Re: Shrinking HFS+ causes fatal out of memory

[K.G.]

On Thu, 20 Oct 2005 20:50:54 +0200 Sven Luther <address@hidden> wrote:
> On Thu, Oct 20, 2005 at 09:00:56PM +0200, K.G. wrote:
> > On Thu, 20 Oct 2005 19:55:41 +0200 Sven Luther <address@hidden> wrote:
> > >
> > > [...]
> > >
> > > Well, then :
> > > 
> > >   rebuild the package with this option.
> > > 
> > >   get the d-i stuff from svn
> > > 
> > >   add the rebuilt packages to localudebs
> > > 
> > >   add libparted .udeb to pkg-lists/(netboot|cdrom|hd-media)/local
> > > 
> > >   rebuild the initrd.
> > > 
> > >   use it with th rest of d-i, preferably netboot or hd-media.
> > > 
> > > Friendly,
> > > 
> > > Sven Luther
> > > 
> > 
> > Waouh, sounds complicated. I doubt I can reasonably ask any user that
> > report a bug with HFS to do that kind of stuff, they will find that
> > too complicated :p
> 
> No, but we could build such a thing and provide it to users, if you want i can
> do one tomorrow. What is exactly the option you need and what does it do ? 

Cool! The option is --enable-hfs-extract-fs for the configure script, and
I already described what it does in a previous mail:

On Thu, 20 Oct 2005 16:21:05 +0200 "K.G." <address@hidden> wrote:
> and then
> you will be able to extract the low level structures of
> HFS+ (and HFS wrapper if any) with the "check" command of
> parted. It will create "vh.hfsplus", "catalog.hfsplus",
> "extents.hfsplus", "bitmap.hfsplus", "attributes.hfsplus",
> "startup.hfsplus" (and maybe "mdb.hfs", "catalog.hfs",
> "extents.hfs", "bitmap.hfs" if there is an HFS Wrapper)
> in the current directory, and then you can send me those
> files for analysis. They will contain some informations
> like file names, date of creation, ..., so if you don't
> want me to be able to read these informations don't do
> it.

Of course this must not be used in a standard Parted build
for a distribution or a distribution installer.

> I also have my own built initrd kernels which include parted, but they are
> geared toward pegasos hardware. I think we may be able to produce something
> using yaird though, more to this later on, need to get 2.6.14-rc kernels out
> first.
ok

Cheers,
Guillaume Knispel