[Bug debug/61013] New: Option parsing difference between < 4.9 and 4.9

[Bug debug/61013] New: Option parsing difference between < 4.9 and 4.9

[andres at anarazel dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

            Bug ID: 61013
           Summary: Option parsing difference between < 4.9 and 4.9
           Product: gcc
           Version: 4.9.1
            Status: UNCONFIRMED
          Severity: minor
          Priority: P3
         Component: debug
          Assignee: unassigned at gcc dot gnu.org
          Reporter: andres at anarazel dot de

In gcc 4.8 a binary compiled with "gcc -g3 ... -g" would include the extended
debug information (e.g. macros), while in gcc 4.9 the second -g seems to lower
the debug level.
That's obviously not a critical issue, but it's annoying enough because several
buildsystems add -g internally, often after the user supplied CFLAGS, making it
harder to build with a sufficient amount of debuginfo.

[Bug debug/61013] Option parsing difference between < 4.9 and 4.9

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |INVALID

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
-g is the same as -g2 and the later option is supposed to override the first
one. Jus like how -O is handled.

[Bug debug/61013] Option parsing difference between < 4.9 and 4.9

[andres at anarazel dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #2 from Andres Freund <andres at anarazel dot de> ---
Hi,

On 2014-04-30 14:54:20 +0000, pinskia at gcc dot gnu.org wrote:
> -g is the same as -g2 and the later option is supposed to override the first
> one. Jus like how -O is handled.

The point is that this has changed between 4.8 and 4.9... And I don't
see anything relevant in http://gcc.gnu.org/gcc-4.9/changes.html

Greetings,

Andres Freund

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
   Last reconfirmed|                            |2014-04-30
                 CC|                            |ccoutant at gcc dot gnu.org,
                   |                            |jakub at gcc dot gnu.org
         Resolution|INVALID                     |---
            Summary|Option parsing difference   |[4.9/4.10 Regression]
                   |between < 4.9 and 4.9       |Option parsing difference
                   |                            |between < 4.9 and 4.9
     Ever confirmed|0                           |1

--- Comment #3 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
It has changed with r205235, and I think that part of the change is
undesirable, -g never stood for -g2 before, it stood for enable debug info, at
whatever level is default or has been previously selected.

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[pinskia at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
It was not on accident, see
http://gcc.gnu.org/ml/gcc-patches/2013-02/msg00260.html and
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02077.html 

And even where I said http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02078.html 

This was all discussed on the list and there was no objections.

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[andres at anarazel dot de]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #5 from Andres Freund <andres at anarazel dot de> ---
Hi,

On 2014-04-30 15:48:33 +0000, pinskia at gcc dot gnu.org wrote:
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013
> 
> --- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
> It was not on accident, see
> http://gcc.gnu.org/ml/gcc-patches/2013-02/msg00260.html and
> http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02077.html 
> 
> And even where I said http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02078.html 
> 
> This was all discussed on the list and there was no objections.

Meh. At the very least you should document such changes in the release
notes. I'd be surprised if I am the only one that wasted time on
debugging this change.

Greetings,

Andres Freund

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I certainly haven't noticed that discussion, if I did, I would object already
by that time.

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[ccoutant at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #7 from Cary Coutant <ccoutant at gcc dot gnu.org> ---
(In reply to Andres Freund from comment #2)
> The point is that this has changed between 4.8 and 4.9... And I don't
> see anything relevant in http://gcc.gnu.org/gcc-4.9/changes.html

Yes, you're right. This change should have been documented there. Sorry!

I did ask twice for consensus, and got no objections either time.

Our build system adds -g1 to the compile options, before user-supplied options,
and if a user adds -g, it's surprising to only get -g1. I wonder if it would be
reasonable for -g to set the debug level to max(2, previous level)? I still
think the simplicity of -g === -g2 is much better, and it's also much better to
be consistent with the -O option.

What should the following combinations do?

-g1 -g
-g1 -g0 -g
-g3 -g
-g3 -g0 -g

-cary

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #8 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
I don't see why there should be any consistency with -O, it is a very different
option, with a very different usage and history.
The 4.8 behavior was that -g set debug level to 2 if the debug level was 0, so
-g1 -g used to be the same as -g1
-g1 -g0 -g used to be the same as -g2
-g3 -g used to be the same as -g3
-g3 -g0 -g used to be the same as -g2
Now, if you want to change a default for your builds, I'd say you'd just tweak
specs so that -g1 is provided if no -g appears on the command line; either
that can be done by changing the default specs, or you simply add a short specs
file which will do that and change say CC to gcc -specs=whatever.

E.g. in Fedora we use:
-specs=/usr/lib/rpm/redhat/redhat-hardened-cc1
and the specs file ensures that -fPIE is supplied by default if no other option
is used on the command line:
*cc1_options:
+ %{!fpie:%{!fPIE:%{!fpic:%{!fPIC:%{!fno-pic:-fPIE}}}}}

So, my preference would be to revert to the 4.8 and older behavior, or if there
really is consensus that -g1 -g should mean -g2 rather than -g1, at least
change
it so that -g3 -g means -g3 (so revert your change and for *arg == '\0' instead
of the 4.8:
      if (!opts->x_debug_info_level)
        opts->x_debug_info_level = DINFO_LEVEL_NORMAL;
do:
      if (opts->x_debug_info_level < DINFO_LEVEL_NORMAL)
        opts->x_debug_info_level = DINFO_LEVEL_NORMAL;

What I'd say would be helpful would be add support for inline specs overrides
which you could specify on the command line rather than having to resort to
loading a file.  So -specsinline='*cc1_options:\n+
%{!fpie:%{!fPIE:%{!fpic:%{!fPIC:%{!fno-pic:-fPIE}}}}}' or so.

