Re: [patch 00/13] devtmpfs patches

[Andrew Morton <akpm@linux-foundation.org>]

On Mon, 11 May 2009 01:47:16 +0200 Kay Sievers <kay.sievers@vrfy.org> wrote:

> On Sun, May 10, 2009 at 23:19, Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
> >> /me boggles, however does one end up with that many partitions on so few
> >> disks?
> >
> > It happens. Tejun did the extending because libata users were really
> > hitting these limits.
> >
> >> /boot
> >> /
> >> /home
> >> /usr
> >> swap
> >> /usr/local
> >> /opt
> >>
> >> and I'm seriously out of sensible ideas.
> >
> > I've seen two usual variants
> >
> > __ __ __ __Fedora & Unbuntu & Debian & ....
> >
> > __ __ __ __Fedora 10 & Rawhide & Fedora 9 & ...
> >
> > for people who like to play with distros or need to test a product on them
> >
> > and once you do that 16 is quite easy
> 
> That's right. Also people with virtualization, running many guest from
> individual partitions, like to do that - and 16 is nothing for them.
> :)
> 
> Some of us run the dynamic sd minors already, and it's not unlikely we
> will enable it pretty soon for the distro.

There are plenty of setups which have lots of partitions, but very very
few of them are booted with any frequency.

The patchset is described as saving "a few seconds" when booting a
typical distro installation?  Out of what, 100 seconds?  That's pretty
thin gruel, IMO, and doesn't obviously justify mucking up the kernel. 

A 1.5 second saving on some gadget which boots in 6 seconds would be a
lot more interesting.