Re: [boost] VC++ 7.1 build options lead to AV

From: Ashok Thirumurthi (athirumurthi.news_at_[hidden])
Date: 2005-10-31 21:12:20

> > Ashok Thirumurthi wrote:
> > Dear Boost,
> >
> > The following call stack leads to an access violation when I link a VC
7.1 DLL
>  > to the release build of the boost signals 1_33_0 DLL.  In comparison,
>  > there is no access violation when I link to the debug build of the same
>  > library.  Similarly, when I add the signals source to my own release
>  > DLL, there is no access violation.  I attempted to disable all compiler
>  > optimizations in msvc.jam as follows.  However, after running bjam -a
>  > in signals\build and installing the output, the AV still occurs.
>
> You were commenting out BBv2 code. Boost.Signals is being build with
> BBv1 (if you were building it from the boost root), use "bjam -a --v2"
> to build with BBv2. Also, you can use:
>
>     bjam -a --v2 optimization=off
>
> to turn off the optimization?
>

Thanks.  Disabling all optimizations prevents the access violation.

> > How can I alter the release build to avoid the access violation?
> >
> > Access violation reading location 0x00000004
> > (*((k2).content)).__vfptr is 0x00000000
> >
> >  bool operator()(const any& k1, const any& k2) const
> >   {
> >     if (k1.type() == typeid(front_type))
> >       return !(k2.type() == typeid(front_type));
> >     if (k1.type() == typeid(back_type))
> >       return false;
> >
> >>  if (k2.type() == typeid(front_type))
>
> This looks to be an issue with either Boost.Any or your code. What is
> any::type() returning?

Actually, the () operator is crashing with the access violation.  When I run
without optimizations, k1.type is front_type and k2.type is back_type.  If I
optimize for speed (/O2), k1.content is null and k2.content.__vfptr is null.
So, type() cannot be evaluated.

If I optimize for size (/O1), I observe an invalid lock sequence exception
(0xC000001E) when constructing a _Tree_ptr.

> Have you made sure that RTTI is enabled for everything (including your
> own project -- is there a difference in the vcproj Configurations?)

Yes.

>
> Are you doing something in the Debug version that is not in the Release
> version?
>
> Are you linking with the right variant of the boost libraries? Is there
> a conflict with the run-time library being used?

I can reproduce the problem and the work-around by compiling the boost
source into a DLL using a VC++ 7.1 project.  Initially, I had used the bjam
scripts without any modification after installing the full 1_33 source.  I
was able to reproduce the problem with boost 1_32.

However, I'm not sure how to determine if a conflict exists.  Thanks to your
advice regarding bjam, I have a work-around in hand,

