In the Linux kernel, the following vulnerability has been resolved:
KVM: x86: Acquire SRCU in KVMGETMP_STATE to protect guest memory accesses
Acquire a lock on kvm->srcu when userspace is getting MP state to handle a rather extreme edge case where "accepting" APIC events, i.e. processing pending INIT or SIPI, can trigger accesses to guest memory. If the vCPU is in L2 with INIT and a TRIPLEFAULT request pending, then getting MP state will trigger a nested VM-Exit by way of ->checknested_events(), and emuating the nested VM-Exit can access guest memory.
The splat was originally hit by syzkaller on a Google-internal kernel, and reproduced on an upstream kernel by hacking the triplefaulteventtest selftest to stuff a pending INIT, store an MSR on VM-Exit (to generate a memory access on VMX), and do vcpumpstateget() to trigger the scenario.
============================= WARNING: suspicious RCU usage 6.14.0-rc3-b112d356288b-vmx/pilockdepfalse_pos-lock #3 Not tainted
include/linux/kvmhost.h:1058 suspicious rcudereference_check() usage!
other info that might help us debug this:
rcuscheduleractive = 2, debuglocks = 1 1 lock held by triplefaultev/1256: #0: ffff88810df5a330 (&vcpu->mutex){+.+.}-{4:4}, at: kvmvcpu_ioctl+0x8b/0x9a0 [kvm]
stack backtrace: CPU: 11 UID: 1000 PID: 1256 Comm: triplefaultev Not tainted 6.14.0-rc3-b112d356288b-vmx #3 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 0.0.0 02/06/2015 Call Trace: <TASK> dumpstacklvl+0x7f/0x90 lockdeprcususpicious+0x144/0x190 kvmvcpugfntomemslot+0x156/0x180 [kvm] kvmvcpureadguest+0x3e/0x90 [kvm] readandcheckmsrentry+0x2e/0x180 [kvmintel] _nestedvmxvmexit+0x550/0xde0 [kvmintel] kvmchecknestedevents+0x1b/0x30 [kvm] kvmapicacceptevents+0x33/0x100 [kvm] kvmarchvcpuioctlgetmpstate+0x30/0x1d0 [kvm] kvmvcpuioctl+0x33e/0x9a0 [kvm] _x64sysioctl+0x8b/0xb0 dosyscall64+0x6c/0x170 entrySYSCALL64afterhwframe+0x4b/0x53 </TASK>