[haiku-development] Re: ATA stack testing

On Wed, 29 Apr 2009 00:46:02 +0200 (CEST), Marcus Overhagen wrote
>> Axel reported that his Eee wouldn't boot anymore with the ata bus_manager and
> Oh that is bad. Is KDL still working?

Nope, completely frozen. It also doesn't seem to be an interrupt storm as a
panic right in the interrupt handler doesn't get triggered. Could be that it's
an interrupt storm on another interrupt line though. Something similar I've
seen/fixed at BeGeistert where not forcing the system management interrupts
off when starting a EHCI controller apparently caused SMIs to get triggered
and completely stalled the system. I don't really see how this would fit in
here though or why the ide bus_manager didn't trigger it.

> Well. I think the clearing of BM interrupt bit in the interrupt 
> handler has a lot of unwanted side effects. Although I think its the 
> only correct way to acknowledge the interrupt, at least the 
> silicon_image_3112 is now completely broken. I think I should just 
> restore ide_adapter.cpp from svn history, and apply the changes to 
> copied source file. I can do so on Wednesday evening, unless you 
> want something else.

What kind of side effects? That you don't have the same status values in
finish_dma() later or actual interrupt related problems?

> My intention certainly wasn't stopping your work. I just tried doing 
> small changes to get it working on my system.

No worries, I'm just a bit busy again (it's on and off lately). I will
continue with the rework then, moving your changes into a new ata_adapter.

Regards
Michael

Other related posts: