Re: 2.9test9-mm1 and DAO ATAPI cd-burning corrupt

[Nick Piggin <piggin@cyberone.com.au>]

Rob Landley wrote:

>On Thursday 06 November 2003 13:45, Linus Torvalds wrote:
>
>>On 6 Nov 2003, bill davidsen wrote:
>>
>>>I'm not sure what you mean by faster, burning runs at device limited
>>>speed in CPU time in the  less than 1% range if you remember to enable
>>>DMA. The last time I looked DMA didn't work in either kernel if write
>>>size was not a multiple of 1k, (or 2k?) has that changed?
>>>
>>DMA works fine
>>
>>	IF YOU DON'T USE IDE-SCSI
>>
>>Don't use it. Please. There's no point.
>>
>>It's much more readable to do
>>
>>	cdrecord dev=/dev/hdc
>>
>>than it is to do some stupid "scan SCSI devices" + "dev=0,1,0" or similar
>>totally incomprehensible crap that doesn't even work right.
>>
>>
>>>I'm not sure what you meant by faster, so don't think I'm disagreeing
>>>with you.
>>>
>>Faster as in "it uses DMA for everything, so you can actually burn at full
>>speed without having to worry about it or sucking up CPU".
>>
>>		Linus
>>
>
>Note this still doesn't mean you can scroll large X windows for two or three 
>seconds at a time without burning a coaster.
>
>I had high hopes with the new scheduler, but no.  (Maybe if I niced the heck 
>out of cdrecord...)
>

RT processes should work well with the scheduler. renicing if it is not
RT will help a little bit, but it doesn't do much to help maximum latency.