public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
* [PATCH] debug: Improve debug info of c++14 deduced return type [PR94459]
@ 2020-04-03 22:47 Jakub Jelinek
  2020-04-03 22:58 ` Hannes Domani
  2020-04-04  4:08 ` Jason Merrill
  0 siblings, 2 replies; 5+ messages in thread
From: Jakub Jelinek @ 2020-04-03 22:47 UTC (permalink / raw)
  To: Jason Merrill; +Cc: gcc-patches

Hi!

On the following testcase, in gdb ptype S<long>::m1 prints long as return
type, but all the other methods show void instead.
PR53756 added code to add_type_attribute if the return type is
auto/decltype(auto), but we actually should look through references,
pointers and qualifiers.
Haven't included there DW_TAG_atomic_type, because I think at least ATM
one can't use that in C++.  Not sure about DW_TAG_array_type or what else
could be deduced.

Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?

2020-04-04  Domani Hannes  <ssbssa@yahoo.de>
	    Jakub Jelinek  <jakub@redhat.com>

	PR debug/94459
	* dwarf2out.c (gen_subprogram_die): Look through references, pointers
	and qualifiers when checking if in-class DIE had an 'auto' or
	'decltype(auto)' return type to emit type again on definition.

	* g++.dg/debug/pr94459.C: New test.

--- gcc/dwarf2out.c.jj	2020-04-03 10:04:44.704972108 +0200
+++ gcc/dwarf2out.c	2020-04-03 17:52:48.909483818 +0200
@@ -22905,11 +22905,19 @@ gen_subprogram_die (tree decl, dw_die_re
 		  != (unsigned) s.column))
 	    add_AT_unsigned (subr_die, DW_AT_decl_column, s.column);
 
-	  /* If the prototype had an 'auto' or 'decltype(auto)' return type,
+	  /* If the prototype had a cv 'auto' or 'decltype(auto)' return type,
 	     emit the real type on the definition die.  */
 	  if (is_cxx () && debug_info_level > DINFO_LEVEL_TERSE)
 	    {
 	      dw_die_ref die = get_AT_ref (old_die, DW_AT_type);
+	      while (die
+		     && (die->die_tag == DW_TAG_reference_type
+			 || die->die_tag == DW_TAG_rvalue_reference_type
+			 || die->die_tag == DW_TAG_pointer_type
+			 || die->die_tag == DW_TAG_const_type
+			 || die->die_tag == DW_TAG_volatile_type
+			 || die->die_tag == DW_TAG_restrict_type))
+		die = get_AT_ref (die, DW_AT_type);
 	      if (die == auto_die || die == decltype_auto_die)
 		add_type_attribute (subr_die, TREE_TYPE (TREE_TYPE (decl)),
 				    TYPE_UNQUALIFIED, false, context_die);
--- gcc/testsuite/g++.dg/debug/pr94459.C.jj	2020-04-03 18:06:28.234725520 +0200
+++ gcc/testsuite/g++.dg/debug/pr94459.C	2020-04-03 18:05:59.287104932 +0200
@@ -0,0 +1,49 @@
+// PR debug/94459
+// { dg-do compile { target c++11 } }
+// { dg-options "-g -dA" }
+
+template <typename T>
+struct S
+{
+  T v;
+  S () : v (0) {}
+  auto m1 () { return v; }
+  auto &m2 () { return v; }
+  auto &&m3 () { return (T&&)v; }
+  const auto m4 () { return v; }
+  const auto &m5 () { return v; }
+  const auto &&m6 () { return (T&&)v; }
+  volatile auto m7 () { return v; }
+  volatile auto &m8 () { return v; }
+  volatile auto &&m9 () { return (T&&)v; }
+  volatile const auto m10 () { return v; }
+  volatile const auto &m11 () { return v; }
+  volatile const auto &&m12 () { return (T&&)v; }
+  const volatile auto m13 () { return v; }
+  const volatile auto &m14 () { return v; }
+  const volatile auto &&m15 () { return (T&&)v; }
+#ifndef __STRICT_ANSI__
+  __restrict const volatile auto &&m16 () { return (T&&)v; }
+  const __restrict auto &m17 () { return v; }
+#endif
+};
+
+S<long> s, u, v;
+
+long
+foo ()
+{
+  return s.m1 () + s.m2 () + s.m3 () + s.m4 () + s.m5 ()
+	 + s.m6 () + s.m7 () + s.m8 () + s.m9 () + s.m10 ()
+	 + s.m11 () + s.m12 () + s.m13 () + s.m14 () + s.m15 ()
+#ifndef __STRICT_ANSI__
+	 + u.m16 () + v.m17 ()
+#endif
+	 + 0;
+}
+
+int
+main ()
+{
+  return foo ();
+}

