telnetd for nommu

[Denis Vlasenko]

On Sunday 25 March 2007 08:46, Mike Frysinger wrote:
> On Saturday 24 March 2007, Denis Vlasenko wrote:
> > On Friday 23 March 2007 15:29, Meihui Fan wrote:
> > > The telnetd.c will not be compiled if enabled on a nommu platform as
> > > blackfin.
> >
> > You NOMMU people are rather hard to find.
> 
> how so ?  the next Blackfin release will be using busybox-1.4.2 ...

Well, I don't see many posts like "bbox fails to do this and that
on NOMMU, you guys suxx", however I suspect it indeed is failing
to do correctly many things on NOMMU.

(For one, entire busybox "all in one" approach shouldn't
work that good for NOMMU - each copy of running busybox
duplicates entire text segment [or I misunderstand something].)

This means that either there is not that much people working
on NOMMU systems or they do not actively reporting problems
for some reason. Which is sad...

> > Does it actually work? I'm asking because telnetd is using fork()
> > in make_new_session() - and that is not supposed to work on NOMMU,
> > right? Thus just fixing compilation by nuking xdaemon() is not enough,
> > you need to get rid of fork().
> 
> correct ... some people leave the fork() symbol in libc.so so users dont 
> notice ... they just compile and maybe run one or two things and assume 
> everything is kosher
> 
> > This leads me to another questions:
> > Will it help with NOMMU bug hunting if we will deliberately make
> > fork() to fail during compilation?
> 
> yes it would

Hmmm
--
vda