Possibly unfreezable system?

Nicholas Knight (tegeran@home.com)
Sat, 4 Aug 2001 14:52:14 -0700


(I'm going to get my ass handed to me for this one, I just know it, but I
have to try.)

I'm not a coder, I can't impliment this, I don't even know for sure if it
IS possible to impliment without a complete rewrite of the kernel.
This is also not something that would likely be added to 2.4, it'd
probably be a 2.5 thing.

Note that I intend for the behavior below to be CONFIGURABLE, NOT the
default behavior of the kernel, and that the exact behavior be (somewhat)
configurable without diving into the code.

I've lately seen many complaints regarding the inability to even access a
system that something (such as kswapd) is going crazy on.
The solution, to me, seems simple, have the kernel reserve some extra RAM
at boot (a few megs), and dictate that it get at least X amount of
processor time, consistantly, to allow for the following:
An alt-sysrq key that switches to a certain virtual console, kills
whatever might already be running there, and allow a person to log in in
order to kill whatever is causing the system to freeze, run out of
memory, etc. The program(s) running here would run in that extra RAM the
kernel reserved at boot.
This obviously degrades performance somewhat, but if properly
implimented, the pros could outweigh the cons on even the most
resource-sensitive systems.

Again, I mean for this to be CONFIGURABLE, and NOT the default behavior.
And please, don't flame me, none of this may be doable, if so, I
apologize for wasting everyone's time.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/