CVE-2026-64527

In the Linux kernel, the following vulnerability has been resolved:

drm/hyperv: validate VMBus packet size in receive callback

hypervreceivesub() reads msg->vidhdr.type and dispatches into one of four message-type branches without knowing how many bytes the host wrote into hv->recvbuf. The completion path then runs memcpy(hv->initbuf, msg, VMBUSMAXPACKETSIZE), so the consumer that wakes on waitforcompletion_timeout() can read up to 16 KiB of residue from a prior message as if it were the response payload.

Pass bytesrecvd into hypervreceivesub() and reject any packet that does not cover the pipe + synthvid header. A single switch on msg->vidhdr.type then computes the type-specific payload size: the three completion-driving types (SYNTHVIDVERSIONRESPONSE, SYNTHVIDRESOLUTIONRESPONSE, SYNTHVIDVRAMLOCATIONACK) fall through to a shared exit that requires that size before memcpy/complete, while SYNTHVIDFEATURECHANGE validates its own payload and returns before reading isdirt_needed. Unknown types are dropped.

SYNTHVIDRESOLUTIONRESPONSE is variable length: the host fills resolutioncount entries, not the full SYNTHVIDMAXRESOLUTIONCOUNT array. Validate the fixed prefix first so resolution_count can be read, bound it against the array, then require only the count-sized array, so the shorter responses the host actually sends are accepted.

Only run the sub-handler when vmbusrecvpacket() returned success. The memcpy length is bytesrecvd, which is bounded by VMBUSMAXPACKETSIZE only on a successful receive; on -ENOBUFS vmbusrecvpacket() instead reports the required length, which can exceed hv->recvbuf, so copying bytesrecvd would read and write past the 16 KiB buffers. Gating on the success return keeps the copy bounded. The nonzero-return path is itself a malformed-message case and is now logged rather than silently skipped; channel recovery is not attempted.

Rejected packets are reported via drmerrratelimited() rather than silently dropped, matching the CoCo-hardened pattern in hvkvponchannelcallback().