Return-Path: From: =?iso-8859-1?q?Jos=E9_Antonio_Santos_Cadenas?= Reply-To: jcaden@libresoft.es To: Elvis =?iso-8859-1?q?Pf=FCtzenreuter?= Subject: Re: HDP proposed API Date: Wed, 5 May 2010 15:36:40 +0200 Cc: "Santiago Carot-Nemesio" , linux-bluetooth@vger.kernel.org References: <1272959741.2182.123.camel@mosquito> <201005051518.45474.jcaden@libresoft.es> In-Reply-To: MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-1" Message-Id: <201005051536.40471.jcaden@libresoft.es> Sender: linux-bluetooth-owner@vger.kernel.org List-ID: El Wednesday 05 May 2010 15:22:16 Elvis Pf?tzenreuter escribi?: > > So you mean do fd-passing with the pipe file descriptor to the client and then HDP will write the data in the L2CAP socket using splice, am I right? > > I guess they meant using splice for the pipe (because as far as I understood, it is for pipes only), so the data exchanged between bluez and application is zero-copy.-- I'm reading splice documentation and I understand that can be used with any type of fd, but one of the fd should be a pipe. This is ok for that case. I see one problem. The pipes are unidirectional so for a full duplex data channel HDP will need two pipes (what means 4 fd and pass 2 to the client, one for read and other for write). We should find an easy way to solve this. Proposals are welcome :)