* [PATCH v2] vhost: add IRQ suppression
@ 2023-08-31 14:44 Maxime Coquelin
2023-08-31 14:45 ` Maxime Coquelin
2023-09-28 8:43 ` David Marchand
0 siblings, 2 replies; 3+ messages in thread
From: Maxime Coquelin @ 2023-08-31 14:44 UTC (permalink / raw)
To: dev, chenbo.xia, david.marchand, echaudro; +Cc: Maxime Coquelin
Guest notifications offloading, which has been introduced
in v23.07, aims at offloading syscalls out of the datapath.
This patch optimizes the offloading by not offloading the
guest notification for a given virtqueue if one is already
being offloaded by the application.
With a single VDUSE device, we can already see few
notifications being suppressed when doing throughput
testing with Iperf3. We can expect to see much more being
suppressed when the offloading thread is under pressure.
Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
---
lib/vhost/vhost.c | 4 ++++
lib/vhost/vhost.h | 27 +++++++++++++++++++++------
2 files changed, 25 insertions(+), 6 deletions(-)
diff --git a/lib/vhost/vhost.c b/lib/vhost/vhost.c
index eb6309b681..7794f29c18 100644
--- a/lib/vhost/vhost.c
+++ b/lib/vhost/vhost.c
@@ -48,6 +48,8 @@ static const struct vhost_vq_stats_name_off vhost_vq_stat_strings[] = {
stats.guest_notifications_offloaded)},
{"guest_notifications_error", offsetof(struct vhost_virtqueue,
stats.guest_notifications_error)},
+ {"guest_notifications_suppressed", offsetof(struct vhost_virtqueue,
+ stats.guest_notifications_suppressed)},
{"iotlb_hits", offsetof(struct vhost_virtqueue, stats.iotlb_hits)},
{"iotlb_misses", offsetof(struct vhost_virtqueue, stats.iotlb_misses)},
{"inflight_submitted", offsetof(struct vhost_virtqueue, stats.inflight_submitted)},
@@ -1516,6 +1518,8 @@ rte_vhost_notify_guest(int vid, uint16_t queue_id)
rte_rwlock_read_lock(&vq->access_lock);
+ __atomic_store_n(&vq->irq_pending, false, __ATOMIC_RELEASE);
+
if (dev->backend_ops->inject_irq(dev, vq)) {
if (dev->flags & VIRTIO_DEV_STATS_ENABLED)
__atomic_fetch_add(&vq->stats.guest_notifications_error,
diff --git a/lib/vhost/vhost.h b/lib/vhost/vhost.h
index 9723429b1c..3e78379e48 100644
--- a/lib/vhost/vhost.h
+++ b/lib/vhost/vhost.h
@@ -156,6 +156,7 @@ struct virtqueue_stats {
uint64_t iotlb_misses;
uint64_t inflight_submitted;
uint64_t inflight_completed;
+ uint64_t guest_notifications_suppressed;
/* Counters below are atomic, and should be incremented as such. */
uint64_t guest_notifications;
uint64_t guest_notifications_offloaded;
@@ -346,6 +347,8 @@ struct vhost_virtqueue {
struct vhost_vring_addr ring_addrs;
struct virtqueue_stats stats;
+
+ bool irq_pending;
} __rte_cache_aligned;
/* Virtio device status as per Virtio specification */
@@ -908,12 +911,24 @@ vhost_need_event(uint16_t event_idx, uint16_t new_idx, uint16_t old)
static __rte_always_inline void
vhost_vring_inject_irq(struct virtio_net *dev, struct vhost_virtqueue *vq)
{
- if (dev->notify_ops->guest_notify &&
- dev->notify_ops->guest_notify(dev->vid, vq->index)) {
