[parted-devel] [PATCH] More correct handling of the HeaderSize field in GPT labels

[Matthew S. Harris]

On 7/8/07, Otavio Salvador <otavio at debian.org> wrote:
> "Matthew S. Harris" <mharris312 at gmail.com> writes:
>
> > In general, if there end up being multiple, backwards-compatible
> > versions of the EFI spec, then Parted should be able to operate on the
> > set of headers it knows about and ignore the rest.
>
> That's the major issue. I personally dislike to believe on that.

I see.  Is your concern that changes to the EFI spec won't be
backwards-compatible, or that Parted will not be careful enough in its
handling of the part of the header it doesn't understand?

> > In this case, the EFI spec shows a table of the fields in the header,
> > and it includes as the last "field" all the remaining space in the
> > block.  OpenSolaris interprets this to mean that HeaderSize is the
> > size of the block; Parted obviously does not.  This is significant
> > because the CRC of the header is supposed to be of HeaderSize bytes.
> > That's why this change is needed.
>
> Sorry but I didn't get what you mean by the way of OpenSolaris to
> work.
>
> You mean that OpenSolaris cannot check the CRC since it interprets the
> size wrongly, is it?

No, I mean that Parted cannot check the CRC of an OpenSolaris
partition because Parted uses a constant rather than the value of
HeaderSize in the disk label to determine how many bytes to compute
the CRC over.


Matthew