[Bug libstdc++/55041] New: prettyprinting/shared_ptr & cxx11 fails on some platforms

[Bug libstdc++/55041] New: prettyprinting/shared_ptr & cxx11 fails on some platforms

[bkoz at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

             Bug #: 55041
           Summary: prettyprinting/shared_ptr & cxx11 fails on some
                    platforms
    Classification: Unclassified
           Product: gcc
           Version: 4.8.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: libstdc++
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: bkoz@gcc.gnu.org


I'm seeing fails on some linux in prettyprinting, but the logs look ok, and
there are no real changes in the primary sources as far as I can tell.

See:
http://gcc.gnu.org/ml/libstdc++/2012-10/msg00149.html

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #1 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-10-23 20:54:38 UTC ---
(on my phone so can't check...)

If the failures are in unordered_set it's almost certainly
http://gcc.gnu.org/ml/libstdc++/2012-10/msg00147.html and the printers need to
go via _M_h

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #2 from Paolo Carlini <paolo.carlini at oracle dot com> 2012-10-23 21:11:11 UTC ---
Nope it must be something else, I saw the failures on some machines already
many days ago (sorry for not having filed a PR, I thought it was my fault, some
sort of config glitch, like defective Python or something)

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

Paolo Carlini <paolo.carlini at oracle dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2012-10-24
     Ever Confirmed|0                           |1

--- Comment #3 from Paolo Carlini <paolo.carlini at oracle dot com> 2012-10-24 10:43:02 UTC ---
Sorry, *only* the shared_ptr fails are relatively old, to my best knowledge.
That makes sense.

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #4 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-10-24 11:21:21 UTC ---
if the recent changes to unordered_set *didn't* break the printers that would
have been surprising ;)

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[paolo.carlini at oracle dot com]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #5 from Paolo Carlini <paolo.carlini at oracle dot com> 2012-10-24 11:58:31 UTC ---
Yes ;) No idea what's going on with shared_ptr.

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #6 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-10-28 12:54:01 UTC ---
I think the reason this doesn't show up everywhere is due to a problem with
older versions of GDB.  Using GDB 7.3.50.20110722-16.fc16 I get this in the
test logs:

48362.gdb:5: Error in sourced command file:^M
No symbol "t1" in current context.^M
skipping: 48362.gdb:5: Error in sourced command file:^M
skipping: No symbol "t1" in current context.^M
UNSUPPORTED: libstdc++-prettyprinters/48362.cc

and I've had the same problem running gdb myself, with gdb saying "No symbol x
in current context" for all local variables, when they clearly do exist!

I also see GDB 7.3 completely confused about the type of std::string:

Temporary breakpoint 1, main () at d.cc:6
6         std::string s1 = "string 1";
(gdb) n
7         std::string s2 = "string 2";
(gdb) p s1
$1 = {i = {1431655765, -1077586603}, x = -0.16666666666666666}
(gdb) ptype s1
type = const union {
    int4 i[2];
    double x;
}

Huh?!

Using GDB from the git mirror or upgrading to rawhide's 7.5.0.20120926-25.fc18
makes the problem go away and now I see the FAILs when running
libstdc++-prettyprinters.exp (which I'm fixing now)

This only happens for code built with trunk, so there seems to be some problem
with trunk and GDB 7.3 - any ideas?

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #7 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-10-28 13:20:38 UTC ---
Author: redi
Date: Sun Oct 28 13:20:31 2012
New Revision: 192894

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=192894
Log:
    PR libstdc++/55041
    * python/libstdcxx/v6/printers.py (Tr1UnorderedMapPrinter): Update
    to handle hashtable as member of unordered_map not base class.
    (Tr1UnorderedSetPrinter): Likewise.

Modified:
    trunk/libstdc++-v3/ChangeLog
    trunk/libstdc++-v3/python/libstdcxx/v6/printers.py

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #8 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-10-28 13:27:20 UTC ---
The shared_ptr tests fail because GDB is getting the variable's type wrong,
seeing it as the base class not the correct type:

(gdb) p sp1
$1 = {<std::__shared_ptr<int, (__gnu_cxx::_Lock_policy)2>> = {_M_ptr =
0x12345678, _M_refcount = {_M_pi = 0x604010}}, <No data fields>}
(gdb) p wp1
$2 = {<std::__weak_ptr<int, (__gnu_cxx::_Lock_policy)2>> = {_M_ptr =
0x12344321, _M_refcount = {_M_pi = 0x604040}}, <No data fields>}
(gdb) p wp2
$3 = {<std::__weak_ptr<int, (__gnu_cxx::_Lock_policy)2>> = {_M_ptr =
0x56788765, _M_refcount = {_M_pi = 0x604070}}, <No data fields>}
(gdb) p/r wp2
$4 = {<std::__weak_ptr<int, (__gnu_cxx::_Lock_policy)2>> = {_M_ptr =
0x56788765, _M_refcount = {_M_pi = 0x604070}}, <No data fields>}
(gdb) p/r sp1
$5 = {<std::__shared_ptr<int, (__gnu_cxx::_Lock_policy)2>> = {_M_ptr =
0x12345678, _M_refcount = {_M_pi = 0x604010}}, <No data fields>}


This doesn't look like a libstdc++ problem.

