Re: PRep-boot partition flag support.

[Sven Luther]

On Mon, Nov 29, 2004 at 09:28:07AM +1100, Andrew Clausen wrote:
> On Sat, Nov 27, 2004 at 10:27:55AM +0100, Sven Luther wrote:
> > Other patches in the debian package which may be of interest are : 
> > 
> > amiga.dpatch:
> >   Adds amiga filesystem support
> 
> It adds both filesystem probe and full partition table support, right?
> The patch looks good... I'm just wondering if you should print
> out a warning about it being unstable?  (How much in-the-wild use has it
> seen?)

It has been part of of the debian parted package since about a year :

parted (1.6.6-1.2) unstable; urgency=low

  * Added amiga filesystem probing support. (Closes: #219947)
  * Fixed wrong linux/fs.h header copying. (Closes: #218980)
  * Now build-depends on the fixed version of linux-kernel-headers, so we
    don't get bit by the __invalid_size_argument_for_IOC signedness problem.
    (See bug #219940)

 -- Sven Luther <address@hidden>  Tue, 16 Dec 2003 14:22:33 +0100

Is the default method for debian-installer partitioning on pegasos, and
possibly also m68k/amiga, but i have no info on that, and mostly people don't
repartition much on 68k/amiga anyway. On the pegasos, i believe it has been
tested on over a 1000 machines at this time, including over 300 automated
installs using a little partition table program from a file i have written.
This is without counting any of the AmigaOne machines which may or may not use
it, and linux installs on pegasos on non-debian distributions (CruxPPC,
Gentoo, Yellow Dog Linux, some random others possibly), as well as testing on
x86 system too. Also, i used to do daily d-i test runs on my machine with it,
during the main development phase of debian-installer.

As such, i believe it has had more testing than other partition table code in
libparted, and i don't really think the below warning is warranted.
Furthermore, i am now the main maintainer of amiga partition table code on all
OSes, and the parted code serves as main documentation :)

> Perhaps put this in amiga_read():
> 
>       ped_exception_throw (
>               PED_EXCEPTION_WARNING,
>               PED_EXCEPTION_OK_CANCEL,
>               _("The software to support for Amiga partition tables "
>                 "is new and has not been widely tested."))
> 
> > no-clobber.dpatch:
> >   Removed a fs->clobber assertion that did break since the
> >   amiga filesystem does not provide a clobber op.
> 
> Remind me why the amiga file system doesn't have a clobber op?

Because i am too lazy to implement them ? Basically the only amiga filesystem
opp is probe, in order to inform the user of the existing filesystems, but i
have not done any further methods.

> > s390.dpatch:
> >   Another s390 bigger sector size patch, replaces Bastian's
> >   more limited sector-size patch.
> 
> Are we looking at the same thing?  The s390.dpatch in the debian/testing
> package is support for s390 partition tables.  We would need a copyright
> disclaimer or assignment for this one...

It is a full disk_s390.c code in fact, and has been in debian since :

parted (1.6.9-2) unstable; urgency=low

  * Add devfs partition detection patch for 2.6 kernels, thanks to
    Joey Hess <address@hidden>. (Closes: #243635)
  * Added a more complete s390 sector size patch by Lucius Leland
    <address@hidden>. (Closes: #243554)

 -- Sven Luther <address@hidden>  Wed, 14 Apr 2004 10:27:58 +0200

is also the main d-i partitioning method on s390, and has probably had about
the same testing as amiga code. That said, there are obviously less s390
machines than amiga/pegasos ones on which it was tested. 

The author is  Lucius Leland <address@hidden>, and i will see if i
can contact him about the copyright assignement stuff, didn't think about
this.

> I just got a USB key image sent to me that has non-512-byte sector
> size.  I'll have to have a look at this...

Mmm, possibly, altough the patch seems to be mainly disk_s390.c support, you
will have to look at the linux.c part of the patch.

> > sun_disklabel.dpatch:
> >   Some older sun disk labels were previously misparsed.
> 
> Looks fine.

Ok, will commit it tomorrow. We also have a bunch of bug reports of note that
i didn't really look over. Involving arm, hppa, sparc and some others
partition table support. Stuff like overlapping partitions, and one cylinder
sized partitions, or partitions overlapping the metadata, if i remember well.

Friendly,

Sven Luther