1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-26 04:32:44 +01:00
llvm-mirror/include/llvm/Support/CrashRecoveryContext.h
Alexandre Ganea 733e59acce [LLD][COFF] When using LLD-as-a-library, always prevent re-entrance on failures
This is a follow-up for D70378 (Cover usage of LLD as a library).

While debugging an intermittent failure on a bot, I recalled this scenario which
causes the issue:

1.When executing lld/test/ELF/invalid/symtab-sh-info.s L45, we reach
  lld:🧝:Obj-File::ObjFile() which goes straight into its base ELFFileBase(),
  then ELFFileBase::init().
2.At that point fatal() is thrown in lld/ELF/InputFiles.cpp L381, leaving a
  half-initialized ObjFile instance.
3.We then end up in lld::exitLld() and since we are running with LLD_IN_TEST, we
  hapily restore the control flow to CrashRecoveryContext::RunSafely() then back
  in lld::safeLldMain().
4.Before this patch, we called errorHandler().reset() just after, and this
  attempted to reset the associated SpecificAlloc<ObjFile<ELF64LE>>. That tried
  to free the half-initialized ObjFile instance, and more precisely its
  ObjFile::dwarf member.

Sometimes that worked, sometimes it failed and was catched by the
CrashRecoveryContext. This scenario was the reason we called
errorHandler().reset() through a CrashRecoveryContext.

But in some rare cases, the above repro somehow corrupted the heap, creating a
stack overflow. When the CrashRecoveryContext's filter (that is,
__except (ExceptionFilter(GetExceptionInformation()))) tried to handle the
exception, it crashed again since the stack was exhausted -- and that took the
whole application down. That is the issue seen on the bot. Locally it happens
about 1 times out of 15.

Now this situation can happen anywhere in LLD. Since catching stack overflows is
not a reliable scenario ATM when using CrashRecoveryContext, we're now
preventing further re-entrance when such failures occur, by signaling
lld::SafeReturn::canRunAgain=false. When running with LLD_IN_TEST=2 (or above),
only one iteration will be executed, instead of two.

Differential Revision: https://reviews.llvm.org/D88348
2020-11-12 08:14:43 -05:00

