bug#70568: 30.0.50; build warnings on 'x86_64-pc-linux-gnu'

bug#70568: 30.0.50; build warnings on 'x86_64-pc-linux-gnu'

[David Ponce via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Hello,

When I build Emacs configured for 'x86_64-pc-linux-gnu' on Fedora 39 with gcc 13.2.1
the below warnings are reported:


1) ./configure --with-x-toolkit=lucid --with-native-compilation=no

..
   CC       xterm.o
xterm.c: In function ‘x_alloc_nearest_color_1’:
xterm.c:9374:27: warning: potential null pointer dereference [-Wnull-dereference]
  9374 |   if (dpyinfo->visual_info.class == DirectColor)
       |       ~~~~~~~~~~~~~~~~~~~~^~~~~~
xterm.c:9374:27: warning: potential null pointer dereference [-Wnull-dereference]
...
   CC       eval.o
In file included from eval.c:25:
lisp.h: In function ‘internal_lisp_condition_case’:
lisp.h:1129:15: warning: variable ‘sym’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
  1129 |   Lisp_Object sym = XSYMBOL_WITH_POS (a)->sym;
       |               ^~~
eval.c:1413:43: warning: argument ‘var’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
  1413 | internal_lisp_condition_case (Lisp_Object var, Lisp_Object bodyform,
       |                               ~~~~~~~~~~~~^~~
...

2) ./configure --with-x-toolkit=gtk3 --with-cairo-xcb --with-native-compilation=no

...
   CC       eval.o
In file included from eval.c:25:
lisp.h: In function ‘internal_lisp_condition_case’:
lisp.h:1129:15: warning: variable ‘sym’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
  1129 |   Lisp_Object sym = XSYMBOL_WITH_POS (a)->sym;
       |               ^~~
eval.c:1413:43: warning: argument ‘var’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
  1413 | internal_lisp_condition_case (Lisp_Object var, Lisp_Object bodyform,
       |                               ~~~~~~~~~~~~^~~
...


In both cases (Lucid, gtk3) the reported warnings don't seem to affect run.
Thanks

bug#70568: 30.0.50; build warnings on 'x86_64-pc-linux-gnu'

[Eli Zaretskii]

> Date: Thu, 25 Apr 2024 17:11:54 +0200
> From:  David Ponce via "Bug reports for GNU Emacs,
>  the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>
> 
> Hello,
> 
> When I build Emacs configured for 'x86_64-pc-linux-gnu' on Fedora 39 with gcc 13.2.1
> the below warnings are reported:
> 
> 
> 1) ./configure --with-x-toolkit=lucid --with-native-compilation=no
> 
> ..
>    CC       xterm.o
> xterm.c: In function ‘x_alloc_nearest_color_1’:
> xterm.c:9374:27: warning: potential null pointer dereference [-Wnull-dereference]
>   9374 |   if (dpyinfo->visual_info.class == DirectColor)
>        |       ~~~~~~~~~~~~~~~~~~~~^~~~~~
> xterm.c:9374:27: warning: potential null pointer dereference [-Wnull-dereference]
> ...
>    CC       eval.o
> In file included from eval.c:25:
> lisp.h: In function ‘internal_lisp_condition_case’:
> lisp.h:1129:15: warning: variable ‘sym’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
>   1129 |   Lisp_Object sym = XSYMBOL_WITH_POS (a)->sym;
>        |               ^~~
> eval.c:1413:43: warning: argument ‘var’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
>   1413 | internal_lisp_condition_case (Lisp_Object var, Lisp_Object bodyform,
>        |                               ~~~~~~~~~~~~^~~
> ...
> 
> 2) ./configure --with-x-toolkit=gtk3 --with-cairo-xcb --with-native-compilation=no
> 
> ...
>    CC       eval.o
> In file included from eval.c:25:
> lisp.h: In function ‘internal_lisp_condition_case’:
> lisp.h:1129:15: warning: variable ‘sym’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
>   1129 |   Lisp_Object sym = XSYMBOL_WITH_POS (a)->sym;
>        |               ^~~
> eval.c:1413:43: warning: argument ‘var’ might be clobbered by ‘longjmp’ or ‘vfork’ [-Wclobbered]
>   1413 | internal_lisp_condition_case (Lisp_Object var, Lisp_Object bodyform,
>        |                               ~~~~~~~~~~~~^~~
> ...
> 
> 
> In both cases (Lucid, gtk3) the reported warnings don't seem to affect run.
> Thanks

