When `CURLOPT_PINNEDPUBLICKEY` is configured alongside options that disable
standard peer verification (`CURLOPT_SSL_VERIFYPEER = 0` and
`CURLOPT_SSL_VERIFYHOST = 0`), libcurl fails to enforce public key pinning on
connections established without a presented server certificate. Bypassing the
pinning check under these disabled-verification conditions allows
unauthenticated connections to succeed when they should be rejected.
When performing transfers via libcurl’s multi interface, pooled TLS
connections can outlive their originating easy handles. In OpenSSL 3 provider
configurations, libcurl attaches an allocated library context to the easy
handle's state and passes it to OpenSSL without acquiring an ownership
reference; destroying the easy handle prematurely frees this context while the
active connection retains a dangling pointer, leading to a heap-use-after-free
upon subsequent I/O or post-handshake operations.
A flaw in libcurl's handling of HTTP/2 Server Push streams, when the parent
handle is set to share connections with other handles, can lead to
use-after-free in the cleanup process.
A flaw in the libcurl SASL negotiation for LDAP authentication allows an
incomplete handshake sequence to be misinterpreted as a successful
cryptographic verification. An attacker executing a Man-in-the-Middle (MITM)
attack can inject a premature or shortcut response that bypasses complete peer
validation.
When a libcurl-based application performs transfers via `SCP://` or `SFTP://`
and utilizes the `CURLOPT_SSH_KEYFUNCTION` callback, it may silently accept an
untrusted server. This vulnerability occurs when a server presents a host key
type that does not match the specific key type already recorded for that host
in the `known_hosts` file. Instead of rejecting the mismatch, the callback
mechanism fails to properly enforce the restriction, allowing the connection
to succeed without warning and risking a potential man-in-the-middle attack.
In this scenario, libcurl first uses a proper HTTP/3 server for the initial
transfers, and when it makes a second transfer to the same site it has been
replaced by the attacker's impostor machine - without a valid certificate.
When libcurl returns to the hostname the second time with a cached SSL session
(`CURLOPT_SSL_SESSIONID_CACHE` is not disabled) and early data enabled (the
`CURLSSLOPT_EARLYDATA` bit is set in `CURLOPT_SSL_OPTIONS`), libcurl might
send off the second request's bytes on that new connection *before* enforcing
the certificate verification failure. Potentially leaking sensitive
information.
Calling `curl_easy_pause()` within the event-based `CURLMOPT_SOCKETFUNCTION`
callback triggers a use-after-free vulnerability, where libcurl attempts to
store a flag using a dangling struct pointer immediately after that pointer's
memory has been freed.
libcurl had a flaw that when instructed to clear proxy authentication
credentials which made it not do so, leaving the old credentials around to get
used for subsequent transfers that should not know nor use them.
libcurl would reuse a previously created connection even when some mTLS config
related option had been changed that should have prohibited reuse.
libcurl keeps previously used connections in a connection pool for subsequent
transfers to reuse if one of them matches the setup. However, some TLS
settings related to client certificates were left out from the configuration
match checks, making them match too easily. In particular options related to
the private key.
When reusing a libcurl handle for sequential transfers driven by
environment-variable proxy configuration, libcurl fails to clear the proxy
authentication state between requests. Specifically, if the initial transfer
authenticates against `proxyA` using Digest auth, a subsequent transfer routed
through `proxyB` erroneously leaks the `Proxy-Authorization:` header intended
solely for `proxyA`.
When asking curl to use a `.netrc` file to find credentials and at the same
time specifying a URL with a username (without a password), like
`https://[email protected]/`, curl could wrongly get and use the password for
*another* user set in the `.netrc` file for that host if such a one exists and
there is no match for the specified user.
The curl logic that works with SASL authentication could end up cleaning up
the GSASL context *twice* without clearing the pointer in between, making it
`free()` the same pointer twice.
A flaw in curl’s cookie parsing logic allows a malicious HTTP server to set
"super cookies" that bypass the Public Suffix List check. This enables an
attacker-controlled origin to inject cookies that curl subsequently scopes and
transmits to unrelated third-party domains.
libcurl might in some circumstances reuse the wrong connection when asked to
do Negotiate-authenticated ones, even when they are set to use different
"services".
libcurl features a pool of recent connections so that subsequent requests can
reuse an existing connection to avoid overhead.
When reusing a connection a range of criteria must be met. Due to a logical
error in the code, a request that was issued by an application could
wrongfully reuse an existing connection to the same server that was
authenticated using different services.
A vulnerability exists where a new transfer that uses STARTTLS to upgrade the
connection might reuse an existing live connection even though the TLS
configuration mismatches so it should not.
When a user invokes curl using a schemeless URL combined with
`--proto-default` sftp (or scp), a disconnect occurs between the tool layer
and libcurl. The tool layer incorrectly infers the URL scheme, which
erroneously bypasses the initialization of critical SSH security options like
CURLOPT_SSH_HOST_PUBLIC_KEY_SHA256 and CURLOPT_SSH_KNOWNHOSTS. Conversely, the
libcurl runtime successfully honors CURLOPT_DEFAULT_PROTOCOL and establishes
the connection via SFTP/SCP as specified. Because the tool layer skipped the
security configuration, these SSH host verification options are silently
omitted, causing curl to connect to an unverified SSH remot
Successfully using libcurl to do a transfer to a specific HTTP origin
(`hostA`) with **Digest** authentication and then changing the origin to a
different one (`hostB`) for a second transfer, reusing the same handle, makes
libcurl wrongly pass on the `Authorization:` header field meant for `hostA`,
to `hostB`.
By default, curl automatically responds to WebSocket PING frames. Because curl
lacks an upper bound on memory allocation for unacknowledged frames, a
malicious server can exhaust all available memory by flooding curl with rapid,
sequential PING messages.
libcurl keeps previously used connections in a connection pool for subsequent
transfers to reuse if one of them matches the setup.
An easy handle that first uses default native CA trust can continue trusting
the native platform store after the application switches that same handle to
custom CA material for a later transfer.
An issue in curl’s QUIC UDP receive function allows a malicious HTTP/3 server
to trigger a remote denial of service against a curl or libcurl client.
Because the helper function discards zero-length UDP datagrams before counting
them toward the per-call packet budget, a connected QUIC peer can continuously
stream empty datagrams to indefinitely stall the client.