[PR,gdb/16959] gdb hangs in infinite recursion

  The original problem was fixed (see related PR 22242). But using a typedef
as the declared type for a static member variable, as commented in this PR,
is still causing gdb to get into infinite loop when printing the static
member's value. This problem can be reproduced as follows:

% cat t.cc
class A {
    typedef A type;
public:
    bool operator==(const type& other) { return true; }

    static const type INSTANCE;
};

const A A::INSTANCE;

int main() {
    A a;
    if (a == A::INSTANCE) {
        return -1;
    }
    return 0;
}
% g++ -g t.cc
% gdb -ex "start" -ex "p a" a.out

The fix is rather trivial - in cp_print_static_field(), should call
check_typedef() to get the static member's real type and use it to
check whether it's a struct or an array.

Tested on both aarch64-linux-gnu and amd64-linux-gnu. No regressions.
---
 gdb/ChangeLog     |    7 +++++++
 gdb/cp-valprint.c |    2 +-
 2 files changed, 8 insertions(+), 1 deletions(-)
  

On 2018-03-15 08:42 PM, Weimin Pan wrote:
> The original problem was fixed (see related PR 22242). But using a typedef
> as the declared type for a static member variable, as commented in this PR,
> is still causing gdb to get into infinite loop when printing the static
> member's value. This problem can be reproduced as follows:
> 
> % cat t.cc
> class A {
>     typedef A type;
> public:
>     bool operator==(const type& other) { return true; }
> 
>     static const type INSTANCE;
> };
> 
> const A A::INSTANCE;
> 
> int main() {
>     A a;
>     if (a == A::INSTANCE) {
>         return -1;
>     }
>     return 0;
> }
> % g++ -g t.cc
> % gdb -ex "start" -ex "p a" a.out
> 
> The fix is rather trivial - in cp_print_static_field(), should call
> check_typedef() to get the static member's real type and use it to
> check whether it's a struct or an array.

Hi Weimin,

Would it be possible to add a test case for this?  I suppose you can quite
easily enhance the test case added by commit

  a43f3893f6cb ("Fix broken recursion detection when printing static members")

> Tested on both aarch64-linux-gnu and amd64-linux-gnu. No regressions.
> ---
>  gdb/ChangeLog     |    7 +++++++
>  gdb/cp-valprint.c |    2 +-
>  2 files changed, 8 insertions(+), 1 deletions(-)
> 
> diff --git a/gdb/ChangeLog b/gdb/ChangeLog
> index d0a8dfd..6fd43de 100644
> --- a/gdb/ChangeLog
> +++ b/gdb/ChangeLog
> @@ -1,3 +1,10 @@
> +2018-02-07  Weimin Pan  <weimin.pan@oracle.com>
> +
> +	PR gdb/16959
> +	* cp-valprint.c: (cp_print_static_field) Use check_typedef() to get 
> +	static member's real type for TYPE_CODE_STRUCT and TYPE_CODE_ARRAY 
> +	comparisons. 
> +
>  2018-01-24  Pedro Alves  <palves@redhat.com>
>  
>  	GCC PR libstdc++/83906
> diff --git a/gdb/cp-valprint.c b/gdb/cp-valprint.c
> index 486653f..0370b56 100644
> --- a/gdb/cp-valprint.c
> +++ b/gdb/cp-valprint.c
> @@ -633,6 +633,7 @@ cp_print_static_field (struct type *type,
>        return;
>      }
>  
> +  type = check_typedef (type);
>    if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
>      {
>        CORE_ADDR *first_dont_print;
> @@ -658,7 +659,6 @@ cp_print_static_field (struct type *type,
>        addr = value_address (val);
>        obstack_grow (&dont_print_statmem_obstack, (char *) &addr,
>  		    sizeof (CORE_ADDR));
> -      type = check_typedef (type);
>        cp_print_value_fields (type, value_enclosing_type (val),
>  			     value_embedded_offset (val), addr,
>  			     stream, recurse, val,
> 

type is passed below to val_print.  I think it would be better to continue
passing the original type to that function instead of the resolved type.  It
could affect how things are printed (if the type name is printed somewhere,
or if pretty printers are involved).  Many functions use a variable "real_type"
to hold the result from check_typedef, you could follow that pattern.

