summaryrefslogtreecommitdiff
path: root/src/lib/libtls/tls.c (follow)
Commit message (Collapse)AuthorAgeFilesLines
...
* Call BIO_sock_init() from tls_init() to ensure sockets are enabled on Windows.bcook2016-01-181-1/+4
| | | | This is of course a no-op on other platforms. Noted by equalsraf from github.
* Allow us to get cipher and version even if there is not a peer certificate.beck2015-10-071-5/+6
| | | | ok doug@
* clean some ugly intendation wartsderaadt2015-09-291-2/+2
|
* Provide tls_config_insecure_noverifytime() in order to be able to disablejsing2015-09-141-1/+6
| | | | | | certificate validity checking. ok beck@
* Expose EOF without close-notify via tls_close().jsing2015-09-141-4/+11
| | | | | | | | | | | Make tls_read(3)/tls_write(3) follow read(2)/write(2) like semantics and return 0 on EOF with and without close-notify. However, if we saw an EOF from the underlying file descriptors without getting a close-notify, save this and make it visible when tls_close(3) is called. This keeps the semantics we want, but makes it possible to detect truncation at higher layers, if necessary. ok beck@ guenther@
* Return an error if tls_handshake() or tls_close() is called on a contextjsing2015-09-141-2/+13
| | | | | | for which they are not valid operations. ok beck@
* work around the stupid semantics of SSL_read and SSL_write to make surebeck2015-09-131-3/+3
| | | | | we can indicate an EOF properly on tls_read and tls_write ok jsing@
* Don't leak conninfo - spotted by marko kreen.beck2015-09-131-2/+3
| | | | ok jsing@
* Move connection info into it's own private structure allocated and filled inbeck2015-09-121-4/+13
| | | | | | at handshake time. change accessors to return const char * to remove need for caller to free memory. ok jsing@
* Ensure that we clear the libssl error stack before we make a function calljsing2015-09-121-3/+6
| | | | | | | | | | that we will pass the result through tls_ssl_error() on failure. Otherwise we can end up reporting spurious errors due to their being unrelated errors already on the error stack. Spotted by Marko Kreen. ok beck@
* Store a reference to the peer certificate (if any) upon completion of thejsing2015-09-111-1/+6
| | | | | | handshake. Free the reference when we reset the TLS context. ok beck@
* Call tls_set_errorx() instead of tls_set_error() injsing2015-09-101-5/+5
| | | | tls_configure_ssl_verify(). Also tweak an error message and unwrap a line.
* comment for errno clobbering, to indicate why we do this.beck2015-09-101-1/+5
| | | | ok deraadt@ jsing@
* change TLS_READ_AGAIN to TLS_WANT_POLLIN and TLS_WRITE_AGAIN to TLS_WANT_POLLOUTbeck2015-09-101-4/+4
| | | | | | | to make it more clear to users of this api what needs to be done in these error cases. Discussed extensively with bluhm@ and jsing@ and others. ok jsing@
* Change tls_read and tls_write semantics to return an ssize_t to betterbeck2015-09-101-17/+11
| | | | | | | | match read() and write() semantics to make porting existing code using read/write easier.. requested by bluhm@ who convinced jsing and I to break the api ok jsing@ bluhm@
* Split tls_handshake() out from tls_accept/tls_connect. By doing this thejsing2015-09-101-5/+30
| | | | | | | | | | | | | | | tls_accept/tls_connect functions can be guaranteed to succeed or fail and will no longer return TLS_READ_AGAIN/TLS_WRITE_AGAIN. This also resolves the semantics of tls_accept_*. The tls_handshake() function now does I/O and can return TLS_READ_AGAIN/TLS_WRITE_AGAIN. Calls to tls_read() and tls_write() will trigger the handshake if it has not already completed, meaning that in many cases existing code will continue to work. Discussed over many coffees at l2k15. ok beck@ bluhm@
* Indent labels with a space so that diff -p is more friendly.jsing2015-09-091-4/+4
| | | | Requested by bluhm@
* Add client certificate support. Still needs a few tweaks but this willbeck2015-09-091-2/+40
| | | | | ride upcoming minor bump ok jsing@
* always clear errno when coming back from tls_read tls_write, and tls_close.beck2015-09-091-9/+20
| | | | | | | this avoids the problem of people checking for return values < 0 and then checking for errno before checking for TLS_READ_AGAIN TLS_WRITE_AGAIN - since we can not guarantee what errno will be set to from the underlying library calls
* nuke trailing whitespacebeck2015-09-091-4/+3
| | | | ok jsing@
* Improve libtls error messages.jsing2015-08-271-22/+65
| | | | | | | | | | | | The tls_set_error() function previously stored the errno but did nothing with it. Change tls_set_error() to append the strerror(3) of the stored errno so that we include useful information regarding failures. Provide a tls_set_errorx() function that does not store the errno or include strerror(3) in the error message. Call this function instead of tls_set_error() for errors where the errno value has no useful meaning. With feedback from and ok doug@
* Split the persistent/configuration flags from temporary state flags andjsing2015-08-271-1/+2
| | | | | | | | | | | ensure that the temporary state flags get cleared in tls_reset(). Fixes a bug spotted by Marko Kreen whereby TLS_CONNECTING could remain on reset. While here, also move the TLS_STATE_CONNECTING check to after the TLS_CLIENT check - if TLS_STATE_CONNECTING was ever set on any other context type it would allow a bypass. ok bluhm@
* Make tls_read()/tls_write always set outlen to zero on error - this gotjsing2015-08-221-5/+5
| | | | | | broken by r1.4. Spotted by Marko Kreen.
