> >Basically it just returns a structure filled out with processor
> >information a la 'cat /proc/cpuinfo', except that this is done via a
> >syscall - no more horrid /proc parsing.
>
> Why not use a sysctl?
This is just the beginning - it's my first bit of kernel hacking so be
gentle with me.
Perhaps I should move the code into the sysctl stuff. I've had quite a few
constructive suggestions and some silly ones.
To sum up for the others posting on the subject
1) Needs to take into account different processors on SMP systems (all we
need is cpu_online_map and cpu_data).
2) Move into the sysctl tree.
3) Modify the procfs code so it does sysctls to get the information (thus
reducing kernel bloat but retaining compatibility, but gives us a
interface to take us into the 21st century).
4) Improve & extend the data structures I've been using to return the data
so that with a little work, utilities can get at the information currently
encoded in the procfs.
5) Provide alternative utilities that reads the sysctl stuff.
6) Port to other architectures - always use a common interface so an
utility that knows about the sysctl will always work across platforms.
One of my biggest bug-bears has been the differences in /proc/cpuinfo on
other architectures. This work will remove this.
I've been using 2.2.10 as it's been reliable. I may do the work on a 2.3.x
kernel as soon as the filesystem instability settles down.
Cheers,
Alex
--
I hate people who gets offended too easily!
http://www.tahallah.demon.co.uk
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/