18:42imirkin: wow, that is an awesome effect (from the nv40 guy)
18:43imirkin: looks like the LUT is all messed up?
19:10Knaldgas: Lenovo X1 with NVidia eGPU and ext. monitor on DP. At X login, the external monitor is active and mouse can be moved to it. After X login, the ext. monitor goes blank. Using Debian 5.4.19-1, xserver-xorg-video-nouveau: 1:1.0.16-1, xorg.0.log at https://pastebin.com/35z4Zgnv : interesting line is likely: Failed to initialize GLX extension (Compatible NVIDIA X driver not found)
19:10Knaldgas: Any hints on how to fix this?
19:10Knaldgas: Btw. xrandr -q does not report any external display, lspci does see nvidia card
19:12Knaldgas: VGA compatible controller: NVIDIA... - Kernel driver in use: nouveau - Kernel modules: nouveau
19:27imirkin: Knaldgas: pastebin dmesg and xorg log
19:33Knaldgas: imirkin, dmesg: https://pastebin.com/MGrAfexS - xorg.0.log: https://pastebin.com/35z4Zgnv
19:46imirkin: Knaldgas: are you trying to use nouveau or nvidia?
19:46imirkin: [ 23.892] (II) Loading /usr/lib/xorg/modules/linux/libglx.so
19:46imirkin: [ 23.917] (II) Module glx: vendor="NVIDIA Corporation"
19:46imirkin: that can't be helping.
19:47Knaldgas: Two modules for the same task you mean?
19:47imirkin: you're using the libglx from nvidia rather than from the open-source collection.
19:47imirkin: nvidia = blob
19:47imirkin: nouveau = open source
19:47imirkin: anyways, nouveau appears to have loaded fine
19:47imirkin: however your issue is in Xorg land
19:49Knaldgas: Strange, aptitude does not show any installed nvidia-anything
19:50Knaldgas: Ahh, is that a mesa lib? - I have internal Intel GPU and Nvidia eGPU
19:50imirkin: you're using the nouveau kernel driver
19:50imirkin: which means you shouldn't have any trace of anything nvidia
19:51imirkin: (or conversely, flip things to use the nvidia kernel driver, in which case you're on your own)
19:51Knaldgas: I can't remember installing anything nvidia
19:51imirkin: ok, well, your xorg log begs to differ
19:51imirkin: NVIDIA Corporation is the vendor of your libglx
19:51Knaldgas: Well, xorg log has probably a better memory than me...
19:53imirkin: or not
19:53imirkin: [ 23.882] (==) Log file: "/var/log/Xorg.0.log", Time: Sat Sep 8 12:57:32 2018
19:53imirkin: perhaps the time is set a bit off on your machine too?
19:54Knaldgas: *slap self*
19:58Knaldgas: imirkin, https://pastebin.com/qmWXr3kj should be the right one - Sorry, didn't check file date, just assumed... :(
19:58Knaldgas: dmesg should be right though
19:58imirkin: yeah, systemd is annoying.
19:58imirkin: [ 21.263] (II) Module glx: vendor="X.Org Foundation"
19:58imirkin: much better :)
20:03imirkin: looks like all is well
20:03imirkin: you may be missing one final step
20:03imirkin: xrandr --setprovideoutputsource 1 0
20:05Knaldgas: Right, that changed things somewaht
20:05Knaldgas: Is that just a command I should fire when having this eGPU attached, or is it something else I should set?
20:09Knaldgas: imirkin, I will change my question: Is it ok to run "xrandr --setprovideroutputsource 1 0" when the eGPU is not attached. I assume it is ok
20:26imirkin: it's basically to set the secondary gpu as a outputsource provider for the main gpu
20:26imirkin: if there is no secondary gpu, no harm no foul
20:33Knaldgas: imirkin, Thanks a lot for your help! - I have some tearing when scrolling, but I'm finding that I should enable vsync - I'll probably manage that on my own, so all is good, thank again :)
20:33imirkin: tearing can't be helped with reverse prime
20:34imirkin: maybe with wayland compositors, but not with xorg's model, i think
20:35imirkin: at least not without introducing some kind of TearFree thing, and even there, i'm not sure.
20:35karolherbst: imirkin: it should be possible with Xorg as well, it just requires a good compositor
20:35imirkin: karolherbst: for reverse prime?
20:35imirkin: i don't think so.
20:35karolherbst: sure, why not?
20:35imirkin: because the copy is async
20:36imirkin: anyways, that's my recollection
20:36karolherbst: mhh, it might be that with X that's indeed more problemativ as we would have to fix X, not the compositor.. right
20:39karolherbst: but I thought other drivers were actually fixing it in the ddx or something... dunno. I only remember that somebody said that there are ways to workaround X limitations here
20:42imirkin: that's the TearFree thing
20:42imirkin: however i'm not sure how that interacts with reverse prime
20:43karolherbst: but yeah.. long term users have to move over to wayland to get good performance (when compositor start to render on the display local GPU instead of just one)
20:44karolherbst: and that's something I don't think one could fix in Xorg
20:44karolherbst: so maybe it really doesn't matter to care much about X here