From patchwork Mon Jan 13 22:26:28 2014 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Keith Busch X-Patchwork-Id: 310391 Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Received: from vger.kernel.org (vger.kernel.org [209.132.180.67]) by ozlabs.org (Postfix) with ESMTP id 5CC5B2C0079 for ; Tue, 14 Jan 2014 09:26:59 +1100 (EST) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751762AbaAMW0z (ORCPT ); Mon, 13 Jan 2014 17:26:55 -0500 Received: from mga11.intel.com ([192.55.52.93]:58033 "EHLO mga11.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751832AbaAMW0e (ORCPT ); Mon, 13 Jan 2014 17:26:34 -0500 Received: from fmsmga002.fm.intel.com ([10.253.24.26]) by fmsmga102.fm.intel.com with ESMTP; 13 Jan 2014 14:26:34 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.95,655,1384329600"; d="scan'208";a="464542477" Received: from dcgshare.lm.intel.com ([10.232.118.254]) by fmsmga002.fm.intel.com with ESMTP; 13 Jan 2014 14:26:33 -0800 Received: by dcgshare.lm.intel.com (Postfix, from userid 1017) id 4937BE012E; Mon, 13 Jan 2014 15:26:33 -0700 (MST) From: Keith Busch To: linux-pci@vger.kernel.org, alex.williamson@redhat.com, bhelgaas@google.com Cc: Keith Busch Subject: [PATCHv2 2/2] NVMe: Implement pci reset callback Date: Mon, 13 Jan 2014 15:26:28 -0700 Message-Id: <1389651988-27392-3-git-send-email-keith.busch@intel.com> X-Mailer: git-send-email 1.7.1 In-Reply-To: <1389651988-27392-1-git-send-email-keith.busch@intel.com> References: <1389651988-27392-1-git-send-email-keith.busch@intel.com> Sender: linux-pci-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-pci@vger.kernel.org Initiates nvme controller reset sequence when pci layer notifies of slot reset. The driver's slot_reset handling is skipped if the nvme polling kthread happened to try accessing a device register while the reset was in progress, which would have scheduled the nvme device to be reset anyway. Signed-off-by: Keith Busch --- drivers/block/nvme-core.c | 36 ++++++++++++++++++++++++++++++++++-- include/linux/nvme.h | 1 + 2 files changed, 35 insertions(+), 2 deletions(-) diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c index b59a93a..1f5c18a 100644 --- a/drivers/block/nvme-core.c +++ b/drivers/block/nvme-core.c @@ -1750,6 +1750,7 @@ static int nvme_kthread(void *data) dev->initialized) { if (work_busy(&dev->reset_work)) continue; + dev->reset = 1; list_del_init(&dev->node); dev_warn(&dev->pci_dev->dev, "Failed status, reset controller\n"); @@ -2088,6 +2089,7 @@ static int nvme_dev_map(struct nvme_dev *dev) } dev->db_stride = 1 << NVME_CAP_STRIDE(readq(&dev->bar->cap)); dev->dbs = ((void __iomem *)dev->bar) + 4096; + dev->reset = 0; return 0; @@ -2265,7 +2267,10 @@ static void nvme_dev_shutdown(struct nvme_dev *dev) list_del_init(&dev->node); spin_unlock(&dev_list_lock); - if (!dev->bar || (dev->bar && readl(&dev->bar->csts) == -1)) { + if (!dev->bar || (dev->bar && readl(&dev->bar->csts) == -1) || + dev->reset) { + if (dev->bar) + nvme_shutdown_ctrl(dev); for (i = dev->queue_count - 1; i >= 0; i--) { struct nvme_queue *nvmeq = dev->queues[i]; nvme_suspend_queue(nvmeq); @@ -2575,7 +2580,6 @@ static void nvme_remove(struct pci_dev *pdev) #define nvme_error_detected NULL #define nvme_dump_registers NULL #define nvme_link_reset NULL -#define nvme_slot_reset NULL #define nvme_error_resume NULL static int nvme_suspend(struct device *dev) @@ -2599,6 +2603,34 @@ static int nvme_resume(struct device *dev) return 0; } + +/** + * nvme_slot_reset - handle nvme device that has been reset + * @pdev: pci device that has been through a pci slot or function reset + * + * Called by pci error recovery. We assume the device is in a state requiring + * initialization, so set the device reset flag first to prevent wasting time + * taking the controller offline with admin commands that we expect to timeout. + */ +static pci_ers_result_t nvme_slot_reset(struct pci_dev *pdev) +{ + struct nvme_dev *dev = pci_get_drvdata(pdev); + + spin_lock(&dev_list_lock); + if (!work_busy(&dev->reset_work)) { + list_del_init(&dev->node); + dev->reset = 1; + } else { + spin_unlock(&dev_list_lock); + return PCI_ERS_RESULT_RECOVERED; + } + spin_unlock(&dev_list_lock); + + nvme_dev_reset(dev); + return dev->initialized ? PCI_ERS_RESULT_RECOVERED : + PCI_ERS_RESULT_DISCONNECT; +} + static SIMPLE_DEV_PM_OPS(nvme_dev_pm_ops, nvme_suspend, nvme_resume); static const struct pci_error_handlers nvme_err_handler = { diff --git a/include/linux/nvme.h b/include/linux/nvme.h index 69ae03f..febce6f 100644 --- a/include/linux/nvme.h +++ b/include/linux/nvme.h @@ -97,6 +97,7 @@ struct nvme_dev { u16 oncs; u16 abort_limit; u8 initialized; + u8 reset; }; /*