	Jakub


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] debug: Improve debug info of c++14 deduced return type [PR94459]
  2020-04-03 22:47 [PATCH] debug: Improve debug info of c++14 deduced return type [PR94459] Jakub Jelinek
@ 2020-04-03 22:58 ` Hannes Domani
  2020-04-04  4:08 ` Jason Merrill
  1 sibling, 0 replies; 5+ messages in thread
From: Hannes Domani @ 2020-04-03 22:58 UTC (permalink / raw)
  To: gcc-patches

 Am Samstag, 4. April 2020, 00:47:27 MESZ hat Jakub Jelinek via Gcc-patches <gcc-patches@gcc.gnu.org> Folgendes geschrieben:

> Hi!
>
> On the following testcase, in gdb ptype S<long>::m1 prints long as return
> type, but all the other methods show void instead.
> PR53756 added code to add_type_attribute if the return type is
> auto/decltype(auto), but we actually should look through references,
> pointers and qualifiers.
> Haven't included there DW_TAG_atomic_type, because I think at least ATM
> one can't use that in C++.  Not sure about DW_TAG_array_type or what else
> could be deduced.
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
>
> 2020-04-04  Domani Hannes  <ssbssa@yahoo.de>
>         Jakub Jelinek  <jakub@redhat.com>

Minor hiccup from my side, my name should be written as Hannes Domani.

I think when I registered the bugzilla account, I wasn't aware yet that
it should be family name last, I've fixed that now.


Hannes

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH] debug: Improve debug info of c++14 deduced return type [PR94459]
  2020-04-03 22:47 [PATCH] debug: Improve debug info of c++14 deduced return type [PR94459] Jakub Jelinek
  2020-04-03 22:58 ` Hannes Domani
@ 2020-04-04  4:08 ` Jason Merrill
  2020-04-04  7:54   ` [PATCH v2] " Jakub Jelinek
  1 sibling, 1 reply; 5+ messages in thread
From: Jason Merrill @ 2020-04-04  4:08 UTC (permalink / raw)
  To: Jakub Jelinek; +Cc: gcc-patches

On 4/3/20 6:47 PM, Jakub Jelinek wrote:
> Hi!
> 
> On the following testcase, in gdb ptype S<long>::m1 prints long as return
> type, but all the other methods show void instead.
> PR53756 added code to add_type_attribute if the return type is
> auto/decltype(auto), but we actually should look through references,
> pointers and qualifiers.
> Haven't included there DW_TAG_atomic_type, because I think at least ATM
> one can't use that in C++.  Not sure about DW_TAG_array_type or what else
> could be deduced.

http://eel.is/c++draft/dcl.spec.auto#3 says it has to appear as a 
decl-specifier.

http://eel.is/c++draft/temp.deduct.type#8 lists the forms where a 
template argument can be deduced.

Looks like you are missing arrays, pointers to members, and function 
return types.

> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
> 
> 2020-04-04  Domani Hannes  <ssbssa@yahoo.de>
> 	    Jakub Jelinek  <jakub@redhat.com>
> 
> 	PR debug/94459
> 	* dwarf2out.c (gen_subprogram_die): Look through references, pointers
> 	and qualifiers when checking if in-class DIE had an 'auto' or
> 	'decltype(auto)' return type to emit type again on definition.
> 
> 	* g++.dg/debug/pr94459.C: New test.
> 
> --- gcc/dwarf2out.c.jj	2020-04-03 10:04:44.704972108 +0200
> +++ gcc/dwarf2out.c	2020-04-03 17:52:48.909483818 +0200
> @@ -22905,11 +22905,19 @@ gen_subprogram_die (tree decl, dw_die_re
>   		  != (unsigned) s.column))
>   	    add_AT_unsigned (subr_die, DW_AT_decl_column, s.column);
>   
> -	  /* If the prototype had an 'auto' or 'decltype(auto)' return type,
> +	  /* If the prototype had a cv 'auto' or 'decltype(auto)' return type,
>   	     emit the real type on the definition die.  */
>   	  if (is_cxx () && debug_info_level > DINFO_LEVEL_TERSE)
>   	    {
>   	      dw_die_ref die = get_AT_ref (old_die, DW_AT_type);
> +	      while (die
> +		     && (die->die_tag == DW_TAG_reference_type
> +			 || die->die_tag == DW_TAG_rvalue_reference_type
> +			 || die->die_tag == DW_TAG_pointer_type
> +			 || die->die_tag == DW_TAG_const_type
> +			 || die->die_tag == DW_TAG_volatile_type
> +			 || die->die_tag == DW_TAG_restrict_type))
> +		die = get_AT_ref (die, DW_AT_type);
>   	      if (die == auto_die || die == decltype_auto_die)
>   		add_type_attribute (subr_die, TREE_TYPE (TREE_TYPE (decl)),
>   				    TYPE_UNQUALIFIED, false, context_die);
> --- gcc/testsuite/g++.dg/debug/pr94459.C.jj	2020-04-03 18:06:28.234725520 +0200
> +++ gcc/testsuite/g++.dg/debug/pr94459.C	2020-04-03 18:05:59.287104932 +0200
> @@ -0,0 +1,49 @@
> +// PR debug/94459
> +// { dg-do compile { target c++11 } }
> +// { dg-options "-g -dA" }
> +
> +template <typename T>
> +struct S
> +{
> +  T v;
> +  S () : v (0) {}
> +  auto m1 () { return v; }
> +  auto &m2 () { return v; }
> +  auto &&m3 () { return (T&&)v; }
> +  const auto m4 () { return v; }
> +  const auto &m5 () { return v; }
> +  const auto &&m6 () { return (T&&)v; }
> +  volatile auto m7 () { return v; }
> +  volatile auto &m8 () { return v; }
> +  volatile auto &&m9 () { return (T&&)v; }
> +  volatile const auto m10 () { return v; }
> +  volatile const auto &m11 () { return v; }
> +  volatile const auto &&m12 () { return (T&&)v; }
> +  const volatile auto m13 () { return v; }
> +  const volatile auto &m14 () { return v; }
> +  const volatile auto &&m15 () { return (T&&)v; }
> +#ifndef __STRICT_ANSI__
> +  __restrict const volatile auto &&m16 () { return (T&&)v; }
> +  const __restrict auto &m17 () { return v; }
> +#endif
> +};
> +
> +S<long> s, u, v;
> +
> +long
> +foo ()
> +{
> +  return s.m1 () + s.m2 () + s.m3 () + s.m4 () + s.m5 ()
> +	 + s.m6 () + s.m7 () + s.m8 () + s.m9 () + s.m10 ()
> +	 + s.m11 () + s.m12 () + s.m13 () + s.m14 () + s.m15 ()
> +#ifndef __STRICT_ANSI__
> +	 + u.m16 () + v.m17 ()
> +#endif
> +	 + 0;
> +}
> +
> +int
> +main ()
> +{
> +  return foo ();
> +}
> 
> 	Jakub
> 


^ permalink raw reply	[flat|nested] 5+ messages in thread

* [PATCH v2] debug: Improve debug info of c++14 deduced return type [PR94459]
  2020-04-04  4:08 ` Jason Merrill
@ 2020-04-04  7:54   ` Jakub Jelinek
  2020-04-04 15:58     ` Jason Merrill
  0 siblings, 1 reply; 5+ messages in thread
From: Jakub Jelinek @ 2020-04-04  7:54 UTC (permalink / raw)
  To: Jason Merrill; +Cc: gcc-patches

On Sat, Apr 04, 2020 at 12:08:38AM -0400, Jason Merrill wrote:
> http://eel.is/c++draft/dcl.spec.auto#3 says it has to appear as a
> decl-specifier.
> 
> http://eel.is/c++draft/temp.deduct.type#8 lists the forms where a template
> argument can be deduced.
> 
> Looks like you are missing arrays, pointers to members, and function return
> types.

