From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mout.kundenserver.de ([212.227.126.187]:56726 "EHLO mout.kundenserver.de" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751118AbaIDN5j (ORCPT ); Thu, 4 Sep 2014 09:57:39 -0400 From: Arnd Bergmann To: linux-arm-kernel@lists.infradead.org Cc: Minghuan Lian , Hu Mingkai-B21284 , Zang Roy-R61911 , linux-pci@vger.kernel.org Subject: Re: [PATCH 2/2] PCI: Layerscape: Add Layerscape PCIe driver Date: Thu, 04 Sep 2014 15:57:35 +0200 Message-ID: <7601888.KPBqiDOLoa@wuerfel> In-Reply-To: <3055252.byHHPbPjpA@wuerfel> References: <1409856338-1730-1-git-send-email-Minghuan.Lian@freescale.com> <2799937.bNtU9KXR6t@wuerfel> <3055252.byHHPbPjpA@wuerfel> MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Sender: linux-pci-owner@vger.kernel.org List-ID: On Thursday 04 September 2014 15:51:22 Arnd Bergmann wrote: > On Thursday 04 September 2014 14:14:48 Arnd Bergmann wrote: > > > > My impression is that you have two distinct MSI controller > > implementations, one for LS1021A and the other one for everything > > else. How about using separate pcie_host_ops for them, possibly > > also moving them into separate files? > > > > One more thing: you should really use the msi-parent property to > find the MSI controller: Sooner or later this PCI block is likely > to end up in a real product that has GICv2m or GICv3 support, so > you will have to pick which of the two MSI controllers to use. Ah, I missed the fact that LS1021A is a real product already, I was confusing it with the LS2085A patches that are also on the list at the moment and that only work in a simulator at present. Note that LS2085A does have GICv3. I don't know if it has a similar PCIe implementation, but if it does, the DT representation should definitely provide a way to pick the MSI controller (you will always want the GIC in practice). Arnd From mboxrd@z Thu Jan 1 00:00:00 1970 From: arnd@arndb.de (Arnd Bergmann) Date: Thu, 04 Sep 2014 15:57:35 +0200 Subject: [PATCH 2/2] PCI: Layerscape: Add Layerscape PCIe driver In-Reply-To: <3055252.byHHPbPjpA@wuerfel> References: <1409856338-1730-1-git-send-email-Minghuan.Lian@freescale.com> <2799937.bNtU9KXR6t@wuerfel> <3055252.byHHPbPjpA@wuerfel> Message-ID: <7601888.KPBqiDOLoa@wuerfel> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Thursday 04 September 2014 15:51:22 Arnd Bergmann wrote: > On Thursday 04 September 2014 14:14:48 Arnd Bergmann wrote: > > > > My impression is that you have two distinct MSI controller > > implementations, one for LS1021A and the other one for everything > > else. How about using separate pcie_host_ops for them, possibly > > also moving them into separate files? > > > > One more thing: you should really use the msi-parent property to > find the MSI controller: Sooner or later this PCI block is likely > to end up in a real product that has GICv2m or GICv3 support, so > you will have to pick which of the two MSI controllers to use. Ah, I missed the fact that LS1021A is a real product already, I was confusing it with the LS2085A patches that are also on the list at the moment and that only work in a simulator at present. Note that LS2085A does have GICv3. I don't know if it has a similar PCIe implementation, but if it does, the DT representation should definitely provide a way to pick the MSI controller (you will always want the GIC in practice). Arnd