Re: [lwip-devel] SMEMCPY()

[address@hidden]

Am 12.06.11 20:20, schrieb Kieran Mansley:
> > Anyway, this change has been mimal-invasive by defining both MEMCPY() and 
> > SMEMCPY() to memcpy(), so I cannot see the downside of it.
> The downside would be that if we previously had an efficient byte copy inlined 
> in the code, and  that was replaced with a call to SMEMCPY() which maps to 
> memcpy(), then we could have harmed performance.
Oh, OK, now I get it. Sorry for taking so long :-)

I see your point there. In that case, letting SMEMCPY() default to a 
byte copy define might be the best idea? Something like:

#define SMEMCPY(dst, src, len) do { size_t i; for(i = 0; i < len; i++) { 
((u8_t*)dst)[i] = ((u8_t*)src)[i]; } }while(0)

should do it, then.
Of course, it's always size vs. speed, I guess. On some platforms, 
calling memcpy is smaller than a local byte-copy loop, on other 
platforms it may be the other way round. However, I think the 
define-version is better than defaulting to memcpy here.

Simon