Re: ZONE_PADDING wastes 4 bytes of the new cacheline

[Andrea Arcangeli]

On Fri, Oct 22, 2004 at 01:02:24PM +1000, Nick Piggin wrote:
> I don't agree, there are times when you need to know the bare pages_xxx
> watermark, and times when you need to know the whole ->protection thing.

we'll see, I agree current alloc_pages is quite clean but I'm quite
tempted to have a strightforward alloc_pages as clean as 2.4:

	for (;;) {
		zone_t *z = *(zone++);
		if (!z)
			break;

		if (zone_free_pages(z, order) > z->watermarks[class_idx].low) {
			page = rmqueue(z, order);
			if (page)
				return page;
		}
	}

2.6 is like this:

	/* Go through the zonelist once, looking for a zone with enough * free */
	for (i = 0; (z = zones[i]) != NULL; i++) {
		min = z->pages_low + (1<<order) + z->protection[alloc_type];

		if (z->free_pages < min)
			continue;

		page = buffered_rmqueue(z, order, gfp_mask);
		if (page)
			goto got_pg;
	}


I don't see any benefit in limiting the high order, infact it seems a
bad bug. If something you should limit the _small_ order, so that the
high order will have a slight chance to succeed. You're basically doing
the opposite.

The pages_low is completely useless too for example and it could go.
pages_min has some benefit for some more feature 2.6 provides (that
could be translated in more watermarks, to separate the "settings of
the watermarks" from the alloc_page user of the watermarks).

> OK I dont disagree that your setup calculations are much nicer, and
> the current ones are pretty broken...

ok cool.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/