Implementing missing macros to openssl compatibility layer by Roy-Carter · Pull Request #10520 · wolfSSL/wolfssl

Roy-Carter · 2026-05-22T19:02:23Z

Description

Implementing as part of migration to wolfssl
SSL_set_ciphersuites
X509_add1_ext_i2d
sk_GENERAL_NAME_new_null
EC_KEY_oct2key

Testing

unitests to each functionality added

Checklist

[ *] added tests
[ *] updated/added doxygen
updated appropriate READMEs
Updated manual and documentation

wolfSSL-Bot · 2026-05-22T19:03:08Z

Can one of the admins verify this patch?

Roy-Carter · 2026-05-22T19:04:34Z

@julek-wolfssl second part of the functions i've implemented , split the PR's for clarity of checks .

Roy-Carter · 2026-05-24T11:04:45Z

@dgarske errors seemed related to master diff. both tests worked for me locally , I rebased , can we re-run workflow?

Roy-Carter · 2026-05-29T12:44:12Z

@julek-wolfssl @dgarske I've rebased to master , seems related to new changes added .

dgarske · 2026-05-29T15:25:41Z

Jenkins okay to test

Copilot

Pull request overview

This PR expands the OpenSSL compatibility layer with aliases and implementations for several missing APIs used during migration to wolfSSL.

Changes:

Adds compatibility mappings for SSL_set_ciphersuites, X509_add1_ext_i2d, sk_GENERAL_NAME_new_null, and EC_KEY_oct2key.
Implements wolfSSL_X509_add1_ext_i2d and wolfSSL_EC_KEY_oct2key.
Adds unit tests covering the new compatibility APIs.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
`wolfssl/ssl.h`	Declares the new X509 extension helper.
`wolfssl/openssl/ssl.h`	Adds OpenSSL compatibility macros for SSL, X509, and GENERAL_NAME APIs.
`wolfssl/openssl/ec.h`	Declares and maps `EC_KEY_oct2key`.
`src/x509.c`	Implements `wolfSSL_X509_add1_ext_i2d`.
`src/pk_ec.c`	Implements `wolfSSL_EC_KEY_oct2key`.
`tests/api.c`	Adds API tests for the new compatibility additions.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

julek-wolfssl

@Roy-Carter please address copilot review. Thanks.

Roy-Carter · 2026-06-07T07:54:10Z

@julek-wolfssl can we re-run workflow ?

julek-wolfssl · 2026-06-08T15:17:35Z

$ ./autogen.sh && ./configure --enable-all && make check
...
FAILURES:
   871: test_wolfSSL_X509_add_ext
   929: test_wolfSSL_X509_add1_ext_i2d_flags

 End API Tests
 Failed/Skipped/Passed/All: 2/182/1432/1616

Roy-Carter · 2026-06-21T12:49:41Z

can we re-run workflow @julek-wolfssl

also writing here about our ticket , I think it might've been closed / I lost perms to write on it even though I did send emails there it seems as if zendesk link is unavailable

dgarske · 2026-06-22T14:39:02Z

+    ExpectNotNull(ctx = SSL_CTX_new(wolfSSLv23_client_method()));
+    ExpectNotNull(ssl = SSL_new(ctx));
+
+    ExpectIntEQ(SSL_set_ciphersuites(ssl, "TLS_AES_128_GCM_SHA256"), 1);


With --disable-errorstrings

1309: test_wolfSSL_SSL_set_ciphersuites : ERROR - tests/api.c line 2537 failed with: expected: wolfSSL_set_cipher_list(ssl, "TLS_AES_128_GCM_SHA256") == 1 result: 0 != 1 failed ( 0.00002) ERROR - tests/api.c line 36346 failed with: expected: Test failed result: ret 0

github-actions · 2026-06-22T15:13:23Z

MemBrowse Memory Report

gcc-arm-cortex-m4-openssl-compat

FLASH: .text +1,216 B (+0.2%, 782,060 B / 1,048,576 B, total: 75% used)

