CVE-2026-33145
MediumCVSS 6.3Exploitation Probability (EPSS)
Low risk29th percentile - higher than 29% of all known CVEs
Summary
xrdp through 0.10.5 allows an authenticated remote user to execute arbitrary commands due to unsafe handling of the AlternateShell parameter in xrdp-sesman. When AllowAlternateShell is enabled (default), the client-supplied value is executed via /bin/sh -c during session initialization. This is fixed in version 0.10.6.
Risk Assessment
An authenticated user can execute arbitrary commands within their session context, bypassing expected initialization flows. This may lead to session compromise or escalation within the user's privileges.
Recommendation
Update xrdp to version 0.10.6 or later. If updating is not possible, disable the AllowAlternateShell setting.
Other vulnerabilities in xrdp
See all- CVE-2017-16927High
The scp_v0s_accept function in the session manager of xrdp up to version 0.9.4 uses an untrusted integer as a write length, which can lead to buffer overflow and application crash.
- CVE-2017-6967High
xrdp version 0.9.1 calls the PAM function auth_start_session() in an incorrect location, leading to improper initialization of PAM session modules. This may result in incorrect configurations or elevation of privileges.
- CVE-2026-41252Critical
xrdp versions up to 0.10.6 have a missing bounds check in vnc-any mode, leading to a heap-based buffer overflow when handling RFB color map messages from a VNC server. A malicious VNC server can send crafted messages with out-of-range values, causing an out-of-bounds write on the heap.
- CVE-2013-1430Critical
An issue was discovered in xrdp before 0.9.1, where a file ~/.vnc/sesman_${username}_passwd is created upon successful RDP login. Its content is the user's cleartext password, DES encrypted with a known key.
- CVE-2026-55639Medium
xrdp versions 0.10.6 and earlier contain a vulnerability in parsing client security data. An unauthenticated attacker can send a crafted RDP packet, causing out-of-bounds memory read.
- CVE-2026-55626High
xrdp versions up to 0.10.6 launch Xvnc over UNIX sockets with insufficient authentication. A local authenticated attacker can bypass session isolation and unauthorizedly view or control other users' sessions. Only the Xvnc backend over UNIX sockets is affected.
- CVE-2026-55645Medium
xrdp versions up to 0.10.6 have a vulnerability in processing Client Control PDUs. Insufficient length validation before reading data can lead to out-of-bounds memory reads by a remote, unauthenticated attacker. This results in a crash of the xrdp process (Denial of Service), but not the entire service, as xrdp forks a new process per connection by default.
- CVE-2026-55238Medium
xrdp versions up to 0.10.6 contain an out-of-bounds read vulnerability during processing of RDP Confirm Active PDU. Insufficient length validation of capability sets allows an unauthenticated remote attacker to send a crafted RDP packet, potentially causing a process crash and denial of service (DoS).
- CVE-2026-54538High
Vulnerability in xrdp due to improper validation of the totalLength field in the RDP protocol header. An unauthenticated attacker can send a crafted packet, causing an infinite CPU loop and resource exhaustion.
- CVE-2026-44978Medium
xrdp versions up to 0.10.6 contain a heap out-of-bounds read vulnerability in FIPS-specific receive paths. The vulnerability does not affect the default configuration and requires security_layer=negotiate or security_layer=rdp and crypt_level=fips in xrdp.ini. An unauthenticated remote attacker can send a crafted FIPS-protected PDU, causing a process crash and denial of service (DoS).
Original NVD description (English source)
xrdp is an open source RDP server. Versions through 0.10.5 allow an authenticated remote user to execute arbitrary commands on the server due to unsafe handling of the AlternateShell parameter in xrdp-sesman. When the AllowAlternateShell setting is enabled (which is the default when not explicitly configured), xrdp accepts a client-supplied AlternateShell value and executes it via /bin/sh -c during session initialization. This results in shell-interpreted execution of unsanitized, user-controlled input. This behavior effectively provides a scriptable remote command execution primitive over RDP within the security context of the authenticated user, occurring prior to normal window manager startup. This can bypass expected session initialization flows and operational assumptions that restrict execution to interactive desktop environments. This issue has been fixed in version 0.10.6.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS

