[coreboot] porting Coreboot to a new motherboard....

[Stefan Reinauer]

* ali hagigat <hagigatali at gmail.com> [120214 18:48]:
> I have initialized Intel 82815 SDRAM controller but for a special type
> of RAM without reading SPD registers. I know that DRP register is :
> 0xCC. I added the code to Coreboot.
> 
> I tested much and concluded that RAM has a problem in a way as you
> guys told me before. When i change the hardwaremain() , even adding
> some code which are bypassed, the results of the post code numbers
> become different!
> 
> I do not know what to do! c_start.S is OK and it is executed and the
> program starts executing hardwaremain() but adding code to the
> hardwaremain() any where seems to disorder printed post code values.
> 
> Is there any interrupt or timer set immediately after hardwaremain()?
> 
> I checked coreboot_ram in build/. The assembly code of hardwaremian() seems OK.
> 
> 
> void sdram_enable(void)
> {
> 	asm("pushal");
> 	asm("push %eax");
> 	asm("push %ebx");
> 	asm("push %ecx");
> 	asm("push %edx");
> 	asm("push %esi");
> 	asm("push %edi");
> 	asm("push %esp");
> 	asm("push %ebp");
> 	asm("pushfl");
> 	asm("jmp  firstlbl");
> 	asm("mov  %ax, %ax");
> 	asm("nop");

[..]

There's a number of things wrong with this approach:

 * if you don't use asm volatile instead of asm, gcc is free to reorder
   any of those statements as it sees fit.

 * you need to put all of the code in one asm volatile statement, or gcc
   will attempt to clean up after each of the statements or overwrite
   registers as it sees fit.

But generally, I suggest you rewrite this in C, so it's a lot more
readable...

Stefan