summaryrefslogtreecommitdiff
path: root/src/lib/libssl (follow)
Commit message (Collapse)AuthorAgeFilesLines
* The undocumented public function SSL_set_SSL_CTX(3) changes theschwarze2020-09-151-4/+4
| | | | | | context used by an SSL object, so do not talk about the SSL_CTX that "an SSL object was created from"; fixing an inaccuracy pointed out by jsing@.
* Split the tls12_record_layer_write_mac() function.jsing2020-09-151-10/+19
| | | | | | | | | Split the existing tls12_record_layer_write_mac() function so that we can soon reuse part of it for the read side. No functional change. ok tb@
* Do not destroy an existing cipher list when ssl_parse_ciphersuites()schwarze2020-09-151-4/+2
| | | | | | | fails, to match the behaviour of ssl_create_cipher_list(). This also agrees with the behaviour of SSL_set_ciphersuites(3) in OpenSSL. Issue found while writing documentation. OK jsing@
* Correct a failure case in tls12_record_layer_seal_record_protected()jsing2020-09-151-2/+2
| | | | This should be a 'goto err' rather than returning.
* Create the missing RETURN VALUES section and move the appropriateschwarze2020-09-151-57/+103
| | | | | | | | content there. Clarify when the returned pointers become invalid, which is far from obvious but sets surprising traps for the user. For three of the functions, correct statements about when they fail. Also improve a number of wordings while here. OK beck@
* Cleanup/simplify SSL_set_ssl_method().jsing2020-09-151-18/+18
| | | | | | | In particular, figure what the handshake_func should be early on, so we can just assign later. ok beck@
* Mop up the get_ssl_method function pointer.jsing2020-09-152-50/+2
| | | | | | | Now that get_ssl_method is no longer used, we can garbage collect the function pointer and some associated machinery. ok beck@
* Move state initialisation from SSL_clear() to ssl3_clear().jsing2020-09-142-4/+4
| | | | | | | | | | | | | | | | | | | | | | If we use the default method (now TLSv1.3) and end up talking to a TLSv1.2 server that gives us a session ticket, then try to resume that session, we end up trying to talk TLS without doing a handshake. This is caused by the state (S3I(s)->hs.state) getting cleared, which results in SSL_do_handshake() and others thinking they do not need to do anything (as SSL_in_init() and SSL_in_before() are not true). The reason this occurs is due to SSL_set_ssl_method() calling ssl_free() and ssl_new() when switching methods. The end result is that the S3I(s) has been freed and reallocated, losing the state in the process. Since the state is part of the S3I(s) structure, move its initialisation into ssl3_clear() - this ensures it gets correctly reinitialised across a SSL_set_ssl_method() call. Issue noticed by sthen@ with nginx and unifi. ok beck@ tb@
* Cleanup and simplify SSL_set_session().jsing2020-09-141-36/+22
| | | | | | | | | | SSL_set_ssl_method() checks to see if the method is already the same, so we do not need to do this check in three different places. Switch to dtls1_get_client_method()/tls1_get_client_method() to find the method - this is a slight change in behaviour, however there is not much point trying to resume a session on something other than a client. ok beck@
* Avoid NULL deref SSL_{,CTX_}set_ciphersuitestb2020-09-141-2/+2
| | | | | | | | Move assignment to the correct place so that the run continuation condition actually checks what it is supposed to. Found by getting lucky when running regress. ok beck jsing
* Implement SSL_{CTX_,}set_ciphersuites().jsing2020-09-135-13/+211
| | | | | | | | | | OpenSSL added a separate API for configuring TLSv1.3 ciphersuites. Provide this API, while retaining the current behaviour of being able to configure TLSv1.3 via the existing interface. Note that this is not currently exposed in the headers/exported symbols. ok beck@ inoguchi@ tb@
* Improve handling of BIO_read()/BIO_write() failures in the TLSv1.3 stack.jsing2020-09-131-1/+9
| | | | | | | | | | | | | | | | When BIO returns a failure, it does not always add an error to the error stack. In the case of the legacy stack, this was generally handled by the guesswork performed by SSL_get_error(). However, in the case of the new stack we push an 'unknown' error onto the stack. Improve this situation by specifically checking errno in the case of a BIO_read() or BIO_write() failure. If the error stack is empty then push a SYSerror() with the errno which is preferable to the 'unknown' error later. Noted by bluhm@ via syslogd regress. ok beck@ tb@
* Use the correct type for tls1_set_ec_id()tb2020-09-121-3/+3
| | | | | | The curve_id is a uint16, not an int. ok beck jsing
