https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64164
--- Comment #21 from Alexandre Oliva <aoliva at gcc dot gnu.org> --- (In reply to rguent...@suse.de from comment #19) > For some reason > into-SSA inserts a debug stmt, creating an extra copy _5 = i_4?! I bet the extra copy is there even without -g. _1, _3 and _5 seem to be SSA names for the result decl. Presumably _5 might be optimized out, replaced with i_4 in the _1 PHI node, at a later pass. > But are you sure > that we are never using REG_EXPR for debug? We use them for non-gimple_register auto variables. > With > -fno-var-tracking-assignments for sure we do (I suppose we don't care), Yeah, I'd say we don't care. If one asks for debug info degradation, we obey ;-) (now, sometimes we disable VTA on our own because some internal limit is exceeded; that's unfortunate, but what are you gonna do? :-)
