Bug 1397

Can you provide an output of nvidia-bug-report.sh ?
Once that said, you could probably better try the driver from
rpmfusion-nonfree-updates-testing once it will hit the repository.

Created attachment 480 [details]
nvidia-bug-report.log

this log was generated running the xserver with vesa driver, because I cant bootup.

Can you show /var/log/Xorg.0.log.old then ?

Created attachment 481 [details]
nvidia-bug-report.log (correct)

I remenber that I could press ctrl+alt+fx to change terminal, this is a log running nvidia driver (I enabled testing repo, to install that kmod)

Comment on attachment 480 [details]
nvidia-bug-report.log

wrong log

Please update to rpmfusion-nonfree-updates-testing version.
If this doesn't solve your problem, please re-open.

(In reply to comment #6)
> Please update to rpmfusion-nonfree-updates-testing version.
> If this doesn't solve your problem, please re-open.
> 

I made this running the correct nvidia log!

(In reply to comment #7)
> (In reply to comment #6)
> > Please update to rpmfusion-nonfree-updates-testing version.
> > If this doesn't solve your problem, please re-open.
> > 
> 
> I made this running the correct nvidia log!
> 

sorry now I see that is a new driver version on testing, yesterday isnt available.
I will test it...

Created attachment 482 [details]
nvidia-bug-report.log, driver from testing (newer version)

The error persists in the driver from the testing repo. Here is the log

You report is invalid, you need to reboot in order to have a driver that match you kernel module.

Created attachment 483 [details]
A bug report when the pc doesnt boot properly

The problem persists, but in a intermittent way, sometimes it can boot, sometimes doesnt. Here is the log when a crash hapens

Created attachment 484 [details]
A bug report when the pc boot properly

This is the report when fedora is initialized correctly.

Problem still there...

(In reply to comment #13)
> Problem still there...
> 

There is indeed a problem with the nvidia module been loaded into memory but not showing into /proc/interrups.
You can give a try with adding options nvidia NVreg_EnableMSI=1 into /etc/modprobe.d/nvidia.conf
and reboot

Please forward this report to nvidia. You can reference this bug. And forward us the answear.
Thx for the report.

Created attachment 488 [details]
log error with official nvidia 256.53 driver

Ok, I send them a error but no reply until now...

still reproducible in kernel 2.6.34.6.53

(In reply to comment #16)
> still reproducible in kernel 2.6.34.6.53
Where you able to reproduce with kernel-2.6.33.8-149.fc13 ?
http://koji.fedoraproject.org/koji/buildinfo?buildID=190700

(In reply to comment #17)
> (In reply to comment #16)
> > still reproducible in kernel 2.6.34.6.53
> Where you able to reproduce with kernel-2.6.33.8-149.fc13 ?
> http://koji.fedoraproject.org/koji/buildinfo?buildID=190700
> 

NO!

(I put the "option nvidia NVregEnableMSI=1" on /etc/modprobe.d/options.conf, and seems to be working properly in kernel 2.6.34.6.53)

Seems that the problem is not solved with the NVreg_EnableMSI=1 option, but is happening less often

(In reply to comment #19)
> Seems that the problem is not solved with the NVreg_EnableMSI=1 option, but is
> happening less often
> 

know, with some fedora updates... nvidia driver dont boot at all. even with NVreg_EnableMSI=1

Did you get an answear from nvidia about this issue ?
Did you send the nvidia-bug-report.sh by email or via nvnews.net ?

(In reply to comment #21)
> Did you get an answear from nvidia about this issue ?
> Did you send the nvidia-bug-report.sh by email or via nvnews.net ?
> 

I send them by email this bug and posted it in nvnews.
No answer so far...

I was digging on the web and find this topic

http://ubuntuforums.org/showthread.php?t=1002287&page=2

Looking in the nvidia log error I find this messages:

vmap allocation for size 16781312 failed: use vmalloc=<size> to increase size.
NVRM: RmInitAdapter failed! (0x26:0xffffffff:1027)

Could this be a vmalloc problem too?

adding in the kernel line (/etc/grub.conf) vmalloc=256MB it works..

Couldnt fedora find the right vmalloc size on the fly?

Does it behave better with the driver from rpmfusion-nonfree-updates-testing ?

NO, new nvidia driver (260.12) from testing didnt solve this.. I need to put vmalloc=<256MB> in the boot kernel line to work.

but this new driver solves a bug that was showing in openoffice:
https://bugzilla.redhat.com/show_bug.cgi?id=643652

how fedora discover the right vmalloc size without this flag? shouldnt fedora discover this on-the-fly?

(In reply to comment #27)
> how fedora discover the right vmalloc size without this flag? shouldnt fedora
> discover this on-the-fly?
I don't know. Might worth to ask on nvnews.net
Do you have an updated bios ?

(In reply to comment #28)
> I don't know. Might worth to ask on nvnews.net
> Do you have an updated bios ?

I update the motherboard bios. In the website of the manufactor of my notebook there isnt a video card bios update.

Can you reproduce (again) with lastest 260.19.21 from rpmfusion-nonfree-updates-testing ?
If yes, then please report this to nvnews.net there is nothing I can do about this as a package maintainer.

Thx

(In reply to comment #30)
> Can you reproduce (again) with lastest 260.19.21 from
> rpmfusion-nonfree-updates-testing ?
> If yes, then please report this to nvnews.net there is nothing I can do about
> this as a package maintainer.
> 
> Thx
> 

I think that nvidia guys wont reply this, because is a documented bug. Fedora that needs to know how vmalloc size was supossed to be. Nvidia drivers break after a update on the kernel, this vmalloc size maybe was changed in the kernel config.

It still weird because your are the only one affected by the problem.
Is there a non-default option you have made with one or another kernel module ?

(In reply to comment #32)
> It still weird because your are the only one affected by the problem.
> Is there a non-default option you have made with one or another kernel module ?
> 

No, no odd option.

Maybe the size of memory of the video card. My card have 512mb of video memory. Could be this? Or my card in specific?

I haven't seen in this report that you have forwarded this problem to nvidia for support from their side (as described in nvidia-bug-report.sh tool). Is it done already ?

(In reply to comment #34)
> I haven't seen in this report that you have forwarded this problem to nvidia
> for support from their side (as described in nvidia-bug-report.sh tool). Is it
> done already ?
> 

Are you asking if I send this bug to nvidia? 

Yes I do, but I think it wont get a answer, because is a documented bug. Maybe from kernel side, that doesnt get the right vmalloc size for this gpu.

BTW, Have you tried using the kernel-PAE since you are on i686 IIRC ?

You can even try with the lastest nvidia driver:
yum install kmod-nvidia-PAE --enablerepo=rpmfusion-nonfree-updates-testing

Thx for your feeback.

(In reply to comment #36)
> BTW, Have you tried using the kernel-PAE since you are on i686 IIRC ?
> 
> You can even try with the lastest nvidia driver:
> yum install kmod-nvidia-PAE --enablerepo=rpmfusion-nonfree-updates-testing
> 
> Thx for your feeback.
> 

I update to fedora 14 (x86_64), and the problem doesnt exists anymore. But I think this can be only related to 32bits arquiteture. As I sayed, it can be a problem on the kernel side.

OK, so closing as worksfor'us'