[BusyBox] pre-commit feedback, please

Mike Frysinger vapier at gentoo.org
Fri Jul 29 05:37:12 UTC 2005


On Friday 29 July 2005 01:08 am, Rob Landley wrote:
> > 0000132 03-16-05 Implement fork using longjmp
> >     [ somewhat controversial, i recall... ]
>
> It is.  This is a libc thing, not a busybox thing.  I object to having it
> in busybox on that grounds.
>
> That said: the whole mess of figuring out when we need to vfork and when we
> need a real fork is a bit of a nightmare right now anyway.  If we wanted to
> clean up ALL that (I could see "use vfork instead of real fork" under
> "general configuration" next to "free memory before exiting"), and add this
> as a third option to a three way (fork/vfork/longjmp) selector that selects
> a bb_spork() helper function in its own darn file, then it becomes a lot
> more palatable.
>
> This could tangentially be related to the "compiler.h" cleanup I mentioned
> earlier, and the fork/exec/wait combo helper function.  I.E. this is 1.1
> material, but it's not a no-brainer to check in in its present form.

erik would be able to add more info to this topic in general ...

here's the situation as i understand (i may be wrong, i dont use uClinux :P):
busybox is already full of __uClinux__ checks simply because fork is not 
support on that kernel ... if we had this sort of option (and the 
implementation does look like a nice/clean drop-in feature), we could 
probably punt a lot of that ...
for example, check out these applets:
crond
udhcp
inetd
ash (100% not supported)
syslogd
and these are just the ones that have been checked ... i'm sure theres plenty 
of other applets which currently wont work properly on uClinux

then there's the topic of libgloss ... some people wish to run busybox as a 
standalone program (that is, no kernel at all) and i *think* that this kind 
of feature support would be a boon to them as well
-mike



More information about the busybox mailing list