[ 
https://issues.apache.org/jira/browse/AXIS2C-1472?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12922591#action_12922591
 ] 

Ivan Pechorin commented on AXIS2C-1472:
---------------------------------------

Hello

It seems like this particular change results in double-freeing of freed memory. 
I hit it every time when processing a request with wrong URI (such as request 
to non-existent or misspelled service).

The first time is when we free fault_ctx->soap_envelope (thanks to this patch), 
here is the call stack:

>       axiom.dll!axiom_soap_envelope_free(axiom_soap_envelope * 
> soap_envelope=0x03a51020, const axutil_env * env=0x03a48478) Line 155  C 
        axis2_engine.dll!axis2_msg_ctx_free(axis2_msg_ctx * msg_ctx=0x03a52de0, 
const axutil_env * env=0x03a48478) 
msg_ctx.c:426 
        axis2_engine.dll!axis2_http_worker_process_request(axis2_http_worker * 
http_worker=0x03a47700, const axutil_env * env=0x03a48478, 
axis2_simple_http_svr_conn * svr_conn=0x03a484d0, axis2_http_simple_request * 
simple_request=0x03a48580) 
http_worker.c:1076 
        axis2_http_receiver.dll!axis2_svr_thread_worker_func(axutil_thread_t * 
thd=0x03a48130, void * data=0x03a480e0) Line 270 + 0x15 bytes    C 
        axutil.dll!dummy_worker(void * opaque=0x03a48130) Line 88       C 


The second time we free the same soap envelope as msg_ctx->fault_soap_envelope 
when called from http_worker.c, line 1830:

        else
        {
                /* cases like HEAD, WSDL */
                axis2_msg_ctx_free(msg_ctx, env);
        }

Here is the corresponding call stack:

>       axiom.dll!axiom_soap_envelope_free(axiom_soap_envelope * 
> soap_envelope=0x03a51020, const axutil_env * env=0x03a48478) Line 155  C 
        axis2_engine.dll!axis2_msg_ctx_free(axis2_msg_ctx * msg_ctx=0x03a49e78, 
const axutil_env * env=0x03a48478) Line 434     C 
msg_ctx.c:431 
        axis2_engine.dll!axis2_http_worker_process_request(axis2_http_worker * 
http_worker=0x03a47700, const axutil_env * env=0x03a48478, 
axis2_simple_http_svr_conn * svr_conn=0x03a484d0, axis2_http_simple_request * 
simple_request=0x03a48580) Line 1833    C 
http_worker.cpp:1830 
        axis2_http_receiver.dll!axis2_svr_thread_worker_func(axutil_thread_t * 
thd=0x03a48130, void * data=0x03a480e0) Line 270 + 0x15 bytes    C 
        axutil.dll!dummy_worker(void * opaque=0x03a48130) Line 88       C 


I had to revert this particular change from revision #961258 in my working 
copy. And there are no problems with double freeing anymore after reverting.

Details of my platform: 
 - compiler: Visual C++ 2010, 32-bit release build
 - OS: Windows 7 64-bit

P.S. If there is such need, I'll try to reproduce the problem on other 
platforms, such as Linux, Solaris or HP-UX.

> in http_worker.c function axis2_http_worker_process_request(): if status == 
> AXIS2_FAILURE then engine and fault_ctx are not freed.
> ----------------------------------------------------------------------------------------------------------------------------------
>
>                 Key: AXIS2C-1472
>                 URL: https://issues.apache.org/jira/browse/AXIS2C-1472
>             Project: Axis2-C
>          Issue Type: Bug
>          Components: core/transport
>    Affects Versions: 1.7.0
>         Environment: linux
>            Reporter: Robert Springer
>            Priority: Minor
>             Fix For: 1.7.0
>
>
> in http_worker.c function axis2_http_worker_process_request(): if status == 
> AXIS2_FAILURE then engine and fault_ctx are not freed:
>                  axis2_http_status_line_free(tmp_stat_line, env);
>                  tmp_stat_line = NULL;
>              }
> +            axis2_engine_free(engine, env);
> +
> +            if (fault_ctx)
> +            {
> +                axis2_msg_ctx_free(fault_ctx, env);
> +            }            
>          }
>      }
>      else

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to