linuxkm-standard

Data: __patchable_function_entries +32 B (+0.1%, 49,792 B)
No memory changes detected for:
gcc-arm-cortex-m0plus
gcc-arm-cortex-m3
gcc-arm-cortex-m4
gcc-arm-cortex-m4-baremetal
gcc-arm-cortex-m4-crypto-only
gcc-arm-cortex-m4-dtls13
gcc-arm-cortex-m4-min-ecc
gcc-arm-cortex-m4-pkcs7
gcc-arm-cortex-m4-pq
gcc-arm-cortex-m4-rsa-only
gcc-arm-cortex-m4-sp-math
gcc-arm-cortex-m4-tls12
gcc-arm-cortex-m4-tls13
gcc-arm-cortex-m7
gcc-arm-cortex-m7-pq
gcc-arm-cortex-m7-tls13
linuxkm-pie
stm32-sim-stm32h753

dgarske

Skoll Code Review

Scan type: reviewOverall recommendation: COMMENT
Findings: 13 total — 4 posted, 9 skipped
4 finding(s) posted as inline comments (see file-level comments below)

Posted findings

[Medium] wolfSSL_EC_KEY_oct2key does not record the point conversion form — src/pk_ec.c:4731-4775
[Medium] wolfssl_x509_ext_is_set() silently reports 'absent' for every unrecognized NID — src/x509.c:3601-3622
[Medium] wolfSSL_X509_add_ext() now fails for extKeyUsage object stacks that previously returned success — src/x509.c:1477-1488
[Low] X509V3_ADD_SILENT has no observable effect — src/x509.c:3740-3778

Skipped findings

[High] ~~X509V3_ADD_REPLACE destroys the existing extension when the subsequent add fails~~
[High] ~~basicConstraints pathLenConstraint is taken from ASN1_INTEGER-length (DER size), not the integer value~~
[High] ~~Removing subjectAltName leaves the altName iterator permanently exhausted~~
[Medium] ~~basicConstraints test does not verify the pathlen round-trip~~
[Medium] ~~EC_KEY_oct2key test only covers the uncompressed form and hardcodes the P-256 size~~
[Medium] ~~X509V3_ADD_APPEND does not append for most NIDs~~
[Low] ~~Flag test computes dnsName/dnsLen but never asserts on them for APPEND and KEEP_EXISTING~~
[Low] ~~SSL_set_ciphersuites alias accepts TLS 1.2 cipher names~~
[Low] ~~test_san_first_dns() truncates into a static buffer while reporting the untruncated length~~

Review generated by Skoll

dgarske · 2026-08-03T21:34:03Z

+ * @return  1 on success.
+ * @return  0 on failure.
+ */
+int wolfSSL_EC_KEY_oct2key(WOLFSSL_EC_KEY *key, const unsigned char *buf,


🟠 [Medium] wolfSSL_EC_KEY_oct2key does not record the point conversion form

OpenSSL's EC_KEY_oct2key() saves the encoding form on the key (key->conv_form = (point_conversion_form_t)(buf[0] & ~0x01)), so that a later i2o_ECPublicKey() / EC_KEY_key2buf() re-emits the same form. The wolfSSL implementation never touches key->form, so a compressed input silently round-trips back as uncompressed.

Verified empirically on this branch:

compressed enc len=33 first=0x03 oct2key(compressed)=1 conv_form after=4 (OpenSSL sets 2)

This matters for applications that hash or compare the re-encoded public key (e.g. JWK/COSE key thumbprints, TLS raw public keys), because they will get a different byte string than OpenSSL would.

Fix: Call wolfSSL_EC_KEY_set_conv_form(key, buf[0] & ~0x01) on success, and document the behaviour in the function's doxygen block.

dgarske · 2026-08-03T21:34:03Z

    return NULL;
 }

+/* Return 1 if an extension of type @nid is already present in the in-memory


🟠 [Medium] wolfssl_x509_ext_is_set() silently reports 'absent' for every unrecognized NID

The default: arm returns 0 for any NID outside the six recognized ones. wolfSSL_X509V3_EXT_i2d() also accepts WC_NID_issuer_alt_name and WC_NID_info_access, and with WOLFSSL_CUSTOM_OID wolfSSL_X509_add_ext() will happily store arbitrary NIDs in x509->custom_exts[]. For those NIDs the presence-dependent operations quietly do the wrong thing rather than reporting that they are unsupported:

X509V3_ADD_DEFAULT never detects a duplicate, so repeated calls keep appending custom extensions until NUM_CUSTOM_EXT is exhausted (then wolfssl_x509_add_custom_ext() fails with a confusing 'Bad value for customExtCount').

