virtio — archive

[Michael S. Tsirkin <mst@redhat.com>]

On Tue, Jun 06, 2023 at 10:46:53PM +0000, Parav Pandit wrote: > > > > From: Michael S. Tsirkin <mst@redhat.com> > > Sent: Tuesday, June 6, 2023 6:37 PM > > > > # 3. Requirements > > > ## 3.1 Device counters > > > @@ -143,3 +144,10 @@ struct vnet_rx_completion { 7. The device should > > > be able to indicate when a full page buffer is consumed, > > > which can be recycled by the driver when the packets from the completed > > > page is fully consumed. > > > + > > > +## 3.3 Virtqueue notification coalescing re-enable support 1. Tx and > > > +Rx virtqueue notification coalescing should auto-disable on > > > + notification reporting to the driver. The driver should be able to enable > > > + coalescing after processing the packets per VQ. This ensures that when > > > + networking stack decides to poll, no new notifications are generated when > > > + per VQ notification coalescing is used. > > > > I don't know what does auto-disable and enable coalescing mean. > > Does this refer to VIRTIO_NET_F_NOTF_COAL? > > > what is the problem this is trying to solve? > > > Current driver when uses F_VQ_NOTF_COAL, and polling, it still receives the event (interrupt) from the device, because the time is always running. > (event is not auto disabled on generation of the event). > So number of interrupts are not fully controlled yet by the driver. could we please speak virtio terms? you are not new in this area.. so this is asking for a way to disable used buffer notifications? we actually have this with packet vq always and with split without EVENT_IDX. what is missing? -- MST