[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: powermac boot-floppy patches

On Thu, May 31, 2001 at 02:45:28PM -0800, Ethan Benson wrote:
> On Thu, May 31, 2001 at 12:21:38PM -0500, Stephen R Marenka wrote:
> > On Wed, May 30, 2001 at 11:07:30PM -0800, Ethan Benson wrote:
> > > On Wed, May 30, 2001 at 08:39:27PM -0500, Stephen R Marenka wrote:

> i have a test disk that i can and do routinly erase for testing like
> this, i can do most of the newworld testing if you prefer.  it would
> certianly reduce risk of trashing your working system.. 

That's the problem with having a laptop for testing. Please, do test. 

> im tinkering with it now, i cleaned up some of the useless cruft in
> the yaboot function (quik cruft that wasn't used).  i gave it a run
> yesterday and the partition checking stuff seems broken.  when i have
> no Apple_Bootstrap it tells me / and Apple_Bootstrap must be on the
> same partition.  not that i need an 800K Apple_Bootstrap.  
> as a sidenote i would prefer for dbootstrap to notify the user they
> screwed up immediatly after they exit the partitioning program that
> they no/incorrect Apple_Bootstrap partition, rather then waiting until
> after they go to all the trouble of installing the base system to tell
> them they must start over...  (believe me ALOT of people screw up the
> bootstrap partition creation, they think they did it right but
> didn't).  

I stripped out the yaboot.conf file I was generating to add the
yabootconfig stuff. I didn't go back and clean out the rest of the junk.
Ooops. Primarily, I wanted to make sure my patches got in to prevent 
someone from duplicating my effort. Thanks for the cleanup.

I'll look through the logic again. I probably hosed up something

> i still don't fully get how the find_partition_() functions work, but
> p->disk->name seems to resolve to the whole disk (/dev/sda for
> example) not the Apple_Bootstrap (even when one is there).
> this is obviously a Bad Thing. 

Right, it should probably be p->name (if I remember correctly). I must
not have been having a good coding night.

> let me work with the yaboot code a bit and ill see what i come up
> with..  a suggestion/clarification about the
> find_partition/fdisk_partition stuff would be welcome if i don't
> figure it out soon enough.
> > > one thing i think we should add is a dialog explaining that making an
> > > oldworld bootable usually requires some effort on the part of the user
> > > no matter what, typically running apple's System Disk control panel to
> > > get the nvramrc patches, and of course the ritual goat sacrifice. 
> > 
> > Hmmm. I never had to do this, perhaps I'd better thoroughly read your
> > documentation and then go back over what I've written. The only real 
> > trick was the goat sacrifice and I'm use to that from supporting 
> > Microsoft stuff. 
> not all machines need anything more then boot-device.  beige g3s for
> example are not one of them...

Do you mean some machines need more or that not all machines work using
this approach? I've seen references to the beige g3s being problematic.

> > I can do the nvsetenv from C, the only odd bit was that I had to use,
> > for example,  ofpath '/dev/sda' and append a 0 instead of '/dev/sda6' 
> > for boot-device. Fortunately, this was documented several places because
> > it's a known "feature."
> its a limitation of ofpath.. it won't return a path with partition
> 0. it could be fixed i suppose but it would change the semantics in a
> way i don't really like.  
> can't you read ofpath's stdout right into a variable or something?
> dumping it into a file, then reading the file strikes me as rather
> kludgy.. no offence ;-)

Oh hey, it's definitely a big ugly kludge! I was writing a shell file 
to grab ofpath and set boot-device, boot-file, input-device, and
output-device. I'm not sure that isn't a bit more elegant. When it
turned out I didn't need anything but boot-device, I scrapped the rest
of it. It might be useful to try to set input-device and output-device 
based on model anyway. That would give us a better failure mode in case 
things go awry (at least on the models we can help on). What do you think?

> > > it should probably be a dialog letting the user backout, since if OF
> > > fails to boot they will get the notorious `black screen syndrome' 
> > 
> > I should definitely clarify that bit about the black screen and probably 
> > include a note about cmd-opt-P-R.
> yes definitly.  

Then, I'll make it happen. Hopefully tomorrow.



Stephen R. Marenka     If life's not fun, you're not doing it right!

Reply to: