jhuber6 created this revision.
jhuber6 added reviewers: jdoerfert, tra, yaxunl, JonChesterfield.
Herald added a project: All.
jhuber6 requested review of this revision.
Herald added subscribers: cfe-commits, MaskRay.
Herald added a project: clang.
The new driver primarily allows us to support RDC-mode compilations with
proper linking. This is not needed for non-RDC mode compilation, but we
still would like the new driver to be able to handle this mode so we can
transition away from the old driver in the future. This patch adds the
necessary code to support creating a fatbinary for HIP code generation.
Repository:
rG LLVM Github Monorepo
https://reviews.llvm.org/D129784
Files:
clang/lib/Driver/Driver.cpp
clang/lib/Driver/ToolChains/Clang.cpp
clang/test/Driver/hip-binding.hip
clang/test/Driver/hip-phases.hip
clang/test/Driver/hip-toolchain-no-rdc.hip
Index: clang/test/Driver/hip-toolchain-no-rdc.hip
===================================================================
--- clang/test/Driver/hip-toolchain-no-rdc.hip
+++ clang/test/Driver/hip-toolchain-no-rdc.hip
@@ -21,6 +21,16 @@
// RUN: %S/Inputs/hip_multiple_inputs/b.hip \
// RUN: 2>&1 | FileCheck -check-prefixes=CHECK %s
+// RUN: %clang -### -target x86_64-linux-gnu -fno-gpu-rdc \
+// RUN: -x hip --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 \
+// RUN: --hip-device-lib=lib1.bc --hip-device-lib=lib2.bc \
+// RUN: --hip-device-lib-path=%S/Inputs/hip_multiple_inputs/lib1 \
+// RUN: --hip-device-lib-path=%S/Inputs/hip_multiple_inputs/lib2 \
+// RUN: -fuse-ld=lld -nogpuinc --offload-new-driver -c \
+// RUN: %S/Inputs/hip_multiple_inputs/a.cu \
+// RUN: %S/Inputs/hip_multiple_inputs/b.hip \
+// RUN: 2>&1 | FileCheck -check-prefixes=CHECK %s
+
// RUN: touch %T/a.o
// RUN: touch %T/b.o
// RUN: %clang -### -target x86_64-linux-gnu \
Index: clang/test/Driver/hip-phases.hip
===================================================================
--- clang/test/Driver/hip-phases.hip
+++ clang/test/Driver/hip-phases.hip
@@ -11,7 +11,10 @@
//
// RUN: %clang -x hip -target x86_64-unknown-linux-gnu -ccc-print-phases \
// RUN: --cuda-gpu-arch=gfx803 %s 2>&1 \
-// RUN: | FileCheck -check-prefixes=BIN,NRD %s
+// RUN: | FileCheck -check-prefixes=BIN,NRD,OLD %s
+// RUN: %clang -x hip -target x86_64-unknown-linux-gnu -ccc-print-phases \
+// RUN: --offload-new-driver --cuda-gpu-arch=gfx803 %s 2>&1 \
+// RUN: | FileCheck -check-prefixes=BIN,NRD,NEW %s
//
// RUN: %clang -x hip -target x86_64-unknown-linux-gnu -ccc-print-phases \
// RUN: --cuda-gpu-arch=gfx803 -fgpu-rdc %s 2>&1 \
@@ -38,7 +41,8 @@
// RDC-DAG: [[P11:[0-9]+]]: offload, "device-[[T]] (amdgcn-amd-amdhsa)" {[[P10]]}, object
// NRD-DAG: [[P12:[0-9]+]]: backend, {[[P11]]}, assembler, (host-[[T]])
// NRD-DAG: [[P13:[0-9]+]]: assembler, {[[P12]]}, object, (host-[[T]])
-// NRD-DAG: [[P14:[0-9]+]]: linker, {[[P13]]}, image, (host-[[T]])
+// OLD-DAG: [[P14:[0-9]+]]: linker, {[[P13]]}, image, (host-[[T]])
+// NEW-DAG: [[P14:[0-9]+]]: clang-linker-wrapper, {[[P13]]}, image, (host-[[T]])
// RDC-DAG: [[P14:[0-9]+]]: linker, {[[P13]], [[P11]]}, image, (host-[[T]])
//
Index: clang/test/Driver/hip-binding.hip
===================================================================
--- clang/test/Driver/hip-binding.hip
+++ clang/test/Driver/hip-binding.hip
@@ -4,6 +4,9 @@
// RUN: %clang -ccc-print-bindings -target x86_64-linux-gnu \
// RUN: --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 %s \
// RUN: -c 2>&1 | FileCheck -check-prefix=NRDCS %s
+// RUN: %clang -ccc-print-bindings -target x86_64-linux-gnu --offload-new-driver \
+// RUN: --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 %s \
+// RUN: -c 2>&1 | FileCheck -check-prefix=NRDCS %s
// NRDCS: # "amdgcn-amd-amdhsa" - "clang", inputs: ["[[IN:.*hip-binding.hip]]"], output: "[[OBJ1:.*o]]"
// NRDCS: # "amdgcn-amd-amdhsa" - "AMDGCN::Linker", inputs: ["[[OBJ1]]"], output: "[[IMG1:.*]]"
// NRDCS: # "amdgcn-amd-amdhsa" - "clang", inputs: ["[[IN:.*hip-binding.hip]]"], output: "[[OBJ2:.*o]]"
