On Sat, 26 Aug 2000, John Summerfield wrote:
>> It's not as easy as I thought. Not all redhat kernels are equal.
>> Some of them has an extra argument to kill_fasync(). It seems this
>> has been added since 2.2.14 but I don't have all redhat kernels to
>> check. Can somebody comfirm this and tell why this change is necessary?
>
>That sort of problem is the reason you need to update modutils (and some other
>software) from time to rim, and why Mike should be using /proc and not talking
>directly to the kernel.
If you're refering to me (Mike above), and my network interface
detection, I should clear up that it is not talking directly to
the kernel. Although that is what I said, it isn't entirely
accurate of me. I'm not doing direct talking via syscall or
ioctl, etc.. Rather, it uses common libc functions to do the
work. Using stuff that requires kernel headers to be included
such as "linux/whatever.h" would indeed be bad for compatibility
between kernel's, but using standard libc functions is a
different story IMHO.
In general though, staying away from kernel internal stuff is a
good idea indeed. Try to use a userland solution such as libc
(like I'm doing), or /proc if possible. /proc itself has been
known to change between kernel revs though so it is not the most
stable way of talking to the kernel either.
Take care,
TTYL
--
Mike A. Harris Linux advocate
Computer Consultant GNU advocate
Capslock Consulting Open Source advocate
If you're interested in computer security, and want to stay on top of the
latest security exploits, and other information, visit:
http://www.securityfocus.com
_______________________________________________
Redhat-devel-list mailing list
[EMAIL PROTECTED]
https://listman.redhat.com/mailman/listinfo/redhat-devel-list
