Messages in this thread | | | Date | Mon, 25 Sep 2000 19:03:46 +0200 (CEST) | From | Ingo Molnar <> | Subject | Re: the new VMt |
| |
On Mon, 25 Sep 2000, Linus Torvalds wrote:
> Yes, I'm inclined to agree. Or at least not disagree. I'm more arguing > that the order itself may not be the most interesting thing, and that > I don't think the balancing has to take the order of the allocation > into account - because it should be equivalent to just tell that it's > a soft allocation (whether though the current !__GFP_HIGH or through a > new __GFP_SOFT with slightly different logic).
yep, and there is another problem with pure order-based distinction: if i do kmalloc(5k), and write the code on Alpha and expect it to never fail, shouldnt i expect this to never fail on x86 as well? Along with the fork() failure. __GFP_SOFT solves this all very nicely - the *allocator* decides what allocation policy to follow. Great!
Ingo
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org Please read the FAQ at http://www.tux.org/lkml/
| |