summaryrefslogtreecommitdiff
path: root/src/lib/libssl (follow)
Commit message (Collapse)AuthorAgeFilesLines
...
* 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
* Check the return value of tls1_enc() in the write path.jsing2020-08-022-6/+6
| | | | | | | | | The write path can return a failure in the AEAD path and there is no reason not to check a return value. Spotted by tb@ during another review. ok tb@
* Clean up/simplify more of the dtls1/ssl3 record writing code:jsing2020-08-012-73/+34
| | | | | | | | | | | | - Make the DTLS code much more consistent with the ssl3 code. - Avoid assigning wr->input and wr->length just so they can be used as arguments to memcpy(). - Remove the arc4random_buf() call for the explicit IV, since tls1_enc() already does this for us. ok tb@
* Pull record version selection code up and pass it as an argument tojsing2020-08-011-15/+15
| | | | | | ssl3_create_record(). ok tb@
* Have ssl_init_wbio_buffer() push the buffering BIO rather than doing itjsing2020-07-301-5/+2
| | | | | | ourselves. Spotted by tb@ during a previous review.
* Clean up and simplify some of the SSL3/DTLS1 record writing code.jsing2020-07-302-76/+72
| | | | | | | | | | | This will allow for further changes to be made with less complexity and easier review. In particular, decide if we need an empty fragment early on and only do the alignment calculation once (rather than in two separate parts of the function. ok tb@ inoguchi@
* Add minimal info callback support for TLSv1.3tb2020-07-303-3/+32
| | | | | | | | | | | | | | As abieber@ found the hard way, some python frameworks (twisted, synapse) thought it a great idea to use the info callback mechanism (designed to get state information about SSL objects) to modify state information such as setting and verifying the SNI. The switch of TLS_method() to default to TLSv1.3 broke these contraptions. Further bits of the info callback mechanism will likely metastasize throughout the TLSv1.3 stack if we need them, so we only do what's really necessary now. Lots of debugging, crucial hint and testing by abieber input & ok jsing
* Handle SSL_MODE_AUTO_RETRY being changed during a TLSv1.3 session.jsing2020-07-251-1/+4
| | | | | | | | | | | | | | | Both Perl's HTTP::Tiny and IO::Socket::SSL know about SSL_MODE_AUTO_RETRY and try to work around the fact that OpenSSL enabled it by default. However, this can lead to the mode being disabled prior to the TLSv1.3 handshake and then enabled after the handshake has completed. In order to handle this correctly we have to check the mode and inform the record layer prior to every read. Issue reported and test case provided by Nathanael Rensen <nathanael@polymorpheus.com>. ok inoguchi@ tb@
* Dedup the use legacy stack code.jsing2020-07-141-56/+25
| | | | ok inoguchi@ tb@
* Revert the TLSv1.3 version switching fix/hack.jsing2020-07-141-10/+1
| | | | | | | | This is no longer necessary since the TLS_method() now supports TLSv1.3. Reverts r1.211 of ssl_lib.c. ok beck@ inoguchi@ tb@
* Remove some unnecessary function pointers from SSL_METHOD_INTERNAL.jsing2020-07-075-64/+17
| | | | | | ssl_version is completely unused and get_timeout is the same everywhere. ok beck@ inoguchi@ tb@
* Enable TLSv1.3 for the generic TLS_method().jsing2020-07-072-5/+52
| | | | | | This can be done now that we have both TLSv1.3 client and server. ok beck@ inoguchi@ tb@
* zap trailing whitespace on one linetb2020-07-031-2/+2
|
* Make the message type available to the extension functionstb2020-07-032-167/+181
| | | | | | | | | | | | | | Some TLS extensions need to be treated differently depending on the handshake message they appear in. Over time, various workarounds and hacks were used to deal with the unavailability of the message type in these functions, but this is getting fragile and unwieldy. Having the message type available will enable us to clean this code up and will allow simple fixes for a number of bugs in our handling of the status_request extension reported by Michael Forney. This approach was suggested a while ago by jsing. ok beck jsing
* Improve argument order for the internal tlsext APItb2020-07-038-39/+39
| | | | | | | | Move is_server and msg_type right after the SSL object so that CBS and CBB and alert come last. This brings these functions more in line with other internal functions and separates state from data. requested by jsing
* Switch the order of the two tests in tls13_client_hello_required_extensionstb2020-06-251-9/+9
| | | | to match the order they are listed in the RFC. No functional change.
* Make tls13_legacy_shutdown() match ssl3_shutdown() semantics.jsing2020-06-241-21/+22
| | | | | | | | | | | | | When first called, queue and send a close notify, before returning 0 or 1 to indicate if a close notify has already been received from the peer. If called again only attempt to read a close notify if there is no pending application data and only read one record from the wire. In particular, this avoids continuing to read application data where the peer continues to send application data. Issue noted by naddy@ with ftp(1). ok jca@ tb@
* Enforce restrictions for ClientHello extensionstb2020-06-241-1/+44
| | | | | | | | | | | | | | | RFC 8446 section 9.2 imposes some requirements on the extensions sent in the ClientHello: key_share and supported_groups must either both be present or both be absent. If no pre_shared_key was sent, the CH must contain both signature_algorithms and supported_groups. If either of these conditions is violated, servers must abort the handshake with a missing_extensions alert. Add a function that enforces this. If we are going to enforce that clients send an SNI, we can also do this in this function. Fixes failing test case in tlsfuzzer's test-tls13-keyshare-omitted.py ok beck inoguchi jsing
* We inherited the constant time CBC padding removal from BoringSSL, buttb2020-06-191-4/+4
| | | | | | | | | | | missed a subsequent fix for an off-by-one in that code. If the first byte of a CBC padding of length 255 is mangled, we don't detect that. Adam Langley's BoringSSL commit 80842bdb44855dd7f1dde64a3fa9f4e782310fc7 Fixes the failing tlsfuzzer lucky 13 test case. ok beck inoguchi
* The check_includes step is incorrect dependency management model forderaadt2020-06-091-11/+1
| | | | | | how our tree gets built. If this was done in all the libraries (imagine sys/dev), it would disrupt the development process hugely. So it should not be done here either. use 'make includes' by hand instead.
* Implement a rolling hash of the ClientHello message, Enforce RFC 8446beck2020-06-066-7/+179
| | | | | | | | section 4.1.2 to ensure subsequent ClientHello messages after a HelloRetryRequest messages must be unchanged from the initial ClientHello. ok tb@ jsing@
* Use IANA allocated GOST ClientCertificateTypes.jsing2020-06-053-9/+15
| | | | | | | | | | | IANA has allocated numbers for GOST ClientCertificateType. Use them in addition to private values (left in place for compatibility). Diff from Dmitry Baryshkov <dbaryshkov@gmail.com> Sponsored by ROSA Linux ok inoguchi@ tb@
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-03-20 00:42:02 +0000