FreeRDP vulnerabilities

FreeRDP 3.0.0 before 3.32.0 contains an insufficiently protected credentials vulnerability where the /restricted-admin option fails to suppress CredSSP credential delegation, sending the cleartext password to the RDP server. A compromised or man-in-the-middle destination host can decode TSPasswordCreds and the Client Info PDU to obtain reusable administrative domain, username, and password values.

The FreeRDP client 3.0.0 before 3.32.0 does not check a server-selected PROTOCOL_RDSTLS or PROTOCOL_HYBRID_EX against its enabled protocol set in nego_process_negotiation_response(), allowing a security protocol downgrade. A malicious server or man-in-the-middle can select RDSTLS to override /sec:nla pinning, complete authentication with an all-zero cookie, and receive the user's cleartext password in the Client Info PDU.

FreeRDP 3.17.1 before 3.32.0 contains a heap buffer over-read vulnerability in DynvcTrackerHandleCreateBack of the freerdp-proxy DRDYNVC channel handler that allows malicious backend RDP servers to read heap memory out of bounds. Attackers can send a CREATE_REQUEST_PDU whose channel name lacks a NUL byte, causing an unbounded snprintf read that leaks memory into proxy logs or crashes the process.

FreeRDP from 3.0.0 before 3.32.0 contains a heap out-of-bounds read vulnerability. A malicious RD Gateway can set MsgBytes far larger than the NDR string ActualCount in consent or service messages, causing UTF-8 conversion to read past the stream allocation and crash the client.

FreeRDP 3.0.0 before 3.32.0 contains an improper restriction of excessive authentication attempts in its virtual smartcard GIDS implementation, where the PIN retry counter is stored per-handle. A malicious RDP server can exhaust the counter with VERIFY APDUs, then reset attempts via a new SCardConnect, enabling unbounded online brute-forcing of the smartcard PIN.

FreeRDP from 3.0.0 before 3.32.0 contains an uninitialized memory disclosure vulnerability in smartcard_Transmit_Call. A malicious RDP server can send SCardTransmit requests with an invalid card handle and nonzero receive length, causing smartcard_pack_transmit_return to serialize up to 66,560 uninitialized heap bytes.

FreeRDP from 3.20.1 before 3.32.0 contains an uncontrolled resource allocation vulnerability. A malicious RDP server can send IRP_MJ_READ with Length 0xffffffff, causing roughly 4 GiB of zeroed allocation on 64-bit builds or an under-reserved buffer on 32-bit builds.

FreeRDP 3.11.0 before 3.32.0 does not validate that AVC444 and AVC444v2 region widths are even before YUV444 conversion, so an odd width reaches an assertion and aborts the client. A malicious or compromised RDP server can terminate a connected FreeRDP client after RDPGFX AVC444 negotiation by sending an in-bounds region rectangle with an odd width.

FreeRDP 3.0.0 before 3.32.0 contains a heap use-after-free vulnerability in smartcard emulation when SCARD_IOCTL_DISCONNECT and SCARD_IOCTL_TRANSMIT operate concurrently on one card handle. A malicious RDP server can overlap these IRPs through separate context workers to free and reuse the SCardHandle, causing client-process heap corruption and termination.

FreeRDP 3.0.0 before 3.32.0 does not remove card handles when Emulate_SCardReleaseContext releases a smartcard-emulation context, so Emulate_SCardTransmit later looks up the deleted parent context and reaches a terminating WINPR_ASSERT. A malicious RDP server can crash a connecting client with smartcard emulation and redirection enabled by releasing a context, establishing another, and issuing SCardTransmit with the stale card handle.

FreeRDP 3.29.0 before 3.32.0 contains a double free in the client's async-update message proxy, which shallow-copies the parser-owned PolygonSC or PolygonCB points array when numPoints is zero, so both the queue consumer and parser cleanup free it. A malicious RDP server can send a point-bearing polygon order followed by a zero-point order to crash a client with +async-update enabled.

