[Bug c++/59821] __builtin_LINE and __builtin_FILE for new'd objects is wrong

[Bug c++/59821] __builtin_LINE and __builtin_FILE for new'd objects is wrong

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59821

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |wrong-debug
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-01-15
     Ever confirmed|0                           |1

--- Comment #1 from Richard Biener <rguenth at gcc dot gnu.org> ---
    [t.C : 12:3] Foo::Foo ([t.C : 12] &D.2242, 12);
    [t.C : 12:42] try
      {
        [t.C : 12:3] D.2247 = D.2242.line;
        [t.C : 12:3] D.2244 = 5;
        [t.C : 12:3] D.2243 = operator new (4);
        [t.C : 12:3] try
          {
            [t.C : 12:3] Foo::Foo (D.2243, D.2244);

looks like location info is broken for the new case.

[Bug c++/59821] __builtin_LINE and __builtin_FILE for new'd objects is wrong

[jason at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59821

Jason Merrill <jason at gcc dot gnu.org> changed:

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

--- Comment #3 from Jason Merrill <jason at gcc dot gnu.org> ---
I think gimplify_arg clobbering a valid EXPR_LOCATION is wrong; both cases
should give the location of the default argument definition.

Aldy, what was the rationale for that change (r140917)?

[Bug c++/59821] __builtin_LINE and __builtin_FILE for new'd objects is wrong

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59821

--- Comment #4 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
While I agree with that, for _builtin_{LINE,FILE,FUNCTION} (), those were added
specifically for the use in default arguments and were I think from the start
meant to give you the location of the call, not of the token where it appears,
otherwise those builtins don't make much sense (you could use __LINE__,
__FILE__
or __FUNCTION__ instead).

[Bug c++/59821] __builtin_LINE and __builtin_FILE for new'd objects is wrong

[aldyh at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59821

--- Comment #5 from Aldy Hernandez <aldyh at gcc dot gnu.org> ---
(In reply to Jason Merrill from comment #3)
> I think gimplify_arg clobbering a valid EXPR_LOCATION is wrong; both cases
> should give the location of the default argument definition.
> 
> Aldy, what was the rationale for that change (r140917)?

Unfortunately this was by design to make multi-line non-trivial arguments play
nice with the debugger.  Perhaps it's time to revisit the ``is_stmt'' DWARF
flag for this?

More details here:

http://gcc.gnu.org/ml/gcc-patches/2008-10/msg00191.html

[Bug c++/59821] __builtin_LINE and __builtin_FILE for new'd objects is wrong

[jason at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59821

Jason Merrill <jason at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |jason at gcc dot gnu.org

[Bug c++/59821] __builtin_LINE and __builtin_FILE for new'd objects is wrong

[jason at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59821

--- Comment #7 from Jason Merrill <jason at gcc dot gnu.org> ---
Author: jason
Date: Thu Jan 16 19:55:12 2014
New Revision: 206686

URL: http://gcc.gnu.org/viewcvs?rev=206686&root=gcc&view=rev
Log:
    PR c++/59821
    * tree.c (bot_manip): Update the location of builtin_LINE and
    builtin_FILE calls.

Added:
    trunk/gcc/testsuite/g++.dg/ext/builtin-line1.C
Modified:
    trunk/gcc/cp/ChangeLog
    trunk/gcc/cp/tree.c
    trunk/gcc/doc/extend.texi

[Bug c++/59821] __builtin_LINE and __builtin_FILE for new'd objects is wrong

[jason at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59821

Jason Merrill <jason at gcc dot gnu.org> changed:

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

--- Comment #8 from Jason Merrill <jason at gcc dot gnu.org> ---
Fixed for 4.9.

[Bug c++/59821] __builtin_LINE and __builtin_FILE for new'd objects is wrong

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=59821

--- Comment #9 from Richard Biener <rguenth at gcc dot gnu.org> ---
Yeah - sorry for this.  As it "worked" for the testcases I tested I was
assuming
that the locations for the default argument computation (!) stmts should be
that of the call, not that of the definition.  Appearantly that isn't so.

Thanks for fixing it.