[PATCH] D76514: [ARM] Avoid pointless vrev of element-wise vmov

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added inline comments.


================
Comment at: llvm/test/CodeGen/ARM/vmov.ll:1
-; RUN: llc -mtriple=arm-eabi -mattr=+neon %s -o - | FileCheck %s
+; RUN: llc -mtriple=arm-eabi -mattr=+neon,+fullfp16 %s -o - | FileCheck %s
+; RUN: llc -mtriple=armeb-eabi -mattr=+neon,+fullfp16 %s -o - | FileCheck %s
----------------
john.brawn wrote:
> dmgreen wrote:
> > Can you please run the update script on this with old codegen, pre-commit the result and then show the diffs against new codegen here.
> > 
> > It might help to make them hard-float too.
> I tried using update_llc_test_checks.py, but it doesn't really work well here because the compiler output is different when big-endian and little-endian (for the llc before this patch, and also in the tests that aren't to do with this patch) and the only way the script can cope with that is to have entirely separate check prefixes, which doesn't do a good job of checking that the code generated doesn't change depending on endianness.
you can add common prefixes:

--check-prefixes=ALL,LE
--check-prefixes=ALL,BE


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D76514/new/

https://reviews.llvm.org/D76514