Re: About usage of floppy with linux guests on gnome-boxes

[Lucas Meneghel Rodrigues <lmr redhat com>]

On 12/02/2011 11:09 AM, Zeeshan Ali (Khattak) wrote:
> On Fri, Dec 2, 2011 at 2:57 PM, Lucas Meneghel Rodrigues<lmr redhat com>  wrote:
> > Fortunately, I know what this might be, since the first windows guest I've
> > tried my initial unattended install code is windows XP 2 years ago and I've
> > hit this issue.
>
>    Oh, fortunate we are indeed. :)
>
> > What most likely is going on here is a slight mismatch on the floppy image
> > size used to hold the unattended install file. If I recall correctly, I was
> > trying to provide 1.44 M to dd at the time to create the floppy.
>
>    Oh, I would never have guessed the floppy size would be the issue
> but rather the unattended file itself.
>
> > [lmr@freedom data]$ file floppy.img
> > floppy.img: x86 boot sector, mkdosfs boot message display, code offset 0x3c,
> > OEM-ID " mkdosfs", sectors/cluster 4, root entries 512, sectors 4096
> > (volumes<=32 MB) , Media descriptor 0xf8, sectors/FAT 3, heads 64, serial
> > number 0x46d6dab4, label: "           ", FAT (12 bit)
> > [lmr@freedom data]$ du -sh floppy.img
> > 20K     floppy.img
> >
> > This certainly doesn't look like a well formed floppy,
>
>    Yikes, I'll try creating the floppy image using the method you proposed then.
>
> > XP standards. I see that you copy the unattended file using mcopy
> >
> >         debug ("Copying unattended file '%s' into floppy drive/image '%s'",
> > unattended_dest_name, floppy_path);
> >         // FIXME: Perhaps we should use libarchive for this?
> >         string[] argv = { "mcopy", "-i", floppy_path,
> >                                    unattended_tmp_path,
> >                                    "::" + unattended_dest_name };
> >         yield exec (argv, cancellable);
> >
> > So, although this seems fine at a first look, it's not good enough for WinXP
> > standards :)
>
>    I don't think mcopy does anything different than what you are doing.
> I'll try first to fix the size of template/empty image first.
> Hopefully that helps.

Yes, I was thinking about it and yeah, shouldn't make a difference. I 
could send a patch, but then it'd be a huge binary blob, so it wouldn't 
help. You fix the floppy and I believe the issue should be resolved.

> > Perhaps you guys are open to a more traditional approach like
> > the one we use on our unattended code?
>
>    The issue with mounting is permissions. Boxes need to run
> unprivileged and you need to be root to mount.

Yes, nevermind. As we are a test framework we do need root privileges to 
do a lot more stuff, so most of our code runs as root anyway. Please 
spare me from the rotten eggs :)