Something that will be useful is if you capture a trace. Particularly please select "Input Latency" category. https://www.chromium.org/developers/how-tos/submitting-a-performance-bug


Also, do you know if restarting you machine updated chrome or were you using the same version before the restart?

I was using the same version of Chrome before the reboot, at least according to my local Apache log.  Web server access from localhost before the reboot had the same user agent as after the reboot: Chrome/53.0.2785.116

I've attached an "Input Latency" trace.  During this trace, I foregrounded a window that had the w3schools page mentioned above.  I moused over the image, saw that nothing happened, clicked, and saw that the mouseover event took effect.  I then moused away from the image, saw that nothing happened, clicked, and saw that the mouseout event took effect.

Let me know if this information is helpful or if I can provide anything else.  From my point of view interacting with it, it doesn't seem to be a latency issue per se.  The events never fire at all if I don't click.

Also, the same effect can be seen on the Chrome UI itself.  For example, when the cursor is the text cursor, I can mouse up to the reload button to the left of the URL bar, and the cursor stays the text cursor.

Deleted: trace_mouseover.json.gz 1.4 MB

trace_mouseover.json.gz 1.4 MB Download

Just to get some idea on your environment. Do you run the chrome with any particular flag in command line or do you have a particular flag on/off in chrome://flags? If so can you also try chrome://flags -> "restore all to default" option as well and see anything happens?

I don't use any command-line args.  Only this flag is non-default: enable-devtools-experiments

There's now an update pending, so I'll first try restarting Chrome to update.  If I still have issues, I will then try resetting flags to default.  Thanks!

Restarted Chrome to update from 53.0.2785.116 to 53.0.2785.143.  Issue remains.

Reset flags to default and restarted Chrome.  Issue remains.

Once in a while, I download a Chrome build manually for testing purposes.  The most recent I have is a build of Chrome 51.0.2704.84 that I downloaded on July 1.  I'm having the same issue there, which is surprising.

This makes me think that although the issue seems to be specific to Chrome (can't repro in Firefox or any other app), it's perhaps caused somehow by the Window manager or some other part of my Goobuntu system.  I would guess that Firefox has some other way of interacting with mouse events that makes it immune to whatever is wrong.

Are you using any special window manager or X server changes?

No, nothing special.  Standard Goobuntu config AFAIK, and I didn't change anything X-related on purpose.  Just rebooted to install some kernel updates.

I may try rebooting again later today to see if it was something ephemeral.  I'll update here with results if that solves it.

Since I see the problem even on very old versions of Chrome I have sitting around, I'm going to assume for now that it's not a Chrome bug and mark it WontFix.  If I find evidence to the contrary, we can always reopen.

Thanks for helping me debug!