I am sorry, the previous email was sent incomplete by accident:
To cut a long story short: I am doing the authentication in the "if
(NULL == *con_cls)" part. If the user passed valid credentials the
connection is serviced later on by returning MHD_YES (without queueing
any response) otherwise the unauthorized response is generated.
Best
Martin
On Mon, Feb 17, 2014 at 3:23 PM, Martin Velek <martin.ve...@gmail.com> wrote:
> Hi,
>
> because this email is not an answer to Your question, please consider
> it like a hint, not advice.
>
> You are not obligated to return only MHD_YES in "if (NULL ==
> *con_cls)". You may also return MHD_NO or even process the connection.
> According to my stack trace,
> Thread [12] 4230 [core: 1] (Suspended : Breakpoint)
> AccessHandlerCallback() at http_server.c:303 0x43a281
> call_connection_handler() at connection.c:1,303 0x43042a
> MHD_connection_handle_idle() at connection.c:2,207 0x431baa
> MHD_run_from_select() at daemon.c:1,814 0x43498c
> MHD_select() at daemon.c:1,909 0x434e15
> MHD_select_thread() at daemon.c:2,417 0x434e7e
> rtos_mhd_thread_wrapper() at platform.h:168 0x4327a6
> ITaskStartup() at rtos_pthread.c:228 0x40895c
> start_thread() at pthread_create.c:308 0x7ffff76c0e9a
> clone() at clone.S:112 0x7ffff70f83fd
> <...more frames...>
> the connection_handler is called after
> MHD_CONNECTION_HEADERS_PROCESSED (the first call). At this time, the
> *upload_data_size is equal to zero even if the TCP packet contains
> data following (behind) the end mark of HTTP headers.
>
> I am using this checking in my url handler:
> // New connection?
> if (NULL == *con_cls)
> {
> // Data should never be greater than zero because of first call.
> assert((*upload_data_size) == 0);
>
> // Default behavior for authentication is OK.
> ret = MHD_YES;
>
> // get username and password
> user = MHD_basic_auth_get_username_password(connection, &pass);
>
> // check if it is valid
> if (false == credentials_check)
> {
> ret = generate_unauthorized_response(connection,
> authentification_failed_size_,
> (void *)authentification_failed_page_,
> MHD_RESPMEM_PERSISTENT);
> // Return value.
> return ret;
> }
> else
> {
> *con_cls = Something;
>
> }
>
>
> }
> // already authenticated ....
>
>
>
> On Mon, Feb 17, 2014 at 2:13 PM, Sebastian Hammerl
> <list-libmicroht...@omoco.de> wrote:
>> Hi,
>>
>> Am 17.02.2014 13:58, schrieb Christian Grothoff:
>>
>>> if (*upload_data_size > 0)
>>> {
>>> *upload_data_size = 0; // consume
>>> return MHD_YES;
>>> }
>>
>>
>> that did it!
>>
>>
>> Thank you very much,
>> Sebastian
>>
