Hi!
I am working on a BIO-pair setup and ran into a question regarding the
process of free()ing the objects after usage.
Considered the following sequence:
typedef struct {
SSL *con;
BIO *internal_bio;
BIO *network_bio;
... } TLScontext_t;
TLScontext_t *TLScontext;
...
TLScontext->con = SSL_new(ctx);
BIO_new_bio_pair(&TLScontext->internal_bio, BIO_bufsiz,
&TLScontext->network_bio, BIO_bufsiz));
SSL_set_bio(TLScontext->con, TLScontext->internal_bio,
TLScontext->internal_bio);
[perform networking with SSL_accept()/read()/write()/shutdown() on the
internal side and the TLScontext->network_bio for the socket communication]
I now would call:
BIO_free(TLScontext->internal_bio);
BIO_free(TLScontext->network_bio);
SSL_free(TLScontext->con);
But this seems to be wrong, since the TLScontext->internal_bio is connected
to the SSL connection TLScontext->con.
So when I call SSL_free() it tries to free again the already BIO_free()ed
TLScontext->internal_bio...
The other way round (SSL_free() first) the same problem applies.
Following my analysis, the _correct_ way of releasing the BIOs is:
/*
* Free the SSL structure and the BIOs. Warning: the internal_bio is
* connected to the SSL structure and is automatically freed with
* it. Do not free it again (core dump)!!
* Only free the network_bio.
*/
SSL_free(TLScontext->con);
BIO_free(TLScontext->network_bio);
with the sequencing (SSL_free() or BIO_free() first) being of no importance...
Did I get this right?
Best regards,
Lutz
PS. To be part of Postfix/TLS-0.6.10, to be released after straightening
out this issue.
--
Lutz Jaenicke [EMAIL PROTECTED]
BTU Cottbus http://www.aet.TU-Cottbus.DE/personen/jaenicke/
Lehrstuhl Allgemeine Elektrotechnik Tel. +49 355 69-4129
Universitaetsplatz 3-4, D-03044 Cottbus Fax. +49 355 69-4153
______________________________________________________________________
OpenSSL Project http://www.openssl.org
Development Mailing List [EMAIL PROTECTED]
Automated List Manager [EMAIL PROTECTED]
