summaryrefslogtreecommitdiff
path: root/src/lib/libcrypto/x509 (follow)
Commit message (Collapse)AuthorAgeFilesLines
* Use a sensible variable name (i.e. nid) instead of i for a NIDtb2023-11-131-5/+6
|
* Use X509_get_signature_nid() instead of inlining ittb2023-11-131-2/+2
| | | | ok beck jsing
* X509_certificate_type() needs to know about RSA-PSStb2023-11-131-1/+4
| | | | | | | This doesn't do much right now, but is part of the tangle that is adding RSA-PSS support. ok beck jsing
* Eliminate the timegm(3) dependency in libcryptotb2023-11-134-19/+36
| | | | | | | | | | | | | | | | | timegm(3) is not available on some operating systems we support in portable. We currently use musl's implementation, for which gcc-13 decided to emit warnings (which seem incorrect in general and are irrelevant in this case anyway). Instead of patching this up and diverge from upstream, we can avoid reports about compiler warnings by simply not depending on this function. Rework the caching of notBefore and notAfter by replacing timegm(3) with asn1_time_tm_to_time_t(3). Also make this API properly error checkable since at the time x509v3_cache_extensions(3) is called, nothing is known about the cert, in particular not whether it isn't malformed one way or the other. suggested by and ok beck
* Fix a few bugs in X509v3_asid_add*()tb2023-11-111-38/+96
| | | | | | | | | | | | | | | | | | | | | | | | | These 'builder' functions, usually used together, can result in corrupt ASIdentifiers on failure. In general, no caller should ever try to recover from OpenSSL API failure. There are simply too many traps. We can still make an effort to leave the objects in unmodified state on failure. This is tricky because ownership transfer happens. Unfortunately a really clean version of this seems impossible, maybe a future iteration will bring improvements... The nasty bit here is that the caller of X509v3_asid_add_id_or_range() can't know from the return value whether ownership of min and max was transferred or not. An inspection of (*choice)->u.range is required. If a caller frees min and max after sk_ASIdOrRange_push() failed, there is a double free. All these complications could have been avoided if the API interface had simply used uint32_t instead of ASN1_INTEGERs. The entire RFC 3779 API was clearly written without proper review. I don't know if there ever was an actual consumer before rpki-client. If it existed, nobody with the requisite skill set looked at it in depth. ok beck for the general direction with a lot of input and ok jsing
* Fix a few bizarre line wraps in x509.htb2023-11-021-8/+5
|
* Add X509_ALGOR_set0_by_nid()tb2023-11-011-1/+3
| | | | | | | | | | | | | X509_ALGOR_set0() is annoyingly unergonomic since it takes an ASN1_OBJECT rather than a nid. This means that almost all callers call OBJ_obj2nid() and they often do this inline without error checking so that the resulting X509_ALGOR object is corrupted and may lead to incorrect encodings. Provide an internal alternative X509_ALGOR_set0_by_nid() that takes a nid instead of an ASN1_OBJECT and performs proper error checking. This will be used to convert callers of X509_ALGOR_set0() in the library. ok jsing
* Fix an error exit in X509v3_addr_validate_path()tb2023-10-291-3/+6
| | | | | | | | If the topmost cert is invalid, this should result in a validation failure. Do the same dance as elsewhere permitting the verify callback to intercept the error but ensuring that we throw an error. ok jsing
* Hide X509_ALGOR_set_md() for LIBRESSL_INTERNALtb2023-10-261-1/+3
|
* Add internal version of X509_ALGOR_set_md()tb2023-10-111-1/+3
