Georg-Johann Lay <avr@xxxxxxxx> writes: > I tried the REG_EXPR of the address and it looked good, but the > internals say that there may be cases where REG_EXPR is 0, so that > there would be no way to see if the callee is tagged with the > attribute or not. REG_EXPR will be 0 for compiled-created temporary variables, but that is not what you have here. I think that if you find that REG_EXPR is present in your tests, then it should be reliable. > Then I am confused about the internals. In "9.3 Types" > http://gcc.gnu.org/onlinedocs/gccint/Types.html#Types > Macros like TYPE_PTRFN_P are documented, but are not present in the > gcc source tree. I am using gcc core for the port, but these macros > are contained only in the C++ part in ./cp/cp-tree.h. Maybe this could > be moved when reorganizing the gcc source tree, as such macros are > useful for C, too? Moving a macro like TYPE_PTRFN_P into tree.h sounds reasonable, whether or not the source tree is reorganized. Ian
