as of today I am booting linux on my AMD rumba! i am using busybox and getting to a /bin/ash prompt. Things all seem to work.
I can insmod the rtl8139 driver, but I'm not sure yet how INTA hooks up to the SB. That's for today.
Thanks to Steve Goodrich, who came down and got us off the ground, and all the great folks at AMD who have given us so much help. This week's work will be ethernet and graphics.
We're getting there. Anyone else out there doing GX2 at this point?
ron
On 4/23/06, Ronald G Minnich rminnich@lanl.gov wrote:
We're getting there. Anyone else out there doing GX2 at this point?
I've got my eval board. But I was waiting for the VSA stuff to get worked out. Can you send me a VSA bin that works and quick instructions for putting it into my build. I'll pull out my eval board and give it a test.
-- Richard A. Smith
Richard Smith wrote:
On 4/23/06, Ronald G Minnich rminnich@lanl.gov wrote:
We're getting there. Anyone else out there doing GX2 at this point?
I've got my eval board. But I was waiting for the VSA stuff to get worked out. Can you send me a VSA bin that works and quick instructions for putting it into my build. I'll pull out my eval board and give it a test.
good point. Vesa is working, but Ollie has the modules. We'll try to get those to you tomorrow.
We're doing very minimal usage of VSA, with luck tending to 0 over time. We mainly use the vas init stuff and SYSMGR, and we only use SYSMGR so we can get pci config cycles.
thanks
ron
We're doing very minimal usage of VSA, with luck tending to 0 over time. We mainly use the vas init stuff and SYSMGR, and we only use SYSMGR so we can get pci config cycles.
Good plan. The inclusion of yet another hunk of 16-bit binary code in to the mix doesn't seem like a good long-term plan. We have enough trouble with the VGA binaries.
-- Richard A. Smith
Richard Smith wrote:
Good plan. The inclusion of yet another hunk of 16-bit binary code in to the mix doesn't seem like a good long-term plan. We have enough trouble with the VGA binaries.
we did confirm during our discussions last week that there is nothing in principle that disallows SMI in 32-bit mode. So we could do the same trick we do for vgabios, where we pop into 32 bit from 16 bit mode (yes, I wrote that correctly!) and back to 16-bit mode for returning from SMI.
The better option is to write adapters for linuxbios and linux that use the GX2 MSRs directly and don't try to use 0xcfc/0xcf8 for PCI enumeration etc. But we'd have to convince linux, plan 9, bsd, etc. to pick this up ... that's definitely a longer-term issue. But I do think that, in the long run, we don't absolutely have to have vsa and SMI Int at all.
ron
-
Richard Smith -
Ronald G Minnich