Bug 8092 - [Update Request] nvidia340 340.102
: [Update Request] nvidia340 340.102
Status: RESOLVED INVALID
Product: Desktop Bugs
Classification: ROSA Desktop
Component: Main Packages
: Fresh
: All Linux
: Normal normal
: ---
Assigned To: ROSA Linux Bugs
: ROSA Linux Bugs
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2017-06-25 00:46 MSD by Zombie Ryushu
Modified: 2017-06-28 21:15 MSD (History)
4 users (show)

See Also:
RPM Package: nvidia340
ISO-related:
Bad POT generating:
Upstream:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Zombie Ryushu 2017-06-25 00:46:25 MSD
Nvidia 340.101 fails with a bad exit status of 10. This is likely to be due to the Kernel 4.9 headers. Should be upgraded to 340.102.
Comment 1 Eugene Shatokhin 2017-06-25 19:52:48 MSD
Version 340.102 is already available for testing in x11_backports repo for rosa2014.1: http://abf-downloads.rosalinux.ru/x11_backports_personal/repository/

And, as usual: if there are hardware and/or driver problems, please include an URL to hw-probe data in your bug report. It usually takes only 10-20 seconds on your part to run hw-probe but makes bug analysis MUCH easier.

In addition, if build of the driver fails, please attach the log from /var/lib/dkms. Again, this is easy to retrieve but would save me a considerable amount of time.

The bugs with all the necessary info provided are often fixed much faster.
Comment 2 Zombie Ryushu 2017-06-25 21:46:37 MSD
I'll get right on that.
Comment 3 Zombie Ryushu 2017-06-25 22:08:57 MSD
ernel preparation unnecessary for this kernel.  Skipping...

Building module:
cleaning build area....
make KERNELRELEASE=4.1.38-nrj-desktop-2rosa-x86_64 module KERNEL_UNAME=4.1.38-nrj-desktop-2rosa-x86_64 && cd uvm && make module KERNEL_UNAME=4.1.38-nrj-desktop-2rosa-x86_64....(bad exit status: 2)

Error! Bad return status for module build on kernel: 4.1.38-nrj-desktop-2rosa-x86_64 (x86_64)
Consult the make.log in the build directory
/var/lib/dkms/nvidia340/340.102-1/build/ for more information.

Kernel preparation unnecessary for this kernel.  Skipping...

Building module:
cleaning build area....
make KERNELRELEASE=4.9.20-nrj-desktop-1rosa-x86_64 module KERNEL_UNAME=4.9.20-nrj-desktop-1rosa-x86_64 && cd uvm && make module KERNEL_UNAME=4.9.20-nrj-desktop-1rosa-x86_64....(bad exit status: 2)

Error! Bad return status for module build on kernel: 4.9.20-nrj-desktop-1rosa-x86_64 (x86_64)
Comment 4 Eugene Shatokhin 2017-06-25 22:52:57 MSD
(In reply to comment #3)
Please attach the build log from /var/lib/dkms/ like I said before. Look for all *.log files from /var/lib/dkms/nvidia340/ directory tree.

I need to see the full log.

As you said about "include/generated/autoconf.h or include/config/auto.conf are missing." (it is just an echo of a command, not always an actual error, btw), I suppose, you found it.

So, please attach it here.

I cannot reproduce the problem on my system though. Version 340.102 of nvidia340 has been built and installed OK for the kernels 4.9.x and 4.1.x on my machine with ROSA R8.

The message about 'make oldconfig && make prepare' is irrelevant. If our kernel-devel package was installed and the files were not damaged, one should not run 'make oldconfig' there.
Comment 5 Zombie Ryushu 2017-06-25 23:00:49 MSD
I'm working on getting you a make.log. But this was on Rosa R9. This was an upgrade from Rosa 8.1
Comment 6 Zombie Ryushu 2017-06-25 23:08:38 MSD
The make.log file is embarrassingly empty

DKMS make.log for nvidia340-340.101-4 for kernel 4.9.20-nrj-desktop-1rosa-x86_64 (x86_64)
Sun Jun 25 15:05:58 EDT 2017

thats all it says.
Comment 7 Eugene Shatokhin 2017-06-26 12:24:30 MSD
(In reply to comment #6)
> The make.log file is embarrassingly empty
> 
> DKMS make.log for nvidia340-340.101-4 for kernel
> 4.9.20-nrj-desktop-1rosa-x86_64 (x86_64)
> Sun Jun 25 15:05:58 EDT 2017
> 
> thats all it says.

That is the old one. The log for nvidia340-340.102-1 should also be there somewhere.

If you cannot find it (yes, dkms sometimes deletes these for little reason or does not store them at all), you can try to rebuild the driver and get the log after that. I assume you use kernel 4.9.20-nrj-desktop-1rosa-x86_64, if not - change the commands appropriately.

dkms remove -m nvidia340 -v 340.102-1 -k 4.9.20-nrj-desktop-1rosa-x86_64
dkms build -m nvidia340 -v 340.102-1 -k 4.9.20-nrj-desktop-1rosa-x86_64

If /var/lib/dkms/nvidia340/340.102-1/build/make.log is empty again, run the build explicitly:

KVER=4.9.20-nrj-desktop-1rosa-x86_64
cd /var/lib/dkms/nvidia340/340.102-1/build/
make KERNELRELEASE=${KVER} module KERNEL_UNAME=${KVER} && cd uvm && make module KERNEL_UNAME=${KVER}

and attach the output here.
Comment 8 Zombie Ryushu 2017-06-27 16:27:53 MSD
Kernel preparation unnecessary for this kernel.  Skipping...

