Hi pekka I test presentation on my side. The presentation results is not accurate because our pageflip event send last frame's timestamp every time. So weston_output_finish_frame send last frame's present tine to current frame's feedback. For this situation, I just delay one frame to send present event. Does this get the right c2p time? Thank you.
Best Regards Nancy 2018-05-02 18:17 GMT+08:00 zou lan <nancy.lan....@gmail.com>: > Hi pekka > > Thank you for your clearly analysis. I understand the presentation > protocol is enough for the requirement now. > > I used to not study the protocol deeply because the code is a little more > than just record the flip timestamp. The > test want to do this as simple as possible. > > I will try the presentation first. Thank you! > > Best Regards > Nancy > > 2018-05-02 16:59 GMT+08:00 Pekka Paalanen <ppaala...@gmail.com>: > >> On Wed, 2 May 2018 16:33:43 +0800 >> zou lan <nancy.lan....@gmail.com> wrote: >> >> > Add to wayland mail list. >> > >> > Hi pekka >> > >> > I used to consider the time line and presentation protocol. But they are >> > all focus on surface, not buffer. The test want to know the buffer >> latency, >> > although they are the same as surface's on >> > most cases. Applications think maybe some buffers doesn't display and >> used >> > again. >> > >> > By the way, could add a event to wl_buffer's protocol achieve the >> > requirement? Thank you. >> >> Hi Nancy, >> >> when a client creates a wp_presentation_feedback object, that object >> tracks the particular wl_surface.commit carrying an attached wl_buffer. >> So it is specific to the particular commit, not just to the surface >> or even to the buffer (buffers get re-used or could theoretically be >> used on multiple surfaces). >> >> Can you not have the client record both the current time at sending >> wl_surface.commit and the corresponding timestamp it gets from >> wp_presentation_feedback? Why would this not be enough? >> >> Please, see clients/presentation-shm.c in the Weston repository for an >> example that tracks this kind of presentation latencies. >> >> >> Thanks, >> pq >> >> > ---------- Forwarded message ---------- >> > From: Pekka Paalanen <ppaala...@gmail.com> >> > Date: 2018-05-02 15:48 GMT+08:00 >> > Subject: Re: unique id for wayland objects >> > To: zou lan <nancy.lan....@gmail.com> >> > >> > >> > On Wed, 2 May 2018 10:57:17 +0800 >> > zou lan <nancy.lan....@gmail.com> wrote: >> > >> > > Hi pekka >> > > >> > > I want to get the accurate buffer latency from a buffer commit by >> client >> > to >> > > display on the screen. The client can get the timestamp when buffer >> commit >> > > and release. So weston only need to give a >> > > timestamp when call on_pageflip. >> > > >> > > When do the real time analysis, there are many clients, I want to >> use one >> > > unique id to distinguish from each other quickly. >> > >> > Hi, >> > >> > is there any reason you don't use presentation_feedback interface? To >> > me it sounds like that would provide exactly what you need. >> > >> > https://cgit.freedesktop.org/wayland/wayland-protocols/ >> > tree/stable/presentation-time >> > >> > Also, could we have this coversation on the mailing list, please? >> > >> > >> > Thanks, >> > pq >> >> >
_______________________________________________ wayland-devel mailing list wayland-devel@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/wayland-devel
