On Thu, 18 Oct 2001 12:35:23 -0700, Thomas J. Merritt wrote:

>|I speculate that their code is obfuscated not because anyone wrote it that 
>|way deliberately, but because of historical accident, i.e., their legacy.
>
>Their code is obfuscated in large part because they use cleanly nested
>subroutine calls while running in an environment that doesn't have a 
>memory controller enabled.  There are some pretty sick tricks to do
>four levels of subroutine calls without touching memory.

I'll support that statement.  The BIOS code I have looked at so far
does all kinds of things to be able to support return linkage when
memory isn't up.

Nothing I have seen in the source indicates any deliberate intention
to make the code difficult to reverse engineer.

Aside from haveing to go through over megabyte of assembly code when
I am trying to find a certain area its really not that bad.  It could
be MUCH worse.

>Linuxbios assumes that the hardware works and the you want to run linux
>as soon as possible.  If you have linuxbios installed and have a hardware

Linuxbios also leaves a bunch of stuff to the linux kernel.

>I think that PC BIOS code probably doesn't look as bad in source form as
>it does in object form, but it stall has to be pretty bad.

My experience agrees with this.  Course I have only seen 1 vendors
code.  In the near future though that will expand to 2.  

>|But so long as they can make a buck on it, it'll be here to stay, until 
>|that Great Day comes when they can't hire anyone to work on it.

And what a buck it is... The ammount in per copy royality alone is
huge.  Not to mention that  it takes $30k to get started.   Man I
wish I had gotten into BIOS dev long ago.  Probally could have made
some big money.


--
Richard A. Smith                         Bitworks, Inc.               
[EMAIL PROTECTED]               501.846.5777                        
Sr. Design Engineer        http://www.bitworks.com   


Reply via email to