Page: 1, 2  Next

nVidia drivers fail to load
Mono
Status: Contributor
Joined: 21 Jun 2012
Posts: 115
Reply Quote
Hello. Yet another problem with graphics after an update. Sigh.

I did a dist-upgrade and have been trying to get graphics since yesterday. I discovered the debugger option in sgfxi yesterday so I sent a bug report with that.

My logs seem to be telling me that some other driver is not letting the nVidia driver get to the graphics card.

I looked through Aptitude and found a Noveau package (X11-blah-blah-noveau or something) that wasn't removed by smxi or sgfxi, so I purged it. But the problem remained.

I'm running an AMD FX8350 with an Asus M5A99FX Pro R2.0 with a GeForce 7600GS, maybe with that you can tell which bug report came from me.

I'm not sure what to do at this point, please help.
Back to top
sedonix
Status: Contributor
Joined: 09 Sep 2008
Posts: 56
Location: Sedona, AZ
Reply Quote
Since it's a holiday weekend and H2 is not getting any younger, I'll just throw this at you while you're waiting. You should be using the same driver 304.121 as I am. No problems at all and I am on Debian sid with, it seems, a weekly kernel change from towo at Siduction. Purging nouveau will not change any config settings. I'm not a code person so I will stop here and hopefully someone will come along to help you.
Back to top
Mono
Status: Contributor
Joined: 21 Jun 2012
Posts: 115
Reply Quote
Oops, how rude of me. :/

I am running Debian Testing I think and am using the Liquorix kernel.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4124
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
sources show sid + wheezy volatile (why? what? and you expect the system to work for what reason?) + some ubuntu stuff (always a fine way to pull in breaks, but should not matter in this case). Assuming I'm looking at the right gz debugger, which I think I am, your system is not debian testing, it's debian mutant, and that would go a long way to explaining why you have issues. If you are running testing, your sources should be for jesse, period. Maybe plus sid and have it pinned to testing. I guess you could use the volatile jesse sources.

Really there is only one question, does nvidia 304, which of courser is the only driver you should be using, install and work on 3.14 kernel?

That's a yes or no question. If it does, and if liquorix 3.15 fails, then it looks like an issue with with 304 + 3.15, which is hardly surprising since 304 support is close to being ended if I read right, not sure when, maybe next year or this. Certainly have not been new 304 releases since 3.14/3.15 I believe, could be wrong on 3.14.

Specific cards in the 6-7xxx series may also trigger bugs.

No point in further talk until you post the results of installing to 3.14.

If it works, with nothing else changed at all, then the issue is a bug with nvidia I believe, so you would then have to file a bug report with nvidia, using their bug report generator. To generate that bug report, make sure to install to 3.15, install 304, then, generate the bug report and upload it to devtalk.nvidia.com/default/board/63/gpu-unix-graphics/ forums.

Their bug tool is: nvidia-bug-report.sh

If you forget that command, which is, unlike their drivers and installers and bug report tool, badly thought out, just do:
ls /usr/bin | grep nvidia
that's how I find it, I always forget the specific name.


However, do NOT do that until you have confirmed or denied that 3.14 works. Don't do anything tricky or otherwise overly clever, simply boot to 3.14, install the driver, then see if your desktop starts.

:: Code ::
Repos:     Active apt sources in file: /etc/apt/sources.list
           deb http://security.debian.org/ wheezy/updates main contrib non-free
           deb-src http://security.debian.org/ wheezy/updates non-free contrib main
           deb http://ftp.us.debian.org/debian/ unstable non-free contrib main
           deb-src http://ftp.us.debian.org/debian/ unstable non-free contrib main
           deb http://liquorix.net/debian/ unstable main
           deb http://ppa.launchpad.net/yannubuntu/boot-repair/ubuntu/ saucy main
           deb [trusted=yes] file:///usr/src/keantoken/cinnamon-debian/aaa-debian-repo public unstable
           Active apt sources in file: /etc/apt/sources.list.d/apt-spy.list
           deb http://mirrors.gigenet.com/debian/ unstable main
           deb-src http://mirrors.gigenet.com/debian/ unstable main


Q: how to break debian? And create a virtually unsupportable system?

A: Run this set of apt sources.

Whatever did these did it wrong, your 'testing' system should have one line:
:: Code ::
# if pinned to testing in apt config:
deb http://mirrors.kernel.org/debian unstable main contrib non-free
# otherwise just this one. You can if you must use the
# volatlle for jesse I guess, but that only makes sense if
# the system does not have sid
deb http://mirrors.kernel.org/debian testing main contrib non-free


I stopped using apt-spy a few hours after starting, it scared me. I prefer to find and use a good repo and stick to it.
Back to top
Mono
Status: Contributor
Joined: 21 Jun 2012
Posts: 115
Reply Quote
Yesterday I tried it with my last kernel (3.13), which it worked on before, and that no longer works. So it's not a problem with 3.15. It could be a problem with the driver or it could be a problem with the upgrade.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4124
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
I would have to consider your system as is largely unsupportable.

However, if a driver that installed before, and which works on sedonix sid with 304 and latest towo kernel, 3.15 I assume, is now not installing, the odds are that it's the system, not the driver.

