Map of the memory

[Eric W. Biederman]

Xavier Pegenaute <xpegenaute at telepolis.es> writes:

> Hi,
> 
> i was trying to make a memory map in the eeprom but i have some
> questions when i saw linuxbios_c.map and linuxbios.map ...
> 
> 
> We "compile" linuxbios_c with c_start.o and linuxbios.a until generate
> linuxbios_payload and linuxbios_c.map.
> 
> We "compile" crt0.S and we make the crt0.o.
> 
> After we "compile" crt0.o with linuxbios_payload generating linuxbios
> binary and linuxbios.map
> 
> If we see inside the map files, i can see that there are labels at the
> same position in the both map files for example:
> 
> linuxbios_c.map :
> 00004000 A _RAMBASE
> 00080000 A _ROMBASE
> 
> linuxbios.map:
> 00004000 A _RAMBASE
> 00080000 A _ROMBASE

At that point those are just hard coded symbols, that set of
symbols lives nowhere.
 
> How is it possible ? when in theory al linuxbios_c is in the payload of
> linuxbios?

Only when they are symbols that refer to actual code or data is
it a problem.  A constant address is not.
 
> linuxbios.map:
> 000805a0 D _payload
> 00085e5d D _epayload
> 
> also i saw in some file (.ld) in the tree of freebios a picture of memory:
> 
>   	Memory map:
> 
>   	0x00000	(4*4096 bytes)	: stack
>   	0x04000	(4096 bytes)	: private data
>   	0x05000			: data space
>   	0x90000			: kernel stack
>   	0xf0000	(64 Kbyte)	: EPROM
> 
> Is it right ? 

No.  At one point it was correct but the comment is now out
of date.

> i think that i'm confusing the memory in RAM and the
> memory in EPROM. But i don't know how build a correct idea.

Please also note the differences in the physical addresses
and the virtual addresses that sometimes show up in the code.

Roughly the 64KB of the ROM lives at either:
0xf0000 - 0xfffff
or
0xffff0000 - 0xfffffffff

And linuxbios_c lives in the first 640K I would have to look to be
more precise.

Eric