KVM: x86/hyper-v: Skip non-canonical addresses during PV TLB flush

JIRA: https://issues.redhat.com/browse/RHEL-104731
CVE: CVE-2025-38351
Upstream: Merged

commit fa787ac
Author: Manuel Andreas <[email protected]>
Date:   Wed Jun 25 15:53:19 2025 +0200

    KVM: x86/hyper-v: Skip non-canonical addresses during PV TLB flush

    In KVM guests with Hyper-V hypercalls enabled, the hypercalls
    HVCALL_FLUSH_VIRTUAL_ADDRESS_LIST and HVCALL_FLUSH_VIRTUAL_ADDRESS_LIST_EX
    allow a guest to request invalidation of portions of a virtual TLB.
    For this, the hypercall parameter includes a list of GVAs that are supposed
    to be invalidated.

    However, when non-canonical GVAs are passed, there is currently no
    filtering in place and they are eventually passed to checked invocations of
    INVVPID on Intel / INVLPGA on AMD.  While AMD's INVLPGA silently ignores
    non-canonical addresses (effectively a no-op), Intel's INVVPID explicitly
    signals VM-Fail and ultimately triggers the WARN_ONCE in invvpid_error():

      invvpid failed: ext=0x0 vpid=1 gva=0xaaaaaaaaaaaaa000
      WARNING: CPU: 6 PID: 326 at arch/x86/kvm/vmx/vmx.c:482
      invvpid_error+0x91/0xa0 [kvm_intel]
      Modules linked in: kvm_intel kvm 9pnet_virtio irqbypass fuse
      CPU: 6 UID: 0 PID: 326 Comm: kvm-vm Not tainted 6.15.0 #14 PREEMPT(voluntary)
      RIP: 0010:invvpid_error+0x91/0xa0 [kvm_intel]
      Call Trace:
        vmx_flush_tlb_gva+0x320/0x490 [kvm_intel]
        kvm_hv_vcpu_flush_tlb+0x24f/0x4f0 [kvm]
        kvm_arch_vcpu_ioctl_run+0x3013/0x5810 [kvm]

    Hyper-V documents that invalid GVAs (those that are beyond a partition's
    GVA space) are to be ignored.  While not completely clear whether this
    ruling also applies to non-canonical GVAs, it is likely fine to make that
    assumption, and manual testing on Azure confirms "real" Hyper-V interprets
    the specification in the same way.

    Skip non-canonical GVAs when processing the list of address to avoid
    tripping the INVVPID failure.  Alternatively, KVM could filter out "bad"
    GVAs before inserting into the FIFO, but practically speaking the only
    downside of pushing validation to the final processing is that doing so
    is suboptimal for the guest, and no well-behaved guest will request TLB
    flushes for non-canonical addresses.

    Fixes: 2609708 ("KVM: x86: hyper-v: Handle HVCALL_FLUSH_VIRTUAL_ADDRESS_LIST{,EX} calls gently")
    Cc: [email protected]
    Signed-off-by: Manuel Andreas <[email protected]>
    Suggested-by: Vitaly Kuznetsov <[email protected]>
    Link: https://lore.kernel.org/r/[email protected]
    Signed-off-by: Sean Christopherson <[email protected]>

Signed-off-by: Jon Maloy <[email protected]>

Author: lmcjoma

arch/x86/kvm/hyperv.c

@@ -1980,6 +1980,9 @@ int kvm_hv_vcpu_flush_tlb(struct kvm_vcpu *vcpu)
 		if (entries[i] == KVM_HV_TLB_FLUSHALL_ENTRY)
 			goto out_flush_all;
 
+		if (is_noncanonical_invlpg_address(entries[i], vcpu))
+			continue;
+
 		/*
 		 * Lower 12 bits of 'address' encode the number of additional
 		 * pages to flush.