Comment 36 for bug 285323

On further analysis, it's only the key release event that doesn't produce output. Key down generates a keycode press.

And I see what's going on in X - it's not that the device itself is stolen, just the focus. The mouse still works (you can click on the firefox icon and launch it, interactic with its buttons, etc.) but you just can't get the keyboard focus returned to it.

I'm not seeing the OCD displayed when gnome-power-manager running, but the issue clearly occurs only when g-p-m is running, so it's looking fairly solidly that g-p-m is taking the keyboard focus on the keydown brightness key events, and hanging on to it expecting a keyup event which never comes.

So I think the key up/down handling logic in g-p-m needs additional thought in light of situations like this, where the kernel isn't passing up the key-release events.

I'm unclear of the role the patch for 280646 is playing here. I *think* it is not causing the bug, but rather exposing the situation where the bug occurs. My guess is that the patch enables g-p-m to handle the key when previously it ignored it and let the hardware handle it.

In any case, I think we need g-p-m folks to analyze this more thoroughly. Ted?