- if (dev->flags & VIRTIO_DEV_STATS_ENABLED)
- __atomic_fetch_add(&vq->stats.guest_notifications_offloaded,
- 1, __ATOMIC_RELAXED);
- return;
+ bool expected = false;
+
+ if (dev->notify_ops->guest_notify) {
+ if (__atomic_compare_exchange_n(&vq->irq_pending, &expected, true, 0,
+ __ATOMIC_RELEASE, __ATOMIC_RELAXED)) {
+ if (dev->notify_ops->guest_notify(dev->vid, vq->index)) {
+ if (dev->flags & VIRTIO_DEV_STATS_ENABLED)
+ __atomic_fetch_add(&vq->stats.guest_notifications_offloaded,
+ 1, __ATOMIC_RELAXED);
+ return;
+ }
+
+ /* Offloading failed, fallback to direct IRQ injection */
+ __atomic_store_n(&vq->irq_pending, 0, __ATOMIC_RELEASE);
+ } else {
+ vq->stats.guest_notifications_suppressed++;
+ return;
+ }
}
if (dev->backend_ops->inject_irq(dev, vq)) {
--
2.41.0
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v2] vhost: add IRQ suppression
2023-08-31 14:44 [PATCH v2] vhost: add IRQ suppression Maxime Coquelin
@ 2023-08-31 14:45 ` Maxime Coquelin
2023-09-28 8:43 ` David Marchand
1 sibling, 0 replies; 3+ messages in thread
From: Maxime Coquelin @ 2023-08-31 14:45 UTC (permalink / raw)
To: dev, chenbo.xia, david.marchand, echaudro
On 8/31/23 16:44, Maxime Coquelin wrote:
> Guest notifications offloading, which has been introduced
> in v23.07, aims at offloading syscalls out of the datapath.
>
> This patch optimizes the offloading by not offloading the
> guest notification for a given virtqueue if one is already
> being offloaded by the application.
>
> With a single VDUSE device, we can already see few
> notifications being suppressed when doing throughput
> testing with Iperf3. We can expect to see much more being
> suppressed when the offloading thread is under pressure.
>
> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
> ---
> lib/vhost/vhost.c | 4 ++++
> lib/vhost/vhost.h | 27 +++++++++++++++++++++------
> 2 files changed, 25 insertions(+), 6 deletions(-)
>
No functionnal change in the V2, just rebased so that CI can apply.
Maxime
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH v2] vhost: add IRQ suppression
2023-08-31 14:44 [PATCH v2] vhost: add IRQ suppression Maxime Coquelin
2023-08-31 14:45 ` Maxime Coquelin
@ 2023-09-28 8:43 ` David Marchand
1 sibling, 0 replies; 3+ messages in thread
From: David Marchand @ 2023-09-28 8:43 UTC (permalink / raw)
To: Maxime Coquelin; +Cc: dev, chenbo.xia, echaudro
On Thu, Aug 31, 2023 at 4:44 PM Maxime Coquelin
<maxime.coquelin@redhat.com> wrote:
>
> Guest notifications offloading, which has been introduced
> in v23.07, aims at offloading syscalls out of the datapath.
>
> This patch optimizes the offloading by not offloading the
> guest notification for a given virtqueue if one is already
> being offloaded by the application.
>
> With a single VDUSE device, we can already see few
> notifications being suppressed when doing throughput
> testing with Iperf3. We can expect to see much more being
> suppressed when the offloading thread is under pressure.
