1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2025-02-01 05:01:59 +01:00
llvm-mirror/lib/ExecutionEngine/Orc/ObjectLinkingLayer.cpp

653 lines
22 KiB
C++
Raw Normal View History

Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
//===------- ObjectLinkingLayer.cpp - JITLink backed ORC ObjectLayer ------===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#include "llvm/ExecutionEngine/Orc/ObjectLinkingLayer.h"
#include "llvm/ADT/Optional.h"
#include "llvm/ExecutionEngine/JITLink/EHFrameSupport.h"
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
#include <vector>
#define DEBUG_TYPE "orc"
using namespace llvm;
using namespace llvm::jitlink;
using namespace llvm::orc;
namespace llvm {
namespace orc {
class ObjectLinkingLayerJITLinkContext final : public JITLinkContext {
public:
ObjectLinkingLayerJITLinkContext(
ObjectLinkingLayer &Layer,
std::unique_ptr<MaterializationResponsibility> MR,
std::unique_ptr<MemoryBuffer> ObjBuffer)
: JITLinkContext(&MR->getTargetJITDylib()), Layer(Layer),
MR(std::move(MR)), ObjBuffer(std::move(ObjBuffer)) {}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
~ObjectLinkingLayerJITLinkContext() {
// If there is an object buffer return function then use it to
// return ownership of the buffer.
if (Layer.ReturnObjectBuffer && ObjBuffer)
Layer.ReturnObjectBuffer(std::move(ObjBuffer));
}
JITLinkMemoryManager &getMemoryManager() override { return Layer.MemMgr; }
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
void notifyFailed(Error Err) override {
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
for (auto &P : Layer.Plugins)
Err = joinErrors(std::move(Err), P->notifyFailed(*MR));
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
Layer.getExecutionSession().reportError(std::move(Err));
MR->failMaterialization();
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
[ORC][JITLink] Add support for weak references, and improve handling of static libraries. This patch substantially updates ORCv2's lookup API in order to support weak references, and to better support static archives. Key changes: -- Each symbol being looked for is now associated with a SymbolLookupFlags value. If the associated value is SymbolLookupFlags::RequiredSymbol then the symbol must be defined in one of the JITDylibs being searched (or be able to be generated in one of these JITDylibs via an attached definition generator) or the lookup will fail with an error. If the associated value is SymbolLookupFlags::WeaklyReferencedSymbol then the symbol is permitted to be undefined, in which case it will simply not appear in the resulting SymbolMap if the rest of the lookup succeeds. Since lookup now requires these flags for each symbol, the lookup method now takes an instance of a new SymbolLookupSet type rather than a SymbolNameSet. SymbolLookupSet is a vector-backed set of (name, flags) pairs. Clients are responsible for ensuring that the set property (i.e. unique elements) holds, though this is usually simple and SymbolLookupSet provides convenience methods to support this. -- Lookups now have an associated LookupKind value, which is either LookupKind::Static or LookupKind::DLSym. Definition generators can inspect the lookup kind when determining whether or not to generate new definitions. The StaticLibraryDefinitionGenerator is updated to only pull in new objects from the archive if the lookup kind is Static. This allows lookup to be re-used to emulate dlsym for JIT'd symbols without pulling in new objects from archives (which would not happen in a normal dlsym call). -- JITLink is updated to allow externals to be assigned weak linkage, and weak externals now use the SymbolLookupFlags::WeaklyReferencedSymbol value for lookups. Unresolved weak references will be assigned the default value of zero. Since this patch was modifying the lookup API anyway, it alo replaces all of the "MatchNonExported" boolean arguments with a "JITDylibLookupFlags" enum for readability. If a JITDylib's associated value is JITDylibLookupFlags::MatchExportedSymbolsOnly then the lookup will only match against exported (non-hidden) symbols in that JITDylib. If a JITDylib's associated value is JITDylibLookupFlags::MatchAllSymbols then the lookup will match against any symbol defined in the JITDylib.
2019-11-25 21:57:27 -08:00
void lookup(const LookupMap &Symbols,
std::unique_ptr<JITLinkAsyncLookupContinuation> LC) override {
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
JITDylibSearchOrder LinkOrder;
MR->getTargetJITDylib().withLinkOrderDo(
[&](const JITDylibSearchOrder &LO) { LinkOrder = LO; });
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
auto &ES = Layer.getExecutionSession();
[ORC][JITLink] Add support for weak references, and improve handling of static libraries. This patch substantially updates ORCv2's lookup API in order to support weak references, and to better support static archives. Key changes: -- Each symbol being looked for is now associated with a SymbolLookupFlags value. If the associated value is SymbolLookupFlags::RequiredSymbol then the symbol must be defined in one of the JITDylibs being searched (or be able to be generated in one of these JITDylibs via an attached definition generator) or the lookup will fail with an error. If the associated value is SymbolLookupFlags::WeaklyReferencedSymbol then the symbol is permitted to be undefined, in which case it will simply not appear in the resulting SymbolMap if the rest of the lookup succeeds. Since lookup now requires these flags for each symbol, the lookup method now takes an instance of a new SymbolLookupSet type rather than a SymbolNameSet. SymbolLookupSet is a vector-backed set of (name, flags) pairs. Clients are responsible for ensuring that the set property (i.e. unique elements) holds, though this is usually simple and SymbolLookupSet provides convenience methods to support this. -- Lookups now have an associated LookupKind value, which is either LookupKind::Static or LookupKind::DLSym. Definition generators can inspect the lookup kind when determining whether or not to generate new definitions. The StaticLibraryDefinitionGenerator is updated to only pull in new objects from the archive if the lookup kind is Static. This allows lookup to be re-used to emulate dlsym for JIT'd symbols without pulling in new objects from archives (which would not happen in a normal dlsym call). -- JITLink is updated to allow externals to be assigned weak linkage, and weak externals now use the SymbolLookupFlags::WeaklyReferencedSymbol value for lookups. Unresolved weak references will be assigned the default value of zero. Since this patch was modifying the lookup API anyway, it alo replaces all of the "MatchNonExported" boolean arguments with a "JITDylibLookupFlags" enum for readability. If a JITDylib's associated value is JITDylibLookupFlags::MatchExportedSymbolsOnly then the lookup will only match against exported (non-hidden) symbols in that JITDylib. If a JITDylib's associated value is JITDylibLookupFlags::MatchAllSymbols then the lookup will match against any symbol defined in the JITDylib.
2019-11-25 21:57:27 -08:00
SymbolLookupSet LookupSet;
for (auto &KV : Symbols) {
orc::SymbolLookupFlags LookupFlags;
switch (KV.second) {
case jitlink::SymbolLookupFlags::RequiredSymbol:
LookupFlags = orc::SymbolLookupFlags::RequiredSymbol;
break;
case jitlink::SymbolLookupFlags::WeaklyReferencedSymbol:
LookupFlags = orc::SymbolLookupFlags::WeaklyReferencedSymbol;
break;
}
LookupSet.add(ES.intern(KV.first), LookupFlags);
}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
// OnResolve -- De-intern the symbols and pass the result to the linker.
auto OnResolve = [LookupContinuation =
std::move(LC)](Expected<SymbolMap> Result) mutable {
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
if (!Result)
LookupContinuation->run(Result.takeError());
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
else {
AsyncLookupResult LR;
for (auto &KV : *Result)
LR[*KV.first] = KV.second;
LookupContinuation->run(std::move(LR));
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
};
for (auto &KV : InternalNamedSymbolDeps) {
SymbolDependenceMap InternalDeps;
InternalDeps[&MR->getTargetJITDylib()] = std::move(KV.second);
MR->addDependencies(KV.first, InternalDeps);
}
ES.lookup(LookupKind::Static, LinkOrder, std::move(LookupSet),
[ORC][JITLink] Add support for weak references, and improve handling of static libraries. This patch substantially updates ORCv2's lookup API in order to support weak references, and to better support static archives. Key changes: -- Each symbol being looked for is now associated with a SymbolLookupFlags value. If the associated value is SymbolLookupFlags::RequiredSymbol then the symbol must be defined in one of the JITDylibs being searched (or be able to be generated in one of these JITDylibs via an attached definition generator) or the lookup will fail with an error. If the associated value is SymbolLookupFlags::WeaklyReferencedSymbol then the symbol is permitted to be undefined, in which case it will simply not appear in the resulting SymbolMap if the rest of the lookup succeeds. Since lookup now requires these flags for each symbol, the lookup method now takes an instance of a new SymbolLookupSet type rather than a SymbolNameSet. SymbolLookupSet is a vector-backed set of (name, flags) pairs. Clients are responsible for ensuring that the set property (i.e. unique elements) holds, though this is usually simple and SymbolLookupSet provides convenience methods to support this. -- Lookups now have an associated LookupKind value, which is either LookupKind::Static or LookupKind::DLSym. Definition generators can inspect the lookup kind when determining whether or not to generate new definitions. The StaticLibraryDefinitionGenerator is updated to only pull in new objects from the archive if the lookup kind is Static. This allows lookup to be re-used to emulate dlsym for JIT'd symbols without pulling in new objects from archives (which would not happen in a normal dlsym call). -- JITLink is updated to allow externals to be assigned weak linkage, and weak externals now use the SymbolLookupFlags::WeaklyReferencedSymbol value for lookups. Unresolved weak references will be assigned the default value of zero. Since this patch was modifying the lookup API anyway, it alo replaces all of the "MatchNonExported" boolean arguments with a "JITDylibLookupFlags" enum for readability. If a JITDylib's associated value is JITDylibLookupFlags::MatchExportedSymbolsOnly then the lookup will only match against exported (non-hidden) symbols in that JITDylib. If a JITDylib's associated value is JITDylibLookupFlags::MatchAllSymbols then the lookup will match against any symbol defined in the JITDylib.
2019-11-25 21:57:27 -08:00
SymbolState::Resolved, std::move(OnResolve),
[this](const SymbolDependenceMap &Deps) {
registerDependencies(Deps);
});
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
Error notifyResolved(LinkGraph &G) override {
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
auto &ES = Layer.getExecutionSession();
SymbolFlagsMap ExtraSymbolsToClaim;
bool AutoClaim = Layer.AutoClaimObjectSymbols;
SymbolMap InternedResult;
for (auto *Sym : G.defined_symbols())
if (Sym->hasName() && Sym->getScope() != Scope::Local) {
auto InternedName = ES.intern(Sym->getName());
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
JITSymbolFlags Flags;
if (Sym->isCallable())
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
Flags |= JITSymbolFlags::Callable;
if (Sym->getScope() == Scope::Default)
Flags |= JITSymbolFlags::Exported;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
InternedResult[InternedName] =
JITEvaluatedSymbol(Sym->getAddress(), Flags);
if (AutoClaim && !MR->getSymbols().count(InternedName)) {
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
assert(!ExtraSymbolsToClaim.count(InternedName) &&
"Duplicate symbol to claim?");
ExtraSymbolsToClaim[InternedName] = Flags;
}
}
for (auto *Sym : G.absolute_symbols())
if (Sym->hasName()) {
auto InternedName = ES.intern(Sym->getName());
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
JITSymbolFlags Flags;
Flags |= JITSymbolFlags::Absolute;
if (Sym->isCallable())
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
Flags |= JITSymbolFlags::Callable;
if (Sym->getLinkage() == Linkage::Weak)
Flags |= JITSymbolFlags::Weak;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
InternedResult[InternedName] =
JITEvaluatedSymbol(Sym->getAddress(), Flags);
if (AutoClaim && !MR->getSymbols().count(InternedName)) {
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
assert(!ExtraSymbolsToClaim.count(InternedName) &&
"Duplicate symbol to claim?");
ExtraSymbolsToClaim[InternedName] = Flags;
}
}
if (!ExtraSymbolsToClaim.empty())
if (auto Err = MR->defineMaterializing(ExtraSymbolsToClaim))
return Err;
[ORC] Add generic initializer/deinitializer support. Initializers and deinitializers are used to implement C++ static constructors and destructors, runtime registration for some languages (e.g. with the Objective-C runtime for Objective-C/C++ code) and other tasks that would typically be performed when a shared-object/dylib is loaded or unloaded by a statically compiled program. MCJIT and ORC have historically provided limited support for discovering and running initializers/deinitializers by scanning the llvm.global_ctors and llvm.global_dtors variables and recording the functions to be run. This approach suffers from several drawbacks: (1) It only works for IR inputs, not for object files (including cached JIT'd objects). (2) It only works for initializers described by llvm.global_ctors and llvm.global_dtors, however not all initializers are described in this way (Objective-C, for example, describes initializers via specially named metadata sections). (3) To make the initializer/deinitializer functions described by llvm.global_ctors and llvm.global_dtors searchable they must be promoted to extern linkage, polluting the JIT symbol table (extra care must be taken to ensure this promotion does not result in symbol name clashes). This patch introduces several interdependent changes to ORCv2 to support the construction of new initialization schemes, and includes an implementation of a backwards-compatible llvm.global_ctor/llvm.global_dtor scanning scheme, and a MachO specific scheme that handles Objective-C runtime registration (if the Objective-C runtime is available) enabling execution of LLVM IR compiled from Objective-C and Swift. The major changes included in this patch are: (1) The MaterializationUnit and MaterializationResponsibility classes are extended to describe an optional "initializer" symbol for the module (see the getInitializerSymbol method on each class). The presence or absence of this symbol indicates whether the module contains any initializers or deinitializers. The initializer symbol otherwise behaves like any other: searching for it triggers materialization. (2) A new Platform interface is introduced in llvm/ExecutionEngine/Orc/Core.h which provides the following callback interface: - Error setupJITDylib(JITDylib &JD): Can be used to install standard symbols in JITDylibs upon creation. E.g. __dso_handle. - Error notifyAdding(JITDylib &JD, const MaterializationUnit &MU): Generally used to record initializer symbols. - Error notifyRemoving(JITDylib &JD, VModuleKey K): Used to notify a platform that a module is being removed. Platform implementations can use these callbacks to track outstanding initializers and implement a platform-specific approach for executing them. For example, the MachOPlatform installs a plugin in the JIT linker to scan for both __mod_inits sections (for C++ static constructors) and ObjC metadata sections. If discovered, these are processed in the usual platform order: Objective-C registration is carried out first, then static initializers are executed, ensuring that calls to Objective-C from static initializers will be safe. This patch updates LLJIT to use the new scheme for initialization. Two LLJIT::PlatformSupport classes are implemented: A GenericIR platform and a MachO platform. The GenericIR platform implements a modified version of the previous llvm.global-ctor scraping scheme to provide support for Windows and Linux. LLJIT's MachO platform uses the MachOPlatform class to provide MachO specific initialization as described above. Reviewers: sgraenitz, dblaikie Subscribers: mgorny, hiraditya, mgrang, ributzka, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D74300
2019-12-16 02:50:40 -08:00
{
// Check that InternedResult matches up with MR->getSymbols().