X509V3_ADD_KEEP_EXISTING always adds instead of keeping.

X509V3_ADD_DELETE / X509V3_ADD_REPLACE_EXISTING always fail with 'No extension to delete', which is indistinguishable from the genuinely-absent case.

None of these NIDs are covered by the new tests.

Fix: Have wolfssl_x509_ext_is_set() report 'unknown NID' distinctly and make wolfSSL_X509_add1_ext_i2d() fail the presence-dependent operations (DEFAULT / KEEP_EXISTING / REPLACE_EXISTING / DELETE) for those NIDs, or at minimum document the limitation in the doxygen block alongside the extKeyUsage note. Add a test for NID_info_access / a custom OID.

dgarske · 2026-08-03T21:34:03Z

                return WOLFSSL_FAILURE;
            }
        }
+        else if (ext && ext->ext_sk != NULL) {


🟠 [Medium] wolfSSL_X509_add_ext() now fails for extKeyUsage object stacks that previously returned success

wolfSSL_X509_add_ext(x509, ext, -1) with an extKeyUsage extension carrying an ext_sk object stack used to return WOLFSSL_SUCCESS (a silent no-op); it now returns WOLFSSL_FAILURE. Turning a silent no-op into an explicit failure is the right call, but it is a visible behaviour change for existing OPENSSL_ALL applications that build an EKU extension with X509V3_EXT_i2d(NID_ext_key_usage, ...) and then X509_add_ext() it - those calls will now start failing.

Blast radius checked: wolfSSL_X509_set_ext() (the source of extensions returned by X509_get_ext()) only sets ext_sk for AUTH_INFO_OID and ALT_NAMES_OID and always populates ext->value.data, so the common 'copy extensions from one cert to another' pattern still takes the first branch and is unaffected. tests/api/test_ossl_x509_ext.c:391 (No Data - no change -> SUCCESS) also still passes because that ext has neither data nor ext_sk.

The underlying gap remains: wolfSSL cannot consume an EKU object stack at all. The OIDs in the stack could be mapped to the EXTKEYUSE_* bits the same way ParseExtKeyUsageStr() maps names.

Fix: Either implement the OID -> EXTKEYUSE_* mapping so extKeyUsage actually works through X509_add1_ext_i2d(), or call out the behaviour change in the PR description / release notes so downstream users of X509_add_ext() are not surprised.

dgarske · 2026-08-03T21:34:03Z

+    op = flags & WOLFSSL_X509V3_ADD_OP_MASK;
+    exists = wolfssl_x509_ext_is_set(x, nid);
+
+    switch (op) {


🔵 [Low] X509V3_ADD_SILENT has no observable effect

WOLFSSL_X509V3_ADD_SILENT only gates WOLFSSL_MSG() calls, which are already compiled out unless DEBUG_WOLFSSL is defined, and wolfSSL does not push to an OpenSSL-style error queue here. The three if ((flags & WOLFSSL_X509V3_ADD_SILENT) == 0) wrappers therefore add branching and indentation for no behavioural difference. The doxygen block already states that SILENT does not change the result, so a single note would be clearer than three dead conditionals.

Fix: Either drop the three SILENT conditionals and note once that the flag is accepted-but-inert, or keep them and add a brief comment explaining they are placeholders for a future error-queue implementation.

dgarske

#10520 (review)

julek-wolfssl · 2026-09-03T10:42:38Z

Retest this please

julek-wolfssl · 2026-09-03T13:35:46Z

Retest this please

… of wolfssl

cipher_names[] only holds the IANA suite name when error strings are built in, so lean builds could not match "TLS_AES_128_GCM_SHA256". Gate the test on BUILD_TLS_AES_128_GCM_SHA256 and fall back to the wolfSSL suite name. Also build the basicConstraints ASN1_OBJECT with OBJ_nid2obj() so it carries a real OID encoding, not just a hand-set type.

