Database/Firmware, BMC & network fabric
Linux kernel ACPI pfr_update: runtime version checked instead of security version on firmware update
Impact
The Platform Firmware Runtime Update driver gated capsule updates on the runtime version number rather than the security version number. The practical effect reported in the record is that the check is the wrong one: an update binary is accepted or rejected on the basis of a field that does not express firmware security level, so the driver's own anti-downgrade reasoning does not hold. On a GPU node, runtime firmware update is the mechanism you rely on to apply microcode and platform fixes without a maintenance window, and a version gate that consults the wrong field weakens the assurance that what got applied is not older than what was there. The record does not describe a working downgrade attack or name an attacker; it describes the check being wrong, and the upstream commit is a correctness fix. Exposure requires the ability to feed the driver a capsule in the first place.
Who can reach it
Local, privileged: a caller able to open the PFRU device node and submit a firmware capsule - in practice root or a holder of the equivalent capability on the host. Not reachable from a tenant container or over the network. Hosts without CONFIG_ACPI_PFRUT or without Intel platform firmware runtime update support are unaffected.
What to do
Pick up the stable kernel commits listed in the record, or the distribution kernel that carries them; the fix is confined to the pfr_update driver's version comparison. Applying a kernel update means draining and rebooting the node. The record names no fixed release version beyond the commits, and no vendor advisory accompanies it.
References
Related entries
- Linux bnxt_en driver (ring defaults vs traffic classes on ifdown): Memory corruption when firmware resources changeCVE-2025-39810 · Linux bnxt_en driver (ring defaults vs traffic classes on ifdown)High
- Linux kernel (drivers/infiniband/sw/rxe): Use-after-free from a race between a busy soft-RoCE task and its ownCVE-2025-40061 · Linux kernel (drivers/infiniband/sw/rxe)High
- The Linux kernel's IPMI driver message-handling layer: A use-after-free in a kernel driver reachable from the host'sCVE-2025-40202 · The Linux kernel's IPMI driver message-handling layerHigh
- Arista DANZ Monitoring Fabric: restricted CLI user can break out to the system shell and escalate privilegesCVE-2025-54545 · Arista DANZ Monitoring Fabric (restricted-user CLI sandbox)High
- Eaton UPS Companion (EUC) executable - library loading: Insecure library loading in the shipped executable givesCVE-2025-67450 · Eaton UPS Companion (EUC) executable - library loadingHigh
- Linux kernel (drivers/infiniband/sw/rxe): Two failed shared-receive-queue resizes in a row panic the node. The firstCVE-2025-68379 · Linux kernel (drivers/infiniband/sw/rxe)High
This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.