summaryrefslogtreecommitdiff
path: root/src/lib/libcrypto/x509 (follow)
Commit message (Collapse)AuthorAgeFilesLines
* Add a comment on abuse of EXFLAG_INVALIDtb2024-04-091-1/+8
| | | | | | | We added things we probably shouldn't have, and so did BoringSSL and OpenSSL. Terrible API is terrible. discussed with jsing
* Remove notBefore and notAfter cacheing.beck2024-04-084-63/+16
| | | | | | | | | | | | | | | | | | | | | | This cache was added because our time conversion used timegm() and gmtime() which aren't very cheap. These calls were noticably expensive when profiling things like rpki-client which do many X.509 validations. Now that we convert times using julien seconds from the unix epoch, BoringSSL style, instead of a julien days from a Byzantine date, we no longer use timegm() and gmtime(). Since the julien seconds calculaitons are cheap for conversion, we don't need to bother caching this, it doesn't have a noticable performance impact. While we are at this correct a bug where x509_verify_asn1_time_to_time_t was not NULL safe. Tested for performance regressions by tb@ and job@ ok tb@ job@
* Make X509_VERIFY_PARAM_set1_policies() less badtb2024-03-291-22/+34
| | | | | | | | | | | If any OBJ_dup() fails along the way, a partially copied policy stack would remain on the params object. This makes no sense. Implement and use an sk_ASN1_OBJECT_deep_copy(), that copies the full stack or else returns NULL. Remove unnecessary NULL check and streamline some other logic. ok jsing
* Clean up X509_VERIFY_PARAM_add0_policy()tb2024-03-291-6/+5
| | | | | | Streamline some checks and use more idiomatic sk_push() error check ok jsing
* Fix indenttb2024-03-291-2/+2
|
* Unify *_up_ref() implementationstb2024-03-271-3/+2
| | | | | | No need for an inconsistently named local variable and a ternary operator. ok jsing
* Remove superfluous parentheses in X509_REQ setterstb2024-03-261-10/+10
| | | | No change in generated assembly
* Cosmetics, mostly removal of silly parenthesestb2024-03-261-21/+19
| | | | No change in generated assembly
* Drop superfluous parentheses in X509_set_version()tb2024-03-261-5/+5
|
* Tweak versions comment for CRLstb2024-03-261-1/+2
|
* piuid and psuid have annoyed me for long enoughtb2024-03-261-7/+7
|
* x509 trust: remove unneeded headerstb2024-03-261-3/+1
|
* Reject setting invalid versions for certs, CRLs and CSRstb2024-03-263-3/+18
| | | | | | | | | | | | The toolkit aspect bites again. Lots of invalid CRLs and CSRs are produced because people neither read the RFCs nor does the toolkit check anything it is fed. Reviewers apparently also aren't capable of remembering that they have three copy-pasted versions of the same API and that adding a version check to one of the might suggest adding one for the other two. This requires ruby-openssl 20240326p0 to pass ok beck job jsing
* PKCS5_pbe2_set_iv() can be local to p5_pbev2tb2024-03-261-3/+1
| | | | quoth the muppet "yes I know this is horrible!"
