[llvm] [VectorCombine] Fold vector.interleave2 with two constant splats (PR #125144)

[Craig Topper via llvm-commits]

================
@@ -0,0 +1,17 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py UTC_ARGS: --version 5
+; RUN: opt -S -mtriple=riscv64 -mattr=+v,+m,+zvfh %s -passes=vector-combine | FileCheck %s
+; RUN: opt -S -mtriple=riscv32 -mattr=+v,+m,+zvfh %s -passes=vector-combine | FileCheck %s
+
+; We should not form a i128 vector.
+
+define void @interleave2_const_splat_nxv8i64(ptr %dst) {
----------------
topperc wrote:

Does this have some bad interaction with zve32x that required a separate test?

https://github.com/llvm/llvm-project/pull/125144