Linux NFS client: missing minimum-length check when decoding a filehandle lets a crafted handle read out of bounds

Impact

nfs_fh_to_dentry() accessed the embedded NFS filehandle before checking that the supplied handle was long enough to contain one, so a short or crafted handle causes an out-of-bounds read of kernel memory and can corrupt state during dentry reconstruction. Upstream rates it local with high confidentiality, integrity and availability impact. On GPU fleets this matters because NFS is the ordinary home-directory and dataset mount on training and inference nodes, and the file_handle decode path is reachable from open_by_handle_at() and from a host that re-exports an NFS mount - so the trigger can come from a tenant process or from filehandles a re-exporting server accepts, not only from the storage admin. A fault here lands in the kernel on a node whose running jobs are expensive to interrupt. The record gives no evidence of exploitation in the wild and no public exploit.

Who can reach it

Local: a process able to call open_by_handle_at() with an attacker-chosen handle (needs CAP_DAC_READ_SEARCH in its namespace), or a node re-exporting an NFS filesystem that decodes filehandles supplied by its own clients. No remote unauthenticated path is described in the record.

What to do

Pick up the fix from the stable trees listed in the record (the change adds the minimum filehandle length check before the embedded handle is used); there is no runtime toggle and no module-only fix, so each node needs the patched kernel and a reboot. Drain GPU nodes and roll them through a maintenance window. Where a reboot cannot be scheduled quickly, the practical interim step is to deny the trigger: avoid re-exporting NFS mounts from those nodes and withhold CAP_DAC_READ_SEARCH from tenant workloads. The record names no vendor-specific fixed release beyond the stable commits.

References

Related entries

• Linux kernel (drivers/gpu/drm/xe): On the migration error path the previous fence is released before the code waits onCVE-2025-39740 · Linux kernel (drivers/gpu/drm/xe)High
• Linux kernel perf: throttling a group with disabled members calls PMU start/stop with hw.idx == -1CVE-2025-39821 · Linux kernel perf core (group throttling of inactive events)High
• Linux kernel lpfc: use-after-free in the deferred receive path on Emulex Fibre Channel HBAsCVE-2025-39841 · Linux kernel lpfc driver (Emulex/Broadcom Fibre Channel HBA, deferred receive path)High
• Linux kernel ice: NULL dereference in the E810 low-latency PTP timestamp interrupt during resetCVE-2025-39854 · Linux kernel ice driver (Intel E810 PTP low-latency Tx timestamp interrupt)High
• Linux kernel ice: NULL dereference in the E810 PTP Tx timestamp IRQ racing driver resetCVE-2025-39855 · Linux kernel ice driver (Intel E810 PTP Tx timestamp IRQ, ice_ptp_ts_irq)High
• Linux kernel idxd: double free in idxd_setup_wqs() error path during DSA device probeCVE-2025-39870 · Linux kernel idxd driver (Intel Data Streaming Accelerator / DSA work-queue setup)High