Re: [PATCH net-next 4/6] vsock: add vsock_loopback transport

From: Stefano Garzarella
Date: Thu Nov 21 2019 - 05:00:00 EST


On Thu, Nov 21, 2019 at 09:34:58AM +0000, Stefan Hajnoczi wrote:
> On Tue, Nov 19, 2019 at 12:01:19PM +0100, Stefano Garzarella wrote:
>
> Ideas for long-term changes below.
>
> Reviewed-by: Stefan Hajnoczi <stefanha@xxxxxxxxxx>
>

Thanks for reviewing!

> > diff --git a/MAINTAINERS b/MAINTAINERS
> > index 760049454a23..c2a3dc3113ba 100644
> > --- a/MAINTAINERS
> > +++ b/MAINTAINERS
> > @@ -17239,6 +17239,7 @@ F: net/vmw_vsock/diag.c
> > F: net/vmw_vsock/af_vsock_tap.c
> > F: net/vmw_vsock/virtio_transport_common.c
> > F: net/vmw_vsock/virtio_transport.c
> > +F: net/vmw_vsock/vsock_loopback.c
> > F: drivers/net/vsockmon.c
> > F: drivers/vhost/vsock.c
> > F: tools/testing/vsock/
>
> At this point you are most active in virtio-vsock and I am reviewing
> patches on a best-effort basis. Feel free to add yourself as
> maintainer.
>

Sure, I'd be happy to maintain it.

> > diff --git a/net/vmw_vsock/vsock_loopback.c b/net/vmw_vsock/vsock_loopback.c
> > new file mode 100644
> > index 000000000000..3d1c1a88305f
> > --- /dev/null
> > +++ b/net/vmw_vsock/vsock_loopback.c
> > @@ -0,0 +1,217 @@
> > +// SPDX-License-Identifier: GPL-2.0-only
> > +/*
> > + * loopback transport for vsock using virtio_transport_common APIs
> > + *
> > + * Copyright (C) 2013-2019 Red Hat, Inc.
> > + * Author: Asias He <asias@xxxxxxxxxx>
> > + * Stefan Hajnoczi <stefanha@xxxxxxxxxx>
> > + * Stefano Garzarella <sgarzare@xxxxxxxxxx>
> > + *
> > + */
> > +#include <linux/spinlock.h>
> > +#include <linux/module.h>
> > +#include <linux/list.h>
> > +#include <linux/virtio_vsock.h>
>
> Is it time to rename the generic functionality in
> virtio_transport_common.c? This doesn't have anything to do with virtio
> :).
>

Completely agree, new transports could use it to handle the protocol without
reimplementing things already done.

> > +
> > +static struct workqueue_struct *vsock_loopback_workqueue;
> > +static struct vsock_loopback *the_vsock_loopback;
>
> the_vsock_loopback could be a static global variable (not a pointer) and
> vsock_loopback_workqueue could also be included in the struct.
>
> The RCU pointer is really a way to synchronize vsock_loopback_send_pkt()
> and vsock_loopback_cancel_pkt() with module exit. There is no other
> reason for using a pointer.
>
> It's cleaner to implement the synchronization once in af_vsock.c (or
> virtio_transport_common.c) instead of making each transport do it.
> Maybe try_module_get() and related APIs provide the necessary semantics
> so that core vsock code can hold the transport module while it's being
> used to send/cancel a packet.

Right, the module cannot be unloaded until open sockets, so here the
synchronization is not needed.

The synchronization come from virtio-vsock device that can be
hot-unplugged while sockets are still open, but that can't happen here.

I will remove the pointers and RCU in the v2.

Can I keep your R-b or do you prefer to watch v2 first?

>
> > +MODULE_ALIAS_NETPROTO(PF_VSOCK);
>
> Why does this module define the alias for PF_VSOCK? Doesn't another
> module already define this alias?

It is a way to load this module when PF_VSOCK is starting to be used.
MODULE_ALIAS_NETPROTO(PF_VSOCK) is already defined in vmci_transport
and hyperv_transport. IIUC it is used for the same reason.

In virtio_transport we don't need it because it will be loaded when
the PCI device is discovered.

Do you think there's a better way?
Should I include the vsock_loopback transport directly in af_vsock
without creating a new module?

Thanks,
Stefano