Re: Parted 1.5.5-pre4

[Andrew Clausen]

On Sat, Dec 01, 2001 at 10:48:56PM -0600, address@hidden wrote:
> * gpt_clobber needs to read the gpt first, and use those values if valid
> else the defaults.

Ah, good point.  BTW, there's probably no need to clobber the
ptes, since they have no (possibly) confusing signatures.

In general, it's probably best to do as little damage as possible
during clobber.

Anyway, I fixed it here...

> * references to sectors being 512 bytes should be changed to use
> dev->sector_size, and test that it's at least 512, maybe in gpt_probe.  The
> only place 512 need be coded is in defining the reserved space at the end of
> GuidPartitionTableHeader_t.

So, what happens when the sector size != 512?

Must partitions be aligned to the sector size?

Remember, a lot of the hard-coded 512 references are because that's
what the libparted API uses (i.e. ped_device_{read,write}).
So, you need to divide by 512 to get the number of *parted* 512
byte sectors to write.

It might be nicer to switch to byte addressing at some stage.
Anyway, this is a side-issue.

> * I've got an equivalent to the crc32 table that can generate the table at
> runtime.  I submitted that code to Linux kernel.  I'll de-kernel-ify it and
> send.

Cool :)

Andrew