scsi: mpi3mr: Add support for recovering controller

Detection of firmware fault or any kind of unresponsiveness in the
controller (any admin command which times out) results in resetting the
controller. The primary reset mechanisms used are either soft reset or diag
fault reset. A reset is performed if the host sets the ResetAction field in
the HostDiagnostic register to either 001b (soft reset) or 007b (diag fault
reset). After successfully resetting the controller the driver
reinitializes the controller by going through start of the day
initialization procedure. Pending I/Os during the reset are returned back
to the SCSI midlayer for retry.

Link: https://lore.kernel.org/r/20210520152545.2710479-10-kashyap.desai@broadcom.com
Cc: sathya.prakash@broadcom.co
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Tomas Henzl <thenzl@redhat.com>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Kashyap Desai <kashyap.desai@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

1 files changed, 379 insertions, 36 deletions

diff --git a/drivers/scsi/mpi3mr/mpi3mr_fw.c b/drivers/scsi/mpi3mr/mpi3mr_fw.c
index 7816e9e154b5..0a6e5246ad84 100644
--- a/drivers/scsi/mpi3mr/mpi3mr_fw.c
+++ b/drivers/scsi/mpi3mr/mpi3mr_fw.c
@@ -1595,6 +1595,41 @@ void mpi3mr_stop_watchdog(struct mpi3mr_ioc *mrioc)
 }
 
 /**
+ * mpi3mr_kill_ioc - Kill the controller
+ * @mrioc: Adapter instance reference
+ * @reason: reason for the failure.
+ *
+ * If fault debug is enabled, display the fault info else issue
+ * diag fault and freeze the system for controller debug
+ * purpose.
+ *
+ * Return: Nothing.
+ */
+static void mpi3mr_kill_ioc(struct mpi3mr_ioc *mrioc, u32 reason)
+{
+	enum mpi3mr_iocstate ioc_state;
+
+	if (!mrioc->fault_dbg)
+		return;
+
+	dump_stack();
+
+	ioc_state = mpi3mr_get_iocstate(mrioc);
+	if (ioc_state == MRIOC_STATE_FAULT)
+		mpi3mr_print_fault_info(mrioc);
+	else {
+		ioc_err(mrioc, "Firmware is halted due to the reason %d\n",
+		    reason);
+		mpi3mr_diagfault_reset_handler(mrioc, reason);
+	}
+	if (mrioc->fault_dbg == 2)
+		for (;;)
+			;
+	else
+		panic("panic in %s\n", __func__);
+}
+
+/**
  * mpi3mr_setup_admin_qpair - Setup admin queue pair
  * @mrioc: Adapter instance reference
  *
@@ -2550,6 +2585,7 @@ out_failed:
 /**
  * mpi3mr_init_ioc - Initialize the controller
  * @mrioc: Adapter instance reference
+ * @re_init: Flag to indicate is this fresh init or re-init
  *
  * This the controller initialization routine, executed either
  * after soft reset or from pci probe callback.
@@ -2562,7 +2598,7 @@ out_failed:
  *
  * Return: 0 on success and non-zero on failure.
  */
-int mpi3mr_init_ioc(struct mpi3mr_ioc *mrioc)
+int mpi3mr_init_ioc(struct mpi3mr_ioc *mrioc, u8 re_init)
 {
 	int retval = 0;
 	enum mpi3mr_iocstate ioc_state;
@@ -2572,13 +2608,16 @@ int mpi3mr_init_ioc(struct mpi3mr_ioc *mrioc)
 	struct mpi3_ioc_facts_data facts_data;
 
 	mrioc->change_count = 0;
-	mrioc->cpu_count = num_online_cpus();
-	retval = mpi3mr_setup_resources(mrioc);
-	if (retval) {
-		ioc_err(mrioc, "Failed to setup resources:error %d\n",
-		    retval);
-		goto out_nocleanup;
+	if (!re_init) {
+		mrioc->cpu_count = num_online_cpus();
+		retval = mpi3mr_setup_resources(mrioc);
+		if (retval) {
+			ioc_err(mrioc, "Failed to setup resources:error %d\n",
+			    retval);
+			goto out_nocleanup;
+		}
 	}
+
 	ioc_status = readl(&mrioc->sysif_regs->ioc_status);
 	ioc_config = readl(&mrioc->sysif_regs->ioc_configuration);
 
@@ -2653,12 +2692,15 @@ int mpi3mr_init_ioc(struct mpi3mr_ioc *mrioc)
 		goto out_failed;
 	}
 
