summaryrefslogtreecommitdiff
Commit message (Collapse)AuthorAgeFilesLines
* Convert tls13_record_layer.c to tls13_secret_{init,cleanup}()tb2021-01-051-21/+11
| | | | ok jsing
* Convert tls13_exporter() to tls13_secret_{init,cleanup}()tb2021-01-051-8/+5
| | | | ok jsing
* Use KNF for licence commenttb2021-01-051-2/+3
| | | | ok jsing
* Move tls13_secrets_destroy() below _create()tb2021-01-051-29/+29
| | | | ok jsing
* Convert tls13_secrets_{create,destroy}() to tls13_secret_{init,cleanup}()tb2021-01-051-71/+34
| | | | ok jsing
* Add tls13_secret_{init,cleanup}()tb2021-01-052-2/+29
| | | | | | | | These are two functions that will help streamlining various functions in the TLSv1.3 code that do not need to know about the interna of this struct. input/ok jsing
* Fix indent.jsing2021-01-051-2/+2
|
* Use consistent names in tls13_{client,server}_finished_{recv,send}().jsing2021-01-052-12/+12
| | | | | | | | | In tls13_{client,server}_finished_recv() we use verify_data_len, which makes more sense than hmac_len. Use the same name in tls13_{client,server}_finished_send(), keeping things consistent between functions. ok tb@
* Use legacy verifier when building auto chains.jsing2021-01-052-2/+6
| | | | | | | | | | | | | | | | | | The new verifier builds all chains, starting with the shortest possible path. It also does not currently return partial chains. Both of these things conflict with auto chain, where we want to build the longest possible chain (to include all intermediates, and probably the root unnecessarily), as well as using an incomplete chain when a trusted chain is not known. Depending on software configuration, we can end up building a chain consisting only of a leaf certificate, rather than a longer chain. This results in auto chain not including intermediates, which is undesireable. For now, switch auto chain building to use the legacy verifier. This should resolve the issues encountered by ajacoutot@ with sendmail. ok tb@
* Handle X509_V_ERR_UNABLE_TO_VERIFY_LEAF_SIGNATURE in new verifier.jsing2021-01-051-1/+4
| | | | | | Yet another mostly meaningless error value... Noted by and ok tb@
* Gracefully handle root certificates being both trusted and untrusted.jsing2021-01-052-4/+16
| | | | | | | | | | | | | | | | | | | | | | | | | | When a certificate (namely a root) is specified as both a trusted and untrusted certificate, the new verifier will find multiple chains - the first being back to the trusted root certificate and a second via the root that is untrusted, followed by the trusted root certificate. This situation can be triggered by a server that (unnecessarily) includes the root certificate in its certificate list. While this validates correctly (using the first chain), it means that we encounter a failure while building the second chain due to the root certificate already being in the chain. When this occurs we call the verify callback indicating a bad certificate. Some sensitive software (including bacula and icinga), treat this single bad chain callback as terminal, even though we successfully verify the certificate. Avoid this problem by simply dumping the chain if we encounter a situation where the certificate is already in the chain and also a trusted root - we'll have already picked up the trusted root as a shorter path. Issue with icinga2 initially reported by Theodore Wynnychenko. Fix tested by sthen@ for both bacula and icinga2. ok tb@
* Remove memset that was made redundant with the ASN1_time_parse()tb2021-01-051-4/+1
| | | | | | fix in libcrypto/asn1/a_time_tm.c r1.16. Suggested by jsing
* double word fix; from martin vahlensieckjmc2021-01-052-6/+6
|
* whitespacetb2021-01-043-10/+10
|
* Keep the various free calls of tls13_record_layer_free() in thetb2021-01-041-4/+4
| | | | | | order of the struct members for reviewability. ok jsing
* Tweak previous:schwarze2021-01-021-19/+34
| | | | | | | * Do not abuse .Bl -tag for lists without bodies, use .Bl -item instead. * In tagged lists, put bodies into bodies, not into heads. * Add a few missing macros. * Drop some useless quoting.
