gdbsupport: fix scoped_debug_start_end's move constructor

  I spotted a problem with scoped_debug_start_end's move constructor.
When constructing a scoped_debug_start_end through it, it doesn't
disable the moved-from object, meaning there are now two objects that
will do the side-effects of decrementing the debug_print_depth global
and printing the "end" message.  Decrementing the debug_print_depth
global twice is actually problematic, because the increments and
decrements get out of sync, meaning we should hit this assertion, in
theory:

    gdb_assert (debug_print_depth > 0);

However, in practice, we don't see that.  This is because despite the
move constructor being required for this to compile:

    template<typename PT>
    static inline scoped_debug_start_end<PT &> ATTRIBUTE_NULL_PRINTF (6, 7)
    make_scoped_debug_start_end (PT &&pred, const char *module, const char *func,
    			     const char *start_prefix,
    			     const char *end_prefix, const char *fmt, ...)
    {
      va_list args;
      va_start (args, fmt);
      auto res = scoped_debug_start_end<PT &> (pred, module, func, start_prefix,
    					   end_prefix, fmt, args);
      va_end (args);

      return res;
    }

... it is never actually called, because compilers elide the move
constructors all the way (the scoped_debug_start_end gets constructed
directly in the instance of the top-level caller).  To confirm this, I
built GDB with -fno-elide-constructors, and now I see it:

    /home/simark/src/binutils-gdb/gdb/../gdbsupport/common-debug.h:147: internal-error: ~scoped_debug_start_end: Assertion `debug_print_depth > 0' failed.

    #9  0x00005614ba5f17c3 in internal_error_loc (file=0x5614b8749960 "/home/simark/src/binutils-gdb/gdb/../gdbsupport/common-debug.h", line=147, fmt=0x5614b8733fa0 "%s: Assertion `%s' failed.") at /home/simark/src/binutils-gdb/gdbsupport/errors.cc:58
    #10 0x00005614b8e1b2e5 in scoped_debug_start_end<bool&>::~scoped_debug_start_end (this=0x7ffc6c5e7b40, __in_chrg=<optimized out>) at /home/simark/src/binutils-gdb/gdb/../gdbsupport/common-debug.h:147
    #11 0x00005614b96dbe34 in make_scoped_debug_start_end<bool&> (pred=@0x5614baad7200: true, module=0x5614b891d840 "infrun", func=0x5614b891d800 "infrun_debug_show_threads", start_prefix=0x5614b891d7c0 "enter", end_prefix=0x5614b891d780 "exit", fmt=0x0) at /home/simark/src/binutils-gdb/gdb/../gdbsupport/common-debug.h:235

Fix this by adding an m_disabled field to scoped_debug_start_end, and
setting it in the move constructor.

Change-Id: Ie5213269c584837f751d2d11de831f45ae4a899f
---
 gdbsupport/common-debug.h | 21 ++++++++++++++++++++-
 1 file changed, 20 insertions(+), 1 deletion(-)


