Well it's a function for kernel developers. And kernel developers usually don't work on public shell access boxes. Why would anyone want to compile this into the kernel on such a box?

So, I would say that it's not a security issue at all.

well, what can i say?
If some of you remember hobbiton.org, they ha sysrq activated, and thats why peple thoght its a joke, but its good to halt the system on intruder alert (after the sysrq call the system is unusable), so i think its a good option, i just wanned to see what you guys have to say about that:)
c ya soon:)

I have just installed kernel 2.4.16 on my machine, and when it came to Kernel Hacking and SysRq, the kernel help file explicitly stated not to enable this option unless you are 100% sure you know what you are doing, and why are you enabling it. Not to mention that kernel hacking options will enlarge your kernel by at least 120 kb, if not more.

140Ks are really a threat to your host if you have 16MB ram.
And i think that mostly sysadmins and testers hack their kernels on more sophisticated machines, no doubt they didnt explained what that oprion does in the HELP button on the xconfig, think about it:). If someone is on your host, and up your ass to hack/crack it, it will be very good to dump all the PID's with just a little call, and fuck all Xserver,inet.d,httpd,telnetd,sql/mysql etc., because this way there is ALMOST no harm done, the serv is down, and you redirect the http requests to another host using some sord of dns manager with the msg "Site Down For Maintance" or something, that way your customers wont loose trust in you, which is very immportant these days.
well, thats from me:)

If you want to run a secure open server you should install good firewall, process accounting, have a good logging facility and monitor the system, instead of relying on some not so well documented option in the kernel that is known to kernel hackers and kernel hackers only. Its a lot easier to prevent problems then to fix the problems.