Re: [tree-ssa][rfc] -finstrument-functions at tree level

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

On Dec 10, 2003, at 17:50, Richard Henderson wrote:

> Even after looking at this and preserving apparently intended
> semantics, I don't see that those semantics are useful.  Given
>
> 	void a() { }
> 	void noinline b() { a(); }
> 	void c() { b(); }
>
> is __cyg_profile_func_start(a, c) really meaningful?  I'm not
> sure that it is.  I would sorta expect that inlined functions
> don't get noticed by this hook at all.
>
> Thoughts?

Yes, is there any way to get this into the mainline as 
-finstrument-functions
is broken on the mainline, see PR 10584.
Also I will close the bug that said GCC should do 
-finstrument-functions at the tree-level.

Thanks,
Andrew Pinski