Re: [gentoo-ppc-user] yaboot has kicked my butt - 5 times?!

[Joseph Jezak]

On 09/25/10 07:45, Mark Knecht wrote:
> On Thu, Sep 23, 2010 at 10:04 PM, Joseph Jezak <josejx@g.o> wrote:
>>  My responses are inline this time. It's easier when there's so much
>> going on!
>>
>> On 09/23/10 16:41, Mark Knecht wrote:
>>> Two pictures posted:
>>>
>>> Top half of boot screen:
>>> http://www.flickr.com/photos/29328985@N03/5018717650/
>>>
>>> Bottom half of boot screen
>>> http://www.flickr.com/photos/29328985@N03/5018718202/
>>>
>> Okay, these look exactly as expected. You've booted into the shell fine
>> and the kernel does detect the hard drive fine. It appears that the disk
>> was not cleanly unmounted, which is what the messages in the bottom
>> picture indicate. Once you get USB working so we can type into the
>> console, we'll take a look at what's actually going on.
>>
>>> Full USB HID support is built as modular. I don't seem to be able to
>>> change it to built in. make menuconfig is only giving me modular or
>>> not set.  (Kernel config USB info this is set is at the end)
>> If you use menuconfig and you go to the "Help" option, it will tell you
>> what dependencies need to be set in order to build the module. Most
>> likely, you did not set the USB subsystem itself to be built in.
>>> lspci says the controller is an Apple controller and the driver is
>>> 'macio' which seems sensible. I see it in the boot screen I think.
>>> That driver is built in, but the PATA_MACIO driver is not:
>>>
>>> (chroot) livecd linux # cat .config | grep MACIO
>>> # CONFIG_PATA_MACIO is not set
>>> CONFIG_ADB_MACIO=y
>>> (chroot) livecd linux #
>>>
>>> Maybe I've mistakenly left the right disk driver out of the kernel
>>> thinking the hardware was SATA based? Does the PATA_MACIO option need
>>> to be set for the Mac Mini? I don't understand how this kernel config
>>> would have ever worked befor unless I'm confusing where it came from.
>>>
>> You're using the old style driver which results in devices named hdX#.
>> It's called IDE_PMAC. The new driver which uses the sdX# naming
>> convention (and uses libpata), is called PATA_MACIO.
>>
>>> Does the append="init=/bin/bash" command allow the kernel to load
>>> drivers or do I need to build USBHID into the kernel to get the
>>> keyboard to work at this level of boot?
>> I would built it in for now, it'll be easier since there's no good way
>> to get into the system to tell it to load the drivers.
>>
>> -Joe
>>
>>
> Hi Joe,
>    OK, I finished the emerge -e @world. No changes. Still doesn't boot.
>
>    I've put the append="init=/bin/bash" back in and booted. I'm at the
> console and this is confusing. It seems that /dev/hda4 is probably
> mounted. I can do
>
> ls -al /
>
> I see all the stuff I'd expect to see - /bin, /boot, etc. - and also
> the two downloads necessary to do the install - portage-latest.tar.bz2
> & stage3-ppc-20100919.tar.bz2 - so I must be looking at the right
> physical hard drive.
>
> However even though I see that stuff simple commands like df don't
> work yielding something like this:
>
> (none)/ #df
> df: cannot read table of mounted file systems: No such file or directory
>
> Additionally, there is nothing at all under /proc. It's empty!
>
> Cheers,
> Mark
>
>
None of these things are weird, they're expected!

When you start the system with "init=/bin/bash", the only thing your
system does is start bash as the only process. Things like mounting
proc, remounting the root file system r/w (and setting up mtab) aren't
done because they're part of the normal startup process.

Can you run fsck.ext3 from the shell that starts when you replace init?

Also, which version of udev are you using? I forgot to ask if you're
running ~ppc instead of ppc, you may want to try switching to the
PATA_MACIO driver in order for the system to work properly.

-Joe