[parted-devel] GPT-MBR hybrid
Dan Knapp
dankna at gmail.com
Sun Dec 23 23:00:06 UTC 2007
Well, really, I'm pretty sure that what you describe is what parted does
now: TheMBR is effectively a "dummy". The EFI specs, which are where GPT is
defined,
state that there will be an MBR and that it will have (if I remember...),
exactly one
entry, the "protective partition" which covers the entire disk so that
non-GPT-aware
tools won't treat the disk as blank. This so far is not really a hybrid
scheme, it's
how GPT is supposed to work. Any GPT-aware OS or program is supposed to
use the GPT entries and ignore the MBR ones.
True "hybrid" GPT/MBR partitioning schemes are basically an informal hack,
and
the gptsync program is the operative definition of what they are and how
they work.
They were created to allow non-GPT-aware versions of Windows to boot from
devices with non-MBR-aware firmware; namely, Intel Macs. The MBR therefore
exists solely for Windows's benefit, and there's no reason it strictly has
to reflect
every partition that the GPT does, but that's the idea right now. Since GPT
can
have arbitrarily many entries whereas MBR is strictly limited to four, it's
not really
clear what Parted's role in this should be...
While people are paying attention, I'd like to just point out another GPT
issue, which
is that as things stand there is no way I'm aware of to use parted to
correct a broken
GPT table which has an invalid uuid for the type of one partition. I never
saw any
response to my email on this topic a couple weeks ago. :) I did already
write a tool
to do that, but it would be great if the functionality could be rolled into
parted...
On Dec 23, 2007 5:40 PM, H. Peter Anvin <hpa at zytor.com> wrote:
> Matt Davis wrote:
> > On Sun, Dec 23, 2007 at 02:27:57PM -0800, H. Peter Anvin wrote:
> >> Matt Davis wrote:
> >>> On Sat, Dec 22, 2007 at 10:16:21PM -0800, H. Peter Anvin wrote:
> >>>> Is there any sane way to support GPT-MBR hybrid partitioning in
> parted?
> >>>> This is required for booting on BIOS-based machines (the standard
> MBR
> >>>> handover protocol doesn't really work for GPT), but GPT is needed to
> >>>> support > 2 TB.
> >>>>
> >>>> At least /boot should be in the MBR; some "other OSes" may need their
> >>>> boot partition MBR-side as well.
> >>> hpa,
> >>> While I do not have an answer for you, I have been working, in my free
> >>> time, to
> >>> try to allow Parted to convert from one partition table type to
> another,
> >>> such as
> >>> gpt to mbr, or vice versa.
> >>>
> >> Well, the hybrid is really another partitioning scheme, in many ways.
> >> It's a little different than either.
> >
> > Yep, I am not familiar with that. However, I did notice the opportunity
> to keep
> > people up-to-date with what is going on regarding similar issue.
> >
>
> The idea with hybrid partition is that you have both a GPT and an MBR.
> The MBR has the boot partition(s), or possibly all partitions below 2
> TB, plus a "protective partition" (type 0xEE); the GPT has the other ones.
>
> -hpa
>
> _______________________________________________
> parted-devel mailing list
> parted-devel at lists.alioth.debian.org
> http://lists.alioth.debian.org/mailman/listinfo/parted-devel
>
--
Dan Knapp
"An infallible method of conciliating a tiger is to allow oneself to be
devoured." (Konrad Adenauer)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.alioth.debian.org/pipermail/parted-devel/attachments/20071223/c612ae03/attachment.htm
More information about the parted-devel
mailing list