[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Problem with Micron Transport Trek 2 and DMA



Sorry about not listing the kernel version it is a 2.4.27-i686.  I do
not have anywhere to disable DMA for the hardware.  It tried disabling
it via the kernel option at boot time but that did not help.  Turning
off DMA using hdparm did not help either.  I have not tried 2.6.x yet
but I guess that is my next step.

On Fri, 2004-09-17 at 20:06, Bob Proulx wrote:
> Robert Goley wrote:
> > I have a Micron Transport Trek 2 that I have loaded Debian (Sarge) on. 
> > I repeatedly get DMA errors on the main console.  It complains then says
> > interrupt lost and inmost cases it resets and works for a while.
> 
> I have a desktop motherboard with a via chipset that has similar
> issues.  I can't say as I am happy with Via at the moment.  But they
> are one of the more popular chipsets.  On any 2.4 kernel on my
> motherboard I had DMA errors just like you report.  Miraculously with
> 2.6 the errors have disappeared.  On 2.4 I needed to disable DMA at
> boot time.
> 
> > I did not see anywhere in the bios to turn DMA off.  I tried turning
> > it off via "hdparm -d0 /dev/hda".  That stabilized it some.
> > Speed/performance was lost but because of the pauses DMA errors were
> > giving it was faster.  The down side was that then it had an DMA
> > error and did not recover.  I am not sure what else to try.  Anyone
> > have any ideas?
> 
> You did not say what kernel you were using.  Saying Sarge could mean
> many things.
> 
>   http://packages.debian.org/cgi-bin/search_packages.pl?keywords=kernel-image&searchon=names&subword=1&version=testing&release=all
> 
> So let me suggest installing the latest available sarge kernel.  No
> guarentee that it will have a driver that has more knowledge of your
> chipset.  But it gives the best chance of it.  If that does not solve
> the problem I would point to sid and installing the latest 2.6.8
> kernel.
> 
> If neither of those solve the problem then I think you have no other
> option than to disable DMA.  Other than digging into the root cause of
> the problem at the level of the source code of the kernel driver.
> 
> Bob



Reply to: