| Message ID | 20180605074046.20709-5-tiwei.bie@intel.com |
|---|---|
| State | RFC, archived |
| Delegated to: | David Miller |
| Headers | show |
| Series | virtio: support packed ring | expand |
On 2018年06月05日 15:40, Tiwei Bie wrote: > static bool virtqueue_enable_cb_delayed_packed(struct virtqueue *_vq) > { > struct vring_virtqueue *vq = to_vvq(_vq); > + u16 bufs, used_idx, wrap_counter; > > START_USE(vq); > > /* We optimistically turn back on interrupts, then check if there was > * more to do. */ > + /* Depending on the VIRTIO_RING_F_EVENT_IDX feature, we need to > + * either clear the flags bit or point the event index at the next > + * entry. Always update the event index to keep code simple. */ > + Maybe for packed ring, it's time to treat event index separately to avoid a virtio_wmb() for event idx is off. > + /* TODO: tune this threshold */ > + if (vq->next_avail_idx < vq->last_used_idx) > + bufs = (vq->vring_packed.num + vq->next_avail_idx - > + vq->last_used_idx) * 3 / 4; > + else > + bufs = (vq->next_avail_idx - vq->last_used_idx) * 3 / 4; vq->next_avail-idx could be equal to vq->last_usd_idx when the ring is full. Though virito-net is the only user now and it can guarantee this won't happen. But consider this is a core API, we should make sure it can work for any cases. It looks to me that bufs is just vq->vring_packed.num - vq->num_free? > + > + wrap_counter = vq->used_wrap_counter; > + > + used_idx = vq->last_used_idx + bufs; > + if (used_idx >= vq->vring_packed.num) { > + used_idx -= vq->vring_packed.num; > + wrap_counter ^= 1; > + } > + > + vq->vring_packed.driver->off_wrap = cpu_to_virtio16(_vq->vdev, > + used_idx | (wrap_counter << 15)); > > if (vq->event_flags_shadow == VRING_EVENT_F_DISABLE) { > - vq->event_flags_shadow = VRING_EVENT_F_ENABLE; > + /* We need to update event offset and event wrap > + * counter first before updating event flags. */ > + virtio_wmb(vq->weak_barriers); > + vq->event_flags_shadow = vq->event ? VRING_EVENT_F_DESC : > + VRING_EVENT_F_ENABLE; > vq->vring_packed.driver->flags = cpu_to_virtio16(_vq->vdev, > vq->event_flags_shadow); > - /* We need to enable interrupts first before re-checking > - * for more used buffers. */ > - virtio_mb(vq->weak_barriers); > } > > + /* We need to update event suppression structure first > + * before re-checking for more used buffers. */ > + virtio_mb(vq->weak_barriers); > + > if (more_used_packed(vq)) { > END_USE(vq); > return false; I think what we need to to make sure the descriptor used_idx is used? Otherwise we may stop and restart qdisc too frequently? Thanks > --
On Thu, Jun 07, 2018 at 05:50:32PM +0800, Jason Wang wrote: > On 2018年06月05日 15:40, Tiwei Bie wrote: > > static bool virtqueue_enable_cb_delayed_packed(struct virtqueue *_vq) > > { > > struct vring_virtqueue *vq = to_vvq(_vq); > > + u16 bufs, used_idx, wrap_counter; > > START_USE(vq); > > /* We optimistically turn back on interrupts, then check if there was > > * more to do. */ > > + /* Depending on the VIRTIO_RING_F_EVENT_IDX feature, we need to > > + * either clear the flags bit or point the event index at the next > > + * entry. Always update the event index to keep code simple. */ > > + > > Maybe for packed ring, it's time to treat event index separately to avoid a > virtio_wmb() for event idx is off. Okay. I'll do it. > > > + /* TODO: tune this threshold */ > > + if (vq->next_avail_idx < vq->last_used_idx) > > + bufs = (vq->vring_packed.num + vq->next_avail_idx - > > + vq->last_used_idx) * 3 / 4; > > + else > > + bufs = (vq->next_avail_idx - vq->last_used_idx) * 3 / 4; > > vq->next_avail-idx could be equal to vq->last_usd_idx when the ring is full. > Though virito-net is the only user now and it can guarantee this won't > happen. But consider this is a core API, we should make sure it can work for > any cases. > > It looks to me that bufs is just vq->vring_packed.num - vq->num_free? I'll try it! Thanks for the suggestion! :) > > > + > > + wrap_counter = vq->used_wrap_counter; > > + > > + used_idx = vq->last_used_idx + bufs; > > + if (used_idx >= vq->vring_packed.num) { > > + used_idx -= vq->vring_packed.num; > > + wrap_counter ^= 1; > > + } > > + > > + vq->vring_packed.driver->off_wrap = cpu_to_virtio16(_vq->vdev, > > + used_idx | (wrap_counter << 15)); > > if (vq->event_flags_shadow == VRING_EVENT_F_DISABLE) { > > - vq->event_flags_shadow = VRING_EVENT_F_ENABLE; > > + /* We need to update event offset and event wrap > > + * counter first before updating event flags. */ > > + virtio_wmb(vq->weak_barriers); > > + vq->event_flags_shadow = vq->event ? VRING_EVENT_F_DESC : > > + VRING_EVENT_F_ENABLE; > > vq->vring_packed.driver->flags = cpu_to_virtio16(_vq->vdev, > > vq->event_flags_shadow); > > - /* We need to enable interrupts first before re-checking > > - * for more used buffers. */ > > - virtio_mb(vq->weak_barriers); > > } > > + /* We need to update event suppression structure first > > + * before re-checking for more used buffers. */ > > + virtio_mb(vq->weak_barriers); > > + > > if (more_used_packed(vq)) { > > END_USE(vq); > > return false; > > I think what we need to to make sure the descriptor used_idx is used? > Otherwise we may stop and restart qdisc too frequently? Good catch! Yeah. It's not the best choice to check the descriptor at last_used_idx. I'll try to fix this! Best regards, Tiwei Bie > > Thanks > > > -- >
diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c index 983ce5ffda1b..aefd7ac40928 100644 --- a/drivers/virtio/virtio_ring.c +++ b/drivers/virtio/virtio_ring.c @@ -1048,7 +1048,7 @@ static inline int virtqueue_add_packed(struct virtqueue *_vq, static bool virtqueue_kick_prepare_packed(struct virtqueue *_vq) { struct vring_virtqueue *vq = to_vvq(_vq); - u16 flags; + u16 new, old, off_wrap, flags, wrap_counter, event_idx; bool needs_kick; u32 snapshot; @@ -1057,9 +1057,19 @@ static bool virtqueue_kick_prepare_packed(struct virtqueue *_vq) * suppressions. */ virtio_mb(vq->weak_barriers); + old = vq->next_avail_idx - vq->num_added; + new = vq->next_avail_idx; + vq->num_added = 0; + snapshot = *(u32 *)vq->vring_packed.device; + off_wrap = virtio16_to_cpu(_vq->vdev, (__virtio16)(snapshot & 0xffff)); flags = virtio16_to_cpu(_vq->vdev, (__virtio16)(snapshot >> 16)) & 0x3; + wrap_counter = off_wrap >> 15; + event_idx = off_wrap & ~(1<<15); + if (wrap_counter != vq->avail_wrap_counter) + event_idx -= vq->vring_packed.num; + #ifdef DEBUG if (vq->last_add_time_valid) { WARN_ON(ktime_to_ms(ktime_sub(ktime_get(), @@ -1068,7 +1078,10 @@ static bool virtqueue_kick_prepare_packed(struct virtqueue *_vq) vq->last_add_time_valid = false; #endif - needs_kick = (flags != VRING_EVENT_F_DISABLE); + if (flags == VRING_EVENT_F_DESC) + needs_kick = vring_need_event(event_idx, new, old); + else + needs_kick = (flags != VRING_EVENT_F_DISABLE); END_USE(vq); return needs_kick; } @@ -1177,6 +1190,15 @@ static void *virtqueue_get_buf_ctx_packed(struct virtqueue *_vq, ret = vq->desc_state_packed[id].data; detach_buf_packed(vq, id, ctx); + /* If we expect an interrupt for the next entry, tell host + * by writing event index and flush out the write before + * the read in the next get_buf call. */ + if (vq->event_flags_shadow == VRING_EVENT_F_DESC) + virtio_store_mb(vq->weak_barriers, + &vq->vring_packed.driver->off_wrap, + cpu_to_virtio16(_vq->vdev, vq->last_used_idx | + ((u16)vq->used_wrap_counter << 15))); + #ifdef DEBUG vq->last_add_time_valid = false; #endif @@ -1204,9 +1226,20 @@ static unsigned virtqueue_enable_cb_prepare_packed(struct virtqueue *_vq) /* We optimistically turn back on interrupts, then check if there was * more to do. */ + /* Depending on the VIRTIO_RING_F_EVENT_IDX feature, we need to + * either clear the flags bit or point the event index at the next + * entry. Always update the event index to keep code simple. */ + + vq->vring_packed.driver->off_wrap = cpu_to_virtio16(_vq->vdev, + vq->last_used_idx | + ((u16)vq->used_wrap_counter << 15)); if (vq->event_flags_shadow == VRING_EVENT_F_DISABLE) { - vq->event_flags_shadow = VRING_EVENT_F_ENABLE; + /* We need to update event offset and event wrap + * counter first before updating event flags. */ + virtio_wmb(vq->weak_barriers); + vq->event_flags_shadow = vq->event ? VRING_EVENT_F_DESC : + VRING_EVENT_F_ENABLE; vq->vring_packed.driver->flags = cpu_to_virtio16(_vq->vdev, vq->event_flags_shadow); } @@ -1232,21 +1265,48 @@ static bool virtqueue_poll_packed(struct virtqueue *_vq, unsigned last_used_idx) static bool virtqueue_enable_cb_delayed_packed(struct virtqueue *_vq) { struct vring_virtqueue *vq = to_vvq(_vq); + u16 bufs, used_idx, wrap_counter; START_USE(vq); /* We optimistically turn back on interrupts, then check if there was * more to do. */ + /* Depending on the VIRTIO_RING_F_EVENT_IDX feature, we need to + * either clear the flags bit or point the event index at the next + * entry. Always update the event index to keep code simple. */ + + /* TODO: tune this threshold */ + if (vq->next_avail_idx < vq->last_used_idx) + bufs = (vq->vring_packed.num + vq->next_avail_idx - + vq->last_used_idx) * 3 / 4; + else + bufs = (vq->next_avail_idx - vq->last_used_idx) * 3 / 4; + + wrap_counter = vq->used_wrap_counter; + + used_idx = vq->last_used_idx + bufs; + if (used_idx >= vq->vring_packed.num) { + used_idx -= vq->vring_packed.num; + wrap_counter ^= 1; + } + + vq->vring_packed.driver->off_wrap = cpu_to_virtio16(_vq->vdev, + used_idx | (wrap_counter << 15)); if (vq->event_flags_shadow == VRING_EVENT_F_DISABLE) { - vq->event_flags_shadow = VRING_EVENT_F_ENABLE; + /* We need to update event offset and event wrap + * counter first before updating event flags. */ + virtio_wmb(vq->weak_barriers); + vq->event_flags_shadow = vq->event ? VRING_EVENT_F_DESC : + VRING_EVENT_F_ENABLE; vq->vring_packed.driver->flags = cpu_to_virtio16(_vq->vdev, vq->event_flags_shadow); - /* We need to enable interrupts first before re-checking - * for more used buffers. */ - virtio_mb(vq->weak_barriers); } + /* We need to update event suppression structure first + * before re-checking for more used buffers. */ + virtio_mb(vq->weak_barriers); + if (more_used_packed(vq)) { END_USE(vq); return false;
This commit introduces the EVENT_IDX support in packed ring. Signed-off-by: Tiwei Bie <tiwei.bie@intel.com> --- drivers/virtio/virtio_ring.c | 74 ++++++++++++++++++++++++++++++++---- 1 file changed, 67 insertions(+), 7 deletions(-)