FreeRDP 2.3.0 before 3.32.0 dereferences a NULL stream in rdg_process_close_packet in libfreerdp/core/gateway/rdg.c because rdg_process_control_packet accepts a header-only PKT_TYPE_CLOSE_CHANNEL packet and passes the null payload without a check. A malicious or compromised RD Gateway can send an eight-byte CLOSE_CHANNEL header to crash a connecting FreeRDP client.

FreeRDP from 2.3.0 before 3.32.0 contains an infinite loop vulnerability in http_chuncked_read that allows malicious RD Gateways to hang clients by sending incomplete chunked responses. Attackers can send a partial chunk-size line or footer followed by a clean TLS shutdown, causing zero-byte reads to loop forever and freeze the client decoder thread.

FreeRDP 3.0.0-beta1 through 3.32.0 on Unix builds with PKCS#11 support has a stack-based buffer overflow in wprintKeyName() in WinPR's NCrypt PKCS#11 provider. It hex-encodes a key's CKA_ID, which can be up to 64 bytes, into a fixed 128-byte stack buffer. An attacker supplying a malicious PKCS#11 token or module whose key and matching certificate use a 55- to 64-byte CKA_ID can overflow the client's stack during key enumeration, causing a crash.

FreeRDP 3.27.1 through 3.32.0 contains a NULL pointer dereference in update_read_cache_bitmap_v3_order() in libfreerdp/core/orders.c, which accesses the client bitmap cache when the CBR3_DO_NOT_CACHE flag is set, even though the cache is not yet initialized before Demand Active. A malicious or compromised RDP server can send a Cache Bitmap V3 order during connection setup to crash a connecting client that has enabled the non-default Bitmap Cache V3 feature (/cache:codec:<rfx|nsc|jpeg>).

Positive mapped-video coordinates overflow and abort FreeRDP clients. A malicious or compromised RDP server can terminate a FreeRDP client after the user connects with video redirection enabled.

FreeRDP 3.18.0 through 3.32.0 has a heap-based out-of-bounds write in freerdp_interruptible_get_line, which shrinks the buffer to 32 bytes while keeping a pre-filled suggested username's length as its write offset. A malicious or compromised ARM routing service can return a long redirectedUsername that reaches the CLI credential prompt, crashing the client when the user types input.

FreeRDP 3.0.0-beta4 through 3.32.0 contains a NULL pointer dereference in arm_handle_request_ok when an ARM endpoint returns an empty HTTP 200 body. Attackers controlling the configured ARM gateway can return a zero-length body to pass NULL to strnlen and crash the connecting client.

FreeRDP 3.0.0-beta1 through 3.32.0, when built with verbose WinPR assertions and using smartcard emulation, has a reachable assertion: the emulated SCardStatusA passes a NULL reader-name destination to scard_copy_strings. A malicious RDP server can send an SCardStatusA request with fmszReaderNamesIsNULL=1 to abort a connecting client.

FreeRDP from 2.0.0-beta1 before 3.28.0 contains an out-of-bounds read vulnerability in smartcard_LocateCardsByATRA_Call because the ATR pattern cbAtr byte count is not capped. Malicious RDP servers can send crafted ATR mask arrays to use query timing as an oracle, leaking heap memory bit-by-bit to defeat ASLR and aid RCE exploits.

FreeRDP from 3.0.0-beta1 before 3.32.0 contains a use-after-free vulnerability in smartcard handling that allows attackers to access a freed smartcard handle object after an invalid connect operation. Attackers can trigger an invalid smartcard connect operation to cause reuse of the freed handle, leading to memory corruption or crashes.

FreeRDP before 3.32.0 contains a double-free vulnerability in nsc_context_initialize_encode() that leaves dangling plane pointers after a partial buffer growth failure. Attackers using a crafted RDP client can negotiate NSCodec with a large desktop geometry to trigger allocation failure, crashing the server during context teardown.

FreeRDP 3.10.0 through 3.32.2 contains a heap buffer overflow in the WinPR png_write_data callback that writes past the buffer after a realloc failure. A malicious RDP clipboard peer can supply a large CF_DIB image that triggers out-of-bounds memcpy during PNG conversion, causing a client crash.

