Re: [Boost-users] Boost Serlialization on XScale problem

Subject: Re: [Boost-users] Boost Serlialization on XScale problem
From: lee nookx (lnookx_at_[hidden])
Date: 2008-09-29 04:15:25

Thanks for the try Robert, but that still gives the same problem...

I suspect that its a deep seated libc or compiler problem, but the
serialization libraries are the first thing to manifest this problem.
weird.

On Mon, Sep 29, 2008 at 12:52 AM, Robert Ramey <ramey_at_[hidden]> wrote:
> I doubt its a problem with the library but try this:
>
> {    std::ifstream ifs("filename");
>    {
>        boost::archive::text_iarchive ia(ifs); // <--- causes a segfault.
>    }
> }
>
> platforms which don't destroy everything in the reverse sequence
> of creation would have problems which would be addressed
> by this change.
>
> Robert Ramey
>
>
>
> lee nookx wrote:
>> Hi,
>>
>> I am having a strange problem with the Boost Serilalizartion library
>> on an XScale platform. I am using v1.34.1 of the libraries and I'm
>> seeing segfaults when creating an archive object, but only under some
>> circumstances.
>>
>> The problem occurs in several different projects, but I've been able
>> to reproduce it using the most simple case from the tutorial
>> (http://www.boost.org/doc/libs/1_36_0/libs/serialization/doc/index.html).
>> The problem that I see is that the code segfaults at these lines, and
>> a core dump shows this error is occuring down on malloc_trim():
>>
>>    std::ifstream ifs("filename");
>>    boost::archive::text_iarchive ia(ifs); // <--- causes a segfault.
>>
>>
>> However, if I move the "std::ifstream ifs("filename");" further up the
>> code (next to the initialisation of the ofs object, then the code
>> works fine.
>>
>> Has anyone else seen any problems like this before?
>>
>> We have been using Boost::Thread, and Boost::Date-Time without any
>> problems for a while, so I believe that this may just be a
>> serialization library problem. I am also going to give v1.36.0 a try
>> to see if it's something subtle that's been fixed already.
>>
>> (I have a core dump that I can make available if anyone needs it....)
>>
>>
>>
>> --lee
>
>
>
> _______________________________________________
> Boost-users mailing list
> Boost-users_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-users
>