On Thu, Aug 18, 2022 at 11:13 PM Guo Zhi <qtxuning1...@sjtu.edu.cn> wrote: > > In virtio-spec 1.1, new feature bit VIRTIO_F_IN_ORDER was introduced. > When this feature has been negotiated, virtio driver will use > descriptors in ring order: starting from offset 0 in the table, and > wrapping around at the end of the table. Virtio devices will always use > descriptors in the same order in which they have been made available. > This can reduce virtio accesses to used ring.
So the bached used buffer seems like an optimization, I'd like re-order the patches 1) Just enable virtio-net in-order without the batching, to see the performance 2) Add the batching optimization on top. > > Based on updated virtio-spec, this series realized IN_ORDER prototype > for virtio-net device in QEMU. Please share with us some performance numbers. E.g pps. > > Some work haven't been done in this patch series: > 1. Virtio device in_order support for packed vq is left for the future. Anything makes packed virtqueue special? Thanks > > Related patches: > In order feature in Linux(support virtio driver, vhost_test and vsock > device): https://lkml.org/lkml/2022/8/17/643 > > Guo Zhi (2): > virtio: expose used buffers > virtio: enable f_in_order feature for virtio-net > > hw/net/virtio-net.c | 30 +++++++++++++++++++++++++++--- > include/hw/virtio/virtio.h | 4 +++- > 2 files changed, 30 insertions(+), 4 deletions(-) > > -- > 2.17.1 >