>
> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
> ---
> lib/vhost/vhost.c | 4 ++++
> lib/vhost/vhost.h | 27 +++++++++++++++++++++------
> 2 files changed, 25 insertions(+), 6 deletions(-)
>
> diff --git a/lib/vhost/vhost.c b/lib/vhost/vhost.c
> index eb6309b681..7794f29c18 100644
> --- a/lib/vhost/vhost.c
> +++ b/lib/vhost/vhost.c
> @@ -48,6 +48,8 @@ static const struct vhost_vq_stats_name_off vhost_vq_stat_strings[] = {
> stats.guest_notifications_offloaded)},
> {"guest_notifications_error", offsetof(struct vhost_virtqueue,
> stats.guest_notifications_error)},
> + {"guest_notifications_suppressed", offsetof(struct vhost_virtqueue,
> + stats.guest_notifications_suppressed)},
> {"iotlb_hits", offsetof(struct vhost_virtqueue, stats.iotlb_hits)},
> {"iotlb_misses", offsetof(struct vhost_virtqueue, stats.iotlb_misses)},
> {"inflight_submitted", offsetof(struct vhost_virtqueue, stats.inflight_submitted)},
> @@ -1516,6 +1518,8 @@ rte_vhost_notify_guest(int vid, uint16_t queue_id)
>
> rte_rwlock_read_lock(&vq->access_lock);
>
> + __atomic_store_n(&vq->irq_pending, false, __ATOMIC_RELEASE);
> +
> if (dev->backend_ops->inject_irq(dev, vq)) {
> if (dev->flags & VIRTIO_DEV_STATS_ENABLED)
> __atomic_fetch_add(&vq->stats.guest_notifications_error,
> diff --git a/lib/vhost/vhost.h b/lib/vhost/vhost.h
> index 9723429b1c..3e78379e48 100644
> --- a/lib/vhost/vhost.h
> +++ b/lib/vhost/vhost.h
> @@ -156,6 +156,7 @@ struct virtqueue_stats {
> uint64_t iotlb_misses;
> uint64_t inflight_submitted;
> uint64_t inflight_completed;
> + uint64_t guest_notifications_suppressed;
> /* Counters below are atomic, and should be incremented as such. */
> uint64_t guest_notifications;
> uint64_t guest_notifications_offloaded;
> @@ -346,6 +347,8 @@ struct vhost_virtqueue {
>
> struct vhost_vring_addr ring_addrs;
> struct virtqueue_stats stats;
> +
> + bool irq_pending;
> } __rte_cache_aligned;
>
> /* Virtio device status as per Virtio specification */
> @@ -908,12 +911,24 @@ vhost_need_event(uint16_t event_idx, uint16_t new_idx, uint16_t old)
> static __rte_always_inline void
> vhost_vring_inject_irq(struct virtio_net *dev, struct vhost_virtqueue *vq)
> {
> - if (dev->notify_ops->guest_notify &&
> - dev->notify_ops->guest_notify(dev->vid, vq->index)) {
> - if (dev->flags & VIRTIO_DEV_STATS_ENABLED)
> - __atomic_fetch_add(&vq->stats.guest_notifications_offloaded,
> - 1, __ATOMIC_RELAXED);
> - return;
> + bool expected = false;
> +
> + if (dev->notify_ops->guest_notify) {
> + if (__atomic_compare_exchange_n(&vq->irq_pending, &expected, true, 0,
> + __ATOMIC_RELEASE, __ATOMIC_RELAXED)) {
> + if (dev->notify_ops->guest_notify(dev->vid, vq->index)) {
> + if (dev->flags & VIRTIO_DEV_STATS_ENABLED)
> + __atomic_fetch_add(&vq->stats.guest_notifications_offloaded,
> + 1, __ATOMIC_RELAXED);
> + return;
> + }
> +
> + /* Offloading failed, fallback to direct IRQ injection */
> + __atomic_store_n(&vq->irq_pending, 0, __ATOMIC_RELEASE);
Nit: s/0/false/
> + } else {
> + vq->stats.guest_notifications_suppressed++;
> + return;
> + }
> }
>
> if (dev->backend_ops->inject_irq(dev, vq)) {
> --
> 2.41.0
>
--
David Marchand
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2023-09-28 8:44 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2023-08-31 14:44 [PATCH v2] vhost: add IRQ suppression Maxime Coquelin
2023-08-31 14:45 ` Maxime Coquelin
2023-09-28 8:43 ` David Marchand
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).