Radeon DRM

Martin Ward macros_the_black at ntlworld.com
Fri Mar 5 15:32:52 PST 2010


On 05/03/10 20:59, Duncan Baynes wrote:
> On Fri, 05 Mar 2010 14:29:44 +0000, Andrew Benton<b3nton at gmail.com>
> wrote:
>    
>> On 05/03/10 11:33, Duncan Baynes wrote:
>>      
>>> I couldn't get the kernel to find the firmware so early in the boot
>>> sequence without using initrd, however I'll have another run at getting
>>> it
>>> working sans initrd if you think that will work.
>>>
>>>        
>> Another option I've tried (which also hung the system if kernel mode
>> setting was enabled)
>> was to build the radeon driver as a module. That way doesn't need the
>> firmware in the kernel,
>> it's enough to put it in /lib/firmware
>>
>>      
>>> Does your system boot if you leave it for about 60 seconds?
>>>
>>> I'm using an ATI 4890 without issue ... so far
>>>
>>>        
>> Yes it does, I'd never waited that long before. I can't see myself
>>      
> waiting
>    
>> that long at each
>> boot. Interesting to see that it does work though.
>> Thanks
>>
>>      
> There is a 60 second timeout period where the kernel strives to locate the
> firmware... so far the only way I could get rid of this delay was to use
> initrd...
>    
its a bug in udev, if relevant firmware is missing it hangs for about 60 
secs trying to load the missing module, two ways to fix this

either grab the missing firm ware from  
http://people.freedesktop.org/~agd5f/radeon_ucode/

or patch udev  with this patch from arch
http://repos.archlinux.org/wsvn/packages/udev/repos/core-i686/firmware-loader-fix-missing-firmware.patch

the missing firm ware is to do with irq handling i believe

martin



More information about the lfs-dev mailing list