Re: IPA merge: cgraph_add_function improvement

[Andrew Pinski <pinskia at physics dot uc dot edu>]

> 
> Hi,
> this patch make cgraph_add_new_function to work in more contextes than gomp.
> We currently do add functions via cgraph_build_static_cdtor, via
> cgraph_add_function and via clonning.  THis is somewhat painful as the
> logic of what needs to be done is duplicated across those places and also
> dependent on stage of compilation. 
> 
> This patch moves all the logic into one place so it is not dependent on
> stage of compilation the function is called and form of function passed
> to it and thus we can move passes around more easilly (on IPA-SSA I do
> GOMP before inlining).  The function is inherently somewhat add-hoc (ie
> whole plan behind cgraph was to make all functions to be noticed early),
> but I hope to simplify it once IPA is more converted to pass manager and
> add some ability to execute all local passes that should be done up to
> current IPA pass.
> 
> Bootstrapped/regtested i686-linux, will commit it tomorrow if there are no
> complains.

This causes a bootstrap failure on powerpc-aix and maybe other targets which don't
have support for the init section.

A reduced testcase is:
static void f(void) __attribute__((constructor));
static void f(void) {}


I tested this with a cross compiler to powerpc-aix5.3.

The ICE we get is:
t.c:2: internal compiler error: in cgraph_expand_all_functions, at cgraphunit.c:1220
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.


Thanks,
Andrew Pinski