[llvm] [ARM][Codegen] Fix vector data miscompilation in arm32be (PR #105519)

[David Green via llvm-commits]

================
@@ -227,7 +226,6 @@ define arm_aapcs_vfpcc <8 x i16> @xor_v8i16_m1(<8 x i16> %a) {
 ; CHECK-BE:       @ %bb.0:
 ; CHECK-BE-NEXT:    vmvn.i32 q8, #0x10000
 ; CHECK-BE-NEXT:    vrev64.16 q9, q0
-; CHECK-BE-NEXT:    vrev32.16 q8, q8
 ; CHECK-BE-NEXT:    veor q8, q9, q8
----------------
davemgreen wrote:

This looks more correct now, nice. Can you remove the FIXME above the function.

There are some others comments on other routines that can hopefully be removed if they are fixed now too.

https://github.com/llvm/llvm-project/pull/105519