Messages in this thread | | | Date | Tue, 27 Oct 1998 20:16:30 -0800 | From | "David S. Miller" <> | Subject | Re: 4MB pages and framebuffer access, x11perf results, 2.1.125 |
| |
Date: Wed, 28 Oct 1998 05:13:44 +0100 (CET) From: MOLNAR Ingo <mingo@chiara.csoma.elte.hu>
On Tue, 27 Oct 1998, David S. Miller wrote:
> Post your patches so we can actually have a chance to figure something > like that out :-)
uhm, was too embarrased to do that :) Be prepared, it's Really Ugly. The main goal was to get a prototype working fast ...
Ok, here is one suggestion.
Arguably mmap_mem() could be dealing with either real memory or I/O memory, the latter is the case we care about here.
So what we could do is introduce macros to test this attribute for a physical page. Using this we could also instroduce the rule that you can't mmap memory which half straddles physical real memory and I/O space.
Anyways, if the macro says this is I/O space we can call ioremap_page_range() or whatever, which is completely implemented already in per-architecture MM support files.
Go from there.
Just an idea.
Later, David S. Miller davem@dm.cobaltmicro.com
- 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.tux.org/lkml/
| |