FreeRDP from 3.9.0 before 3.32.0 contains a NULL pointer dereference vulnerability in the RDPEAR NDR reader that allows malicious servers to crash clients via truncated KERB_RPC_INTERNAL_NAME data. Attackers can send a CreateApReqAuthenticatorReq with NameCount of two or more ending before the Names pointer, causing cleanup to dereference a null base and terminate the client.

FreeRDP before 3.32.0 stores the server-supplied signed MS-RDPEGT geometry origin unchecked in an unsigned video surface coordinate, so the blit clamp in gdiVideoShowSurface() wraps in 32-bit arithmetic and is skipped. A malicious RDP server can send a negative topLevelLeft to clients connecting with /video, causing an out-of-bounds write of decoded H.264 picture data or a client abort.

FreeRDP before 3.32.0 contains a path traversal vulnerability in convert_filedescriptors_to_file_list() in winpr/libwinpr/clipboard/synthetic_file.c. A malicious RDP server can supply '../' sequences so the user's next clipboard paste copies an arbitrary local file instead of the expected remote file.

FreeRDP before 3.32.0 contains a divide-by-zero vulnerability in the freerdp-shadow-cli shadow server: the frame id in a client's Frame Acknowledge PDU is stored without a range check, so the in-flight frame count in shadow_encoder_inflight_frames() can underflow. An authenticated client can acknowledge a frame id one greater than the last frame sent, causing shadow_encoder_create_frame_id() to divide by zero (SIGFPE) and terminate the process, disconnecting all viewers.

FreeRDP before 3.0.0 and 3.6.0 before 3.32.0 renders uninitialized heap memory from un-decoded progressive tiles because the tile->current buffer is allocated without zeroing. A malicious RDP server can send a crafted WBT_TILE_FIRST then WBT_TILE_UPGRADE sequence to paint heap bytes onscreen or abort the client process.

FreeRDP before 3.32.0 contains a NULL pointer dereference vulnerability in the client video redirection channel. A malicious server can send a Geometry CLEAR message that nulls presentation->geometry, then send VideoData that dereferences it, crashing the client.

FreeRDP before version 3.21.0 has a vulnerability in FastGlyph parsing where data length is not validated against the minimum size implied by cx/cy. A malicious server can trigger a client-side global buffer overflow, causing a crash (DoS).

FreeRDP is a free implementation of the Remote Desktop Protocol. Prior to 3.20.1, a race in the serial channel IRP thread tracking allows a heap use‑after‑free when one thread removes an entry from serial->IrpThreads while another reads it. This vulnerability is fixed in 3.20.1.

FreeRDP before 1.1.0-beta+2013071101 allows remote attackers to cause a denial of service (NULL pointer dereference and application crash) by disconnecting before authentication has finished.

FreeRDP before 1.1.0-beta1 allows remote attackers to cause a denial of service via NULL pointer dereference, leading to application crash.

Multiple integer overflows in client/X11/xf_graphics.c in FreeRDP allow remote attackers to have an unspecified impact via the width and height to the xf_Pointer_New or xf_Bitmap_Decompress function, which causes an incorrect amount of memory to be allocated.

FreeRDP server versions before 3.31.0 contain a protocol negotiation bypass vulnerability that allows unauthenticated attackers to establish RDSTLS connections despite server policy disabling them. Attackers can send incompatible protocol requests, receive negotiation failures, then complete TLS handshake and enter RDSTLS to bypass pre-authentication transport restrictions.

In FreeRDP before 3.28.0, freerdp_dsp_decode_opus calls Stream_EnsureRemainingCapacity on the context buffer even though opus_decode writes to the caller-supplied stream. A malicious RDP server negotiating WAVE_FORMAT_OPUS can cause large decoded frames to be written beyond the allocated buffer, leading to client heap corruption, crash, and potential code execution.

In FreeRDP before 3.27.0, rpc_client_recv_fragment ensures response reassembly stream capacity using only the server-declared alloc_hint rather than the actual StubLength. A malicious TS Gateway can send a response with a small alloc_hint and a much larger frag_length, causing a buffer overflow in the pdu->s buffer. This can crash the client and may permit code execution.

