On Nov 21, 2007 7:46 PM, Uwe Hermann uwe@hermann-uwe.de wrote:
On Tue, Nov 20, 2007 at 08:00:14PM -0500, Corey Osgood wrote:
On 11/20/07, Coquelicot coquelicot408@gmail.com wrote:
Hi all,
what can be responsible for long delay before starting LinuxBIOS? I have it on VIA EPIA M-II with LB preppended by original VGA bios extracted from the original BIOS. The delay is ~20 seconds before I get first message on serial console (..."In auto.c:main()" messages).
Correct me if I am wrong: do I need VGA BIOS at all? I want my booting
No. On a head-less server for example, you usually don't care about VGA, and you can easily leave it away (no VGA blob to prepend to linuxbios.rom).
Disable VGA support in LinuxBIOS with CONFIG_CONSOLE_VGA=0 in Options.lb.
What about a carputer which need to be able to display graphics (X or DirectFB)? can I skip the VGA bios in such setup altogether? I'm talking about VIA EPIA M-II so I am not sure if two LB "drivers" you mention below (RageXL and GeoDE) can work?
Linux kernel to use framebuffer but I guess that original VGA BIOS is not really needed for that, just to be able to "control" LB/FILO from the local console, right?
Depends. LinuxBIOS itself has (AFAIK) two "drivers" which can enable graphics without executing any VGA blob, one for RageXL and one for Geode GX1.
I'm not sure if the Linux framebuffer drivers depend on a VGA blob which has to be executed before they work. Most probably do, but maybe not all.
If I had to guess, I'd say your problem is not with the VGA bios, but with ram that doesn't work quite right with the current code for the cle266. Can
Yep, possible. I've seen the same effect on a box, but I can't really explain it. It happens _before_ the UART init, i.e. before the first byte of debugging output can be sent, and way before RAM init, though. Haven't looked further into it.
you try a different ram stick (of a different size/timings) and see if you get the same results?
Thanks for the tip, I will try to find some other memory to test...