--- In l...@yahoogroups.com, David Bermingham <crwpulsar@y...>
wrote:
> Yeah, I call iap_entry(command,result) and the next line of the code
is printf something and it never calls the printf. I pretty much
copied the philips code from the app note because this stuff is new to
me!(Like most of it!!!) so it must be incorrect parameter passing??

I guess you are using the GCC compiler and the ARM instruction set?
Because I'm using that (along with AsIDE) and had exactly the same
problem.

After some reading, and googling and some more reading and some
assembly debugging we found some clues that lead to the solution.

In short: turn on the compiler option '-mthumb-interwork'. This is
nessecary since the bootloader function you are accessing is written
in Thumb code, see page 193 of the user manual of 2003 okt 02.

You should jump from ARM to Thumb using the BX instruction from the
ARM instruction set. This is automatically done by your compiler but
GCC has this function (thumb-interwork) disabled by default since it
generates slightly bigger code. When you switch it on you shuld be
fine...

Hope that helped!
Greetings,
Herman

--- In l...@yahoogroups.com, David Bermingham <crwpulsar@y...>
wrote:
> Thanks for the reply, Im using ARM ADS 1.2 to compile the code!
Does anybody know how or if I have to turn on interworking for ADS?

Hi David, yes you need to turn on interworking, however this is done
for you automatically if you create an application using the
Thumb/ARM interworking template. This is convenient as you simply
place source files to be built as ARM code in the ARM subproject,
and the IDE takes care of all the other details.

Alternatively, you'll find the interworking tickbox in the project
settings dialog, under the ATPCS tab when you select the assembler,
ARM C compiler or Thumb C compiler. A major irritation (for me
anyway) is that you have to set this for each subproject (Thumb and
ARM), each target (Debug, DebugRel, Release) and each tool (asm, C,
C++) that you use.

> 
> Also, I have a routine which has to be executed out of RAM instead
of Flash to operate properly. When I change the scatterloading file
so that aes.o is loaded in RAM the program never branches to
it...its like its not there!!
> When I check the map file generated by ADS it says the code is
located in RAM.
> 
> Is there a way to set up the scatterloading file so that the
routine is copied from Flash to RAM and executed from RAM?

Can I ask first if you have a main() function in your project? If
you have, then the linker pulls in some C-startup code, which
includes a section to move code and initialised data from their load
positions (in Flash) to their execute positions in RAM.

If not, you have to write your own code snippets in the reset
handler to move the code and data around. Conveniently, the linker
will generate symbols giving you the load address, execution
address, and length of each region to be moved. See page 4-2 of the
Linker and Utilities guide, for a description of the format for
these symbols.

HTH, Peter.

--- In l...@yahoogroups.com, David Bermingham <crwpulsar@y...>
wrote:
> Hi Peter,
> Thanks for the info....that solved the IAP issue!!Works prefectly
now!
> 
> As regards the second problem/issue, The startup file Im using
just initialises the exception vectors, stacks and branches to main
(). Do I need another assembly routine to move the data from Flash
to RAM?? I thought the scatterloading file set that up?
> The problem I have is that any global variables or buffers in the
code seem to cause the error! I can remove the problem by using
local variables and passing them between functions but this is less
than pretty!!!
> 
> Thanks,
> David
> 

Hi David,

Excellent - glad the IAP is working now. For the scatterloading
problem, try branching to __main instead of main. Have you also
written your own version of __user_initial_stackheap, to tell the C
startup code where to place the heap base and stack limit?

Peter.