Re: AIC7xxx ?
Adam Di Carlo a écrit :
>
> Mark van Walraven <markv@wave.co.nz> writes:
>
> > On Wed, Apr 19, 2000 at 09:56:23AM +0200, Stephane Bunel wrote:
> > > I'just receive 10 DELL servers. Usually I have no probleme to
> > > install the Unofficial Potato. But this server are shipped with
> > > the latest SCSI Adaptec controlers ( 7899 & 7880 ). Those chips
> > > aren't recognized by the boot disk ;( When I boot with a one disk
> > > distribution (TomsRTBT, Kernel 2.0.36), the controlers are
> > > finded and disk are mountables !!
> >
> > Just a long-shot, but when I installed slink on a Dell with 7860 and 7890,
> > I needed aic7xxx=no_probe when booting the rescue disk. Another identical
> > machine didn't need it, though I suspect the BIOS settings were different.
>
> Can someone confirm this?
The aic7xxx=no_probe problem is well known and explained in the
AIC7xxx mailing list. I personaly had this problem on a DELL 6300
without ISA ports. The error comes from DELL (as explained to me)
because they use an overlapped IO addressing map between ISA en PCI
slots. In this case it lead to a complete freeze of the server.
[quoted from ./linux/drivers/scsi/aic7xxx.c, Version 3.2.4/kernel
2.2.14]
/*
* Certain newer motherboards have put new PCI based devices into the
* IO spaces that used to typically be occupied by VLB or EISA cards.
* This overlap can cause these newer motherboards to lock up when
scanned
* for older EISA and VLB devices. Setting this option to non-0 will
* cause the driver to skip scanning for any VLB or EISA controllers and
* only support the PCI controllers. NOTE: this means that if the
kernel
* os compiled with PCI support disabled, then setting this to non-0
* would result in never finding any devices :)
*/
static int aic7xxx_no_probe = 0;
[...]
Stephane.
Reply to: