[SOLVED] After upgrading to kernel 3.8 module-assistant can't find kernel headers (Debian Sid)

Hungry ghost · 05-11-2013, 08:03 PM

Hi folks,

I'm using Debian Sid and a few days ago I upgraded to kernel 3.8 (from the official Debian repositories). The point is, after this upgrade, module-assistant can't find the kernel headers, even though they are installed. If for example, I try to compile virtualbox I get this:

Code:

# m-a -t a-i virtualbox
.
Updated infos about 1 packages
Getting source for kernel version: 3.8-1-amd64
apt-get install linux-headers-3.8-1-amd64 
Reading package lists... Done
Building dependency tree       
Reading state information... Done
linux-headers-3.8-1-amd64 is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded.
apt-get install build-essential 
Reading package lists... Done
Building dependency tree       
Reading state information... Done
build-essential is already the newest version.
0 upgraded, 0 newly installed, 0 to remove and 1 not upgraded.

Done!


Bad luck, the kernel headers for the target kernel version could not be found and you did not specify other valid kernel headers to use.
      

However, you can install the header files for your kernel which are provided by the linux-headers-3.8-1-amd64 package. For most modules packages, these files are 
perfectly sufficient without having the original kernel source.

To install the package, run:

module-assistant prepare

or

apt-get install linux-headers-3.8-1-amd64

But the kernel headers are installed, as you can see in the following output:

Code:

# apt-cache policy linux-headers-3.8-1-amd64
linux-headers-3.8-1-amd64:
  Installed: 3.8.12-1
  Candidate: 3.8.12-1
  Version table:
 *** 3.8.12-1 0
        500 http://ftp.us.debian.org/debian/ unstable/main amd64 Packages
        100 /var/lib/dpkg/status

If I specify the kernel-headers directory when invoking module-assistant, it throws this error:

Code:

# m-a --kernel-dir /usr/src/linux-headers-3.8-1-amd64 a-i virtualbox

Updated infos about 1 packages
Bad kernel version specification at /usr/bin/m-a line 568, <$apt> line 7.

This is the output of uname -a, just in case:

Code:

# uname -a
Linux debian 3.8-1-amd64 #1 SMP Debian 3.8.12-1 x86_64 GNU/Linux

I just found this related bug report for the same version of module-assistant I'm using, but I'm not sure about how to apply the patch (and whether or not it will fix this issue). Does anyone one know how to apply this patch?

Thanks!

cynwulf · 05-12-2013, 10:23 AM

The patch should do it, you will need to

Code:

$ apt-get source module-assistant

Copy the .patch file to the downloaded source directory, apply the patch e.g.

Code:

$ patch <detection.patch

Code:

$ cd module-assistant*

Get the build dependencies

Code:

# apt-get build-dep module-assistant

Then rebuild the package

Code:

$ dpkg-buildpackage -rfakeroot -us -uc

(you might need to install fakeroot - though I believe it's part of build-essential(?), which I'd suspect is already installed)

Install

Code:

$ cd ..

Code:

# dpkg -i module-assistant*deb

Hungry ghost · 05-12-2013, 07:01 PM

Hello caravel, thanks for your help.

I followed the steps you suggested, and everything seems to go fine but when I get to the dpkg-buildpackage -rfakeroot -us -uc command, I get long output that eventually ends in the following error:

Code:

nsgmls:I: maximum number of errors (200) reached; change with -E option
Can't call method "value" on an undefined value at /usr/share/perl5/sgmlspl-specs/docbook2man-spec.pl line 249, <STDIN> line 1.
make[1]: *** [module-assistant.8] Error 8
make[1]: Leaving directory `/home/vicente/Desktop/ma/module-assistant-0.11.4'
dh_auto_build: make -j1 returned exit code 2
make: *** [build-indep] Error 2
dpkg-buildpackage: fallo: debian/rules build devolvió un estado de salida de error 2

The translation of the last line is pretty much this: "dpkg-buildpackage: fail: debian/rules returned an error exit status 2."

