Dear Maintainer,

After some updates I cannot use primusrun/optirun/pvkrun on my Sid:

```
$ primusrun glxgears
primus: fatal: Bumblebee daemon reported: error: [XORG] (EE) Unable to
locate/open config directory: "/etc/bumblebee/xorg.conf.d"
```

```
$ optirun glxgears
[  444.023511] [ERROR]Cannot access secondary GPU - error: [XORG] (EE)
Unable to locate/open config directory: "/etc/bumblebee/xorg.conf.d"

[  444.023570] [ERROR]Aborting because fallback start is disabled.
```

```
$ pvkrun vkcube
primus: fatal: Bumblebee daemon reported: error: [XORG] (EE) Unable to
locate/open config directory: "/etc/bumblebee/xorg.conf.d"
<hangs>
```

journalctl shows this:

```
$ sudo journalctl -f -u bumblebeed -n0
-- Logs begin at Mon 2020-06-29 19:49:28 EEST. --
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277010] [ERROR][XORG] (EE) Unable to locate/open config directory: "/etc/bumblebee/xorg.conf.d"
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277034] [WARN][XORG] (WW) Warning, couldn't open module mouse
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277039] [ERROR][XORG] (EE) Failed to load module "mouse" (module does not exist, 0)
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277045] [ERROR][XORG] (EE) NOUVEAU(0): [drm] failed to set drm interface version.
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277048] [ERROR][XORG] (EE) NOUVEAU(0): [drm] error opening the drm
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277051] [ERROR][XORG] (EE) NOUVEAU(0): 910:
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277054] [ERROR][XORG] (EE) Screen(s) found, but none have a usable configuration.
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277057] [ERROR][XORG] (EE)
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277060] [ERROR][XORG] (EE) no screens found(EE)
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277062] [ERROR][XORG] (EE)
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277068] [ERROR][XORG] (EE) Please also check the log file at "/var/log/Xorg.8.log" for additional information.
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277073] [ERROR][XORG] (EE)
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277077] [ERROR][XORG] (EE) Server terminated with error (1). Closing log file.
birž. 29 19:57:47 vinco bumblebeed[12788]: [  504.277728] [ERROR]X did not start properly
```

/etc/bumblebee/xorg.conf.d directory is empty :/

In bumblebee changelog I see:

   * Simplify rules and use bumblebee.install etc. for installation.
   * Remove obsolete conffile /etc/bumblebee/xorg.conf.d/10-dummy.conf.

Could these introduce some sort of regression in my case?

$ sudo strace -efile -f -p $(pgrep bumblebeed) 2>&1 | fgrep /etc/bumblebee
[pid  7068] execve("/usr/lib/xorg/Xorg", ["/usr/lib/xorg/Xorg", ":8", "-config",
"/etc/bumblebee/xorg.conf.nouveau", "-configdir", "/etc/bumblebee/xorg.conf.d", "-sharevts",
"-nolisten", "tcp", "-noreset", "-verbose", "3", "-isolateDevice", "PCI:01:00:0"], 0x7fff6ca8f198 /*
4 vars */ <unfinished ...>
[pid  7068] openat(AT_FDCWD, "/etc/bumblebee/xorg.conf.d",
O_RDONLY|O_NONBLOCK|O_CLOEXEC|O_DIRECTORY) = 16
[pid  7068] openat(AT_FDCWD, "/etc/bumblebee/xorg.conf.nouveau", O_RDONLY) = 16

I just this:

sudo apt purge --autoremove nvidia*
sudo apt install nvidia-driver bumblebee-nvidia primus-nvidia primus-vk-nvidia

And after reinstalling driver and all other bumbleblee-related packages, it started to work again.

Control: tag -1 moreinfo unreproducible

Thanks for this "solution". You probably can't tell what changed (likely
in /etc) after purge+reinstall?
Anyway, I'll keep this bug open as unreproducible in case someone else
stumbles upon it ...

Andreas

Sadly, no. I could have taken copy if /etc, just haven't thought of it, sorry.

All I see is that `/etc/bumblebee/xorg.conf.d/` is still empty as it where, can't comment any more.

