[Bug c++/42814] New: gcc segfaults on gch

[Bug c++/42814] New: gcc segfaults on gch

[JamesMikeDuPont at googlemail dot com]

/usr/local/bin/g++ -DHAVE_CONFIG_H -I. -I.. -I../ -DHAVE_CONFIG_H
-DMAPNIK_DEBUGinclude -ggdb -O1 -DDEBUG -pg -DMAPNIK_DEBUG
-I/usr/include/libpng12 -I/usr/include/freetype2
-I/media/NewFoundSpace/2010/01/mapnik/agg/include -I/usr/local/include
-I/usr/include/libxml2 -D_REENTRANT -I/usr/include/cairomm-1.0
-I/usr/include/cairo -I/usr/include/sigc++-2.0 -I/usr/lib/sigc++-2.0/include
-I/usr/include/pixman-1 -I/usr/include/freetype2 -I/usr/include/directfb
-I/usr/include/libpng12 -I../include -save-temps -MT
libmapnik_la-filter_factory.lo -MD -MP -MF
.deps/libmapnik_la-filter_factory.Tpo -c src/filter_factory.ii 
g++: Internal error: Killed (program cc1plus)
Please submit a full bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.


-- 
           Summary: gcc segfaults on gch
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: JamesMikeDuPont at googlemail dot com


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[paolo dot carlini at oracle dot com]

------- Comment #1 from paolo dot carlini at oracle dot com  2010-01-20 19:11 -------
Please provide all the required information, in particular a pre-processed
file. See this page for details:

  http://gcc.gnu.org/bugs/#report


-- 

paolo dot carlini at oracle dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[rguenth at gcc dot gnu dot org]

------- Comment #2 from rguenth at gcc dot gnu dot org  2010-01-20 19:13 -------
Also make sure to include the PCH from the toplevel source file and not from
some header.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[JamesMikeDuPont at googlemail dot com]

------- Comment #3 from JamesMikeDuPont at googlemail dot com  2010-01-20 19:17 -------
Created an attachment (id=19667)
 --> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=19667&action=view)
precompiled headers

/usr/local/bin/g++ -DHAVE_CONFIG_H -I. -I.. -I../ -DHAVE_CONFIG_H
-DMAPNIK_DEBUGinclude -ggdb -O1 -DDEBUG -pg -DMAPNIK_DEBUG
-I/usr/include/libpng12 -I/usr/include/freetype2
-I/media/NewFoundSpace/2010/01/mapnik/agg/include -I/usr/local/include
-I/usr/include/libxml2 -D_REENTRANT -I/usr/include/cairomm-1.0
-I/usr/include/cairo -I/usr/include/sigc++-2.0 -I/usr/lib/sigc++-2.0/include
-I/usr/include/pixman-1 -I/usr/include/freetype2 -I/usr/include/directfb
-I/usr/include/libpng12 -I../include -save-temps -MT
libmapnik_la-filter_factory.lo -MD -MP -MF
.deps/libmapnik_la-filter_factory.Tpo -c src/filter_factory.ii 
g++: Internal error: Killed (program cc1plus)
Please submit a full bug report.
See <http://gcc.gnu.org/bugs.html> for instructions.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[JamesMikeDuPont at googlemail dot com]

------- Comment #4 from JamesMikeDuPont at googlemail dot com  2010-01-20 19:27 -------
Delta compression using up to 4 threads.
Compressing objects: 100% (3/3), done.
Writing objects: 100% (3/3), 326 bytes, done.
Total 3 (delta 2), reused 0 (delta 0)
To git@github.com:h4ck3rm1k3/MapNickAutotools.git
   1ae752a..6803b1b  master -> master


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[JamesMikeDuPont at googlemail dot com]

------- Comment #5 from JamesMikeDuPont at googlemail dot com  2010-01-20 19:41 -------
Sorry that is not the precompiled header, but the preprocessed,
all you need to get the crash i hope.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[paolo dot carlini at oracle dot com]

------- Comment #6 from paolo dot carlini at oracle dot com  2010-01-20 19:48 -------
We are still missing the SVN Rev of the GCC 4.5 you are using and how you built
it.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[rguenth at gcc dot gnu dot org]

------- Comment #7 from rguenth at gcc dot gnu dot org  2010-01-20 20:01 -------
Which GCC version are you using?  The preprocessed source does not use
a precompiled header.  Compiling the file works ok for me with GCC 4.4.


