lkml.org 
[lkml]   [2004]   [Oct]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: ATA/133 Problems with multiple cards
On Sun, 17 Oct 2004, Bartlomiej Zolnierkiewicz wrote:

> > i did actually kind of get the card's working together but ran into
> > another problem.
> >
> > when i boot with ide=nodma and then turn on dma manually on all the other
> > cards / board chipset etc... they all function fine
> >
> > then i can only turn the dma up to ATA/100 if i set it to ata/133 it will
> > cause the errors. I assume this is something todo with the promise bois
> > not setting up the 3rd card at boot time. It only shows drive listing for
> > 2 of the 3 cards.
>
> There were very similar problems reported in the past and they were
> fixed by replacing power supply with a better one.

I dont beleave the power supply to be a problem here. Since the problem
is local to just this 1 card. No matter where i place it on the pci bus
or swap it with the other promise cards the 3rd promise card on the bus
always fails. I have placed this also below / above a network card + scsi
card in the same machine. A volt meter on the power supply is giving
correct results. If the power supply was to be failing under load would
this not leed to more unpredictable lockups ?

eg running both commands at the same time.
cat /dev/zero > /dev/hdi
cat /dev/zero > /dev/hdk
will not cause a lockup
cat /dev/zero > /dev/hdm
cat /dev/zero > /dev/hdo
will always cause a lockup with as soon as the disks start writing.
I have tried with all combinations of the disks / cables.

running a single command cat /dev/zero > /dev/hdm will not cause a lockup
nor will the other command. Only when they are both running together will
it cause a problem. I can even run
cat /dev/zero > /dev/hdi
cat /dev/zero > /dev/hdk
cat /dev/zero > /dev/hdm
together without causing a lockup.
hdi and hdk are located on the 2nd promise card. hdm hdo are on the 3rd
card.
> Also you shouldn't need to use "ide=nodma" and play with hdparm,
> driver should tune the best mode available. If this doesn't work then
> something needs fixing.

> You can find out if BIOS/driver configures cards correctly by
> comparing PCI config space (lspci -xxx) for working/non-working
> controller.

The output of the 3 cards matches except for the memory address.


James

> > Unfortunatly this generated another problem.
> > When read from both drives at the same time it functions normally and
> > see resonable performance. When i attempt to write to both drives it will
> > cause the machine to lockup.
> >
> > James
> >
> >
> >
> > On Thu, 14 Oct 2004, Bartlomiej Zolnierkiewicz wrote:
> >
> > > On Thu, 14 Oct 2004 13:12:42 -0500, Ian Pilcher <i.pilcher@comcast.net> wrote:
> > > > James Stevenson wrote:
> > > > >
> > > > > i seem to have run into an annoying problem with a machine which has
> > > > > 3 promise ata/133 card the PDC20269 type.
> > > > >
> > > >
> > > > ....
> > > >
> > > > >
> > > > > Does anyone have an explenation of why this can happen ?
> > >
> > > * check power supply
> > > * compare PCI config space of the "failing" controller to the one which
> > > is "working" (assuming that identical devices are connected to each),
> > > maybe firmware/driver forgets to setup some settings
> > >
> > > > Promise cards don't support more than two per machine. If you can get a
> > > > third card to work in PIO mode, consider it an added (but unsupported)
> > > > bonus.
> > >
> > > AFAIR people have been running 4-5 cards just fine

--
--------------------------
Mobile: +44 07779080838
http://www.stev.org
6:30pm up 1 day, 3:12, 2 users, load average: 0.07, 0.03, 0.00
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 14:07    [from the cache]
©2003-2011 Jasper Spaans