You are not logged in.
Environment:
Thinkpad T61p Notebook with Nvidia Quadro FX 570M
Standard kernel with i3 as Window Manager, no login manager (i3 is started via startx)
Today I changed to nvidia-340xx as my card is no longer supported by 343.
Now the monitor detection is weird. It detects 3 Displays although there is only the LVDS one from the T61p. The other two are VGA and DVI and xrandr shows them as connected.
$ xrandr
Screen 0: minimum 8 x 8, current 3360 x 1050, maximum 8192 x 8192
VGA-0 connected primary 1680x1050+1680+0 (normal left inverted right x axis y axis) 331mm x 207mm
1680x1050 60.11*+ 50.09
LVDS-0 connected 1680x1050+0+0 (normal left inverted right x axis y axis) 331mm x 207mm
1680x1050 60.11*+ 50.09
DVI-D-0 connected (normal left inverted right x axis y axis)
1680x1050 60.11 + 50.09
As you see the Desktop is expanded to one of the "external" displays, too. I can move my mouse to the right out of the screen.
The main problem is that my external Monitor isn't detected anymore when I connect it via DVI. The output of xrandr doesn't change, there is nothing in dmesg and there is no output on the screen.
Downgrading to nvidia-340.32-2 (The one before the split) fixes the problem.
Did the configuration change inbetween nvidia-340.32-2 and nvidia-340xx-340.46-3 ? Do you have any suggestions how to fix that or is that a bug?
Thanks
reisub
Offline
I have essentially the same problem with an HP8540w with Nvidia Quadro FX 1800M. In my case two displays (CRT-0 and DFP-0) are detected and the virtual screen resolution is
expanded accordingly. As a consequence, lightDM is started in the wrong screen and log in is not possible without further tweaking.
Unfortunately I cannot downgrade the driver without downgrading the kernel, since the last version that I have is not compatible with the current kernel. As a temporary solution,
I disabled (ignore) the second monitor in the xorg.conf configuration file but again, external displays are no longer recognized. Since the situation is so similar to the one posted above,
I am inclined to believe it is a bug. Should we start complaining in the NVIDIA forums?
Offline
Reposted to nvidia forums: https://forums.geforce.com/default/topic/780903
I'm new to their forums, hopefully I chose the correct subforum. Legacy Hardware seemed to fit.
Offline
I'll chime in to report that I have the same issue. Thinkpad T61 with Nvidia Quadro NVS 140M. I get essentially the same result from xrandr that reisub posted above, only I had to include the display flag, ie:
$ xrandr -d :0
Any attempts to run xrandr without that flag result in an error like "Unable to open display." I don't recall the exact text, because I get a similar error when trying to run xset, nvidia-settings, or anything else that reports on an X display.
The problem I have as a result is that my window manager (openbox, via startx) is started on a screen twice as wide as it should be, and the only thing I can get to work is the logout script. I have no external monitor connected.
I have also posted on the Nvidia forums, but I would love to hear from any of the Arch veterans who have any questions or tips on resolving this issue.
edited: Videocard is an NVS, not FX. Whoops.
Last edited by bifkit (2014-10-10 22:53:19)
Offline
I have the same problem on ThinkPad R61 with Quadro NVS 140M. Problem occured after switching to nvidia-340xx package.
My laptop is not docked and no external monitors are connected, but in Xorg.0.log I found:
[ 76.529] (--) NVIDIA(0): IBM (CRT-0) (connected)
[ 76.529] (--) NVIDIA(0): TV-0
[ 76.529] (--) NVIDIA(0): IBM (DFP-0) (boot, connected)
[ 76.529] (--) NVIDIA(0): IBM (DFP-1) (connected)
...
[ 76.531] (II) NVIDIA(0): Validated MetaModes:
[ 76.531] (II) NVIDIA(0): "DFP-0:nvidia-auto-select,CRT-0:nvidia-auto-select"
[ 76.531] (II) NVIDIA(0): Virtual screen size determined to be 3360 x 1050
Offline
Can confirm this issue is also present on a Vaio S13S9 (GeForce 310M). 343xx no longer allowed me to boot into Gnome, installing 340xx solved the issue but it assumes I have another montior connected.
Offline
Can confirm this issue is also present on a Vaio S13S9 (GeForce 310M). 343xx no longer allowed me to boot into Gnome, installing 340xx solved the issue but it assumes I have another montior connected.
Offline
Same issue on a Lenovo T61. 340xx has issues detecting the monitor.
Offline
I have the same issue here : I just upgraded my NVidia driver to nvidia-340xx and Xorg detects 2 screens. I tried to reinstall the package using ABS but it didn't work.
(I disconnected the second screen using nvidia-settings here ...)
$ xrandr
Screen 0: minimum 8 x 8, current 1920 x 1080, maximum 8192 x 8192
VGA-0 connected primary (normal left inverted right x axis y axis)
1920x1080 59.93 + 40.09
LVDS-0 connected 1920x1080+0+0 (normal left inverted right x axis y axis) 345mm x 194mm
1920x1080 59.93*+ 40.09
My card is an NVS 5100M and I'm running Linux 3.16.4.