Not 100%, since if you got a new xorg version in upgrade, to 1.15, that can also be an issue, though I don't believe it is.

Looking at your sgfxi.log, it looks like you used -d? don't do that, it's impossible to debug debian driver issues.

Run this command, and only this, on 3.13 and 3.15:

:: Code ::
sgfxi


nothing else.

Do this first on 3.15, see if desktop start, then do it on 3.14, ideally, or 3.13 if that's all you have. If it doesn't start on 3.13, then the issue is almost impossible to fix or debug. I would recommend starting off by fixing your apt sources list, that way at least there's a hope of getting support, what you are running now is not supportable at all. And that assumes you actually can even use wheezy volatile / security repos at all with sid current, no idea what happens there. Whereever you got that idea from, or if you simply forgot, you can fix that issue by fixing your sources, then doing a full upgrade. Note that if you are running in your mind 'testing', then you have to have testing sources in your apt repos. Again, you have created a non supportable scenario, so you can't expect to get support.

To fix the list, remove apt spy, remove all hint of wheezy repos, totally remove, comment out, then do apt-get update, and then apt-get dist-upgrade. Also comment out the ubuntu repo, again, you can't expect support when running any type of hybrid repo setup.

If you hit weird failures and issues on dist-upgrade, your system is broken and needs to be fixed before you ask for any further support.

It's possible there's some other issues with cruft from the debian nvidia package, but I haven't seen issues about that recently.

If 304 does not install and start your desktop, when it did before, then something broke in the system, and it's unlikely to be anything nvidia can fix for you.
Back to top
sedonix
Status: Contributor
Joined: 09 Sep 2008
Posts: 56
Location: Sedona, AZ
Reply Quote
I can assure you that 304 works on 3.15 kernels in sid.

I can also assure you that if you update to the latest xserver.xorg in sid, 304 will no longer work until nvidia comes up with an updated driver.

I rarely make errors in judgement but one of my installs no longer has a display.

I guess i can wait.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4124
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
Ah, ok, so it's the xserver 1.15 causing the issues, that's good to know.

I will update the sgfxi xorg detection and have it exit with error on 1.15.

On the brighter side, this suggests that mono has not destroyed his system by running weird sources, and that he is running debian sid.

Sometimes there are patches.

When in doubt, do what I do, save me the work, google nvidia 304.121 xorg 1.15 patch.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4124
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
devtalk.nvidia.com/default/topic/754151/304-121-with-xorg-1-16/

:: Quote ::
Prerelease and release candidate X servers are technically not supported. Please stick with 1.15.* until 1.16.0 is released. That said, support for the new X server ABI will be available in the next release.

Aaron Plattner
NVIDIA Linux Graphics


for those of you who don't know him, aaron is the man at nvidia for linux.

However, mono's issue is not related because his system is running:
X.Org X Server 1.15.1
Release Date: 2014-04-13

which is handled by this legacy nvidia 304 as aaron notes. The new sgfxi debugger is a great headache remover, that's for sure.

:: Code ::
apt-cache policy xserver-xorg-core
xserver-xorg-core:
  Installed: 2:1.14.4-1
  Candidate: 2:1.15.99.904-1
  Version table:
     2:1.15.99.904-1 0
        500 http://mirrors.kernel.org/debian/ unstable/main i386 Packages
     2:1.15.1-1 0
        200 http://mirrors.kernel.org/debian/ testing/main i386 Packages
 *** 2:1.14.4-1 0
        100 /var/lib/dpkg/status


so there you have it.Sid has prerelease 1.16, aka, 1.15.99, but mono has 1.15.1, so the xserver is NOT the issue, and in fact, he should not upgrade or his actual issue will be hidden by the xserver update, which will in all cases make 304 non uninstallable unless a patch appears.
Back to top
techAdmin
Status: Site Admin
Joined: 26 Sep 2003
Posts: 4124
Location: East Coast, West Coast? I know it's one of them.
Reply Quote
because the case were xorg version 1.x.99 == 1.(x+1.0 functionally was not handled, now sgfxi has that handled, so that means, if you have xorg 1.5.99, which previously would have been treated as 1.15 simply in sgfxi, it will now be more fine grained in the tests, and will trigger the 1.16 tests, which means, 304 driver plus xorg 1.15.99 will now in theory trip the unsupported xorg driver.

However, I would like to remind legacy nvidia driver users, do not as a rule surf the bleeding edge, nvidia doesn't release early legacy drivers as a rule for abi changes like xorg 1.16 has, so you are taking a chance.

Best strategy is to hold off updates and new kernels until a month or two have passed when using old hardware. Of course, amd/fglrx users, you can ignore this advice, the odds of amd actually offering legacy driver support are essentially zero, so just install radeon and forget about it.

nouveau isn't a horrible bet either for older legacy nvidia, though I think I'd hold off on the 6-7xxx series until nvidia drops support in the future there.

8-9xxx cards for those who don't know, are next in line for legacy status with nvidia.
Back to top
Display posts from previous:   
Page: 1, 2  Next
All times are GMT - 8 Hours