Re: RFA (cgraph): C++ 'structor decloning patch, Mark III

[Jason Merrill <jason at redhat dot com>]

On 12/12/2013 03:08 PM, Jan Hubicka wrote:
> So only reason why this is optimize_size only is the fact that we can't rely on inliner
> to fix up the wrappers?

I was just being conservative.  In fact, the inliner seems to handle 
small [cd]tors well, inlining them into the wrappers and then into 
callers.  Do you think I should enable it more generally?

> In longer term, it would be nice to disable frontend driven clonning and just
> produce the wrappers/virtual clones to be handled by middle-end.

Agreed.  The only snag is handling variadic functions.

How does this look?

Jason

commit 71e2303c3b09b284551f149fc266a00a09b8de0f
Author: Jason Merrill <jason@redhat.com>
Date:   Thu Jan 12 14:04:42 2012 -0500

    	PR c++/41090
    	Add -fdeclone-ctor-dtor.
    gcc/cp/
    	* optimize.c (can_alias_cdtor, populate_clone_array): Split out
    	from maybe_clone_body.
    	(maybe_thunk_body): New function.
    	(maybe_clone_body): Call it.
    	* mangle.c (write_mangled_name): Remove code to suppress
    	writing of mangled name for cloned constructor or destructor.
    	(write_special_name_constructor): Handle decloned constructor.
    	(write_special_name_destructor): Handle decloned destructor.
    	* method.c (trivial_fn_p): Handle decloning.
    	* semantics.c (expand_or_defer_fn_1): Clone after setting linkage.
    gcc/c-family/
    	* c.opt: Add -fdeclone-ctor-dtor.
    	* c-opts.c (c_common_post_options): Default to on iff -Os.
    gcc/
    	* cgraph.h (struct cgraph_node): Add calls_comdat_local.
    	(symtab_comdat_local_p, symtab_in_same_comdat_p): New.
    	* cif-code.def: Add USES_COMDAT_LOCAL.
    	* symtab.c (verify_symtab_base): Make sure we don't refer to a
    	comdat-local symbol from outside its comdat.
    	* cgraph.c (verify_cgraph_node): Likewise.
    	* cgraphunit.c (mark_functions_to_output): Don't mark comdat-locals.
    	* ipa.c (symtab_remove_unreachable_nodes): Likewise.
    	(function_and_variable_visibility): Handle comdat-local fns.
    	* ipa-cp.c (decide_about_value): Don't clone comdat-locals.
    	* ipa-inline-analysis.c (compute_inline_parameters): Update
    	calls_comdat_local.
    	* ipa-inline-transform.c (inline_call): Likewise.
    	* ipa-inline.c (can_inline_edge_p): Check calls_comdat_local.
    	* lto-cgraph.c (input_overwrite_node): Read calls_comdat_local.
    	(lto_output_node): Write it.
    	* symtab.c (symtab_dissolve_same_comdat_group_list): Clear
    	DECL_COMDAT_GROUP.
    include/
    	* demangle.h (enum gnu_v3_ctor_kinds):
    	Added literal gnu_v3_unified_ctor.
    	(enum gnu_v3_ctor_kinds):
    	Added literal gnu_v3_unified_dtor.
    libiberty/
    	* cp-demangle.c (cplus_demangle_fill_ctor,cplus_demangle_fill_dtor):
    	Handle unified ctor/dtor.
    	(d_ctor_dtor_name): Handle unified ctor/dtor.

diff --git a/gcc/c-family/c-opts.c b/gcc/c-family/c-opts.c
index f368cab..3576f7d 100644
--- a/gcc/c-family/c-opts.c
+++ b/gcc/c-family/c-opts.c
@@ -899,6 +899,10 @@ c_common_post_options (const char **pfilename)
   if (warn_implicit_function_declaration == -1)
     warn_implicit_function_declaration = flag_isoc99;
 
+  /* Declone C++ 'structors if -Os.  */
+  if (flag_declone_ctor_dtor == -1)
+    flag_declone_ctor_dtor = optimize_size;
+
   if (cxx_dialect >= cxx11)
     {
       /* If we're allowing C++0x constructs, don't warn about C++98
diff --git a/gcc/c-family/c.opt b/gcc/c-family/c.opt
index bfca1e0..d270f77 100644
--- a/gcc/c-family/c.opt
+++ b/gcc/c-family/c.opt
@@ -890,6 +890,10 @@ fdeduce-init-list
 C++ ObjC++ Var(flag_deduce_init_list) Init(0)
 -fdeduce-init-list	enable deduction of std::initializer_list for a template type parameter from a brace-enclosed initializer-list
 