base-commit: aa036eccf094c4d85b953cf1cc2892f0ff746fd9
  

On 1/4/23 16:22, Simon Marchi wrote:
> I spotted a problem with scoped_debug_start_end's move constructor.
> When constructing a scoped_debug_start_end through it, it doesn't
> disable the moved-from object, meaning there are now two objects that
> will do the side-effects of decrementing the debug_print_depth global
> and printing the "end" message.  Decrementing the debug_print_depth
> global twice is actually problematic, because the increments and
> decrements get out of sync, meaning we should hit this assertion, in
> theory:
> 
>     gdb_assert (debug_print_depth > 0);
> 
> However, in practice, we don't see that.  This is because despite the
> move constructor being required for this to compile:
> 
>     template<typename PT>
>     static inline scoped_debug_start_end<PT &> ATTRIBUTE_NULL_PRINTF (6, 7)
>     make_scoped_debug_start_end (PT &&pred, const char *module, const char *func,
>     			     const char *start_prefix,
>     			     const char *end_prefix, const char *fmt, ...)
>     {
>       va_list args;
>       va_start (args, fmt);
>       auto res = scoped_debug_start_end<PT &> (pred, module, func, start_prefix,
>     					   end_prefix, fmt, args);
>       va_end (args);
> 
>       return res;
>     }
> 
> ... it is never actually called, because compilers elide the move
> constructors all the way (the scoped_debug_start_end gets constructed
> directly in the instance of the top-level caller).  To confirm this, I
> built GDB with -fno-elide-constructors, and now I see it:
> 
>     /home/simark/src/binutils-gdb/gdb/../gdbsupport/common-debug.h:147: internal-error: ~scoped_debug_start_end: Assertion `debug_print_depth > 0' failed.
> 
>     #9  0x00005614ba5f17c3 in internal_error_loc (file=0x5614b8749960 "/home/simark/src/binutils-gdb/gdb/../gdbsupport/common-debug.h", line=147, fmt=0x5614b8733fa0 "%s: Assertion `%s' failed.") at /home/simark/src/binutils-gdb/gdbsupport/errors.cc:58
>     #10 0x00005614b8e1b2e5 in scoped_debug_start_end<bool&>::~scoped_debug_start_end (this=0x7ffc6c5e7b40, __in_chrg=<optimized out>) at /home/simark/src/binutils-gdb/gdb/../gdbsupport/common-debug.h:147
>     #11 0x00005614b96dbe34 in make_scoped_debug_start_end<bool&> (pred=@0x5614baad7200: true, module=0x5614b891d840 "infrun", func=0x5614b891d800 "infrun_debug_show_threads", start_prefix=0x5614b891d7c0 "enter", end_prefix=0x5614b891d780 "exit", fmt=0x0) at /home/simark/src/binutils-gdb/gdb/../gdbsupport/common-debug.h:235
> 
> Fix this by adding an m_disabled field to scoped_debug_start_end, and
> setting it in the move constructor.
> 
> Change-Id: Ie5213269c584837f751d2d11de831f45ae4a899f
> ---
>  gdbsupport/common-debug.h | 21 ++++++++++++++++++++-
>  1 file changed, 20 insertions(+), 1 deletion(-)
> 
> diff --git a/gdbsupport/common-debug.h b/gdbsupport/common-debug.h
> index ec36d88fdea2..33b15a005f11 100644
> --- a/gdbsupport/common-debug.h
> +++ b/gdbsupport/common-debug.h
> @@ -138,10 +138,25 @@ struct scoped_debug_start_end
>  
>    DISABLE_COPY_AND_ASSIGN (scoped_debug_start_end);
>  
> -  scoped_debug_start_end (scoped_debug_start_end &&other) = default;
> +  scoped_debug_start_end (scoped_debug_start_end &&other)
> +    : m_debug_enabled (other.m_debug_enabled),
> +      m_module (other.m_module),
> +      m_func (other.m_func),
> +      m_end_prefix (other.m_end_prefix),
> +      m_msg (other.m_msg),

Just found this nit... not that it changes anything (because this ctor
isn't called in practice), but we should std::move m_msg.  I'll change
it locally.

Well, we could std::move all fields, but it would be a bit verbose.

Simon
  

>>>>> "Simon" == Simon Marchi via Gdb-patches <gdb-patches@sourceware.org> writes:

>> +  scoped_debug_start_end (scoped_debug_start_end &&other)
>> +    : m_debug_enabled (other.m_debug_enabled),
>> +      m_module (other.m_module),
>> +      m_func (other.m_func),
>> +      m_end_prefix (other.m_end_prefix),
>> +      m_msg (other.m_msg),

Simon> Just found this nit... not that it changes anything (because this ctor
Simon> isn't called in practice), but we should std::move m_msg.  I'll change
Simon> it locally.

I think it's also fine to just leave it as-is.

Simon> Well, we could std::move all fields, but it would be a bit verbose.

If we think we may need this kind of behavior again, one way would be a
sort of "move token" object that wraps a bool, and that sets the bool on
construction and clears it on move.  Then scoped_debug_start_end could
just use the default move constructor again, and check the token's value
in its destructor.

Probably overkill for just the one case.  I think your patch is ok.

Tom
  

On 1/5/23 15:05, Tom Tromey wrote:
>>>>>> "Simon" == Simon Marchi via Gdb-patches <gdb-patches@sourceware.org> writes:
> 
>>> +  scoped_debug_start_end (scoped_debug_start_end &&other)
>>> +    : m_debug_enabled (other.m_debug_enabled),
>>> +      m_module (other.m_module),
>>> +      m_func (other.m_func),
>>> +      m_end_prefix (other.m_end_prefix),
>>> +      m_msg (other.m_msg),
> 
> Simon> Just found this nit... not that it changes anything (because this ctor
> Simon> isn't called in practice), but we should std::move m_msg.  I'll change
> Simon> it locally.
> 
> I think it's also fine to just leave it as-is.
> 
> Simon> Well, we could std::move all fields, but it would be a bit verbose.
> 
> If we think we may need this kind of behavior again, one way would be a
> sort of "move token" object that wraps a bool, and that sets the bool on
> construction and clears it on move.  Then scoped_debug_start_end could
> just use the default move constructor again, and check the token's value
> in its destructor.

Interesting, I'll keep this in mind.  I think it could apply to some
scoped_restore_* objects that have global side effects, that we want to
be movable, because we want to be able to have functions that return
them.

> Probably overkill for just the one case.  I think your patch is ok.

Ok, thanks, will push.

Simon