GPT size-of-disk conflict?

[Andrew Clausen]

Hi all,

I've been thinking about what will happen if two different OS's
disagree on the size of the disk.

Usually, pgpt->AlternativeLBA == dev->length - 1

What if it doesn't?  Should parted continue to use the existing location,
or use dev->length - 1?  What if dev->length - 1 < pgpt->AlternativeLBA?
Saying "invalid, fix?" is a bit misleading... it could mean that Linux is
underreporting the size of the disk, and writing something at
dev->length - 1 might actually be overwriting real data.

I'm leaning towards "move to dev->length - 1", but giving a warning.
We can probably make that an assertion for now, and we'll hear about
any problems if they exist ;)  (The assertion message asks for a bug
report ;)

Andrew