CVE-2026-99635: vhost/vsock: size receive SKBs from declared payload
In the Linux kernel, the following vulnerability has been resolved:
vhost/vsock: size receive SKBs from declared payload
vhostvsockallocskb() sizes the skb from iovlength(), the total guest descriptor length. virtiotransportincrxpkt() instead adds hdr->len to rxbytes and bufused, and budgets each queued skb as SKBTRUESIZE(0). The guest controls the descriptor length and hdr->len independently.
A guest can therefore provide a 64 KiB descriptor while declaring a zero payload. The descriptor-sized skb remains queued, rxbytes and bufused remain unchanged, and only the fixed queue-entry budget limits retention. This is not a lifetime leak: the skb is freed when dequeued or at socket teardown. Repeating this across connections can nevertheless exhaust host memory. On Linux v7.2, 455 such descriptors retained 30,255,680 bytes for a 262,144-byte receive buffer; a full-payload control retained 265,984 bytes.
Vhost previously read hdr->len and allocated pkt->buf from that value. The conversion from virtiovsockpkt to skbuff instead allocated receive skbs from the descriptor length, introducing the allocation-accounting mismatch.
Copy the header into a stack object and validate hdr->len before allocating the skb. Size the allocation as sizeof(hdr) + payloadlen so descriptor capacity no longer determines queued memory. Check payloadlen against len - sizeof(hdr) to avoid addition overflow on 32-bit hosts.
Affected Software
Remediation
Recommended actions to resolve this vulnerability, in priority order.
- Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 6.2 - Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 6.12.118 - Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 6.18.64 - Upgrade
Upgrade
linux/Kernelto a version that resolves this vulnerability.Fixed in 7.3.5 - Compensating control
In vhost/vsock, copy the header into a stack object and validate hdr->len before allocating. Check payload_len against hdr->len - sizeof(hdr) to avoid addition overflow on 32-bit hosts, then size the receive skb allocation as sizeof(hdr) + payload_len rather than from the descriptor length.
Frequently Asked Questions
Who is exposed to this issue?
Hosts using the Linux kernel vhost/vsock implementation are exposed to a malicious or compromised guest that can supply virtio-vsock receive descriptors. The guest controls both the descriptor length and the declared payload length.
What does an attacker need to do to trigger the memory exhaustion?
The guest can provide a large descriptor, such as 64 KiB, while declaring a zero-length payload. The resulting descriptor-sized skb can remain queued without corresponding rx_bytes or buf_used accounting, and repetition across connections can exhaust host memory.
Does this create a permanent memory leak?
No. Queued skbs are freed when they are dequeued or when the socket is torn down. However, an attacker can retain substantial host memory while the skbs remain queued and repeat the condition across connections.
How can affected behavior be recognized?
A vulnerable host can show disproportionately high memory retained by queued vhost/vsock receive buffers despite little or no declared payload and unchanged receive-buffer accounting. The described test retained 30,255,680 bytes with 455 zero-payload 64 KiB descriptors for a 262,144-byte receive buffer.