[Cbe-oss-dev] ps3 petitboot rc2 release

[Geoff Levand]

Stefan Assmann wrote:
> Geoff Levand schrieb:
>> We have been working toward that, but I think some cleanup is needed on
>> our PS3 patches.  To make things easier, I made a git repository for the
>> PS3 + openSUSE patches here:
>> 
>>   http://git.infradead.org/users/geoff/petitboot-patches.git


Just FYI, I moved this to 

  http://git.infradead.org/ps3/petitboot-patches.git


> Currently we only suppy full dvd images for ppc, sorry. I'll see what I
> can do. This issue has been addressed by patch
> petitboot.yaboot-suse.patch in the patches archive I sent earlier. The
> yaboot file location is /suseboot/yaboot.cnf on the openSUSE install
> media and the relevant entries are:
> 
> message=yaboot.txt
> image[64bit]=inst64
>   label=install
>   append="quiet sysrq=1 insmod=sym53c8xx insmod=ipr            "
> image[64bit]=inst64
>   label=slp
>   append="quiet sysrq=1 install=slp           "
> image[64bit]=inst64
>   label=rescue
>   append="quiet sysrq=1 rescue=1              "
> [...]


I worked it in as this, could you check it?

  http://git.infradead.org/ps3/petitboot-patches.git?a=blob;f=053-yaboot-suse.patch;hb=HEAD


>>> Another part I'm not so fond
>>> of is that if you have multiple installations the order in which
>>> devices are discovered is random.
>> 
>> 
>> Do you have any ideas of how you would handle that?
>> 
>> 
> 
> Devices are discovered via udev and udev doesn't guarantee any
> particular order, so I reorder them on discovery.
> 
> discovered (udev):	reordered:
> /dev/ps3da3		/dev/ps3da1
> /dev/ps3da1     =>	/dev/ps3da2
> /dev/ps3da2		/dev/ps3da3
> 


I'll let you work on that, since it is not critical to operation.


>>> I see the parameter root=/dev/ram0 added to the kernel commandline,
>>> what's the reason for that?
>> 
>> I don't know why.
>> 
>> -Geoff
>> 
>> 
> 
> My guess is that this happens because the openSUSE entry has no root=
> parameter specified so this one is faked. Why would we do that? Maybe
> Jeremy or Ben know more, but in my opinion this should be removed. In
> case it's not specified it's simply not specified. This causes delays on
> booting as there is no /dev/ram0 and udev waits for it to appear,
> ultimately falling back to the real device.

It seems it should behave as you say.  I wonder if some other distro's
initrd expects a 'root=' option.  I'll try some tests next week.

-Geoff