> I think I've got enough spare time at present to contribute something
> back linux but I'm finding it hard to think of anything worth doing
> in the kernel. Has anyone got any ideas / suggestions I might consider?
Here is a list of interesting things you might do:
http://www.cs.uml.edu/~acahalan/linux/wishlist.html
Things that would help make /bin/ps work better:
There is a combined kernel + libc problem that just hit me. The kernel
does not seem to support sysconf(_SC_NPROCESSORS_CONF), so libc does
a disgusting /proc-based hack. (problem #1) This doesn't work on SPARC
and 680x0 systems because /proc/cpuinfo is inconsistent. (problem #2)
The standards require that sysconf() return -1 for an error, but libc
sometimes returns 0 instead. (problem #3)
It would be very helpful to have /proc/*/tty symlinks. The current code
just uses device numbers. You might need to have the kernel open an
initial tty device in an odd way. (you could fake it)
The kernel should have a large unique ID in System.map, so that libraries
can verify that a System.map file matches the current kernel. An alternate
(better) solution would just put System.map into the kernel itself.
The 200 kB or 50 kB (compressed) is a small price to pay for accurate data.
-
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/