Re: HPT374 IDE problem with 2.6.21.* kernels

[Andrew Morton <akpm@linux-foundation.org>]

On Wed, 30 May 2007 11:30:00 +0200 (CEST)
Geller Sandor <wildy@petra.hos.u-szeged.hu> wrote:

> Hi,
> 
> I saw a similar report yesterday with '2.6.21.1 - 97% wait time on IDE 
> operations' subject.
> 
> After upgrading from 2.6.20.7 kernel to 2.6.21.1 my system started to 
> reset infrequenly the IDE bus. In the syslog DMA timeout, resetting IDE 
> bus messages appeared. I've changed the two disks attached to the HPT374 
> controller, and always the first disk had problems. I've replaced cables, 
> plugged the disks into other IDE ports, but it was only a matter of time 
> to experience an IDE reset. When I upgraded to 2.6.21.3 the resets became 
> much more frequent, this time even DMA was disabled too on the first disk. 
> I turned DMA back Manually with hdparm, and a few seconds of intense IO 
> activity resulted in another IDE reset.
> 
> Reverting back to 2.6.20.12 the problem seems to be gone. BTW I'm using 
> the PATA driver for the HTP374, not the libata one.
> 
> Is this a known problem/ is there a way I can help locating the cause of 
> the problem?
> 

(cc's added)