+fdeclone-ctor-dtor
+C++ ObjC++ Var(flag_declone_ctor_dtor) Init(-1)
+Factor complex constructors and destructors to favor space over speed
+
 fdefault-inline
 C++ ObjC++ Ignore
 Does nothing.  Preserved for backward compatibility.
diff --git a/gcc/cgraph.c b/gcc/cgraph.c
index 9501afa..ccd150c 100644
--- a/gcc/cgraph.c
+++ b/gcc/cgraph.c
@@ -2666,10 +2666,18 @@ verify_cgraph_node (struct cgraph_node *node)
 	  error_found = true;
 	}
     }
+  bool check_comdat = symtab_comdat_local_p (node);
   for (e = node->callers; e; e = e->next_caller)
     {
       if (verify_edge_count_and_frequency (e))
 	error_found = true;
+      if (check_comdat
+	  && !symtab_in_same_comdat_p (e->caller, node))
+	{
+	  error ("comdat-local function called by %s outside its comdat",
+		 identifier_to_locale (e->caller->name ()));
+	  error_found = true;
+	}
       if (!e->inline_failed)
 	{
 	  if (node->global.inlined_to
diff --git a/gcc/cgraph.h b/gcc/cgraph.h
index 0a88da3..69b97a7 100644
--- a/gcc/cgraph.h
+++ b/gcc/cgraph.h
@@ -428,6 +428,9 @@ public:
   unsigned tm_clone : 1;
   /* True if this decl is a dispatcher for function versions.  */
   unsigned dispatcher_function : 1;
+  /* True if this decl calls a COMDAT-local function.  This is set up in
+     compute_inline_parameters and inline_call.  */
+  unsigned calls_comdat_local : 1;
 };
 
 
@@ -1490,4 +1493,22 @@ symtab_can_be_discarded (symtab_node *node)
 	      && node->resolution != LDPR_PREVAILING_DEF_IRONLY
 	      && node->resolution != LDPR_PREVAILING_DEF_IRONLY_EXP));
 }
+
+/* Return true if NODE is local to a particular COMDAT group, and must not
+   be named from outside the COMDAT.  This is used for C++ decloned
+   constructors.  */
+
+static inline bool
+symtab_comdat_local_p (symtab_node *node)
+{
+  return (node->same_comdat_group && !TREE_PUBLIC (node->decl));
+}
+
+/* Return true if ONE and TWO are part of the same COMDAT group.  */
+
+static inline bool
+symtab_in_same_comdat_p (symtab_node *one, symtab_node *two)
+{
+  return DECL_COMDAT_GROUP (one->decl) == DECL_COMDAT_GROUP (two->decl);
+}
 #endif  /* GCC_CGRAPH_H  */
diff --git a/gcc/cgraphunit.c b/gcc/cgraphunit.c
index 44f3afd..637979e 100644
--- a/gcc/cgraphunit.c
+++ b/gcc/cgraphunit.c
@@ -1244,7 +1244,8 @@ mark_functions_to_output (void)
 	      for (next = cgraph (node->same_comdat_group);
 		   next != node;
 		   next = cgraph (next->same_comdat_group))
-		if (!next->thunk.thunk_p && !next->alias)
+		if (!next->thunk.thunk_p && !next->alias
+		    && !symtab_comdat_local_p (next))
 		  next->process = 1;
 	    }
 	}
diff --git a/gcc/cif-code.def b/gcc/cif-code.def
index e71123d..be4af6a 100644
--- a/gcc/cif-code.def
+++ b/gcc/cif-code.def
@@ -106,3 +106,6 @@ DEFCIFCODE(TARGET_OPTION_MISMATCH, N_("target specific option mismatch"))
 
 /* We can't inline because of mismatched optimization levels.  */
 DEFCIFCODE(OPTIMIZATION_MISMATCH, N_("optimization level attribute mismatch"))
+
+/* We can't inline because the callee refers to comdat-local symbols.  */
+DEFCIFCODE(USES_COMDAT_LOCAL, N_("callee refers to comdat-local symbols"))
diff --git a/gcc/cp/decl.c b/gcc/cp/decl.c
index 5d06b37..a184dcc 100644
--- a/gcc/cp/decl.c
+++ b/gcc/cp/decl.c
@@ -10185,7 +10185,9 @@ grokdeclarator (const cp_declarator *declarator,
 	    /* The TYPE_DECL is "abstract" because there will be
 	       clones of this constructor/destructor, and there will
 	       be copies of this TYPE_DECL generated in those
-	       clones.  */
+	       clones.  The decloning optimization (for space) may
+               revert this subsequently if it determines that
+               the clones should share a common implementation.  */
 	    DECL_ABSTRACT (decl) = 1;
 	}
       else if (current_class_type
diff --git a/gcc/cp/mangle.c b/gcc/cp/mangle.c
index 8a24d6c..d99062d 100644
--- a/gcc/cp/mangle.c
+++ b/gcc/cp/mangle.c
@@ -689,13 +689,6 @@ write_mangled_name (const tree decl, bool top_level)
     mangled_name:;
       write_string ("_Z");
       write_encoding (decl);
-      if (DECL_LANG_SPECIFIC (decl)
-	  && (DECL_MAYBE_IN_CHARGE_DESTRUCTOR_P (decl)
-	      || DECL_MAYBE_IN_CHARGE_CONSTRUCTOR_P (decl)))
-	/* We need a distinct mangled name for these entities, but
-	   we should never actually output it.  So, we append some
-	   characters the assembler won't like.  */
-	write_string (" *INTERNAL* ");
     }
 }
 
@@ -1653,25 +1646,21 @@ write_identifier (const char *identifier)
 		    ::= C2   # base object constructor
 		    ::= C3   # complete object allocating constructor
 
