Building current gdb on centos-5

Building current gdb on centos-5

[Julian Smith]

Hello

Is the latest gdb in cvs, expected to build on centos-5 ?

I'm seeing this build failure:

...
gcc -g -O2   -I. -I. -I./common -I./config -DLOCALEDIR="\"/usr/local/share/locale\"" -DHAVE_CONFIG_H -I./../include/opcode -I./../opcodes/.. -I./../readline/.. -I../bfd -I./../bfd -I./../include -I../libdecnumber -I./../libdecnumber  -I./gn
ulib/import -Ibuild-gnulib/import   -DTUI=1  -I/usr/include/python2.4 -I/usr/include/python2.4 -Wall -Wdeclaration-after-statement -Wpointer-arith -Wpointer-sign -Wno-unused -Wunused-value -Wunused-function -Wno-switch -Wno-char-subscripts 
-Wmissing-prototypes -Wdeclaration-after-statement -Wformat-nonliteral  -c -o mi-main.o -MT mi-main.o -MMD -MP -MF .deps/mi-main.Tpo ./mi/mi-main.c
In file included from ./mi/mi-main.c:56:
./python/python-internal.h: In function 'gdb_Py_DECREF':
./python/python-internal.h:179: warning: dereferencing 'void *' pointer
./python/python-internal.h:179: error: request for member 'ob_refcnt' in something not a structure or union

I'm building with:

    cd ~/gdb_cvs_dir/src
    ./configure --disable-werror
    make

My centos-5 has python-2.4 (python-devel.i386 2.4.3-24.el5).

Many thanks,

- Julian

-- 
http://op59.net/

Re: Building current gdb on centos-5

[Pedro Alves]

On 05/29/2013 09:50 PM, Julian Smith wrote:

> Is the latest gdb in cvs, expected to build on centos-5 ?

It should, but looks like I broke it.

> I'm seeing this build failure:
> 
> ...
> gcc -g -O2   -I. -I. -I./common -I./config -DLOCALEDIR="\"/usr/local/share/locale\"" -DHAVE_CONFIG_H -I./../include/opcode -I./../opcodes/.. -I./../readline/.. -I../bfd -I./../bfd -I./../include -I../libdecnumber -I./../libdecnumber  -I./gn
> ulib/import -Ibuild-gnulib/import   -DTUI=1  -I/usr/include/python2.4 -I/usr/include/python2.4 -Wall -Wdeclaration-after-statement -Wpointer-arith -Wpointer-sign -Wno-unused -Wunused-value -Wunused-function -Wno-switch -Wno-char-subscripts 
> -Wmissing-prototypes -Wdeclaration-after-statement -Wformat-nonliteral  -c -o mi-main.o -MT mi-main.o -MMD -MP -MF .deps/mi-main.Tpo ./mi/mi-main.c
> In file included from ./mi/mi-main.c:56:
> ./python/python-internal.h: In function 'gdb_Py_DECREF':
> ./python/python-internal.h:179: warning: dereferencing 'void *' pointer
> ./python/python-internal.h:179: error: request for member 'ob_refcnt' in something not a structure or union

Thanks.  gdb_Py_DECREF is a recent addition:

/* Python 2.6 did not wrap Py_DECREF in 'do {...} while (0)', leading
   to 'suggest explicit braces to avoid ambiguous ‘else’' gcc errors.
   Wrap it ourselves, so that callers don't need to care.  */

static inline void
gdb_Py_DECREF (void *op) /* ARI: editCase function */
{
  Py_DECREF (op);
}

Making that:

-  Py_DECREF (op);
+  Py_DECREF ((PyObject*) op);

should fix this.  I'll push a fix.

On my system's Python 2.7, Py_DECREF always casts its argument
to (PyObject*), so it didn't seem necessary to cast it ourselves:

#define Py_DECREF(op)                                   \
    do {                                                \
        if (_Py_DEC_REFTOTAL  _Py_REF_DEBUG_COMMA       \
        --((PyObject*)(op))->ob_refcnt != 0)            \
            _Py_CHECK_REFCNT(op)                        \
        else                                            \
        _Py_Dealloc((PyObject *)(op));                  \
    } while (0)

> My centos-5 has python-2.4 (python-devel.i386 2.4.3-24.el5).

For the archives, could you please paste what Py_DECREF looks
like in 2.4?  Thanks!

-- 
Pedro Alves

Re: Building current gdb on centos-5

[Pedro Alves]

On 05/30/2013 09:08 AM, Pedro Alves wrote:

> For the archives, could you please paste what Py_DECREF looks
> like in 2.4?  Thanks!

