FS#67762 - [bumblebee] is not working after upgrading xorg-server to 1.20.9

Attached to Project: Arch Linux
Opened by vyacheslav (galdralag) - Sunday, 30 August 2020, 09:31 GMT
Last edited by Giancarlo Razzolini (grazzolini) - Monday, 23 November 2020, 11:36 GMT
Task Type Bug Report
Category Packages: Extra
Status Closed
Assigned To Laurent Carlier (lordheavy)
Giancarlo Razzolini (grazzolini)
Architecture All
Severity Low
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 3
Private No

Details

Description:
Applications does not start when using optirun with xorg-server 1.20.9.
With xorg-server 1.20.8 all ok.
Bumblebeed service starts his own X server and it is failed.

I attach xorg logs for working X server started by bumblebeed and not working.


Additional info:
* package version(s)
xorg-server 1.20.9

Steps to reproduce:
Run
optirun -vv wine cmd
when xorg-server upgraded to 1.20.9
   Working__1.20.8__Xorg.8.log (14.8 KiB)
   NotWorking__1.20.9__Xorg.8.log (5 KiB)
This task depends upon

Closed by  Giancarlo Razzolini (grazzolini)
Monday, 23 November 2020, 11:36 GMT
Reason for closing:  Fixed
Additional comments about closing:  I can confirm that xorg-1.20.9.r21.g5c400cae1-2 fixes it
Comment by Giancarlo Razzolini (grazzolini) - Monday, 31 August 2020, 19:57 GMT
I can confirm this. Not sure yet what the issue is at, but looks like the xorg.conf.nvidia file needs to be changed.
Comment by Philipp Leclercq (philippleclercq) - Saturday, 05 September 2020, 13:45 GMT
I have the same issue with intel-virtual-output using the setup described here: https://wiki.archlinux.org/index.php/Bumblebee#Output_wired_to_the_NVIDIA_chip

It was caused by the upgrade of xorg-server and xorg-server-common from 1.20.8-3 to 1.20.9-1 and persists in version 1.20.9-2.
   Xorg.8.log.success (36.2 KiB)
   Xorg.8.log.fail (4.7 KiB)
Comment by Giancarlo Razzolini (grazzolini) - Monday, 07 September 2020, 14:57 GMT
I tried to make changes to the nvidia xorg.conf but, so far, unsuccessful.
Comment by Gabriel Caffaratti (sxsoft) - Monday, 07 September 2020, 17:17 GMT
I can confirm it's not working for me either. I've also tried with different setups in bumblebee nvidia.xorg.conf but still not working. However nvidia-smi still detects the card. My xorg-server version is 1.20.9-2.
Comment by Craig (Hazey) - Thursday, 10 September 2020, 19:49 GMT
Same issue here as well - anyone narrow down the issue or find anything? Hopefully can get sorted pretty quick as it's impacting 2/2 of my laptops ha.

nvidia-smi detects them all good, looks like bumblebee/xorg just aren't seeing anything.

*Edit - BANDAID:*
Looks like manually specifying the bus ID in /etc/bumblebee/xorg.conf.nvidia file kicks things back into place, odd it's suddenly not being able to be detected so likely needs some digging but at least that will bandaid the problem.

Open /etc/bumblebee/xorg.conf.nvidia and uncommont the BusID line + ensure to specify the proper bus ID (from lspci)

Comment by Giancarlo Razzolini (grazzolini) - Thursday, 10 September 2020, 19:56 GMT
I was not able to diagnose this further, but the Xorg from bumblebee is not starting
Comment by Craig (Hazey) - Thursday, 10 September 2020, 20:00 GMT
Yea very weird, found a bandaid for time being - edited previous comment with the details so at least that gets things working again without downgrading.
Comment by loqs (loqs) - Thursday, 10 September 2020, 20:06 GMT
Could be the bus detection changes mentioned in  FS#67831  which have been reverted upstream.
Comment by Giancarlo Razzolini (grazzolini) - Thursday, 10 September 2020, 20:11 GMT
@loqs, looks it is precisely that. I tried with fixed busid here and it worked.
Comment by Philipp Leclercq (philippleclercq) - Friday, 11 September 2020, 07:37 GMT
Yep, the BusID fixes it for me as well. However, it's important to not just copy the BusID from lscpi, as that uses a different format (lspci says 01:00.0, xorg needs PCI:01:00:0). Thanks @Craig
Comment by Giancarlo Razzolini (grazzolini) - Friday, 11 September 2020, 11:39 GMT
The only difference is the dot at the end. But Xorg can also take the BusID like this: 1:0:0, for example. I guess now we have to wait the next Xorg to land.
Comment by vyacheslav (galdralag) - Monday, 23 November 2020, 11:05 GMT
It seems fixed in xorg-server 1.20.9.r21.g5c400cae1-2

Loading...