- Ashok

>
> - Reece
>
>
>
> ------------------------------
>
> Message: 4
> Date: Sat, 29 Oct 2005 10:43:59 +0100
> From: "Paul A Bristow" <pbristow_at_[hidden]>
> Subject: Re: [boost] Documentation Wizard
> To: <boost_at_[hidden]>
> Message-ID: <E1EVnFw-0007Bf-Oz_at_[hidden]>
> Content-Type: text/plain; charset="us-ascii"
>
>
>
> | -----Original Message-----
> | From: boost-bounces_at_[hidden]
> | [mailto:boost-bounces_at_[hidden]] On Behalf Of Rob Stewart
> | Sent: 27 October 2005 23:21
> | To: boost_at_[hidden]
> | Subject: [boost] Documentation Wizard
> |
> | Quite some time ago, Dave Abrahams announced that I was the new
> | Boost Documentation Wizard.  I said I'd write something on the
> | subject and that is long overdue.  Herewith, I describe what I
> | think the DocWiz should do and what is needed from volunteers to
> | make it happen.  That is, I'm trying to document the new DocWiz
> | role and solicit your help.
>
> My first impression is that this is a bit over-bureaucratic.  You
obviously
> need this sort of style managing 8 offspring ;-) , but Boosters are all
> volunteers and we have to be greatful for all the help we can get.
>
> Authors are severely handicapped when writing documentation by knowing too
> much.
>
> And they are often exhausted by the process of getting the code to work on
> the myriad of inconsistent platforms and versions.
>
> So _any_ documentation editing is likely to be an improvement.
>
> So I think you should just ask for someone (usually only one - we don't
have
> _common_ tools for handling versioning and changes well) willing to
> contribute to editing each library. The code author and the editor will
have
> to try to agree: only if they are uncertain should theygo out to the wider
> Boost audience.
>
> Since I have the advantage of having English as my first language (and not
> the corrupted version of those renegage ex-colonials ;-), and I have
enough
> knowledge of other languages to greatly admire those whose first language
is
> not English struggling with the erudite esoteric elite Boost discussions,
I
> may offer to edit some documentation.
>
> HTH
>
> Paul
>
> PS And I've read Eats, Shoots and Leaves.
>
> -- 
> Paul A Bristow
> Prizet Farmhouse, Kendal, Cumbria UK LA8 8AB
> Phone and SMS text [phone], Mobile and SMS text [phone]
> mailto: pbristow_at_[hidden]  www.hetp.u-net.com
>
>
>
>
>
> ------------------------------
>
> Message: 5
> Date: Sat, 29 Oct 2005 12:01:29 +0100
> From: "Marcin Kalicinski" <kalita_at_[hidden]>
> Subject: [boost] How to run boost regressions?
> To: boost_at_[hidden]
> Message-ID: <djvkq7$ej5$1_at_[hidden]>
>
> Hi,
>
> How can I run boost regression tests at home?
>
> cheers,
> Marcin
>
>
>
>
>
> ------------------------------
>
> Message: 6
> Date: Sat, 29 Oct 2005 05:57:41 -0700
> From: "SourceForge.net" <noreply_at_[hidden]>
> Subject: [boost] [Boost-bugs] [ boost-Bugs-1341560 ] crach build bjam
> from mingw 3.4.4
> To: noreply_at_[hidden]
> Message-ID: <E1EVqHF-00048m-K1_at_[hidden]>
> Content-Type: text/plain; charset="ISO-8859-1"
>
> Bugs item #1341560, was opened at 2005-10-29 19:57
> Message generated for change (Tracker Item Submitted) made by Item
Submitter
> You can respond by visiting:
>
https://sourceforge.net/tracker/?func=detail&atid=107586&aid=1341560&group_id=7586
>
> Please note that this message will contain a full copy of the comment
thread,
> including the initial issue submission, for this request,
> not just the latest update.
> Category: None
> Group: None
> Status: Open
> Resolution: None
> Priority: 5
> Submitted By: Alexandr Zamaraev (shura_zam)
> Assigned to: Nobody/Anonymous (nobody)
> Summary: crach build bjam from mingw 3.4.4
>
> Initial Comment:
> OS:Windows XP pro rus + sp2
> gcc: (GCC) 3.4.4 (mingw special)
> ld: GNU ld version 2.16.91 20050827
> Source: latest csv
>
> Command line:
> >build.bat mingw
