From patchwork Tue Jun 9 07:22:44 2009 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Yanmin Zhang X-Patchwork-Id: 28874 Received: from vger.kernel.org (vger.kernel.org [209.132.176.167]) by demeter.kernel.org (8.14.2/8.14.2) with ESMTP id n597NTXq007850 for ; Tue, 9 Jun 2009 07:23:29 GMT Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1756995AbZFIHWi (ORCPT ); Tue, 9 Jun 2009 03:22:38 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1757672AbZFIHWi (ORCPT ); Tue, 9 Jun 2009 03:22:38 -0400 Received: from mga12.intel.com ([143.182.124.36]:42181 "EHLO azsmga102.ch.intel.com" rhost-flags-OK-OK-OK-FAIL) by vger.kernel.org with ESMTP id S1755570AbZFIHWh (ORCPT ); Tue, 9 Jun 2009 03:22:37 -0400 Received: from azsmga001.ch.intel.com ([10.2.17.19]) by azsmga102.ch.intel.com with ESMTP; 09 Jun 2009 00:22:39 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.41,330,1241420400"; d="scan'208";a="152146120" Received: from ymzhang.sh.intel.com (HELO [10.239.13.146]) ([10.239.13.146]) by azsmga001.ch.intel.com with ESMTP; 09 Jun 2009 00:22:38 -0700 Subject: [PATCH V3: 3/3] pci: Provide Multiple Error Received support on AER From: "Zhang, Yanmin" To: linux-kernel@vger.kernel.org, linux-pci@vger.kernel.org Cc: Jesse Barnes Date: Tue, 09 Jun 2009 15:22:44 +0800 Message-Id: <1244532164.2560.221.camel@ymzhang> Mime-Version: 1.0 X-Mailer: Evolution 2.22.1 (2.22.1-2.fc9) Sender: linux-pci-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-pci@vger.kernel.org When a root port receive the same errors more than once before kernel process them, the Multiple Error Messages Received flags are set by hardware. Because root port could only save one kind of correctable error source id and another uncorrectable error source id at the same time, so the second message sender id is lost if the 2 messages are sent from 2 different devices. Below patch searches all devices under the root port when multiple messages are received. Signed-off-by: Zhang Yanmin --- -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html --- linux-2.6.30-rc3_aernoeid/drivers/pci/pcie/aer/aerdrv_core.c 2009-06-09 01:36:52.000000000 +0800 +++ linux-2.6.30-rc3_aermultierror/drivers/pci/pcie/aer/aerdrv_core.c 2009-06-09 02:04:41.000000000 +0800 @@ -143,13 +143,22 @@ static void set_downstream_devices_error pci_walk_bus(dev->subordinate, set_device_error_reporting, &enable); } +static int add_error_device(struct aer_err_info *e_info, struct pci_dev *dev) +{ + if (e_info->error_dev_num < AER_MAX_MULTI_ERR_DEVICES) { + e_info->dev[e_info->error_dev_num ++] = dev; + return 1; + } else + return 0; +} + static int compare_device_id(struct pci_dev *dev, struct aer_err_info *e_info) { if (e_info->id == ((dev->bus->number << 8) | dev->devfn)) { /* * Device ID match */ - e_info->dev = dev; + add_error_device(e_info, dev); return 1; } @@ -164,20 +173,38 @@ static int find_device_iter(struct pci_d u32 status; u32 mask; u16 reg16; + int result; struct aer_err_info *e_info = (struct aer_err_info *)data; /* * When bus id is equal to 0, it might be a bad id * reported by root port. */ - if (!nosourceid && (PCI_BUS(e_info->id) != 0)) - return compare_device_id(dev, e_info); + if (!nosourceid && (PCI_BUS(e_info->id) != 0)) { + result = compare_device_id(dev, e_info); + /* + * If there is no multiple error, we stop + * or continue based on the id comparing. + */ + if (!(e_info->flags & AER_MULTI_ERROR_VALID_FLAG)) + return result; + + /* + * If there are multiple errors and id does match, + * We need continue to search other devices under + * the root port. Return 0 means that. + */ + if (result) + return 0; + } /* - * Next is to check when bus id is equal to 0 or - * nosourceid==y. Some ports might lose the bus - * id of error source id. We check AER status - * registers to find the initial reporter. + * When either + * 1) nosourceid==y; + * 2) bus id is equal to 0. Some ports might lose the bus + * id of error source id; + * 3) There are multiple errors and prior id comparing fails; + * We check AER status registers to find the initial reporter. */ if (atomic_read(&dev->enable_cnt) == 0) return 0; @@ -206,8 +233,8 @@ static int find_device_iter(struct pci_d pos + PCI_ERR_COR_MASK, &mask); if (status & ERR_CORRECTABLE_ERROR_MASK & ~mask) { - e_info->dev = dev; - return 1; + add_error_device(e_info, dev); + goto added; } } else { pci_read_config_dword(dev, @@ -217,12 +244,18 @@ static int find_device_iter(struct pci_d pos + PCI_ERR_UNCOR_MASK, &mask); if (status & ERR_UNCORRECTABLE_ERROR_MASK & ~mask) { - e_info->dev = dev; - return 1; + add_error_device(e_info, dev); + goto added; } } return 0; + +added: + if (e_info->flags & AER_MULTI_ERROR_VALID_FLAG) { + return 0; + } else + return 1; } /** @@ -703,6 +736,30 @@ static int get_device_error_info(struct return AER_SUCCESS; } +static inline void aer_process_err_devices(struct pcie_device *p_device, + struct aer_err_info *e_info) +{ + int i; + + if (e_info->dev[0] == NULL) { + printk(KERN_DEBUG "%s->can't find device of ID%04x\n", + __func__, e_info->id); + } + + for (i = 0; i < e_info->error_dev_num; i ++) { + if (e_info->dev[i] == NULL) + break; + + if (get_device_error_info(e_info->dev[i], e_info) == + AER_SUCCESS) { + aer_print_error(e_info->dev[i], e_info); + handle_error_source(p_device, + e_info->dev[i], + e_info); + } + } +} + /** * aer_isr_one_error - consume an error detected by root port * @p_device: pointer to error root port service device @@ -745,18 +802,7 @@ static void aer_isr_one_error(struct pci e_info->flags |= AER_MULTI_ERROR_VALID_FLAG; find_source_device(p_device->port, e_info); - if (e_info->dev == NULL) { - printk(KERN_DEBUG "%s->can't find device of ID%04x\n", - __func__, e_info->id); - continue; - } - if (get_device_error_info(e_info->dev, e_info) == - AER_SUCCESS) { - aer_print_error(e_info->dev, e_info); - handle_error_source(p_device, - e_info->dev, - e_info); - } + aer_process_err_devices(p_device, e_info); } kfree(e_info); --- linux-2.6.30-rc3_aernoeid/drivers/pci/pcie/aer/aerdrv.h 2009-04-29 12:44:36.000000000 +0800 +++ linux-2.6.30-rc3_aermultierror/drivers/pci/pcie/aer/aerdrv.h 2009-06-09 01:43:20.000000000 +0800 @@ -56,8 +56,10 @@ struct header_log_regs { unsigned int dw3; }; +#define AER_MAX_MULTI_ERR_DEVICES 5 struct aer_err_info { - struct pci_dev *dev; + struct pci_dev *dev[AER_MAX_MULTI_ERR_DEVICES]; + int error_dev_num; u16 id; int severity; /* 0:NONFATAL | 1:FATAL | 2:COR */ int flags;