aptosid.com

User Scripts & Dragons - Nvidia [and fglrx] broken

etorix - 17.06.2011, 11:14
Post subject: Nvidia [and fglrx] broken
the X-server and mesa have gone multi-arch (which basically means its libraries were moved to a different path)

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=630710

either switch to nouveau or dont upgrade nvidia

for nouveau, see http://aptosid.com/index.php?module=wik ... g=GoNvidia
krisbee - 17.06.2011, 12:01
Post subject:
For the first time I did a DU when I shouldn't have. So I decided to try nouveau to get out of the mess. But I have problems after following the wiki entry.

First, I cannot purge nvidia-kernel-dkms.
      Code:

 apt-get purge nvidia-kernel-dkms
Reading package lists... Done
Building dependency tree       
Reading state information... Done
The following packages will be REMOVED:
  nvidia-kernel-dkms*
0 upgraded, 0 newly installed, 1 to remove and 16 not upgraded.
After this operation, 29.1 MB disk space will be freed.
Do you want to continue [Y/n]? Y
(Reading database ... 148034 files and directories currently installed.)
Removing nvidia-kernel-dkms ...
dkms.conf: Error! No 'DEST_MODULE_LOCATION' directive specified.
dkms.conf: Error! No 'PACKAGE_NAME' directive specified.
dkms.conf: Error! No 'PACKAGE_VERSION' directive specified.
dkms.conf: Error! No 'DEST_MODULE_LOCATION' directive specified.
dkms.conf: Error! No 'PACKAGE_NAME' directive specified.
dkms.conf: Error! No 'PACKAGE_VERSION' directive specified.

Error! Bad conf file.
File: /var/lib/dkms/nvidia/275.09.07/source/dkms.conf does not represent
a valid dkms.conf file.
dpkg: error processing nvidia-kernel-dkms (--purge):
 subprocess installed pre-removal script returned error exit status 5
configured to not write apport reports
                                      Errors were encountered while processing:
 nvidia-kernel-dkms
E: Sub-process /usr/bin/dpkg returned an error code (1)


Second, I have got nouveau to load and KDE to start but the screen resoultion is stuck on 1024x768 as shown by xrandr.

      Code:
xrandr
xrandr: Failed to get size of gamma for output default
Screen 0: minimum 1024 x 768, current 1024 x 768, maximum 1024 x 768
default connected 1024x768+0+0 0mm x 0mm
   1024x768       76.0*


My monitor is capable of 1280x1024


Third, if I remove the vga entry from /etc/default/grub and update-grub, on re-boot X does start.

Any ideas?
devil - 17.06.2011, 13:08
Post subject:
edit the corresponding prerm file to make it read:
      Code:

#!/bin/sh
set -e
exit 0
......

so what you do is, add the exit 0, save and exit. then purge the app.

greetz
devil
towo - 17.06.2011, 13:19
Post subject:
      Quote:
Second, I have got nouveau to load and KDE to start but the screen resoultion is stuck on 1024x768 as shown by xrandr.

      Code:

xrandr
xrandr: Failed to get size of gamma for output default
Screen 0: minimum 1024 x 768, current 1024 x 768, maximum 1024 x 768
default connected 1024x768+0+0 0mm x 0mm
   1024x768       76.0*


My monitor is capable of 1280x1024

aptosid-paste /var/log/xorg.0.log
ls -al /etc/modprobe.d | aptosid-paste

      Quote:
Third, if I remove the vga entry from /etc/default/grub and update-grub, on re-boot X does start.


vga=foo in grub2 is depricated!
jacmoe - 17.06.2011, 13:41
Post subject: Re: Nvidia [and fglrx] broken
      etorix wrote:
the X-server and mesa have gone multi-arch (which basically means its libraries were moved to a different path)

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=630710

either switch to nouveau or dont upgrade nvidia

Will the nvidia driver be back eventually, or are we forever doomed?

I did a mistake and have no X and three held packages..
Noveau is a pile of crap for me (3D programmer), so it can only be a temporary solution.

One of the best features of aptosid was indeed the elegant way it handled my nvidia-glx driver. Wink

<edit>
It sort of looks like it will be fixed eventually ?
Will it be safe to try and get the noveau driver working and then switch back to nvidia-glx when it's ready?
</edit>
towo - 17.06.2011, 13:45
Post subject: RE: Re: Nvidia [and fglrx] broken
Downgrade libgl1-mesa-glx to testing and your nvidia-driver will work again.
jacmoe - 17.06.2011, 13:47
Post subject: RE: Re: Nvidia [and fglrx] broken
Also after having applied the diversions?
nvidia-glx was removed but three packages held..

I will try and figure out how I do that.

What a mess. Smile
jacmoe - 17.06.2011, 14:06
Post subject:
So, to downgrade I do this:

  1. uncomment unstable entries in apt sources
  2. add testing entries
  3. apt-get update
  4. install package(s)
  5. comment testing
  6. uncomment unstable
  7. apt-get update
  8. hold any packages upon dist-upgrades until issue fixed

Am I on the right track?
dibl - 17.06.2011, 14:36
Post subject: Re: RE: Re: Nvidia [and fglrx] broken
      towo wrote:
Downgrade libgl1-mesa-glx to testing and your nvidia-driver will work again.


Brilliant! Also I had to reinstall nvidia-kernel-source and nvidia-kernel-common, and then run m-a a-i nvidia-kernel-source, then reinstall nvidia-glx and I'm good to go with desktop effects.

Thanks towo!

@jacmoe, the downgrade procedure is spelled out in the manual under "Package Management and Dist-Upgrade" section. Don't forget to put a hold on your testing version of the package.
towo - 17.06.2011, 14:38
Post subject:
      Quote:
Am I on the right track?

