qemu-e2k/util/qemu-thread-win32.c
Paolo Bonzini 7c9b2bf677 qemu-thread: add a fast path to the Win32 QemuEvent
QemuEvents are used heavily by call_rcu.  We do not want them to be slow,
but the current implementation does a kernel call on every invocation
of qemu_event_* and won't cut it.

So, wrap a Win32 manual-reset event with a fast userspace path.  The
states and transitions are the same as for the futex and mutex/condvar
implementations, but the slow path is different of course.  The idea
is to reset the Win32 event lazily, as part of a test-reset-test-wait
sequence.  Such a sequence is, indeed, how QemuEvents are used by
RCU and other subsystems!

The patch includes a formal model of the algorithm.

Tested-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2015-09-24 20:52:28 +02:00

480 lines
12 KiB
C

/*
* Win32 implementation for mutex/cond/thread functions
*
* Copyright Red Hat, Inc. 2010
*
* Author:
* Paolo Bonzini <pbonzini@redhat.com>
*
* This work is licensed under the terms of the GNU GPL, version 2 or later.
* See the COPYING file in the top-level directory.
*
*/
#include "qemu-common.h"
#include "qemu/thread.h"
#include "qemu/notify.h"
#include <process.h>
#include <assert.h>
#include <limits.h>
static bool name_threads;
void qemu_thread_naming(bool enable)
{
/* But note we don't actually name them on Windows yet */
name_threads = enable;
fprintf(stderr, "qemu: thread naming not supported on this host\n");
}
static void error_exit(int err, const char *msg)
{
char *pstr;
FormatMessage(FORMAT_MESSAGE_FROM_SYSTEM | FORMAT_MESSAGE_ALLOCATE_BUFFER,
NULL, err, 0, (LPTSTR)&pstr, 2, NULL);
fprintf(stderr, "qemu: %s: %s\n", msg, pstr);
LocalFree(pstr);
abort();
}
void qemu_mutex_init(QemuMutex *mutex)
{
mutex->owner = 0;
InitializeCriticalSection(&mutex->lock);
}
void qemu_mutex_destroy(QemuMutex *mutex)
{
assert(mutex->owner == 0);
DeleteCriticalSection(&mutex->lock);
}
void qemu_mutex_lock(QemuMutex *mutex)
{
EnterCriticalSection(&mutex->lock);
/* Win32 CRITICAL_SECTIONs are recursive. Assert that we're not
* using them as such.
*/
assert(mutex->owner == 0);
mutex->owner = GetCurrentThreadId();
}
int qemu_mutex_trylock(QemuMutex *mutex)
{
int owned;
owned = TryEnterCriticalSection(&mutex->lock);
if (owned) {
assert(mutex->owner == 0);
mutex->owner = GetCurrentThreadId();
}
return !owned;
}
void qemu_mutex_unlock(QemuMutex *mutex)
{
assert(mutex->owner == GetCurrentThreadId());
mutex->owner = 0;
LeaveCriticalSection(&mutex->lock);
}
void qemu_cond_init(QemuCond *cond)
{
memset(cond, 0, sizeof(*cond));
cond->sema = CreateSemaphore(NULL, 0, LONG_MAX, NULL);
if (!cond->sema) {
error_exit(GetLastError(), __func__);
}
cond->continue_event = CreateEvent(NULL, /* security */
FALSE, /* auto-reset */
FALSE, /* not signaled */
NULL); /* name */
if (!cond->continue_event) {
error_exit(GetLastError(), __func__);
}
}
void qemu_cond_destroy(QemuCond *cond)
{
BOOL result;
result = CloseHandle(cond->continue_event);
if (!result) {
error_exit(GetLastError(), __func__);
}
cond->continue_event = 0;
result = CloseHandle(cond->sema);
if (!result) {
error_exit(GetLastError(), __func__);
}
cond->sema = 0;
}
void qemu_cond_signal(QemuCond *cond)
{
DWORD result;
/*
* Signal only when there are waiters. cond->waiters is
* incremented by pthread_cond_wait under the external lock,
* so we are safe about that.
*/
if (cond->waiters == 0) {
return;
}
/*
* Waiting threads decrement it outside the external lock, but
* only if another thread is executing pthread_cond_broadcast and
* has the mutex. So, it also cannot be decremented concurrently
* with this particular access.
*/
cond->target = cond->waiters - 1;
result = SignalObjectAndWait(cond->sema, cond->continue_event,
INFINITE, FALSE);
if (result == WAIT_ABANDONED || result == WAIT_FAILED) {
error_exit(GetLastError(), __func__);
}
}
void qemu_cond_broadcast(QemuCond *cond)
{
