module-init-tools, nvidia302.07 broken, no X

[Almost-retired]

Greetings all;

I am on my laptop as the main machine has been scuttled.

It got laggy this morning, with htop showing no reason for it, but since quite a few updates had been installed since the last reboot, including nvidia-current-302.07 on the 12th of may.

That reported an error of course nvidia wouldn't be nvidia if it was a simple apt-get install.

So I thought 'd reboot, and try out the 3.2.18 kernel I installed a day or so back.  But no X on the reboot, regardless of which kernel I booted to.  I find that 302.07 seems to be installed ANAICT, but depmod isn't updating the modules files in the root of any /lib/modules tree.

Normally here, this will be fixed with a couple sessions of rm -f `locate 295.40`, or whatever the older nvidia-current version was, in this case cleaning up leftovers from both 295.20 and 295.40.  But I still cannot reinstall to a working X that may not be an nvidia problem as the normal hand reinstall of going into /usr/src/nvidia-current-$VERSION and typing "make module-install" hangs at the end with top showing that depmod is using 100% of the cpu, forever.  But run by hand, or from the makefile, a ctl-c kills it instantly.

This is I believe the real problem, and a re-install of module-init-tools doesn't help.

So I am sitting here, with my laptop which is not configured to get email, pointing 2 confused fingers, the primary one at depmod, it simply hangs, as me or as root, a depmod -a -v is totally silent until I hit a ctl-c.  So I had apt-get install module-init-tools-3.6-6pclos2011.i586.rpm while I went to the shop and retrieved my lappy.  That said it was successful, so I did another "depmod -a" and left it sit.  20 minutes later I see this on the pclos box's screen:

depmod -a
WARNING:  Can't read module /lib/modules/3.2.18-pclos1.pae.bfs/kernel/net/x25/x25.ko.gz: Cannot allocate memory
and the root prompt is back, finally.

That box has 4Gb of dram in it, and 12Gb of swap on 3 different drives, so its for sure not out of memory.

And obviously, without a working depmod, I can't fix it.  Phone calls to friends running other flavors of linux haven't helped either.

Clues folks?

Thanks & Cheers, Gene

[AS]

Hi Gene,

please tell me a couple of things:

Code: [Select]

rpm -qa | grep dkms
dkms status
AS

[AS]

Got a PM ...

I can see your message, but there is no REPLY button.

From the rpm -qa;
dkms-2.2.0.3-1pclos2012
dkms-minimal-2.2.03-1pclos2012
dkms-broadcom-wl-5_100_82_112-1pclos2012
dkms-nvidia-current-302.07-1pclos2012

from dkms status;
broadcom-wl, 5_100_82_112-1pclos2012: added
nvidia-current, 302.07-1pclos2012, 3.2.18-pclos1.pae.bfs, i586: built

Thanks AS, Cheers, Gene

Hi Gene,

Your issue is at least partially due to dkms-2.2.0-3, which has been in testing a few time, and was later removed in favor of dkms-2.0.19-3..

The first step is to reinstall the correct version of dkms, please see this thread for further info:
http://www.pclinuxos.com/forum/index.php/topic,105186.0.html

Once you have got the right dkms version, reinstalling the dkms-nvidia-current should solve the issues for itself, I hope at least ... if not, please PM me again.

Regards
AS

i can't even use caps on this screen,  and the virii warning popups are hitting me about ever 5 words!

anyway, the dkms-2.0.19-3 stuff is not available in the repo's, only the 2.2.0-3

this reading it on one screen with mutt, which i have no clue how to run as it won't scroll with the down arrow key to read the whole message, then finding the link posted in the pm, copying its 100 chars back to ff on the lappy correctly, is a time waster and ultra frustrating.

thanks & cheers,  gene

please find below the links to dkms and dkms-minimal packages

http://distro.ibiblio.org/pclinuxos/pclinuxos/apt/pclinuxos/2010/RPMS.testing/dkms-2.0.19-3pclos2012.noarch.rpm

http://distro.ibiblio.org/pclinuxos/pclinuxos/apt/pclinuxos/2010/RPMS.testing/dkms-minimal-2.0.19-3pclos2012.noarch.rpm

I'm assuming you know the command to remove / install the packages, just in case:

Code: [Select]

rpm -e --nodeps dkms
rpm -e --nodeps dkms-minimal
rpm -ivh dkms-2.0.19-3pclos2012.noarch.rpm  dkms-minimal-2.0.19-3pclos2012.noarch.rpm

Regards
AS

[Almost-retired]

I can see your message, but there is no REPLY button.

