On Tue, 21 Mar 2000, Simon Kirby wrote:
> On Tue, Mar 21, 2000 at 12:16:22PM -0800, Andre Hedrick wrote:
>
> > On Tue, 21 Mar 2000, Simon Kirby wrote:
> >
> > > - hdparm -u 1 vs hdparm -u 0 on all drives does not seem to make a
> > > difference.
> > >
> > > - hdparm -d 0 vs hdparm -d 1 on all drives tends to make the problem
> > > happen even more often (guessing 5 times as often). Tested with
> > > both -u 1 and -u 0.
> >
> > What is the rest of the hardware list?
> > You are fingering the difference between DMAING and PIO.
> > If you are trying to DMA with out autodma........you can and usually will
> > be "SHAFTED"!
>
> What do you mean by trying to DMA without autodma?
>
> I'm using an ASUS P2B-DS (PIIX4 tuning and all enabled in kernel config)
> with two Celerons. 128MB ECC SDRAM. Three WD hard drives are attached:
>
> hda: WDC WD273BA, 26105MB w/1961kB Cache, CHS=3328/255/63, UDMA(33)
> hdb: WDC WD273BA, 26105MB w/1961kB Cache, CHS=3328/255/63, UDMA(33)
> hdc: WDC AC28400R, 8063MB w/512kB Cache, CHS=16383/16/63, UDMA(33)
>
> The first two are UDMA/66 drives, but everything is plugged into the
> UDMA/33 on-board controller.
I missed the beginning of this thread. I have the same board you
have. I upgraded the BIOS to 1012 (maybe 1011 - I forget
exactly). Any time I enable the full set of options (AUTODMA,
PIIX tuning, etc, ie: what Andre says should be used), I eventually
get DMA timeouts. If I'm careful, ie: booted into single user
mode, root mounted r/o, DMA is turned off automatically when the
failure occurs and the system recovers. If I'm running say, X,
when the DMA failure occurs, sometimes the system can hang. I
don't know exactly the result since I don't try testing DMA on a
r/w drive any more. If I use something like dd if=/dev/hda
of=/dev/null, I'll get DMA timeouts within a couple of seconds.
Kernels: lots: 2.2.15pre14, 2.3.99pre2, to name a few.
Drives: Seagate. I can't get to my home system right now, so I
can't be more specific.
Does this match your experiences?
Mike
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Mar 23 2000 - 21:00:36 EST