Building module:
make clean && cd uvm && make clean

/usr/sbin/dkms: line 1386: uvm//Module.symvers: No such file or directory
make KERNELRELEASE=4.9.20-nrj-desktop-1rosa-x86_64 module KERNEL_UNAME=4.9.20-nrj-desktop-1rosa-x86_64 && cd uvm && make module KERNEL_UNAME=4.9.20-nrj-desktop-1rosa-x86_64 >> /var/lib/dkms/nvidia340/340.102-1/build/make.log 2>&1
cd ./..; make module SYSSRC=/lib/modules/4.9.20-nrj-desktop-1rosa-x86_64/source SYSOUT=/lib/modules/4.9.20-nrj-desktop-1rosa-x86_64/build KBUILD_EXTMOD=./..
make[1]: Entering directory '/var/lib/dkms/nvidia340/340.102-1/build'
NVIDIA: calling KBUILD...
make[2]: Entering directory '/usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64'
make -C /lib/modules/4.9.20-nrj-desktop-1rosa-x86_64/build KBUILD_SRC=/usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64 \
-f /usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64/Makefile modules
make[3]: Entering directory '/usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64'
test -e include/generated/autoconf.h -a -e include/config/auto.conf || (                \
echo >&2;                                                       \
echo >&2 "  ERROR: Kernel configuration is invalid.";           \
echo >&2 "         include/generated/autoconf.h or include/config/auto.conf are missing.";\
echo >&2 "         Run 'make oldconfig && make prepare' on kernel src to fix it.";      \
echo >&2 ;                                                      \
/bin/false)
mkdir -p ./../.tmp_versions ; rm -f ./../.tmp_versions/*
make -f /usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64/scripts/Makefile.build obj=./..
/usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64/scripts/Makefile.build:44: /usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64/./../Makefile: No such file or directory
make[4]: *** No rule to make target '/usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64/./../Makefile'.  Stop.
make[3]: *** [/usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64/Makefile:1483: _module_./..] Error 2
make[3]: Leaving directory '/usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64'
make[2]: *** [Makefile:150: sub-make] Error 2
make[2]: Leaving directory '/usr/src/linux-4.9.20-nrj-desktop-1rosa-x86_64'
NVIDIA: left KBUILD.
nvidia.ko failed to build!
make[1]: *** [Makefile:185: nvidia.ko] Error 1
make[1]: Leaving directory '/var/lib/dkms/nvidia340/340.102-1/build'
make: *** [Makefile:219: ../Module.symvers] Error 2
(bad exit status: 2)
Comment 11 Eugene Shatokhin 2017-06-28 11:49:45 MSD
(In reply to comment #10)
> https://devtalk.nvidia.com/default/topic/1005209/linux/fully-working-patch-
> for-nvidia-driver-340-102-compiler-installer-file-and-linux-kernel-4-11/

No, both of the above are different issues. The first one is about *loading* of the module, not building. The second one is about kernel 4.11 while you have kernel 4.9.

nvidia340 v340.102 builds and works OK on my machines, so it might be something in your system that is broken.

Most likely - either the installation of dkms-nvidia340 was not complete or the kernel build files were damaged somehow. If you did 'make oldconfig && make prepare' there, that might have removed some needed files too, although I am not 100% sure.

So, to rule all these options out, I suggest the following:

1. Switch to nouveau and uninstall nvidia340 completely using XFdrake. Check that dkms-nvidia340 is not present in the system after that.

2. Install kernel 4.9.34. For ROSA R8, this can be done as follows:
urpmi http://abf-downloads.rosalinux.ru/kernels_stable_personal/repository/rosa2014.1/x86_64/kernel_4_9/release/kernel-nrj-desktop-4.9.34-1rosa-x86_64-1-1-rosa2014.1.x86_64.rpm http://abf-downloads.rosalinux.ru/kernels_stable_personal/repository/rosa2014.1/x86_64/kernel_4_9/release/kernel-nrj-desktop-devel-4.9.34-1rosa-x86_64-1-1-rosa2014.1.x86_64.rpm

Do not run any commands in the kernel source tree explicitly, no 'make prepare', etc.

3. Use XFdrake again to install nvidia340.

4. Post the output of 'dkms status -m nvidia340' here.

This scenario has worked well for me and nvidia340 was properly installed.
Comment 12 Zombie Ryushu 2017-06-28 17:57:04 MSD
Switched to Noeveau. Didn't fix the Plymouth issue. But Noevau can use the card. Getting the kernel RPMs now.
Comment 13 Zombie Ryushu 2017-06-28 19:38:55 MSD
Okay. Success!

This problem partly turned out to be bad housekeeping. 

Elements of older Kernels stretching back to Mandriva 2.6.31, 2.6.33 and 2.6.36 were remnant in the File system, and unowned by any package. Additionally, there was a kernel-desktop package from 3.14.53 still present contrasted with kernel-nrj-desktop.

The Nvidia 340.101-4 Driver now builds on All Kernels.

The Plymouth Splash Screen is still a problem. 

Kodi no longer gives the DRI2: Failed to Authenticate error. But it does say that OpenCL lacks version information in libavformat.so

Wine can now see CUDA Support. Borderlands runs better/smoother.
Comment 14 Zombie Ryushu 2017-06-28 19:40:16 MSD
Your 4.9.34 Kernel seems to work fine as well.
Comment 15 Eugene Shatokhin 2017-06-28 21:00:50 MSD
Good to know. Closing this one.

The bug report concerning Plymouth should remain open for now.
Comment 16 Zombie Ryushu 2017-06-28 21:05:30 MSD
It is possible that we should recompile the opencl library. I'm not sure what implications that Kodi error has.
Comment 17 Eugene Shatokhin 2017-06-28 21:15:12 MSD
(In reply to comment #15)
> Good to know. Closing this one.
> 
> The bug report concerning Plymouth should remain open for now.

Ah, Andrey is already working on that one. Good.