| | | | | | | | | | | | | | | | | | | X509_ALGOR_set_md() is a void function that cannot easily be error checked. The caller has to jump through hoops to make sure this function doesn't fail. Prepare replacing this internally with X509_ALGOR_set_evp_md(), which allows error checking. There is one slight change of behavior: if the EVP_MD object passed in does not have an OID known to the library, then this new API fails. It is unclear what the library should do with such an object and people who use EVP_MD_meth_new() need to know what they are doing anyway and they are better off teaching the lib about the OID if they're going to be messing with certs. Oh, and the prototype is in x509_local.h because the rest of this API is in x509.h despite being implemented in asn1/. ok jsing
* Allow IP addresses to be specified in a URI.beck2023-09-292-15/+20
| | | | | | | | | | | | Our checking here was a bit too aggressive, and did not permit an IP address in a URI. IP's in a URI are allowed for things like CRLdp's AIA, SAN URI's etc.). The check for this was also slightly flawed as we would permit an IP if memory allocation failed while checking for an IP. Correct both issues. ok tb@
* RFC 3779: stop pretending we support AFIs other than IPv4 and IPv6tb2023-09-271-19/+28
| | | | | | | This code is a complete bug fest and using it with any other AFI is downright dangerous. Such don't arise in this context in practice. ok claudio jsing
* Back out superfluous initializationjob2023-09-111-5/+4
| | | | requested by jsing@
* Initialize afi & safi to zerojob2023-09-061-4/+5
| | | | OK tb@
* Avoid use-of-uninitialized in i2r_IPAddrBlocks()tb2023-09-061-1/+8
| | | | | | | Reported by Viktor Szakats in https://github.com/libressl/portable/issues/910 ok job
* Fix leaks in copy_issuer()tb2023-08-301-8/+14
| | | | | | | | | | | | | | The stack of subject alternative names from the issuer is parsed using X509V3_EXT_d2i(), so it must be freed with sk_GENERAL_NAME_pop_free(). It's not worth doing complicated ownership handling when the individual alternative names can be copied with GENERAL_NAME_dup(). Previously, ialt and its remaining members would be leaked when the call to sk_GENERAL_NAME_push() failed halfway through. This is only reachable via the issuer:copy x509v3.cnf(5) directive. ok jsing
* Check X509_digest() return in x509v3_cache_extensions()tb2023-08-181-2/+3
| | | | | | | | On failure invalidate the cert with EXFLAG_INVALID. It's unlikely that a cert would make it through to the end of this function without setting the flag, but it's bad style anyway. ok jsing
* Stop including ecdsa.h and ecdh.h internallytb2023-07-281-7/+1
| | | | | | | | These headers are now reduced to #include <openssl/ec.h> and are provided for compatiblity only. There's no point in using them. At the same time garbage collect the last uses of OPENSSL_NO_{ECDSA,ECDH} in our tree. ok jsing
* Convert some tables to C99 initializerstb2023-07-022-21/+114
| | | | ok & "happy pirate day" beck
* x509v3.h: unwrap a linetb2023-06-251-3/+2
|
* Check for duplicate X.509v3 extension OIDstb2023-06-251-1/+45
| | | | | | | | | | | | | Per RFC 5280, 4.2: A certificate MUST NOT include more than one instance of a particular extension. This implements such a check in x509v3_cache_extensions() by sorting the list of extensions and looking for duplicate neighbors. This sidesteps complications from extensions we do not know about and keeps algorithmic complexity reasonable. If the check fails, EXFLAG_INVALID is set on the certificate, which means that the verifier will not validate it. ok jsing
* Remove some redundant parenthesestb2023-06-231-17/+17
| | | | This file is already enough of an eyesore without them.
