From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Gunthorpe Subject: Re: [RFC ABI 8/8] IB/core: Implement device_create with the new ABI Date: Thu, 26 May 2016 11:30:07 -0600 Message-ID: <20160526173007.GD27115@obsidianresearch.com> References: <1464100526-31730-1-git-send-email-leonro@mellanox.com> <1464100526-31730-9-git-send-email-leonro@mellanox.com> <20160524220720.GF7950@obsidianresearch.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: Matan Barak Cc: Leon Romanovsky , Doug Ledford , linux-rdma , Matan Barak , Haggai Eran List-Id: linux-rdma@vger.kernel.org On Thu, May 26, 2016 at 04:24:06PM +0300, Matan Barak wrote: > >> +enum ibnl_create_device { > >> + IBNL_CREATE_DEVICE_CORE = IBNL_VENDOR_ATTRS_MAX, > >> + IBNL_CREATE_DEVICE_MAX > >> +}; > > > > Whats all this about? Looks very strange. > > The first attribute ids of every command are reserved for the vendor part. I'd still like to see if we can be happy with globally unique attribute ID's. I would like to see each driver have it's own range of global IDs that it can use. Much like an ioctl number each attribute id should be associated with exactly one structure layout. The obvious downside is how to index them as large ID's cannot be array'd anymore, but I wonder if that approach is overkill anyhow, if there is only going to be one or two arguments then a linear search is OK, or we can require them to be sorted by ID or something. Jason -- To unsubscribe from this list: send the line "unsubscribe linux-rdma" in the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org More majordomo info at http://vger.kernel.org/majordomo-info.html