scsi: mpi3mr: Support Prepare for Reset event

The IOC sends a Prepare for Reset Event to the host to prepare for a Soft
Reset. This event data has two reason codes:

 1. Start - The host is expected to gracefully quiesce all I/O within
    approximately 1 second.

 2. Abort - The IOC is requesting to abort a previous Prepare for Reset
    Event request. Normal I/O may be resumed.

Link: https://lore.kernel.org/r/20211220141159.16117-20-sreekanth.reddy@broadcom.com
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>

diff --git a/drivers/scsi/mpi3mr/mpi3mr.h b/drivers/scsi/mpi3mr/mpi3mr.h
index 2602957..8dd669f 100644 (file)
--- a/drivers/scsi/mpi3mr/mpi3mr.h
+++ b/drivers/scsi/mpi3mr/mpi3mr.h
@@ -114,6 +114,7 @@ extern int prot_mask;
 #define MPI3MR_TSUPDATE_INTERVAL               900
 #define MPI3MR_DEFAULT_SHUTDOWN_TIME           120
 #define        MPI3MR_RAID_ERRREC_RESET_TIMEOUT        180
+#define MPI3MR_PREPARE_FOR_RESET_TIMEOUT       180
 #define MPI3MR_RESET_ACK_TIMEOUT               30
 
 #define MPI3MR_WATCHDOG_INTERVAL               1000 /* in milli seconds */
@@ -693,6 +694,8 @@ struct scmd_priv {
  * @prev_reset_result: Result of previous reset
  * @reset_mutex: Controller reset mutex
  * @reset_waitq: Controller reset  wait queue
+ * @prepare_for_reset: Prepare for reset event received
+ * @prepare_for_reset_timeout_counter: Prepare for reset timeout
  * @diagsave_timeout: Diagnostic information save timeout
  * @logging_level: Controller debug logging level
  * @flush_io_count: I/O count to flush after reset
@@ -825,6 +828,9 @@ struct mpi3mr_ioc {
        struct mutex reset_mutex;
        wait_queue_head_t reset_waitq;
 
+       u8 prepare_for_reset;
+       u16 prepare_for_reset_timeout_counter;
+
        u16 diagsave_timeout;
        int logging_level;
        u16 flush_io_count;

diff --git a/drivers/scsi/mpi3mr/mpi3mr_fw.c b/drivers/scsi/mpi3mr/mpi3mr_fw.c
index b25f8e5..a9491f7 100644 (file)
--- a/drivers/scsi/mpi3mr/mpi3mr_fw.c
+++ b/drivers/scsi/mpi3mr/mpi3mr_fw.c
@@ -2262,7 +2262,8 @@ static void mpi3mr_watchdog_work(struct work_struct *work)
            container_of(work, struct mpi3mr_ioc, watchdog_work.work);
        unsigned long flags;
        enum mpi3mr_iocstate ioc_state;
-       u32 fault, host_diagnostic;
+       u32 fault, host_diagnostic, ioc_status;
+       u32 reset_reason = MPI3MR_RESET_FROM_FAULT_WATCH;
 
        if (mrioc->reset_in_progress || mrioc->unrecoverable)
                return;
@@ -2272,43 +2273,55 @@ static void mpi3mr_watchdog_work(struct work_struct *work)
                mpi3mr_sync_timestamp(mrioc);
        }
 
+       if ((mrioc->prepare_for_reset) &&
+           ((mrioc->prepare_for_reset_timeout_counter++) >=
+            MPI3MR_PREPARE_FOR_RESET_TIMEOUT)) {
+               mpi3mr_soft_reset_handler(mrioc,
+                   MPI3MR_RESET_FROM_CIACTVRST_TIMER, 1);
+               return;
+       }
+
+       ioc_status = readl(&mrioc->sysif_regs->ioc_status);
+       if (ioc_status & MPI3_SYSIF_IOC_STATUS_RESET_HISTORY) {
+               mpi3mr_soft_reset_handler(mrioc, MPI3MR_RESET_FROM_FIRMWARE, 0);
+               return;
+       }
+
        /*Check for fault state every one second and issue Soft reset*/
        ioc_state = mpi3mr_get_iocstate(mrioc);
-       if (ioc_state == MRIOC_STATE_FAULT) {
-               fault = readl(&mrioc->sysif_regs->fault) &
-                   MPI3_SYSIF_FAULT_CODE_MASK;
-               host_diagnostic = readl(&mrioc->sysif_regs->host_diagnostic);
-               if (host_diagnostic & MPI3_SYSIF_HOST_DIAG_SAVE_IN_PROGRESS) {
-                       if (!mrioc->diagsave_timeout) {
-                               mpi3mr_print_fault_info(mrioc);
-                               ioc_warn(mrioc, "Diag save in progress\n");
-                       }
-                       if ((mrioc->diagsave_timeout++) <=
-                           MPI3_SYSIF_DIAG_SAVE_TIMEOUT)
-                               goto schedule_work;
-               } else
-                       mpi3mr_print_fault_info(mrioc);
-               mrioc->diagsave_timeout = 0;
+       if (ioc_state != MRIOC_STATE_FAULT)
+               goto schedule_work;
 
