From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-11.5 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_PATCH,MAILING_LIST_MULTI,SIGNED_OFF_BY, SPF_HELO_NONE,SPF_PASS,USER_AGENT_SANE_2 autolearn=unavailable autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 1952AC433E1 for ; Mon, 27 Jul 2020 17:16:07 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id EDE1D206E7 for ; Mon, 27 Jul 2020 17:16:06 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1728692AbgG0RQG convert rfc822-to-8bit (ORCPT ); Mon, 27 Jul 2020 13:16:06 -0400 Received: from lhrrgout.huawei.com ([185.176.76.210]:2540 "EHLO huawei.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1728021AbgG0RQG (ORCPT ); Mon, 27 Jul 2020 13:16:06 -0400 Received: from lhreml710-chm.china.huawei.com (unknown [172.18.7.106]) by Forcepoint Email with ESMTP id 8D3ED8F437775346153B; Mon, 27 Jul 2020 18:16:04 +0100 (IST) Received: from localhost (10.52.121.176) by lhreml710-chm.china.huawei.com (10.201.108.61) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA256) id 15.1.1913.5; Mon, 27 Jul 2020 18:16:04 +0100 Date: Mon, 27 Jul 2020 18:14:40 +0100 From: Jonathan Cameron To: Sean V Kelley CC: , , , "Raj, Ashok" , , , , Qiuxu Zhuo Subject: Re: [RFC PATCH 7/9] PCI/AER: Add RCEC AER handling Message-ID: <20200727181440.0000614a@Huawei.com> In-Reply-To: <6C5C96C5-0365-48A0-B623-1A4C0CE0D13E@intel.com> References: <20200724172223.145608-1-sean.v.kelley@intel.com> <20200724172223.145608-8-sean.v.kelley@intel.com> <20200727132252.0000644c@Huawei.com> <6C5C96C5-0365-48A0-B623-1A4C0CE0D13E@intel.com> Organization: Huawei Technologies Research and Development (UK) Ltd. X-Mailer: Claws Mail 3.17.4 (GTK+ 2.24.32; i686-w64-mingw32) MIME-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 8BIT X-Originating-IP: [10.52.121.176] X-ClientProxiedBy: lhreml704-chm.china.huawei.com (10.201.108.53) To lhreml710-chm.china.huawei.com (10.201.108.61) X-CFilter-Loop: Reflected Sender: linux-pci-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-pci@vger.kernel.org On Mon, 27 Jul 2020 08:19:39 -0700 Sean V Kelley wrote: > On 27 Jul 2020, at 5:22, Jonathan Cameron wrote: > > > On Fri, 24 Jul 2020 10:22:21 -0700 > > Sean V Kelley wrote: > > > >> The Root Complex Event Collectors(RCEC) appear as peers to Root Ports > >> and also have the AER capability. So add RCEC support to the current > >> AER > >> service driver and attach the AER service driver to the RCEC device. > >> > >> Co-developed-by: Qiuxu Zhuo > >> Signed-off-by: Sean V Kelley > >> Signed-off-by: Qiuxu Zhuo > > > > A few questions and comments for this patch. > > > > See inline. > > > > Jonathan > > > > > >> --- > >> drivers/pci/pcie/aer.c | 34 +++++++++++++++++++++++++++------- > >> 1 file changed, 27 insertions(+), 7 deletions(-) > >> > >> diff --git a/drivers/pci/pcie/aer.c b/drivers/pci/pcie/aer.c > >> index f1bf06be449e..7cc430c74c46 100644 > >> --- a/drivers/pci/pcie/aer.c > >> +++ b/drivers/pci/pcie/aer.c > >> @@ -303,7 +303,7 @@ int pci_aer_raw_clear_status(struct pci_dev *dev) > >> return -EIO; > >> > >> port_type = pci_pcie_type(dev); > >> - if (port_type == PCI_EXP_TYPE_ROOT_PORT) { > >> + if (port_type == PCI_EXP_TYPE_ROOT_PORT || port_type == > >> PCI_EXP_TYPE_RC_EC) { > >> pci_read_config_dword(dev, aer + PCI_ERR_ROOT_STATUS, &status); > >> pci_write_config_dword(dev, aer + PCI_ERR_ROOT_STATUS, status); > >> } > >> @@ -389,6 +389,12 @@ void pci_aer_init(struct pci_dev *dev) > >> pci_add_ext_cap_save_buffer(dev, PCI_EXT_CAP_ID_ERR, sizeof(u32) * > >> n); > >> > >> pci_aer_clear_status(dev); > >> + > >> + if (pci_pcie_type(dev) == PCI_EXP_TYPE_RC_EC) { > >> + if (!pci_find_ext_capability(dev, PCI_EXT_CAP_ID_RCEC)) > >> + return; > >> + pci_info(dev, "AER: RCEC CAP FOUND and cap_has_rtctl = %d\n", n); > > > > It feels like failing to find an RC_EC extended cap in a RCEC deserved > > a nice strong error message. Perhaps this isn't the right place to do > > it > > though. For that matter, why are we checking for it here? > > Sorry, I’ve left an in-development output in the code. Will replace > with a check with more meaningful output elsewhere. > > > > >> + } > >> } > >> > >> void pci_aer_exit(struct pci_dev *dev) > >> @@ -577,7 +583,8 @@ static umode_t aer_stats_attrs_are_visible(struct > >> kobject *kobj, > >> if ((a == &dev_attr_aer_rootport_total_err_cor.attr || > >> a == &dev_attr_aer_rootport_total_err_fatal.attr || > >> a == &dev_attr_aer_rootport_total_err_nonfatal.attr) && > > > > It is a bit ugly to have these called rootport_total_err etc for the > > rcec. > > Perhaps we should just add additional attributes to reflect we are > > looking at > > an RCEC? > > I was trying to avoid any renaming to reduce churn as I did with my > first patch for ACPI / CLX_OSC support. > Will take a look. > > > > >> - pci_pcie_type(pdev) != PCI_EXP_TYPE_ROOT_PORT) > >> + ((pci_pcie_type(pdev) != PCI_EXP_TYPE_ROOT_PORT) && > >> + (pci_pcie_type(pdev) != PCI_EXP_TYPE_RC_EC))) > >> return 0; > >> > >> return a->mode; > >> @@ -894,7 +901,10 @@ static bool find_source_device(struct pci_dev > >> *parent, > >> if (result) > >> return true; > >> > >> - pci_walk_bus(parent->subordinate, find_device_iter, e_info); > >> + if (pci_pcie_type(parent) == PCI_EXP_TYPE_RC_EC) > >> + pcie_walk_rcec(parent, find_device_iter, e_info); > >> + else > >> + pci_walk_bus(parent->subordinate, find_device_iter, e_info); > >> > >> if (!e_info->error_dev_num) { > >> pci_info(parent, "can't find device of ID%04x\n", e_info->id); > >> @@ -1030,6 +1040,7 @@ int aer_get_device_error_info(struct pci_dev > >> *dev, struct aer_err_info *info) > >> if (!(info->status & ~info->mask)) > >> return 0; > >> } else if (pci_pcie_type(dev) == PCI_EXP_TYPE_ROOT_PORT || > >> + pci_pcie_type(dev) == PCI_EXP_TYPE_RC_EC || > >> pci_pcie_type(dev) == PCI_EXP_TYPE_DOWNSTREAM || > >> info->severity == AER_NONFATAL) { > >> > >> @@ -1182,6 +1193,8 @@ static int set_device_error_reporting(struct > >> pci_dev *dev, void *data) > >> int type = pci_pcie_type(dev); > >> > >> if ((type == PCI_EXP_TYPE_ROOT_PORT) || > >> + (type == PCI_EXP_TYPE_RC_EC) || > >> + (type == PCI_EXP_TYPE_RC_END) || > > > > Why add RC_END here? > > I’m not clear on your question. Errors can come from RCEC or RCiEPs. > We still need to enable reporting by the RCiEPs. I was curious to see that we need it in this code path for an RCiEP but not for a normal EP. From a quick glance it looks like that is often done in the drivers for the EPs themselves rather than here. > > > > >> (type == PCI_EXP_TYPE_UPSTREAM) || > >> (type == PCI_EXP_TYPE_DOWNSTREAM)) { > >> if (enable) > >> @@ -1206,9 +1219,11 @@ static void > >> set_downstream_devices_error_reporting(struct pci_dev *dev, > >> { > >> set_device_error_reporting(dev, &enable); > >> > >> - if (!dev->subordinate) > >> - return; > >> - pci_walk_bus(dev->subordinate, set_device_error_reporting, > >> &enable); > >> + if (pci_pcie_type(dev) == PCI_EXP_TYPE_RC_EC) > >> + pcie_walk_rcec(dev, set_device_error_reporting, &enable); > >> + else if (dev->subordinate) > >> + pci_walk_bus(dev->subordinate, set_device_error_reporting, > >> &enable); > >> + > >> } > >> > >> /** > >> @@ -1306,6 +1321,11 @@ static int aer_probe(struct pcie_device *dev) > >> struct device *device = &dev->device; > >> struct pci_dev *port = dev->port; > >> > >> + /* Limit to Root Ports or Root Complex Event Collectors */ > >> + if ((pci_pcie_type(port) != PCI_EXP_TYPE_RC_EC) && > >> + (pci_pcie_type(port) != PCI_EXP_TYPE_ROOT_PORT)) > >> + return -ENODEV; > >> + > >> rpc = devm_kzalloc(device, sizeof(struct aer_rpc), GFP_KERNEL); > >> if (!rpc) > >> return -ENOMEM; > >> @@ -1362,7 +1382,7 @@ static pci_ers_result_t aer_root_reset(struct > >> pci_dev *dev) > >> > >> static struct pcie_port_service_driver aerdriver = { > >> .name = "aer", > >> - .port_type = PCI_EXP_TYPE_ROOT_PORT, > >> + .port_type = PCIE_ANY_PORT, > > > > Why this particular change? Seems that is a lot wider than simply > > adding RCEC. Obviously we'll then drop out in the aer_probe but it > > is still rather inelegant. > > In order to extend the service drivers to non-root-port devices (i.e., > RCEC), the simple path appeared to only require setting the type to > ANY_PORT and catching the needed types arriving in the probe. Would you > prefer extending to a type2? I’m not sure how one is more elegant > than another but open to that approach. However, this seems to require > less code perhaps and seems consistent with most ‘drop-out’ > conditional patterns in the kernel. The same applies to pme. I'd miss understood this bit. It's fine as you have it here. Jonathan > > Thanks, > > Sean > > > > > >> .service = PCIE_PORT_SERVICE_AER, > >> > >> .probe = aer_probe,