* Improve certificate version checks in x509v3_cache_extensions()tb2023-06-201-4/+11
| | | | | | | | | Only allow version v1-v3, disallow issuerUID and subjectUID in v1 certs and require that if X509v3 extensions are present that the cert be v3. Initial diff from job ok job jsing
* Teach the grotty X509_certificate_type() about Ed25519 certstb2023-06-151-1/+4
| | | | ok jsing
* Remove dead code.beck2023-06-081-13/+3
| | | | | | | | must_be_ca can no longer be 0 after the proxy cert code got nuked, so change this to an if. must_be_ca is now -1 for a leaf, or 1 for a non leaf. ok tb@
* Make X509_NAME_get_text_by[NID|OBJ] safer.beck2023-05-291-10/+28
| | | | | | | | | | | | | | | | This is an un-revert with nits of the previously landed change to do this which broke libtls. libtls has now been changed to not use this function. This change ensures that if something is returned it is "text" (UTF-8) and a C string not containing a NUL byte. Historically callers to this function assume the result is text and a C string however the OpenSSL version simply hands them the bytes from an ASN1_STRING and expects them to know bad things can happen which they almost universally do not check for. Partly inspired by goings on in boringssl. ok jsing@ tb@
* Merge X509_VERIFY_PARAM_ID into X509_VERIFY_PARAMtb2023-05-284-110/+73
| | | | | | | | | | Back in the day when essentially every struct was open to all applications, X509_VERIFY_PARAM_ID provided a modicum of opacity. This indirection is now no longer needed with X509_VERIFY_PARAM being opaque itself, so stop using X509_VERIFY_PARAM_ID and merge it into X509_VERIFY_PARAM. This is a first small step towards cleaning up the X509_VERIFY_PARAM mess. ok jsing
* Copy the verify param hostflags independently of the host listtb2023-05-241-3/+4
| | | | | | | | | | | | | | | | Without this, hostflags set on the SSL_CTX would not propagate to newly created SSL. This is surprising behavior that was changed in OpenSSL 1.1 by Christian Heimes after the issue was flagged by Quentin Pradet: https://bugs.python.org/issue43522 This is a version of the fix that landed in OpenSSL. There used to be a workaround in place in urllib3, but that was removed at some point. We haven't fixed this earlier since it wasn't reported. It only showed up after recent fallout of extraordinarily strict library checking in urllib3 coming from their own interpretation of the implications of PEP 644. ok jsing
* Provide X509_VERIFY_PARAM_set_hostflags()tb2023-05-241-1/+8
| | | | | | | | This is needed for an upcoming regress test that needs to access the hostflag. This is public API in OpenSSL but since nothing seems to be using this, this accessor will be kept internal-only for the time being. ok jsing
* Rename arguments of X509_STORE_CTX_init()tb2023-05-141-5/+5
| | | | | | | | | | It is higly confusing to call the list of untrusted certs chain, when you're later going to call X509_STORE_CTX_get0_chain() to get a completely unrelated chain by the verifier. Other X509_STORE_CTX APIs call this list of certs 'untrusted', so go with that. At the same time, rename the x509 into leaf, which is more explicit. suggested by/ok jsing
* Fix X509error() and X509V3error()tb2023-05-141-6/+11
| | | | | | | | | | | | | | | When v3err.c was merged into x509_err.c nearly three years ago, it was overlooked that the code needed two distinct pairs of ERR_FUNC/ERR_REASON, one for ERR_LIB_X509 and one for ERR_LIB_X509V3. The result is that the reason strings for the X509_R_* codes would be overwritten by the ones for X509V3_R_* with the same value while the reason strings for all X509V3_R_* would be left undefined. Fix this by an #undef/#define dance for ERR_LIB_X509V3 once we no longer the ERR_FUNC/ERR_REASON pair for ERR_LIB_X509. reported by job ok jsing
* Bob points out that one error should be an X509V3error()tb2023-05-121-2/+2
|
* x509_utl.c: fix some style nits.tb2023-05-121-4/+3
|
* Rewrite string_to_hex() and hex_to_string() using CBB/CBStb2023-05-121-70/+124
