View Issue Details

Jump to NotesJump to History
IDProjectCategoryView StatusDate SubmittedLast Update
0000737channel: elrepo/el7bumblebeepublic2017-05-13 13:562021-04-16 01:29
Reporterwrthissell Assigned Totoracat  
PrioritynormalSeveritymajorReproducibilityalways
Status resolvedResolutionfixed 
PlatformLenovo Thinkpad P50OSCentos OS Version7.3
Summary0000737: Bumblebeed service error: rmmod: ERROR: Module nvidia is in use by: nvidia_modeset
DescriptionI am trying to get Bumblebee working on both Centos 7.3 and 6.9 on this computer for the first time. I had Bumblebee working on a Lenovo Thinkpad W541 in Centos 6.8, but I no longer have access to that computer.
Steps To Reproduce1. Follow the instructions on the kmod-nvidia and bumblee webpages with either
   A. The latest versions of the required packages on the repos
   B. Rebuilt packages from source rpms using the source kernel on the Product Build line
   C. kmod-nvidia and nvidia-x11-drv version 375.39-2, 375.66-1, 378.13 and 381.09
   D. Tried both the current drm rpms and rebuild latest drm rpms from the Fedora branch.
Additional InformationI tried to build a Bumblebee rpm from the development branch on github, but the computer would not load x11 after I installed that rpm. There are many similar reported issues on the github bumblebee site, with some "fixes" posted to the development branch, but I do not have the required skills to figure out what else to do. X11 currently loads with the configuration shown in the attached file. The screen text is extremely small, requiring a magnifying glass to read, when X11 does not load. The workaround I found when this happens is to yum remove kmod-nvidia, dracut -f, reboot, then remove the rest of the relevant rpms and try again with another build. This takes a lot of time, even with a nvme ssd.
TagsNo tags attached.
Attached Files
20170513 Bumblebee error.tar.xz (28,284 bytes)
20170517 Bumblebee kmod-bbswitch error.tar.xz (87,524 bytes)
20170607 bbswitch and bumblebee testing results.tar.xz (54,676 bytes)
Reported upstream

Relationships

related to 0000742 closedtoracat kmod-bbswitch and bumblebee not working on Lenovo Thinkpad P50 and Centos 6.9 kernel 2.6.32-696.1.1.el6.centos.plus.x86_64 

Activities

pperry

2017-05-13 14:20

administrator   ~0005183

This is the elrepo bug tracker, not a support forum, and you are not using any elrepo packages so there is no bug to report here.

I will give you a hint: try using 'modprobe -r' instead of rmmod to unload the nvidia module as modprobe will resolve module dependencies whereas rmmod does not.

wrthissell

2017-05-17 19:51

reporter   ~0005208

Dear Technical Support,
    I just uploaded a new file. I yum installed the elrepo and elpel packages to replace the rebuilt packages that I used to help me isolate the issue. The kmod-bbswitch elrepo package (0.8-3.el7.elrepo) does not install correctly on the current Centos 7.3 kernel. It does not load. Please refer to the attached 20170517 kmod-bbswitch_elrepo_install.txt file for documentation. I rebuilt it for the current Centos 7.3 kernel and it now works.
    I receive the same errors:

bbswitch: device 0000:01:00.0 is in use by driver 'nvidia', refusing OFF
rmmod: ERROR: Module nvidia is in use by: nvidia_modeset

    The github bumblebee issues webpage 0000762 describes some code fixes for this issue. I am insufficiently knowledgeable to incorporate these fixes into an update. I would greatly appreciate your assistance in a new package that resolves the issues I listed above.
    I am also having issues with getting bumblebee working on the Centos 6.9 boot on this computer and I shall now switch to booting to that OS and isolating, troubleshooting, and isolating the issue for reporting.
    Thank you in advance for your help.

toracat

2017-05-23 23:25

administrator   ~0005234

I built the package against -514.10.2 and compared the list of symbols used that are not on the kABI whitelist among -514, -514.10.2 and -514.16.1. They were identical.