Any ideas about what might be wrong? Maybe a problem in the source package itself?

EDDY1 · 05-13-2013, 12:06 AM

Just a shot in the dark, how many kernels do you have installed if more than 1 are you boot to the latest?

cynwulf · 05-13-2013, 05:24 AM

Install docbook-utils and try again.

Hungry ghost · 05-13-2013, 11:27 AM

Hello again,

The package that was missing was docbook-to-man (your suggestion to install docbook-utils made me suspect that some other docbook package was missing

). After installing it, I got past the first error, but got other errors which seemed to be caused by a problem in module-assistant itself, so I made these changes and it finally compiled fine.

Now I'm having other problem when building the virtualbox module with module-assistant (not the initial problem, though). Judging from the results of a quick web search, the virtualbox module fails to build against new kernels for some reason. However, I will have to wait before attempting to fix it, since this will be a busy week for me. Anyway I managed to install the nVidia driver from the repositories (didn't know I could do without module-assistant to install the non-free driver from the repos, as I've always used it to build the driver).

@EDDY1: For the record, I only have the 3.8 kernel installed at the moment (which is the one I'm currently using).

Since the initial problem with module-assistant is fixed now, I will mark this thread as solved. Thank you both for your help, folks; much appreciated.

cynwulf · 05-13-2013, 03:01 PM

The virtualbox module may be failing to build for the same reason (i.e. lack of include/linux/version.h)

Hungry ghost · 05-13-2013, 06:51 PM

Well, the last lines of the build log are different to the first error messages I got, which makes me think this is another issue:

Code:

/usr/src/modules/virtualbox/vboxdrv/r0drv/linux/memobj-r0drv-linux.c: In function ‘rtR0MemObjNativeMapUser’:
/usr/src/modules/virtualbox/vboxdrv/r0drv/linux/memobj-r0drv-linux.c:1451:38: error: ‘VM_RESERVED’ undeclared (first use in this function)
/usr/src/modules/virtualbox/vboxdrv/r0drv/linux/memobj-r0drv-linux.c:1451:38: note: each undeclared identifier is reported only once for each function it appears in
kmk[5]: *** [/usr/src/modules/virtualbox/vboxdrv/r0drv/linux/memobj-r0drv-linux.o] Error 1
kmk[5]: *** Waiting for unfinished jobs....
kmk[5]: Leaving directory `/usr/src/linux-headers-3.8-1-amd64'
kmk[5]: Entering directory `/usr/src/linux-headers-3.8-1-amd64'
kmk[5]: *** Exiting with status 2
kmk[4]: *** [/usr/src/modules/virtualbox/vboxdrv] Error 2
kmk[4]: Leaving directory `/usr/src/linux-headers-3.8-1-amd64'
kmk[3]: *** [_module_/usr/src/modules/virtualbox] Error 2
kmk[3]: Leaving directory `/usr/src/linux-headers-3.8-1-amd64'
kmk[2]: *** [sub-make] Error 2
kmk[2]: Leaving directory `/usr/src/linux-headers-3.8-1-common'
kmk[1]: *** [all] Error 2
kmk[1]: Leaving directory `/usr/src/linux-headers-3.8-1-amd64'
kmk: *** [binary-modules] Error 2
make: *** [kdist_build] Error 2

There's a similar bug report but is related to kernels 3.6 and 3.7, though I suspect it's the same issue since parts of the errors look quite similar. Will try the patch in the link when I have some time this week.

cynwulf · 05-14-2013, 03:34 AM

Your linked bug report is relevant. It's probably resolved if you use Debian's virtualbox packages and kernels.

Hungry ghost · 05-16-2013, 07:59 PM

Hi again,

This is just to let you know that VirtualBox got upgraded with today's updates and now it's working fine again, so the problem seems to be solved. Thanks a lot for your help, caravel.

Cheers.