[HIP] Allow the new driver to compile HIP in non-RDC mode

[jhuber6]

Summary

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.

[jhuber6]

ping

[yaxunl]

LGTM. Thanks.

clang/lib/Driver/Driver.cpp

for (phases::ID Phase : PL) {
           OffloadAction::DeviceDependences DDep;
           DDep.add(*A, *TCAndArch->first, TCAndArch->second.data(), Kind);
           A = C.MakeAction<OffloadAction>(HDep, DDep);
         }
         ++TCAndArch;
       }
     }
 
+    // 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);
       OffloadAction::DeviceDependences DDep;
       DDep.add(*A, *TCAndArch->first, TCAndArch->second.data(), Kind);
       OffloadActions.push_back(C.MakeAction<OffloadAction>(DDep, A->getType()));
       ++TCAndArch;
     }
   }
 
   if (offloadDeviceOnly())
     return C.MakeAction<OffloadAction>(DDeps, types::TY_Nothing);
 
   if (OffloadActions.empty())
     return HostAction;
 
   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.
     Action *PackagerAction =
         C.MakeAction<OffloadPackagerJobAction>(OffloadActions, types::TY_Image);
     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);
   return C.MakeAction<OffloadAction>(
       HDep, isa<CompileJobAction>(HostAction) ? DDep : DDeps);
 }
 
 Action *Driver::ConstructPhaseAction(

clang/lib/Driver/ToolChains/Clang.cpp

 
   // Host-side offloading compilation receives all device-side outputs. Include
   // them in the host compilation depending on the target. If the host inputs
   // are not empty we use the new-driver scheme, otherwise use the old scheme.
   if ((IsCuda || IsHIP) && CudaDeviceInput) {
     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());
     } else {
       for (const InputInfo Input : HostOffloadingInputs)
         CmdArgs.push_back(Args.MakeArgString("-fembed-offload-object=" +
                                              TC.getInputFilename(Input)));
     }

clang/test/Driver/hip-binding.hip

 // REQUIRES: x86-registered-target
 // REQUIRES: amdgpu-registered-target
 
 // 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]]"
 // NRDCS: # "amdgcn-amd-amdhsa" - "AMDGCN::Linker", inputs: ["[[OBJ2]]"], output: "[[IMG2:.*]]"
 // NRDCS: # "amdgcn-amd-amdhsa" - "AMDGCN::Linker", inputs: ["[[IMG1]]", "[[IMG2]]"], output: "[[FATBIN:.*]]"
 // NRDCS: # "x86_64-unknown-linux-gnu" - "clang", inputs: ["[[IN]]", "[[FATBIN]]"], output: "{{.*}}"
 
 // RUN: %clang -ccc-print-bindings -target x86_64-linux-gnu \
 // RUN:   --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 %s \
 // RUN:   -c -fgpu-rdc 2>&1 | FileCheck -check-prefix=RDCS %s
 // RDCS: # "amdgcn-amd-amdhsa" - "clang", inputs: ["[[IN:.*hip-binding.hip]]"], output: "[[BC1:.*bc]]"
 // RDCS: # "amdgcn-amd-amdhsa" - "clang", inputs: ["[[IN:.*hip-binding.hip]]"], output: "[[BC2:.*bc]]"
 // 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\
 // RUN: 2>&1 | FileCheck %s
 
 // CHECK: # "x86_64-unknown-linux-gnu" - "offload bundler", inputs: ["[[IN:.*o]]"], outputs: ["[[HOSTOBJ:.*o]]", "{{.*o}}", "{{.*o}}"]
 // CHECK: # "amdgcn-amd-amdhsa" - "offload bundler", inputs: ["[[IN]]"], outputs: ["{{.*o}}", "[[DOBJ1:.*o]]", "[[DOBJ2:.*o]]"]
 // CHECK: # "amdgcn-amd-amdhsa" - "AMDGCN::Linker", inputs: ["[[DOBJ1]]"], output: "[[IMG1:.*out]]"

clang/test/Driver/hip-phases.hip

 // Tests the phases generated for a CUDA offloading target for different
 // combinations of:
 // - Number of gpu architectures;
 // - Host/device-only compilation;
 // - User-requested final phase - binary or assembly.
 
 // REQUIRES: x86-registered-target
 // REQUIRES: amdgpu-registered-target
 //
 // Test single gpu architecture with complete compilation.
 //
 // 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 \
 // RUN: | FileCheck -check-prefixes=BIN,RDC %s
 //
 // BIN-DAG: [[P0:[0-9]+]]: input, "{{.*}}hip-phases.hip", [[T:hip]], (host-[[T]])
 // BIN-DAG: [[P1:[0-9]+]]: preprocessor, {[[P0]]}, [[T]]-cpp-output, (host-[[T]])
 // BIN-DAG: [[P2:[0-9]+]]: compiler, {[[P1]]}, ir, (host-[[T]])

 // BIN-DAG: [[P9:[0-9]+]]: offload, "device-[[T]] (amdgcn-amd-amdhsa:[[ARCH]])" {[[P8]]}, image
 // NRD-DAG: [[P10:[0-9]+]]: linker, {[[P9]]}, hip-fatbin, (device-[[T]])
 // RDC-DAG: [[P10:[0-9]+]]: linker, {[[P9]]}, object, (device-[[T]])
 
 // NRD-DAG: [[P11:[0-9]+]]: offload, "host-[[T]] (x86_64-unknown-linux-gnu)" {[[P2]]}, "device-[[T]] (amdgcn-amd-amdhsa)" {[[P10]]}, ir
 // 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]])
 
 //
 // Test single gpu architecture up to the assemble phase.
 //
 // RUN: %clang -x hip -target x86_64-unknown-linux-gnu -ccc-print-phases \
 // RUN: --cuda-gpu-arch=gfx803 %s -S 2>&1 \
 // RUN: | FileCheck -check-prefixes=ASM %s

clang/test/Driver/hip-toolchain-no-rdc.hip

 // 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 -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: %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 \
 // RUN:   --hip-link --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 \
 // RUN:   -fuse-ld=lld -nogpuinc \
 // RUN:   %T/a.o %T/b.o \
 // RUN: 2>&1 | FileCheck -check-prefixes=LKONLY %s