[clang] a551703 - [Offload] Fix the offloading wrapper when merged multiple times. (#79231)
via cfe-commits
cfe-commits at lists.llvm.org
Wed Jan 24 11:50:40 PST 2024
Author: Joseph Huber
Date: 2024-01-24T13:50:35-06:00
New Revision: a551703cb5d171d24303de9d20bfdedfe46d49cf
URL: https://github.com/llvm/llvm-project/commit/a551703cb5d171d24303de9d20bfdedfe46d49cf
DIFF: https://github.com/llvm/llvm-project/commit/a551703cb5d171d24303de9d20bfdedfe46d49cf.diff
LOG: [Offload] Fix the offloading wrapper when merged multiple times. (#79231)
Summary:
The offloading wrapper is a object file that contains code necessary to
register offloading entries for the given runtime. Currently, we
expected only one of these to be present when we make the final
executable. However, in the case of redistributable linking with `-r` we
can end up with multiple of these being generated before finally
creating the executable.
This patch simply changes the defintiions of these globals to be
mergable. This allows multiples of these to participate in a single link
job. For ELF, we just make the dummy variable internal and used so it
sets up the section as expected. For COFF we make the entries weak_odr
so they merge to a single symbol
Added:
Modified:
clang/test/Driver/linker-wrapper-image.c
llvm/lib/Frontend/Offloading/Utility.cpp
Removed:
################################################################################
diff --git a/clang/test/Driver/linker-wrapper-image.c b/clang/test/Driver/linker-wrapper-image.c
index fa2e59e36b3828a..b5d8ae217a9723d 100644
--- a/clang/test/Driver/linker-wrapper-image.c
+++ b/clang/test/Driver/linker-wrapper-image.c
@@ -14,10 +14,10 @@
// OPENMP-ELF: @__start_omp_offloading_entries = external hidden constant [0 x %struct.__tgt_offload_entry]
// OPENMP-ELF-NEXT: @__stop_omp_offloading_entries = external hidden constant [0 x %struct.__tgt_offload_entry]
-// OPENMP-ELF-NEXT: @__dummy.omp_offloading_entries = hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "omp_offloading_entries"
+// OPENMP-ELF-NEXT: @__dummy.omp_offloading_entries = internal constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "omp_offloading_entries"
-// OPENMP-COFF: @__start_omp_offloading_entries = hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "omp_offloading_entries$OA"
-// OPENMP-COFF-NEXT: @__stop_omp_offloading_entries = hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "omp_offloading_entries$OZ"
+// OPENMP-COFF: @__start_omp_offloading_entries = weak_odr hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "omp_offloading_entries$OA"
+// OPENMP-COFF-NEXT: @__stop_omp_offloading_entries = weak_odr hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "omp_offloading_entries$OZ"
// OPENMP: @.omp_offloading.device_image = internal unnamed_addr constant [[[SIZE:[0-9]+]] x i8] c"\10\FF\10\AD{{.*}}", section ".llvm.offloading", align 8
// OPENMP-NEXT: @.omp_offloading.device_images = internal unnamed_addr constant [1 x %__tgt_device_image] [%__tgt_device_image { ptr getelementptr inbounds ([[[BEGIN:[0-9]+]] x i8], ptr @.omp_offloading.device_image, i64 1, i64 0), ptr getelementptr inbounds ([[[END:[0-9]+]] x i8], ptr @.omp_offloading.device_image, i64 1, i64 0), ptr @__start_omp_offloading_entries, ptr @__stop_omp_offloading_entries }]
@@ -47,10 +47,10 @@
// CUDA-ELF: @__start_cuda_offloading_entries = external hidden constant [0 x %struct.__tgt_offload_entry]
// CUDA-ELF-NEXT: @__stop_cuda_offloading_entries = external hidden constant [0 x %struct.__tgt_offload_entry]
-// CUDA-ELF-NEXT: @__dummy.cuda_offloading_entries = hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "cuda_offloading_entries"
+// CUDA-ELF-NEXT: @__dummy.cuda_offloading_entries = internal constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "cuda_offloading_entries"
-// CUDA-COFF: @__start_cuda_offloading_entries = hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "cuda_offloading_entries$OA"
-// CUDA-COFF-NEXT: @__stop_cuda_offloading_entries = hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "cuda_offloading_entries$OZ"
