Bug 21537 - Update request: nvidia-current 384.59 (new long lived branch)
Summary: Update request: nvidia-current 384.59 (new long lived branch)
Status: RESOLVED FIXED
Alias: None
Product: Mageia
Classification: Unclassified
Component: RPM Packages (show other bugs)
Version: 6
Hardware: All Linux
Priority: Normal normal
Target Milestone: ---
Assignee: QA Team
QA Contact:
URL:
Whiteboard: advisory MGA5TOO MGA5-32-OK MGA6-64-O...
Keywords: validated_update
Depends on:
Blocks:
 
Reported: 2017-08-15 09:49 CEST by Thomas Backlund
Modified: 2017-08-21 13:32 CEST (History)
4 users (show)

See Also:
Source RPM: nvidia-current
CVE:
Status comment:


Attachments

Description Thomas Backlund 2017-08-15 09:49:49 CEST
Advisory:
nvidia-current update to new R384 long lived branch

This nvidia-current update provides the new R384 long lived branch,
currently version 384.59, adding support for new hardware and fixes
various bugs.

References:
http://www.nvidia.com/Download/driverResults.aspx/120917/en-us

Mga6:

SRPMS:
ldetect-lst-0.3.7.1-1.mga6.src.rpm

nvidia-current-384.59-1.mga6.nonfree.src.rpm


i586:
ldetect-lst-0.3.7.1-1.mga6.i586.rpm
ldetect-lst-devel-0.3.7.1-1.mga6.i586.rpm

dkms-nvidia-current-384.59-1.mga6.nonfree.i586.rpm
nvidia-current-cuda-opencl-384.59-1.mga6.nonfree.i586.rpm
nvidia-current-devel-384.59-1.mga6.nonfree.i586.rpm
nvidia-current-doc-html-384.59-1.mga6.nonfree.i586.rpm
x11-driver-video-nvidia-current-384.59-1.mga6.nonfree.i586.rpm


x86_64:
ldetect-lst-0.3.7.1-1.mga6.x86_64.rpm
ldetect-lst-devel-0.3.7.1-1.mga6.x86_64.rpm

dkms-nvidia-current-384.59-1.mga6.nonfree.x86_64.rpm
nvidia-current-cuda-opencl-384.59-1.mga6.nonfree.x86_64.rpm
nvidia-current-devel-384.59-1.mga6.nonfree.x86_64.rpm
nvidia-current-doc-html-384.59-1.mga6.nonfree.x86_64.rpm
x11-driver-video-nvidia-current-384.59-1.mga6.nonfree.x86_64.rpm





Mga5:

SRPMS:
ldetect-lst-0.1.346.7-1.mga5.src.rpm

nvidia-current-384.59-1.mga5.nonfree.src.rpm


i586:
ldetect-lst-0.1.346.7-1.mga5.i586.rpm
ldetect-lst-devel-0.1.346.7-1.mga5.i586.rpm

dkms-nvidia-current-384.59-1.mga5.nonfree.i586.rpm
nvidia-current-cuda-opencl-384.59-1.mga5.nonfree.i586.rpm
nvidia-current-devel-384.59-1.mga5.nonfree.i586.rpm
nvidia-current-doc-html-384.59-1.mga5.nonfree.i586.rpm
x11-driver-video-nvidia-current-384.59-1.mga5.nonfree.i586.rpm


x86_64:
ldetect-lst-0.1.346.7-1.mga5.x86_64.rpm
ldetect-lst-devel-0.1.346.7-1.mga5.x86_64.rpm

dkms-nvidia-current-384.59-1.mga5.nonfree.x86_64.rpm
nvidia-current-cuda-opencl-384.59-1.mga5.nonfree.x86_64.rpm
nvidia-current-devel-384.59-1.mga5.nonfree.x86_64.rpm
nvidia-current-doc-html-384.59-1.mga5.nonfree.x86_64.rpm
x11-driver-video-nvidia-current-384.59-1.mga5.nonfree.x86_64.rpm
Thomas Backlund 2017-08-15 09:50:54 CEST

Whiteboard: (none) => MGA5TOO

Comment 1 Len Lawrence 2017-08-16 00:14:34 CEST
Mageia release 6 (Official) for x86_64
4.9.40-desktop-1.mga6
Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz
NVIDIA Corporation GF114 [GeForce GTX 555] 

Installed the listed packages and rebooted to the Mate desktop.