-	retval = mpi3mr_setup_isr(mrioc, 1);
-	if (retval) {
-		ioc_err(mrioc, "Failed to setup ISR error %d\n",
-		    retval);
-		goto out_failed;
-	}
+	if (!re_init) {
+		retval = mpi3mr_setup_isr(mrioc, 1);
+		if (retval) {
+			ioc_err(mrioc, "Failed to setup ISR error %d\n",
+			    retval);
+			goto out_failed;
+		}
+	} else
+		mpi3mr_ioc_enable_intr(mrioc);
 
 	retval = mpi3mr_issue_iocfacts(mrioc, &facts_data);
 	if (retval) {
@@ -2668,11 +2710,13 @@ int mpi3mr_init_ioc(struct mpi3mr_ioc *mrioc)
 	}
 
 	mpi3mr_process_factsdata(mrioc, &facts_data);
-	retval = mpi3mr_check_reset_dma_mask(mrioc);
-	if (retval) {
-		ioc_err(mrioc, "Resetting dma mask failed %d\n",
-		    retval);
-		goto out_failed;
+	if (!re_init) {
+		retval = mpi3mr_check_reset_dma_mask(mrioc);
+		if (retval) {
+			ioc_err(mrioc, "Resetting dma mask failed %d\n",
+			    retval);
+			goto out_failed;
+		}
 	}
 
 	retval = mpi3mr_alloc_reply_sense_bufs(mrioc);
@@ -2683,11 +2727,13 @@ int mpi3mr_init_ioc(struct mpi3mr_ioc *mrioc)
 		goto out_failed;
 	}
 
-	retval = mpi3mr_alloc_chain_bufs(mrioc);
-	if (retval) {
-		ioc_err(mrioc, "Failed to allocated chain buffers %d\n",
-		    retval);
-		goto out_failed;
+	if (!re_init) {
+		retval = mpi3mr_alloc_chain_bufs(mrioc);
+		if (retval) {
+			ioc_err(mrioc, "Failed to allocated chain buffers %d\n",
+			    retval);
+			goto out_failed;
+		}
 	}
 
 	retval = mpi3mr_issue_iocinit(mrioc);
@@ -2704,11 +2750,13 @@ int mpi3mr_init_ioc(struct mpi3mr_ioc *mrioc)
 	writel(mrioc->sbq_host_index,
 	    &mrioc->sysif_regs->sense_buffer_free_host_index);
 
-	retval = mpi3mr_setup_isr(mrioc, 0);
-	if (retval) {
-		ioc_err(mrioc, "Failed to re-setup ISR, error %d\n",
-		    retval);
-		goto out_failed;
+	if (!re_init)  {
+		retval = mpi3mr_setup_isr(mrioc, 0);
+		if (retval) {
+			ioc_err(mrioc, "Failed to re-setup ISR, error %d\n",
+			    retval);
+			goto out_failed;
+		}
 	}
 
 	retval = mpi3mr_create_op_queues(mrioc);
@@ -2718,6 +2766,14 @@ int mpi3mr_init_ioc(struct mpi3mr_ioc *mrioc)
 		goto out_failed;
 	}
 
+	if (re_init &&
+	    (mrioc->shost->nr_hw_queues > mrioc->num_op_reply_q)) {
+		ioc_err(mrioc,
+		    "Cannot create minimum number of OpQueues expected:%d created:%d\n",
+		    mrioc->shost->nr_hw_queues, mrioc->num_op_reply_q);
+		goto out_failed;
+	}
+
 	for (i = 0; i < MPI3_EVENT_NOTIFY_EVENTMASK_WORDS; i++)
 		mrioc->event_masks[i] = -1;
 
@@ -2741,15 +2797,111 @@ int mpi3mr_init_ioc(struct mpi3mr_ioc *mrioc)
 		goto out_failed;
 	}
 
+	if (re_init) {
+		ioc_info(mrioc, "Issuing Port Enable\n");
+		retval = mpi3mr_issue_port_enable(mrioc, 0);
+		if (retval) {
+			ioc_err(mrioc, "Failed to issue port enable %d\n",
+			    retval);
+			goto out_failed;
+		}
+	}
 	return retval;
 
 out_failed:
-	mpi3mr_cleanup_ioc(mrioc);
+	mpi3mr_cleanup_ioc(mrioc, re_init);
 out_nocleanup:
 	return retval;
 }
 
