Bug 1264782

This does not occur for XWayland clients: GEdit shows it when running under Wayland, but unsetting $WAYLAND_DISPLAY has it doing the right thing.

Could this be related?

https://bugzilla.gnome.org/show_bug.cgi?id=751835

I don't necessarily think so. Weston disabled the hardware cursor because of a kernel bug which meant that all cursor updates blocked for another vblank period, halving frame-rate BTW - it's never had difficulties with flickering like Mutter has.

In either case, it's not about the timing of when to draw the cursor: as you can see from the video, the cursor position jumps wildly within the box. So something else is much more seriously wrong with the cursor handling in that release.

I do still see some flickering on animated cursors after running a jhbuild from GNOME master everything, but certainly nothing like this. So I guess it's fixed upstream.

(In reply to Daniel Stone from comment #3)
> I don't necessarily think so. Weston disabled the hardware cursor because of
> a kernel bug which meant that all cursor updates blocked for another vblank
> period, halving frame-rate BTW - it's never had difficulties with flickering
> like Mutter has.

Yes, I agree, weston did not disable HW cursor for this, (which has been reverted recently, btw), but not using HW cursor in mutter did make the issue go away (or rather, hide the problem).

> In either case, it's not about the timing of when to draw the cursor: as you
> can see from the video, the cursor position jumps wildly within the box. So
> something else is much more seriously wrong with the cursor handling in that
> release.

I suspect this is because the cursor was refreshed continuously in mutter even when it did need to which can cause flickering on some hardware but not all, which was making the matter worse of these hardware. 

> I do still see some flickering on animated cursors after running a jhbuild
> from GNOME master everything, but certainly nothing like this. So I guess
> it's fixed upstream.

Jonas' series of patches for https://bugzilla.gnome.org/show_bug.cgi?id=744932 has landed upstream very recently, I reckon that's what helped here.

Ugh, actually I'm wrong. Running gnome-session directly from 'jhbuild run' didn't pick up a load of stuff from my session, seemingly. After running it from a proper GDM session, it's back.

For some reason, I just cannot reproduce anymore after the latest F23 update which included mesa (11.0.0-2.20150913), libwayland (1.9.0-1) and kernel (4.2.1-300) updates from today.

Would it be possible for you to update and retry? (I was pretty confident but now I can't reproduce anymore, that makes the investigation a tad more challenging...)

Pretty certain it's gtk+. 

Downgrading to 3.17.9-1.fc23 makes the issue reproducible at will.
Upgrading to 3.18.0-1.fc23 makes the issue go away.

Yeah, this is fixed in latest git, so let's call it done.