> Output (last 3 line):
> don't know how to make mingw
> ...found 1 target...
> ...can't find 1 target...
>
> ----------------------------------------------------------------------
>
> You can respond by visiting:
>
https://sourceforge.net/tracker/?func=detail&atid=107586&aid=1341560&group_id=7586
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by the JBoss Inc.
> Get Certified Today * Register for a JBoss Training Course
> Free Certification Exam for All Training Attendees Through End of 2005
> Visit http://www.jboss.com/services/certification for more information
> _______________________________________________
> Boost-bugs mailing list
> Boost-bugs_at_[hidden]
> https://lists.sourceforge.net/lists/listinfo/boost-bugs
>
>
> ------------------------------
>
> Message: 7
> Date: Sat, 29 Oct 2005 06:12:26 -0700
> From: "SourceForge.net" <noreply_at_[hidden]>
> Subject: [boost] [Boost-bugs] [ boost-Bugs-1341568 ] crach build bjam
> from mingw 3.4.4 with python
> To: noreply_at_[hidden]
> Message-ID: <E1EVqVW-0005Zb-Rw_at_[hidden]>
> Content-Type: text/plain; charset="ISO-8859-1"
>
> Bugs item #1341568, was opened at 2005-10-29 20:12
> Message generated for change (Tracker Item Submitted) made by Item
Submitter
> You can respond by visiting:
>
https://sourceforge.net/tracker/?func=detail&atid=107586&aid=1341568&group_id=7586
>
> Please note that this message will contain a full copy of the comment
thread,
> including the initial issue submission, for this request,
> not just the latest update.
> Category: None
> Group: None
> Status: Open
> Resolution: None
> Priority: 5
> Submitted By: Alexandr Zamaraev (shura_zam)
> Assigned to: Nobody/Anonymous (nobody)
> Summary: crach build bjam from mingw 3.4.4 with python
>
> Initial Comment:
> OS: Windows XP pro rus + sp2
> gcc: (GCC) 3.4.4 (mingw special)
> ld: GNU ld version 2.16.91 20050827
> Python: Python 2.4.2
> Source: latest csv
>
> I try build bjam with python.
> Bat I have crach in linking.
>
> Command line:
> >bjam -f build.jam --toolset=mingw
> --with-python=C:\Lang\Python24
>
> Output (latest 8 lines)
> C:\TEMP\Shura/ccaSyaaa.o:builtins.c:(.text+0x334e):
> undefined reference to `_imp__PyCallable_Check'
> C:\TEMP\Shura/ccaSyaaa.o:builtins.c:(.text+0x337c):
> undefined reference to `_imp___Py_NoneStruct'
> collect2: ld returned 1 exit status
>
>     "gcc"    -o bin.ntx86\bjam.exe  -DNDEBUG
> -DOPT_HEADER_CACHE_EXT -DOPT_GRAPH_DEBUG_EXT
> -DOPT_SEMAPHORE -DOPT_FIX_TARGET_VARIABLES_EXT
> -DOPT_IMPROVED_PATIENCE_EXT -DNT -DYYSTACKSIZE=5000
> -DHAVE_PYTHON  -s -O3 -finline-functions
> -IC:\Lang\Python24\include
> "C:\Lang\Python24\libs\python24.lib"  command.c
> compile.c debug.c expand.c glob.c hash.c hcache.c
> headers.c hdrmacro.c jam.c jambase.c jamgram.c lists.c
> make.c make1.c newstr.c option.c parse.c regexp.c
> rules.c scan.c search.c subst.c w32_getreg.c
> timestamp.c variable.c modules.c strings.c filesys.c
> builtins.c pwd.c class.c native.c modules/set.c
> modules/path.c modules/regex.c modules/property-set.c
> modules/sequence.c modules/order.c execnt.c filent.c
> pathunix.c
>
> ...failed [COMPILE] bin.ntx86\bjam.exe...
> ...failed updating 1 target...
>
> ----------------------------------------------------------------------
>
> You can respond by visiting:
>
https://sourceforge.net/tracker/?func=detail&atid=107586&aid=1341568&group_id=7586
>
>
> -------------------------------------------------------
> This SF.Net email is sponsored by the JBoss Inc.
> Get Certified Today * Register for a JBoss Training Course
> Free Certification Exam for All Training Attendees Through End of 2005
> Visit http://www.jboss.com/services/certification for more information
> _______________________________________________
> Boost-bugs mailing list
> Boost-bugs_at_[hidden]
> https://lists.sourceforge.net/lists/listinfo/boost-bugs
>
>
> ------------------------------
>
> Message: 8