BOOLEAN result;
/*
* As in pthread_cond_signal, access to cond->waiters and
* cond->target is locked via the external mutex.
*/
if (cond->waiters == 0) {
return;
}
cond->target = 0;
result = ReleaseSemaphore(cond->sema, cond->waiters, NULL);
if (!result) {
error_exit(GetLastError(), __func__);
}
/*
* At this point all waiters continue. Each one takes its
* slice of the semaphore. Now it's our turn to wait: Since
* the external mutex is held, no thread can leave cond_wait,
* yet. For this reason, we can be sure that no thread gets
* a chance to eat *more* than one slice. OTOH, it means
* that the last waiter must send us a wake-up.
*/
WaitForSingleObject(cond->continue_event, INFINITE);
}
void qemu_cond_wait(QemuCond *cond, QemuMutex *mutex)
{
/*
* This access is protected under the mutex.
*/
cond->waiters++;
/*
* Unlock external mutex and wait for signal.
* NOTE: we've held mutex locked long enough to increment
* waiters count above, so there's no problem with
* leaving mutex unlocked before we wait on semaphore.
*/
qemu_mutex_unlock(mutex);
WaitForSingleObject(cond->sema, INFINITE);
/* Now waiters must rendez-vous with the signaling thread and
* let it continue. For cond_broadcast this has heavy contention
* and triggers thundering herd. So goes life.
*
* Decrease waiters count. The mutex is not taken, so we have
* to do this atomically.
*
* All waiters contend for the mutex at the end of this function
* until the signaling thread relinquishes it. To ensure
* each waiter consumes exactly one slice of the semaphore,
* the signaling thread stops until it is told by the last
* waiter that it can go on.
*/
if (InterlockedDecrement(&cond->waiters) == cond->target) {
SetEvent(cond->continue_event);
}
qemu_mutex_lock(mutex);
}
void qemu_sem_init(QemuSemaphore *sem, int init)
{
/* Manual reset. */
sem->sema = CreateSemaphore(NULL, init, LONG_MAX, NULL);
}
void qemu_sem_destroy(QemuSemaphore *sem)
{
CloseHandle(sem->sema);
}
void qemu_sem_post(QemuSemaphore *sem)
{
ReleaseSemaphore(sem->sema, 1, NULL);
}
int qemu_sem_timedwait(QemuSemaphore *sem, int ms)
{
int rc = WaitForSingleObject(sem->sema, ms);
if (rc == WAIT_OBJECT_0) {
return 0;
}
if (rc != WAIT_TIMEOUT) {
error_exit(GetLastError(), __func__);
}
return -1;
}
void qemu_sem_wait(QemuSemaphore *sem)
{
if (WaitForSingleObject(sem->sema, INFINITE) != WAIT_OBJECT_0) {
error_exit(GetLastError(), __func__);
}
}
/* Wrap a Win32 manual-reset event with a fast userspace path. The idea
* is to reset the Win32 event lazily, as part of a test-reset-test-wait
* sequence. Such a sequence is, indeed, how QemuEvents are used by
* RCU and other subsystems!
*
* Valid transitions:
* - free->set, when setting the event
* - busy->set, when setting the event, followed by futex_wake
* - set->free, when resetting the event
* - free->busy, when waiting
*
* set->busy does not happen (it can be observed from the outside but
* it really is set->free->busy).
*
* busy->free provably cannot happen; to enforce it, the set->free transition
* is done with an OR, which becomes a no-op if the event has concurrently
* transitioned to free or busy (and is faster than cmpxchg).
*/
#define EV_SET 0
#define EV_FREE 1
#define EV_BUSY -1
void qemu_event_init(QemuEvent *ev, bool init)
{
/* Manual reset. */
ev->event = CreateEvent(NULL, TRUE, TRUE, NULL);
ev->value = (init ? EV_SET : EV_FREE);
}
void qemu_event_destroy(QemuEvent *ev)
{
CloseHandle(ev->event);
}
void qemu_event_set(QemuEvent *ev)
{
if (atomic_mb_read(&ev->value) != EV_SET) {
if (atomic_xchg(&ev->value, EV_SET) == EV_BUSY) {
/* There were waiters, wake them up. */
SetEvent(ev->event);
}
}
}
void qemu_event_reset(QemuEvent *ev)
{
if (atomic_mb_read(&ev->value) == EV_SET) {
/* If there was a concurrent reset (or even reset+wait),
* do nothing. Otherwise change EV_SET->EV_FREE.
*/
atomic_or(&ev->value, EV_FREE);
}
}
void qemu_event_wait(QemuEvent *ev)