-   Currently, allocating constructors are never used.
-
-   We also need to provide mangled names for the maybe-in-charge
-   constructor, so we treat it here too.  mangle_decl_string will
-   append *INTERNAL* to that, to make sure we never emit it.  */
+   Currently, allocating constructors are never used.  */
 
 static void
 write_special_name_constructor (const tree ctor)
 {
   if (DECL_BASE_CONSTRUCTOR_P (ctor))
     write_string ("C2");
+  /* This is the old-style "[unified]" constructor.
+     In some cases, we may emit this function and call
+     it from the clones in order to share code and save space.  */
+  else if (DECL_MAYBE_IN_CHARGE_CONSTRUCTOR_P (ctor))
+    write_string ("C4");
   else
     {
-      gcc_assert (DECL_COMPLETE_CONSTRUCTOR_P (ctor)
-		  /* Even though we don't ever emit a definition of
-		     the old-style destructor, we still have to
-		     consider entities (like static variables) nested
-		     inside it.  */
-		  || DECL_MAYBE_IN_CHARGE_CONSTRUCTOR_P (ctor));
+      gcc_assert (DECL_COMPLETE_CONSTRUCTOR_P (ctor));
       write_string ("C1");
     }
 }
@@ -1681,11 +1670,7 @@ write_special_name_constructor (const tree ctor)
 
      <special-name> ::= D0 # deleting (in-charge) destructor
 		    ::= D1 # complete object (in-charge) destructor
-		    ::= D2 # base object (not-in-charge) destructor
-
-   We also need to provide mangled names for the maybe-incharge
-   destructor, so we treat it here too.  mangle_decl_string will
-   append *INTERNAL* to that, to make sure we never emit it.  */
+		    ::= D2 # base object (not-in-charge) destructor  */
 
 static void
 write_special_name_destructor (const tree dtor)
@@ -1694,14 +1679,14 @@ write_special_name_destructor (const tree dtor)
     write_string ("D0");
   else if (DECL_BASE_DESTRUCTOR_P (dtor))
     write_string ("D2");
+  else if (DECL_MAYBE_IN_CHARGE_DESTRUCTOR_P (dtor))
+    /* This is the old-style "[unified]" destructor.
+       In some cases, we may emit this function and call
+       it from the clones in order to share code and save space.  */
+    write_string ("D4");
   else
     {
-      gcc_assert (DECL_COMPLETE_DESTRUCTOR_P (dtor)
-		  /* Even though we don't ever emit a definition of
-		     the old-style destructor, we still have to
-		     consider entities (like static variables) nested
-		     inside it.  */
-		  || DECL_MAYBE_IN_CHARGE_DESTRUCTOR_P (dtor));
+      gcc_assert (DECL_COMPLETE_DESTRUCTOR_P (dtor));
       write_string ("D1");
     }
 }
diff --git a/gcc/cp/method.c b/gcc/cp/method.c
index 7405365..e79a922 100644
--- a/gcc/cp/method.c
+++ b/gcc/cp/method.c
@@ -477,7 +477,8 @@ trivial_fn_p (tree fn)
     return false;
 
   /* If fn is a clone, get the primary variant.  */
-  fn = DECL_ORIGIN (fn);
+  if (tree prim = DECL_CLONED_FUNCTION (fn))
+    fn = prim;
   return type_has_trivial_fn (DECL_CONTEXT (fn), special_function_p (fn));
 }
 
diff --git a/gcc/cp/optimize.c b/gcc/cp/optimize.c
index f1b09bf..40494f2 100644
--- a/gcc/cp/optimize.c
+++ b/gcc/cp/optimize.c
@@ -193,30 +193,40 @@ cdtor_comdat_group (tree complete, tree base)
   return get_identifier (grp_name);
 }
 
-/* FN is a function that has a complete body.  Clone the body as
-   necessary.  Returns nonzero if there's no longer any need to
-   process the main body.  */
+/* Returns true iff we can make the base and complete [cd]tor aliases of
+   the same symbol rather than separate functions.  */
 
-bool
-maybe_clone_body (tree fn)
+static bool
+can_alias_cdtor (tree fn)
+{
+#ifndef ASM_OUTPUT_DEF
+  /* If aliases aren't supported by the assembler, fail.  */
+  return false;
+#endif
+  /* We can't use an alias if there are virtual bases.  */
+  if (CLASSTYPE_VBASECLASSES (DECL_CONTEXT (fn)))
+    return false;
+  /* ??? Why not use aliases with -frepo?  */
+  if (flag_use_repository)
+    return false;
+  gcc_assert (DECL_MAYBE_IN_CHARGE_CONSTRUCTOR_P (fn)
+	      || DECL_MAYBE_IN_CHARGE_DESTRUCTOR_P (fn));
+  /* Don't use aliases for weak/linkonce definitions unless we can put both
+     symbols in the same COMDAT group.  */
+  return (DECL_INTERFACE_KNOWN (fn)
+	  && (SUPPORTS_ONE_ONLY || !DECL_WEAK (fn))
+	  && (!DECL_ONE_ONLY (fn)
+	      || (HAVE_COMDAT_GROUP && DECL_WEAK (fn))));
+}
+
+/* FN is a [cd]tor, fns is a pointer to an array of length 3.  Fill fns
+   with pointers to the base, complete, and deleting variants.  */
+
+static void
+populate_clone_array (tree fn, tree *fns)
 {
-  tree comdat_group = NULL_TREE;
   tree clone;
-  tree fns[3];
-  bool first = true;
-  bool in_charge_parm_used;
-  int idx;
-  bool need_alias = false;
 
