[llvm] [AArch64][SVE] - Fix crash Possible incorrect use of EVT::getVectorNumElements(). (PR #218733)

[Simon Pilgrim via llvm-commits]

================
@@ -0,0 +1,57 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN: llc -mtriple=aarch64-linux-gnu -mattr=+sve < %s | FileCheck %s
+
+
+define void @blend_heat_16bit(ptr %0, ptr %1, ptr %2, i64 %3) {
----------------
RKSimon wrote:

You've taken the original IR from the bug report, that feeds into clang (not llc!) - you then need to start from the optimized IR spat out by clang (-emit-llvm) and then reduce that - but the llvm-autoreduce bot has done it for you:

https://github.com/dtcxzyw/llvm-autoreduce/issues/348

https://github.com/llvm/llvm-project/pull/218733