On Sat, 2002-01-26 at 06:31, Borsenkow Andrej wrote:
> On ���, 2002-01-25 at 09:26, Borsenkow Andrej wrote:
> > 
> > > 
> > > i have one of these old HP7200i cd-rw (IDE) devices that need dma
> > > turned-off in order for a successful burn/write session:
> > > 
> > > ie: #hdparm -d0 /dev/hdc
> > > 
> > > Otherwise, I get buffer-underruns and or the kernel locks-up
> > horribely.
> > > 
> > > I believe i also had some issues with devfs & this ide cd-rw. forget
> > > tho.
> > > 
> > > thought i'd post this to ensure that it's atleast known about.
> > > 
> > > Again, yes, i need to turn-off dma for this cd-rw only. (my other
> > hdd's
> > > have dma turned on, etc).  I think in the specs, it even says nothing
> > > about dma, etc.
> > > 
> > 
> > Yes, that irritates me as well. I'll try to contact devfs list, but no
> > promises :-)
> > 
> 
> In the meantime you could try
> 
> echo -n "using_dma:0" >  /proc/ide/ide1/hdd/settings
> 
> as root. Replace hdd with your drive. I would be very interested if it
> works (I too need to turn it off on my Philips so I'll do it anyway).
> 
> -andrej
> 
> 

ok, andrej. my memory is returning on this one. becuase of devfs, i
couldn't used hdparm -d0 /dev/hdc to turn-off dma for this particular
ide-cdrw device.  becuase so, i had to turn off devfs in lilo by doing
the devfs=nomount append option.

since then, i haven't tried devfs again becuase of problems. but since i
*am* on kernel 2.4.17-5mdk, i could give it another shot i guess.

in the mean time, your tip does effectively turn-off dma on /dev/hdc
with devfs=nomount.  i used hdparm -d /dev/hdc to verify that it was
turned-off effectively.

i'm sure it *will* work if devfs allows this.  only way to verify at my
end would be to try to burn another cd-r since i can't verify with
hdparm on the ide-scsi device with devfs=mount.




Attachment: signature.asc
Description: This is a digitally signed message part

Reply via email to