[parted-devel] Possible GPT r/w error

[David Cantrell]

David Cantrell wrote:
> I have not been able to reproduce this on any of my hardware here.  The
> problem described to me seems reasonable.  The GPT area is the first
> 17KB of the disk.  Parted 1.6.x reads in 8KB blocks, so to modify the
> GPT area, it reads 24KB to grab the first 17KB.  Changes are made to the
> GPT area and then written out to the disk.  The problem has to do with
> the extra 7KB of data that has been read, which would be the first 7KB
> of the first partition on the disk if it existed.

I should also point out that the possible partition destruction is
supposed to happen if the first partition is mounted and in use while
you run parted and make changes to the GPT area.  Between the time the
GPT area is initially read in parted to when you type 'q', the user
reports that changes could have been made to the 7KB area at the
beginning of the first partition.  If so, typing 'q' in parted writes
out the old 7KB and the first partition is now hosed.

This seems entirely reasonable to me, but I just can't reproduce it.

-- 
David Cantrell
Red Hat / Westford, MA