Intel ice NIC driver: external timestamp interrupt after PTP release panics the host

CVSS 4.7CVE-2024-42107Kernel, userspace & hypervisorcurated

Impact

ice_ptp_extts_event() can run after ice_ptp_release() has torn down the PTP clock, so it calls ptp_clock_event() with a NULL pointer and panics the node. The ice driver covers Intel E810-class datacenter NICs, which are common as the front-end and storage NICs on GPU servers, and PTP hardware timestamping is exactly what gets enabled on fleets that need synchronised clocks for distributed training telemetry and packet capture. The crash window is driver teardown - module unload, device reset or a failing port - while an external timestamp pin is still delivering events, so it shows up as an unexplained host panic during NIC maintenance rather than as something a tenant does.

Who can reach it

Local and essentially administrative: it requires a PTP-enabled ice interface with an external timestamp source and a concurrent PTP release (driver unbind, reset or remove). Not reachable from the network and not reachable by an unprivileged tenant.

What to do

Update to a kernel carrying the linked fix, which checks PTP readiness before processing the event, and reboot each node; if the fleet does not use PTP external timestamping on ice interfaces there is nothing to schedule. The record lists stable commits, not a named fixed release.

References

https://nvd.nist.gov/vuln/detail/CVE-2024-42107 https://git.kernel.org/stable/c/1c4e524811918600683b1ea87a5e0fc2db64fa9b https://git.kernel.org/stable/c/9aa83a5eb013f8aff20e701aa25ba6b0eaad0b65

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.