Lets make a SDK 0.4 Linux nagging thread!

@lhl: I was briefly skimming the xorg.conf man page and realized that this was actually creating more questions for me. Mainly, I'm a bit confused by the "new" way of configuring X without an xorg.conf file (because it raises the question of how to replicate my current configuration when creating an xorg.conf file). So I though it wouldn't hurt to ask a question over at askubuntu. Let's see if one of the X experts can assist.

Actually jherico's idea of using nvidia-settings to generate a configuration sounds promising. Gotta try this later...

The idea behind xorg.conf-less setups is just that new versions of the X server generally auto-detect pretty much everything. If you want to use an unusual or complicated X setup, you just create an xorg.conf file as usual. In particular, using the nvidia proprietary driver requires an xorg.conf setup with at minimum a Monitor, Device, Screen, and ServerLayout section, where the Device specifies "nvidia" as the Driver.

As for the original question: multiple screens are quite easy. You'll just need multiple Screen sections describing each one of them, and in your ServerLayout section you can have things like:

Screen 0 "my_regular_screen" 0 0
Screen 1 "rift_screen" RightOf "my_regular_screen"

edit: also be warned, as the ancient gnu/linux adage goes: "ask on ubuntu forums, and you shall receive crap answers"

"nuclear" wrote:
In particular, using the nvidia proprietary driver requires an xorg.conf setup with at minimum a Monitor, Device, Screen, and ServerLayout section, where the Device specifies "nvidia" as the Driver.

I believe since a few years the preferred method is putting one Device section into /etc/X11/xorg.conf.d/20-nvidia.conf and that's it (https://wiki.archlinux.org/index.php/Nvidia#Minimal_configuration).

As for the second X server: I wouldn't want to have it running all the time. Maybe I'll try making a script to start stuff with a second X server on the rift and when the application exits, the x server will then be killed too.

In the meantime I'm using kde with kscreen. It saves all connected display configurations and switches automatically when one it knows is connected. Since the power button on the rift also disables its display input, that means I have my usual setup with my two monitors and when I press the button on the rift, it automatically powers off the two monitors and uses the rift only. It's not a perfect solution, but works quite nicely since I usually don't need to use the rift and my monitors at the same time.

"haagch" wrote:

"nuclear" wrote:
In particular, using the nvidia proprietary driver requires an xorg.conf setup with at minimum a Monitor, Device, Screen, and ServerLayout section, where the Device specifies "nvidia" as the Driver.

I believe since a few years the preferred method is putting one Device section into /etc/X11/xorg.conf.d/20-nvidia.conf and that's it (https://wiki.archlinux.org/index.php/Nvidia#Minimal_configuration).

There's absolutely nothing different with making a single /etc/X11/xorg.conf file, or multiple files in the /etc/X11/xorg.conf.d directory with different pieces of one (or both). They all just get processed together in the xserver config parser (hw/xfree86/parser/scan.c).

"jherico" wrote:

"lhl" wrote:
I just got a chance to try this out but can only attach to a 60 Hz screen and am getting judder. Are there docs on setting up the DK2 w/ a second X server instead as the README suggests? It looks like the wiki, along w/ a bunch of old Linux config writeups are gone. Does anyone have a copy of that old stuff/a recommendation on where to put new stuff?

Documentation on how to setup the DK2 as a second X screen would't actually be specific to the Rift. You should do a general google search on setting up X screens. With nVidia drivers and the 'nvidia-settings' utility you can actually just select the screen and then select 'new X screen' from a dropdown.

Ah, yeah, I was trying to do it the hard way and edit the xorg.conf - most of my Linux usage is headless or embedded. Using nvidia-settings worked. To run the ConfigUtil test, you have to specify the DISPLAY when launching (which puts the UI on the DK2). It's not juddering, but I notice a lot more jitter w/ the positional tracking than when I had tested it on OS X.

The OculusWorldDemo seem to run pretty rock solid though (75Hz, very accurate tracking). Anyway, looks good enough, time to write some code.

Thanks for the linux support :)

Is anyone able to run config utility or OculusWorldDemo on Debian jessie?
I can't even run config utility because I get sigsegv within a second :(

Below is the info from syslog

CameraThread[23256]: segfault at 7ff7335ba008 ip 00007ff731b044c1 sp 00007ff7335d9a68 error 4 in libc-2.19.so[7ff7319d5000+19f000]

"thorgh" wrote:
Is anyone able to run config utility or OculusWorldDemo on Debian jessie?
I can't even run config utility because I get sigsegv within a second :(

I'm running Debian sid (unstable) with the nvidia proprtietary drivers, and the config utility with its demo scene work fine. I didn't try the world demo yet.

Thanks, it turned out that there was a problem with my camera. After I replugged it everything started working (more or less).

Thanks for the support! :)