[PlanetCCRMA] x fails to start with real-time kernel

[Fernando Lopez-Lezcano]

On 05/03/2011 08:18 PM, craig stephenson wrote:
> So if my kernel is 2.6.33.12-149.rt31.1.fc14.ccrma.x86_64.rt, I should
> have "kmod-nvidia-2.6.33....etc"? All I have other than packages for
> the standard Fedora kernels is kmod-nvidia-260.19.36-1.fc14.3.x86_64
> and akmod-nvidia-260.19.36-1.fc14.ccrma.1.1.x86_64.  I don't see much
> in /var/log/messages that would help.

If you don't see a kmod-nvidia for your kernel then X will not work (of 
course).
In my (fc13) laptop I see, for example:

$ rpm -q -a|grep nvidia
kmod-nvidia-2.6.34.7-66.fc13.i686.PAE-260.19.36-1.fc13.i686
kmod-nvidia-2.6.34.8-68.rt0.3.fc13.ccrma.i686.rtPAE-260.19.36-1.fc13.2.1.i686
nvidia-xconfig-1.0-6.fc13.i686
akmod-nvidia-260.19.36-1.fc13.ccrma.2.1.i686
kmod-nvidia-2.6.34.8-68.fc13.i686.PAE-260.19.36-1.fc13.2.1.i686
kmod-nvidia-2.6.33.12-149.rt31.1.fc13.ccrma.i686.rtPAE-260.19.36-1.fc13.2.1.i686
nvidia-settings-1.0-9.fc13.i686
xorg-x11-drv-nvidia-260.19.36-1.fc13.i686
kmod-nvidia-2.6.33.7-149.rt30.1.fc13.ccrma.i686.rtPAE-260.19.36-1.fc13.2.1.i686
xorg-x11-drv-nvidia-libs-260.19.36-1.fc13.i686

What do you see if you run that command?

You could try to run (as root) the script that gets called when the 
system boots and builds the custom package for the kernel being booted, 
this is what is printed in my laptop:

# /usr/sbin/akmods
Checking kmods exist for 2.6.34.8-68.rt0.3.fc13.ccrma.i686.[  OK  ]

If the kmod is not there it will try to build and install it, maybe the 
error messages will give you some clues.

It is strange that you don't seen any error messages during boot (you 
did try to boot without the "rghb quiet" option?). A service gets 
started during boot that will build the kmod if it does not exist - that 
should have failed in your case (otherwise you would have the kmod). 
Maybe the service is disabled? Maybe the akmod is not properly installed?

-- Fernando


>  There is this....
>   "May  3 13:37:04 ProteusIV /usr/sbin/gpm[1570]: Started gpm
> successfully. Entered daemon mode.
> May  3 13:37:04 ProteusIV abrtd: Done checking for unsaved crashes
> May  3 13:37:04 ProteusIV abrtd: Init complete, entering main loop
> May  3 13:37:06 ProteusIV gdm-binary[1804]: WARNING: GdmDisplay:
> display lasted 0.718627 seconds
> May  3 13:37:06 ProteusIV gdm-binary[1804]: WARNING: GdmDisplay:
> display lasted 0.056100 seconds
> May  3 13:37:06 ProteusIV gdm-binary[1804]: WARNING: GdmDisplay:
> display lasted 0.055372 seconds
> May  3 13:37:06 ProteusIV gdm-binary[1804]: WARNING: GdmDisplay:
> display lasted 0.054508 seconds
> May  3 13:37:06 ProteusIV gdm-binary[1804]: WARNING: GdmDisplay:
> display lasted 0.053880 seconds
> May  3 13:37:06 ProteusIV gdm-binary[1804]: WARNING: GdmDisplay:
> display lasted 0.054166 seconds
> May  3 13:37:06 ProteusIV gdm-binary[1804]: WARNING:
> GdmLocalDisplayFactory: maximum number of X display failures reached:
> check X server log for errors"
>
> On Tue, May 3, 2011 at 3:30 PM, Fernando Lopez-Lezcano
> <nando at ccrma.stanford.edu>  wrote:
>> On 05/03/2011 02:00 PM, craig stephenson wrote:
>>>
>>> I tried again, first adding nouveau.modeset=0 to the grub.conf entry,
>>> and then adding vmalloc=512M.  No luck.  I did notice the message
>>> "Ignoring nvidia-kmod as it failed earlier".  Overall the same result
>>> as earlier, with "module nvidia not found".
>>
>> What errors do you see in /var/log/messages?
>> Maybe the binary akmod is not being created on boot?
>>
>> Try booting without the "rghb quiet" part of the kernel boot line to see all
>> the progress reports that happen during the boot process.
>>
>> Do an "rpm -q -a |grep nvidia", you should see "kmod-nvidia" packages for
>> all kernels you want to boot. If there is one missing for the kernel you
>> want to boot you want to see why the build failed, I think there will be
>> messages in the logs.
...