maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/Driver/hip-toolchain-rdc-separate.hip

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

153 lines
7.1 KiB
Plaintext
Raw Normal View History

[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// REQUIRES: x86-registered-target
// REQUIRES: amdgpu-registered-target
// RUN: %clang -c -### -target x86_64-linux-gnu \
// 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 -fgpu-rdc -nogpuinc \
// RUN: %S/Inputs/hip_multiple_inputs/a.cu \
// RUN: %S/Inputs/hip_multiple_inputs/b.hip \
// RUN: 2>&1 | FileCheck %s
[clang][amdgpu] Use implicit code object version [clang][amdgpu] Use implicit code object version At present, clang always passes amdhsa-code-object-version on to -cc1. That is great for certainty over what object version is being used when debugging. Unfortunately, the command line argument is in AMDGPUBaseInfo.cpp in the amdgpu target. If clang is used with an llvm compiled with DLLVM_TARGETS_TO_BUILD that excludes amdgpu, this will be diagnosed (as discovered via D98658): - Unknown command line argument '--amdhsa-code-object-version=4' This means that clang, built only for X86, can be used to compile the nvptx devicertl for openmp but not the amdgpu one. That would shortly spawn fragile logic in the devicertl cmake to try to guess whether the clang used will work. This change omits the amdhsa-code-object-version parameter when it matches the default that AMDGPUBaseInfo.cpp specifies, with a comment to indicate why. As this is the only part of clang's codegen for amdgpu that depends on the target in the back end it suffices to build the openmp runtime on most (all?) systems. It is a non-functional change, though observable in the updated tests and when compiling with -###. It may cause minor disruption to the amd-stg-open branch. Revision of D98746, builds on refactor in D101077 Reviewed By: yaxunl Differential Revision: https://reviews.llvm.org/D101095
2021-04-24 06:52:49 +08:00
// CHECK: [[CLANG:".*clang.*"]] "-cc1" "-triple" "amdgcn-amd-amdhsa"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: "-aux-triple" "x86_64-unknown-linux-gnu"
// CHECK-SAME: "-emit-llvm-bc"
// CHECK-SAME: {{.*}} "-main-file-name" "a.cu"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fcuda-is-device" "-fcuda-allow-variadic-functions" "-fvisibility" "hidden"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: "-fapply-global-visibility-to-externs"
// CHECK-SAME: "{{.*}}lib1.bc" "{{.*}}lib2.bc"
// CHECK-SAME: "-target-cpu" "gfx803"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fgpu-rdc"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: {{.*}} "-o" "[[A_BC1:.*bc]]" "-x" "hip"
// CHECK-SAME: {{.*}} [[A_SRC:".*a.cu"]]
[clang][amdgpu] Use implicit code object version [clang][amdgpu] Use implicit code object version At present, clang always passes amdhsa-code-object-version on to -cc1. That is great for certainty over what object version is being used when debugging. Unfortunately, the command line argument is in AMDGPUBaseInfo.cpp in the amdgpu target. If clang is used with an llvm compiled with DLLVM_TARGETS_TO_BUILD that excludes amdgpu, this will be diagnosed (as discovered via D98658): - Unknown command line argument '--amdhsa-code-object-version=4' This means that clang, built only for X86, can be used to compile the nvptx devicertl for openmp but not the amdgpu one. That would shortly spawn fragile logic in the devicertl cmake to try to guess whether the clang used will work. This change omits the amdhsa-code-object-version parameter when it matches the default that AMDGPUBaseInfo.cpp specifies, with a comment to indicate why. As this is the only part of clang's codegen for amdgpu that depends on the target in the back end it suffices to build the openmp runtime on most (all?) systems. It is a non-functional change, though observable in the updated tests and when compiling with -###. It may cause minor disruption to the amd-stg-open branch. Revision of D98746, builds on refactor in D101077 Reviewed By: yaxunl Differential Revision: https://reviews.llvm.org/D101095