-  /* We only clone constructors and destructors.  */
-  if (!DECL_MAYBE_IN_CHARGE_CONSTRUCTOR_P (fn)
-      && !DECL_MAYBE_IN_CHARGE_DESTRUCTOR_P (fn))
-    return 0;
-
-  /* Emit the DWARF1 abstract instance.  */
-  (*debug_hooks->deferred_inline_function) (fn);
-
-  in_charge_parm_used = CLASSTYPE_VBASECLASSES (DECL_CONTEXT (fn)) != NULL;
   fns[0] = NULL_TREE;
   fns[1] = NULL_TREE;
   fns[2] = NULL_TREE;
@@ -234,6 +244,206 @@ maybe_clone_body (tree fn)
       fns[2] = clone;
     else
       gcc_unreachable ();
+}
+
+/* FN is a constructor or destructor, and there are FUNCTION_DECLs
+   cloned from it nearby.  Instead of cloning this body, leave it
+   alone and create tiny one-call bodies for the cloned
+   FUNCTION_DECLs.  These clones are sibcall candidates, and their
+   resulting code will be very thunk-esque.  */
+
+static bool
+maybe_thunk_body (tree fn, bool force)
+{
+  tree bind, block, call, clone, clone_result, fn_parm, fn_parm_typelist;
+  tree last_arg, modify, *args;
+  int parmno, vtt_parmno, max_parms;
+  tree fns[3];
+
+  if (!force && !flag_declone_ctor_dtor)
+    return 0;
+
+  /* If function accepts variable arguments, give up.  */
+  last_arg = tree_last (TYPE_ARG_TYPES (TREE_TYPE (fn)));
+  if (last_arg != void_list_node)
+    return 0;
+
+  /* If we got this far, we've decided to turn the clones into thunks.  */
+
+  /* We're going to generate code for fn, so it is no longer "abstract."
+     Also make the unified ctor/dtor private to either the translation unit
+     (for non-vague linkage ctors) or the COMDAT group (otherwise).  */
+
+  populate_clone_array (fn, fns);
+  DECL_ABSTRACT (fn) = false;
+  if (!DECL_WEAK (fn))
+    {
+      TREE_PUBLIC (fn) = false;
+      DECL_EXTERNAL (fn) = false;
+      DECL_INTERFACE_KNOWN (fn) = true;
+    }
+  else if (HAVE_COMDAT_GROUP)
+    {
+      tree comdat_group = cdtor_comdat_group (fns[1], fns[0]);
+      DECL_COMDAT_GROUP (fns[0]) = comdat_group;
+      symtab_add_to_same_comdat_group (cgraph_get_create_node (fns[1]),
+				       cgraph_get_create_node (fns[0]));
+      symtab_add_to_same_comdat_group (symtab_get_node (fn),
+				       symtab_get_node (fns[0]));
+      if (fns[2])
+	/* If *[CD][12]* dtors go into the *[CD]5* comdat group and dtor is
+	   virtual, it goes into the same comdat group as well.  */
+	symtab_add_to_same_comdat_group (cgraph_get_create_node (fns[2]),
+					 symtab_get_node (fns[0]));
+      TREE_PUBLIC (fn) = false;
+      DECL_EXTERNAL (fn) = false;
+      DECL_INTERFACE_KNOWN (fn) = true;
+      /* function_and_variable_visibility doesn't want !PUBLIC decls to
+	 have these flags set.  */
+      DECL_WEAK (fn) = false;
+      DECL_COMDAT (fn) = false;
+    }
+
+  /* Find the vtt_parm, if present.  */
+  for (vtt_parmno = -1, parmno = 0, fn_parm = DECL_ARGUMENTS (fn);
+       fn_parm;
+       ++parmno, fn_parm = TREE_CHAIN (fn_parm))
+    {
+      if (DECL_ARTIFICIAL (fn_parm)
+	  && DECL_NAME (fn_parm) == vtt_parm_identifier)
+	{
+	  /* Compensate for removed in_charge parameter.  */
+	  vtt_parmno = parmno;
+	  break;
+	}
+    }
+
+  /* Allocate an argument buffer for build_cxx_call().
+     Make sure it is large enough for any of the clones.  */
+  max_parms = 0;
+  FOR_EACH_CLONE (clone, fn)
+    {
+      int length = list_length (DECL_ARGUMENTS (fn));
+      if (length > max_parms)
+        max_parms = length;
+    }
+  args = (tree *) alloca (max_parms * sizeof (tree));
+
+  /* We know that any clones immediately follow FN in TYPE_METHODS.  */
+  FOR_EACH_CLONE (clone, fn)
+    {
+      tree clone_parm;
+
+      /* If we've already generated a body for this clone, avoid
+	 duplicating it.  (Is it possible for a clone-list to grow after we
+	 first see it?)  */
+      if (DECL_SAVED_TREE (clone) || TREE_ASM_WRITTEN (clone))
+	continue;
+
+      /* Start processing the function.  */
+      start_preparsed_function (clone, NULL_TREE, SF_PRE_PARSED);
+
+      if (clone == fns[2])
+	{
+	  for (clone_parm = DECL_ARGUMENTS (clone); clone_parm;
+	       clone_parm = TREE_CHAIN (clone_parm))
+	    DECL_ABSTRACT_ORIGIN (clone_parm) = NULL_TREE;
+	  /* Build the delete destructor by calling complete destructor and
+	     delete function.  */
+	  build_delete_destructor_body (clone, fns[1]);
+	}
+      else
+	{
+	  /* Walk parameter lists together, creating parameter list for
+	     call to original function.  */
+	  for (parmno = 0,
+		 fn_parm = DECL_ARGUMENTS (fn),
+		 fn_parm_typelist = TYPE_ARG_TYPES (TREE_TYPE (fn)),
+		 clone_parm = DECL_ARGUMENTS (clone);
+	       fn_parm;
+	       ++parmno,
+		 fn_parm = TREE_CHAIN (fn_parm))
+	    {
+	      if (parmno == vtt_parmno && ! DECL_HAS_VTT_PARM_P (clone))
+		{
+		  gcc_assert (fn_parm_typelist);
+		  /* Clobber argument with formal parameter type.  */
+		  args[parmno]
+		    = convert (TREE_VALUE (fn_parm_typelist),
+			       null_pointer_node);
+		}
+	      else if (parmno == 1 && DECL_HAS_IN_CHARGE_PARM_P (fn))
+		{
+		  tree in_charge
+		    = copy_node (in_charge_arg_for_name (DECL_NAME (clone)));
+		  args[parmno] = in_charge;
+		}
+	      /* Map other parameters to their equivalents in the cloned
+		 function.  */
+	      else
+		{
+		  gcc_assert (clone_parm);
+		  DECL_ABSTRACT_ORIGIN (clone_parm) = NULL;
+		  args[parmno] = clone_parm;
+		  clone_parm = TREE_CHAIN (clone_parm);
+		}
+	      if (fn_parm_typelist)
+		fn_parm_typelist = TREE_CHAIN (fn_parm_typelist);
+	    }
+
+	  /* We built this list backwards; fix now.  */
+	  mark_used (fn);
+	  call = build_cxx_call (fn, parmno, args, tf_warning_or_error);
+	  /* Arguments passed to the thunk by invisible reference should
+	     be transmitted to the callee unchanged.  Do not create a
+	     temporary and invoke the copy constructor.  The thunking
+	     transformation must not introduce any constructor calls.  */
+	  CALL_FROM_THUNK_P (call) = 1;
+	  block = make_node (BLOCK);
+	  if (targetm.cxx.cdtor_returns_this ())
+	    {
+	      clone_result = DECL_RESULT (clone);
+	      modify = build2 (MODIFY_EXPR, TREE_TYPE (clone_result),
+			       clone_result, call);
+	      add_stmt (modify);
+	      BLOCK_VARS (block) = clone_result;
+	    }
+	  else
+	    {
+	      add_stmt (call);
+	    }
+	  bind = c_build_bind_expr (DECL_SOURCE_LOCATION (clone),
+				    block, cur_stmt_list);
+	  DECL_SAVED_TREE (clone) = push_stmt_list ();
+	  add_stmt (bind);
+	}
+
+      DECL_ABSTRACT_ORIGIN (clone) = NULL;
+      expand_or_defer_fn (finish_function (0));
+    }
+  return 1;
+}
+
+/* FN is a function that has a complete body.  Clone the body as
+   necessary.  Returns nonzero if there's no longer any need to
+   process the main body.  */
+
+bool
+maybe_clone_body (tree fn)
+{
+  tree comdat_group = NULL_TREE;
+  tree clone;
+  tree fns[3];
+  bool first = true;
+  int idx;
+  bool need_alias = false;
+
+  /* We only clone constructors and destructors.  */
+  if (!DECL_MAYBE_IN_CHARGE_CONSTRUCTOR_P (fn)
+      && !DECL_MAYBE_IN_CHARGE_DESTRUCTOR_P (fn))
+    return 0;
+
+  populate_clone_array (fn, fns);
 
