-
Notifications
You must be signed in to change notification settings - Fork 10.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[OpenMP] Change __tgt_device_image
to point to the image
#77003
Conversation
Summary: We use the OffloadBinary to contain bundled offloading objects used to support many images / targets at the same time. The `__tgt_device_info` struct used to contain a pointer to this underlying binary format, which contains information about the triple and architecture. We used to parse this in the runtime to do image verification. Recent changes removed the need for this to be used internally, as we just parse it out of the ELF directly. This patch sets the pointers up so they point to the ELF without requiring any further parsing.
@llvm/pr-subscribers-clang-driver @llvm/pr-subscribers-clang Author: Joseph Huber (jhuber6) ChangesSummary: Recent changes removed the need for this to be used internally, as we Full diff: https://github.com/llvm/llvm-project/pull/77003.diff 2 Files Affected:
diff --git a/clang/test/Driver/linker-wrapper-image.c b/clang/test/Driver/linker-wrapper-image.c
index a2a1996f664309..40dde2e0291800 100644
--- a/clang/test/Driver/linker-wrapper-image.c
+++ b/clang/test/Driver/linker-wrapper-image.c
@@ -19,8 +19,8 @@
// 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: @.omp_offloading.device_image = internal unnamed_addr constant [[[SIZE:[0-9]+]] x i8] c"\10\FF\10\AD{{.*}}"
-// OPENMP-NEXT: @.omp_offloading.device_images = internal unnamed_addr constant [1 x %__tgt_device_image] [%__tgt_device_image { ptr @.omp_offloading.device_image, ptr getelementptr inbounds ([[[SIZE]] x i8], ptr @.omp_offloading.device_image, i64 1, i64 0), ptr @__start_omp_offloading_entries, ptr @__stop_omp_offloading_entries }]
+// OPENMP-NEXT: @.omp_offloading.device_image = internal unnamed_addr constant [[[SIZE:[0-9]+]] x i8] c"\10\FF\10\AD\01{{.*}}", 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 }]
// OPENMP-NEXT: @.omp_offloading.descriptor = internal constant %__tgt_bin_desc { i32 1, ptr @.omp_offloading.device_images, ptr @__start_omp_offloading_entries, ptr @__stop_omp_offloading_entries }
// OPENMP-NEXT: @llvm.global_ctors = appending global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @.omp_offloading.descriptor_reg, ptr null }]
// OPENMP-NEXT: @llvm.global_dtors = appending global [1 x { i32, ptr, ptr }] [{ i32, ptr, ptr } { i32 1, ptr @.omp_offloading.descriptor_unreg, ptr null }]
diff --git a/clang/tools/clang-linker-wrapper/OffloadWrapper.cpp b/clang/tools/clang-linker-wrapper/OffloadWrapper.cpp
index f4f500b173572d..161374ae555233 100644
--- a/clang/tools/clang-linker-wrapper/OffloadWrapper.cpp
+++ b/clang/tools/clang-linker-wrapper/OffloadWrapper.cpp
@@ -8,6 +8,7 @@
#include "OffloadWrapper.h"
#include "llvm/ADT/ArrayRef.h"
+#include "llvm/BinaryFormat/Magic.h"
#include "llvm/Frontend/Offloading/Utility.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/GlobalVariable.h"
@@ -121,19 +122,36 @@ GlobalVariable *createBinDesc(Module &M, ArrayRef<ArrayRef<char>> Bufs) {
SmallVector<Constant *, 4u> ImagesInits;
ImagesInits.reserve(Bufs.size());
for (ArrayRef<char> Buf : Bufs) {
+ // We embed the full offloading entry so the binary utilities can parse it.
auto *Data = ConstantDataArray::get(C, Buf);
- auto *Image = new GlobalVariable(M, Data->getType(), /*isConstant*/ true,
+ auto *Image = new GlobalVariable(M, Data->getType(), /*isConstant=*/true,
GlobalVariable::InternalLinkage, Data,
".omp_offloading.device_image");
Image->setUnnamedAddr(GlobalValue::UnnamedAddr::Global);
Image->setSection(".llvm.offloading");
Image->setAlignment(Align(object::OffloadBinary::getAlignment()));
- auto *Size = ConstantInt::get(getSizeTTy(M), Buf.size());
+ StringRef Binary(Buf.data(), Buf.size());
+ assert(identify_magic(Binary) == file_magic::offload_binary &&
+ "Invalid binary format");
+
+ // The device image struct contains the pointer to the beginning and end of
+ // the image stored inside of the offload binary. There should only be one
+ // of these for each buffer so we parse it out manually.
+ const auto *Header =
+ reinterpret_cast<const object::OffloadBinary::Header *>(
+ Binary.bytes_begin());
+ const auto *Entry = reinterpret_cast<const object::OffloadBinary::Entry *>(
+ Binary.bytes_begin() + Header->EntryOffset);
+
+ auto *Begin = ConstantInt::get(getSizeTTy(M), Entry->ImageOffset);
+ auto *Size =
+ ConstantInt::get(getSizeTTy(M), Entry->ImageOffset + Entry->ImageSize);
+ Constant *ZeroBegin[] = {Zero, Begin};
Constant *ZeroSize[] = {Zero, Size};
auto *ImageB =
- ConstantExpr::getGetElementPtr(Image->getValueType(), Image, ZeroZero);
+ ConstantExpr::getGetElementPtr(Image->getValueType(), Image, ZeroBegin);
auto *ImageE =
ConstantExpr::getGetElementPtr(Image->getValueType(), Image, ZeroSize);
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LG
@@ -19,8 +19,8 @@ | |||
// 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: @.omp_offloading.device_image = internal unnamed_addr constant [[[SIZE:[0-9]+]] x i8] c"\10\FF\10\AD{{.*}}" | |||
// OPENMP-NEXT: @.omp_offloading.device_images = internal unnamed_addr constant [1 x %__tgt_device_image] [%__tgt_device_image { ptr @.omp_offloading.device_image, ptr getelementptr inbounds ([[[SIZE]] x i8], ptr @.omp_offloading.device_image, i64 1, i64 0), ptr @__start_omp_offloading_entries, ptr @__stop_omp_offloading_entries }] | |||
// OPENMP-NEXT: @.omp_offloading.device_image = internal unnamed_addr constant [[[SIZE:[0-9]+]] x i8] c"\10\FF\10\AD\01{{.*}}", section ".llvm.offloading", align 8 |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This test now fails on big-endian architectures (where the \01
) will not be at the start.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Whoops, that was an accident to not cut it off after the magic bytes. I'll fix it, thanks for bringing this to my attention.
Summary: We use the OffloadBinary to contain bundled offloading objects used to support many images / targets at the same time. The `__tgt_device_info` struct used to contain a pointer to this underlying binary format, which contains information about the triple and architecture. We used to parse this in the runtime to do image verification. Recent changes removed the need for this to be used internally, as we just parse it out of the ELF directly. This patch sets the pointers up so they point to the ELF without requiring any further parsing.
Summary:
We use the OffloadBinary to contain bundled offloading objects used to
support many images / targets at the same time. The
__tgt_device_info
struct used to contain a pointer to this underlying binary format, which
contains information about the triple and architecture. We used to parse
this in the runtime to do image verification.
Recent changes removed the need for this to be used internally, as we
just parse it out of the ELF directly. This patch sets the pointers up
so they point to the ELF without requiring any further parsing.