2021-04-24 06:52:49 +08:00
// CHECK: [[CLANG]] "-cc1" "-triple" "amdgcn-amd-amdhsa"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: "-aux-triple" "x86_64-unknown-linux-gnu"
// CHECK-SAME: "-emit-llvm-bc"
// CHECK-SAME: {{.*}} "-main-file-name" "a.cu"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fcuda-is-device"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: "{{.*}}lib1.bc" "{{.*}}lib2.bc"
// CHECK-SAME: "-target-cpu" "gfx900"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fgpu-rdc"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: {{.*}} "-o" "[[A_BC2:.*bc]]" "-x" "hip"
// CHECK-SAME: {{.*}} [[A_SRC]]
// CHECK: [[CLANG]] "-cc1" "-triple" "x86_64-unknown-linux-gnu"
// CHECK-SAME: "-aux-triple" "amdgcn-amd-amdhsa"
// CHECK-SAME: "-emit-obj"
// CHECK-SAME: {{.*}} "-main-file-name" "a.cu"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fgpu-rdc"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: {{.*}} "-o" "[[A_OBJ_HOST:.*o]]" "-x" "hip"
// CHECK-SAME: {{.*}} [[A_SRC]]
// CHECK: [[BUNDLER:".*clang-offload-bundler"]] "-type=o"
[clang-offload-bundler] Make Bundle Entry ID backward compatible Earlier BundleEntryID used to be <OffloadKind>-<Triple>-<GPUArch>. This used to work because the clang-offload-bundler didn't need GPUArch explicitly for any bundling/unbundling action. With unbundleArchive it needs GPUArch to ensure compatibility between device specific code objects. D93525 enforced triples to have separators for all 4 components irrespective of number of components, like "amdgcn-amd-amdhsa--". It was required to to correctly parse a possible 4th environment component or a GPU. But, this condition is breaking backward compatibility with archive libraries compiled with compilers older than D93525. This patch allows triples to have any number of components with and without extra separator for empty environment field. Thus, both the following bundle entry IDs are same: openmp-amdgcn-amd-amdhsa--gfx906 openmp-amdgcn-amd-amdhsa-gfx906 Reviewed By: yaxunl, grokos Differential Revision: https://reviews.llvm.org/D106809
2021-07-27 00:39:41 +08:00
// CHECK-SAME: "-targets=hip-amdgcn-amd-amdhsa-gfx803,hip-amdgcn-amd-amdhsa-gfx900,host-x86_64-unknown-linux-gnu"
[clang-offload-bundler] add -input/-output options Currently, clang-offload-bundler has -inputs and -outputs options that accept values with comma as the delimiter. This causes issues with file paths containing commas, which are valid file paths on Linux. This add two new options -input and -output, which accept one single file, and allow multiple instances. This allows arbitrary file paths. The old -inputs and -outputs options will be kept for backward compatibility, but are not allowed to be used with -input and -output options for simplicity. In the future, -inputs and -outputs options will be phasing out. RFC: https://discourse.llvm.org/t/rfc-adding-input-and-output-options-to-clang-offload-bundler/60049 Patch by: Siu Chi Chan Reviewed by: Yaxun Liu Differential Revision: https://reviews.llvm.org/D120662
2022-04-05 22:44:13 +08:00
// CHECK-SAME: "-output=[[A_O:.*a.o]]" "-input=[[A_BC1]]" "-input=[[A_BC2]]" "-input=[[A_OBJ_HOST]]"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
