[all-commits] [llvm/llvm-project] 97f3be: [CUDA][HIP] Improve variable registration with the...

[Joseph Huber via All-commits]

  Branch: refs/heads/main
  Home:   https://github.com/llvm/llvm-project
  Commit: 97f3be2c5a0295632321141bdc001d4f81821958
      https://github.com/llvm/llvm-project/commit/97f3be2c5a0295632321141bdc001d4f81821958
  Author: Joseph Huber <huberjn at outlook.com>
  Date:   2023-12-07 (Thu, 07 Dec 2023)

  Changed paths:
    M clang/lib/CodeGen/CGCUDANV.cpp
    M clang/lib/CodeGen/CGCUDARuntime.h
    M clang/test/CodeGenCUDA/offloading-entries.cu
    M clang/test/Driver/linker-wrapper-image.c
    M clang/tools/clang-linker-wrapper/OffloadWrapper.cpp
    M llvm/include/llvm/Frontend/Offloading/Utility.h
    M llvm/lib/Frontend/Offloading/Utility.cpp
    M llvm/lib/Frontend/OpenMP/OMPIRBuilder.cpp

  Log Message:
  -----------
  [CUDA][HIP] Improve variable registration with the new driver (#73177)

Summary:
This patch adds support for registering texture / surface variables from
CUDA / HIP. Additionally, we now properly track the `extern` and `const`
flags that are also used in these runtime functions.

This does not implement the `managed` variables yet as those seem to
require some extra handling I'm not familiar with. The issue is that the
current offload entry isn't large enough to carry size and alignment
information along with an extra global.