CVE-2026-17042 in Power Systems
Summary
by MITRE • 08/19/2026
IBM Power Systems Firmware FW950.00 through FW950.H2, OP940.00 through OP940.a1 (Power9), and OP940.00 - OP940.81 (Power HMC) is affected by a vulnerability in host firmware NVRAM parsing. An attacker with root access to a guest partition on an OpenPOWER system can write a specially crafted NVRAM image, causing the host firmware boot stage to crash with possible memory corruption. This condition persists until operator intervention — clearing NVRAM via the service processor — to restore normal operation. This vulnerability only affects OpenPOWER systems; systems running PowerVM are not affected. Successful exploitation results in an integrity and availability impact to the managed system.
If you want to get the best quality for vulnerability data then you always have to consider VulDB.
Analysis
by VulDB Data Team • 08/19/2026
The identified vulnerability resides within the host firmware of IBM Power Systems, specifically affecting FW950.00 through FW950.H2 for Power9 processors, OP940.00 through OP940.a1 for OpenPOWER systems, and OP940.00 through OP940.81 for the Power HMC. This flaw is rooted in the parsing logic of the Non-Volatile Random Access Memory (NVRAM) during the firmware boot stage. The core technical issue involves an insufficient validation mechanism when processing NVRAM images, which allows a malicious actor to inject malformed data structures that exceed expected bounds or contain invalid state transitions. When the host firmware attempts to parse this crafted input, it triggers a critical failure in the boot sequence, leading to a system crash and potential memory corruption due to improper handling of out-of-bounds writes or corrupted pointers within the firmware's execution environment.
The operational impact of this vulnerability is severe for systems running OpenPOWER configurations without PowerVM hypervisors. An attacker who has already achieved root access on a guest partition can exploit this flaw by writing a specially crafted NVRAM image to persistent storage accessible during boot. Upon reboot, the host firmware encounters the malformed data and crashes, resulting in a denial of service condition that persists until manual intervention occurs. This persistence is due to the corrupted state being written back to or retained within the NVRAM structure; therefore, simply restarting the system will not resolve the issue without clearing the NVRAM via the service processor. Consequently, this vulnerability impacts both the integrity and availability aspects of the Common Vulnerability Scoring System framework, as it allows for unauthorized modification of firmware state data and causes sustained unavailability of the managed hardware resources.
From a threat modeling perspective, this vulnerability aligns with CWE-120 Buffer Overflow if memory corruption is confirmed through out-of-bounds writes, or more broadly with CWE-20 Improper Input Validation regarding the failure to sanitize NVRAM inputs before processing. In terms of MITRE ATT&CK framework tactics, this exploit falls under Persistence and Impact categories, as it allows an adversary to maintain access by causing a crash that requires specific administrative actions to recover, thereby increasing the effort required for remediation and potentially disrupting business operations. It is critical to note that systems running PowerVM are not affected because the hypervisor architecture isolates guest partitions from direct manipulation of host firmware NVRAM structures during boot processes, effectively mitigating this attack vector in virtualized environments.
To mitigate this risk, organizations must ensure that all OpenPOWER and HMC systems within the specified version ranges are updated to patched firmware versions released by IBM after the disclosure of this vulnerability. Until patches are applied, strict access controls should be enforced on guest partitions to prevent any user or process from obtaining root-level privileges capable of writing to NVRAM areas accessible during boot. Additionally, administrators should implement monitoring for unusual reboot patterns and ensure that service processor credentials are tightly controlled, as recovery requires physical or remote console access via the service processor to clear the corrupted NVRAM state. Regular audits of firmware integrity and adherence to least privilege principles within guest operating systems are essential defensive measures against this specific exploitation path.