[clang][amdgpu] Use implicit code object version [clang][amdgpu] Use implicit code object version At present, clang always passes amdhsa-code-object-version on to -cc1. That is great for certainty over what object version is being used when debugging. Unfortunately, the command line argument is in AMDGPUBaseInfo.cpp in the amdgpu target. If clang is used with an llvm compiled with DLLVM_TARGETS_TO_BUILD that excludes amdgpu, this will be diagnosed (as discovered via D98658): - Unknown command line argument '--amdhsa-code-object-version=4' This means that clang, built only for X86, can be used to compile the nvptx devicertl for openmp but not the amdgpu one. That would shortly spawn fragile logic in the devicertl cmake to try to guess whether the clang used will work. This change omits the amdhsa-code-object-version parameter when it matches the default that AMDGPUBaseInfo.cpp specifies, with a comment to indicate why. As this is the only part of clang's codegen for amdgpu that depends on the target in the back end it suffices to build the openmp runtime on most (all?) systems. It is a non-functional change, though observable in the updated tests and when compiling with -###. It may cause minor disruption to the amd-stg-open branch. Revision of D98746, builds on refactor in D101077 Reviewed By: yaxunl Differential Revision: https://reviews.llvm.org/D101095
2021-04-24 06:52:49 +08:00
// CHECK: [[CLANG]] "-cc1" "-triple" "amdgcn-amd-amdhsa"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: "-aux-triple" "x86_64-unknown-linux-gnu"
// CHECK-SAME: "-emit-llvm-bc"
// CHECK-SAME: {{.*}} "-main-file-name" "b.hip"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fcuda-is-device" "-fcuda-allow-variadic-functions" "-fvisibility" "hidden"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: "-fapply-global-visibility-to-externs"
// CHECK-SAME: "{{.*}}lib1.bc" "{{.*}}lib2.bc"
// CHECK-SAME: "-target-cpu" "gfx803"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fgpu-rdc"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: {{.*}} "-o" "[[B_BC1:.*bc]]" "-x" "hip"
// CHECK-SAME: {{.*}} [[B_SRC:".*b.hip"]]
[clang][amdgpu] Use implicit code object version [clang][amdgpu] Use implicit code object version At present, clang always passes amdhsa-code-object-version on to -cc1. That is great for certainty over what object version is being used when debugging. Unfortunately, the command line argument is in AMDGPUBaseInfo.cpp in the amdgpu target. If clang is used with an llvm compiled with DLLVM_TARGETS_TO_BUILD that excludes amdgpu, this will be diagnosed (as discovered via D98658): - Unknown command line argument '--amdhsa-code-object-version=4' This means that clang, built only for X86, can be used to compile the nvptx devicertl for openmp but not the amdgpu one. That would shortly spawn fragile logic in the devicertl cmake to try to guess whether the clang used will work. This change omits the amdhsa-code-object-version parameter when it matches the default that AMDGPUBaseInfo.cpp specifies, with a comment to indicate why. As this is the only part of clang's codegen for amdgpu that depends on the target in the back end it suffices to build the openmp runtime on most (all?) systems. It is a non-functional change, though observable in the updated tests and when compiling with -###. It may cause minor disruption to the amd-stg-open branch. Revision of D98746, builds on refactor in D101077 Reviewed By: yaxunl Differential Revision: https://reviews.llvm.org/D101095
2021-04-24 06:52:49 +08:00
// CHECK: [[CLANG]] "-cc1" "-triple" "amdgcn-amd-amdhsa"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: "-aux-triple" "x86_64-unknown-linux-gnu"
// CHECK-SAME: "-emit-llvm-bc"
// CHECK-SAME: {{.*}} "-main-file-name" "b.hip"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fcuda-is-device"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: "{{.*}}lib1.bc" "{{.*}}lib2.bc"
// CHECK-SAME: "-target-cpu" "gfx900"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fgpu-rdc"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: {{.*}} "-o" "[[B_BC2:.*bc]]" "-x" "hip"
// CHECK-SAME: {{.*}} [[B_SRC]]
// CHECK: [[CLANG]] "-cc1" "-triple" "x86_64-unknown-linux-gnu"
// CHECK-SAME: "-aux-triple" "amdgcn-amd-amdhsa"