        __dynamic_pr_debug
        __fentry__
        __pm_runtime_idle
        __pm_runtime_resume
        __stack_chk_fail
        acpi_bus_get_device
        acpi_format_exception
        acpi_get_name
        acpi_root_dir
        device_rh_alloc
        pci_read_config_dword
        pci_restore_state
        pci_save_state
        pci_set_power_state
        register_pm_notifier
        remove_proc_entry
        seq_lseek
        seq_read
        unregister_pm_notifier
        warn_slowpath_fmt

Next, I looked at the output from 'rpm -R'. -514 and -514.10.2 were different:

< kernel(acpi_bus_get_device) = 0x459997bb
---
> kernel(acpi_bus_get_device) = 0x03d75282

-514.10.2 and -514.16.1 were the same.

wrthissell

2017-05-24 17:13

reporter   ~0005235

Dear toracat,
   Thank you for working on this issue. Here is the URL you asked for in the related issue 0000742:

https://github.com/Bumblebee-Project/Bumblebee/pull/762/files [^]

https://github.com/Bumblebee-Project/Bumblebee/pull/762 [^]

   I apologize for being insufficiently skilled to properly understand the meaning of your note above. The tar file I attached has the rebuilt kmod-bbswitch rpm and source rpm for the current kernel-plus.

toracat

2017-06-01 18:23

administrator   ~0005245

I have rebuilt the kmod-bbswitch-0.8.x package against the latest kernel (3.10.0-514.21.1.el7) and released it to the elrepo-testing repository. This one is backward compatible with 3.10.0-514.16.1.el7 but not with -514.el7. Other kernel releases have not been tested. Please note that this is not from a developer branch.

kmod-bbswitch-0.8-4.el7.elrepo.x86_64.rpm
bbswitch-kmod-0.8-4.el7.elrepo.src.rpm

toracat

2017-06-04 23:43

administrator   ~0005250

I built the bumblebee package using the code from the development branch of the bumblebee git tree and released it to the elrepo-testing repository.

bumblebee-develop-20170130.el7.elrepo.x86_64.rpm
bumblebee-develop-20170130.el7.elrepo.src.rpm

Please test and report back with the result.

nanga

2017-06-07 13:46

reporter   ~0005255

Hi,

I'm having problems with bumblebeed. It says that "Module nvidia is in use by: nvidia_modeset".

Environment:
... CentOS 7.3.1611
... kmod-bbswitch-0.8-4.el7.elrepo.x86_64 from elrepo-testing
... bumblebee-3.2.1-10.el7.elrepo.x86_64 from elrepo-testing
... kernel-3.10.0-514.21.1.el7.x86_64 from base
... kmod-nvidia-375.66-1.el7.elrepo.x86_64 from elrepo
... kmod-nvidia-375.66-1.el7.elrepo.x86_64 from elrepo

Also nvidia's driver says that no card was detected. However, if I search for a card, I got a result:
# nvidia-debugdump --list
Found 1 NVIDIA devices
        Device ID: 0
        Device name: GeForce 910M
        GPU internal ID: GPU-0feb40fd-4440-16a4-dfd2-bb61948a8a7b

But Xorg is not happy:
[ 150.596] (II) xfree86: Adding drm device (/dev/dri/card0)
[ 150.596] (II) xfree86: Adding drm device (/dev/dri/card1)
[ 150.597] (--) PCI:*(0:0:2:0) 8086:1616:144d:c782 rev 9, Mem @ 0xf5000000/16777216, 0xd0000000/268435456, I/O @ 0x0000f000/64, BIOS @ 0x????????/131072
[ 150.597] (--) PCI: (0:3:0:0) 10de:129a:144d:c782 rev 161, Mem @ 0xf6000000/16777216, 0xe0000000/268435456, 0xf0000000/33554432, I/O @ 0x0000d000/128, BIOS @ 0x????????/524288
[ 150.597] (II) LoadModule: "glx"
[ 150.597] (II) Loading /usr/lib64/xorg/modules/extensions/nvidia/libglx.so
[ 150.602] (II) Module glx: vendor="NVIDIA Corporation"
[ 150.602] compiled for 4.0.2, module version = 1.0.0
[ 150.602] Module class: X.Org Server Extension
[ 150.602] (II) NVIDIA GLX Module 375.66 Mon May 1 14:28:39 PDT 2017
[ 150.602] (II) LoadModule: "nvidia"
[ 150.602] (II) Loading /usr/lib64/xorg/modules/drivers/nvidia_drv.so
[ 150.602] (II) Module nvidia: vendor="NVIDIA Corporation"
[ 150.602] compiled for 4.0.2, module version = 1.0.0
[ 150.602] Module class: X.Org Video Driver
[ 150.602] (II) NVIDIA dlloader X Driver 375.66 Mon May 1 14:03:26 PDT 2017
[ 150.602] (II) NVIDIA Unified Driver for all Supported NVIDIA GPUs
[ 150.602] (++) using VT number 2

