2001-04-26 16:45:43 +00:00
|
|
|
/*
|
|
|
|
SDL - Simple DirectMedia Layer
|
2011-02-11 22:37:15 -08:00
|
|
|
Copyright (C) 1997-2011 Sam Lantinga
|
2001-04-26 16:45:43 +00:00
|
|
|
|
|
|
|
This library is free software; you can redistribute it and/or
|
2006-02-01 06:32:25 +00:00
|
|
|
modify it under the terms of the GNU Lesser General Public
|
2001-04-26 16:45:43 +00:00
|
|
|
License as published by the Free Software Foundation; either
|
2006-02-01 06:32:25 +00:00
|
|
|
version 2.1 of the License, or (at your option) any later version.
|
2001-04-26 16:45:43 +00:00
|
|
|
|
|
|
|
This library is distributed in the hope that it will be useful,
|
|
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
2006-02-01 06:32:25 +00:00
|
|
|
Lesser General Public License for more details.
|
2001-04-26 16:45:43 +00:00
|
|
|
|
2006-02-01 06:32:25 +00:00
|
|
|
You should have received a copy of the GNU Lesser General Public
|
|
|
|
License along with this library; if not, write to the Free Software
|
|
|
|
Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
|
2001-04-26 16:45:43 +00:00
|
|
|
|
|
|
|
Sam Lantinga
|
2001-12-14 12:38:15 +00:00
|
|
|
slouken@libsdl.org
|
2001-04-26 16:45:43 +00:00
|
|
|
*/
|
2006-02-21 08:46:50 +00:00
|
|
|
#include "SDL_config.h"
|
2001-04-26 16:45:43 +00:00
|
|
|
|
|
|
|
/* Mutex functions using the Win32 API */
|
|
|
|
|
2011-01-24 21:20:30 -08:00
|
|
|
#include "../../core/windows/SDL_windows.h"
|
2001-04-26 16:45:43 +00:00
|
|
|
|
|
|
|
#include "SDL_mutex.h"
|
|
|
|
|
|
|
|
|
2006-07-10 21:04:37 +00:00
|
|
|
struct SDL_mutex
|
|
|
|
{
|
Fixed bug 1128
Patrick Baggett 2011-02-16 22:58:33 PST
This enhancement is for both x86/x64 Windows.
The SDL implementation of mutexes uses the Win32 API interprocess
synchronization primitive called a "Mutex". This implementation is subpar
because it has a much higher overhead than an intraprocess mutex. The exact
technical details are below, but my tests have shown that for reasonably high
contention (10 threads on 4 physical cores), it has 13x higher overhead than
the Win32 CriticalSection API.
If this enhancement is accepted, I will write a patch to implement SDL mutexes
using the critical section API, which should dramatically reduce overhead and
improve scalability.
-- Tech details --
Normally, Win32 Mutexes are used across process boundaries to synchronize
separate processes. In order to lock or unlock them, a user->kernel space
transition is necessary, even in the uncontented case on a single CPU machine.
Win32 CriticalSection objects can only be used within the same process virtual
address space and thus to lock one, does not require a user->kernel space
transition for the uncontended case, and additionally may spin a short while
before going into kernel wait. This small spin allows a thread to obtain the
lock if the mutex is released shortly after the thread starts spinning, in
effect bypassing the overhead of user->kernel space transition which has higher
overhead than the spinning itself.
2011-02-17 09:13:12 -08:00
|
|
|
CRITICAL_SECTION cs;
|
2001-04-26 16:45:43 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
/* Create a mutex */
|
2006-07-10 21:04:37 +00:00
|
|
|
SDL_mutex *
|
|
|
|
SDL_CreateMutex(void)
|
2001-04-26 16:45:43 +00:00
|
|
|
{
|
2006-07-10 21:04:37 +00:00
|
|
|
SDL_mutex *mutex;
|
Fixed bug 1128
Patrick Baggett 2011-02-16 22:58:33 PST
This enhancement is for both x86/x64 Windows.
The SDL implementation of mutexes uses the Win32 API interprocess
synchronization primitive called a "Mutex". This implementation is subpar
because it has a much higher overhead than an intraprocess mutex. The exact
technical details are below, but my tests have shown that for reasonably high
contention (10 threads on 4 physical cores), it has 13x higher overhead than
the Win32 CriticalSection API.
If this enhancement is accepted, I will write a patch to implement SDL mutexes
using the critical section API, which should dramatically reduce overhead and
improve scalability.
-- Tech details --
Normally, Win32 Mutexes are used across process boundaries to synchronize
separate processes. In order to lock or unlock them, a user->kernel space
transition is necessary, even in the uncontented case on a single CPU machine.