// CHECK-SAME: "-emit-obj"
// CHECK-SAME: {{.*}} "-main-file-name" "b.hip"
[CUDA][HIP] Pass -fgpu-rdc to host clang -cc1 Currently -fgpu-rdc is not passed to host clang -cc1. This causes issue because -fgpu-rdc affects shadow variable linkage in host compilation. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D96105
2021-02-09 07:51:31 +08:00
// CHECK-SAME: "-fgpu-rdc"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// CHECK-SAME: {{.*}} "-o" "[[B_OBJ_HOST:.*o]]" "-x" "hip"
// CHECK-SAME: {{.*}} [[B_SRC]]
// CHECK: [[BUNDLER:".*clang-offload-bundler"]] "-type=o"
[clang-offload-bundler] Make Bundle Entry ID backward compatible Earlier BundleEntryID used to be <OffloadKind>-<Triple>-<GPUArch>. This used to work because the clang-offload-bundler didn't need GPUArch explicitly for any bundling/unbundling action. With unbundleArchive it needs GPUArch to ensure compatibility between device specific code objects. D93525 enforced triples to have separators for all 4 components irrespective of number of components, like "amdgcn-amd-amdhsa--". It was required to to correctly parse a possible 4th environment component or a GPU. But, this condition is breaking backward compatibility with archive libraries compiled with compilers older than D93525. This patch allows triples to have any number of components with and without extra separator for empty environment field. Thus, both the following bundle entry IDs are same: openmp-amdgcn-amd-amdhsa--gfx906 openmp-amdgcn-amd-amdhsa-gfx906 Reviewed By: yaxunl, grokos Differential Revision: https://reviews.llvm.org/D106809
2021-07-27 00:39:41 +08:00
// CHECK-SAME: "-targets=hip-amdgcn-amd-amdhsa-gfx803,hip-amdgcn-amd-amdhsa-gfx900,host-x86_64-unknown-linux-gnu"
[clang-offload-bundler] add -input/-output options Currently, clang-offload-bundler has -inputs and -outputs options that accept values with comma as the delimiter. This causes issues with file paths containing commas, which are valid file paths on Linux. This add two new options -input and -output, which accept one single file, and allow multiple instances. This allows arbitrary file paths. The old -inputs and -outputs options will be kept for backward compatibility, but are not allowed to be used with -input and -output options for simplicity. In the future, -inputs and -outputs options will be phasing out. RFC: https://discourse.llvm.org/t/rfc-adding-input-and-output-options-to-clang-offload-bundler/60049 Patch by: Siu Chi Chan Reviewed by: Yaxun Liu Differential Revision: https://reviews.llvm.org/D120662
2022-04-05 22:44:13 +08:00
// CHECK-SAME: "-output=[[B_O:.*b.o]]" "-input=[[B_BC1]]" "-input=[[B_BC2]]" "-input=[[B_OBJ_HOST]]"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// RUN: touch %T/a.o
// RUN: touch %T/b.o
// RUN: %clang --hip-link -### -target x86_64-linux-gnu \
// RUN: --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 \
// RUN: -fuse-ld=lld -fgpu-rdc -nogpuinc \
// RUN: %T/a.o %T/b.o \
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// RUN: 2>&1 | FileCheck -check-prefixes=LINK,LINK-HOST-UNBUNDLE,LLD-TMP,LINK-BUNDLE,LINK-EMBED %s
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// RUN: %clang --hip-link -### -target x86_64-linux-gnu \
// RUN: --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 \
// RUN: -fuse-ld=lld -fgpu-rdc -nogpuinc \
// RUN: %T/a.o %T/b.o --cuda-device-only \
// RUN: 2>&1 | FileCheck -check-prefixes=LINK,LLD-TMP,LINK-BUNDLE,LINK-NOEMBED %s
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// RUN: %clang --hip-link -### -target x86_64-linux-gnu \
// RUN: --cuda-gpu-arch=gfx803 --cuda-gpu-arch=gfx900 \
// RUN: -fuse-ld=lld -fgpu-rdc -nogpuinc \
// RUN: %T/a.o %T/b.o --cuda-device-only --no-gpu-bundle-output \
// RUN: 2>&1 | FileCheck -check-prefixes=LINK,LLD-FIN,LINK-NOBUNDLE,LINK-NOEMBED %s
