[PATCH] Fix regression in pointer-to-member printing

[PATCH] Fix regression in pointer-to-member printing

[Tom Tromey]

PR c++/29243 points out that "info func" on a certain C++ executable
will cause an infinite loop in gdb.

I tracked this down to a bug introduced by commit 6b5a7bc76 ("Handle
member pointers directly in generic_value_print").  Before this
commit, the C++ code to print a member pointer would wind up calling
value_print_scalar_formatted; but afterward it simply calls
generic_value_print and gets into a loop.

This patch restores the previous behavior and adds a regression test.
---
 gdb/testsuite/gdb.cp/member-ptr.cc  | 8 ++++++++
 gdb/testsuite/gdb.cp/member-ptr.exp | 4 ++++
 gdb/valprint.c                      | 2 +-
 3 files changed, 13 insertions(+), 1 deletion(-)

diff --git a/gdb/testsuite/gdb.cp/member-ptr.cc b/gdb/testsuite/gdb.cp/member-ptr.cc
index 22ffcc58d39..a563d2e6af3 100644
--- a/gdb/testsuite/gdb.cp/member-ptr.cc
+++ b/gdb/testsuite/gdb.cp/member-ptr.cc
@@ -142,6 +142,11 @@ int Diamond::vget_base ()
   return this->Left::x + 2000;
 }
 
+struct Container
+{
+  PMI member;
+};
+
 int
 func (int x)
 {
@@ -205,6 +210,9 @@ int main ()
   null_pmi = NULL;
   null_pmf = NULL;
 
+  Container contain;
+  contain.member = &A::j;
+
   pmi = NULL; /* Breakpoint 1 here.  */
 
   (diamond.*diamond_pfunc_ptr) (20);
diff --git a/gdb/testsuite/gdb.cp/member-ptr.exp b/gdb/testsuite/gdb.cp/member-ptr.exp
index 110497b893b..5b91bcb493a 100644
--- a/gdb/testsuite/gdb.cp/member-ptr.exp
+++ b/gdb/testsuite/gdb.cp/member-ptr.exp
@@ -628,3 +628,7 @@ gdb_test "print null_pmi = 0" "$vhn = NULL"
 gdb_test "print null_pmf" "$vhn = NULL"
 gdb_test "print null_pmf = &A::foo" "$vhn = \\(int \\(A::\\*\\)\\(A \\*( const)?, int\\)\\) $hex <A::foo ?\\(int\\)>"
 gdb_test "print null_pmf = 0" "$vhn = NULL"
+
+# Print with a format, bypassing the direct call to the scalar
+# printer.  See PR c++/29243.
+gdb_test "print/x contain" " = {member = $hex}"
diff --git a/gdb/valprint.c b/gdb/valprint.c
index f079f31fa7b..585af69cb27 100644
--- a/gdb/valprint.c
+++ b/gdb/valprint.c
@@ -879,7 +879,7 @@ generic_value_print_memberptr
       cp_print_class_member (valaddr, type, stream, "&");
     }
   else
-    generic_value_print (val, stream, recurse, options, decorations);
+    value_print_scalar_formatted (val, options, 0, stream);
 }
 
 /* See valprint.h.  */
-- 
2.34.3

Re: [PATCH] Fix regression in pointer-to-member printing

[Joel Brobecker]

Hi Tom,

On Tue, Oct 18, 2022 at 11:53:04AM -0600, Tom Tromey via Gdb-patches wrote:
> PR c++/29243 points out that "info func" on a certain C++ executable
> will cause an infinite loop in gdb.
> 
> I tracked this down to a bug introduced by commit 6b5a7bc76 ("Handle
> member pointers directly in generic_value_print").  Before this
> commit, the C++ code to print a member pointer would wind up calling
> value_print_scalar_formatted; but afterward it simply calls
> generic_value_print and gets into a loop.

Thanks for the fix :-).

The patch looks good to me. Thanks also for including a regression
test!

> This patch restores the previous behavior and adds a regression test.
> ---
>  gdb/testsuite/gdb.cp/member-ptr.cc  | 8 ++++++++
>  gdb/testsuite/gdb.cp/member-ptr.exp | 4 ++++
>  gdb/valprint.c                      | 2 +-
>  3 files changed, 13 insertions(+), 1 deletion(-)
> 
> diff --git a/gdb/testsuite/gdb.cp/member-ptr.cc b/gdb/testsuite/gdb.cp/member-ptr.cc
> index 22ffcc58d39..a563d2e6af3 100644
> --- a/gdb/testsuite/gdb.cp/member-ptr.cc
> +++ b/gdb/testsuite/gdb.cp/member-ptr.cc
> @@ -142,6 +142,11 @@ int Diamond::vget_base ()
>    return this->Left::x + 2000;
>  }
>  
> +struct Container
> +{
> +  PMI member;
> +};
> +
>  int
>  func (int x)
>  {
> @@ -205,6 +210,9 @@ int main ()
>    null_pmi = NULL;
>    null_pmf = NULL;
>  
> +  Container contain;
> +  contain.member = &A::j;
> +
>    pmi = NULL; /* Breakpoint 1 here.  */
>  
>    (diamond.*diamond_pfunc_ptr) (20);
> diff --git a/gdb/testsuite/gdb.cp/member-ptr.exp b/gdb/testsuite/gdb.cp/member-ptr.exp
> index 110497b893b..5b91bcb493a 100644
> --- a/gdb/testsuite/gdb.cp/member-ptr.exp
> +++ b/gdb/testsuite/gdb.cp/member-ptr.exp
> @@ -628,3 +628,7 @@ gdb_test "print null_pmi = 0" "$vhn = NULL"
>  gdb_test "print null_pmf" "$vhn = NULL"
>  gdb_test "print null_pmf = &A::foo" "$vhn = \\(int \\(A::\\*\\)\\(A \\*( const)?, int\\)\\) $hex <A::foo ?\\(int\\)>"
>  gdb_test "print null_pmf = 0" "$vhn = NULL"
> +
> +# Print with a format, bypassing the direct call to the scalar
> +# printer.  See PR c++/29243.
> +gdb_test "print/x contain" " = {member = $hex}"
> diff --git a/gdb/valprint.c b/gdb/valprint.c
> index f079f31fa7b..585af69cb27 100644
> --- a/gdb/valprint.c
> +++ b/gdb/valprint.c
> @@ -879,7 +879,7 @@ generic_value_print_memberptr
>        cp_print_class_member (valaddr, type, stream, "&");
>      }
>    else
> -    generic_value_print (val, stream, recurse, options, decorations);
> +    value_print_scalar_formatted (val, options, 0, stream);
>  }
>  
>  /* See valprint.h.  */
> -- 
> 2.34.3
> 

-- 
Joel