Win32 CriticalSection objects can only be used within the same process virtual
address space and thus to lock one, does not require a user->kernel space
transition for the uncontended case, and additionally may spin a short while
before going into kernel wait. This small spin allows a thread to obtain the
lock if the mutex is released shortly after the thread starts spinning, in
effect bypassing the overhead of user->kernel space transition which has higher
overhead than the spinning itself.
2011-02-17 09:13:12 -08:00
|
|
|
static DWORD (WINAPI*pf_SetCriticalSectionSpinCount)(LPCRITICAL_SECTION, DWORD) = NULL;
|
|
|
|
static HMODULE kernel32 = NULL;
|
|
|
|
|
|
|
|
/* One time logic - detect WinNT */
|
|
|
|
if(kernel32 == NULL) {
|
|
|
|
kernel32 = GetModuleHandleA("kernel32.dll");
|
|
|
|
if(kernel32) {
|
|
|
|
/* Attempt to resolve symbol -- Win9x gets NULL */
|
|
|
|
pf_SetCriticalSectionSpinCount = (DWORD (WINAPI*)(LPCRITICAL_SECTION, DWORD))GetProcAddress(kernel32, "SetCriticalSectionSpinCount");
|
|
|
|
}
|
|
|
|
else
|
|
|
|
kernel32 = (HMODULE)0x01; /* don't try to init again */
|
|
|
|
}
|
|
|
|
|
2001-04-26 16:45:43 +00:00
|
|
|
|
2006-07-10 21:04:37 +00:00
|
|
|
/* Allocate mutex memory */
|
|
|
|
mutex = (SDL_mutex *) SDL_malloc(sizeof(*mutex));
|
|
|
|
if (mutex) {
|
Fixed bug 1128
Patrick Baggett 2011-02-16 22:58:33 PST
This enhancement is for both x86/x64 Windows.
The SDL implementation of mutexes uses the Win32 API interprocess
synchronization primitive called a "Mutex". This implementation is subpar
because it has a much higher overhead than an intraprocess mutex. The exact
technical details are below, but my tests have shown that for reasonably high
contention (10 threads on 4 physical cores), it has 13x higher overhead than
the Win32 CriticalSection API.
If this enhancement is accepted, I will write a patch to implement SDL mutexes
using the critical section API, which should dramatically reduce overhead and
improve scalability.
-- Tech details --
Normally, Win32 Mutexes are used across process boundaries to synchronize
separate processes. In order to lock or unlock them, a user->kernel space
transition is necessary, even in the uncontented case on a single CPU machine.
Win32 CriticalSection objects can only be used within the same process virtual
address space and thus to lock one, does not require a user->kernel space
transition for the uncontended case, and additionally may spin a short while
before going into kernel wait. This small spin allows a thread to obtain the
lock if the mutex is released shortly after the thread starts spinning, in
effect bypassing the overhead of user->kernel space transition which has higher
overhead than the spinning itself.
2011-02-17 09:13:12 -08:00
|
|
|
/* Initialize */
|
|
|
|
InitializeCriticalSection(&mutex->cs);
|
|
|
|
|
|
|
|
/* On SMP systems, a non-zero spin count generally helps performance */
|
|
|
|
if(pf_SetCriticalSectionSpinCount) pf_SetCriticalSectionSpinCount(&mutex->cs, 2000);
|
2006-07-10 21:04:37 +00:00
|
|
|
} else {
|
|
|
|
SDL_OutOfMemory();
|
|
|
|
}
|
|
|
|
return (mutex);
|
2001-04-26 16:45:43 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Free the mutex */
|
2006-07-10 21:04:37 +00:00
|
|
|
void
|
|
|
|
SDL_DestroyMutex(SDL_mutex * mutex)
|
2001-04-26 16:45:43 +00:00
|
|
|
{
|
2006-07-10 21:04:37 +00:00
|
|
|
if (mutex) {
|
Fixed bug 1128
Patrick Baggett 2011-02-16 22:58:33 PST
This enhancement is for both x86/x64 Windows.
The SDL implementation of mutexes uses the Win32 API interprocess
synchronization primitive called a "Mutex". This implementation is subpar
because it has a much higher overhead than an intraprocess mutex. The exact
technical details are below, but my tests have shown that for reasonably high
contention (10 threads on 4 physical cores), it has 13x higher overhead than
the Win32 CriticalSection API.
If this enhancement is accepted, I will write a patch to implement SDL mutexes
using the critical section API, which should dramatically reduce overhead and
improve scalability.
-- Tech details --
Normally, Win32 Mutexes are used across process boundaries to synchronize
separate processes. In order to lock or unlock them, a user->kernel space
transition is necessary, even in the uncontented case on a single CPU machine.
