Open WLCIssuesBot opened 7 years ago
Comment by Cloudef Wednesday Jan 04, 2017 at 08:34 GMT
In the wlc sources I do see some drmDropMaster in sensible places, but looking with gdb it doesn't seem to be ever called - not even in the 'successful' case where I go via vt2.
This sounds interesting. AFAIK the calls should always be before TTY switch signal is sent to kernel.
Comment by raboof Wednesday Jan 04, 2017 at 09:01 GMT
I did some more digging but don't have final conclusions yet.
I'm using logind, wlc correctly sees that and goes into the logind code instead of the tty code.
Looking at the logind code (and learning about VT switching :) ) I see some code that looks like it should do the right thing, but poking around with gdb/printf statements reveals in my case this code doesn't seem to be called at all.
I'm still kind of confused what's going on, I wrote up my observations and asked for help at https://plus.google.com/110534994903711886928/posts/GgVKKtU5d1u
Issue by raboof Thursday Dec 29, 2016 at 16:45 GMT Originally opened as https://github.com/Cloudef/wlc/issues/217
When running a wlc-based session on vt1 and x11 on vt7, using nouveau as the display driver, switching from vt1 to vt7 crashes x11. The Xorg.log reveals:
It seems only one program can be DRM master at a time, and the wlc-based session hasn't given up its control before nouveau tried to get it. Indeed, when I switch from vt1 to vt2 (giving it time to release DRM master) and then to vt7, x11 does not crash.
In the wlc sources I do see some
drmDropMaster
in sensible places, but looking with gdb it doesn't seem to be ever called - not even in the 'successful' case where I go via vt2.(I originally encountered this issue with way-cooler (https://github.com/Immington-Industries/way-cooler/issues/190) but can also reproduce it with orbment, leading to think this is a wlc issue)