Comment 18 for bug 1523088

I can confirm this, with the following setup:

- Linux 4.3.0 (vanilla source tree, built with Debian's make-kpkg, kernel config derived from Debian distro kernels)
- Debian testing (stretch), amd64
- Intel(R) Core(TM) i5-6600K CPU @ 3.50GHz
- Dual-monitor setup, one connected directly to DVI, the other attached via DisplayPort<->DVI adapter

It seems I can reliably trigger the WARN_ON(!wm_changed) condition by switching between the monitors.

Besides the backtrace-spam in my log, the system seems to run fine, no hangs, crashes, or the like (running Xorg from stretch with i3-wm).

I notice scrolling-related artifacts in conkeror, but that may be a totally unrelated issue.