| | | | | | | | | | | These helpers used to contain messy pointer bashing some with weird logic for NUL termination. This can be written more safely and cleanly using CBB/CBS, so do that. The result is nearly but not entirely identical to code used elsewhere due to some strange semantics. Apart from errors pushed on the stack due to out-of-memory conditions, care was taken to preserve error codes. ok jsing
* Rename the other_ctx in X509_STORE_CTX into trustedtb2023-05-082-12/+12
| | | | | | | | | | | | | | | | The other_ctx is a strong contender for the worst name of a struct member in OpenSSL. It's a void * member whose only purpose ever was to be set to a STACK_OF(X509) * via X509_STORE_CTX_trusted_stack() (yes, this is obviously a setter, why do you ask?) and then to be used by the get_issuer() callback (which of course isn't there to find any old issuer, but only to look for issuers among the 'trusted' certs). Anyway, we may want to rename untrusted into intermediates and trusted into roots later on, but for now let's match the lovely public API. While there rename get_issuer_sk() into get_trusted_issuer() which is a more accurate and slightly less silly name. ok jsing
* X509_verify_cert(): Garbage collect the unused roots variabletb2023-05-081-4/+1
| | | | | | | | | roots was used to store the trusted stack or pull the roots out of the X509_STORE before beck unmooned Ethel in x509_vfy.c r1.88. Since then this variable is effectively unused. It seems the STACK_OF(3) madness is too complicated for -Wunused-but-set-variable to notice. ok miod
* Avoid trailing whitespace in extension printingtb2023-05-081-2/+2
| | | | | | | If an extension is non-critical, X509V3_extensions_print() would leave trailing whitespace. This can be trivially avoided. ok miod
* Remove a misplaced empty linetb2023-05-071-2/+1
|
* Revert utf-8 fix for X509_NAME_get_index_by_NID to avoid libtlsbeck2023-05-031-27/+10
| | | | | | | regress for the moment. this will come back after we rethink the failure versus not there case. ok tb@ jsing@
* Bring back length check tb ok'ed and I managed to remove whilebeck2023-05-031-2/+2
| | | | | | changing tests. ok tb@
* Change X509_NAME_get_index_by[NID|OBJ] to be safer.beck2023-05-021-10/+27
| | | | | | | | | | | | | | | | | | | | | | Currently these functions return raw ASN1_STRING bytes as a C string and ignore the encoding in a "hold my beer I am a toolkit not a functioning API surely it's just for testing and you'd never send nasty bytes" kind of way. Sadly some callers seem to use them to fetch things liks subject name components for comparisons, and often just use the result as a C string. Instead, encode the resulting bytes as UTF-8 so it is something like "text", Add a failure case if the length provided is inadequate or if the resulting text would contain an nul byte. based on boringssl. nits by dlg@ ok tb@
* Send x509_subject_cmp() to the attictb2023-04-301-9/+1
| | | | | | This helper has been inside #if 0 for nearly 25 years. Let it go. If we should ever need it, I'm quite confident that we will be able to come up with its one line body on our own.
* Zap extra blank linetb2023-04-301-2/+1
|
* Enable policy checking by default now that we are DAG implementation based.beck2023-04-284-12/+11
| | | | | | | This ensures that we will no longer silently ignore a certificate with a critical policy extention by default. ok tb@
* Unifdef LIBRESSL_HAS_POLICY_DAG and remove it from the Makefiletb2023-04-283-88/+3
| | | | with beck
* Take the old policy code behind the barntb2023-04-287-1900/+0
| | | | | | | | It can go play in the fields with all the other exponential time policy "code". discussed with jsing ok & commit message beck
* Silence gcc-4 warnings about sk_sort()tb2023-04-281-5/+6
| | | | | Tell it we deliberately ignore the return value, (we really don't care what the old comparison function was).
* Remove now no longer needed <assert.h>; sort headerstb2023-04-281-4/+2
| | | | ok jsing
* Deassert has_explicit_policy()tb2023-04-281-3/+4
| | | | | | | The only caller is X509_policy_check() which goes straight to error. with beck ok jsing
* Deassert delete_if() callbackstb2023-04-281-5/+7
| | | | | | | | Add sk_is_sorted() checks to the callers of sk_X509_POLICY_NODE_delete_if() and add a comment that this is necessary. with beck ok jsing
generated by cgit v1.2.3-55-g6feb (git 2.39.1) at 2025-08-26 18:28:32 +0000