   /* Remember if we can't have multiple clones for some reason.  We need to
      check this before we remap local static initializers in clone_body.  */
@@ -247,9 +457,6 @@ maybe_clone_body (tree fn)
     {
       tree parm;
       tree clone_parm;
-      int parmno;
-      bool alias = false;
-      struct pointer_map_t *decl_map;
 
       clone = fns[idx];
       if (!clone)
@@ -296,26 +503,44 @@ maybe_clone_body (tree fn)
 	   parm = DECL_CHAIN (parm), clone_parm = DECL_CHAIN (clone_parm))
 	/* Update this parameter.  */
 	update_cloned_parm (parm, clone_parm, first);
+    }
+
+  bool can_alias = can_alias_cdtor (fn);
+
+  /* If we decide to turn clones into thunks, they will branch to fn.
+     Must have original function available to call.  */
+  if (!can_alias && maybe_thunk_body (fn, need_alias))
+    {
+      pop_from_top_level ();
+      /* We still need to emit the original function.  */
+      return 0;
+    }
+
+  /* Emit the DWARF1 abstract instance.  */
+  (*debug_hooks->deferred_inline_function) (fn);
+
+  /* We know that any clones immediately follow FN in the TYPE_METHODS list. */
+  for (idx = 0; idx < 3; idx++)
+    {
+      tree parm;
+      tree clone_parm;
+      int parmno;
+      struct pointer_map_t *decl_map;
+      bool alias = false;
+
+      clone = fns[idx];
+      if (!clone)
+	continue;
 