I have added that to dwarf2out.c but have a hard time figuring out how to
write the array case (added just the pointer, pointer-to-member (both data
and function) and pointer to function cases to the testcase so far).
Tried
  auto (&m22 ())[2] { return w; }
but that is rejected with:
pr94459.C:40:10: error: ‘m22’ declared as array of ‘auto’
   40 |   auto (&m22 ())[2] { return w; }
      |          ^~~

2020-04-04  Hannes Domani  <ssbssa@yahoo.de>
	    Jakub Jelinek  <jakub@redhat.com>

	PR debug/94459
	* dwarf2out.c (gen_subprogram_die): Look through references, pointers,
	arrays, pointer-to-members, function types and qualifiers when
	checking if in-class DIE had an 'auto' or 'decltype(auto)' return type
	to emit type again on definition.

	* g++.dg/debug/pr94459.C: New test.

--- gcc/dwarf2out.c.jj	2020-04-04 00:35:03.450199722 +0200
+++ gcc/dwarf2out.c	2020-04-04 09:45:27.821396824 +0200
@@ -22905,11 +22905,22 @@ gen_subprogram_die (tree decl, dw_die_re
 		  != (unsigned) s.column))
 	    add_AT_unsigned (subr_die, DW_AT_decl_column, s.column);
 
-	  /* If the prototype had an 'auto' or 'decltype(auto)' return type,
-	     emit the real type on the definition die.  */
+	  /* If the prototype had an 'auto' or 'decltype(auto)' in
+	     the return type, emit the real type on the definition die.  */
 	  if (is_cxx () && debug_info_level > DINFO_LEVEL_TERSE)
 	    {
 	      dw_die_ref die = get_AT_ref (old_die, DW_AT_type);
+	      while (die
+		     && (die->die_tag == DW_TAG_reference_type
+			 || die->die_tag == DW_TAG_rvalue_reference_type
+			 || die->die_tag == DW_TAG_pointer_type
+			 || die->die_tag == DW_TAG_const_type
+			 || die->die_tag == DW_TAG_volatile_type
+			 || die->die_tag == DW_TAG_restrict_type
+			 || die->die_tag == DW_TAG_array_type
+			 || die->die_tag == DW_TAG_ptr_to_member_type
+			 || die->die_tag == DW_TAG_subroutine_type))
+		die = get_AT_ref (die, DW_AT_type);
 	      if (die == auto_die || die == decltype_auto_die)
 		add_type_attribute (subr_die, TREE_TYPE (TREE_TYPE (decl)),
 				    TYPE_UNQUALIFIED, false, context_die);
--- gcc/testsuite/g++.dg/debug/pr94459.C.jj	2020-04-04 09:24:19.655238506 +0200
+++ gcc/testsuite/g++.dg/debug/pr94459.C	2020-04-04 09:44:07.572588212 +0200
@@ -0,0 +1,63 @@
+// PR debug/94459
+// { dg-do compile { target c++14 } }
+// { dg-options "-g -dA" }
+
+auto
+baz ()
+{
+  return 0L;
+}
+
+template <typename T>
+struct S
+{
+  T v;
+  T w[2];
+  S () : v (0), w { 0, 0 } {}
+  auto m1 () { return v; }
+  auto &m2 () { return v; }
+  auto &&m3 () { return (T&&)v; }
+  const auto m4 () { return v; }
+  const auto &m5 () { return v; }
+  const auto &&m6 () { return (T&&)v; }
+  volatile auto m7 () { return v; }
+  volatile auto &m8 () { return v; }
+  volatile auto &&m9 () { return (T&&)v; }
+  volatile const auto m10 () { return v; }
+  volatile const auto &m11 () { return v; }
+  volatile const auto &&m12 () { return (T&&)v; }
+  const volatile auto m13 () { return v; }
+  const volatile auto &m14 () { return v; }
+  const volatile auto &&m15 () { return (T&&)v; }
+#ifndef __STRICT_ANSI__
+  __restrict const volatile auto &&m16 () { return (T&&)v; }
+  const __restrict auto &m17 () { return v; }
+#endif
+  auto *m18 () { return &v; }
+  auto (S::* (m19 ())) () { return &S::m1; }
+  auto (S::* (m20 ())) { return &S::v; }
+  auto (*m21 ()) () { return baz; }
+};
+
+S<long> s, u, v;
+
+long
+foo ()
+{
+  auto x = s.m19 ();
+  auto y = s.m20 ();
+  auto z = s.m21 ();
+  return s.m1 () + s.m2 () + s.m3 () + s.m4 () + s.m5 ()
+	 + s.m6 () + s.m7 () + s.m8 () + s.m9 () + s.m10 ()
+	 + s.m11 () + s.m12 () + s.m13 () + s.m14 () + s.m15 ()
+#ifndef __STRICT_ANSI__
+	 + u.m16 () + v.m17 ()
+#endif
+	 + *s.m18 () + (s.*x) () + s.*y + z ();
+}
+
+int
+main ()
+{
+  return foo ();
+}