Nevermind, downloading the 2.4.6's sources only took a minute.  :-)

2.4:

#define Py_DECREF(op)                                   \
        if (_Py_DEC_REFTOTAL  _Py_REF_DEBUG_COMMA       \
            --(op)->ob_refcnt != 0)                     \
                _Py_CHECK_REFCNT(op)                    \
        else                                            \
                _Py_Dealloc((PyObject *)(op))

2.7:

#define Py_DECREF(op)                                   \
    do {                                                \
        if (_Py_DEC_REFTOTAL  _Py_REF_DEBUG_COMMA       \
        --((PyObject*)(op))->ob_refcnt != 0)            \
            _Py_CHECK_REFCNT(op)                        \
        else                                            \
        _Py_Dealloc((PyObject *)(op));                  \
    } while (0)

-- 
Pedro Alves

Re: Building current gdb on centos-5

[Julian Smith]

On Thu, 30 May 2013 09:08:26 +0100
Pedro Alves <palves@redhat.com> wrote:

> On 05/29/2013 09:50 PM, Julian Smith wrote:
> 
> > Is the latest gdb in cvs, expected to build on centos-5 ?
> 
> It should, but looks like I broke it.
> 
> > I'm seeing this build failure:
> > 
> > ...
> > gcc -g -O2   -I. -I. -I./common -I./config -DLOCALEDIR="\"/usr/local/share/locale\"" -DHAVE_CONFIG_H -I./../include/opcode -I./../opcodes/.. -I./../readline/.. -I../bfd -I./../bfd -I./../include -I../libdecnumber -I./../libdecnumber  -I./gn
> > ulib/import -Ibuild-gnulib/import   -DTUI=1  -I/usr/include/python2.4 -I/usr/include/python2.4 -Wall -Wdeclaration-after-statement -Wpointer-arith -Wpointer-sign -Wno-unused -Wunused-value -Wunused-function -Wno-switch -Wno-char-subscripts 
> > -Wmissing-prototypes -Wdeclaration-after-statement -Wformat-nonliteral  -c -o mi-main.o -MT mi-main.o -MMD -MP -MF .deps/mi-main.Tpo ./mi/mi-main.c
> > In file included from ./mi/mi-main.c:56:
> > ./python/python-internal.h: In function 'gdb_Py_DECREF':
> > ./python/python-internal.h:179: warning: dereferencing 'void *' pointer
> > ./python/python-internal.h:179: error: request for member 'ob_refcnt' in something not a structure or union
> 
> Thanks.  gdb_Py_DECREF is a recent addition:
> 
> /* Python 2.6 did not wrap Py_DECREF in 'do {...} while (0)', leading
>    to 'suggest explicit braces to avoid ambiguous ‘else’' gcc errors.
>    Wrap it ourselves, so that callers don't need to care.  */
> 
> static inline void
> gdb_Py_DECREF (void *op) /* ARI: editCase function */
> {
>   Py_DECREF (op);
> }
> 
> Making that:
> 
> -  Py_DECREF (op);
> +  Py_DECREF ((PyObject*) op);
> 
> should fix this.  I'll push a fix.

Thanks.

> 
> On my system's Python 2.7, Py_DECREF always casts its argument
> to (PyObject*), so it didn't seem necessary to cast it ourselves:
> 
> #define Py_DECREF(op)                                   \
>     do {                                                \
>         if (_Py_DEC_REFTOTAL  _Py_REF_DEBUG_COMMA       \
>         --((PyObject*)(op))->ob_refcnt != 0)            \
>             _Py_CHECK_REFCNT(op)                        \
>         else                                            \
>         _Py_Dealloc((PyObject *)(op));                  \
>     } while (0)
> 
> > My centos-5 has python-2.4 (python-devel.i386 2.4.3-24.el5).
> 
> For the archives, could you please paste what Py_DECREF looks
> like in 2.4?  Thanks!

Here it is, in /usr/include/python2.4/object.h:

#define Py_DECREF(op)                                   \
        if (_Py_DEC_REFTOTAL  _Py_REF_DEBUG_COMMA       \
            --(op)->ob_refcnt != 0)                     \
                _Py_CHECK_REFCNT(op)                    \
        else                                            \
                _Py_Dealloc((PyObject *)(op))


Thanks,

- Julian

-- 
http://op59.net/

Re: Building current gdb on centos-5

[Pedro Alves]

On 05/30/2013 09:08 AM, Pedro Alves wrote:

> Making that:
> 
> -  Py_DECREF (op);
> +  Py_DECREF ((PyObject*) op);
> 
> should fix this.  I'll push a fix.

Done.  http://sourceware.org/ml/gdb-patches/2013-05/msg01045.html

Thanks for reporting.

-- 
Pedro Alves