No, simple do
      Code:

apt-get install libgl1-mesa-glx/testing libgl1-mesa-dri/testing

after you have don an apt-get update with activated testing entries in your sources.
krisbee - 17.06.2011, 14:38
Post subject:
Thanks devil, never knew about prerm files before now. Purge done.

towo, seemed I had some other nvidia cruff still on my system which was screwing things up. All is fine now with nouveau driver. (nice to learn about apotsid-paste, that's a very nifty app)

      Code:
xrandr
Screen 0: minimum 320 x 200, current 1280 x 1024, maximum 4096 x 4096
VGA-1 disconnected (normal left inverted right x axis y axis)
DVI-I-1 connected 1280x1024+0+0 (normal left inverted right x axis y axis) 338mm x 270mm
   1280x1024      60.0 +   75.0*
   1400x1050      60.0 
   1440x900       75.0     59.9 
   1280x960       60.0 
   1360x768       60.0 
   1280x800       74.9     59.8 
   1152x864       75.0 
   1280x768       74.9     59.9 
   1024x768       75.1     75.0     70.1     60.0 
   832x624        74.6 
   800x600        72.2     75.0     60.3     56.2 
   848x480        60.0 
   640x480        75.0     72.8     72.8     66.7     60.0     59.9 
   720x400        70.1 
TV-1 disconnected (normal left inverted right x axis y axis)


I think I can live without nvidia for now.

Thanks again for the rapid help.
drb - 17.06.2011, 14:59
Post subject:
I'm struggling to understand the problem here. I'm running the latest libgl1-mesa-glx and nvidia 275.09.07 here without any issues. Am I missing something?

      Code:
apt-cache policy libgl1-mesa-glx
libgl1-mesa-glx:
  Installed: 7.10.3-1
  Candidate: 7.10.3-1
  Version table:
 *** 7.10.3-1 0
        500 http://ftp.de.debian.org/debian/ sid/main i386 Packages
        100 /var/lib/dpkg/status
     7.10.2-3 0
        500 http://ftp.de.debian.org/debian/ testing/main i386 Packages
drb@hypnos32:~$ infobash -v3
Host/Kernel/OS  "hypnos32" running Linux 2.6.39-1.slh.8-aptosid-686 i686 [ sidux 2010-01 Ύπνος - kde-full - (201006131622) ]
CPU Info        2x AMD Athlon 64 X2 Dual Core 5600+ 1024 KB cache flags( sse3 ht nx lm svm ) clocked at [ 2800.000 MHz ]
Videocard       nVidia G73 [GeForce 7600 GS]  X.Org 1.10.2  [ 1680x1050@50.0hz ]
Network cards   nVidia MCP55
Processes 154 | Uptime 8:56 | Memory 722.8/3295.8MB | HDD WDC WD3200AAKS-0,WDC WD5000AAKS-0,WDC WD6400AAKS-0,WDC WD15EADS-00P Size 2960GB (37%used) | GLX Renderer GeForce 7600 GS/PCI/SSE2/3DNOW! | GLX Version 2.1.2 NVIDIA 275.09.07 | Client Shell | Infobash v3.36

jacmoe - 17.06.2011, 15:06
Post subject:
      towo wrote:
      Quote:
Am I on the right track?

No, simple do
      Code:

apt-get install libgl1-mesa-glx/testing libgl1-mesa-dri/testing

after you have don an apt-get update with activated testing entries in your sources.

That wasn't enough for me as the nvidia driver was already wiped out.

I ended up changing sources to testing and did this:
      Code:
apt-get install libgl1-mesa-dev=7.10.2-3 libgl1-mesa-dri=7.10.2-3 libgl1-mesa-glx=7.10.2-3 libglu1-mesa=7.10.2-3 libglu1-mesa-dev=7.10.2-3 mesa-common-dev=7.10.2-3

And then, to get my nvidia back:
      Code:
apt-get install nvidia-glx


It would be enough to hold 'nvidia-glx' ?
<edit>I'll hold it and see what happens. Wink</edit>

I am not touching anything until there is a fix from the Debian guys. I am just happy to have my X server back up and running. Smile
devil - 17.06.2011, 16:25
Post subject:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=630710 is probably fixed with mesa* 7.10.3-2.

greetz
devil
jacmoe - 17.06.2011, 17:16
Post subject:
Not yet:
      Quote:
Calculating upgrade... Done
The following packages will be REMOVED:
libgl1-nvidia-glx nvidia-glx
The following packages have been kept back:
libgl1-nvidia-alternatives
The following packages will be upgraded:
bluetooth bluez bluez-pcmcia-support cpp g++ gcc libbluetooth3 libgl1-mesa-dev libgl1-mesa-dri
libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev mesa-common-dev

Or, am I missing something?
devil - 17.06.2011, 17:20
Post subject:
you are missing mesa* 7.10.3-2 packages.
they should be uploaded soon.

greetz
devil
piper - 18.06.2011, 01:09
Post subject: Re: RE: Re: Nvidia [and fglrx] broken
      dibl wrote:
      towo wrote:
Downgrade libgl1-mesa-glx to testing and your nvidia-driver will work again.


Brilliant! Also I had to reinstall nvidia-kernel-source and nvidia-kernel-common, and then run m-a a-i nvidia-kernel-source, then reinstall nvidia-glx and I'm good to go with desktop effects.

Thanks towo!



same here, danke towo! Wink
neosid - 23.06.2011, 08:00
Post subject:
All theses posts about a workaround, with the testing packages, but not one of these worked here. I tried them exactly, in combo, started over, etc, etc.

Can the downgrade work? I mean has anyone seen that work?

I had just installed aptosid kde-lite. Nvidia worked; at a very high frame rate. Then I upgraded, and it seemed to really need it (lots of packs). Now, I got diddly. No nvidia, no nv, no neouveo-ee-o-foo. looks like this has been going on, for 6 days. Has anyone, a workaround; that's working? Could you please post the commands that are verified to work? Is the testing downgrade, just a theory? Trust me, I put the above suggestions in, and well. No install, of nvidia-glx. I just get, unmet libgl1-nivida-glx.

