On 14/12/16 06:58 PM, Carsten Haitzler (The Rasterman) wrote: > On Tue, 13 Dec 2016 11:54:32 -0800 Derek Foreman <der...@osg.samsung.com> > said: > > just a note. i know the original code was wrong and used unix time... unix > time > is gettimeofday and thus can go forward and back (eg ntp changes) and also > thus > its a huge number rather than ecore_time_get(). still it'll wrap at about 49.7 > days...
Ahh, I just assumed the original code used an apropriate efl time function. :/ I've updated to use ecore_time_get() - I still get a base time at the first use of a wayland pixmap because I notice ecore_time_get()'s start time is undefined. Thanks! > why not just keep a global serial # and ++ it each time? at least it'll wrap > only after 4 billion or so pixmap frees (that's going to be an exceedingly > long > time i would guess vs using ms)? isn't this just meant to be some unique > serial > number not a timestamp? technically this also means multiple pixmap frees in a > row there will get the same serial... The wayland spec for surface.frame states: "The callback_data passed in the callback is the current time, in milliseconds, with an undefined base." I'm actually pinging a few people about this since I'm pretty sure that's not what anyone meant, but it appears to be what we've implemented. Surely this should be the time that a frame submitted immediately will actually hit scanout (this can be checked with an asynchronous vblank wait ioctl). Will try to sort this out today, need to confer with some other wayland devs and maybe have the spec text clarified. Thanks, Derek >> derekf pushed a commit to branch master. >> >> http://git.enlightenment.org/core/enlightenment.git/commit/?id=0493abf457331fc604e2aa64eb51448fada09913 >> >> commit 0493abf457331fc604e2aa64eb51448fada09913 >> Author: Derek Foreman <der...@osg.samsung.com> >> Date: Tue Dec 13 13:53:18 2016 -0600 >> >> Fix wayland frame callback times >> >> ecore_time_unix_get() * 1000 is too big to fit in a uint32, so take the >> time we instantiate the wayland pixmap hash and use that as a base. >> --- >> src/bin/e_pixmap.c | 10 ++++++++-- >> 1 file changed, 8 insertions(+), 2 deletions(-) >> >> diff --git a/src/bin/e_pixmap.c b/src/bin/e_pixmap.c >> index f46a994..8e072a0 100644 >> --- a/src/bin/e_pixmap.c >> +++ b/src/bin/e_pixmap.c >> @@ -64,6 +64,8 @@ struct _E_Pixmap >> >> #ifdef HAVE_WAYLAND >> >> +double wayland_time_base; >> + >> static void >> _e_pixmap_cb_deferred_buffer_destroy(struct wl_listener *listener, void >> *data EINA_UNUSED) { >> @@ -342,7 +344,10 @@ e_pixmap_new(E_Pixmap_Type type, ...) >> } >> } >> else >> - pixmaps[type] = eina_hash_int64_new((Eina_Free_Cb)_e_pixmap_free); >> + { >> + pixmaps[type] = eina_hash_int64_new >> ((Eina_Free_Cb)_e_pixmap_free); >> + wayland_time_base = ecore_time_unix_get(); >> + } >> cp = _e_pixmap_new(type); >> cp->win = id; >> eina_hash_add(pixmaps[type], &id, cp); >> @@ -798,7 +803,8 @@ e_pixmap_image_clear(E_Pixmap *cp, Eina_Bool cache) >> cd->frames = NULL; >> EINA_LIST_FREE(free_list, cb) >> { >> - wl_callback_send_done(cb, ecore_time_unix_get() * 1000); >> + double t = ecore_time_unix_get() - wayland_time_base; >> + wl_callback_send_done(cb, t * 1000); >> wl_resource_destroy(cb); >> } >> } >> >> -- >> >> > > ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot _______________________________________________ enlightenment-devel mailing list enlightenment-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-devel