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 GCC target triplet|                            |i?86-*-*


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[JamesMikeDuPont at googlemail dot com]

------- Comment #8 from JamesMikeDuPont at googlemail dot com  2010-01-20 20:05 -------
got it again after make clean,
maybe it ran out of memory:
mdupont@introspector-desktop:/media/NewFoundSpace/2010/01/mapnik$ gunzip
src/filter_factory.ii.gz
mdupont@introspector-desktop:/media/NewFoundSpace/2010/01/mapnik$
/usr/local/bin/g++ -DHAVE_CONFIG_H -I. -I.. -I../ -DHAVE_CONFIG_H
-DMAPNIK_DEBUGinclude -ggdb -O1 -DDEBUG -pg -DMAPNIK_DEBUG
-I/usr/include/libpng12 -I/usr/include/freetype2
-I/media/NewFoundSpace/2010/01/mapnik/agg/include -I/usr/local/include
-I/usr/include/libxml2 -D_REENTRANT -I/usr/include/cairomm-1.0
-I/usr/include/cairo -I/usr/include/sigc++-2.0 -I/usr/lib/sigc++-2.0/include
-I/usr/include/pixman-1 -I/usr/include/freetype2 -I/usr/include/directfb
-I/usr/include/libpng12 -I../include -save-temps -MT
libmapnik_la-filter_factory.lo -MD -MP -MF
.deps/libmapnik_la-filter_factory.Tpo -c src/filter_factory.ii 
g++: Internal error: Killed (program cc1plus)
Please submit a full bug report.
See <http://gcc.gnu.org/bugs.html> for instructions


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[JamesMikeDuPont at googlemail dot com]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1091 bytes --]



------- Comment #9 from JamesMikeDuPont at googlemail dot com  2010-01-20 20:09 -------
Subject: Re:  gcc segfaults on gch

/usr/local/bin/c++ --version
c++ (GCC) 4.5.0 20100113 (experimental)
Copyright (C) 2010 Free Software Foundation, Inc.

On Wed, Jan 20, 2010 at 9:01 PM, rguenth at gcc dot gnu dot org
<gcc-bugzilla@gcc.gnu.org> wrote:
>
>
> ------- Comment #7 from rguenth at gcc dot gnu dot org  2010-01-20 20:01 -------
> Which GCC version are you using?  The preprocessed source does not use
> a precompiled header.  Compiling the file works ok for me with GCC 4.4.
>
>
> --
>
> rguenth at gcc dot gnu dot org changed:
>
>           What    |Removed                     |Added
> ----------------------------------------------------------------------------
>  GCC target triplet|                            |i?86-*-*
>
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814
>
> ------- You are receiving this mail because: -------
> You reported the bug, or are watching the reporter.
>


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[rguenth at gcc dot gnu dot org]

------- Comment #10 from rguenth at gcc dot gnu dot org  2010-01-20 20:47 -------
With unoptimized trunk I see us blowing the stack during garbage-collecting.
Note that we have a load of templates in this testcase and we walk them in
unfortunate order.

(gdb) 
#79368 0x0835b9fb in ggc_collect ()
    at /home/richard/src/trunk/gcc/ggc-page.c:1962
1962      ggc_mark_roots ();


ugh.  We reach stuff from the function decl type.

