[all-commits] [llvm/llvm-project] 69040d: [MLIR] Allow for multiple gpu modules during trans...

[Stephan Herhut via All-commits]

  Branch: refs/heads/master
  Home:   https://github.com/llvm/llvm-project
  Commit: 69040d5b0bfa59edacc2ad10d517b4270bf76845
      https://github.com/llvm/llvm-project/commit/69040d5b0bfa59edacc2ad10d517b4270bf76845
  Author: Stephan Herhut <herhut at google.com>
  Date:   2020-04-16 (Thu, 16 Apr 2020)

  Changed paths:
    M mlir/include/mlir/Dialect/LLVMIR/LLVMDialect.h
    M mlir/include/mlir/Dialect/LLVMIR/LLVMOpBase.td
    M mlir/include/mlir/Target/LLVMIR/ModuleTranslation.h
    M mlir/lib/Conversion/GPUToCUDA/ConvertKernelFuncToCubin.cpp
    M mlir/lib/Conversion/GPUToCUDA/ConvertLaunchFuncToCudaCalls.cpp
    M mlir/lib/Dialect/LLVMIR/CMakeLists.txt
    M mlir/lib/Dialect/LLVMIR/IR/LLVMDialect.cpp
    M mlir/lib/ExecutionEngine/CMakeLists.txt
    M mlir/lib/ExecutionEngine/ExecutionEngine.cpp
    M mlir/lib/Target/LLVMIR/ModuleTranslation.cpp
    A mlir/test/mlir-cuda-runner/two-modules.mlir

  Log Message:
  -----------
  [MLIR] Allow for multiple gpu modules during translation.

This change makes the ModuleTranslation threadsafe by locking on the
LLVMContext. Furthermore, we now clone the llvm module into a new
context when compiling to PTX similar to what the OrcJit does.

Differential Revision: https://reviews.llvm.org/D78207