	Jakub


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [PATCH v2] debug: Improve debug info of c++14 deduced return type [PR94459]
  2020-04-04  7:54   ` [PATCH v2] " Jakub Jelinek
@ 2020-04-04 15:58     ` Jason Merrill
  0 siblings, 0 replies; 5+ messages in thread
From: Jason Merrill @ 2020-04-04 15:58 UTC (permalink / raw)
  To: Jakub Jelinek; +Cc: gcc-patches

On 4/4/20 3:54 AM, Jakub Jelinek wrote:
> On Sat, Apr 04, 2020 at 12:08:38AM -0400, Jason Merrill wrote:
>> http://eel.is/c++draft/dcl.spec.auto#3 says it has to appear as a
>> decl-specifier.
>>
>> http://eel.is/c++draft/temp.deduct.type#8 lists the forms where a template
>> argument can be deduced.
>>
>> Looks like you are missing arrays, pointers to members, and function return
>> types.
> 
> I have added that to dwarf2out.c but have a hard time figuring out how to
> write the array case (added just the pointer, pointer-to-member (both data
> and function) and pointer to function cases to the testcase so far).
> Tried
>    auto (&m22 ())[2] { return w; }
> but that is rejected with:
> pr94459.C:40:10: error: ‘m22’ declared as array of ‘auto’
>     40 |   auto (&m22 ())[2] { return w; }
>        |          ^~~

Hmm, that is in fact prohibited by http://eel.is/c++draft/dcl.array#4

"U is called the array element type; this type shall not be a 
placeholder type...."

That restriction seems misplaced, but clearly interferes with testing 
that pattern.  Let's keep the handling in dwarf2out, though.  The patch 
is OK as is.

