Re: [PATCH 2/5] KVM: Bail from the dirty ring reset flow if a signal is pending

[Sean Christopherson <seanjc@google.com>]

On Mon, Jan 13, 2025, Yan Zhao wrote:
> On Fri, Jan 10, 2025 at 05:04:06PM -0800, Sean Christopherson wrote:
> > Abort a dirty ring reset if the current task has a pending signal, as the
> > hard limit of INT_MAX entries doesn't ensure KVM will respond to a signal
> > in a timely fashion.
> > 
> > Fixes: fb04a1eddb1a ("KVM: X86: Implement ring-based dirty memory tracking")
> > Signed-off-by: Sean Christopherson <seanjc@google.com>
> > ---
> >  virt/kvm/dirty_ring.c | 3 +++
> >  1 file changed, 3 insertions(+)
> > 
> > diff --git a/virt/kvm/dirty_ring.c b/virt/kvm/dirty_ring.c
> > index 2faf894dec5a..a81ad17d5eef 100644
> > --- a/virt/kvm/dirty_ring.c
> > +++ b/virt/kvm/dirty_ring.c
> > @@ -117,6 +117,9 @@ int kvm_dirty_ring_reset(struct kvm *kvm, struct kvm_dirty_ring *ring,
> >  	cur_slot = cur_offset = mask = 0;
> >  
> >  	while (likely((*nr_entries_reset) < INT_MAX)) {
> > +		if (signal_pending(current))
> > +			return -EINTR;
> Will it break the userspace when a signal is pending? e.g. QEMU might report an
> error like
> "kvm_dirty_ring_reap_locked: Assertion `ret == total' failed".

Ugh.  In theory, yes.  In practice, I hope not?  If it's a potential problem for
QEMU, the only idea have is to only react to fatal signals by default, and then
let userspace opt-in to reacting to non-fatal signals.

> 
> >  		entry = &ring->dirty_gfns[ring->reset_index & (ring->size - 1)];
> >  
> >  		if (!kvm_dirty_gfn_harvested(entry))
> > -- 
> > 2.47.1.613.gc27f4b7a9f-goog
> >