> Date: Sat, 29 Oct 2005 09:40:28 -0500
> From: Douglas Gregor <doug.gregor_at_[hidden]>
> Subject: Re: [boost] How to run boost regressions?
> To: boost_at_[hidden]
> Message-ID: <66D002ED-12FC-40F7-A0AC-BD3AE3A8CA0C_at_[hidden]>
> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
>
>
> On Oct 29, 2005, at 6:01 AM, Marcin Kalicinski wrote:
> > How can I run boost regression tests at home?
>
> There are some instructions here:
>
>      http://www.boost.org/tools/regression/xsl_reports/runner/
> instructions.html
>
>      Doug
>
>
>
> ------------------------------
>
> Message: 9
> Date: Sat, 29 Oct 2005 09:51:08 -0500
> From: Douglas Gregor <doug.gregor_at_[hidden]>
> Subject: Re: [boost] [GCC 2.95.3] Boost regression notification
> (2005-10-28 [HEAD])
> To: boost_at_[hidden]
> Message-ID: <C805154E-5675-42B4-A0D9-2F42C8920BF1_at_[hidden]>
> Content-Type: text/plain; charset=US-ASCII; delsp=yes; format=flowed
>
>
> On Oct 28, 2005, at 5:11 PM, Beman Dawes wrote:
> > A lot of these failures are for gcc-2.95.3-linux. Do we care enough
> > about
> > 2.95.3 nowadays to consider it a release platform?
>
> I vageuely remember trying to kill GCC 2.95.3 support at one point,
> but was convinced otherwise. We could just mark it as unusable on new
> libraries/tests that support it but retain backward compatibility on
> those libraries that do.
>
>      Doug
>
>
> ------------------------------
>
> Message: 10
> Date: Sat, 29 Oct 2005 17:04:16 +0100
> From: "John Maddock" <john_at_[hidden]>
> Subject: Re: [boost] [range][1.33.1] last few regressions.
> To: <boost_at_[hidden]>
> Message-ID: <015001c5dca2$7118cab0$6f2e0d52_at_fuji>
> Content-Type: text/plain; format=flowed; charset="iso-8859-1";
> reply-type=original
>
> > In that case we might as well just roll-back the changes from the RC
> > branch to
> > the main branch.
>
> There haven't been any changes to the test itself, and the only changes in
> the headers appear to be:
>
http://cvs.sourceforge.net/viewcvs.py/boost/boost/boost/range/end.hpp?r1=1.18&r2=1.18.2.1&only_with_tag=RC_1_33_0
> and related changes.
>
> I'd still rather see this fixed than just rolled back (I presume these
> changes were in there for a reason?), or we could disable them for broken
> compilers if that works (but we need to know why they were in there in the
> first place, this could be a fix for VC6/7 for all I know).
>
> BTW I don't have a Boost source tree (or compiler) to work with at present
> to test this out, since I've just replaced my hard drive!
>
> Hopefully this narrows the problem down a bit?
>
> John.
>
>
>
> ------------------------------
>
> Message: 11
> Date: Sat, 29 Oct 2005 12:24:53 -0400
> From: David Abrahams <dave_at_[hidden]>
> Subject: Re: [boost] [GCC 2.95.3] Boost regression notification
> (2005-10-28 [HEAD])
> To: boost_at_[hidden]
> Message-ID: <uy84cl296.fsf_at_[hidden]>
> Content-Type: text/plain; charset=us-ascii
>
> Douglas Gregor <doug.gregor_at_[hidden]> writes:
>
> > On Oct 28, 2005, at 5:11 PM, Beman Dawes wrote:
> >> A lot of these failures are for gcc-2.95.3-linux. Do we care enough
> >> about
> >> 2.95.3 nowadays to consider it a release platform?
> >
> > I vageuely remember trying to kill GCC 2.95.3 support at one point,
> > but was convinced otherwise. We could just mark it as unusable on new
> > libraries/tests that support it but retain backward compatibility on
>                       ^
>                       don't??
>
> > those libraries that do.
>
> If not "don't," how will we know we're retaining backward
> compatibility?
>
> -- 
> Dave Abrahams
> Boost Consulting
> www.boost-consulting.com
>
>
>
> ------------------------------
>
> _______________________________________________
> Unsubscribe &amp; other changes:
http://lists.boost.org/mailman/listinfo.cgi/boost
>
> End of Boost Digest, Vol 1260, Issue 1
> **************************************
>