[llvm] LoopVectorize/X86: add test for crash in #72969 (PR #74111)

[Ramkumar Ramachandra via llvm-commits]

================
@@ -0,0 +1,26 @@
+; RUN: not --crash opt -passes=loop-vectorize -S < %s
+
+target triple = "x86_64-unknown-linux-gnu"
----------------
artagnon wrote:

Unfortunately, it can't be made target-independent, as it depends on the register bitwidth and instruction costs. The bug reproduces on AArch64 as well with `-mattr=-sve,+neon`: should I duplicate this test under that target?

https://github.com/llvm/llvm-project/pull/74111