[Clang] Only emit textual LLVM-IR in device only mode

Currently, we embed device code into the host to perform
multi-architecture linking and handling of device code. If the user
specified `-S -emit-llvm` then the embedded output will be textual
LLVM-IR. This is a problem because it can't be used by the LTO backend
and it makes reading the file confusing.

This patch changes the behaviour to only emit textual device IR if we
are in device only mode, that is, if the device code is presented
directly to the user instead of being embedded. Otherwise we should
always embed device bitcode instead.

Reviewed By: tra

Differential Revision: https://reviews.llvm.org/D141717

diff --git a/clang/lib/Driver/Driver.cpp b/clang/lib/Driver/Driver.cpp
index 49faa19..661d997 100644 (file)
--- a/clang/lib/Driver/Driver.cpp
+++ b/clang/lib/Driver/Driver.cpp
@@ -4634,7 +4634,14 @@ Action *Driver::ConstructPhaseAction(
                        false) ||
           TargetDeviceOffloadKind == Action::OFK_OpenMP))) {
       types::ID Output =
-          Args.hasArg(options::OPT_S) ? types::TY_LLVM_IR : types::TY_LLVM_BC;
+          Args.hasArg(options::OPT_S) &&
+                  (TargetDeviceOffloadKind == Action::OFK_None ||
+                   offloadDeviceOnly() ||
+                   (TargetDeviceOffloadKind == Action::OFK_HIP &&
+                    !Args.hasFlag(options::OPT_offload_new_driver,
+                                  options::OPT_no_offload_new_driver, false)))
+              ? types::TY_LLVM_IR
+              : types::TY_LLVM_BC;
       return C.MakeAction<BackendJobAction>(Input, Output);
     }
     return C.MakeAction<BackendJobAction>(Input, types::TY_PP_Asm);

diff --git a/clang/test/Driver/openmp-offload-gpu.c b/clang/test/Driver/openmp-offload-gpu.c
index 9a05daf..b1e1954 100644 (file)
--- a/clang/test/Driver/openmp-offload-gpu.c
+++ b/clang/test/Driver/openmp-offload-gpu.c
@@ -304,9 +304,16 @@
 // CHECK-IR: "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[INPUT_IR:.+]]"], output: "[[OBJECT:.+]]"
 // CHECK-IR: "x86_64-unknown-linux-gnu" - "Offload::Linker", inputs: ["[[OBJECT]]"], output: "a.out"
 
-// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -emit-llvm -S -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 -nogpulib %s 2>&1 | FileCheck %s --check-prefix=CHECK-EMIT-LLVM-IR
+// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -emit-llvm -S -fopenmp=libomp --offload-device-only \
+// RUN:     -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 -nogpulib %s 2>&1 \
+// RUN:   | FileCheck %s --check-prefix=CHECK-EMIT-LLVM-IR
 // CHECK-EMIT-LLVM-IR: "-cc1"{{.*}}"-triple" "nvptx64-nvidia-cuda"{{.*}}"-emit-llvm"
 
+// RUN:   %clang -### --target=x86_64-unknown-linux-gnu -emit-llvm -S -fopenmp=libomp \
+// RUN:     -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvidia-cuda -march=sm_52 -nogpulib %s 2>&1 \
+// RUN:   | FileCheck %s --check-prefix=CHECK-EMIT-LLVM-IR-BC
+// CHECK-EMIT-LLVM-IR-BC: "-cc1"{{.*}}"-triple" "nvptx64-nvidia-cuda"{{.*}}"-emit-llvm-bc"
+
 // RUN:   %clang -### -fopenmp=libomp -fopenmp-targets=nvptx64-nvidia-cuda -Xopenmp-target=nvptx64-nvida-cuda -march=sm_70 \
 // RUN:          --libomptarget-nvptx-bc-path=%S/Inputs/libomptarget/libomptarget-new-nvptx-test.bc \
 // RUN:          -nogpulib %s -o openmp-offload-gpu 2>&1 \