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #9 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-10-28 13:37:31 UTC ---
Sorry, I misread that output, it does know the right type, it just doesn't
match the printer for some reason

(gdb) ptype sp1
type = class std::shared_ptr<int> : public std::__shared_ptr<int,
(__gnu_cxx::_Lock_policy)2> {
  public:
    void shared_ptr(void);
    void shared_ptr(const std::shared_ptr<int> &);
    void shared_ptr(<unknown type in /dev/shm/a.out, CU 0x0, DIE 0x127f>);
    void shared_ptr(std::nullptr_t);
    std::shared_ptr<int> & operator=(const std::shared_ptr<int> &);
    std::shared_ptr<int> & operator=(<unknown type in /dev/shm/a.out, CU 0x0,
DIE 0x12e3>);
    void shared_ptr<int, deleter>(int *, deleter);
}

(I'll file a separate bug about the "unknown type" error for the shared_ptr&&
type)

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #10 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-10-28 13:57:00 UTC ---
(In reply to comment #9)
> (I'll file a separate bug about the "unknown type" error for the shared_ptr&&
> type)

That's http://sourceware.org/PR14441

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #11 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-11-16 18:14:02 UTC ---
Tom, do you have any idea what's going on in comment 6 and comment 8 of this
bug?

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[tromey at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #12 from Tom Tromey <tromey at gcc dot gnu.org> 2012-11-16 20:46:33 UTC ---
(In reply to comment #11)
> Tom, do you have any idea what's going on in comment 6 and comment 8 of this
> bug?

Not offhand.
If you send me the failing executable(s) I can take a look at their
debuginfo.  Maybe that will be useful.

I don't think gdb 7.3 really works with trunk after this patch, though:

commit 2dd133d84c8991217864523ea79d11f1885f1388
Author: mark <mark@138bc75d-0d04-0410-961f-82ee72b054a4>
Date:   Mon Aug 20 20:27:46 2012 +0000

    dwarf2out.c: For DWARF 4+, output DW_AT_high_pc as constant offset.


This required a gdb change.

Another thing to try is turning on complaints before running the program
in gdb:

(gdb) set complaints 10000
(gdb) start

For me this prints tons of junk, some of which could actually be gcc
bugs:

DW_AT_low_pc 0x0 is zero for DIE at 0xf055a [in module
/home/tromey/Space/Trunk/Git/build/x86_64-unknown-linux-gnu/./libstdc++-v3/src/.libs/libstdc++.so.6]...

debug_line address at offset 0x449b7 is 0 [in module
/home/tromey/Space/Trunk/Git/build/x86_64-unknown-linux-gnu/./libstdc++-v3/src/.libs/libstdc++.so.6]

It's hard to say without debugging further, this is just a first glance.
I'm not sure if these could cause the problems you're seeing.

FWIW, the tests mostly work for me, except:

UNSUPPORTED: libstdc++-prettyprinters/shared_ptr.cc

In my case the issue seems to be weird debuginfo emitted by gcc;
look at what the breakpoint reports:

Breakpoint 1, _GLOBAL__sub_I__Z4makem () at
/home/tromey/Space/Trunk/Git/gcc/lib
stdc++-v3/testsuite/libstdc++-prettyprinters/shared_ptr.cc:81

Setting it by hand with "gdb shared_ptr.exe" then "break 72":

(gdb) b 72
Breakpoint 1 at 0x400deb:
/home/tromey/Space/Trunk/Git/gcc/libstdc++-v3/testsuite/libstdc++-prettyprinters/shared_ptr.cc:72.
(3 locations)
(gdb) info b
Num     Type           Disp Enb Address            What
1       breakpoint     keep y   <MULTIPLE>         
1.1                         y     0x0000000000400deb in main() 
                                                   at
/home/tromey/Space/Trunk/Git/gcc/libstdc++-v3/testsuite/libstdc++-prettyprinters/shared_ptr.cc:72
1.2                         y     0x0000000000400ec0 in
_GLOBAL__sub_I__Z4makem() 
                                                   at
/home/tromey/Space/Trunk/Git/gcc/libstdc++-v3/testsuite/libstdc++-prettyprinters/shared_ptr.cc:72
1.3                         y     0x0000000000401070 in
std::_Sp_counted_deleter<int*, deleter, std::allocator<int>,
(__gnu_cxx::_Lock_policy)2>::_M_get_deleter(std::type_info const&) 
                                                   at
/home/tromey/Space/Trunk/Git/gcc/libstdc++-v3/testsuite/libstdc++-prettyprinters/shared_ptr.cc:72


... very weird, looks like a gcc bug to me.

