On Tue, Oct 7, 2008 at 9:44 AM, Myles Watson mylesgw@gmail.com wrote:
Here is a proposed change to dts. I've forgotten all I ever knew about the 8111 so I am pretty sure this is incomplete, but it's something like what we want.
Index: mainboard/amd/serengeti/dts
--- mainboard/amd/serengeti/dts (revision 904) +++ mainboard/amd/serengeti/dts (working copy) @@ -28,18 +28,17 @@ /config/("northbridge/amd/k8/domain"); pci@1,0{ };
/* guesses; we need a real lspci */ pci0@18,0 { /config/("northbridge/amd/k8/pci");
pci@0,0 {
pci@0,0 {
/config/("southbridge/amd/amd8111/amd8111.dts");
pci@1,0 {
/config/("southbridge/amd/amd8111/nic.dts");
}; }; pci@4,0 { /config/("southbridge/amd/amd8111/ide.dts"); };
pci@5,0 {
/config/("southbridge/amd/amd8111/nic.dts");
}; }; pci1@18,0 { /config/("northbridge/amd/k8/pci");
It still doesn't seem quite right. Segher?
I've played with it a little more, no luck yet. Could you help me understand the naming convention?
pci@5,0 = pci at device 5 function 0 ?
yes
I don't see how we say there's a pci bridge at device 0 function 0 on this bus, then specify the devices on the pci bus from there.
the bridge is actually implied by the fact that the node has child nodes (as in v2). But yes, I see your point. I don't have time to look today but there is some simple thing wrong. The devices are straight from v2.
Should we be using the amd8111/pci.dts here too? Maybe that's why you had the phase3_scan = 0, since the pci bus should take care of its scan?
Hmm, not sure, ref. v2 again.
I think we should take a careful look at v2 again but I think this is a simple problem that just needs close looking at ...
ron