* Simplify tls1_set_ec_id() a bittb2020-09-121-24/+19
| | | | | | | Use more descriptive variable names, explain why NID_undef is fine and simplify the logic. ok beck jsing
* Remove cipher_list_by_id.jsing2020-09-117-89/+32
| | | | | | | | | | | | | | | | | When parsing a cipher string, a cipher list is created, before being duplicated and sorted - the second copy being stored as cipher_list_by_id. This is done only so that a client can ensure that the cipher selected by a server is in the cipher list. This is pretty pointless given that most clients are short-lived and that we already had to iterate over the cipher list in order to build the client hello. Additionally, any update to the cipher list requires that cipher_list_by_id also be updated and kept in sync. Remove all of this and replace it with a simple linear scan - the overhead of duplicating and sorting the cipher list likely exceeds that of a simple linear scan over the cipher list (64 maximum, more typically ~9 or so). ok beck@ tb@
* Simplify SSL_get_ciphers().jsing2020-09-111-13/+7
| | | | ok beck@, tb@
* Rename ssl_cipher_is_permitted()jsing2020-09-113-10/+10
| | | | | | | | | | The name ssl_cipher_is_permitted() is not entirely specific - what it really means is "can this cipher be used with a given version range". Use ssl_cipher_allowed_in_version_range() to more clearly indicate this. Bikeshedded with tb@ ok tb@
* Some SSL_AD_* defines snuck into the TLSv1.3 code - replace them withjsing2020-09-112-10/+10
| | | | | | TLS13_ALERT_* defines. ok beck@ tb@
* Various ciphers related clean up.jsing2020-09-111-41/+36
| | | | | | | Consistently use the names 'ciphers' and 'cipher' instead of 'sk' and 'c'. Remove some redundant code, unnecessary parentheses and fix some style(9). ok inoguchi@ tb@
* Set alpn_selected_len = 0 when alpn_selected is NULLinoguchi2020-09-091-1/+4
| | | | ok jsing@ tb@
* Garbage collect renew_ticket in tls_decrypt_tickettb2020-09-071-8/+5
| | | | | | | | | | | | This is only set in one place and read in one place to set the badly named tlsext_ticket_expected flag. It seems preferable to set this flag directly, thus simplifying the logic. This slightly changes the behavior in that this flag is now set earlier, but this seems preferable anyway. Any error between the old and the new position where the flag is set is either fatal (so the connection will be closed) or a decrypt error (so the flag will be set). discussed with jsing
* KNF and comment tweakstb2020-09-021-8/+10
|
* Zero out data to avoid leaving stack garbage in the tail oftb2020-09-011-1/+3
| | | | | | | the session id in case the copied session id is shorter than SSL_MAX_SESSION_ID_LENGTH. long standing bug pointed out by jsing
* The bumping of sess_cb_hit stats can wait until handling oftb2020-09-011-4/+3
| | | | get_session_cb is completed.
* In the explanatory comment of ssl_get_prev_session fix the spelling oftb2020-09-011-5/+6
| | | | the function name, document alert and make it fit into 80 columns.
* Split session retrieval out of ssl_get_prev_session()tb2020-09-011-78/+92
| | | | | | | | | | | | | | | | In case the session ticket was empty or missing, an attempt is made to retrieve the session from the internal cache or via a callback. This code can easily be flattened a bit and factored into two functions. I decided to wrap those into a third function to make the call from the switch easier on the eye. I could have kept the try_session_cache flag, but it now seems rather pointless and awkwardly named anyway, so I took its negation and named it ticket_decrypted. To top things off, a little bit of polish in the exit path. ok beck inoguchi jsing (with the usual healthy dose of nits)
* copy session id directly in ssl_get_prev_sessiontb2020-09-013-27/+23
| | | | | | | | | | | | ssl_get_prev_session() hands the session id down to tls_decrypt_ticket() which then copies it into the session pointer that it is about to return. It's a lot simpler to retrieve the session pointer and copy the session id inside ssl_get_prev_session(). Also, 'goto err' directly in TLS1_TICKET_NOT_DECRYPTED instead of skipping a couple of long if clauses before doing so. ok inoguchi jsing
* indent the only other label in this filetb2020-09-011-2/+2
|
* Indent label and remove dangling elsetb2020-09-011-4/+4
|
* Zap NULL check before SSL_SESSION_free()tb2020-09-011-3/+2
|
* Rename the session pointer ret to sesstb2020-09-011-25/+25
| | | | | | ret is a confusing name for a pointer in a function that returns int. ret is only returned in the sense that it ultimately replaces the current s->session on success.
