Re: Bug#475718: Bug #475718 grub gets confused by hybrid apple/pc partmap

[Chris Knadle]

On Tuesday 05 August 2008, Marco Gerards wrote:
> Robert Millan <address@hidden> writes:
> > On Fri, Jul 25, 2008 at 07:36:25PM -0400, Chris Knadle wrote:
> >>    If you ask me, I think this unfortunately looks like a complex
> >> detection problem -- which I think is eventually going to start with a
> >> *successful* detection of the Apple partition (because Apple + PC
> >> partitions can co-exist), followed by somehow *rejecting* that based on
> >> not finding any Apple HFS+ partitions (avoiding scanning partition 0),
> >> re-detecting partmaps (avoiding the failed Apple detection), and then
> >> going from there.  :-/  I've been thinking about that problem.  No
> >> matter how I look at this issue, it's messy.
> >
> > It can't be so complicated;  your layout is properly detected on Linux,
> > isn't it?
> >
> > Maybe we just need to give preference to MSDOS partitions.
> >
> > Btw, adding grub-devel to CC.  Let's try to have discussions in upstream
> > list..
>
> If I only know what you were talking about here... ;-)
>
> --
> Marco

   Who, me or Robert?  ;-)

   If you mean me -- I was confused.  Pavel made the suggestion of fixing the 
Grub2 issues with hybrid Apple/PC partition maps by searching for "the Apple 
magic" of 0x4552.  I've never dealt with Apple partition maps before, and so 
I didn't realize that Apple machines have a special "Apple" partition on the 
first partition of the drive -- so I mistakenly took Pavel's words to mean 
looking for an Apple *filesystem* (meaning searching on partitions > 0), so 
in my mind 0x4552 was for an HFS or HFS+ magic number.  I found some Apple 
documentatino that mentioned it, but if it mentioned that the 0x4552 magic 
was on partition 0, I missed it, so the logical train of thought led me to a 
bunch of erroneous conclusions.

   So yeah, if you don't understand what I wrote, it's because it's nonsense.

   Robert committed a fix for this bug and uploaded a new package, and the bug 
is closed.

   -- Chris

-- 

Chris Knadle
address@hidden