Re: 9.0 beta2 & the new bsdinstaller

From: Warren Block <wblock_at_wonkity.com>
Date: Wed, 21 Sep 2011 07:53:05 -0600 (MDT)
On Wed, 21 Sep 2011, Thomas Mueller wrote:

>>> From "Matthew D. Fuller" <fullermd_at_over-yonder.net>:
>
>> I've been meaning to mention this, but we really should document
>> somewhere that it has a _MAXIMUM_ size.
>
>> I setup a system a few weeks back with GPT, and figured I'd just make
>> the first 'real' partition start at the 1 meg mark.  And make
>> everything before that (1 meg - the however many sectors for the pmbr)
>> the freebsd-boot partition.
>
>> It worked fine, up 'till the point that I tried to boot, and it
>> completely failed to, complaining that the boot code was too big.  I
>> had to track around in pmbr to find
>
>> .   .   cmp $0x9000,%ax..   .   # Don't load past 0x90000,
>> .   .   jae err_big..   .   #  545k should be enough for
>> .   .   mov %ax,%es..   .   #  any boot code. :)
>
>> and redo the partition to 512k (leaving a few hundred k unused before
>> the next partition started) before it would boot.  That's a little
>> nerve-wracking to hit on a critical system...
>
> I don't think there is any particular advantage in aligning GPT partitions on 1 MB boundaries.

Agreed.  But Windows 7 also starts the main partition at 1M.  Taking 
that as a standard could provide compatibility with other (admittedly 
poorly-written) disk partitioning software.  And it might not, but if it 
helps with POLA for people used to using GPT elsewhere, that's not a bad 
reason either.

The bug shown above means the freebsd-boot partition should be limited 
to 512K at present.  Another 512K of space after that doesn't really 
cost anything.  If that extra space is needed later, it can be used 
without repartitioning.
Received on Wed Sep 21 2011 - 11:53:07 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:18 UTC