You are viewing a plain text version of this content. The canonical link for it is here.
Posted to log4cxx-user@logging.apache.org by Anand Sherkhane <an...@gmail.com> on 2007/06/21 07:38:50 UTC

Crash in log4cxx v0.10.0 - forward_iterator_tag

Hi,

Any updates?

Regards,
Anand.

On 6/20/07, Anand Sherkhane <an...@gmail.com> wrote:
>
> Hi,
>
> I'm seeing a crash in my app that uses log4cxx v0.10.0. Relevant stack
> trace is produced below:
> <snip>
> fee20218 _lwp_kill (6, 0, fa77a410, fee88ef8, 1, fa77a47c) + 8
> fedb6e98 abort (feee1afc, 1bf, feead104, fee88248, 1, fa77a1a5) + 100
> feee1ac4 _ZN10__cxxabiv111__terminateEPFvvE (7178ac, fa779e38, 474e5543,
> 432b2b00, 0, fef099c8) + 4
> feee1afc _ZSt9terminatev (0, 7178ac, feee1ae0, fef099cc, 474e5543,
> 432b2b00) + 1c
> feee1c6c __cxa_throw (fa0748, fef05614, feed65c4, 0, 1, 34c) + 8c
> feeba748 _ZSt19__throw_logic_errorPKc (feee3f98, 1, fa0760, 19089c0,
> bd0368, bd0370) + 6c
> feedb754 _ZNSs12_S_constructIPKcEEPcT_S3_RKSaIcESt20forward_iterator_tag
> (0, ffffffff, fa77a7a8, fa77a6f7, 2710, f71bd080) + 5c
> feed8a44 _ZNSsC1EPKcRKSaIcE (fa77a7a0, 0, fa77a7a8, 800, fa77a3a0,
> fa77a7a8) + 24
> </snip>
>
> Any idea where that points to?
>
> Is it happening because the message being passed to the logger is more
> than it can handle?
>
> By the way: What is the maximum size of message that log4cxx v0.10.0 can
> handle?
>
> Regards,
> Anand.
>
>

Re: Crash in log4cxx v0.10.0 - forward_iterator_tag

Posted by Anand Sherkhane <an...@gmail.com>.
Hi,

Yes, it is possible that 'plainMsg' was NULL.

I checked this behavior in a test program(passed NULL to LogRawMesg) and I
see similar crash.

Thanks for pointing it out.

Regards,
Anand.

On 6/23/07, Curt Arnold <ca...@apache.org> wrote:
>
> On Jun 22, 2007, at 12:34 AM, Anand Sherkhane wrote:
>
> Right. So here I produce some more details.
>
> Stack trace with some more info:
> <snip>
> fee20218 _lwp_kill (6, 0, fa77a410, fee88ef8, 1, fa77a47c) + 8
> fedb6e98 abort (feee1afc, 1bf, feead104, fee88248, 1, fa77a1a5) + 100
> feee1ac4 _ZN10__cxxabiv111__terminateEPFvvE (7178ac, fa779e38, 474e5543,
> 432b2b00, 0, fef099c8) + 4
> feee1afc _ZSt9terminatev (0, 7178ac, feee1ae0, fef099cc, 474e5543,
> 432b2b00) + 1c
> feee1c6c __cxa_throw (fa0748, fef05614, feed65c4, 0, 1, 34c) + 8c
> feeba748 _ZSt19__throw_logic_errorPKc (feee3f98, 1, fa0760, 19089c0,
> bd0368, bd0370) + 6c
> feedb754 _ZNSs12_S_constructIPKcEEPcT_S3_RKSaIcESt20forward_iterator_tag
> (0, ffffffff, fa77a7a8, fa77a6f7, 2710, f71bd080) + 5c
> feed8a44 _ZNSsC1EPKcRKSaIcE (fa77a7a0, 0, fa77a7a8, 800, fa77a3a0,
> fa77a7a8) +
> 24
> fe3a9430 _ZN9ApacheLog10LogRawMesgEiPKc (b641c0, 64, 0, 1bc37f8, fd96eea0,
> 8000) + 644
> f16c8d68 ObAnPluginFn (92bfb8, 13153d0, f41818, fa77ad1c, 7567696e, 9000)
> + 1c3c
> </snip>
>
> ApacheLog is a wrapper class and LogRawMesg is a method of that class.
> Implementation:
> void ApacheLog::LogRawMesg(int level, const char* plainMsg)
> {
>         switch(level) {
>             case LOGFATAL:  LOG4CXX_FATAL(logger, plainMsg ); break;
>             case LOGINFO:   LOG4CXX_INFO(logger, plainMsg ); break;
>             case LOGWARN:   LOG4CXX_WARN(logger, plainMsg ); break;
>             case LOGERROR:  LOG4CXX_ERROR(logger, plainMsg ); break;
>             case LOGDEBUG:  LOG4CXX_DEBUG(logger, plainMsg ); break;
>         }
> }
> ApacheLog::logger is a member variable that is initialized in constructor
> of ApacheLog using Logger::getLogger()
>
> ApacheLog::LogRawMesg() is called from a shared object.
>
> BTW: I checked if length message size was an issue and it doesn't look
> like it is.
>
> Environment details:
> OS: Solaris 9(SPARC)
> Compiler: g++ 3.3.2
> log4cxx v0.10.0
>
> Regards,
> Anand.
>
>
> Any chance that plainMsg is NULL?  The LOG4CXX_ macros will eventually
> result in a call to
>
>         void Logger::forcedLog(const LevelPtr& level, const std::string&
> message,
>                         const log4cxx::spi::LocationInfo& location);
>
> which will result in the transparent construction and destruction of a
> std::string to contain the message.  The std::string constructor would
> likely not like a NULL argument and could die with a stack trace something
> like you are seeing.
>

Re: Crash in log4cxx v0.10.0 - forward_iterator_tag

Posted by Curt Arnold <ca...@apache.org>.
On Jun 22, 2007, at 12:34 AM, Anand Sherkhane wrote:

> Right. So here I produce some more details.
>
> Stack trace with some more info:
> <snip>
> fee20218 _lwp_kill (6, 0, fa77a410, fee88ef8, 1, fa77a47c) + 8
> fedb6e98 abort (feee1afc, 1bf, feead104, fee88248, 1, fa77a1a5) + 100
> feee1ac4 _ZN10__cxxabiv111__terminateEPFvvE (7178ac, fa779e38,  
> 474e5543, 432b2b00, 0, fef099c8) + 4
> feee1afc _ZSt9terminatev (0, 7178ac, feee1ae0, fef099cc, 474e5543,  
> 432b2b00) + 1c
> feee1c6c __cxa_throw (fa0748, fef05614, feed65c4, 0, 1, 34c) + 8c
> feeba748 _ZSt19__throw_logic_errorPKc (feee3f98, 1, fa0760,  
> 19089c0, bd0368, bd0370) + 6c
> feedb754  
> _ZNSs12_S_constructIPKcEEPcT_S3_RKSaIcESt20forward_iterator_tag (0,  
> ffffffff, fa77a7a8, fa77a6f7, 2710, f71bd080) + 5c
> feed8a44 _ZNSsC1EPKcRKSaIcE (fa77a7a0, 0, fa77a7a8, 800, fa77a3a0,  
> fa77a7a8) +
> 24
> fe3a9430 _ZN9ApacheLog10LogRawMesgEiPKc (b641c0, 64, 0, 1bc37f8,  
> fd96eea0, 8000) + 644
> f16c8d68 ObAnPluginFn (92bfb8, 13153d0, f41818, fa77ad1c, 7567696e,  
> 9000) + 1c3c
> </snip>
>
> ApacheLog is a wrapper class and LogRawMesg is a method of that  
> class. Implementation:
> void ApacheLog::LogRawMesg(int level, const char* plainMsg)
> {
>         switch(level) {
>             case LOGFATAL:  LOG4CXX_FATAL(logger, plainMsg ); break;
>             case LOGINFO:   LOG4CXX_INFO(logger, plainMsg ); break;
>             case LOGWARN:   LOG4CXX_WARN(logger, plainMsg ); break;
>             case LOGERROR:  LOG4CXX_ERROR(logger, plainMsg ); break;
>             case LOGDEBUG:  LOG4CXX_DEBUG(logger, plainMsg ); break;
>         }
> }
> ApacheLog::logger is a member variable that is initialized in  
> constructor of ApacheLog using Logger::getLogger()
>
> ApacheLog::LogRawMesg() is called from a shared object.
>
> BTW: I checked if length message size was an issue and it doesn't  
> look like it is.
>
> Environment details:
> OS: Solaris 9(SPARC)
> Compiler: g++ 3.3.2
> log4cxx v0.10.0
>
> Regards,
> Anand.
>

