On Fri, Feb 3, 2012 at 8:55 AM, Russell King - ARM Linux <linux@xxxxxxxxxxxxxxxx> wrote: > On Fri, Feb 03, 2012 at 08:12:02AM -0800, Bjorn Helgaas wrote: >> On Fri, Feb 3, 2012 at 8:01 AM, Russell King - ARM Linux >> <linux@xxxxxxxxxxxxxxxx> wrote: >> > On Mon, Jan 30, 2012 at 09:57:45AM -0700, Bjorn Helgaas wrote: >> >> Tell the PCI core about host bridge address translation so it can take >> >> care of bus-to-resource conversion for us. >> > >> > Are the rest of these patches (in particular whereever is introduced >> > the generic versions) available somewhere? >> >> Oh, sorry, this doesn't mean much without the context. Here's a >> pointer to the overview, with a git tree and so on: >> >> http://marc.info/?l=linux-arch&m=132794281821723&w=2 > > Trying to read github in elinks isn't very nice. But from what I could > see it looks fine. When I get around to pulling my private patchsets > forward, I'll try to remember to give this a test. Thanks for taking a look. Let me know if you find issues. > The thing which was concerning me is that I have been carrying the > following patches for the last 10 years or more in various forms to > make PCMCIA work on a couple of platforms I have here. It's more > or less the same problem, and it uses the bus_to_resource stuff to > fix it for Yenta. > > These patches address PCMCIA's idea of 'lowmem' windows always being > at the low 640K of _CPU_ memory rather than the low 640K of _bus_ > memory. When your bus starts at an offset of more than 640K in CPU > memory, the resource handling doesn't work as well. My patches only affect devices below a PCI host bridge, and from the device's point of view, it should see no difference -- the values in its BARs shouldn't change. I know very little about PCMCIA. The dependency on the CPU addresses is interesting. I would think such a dependency would mean we're missing a bus/resource conversion somewhere, but it must be more complicated than that. But in any event, my *intent* is that neither CPU addresses nor bus addresses should change. If they do change, that's a bug in my series. > I've never pushed these upstream because I'm not entirely sure why > PCMCIA deals with lowmem windows on a card-by-card basis, rather than > a socket basis (surely its more to do with the capabilities of the > socket interface than the card itself...) Bjorn -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