From the rpm -qa;
dkms-2.2.0.3-1pclos2012
dkms-minimal-2.2.03-1pclos2012
dkms-broadcom-wl-5_100_82_112-1pclos2012
dkms-nvidia-current-302.07-1pclos2012

from dkms status;
broadcom-wl, 5_100_82_112-1pclos2012: added

nvidia-current, 302.07-1pclos2012, 3.2.18-pclos1.pae.bfs, i586: built

Now I have no clue the threading is afu now.

You gave a couple links to get the files, but Ihave no x on that machine so I tried wget, but both files on the ibiblio site are 404.

Next?

Thanks for moving this back to where I can post again.

Cheers, Gene

Thanks AS, Cheers, Gene

Hi Gene,

yeah, I see, some bureaucrat has moved the thread 

Your issue is at least partially due to dkms-2.2.0-3, which has been in testing a few time, and was later removed in favor of dkms-2.0.19-3..

The first step is to reinstall the correct version of dkms, please see this thread for further info:
http://www.pclinuxos.com/forum/index.php/topic,105186.0.html

Once you have got the right dkms version, reinstalling the dkms-nvidia-current should solve the issues for itself, I hope at least ... if not, please PM me again.

Regards
AS

i can't even use caps on this screen,  and the virii warning popups are hitting me about ever 5 words!

[Almost-retired]

Ok, I have those 2 files in the Downloads dir here.  How is the best way to move them to the big box?

Thanks & Cheers, Gene

[AS]

Ok, I have those 2 files in the Downloads dir here.  How is the best way to move them to the big box?

Thanks & Cheers, Gene

USB devices ?

guess the network is not working on your main machine if you are using the broadcom-wl driver ...

AS

[Almost-retired]

The network is fine, the broadcom stuff is because I do have a card in that box that I've given up ever making work.
My wired circuit is shentel cable, about 400k/sec.

I have those 2 rpms installed now, do I dare do a "dkms_autoinstall start"?  Humm, probably not, I don't believe the nvidia-current-295.40 (and thats old I think, shouldn't it be .49?) survived my nuking of anything that would compete with the new one.

I can't find the .49 on ibiblio, so I'll chance it, as this is now close to 11 hours I've been putzing with this.

Its munching along on the broadcom-wl thingy, obviously waiting for depmod to run, which for some reason takes about 20-30 minutes, and if you set the -v option, generates over 50k lines of output just for the /lib/modules/3.2.18 tree!  depmod used to run in 2 or 3 seconds, what the heck is this 30 minutes and occasional  OOM error exits all about now?  Something has definitely stopped up the plumbing for depmod IMO.

4 full rows of dots under the broadcom install line, and when it went to do the nvidia, install, small splat:
It tried to use the 302.07 version, but:
....(bad exit status: 4)
Installation failed!

Next?

Thanks & Cheers, Gene

[AS]

The network is fine, the broadcom stuff is because I do have a card in that box that I've given up ever making work.
My wired circuit is shentel cable, about 400k/sec.

I have those 2 rpms installed now, do I dare do a "dkms_autoinstall start"?  Humm, probably not, I don't believe the nvidia-current-295.40 (and thats old I think, shouldn't it be .49?) survived my nuking of anything that would compete with the new one.

I can't find the .49 on ibiblio, so I'll chance it, as this is now close to 11 hours I've been putzing with this.

Its munching along on the broadcom-wl thingy, obviously waiting for depmod to run, which for some reason takes about 20-30 minutes, and if you set the -v option, generates over 50k lines of output just for the /lib/modules/3.2.18 tree!  depmod used to run in 2 or 3 seconds, what the heck is this 30 minutes and occasional  OOM error exits all about now?  Something has definitely stopped up the plumbing for depmod IMO.

4 full rows of dots under the broadcom install line, and when it went to do the nvidia, install, small splat:
It tried to use the 302.07 version, but:
....(bad exit status: 4)
Installation failed!

Next?

Thanks & Cheers, Gene

guess there is some residual of dkms modules around ... let see what is it ...

Code: [Select]

ls -l /var/lib/dkms
ls -l /var/lib/dkms/*
find /lib/modules -name nvidia-current.ko\*
find /lib/modules -name wl.ko\*


[JohnW_57]

In 2010 repository, section non-free is:  dkms-nvidia-current-295.53-1pclos2012.i586.rpm   and  x11-driver-video-nvidia-current-295.53-1pclos2012.i586.rpm  is actual.

Nvidia302.07 (beta) was only for to test DKMS  and flash stuff.

JohnW

[Almost-retired]

Got it finally, and working well.  First thing I did was go through the repo list and get rid of the testing branch.  I don't mind bleeding a little from time to time but that was a bear.

Thanks to both AS and John I believe it is sorted now.

Cheers, Gene

[AS]

Glad you got it sorted out! 

AS