400                   gt_ggc_m_9tree_node ((*x).generic.type.size_unit);
(gdb) 
#79358 0x0828fdc9 in gt_ggc_mx_lang_tree_node (x_p=0xb772f378)
    at ./gt-cp-tree.h:163
163                   gt_ggc_m_9tree_node ((*x).generic.int_cst.common.type);
(gdb) 
#79357 0x0829106b in gt_ggc_mx_lang_tree_node (x_p=0xb7743000)
    at ./gt-cp-tree.h:398
398                   gt_ggc_m_9tree_node ((*x).generic.type.values);

hmm ...

#79349 0x0829115c in gt_ggc_mx_lang_tree_node (x_p=0xa83ff180)
    at ./gt-cp-tree.h:417
417                   gt_ggc_m_9tree_node ((*x).generic.type.name);
(gdb) 

#79348 0x08290cb9 in gt_ggc_mx_lang_tree_node (x_p=0xa8400138)
    at ./gt-cp-tree.h:355
355                   gt_ggc_m_9tree_node
((*x).generic.type_decl.common.common.common.common.common.context);
(gdb) 
#79347 0x0829106b in gt_ggc_mx_lang_tree_node (x_p=0xa83ff000)
    at ./gt-cp-tree.h:398
398                   gt_ggc_m_9tree_node ((*x).generic.type.values);
(gdb) 
#79346 0x08290229 in gt_ggc_mx_lang_tree_node (x_p=0xa84003a8)
    at ./gt-cp-tree.h:228
228                   gt_ggc_m_9tree_node
((*x).generic.decl_non_common.common.common.common.common.common.type);
(gdb) 
#79345 0x0829121c in gt_ggc_mx_lang_tree_node (x_p=0xa83ff420)
    at ./gt-cp-tree.h:425
425                   gt_ggc_m_9lang_type ((*x).generic.type.lang_specific);
(gdb) 
#79344 0x08292154 in gt_ggc_mx_lang_type (x_p=0xa83fdaa0) at ./gt-cp-tree.h:713
713               gt_ggc_m_9tree_node ((*x).u.c.template_info);


hmm - we reach templates via integer constants ...


-- 

rguenth at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |NEW
     Ever Confirmed|0                           |1
   Last reconfirmed|0000-00-00 00:00:00         |2010-01-20 20:47:22
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[JamesMikeDuPont at googlemail dot com]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 3945 bytes --]



------- Comment #11 from JamesMikeDuPont at googlemail dot com  2010-01-20 20:54 -------
Subject: Re:  gcc segfaults on gch

so I am thinking about a way to wrap the existing templates that I use.
there must be a way to reduce the size of the gch by specifing a list
of classes we use. that would reduce the build size.

this boost lib takes forever to process.
We only use %1 of it at a time. There has got to be a way to
precompile all these templates. A nice fine grained, for each header,
external pch database for stl and boost would make working with them
fun again.

I have been experimenting with grouping the headers into groups of
common dependencies. Basically I made a type level dependency tree
that is sliced up into pools of common types and topologically sorted
so that we can compile the the gch files in the right order and
include the previous gch.

I would consider putting some time into helping make the gch better if
you give me direction.

