Re: PATCH Prevent segfaults in connection state

Ping, in anticipation of a 2.8.26 for 1.3.35 (are we expecting one?)

I didn't see any feedback, do any of the Apache guru's have comments?  Any
clarifications?  (I hope I made the specifics clear enough, below.)

Yours,

Bill


William A. Rowe, Jr. wrote:
> Maintainers,
> 
> This patch addresses a still-outstanding flaw in mod_ssl, on *all* platforms.
> However it's rarely evident on any platform other than Win32, because only Win32
> recycles memory -so quickly- on other threads, that the cleanup cannot be
> invoked.
> 
> Instead, in the LogRequest (request-is-done) hook is used to clean everything
> up before r->pool goes poof.
> 
> Please, again consider this patch... I'm finished submitting through private
> channels and would like the end-users to be able to take advantage of it 
> already.
> 
> Thanks.
> 
> Bill
> 
> 
>   --- mod_ssl.h    25 Oct 2005 04:32:42 -0000    1.1
>   +++ mod_ssl.h    25 Oct 2005 05:54:19 -0000    1.2
>   @@ -711,6 +711,7 @@
>    int          ssl_hook_Fixup(request_rec *);
>    int          ssl_hook_ReadReq(request_rec *);
>    int          ssl_hook_Handler(request_rec *);
>   +int          ssl_hook_LogRequest(request_rec *r);
> 
>    /*  OpenSSL callbacks */
>    RSA         *ssl_callback_TmpRSA(SSL *, int, int);
>   --- mod_ssl.c    25 Oct 2005 04:32:46 -0000    1.1
>   +++ mod_ssl.c    25 Oct 2005 05:52:20 -0000    1.2
>   @@ -231,7 +231,7 @@
>        ssl_hook_Access,          /* [#3] check access by host address   */
>        NULL,                     /* [#6] determine MIME type            */
>        ssl_hook_Fixup,           /* [#7] pre-run fixups                 */
>   -    NULL,                     /* [#9] log a transaction              */
>   +    ssl_hook_LogRequest,      /* [#9] log a transaction              */
>        NULL,                     /* [#2] header parser                  */
>        ssl_init_Child,           /* child_init                          */
>        NULL,                     /* child_exit                          */
>   --- ssl_engine_io.c    25 Oct 2005 04:32:28 -0000    1.1
>   +++ ssl_engine_io.c    25 Oct 2005 05:52:20 -0000    1.2
>   @@ -263,7 +263,7 @@
>            r = (request_rec *)ap_ctx_get(actx, "ssl::request_rec");
> 
>        rv = -1;
>   -    if (r != NULL) {
>   +    if (r != NULL && r->ctx != NULL) {
>            ss = ap_ctx_get(r->ctx, "ssl::io::suck");
>            if (ss != NULL) {
>                if (ss->active && ss->pendlen > 0) {
>   --- ssl_engine_kernel.c    25 Oct 2005 04:32:41 -0000    1.1
>   +++ ssl_engine_kernel.c    25 Oct 2005 05:52:20 -0000    1.2
>   @@ -542,6 +542,28 @@
>    }
> 
>    /*
>   + *  Logging Handler, last chance at request_rec
>   + */
>   +int ssl_hook_LogRequest(request_rec *r)
>   +{
>   +    SSL *ssl;
>   +    ap_ctx *apctx;
>   +
>   +    /* Mitigate potential damage of any invalid ssl::request_rec
>   +     * by clearing this datum prior to child_sub_main destroying
>   +     * our r->pool (and within in, our request_rec!!!)
>   +     */
>   +    ssl = ap_ctx_get(r->connection->client->ctx, "ssl");
>   +    if (ssl != NULL) {
>   +        apctx = SSL_get_app_data2(ssl);
>   +        if (apctx && ap_ctx_get(apctx, "ssl::request_rec")) {
>   +            ap_ctx_set(apctx, "ssl::request_rec", NULL);
>   +        }
>   +    }
>   +    return OK;
>   +}
>   +
>   +/*
>     *  Post Read Request Handler
>     */
>    int ssl_hook_ReadReq(request_rec *r)
> 
> 
> 
> ______________________________________________________________________
> Apache Interface to OpenSSL (mod_ssl)                   www.modssl.org
> User Support Mailing List                      [email protected]
> Automated List Manager                            [email protected]
> 
> 
______________________________________________________________________
Apache Interface to OpenSSL (mod_ssl)                   www.modssl.org
User Support Mailing List                      [email protected]
Automated List Manager                            [email protected]