Should work and does work, are two different things. Smile
neosid - 23.06.2011, 08:52
Post subject:
Oh foo!

Success, but by other means. Full DRI/3D.

      Code:
apt-get install xorg


      Code:
cd /usr/local/bin && wget -O sgfxi smxi.org/sgfxi && chmod +x sgfxi


Then ran the sgfxi script, of course.

Smile
piper - 23.06.2011, 08:59
Post subject:
      neosid wrote:
All theses posts about a workaround, with the testing packages, but not one of these worked here. I tried them exactly, in combo, started over, etc, etc.

Can the downgrade work? I mean has anyone seen that work?

I had just installed aptosid kde-lite. Nvidia worked; at a very high frame rate. Then I upgraded, and it seemed to really need it (lots of packs). Now, I got diddly. No nvidia, no nv, no neouveo-ee-o-foo. looks like this has been going on, for 6 days. Has anyone, a workaround; that's working? Could you please post the commands that are verified to work? Is the testing downgrade, just a theory? Trust me, I put the above suggestions in, and well. No install, of nvidia-glx. Unmet libgl1-nivida-glx. Should work and does work, are two different things. Smile




      Code:
ctrl+alt+F1

init 3 && apt-get update && apt-get dist-upgrade

(let the 3 nvidia packages get removed)

nano /etc/apt/sources/.list.d/debian.list

enable/add testing

deb http://http.us.debian.org/debian/ testing main contrib non-free

apt-get update && apt-get install libgl1-mesa-glx=7.10.2-3

disable testing

apt-get update && apt-get install nvidia-kernel-source- nvidia-kernel-common nvidia-glx

reboot

Success and faster than smxi/sgfxi

edit: made it more clear
jacmoe - 23.06.2011, 09:10
Post subject:
Bollocks.
That maneuver caused apt to remove 140 packages. KDE and X server/core.
I am just waiting it out.
And seeing that it's got low priority, it can take a long while until we get our nVidia back.
piper - 23.06.2011, 09:18
Post subject:
      jacmoe wrote:
Bollocks.
That maneuver caused apt to remove 140 packages. KDE and X server/core.
I am just waiting it out.
And seeing that it's got low priority, it can take a long while until we get our nVidia back.


you havent done a du in a while then huh, 140 packages is a lot, I du at least twice a day, everything smooth,
jacmoe - 23.06.2011, 09:43
Post subject:
Well, I do dist-upgrade at least twice a week.
That's not the problem.
I will give it another shot. Wink
jacmoe - 23.06.2011, 10:00
Post subject:
@piper:
Thanks! Very Happy
Worked like a charm.

IIRC, apt would remove 140 packages when I tried to install libgl1-nvidia-blahblah..

Now all I have to do is hold back libgl1-mesa-glx and wait. Wink
OppaErich - 23.06.2011, 12:12
Post subject:
Lucky you. Still no X for me. http://paste.pocoo.org/show/417401
And no, nvidia is installed, I had to use the force option to make module-assistent build another one. Confused

Well, I'm starting to love LMDE Gnome...
ghstryder - 23.06.2011, 14:03
Post subject:
OppaErich, you are not alone. My laptop isn't working right now. I lost nvidia with the DU, and decided to try Nouveau for a second time. The last time we had nvidia issues, I tried Nouveau unsucessfully, but was able to get nvidia working again. This time, I still can't get Nouveau to work, nor can I get nvidia to work. I have tried every fix that has shown up, to no avail. I haven't had time to dive into it, and have not needed the laptop, so it isn't a huge issue for me, thankfully. I'll get it resolved one of these days, or introduce it to 2011-next.
piper - 23.06.2011, 19:14
Post subject:
These are the problem packages

      Code:
libgl1-nvidia-alternatives libgl1-nvidia-glx nvidia-glx libgl1-mesa-glx


On a dist-upgrade these 3 will be removed

      Code:
libgl1-nvidia-alternatives libgl1-nvidia-glx nvidia-glx


After those 3 are removed, enable testing and
      Code:
apt-get update && apt-get install libgl1-mesa-glx=7.10.2-3
then disable testing and
      Code:
apt-get update && apt-get install nvidia-kernel-source- nvidia-kernel-common nvidia-glx


Some people might have to do this also, then the steps above

      Code:
apt-get purge --remove dkms xserver-xorg-video-nouveau
m-a a-i nvidia-kernel-source


      Code:
echo blacklist nouveau > /etc/modprobe.d/nouveau.conf


as root and reboot.
ghstryder - 23.06.2011, 19:56
Post subject:
      piper wrote:
      Code:
echo blacklist nouveau > /etc/modprobe.d/nouveau.conf
as root and reboot.
Ah, that is one I must have missed. I'll try running through the whole thing again and see what happens.

Thanks Piper!
neosid - 24.06.2011, 11:48
Post subject:
Don't doubt that I tried all that, or think that I just miss typed something. Plus, I'm not really new at this stuff anymore. As many are now reporting, it was a no go (on my GeForce 7). Congrats, if downgrading worked for you. I'm just saying wait, or try sgfxi. I dist-upgraded again (after my initial install, and first dist-upgrade, as referenced in my last post) and the new kernel, of course, needed a fix again, which sgfxi did quickly.

I don't think we should need sgfxi either; it's just that it worked.

Thanks for the list of commands/directions Piper. Are those just suggestions, or what worked to fix yours?