Last edited by Unda (2014-10-14 21:55:55)
Offline
Sony Vaio S11, Nvidia 310M. Same problem, but it shows a total of 4 monitors
$ xrandr
Screen 0: minimum 8 x 8, current 1366 x 768, maximum 8192 x 8192
VGA-0 connected (normal left inverted right x axis y axis)
1366x768 59.95 + 50.00
LVDS-0 connected primary 1366x768+0+0 (normal left inverted right x axis y axis) 290mm x 170mm
1366x768 59.95*+ 50.00
HDMI-0 connected (normal left inverted right x axis y axis)
1366x768 59.95 + 50.00
DVI-D-0 connected (normal left inverted right x axis y axis)
1366x768 59.95 + 50.00
Offline
Same in my case vaio and nvidia gt330m. Downgrading to 304xx fixes the issue
Offline
Hi,
Sony Vaio VPCS13C5E
NVIDIA Corporation GT218M [GeForce 310M]
same problem here with latest gentoo system, Kernel 3.14.14-gentoo and nvidia-drivers-340.46
=> screen resolution of attached external monitor isn't detected correctly, it assumes the "same" resolution as for the internal laptop screen (1366x768) instead of 1920x1080.
=> same problem on VGA and HDMI port...
Dualscreen isn't possible anymore...
workaround:
downgrade to nvidia-drivers-337.25 works like a charm...
Kontr-Olli
Last edited by kontr-olli (2014-10-20 19:17:31)
Offline
---
Last edited by septigation (2016-07-31 14:28:59)
Offline
Same problem here with a NVIDIA Corporation GT216M [NVS 5100M] (rev a2). Got 1 phantom monitor, sadly the gdm login screen goes to this screen... When I type in my password blindly, I can disable the phantom monitor in gnome settings, but it fails to enable second or third screen (my usual configuration is my laptop on a docking station with 2 monitors, disabling the internal notebook screen). I must admit that I did no system updates for a few months, but when upgrading packages I got the warning that my GPU is no longer supported and that I should switch to 340xx legacy branch. After doing this, I'm on 340.46-4 and got these problems....
Last edited by nnscr (2014-10-30 19:21:14)
Offline
Same problem here on a Sony Vaio VPCS137GG (01:00.0 VGA compatible controller: NVIDIA Corporation GT218M [GeForce 310M] (rev a2)).
Upgrading 343 broke after a system upgrade because my card is now in nvidia-340xx. Installed this driver, but xrandr is showing all 4 outputs connected simultaneously
I ended up going back to 340.32-1, however I also ended up having to go back to linux 3.16.3-1. The problem here is, I'm not sure how to work around this (I'm relatively new to Arch, but have used linux on and off for years) given Arch is a rolling release distro. I now want to install other packages (e.g. virtualbox), but I require linux-3.17.1-1.
I have the following in my pacman.conf:
IgnorePkg = linux nvidia nvidia-libgl nvidia-utils
My question is, is there a way I can upgrade linux to 3.17.1-1 (or whatever happens to be current at the time), and rebuild the nvidia driver against the new kernel version? Can I achieve this while maintaining visibility of the package in pacman?
Would I be on the right track if I start looking at DKMS? So, I remove my existing (older) nvidia-340xx package, install nvidia-340xx-dkms (using a snapshotted 340.32-1 package via pacman -U), remove linux from my IgnorePkg and upgrade it, and run dkms install nvidia/340.32 -k $(uname -r) ?
I'm hoping this means I can keep my 340xx on the older 340.32-1 version which is working, whilst upgrading everything else (kernel, etc) as per normal and just run dkms whenever my kernel is updated to rebuild the nvidia 340.32-1 module. Does this sound right?
I'm not sure how to grab an older version of an AUR package, so I'm guessing I Just grab the most recent tarball and change pkgver=340.46 to pkgver=340.42 in the PKGBUILD, makepkg and then pacman -U xxx.tar.gz. I guess I'd probably need to update the sha256sums, too.
(Sorry for the many questions -- just wanted to provide as much information as possible and see if I'm on the right track)
Many thanks
Last edited by syner (2014-11-01 12:44:41)
Offline
Same problem. Thinkpad T61 with Quadro NVS140M. https://bbs.archlinux.org/viewtopic.php … 7#p1471567
I could start graphical environment with external monitor today (by booting laptop without it and then plugging it in), but I don't know if it's anyway possible to configure it. As I wrote in my thread - image on external monitor was really bad, flickering and instead 1650x1080 resolution was set to 1440x900 (laptop screen resolution). I also couldn't configure it with nvidia-settings as monitor section was messed up.
Is it possible to configure everything manually and then use this configuration with 340xx? I have read somewhere that currently driver is not making so much use of xorg.conf anymore. Is that right?
Last edited by jakub (2014-11-01 12:27:41)
Offline
If this is a general problem consider opening a thread at: https://devtalk.nvidia.com/default/board/98/linux/
Offline
Thanks for the link. Problem described under https://devtalk.nvidia.com/default/topi … x-ubunt/2/ looks familiar...