+// CUDA-COFF: @__start_cuda_offloading_entries = weak_odr hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "cuda_offloading_entries$OA"
+// CUDA-COFF-NEXT: @__stop_cuda_offloading_entries = weak_odr hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "cuda_offloading_entries$OZ"
// CUDA: @.fatbin_image = internal constant [0 x i8] zeroinitializer, section ".nv_fatbin"
// CUDA-NEXT: @.fatbin_wrapper = internal constant %fatbin_wrapper { i32 1180844977, i32 1, ptr @.fatbin_image, ptr null }, section ".nvFatBinSegment", align 8
@@ -145,10 +145,10 @@
// HIP-ELF: @__start_hip_offloading_entries = external hidden constant [0 x %struct.__tgt_offload_entry]
// HIP-ELF-NEXT: @__stop_hip_offloading_entries = external hidden constant [0 x %struct.__tgt_offload_entry]
-// HIP-ELF-NEXT: @__dummy.hip_offloading_entries = hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "hip_offloading_entries"
+// HIP-ELF-NEXT: @__dummy.hip_offloading_entries = internal constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "hip_offloading_entries"
-// HIP-COFF: @__start_hip_offloading_entries = hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "hip_offloading_entries$OA"
-// HIP-COFF-NEXT: @__stop_hip_offloading_entries = hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "hip_offloading_entries$OZ"
+// HIP-COFF: @__start_hip_offloading_entries = weak_odr hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "hip_offloading_entries$OA"
+// HIP-COFF-NEXT: @__stop_hip_offloading_entries = weak_odr hidden constant [0 x %struct.__tgt_offload_entry] zeroinitializer, section "hip_offloading_entries$OZ"
// HIP: @.fatbin_image = internal constant [0 x i8] zeroinitializer, section ".hip_fatbin"
// HIP-NEXT: @.fatbin_wrapper = internal constant %fatbin_wrapper { i32 1212764230, i32 1, ptr @.fatbin_image, ptr null }, section ".hipFatBinSegment", align 8
diff --git a/llvm/lib/Frontend/Offloading/Utility.cpp b/llvm/lib/Frontend/Offloading/Utility.cpp
index 531919bccb94e3b..bd5b89b917706da 100644
--- a/llvm/lib/Frontend/Offloading/Utility.cpp
+++ b/llvm/lib/Frontend/Offloading/Utility.cpp
@@ -11,6 +11,7 @@
#include "llvm/IR/GlobalValue.h"
#include "llvm/IR/GlobalVariable.h"
#include "llvm/IR/Value.h"
+#include "llvm/Transforms/Utils/ModuleUtils.h"
using namespace llvm;
using namespace llvm::offloading;
@@ -86,14 +87,16 @@ offloading::getOffloadEntryArray(Module &M, StringRef SectionName) {
ConstantAggregateZero::get(ArrayType::get(getEntryTy(M), 0u));
auto *EntryInit = Triple.isOSBinFormatCOFF() ? ZeroInitilaizer : nullptr;
auto *EntryType = ArrayType::get(getEntryTy(M), 0);
+ auto Linkage = Triple.isOSBinFormatCOFF() ? GlobalValue::WeakODRLinkage
+ : GlobalValue::ExternalLinkage;
- auto *EntriesB = new GlobalVariable(M, EntryType, /*isConstant=*/true,
- GlobalValue::ExternalLinkage, EntryInit,
- "__start_" + SectionName);
+ auto *EntriesB =
+ new GlobalVariable(M, EntryType, /*isConstant=*/true, Linkage, EntryInit,
+ "__start_" + SectionName);
EntriesB->setVisibility(GlobalValue::HiddenVisibility);
- auto *EntriesE = new GlobalVariable(M, EntryType, /*isConstant=*/true,
- GlobalValue::ExternalLinkage, EntryInit,
- "__stop_" + SectionName);
+ auto *EntriesE =
+ new GlobalVariable(M, EntryType, /*isConstant=*/true, Linkage, EntryInit,
+ "__stop_" + SectionName);
EntriesE->setVisibility(GlobalValue::HiddenVisibility);
if (Triple.isOSBinFormatELF()) {
@@ -102,10 +105,10 @@ offloading::getOffloadEntryArray(Module &M, StringRef SectionName) {
// valid C-identifier is present. We define a dummy variable here to force
// the linker to always provide these symbols.
auto *DummyEntry = new GlobalVariable(
- M, ZeroInitilaizer->getType(), true, GlobalVariable::ExternalLinkage,
+ M, ZeroInitilaizer->getType(), true, GlobalVariable::InternalLinkage,
ZeroInitilaizer, "__dummy." + SectionName);
DummyEntry->setSection(SectionName);
- DummyEntry->setVisibility(GlobalValue::HiddenVisibility);
+ appendToCompilerUsed(M, DummyEntry);
} else {
// The COFF linker will merge sections containing a '$' together into a
// single section. The order of entries in this section will be sorted
More information about the cfe-commits
mailing list