> From: Alvaro Karsz <[email protected]> > Sent: Saturday, February 11, 2023 11:14 AM > > I think that I wasn't clear enough. > > I'm not saying that we should not define in the spec how to handle a situation > when a device receives both RX_SET and VQ_SET (or a driver sends both). > I'm saying that I don't think that the driver should handle the situation the > way > you described it: >
> > When the driver prefers to use per virtqueue notifications coalescing, and > > if > queue group (transmit or receive) level notification coalescing is enabled, > driver > SHOULD first disable device level notification coalescing. > > Or it should be, > > > > Virtqueue level notifications coalescing, and device level notifications > > can be > enabled together. > > When both of them are enabled, per virtqueue notifications coalescing take > priority over queue group level. > These are the two options I proposed. In second option missed the case of per vq followed by group level. Which I further acked in [1] to have it as last configured. > I think that if we want to refer to this situation in the spec, it should be > something like: > "A Device should use the last coalescing parameters received for a virtqueue, > regardless of the command used to deliver the parameters." > (just an example to make the point). Yes, as acked in [1]. :) We both agree on #b which you described in above example. So things looks good. Lets wait for Heng to generate v3 covering above case. [1] https://lists.oasis-open.org/archives/virtio-dev/202302/msg00227.html