* Revert r1.13 since it currently breaks openssl-ruby regress tests.jsing2024-03-251-2/+2
| | | | ok tb@
* Error on setting an invalid CSR versionjob2024-03-251-2/+2
| | | | | | Reported by David Benjamin (BoringSSL) OK tb@
* Final tweaks in x509_trs.c for nowtb2024-03-251-10/+11
| | | | looked over by jsing
* Kill X509_TRUSTtb2024-03-251-79/+31
| | | | | | | | After peeling off enough layers, the entire wacky abstraction turns out to be nothing but dispatching from a trust_id to a trust handler and passing the appropriate nid and the cert. ok beck jsing
* Clean up EVP_MD_CTX_{legacy_clear,cleanup}() usage in x509/x509_cmp.cjoshua2024-03-251-21/+25
| | | | ok tb@
* Add back trust member of X509_TRUST that I accidentally deletedtb2024-03-251-1/+2
|
* Const correct the trust handlerstb2024-03-251-8/+7
| | | | | | | The certificates no longer need to be modified since we cache the extensions up front. ok beck
* Pass the nid instead of the entire trust structuretb2024-03-251-10/+10
| | | | | | | This code is so ridiculously overengineered that it is an achievement even by early OpenSSL standards. ok beck
* Pull extension caching into X509_check_trust()tb2024-03-251-2/+7
| | | | | | This way the trust handlers can stop modifying the certificates. ok beck
* Remove unnecessary stat() calls from by_dirbeck2024-03-251-14/+15
| | | | | | | | | | | | | | | | | | | | | When searching for a CA or CRL file in by_dir, this stat() was used to short circuit attempting to open the file with X509_load_cert_file(). This was a deliberate TOCTOU introduced to avoid setting an error on the error stack, when what you really want to say is "we couldn't find a CA" and continue merrily on your way. As it so happens you really do not care why the load_file failed in any of these cases, it all boils down to "I can't find the CA or CRL". Instead we just omit the stat call, and clear the error stack if the load_file fails. The fact that you don't have a CA or CRL is caught later in the callers and is what you want, mimicing the non by_dir behaviour instead of possibly some bizzaro file system error. Based on a similar change in Boring. ok tb@
* Rename ax into auxtb2024-03-241-8/+7
|
* Drop redundant NULL checks around two for loopstb2024-03-241-15/+14
| | | | ok beck
* Rename arg1 to nid because that's what it istb2024-03-241-11/+11
| | | | | "Yeah, arg1 is always such an imaginative name" ian ok beck
* Drop the unused 'name' member from X509_TRUSTtb2024-03-241-10/+1
| | | | | | | This used to be exposed via an accessor, but this accessor is no longer part of the library, so nuke it. ok beck
* Remove unused flags and arg2 members from X509_TRUSTtb2024-03-231-3/+1
| | | | ok jsing
* Remove unused flags argument from the trust handlerstb2024-03-231-13/+13
| | | | | | | | The public X509_check_trust() takes a flag parameter which we must leave in place. However, we can stop passing the flag parameter around without ever looking at it. ok jsing
* Simplify X509_STORE_free()tb2024-03-221-11/+3
| | | | | | | This had an inlined version of sk_pop_free(). We can just call it the right way. ok jsing
* Remove sk_find_ex()tb2024-03-021-3/+1
| | | | | | | This API intends to find the closest match to the needle. M2Crypto exposes it because it can. This will be fixed by patching the port. ok jsing
* Make LHASH_OF() and STACK_OF() use opaque structstb2024-03-021-1/+2
| | | | | | | | | | | | | This removes internals of these two special snowflakes and will allow further simplifications. Unfortunately, there are some pieces of software that actually use LHASH_OF() (looking at you, pound, Ruby, and openssl(1)), so we get to keep exposing this garbage, at least for now. Expose lh_error() as a symbol to replace a macro reaching into _LHASH. lh_down_load() is no longer available. _LHASH and _STACK are now opaque, LHASH_NODE becomes internal-only. from jsing
* Remove X509_CRL_METHOD APItb2024-03-021-13/+1
| | | | | | I would keep repeating myself... In the bit bucket you go. ok jsing
* Garbage collect most of the public LOOKUP APItb2024-03-022-62/+4
| | | | | | | | Yet another bit of extensibility that no one ever really used. X509_LOOKUP_free() needs to stay because of ... rust-openssl (and kdelibs4support). ok jsing
