[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[PULL v2 47/53] virtio: use shadow_avail_idx while checking number of he
|
From: |
Michael S. Tsirkin |
|
Subject: |
[PULL v2 47/53] virtio: use shadow_avail_idx while checking number of heads |
|
Date: |
Wed, 4 Oct 2023 23:45:36 -0400 |
From: Ilya Maximets <i.maximets@ovn.org>
We do not need the most up to date number of heads, we only want to
know if there is at least one.
Use shadow variable as long as it is not equal to the last available
index checked. This avoids expensive qatomic dereference of the
RCU-protected memory region cache as well as the memory access itself.
The change improves performance of the af-xdp network backend by 2-3%.
Signed-off-by: Ilya Maximets <i.maximets@ovn.org>
Message-Id: <20230927135157.2316982-1-i.maximets@ovn.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
---
hw/virtio/virtio.c | 18 +++++++++++++++---
1 file changed, 15 insertions(+), 3 deletions(-)
diff --git a/hw/virtio/virtio.c b/hw/virtio/virtio.c
index c727e9201b..2058b838e9 100644
--- a/hw/virtio/virtio.c
+++ b/hw/virtio/virtio.c
@@ -997,7 +997,12 @@ void virtqueue_push(VirtQueue *vq, const VirtQueueElement
*elem,
/* Called within rcu_read_lock(). */
static int virtqueue_num_heads(VirtQueue *vq, unsigned int idx)
{
- uint16_t num_heads = vring_avail_idx(vq) - idx;
+ uint16_t avail_idx, num_heads;
+
+ /* Use shadow index whenever possible. */
+ avail_idx = (vq->shadow_avail_idx != idx) ? vq->shadow_avail_idx
+ : vring_avail_idx(vq);
+ num_heads = avail_idx - idx;
/* Check it isn't doing very strange things with descriptor numbers. */
if (num_heads > vq->vring.num) {
@@ -1005,8 +1010,15 @@ static int virtqueue_num_heads(VirtQueue *vq, unsigned
int idx)
idx, vq->shadow_avail_idx);
return -EINVAL;
}
- /* On success, callers read a descriptor at vq->last_avail_idx.
- * Make sure descriptor read does not bypass avail index read. */
+ /*
+ * On success, callers read a descriptor at vq->last_avail_idx.
+ * Make sure descriptor read does not bypass avail index read.
+ *
+ * This is necessary even if we are using a shadow index, since
+ * the shadow index could have been initialized by calling
+ * vring_avail_idx() outside of this function, i.e., by a guest
+ * memory read not accompanied by a barrier.
+ */
if (num_heads) {
smp_rmb();
}
--
MST
- [PULL v2 36/53] hw/cxl: Add utility functions decoder interleave ways and target count., (continued)
- [PULL v2 36/53] hw/cxl: Add utility functions decoder interleave ways and target count., Michael S. Tsirkin, 2023/10/04
- [PULL v2 37/53] hw/cxl: Fix and use same calculation for HDM decoder block size everywhere, Michael S. Tsirkin, 2023/10/04
- [PULL v2 35/53] hw/cxl: Push cxl_decoder_count_enc() and cxl_decode_ig() into .c, Michael S. Tsirkin, 2023/10/04
- [PULL v2 38/53] hw/cxl: Support 4 HDM decoders at all levels of topology, Michael S. Tsirkin, 2023/10/04
- [PULL v2 41/53] vdpa net: stop probing if cannot set features, Michael S. Tsirkin, 2023/10/04
- [PULL v2 46/53] libvhost-user.c: add assertion to vu_message_read_default, Michael S. Tsirkin, 2023/10/04
- [PULL v2 43/53] amd_iommu: Fix APIC address check, Michael S. Tsirkin, 2023/10/04
- [PULL v2 44/53] hw/i386/pc: improve physical address space bound check for 32-bit x86 systems, Michael S. Tsirkin, 2023/10/04
- [PULL v2 47/53] virtio: use shadow_avail_idx while checking number of heads,
Michael S. Tsirkin <=
- [PULL v2 50/53] util/uuid: add a hash function, Michael S. Tsirkin, 2023/10/04
- [PULL v2 48/53] virtio: remove unnecessary thread fence while reading next descriptor, Michael S. Tsirkin, 2023/10/04
- [PULL v2 52/53] vhost-user: add shared_object msg, Michael S. Tsirkin, 2023/10/04
- [PULL v2 53/53] libvhost-user: handle shared_object msg, Michael S. Tsirkin, 2023/10/04
- [PULL v2 51/53] hw/display: introduce virtio-dmabuf, Michael S. Tsirkin, 2023/10/04
- [PULL v2 49/53] virtio: remove unused next argument from virtqueue_split_read_next_desc(), Michael S. Tsirkin, 2023/10/04
- Re: [PULL v2 00/53] virtio,pci: features, cleanups, Stefan Hajnoczi, 2023/10/05