[Bug preprocessor/51391] New: pragma GCC diag ignored then warning of Winline activates -fno-inline

[Bug preprocessor/51391] New: pragma GCC diag ignored then warning of Winline activates -fno-inline

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

             Bug #: 51391
           Summary: pragma GCC diag ignored then warning of Winline
                    activates -fno-inline
    Classification: Unclassified
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: preprocessor
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: pnewell@cs.cmu.edu


In the test program I will be submitting after the bug is in bugzilla, I am
using -Winline with g++ -g. Everything works until I try to disable the check
in a file with #pragma GCC diagnostic {ignored,warning} "-Winline" ... then it
complains that it can't inline as -fno-inline is active ... which it isn't (per
what I can see).

The two files I have included are inlineBug.cpp and inlineBug.h. The offending
pragmas are in inlineBug.cpp. I am sending the file with them active. I think
the problem is with the second one as it seems that it doesn't matter whether I
ask it to ignore, its the re-activation as warning that seems to do it.

The command I am using to compile is: g++ -g -Winline -o inlineBug
inlineBug.cpp

I have also attached the output of a bad make/build

This is being run on a Fedora14 x86_64 system with their latest yum update of
4.5.1 which g++ -v give as "gcc version 4.5.1 20100924 (Red Hat 4.5.1-4) (GCC)"

As this is the first bug I have submitted to gcc bugzilla, please let me know
if I have forgotten anything that I need to tell you.

And, of course, let me know if I can run any more tests to help resolve this
(assuming that it is a bug and not already fixed -- I didn't see anything when
searching under "Winline fno-inline")

Thanks,
Paul

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #1 from pnewell at cs dot cmu.edu 2011-12-03 03:10:05 UTC ---
Created attachment 25973
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25973
this is the source file

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #2 from pnewell at cs dot cmu.edu 2011-12-03 03:10:37 UTC ---
Created attachment 25974
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25974
this is the header file

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #3 from pnewell at cs dot cmu.edu 2011-12-03 03:11:11 UTC ---
Created attachment 25975
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=25975
this is the bad output I am getting

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #4 from pnewell at cs dot cmu.edu 2011-12-03 03:13:49 UTC ---
Forgot to mention ... in my original discovery, I was seeing it occurring when
inlines were inside inlines. That is what I submitted, I don't know if it is
part of the problem

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #5 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-03 03:14:27 UTC ---
The warnings are correct.  You declared some functions as inline but they are
not being inlined because optimization is not turned on (that is -fno-inline).

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #6 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-03 03:15:48 UTC ---
inlineBug.h:16:8: warning: function ‘inlineBug::inlineBug()’ can never be
inlined because it is suppressed using -fno-inline

Without any -On (or -O or -Os), -fno-inline is the default.  The warnings are
correct based on what you are asking of the compiler.

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #7 from pnewell at cs dot cmu.edu 2011-12-03 03:18:50 UTC ---
Wow ... I never picked that up when trying to find the answer in the online GCC
docs. Thanks for the advice.

That being said, why would it work when I didn't have the pragma active inside
the code. It seems that changing the state of the warnings and changing back is
giving me a different result

Thanks,
Paul

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #8 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-03 03:24:26 UTC ---
Because there is code to explicitly turn off the warning for -O0 but not when
using the pragma:
      /* Inlining does not work if not optimizing,
         so force it not to be done.  */
      opts->x_warn_inline = 0;
      opts->x_flag_no_inline = 1;

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #9 from pnewell at cs dot cmu.edu 2011-12-03 03:29:03 UTC ---
I just tested on my Linux box and I am getting the results you suggest for -O2
and -O3. It is finding issues with -O, -Os, and -O1, but I think that is my
problem (barking on the inlined destructor).

However, -O0 still gives me the -fno-inline warnings

I just got your email about "Because there is code to explicitly turn off the
warning for -O0 but not when using the pragma". I've never seen any
documentation about the pragma GCC diagnostic warning having different actions
than just on the compile line. Is there any more info on this?

Please pardon my questioning this, but does that seem right? I would expect
same behavior between a compile line -Winline and a pragma in the code stating
"#pragma GCC diagnostic warning "-Winline"?

Thanks,
Paul

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #10 from pnewell at cs dot cmu.edu 2011-12-03 03:31:06 UTC ---
off-topic, but I added myself to the cc-list but don't get copies of my own
submissions, is that the way gcc bugzilla behaves (would be different than
redhat's for fedora)?

