On 5/24/06, Quentin Garnier <cube@cubidou.net> wrote:
> On Wed, May 24, 2006 at 10:16:39AM +0200, Pavel Cahyna wrote:
> > On Wed, May 24, 2006 at 02:54:07AM +0200, Zbigniew Baniewski wrote:
> > > On Wed, May 24, 2006 at 02:23:53AM +0200, Pavel Cahyna wrote:
> > >
> > > > Please send us your dmesg.
> > >
> > > Below. Of course, it's "normal" dmesg - I cannot send the one after I=
'll
> > > attach the second drive, from the reason I wrote about.
> > >
> > > > The problematic drive does not have to be attached.
> > >
> > > What you mean? I want to copy a files from there...
> >
> > I mean that the dmesg without the drive is enough.
> >
> > > piixide0 at pci0 dev 7 function 1
> > > piixide0: Intel 82371AB IDE controller (PIIX4) (rev. 0x01)
> > > piixide0: bus-master DMA support present
> > > piixide0: primary channel wired to compatibility mode
> > > piixide0: primary channel ignored (disabled)
> > > piixide0: secondary channel wired to compatibility mode
> > > piixide0: secondary channel interrupting at irq 15
> > > atabus0 at piixide0 channel 1
> >
> > > hptide0 at pci0 dev 19 function 0
> > > hptide0: Triones/Highpoint HPT370 IDE Controller
> > > hptide0: bus-master DMA support present
> > > hptide0: primary channel wired to native-PCI mode
> > > hptide0: using irq 10 for native-PCI interrupt
> > > atabus1 at hptide0 channel 0
> > > hptide0: secondary channel wired to native-PCI mode
> > > atabus2 at hptide0 channel 1
> >
> > So, you have several ATA busses (channels). And the hptide ones have
> > higher numbers, so all devices on them will have higher numbers.
> >
> > Add a line
> > atabus0 at hptide0 channel 0
> >
> > to your kernel config file and recompile. This should work and should b=
e
> > enough.
>
> No;  that will just wire down atabus0, not wd0.  If you only fix
> atabus0, you'll indeed get an atabus1 attaching to piixide0, but you
> will still have a wd0 attaching to that atabus1.
>
> What amuses me greatly is this thread is how Linux mostly lacks that
> ability of wiring this down.  I'm surprised the initial poster got so
> much attention with his issue, considering Linux has the exact same
> issue with SCSI drives (which include SATA drives, due to the way Linux
> handles those...).  Also, RedHat Linux has a "solution" for that, using
> file-system labels.  Of course, it uses the same labels for all
> installations, so that mixing drives from two different RedHat
> installations leads to interesting results.
>

Solaris has /etc/path_to_inst, which is really easy to use for just
this sort of thing.