[En-Nut-Discussion] bootloader patches submitted for approval

[Dave]

Hi Harald,

May I suggest you also include a pre-compiled .hex file so those that
don't have the latest compiler can still install it without being forced
to upgrade their compiler. They would only be forced to upgrade
compilers if they wanted to modify it. I suspect that some users that
use the bootloader feature would want to upgrade, otherwise I wouldn't
bother to make that suggestion.

Also, if you compile and compare it to my eboot.hex file I attached
before, there will be a slight difference, since in main() I moved the 2
asm lines before the 2 usart lines, but didn't include that change in my
patch file.

Oops, I forgot to change the docs in tftp.c, so the line that says
" * \param page The page number to program, 0..479." should have '479'
changed to '495'

On a different subject, I want to personally thank you and everyone else
who has contributed countless time and effort to make Nut/OS such a nice
operating system. I'm glad I can contribue a little something back to
the ethernut and Nut/OS community I've gotten so much from. I wished I
could do more, but this is a start, and I'm sure I'll be contributing
more in the near future.

-Dave

On Thu, 2005-05-26 at 17:13 +0200, Harald Kipp wrote:
> Dave,
> 
> Probably avr-gcc 3.4.3 provides better code optimization.
> 
>  From experience with the LAN91C111 bootloader I'm sure,
> that there is some potential to optimize the code.
> It's just 40 bytes overflow. Because of PHY initialization,
> the LAN91C111 bootloader needs more functions, but fits
> in 4094 bytes (using gcc 3.3.1).
> 
> I'd not like to force everyone to upgrade the compiler.
> Specially in existing commercial projects this might be
> a big problem.
> 
> For now we can leave it as it. If someone wants to
> build this very special part of the distribution, he
> may upgrade the compiler.
> 
> Harald