Win32 CriticalSection objects can only be used within the same process virtual
address space and thus to lock one, does not require a user->kernel space
transition for the uncontended case, and additionally may spin a short while
before going into kernel wait. This small spin allows a thread to obtain the
lock if the mutex is released shortly after the thread starts spinning, in
effect bypassing the overhead of user->kernel space transition which has higher
overhead than the spinning itself.
2011-02-17 09:13:12 -08:00
|
|
|
DeleteCriticalSection(&mutex->cs);
|
2006-07-10 21:04:37 +00:00
|
|
|
SDL_free(mutex);
|
|
|
|
}
|
2001-04-26 16:45:43 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Lock the mutex */
|
2006-07-10 21:04:37 +00:00
|
|
|
int
|
|
|
|
SDL_mutexP(SDL_mutex * mutex)
|
2001-04-26 16:45:43 +00:00
|
|
|
{
|
2006-07-10 21:04:37 +00:00
|
|
|
if (mutex == NULL) {
|
|
|
|
SDL_SetError("Passed a NULL mutex");
|
|
|
|
return -1;
|
|
|
|
}
|
Fixed bug 1128
Patrick Baggett 2011-02-16 22:58:33 PST
This enhancement is for both x86/x64 Windows.
The SDL implementation of mutexes uses the Win32 API interprocess
synchronization primitive called a "Mutex". This implementation is subpar
because it has a much higher overhead than an intraprocess mutex. The exact
technical details are below, but my tests have shown that for reasonably high
contention (10 threads on 4 physical cores), it has 13x higher overhead than
the Win32 CriticalSection API.
If this enhancement is accepted, I will write a patch to implement SDL mutexes
using the critical section API, which should dramatically reduce overhead and
improve scalability.
-- Tech details --
Normally, Win32 Mutexes are used across process boundaries to synchronize
separate processes. In order to lock or unlock them, a user->kernel space
transition is necessary, even in the uncontented case on a single CPU machine.
Win32 CriticalSection objects can only be used within the same process virtual
address space and thus to lock one, does not require a user->kernel space
transition for the uncontended case, and additionally may spin a short while
before going into kernel wait. This small spin allows a thread to obtain the
lock if the mutex is released shortly after the thread starts spinning, in
effect bypassing the overhead of user->kernel space transition which has higher
overhead than the spinning itself.
2011-02-17 09:13:12 -08:00
|
|
|
|
|
|
|
EnterCriticalSection(&mutex->cs);
|
2006-07-10 21:04:37 +00:00
|
|
|
return (0);
|
2001-04-26 16:45:43 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* Unlock the mutex */
|
2006-07-10 21:04:37 +00:00
|
|
|
int
|
|
|
|
SDL_mutexV(SDL_mutex * mutex)
|
2001-04-26 16:45:43 +00:00
|
|
|
{
|
2006-07-10 21:04:37 +00:00
|
|
|
if (mutex == NULL) {
|
|
|
|
SDL_SetError("Passed a NULL mutex");
|
|
|
|
return -1;
|
|
|
|
}
|
Fixed bug 1128
Patrick Baggett 2011-02-16 22:58:33 PST
This enhancement is for both x86/x64 Windows.
The SDL implementation of mutexes uses the Win32 API interprocess
synchronization primitive called a "Mutex". This implementation is subpar
because it has a much higher overhead than an intraprocess mutex. The exact
technical details are below, but my tests have shown that for reasonably high
contention (10 threads on 4 physical cores), it has 13x higher overhead than
the Win32 CriticalSection API.
If this enhancement is accepted, I will write a patch to implement SDL mutexes
using the critical section API, which should dramatically reduce overhead and
improve scalability.
-- Tech details --
Normally, Win32 Mutexes are used across process boundaries to synchronize
separate processes. In order to lock or unlock them, a user->kernel space
transition is necessary, even in the uncontented case on a single CPU machine.
Win32 CriticalSection objects can only be used within the same process virtual
address space and thus to lock one, does not require a user->kernel space
transition for the uncontended case, and additionally may spin a short while
before going into kernel wait. This small spin allows a thread to obtain the
lock if the mutex is released shortly after the thread starts spinning, in
effect bypassing the overhead of user->kernel space transition which has higher
overhead than the spinning itself.
2011-02-17 09:13:12 -08:00
|
|
|
|
|
|
|
LeaveCriticalSection(&mutex->cs);
|
2006-07-10 21:04:37 +00:00
|
|
|
return (0);
|
2001-04-26 16:45:43 +00:00
|
|
|
}
|
2006-07-10 21:04:37 +00:00
|
|
|
|
|
|
|
/* vi: set ts=4 sw=4 expandtab: */
|