[oe] Are we starting udev too early in the boot process?

[Graeme Gregory]

On Sat, Jan 09, 2010 at 09:47:41PM +0000, Graeme Gregory wrote:
> On Sat, Jan 09, 2010 at 10:20:20PM +0100, Rolf Leggewie wrote:
> > Hi,
> > 
> > could it be that we are starting udev too early in the boot process?
> > 
> > http://bugs.openembedded.org/show_bug.cgi?id=5090
> > http://bugs.openembedded.org/show_bug.cgi?id=5246
> > 
> > Above two bug reports whose effects I can confirm would suggest that to
> > me.  But I'm not sure about the side-effects of starting udev later than
> > where it currently is at.  Anybody care to comment?
> > 
> My opinion is the dev cache code is wrong, it should cache into /dev which
> is the only place guaranteed to be writeable. There then should be a second
> init script later in the boot process which moves this to /etc/
> 
In fact to follow on from my mail, as 2.6.32 devfs has reapperred in the
kernel we could do away with the caching on modern kernels totally.

Graeme