ATA driver races with interrupts

[Søren Schmidt]

Ville-Pertti Keinonen wrote:
> Søren Schmidt wrote:
> 
>> Reliable solutions needs reliable HW, we can narrow down the race 
>> window here, but without HW support we cannot close it completely 
>> unless we serialises access to the channels (what lots of OS's does 
>> btw)..
> 
> For now, I moved my SATA disks to the Promise controller, which seems to 
> work with an unpatched -current.

Yeah, Promise makes HW thats designed to work ;)

> If serialization is the only reliable solution, shouldn't it be enabled 
> for all controllers with multiple channels on the same irq that use 
> ata_generic_intr?  It isn't going to hurt performance when only using 
> one channel, and when using multiple channels...currently it's pretty 
> much guaranteed to blow up.

Well, until very recently such controllers didn't exist. I've fixed the 
Intel SATA part, but I have no docs for the VIA SATA part so I have 
nothing to go by there so maybe I should mark that one as broken and 
serialize the accesses...


-- 
-Søren