While I prefer the nvidia driver, and would like to see neuvo, or whatever it's called, work better, it doesn't, and I understand Nvidia is dragging tail, by not open sourcing their drivers, like many others have done. Thing would be so much simpler. This will be their downfall; unless they learn to play together, better.
piper - 24.06.2011, 12:00
Post subject:
      neosid wrote:
Don't doubt that I tried all that, or think that I just miss typed something. Plus, I'm not really new at this stuff anymore. As many are now reporting, it was a no go (on my GeForce 7). Congrats, if downgrading worked for you. I'm just saying wait, or try sgfxi. I dist-upgraded again (after my initial install, and first dist-upgrade, as referenced in my last post) and the new kernel, of course, needed a fix again, which sgfxi did quickly.

I don't think we should need sgfxi either; it's just that it worked.

Thanks for the list of commands/directions Piper. Are those just suggestions, or what worked to fix yours?

While I prefer the nvidia driver, and would like to see neuvo, or whatever it's called, work better, it doesn't, and I understand Nvidia is dragging tail, by not open sourcing their drivers, like many others have done. Thing would be so much simpler. This will be their downfall; unless they learn to play together, better.


works on all 3 of my nvidia boxes, my ati box was not effected
OppaErich - 24.06.2011, 14:07
Post subject:
      neosid wrote:
Don't doubt that I tried all that, or think that I just miss typed something.

Same here and all went through without error messages but still Xorg.0.log says "module nvidia not found, module does not exist". I have Pipers tips in print now and will give it a last try. If it's still not working, I give sgfxi a shot.
neosid - 24.06.2011, 14:11
Post subject:
UPDATE: Fixed now; in sid!

Working "testing" downgrade directions

This will be not be unnecessary, when the fix, now in experimental, passes to sid.

Thank you piper for the verification (downgrading; as seen working).
I ran your commands to clear nouveau:
      Quote:
apt-get purge --remove dkms xserver-xorg-video-nouveau
m-a a-i nvidia-kernel-source
echo blacklist nouveau > /etc/modprobe.d/nouveau.conf
...but don't think that was it. Who knows?
============================================================

For the sake of others, I took my perfectly working system, had some coffee, and ran your two (below) command strings in a root terminal first, just to get them into the buffer, where I could up arrow to them easily, after init 3.

Changed "sid" to "testing" with:
      Quote:
mcedit /etc/apt/sources.list.d/debian.list

I think mcedit, is the aptosid recommended editor.
As root
      Quote:
init 3
apt-get update && apt-get install libgl1-mesa-glx=7.10.2-3

Changed from testing; back to sid again; with:
      Quote:
mcedit /etc/apt/sources.list.d/debian.list

THEN, look and make sure your command (as below) is updating back to sid, first:
      Quote:
apt-get update && apt-get install nvidia-kernel-source nvidia-kernel-common nvidia-glx
Noting your extra "-" in there, I removed.

With approval (Y), it installed packages to clear-out other nvidia stuff! It did.

Then reboot completely!

Also, to hold this pack:
      Quote:
echo libgl1-mesa-glx hold|dpkg --set-selections

Check this hold, and list packages on hold:
      Quote:
dpkg --get-selections | grep hold

Later, to take this package off Hold:
      Quote:
echo libgl1-mesa-glx install|dpkg --set-selections


Verified Working on mine, here today.
Thank you Piper. Cool

Verified, here! This will be unnecessary, when the fix, now in experimental, passes to sid.
towo - 24.06.2011, 15:42
Post subject:
No need for testing anymore, fixed packages are in experimental.
neosid - 24.06.2011, 15:56
Post subject:
I did verify, still broken in sid. Now, what other commands work for using experimental. Nice to see the progress! It's hard to just wait, without X, so I hope the downgrade workaround helped, in the interim (day 7).
towo - 24.06.2011, 16:00
Post subject:
apt-get purge $(dpkg -l | awk '/nvidia/{print $2}')
apt-get update && apt-get dist-upgrade
apt-get -t experimental install nvidia-kernel-source
m-a clean nvidia-kernel-source && m-a a-i nvidia-kernel-source
apt-get -t experimental install nvidia-glx
neosid - 24.06.2011, 16:10
Post subject:
Thank you towo, but Opps. Didn't work here.

...Unheld and dist-upgraded; and entered all your commands.

dpkg: dependency problems prevent configuration of nvidia-kernel-2.6.39-2.slh.1-aptosid-amd64:
nvidia-kernel-2.6.39-2.slh.1-aptosid-amd64 depends on nvidia-kernel-common (>= 20100522); however:
Package nvidia-kernel-common is not installed.
dpkg: error processing nvidia-kernel-2.6.39-2.slh.1-aptosid-amd64 (--install):
dependency problems - leaving unconfigured
Errors were encountered while processing:
nvidia-kernel-2.6.39-2.slh.1-aptosid-amd64


The following packages have unmet dependencies:
nvidia-glx : Depends: libgl1-nvidia-glx (= 275.09.07-1)
E: Broken packages
towo - 24.06.2011, 16:13
Post subject:
275.09.07-1 is not the version in experimental!
neosid - 24.06.2011, 16:17
Post subject:
So, I need to enable experimental somehow, as well?
towo - 24.06.2011, 16:20
Post subject:
Are you joking?
OppaErich - 24.06.2011, 16:21
Post subject:
      towo wrote:
No need for testing anymore, fixed packages are in experimental.

Ah, just coincidence ? I've done as Piper wrote on my 32bit install, all went fine without error messages. But still "Failed to load module 'nvidia' (module does not exist, 0)".

Then I took my 64bit install on the same machine. All fine without errors again but X came up after a reboot. The only difference I noticed (besides 32<->64) was dkms kicking in. I've purged dkms on 32bit a few weeks ago... Confused
neosid - 24.06.2011, 16:23
Post subject:
Sadly no. I'm not joking. LOL. I would think your commands would pull it from experimental, with no other changes needed, but here, we got those errors. What next?
towo - 24.06.2011, 16:25
Post subject:
      Quote:
Ah, just coincidence ? I've done as Piper wrote on my 32bit install, all went fine without error messages. But still "Failed to load module 'nvidia' (module does not exist, 0)".

