Created attachment 150221 [details] kernel panic Apologies if the title is misleading, there is some magic going on behind the scene that I am unaware of. Problem: When "switching users" in gnome3, Xorg will crash and a kernel panic will occur. I suspect gnome is spawning a second X server. Operating System: Fedora 20 x86_64 Hardware: X1 Carbon 2nd Generation / 2014 Thinkpad One Link Pro Dock Physical Displays eDP1 connected 2560x1440+0+0 (laptop LCD) DP3 connected primary 1680x1050+2560+0 (Acer 22" Wide) DP4 connected 1440x900+4240+0 (Dell 19" Wide) Reproduction 1: Intel Driver: xorg-x11-drv-intel-2.99.916-1 (I compiled from the latest source code) Kernel: 3.17.0-0.rc4.git1.1.fc22.x86_64 1) Lock Screen 2) Select "Switch User" on Lock Screen 3) Kernel Panic / X crash Reproduction 2 (some improvement with latest git kernel): Intel Driver: xorg-x11-drv-intel-2.99.916-1 (I compiled from the latest source code) Kernel: 3.17.0-rc4git+ (1536340e7c67694b134cbbc07168e5a524e49d08) 1) Lock Screen 2) Select "Switch User" on Lock Screen 3) Log in as new user 4) Do stuff 5) Logout 6) Kernel Panic / X Crash Attached is the kernel panic trace. Please let me know what additional information I can include and if there is anything I can do further to help.
I forgot to mention I am running Gnome 3.12 (due to better support for HiDPI displays).
Chris, any ideas?
BUG_ON(vma->pin_count == 0) in i915_gem_object_ggtt_unpin().
It either begins with the erroneous modeset on the stale connector, or before when the MST connector is unplugged. I expect that drm.debug=0x6 would show when that framebuffer first disappears.
Created attachment 150451 [details] kernel crash with drm.debug=0x6
I booted with drm.debug=0x6. I found it was a little bit harder to reproduce the bug, but switching users a couple of times triggered it. I have attached the log. I debated whether or not to compress the log, I hope the uncompressed version is acceptable.
Uncompressed version is not only acceptable, it's preferrable. Thanks. Please attach the dmesg all the way from boot to ensure we're seeing the first reported problem.
Timeout, closing. Please feel free to reopen with results on latest kernels. Thanks.