[I think this was only sent to me, not the list, so I'm forwarding...]
----- Forwarded message from Marc Jones Marc.Jones@AMD.com -----
From: Marc Jones Marc.Jones@AMD.com To: Uwe Hermann uwe@hermann-uwe.de Subject: Re: [LinuxBIOS] r386 - in LinuxBIOSv3/arch/x86: . geodelx Date: Wed, 27 Jun 2007 23:11:50 -0600
Uwe Hermann wrote:
On Thu, Jun 28, 2007 at 01:08:28AM +0200, Peter Stuge wrote:
On Wed, Jun 27, 2007 at 11:01:01PM +0200, svn@openbios.org wrote:
+struct constructor geodelx_constructors[] = {
- {.id = {.type = DEVICE_ID_PCI,
.u = {.pci = {.vendor = X86_VENDOR_AMD,.device = 0x05A2}}},- .ops = &geodelx_cpuops},
- {.ops = 0},
+};
Did this get changed to better PCI ID #defines already?
I'm not even sure it works this way. X86_VENDOR_AMD is not a PCI vendor ID, it's simply a 3, see cpu.h. As this uses ".type = DEVICE_ID_PCI" I doubt it'll work. We probably need some DEVICE_ID_CPU or something?
Uwe is correct. The 0x5a2 is returned from a CPUID eax=0x01 instructions. I think that it is defined lxdef.h. Marc
----- End forwarded message -----
Uwe.