This does say nothing!
Where you get this error?
towo - 24.06.2011, 16:25
Post subject:
      neosid wrote:
Sadly no. I'm not joking. LOL. I would think your commands would pull it from experimental, with no other changes needed, but here, we got those errors. What next?

Read the manual about managing different repos.
neosid - 24.06.2011, 16:31
Post subject:
Aw, come on now. I guess your saying I need to do like we did before, and temp change sid to experimental, in the sources, but then remember to change it back. I'm asking is that correct, and is that also advised?

I'm also thinking, if I can get this done, via experimental now, then future sid upgrades (regression repairs) will leave nothing else to do, right? Seems obvious, but that's not always the case, now is it?
towo - 24.06.2011, 16:33
Post subject:
      Quote:
I guess your saying I need to do like we did before, and temp change sid to experimental,

No! You have to add experimental!
towo - 24.06.2011, 16:35
Post subject:
Btw, the experimental packages only works in 64bit.
In 32bit glx-alternative-mesa is broken.
OppaErich - 24.06.2011, 16:38
Post subject:
      towo wrote:

This does say nothing!
Where you get this error?

Xorg.0.log http://paste.pocoo.org/show/417401/
towo - 24.06.2011, 16:39
Post subject:
Then, there is no xserver-xorg-video-nvidia installed.
neosid - 24.06.2011, 16:46
Post subject:
New and additional experimental line, in sources? Then your commands? Then remove that experimental line, before my next dist-upgrade; after the fix falls to sid?

64bit here; thanks.
towo - 24.06.2011, 16:53
Post subject:
There is no need to remove the experimental entrie from the sources.
sng - 24.06.2011, 17:55
Post subject:
I have successfully applied ripper's instructions and everything is ok now on 32bit

Thanks again man
neosid - 24.06.2011, 18:22
Post subject:
Tried fixing via experimental, and failed:

What I did
..added duplicate line, in sources, and changed sid, to experimental (left sid line), and dist-upgraded. Still broken nvidia. I broke my repaired downgrade (page 3), to test this....

Ran Towo's commands:
      Quote:
apt-get purge $(dpkg -l | awk '/nvidia/{print $2}')
apt-get update && apt-get dist-upgrade
apt-get -t experimental install nvidia-kernel-source
m-a clean nvidia-kernel-source && m-a a-i nvidia-kernel-source
apt-get -t experimental install nvidia-glx

Failed installing packs. Trying to recover...

I understand, some people posted that it worked for them, but for me, and now others...

errors:

      Quote:
Package glx-alternative-mesa is not configured yet.
dpkg: error processing glx-alternative-nvidia (--configure):
dependency problems - leaving unconfigured


Removed experimental source, left sid.

Trying then to fix the two stuck, and loaded (selected?) experimental packages. I set:
(This was wrong, I think)
      Quote:
Package: *
Pin: release a=testing
Pin-Priority: 1001

Package: *
Pin: release a=unstable
Pin-Priority: 60

Package: *
Pin: release a=experimental
Pin-Priority: 50
Updated and ran:

# dpkg --force-depends -r glx-alternative-mesa

...DID NOT WORK, strangely.

got

      Quote:
(Reading database ... 131314 files and directories currently installed.)
Removing glx-alternative-mesa ...
update-alternatives: error: /var/lib/dpkg/alternatives/glx corrupt: duplicate slave link /usr/lib/x86_64-linux-gnu/libGL.so.1
dpkg: error processing glx-alternative-mesa (--remove):
subprocess installed pre-removal script returned error exit status 2
Errors were encountered while processing:
glx-alternative-mesa

How do I get back?
towo - 24.06.2011, 18:42
Post subject:
I have installed the driver from experimental on 2 machines without any problem.
I don't know, what are you doing.
neosid - 24.06.2011, 18:56
Post subject:
I am trying to remove glx-alternative-mesa, that is left as "unconfigured"; so, I can purge the two installed, and difficult to remove, experimental packages...

glx-alternative-mesa
glx-alternative-nvidia

      Quote:
apt-get -f install
Reading package lists... Done
Building dependency tree
Reading state information... Done
0 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
2 not fully installed or removed.
After this operation, 0 B of additional disk space will be used.
Setting up glx-alternative-mesa (0.1) ...
Processing triggers for glx-alternative-mesa ...
update-alternatives: error: /var/lib/dpkg/alternatives/glx corrupt: duplicate slave link /usr/lib/x86_64-linux-gnu/libGL.so.1
dpkg: error processing glx-alternative-mesa (--configure):
subprocess installed post-installation script returned error exit status 2
configured to not write apport reports
dpkg: dependency problems prevent configuration of glx-alternative-nvidia:
glx-alternative-nvidia depends on glx-alternative-mesa; however:
Package glx-alternative-mesa is not configured yet.
dpkg: error processing glx-alternative-nvidia (--configure):
dependency problems - leaving unconfigured
configured to not write apport reports
Errors were encountered while processing:
glx-alternative-mesa
glx-alternative-nvidia
E: Sub-process /usr/bin/dpkg returned an error code (1)

, per your commands, and exactly what I said, I did above.

What is "duplicate slave link", and why is that stopping the install and removal, of theses packages?

Addendum:

OK. I fixed it myself! ...Got myself in a mess, just testing for everyone, and got myself back out!

The error on the "mesa" pack (the mesa one, or some dependency, of a dependency), was the file "glx". I renamed it glx.HOLD.

I went back to my working directions, earlier in this thread and followed them, again. The point, after changing back from testing, to sid sources, I could then (and only then, after renaming glx), install dependences (the mesa one, or some dependency, of a dependency) to start, one by one, (about three) until nvidia-glx (in my line; after setting sources back) would install, along with the others. I also deleted my test, preferences file. I had none before.

