NextGraph
/
rocksdb
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fixed a snappy compression wrapper bug (passing wrong variable).

Browse Source 
Change atomic_pointer.h to prefer a memory barrier based
implementation over a <cstdatomic> based implementation for
the following reasons:
(1) On a x86-32-bit gcc-4.4 build, <ctdatomic> was corrupting
    the AtomicPointer.
(2) On a x86-64-bit gcc build, a <ctstdatomic> based acquire-load
    takes ~15ns as opposed to the ~1ns for a memory-barrier
    based implementation.

Fixes issue 9 (corruption_test fails)
http://code.google.com/p/leveldb/issues/detail?id=9

Fixes issue 16 (CorruptionTest.MissingDescriptor fails)
http://code.google.com/p/leveldb/issues/detail?id=16



git-svn-id: https://leveldb.googlecode.com/svn/trunk@36 62dab493-f737-651d-591e-8d6aee1b9529
main
gabor@google.com 13 years ago
parent 85f0ab1975
commit ed154f6dc4
2 changed files with 66 additions and 144 deletions
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Show Stats Download Patch File Download Diff File
  1. 208
      port/atomic_pointer.h
  2. 2
      port/port_posix.h

208
port/atomic_pointer.h
Unescape View File

@ -4,57 +4,31 @@
// AtomicPointer provides storage for a lock-free pointer.
// Platform-dependent implementation of AtomicPointer:
// - If the platform provides a cheap barrier, we use it with raw pointers
// - If cstdatomic is present (on newer versions of gcc, it is), we use
// a cstdatomic-based AtomicPointer
// - If it is not, we define processor-dependent AtomicWord operations,
// and then use them to build AtomicPointer
//
// a cstdatomic-based AtomicPointer. However we prefer the memory
// barrier based version, because at least on a gcc 4.4 32-bit build
// on linux, we have encountered a buggy <cstdatomic>
// implementation. Also, some <cstdatomic> implementations are much
// slower than a memory-barrier based implementation (~16ns for
// <cstdatomic> based acquire-load vs. ~1ns for a barrier based
// acquire-load).
// This code is based on atomicops-internals-* in Google's perftools:
// http://code.google.com/p/google-perftools/source/browse/#svn%2Ftrunk%2Fsrc%2Fbase
#ifndef PORT_ATOMIC_POINTER_H_
#define PORT_ATOMIC_POINTER_H_
#include <stdint.h>
#ifdef LEVELDB_CSTDATOMIC_PRESENT
///////////////////////////////////////////////////////////////////////////////
// WE HAVE <cstdatomic>
// Use a <cstdatomic>-based AtomicPointer
#include <cstdatomic>
#include <stdint.h>
namespace leveldb {
namespace port {
// Storage for a lock-free pointer
class AtomicPointer {
private:
std::atomic<void*> rep_;
public:
AtomicPointer() { }
explicit AtomicPointer(void* v) : rep_(v) { }
inline void* Acquire_Load() const {
return rep_.load(std::memory_order_acquire);
}
inline void Release_Store(void* v) {
rep_.store(v, std::memory_order_release);
}
inline void* NoBarrier_Load() const {
return rep_.load(std::memory_order_relaxed);
}
inline void NoBarrier_Store(void* v) {
rep_.store(v, std::memory_order_relaxed);
}
};
} // namespace leveldb::port
} // namespace leveldb
#else
///////////////////////////////////////////////////////////////////////////////
// NO <cstdatomic>
// The entire rest of this file covers that case
#endif
#ifdef OS_WIN
#include <windows.h>
#endif
#ifdef OS_MACOSX
#include <libkern/OSAtomic.h>
#endif
#if defined(_M_X64) || defined(__x86_64__)
#define ARCH_CPU_X86_FAMILY 1
@ -62,152 +36,100 @@ class AtomicPointer {
#define ARCH_CPU_X86_FAMILY 1
#elif defined(__ARMEL__)
#define ARCH_CPU_ARM_FAMILY 1
#else
#warning Please add support for your architecture in atomicpointer.h
#endif
namespace leveldb {
namespace port {
namespace internal {
// AtomicWord is a machine-sized pointer.
typedef intptr_t AtomicWord;
} // namespace leveldb::port::internal
} // namespace leveldb::port
} // namespace leveldb
// Include our platform specific implementation.
///////////////////////////////////////////////////////////////////////////////
// Define MemoryBarrier() if available
// Windows on x86
#if defined(OS_WIN) && defined(COMPILER_MSVC) && defined(ARCH_CPU_X86_FAMILY)
// void MemoryBarrier(void) macro is defined in windows.h:
// windows.h already provides a MemoryBarrier(void) macro
// http://msdn.microsoft.com/en-us/library/ms684208(v=vs.85).aspx
// Including windows.h here; MemoryBarrier() gets used below.
#include <windows.h>
///////////////////////////////////////////////////////////////////////////////
// Mac OS on x86
#elif defined(OS_MACOSX) && defined(ARCH_CPU_X86_FAMILY)
#include <libkern/OSAtomic.h>
namespace leveldb {
namespace port {
