Re: Unnecessary hash lookup in cgraph_finalize_function?

[Richard Sandiford <rsandifo at redhat dot com>]

Jan Hubicka <hubicka@ucw.cz> writes:
> > Couldn't it just call cgraph_remove_edge directly, as per the patch below?
> 
> Yes, this patch is fine.  Thanks!

Thanks, it's in.

> Unfortunately the patch just works around the problem we need to solve
> anyway.

Feared as much.

>  I have both patches - to declare this invalid (as sugested by
> Geoff) and to update the tables accordingly (I currently use on hammer branch).
> We should make the decision on what path we would
> like to go.  The testcas above looks pretty sane to me, so perhaps we
> really want to go the renaming way as separating sane from insane
> testcases is insanely dificult.

Sane by glibc standards maybe. ;)  How much contortion does
the hammer version need to do?  It certainly sounds better from a
user-friendliness POV.

Richard