mike

On Wed, Jan 20, 2010 at 9:47 PM, rguenth at gcc dot gnu dot org
<gcc-bugzilla@gcc.gnu.org> wrote:
>
>
> ------- Comment #10 from rguenth at gcc dot gnu dot org  2010-01-20 20:47 -------
> With unoptimized trunk I see us blowing the stack during garbage-collecting.
> Note that we have a load of templates in this testcase and we walk them in
> unfortunate order.
>
> (gdb)
> #79368 0x0835b9fb in ggc_collect ()
>    at /home/richard/src/trunk/gcc/ggc-page.c:1962
> 1962      ggc_mark_roots ();
>
>
> ugh.  We reach stuff from the function decl type.
>
> 400                   gt_ggc_m_9tree_node ((*x).generic.type.size_unit);
> (gdb)
> #79358 0x0828fdc9 in gt_ggc_mx_lang_tree_node (x_p=0xb772f378)
>    at ./gt-cp-tree.h:163
> 163                   gt_ggc_m_9tree_node ((*x).generic.int_cst.common.type);
> (gdb)
> #79357 0x0829106b in gt_ggc_mx_lang_tree_node (x_p=0xb7743000)
>    at ./gt-cp-tree.h:398
> 398                   gt_ggc_m_9tree_node ((*x).generic.type.values);
>
> hmm ...
>
> #79349 0x0829115c in gt_ggc_mx_lang_tree_node (x_p=0xa83ff180)
>    at ./gt-cp-tree.h:417
> 417                   gt_ggc_m_9tree_node ((*x).generic.type.name);
> (gdb)
>
> #79348 0x08290cb9 in gt_ggc_mx_lang_tree_node (x_p=0xa8400138)
>    at ./gt-cp-tree.h:355
> 355                   gt_ggc_m_9tree_node
> ((*x).generic.type_decl.common.common.common.common.common.context);
> (gdb)
> #79347 0x0829106b in gt_ggc_mx_lang_tree_node (x_p=0xa83ff000)
>    at ./gt-cp-tree.h:398
> 398                   gt_ggc_m_9tree_node ((*x).generic.type.values);
> (gdb)
> #79346 0x08290229 in gt_ggc_mx_lang_tree_node (x_p=0xa84003a8)
>    at ./gt-cp-tree.h:228
> 228                   gt_ggc_m_9tree_node
> ((*x).generic.decl_non_common.common.common.common.common.common.type);
> (gdb)
> #79345 0x0829121c in gt_ggc_mx_lang_tree_node (x_p=0xa83ff420)
>    at ./gt-cp-tree.h:425
> 425                   gt_ggc_m_9lang_type ((*x).generic.type.lang_specific);
> (gdb)
> #79344 0x08292154 in gt_ggc_mx_lang_type (x_p=0xa83fdaa0) at ./gt-cp-tree.h:713
> 713               gt_ggc_m_9tree_node ((*x).u.c.template_info);
>
>
> hmm - we reach templates via integer constants ...
>
>
> --
>
> rguenth at gcc dot gnu dot org changed:
>
>           What    |Removed                     |Added
> ----------------------------------------------------------------------------
>             Status|WAITING                     |NEW
>     Ever Confirmed|0                           |1
>   Last reconfirmed|0000-00-00 00:00:00         |2010-01-20 20:47:22
>               date|                            |
>
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814
>
> ------- You are receiving this mail because: -------
> You reported the bug, or are watching the reporter.
>


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814

[Bug c++/42814] gcc segfaults on gch

[rguenth at gcc dot gnu dot org]

------- Comment #12 from rguenth at gcc dot gnu dot org  2010-01-20 21:09 -------
(In reply to comment #11)
> Subject: Re:  gcc segfaults on gch
> 
> so I am thinking about a way to wrap the existing templates that I use.
> there must be a way to reduce the size of the gch by specifing a list
> of classes we use. that would reduce the build size.
> 
> this boost lib takes forever to process.
> We only use %1 of it at a time. There has got to be a way to
> precompile all these templates. A nice fine grained, for each header,
> external pch database for stl and boost would make working with them
> fun again.
> 
> I have been experimenting with grouping the headers into groups of
> common dependencies. Basically I made a type level dependency tree
> that is sliced up into pools of common types and topologically sorted
> so that we can compile the the gch files in the right order and
> include the previous gch.

You can at most include a single gch file.

> I would consider putting some time into helping make the gch better if
> you give me direction.

I think the issue is the following enum:

  template<class _Sp, class _Tp>
    struct __traitor
    {
      enum { __value = bool(_Sp::__value) || bool(_Tp::__value) };


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42814