ATA driver races with interrupts

[Daniel Eriksson]

Ville-Pertti Keinonen wrote:

> The attached patch should enable serialization for the 
> controller, which is the only completely reliable fix
> (without chipset documentation) according to Søren.
> Obviously it reduces performance since it doesn't 
> allow both channels to operate simultaneously.

After applying the patch to sources cvsuped 2004.08.04.01.00.00, everything
seems to be working correctly. I've run a few stress-tests successfully, and
I also started smartd (/usr/ports/sysutils/smartmontools/) which previously
always managed to lock up one of the SATA channels.

I am going to run some more stress-tests later today, but it looks pretty
promising.

/Daniel Eriksson