Can't say as I can recommend the experimental method, even though Towo got it to work. I'd say, just use my testing downgrade post, until the fix is in sid. Then release it.

As a fall-back, you have the sgfxi script. I'm not using that currently.

Working; as testing downgrade, and held. See my instructions, on page 3, of this thread. 4+ times verified, here; on one of my systems.
Spaceferret - 25.06.2011, 01:48
Post subject:
Thank you towo, I have followed your recommendations installing from experimental and it works 100%. Thank you again !
Spaceferret - 25.06.2011, 02:13
Post subject:
Just curious- is it OK to update to the newer versions of glx in experimental ?Or is it better to comment experimental out and wait ?
jacmoe - 25.06.2011, 10:43
Post subject:
Why did you move this topic out of upgrade warnings?
To create drama? Or was it a mistake?
DonKult - 25.06.2011, 11:23
Post subject:
      jacmoe wrote:
Or was it a mistake?

I haven't moved it myself, but the end of the thread is FAAAAAR away from being close to be supportable. If the answer is --force-depends the question was properly: "After which commands should i reinstall my system?" The showed preference entry is also a good answer to this question…

All in all the only sane solution i have seen here is installing the package from experimental which is now in unstable, so no need for all these kind of trickery…

And i heard a few people saying it works now (irc, this thread), so if it doesn't for you, please open a new thread as it would be a different issue with the same endresult - and none of the included """solutions""" is good anymore… (if they were at any time).


P.S.: I can't help with the issue at hand as i don't have radeon or nvidia cards, but if i would read one of the posts above i wouldn't even bother to help as i would prefer to work on my skill to juggle with running chainsaws. That would be at least saner and easier… so splitting your issue to a new one is a very good idea…
jacmoe - 25.06.2011, 19:19
Post subject:
I agree that the topic was invaded by dragons. Smile
My problem is solved (downgraded), and I am just going to wait until updated nVidia/glx stuff hits the unstable branch.
mz - 25.06.2011, 23:21
Post subject:
      neosid wrote:
The error on the "mesa" pack (the mesa one, or some dependency, of a dependency), was the file "glx". I renamed it glx.HOLD.

I upgraded two boxes; one went smoothly and the other had "the neosid effect". Thanks for this hint; both are OK now.
neosid - 27.06.2011, 05:12
Post subject:
"the neosid effect"

LOL.

      Quote:
All in all the only sane solution i have seen here is installing the package from experimental which is now in unstable, so no need for all these kind of trickery…

And i heard a few people saying it works now (irc, this thread), so if it doesn't for you, please open a new thread as it would be a different issue with the same endresult - and none of the included """solutions""" is good anymore… (if they were at any time).


P.S.: I can't help with the issue at hand as i don't have radeon or nvidia cards, but if i would read one of the posts above i wouldn't even bother to help as i would prefer to work on my skill to juggle with running chainsaws. That would be at least saner and easier… so splitting your issue to a new one is a very good idea…


The neosid Effects:

1. I have received thank you messages, about how my testing and posted results have saved people from giving up on aptosid.

2. The upgrade to sid, is not resolved yet, as was inferred above. I had to revert to my own posted commands, on page 3 of this thread, yet again, and we are working fine, here. Others have reported the same; with the downgrade.

3. As to "sanity" and "trickery", when X is broken, waiting over Ten days (now) is far from acceptable behavior, even for sid. A workaround, if worked out, and easily instructed, as I did, would be in order. While each may have their own, and sometimes differing results, other people have corroborated the same results, as I experienced. I do not, generally, doubt posts of other methods, as working. However, on my system, those posted experimental usage commands worked out to be an insanely difficult, to get back out of. I actually reverted back from experimental, the hard way, and which does not always work. I have reports, that had I not posted a solution, to work back out of the failed experimental instructions, they would have left aptosid, for another, more reasonable distro, and all that is involved with it.

4. I do not think, ..."is good anymore… (if they were at any time)." is appropriate forum statements. This opens the door to my next point.

5. It should be obvious, at this late date, that we should state whether our offered help, is seen validated, or not; along with an easy to follow, to a point, command instructions. Of course, an efforts success, or failure, may not be everyones same experience.

6. If you do not have the hardware in question, then you may certainly suggest better hardware, or even policy changes, from feet dragging, non-free companies, but why would you even attempt to in effect, call someone a lier, for hard working out a throughly tested workaround, that gets X up now?

7. As far as this thread, and its move, I have no control over this. If you believe we should take up another thread, I await your detailed recommendation, as I do not understand why a working, hard-fought workaround, would be moved; before the fix truly falls through to sid, and my dist-upgrades.

8. As far as I knew, this was a thread, and warning about nvidia (and ati I read), not working; after dist-upgrade. I also wonder why, as this was a "multi-arch" restructuring, and thus folder foul-up, why after 10 days, it is not in sid? By the time I write this, it may indeed be fixed. I stated this with my, quick and easy, testing downgrade commands, but the bug was still not fixed, a few minutes minute ago, here. I checked.

9. I think some of us have forgotten, that we are all in this together. What you post, reflects on the whole of open software. "Everything is permissible, but not everything is advisable." Please do not post guesses; without testing them, or stating that it's your best guess. Please do not infer, that the testing results of others is disingenuous, and please consider posting solutions that are easy to implement.

10. Most of all, what are we doing to reduce the amount of time required, for theses fixes, and also their temporary workarounds?
ghstryder - 27.06.2011, 10:33
Post subject:
Ah, the knight protector of the downtrodden nvidia class.
DonKult - 27.06.2011, 14:00
Post subject:
      neosid wrote:
6. If you do not have the hardware in question, then you may certainly suggest better hardware, or even policy changes, from feet dragging, non-free companies, but why would you even attempt to in effect, call someone a lier, for hard working out a throughly tested workaround, that gets X up now?