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

Jonathan Wakely <redi at gcc dot gnu.org> changed:

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

--- Comment #13 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-11-17 00:47:40 UTC ---
(In reply to comment #12)
> (In reply to comment #11)
> > Tom, do you have any idea what's going on in comment 6 and comment 8 of this
> > bug?
> 
> Not offhand.
> If you send me the failing executable(s) I can take a look at their
> debuginfo.  Maybe that will be useful.

OK, will do so asap

> I don't think gdb 7.3 really works with trunk after this patch, though:
> 
> commit 2dd133d84c8991217864523ea79d11f1885f1388
> Author: mark <mark@138bc75d-0d04-0410-961f-82ee72b054a4>
> Date:   Mon Aug 20 20:27:46 2012 +0000
> 
>     dwarf2out.c: For DWARF 4+, output DW_AT_high_pc as constant offset.
> 
> 
> This required a gdb change.

Ah, that should be in http://gcc.gnu.org/gcc-4.8/changes.html then. Mark?

> ... very weird, looks like a gcc bug to me.

Thanks, I thought it might be but don't know enough to confirm it or know where
to start looking.

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[mark at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #14 from Mark Wielaard <mark at gcc dot gnu.org> 2012-11-17 13:27:03 UTC ---
>>     dwarf2out.c: For DWARF 4+, output DW_AT_high_pc as constant offset.
>> 
>> 
>> This required a gdb change.
>
> Ah, that should be in http://gcc.gnu.org/gcc-4.8/changes.html then. Mark?

How about something like:

diff -u -r1.57 changes.html
--- htdocs/gcc-4.8/changes.html    14 Nov 2012 04:45:41 -0000    1.57
+++ htdocs/gcc-4.8/changes.html    17 Nov 2012 13:24:24 -0000
@@ -53,6 +53,14 @@
 <h2>General Optimizer Improvements (and Changes)</h2>

   <ul>
+    <li>DWARF4 is now the default when generating DWARF debug information.
+      When -g is used on a platform that uses DWARF debugging information,
+      GCC will now default to -gdwarf-4 -fno-debug-types-section.
+      GDB 7.5, Valgrind 3.8.0 and elfutils 0.154 debug information consumers
+      support DWARF4 by default. Before GCC 4.8 the default version used
+      was DWARF2. To make gcc 4.8 generate an older DWARF version use -g
+      together with -gdwarf-2 or -gdwarf-3.
+    </li>
     <li>A new general optimization level, <code>-Og</code>, has been
       introduced.  It addresses the need for fast compilation and a
       superior debugging experience while providing a reasonable level

On which list should website changes be proposed?

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[redi at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #15 from Jonathan Wakely <redi at gcc dot gnu.org> 2012-11-20 00:10:01 UTC ---
(In reply to comment #14)
> [...]

Looks good to me, thanks!

> On which list should website changes be proposed?

On gcc-patches, I usually just commit it and put [wwwdocs] in the subject line,
which gets Gerald's attention.

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[tromey at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

--- Comment #16 from Tom Tromey <tromey at gcc dot gnu.org> 2013-01-24 18:50:58 UTC ---
(In reply to comment #12)

> In my case the issue seems to be weird debuginfo emitted by gcc;
> look at what the breakpoint reports:
> 
> Breakpoint 1, _GLOBAL__sub_I__Z4makem () at
> /home/tromey/Space/Trunk/Git/gcc/lib
> stdc++-v3/testsuite/libstdc++-prettyprinters/shared_ptr.cc:81


FWIW this problem doesn't appear with a more recent trunk gcc.
I tried one from Jan 17 (c1f9696864b1b8ab20a6798751eec48986a2e98a)
just now.

Are there still unresolved problems in this bug?

[Bug libstdc++/55041] prettyprinting/shared_ptr & cxx11 fails on some platforms

[bkoz at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55041

Benjamin Kosnik <bkoz at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED

--- Comment #17 from Benjamin Kosnik <bkoz at gcc dot gnu.org> 2013-01-24 18:54:51 UTC ---

Thanks for the ping Tom. Confirmed: let's go ahead and close this.