Re: scsi and windows boot problem

[Szakacsits Szabolcs]

On Mon, 19 Jul 2004, Andrew Clausen wrote:
> On Fri, Jul 16, 2004 at 02:20:16PM -0700, Anthony Ewell wrote:
> > On this parted bug:
> > 
> >     https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=115980
> > 
> > where you can no longer dual boot to Windows after installing
> > FC2, can anyone tell me if this also affects SCSI drives?
> 
> As I haven't been able to reproduce the bug, I'm not sure what causes it,
> and therefore not sure what circumstances are likely to lead to it.

Reproducing the problem is a no-brainer on any computer. Over the last 8
months several people told you several times what's the problem in parted
and how to fix it. There were over 100(!) emails sent to you in the
subject. Everybody seems to understand except you. Even today. You keep
repeating "nobody tells me anything, I can't reproduce anything".

But you've also made many completely _non-sense_ statements, questions and
produced new but also broken code. Andries Brouwer also pointed these out
to you. Based on the last public info Andries sent, before he left for a
long trip till September, you were working on some also non-sense solution
but this time together with him. It seems you could fool him too but given
that he is also partly lost in this area, it's understandable.

Unfortunately I don't wonder anymore why parted trashes also reiserfs
sometimes, Windows dynamic disks, mounted filesystems and has several
severe design and implementation problems.

As I see and think for over a year but I've never wrote down publicly
before, the fundamental problem is that you were unable to grown up for
the responsibilities and competence what parted requires today. Much
worse, you still can't realize this. Hence you're causing even more damage
then if you just passed over the maintainership to somebody who knows and
understands software engineering better. There were volunteers.

It's also pitty FSF failed to handle and resolve such a serious problem.
At least tens, hundreds of thousands people trash their data for years due
to bugs in the neglected http://www.gnu.org/software/parted/parted.html.
The current situation isn't something FSF should be very proud of.
 
> However, I can't see any reason why SCSI would be more/less susceptible
> than any other device.

You were told several times, too.

Anthony, parted uses a kernel function (HDIO_GETGEO ioctl) what it should
have _never_ used. This kernel function never worked reliable for the
purpose parted wants to use. It was always broken, people always lost data
but it wasn't very visible and there was always something else to be
mistakenly blamed.

In 2.6 kernels the HDIO_GETGEO ioctl was changed only for IDE and returns
values what made the parted problem much more visible.

The HDIO_GETGEO values are the same for SCSI in 2.6 kernels than earlier.
This means parted will quite _probably_ work for you. At least I can't
remember anybody ever had this bootability problem with SCSI (I've seen or
investigated over a hundred trashed partition tables by [lib]parted).

It's also possible that either one or more of these are true,

   - since less people use SCSI thus this problem is less visible

   - booting from SCSI doesn't need the geometries (usually) so it was
     never (usually) a problem on SCSI

   - HDIO_GETGEO always returned the parted expected value, so it always
     worked

        Szaka