// This guards against faulty transformations / compilers / object caches.
// First check that there aren't any missing symbols.
size_t NumMaterializationSideEffectsOnlySymbols = 0;
SymbolNameVector ExtraSymbols;
SymbolNameVector MissingSymbols;
for (auto &KV : MR->getSymbols()) {
// If this is a materialization-side-effects only symbol then bump
// the counter and make sure it's *not* defined, otherwise make
// sure that it is defined.
if (KV.second.hasMaterializationSideEffectsOnly()) {
++NumMaterializationSideEffectsOnlySymbols;
if (InternedResult.count(KV.first))
ExtraSymbols.push_back(KV.first);
continue;
} else if (!InternedResult.count(KV.first))
MissingSymbols.push_back(KV.first);
}
// If there were missing symbols then report the error.
if (!MissingSymbols.empty())
return make_error<MissingSymbolDefinitions>(G.getName(),
std::move(MissingSymbols));
// If there are more definitions than expected, add them to the
// ExtraSymbols vector.
if (InternedResult.size() >
MR->getSymbols().size() - NumMaterializationSideEffectsOnlySymbols) {
for (auto &KV : InternedResult)
if (!MR->getSymbols().count(KV.first))
ExtraSymbols.push_back(KV.first);
}
// If there were extra definitions then report the error.
if (!ExtraSymbols.empty())
return make_error<UnexpectedSymbolDefinitions>(G.getName(),
std::move(ExtraSymbols));
}
if (auto Err = MR->notifyResolved(InternedResult))
return Err;
Layer.notifyLoaded(*MR);
return Error::success();
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
void notifyFinalized(
std::unique_ptr<JITLinkMemoryManager::Allocation> A) override {
if (auto Err = Layer.notifyEmitted(*MR, std::move(A))) {
Layer.getExecutionSession().reportError(std::move(Err));
MR->failMaterialization();
return;
}
if (auto Err = MR->notifyEmitted()) {
Layer.getExecutionSession().reportError(std::move(Err));
MR->failMaterialization();
}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
LinkGraphPassFunction getMarkLivePass(const Triple &TT) const override {
return [this](LinkGraph &G) { return markResponsibilitySymbolsLive(G); };
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
Error modifyPassConfig(const Triple &TT, PassConfiguration &Config) override {
// Add passes to mark duplicate defs as should-discard, and to walk the
// link graph to build the symbol dependence graph.
Config.PrePrunePasses.push_back([this](LinkGraph &G) {
return claimOrExternalizeWeakAndCommonSymbols(G);
});
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
Layer.modifyPassConfig(*MR, TT, Config);
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
Config.PostPrunePasses.push_back(
[this](LinkGraph &G) { return computeNamedSymbolDependencies(G); });
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
return Error::success();
}
private:
[ORC] Add generic initializer/deinitializer support. Initializers and deinitializers are used to implement C++ static constructors and destructors, runtime registration for some languages (e.g. with the Objective-C runtime for Objective-C/C++ code) and other tasks that would typically be performed when a shared-object/dylib is loaded or unloaded by a statically compiled program. MCJIT and ORC have historically provided limited support for discovering and running initializers/deinitializers by scanning the llvm.global_ctors and llvm.global_dtors variables and recording the functions to be run. This approach suffers from several drawbacks: (1) It only works for IR inputs, not for object files (including cached JIT'd objects). (2) It only works for initializers described by llvm.global_ctors and llvm.global_dtors, however not all initializers are described in this way (Objective-C, for example, describes initializers via specially named metadata sections). (3) To make the initializer/deinitializer functions described by llvm.global_ctors and llvm.global_dtors searchable they must be promoted to extern linkage, polluting the JIT symbol table (extra care must be taken to ensure this promotion does not result in symbol name clashes). This patch introduces several interdependent changes to ORCv2 to support the construction of new initialization schemes, and includes an implementation of a backwards-compatible llvm.global_ctor/llvm.global_dtor scanning scheme, and a MachO specific scheme that handles Objective-C runtime registration (if the Objective-C runtime is available) enabling execution of LLVM IR compiled from Objective-C and Swift. The major changes included in this patch are: (1) The MaterializationUnit and MaterializationResponsibility classes are extended to describe an optional "initializer" symbol for the module (see the getInitializerSymbol method on each class). The presence or absence of this symbol indicates whether the module contains any initializers or deinitializers. The initializer symbol otherwise behaves like any other: searching for it triggers materialization. (2) A new Platform interface is introduced in llvm/ExecutionEngine/Orc/Core.h which provides the following callback interface: - Error setupJITDylib(JITDylib &JD): Can be used to install standard symbols in JITDylibs upon creation. E.g. __dso_handle. - Error notifyAdding(JITDylib &JD, const MaterializationUnit &MU): Generally used to record initializer symbols. - Error notifyRemoving(JITDylib &JD, VModuleKey K): Used to notify a platform that a module is being removed. Platform implementations can use these callbacks to track outstanding initializers and implement a platform-specific approach for executing them. For example, the MachOPlatform installs a plugin in the JIT linker to scan for both __mod_inits sections (for C++ static constructors) and ObjC metadata sections. If discovered, these are processed in the usual platform order: Objective-C registration is carried out first, then static initializers are executed, ensuring that calls to Objective-C from static initializers will be safe. This patch updates LLJIT to use the new scheme for initialization. Two LLJIT::PlatformSupport classes are implemented: A GenericIR platform and a MachO platform. The GenericIR platform implements a modified version of the previous llvm.global-ctor scraping scheme to provide support for Windows and Linux. LLJIT's MachO platform uses the MachOPlatform class to provide MachO specific initialization as described above. Reviewers: sgraenitz, dblaikie Subscribers: mgorny, hiraditya, mgrang, ributzka, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D74300
2019-12-16 02:50:40 -08:00
struct LocalSymbolNamedDependencies {
SymbolNameSet Internal, External;
};
using LocalSymbolNamedDependenciesMap =
DenseMap<const Symbol *, LocalSymbolNamedDependencies>;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
Error claimOrExternalizeWeakAndCommonSymbols(LinkGraph &G) {
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
auto &ES = Layer.getExecutionSession();
SymbolFlagsMap NewSymbolsToClaim;
std::vector<std::pair<SymbolStringPtr, Symbol *>> NameToSym;
auto ProcessSymbol = [&](Symbol *Sym) {
if (Sym->hasName() && Sym->getLinkage() == Linkage::Weak) {
auto Name = ES.intern(Sym->getName());
if (!MR->getSymbols().count(ES.intern(Sym->getName()))) {
JITSymbolFlags SF = JITSymbolFlags::Weak;
if (Sym->getScope() == Scope::Default)
SF |= JITSymbolFlags::Exported;
NewSymbolsToClaim[Name] = SF;
NameToSym.push_back(std::make_pair(std::move(Name), Sym));
}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
};
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
for (auto *Sym : G.defined_symbols())
ProcessSymbol(Sym);
for (auto *Sym : G.absolute_symbols())
ProcessSymbol(Sym);
// Attempt to claim all weak defs that we're not already responsible for.