OK so it looks like after 2020-09-01 updates that had lot's of NVIDIA stuff, same error reproduces again:

```
$ pvkrun vkcube
primus: fatal: Bumblebee daemon reported: error: [XORG] (EE) Unable to locate/open config directory:
"/etc/bumblebee/xorg.conf.d"
```

```
$ optirun glxgears
[ 5856.436157] [ERROR]Cannot access secondary GPU - error: [XORG] (EE) Unable to locate/open config directory:
"/etc/bumblebee/xorg.conf.d"

[ 5856.436198] [ERROR]Aborting because fallback start is disabled.
```


```
$ primusrun glxgears
primus: fatal: Bumblebee daemon reported: error: [XORG] (EE) Unable to locate/open config directory:
"/etc/bumblebee/xorg.conf.d
```

optirun/primusrun/pvkrun does not work. glxgears does run with integrated graphics, but not with NVIDIA.

Attached /etc/bumblebee directory contents.

Here's what /var/log/apt/history.log tells:

Start-Date: 2020-09-01  18:26:20
Commandline: apt full-upgrade
Requested-By: vincas (1000)
Upgrade: libmpc3:amd64 (1.2.0~rc1-1, 1.2.0-1), libnvidia-glvkspirv:amd64 (450.57-3, 450.66-1), libnvidia-glvkspirv:i386
(450.57-3, 450.66-1), libstdc++-10-dev:amd64 (10.2.0-5, 10.2.0-6), liblouisutdml-data:amd64 (2.8.0-4, 2.9.0-1),
libwayland-egl1:amd64 (1.18.0-2~exp1, 1.18.0-2~exp1.1), libwayland-egl1:i386 (1.18.0-2~exp1, 1.18.0-2~exp1.1),
mesa-opencl-icd:amd64 (20.1.5-1, 20.1.6-1), libgles-nvidia1:amd64 (450.57-3, 450.66-1), libgles-nvidia1:i386 (450.57-3,
450.66-1), libgles-nvidia2:amd64 (450.57-3, 450.66-1), libgles-nvidia2:i386 (450.57-3, 450.66-1), xserver-common:amd64
(2:1.20.8-2, 2:1.20.9-1), g++-10:amd64 (10.2.0-5, 10.2.0-6), libnvidia-ml1:amd64 (450.57-3, 450.66-1),
nvidia-vulkan-icd:amd64 (450.57-3, 450.66-1), nvidia-vulkan-icd:i386 (450.57-3, 450.66-1), xserver-xorg-core:amd64
(2:1.20.8-2, 2:1.20.9-1), nvidia-egl-icd:amd64 (450.57-3, 450.66-1), nvidia-egl-icd:i386 (450.57-3, 450.66-1),
libegl-mesa0:amd64 (20.1.5-1, 20.1.6-1), libegl-mesa0:i386 (20.1.5-1, 20.1.6-1), lib32stdc++6:amd64 (10.2.0-5,
10.2.0-6), libaprutil1:amd64 (1.6.1-4+b1, 1.6.1-5), gcc-10-base:amd64 (10.2.0-5, 10.2.0-6), gcc-10-base:i386 (10.2.0-5,
10.2.0-6), electrum:amd64 (4.0.2-1, 4.0.2-2), libobjc4:amd64 (10.2.0-5, 10.2.0-6), nvidia-driver:amd64 (450.57-3,
450.66-1), libglapi-mesa:amd64 (20.1.5-1, 20.1.6-1), libglapi-mesa:i386 (20.1.5-1, 20.1.6-1), mesa-common-dev:amd64
(20.1.5-1, 20.1.6-1), lib32gcc-s1:amd64 (10.2.0-5, 10.2.0-6), mesa-vulkan-drivers-dbgsym:amd64 (20.1.5-1, 20.1.6-1),
liblouis20:amd64 (3.14.0-1, 3.15.0-1), libitm1:amd64 (10.2.0-5, 10.2.0-6), nvidia-vulkan-common:amd64 (450.57-3,
450.66-1), xserver-xorg-legacy:amd64 (2:1.20.8-2, 2:1.20.9-1), libxatracker2:amd64 (20.1.5-1, 20.1.6-1),
python3-louis:amd64 (3.14.0-1, 3.15.0-1), nvidia-vdpau-driver:amd64 (450.57-3, 450.66-1), libgl1-nvidia-glvnd-glx:amd64
(450.57-3, 450.66-1), libgl1-nvidia-glvnd-glx:i386 (450.57-3, 450.66-1), libegl1-mesa:amd64 (20.1.5-1, 20.1.6-1),
cmake-data:amd64 (3.16.3-3, 3.18.2-1), libasan6:amd64 (10.2.0-5, 10.2.0-6), libglx-nvidia0:amd64 (450.57-3, 450.66-1),
libglx-nvidia0:i386 (450.57-3, 450.66-1), libquadmath0:amd64 (10.2.0-5, 10.2.0-6), orca:amd64 (3.36.5-1, 3.36.6-1),