namespace internal {
#define LEVELDB_HAVE_MEMORY_BARRIER
// Gcc on x86
#elif defined(__GNUC__) && defined(ARCH_CPU_X86_FAMILY)
inline void MemoryBarrier() {
#if defined(__GNUC__) && (defined(__i386__) || defined(__x86_64__))
// See http://gcc.gnu.org/ml/gcc/2003-04/msg01180.html for a discussion on
// this idiom. Also see http://en.wikipedia.org/wiki/Memory_ordering.
__asm__ __volatile__("" : : : "memory");
#else
OSMemoryBarrier();
#endif
}
#define LEVELDB_HAVE_MEMORY_BARRIER
} // namespace leveldb::port::internal
} // namespace leveldb::port
} // namespace leveldb
///////////////////////////////////////////////////////////////////////////////
// Any x86 CPU
#elif defined(ARCH_CPU_X86_FAMILY)
namespace leveldb {
namespace port {
namespace internal {
// Mac OS
#elif defined(OS_MACOSX)
inline void MemoryBarrier() {
__asm__ __volatile__("" : : : "memory");
OSMemoryBarrier();
}
#define LEVELDB_HAVE_MEMORY_BARRIER
} // namespace leveldb::port::internal
} // namespace leveldb::port
} // namespace leveldb
#undef ATOMICOPS_COMPILER_BARRIER
///////////////////////////////////////////////////////////////////////////////
// ARM
#elif defined(ARCH_CPU_ARM_FAMILY)
namespace leveldb {
namespace port {
namespace internal {
typedef void (*LinuxKernelMemoryBarrierFunc)(void);
LinuxKernelMemoryBarrierFunc pLinuxKernelMemoryBarrier __attribute__((weak)) =
(LinuxKernelMemoryBarrierFunc) 0xffff0fa0;
inline void MemoryBarrier() {
pLinuxKernelMemoryBarrier();
}
#define LEVELDB_HAVE_MEMORY_BARRIER
} // namespace leveldb::port::internal
} // namespace leveldb::port
} // namespace leveldb
#else
#error "Atomic operations are not supported on your platform"
#endif
///////////////////////////////////////////////////////////////////////////////
// Implementation of AtomicPointer based on MemoryBarriers above
namespace leveldb {
namespace port {
namespace internal {
// Atomic operations using per-system MemoryBarrier()s
inline AtomicWord Acquire_Load(volatile const AtomicWord* ptr) {
AtomicWord value = *ptr;
MemoryBarrier();
return value;
}
inline void Release_Store(volatile AtomicWord* ptr, AtomicWord value) {
MemoryBarrier();
*ptr = value;
}
inline AtomicWord NoBarrier_Load(volatile const AtomicWord* ptr) {
return *ptr;
}
inline void NoBarrier_Store(volatile AtomicWord* ptr, AtomicWord value) {
*ptr = value;
}
} // namespace leveldb::port::internal
// AtomicPointer built using platform-specific MemoryBarrier()
#if defined(LEVELDB_HAVE_MEMORY_BARRIER)
class AtomicPointer {
private:
void* rep_;
public:
AtomicPointer() { }
explicit AtomicPointer(void* p) : rep_(p) {}
inline void* NoBarrier_Load() const { return rep_; }
inline void NoBarrier_Store(void* v) { rep_ = v; }
inline void* Acquire_Load() const {
void* result = rep_;
MemoryBarrier();
return result;
}
inline void Release_Store(void* v) {
MemoryBarrier();
rep_ = v;
}
};
// AtomicPointer definition for systems without <cstdatomic>.
// AtomicPointer based on <cstdatomic>
#elif defined(LEVELDB_CSTDATOMIC_PRESENT)
class AtomicPointer {
private:
typedef internal::AtomicWord Rep;
Rep rep_;
std::atomic<void*> rep_;
public:
AtomicPointer() { }
explicit AtomicPointer(void* p) : rep_(reinterpret_cast<Rep>(p)) {}
explicit AtomicPointer(void* v) : rep_(v) { }
inline void* Acquire_Load() const {
return reinterpret_cast<void*>(internal::Acquire_Load(&rep_));
return rep_.load(std::memory_order_acquire);
}
inline void Release_Store(void* v) {
internal::Release_Store(&rep_, reinterpret_cast<Rep>(v));
rep_.store(v, std::memory_order_release);
}
inline void* NoBarrier_Load() const {
return reinterpret_cast<void*>(internal::NoBarrier_Load(&rep_));
return rep_.load(std::memory_order_relaxed);
}
inline void NoBarrier_Store(void* v) {
internal::NoBarrier_Store(&rep_, reinterpret_cast<Rep>(v));
rep_.store(v, std::memory_order_relaxed);
}
};
// We have neither MemoryBarrier(), nor <cstdatomic>
#else
#error Please implement AtomicPointer for this platform.
#endif
#undef LEVELDB_HAVE_MEMORY_BARRIER
#undef ARCH_CPU_X86_FAMILY
#undef ARCH_CPU_ARM_FAMILY
} // namespace leveldb::port
} // namespace leveldb
#endif // LEVELDB_CSTDATOMIC_PRESENT
#endif // PORT_ATOMIC_POINTER_H_

2
port/port_posix.h
Unescape View File

@ -97,7 +97,7 @@ inline bool Snappy_Uncompress(const char* input_data, size_t input_length,
::std::string* output) {
#ifdef SNAPPY
size_t ulength;
if (!snappy::GetUncompressedLength(input_data, ulength, &ulength)) {
if (!snappy::GetUncompressedLength(input_data, input_length, &ulength)) {
return false;
}
output->resize(ulength);

Write Preview
Loading…
Cancel
Save