[all-commits] [llvm/llvm-project] e53175: clang: Add -fconvergent-functions flag

[Matt Arsenault via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: e531750c6cf9ab6ca987ffbfe100b1d766269eb5
      https://github.com/llvm/llvm-project/commit/e531750c6cf9ab6ca987ffbfe100b1d766269eb5
  Author: Matt Arsenault <Matthew.Arsenault at amd.com>
  Date:   2019-11-19 (Tue, 19 Nov 2019)

  Changed paths:
    M clang/include/clang/Basic/LangOptions.def
    M clang/include/clang/Basic/LangOptions.h
    M clang/include/clang/Driver/Options.td
    M clang/lib/Frontend/CompilerInvocation.cpp
    A clang/test/CodeGen/convergent-functions.cpp
    M clang/test/CodeGenCUDA/propagate-metadata.cu

  Log Message:
  -----------
  clang: Add -fconvergent-functions flag

The CUDA builtin library is apparently compiled in C++ mode, so the
assumption of convergent needs to be made in a typically non-SPMD
language. The functions in the library should still be assumed
convergent. Currently they are not, which is potentially incorrect and
this happens to work after the library is linked.