[Bug debug/69077] [6 Regression] omnetpp ICEs with -flto -g

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69077

--- Comment #15 from rguenther at suse dot de <rguenther at suse dot de> ---
On Mon, 11 Jan 2016, hubicka at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69077
> 
> --- Comment #14 from Jan Hubicka <hubicka at gcc dot gnu.org> ---
> I suppose the problem here is that lto-symtab is not merging symtab nodes but
> only the FUNCTION_DECLs so the flag is not merged. Something like this may
> help.
> Index: lto-symtab.c
> ===================================================================
> --- lto-symtab.c        (revision 232227)
> +++ lto-symtab.c        (working copy)
> @@ -997,6 +1005,8 @@ lto_symtab_prevailing_virtual_decl (tree
>      n = n->next_sharing_asm_name;
>    if (n)
>      {
> +      DECL_POSSIBLY_INLINED (n->decl) |= DECL_POSSIBLY_INLINED (decl);
> +      DECL_POSSIBLY_INLINED (decl) |= DECL_POSSIBLY_INLINED (n->decl);
>        lto_symtab_prevail_decl (n->decl, decl);
>        decl = n->decl;
>      }

Works.  Testing

Index: lto/lto-symtab.c
===================================================================
--- lto/lto-symtab.c    (revision 232261)
+++ lto/lto-symtab.c    (working copy)
@@ -997,6 +997,18 @@ lto_symtab_prevailing_virtual_decl (tree
     n = n->next_sharing_asm_name;
   if (n)
     {
+      /* Merge decl state in both directions, we may still end up using
+        the other decl.  */
+      TREE_ADDRESSABLE (n->decl) |= TREE_ADDRESSABLE (decl);
+      TREE_ADDRESSABLE (decl) |= TREE_ADDRESSABLE (n->decl);
+
+      if (TREE_CODE (decl) == FUNCTION_DECL)
+       {
+         /* Merge decl state in both directions, we may still end up 
using
+            the other decl.  */
+         DECL_POSSIBLY_INLINED (n->decl) |= DECL_POSSIBLY_INLINED (decl);
+         DECL_POSSIBLY_INLINED (decl) |= DECL_POSSIBLY_INLINED (n->decl);
+       }
       lto_symtab_prevail_decl (n->decl, decl);
       decl = n->decl;
     }