Re: [PATCH] MSVC: Windows-native implementation for subset of Pthreads API

From: Nicolas Pitre <nico@fluxnic.net>
Date: Nov 5, 2009, 02:47 UTC
Message-ID: <alpine.LFD.2.00.0911042111270.10340@xanadu.home>
In-Reply-To: <4AF20534.2030004@gmail.com>

On Wed, 4 Nov 2009, Andrzej K. Haczewski wrote:

Show 8 quoted lines

> Erik Faye-Lund pisze:
> > Couldn't the windows version of pthread_create have a wrapper
> > function, that corrected the calling convention, much like the
> > function run_thread that start_async in run-command.c has?
> 
> Can't be done without allocations. I'd have to pass to that wrapping
> thread function an address of original function *and* an original
> argument, and there's no way to pack that as one void*.

What about:

typedef struct {
	HANDLE handle;
	void *(*start_routine)(void *);
	void *arg;
} pthread_t;

DWORD __stdcall windows_thread_start(LPVOID _self)
{
	pthread_t *self = _self;
	void *ret = self->start_routine(self->arg);
	return (DWORD)ret;
}

static inline int pthread_create(pthread_t *thread, const void *unused,
                                 void *(*start_routine)(void *), void *arg)
{
        thread->handle = CreateThread(NULL, 0, windows_thread_start, 
                                      thread, 0, NULL);
        [...]
}

Sure this will use 8 to 16 more bytes per thread, but we're dealing with a rather small number of threads anyway (more threads than the number of CPU cores is useless) making this extra memory usage rather insignificant compared to the many megabytes of RAM the rest of the code is using. The advantage is full compatibility with the native pthread interface git is using at the source level while still being much lighter than a full blown pthread implementation.

And thread creation is a relatively rare event compared to e.g. mutex lock/unlock, so the indirection shouldn't be noticeable. For the same reason, I also think that you could make pthread_create() and pthread_join() into a C file instead of being inlined which would reduce the code footprint at every call site, and allow for only one instance of windows_thread_start() which could then be made static.

Nicolas