linux-scsi.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Dan Williams <dan.j.williams@intel.com>
To: linux-scsi@vger.kernel.org
Cc: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Subject: [isci-rnc PATCH v1 22/37] isci: Callbacks to libsas occur under scic_lock and are synchronized.
Date: Thu, 22 Mar 2012 17:29:47 -0700	[thread overview]
Message-ID: <20120323002947.18065.29895.stgit@dwillia2-linux.jf.intel.com> (raw)
In-Reply-To: <20120323002504.18065.45709.stgit@dwillia2-linux.jf.intel.com>

From: Jeff Skirvin <jeffrey.d.skirvin@intel.com>

This patch changes the callback mechanism to libsas to only occur while
the scic_lock is held; the abort path cleanup of I/Os also checks to make
sure IREQ_ABORT_PATH_ACTIVE is clear before proceding.

Signed-off-by: Jeff Skirvin <jeffrey.d.skirvin@intel.com>
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
 drivers/scsi/isci/host.c          |   10 +++------
 drivers/scsi/isci/remote_device.c |   43 ++++++++++++++++++++++++++++++++-----
 2 files changed, 40 insertions(+), 13 deletions(-)

diff --git a/drivers/scsi/isci/host.c b/drivers/scsi/isci/host.c
index 3cc2a23..cd01e39 100644
--- a/drivers/scsi/isci/host.c
+++ b/drivers/scsi/isci/host.c
@@ -1094,8 +1094,6 @@ void isci_host_completion_routine(unsigned long data)
 	list_splice_init(&ihost->requests_to_complete,
 			 &completed_request_list);
 
-	spin_unlock_irq(&ihost->scic_lock);
-
 	/* Process any completions in the list. */
 	list_for_each_safe(current_position, next_position,
 			   &completed_request_list) {
@@ -1104,7 +1102,6 @@ void isci_host_completion_routine(unsigned long data)
 				     completed_node);
 		task = isci_request_access_task(request);
 
-
 		/* Return the task to libsas */
 		if (task != NULL) {
 
@@ -1130,11 +1127,12 @@ void isci_host_completion_routine(unsigned long data)
 				}
 			}
 		}
+		if (test_and_clear_bit(IREQ_ABORT_PATH_ACTIVE, &request->flags))
+			wake_up_all(&ihost->eventq);
 
-		spin_lock_irq(&ihost->scic_lock);
 		isci_free_tag(ihost, request->io_tag);
-		spin_unlock_irq(&ihost->scic_lock);
 	}
+	spin_unlock_irq(&ihost->scic_lock);
 
 	/* the coalesence timeout doubles at each encoding step, so
 	 * update it based on the ilog2 value of the outstanding requests
@@ -2685,8 +2683,6 @@ enum sci_status sci_controller_complete_io(struct isci_host *ihost,
 
 		index = ISCI_TAG_TCI(ireq->io_tag);
 		clear_bit(IREQ_ACTIVE, &ireq->flags);
-		if (test_bit(IREQ_ABORT_PATH_ACTIVE, &ireq->flags))
-			wake_up_all(&ihost->eventq);
 		return SCI_SUCCESS;
 	default:
 		dev_warn(&ihost->pdev->dev, "%s invalid state: %d\n",
diff --git a/drivers/scsi/isci/remote_device.c b/drivers/scsi/isci/remote_device.c
index d1c2a22..21a9800 100644
--- a/drivers/scsi/isci/remote_device.c
+++ b/drivers/scsi/isci/remote_device.c
@@ -145,6 +145,39 @@ static bool isci_compare_suspendcount(
 	return localcount != idev->rnc.suspend_count;
 }
 
+static bool isci_check_reqterm(
+	struct isci_host *ihost,
+	struct isci_remote_device *idev,
+	struct isci_request *ireq,
+	u32 localcount)
+{
+	unsigned long flags;
+	bool res;
+
+	spin_lock_irqsave(&ihost->scic_lock, flags);
+	res = isci_compare_suspendcount(idev, localcount)
+		&& !test_bit(IREQ_ABORT_PATH_ACTIVE, &ireq->flags);
+	spin_unlock_irqrestore(&ihost->scic_lock, flags);
+
+	return res;
+}
+
+static bool isci_check_devempty(
+	struct isci_host *ihost,
+	struct isci_remote_device *idev,
+	u32 localcount)
+{
+	unsigned long flags;
+	bool res;
+
+	spin_lock_irqsave(&ihost->scic_lock, flags);
+	res = isci_compare_suspendcount(idev, localcount)
+		&& idev->started_request_count == 0;
+	spin_unlock_irqrestore(&ihost->scic_lock, flags);
+
+	return res;
+}
+
 enum sci_status isci_remote_device_terminate_requests(
 	struct isci_host *ihost,
 	struct isci_remote_device *idev,
@@ -179,17 +212,15 @@ enum sci_status isci_remote_device_terminate_requests(
 			sci_remote_device_terminate_req(ihost, idev, 0, ireq);
 			spin_unlock_irqrestore(&ihost->scic_lock, flags);
 			wait_event(ihost->eventq,
-				   (isci_compare_suspendcount(idev,
-							      rnc_suspend_count)
-				    && !test_bit(IREQ_ACTIVE, &ireq->flags)));
+				   isci_check_reqterm(ihost, idev, ireq,
+						      rnc_suspend_count));
 		} else {
 			/* Terminate all TCs. */
 			sci_remote_device_terminate_requests(idev);
 			spin_unlock_irqrestore(&ihost->scic_lock, flags);
 			wait_event(ihost->eventq,
-				   (isci_compare_suspendcount(idev,
-							      rnc_suspend_count)
-				    && idev->started_request_count == 0));
+				   isci_check_devempty(ihost, idev,
+						       rnc_suspend_count));
 		}
 		dev_dbg(&ihost->pdev->dev, "%s: idev=%p, wait done\n",
 			__func__, idev);