[ 150.602] (EE) No devices detected.
[ 150.602] (EE)
Fatal server error:
[ 150.602] (EE) no screens found(EE)
[ 150.603] (EE)

wrthissell

2017-06-07 19:26

reporter   ~0005262

Dear Torcat,
    Thank you again for helping us resolve our issues in getting bumblebee to work with the latest or even recent kmod-nvidia drivers.
    I just uploaded a new tar.xz file with my test results from 20170603 with the new build of kmod-bbswitch and from 20170607 with the new build of bumblebee-develop.
    The new build of kmod-bbswitch for the new Centos 7.3 kernel appears to work. But bumblebee does not work. Please refer to the end of 20170603 dmesg.txt:

[wrthissell@LAPTOP-BKIJEPGK ~]$ sudo modprobe -r -f nvidia_drm nvidia_modeset nvidia_uvm
[wrthissell@LAPTOP-BKIJEPGK ~]$ sudo systemctl -l restart bumblebeed
[wrthissell@LAPTOP-BKIJEPGK ~]$ sudo systemctl -l status bumblebeed
● bumblebeed.service - Bumblebee C Daemon
   Loaded: loaded (/usr/lib/systemd/system/bumblebeed.service; enabled; vendor preset: disabled)
   Active: active (running) since Sun 2017-06-04 11:24:12 EDT; 4s ago
 Main PID: 17727 (bumblebeed)
   CGroup: /system.slice/bumblebeed.service
           └─17727 /sbin/bumblebeed

Jun 04 11:24:12 LAPTOP-BKIJEPGK systemd[1]: Started Bumblebee C Daemon.
Jun 04 11:24:12 LAPTOP-BKIJEPGK systemd[1]: Starting Bumblebee C Daemon...
Jun 04 11:24:12 LAPTOP-BKIJEPGK bumblebeed[17727]: [ 257.210614] [INFO]/sbin/bumblebeed 3.2.1 started
[wrthissell@LAPTOP-BKIJEPGK ~]$ optirun --debug glxgears
[ 294.714785] [DEBUG]optirun version 3.2.1 starting...
[ 294.714855] [DEBUG]Active configuration:
[ 294.714877] [DEBUG] bumblebeed config file: /etc/bumblebee/bumblebee.conf
[ 294.714898] [DEBUG] X display: :8
[ 294.714957] [DEBUG] LD_LIBRARY_PATH: /usr/lib64/nvidia:/usr/lib64/vdpau:/usr/lib/nvidia:/usr/lib/vdpau
[ 294.714980] [DEBUG] Socket path: /var/run/bumblebee.socket
[ 294.715022] [DEBUG] Accel/display bridge: auto
[ 294.715042] [DEBUG] VGL Compression: proxy
[ 294.715063] [DEBUG] VGLrun extra options:
[ 294.715084] [DEBUG] Primus LD Path: /usr/lib/primus:/usr/lib32/primus
[ 294.715218] [DEBUG]Using auto-detected bridge virtualgl
[ 298.965444] [INFO]Response: No - error: [XORG] (EE) /dev/dri/card1: failed to set DRM interface version 1.4: Permission denied

[ 298.965490] [ERROR]Cannot access secondary GPU - error: [XORG] (EE) /dev/dri/card1: failed to set DRM interface version 1.4: Permission denied

