cert_type, key, key_type, key_passwd and key_blob lived in
ssl_config_data but not in ssl_primary_config, so they were invisible to
match_ssl_primary_config() and to the TLS session cache peer key.
Two easy handles sharing a connection pool could reuse each other's
authenticated connections when they differed only on SSLKEY, SSLKEYTYPE,
KEYPASSWD, SSLCERTTYPE or SSLKEYBLOB. The second handle would silently
inherit the first handle's authenticated identity.
Promote all five fields into ssl_primary_config so the conn-reuse
predicate and session cache key cover the complete client credential
set. Also replace the fixed ":CCERT" session cache marker with the
actual clientcert path so sessions are not shared across different
client certificates.
Verified by test 3303 and 3304
Reported-By: Joshua Rogers (AISLE Research)
Closes#21667
Both are filesystem paths (or case-sensitive hash strings for
pinned_key). curl_strequal is case-insensitive and would treat
/etc/ssl/Crl.pem and /etc/ssl/crl.pem as the same file, unlike the other
path fields (CApath, CAfile, issuercert, clientcert) which already use
Curl_safecmp.
Closes#21668
Curl_ssl_peer_key_make() omitted ssl->signature_algorithms, although
match_ssl_primary_config() compares the field. Two handles differing
only in CURLOPT_SSL_SIGNATURE_ALGORITHMS therefore shared a peer key and
could resume each other's sessions across a shared CURLSH SSL session
cache.
Add :SIGALGS-%s next to the other ssl_primary_config fields.
Closes#21651
The dh(g) parameter was read from param->beg instead of from the
cursor p returned by parsing dh(p). This caused dh(g) to always
report the same value as dh(p) when inspecting DH certificates
via CURLOPT_CERTINFO on non-OpenSSL backends.
The DSA branch correctly advances the cursor; the DH branch lost
this during what appears to be a copy-paste.
Add unit1676 to verify that dh(p) and dh(g) report distinct values
using a hand-crafted minimal DER certificate.
Assisted by: Claude Opus 4.6
Signed-off-by: Sergio Correia <scorreia@redhat.com>
Closes#21595
To make them more accurate.
Also:
- show Visual Studio version, where missing.
- ease the formatting.
- schannel_int.h: clang-tidy fallback code.
Used: `rg -l --sort=path CERT_FIND_HAS_PRIVATE_KEY`
Closes#21621
- disable DES with nettle 4. It no longer supports it.
```
lib/curl_ntlm_core.c:67:12: fatal error: 'nettle/des.h' file not found
67 | # include <nettle/des.h>
| ^~~~~~~~~~~~~~
```
- fix MD4 support with nettle 4.
```
lib/md4.c:178:36: error: too many arguments to function call, expected 2, have 3
178 | md4_digest(ctx, MD4_DIGEST_SIZE, digest);
| ~~~~~~~~~~ ^~~~~~
```
- fix unused argument compiler warning:
```
lib/vtls/gtls.c:2267:39: error: unused parameter 'sha256len' [clang-diagnostic-unused-parameter,-warnings-as-errors]
2267 | size_t sha256len)
| ^
```
Ref: https://github.com/curl/curl/actions/runs/25710321195/job/75488970143?pr=21557
- GHA/macos: stop enabling NTLM in the GnuTLS job.
It no longer builds due to missing DES support in nettle 4.
```
lib/curl_ntlm_core.c:90:4: error: "cannot compile NTLM support without a crypto library with DES."
90 | # error "cannot compile NTLM support without a crypto library with DES."
| ^
```
Ref: https://github.com/curl/curl/actions/runs/25710321195/job/75488970170?pr=21557
Follow-up to cfadbaa133#21169Closes#21557
The code would previously read one byte past the provided
CURLOPT_CAINFO_BLOB if the blob ends exactly with -----BEGIN
CERTIFICATE-----
Reported-by: Andrew Nesbit
Closes#21543
- passing an unknown string to CURLOPT_ECH now returns error
To properly allow applications to spot if they pass in a typo or
something to libcurl.
