Re: Kernel Compile
Pann McCuaig Writes..
> The packages I've listed above are required to compile (using make-kpkg
> and make menuconfig) a 2.0.34 kernel on a libc6 system.
Right I have 2.0.34 compiling now. Using the source code from the Debian CD
...well at least on ONE of my boxes. (It's a start)
However, using 2.0.35 or 2.0.36 source code still fails. Why is that !
What is different about the normal source code I pull off 'sunsite' and that
which is found on the Debian CD! Now I am really confused ;-)
> All the libc5-related *-dev packages should have been removed from your
> system during the upgrade, and will have to be re-installed from the
> oldlibs directory for your libc5 development environment to work as
> before.
Yup they were, and I have re-installed the ones I had before, selecting them
from the oldlibs section.
Now then, I noticed earlier that other compiles (not kernel) which are
failing, are complaining about /usr/include errors !
At the moment I am sticking to 2.0.34 kernel source from the CD, as I know
it works (although still running 2.0.36 image from 1.3.1).
Why would some things complain about /usr/include !? IS there something
missing here, or needs adding etc.
I have tried both with ASM SCSI and LINUX dirs 'as is'. And I have also used
sym links to /usr/src/linux/include as the README in the kernel source
always suggests.
However, no matter what I do, I still get some code failing to compile.
IS the code looking for something it can't find here - like kernel headers !
> Sorry to hear you're going through this. The libc5-->libc6 upgrade is
> the most perilous we've gone through since a.out to ELF.
I just want it to work again like before. My 1.3.1 box was the Dogs Parts,
and I loved it to bits. But 2.0 has ruined it all at the moment. Just wish I
knew how to fix it :-( Better still Debian *SHOULD* cost a fortune instead
of being Free, then I wouldn't have upgraded in the first place ;-)
> Now if you'd had redhat you'd have had to wipe your hard disk and start
> over to upgrade.
I have now come to the conclusion that is *THE* only way to upgrade. Never
will be an easier way if you ask me. When I first installed Debian, I
re-created a Slackware box of several years, in just one week-end :-)
That impressed me alot - Debian was so damm easy to sort out. But after this
'upgrade' I have spent days trying to mend everything that broke. Would have
been easier to rm -rf and then start again... In fact I might still do at
this rate ;-)
> The upside is, you _can_ recover from your current
Yes, I fear a restore of 1.3.1 is lurking on the horizon now. If I can't
start to sort 2.0 out soon I shall remove it. I have decided that all this
LIB6C stuff is not (yet) for me. It breaks everything it see's of mine !
> and will have a stable libc6
Or even libc5 :-))
> without throwing anything away.
I dunno. I already have alot of broken not worky junk lying around now. In
fact I have just deleted loads, cus it just won't work, and won't bloody
compile either :(
> Just keep on truckin'.
Yeah but having the best truck in the world is a bit sad when you can't
get any wheels to fit it. You just have to sit and look at it, cus it
aint going anywhere without them wheels ?!
Cheerio
--
Nidge Jones
Reply to: