[Mesa-dev] [PATCH 10/15] egl: add eglWaitSync

[Emil Velikov]

On 15/05/15 16:18, Marek Olšák wrote:
> On Fri, May 15, 2015 at 1:23 AM, Emil Velikov <emil.l.velikov at gmail.com> wrote:
>> On 12/05/15 22:54, Marek Olšák wrote:
>>> From: Marek Olšák <marek.olsak at amd.com>
>>>
>>> ---
>>>  src/egl/main/eglapi.c | 12 ++++++++++++
>>>  1 file changed, 12 insertions(+)
>>>
>>> diff --git a/src/egl/main/eglapi.c b/src/egl/main/eglapi.c
>>> index 60df297..544f7e4 100644
>>> --- a/src/egl/main/eglapi.c
>>> +++ b/src/egl/main/eglapi.c
>>> @@ -1162,6 +1162,7 @@ eglGetProcAddress(const char *procname)
>>>        { "eglDestroySync", (_EGLProc) eglDestroySync },
>>>        { "eglClientWaitSync", (_EGLProc) eglClientWaitSync },
>>>        { "eglDestroyImage", (_EGLProc) eglDestroyImage },
>>> +      { "eglWaitSync", (_EGLProc) eglWaitSync },
>>>  #endif /* _EGL_GET_CORE_ADDRESSES */
>>>  #ifdef EGL_MESA_drm_display
>>>        { "eglGetDRMDisplayMESA", (_EGLProc) eglGetDRMDisplayMESA },
>>> @@ -1514,6 +1515,17 @@ eglWaitSyncKHR(EGLDisplay dpy, EGLSync sync, EGLint flags)
>>>
>>>
>>>  EGLBoolean EGLAPIENTRY
>>> +eglWaitSync(EGLDisplay dpy, EGLSync sync, EGLint flags)
>>> +{
>>> +   /* The KHR version returns EGLint, while the core version returns
>>> +    * EGLBoolean. In both cases, the return values can only be EGL_FALSE and
>>> +    * EGL_TRUE.
>>> +    */
>>> +   return eglWaitSyncKHR(dpy, sync, flags);
>> Maybe add the cast, considering that sizeof(EGLenum) != sizeof(EGLInt)
>> on 64 bit systems.
>>
>> Thinking about the assembly - not sure if won't end up with pushing a
>> 64bit value, while a 32bit one being popped. My x86 assembly days were
>> mostly before x86_64 became a thing so ^^ might sound a bit stupid :)
> 
> It's EGLBoolean, not EGLenum. Also, no cast is needed between basic types.
> 
> typedef int32_t khronos_int32_t;
> typedef khronos_int32_t EGLint;
> typedef unsigned int EGLBoolean;
> 
> The types are equivalent on 32-bit and 64-bit, so I could have just
> made eglWaitSyncKHR be an alias of eglWaitSync and it would work, but
> I wanted to document this in the code.
> 
Had a bit of a brain fart. Pardon for the noise.

-Emil