Thanks,

Simon
  

On 3/17/2018 10:13 AM, Simon Marchi wrote:
> On 2018-03-15 08:42 PM, Weimin Pan wrote:
>> The original problem was fixed (see related PR 22242). But using a typedef
>> as the declared type for a static member variable, as commented in this PR,
>> is still causing gdb to get into infinite loop when printing the static
>> member's value. This problem can be reproduced as follows:
>>
>> % cat t.cc
>> class A {
>>      typedef A type;
>> public:
>>      bool operator==(const type& other) { return true; }
>>
>>      static const type INSTANCE;
>> };
>>
>> const A A::INSTANCE;
>>
>> int main() {
>>      A a;
>>      if (a == A::INSTANCE) {
>>          return -1;
>>      }
>>      return 0;
>> }
>> % g++ -g t.cc
>> % gdb -ex "start" -ex "p a" a.out
>>
>> The fix is rather trivial - in cp_print_static_field(), should call
>> check_typedef() to get the static member's real type and use it to
>> check whether it's a struct or an array.
> Hi Weimin,
>
> Would it be possible to add a test case for this?  I suppose you can quite
> easily enhance the test case added by commit
>
>    a43f3893f6cb ("Fix broken recursion detection when printing static members")

Hi Simon,

Will see if I can add this testing to an existing test case or I need to 
create a new
test case for it. Maybe the latter is preferred?

Also I need to study/learn your "commit" command since I'm a new "git" user.

Thank you very much for your comment.

Weimin

>
>> Tested on both aarch64-linux-gnu and amd64-linux-gnu. No regressions.
>> ---
>>   gdb/ChangeLog     |    7 +++++++
>>   gdb/cp-valprint.c |    2 +-
>>   2 files changed, 8 insertions(+), 1 deletions(-)
>>
>> diff --git a/gdb/ChangeLog b/gdb/ChangeLog
>> index d0a8dfd..6fd43de 100644
>> --- a/gdb/ChangeLog
>> +++ b/gdb/ChangeLog
>> @@ -1,3 +1,10 @@
>> +2018-02-07  Weimin Pan  <weimin.pan@oracle.com>
>> +
>> +	PR gdb/16959
>> +	* cp-valprint.c: (cp_print_static_field) Use check_typedef() to get
>> +	static member's real type for TYPE_CODE_STRUCT and TYPE_CODE_ARRAY
>> +	comparisons.
>> +
>>   2018-01-24  Pedro Alves  <palves@redhat.com>
>>   
>>   	GCC PR libstdc++/83906
>> diff --git a/gdb/cp-valprint.c b/gdb/cp-valprint.c
>> index 486653f..0370b56 100644
>> --- a/gdb/cp-valprint.c
>> +++ b/gdb/cp-valprint.c
>> @@ -633,6 +633,7 @@ cp_print_static_field (struct type *type,
>>         return;
>>       }
>>   
>> +  type = check_typedef (type);
>>     if (TYPE_CODE (type) == TYPE_CODE_STRUCT)
>>       {
>>         CORE_ADDR *first_dont_print;
>> @@ -658,7 +659,6 @@ cp_print_static_field (struct type *type,
>>         addr = value_address (val);
>>         obstack_grow (&dont_print_statmem_obstack, (char *) &addr,
>>   		    sizeof (CORE_ADDR));
>> -      type = check_typedef (type);
>>         cp_print_value_fields (type, value_enclosing_type (val),
>>   			     value_embedded_offset (val), addr,
>>   			     stream, recurse, val,
>>
> type is passed below to val_print.  I think it would be better to continue
> passing the original type to that function instead of the resolved type.  It
> could affect how things are printed (if the type name is printed somewhere,
> or if pretty printers are involved).  Many functions use a variable "real_type"
> to hold the result from check_typedef, you could follow that pattern.
>
> Thanks,
>
> Simon
>