Stellarium looks good.

    glmark2 2014.03
=======================================================
    OpenGL Information
    GL_VENDOR:     NVIDIA Corporation
    GL_RENDERER:   GeForce  GTX 555/PCIe/SSE2
    GL_VERSION:    4.5.0 NVIDIA 384.59
=======================================================
                                  glmark2 Score: 3424 
=======================================================

Two instances of vlc playing separate videos, smoothly at normal speed.
Launched two virtualboxes (1920x1080) in different workspaces - working fairly well.  GPU near normal temperature (63°C) - glmark2 had set the fans spinning (90°C).

CC: (none) => tarazed25

Comment 2 Len Lawrence 2017-08-16 01:07:56 CEST
Mageia release 5 (Official) for x86_64
4.4.82-1.mga5
Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz
NVIDIA Corporation GM204 [GeForce GTX 970] 
GL_VERSION:    4.5.0 NVIDIA 384.59

No problems after installation of nvidia 384.59.  Mate desktop running fine.  All graphical applications work as expected.
Comment 3 Len Lawrence 2017-08-16 07:57:50 CEST
Mageia release 6 (Official) for x86_64
4.9.38-desktop-1.mga6
Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz
NVIDIA Corporation GM204 [GeForce GTX 970] 
GL_VERSION: 4.5.0 NVIDIA 384.59

No problems after several hours of running desktop applications.
Driving 4K monitor.
Comment 4 Len Lawrence 2017-08-16 08:35:11 CEST
Mageia release 5 (Official) for x86_64
4.4.79-desktop-1.mga5
Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz
NVIDIA Corporation GK104 [GeForce GTX 770] 
GL_VERSION: 4.5.0 NVIDIA 384.59
3K monitor

Clean install and reboot to Mate, 5 workspaces, 6 terminals, gkrellm, firefox.
Addd two emacs sessions, vlc for HD TV, glmark2
Everything running well.
Comment 5 Herman Viaene 2017-08-16 11:17:00 CEST
MGA5-32 on Asus A6000VM nVidia GeForce7300 Xfce
No installation issues.
Reboot OK, play large mpg file OK with vlc, open doc file, all OK.

Whiteboard: MGA5TOO => MGA5TOO MGA5-32-OK
CC: (none) => herman.viaene

Comment 6 Rémi Verschelde 2017-08-16 23:16:47 CEST
Tested OK on Mageia 6 x86_64, glxinfo/glxspheres64 report relevant info and the games I tried run as well as they should.

Had a small quirk which is IMO unrelated to this update, but show an issue in mageia-prime (hence CC'ing joeghi, but further discussion should likely put in another issue). After installing the nvidia upgrade + the new kernel, X wouldn't start anymore, probably due to `nouveau.modeset=0` being removed from the kernel command line. Re-"installing" with `mageia-prime-install -g` fixed it.

OK'ing for mga6 64 as comments 1 and 3 were also a success.

CC: (none) => ghibomgx
Whiteboard: MGA5TOO MGA5-32-OK => MGA5TOO MGA5-32-OK MGA6-64-OK

Comment 7 Rémi Verschelde 2017-08-16 23:26:05 CEST
Advisory uploaded.

Whiteboard: MGA5TOO MGA5-32-OK MGA6-64-OK => advisory MGA5TOO MGA5-32-OK MGA6-64-OK

Comment 8 Rémi Verschelde 2017-08-16 23:27:12 CEST
Note: a few more tests would be good before validating, as a breakage on such an update means no X server, which can be a deal breaker for non power users.
Comment 9 Len Lawrence 2017-08-17 10:49:28 CEST
Note re comment 6
Install virtualgl for glxspheres64.
Comment 10 Len Lawrence 2017-08-17 12:20:49 CEST
mga6  x86_64
4.9.32-desktop-1.mga6
Intel(R) Core(TM) i7-5700HQ CPU @ 2.70GHz
NVIDIA Corporation GM204M [GeForce GTX 965M] 
twin nvidia running in single mode (non SLI)

glmark2 and glxpheres64 ran fine.
The rest of the desktop is functioning normally.
Comment 11 Len Lawrence 2017-08-17 12:21:51 CEST
s/glxpheres/glxspheres/
Comment 12 Len Lawrence 2017-08-17 15:09:27 CEST
Mageia release 5 (Official) for x86_64
4.4.82-desktop-1.mga5
NVIDIA Corporation GM204M [GeForce GTX 965M] 

