[PATCH 16/18] KVM: Don't take mmu_lock for range invalidation unless necessary

[Sean Christopherson]

On Wed, Mar 31, 2021, Paolo Bonzini wrote:
> On 31/03/21 21:47, Sean Christopherson wrote:
> > Rereading things, a small chunk of the rwsem nastiness can go away.  I don't see
> > any reason to use rw_semaphore instead of rwlock_t.
> 
> Wouldn't it be incorrect to lock a mutex (e.g. inside *another* MMU
> notifier's invalidate callback) while holding an rwlock_t?  That makes sense
> because anybody that's busy waiting in write_lock potentially cannot be
> preempted until the other task gets the mutex.  This is a potential
> deadlock.

Yes?  I don't think I follow your point though.  Nesting a spinlock or rwlock
inside a rwlock is ok, so long as the locks are always taken in the same order,
i.e. it's never mmu_lock -> mmu_notifier_slots_lock.

> I also thought of busy waiting on down_read_trylock if the MMU notifier
> cannot block, but that would also be invalid for the opposite reason (the
> down_write task might be asleep, waiting for other readers to release the
> task, and the down_read_trylock busy loop might not let that task run).
> 
> > And that's _already_ the worst case since notifications are currently
> > serialized by mmu_lock.
> 
> But right now notifications are not a single critical section, they're two,
> aren't they?

Ah, crud, yes.  Holding a spinlock across the entire start() ... end() would be
bad, especially when the notifier can block since that opens up the possibility
of the task sleeping/blocking/yielding while the spinlock is held.  Bummer.