Created attachment 122353 [details] intel_reg dump --all output, before plugging external monitor in On a thinkpad t450s (broadwell), I can plug in an external monitor and it works great. But when I unplug it, then the most common outcome is that I drop into a weird crash state: - I get dropped to a text console on vt1, which functions fine -- I can log in, collect debugging information, etc. - the X server itself is still running -- e.g. I can do "DISPLAY=:0 xrandr" and it happily reports back that the external monitor is disconnected and that the built-in panel is configured and running. - applications connected to the X server in general are still running (e.g. they show up on 'ps', clementine continues to play music out the speakers, etc.) - exception: there's a line in dmesg saying "traps: gnome-shell[1172] trap int3 ip:7f31e41a887b sp:7ffe72d29a70 error:0", so I guess gnome-shell has probably crashed? - aside from this line about gnome-shell, there's nothing at all reported in dmesg -- no BUG or backtrace or anything like that - there's nothing at all reported in Xorg.0.log - if I then press alt-f2 to switch back to vt2, where the X server is running, then the screen goes blank and input locks up. - except, if I press control-alt-del to reboot, then systemd successfully sees the control-alt-del, and successfully prints some stuff on the screen, so I guess it's really showing vt2 in text mode? This is using Debian testing's userspace (xserver-xorg-core 1.18.1, xserver-xorg-video-intel 2.99.917+git20160307-2), and I get the same behavior with both Debian's default 4.3 kernel and with a 4.5.0rc7 from Debian experimental. ~$ uname -a Linux branna 4.5.0-rc7-amd64 #1 SMP Debian 4.5~rc7-1~exp1 (2016-03-07) x86_64 GNU/Linux The external monitor is an HDMI cable attached to an HDMI-to-miniDP dongle, plugged into the miniDP port on the laptop. This is extremely frustrating (basically makes my big work monitor unusable) and extremely reproducible, so if you have any debugging suggestions I'm very motivated to try :-). (Difficulty level: I've written (small) kernel and Xorg patches before; I'd be debugging it myself already if I had any idea where to start :-).)
Created attachment 122354 [details] intel_reg dump --all output, with external monitor plugged in
Created attachment 122355 [details] intel_reg dump --all output, after unplugging external monitor (run from text VT; X is wedged)
Created attachment 122356 [details] xrandr --verbose output, with external monitor plugged in
(I'd add a vbios dump, but following the instructions at https://01.org/linuxgraphics/documentation/development/how-dump-video-bios gives me: cat: '/sys/devices/pci0000:00/0000:00:02.0/rom': Input/output error )
Created attachment 122357 [details] xrandr --verbose output, after unplugging external monitor (run from text VT; X is wedged)
Created attachment 122358 [details] dmesg when running with drm.debug=0xe and performing plug/unplug/wedge cycle
Small correction: I said in the original message that after getting dropped to vt1, then attempting to switch back to vt2 makes the screen go blank. This is not true; in fact, the screen doesn't change at all, except that the blinking cursor disappears.
Created attachment 122359 [details] Xorg.log
Just dredged up some cables and monitors, and can now confirm that I see exactly the same problems regardless of whether I'm connecting via DisplayPort or HDMI, and also on multiple different monitors.
There are two similar bugs in the Debian bug tracker: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=817195 https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=818384
The upgrade to 2.99.917+git20160325 in Debian unstable fixed this issue for me.
Can confirm that that upgrade fixed it for me too, thanks!
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.