// LINK-HOST-UNBUNDLE: [[BUNDLER:".*clang-offload-bundler"]] "-type=o"
// LINK-HOST-UNBUNDLE-SAME: "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx803,hip-amdgcn-amd-amdhsa-gfx900"
[clang-offload-bundler] add -input/-output options Currently, clang-offload-bundler has -inputs and -outputs options that accept values with comma as the delimiter. This causes issues with file paths containing commas, which are valid file paths on Linux. This add two new options -input and -output, which accept one single file, and allow multiple instances. This allows arbitrary file paths. The old -inputs and -outputs options will be kept for backward compatibility, but are not allowed to be used with -input and -output options for simplicity. In the future, -inputs and -outputs options will be phasing out. RFC: https://discourse.llvm.org/t/rfc-adding-input-and-output-options-to-clang-offload-bundler/60049 Patch by: Siu Chi Chan Reviewed by: Yaxun Liu Differential Revision: https://reviews.llvm.org/D120662
2022-04-05 22:44:13 +08:00
// LINK-HOST-UNBUNDLE-SAME: "-input=[[A_O:.*a.o]]" "-output=[[A_OBJ_HOST:.*o]]" "-output={{.*o}}" "-output={{.*o}}"
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// LINK-HOST-UNBUNDLE: "-unbundle" "-allow-missing-bundles"
// LINK-HOST-UNBUNDLE: [[BUNDLER:".*clang-offload-bundler"]] "-type=o"
// LINK-HOST-UNBUNDLE-SAME: "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx803,hip-amdgcn-amd-amdhsa-gfx900"
[clang-offload-bundler] add -input/-output options Currently, clang-offload-bundler has -inputs and -outputs options that accept values with comma as the delimiter. This causes issues with file paths containing commas, which are valid file paths on Linux. This add two new options -input and -output, which accept one single file, and allow multiple instances. This allows arbitrary file paths. The old -inputs and -outputs options will be kept for backward compatibility, but are not allowed to be used with -input and -output options for simplicity. In the future, -inputs and -outputs options will be phasing out. RFC: https://discourse.llvm.org/t/rfc-adding-input-and-output-options-to-clang-offload-bundler/60049 Patch by: Siu Chi Chan Reviewed by: Yaxun Liu Differential Revision: https://reviews.llvm.org/D120662
2022-04-05 22:44:13 +08:00
// LINK-HOST-UNBUNDLE-SAME: "-input=[[B_O:.*b.o]]" "-output=[[B_OBJ_HOST:.*o]]" "-output={{.*o}}" "-output={{.*o}}"
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// LINK-HOST-UNBUNDLE: "-unbundle" "-allow-missing-bundles"
[HIP] Move HIP Linking Logic into HIP ToolChain This patch is a follow up on https://reviews.llvm.org/D78759. Extract the HIP Linker script from generic GNU linker, and move it into HIP ToolChain. Update OffloadActionBuilder Link actions feature to apply device linking and host linking actions separately. Using MC Directives, embed the device images and define symbols. Reviewers: JonChesterfield, yaxunl Subscribers: tra, echristo, jdoerfert, msearles, scchan Differential Revision: https://reviews.llvm.org/D81963
2020-06-12 02:36:10 +08:00
// LINK: [[BUNDLER:".*clang-offload-bundler"]] "-type=o"
[clang-offload-bundler] Make Bundle Entry ID backward compatible Earlier BundleEntryID used to be <OffloadKind>-<Triple>-<GPUArch>. This used to work because the clang-offload-bundler didn't need GPUArch explicitly for any bundling/unbundling action. With unbundleArchive it needs GPUArch to ensure compatibility between device specific code objects. D93525 enforced triples to have separators for all 4 components irrespective of number of components, like "amdgcn-amd-amdhsa--". It was required to to correctly parse a possible 4th environment component or a GPU. But, this condition is breaking backward compatibility with archive libraries compiled with compilers older than D93525. This patch allows triples to have any number of components with and without extra separator for empty environment field. Thus, both the following bundle entry IDs are same: openmp-amdgcn-amd-amdhsa--gfx906 openmp-amdgcn-amd-amdhsa-gfx906 Reviewed By: yaxunl, grokos Differential Revision: https://reviews.llvm.org/D106809
