CVE-2026-98621
NoneCVSS
—
EPSS
0.2%
Published
Oct 18, 2026
CISA due date
—
Cloudanix analysis
A recently fixed flaw in the Linux kernel could let an attacker corrupt memory on systems that handle a certain kind of network packet, potentially crashing machines or allowing code execution.
- Exploitation
- Remote Unauthenticated
- Attacker needs
- None
- Complexity
- Moderate
- Impact
- Full Compromise
- Component
- OS kernel
- Class
- Buffer Overflow
- Remediation
- Upgrade
- Effort
- Medium
- Patch risk
- Low
What it actually does
Invalid REMCSUM offset values in GUE packets cause underflow when computing skb->csum_offset, allowing a later skb_copy_and_csum_dev call on a CHECKSUM_PARTIAL packet to perform a two-byte write far beyond the destination buffer when a NETIF_F_HW_CSUM driver is in use.
Why it matters to the business
Linux systems that process or forward GUE-encapsulated traffic face risk of memory corruption, crashes, or potential remote code execution; the flaw affects kernel networking stacks in routers, hosts, and cloud instances handling this encapsulation.
How an attack unfolds
An attacker crafts a GUE packet containing a REMCSUM option where the checksum-field offset is smaller than the start offset; after length validation the packet is accepted, the underflow populates csum_offset with a huge value, and when the packet reaches a hardware-checksum driver the kernel writes two bytes approximately 64 KiB past the intended buffer.
How to fix it
Apply the upstream kernel commit that adds reversed-tuple rejection inside validate_gue_flags() after length validation; rebuild and deploy kernels containing this check on all affected systems.
How to confirm the fix worked
Inspect the GUE module source for the added ordering test in validate_gue_flags(); confirm that packets with offset < start are now rejected before any checksum metadata is stored.
Typical attack path
- Send a forwarded GUE packet with REMCSUM option carrying offset < start
- Bypass existing length checks in GUE parsing
- Trigger underflow into skb->csum_offset during partial checksum processing
- Force the packet to a NETIF_F_HW_CSUM-capable driver
- Trigger skb_copy_and_csum_dev to perform out-of-bounds write
How to detect it
Search network-capture or kernel logs for GUE packets carrying the REMCSUM option flag; correlate with subsequent checksum-offload errors or skb corruption warnings in dmesg or netdev logs.
What your security team should do
- Prioritize kernel updates for any system exposing GUE termination or forwarding
- Monitor EPSS and exploit-maturity signals for this CVE over the next 30 days
- Temporarily disable GUE encapsulation where operationally feasible until patches are deployed
The record
In the Linux kernel, the following vulnerability has been resolved: net: gue: reject invalid REMCSUM offsets The REMCSUM option carries an absolute checksum start and checksum field offset. gue_remcsum() passes them to skb_remcsum_process(), whose partial path stores offset - start in the u16 skb->csum_offset variable. If offset is less than start, this underflows. A forwarded packet can retain CHECKSUM_PARTIAL and reach a NETIF_F_HW_CSUM driver which trusts the metadata, leading skb_copy_and_csum_dev() to write two bytes about 64 KiB beyond the destination buffer. Reject reversed tuples in validate_gue_flags(), after the existing length validation, so all GUE parsers enforce the ordering in one place.
Affected packages
| Ecosystem | Package | Introduced | Fixed |
|---|