- CURLECH_DISABLE is now a plain zero internally, not a dedicated bit which
simplifies checks for when ECH is enabled
- Dropped the CURLECH_CLA_CFG bit, and just check STRING_ECH_CONFIG
- Turn grease/enable/hard into three different numerical values, no bitmask
needed
- Convert the struct field 'tls_ech' from an int to a byte.
Closes#21532
Unfortunately, mbedtls_pk_parse_key() requires the data to be
null-terminated if the data is PEM encoded (even when provided the exact
length), so this function needs to make a copy that has one.
Reported-by: Elise Vance
Closes#21515
`struct Curl_peer` keeps information about a communication endpoint
together. It will replace `conn->host` and `conn->conn_to_host` and
proxyinfo host. It will also become part of `struct ssl_peer`.
It has a reference counter, so an instance can be shared between
connections and filters.
Elminiates `conn->host` and `conn->connect_to_host`, used in the
proxyinfo structures. Passed to DNS resolution and socks filters, etc.
Pass peer to http proxy and socks tunnel filters. Use peer in dns filter
and resolving. Make `Curl_peer` a member in the `struct ssl_peer`.
Add `docs/internals/PEERS.md` for documentation.
Closes#21472
failf() needs an easy handle to work. This change removes the call since
there is normnally nowhere to show the output if init fails.
Bonus: improve language in an infof() call
Spotted by Copilot
Closes#21441
Several mbedTLS resources (entropy/CTR-DRBG, CA/client certs, keys, CRL)
are initialized and may allocate memory before initialized is set, and
must still be cleaned up.
Follow-up to 1c4813c769
Caught by Codex Security
Closes#21440
Before this fix, Curl_tls_keylog_open() assigned the environment
variable result to a global keylog_file_name without freeing any prior
allocation. If the file cannot be opened (e.g., permission error)
keylog_file_fp stays NULL, so subsequent calls to Curl_tls_keylog_open
will overwrite keylog_file_name and leak the previous allocation.
Spotted by Codex Security
Closes#21427
Seen with unity, H3, wolfssl with `HAVE_EX_DATA`.
Fixing:
```
lib/vtls/wolfssl.c:412:10: error: no previous prototype for function 'Curl_wssl_cache_session' [-Wmissing-prototypes]
412 | CURLcode Curl_wssl_cache_session(struct Curl_cfilter *cf,
| ^
lib/vtls/wolfssl.c:412:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
412 | CURLcode Curl_wssl_cache_session(struct Curl_cfilter *cf,
| ^
| static
1 error generated.
```
Follow-up to cc5c1553fb#19852Closes#21392
Track lock status of session cache, add DEBUGASSERT()s for proper
calling sequences. Add check of lock status.
Use lock status check to prevent reentry of import/export calls.
Deny such calls with CURLE_RECURSIVE_API_CALL.
Closes#21383
Also fix clang-tidy `readability-redundant-control-flow`.
```
lib/vtls/openssl.c:4055:46: error: unused parameter 'reason' [-Wunused-parameter]
4055 | int reason)
| ^
lib/vtls/openssl.c:4279:19: error: variable 'status' set but not used [-Wunused-but-set-variable]
4279 | const char *status = NULL;
| ^
```
Ref: https://github.com/curl/curl/actions/runs/24641388439/job/72045859306?pr=21379
```
lib/vtls/openssl.c:4092:7: error: Value stored to 'rv' is never read [clang-analyzer-deadcode.DeadStores]
4092 | rv = SSL_ech_get1_status(ssl, &inner, &outer);
| ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
lib/vtls/openssl.c:4092:7: note: Value stored to 'rv' is never read
4092 | rv = SSL_ech_get1_status(ssl, &inner, &outer);
| ^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
lib/vtls/openssl.c:4112:3: error: redundant return statement at the end of a function with a void return type [readability-redundant-control-flow]
4111 | #endif
| ~~~~~~
4112 | return;
| ^~~~~~~
4113 | }
```
Ref: https://github.com/curl/curl/actions/runs/24642154148/job/72047919525?pr=21379#step:42:186
Cherry-picked from #21379Closes#21380
- make sure all UNITTEST prototypes mark in which unit test they are used,
with "@unittest" markup
- make sure all UNITTEST functions do not use Curl_ prefix, as that is a
prefix we use for global private functions and these functions are static
and therefore not global and the prefix is wrong
- drop UNITTEST for functions not used in unit tests
- make the extract-unit-protos script highlight the above issues if found
- extract-unit-protos now also outputs the unit test number for all the
generated protos in lib/unitprotos.h to aid readers. It also adds the source
file and line number where the proto originates from.