In FreeRDP before 3.27.0, clients negotiating RDPGFX AVC444 calculate the intermediate YUV444 allocation size using 32-bit multiplication in avc444_ensure_buffer. A malicious RDP server can supply surface dimensions causing integer wrap, leading to an undersized buffer allocation. This can cause a client crash and may permit code execution.

FreeRDP, a free implementation of the Remote Desktop Protocol, prior to version 3.30.0 has a vulnerability in the kerberos_DecryptMessage function that fails to bound the peer-controlled GSS Wrap-token EC field before using it with RRC in IOV pointer offsets. A malicious RDP peer can trigger out-of-bounds reads and in-place writes during CredSSP/NLA Kerberos decryption.

FreeRDP before 3.30.0 has a heap-based buffer overflow in CliprdrStream_Read. A malicious RDP server can send an oversized response, causing an out-of-bounds write.

FreeRDP before version 3.29.0 (affected versions <= 3.28.0) does not validate CRLF and control characters in the server-controlled RDP redirection TargetNetAddress field. This value is copied into the client's ServerHostname and, when the client connects through an HTTP proxy, is written directly into the proxy CONNECT request line and Host header by http_proxy_connect() without filtering. A malicious or compromised RDP server can send a crafted redirection PDU containing embedded control characters to inject arbitrary headers/requests into the HTTP proxy CONNECT request.

FreeRDP before version 3.29.0 (affected versions <= 3.28.0) contains multiple TLS certificate identity validation weaknesses in tls_verify_certificate(), tls_match_hostname(), and x509_utils_get_dns_names(). Because FreeRDP performs custom Common Name and DNS SAN string matching instead of using OpenSSL's length-aware identity validation APIs, it (1) truncates DNS SAN values at embedded NUL bytes (accepting e.g. 'victim.example\0.attacker.example' as 'victim.example'), (2) accepts a matching Common Name even when non-matching DNS SAN entries are present, and (3) accepts IP-literal targets via DNS/CN matching without comparing iPAddress SANs. Under a trusted or misissued certificate chain, an attacker positioned to present such a certificate can bypass server identity verification, weakening TLS server authentication.

A heap-based buffer overflow in FreeRDP before version 3.28.0 (affected <=3.27.1) in the crypto_rsa_common() function. The function writes the modular exponentiation result to the output buffer before checking bounds, allowing out-of-bounds writes.

FreeRDP versions 3.21.0 through 3.27.9 contain an incomplete fix for CVE-2026-23530 in the planar_decompress_plane_rle_only function in planar.c. A malicious RDP server can send a crafted planar RDPGFX_CMDID_WIRETOSURFACE_1 payload that reads one byte past the input buffer.

In 32-bit builds of FreeRDP prior to 3.28.0, an integer overflow occurs in the update_read_delta_points function in libfreerdp/core/orders.c. A malicious RDP server can exploit this to allocate an undersized heap buffer and write beyond its bounds.

FreeRDP prior to 3.26.0 has an out-of-bounds heap write in the planar bitmap decoder. An attacker can bypass coordinate validation and write past the internal buffer.

In FreeRDP prior to 3.24.0, the gdi_surface_bits() function fails to properly validate bmp.width and bmp.height values from the RDP server. A malicious server can supply crafted values exceeding the surface size, causing a heap buffer overflow during bitmap decoding. The attacker controls pixel data, potentially allowing overwriting adjacent heap memory.

In FreeRDP prior to version 3.21.0, offscreen bitmap deletion leaves `gdi->drawing` pointing to freed memory, causing a use-after-free (UAF) when related update packets arrive. A malicious server can trigger this client-side vulnerability, causing a crash (DoS) and potential heap corruption with code-execution risk.

In FreeRDP prior to version 3.21.0, the `xf_Pointer_New` function frees `cursorPixels` on failure, then `pointer_free` calls `xf_Pointer_Free` which frees it again, triggering an ASan-detected use-after-free. A malicious server can trigger a client-side use after free, causing a crash (DoS) and potential heap corruption with code-execution risk depending on allocator behavior and surrounding heap layout.

In FreeRDP prior to version 3.21.0, a client-side heap buffer overflow occurs in the ClearCodec bands decode path when crafted band coordinates allow writes past the end of the destination surface buffer.