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[rth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

Richard Henderson <rth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |rth at gcc dot gnu.org

--- Comment #9 from Richard Henderson <rth at gcc dot gnu.org> ---
(In reply to Jakub Jelinek from comment #8)
> So, my preference would be to revert to the 4.8 and older behavior, or if
> there really is consensus that -g1 -g should mean -g2 rather than -g1, at
> least change it so that -g3 -g means -g3 (so revert your change and
> for *arg == '\0' instead of the 4.8:
>       if (!opts->x_debug_info_level)
>         opts->x_debug_info_level = DINFO_LEVEL_NORMAL;
> do:
>       if (opts->x_debug_info_level < DINFO_LEVEL_NORMAL)
>         opts->x_debug_info_level = DINFO_LEVEL_NORMAL;

I agree on both points.

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[ccoutant at google dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #10 from ccoutant at google dot com ---
>> So, my preference would be to revert to the 4.8 and older behavior, or if
>> there really is consensus that -g1 -g should mean -g2 rather than -g1, at
>> least change it so that -g3 -g means -g3 (so revert your change and
>> for *arg == '\0' instead of the 4.8:
>>       if (!opts->x_debug_info_level)
>>         opts->x_debug_info_level = DINFO_LEVEL_NORMAL;
>> do:
>>       if (opts->x_debug_info_level < DINFO_LEVEL_NORMAL)
>>         opts->x_debug_info_level = DINFO_LEVEL_NORMAL;
>
> I agree on both points.

Sorry, I'm not sure what "both points" are. Does that mean that you
would support changing -g so that -g1 -g means -g2, but -g3 -g means
-g3? (I.e., -g will raise the level to 2 but will not lower it.)

That seems reasonable to me, and it would support both build scenarios
mentioned above (Andres' and mine). It'll leave the meaning of -g3 -g
the same as 4.8, but change the meaning of -g1 -g (which shouldn't be
much of a problem since everyone here seemed to think that -g1 usage
was rare).

-cary

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[tromey at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

Tom Tromey <tromey at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |tromey at gcc dot gnu.org

--- Comment #11 from Tom Tromey <tromey at gcc dot gnu.org> ---
FWIW this regresses a few gdb tests.  It's easy to fix the
gdb test suite, but if this is going to be fixed before the
next gcc release, I'd rather not bother.  Any word on that?

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[ccoutant at google dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #12 from ccoutant at google dot com ---
> FWIW this regresses a few gdb tests.  It's easy to fix the
> gdb test suite, but if this is going to be fixed before the
> next gcc release, I'd rather not bother.  Any word on that?

I'm planning to fix it as proposed above, and backport to the 4.9
branch -- I was just waiting to hear back from rth.

-cary

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[rth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #13 from Richard Henderson <rth at gcc dot gnu.org> ---
(In reply to ccoutant from comment #10)
> Sorry, I'm not sure what "both points" are. Does that mean that you
> would support changing -g so that -g1 -g means -g2, but -g3 -g means
> -g3? (I.e., -g will raise the level to 2 but will not lower it.)

Yes, that's what I meant.

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[ccoutant at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #14 from Cary Coutant <ccoutant at gcc dot gnu.org> ---
Author: ccoutant
Date: Wed May 14 21:48:47 2014
New Revision: 210442

URL: http://gcc.gnu.org/viewcvs?rev=210442&root=gcc&view=rev
Log:
Change -g so that it will override -g1 but not -g3.

gcc/
        PR debug/61013
    * opts.c (common_handle_option): Don't special-case "-g".
    (set_debug_level): Default to at least level 2 with "-g".

Modified:
    trunk/gcc/ChangeLog
    trunk/gcc/opts.c

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[ccoutant at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

--- Comment #15 from Cary Coutant <ccoutant at gcc dot gnu.org> ---
Author: ccoutant
Date: Thu May 15 00:34:20 2014
New Revision: 210456

URL: http://gcc.gnu.org/viewcvs?rev=210456&root=gcc&view=rev
Log:
Change -g so that it will override -g1 but not -g3.

Backported from trunk r210442.

gcc/
        PR debug/61013
    * opts.c (common_handle_option): Don't special-case "-g".
    (set_debug_level): Default to at least level 2 with "-g".

Modified:
    branches/gcc-4_9-branch/gcc/ChangeLog
    branches/gcc-4_9-branch/gcc/opts.c

[Bug debug/61013] [4.9/4.10 Regression] Option parsing difference between < 4.9 and 4.9

[ccoutant at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61013

Cary Coutant <ccoutant at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|REOPENED                    |RESOLVED
         Resolution|---                         |FIXED
           Assignee|unassigned at gcc dot gnu.org      |ccoutant at gcc dot gnu.org

--- Comment #16 from Cary Coutant <ccoutant at gcc dot gnu.org> ---
Fixed on trunk and on 4.9 branch.