[ 298.965505] [DEBUG]Socket closed.
[ 298.965547] [ERROR]Aborting because fallback start is disabled.
[ 298.965559] [DEBUG]Killing all remaining processes.
[wrthissell@LAPTOP-BKIJEPGK ~]$

   Here are some relevant observations on the new test bumblebee-develop rpm build:

20170607 dmesg #2.txt:

[ 4.125039] Request for unknown module key 'The ELRepo Project (http://elrepo.org): ELRepo.org Secure Boot Key: f365ad3481a7b20e3427b61b2a26635b83fe427b' err -11

[ 4.144757] Request for unknown module key 'The ELRepo Project (http://elrepo.org): ELRepo.org Secure Boot Key: f365ad3481a7b20e3427b61b2a26635b83fe427b' err -11

Do the above lines refer to issues with the kmod_nvidia driver rpm?

The files 20170607 check bumblebee_develop rpm.txt and 20170607 SE_Linux Errors.txt ought to studied together.

bbswitch does not load with the new bumblebee develop rpm files, and bumbleebeed fails to load.

Could an issue exist with the kmod-nvidia rpm, SE_Linux, and the loading of nvidia_drm, nvidia_modeset, nvidia_uvm, and drm permissions?

toracat

2017-06-09 23:37

administrator   ~0005272

@wrthissell

You can ignore the "Secure Boot" lines unless your system actually has Secure Boot enabled.

About the selinux-related issues, please try the selinux-bumblebee package and see if they go away.

wrthissell

2017-06-11 18:16

reporter   ~0005275

Dear Toracat,
    Thank you again for helping us resolve the issue(s) with getting Bumblebee to work.
    I have bumblebee-selinux installed:

[wrthissell@LAPTOP-BKIJEPGK ~]$ yum list installed bumblebee-selinux
Loaded plugins: fastestmirror, langpacks, nvidia
Loading mirror speeds from cached hostfile
 * base: mirror.solarvps.com
 * centosplus: mirror.nodesdirect.com
 * elrepo: ord.mirror.rackspace.com
 * epel: mirror.cogentco.com
 * extras: mirror.umd.edu
 * nux-dextop: mirror.li.nux.ro
 * rpmforge: mirror.team-cymru.org
 * rpmfusion-free-updates: mirror.math.princeton.edu
 * rpmfusion-nonfree-updates: mirror.math.princeton.edu
 * updates: mirror.beyondhosting.net
Installed Packages
bumblebee-selinux.x86_64 1.0-1.el7.elrepo @elrepo

See this line in the yum.log that is in the latest tar file I sent you:

May 14 21:59:51 Updated: bumblebee-selinux-1.0-1.el7.elrepo.x86_64

All of the issues described in the latest tar file are with the above version installed.

I have currently disabled Secure Boot because I also have Centos 6.9 installed on this computer.

Issue History

Date Modified Username Field Change
2017-05-13 13:56 wrthissell New Issue
2017-05-13 13:56 wrthissell Status new => assigned
2017-05-13 13:56 wrthissell Assigned To => toracat
2017-05-13 13:56 wrthissell File Added: 20170513 Bumblebee error.tar.xz
2017-05-13 14:20 pperry Note Added: 0005183
2017-05-17 19:37 wrthissell File Added: 20170517 Bumblebee kmod-bbswitch error.tar.xz
2017-05-17 19:51 wrthissell Note Added: 0005208
2017-05-20 00:49 toracat Relationship added related to 0000742
2017-05-23 23:25 toracat Note Added: 0005234
2017-05-24 17:13 wrthissell Note Added: 0005235
2017-06-01 18:23 toracat Note Added: 0005245
2017-06-04 23:43 toracat Note Added: 0005250
2017-06-04 23:43 toracat Status assigned => feedback
2017-06-07 13:46 nanga Note Added: 0005255
2017-06-07 19:09 wrthissell File Added: 20170607 bbswitch and bumblebee testing results.tar.xz
2017-06-07 19:26 wrthissell Note Added: 0005262
2017-06-09 23:37 toracat Note Added: 0005272
2017-06-11 18:16 wrthissell Note Added: 0005275
2021-04-16 01:29 toracat Status feedback => resolved
2021-04-16 01:29 toracat Resolution open => fixed