diff mbox

DWARF: add abstract origin links on lexical blocks DIEs

Message ID 569F5796.3080309@adacore.com
State New
Headers show

Commit Message

Pierre-Marie de Rodat Jan. 20, 2016, 9:47 a.m. UTC
On 01/18/2016 10:47 AM, Pierre-Marie de Rodat wrote:
> Thank you for your inputs! I’m going to try that, then. I hope this test
> will not be too fragile…

Here it is! Re-bootstrapped and regtested successfuly on x86_64-linux. 
I’ve checked that the testcase fails on the mainline.

Comments

Pierre-Marie de Rodat April 26, 2016, 8 a.m. UTC | #1
Ping for the patch submitted at 
<https://gcc.gnu.org/ml/gcc-patches/2016-01/msg01495.html>. It applies 
just fine on the current trunk and still bootstrapps and regtests 
successfuly on x86_64-linux.

Thank you in advance,
Pierre-Marie de Rodat May 4, 2016, 2:23 p.m. UTC | #2
Ping for the patch submitted at 
<https://gcc.gnu.org/ml/gcc-patches/2016-01/msg01495.html>. It applies 
just fine on the current trunk and still bootstrapps and regtests 
successfuly on x86_64-linux.

Thank you in advance,
Richard Biener May 9, 2016, 11:01 a.m. UTC | #3
On Wed, May 4, 2016 at 4:23 PM, Pierre-Marie de Rodat
<derodat@adacore.com> wrote:
> Ping for the patch submitted at
> <https://gcc.gnu.org/ml/gcc-patches/2016-01/msg01495.html>. It applies just
> fine on the current trunk and still bootstrapps and regtests successfuly on
> x86_64-linux.
>
> Thank you in advance,

Ok and sorry for the delay.

Thanks,
Richard.

> --
> Pierre-Marie de Rodat
Pierre-Marie de Rodat May 10, 2016, 8:03 a.m. UTC | #4
On 05/09/2016 01:01 PM, Richard Biener wrote:
> Ok and sorry for the delay.

No problem. :-) Thanks, this is committed!
diff mbox

Patch

From 451d62ff871734727b0f0f570f89b6cfbed922f2 Mon Sep 17 00:00:00 2001
From: Pierre-Marie de Rodat <derodat@adacore.com>
Date: Tue, 12 Jan 2016 14:50:33 +0100
Subject: [PATCH] DWARF: add abstract origin links on lexical blocks DIEs

Track from which abstract lexical block concrete ones come from in DWARF
so that debuggers can inherit the former from the latter. This enables
debuggers to properly handle the following case:

  * function Child2 is nested in a lexical block, itself nested in
    function Child1;
  * function Child1 is inlined into some call site;
  * function Child2 is never inlined.

Here, Child2 is described in DWARF only in the abstract instance of
Child1. So when debuggers decode Child1's concrete instances, they need
to fetch the definition for Child2 in the corresponding abstract
instance: the DW_AT_abstract_origin link on the lexical block that
embeds Child1 enables them to do that.

Bootstrapped and regtested on x86_64-linux.

gcc/ChangeLog:

	* dwarf2out.c (add_abstract_origin_attribute): Adjust
	documentation comment.  For BLOCK nodes, add a
	DW_AT_abstract_origin attribute that points to the DIE generated
	for the origin BLOCK.
	(gen_lexical_block_die): Call add_abstract_origin_attribute for
	blocks from inlined functions.

gcc/testsuite/Changelog:

	* gcc.dg/debug/dwarf2/nested_fun.c: New testcase.
---
 gcc/dwarf2out.c                                | 13 ++++--
 gcc/testsuite/gcc.dg/debug/dwarf2/nested_fun.c | 65 ++++++++++++++++++++++++++
 2 files changed, 75 insertions(+), 3 deletions(-)
 create mode 100644 gcc/testsuite/gcc.dg/debug/dwarf2/nested_fun.c

diff --git a/gcc/dwarf2out.c b/gcc/dwarf2out.c
index f742900..d1503ec 100644
--- a/gcc/dwarf2out.c
+++ b/gcc/dwarf2out.c
@@ -18470,15 +18470,16 @@  add_prototyped_attribute (dw_die_ref die, tree func_type)
 }
 
