Re: [PATCH v7] evtchn: convert domain event lock to an r/w one

["Daniel P. Smith" <dpsmith@xxxxxxxxxxxxxxxxxxxx>]

On 7/27/22 11:56, Jan Beulich wrote:
> Especially for the use in evtchn_move_pirqs() (called when moving a vCPU
> across pCPU-s) and the ones in EOI handling in PCI pass-through code,
> serializing perhaps an entire domain isn't helpful when no state (which
> isn't e.g. further protected by the per-channel lock) changes.
>
> Unfortunately this implies dropping of lock profiling for this lock,
> until r/w locks may get enabled for such functionality.
>
> While ->notify_vcpu_id is now meant to be consistently updated with the
> per-channel lock held, an extension applies to ECS_PIRQ: The field is
> also guaranteed to not change with the per-domain event lock held for
> writing. Therefore the link_pirq_port() call from evtchn_bind_pirq()
> could in principle be moved out of the per-channel locked regions, but
> this further code churn didn't seem worth it.
>
> Signed-off-by: Jan Beulich <jbeulich@xxxxxxxx>
> ---
> v6: Re-vive and re-base.
> v5: Re-base, also over dropped earlier patch.
> v4: Re-base, in particular over new earlier patches. Acquire both
>      per-domain locks for writing in evtchn_close(). Adjust
>      spin_barrier() related comments.
> v3: Re-base.
> v2: Consistently lock for writing in evtchn_reset(). Fix error path in
>      pci_clean_dpci_irqs(). Lock for writing in pt_irq_time_out(),
>      hvm_dirq_assist(), hvm_dpci_eoi(), and hvm_dpci_isairq_eoi(). Move
>      rw_barrier() introduction here. Re-base over changes earlier in the
>      series.
<snip/>

> --- a/xen/xsm/flask/flask_op.c
> +++ b/xen/xsm/flask/flask_op.c
> @@ -530,7 +530,7 @@ static int flask_get_peer_sid(struct xen
>       struct evtchn *chn;
>       struct domain_security_struct *dsec;
>   
> -    spin_lock(&d->event_lock);
> +    read_lock(&d->event_lock);
>   
>       if ( !port_is_valid(d, arg->evtchn) )
>           goto out;
> @@ -548,7 +548,7 @@ static int flask_get_peer_sid(struct xen
>       rv = 0;
>   
>    out:
> -    spin_unlock(&d->event_lock);
> +    read_unlock(&d->event_lock);
>       return rv;
>   }
Acked-by: Daniel P. Smith <dpsmith@xxxxxxxxxxxxxxxxxxxx>