[clang] [llvm] [LLVM][SVE] Remove aarch64.sve.rev intrinsic, using vector.reverse instead. (PR #169654)

[Paul Walker via llvm-commits]

================
@@ -1015,6 +1015,14 @@ static bool upgradeArmOrAarch64IntrinsicFunction(bool IsArm, Function *F,
         }
         return false; // No other 'aarch64.sve.tuple.*'.
       }
+
+      if (Name.starts_with("rev.nxv")) {
----------------
paulwalker-arm wrote:

That's not really how bitcode tests are supposed to work, but I've added aarch64-sve-rev-upgrade.ll[.bc] instead.

https://github.com/llvm/llvm-project/pull/169654