From patchwork Sun Mar 11 07:28:35 2012 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Dan Williams X-Patchwork-Id: 145955 X-Patchwork-Delegate: davem@davemloft.net 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 21E2EB6EF3 for ; Sun, 11 Mar 2012 18:13:21 +1100 (EST) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1751631Ab2CKHNT (ORCPT ); Sun, 11 Mar 2012 03:13:19 -0400 Received: from mga09.intel.com ([134.134.136.24]:46364 "EHLO mga09.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751630Ab2CKHNT (ORCPT ); Sun, 11 Mar 2012 03:13:19 -0400 Received: from orsmga002.jf.intel.com ([10.7.209.21]) by orsmga102.jf.intel.com with ESMTP; 10 Mar 2012 23:13:18 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.67,351,1309762800"; d="scan'208";a="119826701" Received: from dwillia2-linux.jf.intel.com ([10.23.45.110]) by orsmga002.jf.intel.com with ESMTP; 10 Mar 2012 23:13:18 -0800 Received: from dwillia2-linux.jf.intel.com (localhost.localdomain [IPv6:::1]) by dwillia2-linux.jf.intel.com (Postfix) with ESMTP id C7964800DF; Sat, 10 Mar 2012 23:28:35 -0800 (PST) Subject: [isci PATCH v2 10/18] isci: fix interrupt disable To: linux-scsi@vger.kernel.org From: Dan Williams Cc: linux-ide@vger.kernel.org, Richard Boyd , Jacek Danecki Date: Sat, 10 Mar 2012 23:28:35 -0800 Message-ID: <20120311072835.6320.23343.stgit@dwillia2-linux.jf.intel.com> In-Reply-To: <20120311072518.6320.61717.stgit@dwillia2-linux.jf.intel.com> References: <20120311072518.6320.61717.stgit@dwillia2-linux.jf.intel.com> User-Agent: StGit/0.16-1-g7004 MIME-Version: 1.0 Sender: linux-ide-owner@vger.kernel.org Precedence: bulk List-ID: X-Mailing-List: linux-ide@vger.kernel.org There is a (dubious?) lost irq workaround in sci_controller_isr() that effectively nullifies attempts to disable interrupts. Until the workaround can be re-evaluated add some infrastructure to prevent the interrupt handler from inadvertantly re-enabling interrupts. The failure mode was interrupts continuing to run after the driver had been removed and its iomappings torn down. Reported-by: Jacek Danecki Tested-by: Jacek Danecki [richard: clear remaining interrupts at the end of reset] Acked-by: Richard Boyd Signed-off-by: Dan Williams --- drivers/scsi/isci/host.c | 39 ++++++++++++++++++++++++++------------- drivers/scsi/isci/host.h | 1 + 2 files changed, 27 insertions(+), 13 deletions(-) -- To unsubscribe from this list: send the line "unsubscribe linux-ide" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html diff --git a/drivers/scsi/isci/host.c b/drivers/scsi/isci/host.c index 577a836..5832b13 100644 --- a/drivers/scsi/isci/host.c +++ b/drivers/scsi/isci/host.c @@ -192,22 +192,27 @@ static bool sci_controller_completion_queue_has_entries(struct isci_host *ihost) static bool sci_controller_isr(struct isci_host *ihost) { - if (sci_controller_completion_queue_has_entries(ihost)) { + if (sci_controller_completion_queue_has_entries(ihost)) return true; - } else { - /* - * we have a spurious interrupt it could be that we have already - * emptied the completion queue from a previous interrupt */ - writel(SMU_ISR_COMPLETION, &ihost->smu_registers->interrupt_status); - /* - * There is a race in the hardware that could cause us not to be notified - * of an interrupt completion if we do not take this step. We will mask - * then unmask the interrupts so if there is another interrupt pending - * the clearing of the interrupt source we get the next interrupt message. */ + /* we have a spurious interrupt it could be that we have already + * emptied the completion queue from a previous interrupt + * FIXME: really!? + */ + writel(SMU_ISR_COMPLETION, &ihost->smu_registers->interrupt_status); + + /* There is a race in the hardware that could cause us not to be + * notified of an interrupt completion if we do not take this + * step. We will mask then unmask the interrupts so if there is + * another interrupt pending the clearing of the interrupt + * source we get the next interrupt message. + */ + spin_lock(&ihost->scic_lock); + if (test_bit(IHOST_IRQ_ENABLED, &ihost->flags)) { writel(0xFF000000, &ihost->smu_registers->interrupt_mask); writel(0, &ihost->smu_registers->interrupt_mask); } + spin_unlock(&ihost->scic_lock); return false; } @@ -698,14 +703,15 @@ static u32 sci_controller_get_suggested_start_timeout(struct isci_host *ihost) static void sci_controller_enable_interrupts(struct isci_host *ihost) { - BUG_ON(ihost->smu_registers == NULL); + set_bit(IHOST_IRQ_ENABLED, &ihost->flags); writel(0, &ihost->smu_registers->interrupt_mask); } void sci_controller_disable_interrupts(struct isci_host *ihost) { - BUG_ON(ihost->smu_registers == NULL); + clear_bit(IHOST_IRQ_ENABLED, &ihost->flags); writel(0xffffffff, &ihost->smu_registers->interrupt_mask); + readl(&ihost->smu_registers->interrupt_mask); /* flush */ } static void sci_controller_enable_port_task_scheduler(struct isci_host *ihost) @@ -1318,7 +1324,9 @@ void isci_host_deinit(struct isci_host *ihost) */ writel(0, &ihost->scu_registers->peg0.sgpio.interface_control); + spin_lock_irq(&ihost->scic_lock); sci_controller_reset(ihost); + spin_unlock_irq(&ihost->scic_lock); /* Cancel any/all outstanding port timers */ for (i = 0; i < ihost->logical_port_entries; i++) { @@ -1605,6 +1613,9 @@ static void sci_controller_reset_hardware(struct isci_host *ihost) /* The write to the UFQGP clears the UFQPR */ writel(0, &ihost->scu_registers->sdma.unsolicited_frame_get_pointer); + + /* clear all interrupts */ + writel(~SMU_INTERRUPT_STATUS_RESERVED_MASK, &ihost->smu_registers->interrupt_status); } static void sci_controller_resetting_state_enter(struct sci_base_state_machine *sm) @@ -2391,7 +2402,9 @@ int isci_host_init(struct isci_host *ihost) int i, err; enum sci_status status; + spin_lock_irq(&ihost->scic_lock); status = sci_controller_construct(ihost, scu_base(ihost), smu_base(ihost)); + spin_unlock_irq(&ihost->scic_lock); if (status != SCI_SUCCESS) { dev_err(&ihost->pdev->dev, "%s: sci_controller_construct failed - status = %x\n", diff --git a/drivers/scsi/isci/host.h b/drivers/scsi/isci/host.h index 9dc910b..9701c1d 100644 --- a/drivers/scsi/isci/host.h +++ b/drivers/scsi/isci/host.h @@ -200,6 +200,7 @@ struct isci_host { struct pci_dev *pdev; #define IHOST_START_PENDING 0 #define IHOST_STOP_PENDING 1 + #define IHOST_IRQ_ENABLED 2 unsigned long flags; wait_queue_head_t eventq; struct Scsi_Host *shost;