[PATCH] D42993: [AArch64] Fixes for ARMv8.2-A FP16 scalar intrinsic

[Abderrazek Zaafrani via Phabricator via cfe-commits]

az created this revision.
az added a reviewer: SjoerdMeijer.
Herald added subscribers: hiraditya, kristof.beyls, javed.absar, rengolin, aemerson.

A couple of fixes on top of https://reviews.llvm.org/D41792:

- Fixes for freceprical, and fsqrt instructions in the backend.
- The intrinsics that generate builtin calls with i16 data types fails in instruction selection. In a preparation for future fixes in the backend for these, the code generated by the frontend is modified. For example, a builtin that returns i16 is implemented by returning i32 that is truncated to i16. This is done similar to other non-fp16 intrinsics that produces i16 data types.
- Fix frontend code generated for abs.


https://reviews.llvm.org/D42993

Files:
  clang/lib/CodeGen/CGBuiltin.cpp
  clang/test/CodeGen/aarch64-v8.2a-fp16-intrinsics.c
  llvm/lib/Target/AArch64/AArch64InstrInfo.td

-------------- next part --------------
A non-text attachment was scrubbed...
Name: D42993.133101.patch
Type: text/x-patch
Size: 16697 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20180207/7e3b66c1/attachment-0001.bin>