* Make list of DHE parameters more prominentkn2021-01-021-6/+19
| | | | | | | Follow the previous commit and complete the manual page for consistency; better readable and tags for free. OK tb
* Free {alert,phh}_data in tls13_record_layer_free()tb2021-01-021-1/+4
| | | | | | | | | | | | | | httpd(8)'s incorrect tls_close() after closing the underlying socket led to a leak: tls_close()'s attempt to send out the close_notify won't work very well over a closed pipe. This resulted in alert_data still hanging off the TLSv1.3 context's record layer struct. The tls_free() call should have cleaned this up but failed to do so. The record layer's phh_data potentially has the same issue, so free it as well. This diff makes -current httpd(8) run in constant memory over hundreds of thousands TLS connections with a static site. ok inoguchi jsing
* Remove unused categories in re_guts; they are written to but never read.millert2020-12-311-39/+2
| | | | From miod@, OK tb@
* Make the list of supported protocols more prominentkn2020-12-301-5/+15
| | | | | | | | | | Manuals like httpd.conf(5) refer to this for valid protocol strings, but elements inlined into sentences are hard find to spot. Use a list as already done elsewhere in this manual. OK jmc on earlier version Feeback OK tb
* Document meaning of '*' in genrsa outputtb2020-12-301-3/+6
| | | | ok inoguchi jmc kn
* Fix an off-by-one error in the marking of the O_CH operator followingmillert2020-12-281-1/+4
| | | | | an OOR2 operator. Also includes a regress test for the issue. From FreeBSD via miod@
* Start each regress run from scratch with new keys and CA database.bluhm2020-12-261-12/+17
|
* Convert CA regress implementation from shell script to make file.bluhm2020-12-265-141/+100
| | | | Ensure that it works with obj directory and link regress to build.
* Add to *FLAGS and *ADD rather than overwriting themtb2020-12-251-6/+6
| | | | | This makes CFLAGS pick up -O2, which shaves a few seconds runtime off these very slow tests.
* Destroy the mutex in tls_config objects when tls_config_free is called.bcook2020-12-221-1/+3
| | | | | | Add a stub for pthread_mutex_destroy() for installers. ok tb@
* Revert call to pthread_mutex_destroy until installers have a stub.bcook2020-12-221-3/+1
| | | | noted by deraadt@
* Destroy the mutex in a tls_config object when tls_config_free is called.bcook2020-12-211-1/+3
| | | | ok inoguchi@
* Remove echo headlines.bluhm2020-12-1715-46/+15
|
* Remove a redundant memset call.tb2020-12-161-2/+2
|
* Remove two reduntat memset calls.tb2020-12-162-5/+2
| | | | pointed out by jsing
* Avoid potential use of uninitialized in ASN1_time_parsetb2020-12-161-4/+3
| | | | | | | | | | | When parsing an UTCTime into a struct tm that wasn't cleared by the caller, the years would be added to the already present value, which could give an incorrect result. This is an issue in ASN1_UTCTIME_cmp_time_t(), which is practically unused. Fix this by always zeroing the passed struct tm. Issue reported by Olivier Taïbi, thanks! ok jsing
* Fix some KNF issuestb2020-12-161-7/+8
|
* Use natural sizes for S3I(s)->tmp's *_md arraystb2020-12-151-6/+4
| | | | | | | | | | | | | | | | | | | | It is a historical artifact that cert_verify_md[], finish_md[] and peer_finish_md[] are twice as large as they need to be. This is confusing, especially for finish_md[] and peer_finish_md[] which are copied to to previous_client_finished[] and previous_server_finished[] which are only half as large. It is easy to check that they will never get more than EVP_MAX_MD_SIZE data written to them. In 1998, EVP_MAX_MD_SIZE was 20 bytes long (for SHA-1). This got bumped to 16+20 for the SSLv3-specific md5+sha1. Apparently under the impression that EVP_MAX_MD_SIZE was still 20 bytes, someone else doubled finish_md[]'s size to EVP_MAX_MD_SIZE*2 and added /* actually only needs to be 16+20 */. A bit later finish_md[] was split up, and still a bit later the comment was amended for TLSv1. Shortly thereafter SHA-512 required a bump of EVP_MAX_MD_SIZE to 64 by a third person and we have been carrying 192 bytes of untouched memory in each of our SSLs ever since. ok inoguchi jsing (jsing had the same diff)
* Fix return value variable type in tls_keypair_load_certinoguchi2020-12-151-2/+2
| | | | | | | ERR_peek_error() returns unsigned long. Reported by github issue by @rozhuk-im. ok bcook@ jsing@
* Fix SSL_get{,_peer}_finished() with TLSv1.3tb2020-12-142-2/+28
| | | | | | | | | | As reported by Steffen Ullrich and bluhm, the Finished tests in p5-Net-SSLeay's t/local/43_misc_functions.t broke with with TLSv1.3. The reason for this is that we don't copy the MDs over to the SSL, so the API functions can't retrieve them. This commit fixes this part of the test (one unrelated test still fails). ok inoguchi jsing
