https://gcc.gnu.org/bugzilla/show_bug.cgi?id=79776
--- Comment #4 from Richard Biener <rguenth at gcc dot gnu.org> ---
Patch doesn't help.
Note IPA PTA is a simple IPA pass running after inlining decisions are
committed to the IL so there shouldn't be any inlining clones around? Oh, and
it does try to handle them:
/* Build the constraints. */
FOR_EACH_DEFINED_FUNCTION (node)
{
varinfo_t vi;
/* Nodes without a body are not interesting. Especially do not
visit clones at this point for now - we get duplicate decls
there for inline clones at least. */
if (!node->has_gimple_body_p () || node->global.inlined_to)
continue;
obviously it doesn't work...
One way would be to, in this loop, only walk over ultimate_alias_target ()
nodes
(hopefully those will be the ones with body...).
@@ -7785,6 +7793,10 @@ ipa_pta_execute (void)
there for inline clones at least. */
if (!node->has_gimple_body_p () || node->global.inlined_to)
continue;
+
+ if (node->ultimate_alias_target () != node)
+ continue;
+
node->get_body ();
gcc_assert (!node->clone_of);
doesn't work either.
So we have the same decl in multiple alias chains?! And multiple times in the
symbol table?
Sth is fishy with cgraph here.
