Memory leak in gss_init_sec_context when using spnego mechanism

[Markus Moeller]

Did anybody have a chance to confirm my finding ?

Thank you
Markus

"Markus Moeller" <huaraz at moeller.plus.com> wrote in message 
news:f7lpqd$tmm$1 at sea.gmane.org...
>I attach my test program source with the valgrind results.  As it might not 
>get through here is the main part of my test source
>
> static gss_OID_desc _gss_mech_spnego  = {6, (void 
> *)"\x2b\x06\x01\x05\x05\x02"};
> gss_OID gss_mech_spnego = &_gss_mech_spnego;
>
> const char *create_token(int spnego, const char *server) {
> OM_uint32 major_status, minor_status;
> gss_ctx_id_t          gss_context = GSS_C_NO_CONTEXT;
> gss_name_t            server_name = GSS_C_NO_NAME;
> gss_buffer_desc       service = GSS_C_EMPTY_BUFFER;
> gss_buffer_desc       input_token = GSS_C_EMPTY_BUFFER;
> gss_buffer_desc       output_token = GSS_C_EMPTY_BUFFER;
> const char   *token = NULL;
>
>  setbuf(stdout,NULL);
>  setbuf(stdin,NULL);
>
>  if (!server) {
>     fprintf(stderr, "%s| %s: Error: No server name\n", LogTime(), 
> PROGRAM);
>     return NULL;
>  }
>
>  service.value = malloc(strlen("HTTP")+strlen(server)+2);
> 
> snprintf(service.value,strlen("HTTP")+strlen(server)+2,"%s@%s","HTTP",server);
>  service.length = strlen((char *)service.value);
>
>  major_status = gss_import_name(&minor_status, &service,
>                                 gss_nt_service_name, &server_name);
>
>  if (check_gss_err(major_status,minor_status,"gss_import_name()") )
>      goto cleanup;
>  if (spnego) {
>  major_status = gss_init_sec_context(&minor_status,
>                                      GSS_C_NO_CREDENTIAL,
>                                      &gss_context,
>                                      server_name,
>                                      gss_mech_spnego,
>                                      0,
>                                      0,
>                                      GSS_C_NO_CHANNEL_BINDINGS,
>                                      &input_token,
>                                      NULL,
>                                      &output_token,
>                                     NULL,
>                                      NULL);
>  } else {
>  major_status = gss_init_sec_context(&minor_status,
>                                      GSS_C_NO_CREDENTIAL,
>                                      &gss_context,
>                                      server_name,
>                                      GSS_C_NO_OID,
>                                      0,
>                                      0,
>                                      GSS_C_NO_CHANNEL_BINDINGS,
>                                      &input_token,
>                                      NULL,
>                                      &output_token,
>                                      NULL,
>                                      NULL);
>  }
>  if (check_gss_err(major_status,minor_status,"gss_init_sec_context()") )
>      goto cleanup;
>
>  if (output_token.length) {
>
>      token = "OK";
>  }
>
>
> cleanup:
>  gss_delete_sec_context(&minor_status, &gss_context, NULL);
>  gss_release_buffer(&minor_status, &service);
>  gss_release_buffer(&minor_status, &input_token);
>  gss_release_buffer(&minor_status, &output_token);
>  gss_release_name(&minor_status, &server_name);
>
>  return token;
> }
> int main(int argc, char *argv[]) {
>
>   const char *Token;
>   int opt,i=10;
>   int spnego=0;
>
>   while (-1 != (opt = getopt(argc, argv, "s"))) {
>    switch (opt) {
>    case 's':
>      spnego = 1;
>      break;
>    default:
>      fprintf(stderr, "%s| %s: unknown option: -%c.\n", LogTime(), PROGRAM, 
> opt);
>    }
>  }
>
>   fprintf(stdout,"Spnego : %d\n",spnego);
>   while (i--) {
>   Token = (const char *)create_token(spnego,"w2k3.windows2003.home");
>   fprintf(stdout,"Token: %s\n",Token);
>   }
>   return(0);
> }
>
>
>
> "Sam Hartman" <hartmans at mit.edu> wrote in message 
> news:tslhco1re4f.fsf at mit.edu...
>> Hi.  Thanks for your question; as best I know, we have not found this 
>> yet.
>>
>>
>> Is this in a case where gss_init_sec_ctx is successful or in a case
>> where it fails?
>>
>> Presumably in both cases you are deleting the context?
>> _______________________________________________
>> krbdev mailing list             krbdev at mit.edu
>> https://mailman.mit.edu/mailman/listinfo/krbdev
>>
>
>
>


--------------------------------------------------------------------------------


>
>
> _______________________________________________
> krbdev mailing list             krbdev at mit.edu
> https://mailman.mit.edu/mailman/listinfo/krbdev
>