[PATCH 6/8] qedf: Add schedule_hw_err_handler callback for fan failure

From: Javed Hasan <jhasan@marvell.com>
To: <martin.petersen@oracle.com>
Cc: <linux-scsi@vger.kernel.org>,
	<GR-QLogic-Storage-Upstream@marvell.com>, <jhasan@marvell.com>
Subject: [PATCH 6/8] qedf: Add schedule_hw_err_handler callback for fan failure
Date: Mon, 7 Sep 2020 05:14:41 -0700	[thread overview]
Message-ID: <20200907121443.5150-7-jhasan@marvell.com> (raw)
In-Reply-To: <20200907121443.5150-1-jhasan@marvell.com>

From: Saurav Kashyap <skashyap@marvell.com>

 On fan failure, disable the PCI function and initiate recovery for ramrod
 failure.


Signed-off-by: Javed Hasan <jhasan@marvell.com>
---
 drivers/scsi/qedf/qedf.h      |  4 ++++
 drivers/scsi/qedf/qedf_main.c | 45 +++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 49 insertions(+)

diff --git a/drivers/scsi/qedf/qedf.h b/drivers/scsi/qedf/qedf.h
index 15d6cbe..0e2cbb1 100644
--- a/drivers/scsi/qedf/qedf.h
+++ b/drivers/scsi/qedf/qedf.h
@@ -389,6 +389,7 @@ struct qedf_ctx {
 	mempool_t *io_mempool;
 	struct workqueue_struct *dpc_wq;
 	struct delayed_work recovery_work;
+	struct delayed_work board_disable_work;
 	struct delayed_work grcdump_work;
 	struct delayed_work stag_work;
 
@@ -541,6 +542,9 @@ extern void qedf_process_seq_cleanup_compl(struct qedf_ctx *qedf,
 extern void qedf_wq_grcdump(struct work_struct *work);
 void qedf_stag_change_work(struct work_struct *work);
 void qedf_ctx_soft_reset(struct fc_lport *lport);
+extern void qedf_board_disable_work(struct work_struct *work);
+extern void qedf_schedule_hw_err_handler(void *dev,
+		enum qed_hw_err_type err_type);
 
 #define FCOE_WORD_TO_BYTE  4
 #define QEDF_MAX_TASK_NUM	0xFFFF
diff --git a/drivers/scsi/qedf/qedf_main.c b/drivers/scsi/qedf/qedf_main.c
index 50af70a..3a45ca7 100644
--- a/drivers/scsi/qedf/qedf_main.c
+++ b/drivers/scsi/qedf/qedf_main.c
@@ -104,6 +104,12 @@
 MODULE_PARM_DESC(dp_level, " printk verbosity control passed to qed module  "
 	"during probe (0-3: 0 more verbose).");
 
+static bool qedf_enable_recovery = true;
+module_param_named(enable_recovery, qedf_enable_recovery,
+		bool, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(enable_recovery, "Enable/disable recovery on driver/firmware "
+		"interface level errors 0 = Disabled, 1 = Enabled (Default: 1).");
+
 struct workqueue_struct *qedf_io_wq;
 
 static struct fcoe_percpu_s qedf_global;
@@ -668,6 +674,7 @@ static u32 qedf_get_login_failures(void *cookie)
 		.dcbx_aen = qedf_dcbx_handler,
 		.get_generic_tlv_data = qedf_get_generic_tlv_data,
 		.get_protocol_tlv_data = qedf_get_protocol_tlv_data,
+		.schedule_hw_err_handler = qedf_schedule_hw_err_handler,
 	}
 };
 
@@ -3777,6 +3784,44 @@ void qedf_wq_grcdump(struct work_struct *work)
 	qedf_capture_grc_dump(qedf);
 }
 
+void qedf_schedule_hw_err_handler(void *dev, enum qed_hw_err_type err_type)
+{
+	struct qedf_ctx *qedf = dev;
+
+	QEDF_ERR(&(qedf->dbg_ctx),
+			"Hardware error handler scheduled, event=%d.\n",
+			err_type);
+
+	if (test_bit(QEDF_IN_RECOVERY, &qedf->flags)) {
+		QEDF_ERR(&(qedf->dbg_ctx),
+				"Already in recovery, not scheduling board disable work.\n");
+		return;
+	}
+
+	switch (err_type) {
+	case QED_HW_ERR_FAN_FAIL:
+		schedule_delayed_work(&qedf->board_disable_work, 0);
+		break;
+	case QED_HW_ERR_MFW_RESP_FAIL:
+	case QED_HW_ERR_HW_ATTN:
+	case QED_HW_ERR_DMAE_FAIL:
+	case QED_HW_ERR_FW_ASSERT:
+		/* Prevent HW attentions from being reasserted */
+		qed_ops->common->attn_clr_enable(qedf->cdev, true);
+		break;
+	case QED_HW_ERR_RAMROD_FAIL:
+		/* Prevent HW attentions from being reasserted */
+		qed_ops->common->attn_clr_enable(qedf->cdev, true);
+
+		if (qedf_enable_recovery)
+			qed_ops->common->recovery_process(qedf->cdev);
+
+		break;
+	default:
+		break;
+	}
+}
+
 /*
  * Protocol TLV handler
  */
-- 
1.8.3.1

