Ziele Untersuchung
mit Columbo Integrität von
Datenbanken Interaktion und
Portierbarkeit Ergonomie der
Schnittstellen

Angebot Produkte Projekt Beratung

Mittel Analytik Modellierung Sprachen Algebra Logik Hardware Denken Kreativität

Zusammenhänge Gesellschaft Wirtschaft Branche Firma


products/Sources/formale Sprachen/GAP/src/hpc/ (Algebra von RWTH Aachen Version 4.15.1^©) Datei vom 18.9.2025 mit Größe 6 kB

Quelle tls.h Sprache: C

/****************************************************************************
**
**  This file is part of GAP, a system for computational discrete algebra.
**
**  Copyright of GAP belongs to its developers, whose names are too numerous
**  to list here. Please refer to the COPYRIGHT file for details.
**
**  SPDX-License-Identifier: GPL-2.0-or-later
*/

#ifndef GAP_TLS_H
#define GAP_TLS_H

#include "../system.h"

#if !defined(HPCGAP)
#error This header is only meant to be used with HPC-GAP
#endif

// This contains our implementation of thread-local storage (TLS). TLS
// is a feature that is only to varying degrees supported by compilers
// and operating systems. As it is critical for the performance of
// HPC-GAP, we allow the implementation to choose between various
// options.
//
// By default, we use a low-level stack-based approach. Each thread
// stack is allocated on a memory segment whose boundaries are a
// multiple of a fixed power of 2 (see tlsconfig.h). The pages with
// the lowest address in that memory range are reserved for TLS,
// the rest for the stack itself. This allows us to get the base
// address for TLS by taking the stack or frame pointer (or any address
// on the stack) and masking out the lowest bits. This is very fast
// and on many platforms actually faster than global variables.
//
// However, this option only works when GAP is run as an application,
// as it needs to control the creation of threads in order to force
// thread stacks on the required boundaries. When used as a library,
// we need a native implementation. These can be chosen by passing
// the --enable-native-tls option to the configure script.
//
// By default, native TLS support will default to declaring TLS as
// storage type __thread. This is simple, but inefficient on some
// platforms. In particular, Cygwin and macOS require a function call
// to a dynamically loaded library to access __thread storage, which
// furthermore inhibits compiler optimizations.
//
// For these platforms, we use a different approach. As a basic
// alternative, on platforms that support it, we use
// pthread_getspecific() to get the base address of TLS and we
// define pthread_getspecific() as a pure function. This is still
// not faster than __thread by itself, but declaring
// pthread_getspecific() as pure allows compiler optimizations such
// as common subexpression elimination and code hoisting.
//
// As a further improvement, on macOS (and in the future, on Cygwin),
// we use an assembly-based alternative implementation of
// pthread_getspecific(). This assembly code can be treated like
// a normal variable access. In order to check that this is safe,
// we verify both during configuration and at runtime that the
// assembly code for pthread_getspecific() is implemented in a way
// consistent with such code.

#include "tlsconfig.h"

#include <setjmp.h>

typedef struct Region Region;

typedef struct ThreadLocalStorage
{
  int threadID;
  void *threadLock;
  void *threadSignal;
  void *acquiredMonitor;
  unsigned multiplexRandomSeed;
  Region * currentRegion;
  Region * threadRegion;
  Obj threadObject;
  Obj tlRecords;
  Obj lockStack;
  int lockStackPointer;
  Obj copiedObjs;
  Obj interruptHandlers;
  void *CurrentHashLock;
  int DisableGuards;

  // From intrprtr.c
  UInt PeriodicCheckCount;

  // From read.c
  jmp_buf threadExit;

  // Profiling
  UInt CountActive;
  UInt LocksAcquired;
  UInt LocksContended;
} ThreadLocalStorage;

#ifdef USE_NATIVE_TLS
extern __thread ThreadLocalStorage *TLSInstance;
#endif

#ifdef USE_PTHREAD_TLS

#include <pthread.h>

#ifdef HAVE_FUNC_ATTRIBUTE_PURE
#define PURE_FUNC __attribute__((pure))
#else
#define PURE_FUNC
#endif

#ifdef HAVE_FUNC_ATTRIBUTE_CONSTRUCTOR
#define CONSTRUCTOR_FUNC __attribute__((constructor))
#else
#define CONSTRUCTOR_FUNC
#endif

#ifdef ALLOW_PURE_PTHREAD_GETSPECIFIC
// pthread_getspecific() is not defined as pure by default; redeclaring
// it as such works for gcc/clang and allows the compiler to hoist calls
// to pthread_getspecific() out of loops and perform constant
// subexpression elimination on expressions containing TLS values.
PURE_FUNC void * pthread_getspecific(pthread_key_t);
#endif

#ifdef USE_MACOS_PTHREAD_TLS_ASM
static UInt TLSOffsetCopy;
UInt GetTLSOffset(void);
#else
static pthread_key_t TLSKeyCopy;
pthread_key_t GetTLSKey(void);
#endif

// This constructor function will initialize a static variable to
// address thread-local storage via pthread_getspecific() for each C
// module (TLSOffset or TLSKeyCopy, depending on how thread-local
// storage is being addressed). Making the variable static (instead of
// having it global and initialized once) allows optimizations that a
// global non-static variable would not permit. On some architectures,
// it is also faster to access a static than a global non-static
// variable, avoiding one level of indirection.
//
// Constructor functions are executed upon program start/library load.
#ifdef HAVE_FUNC_ATTRIBUTE_CONSTRUCTOR
#ifdef USE_MACOS_PTHREAD_TLS_ASM
CONSTRUCTOR_FUNC static void InitTLSOffset()
{
    TLSOffsetCopy = GetTLSOffset();
}
#else
CONSTRUCTOR_FUNC static void InitTLSKey()
{
    TLSKeyCopy = GetTLSKey();
}
#endif // USE_MACOS_PTHREAD_TLS_ASM
#endif // HAVE_FUNC_ATTRIBUTE_CONSTRUCTOR
#endif // USE_PTHREAD_TLS

#if defined(VERBOSE_GUARDS)
static inline ThreadLocalStorage * GetTLS(void)
#else
static ALWAYS_INLINE ThreadLocalStorage * GetTLS(void)
#endif
{
#if defined(USE_NATIVE_TLS)
    return TLSInstance;
#elif defined(USE_PTHREAD_TLS)
#ifdef USE_MACOS_PTHREAD_TLS_ASM
    long   offset;
    void * p;
    offset = (long)TLSOffsetCopy;
    // The following inline assembly code is the same as that in
    // pthread_getspecific(). That pthread_getspecific() is actually
    // implemented in that way has been verified in the configure
    // script and resulted in defining USE_MACOS_PTHREAD_TLS_ASM.
    // If that test fails, we fall back to the standard implementation
    // of pthread_getspecific().
    asm("movq %%gs:(%1), %0"
        : "=r"(p)  // output = %0
        : "r"(offset) // input = %1
        : /* clobber = none */);
    return (ThreadLocalStorage *)p;
#else
#ifdef HAVE_FUNC_ATTRIBUTE_CONSTRUCTOR
    return (ThreadLocalStorage *)pthread_getspecific(TLSKeyCopy);
#else
    return (ThreadLocalStorage *)pthread_getspecific(GetTLSKey());
#endif
#endif // USE_MACOS_PTHREAD_TLS_ASM
#else
    // use stack mask approach
    void * stack;
#ifdef __GNUC__
    stack = __builtin_frame_address(0);
#else
    int dummy[1];
    stack = dummy;
#endif
    return (ThreadLocalStorage *)(((uintptr_t)stack) & TLS_MASK);
#endif // USE_NATIVE_TLS, USE_PTHREAD_TLS
}

// Convenience helper for accessing TLS members
#define TLS(x) GetTLS()->x

void InitializeTLS(void);

#endif // GAP_TLS_H

quality96%

¤ Dauer der Verarbeitung: 0.39 Sekunden (vorverarbeitet) ¤

Wurzel

Suchen

Beweissystem der NASA

Beweissystem Isabelle

NIST Cobol Testsuite

Cephes Mathematical Library

Wiener Entwicklungsmethode

Haftungshinweis

Die Informationen auf dieser Webseite wurden nach bestem Wissen sorgfältig zusammengestellt. Es wird jedoch weder Vollständigkeit, noch Richtigkeit, noch Qualität der bereit gestellten Informationen zugesichert.