Paul, are those real warnings, or just false alarms?

bug#70568: 30.0.50; build warnings on 'x86_64-pc-linux-gnu'

[Paul Eggert]

On 2024-04-27 02:37, Eli Zaretskii wrote:
> Paul, are those real warnings, or just false alarms?
Hard to say; someone who understands that part of the code would have to 
look at it in more detail.

I typically don't worry about these warnings unless they're generated by 
a default build (no 'configure' options except perhaps 
--enable-gcc-warnings) with the most recent GCC. This is because these 
warnings are usually false alarms these days, except perhaps for 
recently-added code.

On my list of things to do is to get Emacs to build with 
--enable-gcc-warnings and GCC 14, which is sort of out now (it's part of 
Fedora 40).

bug#70568: 30.0.50; build warnings on 'x86_64-pc-linux-gnu'

[Eli Zaretskii]

bug#70568: 30.0.50; build warnings on 'x86_64-pc-linux-gnu'

[Eli Zaretskii]

> Date: Sat, 27 Apr 2024 15:20:36 -0700
> Cc: 70568@debbugs.gnu.org
> From: Paul Eggert <eggert@cs.ucla.edu>
> 
> On 2024-04-27 02:37, Eli Zaretskii wrote:
> > Paul, are those real warnings, or just false alarms?
> Hard to say; someone who understands that part of the code would have to 
> look at it in more detail.

Thanks.  Perhaps Po Lu (CC'ed) can?

bug#70568: 30.0.50; build warnings on 'x86_64-pc-linux-gnu'

[Po Lu via Bug reports for GNU Emacs, the Swiss army knife of text editors]

Eli Zaretskii <eliz@gnu.org> writes:

>> Date: Sat, 27 Apr 2024 15:20:36 -0700
>> Cc: 70568@debbugs.gnu.org
>> From: Paul Eggert <eggert@cs.ucla.edu>
>> 
>> On 2024-04-27 02:37, Eli Zaretskii wrote:
>> > Paul, are those real warnings, or just false alarms?
>> Hard to say; someone who understands that part of the code would have to 
>> look at it in more detail.
>
> Thanks.  Perhaps Po Lu (CC'ed) can?

Definitely a false alarm, if the eassume above wasn't explicit enough.
It's impossible for a display connection not managed by Emacs to enter
functions that call x_display_info_for_display.

bug#70568: 30.0.50; build warnings on 'x86_64-pc-linux-gnu'

[David Ponce via Bug reports for GNU Emacs, the Swiss army knife of text editors]

On 28/04/2024 00:20, Paul Eggert wrote:
> On 2024-04-27 02:37, Eli Zaretskii wrote:
>> Paul, are those real warnings, or just false alarms?
> Hard to say; someone who understands that part of the code would have to look at it in more detail.
> 
> I typically don't worry about these warnings unless they're generated by a default build (no 'configure' options except perhaps --enable-gcc-warnings) with the most recent GCC. This is because these warnings are usually false alarms these days, except perhaps for recently-added code.
> 
> On my list of things to do is to get Emacs to build with --enable-gcc-warnings and GCC 14, which is sort of out now (it's part of Fedora 40).

Since Paul committed changes to pacify GCC 14 warnings, Emacs builds cleanly.
For me this bug can be closed.

Thanks!

bug#70568: 30.0.50; build warnings on 'x86_64-pc-linux-gnu'

[Paul Eggert]

On 5/1/24 00:18, David Ponce wrote:
> Since Paul committed changes to pacify GCC 14 warnings, Emacs builds 
> cleanly.
> For me this bug can be closed.

Thanks for checking; closing the bug report.