kvm: dirty-ring: Fix race with vcpu creation
commit56adee407fc564da19e49cfe18e20e3da92320be
authorPeter Xu <peterx@redhat.com>
Thu, 16 Feb 2023 16:18:32 +0000 (17 00:18 +0800)
committerPaolo Bonzini <pbonzini@redhat.com>
Tue, 4 Apr 2023 16:46:46 +0000 (4 18:46 +0200)
treebc9b62476258378ed2f6713315af295b7d42804a
parentf00506aeca2f6d92318967693f8da8c713c163f3
kvm: dirty-ring: Fix race with vcpu creation

It's possible that we want to reap a dirty ring on a vcpu that is during
creation, because the vcpu is put onto list (CPU_FOREACH visible) before
initialization of the structures.  In this case:

qemu_init_vcpu
    x86_cpu_realizefn
        cpu_exec_realizefn
            cpu_list_add      <---- can be probed by CPU_FOREACH
        qemu_init_vcpu
            cpus_accel->create_vcpu_thread(cpu);
                kvm_init_vcpu
                    map kvm_dirty_gfns  <--- kvm_dirty_gfns valid

Don't try to reap dirty ring on vcpus during creation or it'll crash.

Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=2124756
Reported-by: Xiaohui Li <xiaohli@redhat.com>
Signed-off-by: Peter Xu <peterx@redhat.com>
Message-Id: <1d14deb6684bcb7de1c9633c5bd21113988cc698.1676563222.git.huangy81@chinatelecom.cn>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
accel/kvm/kvm-all.c