- extract-unit-protos now exits with a non-zero value if any of the above
warnings are triggered
- cf-dns: Curl_cf_dns_result => static cf_dns_result
- hostip: Curl_ipv6works => static ipv6works
- url: remove Curl_setup_conn() - not used anymore
- connect: Curl_timeleft_now_ms => UNITTEST timeleft_now_ms
Closes#21330
- enable three checks:
- bugprone-invalid-enum-default-initialization
- bugprone-sizeof-expression
- readability-inconsistent-declaration-parameter-name (strict)
- fix remaining discrepancies with arg names in prototypes
and implementation, in strict mode.
- document reason for some checks tested but not enabled.
Closes#20794
- examples: sync debug output printf masks.
- INSTALL-CMAKE.md: tidy up section for some options.
- curl_sha512_256: delete comment suggesting an optimization.
- vtls/keylog: scope a variable.
- vtls/openssl: make a source code URL a permalink.
- vtls/schannel: drop redundant parentheses.
- test1119.pl: robustify `$1` -> `$s`.
- sync arg names in comments to match the code.
- tidy up and minor fixes to comments.
- fix formatting/indenting/comment/newline/include nits.
- move `UNITTEST` protos next to definitions, sync their argument names.
- make variables static.
- add parentheses to Perl `open()` calls.
- drop unnecessary double quotes in Perl.
- clang-format.
Closes#21000
When HTTPS-RR is needed for the HTTP/3 handshake, delay the connect
until it arrives. Relevant only for TLS backends that support ECH, for
now.
Closes#21253
The making of the TLS session cache key should use the cert blob
independently of verifypeer on/off.
Follow-up to fa0ccd9f1f
Spotted by Codex Security
Closes#21222
It did not require a full-length match, so empty or prefix tokens map to
ECJPAKE would silently add that cipher to the configured list.
Follow-up to fba9afebba
Reported by Codex Security
Closes#21264
To sync names for the same macro logic between lib and src, and to move
it to the curlx namespace, to match `curlx_free()` that it's calling.
Closes#21151
- allow to specify when they are wanted on starting a resolve
- match dns cache entries accordingly. An entry which never
tried to get HTTPS-RRs is no answer for a resolve that wants
it.
- fix late arrivals of resolve answers to match the "async"
records that started them - if it still exists.
- provide for multiple "async" resolves in a transfer at the
same time. We may need to resolve an IP interface while the
main connection resolve has not finished yet.
- allow lookup of HTTPS-RR information as soon as it is
available, even if A/AAAA queries are still ongoing.
For this, the "async" infrastructure is changed:
- Defined bits for DNS queries `CURL_DNSQ_A`, `CURL_DNSQ_AAAA`
and `CURL_DNSQ_HTTPS`. These replace `ip_version` which says
nothing about HTTPS.
Use them in dns cache entries for matching.
- enhance the `async->id` to be a unique `uint32_t` for
resolves inside one multi. This is weak, as the id may
wrap around. However it is combined with the `mid` of
the easy handle, making collisions highly unlikely.
`data->state.async` is only accessed in few places where
the mid/async-id match is performed.
- vtls: for ECH supporting TLS backends (openssl, rustls, wolfssl),
retrieve the HTTPS-RR information from the dns connection filter.
Delay the connect if the HTTPS-RR is needed, but has not
been resolved yet.
The implementation of all this is complete for the threaded
resolver. c-ares resolver and DoH do not take advantage of
all new async features yet. To be done in separate PRs.
Details:
c-ares: cleanup settings and initialisation. Any ares channel
is only being created on starting a resolve and propagating
operations in setopt.c to the channel are not helpful.
Changed threaded+ares pollset handling so that they do not
overwrite each others `ASYNC_NAME` timeouts.
Add trace name 'threads' for tracing thread queue and
pool used by threaded resolver.
Closes#21175
