Steve,
I had that feeling that TI documentation is bit messy (pdf datasheep, page 
14):
So, according to the 16x datesheet, memory organization of 1610 is:

3kB : 24ff - 1100 (which is really 5120 bytes!)
2k :   9ff - 200                (which is 2kB)

1611:
8kB:    38ff - 1100 (which is really 10240 bytes)
2k: the same as above.

and i have no any info on 1612. Any link appreciated :)

There are no any words about memory mapping (say map lower 2kB to upper 5 or 
10 kBytes)

Therefore, the question is:
is it possible to use lower 2k ?
If yes, the it seems to be reasonable to use 2k as main memory and upper 
memory as '.heap' (as I planned before). So, check the file
binutils-XXX/ld/scripttempl/elf32msp430.sc
where you get some clue about heap memory declaration -- 
in file msp430all.sh:

GOT_HEAP_MSP must be defined as non zero
HEAP_START should be start of upper RAM which is 0x1100
HEAP_LENGTH     should be lenght of upper memory
STACK should be a top of lower memory

Then....
gcc will refer to 2k memory for globals, locals and stack.
malloc() already sharpen to work with '.heap' memory and user can use it for 
large arrays.

OR VICE VERSA :) -- 2k for heap and upper memory for stack, locals, globals, 
etc...

What do you think?

Cheers,
Dmitry.



On Monday 23 February 2004 15:01, Steve Underwood wrote:
> Hi,
>
> The attached patch is against today's (040223) snapshot of binutils at
> sources.redhat.com. It adds support for the new 1610, 1611, and 1612
> which should be sampling about......... NOW!
>
> Dmitry:
> This needs to be sent to the binutils maintainer.
>
> Regards,
> Steve

-- 
/*****************************************************************
     ("`-''-/").___..--''"`-._     (\   Dimmy the Wild     UA1ACZ
      `6_ 6  )   `-.  (     ).`-.__.`)  State Polytechnical Univ.
      (_Y_.)'  ._   )  `._ `. ``-..-'   Radio-Physics Departament
    _..`--'_..-_/  /--'_.' ,'           Saint Petersburg,  Russia
   (il),-''  (li),'  ((!.-'             +7 (812) 5403923, 5585314
 *****************************************************************/


Reply via email to