userland boot time

[Carsten Haitzler (The Rasterman)]

On Mon, 21 Jan 2008 17:07:38 +0100 Tilman Baumann <tilman at baumann.name> babbled:

> Carsten Haitzler (The Rasterman) wrote:
> 
> > so we need to address this in 2 parts. sysvinit etc. is only 1. the actual
> > user xsession is another. so /etc/X11/Xserver needs some love and attention
> > too. and thus /etc/X11/Xsession and /etc/X11/Xsession.d/*
> 
> Could some of this not be paralleled via init too?
> Most of the stuff here just needs X up and running, that's the reason 
> why it is there.
> Stick a 'depends on X' on every task there and let it be run by init 
> after X has booted. Most of the things in startup scripts are mandatory, 
> but very less of them need to run sequentially...
> I guess the only problem would be inheriting a X environment for these 
> scripts. Not a problem, more the fact that we would have to hack 
> something here... :)
> 
> One benefit i would see is, that often just the WM crashes but X is 
> running fine. With any init that can restart services, we could just 
> fire up the WM again and be good.
> Sadly this happened some times with my system, i think one of the panels 
> put the wm down with itself.
> 
> Clearly, i'm not suggesting to dissect all of the Xsession scripts to 
> the last line, but some of the things that come up in this stage are 
> more or less just deamons too, which would fit nicely into a modern-init 
> based startup. And could be supervised by a service monitor for 
> restarting them after crash.

if you check the x startup - most things are started in parallel (the session
scripts start them with a &", but all the hammering of disk IO, library
linking, symbol resolving, everyone connecting to x (multiple times) and so on
all at once is not going to be nice on speedy startup.


-- 
Carsten Haitzler (The Rasterman) <raster at openmoko.org>