From mboxrd@z Thu Jan 1 00:00:00 1970 From: Jason Gunthorpe Subject: Re: [ANNOUNCE] Shared Topic Branch For ABI Changes Date: Thu, 21 Apr 2016 14:01:02 -0600 Message-ID: <20160421200102.GB6258@obsidianresearch.com> References: <20160420055221.GT6349@leon.nu> <20160420152924.GA25424@obsidianresearch.com> <20160421065847.GC12730@leon.nu> <20160421173815.GB5102@obsidianresearch.com> <1828884A29C6694DAF28B7E6B8A82373AB04435E@ORSMSX109.amr.corp.intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <1828884A29C6694DAF28B7E6B8A82373AB04435E-P5GAC/sN6hkd3b2yrw5b5LfspsVTdybXVpNB7YpNyf8@public.gmane.org> Sender: linux-rdma-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org To: "Hefty, Sean" Cc: "leon-DgEjT+Ai2ygdnm+yROfE0A@public.gmane.org" , "Weiny, Ira" , RDMA mailing list , Doug Ledford List-Id: linux-rdma@vger.kernel.org On Thu, Apr 21, 2016 at 06:17:18PM +0000, Hefty, Sean wrote: > > Doug was ambivalent, Steve was against. There was a concern the natual > > release schedual of the parts was different, so why link them. > > > > My counter was that isn't really important, the goal is is to reduce > > developer time when working on this mess. > > > > Since we are looking at changing *everything* developer time seems > > more important right now. > > FWIW, I moved the librdmacm and libibacm into a single guthub > project. Even if we don't share a single git tree, this enables > multiple maintainers for the code. I would really like to see > libibcm dropped entirely, and think libibverbs and librdmacm should > merge. +1 > In the past I would have been in favor of more closely associating > libibumad with libibverbs, but now that OPA uses IB-like management, > I'm not so sure. I think OPA management could be done outside of > any verbs emulation. Ira might know for sure. IIRC OPA still uses /dev/umad. I suspect libmad/umad needs to grow new structures and constants primarily. I would call the combined repo 'rdma-plumbing' or something, intended to cover the user space side for all the kernel interfaces. I saw libfabric was having success with this model - is that accurate? 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