Last edited by jakub (2014-11-01 17:06:21)
Offline
Ok, here's my workaround.. I kept my old nvidia-libgl and nvidia-utils (340.32-1), as this driver version was working prior to nvidia-340xx 340.46-4 (or downgrade as required with pacman -U)
I Removed my existing driver with
pacman -Rdd nvidia
I installed dkms with
pacman -S dkms
I grabbed the current nvidia-340xx-dkms.tar.gz and extracted with
tar xzvf nvidia-340xx-dkms.tar.gz
Within the extracted directory, I modified PKGBUILD, changing pkgver to 340.32 and pkgrel to 1 and renamed the dependency from nvidia-340xx-utils to nvidia-utils (old working 340.32-1 was named this way before it was split out into nvidia-340xx-utils):
pkgver=340.32
pkgrel=1
..
depends=('dkms' "nvidia-utils=${pkgver}")
Ran
updpkgsums
to update the checksums in PKGBUILD
Installed my modified nvidia-340xx-dkms package with
makepkg -i
Ran
dkms autoinstall -k $(uname -r)
Rebooted and confirmed nvidia driver was working against my current kernel version
Removed linux from IgnorePkg in my pacman.conf
Ran a
pacman -Syu
and updated to linux-3.17.1-1
Ran
dkms autoinstall -k 3.17.1-1-ARCH
Rebooted
It seems to be working fine..
$ modinfo nvidia
filename: /lib/modules/3.17.1-1-ARCH/kernel/drivers/video/nvidia.ko
alias: char-major-195-*
version: 340.32
Fingers crossed. Is anyone able to confirm this should be okay? I suppose I should enable the dkms systemd service so kernel modules are automatically rebuilt when the kernel is updated so I don't forget to do it manually.
Hopefully for now I can keep updating the rest of my system, particularly the kernel, whilst keeping the working nvidia driver in place without upgrading.. At least until it breaks
I'm hoping this will help others in the same situation who want to stick to the working driver while updating everything else, but I'd like someone to give the okay first. My main concern is that eventually the driver will break against newer kernels, and given it's a 'legacy' driver, nvidia might be reluctant to fix the issues being reported..
Cheers
Last edited by syner (2014-11-02 01:13:32)
Offline
Thanks syner. You are great.
Moving from your soution I further developed a new solution.
1) Install abs and copy the PKGBUILDs of the 340xx packages
sudo pacman -S abs
sudo abs
cd /var/abs
cp -R extra/nvidia-340xx extra/nvidia-340xx-utils multilib/lib32-nvidia-340xx-utils ~/your_folder
2) Install dkms
sudo pacman -S dkms
3) Modify the version of the pkgver to 340.32
pkgver=340.32
pkgrel=1
4) Update the sums
updpkgsums
5) Remove the old packages
sudo pacman -Rdd nvidia-304xx nvidia-304xx-utils nvidia-304xx-libgl opencl-nvidia-304xx lib32-nvidia-304xx-utils lib32-nvidia-304xx-libgl lib32-opencl-nvidia-304xx
6) Compile and install the "new" packages
cd ~/your_folder
cd nvidia-340xx-utils
makepkg -i
cd ..
cd nvidia-340xx
makepkg -i
cd..
cd lib32-nvidia-340xx-utils
makepkg -i
7) Reboot and everything should work
Regards
Gianluca
Offline
Thanks for the update!.. I thought there would be a proper way, and had no idea ABS was the solution.
I'm a little confused though -- doesn't nvidia-340xx-libgl need to be included in the above? Although I noticed the the description of this package is that it's a symbolic link, and it's not in /var/abs/extra.. Does it still need to be installed similar to what you have above, except change the version to 340.32-1?
I guess I'd like to make the install of the appropriate packages as clean as possible. My method seems a bit hacky, and I'm not sure what is/isn't required.. nvidia-libgl also doesn't exist in ABS.
Last edited by syner (2014-11-24 11:18:05)
Offline
I had the same problem, but I found a way to fix it without having to downgrade (I have Fedora but I guess it should work similarly.
I runned "nvidia-settings" as root, and here one can see that the phantom screens are defined (along with their positions, etc.). In "X Server Display Configuration", all monitors (three in my case) are listed in "Selection" (which contains also an option "X Screen 0") but it does not seem to be relevant. The idea is to select "Off" in the menu "Resolution" for all the screens but one (I kept the one marked as "primary display"). Then I clicked on "apply" and now my screen is normal again (with a hard wall on the right, etc.). It seems to also have removed the linking of the screen.
Offline
So you have now one working screen? What about two other screens?
Offline
Yes only one working screen, the others still exist in the configuration, but as they are disabled one can not interact with them (they appear as grey in the setting window).
After rebooting the configuration was reverted to the previous one, so I had to do the same thing to remove the screen. This time I pressed the button "save to X configuration file", so maybe it will remain after reboot.
Offline
Configuration will remain if you'll save it as xorg.conf in /etc/X11/ I guess. Anyway - it's not a workaround to disable additional screens in order to get rid of the problem
Offline