Old livna kmod

[David Timms]

Hadders wrote:
> David Timms wrote:
>> Hadders wrote:
>>> Hi all,
>>>  I'm just wondering how I get an old kernel module from livna kmod?
>>>
>>> I need to get the module for 2.6.15-1.2054_FC5smp, how would I go that?
>> Have you tried:
>> http://fr2.rpmfind.net/linux/rpm2html/search.php?query=kmod-nvidia
>> http://rpm.pbone.net/index.php3
>> http://ftp.upjs.sk/pub/mirrors/rpm.livna.org/5/i386/
>>
>> Often mirrors will keep around old packages {sometimes by accident}. 
>> You might also like to try a kernel half way between that and current, 
>> to see whether a newer but not current kernel resolves the problem 
>> with the current kernel.
>>
>> Also, I have from my yum cache: {if you still cant find what you are 
>> after}
>> kmod-nvidia-1.0.8762-1.2.6.16_1.2122_FC5.i686.rpm
>> kmod-nvidia-1.0.8762-2.2.6.16_1.2133_FC5.i686.rpm
>> kmod-nvidia-1.0.8762-2.2.6.17_1.2145_FC5.i686.rpm
>> kmod-nvidia-1.0.8762-2.2.6.17_1.2157_FC5.i686.rpm
>> kmod-nvidia-1.0.8774-1.2.6.17_1.2174_FC5.i686.rpm
>>
>> If you boot a non-smp kernel, you might be pleasantly surprised that 
>> other stuff might just work.
>>
>> DaveT.
>>
> Thanks, I'll give those a go and may just try some older kernels, but I 
> thought the kmod stuff came as two RPMs?
On fc5/fc6 it is dkms {extras} which is the enabler. Then the package in 
question must be built to use dkms.

> Hmmm, not sure I agree with your smp diagnosis. After all, if it doesn't 
> work, then we're in big trouble as practically every new processor die 
> now has two cores!
Which work OK with the non-smp kernel in my experience. You might have 
noticed that the grub menu provides both the smp and non smp options ? 
On a machine of mine eg pentiumIV with hyperthreading activated, both 
the kernels operate OK. On another box {dell pe2650} and FC3, there was 
often times when smp wouldn't boot after an update but non-smp was fine. 
It didn't really make much difference to the performance of that machine 
{perhaps it also wasn't under much load}.

Just trying to reduce the scope of the problem to something more workable ;)

DaveT.