1. So the kernel need to know Hypertransport HT scan. 2. Also we need to move silicon bug fix code to kernel driver..., or cache_as_ram stage code...
YH
-----Original Message----- From: Ronald G Minnich [mailto:rminnich@lanl.gov] Sent: Wednesday, May 03, 2006 9:14 AM To: Lu, Yinghai Cc: Richard Smith; LinuxBIOS Subject: Re: [LinuxBIOS] Error in FSF high priority list LinuxBIOS description
Lu, Yinghai wrote:
You don't need linuxbios_ram part to scan pci and allocate resource
for
PCI devices? Or let kernel to do these allocation?
I want to find out how to let the kernel do it. It's almost capable. This was our goal, 6 years ago; maybe the kernel is ready.
ron
Lu, Yinghai wrote:
- So the kernel need to know Hypertransport HT scan.
- Also we need to move silicon bug fix code to kernel driver..., or
cache_as_ram stage code...
I'm doing this on OLPC to start. HT is too much for me to think about.
Simple things first.
ron
* Lu, Yinghai yinghai.lu@amd.com [060503 18:28]:
- So the kernel need to know Hypertransport HT scan.
- Also we need to move silicon bug fix code to kernel driver..., or
cache_as_ram stage code...
YH
All in all I like the fact that we take this kind of stuff out of "the kernel", and have it seperately. I agree that the line that is drawn by the proprietary BIOS that makes up the interface today is less than optimal.
On the other hand, Linux is really complex already and for me small parallel and understandable systems are worth a lot more than creating an egg-laying wool-milk-pig (heck, is there something like this in english? all-in-one device suitable for every purpose..)
The linuxbios <--> payload determination is a really good one. For one real reason: I take the same kernel that works with legacy bios and I put it into flash given thats big enough.
When I am going to flash LinuxBIOS, I know that I will not have to look at much more than the ~ 200,000 lines of LinuxBIOS code (wc -l on all files, including documentation).
The idea that we want to or should cope with the several millions of Linux code in the very same code is very ambitious. And scares me a bit, thinking of the mess I had with MILO.
Stefan
-----Original Message----- From: Ronald G Minnich [mailto:rminnich@lanl.gov] Sent: Wednesday, May 03, 2006 9:14 AM To: Lu, Yinghai Cc: Richard Smith; LinuxBIOS Subject: Re: [LinuxBIOS] Error in FSF high priority list LinuxBIOS description
Lu, Yinghai wrote:
You don't need linuxbios_ram part to scan pci and allocate resource
for
PCI devices? Or let kernel to do these allocation?
I want to find out how to let the kernel do it. It's almost capable. This was our goal, 6 years ago; maybe the kernel is ready.
ron
-- linuxbios mailing list linuxbios@linuxbios.org http://www.openbios.org/mailman/listinfo/linuxbios
Stefan Reinauer wrote:
On the other hand, Linux is really complex already and for me small parallel and understandable systems are worth a lot more than creating an egg-laying wool-milk-pig (heck, is there something like this in english? all-in-one device suitable for every purpose..)
ok you win.
I won't do it :-)
ron
-
Lu, Yinghai -
Ronald G Minnich -
Stefan Reinauer