Removing an extension left state behind: the freed altName list was still referenced by the altNamesNext retrieval hint (use after free through X509_get_next_altname()), and the cached subject key ID string, the encoded extended key usage and the per-extension set flags kept the removed extension observable. X509V3_ADD_SILENT only suppresses error reporting in OpenSSL, so X509V3_ADD_DEFAULT on an existing extension fails either way. REPLACE and REPLACE_EXISTING now encode the new extension before dropping the old one, so a failure no longer destroys the existing extension.

OpenSSL saves the encoding form on the key so that a later i2o_ECPublicKey() re-emits it. Without this a compressed point silently round-tripped back as uncompressed.

wolfSSL_X509V3_EXT_i2d() encodes extKeyUsage as a stack of ASN1_OBJECTs, which wolfSSL_X509_add_ext() could not use at all. Map each KeyPurposeId OID to its EXTKEYUSE_* bit and store the list as DER, the layout a parsed certificate carries, so X509_get_ext_d2i() and X509_get_extended_key_usage() both read it back. OIDs wolfSSL has no bit for stay in the list, matching the certificate parser. Entries that are not a single OBJECT IDENTIFIER are rejected before anything is stored.

wolfssl_x509_ext_is_set() answered "absent" for every NID it does not know, so DELETE and REPLACE_EXISTING failed with a reason indistinguishable from a real absence, and DEFAULT and KEEP_EXISTING could not detect a duplicate. It now reports "unknown" separately and only X509V3_ADD_APPEND, which is defined as not checking, is attempted for such a NID. X509V3_ADD_SILENT only gated WOLFSSL_MSG() calls that are compiled out without DEBUG_WOLFSSL, and this API pushes nothing to an error queue. Drop the three dead conditionals and document the flag as accepted but inert.

wolfSSL_X509V3_EXT_i2d() builds the authorityKeyIdentifier object from the caller's issuer name when no key ID is given, and wolfSSL_X509_add_ext() dispatches on that object rather than on the NID that was asked for. A REPLACE therefore removed the certificate's authority key identifier and then failed, or wrote into a different extension's storage. Check the encoded NID before anything is removed.

wolfSSL_X509V3_EXT_i2d() copies the object stack without inspecting it, so a REPLACE dropped the certificate's extKeyUsage and only then discovered that the new stack could not be stored. Split the per-entry check out of wolfssl_x509_add_ext_key_usage_sk() and run it before the removal.

julek-wolfssl · 2026-09-03T16:31:57Z

retest this please

julek-wolfssl · 2026-09-04T08:37:01Z

@dgarske addressed review

julek-wolfssl · 2026-09-04T10:43:51Z

Not sure why this got closed. Re-opened as #11375 to make it easier for me to track.

dgarske assigned wolfSSL-Bot May 22, 2026

Roy-Carter force-pushed the feature/extending_comp_layer branch from 54ef2e4 to 413e8ac Compare May 24, 2026 11:04

Roy-Carter force-pushed the feature/extending_comp_layer branch from 413e8ac to 62138a2 Compare May 29, 2026 12:42

dgarske requested a review from julek-wolfssl May 29, 2026 15:25

julek-wolfssl requested a review from Copilot May 31, 2026 15:01

Copilot started reviewing on behalf of julek-wolfssl May 31, 2026 15:02 View session

Copilot AI reviewed May 31, 2026

View reviewed changes

Comment thread src/x509.c Outdated

Comment thread src/x509.c

Comment thread tests/api.c

Comment thread tests/api.c

julek-wolfssl requested changes Jun 1, 2026

View reviewed changes

Roy-Carter force-pushed the feature/extending_comp_layer branch from 62138a2 to 653ec68 Compare June 7, 2026 07:53

Roy-Carter requested a review from julek-wolfssl June 7, 2026 07:54

julek-wolfssl previously approved these changes Jun 8, 2026

View reviewed changes

julek-wolfssl assigned Roy-Carter and unassigned wolfSSL-Bot Jun 8, 2026

Roy-Carter dismissed julek-wolfssl’s stale review via f2a562a June 21, 2026 12:39