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #11 from Andrew Pinski <pinskia at gcc dot gnu.org> 2011-12-03 03:32:50 UTC ---
(In reply to comment #10)
> off-topic, but I added myself to the cc-list but don't get copies of my own
> submissions, is that the way gcc bugzilla behaves (would be different than
> redhat's for fedora)?

That is the way it acts by default though you can change your preferences.

[Bug preprocessor/51391] pragma GCC diag ignored then warning of Winline activates -fno-inline

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #12 from pnewell at cs dot cmu.edu 2011-12-03 03:36:51 UTC ---
changed preferences, thanks

[Bug c/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[manu at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

Manuel López-Ibáñez <manu at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2011-12-03
                 CC|                            |manu at gcc dot gnu.org
          Component|preprocessor                |c
            Summary|pragma GCC diag ignored     |Differences between setting
                   |then warning of Winline     |Winline in command-line and
                   |activates -fno-inline       |through pragma GCC
                   |                            |diagnostic
     Ever Confirmed|0                           |1

--- Comment #13 from Manuel López-Ibáñez <manu at gcc dot gnu.org> 2011-12-03 10:59:21 UTC ---
As Andrew explained, this is because the pragma uses handle_option, but
command-line options go through a different code path that sets those defaults. 

I am pretty sure there are many more cases of this. The short-term fix is to
duplicate that logic also in common_handle_option in a new case for
OPT_Winline. The long term fix is to encode the dependencies in .opt files and
let the awk scripts to generate the necessary code to handle all cases. But my
awk skills are not so good.

The fix is probably a one-line change, plus testcase and changelog (but of
course, one needs to build gcc and regression test the patches, so don't expect
anyone else to fix this anytime soon).

[Bug c/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[manu at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #14 from Manuel López-Ibáñez <manu at gcc dot gnu.org> 2011-12-03 11:04:48 UTC ---
This patch fixes the issue:

Index: gcc/opts.c
===================================================================
--- gcc/opts.c  (revision 180166)
+++ gcc/opts.c  (working copy)
@@ -1423,10 +1423,20 @@ common_handle_option (struct gcc_options

       enable_warning_as_error (arg, value, lang_mask, handlers,
                               opts, opts_set, loc, dc);
       break;

+    case OPT_Winline:
+      if (opts->x_optimize == 0)
+       {
+         /* Inlining does not work if not optimizing,
+            so force it not to be done.  */
+         opts->x_warn_inline = 0;
+         opts->x_flag_no_inline = 1;
+       }
+      break;
+
     case OPT_Wlarger_than_:
       opts->x_larger_than_size = value;
       opts->x_warn_larger_than = value != -1;
       break;


Feel free to take it and prepare it for submission:

http://gcc.gnu.org/contribute.html#patches

[Bug c/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #15 from pnewell at cs dot cmu.edu 2011-12-03 19:10:39 UTC ---
Manu:

Yes, Andrew did a great job of letting me know what the issue was and gave me
the necessary information to have a work-around within 1/2 hour. Fantastic
turnaround in my book, many thanks to him!

I appreciate the confirm that it is a bug. I can also believe that if they go
through different code paths there is a good chance there are more of these
"opps" situations. I'd rather not use #pragmas, but with third party code it
becomes necessary to get rid of all the warnings one can't do anything about
(except tell the third party ... and that hasn't produced any results yet).

I am not surprised that the fix is quick but the build/testing makes it a bit
of a question whether someone is going to want to do all that legwork. It isn't
exactly a show-stopper (smile) and, since I am stuck on F14 until spring of
2012, I wouldn't be able to make use of the patch until I upgrade. But the fix
is there (thanks) and I'll just see what happens.

Mostly, I wanted to tell both Andrew and you that I very impressed with how
this bug was handled. A very prompt response, solid info on what the issue is,
and an honest assessment of its chances for fixing. This is the first bug I
have submitted to gcc bugzilla and I certainly feel that it was worth the
effort for me to package up an example to send to you.

Thanks,
Paul

[Bug c/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

Richard Guenther <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hubicka at gcc dot gnu.org,
                   |                            |jsm28 at gcc dot gnu.org
           Severity|normal                      |enhancement

--- Comment #16 from Richard Guenther <rguenth at gcc dot gnu.org> 2011-12-05 09:49:20 UTC ---
I think a more appropriate "fix" today is to drop the x_flag_no_inline
setting (it should not be necessary these days) but keep ignoring
-Winline at -O0 (there is probably a better central place to do this,
supposedly at the point we emit the warning, instead of in the option
code).  In fact, the warning

inlineBug.h:16:8: warning: function ‘inlineBug::inlineBug()’ can never be
inlined because it is suppressed using -fno-inline

does not seem useful to me at all (see tree-inline.c:tree_inlinable_function_p,
warning in a predicate seems questionable to me at best, warning during
expand_call_inline isn't better either.  A pass over remaining inline-declared
edges after IPA inline transform sounds more appropriate to me, also enabled
when IPA inline is disabled itself)

[Bug c/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[pnewell at cs dot cmu.edu]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #17 from pnewell at cs dot cmu.edu 2011-12-06 08:00:42 UTC ---
I am a bit afraid to do this, but I think I need to do it for my own education.

I saw the comment by Richard (#16) and have to admit that I would like a
translation into something I might understand. Its sorta like being with a
lawyer and going "Can you please speak English?". I understand that this is
internal conversation, but as the reporter of the bug I would really like to
comment on how gcc wishes to deal with it. Users may not get the big picture,
but I think anyone who reports a bug really wants to understand what the
process is.

I really appreciate (in advance) your understanding,
Thanks,
Paul

[Bug middle-end/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[pnewell0705 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

pnewell0705 at gmail dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pnewell0705 at gmail dot com

--- Comment #18 from pnewell0705 at gmail dot com ---
To "Pinskia" and/or whoever:

The last entry on this issue was in 2011. Tonight, in 2021, I have just gotten
5 emails that imply "something is happening". I open the bugzilla and there are
no new comments. I do a bit of checking and "something has changed that I
wasn't aware of ... I am suddenly no longer on the cc-list. With all due
respect, I was back in 2021.

I am re-adding myself to the cc list but need to make the request that I get an
update as to what the flurry is tonight.

I also note that every email gotten tonight is a "change in status" with no
comments 

My thanks in advance,
Paul

[Bug middle-end/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[pnewell0705 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #19 from pnewell0705 at gmail dot com ---
To "Pinskia" and/or whoever:

The last entry on this issue was in 2011. Tonight, in 2021, I have just gotten
5 emails that imply "something is happening". I open the bugzilla and there are
no new comments. I do a bit of checking and "something has changed that I
wasn't aware of ... I am suddenly no longer on the cc-list. With all due
respect, I was back in 2021.

I am re-adding myself to the cc list but need to make the request that I get an
update as to what the flurry is tonight.

I also note that every email gotten tonight is a "change in status" with no
comments 

My thanks in advance,
Paul

[Bug middle-end/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #20 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to pnewell0705 from comment #19)
> To "Pinskia" and/or whoever:
> 
> The last entry on this issue was in 2011. Tonight, in 2021, I have just
> gotten 5 emails that imply "something is happening". I open the bugzilla and
> there are no new comments. I do a bit of checking and "something has changed
> that I wasn't aware of ... I am suddenly no longer on the cc-list. With all
> due respect, I was back in 2021.

So I was looking into this and it was actually fixed for GCC 4.9.0 (which was
released back in 2014. I was fixing the up the attachments for easier access
and moving the Component to a more reasonible one.  I was just looking for the
patch which would have fixed the bug too.

[Bug middle-end/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[pnewell0705 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #21 from pnewell0705 at gmail dot com ---
Andrew Pinski:

Thank you for the reply so I understand what happened

Best,
Paul

[Bug middle-end/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |4.9.0
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #22 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #20)
> So I was looking into this and it was actually fixed for GCC 4.9.0 (which
> was released back in 2014. I was fixing the up the attachments for easier
> access and moving the Component to a more reasonible one.  I was just
> looking for the patch which would have fixed the bug too.

So this was fixed by r0-125277-gb631d45ac3d9d5 .

[Bug middle-end/51391] Differences between setting Winline in command-line and through pragma GCC diagnostic

[pnewell0705 at gmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51391

--- Comment #23 from pnewell0705 at gmail dot com ---
To Andrew Pinski:

Thank you for the additional post about the fix

Paul