* Switch finish{,_peer}_md_len from int to size_ttb2020-12-141-3/+3
| | | | | | | This is the natural type for these and it simplifies an upcoming commit. The few consumers have been carefully checked to be fine with this. ok inoguchi jsing
* LibreSSL 3.3.1libressl-v3.3.1bcook2020-12-081-3/+3
|
* Fix a NULL dereference in GENERAL_NAME_cmp()tb2020-12-086-11/+94
| | | | | | | | | | | | Comparing two GENERAL_NAME structures containing an EDIPARTYNAME can lead to a crash. This enables a denial of service attack for an attacker who can control both sides of the comparison. Issue reported to OpenSSL on Nov 9 by David Benjamin. OpenSSL shared the information with us on Dec 1st. Fix from Matt Caswell (OpenSSL) with a few small tweaks. ok jsing
* Enable t_mmap-1 test. It is skipped on NetBSD, but works for us.bluhm2020-12-062-3/+10
|
* Mark bitmask_{start,end}_values[] and g_probable_mtu[] const.tb2020-12-051-4/+4
| | | | ok jsing kn
* Mark nid_list[] const. This moves 116 bytes to .rodata.tb2020-12-051-2/+2
| | | | ok jsing kn
* Move point-on-curve check to set_affine_coordinatestb2020-12-044-18/+50
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | Bad API design makes it possible to set an EC_KEY public key to a point not on the curve. As a consequence, it was possible to have bogus ECDSA signatures validated. In practice, all software uses either EC_POINT_oct2point*() to unmarshal public keys or issues a call to EC_KEY_check_key() after setting it. This way, a point on curve check is performed and the problem is mitigated. In OpenSSL commit 1e2012b7ff4a5f12273446b281775faa5c8a1858, Emilia Kasper moved the point-on-curve check from EC_POINT_oct2point to EC_POINT_set_affine_coordinates_*, which results in more checking. In addition to this commit, we also check in the currently unused codepath of a user set callback for setting compressed coordinates, just in case this will be used at some point in the future. The documentation of EC_KEY_check_key() is very vague on what it checks and when checks are needed. It could certainly be improved a lot. It's also strange that EC_KEY_set_key() performs no checks, while EC_KEY_set_public_key_affine_coordinates() implicitly calls EC_KEY_check_key(). It's a mess. Issue found and reported by Guido Vranken who also tested an earlier version of this fix. ok jsing
* grammar fixes from Varik "The Genuine Article!!!" Valefor;jmc2020-12-033-9/+9
|
* Extend the methods test to cover dtls methods as welltb2020-12-011-1/+57
|
* Enable ssl_methods unit test.tb2020-12-011-1/+2
|
* Add an ssl_methods() unit test that currently only covers thetb2020-12-011-0/+192
| | | | | behavior of SSL_is_server(). This would have caught the regression introduced in the method unification.
* Bring back *_client_method() structstb2020-12-013-11/+200
| | | | | | | | | | | | | | | | | | | | | | | | | | The method unification broke an API promise of SSL_is_server(). According to the documentation, calling SSL_is_server() on SSL objects constructed from generic and server methods would result in 1 even before any call to SSL_set_accept_state(). This means the information needs to be available when SSL_new() is called, so must come from the method itself. Prior to the method unification, s->server would be set to 0 or 1 in SSL_new() depending on whether the accept method was undefined or not. Instead, introduce a flag to the internal structs to distinguish client methods from server and generic methods and copy that flag to s->server in SSL_new(). This problem was reported to otto due to breakage of DoH in net/dnsdist. The reason for this is that www/h2o relies on SSL_is_server() to decide whether to call SSL_accept() or SSL_connect(). Thus, the h2o server would end up responding to a ClientHello with another ClientHello, which results in a handshake failure. The bandaid applied to www/h2o can be removed once this fix has made it into snaps. No other breakage is known. This commit brings back only about half of the duplication removed in the method unification, so is preferable to a full revert. ok jsing
* Avoid undefined behavior due to memcpy(NULL, NULL, 0)tb2020-11-251-4/+6
| | | | | | | | | This happens if name->der_len == 0. Since we already have a length check, we can malloc and memcpy inside the conditional. This also makes the code easier to read. agreement from millert ok jsing
* mapalign() only handles allocations >= a page; problem found by and ok semarie@otto2020-11-231-1/+3
|
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-03-28 12:16:52 +0000