 /* Add an 'abstract_origin' attribute below a given DIE.  The DIE is found
-   by looking in either the type declaration or object declaration
-   equate table.  */
+   by looking in the type declaration, the object declaration equate table or
+   the block mapping.  */
 
 static inline dw_die_ref
 add_abstract_origin_attribute (dw_die_ref die, tree origin)
 {
   dw_die_ref origin_die = NULL;
 
-  if (TREE_CODE (origin) != FUNCTION_DECL)
+  if (TREE_CODE (origin) != FUNCTION_DECL
+      && TREE_CODE (origin) != BLOCK)
     {
       /* We may have gotten separated from the block for the inlined
 	 function, if we're in an exception handler or some such; make
@@ -18500,6 +18501,8 @@  add_abstract_origin_attribute (dw_die_ref die, tree origin)
     origin_die = lookup_decl_die (origin);
   else if (TYPE_P (origin))
     origin_die = lookup_type_die (origin);
+  else if (TREE_CODE (origin) == BLOCK)
+    origin_die = BLOCK_DIE (origin);
 
   /* XXX: Functions that are never lowered don't always have correct block
      trees (in the case of java, they simply have no block tree, in some other
@@ -21307,6 +21310,10 @@  gen_lexical_block_die (tree stmt, dw_die_ref context_die)
 	  BLOCK_DIE (stmt) = stmt_die;
 	  old_die = NULL;
 	}
+
+      tree origin = block_ultimate_origin (stmt);
+      if (origin != NULL_TREE && origin != stmt)
+	add_abstract_origin_attribute (stmt_die, origin);
     }
 
   if (old_die)
diff --git a/gcc/testsuite/gcc.dg/debug/dwarf2/nested_fun.c b/gcc/testsuite/gcc.dg/debug/dwarf2/nested_fun.c
new file mode 100644
index 0000000..c783ac0
--- /dev/null
+++ b/gcc/testsuite/gcc.dg/debug/dwarf2/nested_fun.c
@@ -0,0 +1,65 @@ 
+/* As part of inlining, a BLOCK (described as DW_TAG_lexical_block DIE's) may
+   be present both as an abstract instance and a concrete one in the DWARF
+   output.  This testcase attempts to make sure that the concrete ones refer to
+   the abstract ones thanks to the DW_AT_abstract_origin attribute.
+
+   Such a back-link enables debuggers to make entities present in the abstract
+   instance only available in concrete ones.  */
+
+/* { dg-options "-O2 -g -std=gnu99 -gdwarf -dA" } */
+/* { dg-final { scan-assembler-times "\\(DIE \\(0x.*\\) DW_TAG_lexical_block\\)\[^)\]*DW_AT_abstract_origin" 1 } } */
+
+extern void *create (const char *);
+extern void destroy (void *);
+extern void do_nothing (char);
+
+struct string
+{
+  const char *data;
+  int lb;
+  int ub;
+};
+
+int
+main (void)
+{
+  void *o1 = create ("foo");
+
+  void
+  parent (void)
+  {
+    {
+      void *o2 = create ("bar");
+
+      int
+      child (struct string s)
+      {
+	int i = s.lb;
+
+	if (s.lb <= s.ub)
+	  while (1)
+	    {
+	      char c = s.data[i - s.lb];
+	      do_nothing (c);
+	      if (c == 'o')
+		return 1;
+	      if (i == s.ub)
+		break;
+	      ++i;
+	    }
+	return 0;
+      }
+
+      int r;
+
+      r = child ((struct string) {"baz", 1, 3});
+      r = child ((struct string) {"qux", 2, 4});
+      r = child ((struct string) {"foobar", 1, 6});
+    }
+
+    do_nothing (0);
+  }
+
+  parent ();
+  return 0;
+}
-- 
2.3.3.199.g52cae64