// This cannot fail -- any clashes will just result in rejection of our
// claim, at which point we'll externalize that symbol.
cantFail(MR->defineMaterializing(std::move(NewSymbolsToClaim)));
for (auto &KV : NameToSym)
if (!MR->getSymbols().count(KV.first))
G.makeExternal(*KV.second);
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
return Error::success();
}
Error markResponsibilitySymbolsLive(LinkGraph &G) const {
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
auto &ES = Layer.getExecutionSession();
for (auto *Sym : G.defined_symbols())
if (Sym->hasName() && MR->getSymbols().count(ES.intern(Sym->getName())))
Sym->setLive(true);
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
return Error::success();
}
Error computeNamedSymbolDependencies(LinkGraph &G) {
auto &ES = MR->getTargetJITDylib().getExecutionSession();
auto LocalDeps = computeLocalDeps(G);
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
[ORC] Add generic initializer/deinitializer support. Initializers and deinitializers are used to implement C++ static constructors and destructors, runtime registration for some languages (e.g. with the Objective-C runtime for Objective-C/C++ code) and other tasks that would typically be performed when a shared-object/dylib is loaded or unloaded by a statically compiled program. MCJIT and ORC have historically provided limited support for discovering and running initializers/deinitializers by scanning the llvm.global_ctors and llvm.global_dtors variables and recording the functions to be run. This approach suffers from several drawbacks: (1) It only works for IR inputs, not for object files (including cached JIT'd objects). (2) It only works for initializers described by llvm.global_ctors and llvm.global_dtors, however not all initializers are described in this way (Objective-C, for example, describes initializers via specially named metadata sections). (3) To make the initializer/deinitializer functions described by llvm.global_ctors and llvm.global_dtors searchable they must be promoted to extern linkage, polluting the JIT symbol table (extra care must be taken to ensure this promotion does not result in symbol name clashes). This patch introduces several interdependent changes to ORCv2 to support the construction of new initialization schemes, and includes an implementation of a backwards-compatible llvm.global_ctor/llvm.global_dtor scanning scheme, and a MachO specific scheme that handles Objective-C runtime registration (if the Objective-C runtime is available) enabling execution of LLVM IR compiled from Objective-C and Swift. The major changes included in this patch are: (1) The MaterializationUnit and MaterializationResponsibility classes are extended to describe an optional "initializer" symbol for the module (see the getInitializerSymbol method on each class). The presence or absence of this symbol indicates whether the module contains any initializers or deinitializers. The initializer symbol otherwise behaves like any other: searching for it triggers materialization. (2) A new Platform interface is introduced in llvm/ExecutionEngine/Orc/Core.h which provides the following callback interface: - Error setupJITDylib(JITDylib &JD): Can be used to install standard symbols in JITDylibs upon creation. E.g. __dso_handle. - Error notifyAdding(JITDylib &JD, const MaterializationUnit &MU): Generally used to record initializer symbols. - Error notifyRemoving(JITDylib &JD, VModuleKey K): Used to notify a platform that a module is being removed. Platform implementations can use these callbacks to track outstanding initializers and implement a platform-specific approach for executing them. For example, the MachOPlatform installs a plugin in the JIT linker to scan for both __mod_inits sections (for C++ static constructors) and ObjC metadata sections. If discovered, these are processed in the usual platform order: Objective-C registration is carried out first, then static initializers are executed, ensuring that calls to Objective-C from static initializers will be safe. This patch updates LLJIT to use the new scheme for initialization. Two LLJIT::PlatformSupport classes are implemented: A GenericIR platform and a MachO platform. The GenericIR platform implements a modified version of the previous llvm.global-ctor scraping scheme to provide support for Windows and Linux. LLJIT's MachO platform uses the MachOPlatform class to provide MachO specific initialization as described above. Reviewers: sgraenitz, dblaikie Subscribers: mgorny, hiraditya, mgrang, ributzka, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D74300
2019-12-16 02:50:40 -08:00
// Compute dependencies for symbols defined in the JITLink graph.
for (auto *Sym : G.defined_symbols()) {
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
// Skip local symbols: we do not track dependencies for these.
if (Sym->getScope() == Scope::Local)
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
continue;
assert(Sym->hasName() &&
"Defined non-local jitlink::Symbol should have a name");
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
SymbolNameSet ExternalSymDeps, InternalSymDeps;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
// Find internal and external named symbol dependencies.
for (auto &E : Sym->getBlock().edges()) {
auto &TargetSym = E.getTarget();
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
if (TargetSym.getScope() != Scope::Local) {
if (TargetSym.isExternal())
ExternalSymDeps.insert(ES.intern(TargetSym.getName()));
else if (&TargetSym != Sym)
InternalSymDeps.insert(ES.intern(TargetSym.getName()));
} else {
assert(TargetSym.isDefined() &&
"local symbols must be defined");
auto I = LocalDeps.find(&TargetSym);
[ORC] Add generic initializer/deinitializer support. Initializers and deinitializers are used to implement C++ static constructors and destructors, runtime registration for some languages (e.g. with the Objective-C runtime for Objective-C/C++ code) and other tasks that would typically be performed when a shared-object/dylib is loaded or unloaded by a statically compiled program. MCJIT and ORC have historically provided limited support for discovering and running initializers/deinitializers by scanning the llvm.global_ctors and llvm.global_dtors variables and recording the functions to be run. This approach suffers from several drawbacks: (1) It only works for IR inputs, not for object files (including cached JIT'd objects). (2) It only works for initializers described by llvm.global_ctors and llvm.global_dtors, however not all initializers are described in this way (Objective-C, for example, describes initializers via specially named metadata sections). (3) To make the initializer/deinitializer functions described by llvm.global_ctors and llvm.global_dtors searchable they must be promoted to extern linkage, polluting the JIT symbol table (extra care must be taken to ensure this promotion does not result in symbol name clashes). This patch introduces several interdependent changes to ORCv2 to support the construction of new initialization schemes, and includes an implementation of a backwards-compatible llvm.global_ctor/llvm.global_dtor scanning scheme, and a MachO specific scheme that handles Objective-C runtime registration (if the Objective-C runtime is available) enabling execution of LLVM IR compiled from Objective-C and Swift. The major changes included in this patch are: (1) The MaterializationUnit and MaterializationResponsibility classes are extended to describe an optional "initializer" symbol for the module (see the getInitializerSymbol method on each class). The presence or absence of this symbol indicates whether the module contains any initializers or deinitializers. The initializer symbol otherwise behaves like any other: searching for it triggers materialization. (2) A new Platform interface is introduced in llvm/ExecutionEngine/Orc/Core.h which provides the following callback interface: - Error setupJITDylib(JITDylib &JD): Can be used to install standard symbols in JITDylibs upon creation. E.g. __dso_handle. - Error notifyAdding(JITDylib &JD, const MaterializationUnit &MU): Generally used to record initializer symbols. - Error notifyRemoving(JITDylib &JD, VModuleKey K): Used to notify a platform that a module is being removed. Platform implementations can use these callbacks to track outstanding initializers and implement a platform-specific approach for executing them. For example, the MachOPlatform installs a plugin in the JIT linker to scan for both __mod_inits sections (for C++ static constructors) and ObjC metadata sections. If discovered, these are processed in the usual platform order: Objective-C registration is carried out first, then static initializers are executed, ensuring that calls to Objective-C from static initializers will be safe. This patch updates LLJIT to use the new scheme for initialization. Two LLJIT::PlatformSupport classes are implemented: A GenericIR platform and a MachO platform. The GenericIR platform implements a modified version of the previous llvm.global-ctor scraping scheme to provide support for Windows and Linux. LLJIT's MachO platform uses the MachOPlatform class to provide MachO specific initialization as described above. Reviewers: sgraenitz, dblaikie Subscribers: mgorny, hiraditya, mgrang, ributzka, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D74300
