Why is a memory barrier necessary between MONITOR and MWAIT?

Question

Perusing the Linux x86 idle loop, I noticed a memory barrier in between monitor and mwait, and I can't figure out exactly why it's necessary.

void mwait_idle_with_hints(unsigned long ax, unsigned long cx) {
    if (!need_resched()) {
        if (this_cpu_has(X86_FEATURE_CLFLUSH_MONITOR))   // True for Intel Xeon 7400 series CPUs
            clflush((void *)&current_thread_info()->flags);   // Workaround for erratum AAI65 for the Xeon 7400 series

        __monitor((void *)&current_thread_info()->flags, 0, 0);
        smp_mb();
        if (!need_resched())
            __mwait(ax, cx);
    }
}

smp_mb() is a macro for asm volatile("mfence":::"memory").

Why is it necessary here? I understand why a compiler memory barrier would be necessary, but not a hardware memory barrier.