271 lines
9.4 KiB
C++

//===--- CrashRecoveryContext.h - Crash Recovery ----------------*- C++ -*-===//
//
// 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
//
//===----------------------------------------------------------------------===//
#ifndef LLVM_SUPPORT_CRASHRECOVERYCONTEXT_H
#define LLVM_SUPPORT_CRASHRECOVERYCONTEXT_H
#include "llvm/ADT/STLExtras.h"
namespace llvm {
class CrashRecoveryContextCleanup;
/// Crash recovery helper object.
///
/// This class implements support for running operations in a safe context so
/// that crashes (memory errors, stack overflow, assertion violations) can be
/// detected and control restored to the crashing thread. Crash detection is
/// purely "best effort", the exact set of failures which can be recovered from
/// is platform dependent.
///
/// Clients make use of this code by first calling
/// CrashRecoveryContext::Enable(), and then executing unsafe operations via a
/// CrashRecoveryContext object. For example:
///
/// \code
/// void actual_work(void *);
///
/// void foo() {
/// CrashRecoveryContext CRC;
///
/// if (!CRC.RunSafely(actual_work, 0)) {
/// ... a crash was detected, report error to user ...
/// }
///
/// ... no crash was detected ...
/// }
/// \endcode
///
/// To assist recovery the class allows specifying set of actions that will be
/// executed in any case, whether crash occurs or not. These actions may be used
/// to reclaim resources in the case of crash.
class CrashRecoveryContext {
void *Impl = nullptr;
CrashRecoveryContextCleanup *head = nullptr;
public:
CrashRecoveryContext();
~CrashRecoveryContext();
/// Register cleanup handler, which is used when the recovery context is
/// finished.
/// The recovery context owns the handler.
void registerCleanup(CrashRecoveryContextCleanup *cleanup);
void unregisterCleanup(CrashRecoveryContextCleanup *cleanup);
/// Enable crash recovery.
static void Enable();
/// Disable crash recovery.
static void Disable();
/// Return the active context, if the code is currently executing in a
/// thread which is in a protected context.
static CrashRecoveryContext *GetCurrent();
/// Return true if the current thread is recovering from a crash.
static bool isRecoveringFromCrash();
/// Execute the provided callback function (with the given arguments) in
/// a protected context.
///
/// \return True if the function completed successfully, and false if the
/// function crashed (or HandleCrash was called explicitly). Clients should
/// make as little assumptions as possible about the program state when
/// RunSafely has returned false.
bool RunSafely(function_ref<void()> Fn);
bool RunSafely(void (*Fn)(void*), void *UserData) {
return RunSafely([&]() { Fn(UserData); });
}
/// Execute the provide callback function (with the given arguments) in
/// a protected context which is run in another thread (optionally with a
/// requested stack size).
///
/// See RunSafely() and llvm_execute_on_thread().
///
/// On Darwin, if PRIO_DARWIN_BG is set on the calling thread, it will be
/// propagated to the new thread as well.
bool RunSafelyOnThread(function_ref<void()>, unsigned RequestedStackSize = 0);
bool RunSafelyOnThread(void (*Fn)(void*), void *UserData,
unsigned RequestedStackSize = 0) {
return RunSafelyOnThread([&]() { Fn(UserData); }, RequestedStackSize);
}
/// Explicitly trigger a crash recovery in the current process, and
/// return failure from RunSafely(). This function does not return.
LLVM_ATTRIBUTE_NORETURN
void HandleExit(int RetCode);
/// Throw again a signal or an exception, after it was catched once by a
/// CrashRecoveryContext.
static bool throwIfCrash(int RetCode);
/// In case of a crash, this is the crash identifier.
int RetCode = 0;
/// Selects whether handling of failures should be done in the same way as
/// for regular crashes. When this is active, a crash would print the
/// callstack, clean-up any temporary files and create a coredump/minidump.
bool DumpStackAndCleanupOnFailure = false;
};
/// Abstract base class of cleanup handlers.
///
/// Derived classes override method recoverResources, which makes actual work on
/// resource recovery.
///
/// Cleanup handlers are stored in a double list, which is owned and managed by
/// a crash recovery context.
class CrashRecoveryContextCleanup {
protected:
CrashRecoveryContext *context = nullptr;
CrashRecoveryContextCleanup(CrashRecoveryContext *context)
: context(context) {}
public:
bool cleanupFired = false;
virtual ~CrashRecoveryContextCleanup();
virtual void recoverResources() = 0;
CrashRecoveryContext *getContext() const {
return context;
}
private:
friend class CrashRecoveryContext;
CrashRecoveryContextCleanup *prev = nullptr, *next = nullptr;
};
/// Base class of cleanup handler that controls recovery of resources of the
/// given type.
///
/// \tparam Derived Class that uses this class as a base.
/// \tparam T Type of controlled resource.
///
/// This class serves as a base for its template parameter as implied by
/// Curiously Recurring Template Pattern.
///
/// This class factors out creation of a cleanup handler. The latter requires
/// knowledge of the current recovery context, which is provided by this class.
template<typename Derived, typename T>
class CrashRecoveryContextCleanupBase : public CrashRecoveryContextCleanup {
protected:
T *resource;
CrashRecoveryContextCleanupBase(CrashRecoveryContext *context, T *resource)
: CrashRecoveryContextCleanup(context), resource(resource) {}
public:
/// Creates cleanup handler.
/// \param x Pointer to the resource recovered by this handler.
/// \return New handler or null if the method was called outside a recovery
/// context.
static Derived *create(T *x) {
if (x) {
if (CrashRecoveryContext *context = CrashRecoveryContext::GetCurrent())
return new Derived(context, x);
}
return nullptr;
}
};
/// Cleanup handler that reclaims resource by calling destructor on it.
template <typename T>
class CrashRecoveryContextDestructorCleanup : public
CrashRecoveryContextCleanupBase<CrashRecoveryContextDestructorCleanup<T>, T> {
public:
CrashRecoveryContextDestructorCleanup(CrashRecoveryContext *context,
T *resource)
: CrashRecoveryContextCleanupBase<
CrashRecoveryContextDestructorCleanup<T>, T>(context, resource) {}
void recoverResources() override {
this->resource->~T();
}
};
/// Cleanup handler that reclaims resource by calling 'delete' on it.
template <typename T>
class CrashRecoveryContextDeleteCleanup : public
CrashRecoveryContextCleanupBase<CrashRecoveryContextDeleteCleanup<T>, T> {
public:
CrashRecoveryContextDeleteCleanup(CrashRecoveryContext *context, T *resource)
: CrashRecoveryContextCleanupBase<
CrashRecoveryContextDeleteCleanup<T>, T>(context, resource) {}
void recoverResources() override { delete this->resource; }
};
/// Cleanup handler that reclaims resource by calling its method 'Release'.
template <typename T>
class CrashRecoveryContextReleaseRefCleanup : public
CrashRecoveryContextCleanupBase<CrashRecoveryContextReleaseRefCleanup<T>, T> {
public:
CrashRecoveryContextReleaseRefCleanup(CrashRecoveryContext *context,
T *resource)
: CrashRecoveryContextCleanupBase<CrashRecoveryContextReleaseRefCleanup<T>,
T>(context, resource) {}
void recoverResources() override { this->resource->Release(); }
};
/// Helper class for managing resource cleanups.
///
/// \tparam T Type of resource been reclaimed.
/// \tparam Cleanup Class that defines how the resource is reclaimed.
///
/// Clients create objects of this type in the code executed in a crash recovery
/// context to ensure that the resource will be reclaimed even in the case of
/// crash. For example:
///
/// \code
/// void actual_work(void *) {
/// ...
/// std::unique_ptr<Resource> R(new Resource());
/// CrashRecoveryContextCleanupRegistrar D(R.get());
/// ...
/// }
///
/// void foo() {
/// CrashRecoveryContext CRC;
///
/// if (!CRC.RunSafely(actual_work, 0)) {
/// ... a crash was detected, report error to user ...
/// }
/// \endcode
///
/// If the code of `actual_work` in the example above does not crash, the
/// destructor of CrashRecoveryContextCleanupRegistrar removes cleanup code from
/// the current CrashRecoveryContext and the resource is reclaimed by the
/// destructor of std::unique_ptr. If crash happens, destructors are not called
/// and the resource is reclaimed by cleanup object registered in the recovery
/// context by the constructor of CrashRecoveryContextCleanupRegistrar.
template <typename T, typename Cleanup = CrashRecoveryContextDeleteCleanup<T> >
class CrashRecoveryContextCleanupRegistrar {
CrashRecoveryContextCleanup *cleanup;
public:
CrashRecoveryContextCleanupRegistrar(T *x)
: cleanup(Cleanup::create(x)) {
if (cleanup)
cleanup->getContext()->registerCleanup(cleanup);
}
~CrashRecoveryContextCleanupRegistrar() { unregister(); }
void unregister() {
if (cleanup && !cleanup->cleanupFired)
cleanup->getContext()->unregisterCleanup(cleanup);
cleanup = nullptr;
}
};
} // end namespace llvm
#endif // LLVM_SUPPORT_CRASHRECOVERYCONTEXT_H