2019-12-16 02:50:40 -08:00
if (I != LocalDeps.end()) {
for (auto &S : I->second.External)
ExternalSymDeps.insert(S);
for (auto &S : I->second.Internal)
InternalSymDeps.insert(S);
}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
}
if (ExternalSymDeps.empty() && InternalSymDeps.empty())
continue;
auto SymName = ES.intern(Sym->getName());
if (!ExternalSymDeps.empty())
ExternalNamedSymbolDeps[SymName] = std::move(ExternalSymDeps);
if (!InternalSymDeps.empty())
InternalNamedSymbolDeps[SymName] = std::move(InternalSymDeps);
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
[ORC] Add generic initializer/deinitializer support. Initializers and deinitializers are used to implement C++ static constructors and destructors, runtime registration for some languages (e.g. with the Objective-C runtime for Objective-C/C++ code) and other tasks that would typically be performed when a shared-object/dylib is loaded or unloaded by a statically compiled program. MCJIT and ORC have historically provided limited support for discovering and running initializers/deinitializers by scanning the llvm.global_ctors and llvm.global_dtors variables and recording the functions to be run. This approach suffers from several drawbacks: (1) It only works for IR inputs, not for object files (including cached JIT'd objects). (2) It only works for initializers described by llvm.global_ctors and llvm.global_dtors, however not all initializers are described in this way (Objective-C, for example, describes initializers via specially named metadata sections). (3) To make the initializer/deinitializer functions described by llvm.global_ctors and llvm.global_dtors searchable they must be promoted to extern linkage, polluting the JIT symbol table (extra care must be taken to ensure this promotion does not result in symbol name clashes). This patch introduces several interdependent changes to ORCv2 to support the construction of new initialization schemes, and includes an implementation of a backwards-compatible llvm.global_ctor/llvm.global_dtor scanning scheme, and a MachO specific scheme that handles Objective-C runtime registration (if the Objective-C runtime is available) enabling execution of LLVM IR compiled from Objective-C and Swift. The major changes included in this patch are: (1) The MaterializationUnit and MaterializationResponsibility classes are extended to describe an optional "initializer" symbol for the module (see the getInitializerSymbol method on each class). The presence or absence of this symbol indicates whether the module contains any initializers or deinitializers. The initializer symbol otherwise behaves like any other: searching for it triggers materialization. (2) A new Platform interface is introduced in llvm/ExecutionEngine/Orc/Core.h which provides the following callback interface: - Error setupJITDylib(JITDylib &JD): Can be used to install standard symbols in JITDylibs upon creation. E.g. __dso_handle. - Error notifyAdding(JITDylib &JD, const MaterializationUnit &MU): Generally used to record initializer symbols. - Error notifyRemoving(JITDylib &JD, VModuleKey K): Used to notify a platform that a module is being removed. Platform implementations can use these callbacks to track outstanding initializers and implement a platform-specific approach for executing them. For example, the MachOPlatform installs a plugin in the JIT linker to scan for both __mod_inits sections (for C++ static constructors) and ObjC metadata sections. If discovered, these are processed in the usual platform order: Objective-C registration is carried out first, then static initializers are executed, ensuring that calls to Objective-C from static initializers will be safe. This patch updates LLJIT to use the new scheme for initialization. Two LLJIT::PlatformSupport classes are implemented: A GenericIR platform and a MachO platform. The GenericIR platform implements a modified version of the previous llvm.global-ctor scraping scheme to provide support for Windows and Linux. LLJIT's MachO platform uses the MachOPlatform class to provide MachO specific initialization as described above. Reviewers: sgraenitz, dblaikie Subscribers: mgorny, hiraditya, mgrang, ributzka, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D74300
2019-12-16 02:50:40 -08:00
for (auto &P : Layer.Plugins) {
auto SyntheticLocalDeps = P->getSyntheticSymbolLocalDependencies(*MR);
[ORC] Add generic initializer/deinitializer support. Initializers and deinitializers are used to implement C++ static constructors and destructors, runtime registration for some languages (e.g. with the Objective-C runtime for Objective-C/C++ code) and other tasks that would typically be performed when a shared-object/dylib is loaded or unloaded by a statically compiled program. MCJIT and ORC have historically provided limited support for discovering and running initializers/deinitializers by scanning the llvm.global_ctors and llvm.global_dtors variables and recording the functions to be run. This approach suffers from several drawbacks: (1) It only works for IR inputs, not for object files (including cached JIT'd objects). (2) It only works for initializers described by llvm.global_ctors and llvm.global_dtors, however not all initializers are described in this way (Objective-C, for example, describes initializers via specially named metadata sections). (3) To make the initializer/deinitializer functions described by llvm.global_ctors and llvm.global_dtors searchable they must be promoted to extern linkage, polluting the JIT symbol table (extra care must be taken to ensure this promotion does not result in symbol name clashes). This patch introduces several interdependent changes to ORCv2 to support the construction of new initialization schemes, and includes an implementation of a backwards-compatible llvm.global_ctor/llvm.global_dtor scanning scheme, and a MachO specific scheme that handles Objective-C runtime registration (if the Objective-C runtime is available) enabling execution of LLVM IR compiled from Objective-C and Swift. The major changes included in this patch are: (1) The MaterializationUnit and MaterializationResponsibility classes are extended to describe an optional "initializer" symbol for the module (see the getInitializerSymbol method on each class). The presence or absence of this symbol indicates whether the module contains any initializers or deinitializers. The initializer symbol otherwise behaves like any other: searching for it triggers materialization. (2) A new Platform interface is introduced in llvm/ExecutionEngine/Orc/Core.h which provides the following callback interface: - Error setupJITDylib(JITDylib &JD): Can be used to install standard symbols in JITDylibs upon creation. E.g. __dso_handle. - Error notifyAdding(JITDylib &JD, const MaterializationUnit &MU): Generally used to record initializer symbols. - Error notifyRemoving(JITDylib &JD, VModuleKey K): Used to notify a platform that a module is being removed. Platform implementations can use these callbacks to track outstanding initializers and implement a platform-specific approach for executing them. For example, the MachOPlatform installs a plugin in the JIT linker to scan for both __mod_inits sections (for C++ static constructors) and ObjC metadata sections. If discovered, these are processed in the usual platform order: Objective-C registration is carried out first, then static initializers are executed, ensuring that calls to Objective-C from static initializers will be safe. This patch updates LLJIT to use the new scheme for initialization. Two LLJIT::PlatformSupport classes are implemented: A GenericIR platform and a MachO platform. The GenericIR platform implements a modified version of the previous llvm.global-ctor scraping scheme to provide support for Windows and Linux. LLJIT's MachO platform uses the MachOPlatform class to provide MachO specific initialization as described above. Reviewers: sgraenitz, dblaikie Subscribers: mgorny, hiraditya, mgrang, ributzka, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D74300
2019-12-16 02:50:40 -08:00
if (SyntheticLocalDeps.empty())
continue;
for (auto &KV : SyntheticLocalDeps) {
auto &Name = KV.first;
auto &LocalDepsForName = KV.second;
for (auto *Local : LocalDepsForName) {
assert(Local->getScope() == Scope::Local &&
"Dependence on non-local symbol");
auto LocalNamedDepsItr = LocalDeps.find(Local);
if (LocalNamedDepsItr == LocalDeps.end())
continue;
for (auto &S : LocalNamedDepsItr->second.Internal)
InternalNamedSymbolDeps[Name].insert(S);
for (auto &S : LocalNamedDepsItr->second.External)
ExternalNamedSymbolDeps[Name].insert(S);
}
}
}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
return Error::success();
}