rsyslog:amd64 (8.2006.0-2, 8.2008.0-1), nvidia-kernel-dkms:amd64 (450.57-3, 450.66-1), libgbm1:amd64 (20.1.5-1,
20.1.6-1), libgbm1:i386 (20.1.5-1, 20.1.6-1), libegl-nvidia0:amd64 (450.57-3, 450.66-1), libegl-nvidia0:i386 (450.57-3,
450.66-1), libwayland-client0:amd64 (1.18.0-2~exp1, 1.18.0-2~exp1.1), libwayland-client0:i386 (1.18.0-2~exp1,
1.18.0-2~exp1.1), nvidia-egl-common:amd64 (450.57-3, 450.66-1), libapr1:amd64 (1.7.0-1, 1.7.0-3), libgcc-10-dev:amd64
(10.2.0-5, 10.2.0-6), libllvm10:amd64 (1:10.0.1-4, 1:10.0.1-5), libllvm10:i386 (1:10.0.1-4, 1:10.0.1-5),
libnvidia-cfg1:amd64 (450.57-3, 450.66-1), libtsan0:amd64 (10.2.0-5, 10.2.0-6), nvidia-legacy-check:amd64 (450.57-3,
450.66-1), libubsan1:amd64 (10.2.0-5, 10.2.0-6), aspell-de-1901:amd64 (1:2-35, 1:2-36), liblouisutdml9:amd64 (2.8.0-4,
2.9.0-1), libgfortran5:amd64 (10.2.0-5, 10.2.0-6), kinfocenter:amd64 (4:5.17.5-2, 4:5.17.5-3),
nvidia-kernel-support:amd64 (450.57-3, 450.66-1), liblsan0:amd64 (10.2.0-5, 10.2.0-6), libgomp1:amd64 (10.2.0-5,
10.2.0-6), libgomp1:i386 (10.2.0-5, 10.2.0-6), nvidia-driver-libs:amd64 (450.57-3, 450.66-1), nvidia-driver-libs:i386
(450.57-3, 450.66-1), nvidia-driver-bin:amd64 (450.57-3, 450.66-1), libgl1-mesa-dev:amd64 (20.1.5-1, 20.1.6-1),
libgl1-mesa-dri:amd64 (20.1.5-1, 20.1.6-1), libgl1-mesa-dri:i386 (20.1.5-1, 20.1.6-1), liblouisutdml-bin:amd64 (2.8.0-4,
2.9.0-1), libosmesa6:amd64 (20.1.5-1, 20.1.6-1), libosmesa6:i386 (20.1.5-1, 20.1.6-1), libtss2-esys0:amd64 (2.4.1-1+b1,
3.0.0-1), libgcc-s1:amd64 (10.2.0-5, 10.2.0-6), libgcc-s1:i386 (10.2.0-5, 10.2.0-6), libgl1-mesa-glx:amd64 (20.1.5-1,
20.1.6-1), libgl1-mesa-glx:i386 (20.1.5-1, 20.1.6-1), xserver-xorg-video-nvidia:amd64 (450.57-3, 450.66-1), cmake:amd64
(3.16.3-3, 3.18.2-1), libclang-common-10-dev:amd64 (1:10.0.1-4, 1:10.0.1-5), liblouis-data:amd64 (3.14.0-1, 3.15.0-1),
python3-electrum:amd64 (4.0.2-1, 4.0.2-2), libnvidia-rtcore:amd64 (450.57-3, 450.66-1), cpp-10:amd64 (10.2.0-5,
10.2.0-6), mesa-vdpau-drivers:amd64 (20.1.5-1, 20.1.6-1), mesa-vdpau-drivers:i386 (20.1.5-1, 20.1.6-1),
mesa-vulkan-drivers:amd64 (20.1.5-1, 20.1.6-1), mesa-vulkan-drivers:i386 (20.1.5-1, 20.1.6-1), libatomic1:amd64
(10.2.0-5, 10.2.0-6), libatomic1:i386 (10.2.0-5, 10.2.0-6), libcc1-0:amd64 (10.2.0-5, 10.2.0-6), libstdc++6:amd64
(10.2.0-5, 10.2.0-6), libstdc++6:i386 (10.2.0-5, 10.2.0-6), libnvidia-cbl:amd64 (450.57-3, 450.66-1),
libhtml-parser-perl:amd64 (3.73-1, 3.75-1), libnvidia-glcore:amd64 (450.57-3, 450.66-1), libnvidia-glcore:i386
(450.57-3, 450.66-1), libwayland-server0:amd64 (1.18.0-2~exp1, 1.18.0-2~exp1.1), libwayland-server0:i386 (1.18.0-2~exp1,
1.18.0-2~exp1.1), gcc-10:amd64 (10.2.0-5, 10.2.0-6), libnvidia-eglcore:amd64 (450.57-3, 450.66-1),
libnvidia-eglcore:i386 (450.57-3, 450.66-1), mesa-va-drivers:amd64 (20.1.5-1, 20.1.6-1), mesa-va-drivers:i386 (20.1.5-1,
20.1.6-1), libglx-mesa0:amd64 (20.1.5-1, 20.1.6-1), libglx-mesa0:i386 (20.1.5-1, 20.1.6-1), nvidia-alternative:amd64
(450.57-3, 450.66-1), libwayland-cursor0:amd64 (1.18.0-2~exp1, 1.18.0-2~exp1.1), libwayland-cursor0:i386 (1.18.0-2~exp1,
1.18.0-2~exp1.1)
End-Date: 2020-09-01  18:28:17