2021-07-27 00:39:41 +08:00
// LINK-SAME: "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx803,hip-amdgcn-amd-amdhsa-gfx900"
[clang-offload-bundler] add -input/-output options Currently, clang-offload-bundler has -inputs and -outputs options that accept values with comma as the delimiter. This causes issues with file paths containing commas, which are valid file paths on Linux. This add two new options -input and -output, which accept one single file, and allow multiple instances. This allows arbitrary file paths. The old -inputs and -outputs options will be kept for backward compatibility, but are not allowed to be used with -input and -output options for simplicity. In the future, -inputs and -outputs options will be phasing out. RFC: https://discourse.llvm.org/t/rfc-adding-input-and-output-options-to-clang-offload-bundler/60049 Patch by: Siu Chi Chan Reviewed by: Yaxun Liu Differential Revision: https://reviews.llvm.org/D120662
2022-04-05 22:44:13 +08:00
// LINK-SAME: "-input=[[A_O:.*a.o]]" "-output={{.*o}}" "-output=[[A_BC1:.*o]]" "-output=[[A_BC2:.*o]]"
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// LINK-SAME: "-unbundle" "-allow-missing-bundles"
[HIP] Move HIP Linking Logic into HIP ToolChain This patch is a follow up on https://reviews.llvm.org/D78759. Extract the HIP Linker script from generic GNU linker, and move it into HIP ToolChain. Update OffloadActionBuilder Link actions feature to apply device linking and host linking actions separately. Using MC Directives, embed the device images and define symbols. Reviewers: JonChesterfield, yaxunl Subscribers: tra, echristo, jdoerfert, msearles, scchan Differential Revision: https://reviews.llvm.org/D81963
2020-06-12 02:36:10 +08:00
// LINK: [[BUNDLER:".*clang-offload-bundler"]] "-type=o"
[clang-offload-bundler] Make Bundle Entry ID backward compatible Earlier BundleEntryID used to be <OffloadKind>-<Triple>-<GPUArch>. This used to work because the clang-offload-bundler didn't need GPUArch explicitly for any bundling/unbundling action. With unbundleArchive it needs GPUArch to ensure compatibility between device specific code objects. D93525 enforced triples to have separators for all 4 components irrespective of number of components, like "amdgcn-amd-amdhsa--". It was required to to correctly parse a possible 4th environment component or a GPU. But, this condition is breaking backward compatibility with archive libraries compiled with compilers older than D93525. This patch allows triples to have any number of components with and without extra separator for empty environment field. Thus, both the following bundle entry IDs are same: openmp-amdgcn-amd-amdhsa--gfx906 openmp-amdgcn-amd-amdhsa-gfx906 Reviewed By: yaxunl, grokos Differential Revision: https://reviews.llvm.org/D106809
2021-07-27 00:39:41 +08:00
// LINK-SAME: "-targets=host-x86_64-unknown-linux-gnu,hip-amdgcn-amd-amdhsa-gfx803,hip-amdgcn-amd-amdhsa-gfx900"
[clang-offload-bundler] add -input/-output options Currently, clang-offload-bundler has -inputs and -outputs options that accept values with comma as the delimiter. This causes issues with file paths containing commas, which are valid file paths on Linux. This add two new options -input and -output, which accept one single file, and allow multiple instances. This allows arbitrary file paths. The old -inputs and -outputs options will be kept for backward compatibility, but are not allowed to be used with -input and -output options for simplicity. In the future, -inputs and -outputs options will be phasing out. RFC: https://discourse.llvm.org/t/rfc-adding-input-and-output-options-to-clang-offload-bundler/60049 Patch by: Siu Chi Chan Reviewed by: Yaxun Liu Differential Revision: https://reviews.llvm.org/D120662
