GPU VulnDB

Database/Firmware, BMC & network fabric

Arista EOS: crafted BGP message on a DPS-enabled session crashes the BGP process

CVSS 6.9CVE-2026-121293Firmware, BMC & network fabriccurated

Impact

A BGP peer that sends a crafted message over an established session can terminate the BGP process on the switch. Every BGP session on the device resets and routing is disrupted until the process restarts. On a GPU fabric with a BGP underlay this can cut connectivity for every node behind the affected switch for the duration, which drops running collective jobs. Only sessions where the Dynamic Path Selection (DPS) address family is negotiated are affected, so switches that do not use DPS are not exposed according to the record.

Who can reach it

An attacker who controls a BGP peer with an established session that has the DPS address family negotiated. This needs a configured peering relationship (high privileges in the CVSS vector); it cannot be reached by an arbitrary host.

What to do

Check whether the DPS address family is configured on any BGP neighbor. If it is, upgrade EOS to a fixed release listed in Arista Security Advisory 0210 and follow that advisory for any workaround. An EOS image upgrade usually means a switch reload unless your platform supports a hitless upgrade, so schedule it with traffic moved off the switch. Switches without DPS configured do not need an urgent change.

References

Related entries

All Firmware, BMC & network fabric entries

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.