Runtime flow of LinuxBIOS with a payload(etherboot for IDE) for FLASHROM and IDE

Richard Smith rsmith at bitworks.com
Mon May 5 13:11:00 CEST 2003


Deepak Kotian wrote:

>     If I see the Makefile, the romimage is done using
>     cat payload linixbios >romimage.
> 
>     1. Why is the payload stuck before linuxbios, any specific
>     reason ?

Linuxbios has to be the last bit of code in your rom image.  See the 
recient thread on how the x86 starts up.  You have to have startup code 
located at that upper address or you will be attempting to run data.

Plus most chipsets will only assert the BIOS rom chipselect for specific 
address ranges.  So you have to have your bios code located in those 
ranges or your rom won't be selected.

Basically its a hardware thing.  Linuxbios (or at least the startup 
code) needs to be located in the last 64k of address space.

>     2. Is there some specific addressing rule one needs to flow
>     when making/burning the payload and linuxbios , basically the sizing?

That's kinda board specific.  Depends on how the chipset works and the 
size of your rom.  For example on my board I have 2 512KiB roms.  So I 
have to tweak ROM_IMAGE_SIZE and PAYLOAD_SIZE such that they equal 
512KiB which locates linuxbios in the right spot.

>     3. I assume that the entherboot payload is also copied to RAM , is
>         this correct ?

Yes.

>         If yes, is there a way to know, if it is done properly.

Sure.  If it boots its probally done correctly. *grin*  There are a few 
other items you can use to test though.

1) is the ramtest.inc code.  Include it directly after you get ram setup.

see freebios/src/ram/ramtest.inc

When I was using this I had the following in my mainboard config file 
dirctly after my northbridge and southbridge inits

mainboardinit ram/ramtest.inc
mainboardinit mainboard/bitworks/rim/do_ramtest.inc

where do_ramtest.inc calls the ram test routine.

2) Is to build memtest86 as an elfimage and boot that.  Memtest does 
extensive memory testing.  Of course you need elfboot working first.


-- 
Richard A. Smith
rsmith at bitworks.com





More information about the coreboot mailing list