  parent reply	other threads:[~2012-03-23  0:14 UTC|newest]

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-03-23  0:27 [isci-rnc PATCH v1 00/37] remote node context rework Dan Williams
2012-03-23  0:27 ` [isci-rnc PATCH v1 01/37] isci: Manage the link layer hang detect timer for RNC suspensions Dan Williams
2012-03-23  0:27 ` [isci-rnc PATCH v1 02/37] isci: Fixed bug in resumption from RNC Tx/Rx suspend state Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 03/37] isci: Handle all suspending TC completions Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 04/37] isci: Terminate outstanding TCs on TX/RX RNC suspensions Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 05/37] isci: Manage device suspensions during TC terminations Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 06/37] isci: Remote device must be suspended for NCQ cleanup Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 07/37] isci: Remote device stop also suspends the RNC and terminates I/O Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 08/37] isci: Escalate to I_T_Nexus_Reset when the device is gone Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 09/37] isci: Redesign device suspension, abort, cleanup Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 10/37] isci: Add suspension cases for RNC INVALIDATING, POSTING states Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 11/37] isci: Device access in the error path does not depend on IDEV_GONE Dan Williams
2012-03-23  0:28 ` [isci-rnc PATCH v1 12/37] isci: All pending TCs are terminated when the RNC is invalidated Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 13/37] isci: Only set IDEV_GONE in the device stop path Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 14/37] isci: Remove isci_device reqs_in_process and dev_node from isci_device Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 15/37] isci: Distinguish between remote device suspension cases Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 16/37] isci: Fix the terminated I/O to not call sas_task_abort() Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 17/37] isci: Save the suspension hint for upcoming suspensions Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 18/37] isci: Manage the LLHANG timer enable/disable per-device Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 19/37] isci: Make sure all TCs are terminated and cleaned in LUN reset Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 20/37] isci: Implement waiting for suspend in the abort path Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 21/37] isci: When in the abort path, defeat other resume calls until done Dan Williams
2012-03-23  0:29 ` Dan Williams [this message]
2012-03-23  0:29 ` [isci-rnc PATCH v1 23/37] isci: Manage tag releases differently when aborting tasks Dan Williams
2012-03-23  0:29 ` [isci-rnc PATCH v1 24/37] isci: Fix RNC suspend call for SCI_RESUMING state Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 25/37] isci: Wait for RNC resumption before leaving the abort path Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 26/37] isci: Directly control IREQ_ABORT_PATH_ACTIVE when completing TMFs Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 27/37] isci: Add protocol indicator for TMF requests Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 28/37] isci: Added timeouts to RNC suspensions in the abort path Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 29/37] isci: Change the phy control and link reset interface for HW reasons Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 30/37] isci: Don't wait for an RNC suspend if it's being destroyed Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 31/37] isci: Restore the ATAPI device RNC management code Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 32/37] isci: Check IDEV_GONE before performing abort path operations Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 33/37] isci: Remove obviated host callback list Dan Williams
2012-03-23  0:30 ` [isci-rnc PATCH v1 34/37] isci: Manage the IREQ_NO_AUTO_FREE_TAG under scic_lock Dan Williams
2012-03-23  0:31 ` [isci-rnc PATCH v1 35/37] isci: Fix RNC AWAIT_SUSPENSION->INVALIDATING transition Dan Williams
2012-03-23  0:31 ` [isci-rnc PATCH v1 36/37] isci: Fixed RNC bug that lost the suspension or resumption during destroy Dan Williams
2012-03-23  0:31 ` [isci-rnc PATCH v1 37/37] isci: End the RNC resumption wait when the RNC is destroyed Dan Williams

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20120323002947.18065.29895.stgit@dwillia2-linux.jf.intel.com \
    --to=dan.j.williams@intel.com \
    --cc=jeffrey.d.skirvin@intel.com \
    --cc=linux-scsi@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).