Re: [PATCH] 2.5-rmap: VM strict overcommit

[Pavel Machek]

Hi!

> diff -urN linux-2.5.26-rmap/Documentation/vm/overcommit-accounting linux/Documentation/vm/overcommit-accounting
> --- linux-2.5.26-rmap/Documentation/vm/overcommit-accounting	Wed Dec 31 16:00:00 1969
> +++ linux/Documentation/vm/overcommit-accounting	Wed Jul 17 10:45:47 2002
> @@ -0,0 +1,77 @@
> +The Linux kernel supports four overcommit handling modes
> +
> +0	-	Heuristic overcommit handling. Obvious overcommits of
> +		address space are refused. Used for a typical system. It
> +		ensures a seriously wild allocation fails while allowing
> +		overcommit to reduce swap usage.  This is the default.
> +
> +1	-	No overcommit handling. Appropriate for some scientific
> +		applications.
> +
> +2	-	(NEW) swapless strict overcommit. The total address space
> +		commit for the system is not permitted to exceed 95% of
> +		free memory. This mode utilizes the new stricter accounting
> +		but does not impose a very strict rule.  It is possible that
> +		the system could kill a process accessing pages in certain
> +		cases.  If mode 3 is too strict when no swap is	present
> +		this is the best you can do.
> +
> +3	-	(NEW) strict overcommit. The total address space commit
> +		for the system is not permitted to exceed swap + half ram.
> +		In almost all situations this means a process will not be
> +		killed while accessing pages but only by malloc failures
> +		that are reported back by the kernel mmap/brk code.

In what scenario can "strict overcommit" kill?

> +4	-	(NEW) paranoid overcommit. The total address space commit
> +		for the system is not permitted to exceed swap. The machine
> +		will never kill a process accessing pages it has mapped
> +		except due to a bug (ie report it!).

...and why is that scenario impossible on "paranoid overcommit"?
								Pavel
-- 
I'm [e-mail]. "In my country we have almost anarchy and I don't care."
Panos Katsaloulis describing me w.r.t. patents at [e-mail]