2022-04-05 22:44:13 +08:00
// LINK-SAME: "-input=[[B_O:.*b.o]]" "-output={{.*o}}" "-output=[[B_BC1:.*o]]" "-output=[[B_BC2:.*o]]"
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// LINK-SAME: "-unbundle" "-allow-missing-bundles"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// LINK-NOT: "*.llvm-link"
// LINK-NOT: ".*opt"
// LINK-NOT: ".*llc"
[HIP] Add missing options for lto Add -mcpu, -mattr, -mllvm, and -save-temps options for lto when necessary. Differential Revision: https://reviews.llvm.org/D82506
2020-06-25 05:34:21 +08:00
// LINK: {{".*lld.*"}} {{.*}} "-plugin-opt=-amdgpu-internalize-symbols"
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// LINK-SAME: "-plugin-opt=mcpu=gfx803"
// LLD-TMP-SAME: "-o" "[[IMG_DEV1:.*.out]]"
// LLD-FIN-SAME: "-o" "[[IMG_DEV1:a.out-.*gfx803]]"
// LINK-SAME "[[A_BC1]]" "[[B_BC1]]"
[HIP] Do not use llvm-link/opt/llc for -fgpu-rdc This patch is a follow up on https://reviews.llvm.org/D81627. In addition to default -fno-gpu-rdc case, this patches let HIP toolchain not use llvm-link/opt/llc to link device code for -fgpu-rdc case. Instead, uses standard lto. This will eliminate some redundant optimizations and speed up the compilation/linking. Differential Revision: https://reviews.llvm.org/D81861
2020-06-14 21:07:31 +08:00
// LINK-NOT: "*.llvm-link"
// LINK-NOT: ".*opt"
// LINK-NOT: ".*llc"
[HIP] Add missing options for lto Add -mcpu, -mattr, -mllvm, and -save-temps options for lto when necessary. Differential Revision: https://reviews.llvm.org/D82506
2020-06-25 05:34:21 +08:00
// LINK: {{".*lld.*"}} {{.*}} "-plugin-opt=-amdgpu-internalize-symbols"
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// LINK-SAME: "-plugin-opt=mcpu=gfx900"
// LLD-TMP-SAME: "-o" "[[IMG_DEV2:.*.out]]"
// LLD-FIN-SAME: "-o" "[[IMG_DEV1:a.out-.*gfx900]]"
// LINK-SAME "[[A_BC2]]" "[[B_BC2]]"
// LINK-BUNDLE: [[BUNDLER:".*clang-offload-bundler"]] "-type=o"
// LINK-BUNDLE-SAME: "-targets={{.*}},hipv4-amdgcn-amd-amdhsa--gfx803,hipv4-amdgcn-amd-amdhsa--gfx900"
[clang-offload-bundler] add -input/-output options Currently, clang-offload-bundler has -inputs and -outputs options that accept values with comma as the delimiter. This causes issues with file paths containing commas, which are valid file paths on Linux. This add two new options -input and -output, which accept one single file, and allow multiple instances. This allows arbitrary file paths. The old -inputs and -outputs options will be kept for backward compatibility, but are not allowed to be used with -input and -output options for simplicity. In the future, -inputs and -outputs options will be phasing out. RFC: https://discourse.llvm.org/t/rfc-adding-input-and-output-options-to-clang-offload-bundler/60049 Patch by: Siu Chi Chan Reviewed by: Yaxun Liu Differential Revision: https://reviews.llvm.org/D120662
2022-04-05 22:44:13 +08:00
// LINK-BUNDLE-SAME: "-input={{.*}}" "-input=[[IMG_DEV1]]" "-input=[[IMG_DEV2]]" "-output=[[BUNDLE:.*]]"
[HIP] Fix device only linking for -fgpu-rdc Currently when -fgpu-rdc is specified, HIP toolchain always does host linking even if --cuda-device-only is specified. This patch fixes that. Only device linking is performed when --cuda-device-only is specified. Reviewed by: Artem Belevich Differential Revision: https://reviews.llvm.org/D116840
2022-01-08 02:33:00 +08:00
// LINK-NOBUNDLE-NOT: {{".*clang-offload-bundler"}} "-type=o"
// LINK-EMBED: {{".*llvm-mc.*"}} "-o" "[[OBJBUNDLE:.*o]]" "{{.*}}.mcin" "--filetype=obj"
// LINK-NOEMBED-NOT: {{".*llvm-mc.*"}} "-o"
// LINK-EMBED: [[LD:".*ld.*"]] {{.*}} "-o" "a.out" {{.*}} "[[A_OBJ_HOST]]"
// LINK-EMBED-SAME: "[[B_OBJ_HOST]]" "[[OBJBUNDLE]]"
// LINK-NOEMBED-NOT: {{".*ld.*"}} {{.*}} "-o" "a.out"