RE: [RESEND PATCH 4/6] x86/traps: add external_interrupt() to dispatch external interrupts

[Li, Xin3]

> > +__visible noinstr void external_interrupt(struct pt_regs *regs,
> > +					  unsigned int vector)
> > +{
> > +	unsigned int sysvec = vector - FIRST_SYSTEM_VECTOR;
> > +
> > +	BUG_ON(vector < FIRST_EXTERNAL_VECTOR);
> 
> Why not return an error up the stack?  KVM and/or CPU bugs aren't unheard
> of.
> Dropping an IRQ obviously isn't ideal, but there's a non-zero chance that letting
> KVM WARN and kill the VM will keep the host alive and thus other VMs
> running.  A somewhat sophisticated setup might even react to the VM being
> killed by migrating other VMs off the system and initiating host maintenance.

Make sense.

What about having it return a signed integer?