From mboxrd@z Thu Jan 1 00:00:00 1970 Date: Mon, 25 Sep 2000 13:41:02 -0300 (BRST) From: Rik van Riel Subject: Re: the new VMt In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-linux-mm@kvack.org Return-Path: To: Linus Torvalds Cc: Andrea Arcangeli , Andi Kleen , Ingo Molnar , Alan Cox , Marcelo Tosatti , Roger Larsson , MM mailing list , linux-kernel@vger.kernel.org List-ID: On Mon, 25 Sep 2000, Linus Torvalds wrote: > On Mon, 25 Sep 2000, Andrea Arcangeli wrote: > > > > But I'd much prefer to pass not only the classzone from allocator > > to memory balancing, but _also_ the order of the allocation, > > and then shrink_mmap will know it doesn't worth to free anything > > that isn't contigous on the order of the allocation that we need. > > I suspect that the proper way to do this is to just make another gfp_flag, > which is basically another hint to the mm layer that we're doing a multi- > page allocation and that the MM layer should not try forever to handle it. > > In fact, that's independent of whether it is a multi-page > allocation or not. It might be something like __GFP_SOFT - you > could use it with single pages too. > > Thinking about it, we do have it already. It's called > !__GFP_HIGH, and it used by all the GFP_USER allocations. Hmm, I think these two are orthagonal. __GFP_HIGH means that we are allowed to eat deeper into the free list (maybe needed to avoid a deadlock freeing pages) __GFP_SOFT would mean "don't bother waiting for free pages", which is something very different... (I wouldn't want a user process to get killed simply because kswapd is waiting for IO to finish on a swapout, in that case we really do want to sleep for a while) regards, Rik -- "What you're running that piece of shit Gnome?!?!" -- Miguel de Icaza, UKUUG 2000 http://www.conectiva.com/ http://www.surriel.com/ -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@kvack.org. For more info on Linux MM, see: http://www.linux.eu.org/Linux-MM/