[HarfBuzz] [PATCH] Implement threading primitives on Windows; fix compile error and leak

[Behdad Esfahbod]

Hi Bradley,

Thanks for the patches.

On 04/25/11 13:01, Bradley Grainger wrote:
> 0001: This fixes a compile error with MS Visual C++ 2010,

Fixed already.


> 0002: This adds a "hb_mutex_free" macro that is called by hb_blob_destroy  (to free the mutex object created in "hb_blob_create"). If GLib is available, this is defined as g_static_mutex_free.
> This fixes a bug that leaks GStaticMutex objects; from http://developer.gnome.org/glib/unstable/glib-Threads.html#g-static-mutex-free: "if you have a GStaticMutex as a member of a structure and the structure is freed, you should also free the GStaticMutex."

Thanks.  Fixed.


> 0003: This implements the HB threading primitives for MSVC on Windows. The "hb_atomic_int_*" macros are defined to use various interlocked intrinsics supported by the compiler. The "hb_mutex_*" macros are implemented with some new functions that delegate to Initialize/Enter/DeleteCriticalSection. 
> Note that "#include <Windows.h>" is not allowed in hb-private.hh because hb-open-type-private.hh (which includes it) redefines "LONG", "ULONG" and other standard Win32 data types with different meanings. (This naming conflict is rather unfortunate and makes it a little more difficult to port the library to Windows.)

I didn't know that Windows.h defines LONG, etc.  That's unfortunate.  I've
committed your patch for now, but I'm going to find a better way to solve this
problem.  For starters, HB_MUTEX_INIT is not supposed to call a function, and
I'm surprised that it compiles with the MS compiler.

I'll figure out a way to include Windows.h without conflicting with our
hb-open-type.h types and will commit.  Please test afterward.

Thanks again,
behdad


> Bradley
> 
> 
> 
> _______________________________________________
> HarfBuzz mailing list
> HarfBuzz at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/harfbuzz