Hey there! I’m relatively new to the Linux world, and recently switched from Linux Mint to openSuse. I liked Mint, but felt like quite a few things were broken in the latest KDE distro. I was pretty impressed with openSuse from the get go, but ran into a couple of problems when I installed the NVIDIA driver with the suggested “one-click” installation method found on this openSuse site : SDB:NVIDIA drivers - openSUSE. Basically, after restart, the GUI would simply not load anymore. The computer was loading in run level 5, on command line.
As suggested everywhere, I tried a couple of booting options (nomodeset, editing /etc/sysconfig Editor to set “no_kms_in” to “yes” as suggested elswhere, setting “run_para” to “no”, … etc) in order to try and get my system back. These methods were suggested in different places on this forum, but none seemed to work for my system (as a Linux newbie, I must say I was quite happy with myself when I found out how to start Yast from run level 3… not a big achievement, but still…!). Finally, althought it was mentioned in lots of places that Failsafe boot would just not work, I tried it out on a hunch… and what do you know, my system booted almost properly!
Now, first thing I did : flush those evil NVIDIA drivers, and switch back to the opensource nouveau driver. At some point, I was even considering switching distro : not because I did not like Suse, but mostly because I thought I’d have to do a fresh install to get my system back, and in that case, since I’m in an experimenting mood, trying out a new distro sounded like fun. On my Win7 partition, I ended up on the ArchLinux web site… and first thing I see is this message :
“The nvidia-173xx and nvidia-96xx driver packages have been removed from our repositories as they are incompatible with newer xorg servers. This can only be fixed by an upstream update, which has not happened yet.”
Is there any chance that my problem is related to this issue with the NVIDIA driver?
Where did you read statements that the ‘failsafe’ boot will not just work ? Please advise as I would like to check those threads because that is mostly NOT true, and if there are incorrect posts on our forum causing confusion I would like to post to clarify/correct such threads.
When you select the failsafe boot, it will boot your PC with a number of boot codes, two of which are: ‘nomodeset’ and ‘x11failsafe’. In fact, I suspect with just ‘x11failsafe’ boot code alone that would have been sufficient in your PC’s case as it typically forces the load of the FBDEV graphic driver.
If one is using a default openSUSE-11.3 or 11.4 install, without the proprietary nVidia driver installed, and if one uses the boot code ‘nomodeset’ (without the ‘x11failsafe’ option) openSUSE will boot to the ‘nv’ video graphic driver. There has been more development recently in to the ‘nouveau’ video graphic driver and hence newer graphic hardware will typically work better with the ‘nouveau’ than it will with the ‘nv’. Typically the ‘nv’ will work better with a smaller selection of older hardware (in my experience) than the ‘nouveau’ driver.
I have the nvidia-173xx drivers working on a nVidia FX5200 AGP card with the xorg that comes with openSUSE-11.4. I have the nvidia-270xxx drivers working on a different PC with a GTX260 nVidia card on the xorg that comes with openSUSE-11.4 . It ‘just works’ ! I installed in both cases ‘the hardway’ (which is not hard).
Did you do something to your repositories to install unstable versions of X ? Or did you apply a kernel update ? Because this DOES just work with the stock 11.4 Xorg.
Can you point to the arch linux link, because I am concerned that arch linux post is being taken out of context which can be HIGHLY misleading to our users. We want clarity, and not confusion if at all possible.
It also would help if you advised as to your openSUSE version ? YES there are BIG differences in graphics behaviour between openSUSE versions. And indicating EXACTLY what nVidia hardware you have on your PC would help put your post in context. Else given the wide disparity in hardware, its not so useful without such information.
Reference your journey with nomodeset and kms and failsafe, it strikes me from reading your post that you were not clear as to what each of those options do. It might help a bit (at least start to help) if you wish to understand more to read the practical theory guide on openSUSE graphics, where I dumped some of my limited knowledge: openSUSE Graphic Card Practical Theory Guide for Users and also read the openSUSE-11.4 release notes: openSUSE 11.4 Release Notes
To answer this question, yes there is a small chance, but it will be openSUSE version specific and also it will be your hardware specific, and also specific to any custom kernels and versions of X you may have installed, and without that information its not possible to be specific. In the VAST majority of cases this will ‘just work’ , but to confirm if you have encountered a proprietary driver issue can only attempted to be answered with the information noted.
>
> Vinster171;2326290 Wrote:
>> Finally, althought it was mentioned in lots of places that Failsafe boot
>> would just not work, I tried it out on a hunch… and what do you know,
>> my system booted almost properly!
>>
>
> Where did you read statements that the ‘failsafe’ boot will not just
> work ? Please advise as I would like to check those threads because that
> is mostly NOT true, and if there are incorrect posts on our forum
> causing confusion I would like to post to clarify/correct such threads.
>
> When you select the failsafe boot, it will boot your PC with a number
> of boot codes, two of which are: ‘nomodeset’ and ‘x11failsafe’. In
> fact, I suspect with just ‘x11failsafe’ boot code alone that would have
> been sufficient in your PC’s case as it typically forces the load of the
> FBDEV graphic driver.