[ORC] Add generic initializer/deinitializer support. Initializers and deinitializers are used to implement C++ static constructors and destructors, runtime registration for some languages (e.g. with the Objective-C runtime for Objective-C/C++ code) and other tasks that would typically be performed when a shared-object/dylib is loaded or unloaded by a statically compiled program. MCJIT and ORC have historically provided limited support for discovering and running initializers/deinitializers by scanning the llvm.global_ctors and llvm.global_dtors variables and recording the functions to be run. This approach suffers from several drawbacks: (1) It only works for IR inputs, not for object files (including cached JIT'd objects). (2) It only works for initializers described by llvm.global_ctors and llvm.global_dtors, however not all initializers are described in this way (Objective-C, for example, describes initializers via specially named metadata sections). (3) To make the initializer/deinitializer functions described by llvm.global_ctors and llvm.global_dtors searchable they must be promoted to extern linkage, polluting the JIT symbol table (extra care must be taken to ensure this promotion does not result in symbol name clashes). This patch introduces several interdependent changes to ORCv2 to support the construction of new initialization schemes, and includes an implementation of a backwards-compatible llvm.global_ctor/llvm.global_dtor scanning scheme, and a MachO specific scheme that handles Objective-C runtime registration (if the Objective-C runtime is available) enabling execution of LLVM IR compiled from Objective-C and Swift. The major changes included in this patch are: (1) The MaterializationUnit and MaterializationResponsibility classes are extended to describe an optional "initializer" symbol for the module (see the getInitializerSymbol method on each class). The presence or absence of this symbol indicates whether the module contains any initializers or deinitializers. The initializer symbol otherwise behaves like any other: searching for it triggers materialization. (2) A new Platform interface is introduced in llvm/ExecutionEngine/Orc/Core.h which provides the following callback interface: - Error setupJITDylib(JITDylib &JD): Can be used to install standard symbols in JITDylibs upon creation. E.g. __dso_handle. - Error notifyAdding(JITDylib &JD, const MaterializationUnit &MU): Generally used to record initializer symbols. - Error notifyRemoving(JITDylib &JD, VModuleKey K): Used to notify a platform that a module is being removed. Platform implementations can use these callbacks to track outstanding initializers and implement a platform-specific approach for executing them. For example, the MachOPlatform installs a plugin in the JIT linker to scan for both __mod_inits sections (for C++ static constructors) and ObjC metadata sections. If discovered, these are processed in the usual platform order: Objective-C registration is carried out first, then static initializers are executed, ensuring that calls to Objective-C from static initializers will be safe. This patch updates LLJIT to use the new scheme for initialization. Two LLJIT::PlatformSupport classes are implemented: A GenericIR platform and a MachO platform. The GenericIR platform implements a modified version of the previous llvm.global-ctor scraping scheme to provide support for Windows and Linux. LLJIT's MachO platform uses the MachOPlatform class to provide MachO specific initialization as described above. Reviewers: sgraenitz, dblaikie Subscribers: mgorny, hiraditya, mgrang, ributzka, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D74300
2019-12-16 02:50:40 -08:00
LocalSymbolNamedDependenciesMap computeLocalDeps(LinkGraph &G) {
DenseMap<jitlink::Symbol *, DenseSet<jitlink::Symbol *>> DepMap;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
// For all local symbols:
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
// (1) Add their named dependencies.
// (2) Add them to the worklist for further iteration if they have any
// depend on any other local symbols.
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
struct WorklistEntry {
WorklistEntry(Symbol *Sym, DenseSet<Symbol *> LocalDeps)
: Sym(Sym), LocalDeps(std::move(LocalDeps)) {}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
Symbol *Sym = nullptr;
DenseSet<Symbol *> LocalDeps;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
};
std::vector<WorklistEntry> Worklist;
for (auto *Sym : G.defined_symbols())
if (Sym->getScope() == Scope::Local) {
auto &SymNamedDeps = DepMap[Sym];
DenseSet<Symbol *> LocalDeps;
for (auto &E : Sym->getBlock().edges()) {
auto &TargetSym = E.getTarget();
if (TargetSym.getScope() != Scope::Local)
SymNamedDeps.insert(&TargetSym);
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
else {
assert(TargetSym.isDefined() &&
"local symbols must be defined");
LocalDeps.insert(&TargetSym);
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
}
if (!LocalDeps.empty())
Worklist.push_back(WorklistEntry(Sym, std::move(LocalDeps)));
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
// Loop over all local symbols with local dependencies, propagating
// their respective non-local dependencies. Iterate until we hit a stable
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
// state.
bool Changed;
do {
Changed = false;
for (auto &WLEntry : Worklist) {
auto *Sym = WLEntry.Sym;
auto &NamedDeps = DepMap[Sym];
auto &LocalDeps = WLEntry.LocalDeps;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
for (auto *TargetSym : LocalDeps) {
auto I = DepMap.find(TargetSym);
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
if (I != DepMap.end())
for (const auto &S : I->second)
Changed |= NamedDeps.insert(S).second;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
}
} while (Changed);
[ORC] Add generic initializer/deinitializer support. Initializers and deinitializers are used to implement C++ static constructors and destructors, runtime registration for some languages (e.g. with the Objective-C runtime for Objective-C/C++ code) and other tasks that would typically be performed when a shared-object/dylib is loaded or unloaded by a statically compiled program. MCJIT and ORC have historically provided limited support for discovering and running initializers/deinitializers by scanning the llvm.global_ctors and llvm.global_dtors variables and recording the functions to be run. This approach suffers from several drawbacks: (1) It only works for IR inputs, not for object files (including cached JIT'd objects). (2) It only works for initializers described by llvm.global_ctors and llvm.global_dtors, however not all initializers are described in this way (Objective-C, for example, describes initializers via specially named metadata sections). (3) To make the initializer/deinitializer functions described by llvm.global_ctors and llvm.global_dtors searchable they must be promoted to extern linkage, polluting the JIT symbol table (extra care must be taken to ensure this promotion does not result in symbol name clashes). This patch introduces several interdependent changes to ORCv2 to support the construction of new initialization schemes, and includes an implementation of a backwards-compatible llvm.global_ctor/llvm.global_dtor scanning scheme, and a MachO specific scheme that handles Objective-C runtime registration (if the Objective-C runtime is available) enabling execution of LLVM IR compiled from Objective-C and Swift. The major changes included in this patch are: (1) The MaterializationUnit and MaterializationResponsibility classes are extended to describe an optional "initializer" symbol for the module (see the getInitializerSymbol method on each class). The presence or absence of this symbol indicates whether the module contains any initializers or deinitializers. The initializer symbol otherwise behaves like any other: searching for it triggers materialization. (2) A new Platform interface is introduced in llvm/ExecutionEngine/Orc/Core.h which provides the following callback interface: - Error setupJITDylib(JITDylib &JD): Can be used to install standard symbols in JITDylibs upon creation. E.g. __dso_handle. - Error notifyAdding(JITDylib &JD, const MaterializationUnit &MU): Generally used to record initializer symbols. - Error notifyRemoving(JITDylib &JD, VModuleKey K): Used to notify a platform that a module is being removed. Platform implementations can use these callbacks to track outstanding initializers and implement a platform-specific approach for executing them. For example, the MachOPlatform installs a plugin in the JIT linker to scan for both __mod_inits sections (for C++ static constructors) and ObjC metadata sections. If discovered, these are processed in the usual platform order: Objective-C registration is carried out first, then static initializers are executed, ensuring that calls to Objective-C from static initializers will be safe. This patch updates LLJIT to use the new scheme for initialization. Two LLJIT::PlatformSupport classes are implemented: A GenericIR platform and a MachO platform. The GenericIR platform implements a modified version of the previous llvm.global-ctor scraping scheme to provide support for Windows and Linux. LLJIT's MachO platform uses the MachOPlatform class to provide MachO specific initialization as described above. Reviewers: sgraenitz, dblaikie Subscribers: mgorny, hiraditya, mgrang, ributzka, llvm-commits Tags: #llvm Differential Revision: https://reviews.llvm.org/D74300
2019-12-16 02:50:40 -08:00
// Intern the results to produce a mapping of jitlink::Symbol* to internal
// and external symbol names.
