Another inlining idea
Richard Guenther
rguenth@tat.physik.uni-tuebingen.de
Wed May 7 20:46:00 GMT 2003
Hi!
In order to help the user specify inline decisions, how hard would it be
to implement an attribute that forces not inlining of the function itself,
but recursively all called functions from within it? F.i. once we start
processing the marked function, set a global flag to force all functions
inline. I'd suggest __attribute__((force_flat)).
Does this sound feasible?
Richard.
More information about the Gcc
mailing list