Using QP0 for anything other than passing IB management packets is basically a hack, why should this be a model for other devices? Leah ________________________________________ From: linux-rdma-owner@xxxxxxxxxxxxxxx <linux-rdma-owner@xxxxxxxxxxxxxxx> on behalf of Dennis Dalessandro <dennis.dalessandro@xxxxxxxxx> Sent: Wednesday, December 5, 2018 5:40 PM To: Pressman, Gal; Doug Ledford; Jason Gunthorpe Cc: Matushevsky, Alexander; Leybovich, Yossi; linux-rdma@xxxxxxxxxxxxxxx; Tom Tucker Subject: Re: [PATCH rdma-next 02/13] RDMA/efa: Add EFA device definitions On 12/4/2018 7:04 AM, Gal Pressman wrote: > EFA PCIe device implements a single Admin Queue (AQ) and Admin Completion Queue > (ACQ) pair to initialize and communicate configuration with the device (similar > to NVMe and ENA network device). > Through this pair, we run SET/GET commands for querying and configuring the > device, CREATE/MODIFY/DESTROY queues, and IB specific commands like Address > Handler (AH), Memory Registration (MR) and Protection Domains (PD). So this AQ is not like QP0 in IB/OPA then? You send commands to this queue, including IB commands, and don't use the IOCTL interface. Why wouldn't you want to leverage the existing way? -Denny
