[BusyBox] Re: busybox digest, Vol 1 #36 - 6 msgs

[Erik Andersen]

On Tue Jul 11, 2000 at 07:35:55AM +1200, Simon Byrnand wrote:
> >
> >Anybody else have something they think should be fixed?
> 
> One little wish - has anybody attempted to, or is planning to attempt to
> figure out what needs doing for busybox to compile on a 2.0 (libc5) system
> ? Libc5 support was added in 0.45 (I believe) but there still appears to be
> reliance on the 2.2 header files. Are there fundamental things which would
> stop it working on 2.0 (calling functions that dont exist in 2.0 etc) or is
> it just likely to be a difference in available header files that could be
> sorted out with a few #ifdefs ? 2.0 still makes sense in embedded or
> otherwise minimal systems since a kernel can be significantly smaller than
> 2.2.
> 
> I plan to look at this myself soon, but right at the moment I'm knee deep
> in a dhcp client trying to rewrite it, and I'm not quite sure how to go
> about solving the header dependencies anyway - other than just sifting my
> way through the 2.2 and 2.0 headers, trying to compile, see what header its
> not finding, see if 2.0 has an equivilent, #ifdef'ing it in, seeing if it
> works, and so on. But even if I get something that compiles are there
> likely to be subtle bugs introduced that I may not notice ?
> 
> Regards,
> Simon

Could you try the latest from CVS on your 2.0 kernel based system?  Over the
weekend I started trying to remove all the kernel header file dependancies.
Right now only fbset (which won't work with older kernel), fsck.minix,
mkfs.minix, nfsmount, and utility.c (loop.h used by BB_FEATURE_MOUNT_LOOP) use
kernel headers.  Could you give it a try and tell me if it works for you? 

 -Erik

--
Erik B. Andersen   email:  andersen at lineo.com
--This message was written using 73% post-consumer electrons--