On Fri, 17 Mar 2000, Peter Svensson wrote:
> On Fri, 17 Mar 2000, James Sutherland wrote:
>
> > Yes, you COULD have COW without overcommitting - but you still lose one of
> > the major benefits of COW, namely huge savings on VM usage. If I fork()
> > 100 Apache processes of 20Mb each, I need perhaps 30Mb of VM total.
> > WITHOUT overcommitted COW, I end up needing 2Gb of swap space - 1.98Gb of
> > which I will never use! This is certainly not an efficient use of swap
> > space, IMO...
>
> Well, that depends entierly on what you use the computer for. Another
> possibility is to deny writing to the text pages of a process. _Then_ you
> can safely share the pages without making promises the kernel man not be
> able to honour.
One problem: either (a) the code never writes to the text pages anyway, so
you gain nothing (the promise could never have been broken anyway) OR (b)
this change breaks the code totally, because it DOES write to the text
pages.
The only circumstance under which this change would have any effect is
where the kernel's "promise" is put to the test. With the current
behaviour, the promise COULD be broken. With your suggestion implemented,
it GUARANTEES that the problem occurs.
So other than turning the remote possibility of a problem into a
guaranteed problem, the change achieves nothing. I don't think it'll make
it into the tree, then :-)
James.
-
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/
This archive was generated by hypermail 2b29 : Thu Mar 23 2000 - 21:00:21 EST