> 21/09/2022 04:02, Chaoyong He: > > > On 9/15/2022 11:44 AM, Chaoyong He wrote: > > > > Adds the setup/start logic for the ctrl vNIC. This vNIC is used by > > > > the PMD and flower firmware application as a communication channel > > > between > > > > driver and firmware. In the case of OVS it is also used to > > > > communicate flow statistics from hardware to the driver. > > > > > > > > A rte_eth device is not exposed to DPDK for this vNIC as it is > > > > strictly used internally by flower logic. > > > > > > > > > > Hi Chaoyong, > > > > > > Similar comment with previous versions, interface is created using > > > regular 'rte_eth_dev_allocate()' API, I think interface will be > > > visible to application, I can't understand the need of creating an > > > interface > for control. > > You didn't reply to this. > Why the control port should be exposed to the application? > We recommend not using ethdev for this. >
Actually, in the v1--v5 of this patch series, we did create a control port which is not in the rte_eth_devices[] array, so it won't exposed to the application. > > > > What is the communication method between driver and FW? > > > Since one of the following patches (09/12) introduces Rx/Tx for ctrl > > > interface, is device interface is control packets (similar to network data > packets)? > > > > > > > Basically, the 'control message' is exist in the form of normal data > > packets. > > > > When we use the flower firmware application, there exist two types of > > packets for now, and they are identified only from the prepend meta-data. > > > > Bit 3 3 2 2 2 2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 > > -----\ 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 9 8 7 6 5 4 3 2 1 0 > > Word +---------------+---------------+---------------+---------------+ > > 0 | type | type | type | type | > > > > +---------------+---------------+---------------+---------------+ > > The 'control message' packets are processed by the ctrl vNIC. > > The 'normal' packets are processed by the pf vNIC. > > > > The communication method between driver and firmware is decided by > the > > designment of hardware and firmware. > > > > The kernel driver also has the same ctrl vNIC and pf vNIC ethdev and the > usage is same. > > > > > > Because of the add of ctrl vNIC, a new PCItoCPPBar is needed. > > > > Modify the related logics. > > > > > > > > Signed-off-by: Chaoyong He <chaoyong...@corigine.com> > > > > Reviewed-by: Niklas Söderlund <niklas.soderl...@corigine.com> > >
