On Fri, 12 Sep 2003 00:13:07 +0200 (CEST)
Dag Wieers <dag(a)wieers.com> wrote:
On Thu, 11 Sep 2003, Bill Rugolsky Jr. wrote:
> On Thu, Sep 11, 2003 at 11:01:20PM +0200, Dag Wieers wrote:
> > It's not that I'm forcing anyone to use the whole argument space. And
it's
> > not that I'm arguing to make it 1Gb either.
Read this again.
> > I don't see why processing 1Gb arguments would be slower than processing
> > 10 times 100Kb arguments. I'd even wildly guess the latter case is slower
> > than the first.
>
> Wrong question. The question is whether adding support for very large,
> or arbitrarily large (hence swappable) arglist+environment makes the
> common case (i.e., 1 page) significantly slower, or otherwise negatively
> impacts the kernel (e.g., resource starvation). We won't know until someone
> implements it. If you are interested in pursuing this, and seeing
> it done the "right" way, see this post by Jamie Lokier from Mar 2000,
> along with the surrounding thread:
>
>
http://www.ussg.iu.edu/hypermail/linux/kernel/0003.0/0887.html
Well, 1Gb is something very different than I would propose. Replace 1Gb by
256Kb and '10 times 100Kb' by '4 times 64Kb' and you're closer to
home.
But if Jamie Lokier doesn't see any reason to have a limit (!), I rest
my case.
Maybe Jamie has had a change of heart since he hasn't implemented it
in the three and a half years since that was written ;o)
Cheers,
Sean