Message-ID: <20220427022907-mutt-send-email-mst@kernel.org>
Date:   Wed, 27 Apr 2022 02:30:10 -0400
From:   "Michael S. Tsirkin" <mst@...hat.com>
To:     Jason Wang <jasowang@...hat.com>
Cc:     Halil Pasic <pasic@...ux.ibm.com>,
        Cornelia Huck <cohuck@...hat.com>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        virtualization <virtualization@...ts.linux-foundation.org>,
        Stefano Garzarella <sgarzare@...hat.com>,
        eperezma <eperezma@...hat.com>, Cindy Lu <lulu@...hat.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Peter Zijlstra <peterz@...radead.org>,
        "Paul E. McKenney" <paulmck@...nel.org>,
        Marc Zyngier <maz@...nel.org>
Subject: Re: [PATCH V3 6/9] virtio-ccw: implement synchronize_cbs()

On Wed, Apr 27, 2022 at 11:53:25AM +0800, Jason Wang wrote:
> On Tue, Apr 26, 2022 at 2:30 PM Michael S. Tsirkin <mst@...hat.com> wrote:
> >
> > On Tue, Apr 26, 2022 at 12:07:39PM +0800, Jason Wang wrote:
> > > On Tue, Apr 26, 2022 at 11:55 AM Michael S. Tsirkin <mst@...hat.com> wrote:
> > > >
> > > > On Mon, Apr 25, 2022 at 11:53:24PM -0400, Michael S. Tsirkin wrote:
> > > > > On Tue, Apr 26, 2022 at 11:42:45AM +0800, Jason Wang wrote:
> > > > > >
> > > > > > 在 2022/4/26 11:38, Michael S. Tsirkin 写道:
> > > > > > > On Mon, Apr 25, 2022 at 11:35:41PM -0400, Michael S. Tsirkin wrote:
> > > > > > > > On Tue, Apr 26, 2022 at 04:29:11AM +0200, Halil Pasic wrote:
> > > > > > > > > On Mon, 25 Apr 2022 09:59:55 -0400
> > > > > > > > > "Michael S. Tsirkin" <mst@...hat.com> wrote:
> > > > > > > > >
> > > > > > > > > > On Mon, Apr 25, 2022 at 10:54:24AM +0200, Cornelia Huck wrote:
> > > > > > > > > > > On Mon, Apr 25 2022, "Michael S. Tsirkin" <mst@...hat.com> wrote:
> > > > > > > > > > > > On Mon, Apr 25, 2022 at 10:44:15AM +0800, Jason Wang wrote:
> > > > > > > > > > > > > This patch tries to implement the synchronize_cbs() for ccw. For the
> > > > > > > > > > > > > vring_interrupt() that is called via virtio_airq_handler(), the
> > > > > > > > > > > > > synchronization is simply done via the airq_info's lock. For the
> > > > > > > > > > > > > vring_interrupt() that is called via virtio_ccw_int_handler(), a per
> > > > > > > > > > > > > device spinlock for irq is introduced ans used in the synchronization
> > > > > > > > > > > > > method.
> > > > > > > > > > > > >
> > > > > > > > > > > > > Cc: Thomas Gleixner <tglx@...utronix.de>
> > > > > > > > > > > > > Cc: Peter Zijlstra <peterz@...radead.org>
> > > > > > > > > > > > > Cc: "Paul E. McKenney" <paulmck@...nel.org>
> > > > > > > > > > > > > Cc: Marc Zyngier <maz@...nel.org>
> > > > > > > > > > > > > Cc: Halil Pasic <pasic@...ux.ibm.com>
> > > > > > > > > > > > > Cc: Cornelia Huck <cohuck@...hat.com>
> > > > > > > > > > > > > Signed-off-by: Jason Wang <jasowang@...hat.com>
> > > > > > > > > > > >
> > > > > > > > > > > > This is the only one that is giving me pause. Halil, Cornelia,
> > > > > > > > > > > > should we be concerned about the performance impact here?
> > > > > > > > > > > > Any chance it can be tested?
> > > > > > > > > > > We can have a bunch of devices using the same airq structure, and the
> > > > > > > > > > > sync cb creates a choke point, same as registering/unregistering.
> > > > > > > > > > BTW can callbacks for multiple VQs run on multiple CPUs at the moment?
> > > > > > > > > I'm not sure I understand the question.
> > > > > > > > >
> > > > > > > > > I do think we can have multiple CPUs that are executing some portion of
> > > > > > > > > virtio_ccw_int_handler(). So I guess the answer is yes. Connie what do you think?
> > > > > > > > >
> > > > > > > > > On the other hand we could also end up serializing synchronize_cbs()
> > > > > > > > > calls for different devices if they happen to use the same airq_info. But
> > > > > > > > > this probably was not your question
> > > > > > > >
> > > > > > > > I am less concerned about  synchronize_cbs being slow and more about
> > > > > > > > the slowdown in interrupt processing itself.
> > > > > > > >
> > > > > > > > > > this patch serializes them on a spinlock.
> > > > > > > > > >
> > > > > > > > > Those could then pile up on the newly introduced spinlock.
> > > > > > > > >
> > > > > > > > > Regards,
> > > > > > > > > Halil
> > > > > > > > Hmm yea ... not good.
> > > > > > > Is there any other way to synchronize with all callbacks?
> > > > > >
> > > > > >
> > > > > > Maybe using rwlock as airq handler?
> > > > > >
> > > > > > Thanks
> > > > > >
> > > > >
> > > > > rwlock is still a shared cacheline bouncing between CPUs and
> > > > > a bunch of ordering instructions.
> > >
> > > Yes, but it should be faster than spinlocks anyhow.
> > >
> > > > > Maybe something per-cpu + some IPIs to run things on all CPUs instead?
> > >
> > > Is this something like a customized version of synchronzie_rcu_expedited()?
> >
> > With interrupts running in an RCU read size critical section?
> 
> For vring_interrupt(), yes.
> 
> 
> > Quite possibly that is also an option.
> > This will need a bunch of documentation since this is not
> > a standard use of RCU,
> 
> According to Documentation/RCU/requirements.rst, it looks like a legal case:
> 
> "
> The Linux kernel has interrupts, and RCU read-side critical sections are
> legal within interrupt handlers and within interrupt-disabled regions of
> code, as are invocations of call_rcu().
> "

My problem is it is not clear what data is protected by rcu here.
Nothing is tagged with __rcu or uses rcu_dereference.
We need at least an ack from rcu maintainers that this is
a valid use of rcu and not an undocumented side effect.

> And as discussed, synchronize_rcu_expedited() is not friendly to real
> time workload. I think we can simply
> 
> 1) protect vring_interrupt() with rcu_read_lock()
> 2) use synchronize_rcu() in synchronize_cbs for ccw
> 
> And if we care about the long delay we can use per device srcu to reduce that?
> 
> Thanks
> 
> > and probably get a confirmation
> > from RCU maintainers that whatever assumptions we make
> > are guaranteed to hold down the road.
> >
> > > >
> > > > ... and I think classic and device interrupts are different enough
> > > > here ...
> > >
> > > Yes.
> > >
> > > Thanks
> > >
> > > >
> > > > > > >
> > > > > > > > --
> > > > > > > > MST
> > > >
> >

Powered by blists - more mailing lists