* Expose X509_STORE_get1_objects()tb2024-03-021-3/+1
| | | | | | Safer replacement API for the unsafe X509_STORE_get0_objects(). ok jsing
* Make X509_CERT_AUX internaltb2024-03-022-12/+11
| | | | | | Another struct/API that should never have leaked out of the library. ok jsing
* Remove X509_TRUST from the public APItb2024-03-022-13/+11
| | | | | | With API and other users internal, this struct can now go. ok jsing
* Remove DECLARE_STACK_OF(X509_TRUST)tb2024-03-021-3/+1
| | | | | | Unused since the extensibility was neutered. ok jsing
* Remove unused public X509_TRUST APItb2024-03-022-92/+2
| | | | | | | Another thing that should never have leaked out of the library. It will become internal entirely, where the code can be simplified greatly. ok jsing
* const correct X509_PURPOSE_get0{,_{,s}name}()tb2024-03-022-10/+11
| | | | | | | Unfortunately, PHP and rust-openssl still need this API. At least we can make the table read-only now since we disabled its extensibility. ok jsing
* Make X509_PURPOSE opaquetb2024-03-022-11/+13
| | | | | | | Code using details of X509_PURPOSE does so by using API. So we can make this struct opaque. ok jsing
* Remove unused parts of the purpose APItb2024-03-025-54/+11
| | | | | | | | | Most of this is the ability to add custom purposes. Also the astounding X509_STORE_CTX_purpose_inherit(). The names are used by PHP, and M2Crypto exposes X509_check_purpose(), so these remain public. Some weird, most likely invalid, uses also remain in rust-openssl. ok jsing
* Remove X509V3_EXT extensibility APItb2024-03-022-40/+2
| | | | | | | | | | You used to be able to define your own X.509 extension handlers. Great. Even greater: the verifier would ignore any custom extensions. So this was only ever useful for serialization and deserialization. In other words, almost entirely pointless. The API was also unused except for a hack in kore-acme, which was fixed recently. ok jsing
* Remove X509_ALGOR_set_md()tb2024-03-021-4/+1
| | | | | | | | One of those void APIs that are super hard to use safely since they can fail but can't communicate failure. Nothing uses this. Internal uses have been converted to error checked X509_ALGOR_set_evp_md(). ok jsing
* Remove more PBE stuff from the public APItb2024-03-022-16/+13
| | | | | | | This is still needed internally for CMS and its predecessors. This removal will enable disentangling some of its innards. ok jsing
* Remove most PBEPARAM stuff from public visibilitytb2024-03-022-33/+33
| | | | | | The struct itself needs to remain public, unfortunately. ok jsing
* Small tweak for X509_check_purpose()tb2024-02-281-4/+6
| | | | ok jsing
* Use an accessor instead of reaching into X509_PURPOSEtb2024-02-281-2/+2
| | | | ok jsing
* Prepare to provide X509_STORE_get1_objects()tb2024-02-232-2/+70
| | | | | | | | | | | | | | | | | | | | | | | | | | | | The OpenSSL 1.1 API X509_STORE_get0_objects() is not thread safe. It exposes a naked internal pointer containing certificates, CRLs and cached objects added by X509_LOOKUP_hash_dir(). Thus, if the store is shared between threads, it is not possible to inspect this pointer safely since another thread could concurrently add to it. This may happen in particular during certificate verification. This API led to security issues in rust-openssl and is also problematic in current Python. Other consumers of X509_STORE_get0_objects() are haproxy, isync, openvpn. The solution is to take a snapshot of the state under a lock and return that. This is what X509_STORE_get1_objects() does. It returns a newly allocated stack that needs to be freed with sk_X509_OBJECT_pop_free(), passing X509_OBJECT_free as a second argument. Based on a diff by David Benjamin for BoringSSL. https://boringssl-review.googlesource.com/c/boringssl/+/65787 ok beck jsing PS: Variants of this have landed in Python and OpenSSL 3 as well. There the sk_*deep_copy() API is used, which in OpenSSL relies on evaluating function pointers after casts (BoringSSL fixed that). Instead of using this macro insanity and exposing that garbage in public, we can do this by implementing a pedestrian, static sk_X509_OBJECT_deep_copy() by hand.
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-08-26 00:32:12 +0000