public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: Cary Coutant <ccoutant@google.com>
To: gcc-patches <gcc-patches@gcc.gnu.org>
Cc: Jim Wilson <wilson@codesourcery.com>, Jason Merrill <jason@redhat.com>
Subject: [patch] Fix two bugs with -gdwarf-4
Date: Tue, 22 Jun 2010 04:27:00 -0000	[thread overview]
Message-ID: <AANLkTilGVXqJAphOrs7BMfJ1YBgOVq7c5a_BuY4ncR62@mail.gmail.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 673 bytes --]

This patch fixes a couple of bugs with -gdwarf-4. One is an ICE that
occurs when there's a particular kind of local typedef in a member
function, and the other caused an unsatisfied reference at link time
caused by moving a locally-scoped type into a type unit.

Bootstrapped and tested on x86_64. OK?

-cary


gcc/ChangeLog:

        * dwarf2out.c (is_nested_in_subprogram): New function.
        (should_move_die_to_comdat): Use it.
        (copy_ancestor_tree): Don't mark DIEs here.
        (copy_decls_walk): Start walk from root of newly-added tree;
        mark DIEs here instead.

gcc/testsuite/ChangeLog:

        * g++.dg/debug/dwarf2/dwarf4-typedef.C: New test.

[-- Attachment #2: gcc-dwarf4-100621-patch.txt --]
[-- Type: text/plain, Size: 2856 bytes --]

Index: testsuite/g++.dg/debug/dwarf2/dwarf4-typedef.C
===================================================================
--- testsuite/g++.dg/debug/dwarf2/dwarf4-typedef.C	(revision 0)
+++ testsuite/g++.dg/debug/dwarf2/dwarf4-typedef.C	(revision 0)
@@ -0,0 +1,34 @@
+/* { dg-do compile } */
+/* { dg-options "-gdwarf-4" } */
+
+/* Regression test for an ICE in output_die when using -gdwarf-4.  */
+
+namespace {
+
+struct A {
+  virtual ~A();
+};
+
+struct B : public A {
+  template <typename A>
+  bool foo(A x[2]) { }
+};
+
+template <typename T>
+struct C {
+  T v[2];
+};
+
+template <typename T>
+bool X(T &b) {
+  typedef C<int> D;
+  D x[2];
+  return b.foo(x);
+}
+
+void f() {
+  B b;
+  X<B>(b);
+}
+
+}
Index: dwarf2out.c
===================================================================
--- dwarf2out.c	(revision 161107)
+++ dwarf2out.c	(working copy)
@@ -9630,6 +9630,18 @@ is_declaration_die (dw_die_ref die)
   return 0;
 }
 
+/* Return non-zero if this DIE is nested inside a subprogram.  */
+
+static int
+is_nested_in_subprogram (dw_die_ref die)
+{
+  dw_die_ref decl = get_AT_ref (die, DW_AT_specification);
+
+  if (decl == NULL)
+    decl = die;
+  return local_scope_p (decl);
+}
+
 /* Return non-zero if this is a type DIE that should be moved to a
    COMDAT .debug_types section.  */
 
@@ -9642,8 +9654,11 @@ should_move_die_to_comdat (dw_die_ref di
     case DW_TAG_structure_type:
     case DW_TAG_enumeration_type:
     case DW_TAG_union_type:
-      /* Don't move declarations or inlined instances.  */
-      if (is_declaration_die (die) || get_AT (die, DW_AT_abstract_origin))
+      /* Don't move declarations, inlined instances, or types nested in a
+	 subprogram.  */
+      if (is_declaration_die (die)
+          || get_AT (die, DW_AT_abstract_origin)
+          || is_nested_in_subprogram (die))
         return 0;
       return 1;
     case DW_TAG_array_type:
@@ -10055,8 +10070,6 @@ copy_ancestor_tree (dw_die_ref unit, dw_
 
   if (decl_table != NULL)
     {
-      /* Make sure the copy is marked as part of the type unit.  */
-      copy->die_mark = 1;
       /* Record the pointer to the copy.  */
       entry->copy = copy;
     }
@@ -10130,7 +10143,18 @@ copy_decls_walk (dw_die_ref unit, dw_die
                  installed in a previously-added context, it won't
                  get visited otherwise.  */
               if (parent != unit)
-                copy_decls_walk (unit, parent, decl_table);
+		{
+		  /* Find the highest point of the newly-added tree,
+		     mark each node along the way, and walk from there.  */
+		  parent->die_mark = 1;
+		  while (parent->die_parent
+		  	 && parent->die_parent->die_mark == 0)
+		    {
+		      parent = parent->die_parent;
+		      parent->die_mark = 1;
+		    }
+		  copy_decls_walk (unit, parent, decl_table);
+		}
             }
         }
     }

             reply	other threads:[~2010-06-22  0:07 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-06-22  4:27 Cary Coutant [this message]
2010-06-22  7:03 ` Jason Merrill
2010-06-22 18:26   ` Cary Coutant

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=AANLkTilGVXqJAphOrs7BMfJ1YBgOVq7c5a_BuY4ncR62@mail.gmail.com \
    --to=ccoutant@google.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=jason@redhat.com \
    --cc=wilson@codesourcery.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).