       /* Start processing the function.  */
       start_preparsed_function (clone, NULL_TREE, SF_PRE_PARSED);
 
       /* Tell cgraph if both ctors or both dtors are known to have
 	 the same body.  */
-      if (!in_charge_parm_used
+      if (can_alias
 	  && fns[0]
 	  && idx == 1
-	  && !flag_use_repository
-	  && DECL_INTERFACE_KNOWN (fns[0])
-	  && (SUPPORTS_ONE_ONLY || !DECL_WEAK (fns[0]))
-	  && (!DECL_ONE_ONLY (fns[0])
-	      || (HAVE_COMDAT_GROUP
-		  && DECL_WEAK (fns[0])))
-	  && !flag_syntax_only
-	  /* Set linkage flags appropriately before
-	     cgraph_create_function_alias looks at them.  */
-	  && expand_or_defer_fn_1 (clone)
-	  && cgraph_same_body_alias (cgraph_get_node (fns[0]),
+	  && cgraph_same_body_alias (cgraph_get_create_node (fns[0]),
 				     clone, fns[0]))
 	{
 	  alias = true;
diff --git a/gcc/cp/semantics.c b/gcc/cp/semantics.c
index 63f50fb..7be3541 100644
--- a/gcc/cp/semantics.c
+++ b/gcc/cp/semantics.c
@@ -3901,20 +3901,6 @@ expand_or_defer_fn_1 (tree fn)
 
   gcc_assert (DECL_SAVED_TREE (fn));
 
-  /* If this is a constructor or destructor body, we have to clone
-     it.  */
-  if (maybe_clone_body (fn))
-    {
-      /* We don't want to process FN again, so pretend we've written
-	 it out, even though we haven't.  */
-      TREE_ASM_WRITTEN (fn) = 1;
-      /* If this is an instantiation of a constexpr function, keep
-	 DECL_SAVED_TREE for explain_invalid_constexpr_fn.  */
-      if (!is_instantiation_of_constexpr (fn))
-	DECL_SAVED_TREE (fn) = NULL_TREE;
-      return false;
-    }
-
   /* We make a decision about linkage for these functions at the end
      of the compilation.  Until that point, we do not want the back
      end to output them -- but we do want it to see the bodies of
@@ -3962,6 +3948,20 @@ expand_or_defer_fn_1 (tree fn)
 	}
     }
 
+  /* If this is a constructor or destructor body, we have to clone
+     it.  */
+  if (maybe_clone_body (fn))
+    {
+      /* We don't want to process FN again, so pretend we've written
+	 it out, even though we haven't.  */
+      TREE_ASM_WRITTEN (fn) = 1;
+      /* If this is an instantiation of a constexpr function, keep
+	 DECL_SAVED_TREE for explain_invalid_constexpr_fn.  */
+      if (!is_instantiation_of_constexpr (fn))
+	DECL_SAVED_TREE (fn) = NULL_TREE;
+      return false;
+    }
+
   /* There's no reason to do any of the work here if we're only doing
      semantic analysis; this code just generates RTL.  */
   if (flag_syntax_only)
diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
index b655a64..678c352 100644
--- a/gcc/doc/invoke.texi
+++ b/gcc/doc/invoke.texi
@@ -7338,6 +7338,18 @@ branch-less equivalents.
 
 Enabled at levels @option{-O}, @option{-O2}, @option{-O3}, @option{-Os}.
 
+@item -fdeclone-ctor-dtor
+@opindex fdeclone-ctor-dtor
+The C++ ABI requires multiple entry points for constructors and
+destructors: one for a base subobject, one for a complete object, and
+one for a virtual destructor that calls operator delete afterwards.
+For a hierarchy with virtual bases, the base and complete variants are
+clones, which means two copies of the function.  With this option, the
+base and complete variants are changed to be thunks that call a common
+implementation.
+
+Enabled by @option{-Os}.
+
 @item -fdelete-null-pointer-checks
 @opindex fdelete-null-pointer-checks
 Assume that programs cannot safely dereference null pointers, and that
diff --git a/gcc/ipa-cp.c b/gcc/ipa-cp.c
index 305ad2d..9c1b0af 100644
--- a/gcc/ipa-cp.c
+++ b/gcc/ipa-cp.c
@@ -3326,6 +3326,11 @@ decide_about_value (struct cgraph_node *node, int index, HOST_WIDE_INT offset,
 					    &caller_count))
     return false;
 
+  /* Don't clone decls local to a comdat group; for C++ decloned
+     constructors, inlining is always better.  */
+  if (symtab_comdat_local_p (node))
+    return false;
+
   if (dump_file && (dump_flags & TDF_DETAILS))
     {
       fprintf (dump_file, " - considering value ");
diff --git a/gcc/ipa-inline-analysis.c b/gcc/ipa-inline-analysis.c
index 9e9087f..21e52a1 100644
--- a/gcc/ipa-inline-analysis.c
+++ b/gcc/ipa-inline-analysis.c
@@ -2796,6 +2796,11 @@ compute_inline_parameters (struct cgraph_node *node, bool early)
     }
   estimate_function_body_sizes (node, early);
 
