RE: [PATCH] http.c

The problem is with the calling of "parse_url" in the code.
"parse_url" is called twice when a http request is being made.
It is the second call that causes the leak.

the "write_request_thread" function invokes "get_connection", this calls
"parse_url" and populates the trans->host field.
then "write_request_thread" invokes "send_request", this calls "parse_url"
again
and populates the trans->host fields.

However send_request does not destroy the trans->host that has already been
allocated.

cheers,
michael.

ANAM Wireless Internet Solutions
http://www.anam.com mailto:[email protected]
+353 1 284 7555
Castle Yard, Saint Patrick's Road, Dalkey, County Dublin, Ireland

> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of Stipe Tolj
> Sent: 03 December 2002 14:25
> To: Nisan Bloch
> Cc: [email protected]
> Subject: Re: [PATCH] http.c
>
>
> > the check for != NULL is not necessary, yes. But the
> octstr_destroy is, as
> > in http.c:parse_url:1155 , host is created.
> > *host = octstr_copy(url, prefix_len, host_len);
>
> but the host is inside the HTTPServe trans structure, it should be
> free'ed there, see gwlib/http.c:616:
>
> static void server_destroy(void *p)
> {
>     HTTPServer *trans;
>
>     trans = p;
>     octstr_destroy(trans->url);
>     http_destroy_headers(trans->request_headers);
>     trans->request_headers = NULL;
>     octstr_destroy(trans->request_body);
>     entity_destroy(trans->response);
>     octstr_destroy(trans->host);
>     gw_free(trans);
> }
>
> and the previous server_create() garantees that the ->host is NULLed.
>
> Stipe
>
> [email protected]
> -------------------------------------------------------------------
> Wapme Systems AG
>
> Vogelsanger Weg 80
> 40470 Düsseldorf
>
> Tel: +49-211-74845-0
> Fax: +49-211-74845-299
>
> E-Mail: [email protected]
> Internet: http://www.wapme-systems.de
> -------------------------------------------------------------------
> wapme.net - wherever you are