[Bug debug/63572] [5 Regression] ICF breaks user debugging experience

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63572

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
For debugging most important is to get

(gdb) b foo

still work when foo was merged with bar and the program now calls bar

Similarly nice (but probably impossible) is

(gdb) b foo.c:23

with foo.c:23 inside foo.  Of course foo and bar, even if considered equal
by ICF, are by no means equivalent lexically.  So foo.c:23 could at most
break at the start of bar.  (or give a useful diagnostic from gdb)