Roy-Carter force-pushed the feature/extending_comp_layer branch from f2a562a to 99427a3 Compare June 21, 2026 12:44

dgarske requested changes Jun 22, 2026

View reviewed changes

julek-wolfssl force-pushed the feature/extending_comp_layer branch from 99427a3 to 76dec0a Compare July 28, 2026 14:26

julek-wolfssl assigned julek-wolfssl and unassigned Roy-Carter Jul 28, 2026

julek-wolfssl requested a review from Copilot July 28, 2026 16:49

Copilot started reviewing on behalf of julek-wolfssl July 28, 2026 16:53 View session

julek-wolfssl removed their assignment Jul 30, 2026

dgarske self-requested a review August 3, 2026 21:00

dgarske reviewed Aug 3, 2026

View reviewed changes

dgarske assigned Roy-Carter Aug 3, 2026

dgarske requested changes Aug 3, 2026

View reviewed changes

julek-wolfssl assigned julek-wolfssl and unassigned wolfSSL-Bot and Roy-Carter Aug 27, 2026

julek-wolfssl force-pushed the feature/extending_comp_layer branch from 76dec0a to 51dc78f Compare September 2, 2026 17:05

Roy-Carter and others added 11 commits September 3, 2026 13:41

Introduce missing functionality that was found in compatibility layer…

386dda0

… of wolfssl

Refactor - Fix PR notes

0de77c4

Bugfix - fix tests failure

a35531e

Keep new test comment within 80 columns

b340414

Record the point conversion form in EC_KEY_oct2key

60499d6

OpenSSL saves the encoding form on the key so that a later i2o_ECPublicKey() re-emits it. Without this a compressed point silently round-tripped back as uncompressed.

julek-wolfssl force-pushed the feature/extending_comp_layer branch from 51dc78f to 6225adf Compare September 3, 2026 13:50

julek-wolfssl closed this Sep 4, 2026

julek-wolfssl assigned wolfSSL-Bot and unassigned julek-wolfssl Sep 4, 2026

julek-wolfssl mentioned this pull request Sep 4, 2026

Fix X509 extension handling (X509V3_ADD_*, extKeyUsage, EC point form) and cipher suite test #11375

Open

Uh oh!

Conversation

Roy-Carter commented May 22, 2026

Description

Testing

Checklist

Uh oh!

wolfSSL-Bot commented May 22, 2026

Uh oh!

Roy-Carter commented May 22, 2026

Uh oh!

Roy-Carter commented May 24, 2026

Uh oh!

Roy-Carter commented May 29, 2026

Uh oh!

dgarske commented May 29, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

julek-wolfssl left a comment

Choose a reason for hiding this comment

Uh oh!

Roy-Carter commented Jun 7, 2026

Uh oh!

julek-wolfssl commented Jun 8, 2026

Uh oh!

Roy-Carter commented Jun 21, 2026

Uh oh!

dgarske Jun 22, 2026

Choose a reason for hiding this comment

Uh oh!

github-actions Bot commented Jun 22, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

MemBrowse Memory Report

gcc-arm-cortex-m4-openssl-compat

linuxkm-standard

Uh oh!

dgarske left a comment

Choose a reason for hiding this comment

Skoll Code Review

Posted findings

Skipped findings

Uh oh!

dgarske Aug 3, 2026

Choose a reason for hiding this comment

Uh oh!

dgarske Aug 3, 2026

Choose a reason for hiding this comment

Uh oh!

dgarske Aug 3, 2026

Choose a reason for hiding this comment

Uh oh!

dgarske Aug 3, 2026

Choose a reason for hiding this comment

Uh oh!

dgarske left a comment

Choose a reason for hiding this comment

Uh oh!

julek-wolfssl commented Sep 3, 2026

Uh oh!

julek-wolfssl commented Sep 3, 2026

Uh oh!

julek-wolfssl commented Sep 3, 2026

Uh oh!

julek-wolfssl commented Sep 4, 2026

Uh oh!

julek-wolfssl commented Sep 4, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

github-actions Bot commented Jun 22, 2026 •

edited

Loading