Hi Wolfgang,
The only thing I wonder is the filename 'memcpy_mpc5200.c' as the code doesn't really have any 5200 specifics in it. What about
As far as I understand this behaviour is specific to the MPC512x and MPC5200 processors; I did not notice it on other cores yet. Unfortunately we do not have a generic name that includes these processors (5xxx is taken by something else).
I believe that other architectures may also suffer from such unaligned byte copies, but maybe in those cases 'memcpy' already takes car of that, I don't know. The point I wanted to make is that the file in question has no powerpc specific code in it at all.
'memcpy_align32wrap' or something likew that to express the more general nature of the code?
I could not come up with a better name... What is "align32wrap" supposed to mean?
It was meant to mean somehow to express the fact that for some condition the original memcpy was called and for some conditions not - hence "wrap". The condition which was passed on to the original memcpy is connected to correct 32 bit alignment, hence the proposal.
If one thinks further along this line, why not move the wrapper to lib/?
Please re-check what my patch does - exactly that:
arch/powerpc/{cpu/mpc5xxx => lib}/memcpy_mpc5200.c | 0
Actually I was aiming for "lib" outside of any architecure for the reason noted above.
But as I said, this is only place for improvement, I surely will not nak the patch ;)
Cheers Detlev