+  for (e = node->callees; e; e = e->next_callee)
+    if (symtab_comdat_local_p (e->callee))
+      break;
+  node->calls_comdat_local = (e != NULL);
+
   /* Inlining characteristics are maintained by the cgraph_mark_inline.  */
   info->time = info->self_time;
   info->size = info->self_size;
diff --git a/gcc/ipa-inline-transform.c b/gcc/ipa-inline-transform.c
index 7fb4ab9..87c2479 100644
--- a/gcc/ipa-inline-transform.c
+++ b/gcc/ipa-inline-transform.c
@@ -272,6 +272,18 @@ inline_call (struct cgraph_edge *e, bool update_original,
    inline_update_overall_summary (to);
   new_size = inline_summary (to)->size;
 
+  if (callee->calls_comdat_local)
+    to->calls_comdat_local = true;
+  else if (to->calls_comdat_local && symtab_comdat_local_p (callee))
+    {
+      struct cgraph_edge *se = to->callees;
+      for (; se; se = se->next_callee)
+	if (se->inline_failed && symtab_comdat_local_p (se->callee))
+	  break;
+      if (se == NULL)
+	to->calls_comdat_local = false;
+    }
+
 #ifdef ENABLE_CHECKING
   /* Verify that estimated growth match real growth.  Allow off-by-one
      error due to INLINE_SIZE_SCALE roudoff errors.  */
diff --git a/gcc/ipa-inline.c b/gcc/ipa-inline.c
index 38157ca..9d112c9 100644
--- a/gcc/ipa-inline.c
+++ b/gcc/ipa-inline.c
@@ -241,7 +241,7 @@ report_inline_failed_reason (struct cgraph_edge *e)
 
    if REPORT is true, output reason to the dump file.  
 
-   if DISREGARD_LIMITES is true, ignore size limits.*/
+   if DISREGARD_LIMITS is true, ignore size limits.*/
 
 static bool
 can_inline_edge_p (struct cgraph_edge *e, bool report,
@@ -271,6 +271,12 @@ can_inline_edge_p (struct cgraph_edge *e, bool report,
       e->inline_failed = CIF_BODY_NOT_AVAILABLE;
       inlinable = false;
     }
+  else if (callee->calls_comdat_local
+	   && !symtab_in_same_comdat_p (e->caller, callee))
+    {
+      e->inline_failed = CIF_USES_COMDAT_LOCAL;
+      inlinable = false;
+    }
   else if (!inline_summary (callee)->inlinable 
 	   || (caller_cfun && fn_contains_cilk_spawn_p (caller_cfun)))
     {
diff --git a/gcc/ipa.c b/gcc/ipa.c
index 1ec4b5f..3c19288 100644
--- a/gcc/ipa.c
+++ b/gcc/ipa.c
@@ -362,14 +362,17 @@ symtab_remove_unreachable_nodes (bool before_inlining_p, FILE *file)
 	      enqueue_node (origin_node, &first, reachable);
 	    }
 	  /* If any symbol in a comdat group is reachable, force
-	     all other in the same comdat group to be also reachable.  */
+	     all externally visible symbols in the same comdat
+	     group to be reachable as well.  Comdat-local symbols
+	     can be discarded if all uses were inlined.  */
 	  if (node->same_comdat_group)
 	    {
 	      symtab_node *next;
 	      for (next = node->same_comdat_group;
 		   next != node;
 		   next = next->same_comdat_group)
-		if (!pointer_set_insert (reachable, next))
+		if (!symtab_comdat_local_p (next)
+		    && !pointer_set_insert (reachable, next))
 		  enqueue_node (next, &first, reachable);
 	    }
 	  /* Mark references as reachable.  */
@@ -969,14 +972,14 @@ function_and_variable_visibility (bool whole_program)
 	  node->unique_name = ((node->resolution == LDPR_PREVAILING_DEF_IRONLY
 				      || node->resolution == LDPR_PREVAILING_DEF_IRONLY_EXP)
 				      && TREE_PUBLIC (node->decl));
-	  symtab_make_decl_local (node->decl);
 	  node->resolution = LDPR_PREVAILING_DEF_IRONLY;
-	  if (node->same_comdat_group)
+	  if (node->same_comdat_group && TREE_PUBLIC (node->decl))
 	    /* cgraph_externally_visible_p has already checked all other nodes
 	       in the group and they will all be made local.  We need to
 	       dissolve the group at once so that the predicate does not
 	       segfault though. */
 	    symtab_dissolve_same_comdat_group_list (node);
+	  symtab_make_decl_local (node->decl);
 	}
 
