[PATCH] D96248: [OpenMP][AMDGPU] Add support for linking libomptarget bitcode

Jon Chesterfield via Phabricator via cfe-commits cfe-commits at lists.llvm.org
Mon Feb 8 05:04:52 PST 2021


JonChesterfield added a comment.

I like this. Using the same logic, in the same function call, to find this library on either gpu is the right thing to do. Looks like a non functional change on nvptx, though phab doesn't make that obvious.



================
Comment at: clang/include/clang/Driver/Options.td:935
   HelpText<"Path to libomptarget-nvptx bitcode library">;
+def libomptarget_device_bc_path_EQ : Joined<["--"], "libomptarget-device-bc-path=">, Group<i_Group>,
+  HelpText<"Path to libomptarget bitcode library">;
----------------
I think there's an aliasing mechanism in the Options handling, where we can use device_bc_path as the canonical choice and nvptx_bc_path as a backwards-compatible argument that sets device_bc_path 


================
Comment at: clang/lib/Driver/ToolChains/CommonArgs.cpp:1655
+          DriverArgs.getLastArg(options::OPT_libomptarget_nvptx_bc_path_EQ,
+                                options::OPT_libomptarget_device_bc_path_EQ)) {
+    std::string LibOmpTargetName(A->getValue());
----------------
This, I think, can be reduced to checking OPT_libomptarget_device_bc_path_EQ if the aliasing machinery I remember does actually exist


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D96248/new/

https://reviews.llvm.org/D96248



More information about the cfe-commits mailing list