While looking into JDK-8375362 I realized we should also not check for suspension in MountUnmountDisabler::is_start_transition_disabled() for a thread executing within a disableSuspendAndPreempt section, i.e JavaThread::_is_disable_suspend set. Otherwise we might suspend the thread while holding the interruptLock monitor which is what we wanted to avoid when this mechanism was added in JDK-8311218. Looking at the code, I don't see any obvious unmount point after grabbing the monitor. But with the changes in JDK-8369238, it's hard to tell if we might try to initialize some class triggering an unmount attempt (looking at appendCarrierInfo).