> lilo:
> >
> > > > What's next, /proc/cflags? :->
> >
> > I only have one question.
> >
> > WHY??????!!!??!?!?!?!!!!
>
> There are many new gcc's floating around, and many new options to
> gcc (pentium gcc provides one example). In the event that
> people testing their kernel decide to change the default
> CFLAGS through something like
>
> make CFLAGS='-O3 -mfoobar ... '
>
> There is no record of what they used. The diff I wrote
> adds functionality to put this in include/linux/compile.h, as
> well as a /proc entry to display the CFLAGS string.
>
> >From now on you can pick up any of my patches (a long,
> fruitless legacy ;) from embittered.caltech.edu in
> /pub/linux/patches.
I wish something like this existed for configured drivers. Although most
drivers are getting modular, some still don't and unless they did probed
successfully, you don't know if it is there. I have customer who
generally build their kernel with ne2000,3c509 and smc-ultra because this
is the adaptor they use. This allow them to switch ethernet adaptor
without doing anything (pure magic). The problem is that I am never sure
that a give kernel do have this or this feature until I try it. I would
vote for a /proc/config.
It is so easy to move compiled kernel around (unlike some operating
system where the kernel is hidden) from machine to machine, that
/proc/config is the only reliable source of information.
Sure enough, /proc/... and /proc/net/... generally provide a lot of
information about the running kernel.
I have implement this a long time ago but felt that people would consider
a kernel bloat. Any interest, I must have the patch somewhere.
--------------------------------------------------------
Jacques Gelinas (jacques@solucorp.qc.ca)
Use Linux without reformating: Use UMSDOS.