Re: Out of memory kernel deat

Greg Alexander (galexand@sietch.bloomington.in.us)
Mon, 12 May 1997 15:20:58 -0500 (EST)


On Mon, 12 May 1997, Thomas Koenig wrote:

> I'd propose a fairly simple solution: send a SIGURG signal to
> whatever process has a signal handler for it installed.
>
> This process could then do whatever it wants to, such as allocating
> more swap, killing off an errant process or send an E-Mail to
> the admin to go out and buy more RAM.
>
> Policy is for userland.

Taking this to an obscene level...
I'm going to write a usermode daemon that will be summoned every
time the kernel recieves a syscall from a non-root process and gives a
go/no-go return to the kernel. For performance reasons, I'll try to design
the interface such that you could have it run as either a user daemon or a
kernel module (daemon for testing, module when you've had it going a few
months without trouble). Here we can put in the messy stuff that decides
whether or not this process is important, decides if we should halt the
system until the RAM-installing-robot gets here, what have you. This would
be a program that would probably be different just about anywhere -- I'd
write a version designed for systems with hostile users. Someone else may
write one to keep Netscape in check or what have you.
What I'd like to know is if anyone's done this before and if anyone
else has any interest in this.

Greg Alexander
http://www.cia-g.com/~sietch/
----
"I read about monkeys in the encyclopedia as soon as I got home from the
funeral and I wonder if this one throws turds and masturbates all the time
like those monkeys saw it the zoo in San Francisco or if witness monkeys are
more like people."
-- a character in Orson Scott Card and Kathryn H. Kidd's novel,
Lovelock.