[lug] generally compiling problems?

D. Stimits stimits at idcomm.com
Wed Sep 27 19:19:52 MDT 2000


Michael Wegener wrote:
> 
> I don't think the kernel was compiled locally, but it may well have. How do I
> check?

Someone else has answered this, at least in part. It is a good question.
One possibility, is that in /usr/src/linux/, if it was ever configured,
there will be a .config file. You can open that with an editor, and
maybe decide manually if the options listed are the options the machine
is set up with. If it isn't there, you can probably be sure it was not
set up. If it is there, it might not be current. It would be interesting
if someone could create a checksum generator for config files that also
reads the kernel symbols, and can match them.

Every time I compile a kernel, I use the EXTRAVERSION part of the
Makefile to name it uniquely (such as date), along with saving the
config file (named to match) in a root-only archive directory. It saves
so much time not guessing. If you really want to be sure, and you don't
know your configuration, I suspect you'd have to recompile and install a
kernel to know.

> 
> --M
> 
> "D. Stimits" wrote:
> 
> > Someone else already mentioned the sym link possibilities, but
> > apparently that isn't being the problem. One other possibility seems
> > reasonable here. The sources that are linked to in /usr/include/linux/
> > are part of the kernel source tree, and to some extent, parts of that
> > tree are modified by make menuconfig (or other config) of the kernel
> > compile. For example, modules compiled outside of the kernel, such as
> > emu10k1 from soundblaster live, will end up with unresolved symbols if
> > the current source is not also configured to match your existing system.
> > In your kernel source tree, does it still have the exact configuration
> > left from "make menuconfig" (or other config) to match your installed
> > kernel? Or does the kernel compiler configuration no longer match the
> > installed kernel? Some software depends on this, many others don't.
> >
> > Michael Wegener wrote:
> > >
> > > Hello everyone,
> > >
> > > My recent attempt to compile sitecopy on one of my two identical servers
> > > seems to have exposed a mysterious problem. The make output reports a
> > > number of kernel header files missing, yet when I look in the places
> > > listed, there they are. As I look back, I remember never having any luck
> > > compiling on this machine, but always chalked it up to inexperience and
> > > bad luck, and have found work-arounds. This server's twin has no problem
> > > compiling, and they were set up as identical machines by a high-caliber
> > > linux admin.
> > >
> > > Any ideas?
> > >
> > > --Michael Wegener
> > >
> > > --------------------- wegener at zulaware.com
> > > President
> > > --------------------- (voice) 303-402-0364
> > > Zulaware, Inc.
> > > --------------------- (fax) 1-877-540-6114
> > > 5589 Arapahoe Ave
> > > Suite 206A
> > > Boulder, CO 80303
> > >
> > > _______________________________________________
> > > Web Page:  http://lug.boulder.co.us
> > > Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> >
> > _______________________________________________
> > Web Page:  http://lug.boulder.co.us
> > Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug
> 
> --
> Michael Wegener
> --------------------- wegener at zulaware.com
> President
> --------------------- (voice) 303-402-0364
> Zulaware, Inc.
> --------------------- (fax) 1-877-540-6114
> 5589 Arapahoe Ave
> Suite 206A
> Boulder, CO 80303
> 
> _______________________________________________
> Web Page:  http://lug.boulder.co.us
> Mailing List: http://lists.lug.boulder.co.us/mailman/listinfo/lug




More information about the LUG mailing list