{
unsigned value;
value = atomic_mb_read(&ev->value);
if (value != EV_SET) {
if (value == EV_FREE) {
/* qemu_event_set is not yet going to call SetEvent, but we are
* going to do another check for EV_SET below when setting EV_BUSY.
* At that point it is safe to call WaitForSingleObject.
*/
ResetEvent(ev->event);
/* Tell qemu_event_set that there are waiters. No need to retry
* because there cannot be a concurent busy->free transition.
* After the CAS, the event will be either set or busy.
*/
if (atomic_cmpxchg(&ev->value, EV_FREE, EV_BUSY) == EV_SET) {
value = EV_SET;
} else {
value = EV_BUSY;
}
}
if (value == EV_BUSY) {
WaitForSingleObject(ev->event, INFINITE);
}
}
}
struct QemuThreadData {
/* Passed to win32_start_routine. */
void *(*start_routine)(void *);
void *arg;
short mode;
NotifierList exit;
/* Only used for joinable threads. */
bool exited;
void *ret;
CRITICAL_SECTION cs;
};
static bool atexit_registered;
static NotifierList main_thread_exit;
static __thread QemuThreadData *qemu_thread_data;
static void run_main_thread_exit(void)
{
notifier_list_notify(&main_thread_exit, NULL);
}
void qemu_thread_atexit_add(Notifier *notifier)
{
if (!qemu_thread_data) {
if (!atexit_registered) {
atexit_registered = true;
atexit(run_main_thread_exit);
}
notifier_list_add(&main_thread_exit, notifier);
} else {
notifier_list_add(&qemu_thread_data->exit, notifier);
}
}
void qemu_thread_atexit_remove(Notifier *notifier)
{
notifier_remove(notifier);
}
static unsigned __stdcall win32_start_routine(void *arg)
{
QemuThreadData *data = (QemuThreadData *) arg;
void *(*start_routine)(void *) = data->start_routine;
void *thread_arg = data->arg;
qemu_thread_data = data;
qemu_thread_exit(start_routine(thread_arg));
abort();
}
void qemu_thread_exit(void *arg)
{
QemuThreadData *data = qemu_thread_data;
notifier_list_notify(&data->exit, NULL);
if (data->mode == QEMU_THREAD_JOINABLE) {
data->ret = arg;
EnterCriticalSection(&data->cs);
data->exited = true;
LeaveCriticalSection(&data->cs);
} else {
g_free(data);
}
_endthreadex(0);
}
void *qemu_thread_join(QemuThread *thread)
{
QemuThreadData *data;
void *ret;
HANDLE handle;
data = thread->data;
if (data->mode == QEMU_THREAD_DETACHED) {
return NULL;
}
/*
* Because multiple copies of the QemuThread can exist via
* qemu_thread_get_self, we need to store a value that cannot
* leak there. The simplest, non racy way is to store the TID,
* discard the handle that _beginthreadex gives back, and
* get another copy of the handle here.
*/
handle = qemu_thread_get_handle(thread);
if (handle) {
WaitForSingleObject(handle, INFINITE);
CloseHandle(handle);
}
ret = data->ret;
DeleteCriticalSection(&data->cs);
g_free(data);
return ret;
}
void qemu_thread_create(QemuThread *thread, const char *name,
void *(*start_routine)(void *),
void *arg, int mode)
{
HANDLE hThread;
struct QemuThreadData *data;
data = g_malloc(sizeof *data);
data->start_routine = start_routine;
data->arg = arg;
data->mode = mode;
data->exited = false;
notifier_list_init(&data->exit);
if (data->mode != QEMU_THREAD_DETACHED) {
InitializeCriticalSection(&data->cs);
}
hThread = (HANDLE) _beginthreadex(NULL, 0, win32_start_routine,
data, 0, &thread->tid);
if (!hThread) {
error_exit(GetLastError(), __func__);
}
CloseHandle(hThread);
thread->data = data;
}
void qemu_thread_get_self(QemuThread *thread)
{
thread->data = qemu_thread_data;
thread->tid = GetCurrentThreadId();
}
HANDLE qemu_thread_get_handle(QemuThread *thread)
{
QemuThreadData *data;
HANDLE handle;
data = thread->data;
if (data->mode == QEMU_THREAD_DETACHED) {
return NULL;
}
EnterCriticalSection(&data->cs);
if (!data->exited) {
handle = OpenThread(SYNCHRONIZE | THREAD_SUSPEND_RESUME, FALSE,
thread->tid);
} else {
handle = NULL;
}
LeaveCriticalSection(&data->cs);
return handle;
}
bool qemu_thread_is_self(QemuThread *thread)
{
return GetCurrentThreadId() == thread->tid;
}