I have at least 3 machines available to me which use various older Nvidia
video chips on the mb. All 3 run into booting problems because the BIOS
setup configures the Nvidia chip to use an IRQ which is unacceptable to the
kernel as it conflicts with ACPI settings. As a result, monitor info is not
available to the system which cascades to the rest of the configuration and
you get 640x480 displays at best.
The boot parameter work around in these cases is “irqpolling”. I would
suggest adding that to even the failsafe boot. It seems to get me the best
intermediate result - at least I can boot and see what I’m doing to continue
the battle. I’m not sure what the total implications of the irqpolling is
on performance as once I can boot to the point where I can install the
proprietary drivers after which I can remove the parameter and press on.
I’ve have lost too much hair fighting with video so I am really grateful to
helpful soul who pointed out this parameter to me!
>
> If one is using a default openSUSE-11.3 or 11.4 install, without the
> proprietary nVidia driver installed, and if one uses the boot code
> ‘nomodeset’ (without the ‘x11failsafe’ option) openSUSE will boot to the
> ‘nv’ video graphic driver. There has been more development recently in
> to the ‘nouveau’ video graphic driver and hence newer graphic hardware
> will typically work better with the ‘nouveau’ than it will with the
> ‘nv’. Typically the ‘nv’ will work better with a smaller selection of
> older hardware (in my experience) than the ‘nouveau’ driver.
>
> Vinster171;2326290 Wrote:
>> …“The nvidia-173xx and nvidia-96xx driver packages have been removed
>> from our repositories as they are incompatible with newer xorg servers.
>> This can only be fixed by an upstream update, which has not happened
>> yet.”
>>
>> Is there any chance that my problem is related to this issue with the
>> NVIDIA driver?
>
> I have the nvidia-173xx drivers working on a nVidia FX5200 AGP card
> with the xorg that comes with openSUSE-11.4. I have the nvidia-270xxx
> drivers working on a different PC with a GTX260 nVidia card on the xorg
> that comes with openSUSE-11.4 . It ‘just works’ ! I installed in
> both cases ‘the hardway’ (which is not hard).
>
> Did you do something to your repositories to install unstable versions
> of X ? Or did you apply a kernel update ? Because this DOES just work
> with the stock 11.4 Xorg.
>
> Can you point to the arch linux link, because I am concerned that arch
> linux post is being taken out of context which can be HIGHLY misleading
> to our users. We want clarity, and not confusion if at all possible.
>
> It also would help if you advised as to your openSUSE version ? YES
> there are BIG differences in graphics behaviour between openSUSE
> versions. And indicating EXACTLY what nVidia hardware you have on your
> PC would help put your post in context. Else given the wide disparity in
> hardware, its not so useful without such information.
>
> Reference your journey with nomodeset and kms and failsafe, it strikes
> me from reading your post that you were not clear as to what each of
> those options do. It might help a bit (at least start to help) if you
> wish to understand more to read the practical theory guide on openSUSE
> graphics, where I dumped some of my limited knowledge: ‘openSUSE
> Graphic Card Practical Theory Guide for Users’
> (http://tinyurl.com/4wcyz8l) and also read the openSUSE-11.4 release
> notes: ‘openSUSE 11.4 Release Notes’
> (http://www.suse.com/relnotes/i386/openSUSE/11.4/RELEASE-NOTES.en.html)
>
>
I think it would be useful to specify exactly which chipset you encountered these on. It may also be useful to specify exactly what mother boards.
I’ve successfully run the opensource and proprietary nvidia drivers on nivida AGP FX5200 , nVidia PCI GeForce 8400GS, nVidia PCIe GTX260, nVidia PCIe G210 and never seen such a problem. Motherboards were a variety, including ASUS P6T Delux, MSI KT3 Ultra, Asus A78NX, EPOX 8K7A. None had the problem you noted. So why do I mention this ?
Because without specifics, we risk ending up with hearsay, which can lead users who don’t know any better to draw misconclusions thinking such hiccups globally apply every time there is the slightest hiccup. But if we can be specific as to our hardware that either works or fails, we can be scope where the problems may be and where they may not be.
I’ll start by saying that it is not my intention to mislead anyone on this issue, and as mentioned above, since I’m really a beginner as far as Linux is concerned, it is completely possible that the problem I encountered was my own doing. Unfortunately, I don’t seem to be able to locate the posts concerning “nomodeset” and other “workarounds”, but I don’t think these were on the openSuse forum (although now I see that it was first worded that way… wish I could edit that post now!). Hence, I guess I should probably not have tried these things, although in the end, it did not cause further problems.
The computer is a Compaq CQ50, AMD Turion double-core RM-70, with a NVIDIA GeForce 8200M graphic card. As for the openSuse version, I’m running 11.4, and the latest updates were installed. I did not do any changes to the repositories, except adding the Pacman in order to get multimedia codec. Of course, when I used the “one-click” install for the NVIDIA drivers, the NVIDIA repository was installed, but I have since removed it.
