[PlanetCCRMA] FC2 "experimental" kernel install dependency error

Chris Miles (PlanetCCRMA) PlanetCCRMA@ir.compulink.co.uk
Fri Nov 5 11:42:00 2004


I had some trouble with the 6111 drivers when I was running the ll 
kernel on rh9. I found a fix on the web for the 6111 but had to give up 
on the ll kernel as my keyboard would hang.

The "fix" was to do with opengl and tls. there is one libnvidia-tls in 
/usr/lib and one in /usr/lib/tls. Someone had found that overwriting one 
with the other allowed X to come up.

I've no idea what they were really playing with or whether my problems 
came from this change. YMMV.

Chris

Brad Fuller wrote:

> Ok, I got the kernel 2.6.7-1.437.1.11.rhfc2.ccrma loaded. 
> Unfortunately, Xwindows didn't start up. I had to change the 
> nvidia6111 driver to the older one -- and then X started up. Why would 
> this be? I was going to install 6111 again -- the nvidia installer 
> said it was already installed (thought maybe there was some 
> interaction between this kernel and nvidia6111.)
>
> Anyone come across this?
>
> brad
>
> Fernando Pablo Lopez-Lezcano wrote:
>
>>On Wed, 2004-11-03 at 22:43, Brad Fuller wrote:
>>  
>>
>>>Tried to install the new FC2 ccrma kernel:
>>>
>>># apt-get install planetccrma-core
>>>
>>>but received a conflict that 
>>>
>>>    
>>>
>>>>file /usr/lib/libasound.so.2.0.0 from install of
>>>>      
>>>>
>>>alsa-lib-1.0.5-1.cvs.rhfc2.ccrma conflicts with file from package
>>>libasound2.1.0.6-16.rhfc2.at
>>>    
>>>
>>
>>That's a conflict caused by unfortunate non-standard naming of the alsa
>>library packages at atrpms. I should (will in the next release) include
>>an "Obsoletes:" clause for them, but the proper fix will be for atrpms
>>to return to the "standard" naming of those packages ("standard" as in
>>"what all others do, including the base redhat/fedora distribution"). 
>>
>>  
>>
>>>A LOT of packages depend on the libasound2.10.06. Any way to get
>>>around this? 
>>>    
>>>
>>
>>You can either install the planetccrma-core critical components one by
>>one, or get rid of the libasound2 packages and replace them by the
>>equivalent alsa-lib packages (BTW, is there an explicit dependency on
>>libasound2?, there should not be one, try "rpm -q --whatrequires
>>libasound2"). 
>>
>>For the first option, do an "apt-get install kernel" and you will see
>>the choices you have. The most conservative one for fc2 will be
>>kernel#2.6.7-1.437.1.ll.rhfc2.ccrma (that is the one "required" by
>>planetccrma-core), there are other more experimental kernels in the
>>"planetedge" repository. Those have better low latency performance but
>>are riskier to run[*]. 
>>
>>I posted this a while ago:
>>  
>>
>>>The best kernels I currently have for 2.6 are in the "experimental"
>>>repository, they have to be installed explicitly (not through a meta
>>>package), and they are only advertised in the mailing list. 
>>>
>>>Check this post for information on current latest bleeding edge
>>>kernel:
>>>http://ccrma-mail.stanford.edu/pipermail/planetccrma/2004-August/005879.html
>>>And follow the link inside for more details...
>>>
>>>Remember that before trying the apt-get install xxx. you have to add
>>>another line to the /etc/apt/sources.list (same as the planetcore
>>>line, but replacing planetcore with planetedge), and do an apt-get
>>>update. 
>>>    
>>>
>>
>>If you install either (edge or conservative), you also have available a
>>newer alsa than the one included in the kernel itself, for that do an:
>>
>>apt-get install kernel-module-alsa
>>
>>and again you will get choices. Explicitly install the one that matches
>>the kernel you just installed. 
>>
>>-- Fernando
>>
>>[*] I was hoping that by now the experimental voluntary-preempt patches
>>for very good low latency performance would have stabilized a lot, but
>>another approach was taken by Ingo Molnar and we now have a very new
>>"realtime-preempt" patch that promises to be even better. Regretfully it
>>is still quite unstable. I'm trying it out regularly and trying to
>>follow its development but I have not posted experimental packages yet
>>(let me know if any on the list needs to be a guinea pig). 
>>
>>
>>_______________________________________________
>>PlanetCCRMA mailing list
>>PlanetCCRMA@ccrma.stanford.edu
>>http://ccrma-mail.stanford.edu/mailman/listinfo/planetccrma
>>
>>
>>  
>>