* Set SSL_MODE_ENABLE_PARTIAL_WRITE and SSL_MODE_ACCEPT_MOVING_WRITE_BUFFERbluhm2015-07-181-1/+4
| | | | | | | in libtls. This gives tls_write() a similar short write semantics as write(2). So implementing daemons with libevent buffers will be easier and workarounds in syslogd and httpd can be removed. OK tedu@ beck@ reyk@
* Treat SSL_ERROR_ZERO_RETURN as a success, rather than a failure. Alsojsing2015-04-151-5/+6
| | | | | | | ensure that outlen is set to zero so that tls_read() has read(2) like semantics for EOF. Spotted by doug@
* Make tls_close() more robust - do not rely on a close notify being receivedjsing2015-04-151-13/+17
| | | | | | | | | | | from the other side and only return TLS_READ_AGAIN/TLS_WRITE_AGAIN if we failed to send a close notify on a non-blocking socket. Otherwise be more forceful and always shutdown/close the socket regardless of other failures. Also do not consider ENOTCONN or ECONNRESET to be a shutdown failure, since there are various situations where this can occur. ok doug@ guenther@
* Handle the case where multiple calls to SSL_shutdown() are required tojsing2015-04-021-6/+9
| | | | | | close the connection. Also correctly handle the error on failure. Diff from cookieandscream via github.
* Store errors that occur during a tls_accept_socket() call on the contextjsing2015-03-311-5/+5
| | | | | | | for the server, rather than on the context for the connection. This makes more sense than the current behaviour does. Issue reported by Tim van der Molen.
* Convert tls_connect_fds() and tls_accept_socket() to the new OpenSSL errorjsing2015-02-071-2/+2
| | | | | | | dance handling code. This means that we get slightly useful messages when a TLS connection or accept fails. Requested by reyk@
* Attempt to implement the OpenSSL error dance so that TLS read/writejsing2015-02-071-29/+61
| | | | | | failures return something that is actually useful to the caller. ok reyk@
* Rename SSL_CTX_use_certificate_chain() to SSL_CTX_use_certificate_chain_mem().reyk2015-02-061-2/+2
| | | | | As discussed with beck@ jsing@ and others OK beck@
* Add size_t to int checks for SSL functions.doug2014-12-171-1/+22
| | | | | | | | | | libtls accepts size_t for lengths but libssl accepts int. This verifies that the input does not exceed INT_MAX. It also avoids truncating size_t when comparing with int and adds printf-style attributes for tls_set_error(). with input from deraadt@ and tedu@ ok tedu@
* revert previous change for now, adjusting based on comments from jsing@bcook2014-12-071-10/+3
|
* Allow specific libtls hostname validation errors to propagate.bcook2014-12-071-3/+10
| | | | | | | | | | | | | Remove direct calls to printf from the tls_check_hostname() path. This allows NUL byte error messages to bubble up to the caller, to be logged in a program-appropriate way. It also removes non-portable calls to getprogname(). The semantics of tls_error() are changed slightly: the last error message is not necessarily preserved between subsequent calls into the library. When the previous call to libtls succeeds, client programs should treat the return value of tls_error() as undefined. ok tedu@
* Rename libressl to libtls to avoid confusion and to make it easier tojsing2014-10-311-0/+300
distinguish between LibreSSL (the project) and libressl (the library). Discussed with many.
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-03-21 19:11:31 +0000