I may not own the hardware you talked about, but i contribute a fair share to software you have misused in various ways to get to your "throughly tested workaround" - so i think i am entitled to criticism your "solution": It can't be that well tested through as the pinning entry you recommend to use will downgrade every package which is effected by any install/upgrade request you do while it is in effect to testing and forbids the installation of any package from unstable or experimental (beside that you haven't even said that they NEED to remove it later). So, that you got a still working system and even the thing you desired was only possible with a not to small portion of luck, not skill nor through testing as something like --force-depends can't be tested. It usually just destroys systems if it is included in step-by-step guides and even as simple as looking at the manpages would have told you this.


But to get at least some value in this post:

As far as i know the current working solution is (if your system is already broken - at least a few sources suggest that current sid works and btw: even if it would be 10 days, waiting even longer isn't unbearable. At least if you are not an addict. You know what they say: "I can stop it at any time…")
      Code:
echo "deb http://ftp.us.debian.org/debian/ experimental main contrib non-free" > /etc/apt/sources.list.d/debian-experimental.list
apt-get purge .*nvidia.*
apt-get update
apt-get dist-upgrade
apt-get -t experimental install nvidia-kernel-source
m-a clean nvidia-kernel-source
m-a a-i nvidia-kernel-source
apt-get -t experimental install nvidia-glx
echo "blacklist nouveau" > /etc/modprobe.d/blacklist-nouveau.conf
rm /etc/apt/sources.list.d/debian-experimental.list #optional


More about adding sources to your list and how m-a works can be found in the manual of course! Don't be as blind as entering whatever is written on a forum straight into your terminal without checking it.

Also note through that the amount of packages you need from experimental drops further and further. Pulling glx-alternative from experimental is for example a bad idea now as it is older than the sid one and this will become true for more. So watch out for messages about DOWNGRADE and do NOT perform them. Instead use the version dist-upgrade installed for you then and go on.

As you might know downgrades are not supported by the software nor the distribution and can cause a lot of trouble: Some of you have already experienced it. Downgrading with diversions is no fun at all…
For people already fallen into this trap something like that might help:
      Code:
rm -f /var/lib/dpkg/alternatives/glx /etc/alternatives/glx
dpkg --configure -a
apt-get update
apt-get dist-upgrade
apt-get install --reinstall --only-upgrade glx-alternative-mesa glx-alternative-nvidia glx-alternative-fglrx


Through i wouldn't hold my breath as i can't test it…
dibl - 27.06.2011, 15:00
Post subject:
Just to be clear for anyone trying to follow this torturous thread, towo's advice on page 1 to downgrade libgl1-mesa-glx to the testing version (7.10.2-3), and then pin it, continues to allow both my Nvidia systems to work with no video issues. Both systems are fully upgraded as of yesterday evening. Here is one:


      Code:
susan@susansid:~$ infobash -v3
Host/Kernel/OS  "susansid" running Linux 2.6.39-2.slh.1-aptosid-amd64 x86_64 [ aptosid 2011-01 Γῆρας - kde-lite - (201102051540) ]
CPU Info        2x Intel Core2 X6800 @ 4096 KB cache flags( sse3 ht nx lm vmx ) clocked at [ 2926.000 MHz ]
Videocard       nVidia GT200 [GeForce GTX 260]  X.Org 1.10.2  [ 1600x1200@50.0hz ]
Network cards   Intel 82573L Gigabit
Processes 185 | Uptime 2days | Memory 1017.0/3964.8MB | HDD WDC WD1500AHFD-0 Size 150GB (32%used) | GLX Renderer GeForce GTX 260/PCI/SSE2 | GLX Version 3.3.0 NVIDIA 275.09.07 | Client Shell | Infobash v3.36


FWIIW
neosid - 28.06.2011, 03:41
Post subject:
Trolling, or worse, is where you take my stated, and verified post, which others have agreed helped, and then mixed it with my other posts. Where I was asking for help, and clearly stated what did not work. Don't you think I could have deleted my errors. Why do you think I didn't delete my mistakes? Do you never err? Which by the way, those were clearly stated, in a response to asking what I actually did. From the moment they were typed, it was clearly my attempt at guessing how to repair, the failed experimental recommendation. Are you inferring Towo comes on these forums, just to mis-guild people? I think not. As far as my "folly", in trying a "team leader" suggestion, I re-broke my system; just to test it for the good of all. I almost had to reinstall. Especially, due to the lack of much help. I mean what really happened, is commands were posted; that were terse, and when they failed, I was told to "read the manual", and "are you kidding". What about the fact the those commands, cause problems for people? I fixed it myself, and even told you how i did it. Now, we have posts factually stating two things. One, the experimental method failed, with others, just as it did mine, and Two, My self discovered back-out (of experimental packs) worked, not only for me, but also others.

So guess what? Your "ingenious commands" (the new one above; by someone else) failed to work/fix, my dilemma, as they were the first thing I tried. How can you even be here, and not know the difference between theory and fact? How can you not know, that NOT everything, that is supposed to work, does?

Oh, and your inference that people can't, or don't read the texts, in between quoted commands, is insulting to us all, I'm sure. Would that anyone do so, there was never any question about which commands I posted, worked, or did not.

I did my job, and that, I picked. Including the mess I got myself into, and out of. I fully knew, the risks, going in. You are welcome. You see, I offered the working solution, in a speedy way, that was easy to follow, when peoples X was down. This got them going; now! What are you doing?

I actually fixed this temporary problem; 3 different ways.

1. sgfxi. It worked.

2. The vague, "just use testing", I turned into something actually doable; fast.

3. I even showed how to fix the posted, "experimental" method; when it failed, for some people.

Aptosid Knight,

It seems you are in dire need; of a knight. Yet, I don't have time, for this extra, added, horse foo. So, fight the good fight, noble ones.