Any chance that plainMsg is NULL?  The LOG4CXX_ macros will  
eventually result in a call to

         void Logger::forcedLog(const LevelPtr& level, const  
std::string& message,
                         const log4cxx::spi::LocationInfo& location);

which will result in the transparent construction and destruction of  
a std::string to contain the message.  The std::string constructor  
would likely not like a NULL argument and could die with a stack  
trace something like you are seeing.

Re: Crash in log4cxx v0.10.0 - forward_iterator_tag

Posted by Anand Sherkhane <an...@gmail.com>.
Right. So here I produce some more details.

Stack trace with some more info:
<snip>
fee20218 _lwp_kill (6, 0, fa77a410, fee88ef8, 1, fa77a47c) + 8
fedb6e98 abort (feee1afc, 1bf, feead104, fee88248, 1, fa77a1a5) + 100
feee1ac4 _ZN10__cxxabiv111__terminateEPFvvE (7178ac, fa779e38, 474e5543,
432b2b00, 0, fef099c8) + 4
feee1afc _ZSt9terminatev (0, 7178ac, feee1ae0, fef099cc, 474e5543, 432b2b00)
+ 1c
feee1c6c __cxa_throw (fa0748, fef05614, feed65c4, 0, 1, 34c) + 8c
feeba748 _ZSt19__throw_logic_errorPKc (feee3f98, 1, fa0760, 19089c0, bd0368,
bd0370) + 6c
feedb754 _ZNSs12_S_constructIPKcEEPcT_S3_RKSaIcESt20forward_iterator_tag (0,
ffffffff, fa77a7a8, fa77a6f7, 2710, f71bd080) + 5c
feed8a44 _ZNSsC1EPKcRKSaIcE (fa77a7a0, 0, fa77a7a8, 800, fa77a3a0, fa77a7a8)
+
24
fe3a9430 _ZN9ApacheLog10LogRawMesgEiPKc (b641c0, 64, 0, 1bc37f8, fd96eea0,
8000) + 644
f16c8d68 ObAnPluginFn (92bfb8, 13153d0, f41818, fa77ad1c, 7567696e, 9000) +
1c3c
</snip>

ApacheLog is a wrapper class and LogRawMesg is a method of that class.
Implementation:
void ApacheLog::LogRawMesg(int level, const char* plainMsg)
{
        switch(level) {
            case LOGFATAL:  LOG4CXX_FATAL(logger, plainMsg ); break;
            case LOGINFO:   LOG4CXX_INFO(logger, plainMsg ); break;
            case LOGWARN:   LOG4CXX_WARN(logger, plainMsg ); break;
            case LOGERROR:  LOG4CXX_ERROR(logger, plainMsg ); break;
            case LOGDEBUG:  LOG4CXX_DEBUG(logger, plainMsg ); break;
        }
}
ApacheLog::logger is a member variable that is initialized in constructor of
ApacheLog using Logger::getLogger()

ApacheLog::LogRawMesg() is called from a shared object.

BTW: I checked if length message size was an issue and it doesn't look like
it is.

Environment details:
OS: Solaris 9(SPARC)
Compiler: g++ 3.3.2
log4cxx v0.10.0

Regards,
Anand.

