Re: Testing newoldworld pmac miboot 2.6 floppies
On Fri, Sep 17, 2004 at 09:42:09AM -0700, Brad Boyer wrote:
> On Fri, Sep 17, 2004 at 09:55:17AM +0200, Sven Luther wrote:
> > Damnit. But Jens reported it as working on his oldworld box. Could you maybe
> > fill a bug report against kernel-image-2.6.8-powerpc about this ?
> I suspect that the one that worked was SCSI only. All the first generation
> PCI powermacs (7x00,8x00,9x00) had one or two SCSI controllers, but no
> onboard IDE. However, most of the G3 models had IDE and SCSI onboard, and
> used mostly IDE devices.
Ah, maybe ...
> > Ok. We need to know what is your ide controller, and in which udeb it is
> > found, and if discover lists it or not.
> If it is the onboard stuff, it can't be built as a separate module from
> the way the source looks. It appears that ide/ppc/pmac.c is always
> compiled into ide-core if it is enabled.
Yes, indeed it is.
> > Definitively a bug in discover, could you fill a bug report against discover1
> > with your lspci and lspci -n output ?
> You won't get much help out of lspci. These are not PCI devices. The macio
> chip shows up as one huge PCI device, and the macio layer knows how to
/proc/device-tree/aliases then. And this probably means that there is no
chance ever of discover discovering them. Oh well.
> read data from OF to find the sub-devices. I don't think either version
> of discover has any knowledge of macio, which prevents detection of drivers
> like mac53c94, mesh, mace, pmac_zilog, and lots of other older mac stuff.
> The pmac IDE support can handle both PCI and macio devices, but I'm pretty
> sure that only newworld macs came with onboard IDE that showed up as
> separate PCI devices.