 /**
+ * mpi3mr_memset_op_reply_q_buffers - memset the operational reply queue's
+ *					segments
+ * @mrioc: Adapter instance reference
+ * @qidx: Operational reply queue index
+ *
+ * Return: Nothing.
+ */
+static void mpi3mr_memset_op_reply_q_buffers(struct mpi3mr_ioc *mrioc, u16 qidx)
+{
+	struct op_reply_qinfo *op_reply_q = mrioc->op_reply_qinfo + qidx;
+	struct segments *segments;
+	int i, size;
+
+	if (!op_reply_q->q_segments)
+		return;
+
+	size = op_reply_q->segment_qd * mrioc->op_reply_desc_sz;
+	segments = op_reply_q->q_segments;
+	for (i = 0; i < op_reply_q->num_segments; i++)
+		memset(segments[i].segment, 0, size);
+}
+
+/**
+ * mpi3mr_memset_op_req_q_buffers - memset the operational request queue's
+ *					segments
+ * @mrioc: Adapter instance reference
+ * @qidx: Operational request queue index
+ *
+ * Return: Nothing.
+ */
+static void mpi3mr_memset_op_req_q_buffers(struct mpi3mr_ioc *mrioc, u16 qidx)
+{
+	struct op_req_qinfo *op_req_q = mrioc->req_qinfo + qidx;
+	struct segments *segments;
+	int i, size;
+
+	if (!op_req_q->q_segments)
+		return;
+
+	size = op_req_q->segment_qd * mrioc->facts.op_req_sz;
+	segments = op_req_q->q_segments;
+	for (i = 0; i < op_req_q->num_segments; i++)
+		memset(segments[i].segment, 0, size);
+}
+
+/**
+ * mpi3mr_memset_buffers - memset memory for a controller
+ * @mrioc: Adapter instance reference
+ *
+ * clear all the memory allocated for a controller, typically
+ * called post reset to reuse the memory allocated during the
+ * controller init.
+ *
+ * Return: Nothing.
+ */
+static void mpi3mr_memset_buffers(struct mpi3mr_ioc *mrioc)
+{
+	u16 i;
+
+	memset(mrioc->admin_req_base, 0, mrioc->admin_req_q_sz);
+	memset(mrioc->admin_reply_base, 0, mrioc->admin_reply_q_sz);
+
+	memset(mrioc->init_cmds.reply, 0, sizeof(*mrioc->init_cmds.reply));
+	for (i = 0; i < MPI3MR_NUM_DEVRMCMD; i++)
+		memset(mrioc->dev_rmhs_cmds[i].reply, 0,
+		    sizeof(*mrioc->dev_rmhs_cmds[i].reply));
+	memset(mrioc->removepend_bitmap, 0, mrioc->dev_handle_bitmap_sz);
+	memset(mrioc->devrem_bitmap, 0, mrioc->devrem_bitmap_sz);
+
+	for (i = 0; i < mrioc->num_queues; i++) {
+		mrioc->op_reply_qinfo[i].qid = 0;
+		mrioc->op_reply_qinfo[i].ci = 0;
+		mrioc->op_reply_qinfo[i].num_replies = 0;
+		mrioc->op_reply_qinfo[i].ephase = 0;
+		mpi3mr_memset_op_reply_q_buffers(mrioc, i);
+
+		mrioc->req_qinfo[i].ci = 0;
+		mrioc->req_qinfo[i].pi = 0;
+		mrioc->req_qinfo[i].num_requests = 0;
+		mrioc->req_qinfo[i].qid = 0;
+		mrioc->req_qinfo[i].reply_qid = 0;
+		spin_lock_init(&mrioc->req_qinfo[i].q_lock);
+		mpi3mr_memset_op_req_q_buffers(mrioc, i);
+	}
+}
+
+/**
  * mpi3mr_free_mem - Free memory allocated for a controller
  * @mrioc: Adapter instance reference
  *
@@ -2921,6 +3073,7 @@ static void mpi3mr_issue_ioc_shutdown(struct mpi3mr_ioc *mrioc)
 /**
  * mpi3mr_cleanup_ioc - Cleanup controller
  * @mrioc: Adapter instance reference
+ * @re_init: Cleanup due to a reinit or not
  *
  * controller cleanup handler, Message unit reset or soft reset
  * and shutdown notification is issued to the controller and the
@@ -2928,11 +3081,12 @@ static void mpi3mr_issue_ioc_shutdown(struct mpi3mr_ioc *mrioc)
  *
  * Return: Nothing.
  */
-void mpi3mr_cleanup_ioc(struct mpi3mr_ioc *mrioc)
+void mpi3mr_cleanup_ioc(struct mpi3mr_ioc *mrioc, u8 re_init)
 {
 	enum mpi3mr_iocstate ioc_state;
 
-	mpi3mr_stop_watchdog(mrioc);
+	if (!re_init)
+		mpi3mr_stop_watchdog(mrioc);
 
 	mpi3mr_ioc_disable_intr(mrioc);
 
@@ -2946,11 +3100,93 @@ void mpi3mr_cleanup_ioc(struct mpi3mr_ioc *mrioc)
 			    MPI3_SYSIF_HOST_DIAG_RESET_ACTION_SOFT_RESET,
 			    MPI3MR_RESET_FROM_MUR_FAILURE);
 
-		 mpi3mr_issue_ioc_shutdown(mrioc);
+		if (!re_init)
+			mpi3mr_issue_ioc_shutdown(mrioc);
 	}
 
-	mpi3mr_free_mem(mrioc);
-	mpi3mr_cleanup_resources(mrioc);
+	if (!re_init) {
+		mpi3mr_free_mem(mrioc);
+		mpi3mr_cleanup_resources(mrioc);
+	}
+}
+
+/**
+ * mpi3mr_drv_cmd_comp_reset - Flush a internal driver command
+ * @mrioc: Adapter instance reference
+ * @cmdptr: Internal command tracker
+ *
+ * Complete an internal driver commands with state indicating it
+ * is completed due to reset.
+ *
+ * Return: Nothing.
+ */
+static inline void mpi3mr_drv_cmd_comp_reset(struct mpi3mr_ioc *mrioc,
+	struct mpi3mr_drv_cmd *cmdptr)
+{
+	if (cmdptr->state & MPI3MR_CMD_PENDING) {
+		cmdptr->state |= MPI3MR_CMD_RESET;
+		cmdptr->state &= ~MPI3MR_CMD_PENDING;
+		if (cmdptr->is_waiting) {
+			complete(&cmdptr->done);
+			cmdptr->is_waiting = 0;
+		} else if (cmdptr->callback)
+			cmdptr->callback(mrioc, cmdptr);
+	}
+}
+
+/**
+ * mpi3mr_flush_drv_cmds - Flush internaldriver commands
+ * @mrioc: Adapter instance reference
+ *
+ * Flush all internal driver commands post reset
+ *
+ * Return: Nothing.
+ */
+static void mpi3mr_flush_drv_cmds(struct mpi3mr_ioc *mrioc)
+{
+	struct mpi3mr_drv_cmd *cmdptr;
+	u8 i;
+
+	cmdptr = &mrioc->init_cmds;
+	mpi3mr_drv_cmd_comp_reset(mrioc, cmdptr);
+
+	for (i = 0; i < MPI3MR_NUM_DEVRMCMD; i++) {
+		cmdptr = &mrioc->dev_rmhs_cmds[i];
+		mpi3mr_drv_cmd_comp_reset(mrioc, cmdptr);
+	}
+}
+
+/**
+ * mpi3mr_diagfault_reset_handler - Diag fault reset handler
+ * @mrioc: Adapter instance reference
+ * @reset_reason: Reset reason code
+ *
+ * This is an handler for issuing diag fault reset from the
+ * applications through IOCTL path to stop the execution of the
+ * controller
+ *
+ * Return: 0 on success, non-zero on failure.
+ */
+int mpi3mr_diagfault_reset_handler(struct mpi3mr_ioc *mrioc,
+	u32 reset_reason)
+{
+	int retval = 0;
+
+	mrioc->reset_in_progress = 1;
+
+	mpi3mr_ioc_disable_intr(mrioc);
+
+	retval = mpi3mr_issue_reset(mrioc,
+	    MPI3_SYSIF_HOST_DIAG_RESET_ACTION_DIAG_FAULT, reset_reason);
+
+	if (retval) {
+		ioc_err(mrioc, "The diag fault reset failed: reason %d\n",
+		    reset_reason);
+		mpi3mr_ioc_enable_intr(mrioc);
+	}
+	ioc_info(mrioc, "%s\n", ((retval == 0) ? "SUCCESS" : "FAILED"));
+	mrioc->reset_in_progress = 0;
+	return retval;
 }
 
