Re: Strange interrupt behaviour

Rik van Riel (H.H.vanRiel@phys.uu.nl)
Sun, 12 Jul 1998 08:25:33 +0200 (CEST)

On Sat, 11 Jul 1998, Linus Torvalds wrote:
> On Sun, 12 Jul 1998 ak@muc.de wrote:
> >
> > Maybe it is a "balancing issue" on machines with 128MB of memory, but on
> > 8MB machines I see no way to get this to work without making it unacceptable
> > slow or waste lots of memory (barring radical solutions like replacing the
> > buddy page allocator with something that is less prone to fragmentation)
>
> I can see one really _trivial_ way of handling this: make the page size be
> 8kB as far as the page allocator is concerned.
[SNIP]
> However, the big downside (apart from slightly bigger memory use) is that
> it impacts user level (mmap would also be 8kB-constrained). And that is
> probably unacceptable.
>
> So the sligtly uglier version of this is to do all page allocations as 8kB
> chunks, and then the memory management layer has its own "sub-buddy" to
> split a 8kB page into two hardware pages. The sparc already does something

*Bzzt* This will mean that we'll free pages on a per-4k-page
basis too, returning to the current failing behaviour...

We'll need something a little more radical, or just halfway
radical by counting the contiguous buddies (of eg. 3 pages)
and allocating from smaller areas only, giving the larger
areas chance to grow.

But this is so close to a zone allocator that we might as well
do a lot of finetuning right now and start 2.3 real fast...

> However, I'd prefer to still try out some other ways of handling this. For
> example, "__get_free_pages()" currently only re-tries once. It shouldn't
> be hard to make it re-try a few more times, and it might well be enough to
> make the problem go away.

retry = (gfp & GFP_WAIT ? 3 : 1);
do {
stuff();
} while (retry--);

> 2.1.x is not going to be usable on 4MB machines. I didn't even have to

That's a real shame, IMHO. But with the new VM enhancements we
planned, 2.3 might actually be usable on small boxes again...

Rik.
+-------------------------------------------------------------------+
| Linux memory management tour guide. H.H.vanRiel@phys.uu.nl |
| Scouting Vries cubscout leader. http://www.phys.uu.nl/~riel/ |
+-------------------------------------------------------------------+

-
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.altern.org/andrebalsa/doc/lkml-faq.html