On Wed, 23 Oct 2019 14:12:03 +0200 Böszörményi Zoltán <zbos...@pr.hu> wrote:
> 2019. 10. 23. 9:42 keltezéssel, Pekka Paalanen írta: > > On Tue, 22 Oct 2019 17:50:21 +0200 > > Böszörményi Zoltán <zbos...@pr.hu> wrote: > > > >> Hi, > >> > >> I have the below configuration for an Intel based POS system that, > >> while advertises 3 outputs (DP1, VGA1 and HDMI1 with xf86-video-intel), > >> only two are usable. DP1 for the built-in touchscreen and VGA1 for > >> the external VGA connector. > >> > >> I wanted to use an USB DisplayLink device as the 3rd output, with all > >> three output using its own Screen number, i.e. :0.0 :0.1 and :0.2. > > > > ... > > > >> The third observation is that while I am using this configuration below, > >> so the UDL device should be assigned to :0.2 (and active!), it is really > >> assigned to :0[.0] as an inactive output. See that there's no "*" indicator > >> set for any of the supported modes on DVI-I-1-1. > >> > >> How can I set up 3 different Screens correctly for 3 separate fullscreen > >> applications? > >> > >> I am using Xorg 1.20.4 patched with the "autobind GPUs to the screen" > >> patch from Dave Airlie that at least wakes up the UDL device and makes > >> it visible without extra magic with providers/sinks. > > > > Hi, > > > > for your specific use case, auto-bind is exactly what you do not want. > > So drop the patch or (since the patch is in upstream master already) > > use the option it adds to stop auto-binding. > > With Option "AutoBindGPU" "false" in effect (equivalent of backing the > patch out) the UDL device does not get assigned to ANY of the screens. > > I want it to have its own :0.2 bit that doesn't happen. Yes, that's another problem. Autobind=false is a step in the right direction, but apparently not sufficient. Thanks, pq
pgpSeRr6s7Lvh.pgp
Description: OpenPGP digital signature
_______________________________________________ xorg@lists.x.org: X.Org support Archives: http://lists.freedesktop.org/archives/xorg Info: https://lists.x.org/mailman/listinfo/xorg Your subscription address: %(user_address)s