The Arch Linux link can be found here : Arch Linux - News: nvidia-173xx and nvidia-96xx removed from [extra]. As mentioned above, I’m sorry if the information provided at first might have seemed incomplete/misleading : I am still new at this. Also, thank you for the links : I’ll go have a look and try to learn something!
Now, the system is running well with the “nouveau” driver, with only a couple of minor issues. What would you suggest I try in order to get the latest NVIDIA proprietary driver up and running?
>
> Will Honea;2326699 Wrote:
>> I have at least 3 machines available to me which use various older
>> Nvidia video chips on the mb. All 3 run into booting problems because
>> the BIOS setup configures the Nvidia chip to use an IRQ which is
>> unacceptable to the kernel as it conflicts with ACPI settings.
>
> I think it would be useful to specify exactly which chipset you
> encountered these on. It may also be useful to specify exactly what
> mother boards.
>
> I’ve successfully run the opensource and proprietary nvidia drivers on
> nivida AGP FX5200 , nVidia PCI GeForce 8400GS, nVidia PCIe GTX260,
> nVidia PCIe G210 and never seen such a problem. Motherboards were a
> variety, including ASUS P6T Delux, MSI KT3 Ultra, Asus A78NX, EPOX 8K7A.
> None had the problem you noted. So why do I mention this ?
>
> Because without specifics, we risk ending up with hearsay, which can
> lead users who don’t know any better to draw misconclusions thinking
> such hiccups globally apply every time there is the slightest hiccup.
> But if we can be specific as to our hardware that either works or fails,
> we can be scope where the problems may be and where they may not be.
Good points - pardon the omission. My primary system - and the worst of the
bunch - is a Lenovo j3000, Athlon X2-64 3600+, Nvidia gf6100. Vintage 2007,
Lenovo mb with Phoenix bios
The second is a Compaq/HP with Athlon X2-64 4800+, Nvidia gf6150 vintage
2009. Sorry, model the Compaq model escapes me just now. I can check
tomorrow if it is important
The third is a donated box that I suspect is a bit of a mongrel. The case
is a Dell but it has AMD 32-bit processor at 3Gbs which has always struck me
as odd as I’m not familiar with any other Dell machines with AMD processors.
It uses one of the gf6xxx series (I want to say 6250 but I’m not absolutely
sure - it’s later the other 2 but not by much). I’ll definitely check the
DMI on that mb. Now that you bring it up I also questionable whether this
may be a replacement mb since this is supposedly the oldest machine in the
office and that doesn’t match up with the chipset dates. The Dell model
would date it to 2006 or so. As I said, it was a donation from a little old
lady who only used it for email, so who knows?
The sympthom of the problem addressed by the irqpolling parameter is easy to
spot during boot or by inspecting the boot log. After the video chip is
detected, there will be an error flag complaining about not being able to
read the SMB bus due to a conflict between the video chip irq and one of the
ACPI regions. This has the effect of not retrieving the DMI info from the
monitor so the system has no way to determine any video modes and drops to
the failsafe 640x480. By forcing IRQ polling, the system probes for the DMI
info with alternate procedure and, so far, has been successful in selecting
an appropriate video mode on these machines. Once the Nvidia proprietary
driver is installed I remove the polling parameter and the Nvidia module is
able to appropriately read the DMI info without error.
This is very Arch Linux specific. According to Distrowatch, Archlinux is using xorg-server version 1.10.1. However openSUSE-11.4 is using the older xorg-server version 1.9.3. Hence this Arch Linux comment does not apply to openSUSE-11.4. It may ‘eventually’ apply to tumbleweed-11.4 IF tumbleweed updates the xserver, but so far that has not happened.
Reference updating to the latest nVidia proprietary graphic driver, and reference graphic card theory in general, I wrote a very basic practical theory guide for openSUSE users here: openSUSE Graphic Card Practical Theory Guide for Users with the nVidia section here: openSUSE Graphic Card Practical Theory Guide for openSUSE nVidia graphic card Users and in that you will note a link to a proprietary nVidia driver installation guide taking you here: SDB:NVIDIA drivers - openSUSE That talks about an “Easy Way”, a “Repository Way”, a “Repository Legacy Way”, and a “Repository Way with Zypper”. I do not use ANY of those ways. Instead if you look at the bottom of that proprietary installation guide page you will see a link to the “Hardway” (which by the way is not hard) which will take one here: SDB:NVIDIA the hard way - openSUSE
Note the entries in that “Hardway” wiki page are out of order (from my perspective), probably done so to make the method appear easy, but it adds steps that are not IMHO necessary , and it puts only at the end steps that may be (from my perspective) necessary. For my FX5200 (with the nvidia-173xx drivers) and GTX260 (with the nVidia-270xxx driver) with which I built custom drivers I disabled KMS with YaST and built the drivers after booting with nomodeset specified. I did not find blacklisting the nouveau driver necessary on openSUSE-11.4 with the latest nVidia drivers. YMMV.
