This patch mostly documents the functionality of amdk8_util.asl. It also fixes a couple of logical errors that were corrupting the resources returned from _CRS.
Fixes: 1. When adding a MMIO region to the southbridge, the previous region was corrupted. - Save previous region before checking if you're handling the southbridge. - Make sure you're on node 0 before adding the TOM region. - Add in a MMAX value of 0xFFDFFFFF (See the ACPI question at the end.) - Store resulting region with the saved region. - Move increment of Local 4 so it's more obvious what it refers to. Before: [TOM - 0xBFFFF] added to every node with Link 0 enabled After: [0xA0000-0xBFFFF] and [TOM-0xFFDFFFFF] only once 2. Missing RTAG call corrupted return value when there was no resource for the link - Add RTAG (BUF0)
Signed-off-by: Myles Watson mylesgw@gmail.com
But the question is:
Why is this logic here? Why do you want a region from TOM-(4G-2M)* when you explicitly have all the regions that are reachable from the HT links in other resources?
* I got 4G-2M from the ACPI FAQ.
Thanks, Myles
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hi,
Why is this logic here? Why do you want a region from TOM-(4G-2M)*
Tha acpi FAQ is just a example imho. I guess we should add only the PCI mem and not the real mem maybe?
Please have you checked the m2v-mx version of the k8-util.asl? I fixed there also something.
Rudolf
On Mon, Mar 9, 2009 at 1:36 PM, Rudolf Marek r.marek@assembler.cz wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
Hi,
Why is this logic here? Why do you want a region from TOM-(4G-2M)*
Tha acpi FAQ is just a example imho. I guess we should add only the PCI mem and not the real mem maybe?
Please have you checked the m2v-mx version of the k8-util.asl? I fixed there also something.
It looks like you added the Increment so that sizes end up being round numbers. I found that one too. The strange thing is that SimNOW (serengeti_cheetah) loads Windows fine even with the bugs. It's frustrating because even though I'm making it more compliant, I'm not sure I'm getting closer.
Thanks, Myles
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
It looks like you added the Increment so that sizes end up being round numbers. I found that one too. The strange thing is that SimNOW
Yes see the ACPI FAQ windows needs it. I'm now reviewing testing your ACPI patches, stay tuned.
Rudolf