If I mistook towo, as forum leader around here, that was a mistake, on my part. If you read what I did, I effectively showed, how towo commands didn't work for some people. Now, I see "team" people here are saying (I think) "experimental" was never a good idea, if running Sid. I would generally agree.

As far as temporarily downgrading, a pack to testing, so that X works, and right now, that was the suggestion of the people who broke it; in the first place. That's in the bug reports. I just added to what Piper knew (and towo, to be fair), and made it easy (to a point), and fast.
jacmoe - 28.06.2011, 12:36
Post subject:
I just unheld my nvidia stuff and upgraded - so this problem is solved!
nVidia and compatible glx/mesa is in Debian Sid now.
At least on amd64. Smile
neosid - 28.06.2011, 15:23
Post subject:
Unhold with....

echo libgl1-mesa-glx install|dpkg --set-selections

and with init 3

(Actually, check for other warnings here; before upgrade. Also, consider using apt-listbugs. Plus, there's been some issues, with repo's.)

apt-get dist-upgrade

Yes, the wait is over. It works. My post, on page 3, is no longer needed. That was from the 17th, to the 28th, which was about a week, and 3 days, where using those parts, in testing (for sid users), would get you going; until today.

Nvidia; amd64 here.
merlyn - 28.06.2011, 21:07
Post subject:
Just released libgl1-mesa-glx and libgl1-mesa-dri from hold and followed the correct du procedure.

All is working well on my 32 bit install.

Off topic; good to see that it is also working on 64 bit. Time to do a 64 bit install now.

Cheers.
tageloehner - 13.08.2011, 23:12
Post subject:
Hello,

on my system
"Linux eiche 3.0.0-1-amd64 #1 SMP Sun Jul 24 02:24:44 UTC 2011 x86_64 GNU/Linux"
(normally the perfect computer)

updates of some packages still crashes the X server - like reported:
module glx and fglrx not found,
even with correctly loaded modules.


Any installation of one of the packages in the attachment prevents the X server to start (if you keep dependencies).

My Linux level is fairly ok, this is not a beginners remark.

In my opinion this is a severe bug.

Best regards...
dibl - 14.08.2011, 00:44
Post subject:
You have fglrx and nvidia packages -- what the heck? What is your secret mystery GPU?

If you simply follow the manual, you should not have such a problem.

Here the nvidia driver is working fine on 2 systems.
      Code:
Host/Kernel/OS  "aptosidbox" running Linux 2.6.39-3.slh.2-aptosid-amd64 x86_64 [ aptosid 2010-03 Ἀπάτη - kde-lite - (201012262151) ]
CPU Info        8x Intel Core i7 950 @ 8192 KB cache flags( sse3 ht nx lm vmx ) clocked at [ 1600.000 MHz ]
Videocard       nVidia GF100 [GeForce GTX 480]  X.Org 1.10.3  [ 1920x1200@50.0hz ]
Network cards   Marvell 88E8056 PCI-E Gigabit
Processes 232 | Uptime 1day | Memory 984.4/5978.8MB | HDD OCZ-REVODRIVE,OCZ-REVODRIVE,KINGSTON SS100S2,WDC WD1002FAEX-0,WDC WD1002FAEX-0 Size 2136GB (2%used) | GLX Renderer GeForce GTX 480/PCI/SSE2 | GLX Version 4.1.0 NVIDIA 280.13 | Client Shell | Infobash v3.37

tageloehner - 18.08.2011, 11:49
Post subject:
Yes, both, because one image is running on 13 Laptops, if st is broken we just change the disks.

Best regards
slam - 18.08.2011, 12:29
Post subject:
While running the same image/disk on different machines will work quite ok with Linux, there are some points where you need to adapt them. Using closed source graphics drivers from different vendors is one of those points. Running nvidia and fglrx on the same installation will always lead to problems, as both drivers tend to overwrite core Xorg files making them unusable for any other driver.
You may file bug reports at nvidia and amd/ati commercial sites, but I do not think that they will care much.
Adapting fstab, host & mail names is necessary also.
Greetings,
Chris
tageloehner - 19.08.2011, 09:41
Post subject:
I dont want to argue about my choise of installation - it is working fine since years. Its one image fitting all the hardware of different laptops without changes.

What happend: Update of packages appear, the X server fails with message
"failed to load module glx" and
"failed to load module fglrx"
even if these modules are loaded directly.

I set the responsible packages - see last attachment - to "hold" and submit a bug report.
The only answers are flames.

So I stop conversation. Bye.



>While running the same image/disk on different machines will >work quite ok with Linux, there are some points where you need >to adapt them. Using closed source graphics drivers from >different vendors is one of those points. Running nvidia and >fglrx on the same installation will always lead to problems, >as both drivers tend to overwrite core Xorg files making them >unusable for any other driver.
>You may file bug reports at nvidia and amd/ati commercial >sites, but I do not think that they will care much.
>Adapting fstab, host & mail names is necessary also.
>Greetings,
>Chris
slam - 19.08.2011, 12:21
Post subject:
I really do not see flames here, sorry - just people trying to help you. Please re-read my post above. With every upgrade of Xorg or fglrx or nvidia several files of Xorg (specially those -glx ones) will be overwritten. This has nothing to do with your "choice of installation", but is a common problem of the both closed source drivers.
Greetings,
Chris
tageloehner - 02.09.2011, 14:46
Post subject:
Its working again now - bug seems to be fixed.

See

http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=610022

Debian Bug report logs - #610022
"fglrx-glx fails to install in parallel with nvidia-glx"

Thanks to all developers.
All times are GMT - 12 Hours
Powered by PNphpBB2 © 2003-2010 The Zafenio Group
Credits