auto &ES = Layer.getExecutionSession();
LocalSymbolNamedDependenciesMap Result;
for (auto &KV : DepMap) {
auto *Local = KV.first;
assert(Local->getScope() == Scope::Local &&
"DepMap keys should all be local symbols");
auto &LocalNamedDeps = Result[Local];
for (auto *Named : KV.second) {
assert(Named->getScope() != Scope::Local &&
"DepMap values should all be non-local symbol sets");
if (Named->isExternal())
LocalNamedDeps.External.insert(ES.intern(Named->getName()));
else
LocalNamedDeps.Internal.insert(ES.intern(Named->getName()));
}
}
return Result;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
void registerDependencies(const SymbolDependenceMap &QueryDeps) {
for (auto &NamedDepsEntry : ExternalNamedSymbolDeps) {
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
auto &Name = NamedDepsEntry.first;
auto &NameDeps = NamedDepsEntry.second;
SymbolDependenceMap SymbolDeps;
for (const auto &QueryDepsEntry : QueryDeps) {
JITDylib &SourceJD = *QueryDepsEntry.first;
const SymbolNameSet &Symbols = QueryDepsEntry.second;
auto &DepsForJD = SymbolDeps[&SourceJD];
for (const auto &S : Symbols)
if (NameDeps.count(S))
DepsForJD.insert(S);
if (DepsForJD.empty())
SymbolDeps.erase(&SourceJD);
}
MR->addDependencies(Name, SymbolDeps);
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
}
ObjectLinkingLayer &Layer;
std::unique_ptr<MaterializationResponsibility> MR;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
std::unique_ptr<MemoryBuffer> ObjBuffer;
DenseMap<SymbolStringPtr, SymbolNameSet> ExternalNamedSymbolDeps;
DenseMap<SymbolStringPtr, SymbolNameSet> InternalNamedSymbolDeps;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
};
ObjectLinkingLayer::Plugin::~Plugin() {}
ObjectLinkingLayer::ObjectLinkingLayer(ExecutionSession &ES,
JITLinkMemoryManager &MemMgr)
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
: ObjectLayer(ES), MemMgr(MemMgr) {
ES.registerResourceManager(*this);
}
ObjectLinkingLayer::ObjectLinkingLayer(
ExecutionSession &ES, std::unique_ptr<JITLinkMemoryManager> MemMgr)
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
: ObjectLayer(ES), MemMgr(*MemMgr), MemMgrOwnership(std::move(MemMgr)) {
ES.registerResourceManager(*this);
}
ObjectLinkingLayer::~ObjectLinkingLayer() {
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
assert(Allocs.empty() && "Layer destroyed with resources still attached");
getExecutionSession().deregisterResourceManager(*this);
}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
void ObjectLinkingLayer::emit(std::unique_ptr<MaterializationResponsibility> R,
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
std::unique_ptr<MemoryBuffer> O) {
assert(O && "Object must not be null");
auto ObjBuffer = O->getMemBufferRef();
auto Ctx = std::make_unique<ObjectLinkingLayerJITLinkContext>(
*this, std::move(R), std::move(O));
if (auto G = createLinkGraphFromObject(std::move(ObjBuffer)))
link(std::move(*G), std::move(Ctx));
else
Ctx->notifyFailed(G.takeError());
}
void ObjectLinkingLayer::emit(std::unique_ptr<MaterializationResponsibility> R,
std::unique_ptr<LinkGraph> G) {
link(std::move(G), std::make_unique<ObjectLinkingLayerJITLinkContext>(
*this, std::move(R), nullptr));
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
void ObjectLinkingLayer::modifyPassConfig(MaterializationResponsibility &MR,
const Triple &TT,
PassConfiguration &PassConfig) {
for (auto &P : Plugins)
P->modifyPassConfig(MR, TT, PassConfig);
}
void ObjectLinkingLayer::notifyLoaded(MaterializationResponsibility &MR) {
for (auto &P : Plugins)
P->notifyLoaded(MR);
}
Error ObjectLinkingLayer::notifyEmitted(MaterializationResponsibility &MR,
AllocPtr Alloc) {
Error Err = Error::success();
for (auto &P : Plugins)
Err = joinErrors(std::move(Err), P->notifyEmitted(MR));
if (Err)
return Err;
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
return MR.withResourceKeyDo(
[&](ResourceKey K) { Allocs[K].push_back(std::move(Alloc)); });
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
Error ObjectLinkingLayer::handleRemoveResources(ResourceKey K) {
Error Err = Error::success();
for (auto &P : Plugins)
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
Err = joinErrors(std::move(Err), P->notifyRemovingResources(K));
std::vector<AllocPtr> AllocsToRemove;
getExecutionSession().runSessionLocked([&] {
auto I = Allocs.find(K);
if (I != Allocs.end()) {
std::swap(AllocsToRemove, I->second);
Allocs.erase(I);
}
});
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
while (!AllocsToRemove.empty()) {
Err = joinErrors(std::move(Err), AllocsToRemove.back()->deallocate());
AllocsToRemove.pop_back();
}
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
return Err;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
void ObjectLinkingLayer::handleTransferResources(ResourceKey DstKey,
ResourceKey SrcKey) {
auto I = Allocs.find(SrcKey);
if (I != Allocs.end()) {
auto &SrcAllocs = I->second;
auto &DstAllocs = Allocs[DstKey];
DstAllocs.reserve(DstAllocs.size() + SrcAllocs.size());
for (auto &Alloc : SrcAllocs)
DstAllocs.push_back(std::move(Alloc));
// Erase SrcKey entry using value rather than iterator I: I may have been
// invalidated when we looked up DstKey.
