On Tue, Dec 7, 2021 at 10:45 AM Mike Christie <michael.christie at oracle.com> wrote:> > vhost_poll_flush() is a simple wrapper around vhost_work_dev_flush(). > It gives wrong impression that we are doing some work over vhost_poll, > while in fact it flushes vhost_poll->dev.This "problem" is a byproduct of 7235acdb1144 ("vhost: simplify work flushing"). Before that we indeed have per poll flush flush.> It only complicate understanding of the code and leads to mistakes > like flushing the same vhost_dev several times in a row. > > Just remove vhost_poll_flush() and call vhost_work_dev_flush() directly.Not a native speaker but since we don't have an per work flush, is it better to rename this simply as vhost_flush()? Thanks> > Signed-off-by: Andrey Ryabinin <arbn at yandex-team.com> > [merge vhost_poll_flush removal from Stefano Garzarella] > Signed-off-by: Mike Christie <michael.christie at oracle.com> > --- > drivers/vhost/net.c | 4 ++-- > drivers/vhost/test.c | 2 +- > drivers/vhost/vhost.c | 12 ++---------- > drivers/vhost/vhost.h | 1 - > drivers/vhost/vsock.c | 2 +- > 5 files changed, 6 insertions(+), 15 deletions(-) > > diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c > index 28ef323882fb..11221f6d11b8 100644 > --- a/drivers/vhost/net.c > +++ b/drivers/vhost/net.c > @@ -1375,8 +1375,8 @@ static void vhost_net_stop(struct vhost_net *n, struct socket **tx_sock, > > static void vhost_net_flush_vq(struct vhost_net *n, int index) > { > - vhost_poll_flush(n->poll + index); > - vhost_poll_flush(&n->vqs[index].vq.poll); > + vhost_work_dev_flush(n->poll[index].dev); > + vhost_work_dev_flush(n->vqs[index].vq.poll.dev); > } > > static void vhost_net_flush(struct vhost_net *n) > diff --git a/drivers/vhost/test.c b/drivers/vhost/test.c > index a09dedc79f68..1a8ab1d8cb1c 100644 > --- a/drivers/vhost/test.c > +++ b/drivers/vhost/test.c > @@ -146,7 +146,7 @@ static void vhost_test_stop(struct vhost_test *n, void **privatep) > > static void vhost_test_flush_vq(struct vhost_test *n, int index) > { > - vhost_poll_flush(&n->vqs[index].poll); > + vhost_work_dev_flush(n->vqs[index].poll.dev); > } > > static void vhost_test_flush(struct vhost_test *n) > diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c > index 8cf259d798c0..7346fa519eb6 100644 > --- a/drivers/vhost/vhost.c > +++ b/drivers/vhost/vhost.c > @@ -244,14 +244,6 @@ void vhost_work_dev_flush(struct vhost_dev *dev) > } > EXPORT_SYMBOL_GPL(vhost_work_dev_flush); > > -/* Flush any work that has been scheduled. When calling this, don't hold any > - * locks that are also used by the callback. */ > -void vhost_poll_flush(struct vhost_poll *poll) > -{ > - vhost_work_dev_flush(poll->dev); > -} > -EXPORT_SYMBOL_GPL(vhost_poll_flush); > - > void vhost_work_queue(struct vhost_dev *dev, struct vhost_work *work) > { > if (!dev->worker) > @@ -677,7 +669,7 @@ void vhost_dev_stop(struct vhost_dev *dev) > for (i = 0; i < dev->nvqs; ++i) { > if (dev->vqs[i]->kick && dev->vqs[i]->handle_kick) { > vhost_poll_stop(&dev->vqs[i]->poll); > - vhost_poll_flush(&dev->vqs[i]->poll); > + vhost_work_dev_flush(dev->vqs[i]->poll.dev); > } > } > } > @@ -1721,7 +1713,7 @@ long vhost_vring_ioctl(struct vhost_dev *d, unsigned int ioctl, void __user *arg > mutex_unlock(&vq->mutex); > > if (pollstop && vq->handle_kick) > - vhost_poll_flush(&vq->poll); > + vhost_work_dev_flush(vq->poll.dev); > return r; > } > EXPORT_SYMBOL_GPL(vhost_vring_ioctl); > diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h > index 09748694cb66..67b23e178812 100644 > --- a/drivers/vhost/vhost.h > +++ b/drivers/vhost/vhost.h > @@ -56,7 +56,6 @@ void vhost_poll_init(struct vhost_poll *poll, vhost_work_fn_t fn, > __poll_t mask, struct vhost_dev *dev); > int vhost_poll_start(struct vhost_poll *poll, struct file *file); > void vhost_poll_stop(struct vhost_poll *poll); > -void vhost_poll_flush(struct vhost_poll *poll); > void vhost_poll_queue(struct vhost_poll *poll); > void vhost_work_dev_flush(struct vhost_dev *dev); > > diff --git a/drivers/vhost/vsock.c b/drivers/vhost/vsock.c > index d6ca1c7ad513..2339587bcd31 100644 > --- a/drivers/vhost/vsock.c > +++ b/drivers/vhost/vsock.c > @@ -707,7 +707,7 @@ static void vhost_vsock_flush(struct vhost_vsock *vsock) > > for (i = 0; i < ARRAY_SIZE(vsock->vqs); i++) > if (vsock->vqs[i].handle_kick) > - vhost_poll_flush(&vsock->vqs[i].poll); > + vhost_work_dev_flush(vsock->vqs[i].poll.dev); > vhost_work_dev_flush(&vsock->dev); > } > > -- > 2.25.1 >
Mike Christie
2021-Dec-08 16:41 UTC
[PATCH 1/7] vhost: get rid of vhost_poll_flush() wrapper
On 12/7/21 9:49 PM, Jason Wang wrote:> On Tue, Dec 7, 2021 at 10:45 AM Mike Christie > <michael.christie at oracle.com> wrote: >> >> vhost_poll_flush() is a simple wrapper around vhost_work_dev_flush(). >> It gives wrong impression that we are doing some work over vhost_poll, >> while in fact it flushes vhost_poll->dev. > > This "problem" is a byproduct of 7235acdb1144 ("vhost: simplify work flushing"). > > Before that we indeed have per poll flush flush. > >> It only complicate understanding of the code and leads to mistakes >> like flushing the same vhost_dev several times in a row. >> >> Just remove vhost_poll_flush() and call vhost_work_dev_flush() directly. > > Not a native speaker but since we don't have an per work flush, is it > better to rename this simply as vhost_flush()? >What about vhost_dev_flush? For the existing naming when we have a function exported we tend to have "vhost_" then the object/struct it works on then the action. For work we have: vhost_work_queue/init (we also have vhost_has_work which doesn't follow that pattern but would sound strange as vhost_work_has so ignore that one). For dev operations we have: vhost_dev_reset_owner/set_owner/has_owner/cleanup/init For the flush operation I wanted it to reflect it flushed all work on the device, so I mashed up the work and dev naming above and I agree it's a little strange.