 /**
@@ -2959,12 +3195,119 @@ void mpi3mr_cleanup_ioc(struct mpi3mr_ioc *mrioc)
  * @reset_reason: Reset reason code
  * @snapdump: Flag to generate snapdump in firmware or not
  *
- * TBD
+ * This is an handler for recovering controller by issuing soft
+ * reset are diag fault reset.  This is a blocking function and
+ * when one reset is executed if any other resets they will be
+ * blocked. All IOCTLs/IO will be blocked during the reset. If
+ * controller reset is successful then the controller will be
+ * reinitalized, otherwise the controller will be marked as not
+ * recoverable
+ *
+ * In snapdump bit is set, the controller is issued with diag
+ * fault reset so that the firmware can create a snap dump and
+ * post that the firmware will result in F000 fault and the
+ * driver will issue soft reset to recover from that.
  *
  * Return: 0 on success, non-zero on failure.
  */
 int mpi3mr_soft_reset_handler(struct mpi3mr_ioc *mrioc,
 	u32 reset_reason, u8 snapdump)
 {
-	return 0;
+	int retval = 0, i;
+	unsigned long flags;
+	u32 host_diagnostic, timeout = MPI3_SYSIF_DIAG_SAVE_TIMEOUT * 10;
+
+	if (mrioc->fault_dbg) {
+		if (snapdump)
+			mpi3mr_set_diagsave(mrioc);
+		mpi3mr_kill_ioc(mrioc, reset_reason);
+	}
+
+	/*
+	 * Block new resets until the currently executing one is finished and
+	 * return the status of the existing reset for all blocked resets
+	 */
+	if (!mutex_trylock(&mrioc->reset_mutex)) {
+		ioc_info(mrioc, "Another reset in progress\n");
+		return -1;
+	}
+	mrioc->reset_in_progress = 1;
+
+	if ((!snapdump) && (reset_reason != MPI3MR_RESET_FROM_FAULT_WATCH) &&
+	    (reset_reason != MPI3MR_RESET_FROM_CIACTIV_FAULT)) {
+		for (i = 0; i < MPI3_EVENT_NOTIFY_EVENTMASK_WORDS; i++)
+			mrioc->event_masks[i] = -1;
+
+		retval = mpi3mr_issue_event_notification(mrioc);
+
+		if (retval) {
+			ioc_err(mrioc,
+			    "Failed to turn off events prior to reset %d\n",
+			    retval);
+		}
+	}
+
+	mpi3mr_ioc_disable_intr(mrioc);
+
+	if (snapdump) {
+		mpi3mr_set_diagsave(mrioc);
+		retval = mpi3mr_issue_reset(mrioc,
+		    MPI3_SYSIF_HOST_DIAG_RESET_ACTION_DIAG_FAULT, reset_reason);
+		if (!retval) {
+			do {
+				host_diagnostic =
+				    readl(&mrioc->sysif_regs->host_diagnostic);
+				if (!(host_diagnostic &
+				    MPI3_SYSIF_HOST_DIAG_SAVE_IN_PROGRESS))
+					break;
+				msleep(100);
+			} while (--timeout);
+		}
+	}
+
+	retval = mpi3mr_issue_reset(mrioc,
+	    MPI3_SYSIF_HOST_DIAG_RESET_ACTION_SOFT_RESET, reset_reason);
+	if (retval) {
+		ioc_err(mrioc, "Failed to issue soft reset to the ioc\n");
+		goto out;
+	}
+
+	mpi3mr_flush_delayed_rmhs_list(mrioc);
+	mpi3mr_flush_drv_cmds(mrioc);
+	memset(mrioc->devrem_bitmap, 0, mrioc->devrem_bitmap_sz);
+	memset(mrioc->removepend_bitmap, 0, mrioc->dev_handle_bitmap_sz);
+	mpi3mr_cleanup_fwevt_list(mrioc);
+	mpi3mr_flush_host_io(mrioc);
+	mpi3mr_invalidate_devhandles(mrioc);
+	mpi3mr_memset_buffers(mrioc);
+	retval = mpi3mr_init_ioc(mrioc, 1);
+	if (retval) {
+		pr_err(IOCNAME "reinit after soft reset failed: reason %d\n",
+		    mrioc->name, reset_reason);
+		goto out;
+	}
+	ssleep(10);
+
+out:
+	if (!retval) {
+		mrioc->reset_in_progress = 0;
+		scsi_unblock_requests(mrioc->shost);
+		mpi3mr_rfresh_tgtdevs(mrioc);
+		spin_lock_irqsave(&mrioc->watchdog_lock, flags);
+		if (mrioc->watchdog_work_q)
+			queue_delayed_work(mrioc->watchdog_work_q,
+			    &mrioc->watchdog_work,
+			    msecs_to_jiffies(MPI3MR_WATCHDOG_INTERVAL));
+		spin_unlock_irqrestore(&mrioc->watchdog_lock, flags);
+	} else {
+		mpi3mr_issue_reset(mrioc,
+		    MPI3_SYSIF_HOST_DIAG_RESET_ACTION_DIAG_FAULT, reset_reason);
+		mrioc->unrecoverable = 1;
+		mrioc->reset_in_progress = 0;
+		retval = -1;
+	}
+
+	mutex_unlock(&mrioc->reset_mutex);
+	ioc_info(mrioc, "%s\n", ((retval == 0) ? "SUCCESS" : "FAILED"));
+	return retval;
 }