Allocs.erase(SrcKey);
}
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
for (auto &P : Plugins)
P->notifyTransferringResources(DstKey, SrcKey);
}
EHFrameRegistrationPlugin::EHFrameRegistrationPlugin(
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
ExecutionSession &ES, std::unique_ptr<EHFrameRegistrar> Registrar)
: ES(ES), Registrar(std::move(Registrar)) {}
void EHFrameRegistrationPlugin::modifyPassConfig(
MaterializationResponsibility &MR, const Triple &TT,
PassConfiguration &PassConfig) {
PassConfig.PostFixupPasses.push_back(createEHFrameRecorderPass(
TT, [this, &MR](JITTargetAddress Addr, size_t Size) {
if (Addr) {
std::lock_guard<std::mutex> Lock(EHFramePluginMutex);
assert(!InProcessLinks.count(&MR) &&
"Link for MR already being tracked?");
InProcessLinks[&MR] = {Addr, Size};
}
}));
}
Error EHFrameRegistrationPlugin::notifyEmitted(
MaterializationResponsibility &MR) {
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
EHFrameRange EmittedRange;
{
std::lock_guard<std::mutex> Lock(EHFramePluginMutex);
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
auto EHFrameRangeItr = InProcessLinks.find(&MR);
if (EHFrameRangeItr == InProcessLinks.end())
return Error::success();
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
EmittedRange = EHFrameRangeItr->second;
assert(EmittedRange.Addr && "eh-frame addr to register can not be null");
InProcessLinks.erase(EHFrameRangeItr);
}
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
if (auto Err = MR.withResourceKeyDo(
[&](ResourceKey K) { EHFrameRanges[K].push_back(EmittedRange); }))
return Err;
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
return Registrar->registerEHFrames(EmittedRange.Addr, EmittedRange.Size);
}
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
Error EHFrameRegistrationPlugin::notifyFailed(
MaterializationResponsibility &MR) {
std::lock_guard<std::mutex> Lock(EHFramePluginMutex);
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
InProcessLinks.erase(&MR);
return Error::success();
}
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
Error EHFrameRegistrationPlugin::notifyRemovingResources(ResourceKey K) {
std::vector<EHFrameRange> RangesToRemove;
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
ES.runSessionLocked([&] {
auto I = EHFrameRanges.find(K);
if (I != EHFrameRanges.end()) {
RangesToRemove = std::move(I->second);
EHFrameRanges.erase(I);
}
});
Error Err = Error::success();
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
while (!RangesToRemove.empty()) {
auto RangeToRemove = RangesToRemove.back();
RangesToRemove.pop_back();
assert(RangeToRemove.Addr && "Untracked eh-frame range must not be null");
Err = joinErrors(
std::move(Err),
Registrar->deregisterEHFrames(RangeToRemove.Addr, RangeToRemove.Size));
}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
return Err;
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
}
[ORC] Add support for resource tracking/removal (removable code). This patch introduces new APIs to support resource tracking and removal in Orc. It is intended as a thread-safe generalization of the removeModule concept from OrcV1. Clients can now create ResourceTracker objects (using JITDylib::createResourceTracker) to track resources for each MaterializationUnit (code, data, aliases, absolute symbols, etc.) added to the JIT. Every MaterializationUnit will be associated with a ResourceTracker, and ResourceTrackers can be re-used for multiple MaterializationUnits. Each JITDylib has a default ResourceTracker that will be used for MaterializationUnits added to that JITDylib if no ResourceTracker is explicitly specified. Two operations can be performed on ResourceTrackers: transferTo and remove. The transferTo operation transfers tracking of the resources to a different ResourceTracker object, allowing ResourceTrackers to be merged to reduce administrative overhead (the source tracker is invalidated in the process). The remove operation removes all resources associated with a ResourceTracker, including any symbols defined by MaterializationUnits associated with the tracker, and also invalidates the tracker. These operations are thread safe, and should work regardless of the the state of the MaterializationUnits. In the case of resource transfer any existing resources associated with the source tracker will be transferred to the destination tracker, and all future resources for those units will be automatically associated with the destination tracker. In the case of resource removal all already-allocated resources will be deallocated, any if any program representations associated with the tracker have not been compiled yet they will be destroyed. If any program representations are currently being compiled then they will be prevented from completing: their MaterializationResponsibility will return errors on any attempt to update the JIT state. Clients (usually Layer writers) wishing to track resources can implement the ResourceManager API to receive notifications when ResourceTrackers are transferred or removed. The MaterializationResponsibility::withResourceKeyDo method can be used to create associations between the key for a ResourceTracker and an allocated resource in a thread-safe way. RTDyldObjectLinkingLayer and ObjectLinkingLayer are updated to use the ResourceManager API to enable tracking and removal of memory allocated by the JIT linker. The new JITDylib::clear method can be used to trigger removal of every ResourceTracker associated with the JITDylib (note that this will only remove resources for the JITDylib, it does not run static destructors). This patch includes unit tests showing basic usage. A follow-up patch will update the Kaleidoscope and BuildingAJIT tutorial series to OrcV2 and will use this API to release code associated with anonymous expressions.
2020-09-11 09:50:41 -07:00
void EHFrameRegistrationPlugin::notifyTransferringResources(
ResourceKey DstKey, ResourceKey SrcKey) {
auto SI = EHFrameRanges.find(SrcKey);
if (SI != EHFrameRanges.end()) {
auto &SrcRanges = SI->second;
auto &DstRanges = EHFrameRanges[DstKey];
DstRanges.reserve(DstRanges.size() + SrcRanges.size());
for (auto &SrcRange : SrcRanges)
DstRanges.push_back(std::move(SrcRange));
EHFrameRanges.erase(SI);
}
}
Initial implementation of JITLink - A replacement for RuntimeDyld. Summary: JITLink is a jit-linker that performs the same high-level task as RuntimeDyld: it parses relocatable object files and makes their contents runnable in a target process. JITLink aims to improve on RuntimeDyld in several ways: (1) A clear design intended to maximize code-sharing while minimizing coupling. RuntimeDyld has been developed in an ad-hoc fashion for a number of years and this had led to intermingling of code for multiple architectures (e.g. in RuntimeDyldELF::processRelocationRef) in a way that makes the code more difficult to read, reason about, extend. JITLink is designed to isolate format and architecture specific code, while still sharing generic code. (2) Support for native code models. RuntimeDyld required the use of large code models (where calls to external functions are made indirectly via registers) for many of platforms due to its restrictive model for stub generation (one "stub" per symbol). JITLink allows arbitrary mutation of the atom graph, allowing both GOT and PLT atoms to be added naturally. (3) Native support for asynchronous linking. JITLink uses asynchronous calls for symbol resolution and finalization: these callbacks are passed a continuation function that they must call to complete the linker's work. This allows for cleaner interoperation with the new concurrent ORC JIT APIs, while still being easily implementable in synchronous style if asynchrony is not needed. To maximise sharing, the design has a hierarchy of common code: (1) Generic atom-graph data structure and algorithms (e.g. dead stripping and | memory allocation) that are intended to be shared by all architectures. | + -- (2) Shared per-format code that utilizes (1), e.g. Generic MachO to | atom-graph parsing. | + -- (3) Architecture specific code that uses (1) and (2). E.g. JITLinkerMachO_x86_64, which adds x86-64 specific relocation support to (2) to build and patch up the atom graph. To support asynchronous symbol resolution and finalization, the callbacks for these operations take continuations as arguments: using JITLinkAsyncLookupContinuation = std::function<void(Expected<AsyncLookupResult> LR)>; using JITLinkAsyncLookupFunction = std::function<void(const DenseSet<StringRef> &Symbols, JITLinkAsyncLookupContinuation LookupContinuation)>; using FinalizeContinuation = std::function<void(Error)>; virtual void finalizeAsync(FinalizeContinuation OnFinalize); In addition to its headline features, JITLink also makes other improvements: - Dead stripping support: symbols that are not used (e.g. redundant ODR definitions) are discarded, and take up no memory in the target process (In contrast, RuntimeDyld supported pointer equality for weak definitions, but the redundant definitions stayed resident in memory). - Improved exception handling support. JITLink provides a much more extensive eh-frame parser than RuntimeDyld, and is able to correctly fix up many eh-frame sections that RuntimeDyld currently (silently) fails on. - More extensive validation and error handling throughout. This initial patch supports linking MachO/x86-64 only. Work on support for other architectures and formats will happen in-tree. Differential Revision: https://reviews.llvm.org/D58704 llvm-svn: 358818
2019-04-20 17:10:34 +00:00
} // End namespace orc.
} // End namespace llvm.