Sadly, purge & reinstall did not help this time...

from `/var/log/Xorg.8.log`

[  2486.434] (II) NVIDIA dlloader X Driver  450.66  Wed Aug 12 19:44:12 UTC 2020
[  2486.434] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[  2486.435] (EE) No devices detected.
[  2486.435] (EE)
Fatal server error:
[  2486.435] (EE) no screens found(EE)
[  2486.435] (EE)

So maybe this latest 450.66 simply does not support my `GM107M [GeForce GTX 860M]`, or just because of some bug?

In https://wiki.debian.org/Bumblebee#Debian_10_and_older I've found this hint:

```
[ERROR]Cannot access secondary GPU - error: [XORG] (EE) No devices detected

You may have to set the BusID manually, in /etc/bumblebee/xorg.conf.nvidia. To get the BusID, run lspci | egrep 'VGA|3D'
in a terminal. Refer to the comments in that file for further instructions.
```

So I've edited `/etc/bumblebee/xorg.conf.nvidia` and uncommented one line so it's now:

```
BusID "PCI:01:00:0"
```

And it works!

`lspci | egrep 'VGA|3D'` on my machine returns:

```
00:02.0 VGA compatible controller: Intel Corporation 4th Gen Core Processor Integrated Graphics Controller (rev 06)
01:00.0 3D controller: NVIDIA Corporation GM107M [GeForce GTX 860M] (rev ff)
```

So some part fails to detect that only NVIDIA card..? :)

I've just upgraded from nvidia 450.57-2 to 450.66-1 and observe exactly
the same problem. In parallel, I also upgraded bumblebee from 3.2.1-25
to 3.2.1-26. And x-server from 2:1.20.8-2 to 2:1.20.9-1. Also my nvidia
card (NVIDIA Corporation GM206GLM [Quadro M2200 Mobile]) is now not
detected automatically anymore. Adding it manually in the config also
solves this problem for me.

I've experimented a bit with the involved packages and observed this:
- downgrade the nvidia-driver: still no auto-detect
- downgrade bumblebee: still no auto-detect
- downgrade the xserver: auto-detect works again
- upgrade nvidia-driver and bumblebee: auto-detect still works!
- upgrade xserver: auto-detect broken again

So I believe one can workaround this problem as well by doing:
apt install --only-upgrade xserver-common=2:1.20.8-2
xserver-xorg-legacy=2:1.20.8-2 xserver-xorg-core=2:1.20.8-2
xserver-xephyr=2:1.20.8-2 xwayland=2:1.20.8-2

Maybe this information helps someone to figure out what is wrong.

...
...

please try xserver-xorg-core 2:1.20.9-2 first and file a bug against it
if the issue persists.

Andreas

I've just upgraded to 2:1.20.9-2 and this seems to solve the problem for
me. So it seems one of these:
https://salsa.debian.org/xorg-team/xserver/xorg-server/-/commit/9c140c1c65aa05bd2e1b50cc8d6c8c4b9a1c1d9e
has solved this bug for me (which does not seem that unplausible).

