On 02/04/13 23:02, Jeroen Hofstee wrote:
Hello Wolfgang,
On 02/04/2013 09:35 PM, Wolfgang Denk wrote:
Dear Nikita Kiryanov,
In message 510FAB72.3050901@compulab.co.il you wrote:
I did mention that I consider that fix as temporary. I believe this fix is better because it addresses the issue everywhere and does so in a more maintainable way (does not require the address fix to be duplicated everywhere there is a problem).
Can you please explain (again) what exactly you consider the problem?
That I can load a BMP image to anodd address, and that this will result in odd results?
If you use eldk 5.3, a board will be bricked if you set the load address to not (an aligned address _+ 2_). So anyone who didn't know the +2 requirement, has a bricked board (and can't recover). So this is really a different then md trapping, since resetting the board won't help... it is broken / kaput.
Agreed totally, by anyone we mean any _user_ that don't even need to care about the bmp header structure and uses the perfectly 32 bit aligned address.
That said, I am not really in favor of this solution. But I do think it needs to be fixed.
I actually like the approach as it gives an abstraction for: 1) different compilers and their default options, 2) architecture specific "problems" like weather it can deal with unaligned accesses