GPU VulnDB

Database/Control plane, storage & DevOps

Linux nfsd: fh_verify tracepoints size the server address slot wrongly, writing past the trace buffer

CVSS 9.8CVE-2026-89702Control plane, storage & DevOpscurated

Impact

The nfsd_fh_verify and nfsd_fh_verify_err tracepoints reserve the server sockaddr slot using xpt_remotelen but fill it from xpt_local using xpt_locallen. For NFS over UDP xpt_remotelen stays zero for the transport's lifetime, so every fh_verify trace copies 16 or 28 bytes into a zero-length ring-buffer slot; in the reverse case the untouched tail of an oversized slot leaks previous ring-buffer contents to whoever reads the trace. The damage is confined to hosts that have these nfsd tracepoints enabled - an idle tracing configuration is not exposed - so on most production NFS servers this is inert until someone turns on nfsd tracing to debug something.

Who can reach it

Requires the nfsd_fh_verify tracepoints to be enabled on the server. With tracing on, any NFSv2/v3-over-UDP client request reaches the overflow; reading the leaked ring-buffer bytes needs local access to tracefs.

What to do

Take a kernel with one of the linked stable commits and reboot the NFS server. Until then, leave the nfsd_fh_verify and nfsd_fh_verify_err tracepoints disabled, which is the default, and avoid enabling them on servers still exporting NFS over UDP.

References

Related entries

All Control plane, storage & DevOps 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.