Dear Maintainer,


        everything was working as a charm, until few days ago. I can't find
        what has changed but only the following packages has been updated
        between the last time primusrun worked out of the box and the
        failure encountered:
        - libproxy1v5:amd64 (0.4.15-5, 0.4.15-5+deb10u1)
        - libbrotli1:amd64 (1.0.7-2, 1.0.7-2+deb10u1)
        - libproxy1-plugin-gsettings:amd64 (0.4.15-5, 0.4.15-5+deb10u1)
        - libproxy-tools:amd64 (0.4.15-5, 0.4.15-5+deb10u1)
        - libproxy1-plugin-webkit:amd64 (0.4.15-5, 0.4.15-5+deb10u1)
        - libproxy1-plugin-networkmanager:amd64 (0.4.15-5, 0.4.15-5+deb10u1)

        After many tries, following many attempts to workaround
        including the upgrade from stable to backports for bumblebee and
        nvidia propietary drivers I found a post, a bit isolated:
https://github.com/Bumblebee-Project/Bumblebee/issues/652#issuecomment-123401024

        It talks about switching off the network as a workaround to get
        the secondary GPU driver loaded and used, no-one replied.

        It worked for me and I added the following piece of ugly code in script
        in order to use nvidia driver:

        (
            # TODO: manage restart of blutooth if needed
            nmcli radio all off
            while test $(lsmod | grep -c nvidia) -eq 0; do
                sleep 1
            done
            nmcli radio wifi on
        )&
        primusrun what_I_have_to_launch

        It is ugly, but it works (for now), and may be a clue for the
        resolution of the issue.

        I haven't tried from a dummy new user if the issue is still
        there, I haven't tried to upgrade to newer kernel. I've tried
        with nouveau and the issue is the same, but haven't tried yet
        with the network-off workaround.

        I don't have a github account so I can't post this upstream on
        bumblebee project.

        Thanks a lot for your concern, and best regards

        Sebastien MARQUE

I also had this issue on a fresh bullseye install.

I was able to get a different error message by commenting out the reference
to /etc/bumblebee/xorg.conf.d in /etc/bumblebee.conf, but then I get a
different error about a missing module 'mouse'.

Unfortunately I didn't have any time to troubleshoot any further and I
think there are other issues at play outside the scope of this bug (like
the fact that with the legacy Nvidia drivers there is no 'nvidia' module so
it's trying to use nouveau) so I'll probably have to try to solve all of
that before I can offer any more useful feedback on this bug.

I have the same issue with these versions:


bumblebee:  3.2.1-27
init-system-helpers:  1.60
libbsd0:  0.11.3-1
libc6:  2.31-12
libglib2:
libkmod2:  28-1
libx11-6:  2:1.7.1-1
lsb-base:  11.1.0
xserver-xorg-core:  2:1.20.11-1
bbswitch-dkms:  0.8-10
primus-libs:  0~20150328-13
bumblebee-nvidia:  3.2.1-27
nvidia-driver:  460.73.01-1
nvidia-driver-libs:  460.73.01-1
primus-libs:  0~20150328-13
xserver-xorg-video-nvidia:  460.73.01-1

Many thanks to Sébastien MARQUE for the suggestion `nmcli radio all
off`. This resolves the problem. For me, I don't need to wait for nvidia
to no longer be in lsmod before I can start it again.

This is with a Qualcomm Atheros QCA9377 802.11ac Wireless Network
Adapter (rev 31), Intel Corporation UHD Graphics 620 (rev 07), and
NVIDIA Corporation GP108M [GeForce MX150] (rev a1), on an up-to-date
Debian 10 using nouveau drivers if I'm not mistaken (I tried every
possible setup until something worked to run Parkitect a few weeks ago,
not sure what the current setup is, just that the nvidia-detect command
doesn't exist and I have to prefix primusrun and add -force-opengl).

If I can help by providing details about my setup or by trying
something, please let me know what commands to run -- via lgms.nl/email
since I expect I'll be receiving spam on the sending address before too
long which means I'll have to block it.

Best,
Luc

Hello,

Nice to see the hack could help someone.

BTW I "improved" the ugly hack this way:

 (
 path=/proc/driver/nvidia
 if ! test -e $path; then
 IFS=':' read -a radio <<< $(LANG=C nmcli -t radio)
 nmcli radio wifi off
 while ! test -e $path; do
 continue
 done
 test ${radio[1]} = 'enabled' && nmcli radio wifi on
 fi
)&

And it resides now on /usr/local/bin/primusrun (https://seb.lautre.net/git/seb/scripts/blob/master/primusrun (https://seb.lautre.net/git/seb/scripts/blob/master/primusrun))

Best regards,

Seb

I can confirm I have the same issue using a brandly new installed machine
with the nvidia driver.


primus: fatal: Bumblebee daemon reported: error: [XORG] (EE) Unable to
locate/open config directory: "/etc/bumblebee/xorg.conf.d"