Re: [PATCH] remove compat code for old devfs naming scheme

[Richard Gooch]

Christoph Hellwig writes:
> On Tue, May 14, 2002 at 08:59:41AM -0600, Richard Gooch wrote:
> > The reason to support it is because lots of people are depending on
> > it. A lot of systems would break for no real gain. This code is in the
> > init section, so the memory will be freed before init(8) starts.
> > 
> > The code should stay. In 2.5, I can move it into the mini devfsd that
> > will go into the initial rootfs.
> 
> Exactly how many system exist which
> 
>   a) still use the two or three year old devfs naming scheme for
>      their root= command line
>   b) use current 2.5 kernels

Those names are not really "old", but convenient shorthand. And it's
more systems than you think.

> and
> 
>   c) have users not capable of translating that syntax to either
>      the new devfs or traditional Linux syntax?
> 
> I don't think there are more then about ten, and no I _absolutely_
> do not consider that a reason to bloat the kernel, even if it is just
> source and image bloat.

A lot more people are using devfs than you realise. Removing the code
would screw them over. Leave it alone.

				Regards,

					Richard....
Permanent: [e-mail]
Current:   [e-mail]