Re: [PATCH] Patch to Memory Subsystem ... (Needed?)

H.J. Lu (woody@chunnel.oca.udayton.edu)
Mon, 9 Nov 1998 11:17:36 -0500 (EST)


On Sat, 7 Nov 1998, Brian Schau wrote:

> Horst von Brand wrote:
>
> But why do people (in general) run raid-1 on their harddisks? To be
> protected against disk failure.
>
> Why do I want to reserve some pages for root? So that if all memory
> seems to be exhausted, root still have the opportunity to login,
> evaluate and eventually find a solution to the problem. Instead of
> doing a hard reboot as I was forced to do (which meant that I put 14
> people on hold for 15-20 minutes while the server was fighting it's way
> back to life. Fortunately we didn't loose any data ...)

That's what the MagicSysRq key is for. ;) Hehehe

But I do agree with the idea of having "reserved" root memory. Again as
stated before... having kilobytes instead of pages declared makes it
universal... be it 32 or 64 bits. ;)

As for determining if a "root" process is something silly like CRON...
can't you throw in a check to see if the process has a controlling tty?
If it has a tty, it must belong to an interactive user, no? Or am I way
off base here? And if we determine by tty, shouldnt we reserve that tty
just for the "emergency" root action? What happens if someone leaves root
logged in at console... and he's already eaten our last bit of RAM? ;)

-Woodstock
woody@chunnel.oca.udayton.edu

-
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/