From: Jason Wang <jasowang@redhat.com>
To: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Eli Cohen <elic@nvidia.com>,
virtualization <virtualization@lists.linux-foundation.org>,
linux-kernel <linux-kernel@vger.kernel.org>
Subject: Re: [PATCH 2/2] vdpa: mlx5: synchronize driver status with CVQ
Date: Mon, 21 Mar 2022 15:36:16 +0800 [thread overview]
Message-ID: <CACGkMEvQCBTk7WVLvRAsgj1OOcE7rh5DEpN91HOdZLZyagcbuw@mail.gmail.com> (raw)
In-Reply-To: <20220321032130-mutt-send-email-mst@kernel.org>
On Mon, Mar 21, 2022 at 3:24 PM Michael S. Tsirkin <mst@redhat.com> wrote:
>
> On Mon, Mar 21, 2022 at 02:04:29PM +0800, Jason Wang wrote:
> > Currently, CVQ doesn't have any synchronization with the driver
> > status. Then CVQ emulation code run in the middle of:
> >
> > 1) device reset
> > 2) device status changed
> > 3) map updating
> >
> > The will lead several unexpected issue like trying to execute CVQ
> > command after the driver has been teared down.
> >
> > Fixing this by using reslock to synchronize CVQ emulation code with
> > the driver status changing:
> >
> > - protect the whole device reset, status changing and map updating
> > with reslock
> > - protect the CVQ handler with the reslock and check
> > VIRTIO_CONFIG_S_DRIVER_OK in the CVQ handler
> >
> > This will guarantee that:
> >
> > 1) CVQ handler won't work if VIRTIO_CONFIG_S_DRIVER_OK is not set
> > 2) CVQ handler will see a consistent state of the driver instead of
> > the partial one when it is running in the middle of the
> > teardown_driver() or setup_driver().
> >
> > Cc: 5262912ef3cfc ("vdpa/mlx5: Add support for control VQ and MAC setting")
> > Signed-off-by: Jason Wang <jasowang@redhat.com>
> > ---
> > drivers/vdpa/mlx5/net/mlx5_vnet.c | 42 +++++++++++++++++++++++--------
> > 1 file changed, 31 insertions(+), 11 deletions(-)
> >
> > diff --git a/drivers/vdpa/mlx5/net/mlx5_vnet.c b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > index d5a6fb3f9c41..524240f55c1c 100644
> > --- a/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > +++ b/drivers/vdpa/mlx5/net/mlx5_vnet.c
> > @@ -1618,11 +1618,17 @@ static void mlx5_cvq_kick_handler(struct work_struct *work)
> > mvdev = wqent->mvdev;
> > ndev = to_mlx5_vdpa_ndev(mvdev);
> > cvq = &mvdev->cvq;
> > +
> > + mutex_lock(&ndev->reslock);
> > +
> > + if (!(mvdev->status & VIRTIO_CONFIG_S_DRIVER_OK))
> > + goto done;
> > +
> > if (!(ndev->mvdev.actual_features & BIT_ULL(VIRTIO_NET_F_CTRL_VQ)))
> > - return;
> > + goto done;
> >
> > if (!cvq->ready)
> > - return;
> > + goto done;
> >
> > while (true) {
> > err = vringh_getdesc_iotlb(&cvq->vring, &cvq->riov, &cvq->wiov, &cvq->head,
> > @@ -1663,6 +1669,9 @@ static void mlx5_cvq_kick_handler(struct work_struct *work)
> > break;
> > }
> > }
> > +
> > +done:
> > + mutex_unlock(&ndev->reslock);
> > }
> >
> > static void mlx5_vdpa_kick_vq(struct vdpa_device *vdev, u16 idx)
> > @@ -2125,6 +2134,8 @@ static int mlx5_vdpa_change_map(struct mlx5_vdpa_dev *mvdev, struct vhost_iotlb
> > struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev);
> > int err;
> >
> > + mutex_lock(&ndev->reslock);
> > +
> > suspend_vqs(ndev);
> > err = save_channels_info(ndev);
> > if (err)
> > @@ -2137,18 +2148,20 @@ static int mlx5_vdpa_change_map(struct mlx5_vdpa_dev *mvdev, struct vhost_iotlb
> > goto err_mr;
> >
> > if (!(mvdev->status & VIRTIO_CONFIG_S_DRIVER_OK))
> > - return 0;
> > + goto err_mr;
> >
> > restore_channels_info(ndev);
> > err = setup_driver(mvdev);
> > if (err)
> > goto err_setup;
> >
> > + mutex_unlock(&ndev->reslock);
> > return 0;
> >
> > err_setup:
> > mlx5_vdpa_destroy_mr(mvdev);
> > err_mr:
> > + mutex_unlock(&ndev->reslock);
> > return err;
> > }
> >
> > @@ -2157,7 +2170,8 @@ static int setup_driver(struct mlx5_vdpa_dev *mvdev)
> > struct mlx5_vdpa_net *ndev = to_mlx5_vdpa_ndev(mvdev);
> > int err;
> >
> > - mutex_lock(&ndev->reslock);
> > + WARN_ON(!mutex_is_locked(&ndev->reslock));
> > +
> > if (ndev->setup) {
> > mlx5_vdpa_warn(mvdev, "setup driver called for already setup driver\n");
> > err = 0;
>
>
> Maybe also add a comment near function header explaining this must be
> called with lock held.
Will do.
Thanks
>
next prev parent reply other threads:[~2022-03-21 7:36 UTC|newest]
Thread overview: 18+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-03-21 6:04 [PATCH 1/2] vdpa: mlx5: prevent cvq work from hogging CPU Jason Wang
2022-03-21 6:04 ` [PATCH 2/2] vdpa: mlx5: synchronize driver status with CVQ Jason Wang
2022-03-21 6:56 ` Eli Cohen
2022-03-21 7:23 ` Michael S. Tsirkin
2022-03-21 7:36 ` Jason Wang [this message]
2022-03-21 6:31 ` [PATCH 1/2] vdpa: mlx5: prevent cvq work from hogging CPU Eli Cohen
2022-03-21 7:20 ` Michael S. Tsirkin
2022-03-21 7:35 ` Jason Wang
[not found] ` <20220321085317.3148-1-hdanton@sina.com>
2022-03-21 8:59 ` Jason Wang
2022-03-21 9:00 ` Jason Wang
[not found] ` <20220321123420.3207-1-hdanton@sina.com>
2022-03-22 1:59 ` Jason Wang
[not found] ` <20220324005345.3623-1-hdanton@sina.com>
2022-03-24 2:34 ` Jason Wang
[not found] ` <20220324060419.3682-1-hdanton@sina.com>
2022-03-24 6:17 ` Michael S. Tsirkin
2022-03-24 8:20 ` Jason Wang
[not found] ` <20220324120217.3746-1-hdanton@sina.com>
2022-03-24 12:24 ` Eli Cohen
2022-03-25 3:22 ` Jason Wang
2022-03-25 6:45 ` Michael S. Tsirkin
2022-03-25 7:53 ` Jason Wang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CACGkMEvQCBTk7WVLvRAsgj1OOcE7rh5DEpN91HOdZLZyagcbuw@mail.gmail.com \
--to=jasowang@redhat.com \
--cc=elic@nvidia.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mst@redhat.com \
--cc=virtualization@lists.linux-foundation.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).