From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from moutng.kundenserver.de ([212.227.17.10]:62655 "EHLO moutng.kundenserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1757806Ab3CYJaE (ORCPT ); Mon, 25 Mar 2013 05:30:04 -0400 From: Arnd Bergmann To: Thierry Reding Subject: Re: [RFC 0/2] PCI: Introduce MSI chip infrastructure Date: Mon, 25 Mar 2013 09:29:57 +0000 Cc: Thomas Petazzoni , Andrew Murray , Bjorn Helgaas , "linux-pci@vger.kernel.org" , "linux-kernel@vger.kernel.org" References: <1363942307-9327-1-git-send-email-thierry.reding@avionic-design.de> <20130325093847.7474dc0e@skate> <20130325091534.GA21054@avionic-0098.mockup.avionic-design.de> In-Reply-To: <20130325091534.GA21054@avionic-0098.mockup.avionic-design.de> MIME-Version: 1.0 Content-Type: Text/Plain; charset="iso-8859-15" Message-Id: <201303250929.58010.arnd@arndb.de> Sender: linux-pci-owner@vger.kernel.org List-ID: On Monday 25 March 2013, Thierry Reding wrote: > I think you can just make this: > > mpic: interrupt-controller@d0020000 { > ... > }; > > ... > > soc { > pcie-controller { > marvell,msi = <&mpic>; > }; > }; > > And everything else should just work given the APIs I mentioned. But as > you said it'd be good if somebody else could share their opinion about > this. I think the property referring to the msi controller should have a fixed name, such as "msi-parent", to go along with "interrupt-parent". Similarly, I would suggest using an empty "msi-controller" property to mark the controller that is capable of serving MSIs. The Linux implementation doesn't currently require the "interrupt-controller" property, but I think it's good to stay close to the original interrupt binding here for consistency. Arnd