* Hoist ERR_clear_error() call into the derr: labeltb2020-09-011-4/+2
| | | | | | | | | The only path that sets TLS1_TICKET_NOT_DECRPYTED is through this label and the ERR_clear_error() is called conditionally on this. We clear the errors to make decrypt errors non-fatal. The free functions should not set the errors and if they do, we don't want to hide that. discussed with jsing
* simplify tls1_process_ticket() exit pathtb2020-09-012-19/+7
| | | | | | | | | | | | | | | | tls1_process_ticket() - the only caller of tls_decrypt_ticket() - ends in a switch over the return value of tls_decrypt_ticket() to decide whether or not to set s->internal->tlsext_ticket_expected = 1. Since tls_decrypt_ticket() already knows what it will return and partly bases its decision on what to return on whether or not the ticket needs to be renewed, it can also take care of setting this flag. This way we don't need to have a confusing switch that conflates some return values and sets this flag. Moreover, we can get rid of the ugly TLS1_TICKET_DECRYPTED_RENEW whose only purpose is to signal that the flag should be set. ok jsing
* Return code tweaks for session ticket handlerstb2020-08-313-47/+51
| | | | | | | | In tls1_process_ticket() and tls_decrypt_ticket() use #defines with descriptive names instead of hardcoding -1 1 2 3 4 and occasionally explaining the magic numbers with comments. ok beck inoguchi
* Send alert on ssl_get_prev_session failuretb2020-08-314-20/+32
| | | | | | | | | | | | ssl_get_prev_session() can fail for various reasons some of which may be internal_error others decode_error alerts. Propagate the appropriate alert up to the caller so we can abort the handshake by sending a fatal alert instead of rudely closing the pipe. Currently only 28 of 292 test cases of tlsfuzzer's test-extension.py pass. With this diff, 272 pass. The rest will require fixes elsewhere. ok beck inoguchi jsing
* Start replacing the existing TLSv1.2 record layer.jsing2020-08-307-195/+614
| | | | | | | | | | This takes the same design/approach used in TLSv1.3 and provides an opaque struct that is self contained and cannot reach back into other layers. For now this just implements/replaces the writing of records for DTLSv1/TLSv1.0/TLSv1.1/TLSv1.2. In doing so we stop copying the plaintext into the same buffer that is used to transmit to the wire. ok inoguchi@ tb@
* Send an unexpected message alert if no valid content type is found.jsing2020-08-111-2/+5
| | | | | | | | | When record protection is engaged, the plaintext must be followed by a non-zero content type and optional zero padding. If the plaintext is zero length or only consists of zero bytes then it is not a valid message, since the content type is unspecified. ok tb@
* Increment the epoch in the same place for both read and write.jsing2020-08-111-3/+3
| | | | ok inoguchi@ tb@
* Use 0 instead of 0x00 for memset() calls.jsing2020-08-112-8/+8
| | | | ok inoguchi@ tb@
* Use SSL3_SEQUENCE_SIZE for last_write_sequence[] rather than hardcoding.jsing2020-08-111-2/+2
| | | | ok inoguchi@ tb@
* In SSL_new() just 'goto err' on allocation failure.jsing2020-08-111-11/+6
| | | | | | The error path does the same as the currently duplicated code. ok inoguchi@ tb@
* Avoid passing -1 to freezero.tb2020-08-101-9/+10
| | | | | | | | If a peer sends a bogus record consisting of all-zero plaintext, the content_len would be decremented to -1 and cause a crash in freezero. ok inoguchi jsing
* Fix some wrapping/indent.jsing2020-08-091-4/+3
|
* Add P-521 to the list of curves supported by default in the client.jsing2020-08-091-5/+18
| | | | | | | | | | | | | | | A certain VPN provider appears to have configured their servers to only accept P-521 for TLSv1.3 key exchange. The particular VPN software in use also does not currently allow for the TLSv1.3 key share groups to be configured, which means that there is no way to easily use LibreSSL in this situation. Include P-521 in the list of curves that are supported by default in the client, in order to increase interoperability. Discussed at length with beck@, inoguchi@ and tb@. ok tb@
* Use CBB more correctly when writing SSL3/DTLS records.jsing2020-08-092-66/+92
| | | | | | | | | | | | Previously we used CBB to build the record headers, but not the entire record. Use CBB_init_fixed() upfront, then build the record header and add space for the record content. However, in order to do this we need to determine the length of the record upfront. This simplifies the code, removes a number of manual bounds checks and makes way for further improvements. ok inoguchi@ tb@
* Make the explicit IV length handling in DTLS the same as SSL3/TLS.jsing2020-08-091-8/+13
| | | | ok inoguchi@ tb@
* Cleanup aead_ctxinoguchi2020-08-041-1/+3
| | | | ok jsing@ tb@
* Only parse a client's status_request in the CHtb2020-08-031-1/+4
| | | | | | | | A client should only send a status_request as part of the CH. Pointed out by Michael Forney ok inoguchi jsing
* Ensure clients only send a status_request in the CHtb2020-08-031-3/+7
| | | | | | | | | | The current code might cause a client to send a status_request containing a CertificateStatusRequest with its certificate. This makes no sense. Pointed out by Michael Forney ok inoguchi jsing
* Correctly handle server requests for an OCSP responsetb2020-08-031-1/+12
| | | | | | | | | | | | | | | | | According to RFC 8446, 4.4.2.1, a server may request that a client present an OCSP response with its certificate by sending an empty status_request extension as part of the certificate request. The current code expects a full CertificateStatus structure, which is only sent if the server sends an OCSP response with its certificate. This causes interoperability issues with Go's TLS server and with newer GnuTLS where we would abort the handshake with a decode_error alert and length mismatch error. Issue reported and diagnosed by Michael Forney Problem also found by Mikolaj Kucharski and inoguchi. ok inoguchi jsing
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-05-28 08:52:36 +0000