https://gcc.gnu.org/bugzilla/show_bug.cgi?id=90889
Martin Jambor <jamborm at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |jamborm at gcc dot gnu.org
--- Comment #8 from Martin Jambor <jamborm at gcc dot gnu.org> ---
(In reply to Jan Hubicka from comment #3)
> The failing check is Martin (Jambor's) sanity check of ipa-cp, so I am
> adding him to CC: at very first glance it looks like ipa bug exposed by LTO.
>
The gcc_unreachable in ipcp_verify_propagated_values gets tripped over
if there is a TOP lattice in any of the IPA-CP scalar constant
propagation lattices after the propagation.
All non-local nodes get all their lattices set to VARIABLE and all
local nodes should either be removed by unreachable code or get
another lattice value from their callers. Therefore, this assert
triggers when a function should have been removed by unreachable node
removal or if IPA-CP has a bad notion of about locality of a node,
usually the former.
Yesterday I ran out of space in /tmp when trying to bootstrap Ada, I
will try again today and have a quick look.