Installed and rebooted smootly to Mate desktop: nvidia 384.59.  Desktop graphics working as expected.  Launched a few guis; vlc, glxspheres64, glmark2, gqview, the gimp, gwenview, stellarium.  All running fine.
Comment 13 Len Lawrence 2017-08-17 16:10:15 CEST
Mageia release 6 (Official) for x86_64
4.9.40-1.mga6
NVIDIA Corporation GK107M [GeForce GT 650M] 

Installed all the packages and rebooted to Mate.
nvidia 384.59
glxspheres increased its data rate when switching from default to fullscreen mode in order to maintain 60 fps.  Ran glmark2 and tried various graphical applications.  glmark2 uses one cpu core to full capacity.  Played a video in vlc over the LAN, fullscreen without stuttering.
Comment 14 Len Lawrence 2017-08-17 17:13:29 CEST
Mageia release 5 (Official) for x86_64
4.4.74-desktop-1.mga5
NVIDIA Corporation GK107M [GeForce GT 650M] 
Starting with nvidia 375.26

Installed all the packages and rebooted to KDE4.
nvidia 384.59

Ran the usual tests and checked out the desktop.  Logged in to another machine on the LAN and viewed images using gqview which responded to keyboard events on the local machine.  Everything working.

That just about exhausts my hardware.
Comment 15 Giuseppe Ghibò 2017-08-17 22:33:39 CEST
you can also try installing cuda-z (it's in nonfree).
Comment 16 Giuseppe Ghibò 2017-08-19 13:55:22 CEST
Actually the upgrade should work with mageia-prime, as soon as the configuration will be keepen always "working". If for instance one typical scenario is this: 1) upgrade drivers while being in nvidia-mode, then kill X11 or the dkms new modules for some reasons not built. At this point the X11 configuration is temporarely not working because there is a mismatch between the older loaded in memory nvidia kernel modules and the nvidia libraries which are changed, so X11 won't start. As fallback some drakxutils (or inside prefdm I'm not sure) will detect this and will change continuosly xorg.conf to something other (usually forcing nouveau, while IMHO this is wrong, as a fallback should be at most vesa). I'm not sure where it is but I suspect it is controlled by the line AUTORECONFIGURE_RIGHT_XORG_DRIVER=yes|no in /etc/sysconfig/harddrake2/service.conf; I suspect also /etc/sysconfig/harddrake2/xorg XORG_DRV=<something> might influence this.

IMHO the right way to do the upgrade without the reboot is to do the drivers upgrade, ensuring the newer nvidia kernel modules from dkms are correctly built (there might be some configuration where kernel-devel is not forced to be installed [e.g. because of a kernel upgrade] so in this case dkms fails to build the modules). Then switch to init level 3 (or text mode) with systemctl, then remove the older nvidia kernel modules, modprobe all the newer kernel modules then bring back the system to level 5 (or graphical.target).
Comment 17 Herman Viaene 2017-08-21 11:31:24 CEST
MGA6-32 on Asus A6000VM MATE nVidia GeForce7300
No installation issues.
Reboot OK, played locally video file (thus sound is OK), view pictures, read pdf file. Opened rather large odt file from NFS share.
OK for me.
Note to Len: you tested apparently both M5-64 and M6-64, but you did not OK the M5-64, akthough I cann't see a problem in your comments???
So I do not validate, as not knowing whether there is still an issue left.

Whiteboard: advisory MGA5TOO MGA5-32-OK MGA6-64-OK => advisory MGA5TOO MGA5-32-OK MGA6-64-OK MGA6-32-OK

Comment 18 Len Lawrence 2017-08-21 12:51:08 CEST
Sorry Herman.  I was just waiting for others to repeat the tests but you are right the bug can be validated for both architectures.
Len Lawrence 2017-08-21 12:51:26 CEST

Keywords: (none) => validated_update
CC: (none) => sysadmin-bugs

Comment 19 Mageia Robot 2017-08-21 13:32:19 CEST
An update for this issue has been pushed to the Mageia Updates repository.

http://advisories.mageia.org/MGAA-2017-0064.html

Resolution: (none) => FIXED
Status: NEW => RESOLVED


Note You need to log in before you can comment on or make changes to this bug.