On 6/21/07, Curt Arnold <ca...@apache.org> wrote:
>
>
> On Jun 21, 2007, at 12:38 AM, Anand Sherkhane wrote:
>
> Hi,
>
> Any updates?
>
> Regards,
> Anand.
>
>
> On 6/20/07, Anand Sherkhane <an...@gmail.com> wrote:
> >
> > Hi,
> >
> > I'm seeing a crash in my app that uses log4cxx v0.10.0. Relevant stack
> > trace is produced below:
> > <snip>
> > fee20218 _lwp_kill (6, 0, fa77a410, fee88ef8, 1, fa77a47c) + 8
> > fedb6e98 abort (feee1afc, 1bf, feead104, fee88248, 1, fa77a1a5) + 100
> > feee1ac4 _ZN10__cxxabiv111__terminateEPFvvE (7178ac, fa779e38, 474e5543,
> > 432b2b00, 0, fef099c8) + 4
> > feee1afc _ZSt9terminatev (0, 7178ac, feee1ae0, fef099cc, 474e5543,
> > 432b2b00) + 1c
> > feee1c6c __cxa_throw (fa0748, fef05614, feed65c4, 0, 1, 34c) + 8c
> > feeba748 _ZSt19__throw_logic_errorPKc (feee3f98, 1, fa0760, 19089c0,
> > bd0368, bd0370) + 6c
> > feedb754 _ZNSs12_S_constructIPKcEEPcT_S3_RKSaIcESt20forward_iterator_tag
> > (0, ffffffff, fa77a7a8, fa77a6f7, 2710, f71bd080) + 5c
> > feed8a44 _ZNSsC1EPKcRKSaIcE (fa77a7a0, 0, fa77a7a8, 800, fa77a3a0,
> > fa77a7a8) + 24
> > </snip>
> >
> > Any idea where that points to?
> >
> > Is it happening because the message being passed to the logger is more
> > than it can handle?
> >
> > By the way: What is the maximum size of message that log4cxx v0.10.0 can
> > handle?
> >
> > Regards,
> > Anand.
> >
> >
>
> Isn't obvious that the stack trace is log4cxx related.  Was next line
> snipped of the stack trace an identifiable line in your code that was making
> a call to log4cxx?
>
> Knowing platform, compiler, et al would be helpful.  But without either a
> stack trace that clearly identifies a location in log4cxx or a test, there
> is not enough to go on.
>
> I don't believe that log4cxx imposes its own limit of message size, so you
> are probably limited to available memory or the max size of
> std::basic_string.
>

Re: Crash in log4cxx v0.10.0 - forward_iterator_tag

Posted by Curt Arnold <ca...@apache.org>.
On Jun 21, 2007, at 12:38 AM, Anand Sherkhane wrote:

> Hi,
>
> Any updates?
>
> Regards,
> Anand.
>

> On 6/20/07, Anand Sherkhane <an...@gmail.com> wrote:
> Hi,
>
> I'm seeing a crash in my app that uses log4cxx v0.10.0. Relevant  
> stack trace is produced below:
> <snip>
> fee20218 _lwp_kill (6, 0, fa77a410, fee88ef8, 1, fa77a47c) + 8
> fedb6e98 abort (feee1afc, 1bf, feead104, fee88248, 1, fa77a1a5) + 100
> feee1ac4 _ZN10__cxxabiv111__terminateEPFvvE (7178ac, fa779e38,  
> 474e5543, 432b2b00, 0, fef099c8) + 4
> feee1afc _ZSt9terminatev (0, 7178ac, feee1ae0, fef099cc, 474e5543,  
> 432b2b00) + 1c
> feee1c6c __cxa_throw (fa0748, fef05614, feed65c4, 0, 1, 34c) + 8c
> feeba748 _ZSt19__throw_logic_errorPKc (feee3f98, 1, fa0760,  
> 19089c0, bd0368, bd0370) + 6c
> feedb754  
> _ZNSs12_S_constructIPKcEEPcT_S3_RKSaIcESt20forward_iterator_tag (0,  
> ffffffff, fa77a7a8, fa77a6f7, 2710, f71bd080) + 5c
> feed8a44 _ZNSsC1EPKcRKSaIcE (fa77a7a0, 0, fa77a7a8, 800, fa77a3a0,  
> fa77a7a8) + 24
> </snip>
>
> Any idea where that points to?
>
> Is it happening because the message being passed to the logger is  
> more than it can handle?
>
> By the way: What is the maximum size of message that log4cxx  
> v0.10.0 can handle?
>
> Regards,
> Anand.
>
>

Isn't obvious that the stack trace is log4cxx related.  Was next line  
snipped of the stack trace an identifiable line in your code that was  
making a call to log4cxx?

Knowing platform, compiler, et al would be helpful.  But without  
either a stack trace that clearly identifies a location in log4cxx or  
a test, there is not enough to go on.

I don't believe that log4cxx imposes its own limit of message size,  
so you are probably limited to available memory or the max size of  
std::basic_string.