 5
-; RUN: opt -passes="print<cost-model>" -cost-kind=throughput 2>&1 -disable-output -mtriple=aarch64--linux-gnu < %s | FileCheck %s
+; RUN: opt -passes="print<cost-model>" -cost-kind=throughput 2>&1 -disable-output -mtriple=aarch64--linux-gnu -mattr=+sve2p1 < %s | FileCheck %s
+; RUN: opt -passes="print<cost-model>" -cost-kind=throughput 2>&1 -disable-output -mtriple=aarch64--linux-gnu -mattr=+sme2p1 -force-streaming < %s | FileCheck %s
 
 ;; Broadcast indexed lane within 128b segments (dupq zd.t, zn.t[idx])
 define void @dup_within_each_segment() #0 {
@@ -22,4 +23,4 @@ define void @dup_within_each_segment() #0 {
   ret void
 }
 
-attributes #0 = { noinline vscale_range(2,2) "target-features"="+sve2p1" }
+attributes #0 = { noinline vscale_range(2,2) }