> 2020-04-04  Hannes Domani  <ssbssa@yahoo.de>
> 	    Jakub Jelinek  <jakub@redhat.com>
> 
> 	PR debug/94459
> 	* dwarf2out.c (gen_subprogram_die): Look through references, pointers,
> 	arrays, pointer-to-members, function types and qualifiers when
> 	checking if in-class DIE had an 'auto' or 'decltype(auto)' return type
> 	to emit type again on definition.
> 
> 	* g++.dg/debug/pr94459.C: New test.
> 
> --- gcc/dwarf2out.c.jj	2020-04-04 00:35:03.450199722 +0200
> +++ gcc/dwarf2out.c	2020-04-04 09:45:27.821396824 +0200
> @@ -22905,11 +22905,22 @@ gen_subprogram_die (tree decl, dw_die_re
>   		  != (unsigned) s.column))
>   	    add_AT_unsigned (subr_die, DW_AT_decl_column, s.column);
>   
> -	  /* If the prototype had an 'auto' or 'decltype(auto)' return type,
> -	     emit the real type on the definition die.  */
> +	  /* If the prototype had an 'auto' or 'decltype(auto)' in
> +	     the return type, emit the real type on the definition die.  */
>   	  if (is_cxx () && debug_info_level > DINFO_LEVEL_TERSE)
>   	    {
>   	      dw_die_ref die = get_AT_ref (old_die, DW_AT_type);
> +	      while (die
> +		     && (die->die_tag == DW_TAG_reference_type
> +			 || die->die_tag == DW_TAG_rvalue_reference_type
> +			 || die->die_tag == DW_TAG_pointer_type
> +			 || die->die_tag == DW_TAG_const_type
> +			 || die->die_tag == DW_TAG_volatile_type
> +			 || die->die_tag == DW_TAG_restrict_type
> +			 || die->die_tag == DW_TAG_array_type
> +			 || die->die_tag == DW_TAG_ptr_to_member_type
> +			 || die->die_tag == DW_TAG_subroutine_type))
> +		die = get_AT_ref (die, DW_AT_type);
>   	      if (die == auto_die || die == decltype_auto_die)
>   		add_type_attribute (subr_die, TREE_TYPE (TREE_TYPE (decl)),
>   				    TYPE_UNQUALIFIED, false, context_die);
> --- gcc/testsuite/g++.dg/debug/pr94459.C.jj	2020-04-04 09:24:19.655238506 +0200
> +++ gcc/testsuite/g++.dg/debug/pr94459.C	2020-04-04 09:44:07.572588212 +0200
> @@ -0,0 +1,63 @@
> +// PR debug/94459
> +// { dg-do compile { target c++14 } }
> +// { dg-options "-g -dA" }
> +
> +auto
> +baz ()
> +{
> +  return 0L;
> +}
> +
> +template <typename T>
> +struct S
> +{
> +  T v;
> +  T w[2];
> +  S () : v (0), w { 0, 0 } {}
> +  auto m1 () { return v; }
> +  auto &m2 () { return v; }
> +  auto &&m3 () { return (T&&)v; }
> +  const auto m4 () { return v; }
> +  const auto &m5 () { return v; }
> +  const auto &&m6 () { return (T&&)v; }
> +  volatile auto m7 () { return v; }
> +  volatile auto &m8 () { return v; }
> +  volatile auto &&m9 () { return (T&&)v; }
> +  volatile const auto m10 () { return v; }
> +  volatile const auto &m11 () { return v; }
> +  volatile const auto &&m12 () { return (T&&)v; }
> +  const volatile auto m13 () { return v; }
> +  const volatile auto &m14 () { return v; }
> +  const volatile auto &&m15 () { return (T&&)v; }
> +#ifndef __STRICT_ANSI__
> +  __restrict const volatile auto &&m16 () { return (T&&)v; }
> +  const __restrict auto &m17 () { return v; }
> +#endif
> +  auto *m18 () { return &v; }
> +  auto (S::* (m19 ())) () { return &S::m1; }
> +  auto (S::* (m20 ())) { return &S::v; }
> +  auto (*m21 ()) () { return baz; }
> +};
> +
> +S<long> s, u, v;
> +
> +long
> +foo ()
> +{
> +  auto x = s.m19 ();
> +  auto y = s.m20 ();
> +  auto z = s.m21 ();
> +  return s.m1 () + s.m2 () + s.m3 () + s.m4 () + s.m5 ()
> +	 + s.m6 () + s.m7 () + s.m8 () + s.m9 () + s.m10 ()
> +	 + s.m11 () + s.m12 () + s.m13 () + s.m14 () + s.m15 ()
> +#ifndef __STRICT_ANSI__
> +	 + u.m16 () + v.m17 ()
> +#endif
> +	 + *s.m18 () + (s.*x) () + s.*y + z ();
> +}
> +
> +int
> +main ()
> +{
> +  return foo ();
> +}
> 
> 	Jakub
> 


^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2020-04-04 15:58 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-04-03 22:47 [PATCH] debug: Improve debug info of c++14 deduced return type [PR94459] Jakub Jelinek
2020-04-03 22:58 ` Hannes Domani
2020-04-04  4:08 ` Jason Merrill
2020-04-04  7:54   ` [PATCH v2] " Jakub Jelinek
2020-04-04 15:58     ` Jason Merrill

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).