@@ -19,6 +22,14 @@
// RDCS: # "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[IN]]"], output: "[[HOSTOBJ:.*o]]"
// RDCS: # "x86_64-unknown-linux-gnu" - "offload bundler", inputs: ["[[BC1]]", "[[BC2]]", "[[HOSTOBJ]]"], output: "{{.*}}"
+// RUN: %clang -ccc-print-bindings -target x86_64-linux-gnu --offload-new-driver \
+// RUN: --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 %s \
+// RUN: -c -fgpu-rdc 2>&1 | FileCheck -check-prefix=RDCS-NEW %s
+// RDCS-NEW: # "amdgcn-amd-amdhsa" - "clang", inputs: ["[[INPUT:.+]]"], output: "[[HIP803:.+]]"
+// RDCS-NEW: # "amdgcn-amd-amdhsa" - "clang", inputs: ["[[INPUT]]"], output: "[[HIP900:.+]]"
+// RDCS-NEW: # "x86_64-unknown-linux-gnu" - "Offload::Packager", inputs: ["[[HIP803]]", "[[HIP900]]"], output: "[[HIPFB:.+]]"
+// RDCS-NEW: # "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[INPUT]]", "[[HIPFB]]"], output: "{{.*}}"
+
// RUN: touch %t.o
// RUN: %clang --hip-link -ccc-print-bindings -target x86_64-linux-gnu \
// RUN: --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 -fgpu-rdc %t.o\
Index: clang/lib/Driver/ToolChains/Clang.cpp
===================================================================
--- clang/lib/Driver/ToolChains/Clang.cpp
+++ clang/lib/Driver/ToolChains/Clang.cpp
@@ -6986,7 +6986,7 @@
CmdArgs.push_back("-fcuda-include-gpubinary");
CmdArgs.push_back(CudaDeviceInput->getFilename());
} else if (!HostOffloadingInputs.empty()) {
- if (IsCuda && !IsRDCMode) {
+ if ((IsCuda || IsHIP) && !IsRDCMode) {
assert(HostOffloadingInputs.size() == 1 && "Only one input expected");
CmdArgs.push_back("-fcuda-include-gpubinary");
CmdArgs.push_back(HostOffloadingInputs.front().getFilename());
Index: clang/lib/Driver/Driver.cpp
===================================================================
--- clang/lib/Driver/Driver.cpp
+++ clang/lib/Driver/Driver.cpp
@@ -4473,6 +4473,15 @@
}
}
+ // Compiling HIP in non-RDC mode requires linking each action individually.
+ for (Action *&A : DeviceActions) {
+ if (A->getType() != types::TY_Object || Kind != Action::OFK_HIP ||
+ Args.hasFlag(options::OPT_fgpu_rdc, options::OPT_fno_gpu_rdc, false))
+ continue;
+ ActionList LinkerInput = {A};
+ A = C.MakeAction<LinkJobAction>(LinkerInput, types::TY_Image);
+ }
+
auto TCAndArch = TCAndArchs.begin();
for (Action *A : DeviceActions) {
DDeps.add(*A, *TCAndArch->first, TCAndArch->second.data(), Kind);
@@ -4489,12 +4498,21 @@
OffloadAction::DeviceDependences DDep;
if (C.isOffloadingHostKind(Action::OFK_Cuda) &&
!Args.hasFlag(options::OPT_fgpu_rdc, options::OPT_fno_gpu_rdc, false)) {
- // If we are not in RDC-mode we just emit the final CUDA fatbinary for each
- // translation unit without requiring any linking.
+ // If we are not in RDC-mode we just emit the final CUDA fatbinary for
+ // each translation unit without requiring any linking.
Action *FatbinAction =
C.MakeAction<LinkJobAction>(OffloadActions, types::TY_CUDA_FATBIN);
DDep.add(*FatbinAction, *C.getSingleOffloadToolChain<Action::OFK_Cuda>(),
nullptr, Action::OFK_Cuda);
+ } else if (C.isOffloadingHostKind(Action::OFK_HIP) &&
+ !Args.hasFlag(options::OPT_fgpu_rdc, options::OPT_fno_gpu_rdc,
+ false)) {
+ // If we are not in RDC-mode we just emit the final HIP fatbinary for each
+ // translation unit, linking each input individually.
+ Action *FatbinAction =
+ C.MakeAction<LinkJobAction>(OffloadActions, types::TY_HIP_FATBIN);
+ DDep.add(*FatbinAction, *C.getSingleOffloadToolChain<Action::OFK_HIP>(),
+ nullptr, Action::OFK_HIP);
} else {
// Package all the offloading actions into a single output that can be
// embedded in the host and linked.
@@ -4503,6 +4521,7 @@
DDep.add(*PackagerAction, *C.getSingleOffloadToolChain<Action::OFK_Host>(),
nullptr, Action::OFK_None);
}
+
OffloadAction::HostDependence HDep(
*HostAction, *C.getSingleOffloadToolChain<Action::OFK_Host>(),
/*BoundArch=*/nullptr, isa<CompileJobAction>(HostAction) ? DDep : DDeps);
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