       if (node->thunk.thunk_p
diff --git a/gcc/lto-cgraph.c b/gcc/lto-cgraph.c
index 44cfa28..7834ed0 100644
--- a/gcc/lto-cgraph.c
+++ b/gcc/lto-cgraph.c
@@ -518,6 +518,7 @@ lto_output_node (struct lto_simple_output_block *ob, struct cgraph_node *node,
   bp_pack_value (&bp, node->only_called_at_startup, 1);
   bp_pack_value (&bp, node->only_called_at_exit, 1);
   bp_pack_value (&bp, node->tm_clone, 1);
+  bp_pack_value (&bp, node->calls_comdat_local, 1);
   bp_pack_value (&bp, node->thunk.thunk_p && !boundary_p, 1);
   bp_pack_enum (&bp, ld_plugin_symbol_resolution,
 	        LDPR_NUM_KNOWN, node->resolution);
@@ -993,6 +994,7 @@ input_overwrite_node (struct lto_file_decl_data *file_data,
   node->only_called_at_startup = bp_unpack_value (bp, 1);
   node->only_called_at_exit = bp_unpack_value (bp, 1);
   node->tm_clone = bp_unpack_value (bp, 1);
+  node->calls_comdat_local = bp_unpack_value (bp, 1);
   node->thunk.thunk_p = bp_unpack_value (bp, 1);
   node->resolution = bp_unpack_enum (bp, ld_plugin_symbol_resolution,
 				     LDPR_NUM_KNOWN);
diff --git a/gcc/symtab.c b/gcc/symtab.c
index dc700e7..a448dea 100644
--- a/gcc/symtab.c
+++ b/gcc/symtab.c
@@ -538,6 +538,7 @@ symtab_dissolve_same_comdat_group_list (symtab_node *node)
     {
       next = n->same_comdat_group;
       n->same_comdat_group = NULL;
+      DECL_COMDAT_GROUP (n->decl) = NULL_TREE;
       n = next;
     }
   while (n != node);
@@ -844,6 +845,21 @@ verify_symtab_base (symtab_node *node)
 	  n = n->same_comdat_group;
 	}
       while (n != node);
+      if (symtab_comdat_local_p (node))
+	{
+	  struct ipa_ref_list *refs = &node->ref_list;
+	  struct ipa_ref *ref;
+	  for (int i = 0; ipa_ref_list_referring_iterate (refs, i, ref); ++i)
+	    {
+	      if (!symtab_in_same_comdat_p (ref->referring, node))
+		{
+		  error ("comdat-local symbol referred to by %s outside its "
+			 "comdat",
+			 identifier_to_locale (ref->referring->name()));
+		  error_found = true;
+		}
+	    }
+	}
     }
   return error_found;
 }
@@ -911,6 +927,10 @@ symtab_make_decl_local (tree decl)
 {
   rtx rtl, symbol;
 
+  /* Avoid clearing DECL_COMDAT_GROUP on comdat-local decls.  */
+  if (TREE_PUBLIC (decl) == 0)
+    return;
+
   if (TREE_CODE (decl) == VAR_DECL)
     DECL_COMMON (decl) = 0;
   else gcc_assert (TREE_CODE (decl) == FUNCTION_DECL);
diff --git a/gcc/testsuite/g++.dg/ext/label13a.C b/gcc/testsuite/g++.dg/ext/label13a.C
new file mode 100644
index 0000000..3be8e31
--- /dev/null
+++ b/gcc/testsuite/g++.dg/ext/label13a.C
@@ -0,0 +1,25 @@
+// PR c++/41090
+// { dg-do run }
+// { dg-options "-save-temps" }
+// { dg-final { scan-assembler "_ZN1CC4Ev" } }
+// { dg-final cleanup-saved-temps }
+
+int i;
+struct A { A() {} };
+struct C: virtual A
+{
+  C();
+};
+
+C::C()
+{
+  static void *labelref = &&label;
+  goto *labelref;
+ label: i = 1;
+}
+
+int main()
+{
+  C c;
+  return (i != 1);
+}
diff --git a/include/demangle.h b/include/demangle.h
index 58bf547..bbad71b 100644
--- a/include/demangle.h
+++ b/include/demangle.h
@@ -173,6 +173,10 @@ enum gnu_v3_ctor_kinds {
   gnu_v3_complete_object_ctor = 1,
   gnu_v3_base_object_ctor,
   gnu_v3_complete_object_allocating_ctor,
+  /* These are not part of the V3 ABI.  Unified constructors are generated
+     as a speed-for-space optimization when the -fdeclone-ctor-dtor option
+     is used, and are always internal symbols.  */
+  gnu_v3_unified_ctor,
   gnu_v3_object_ctor_group
 };
 
@@ -188,6 +192,10 @@ enum gnu_v3_dtor_kinds {
   gnu_v3_deleting_dtor = 1,
   gnu_v3_complete_object_dtor,
   gnu_v3_base_object_dtor,
+  /* These are not part of the V3 ABI.  Unified destructors are generated
+     as a speed-for-space optimization when the -fdeclone-ctor-dtor option
+     is used, and are always internal symbols.  */
+  gnu_v3_unified_dtor,
   gnu_v3_object_dtor_group
 };
 
diff --git a/libiberty/cp-demangle.c b/libiberty/cp-demangle.c
index 029151e..7f51268 100644
--- a/libiberty/cp-demangle.c
+++ b/libiberty/cp-demangle.c
@@ -2107,6 +2107,9 @@ d_ctor_dtor_name (struct d_info *di)
 	  case '3':
 	    kind = gnu_v3_complete_object_allocating_ctor;
 	    break;
+          case '4':
+	    kind = gnu_v3_unified_ctor;
+	    break;
 	  case '5':
 	    kind = gnu_v3_object_ctor_group;
 	    break;
@@ -2132,6 +2135,10 @@ d_ctor_dtor_name (struct d_info *di)
 	  case '2':
 	    kind = gnu_v3_base_object_dtor;
 	    break;
+          /*  digit '3' is not used */
+	  case '4':
+	    kind = gnu_v3_unified_dtor;
+	    break;
 	  case '5':
 	    kind = gnu_v3_object_dtor_group;
 	    break;