-               if (fault == MPI3_SYSIF_FAULT_CODE_POWER_CYCLE_REQUIRED) {
-                       ioc_info(mrioc,
-                           "Factory Reset fault occurred marking controller as unrecoverable"
-                           );
-                       mrioc->unrecoverable = 1;
-                       goto out;
+       fault = readl(&mrioc->sysif_regs->fault) & MPI3_SYSIF_FAULT_CODE_MASK;
+       host_diagnostic = readl(&mrioc->sysif_regs->host_diagnostic);
+       if (host_diagnostic & MPI3_SYSIF_HOST_DIAG_SAVE_IN_PROGRESS) {
+               if (!mrioc->diagsave_timeout) {
+                       mpi3mr_print_fault_info(mrioc);
+                       ioc_warn(mrioc, "diag save in progress\n");
                }
+               if ((mrioc->diagsave_timeout++) <= MPI3_SYSIF_DIAG_SAVE_TIMEOUT)
+                       goto schedule_work;
+       }
 
-               if ((fault == MPI3_SYSIF_FAULT_CODE_DIAG_FAULT_RESET) ||
-                   (fault == MPI3_SYSIF_FAULT_CODE_SOFT_RESET_IN_PROGRESS) ||
-                   (mrioc->reset_in_progress))
-                       goto out;
-               if (fault == MPI3_SYSIF_FAULT_CODE_CI_ACTIVATION_RESET)
-                       mpi3mr_soft_reset_handler(mrioc,
-                           MPI3MR_RESET_FROM_CIACTIV_FAULT, 0);
-               else
-                       mpi3mr_soft_reset_handler(mrioc,
-                           MPI3MR_RESET_FROM_FAULT_WATCH, 0);
+       mpi3mr_print_fault_info(mrioc);
+       mrioc->diagsave_timeout = 0;
+
+       switch (fault) {
+       case MPI3_SYSIF_FAULT_CODE_POWER_CYCLE_REQUIRED:
+               ioc_info(mrioc,
+                   "controller requires system power cycle, marking controller as unrecoverable\n");
+               mrioc->unrecoverable = 1;
+               return;
+       case MPI3_SYSIF_FAULT_CODE_SOFT_RESET_IN_PROGRESS:
+               return;
+       case MPI3_SYSIF_FAULT_CODE_CI_ACTIVATION_RESET:
+               reset_reason = MPI3MR_RESET_FROM_CIACTIV_FAULT;
+               break;
+       default:
+               break;
        }
+       mpi3mr_soft_reset_handler(mrioc, reset_reason, 0);
+       return;
 
 schedule_work:
        spin_lock_irqsave(&mrioc->watchdog_lock, flags);
@@ -2317,7 +2330,6 @@ schedule_work:
                    &mrioc->watchdog_work,
                    msecs_to_jiffies(MPI3MR_WATCHDOG_INTERVAL));
        spin_unlock_irqrestore(&mrioc->watchdog_lock, flags);
-out:
        return;
 }
 
@@ -3488,6 +3500,7 @@ static int mpi3mr_enable_events(struct mpi3mr_ioc *mrioc)
        mpi3mr_unmask_events(mrioc, MPI3_EVENT_SAS_BROADCAST_PRIMITIVE);
        mpi3mr_unmask_events(mrioc, MPI3_EVENT_PCIE_TOPOLOGY_CHANGE_LIST);
        mpi3mr_unmask_events(mrioc, MPI3_EVENT_PCIE_ENUMERATION);
+       mpi3mr_unmask_events(mrioc, MPI3_EVENT_PREPARE_FOR_RESET);
        mpi3mr_unmask_events(mrioc, MPI3_EVENT_CABLE_MGMT);
        mpi3mr_unmask_events(mrioc, MPI3_EVENT_ENERGY_PACK_CHANGE);
 
@@ -4223,6 +4236,10 @@ int mpi3mr_soft_reset_handler(struct mpi3mr_ioc *mrioc,
        mpi3mr_cleanup_fwevt_list(mrioc);
        mpi3mr_flush_host_io(mrioc);
        mpi3mr_invalidate_devhandles(mrioc);
+       if (mrioc->prepare_for_reset) {
+               mrioc->prepare_for_reset = 0;
+               mrioc->prepare_for_reset_timeout_counter = 0;
+       }
        mpi3mr_memset_buffers(mrioc);
        retval = mpi3mr_reinit_ioc(mrioc, 0);
        if (retval) {

diff --git a/drivers/scsi/mpi3mr/mpi3mr_os.c b/drivers/scsi/mpi3mr/mpi3mr_os.c
index 728d6ce..192986f 100644 (file)
--- a/drivers/scsi/mpi3mr/mpi3mr_os.c
+++ b/drivers/scsi/mpi3mr/mpi3mr_os.c
@@ -1989,6 +1989,40 @@ out:
 }
 
 /**
+ * mpi3mr_preparereset_evt_th - Prepare for reset event tophalf
+ * @mrioc: Adapter instance reference
+ * @event_reply: event data
+ *
+ * Blocks and unblocks host level I/O based on the reason code
+ *
+ * Return: Nothing
+ */
+static void mpi3mr_preparereset_evt_th(struct mpi3mr_ioc *mrioc,
+       struct mpi3_event_notification_reply *event_reply)
+{
+       struct mpi3_event_data_prepare_for_reset *evtdata =
+           (struct mpi3_event_data_prepare_for_reset *)event_reply->event_data;
+
+       if (evtdata->reason_code == MPI3_EVENT_PREPARE_RESET_RC_START) {
+               dprint_event_th(mrioc,
+                   "prepare for reset event top half with rc=start\n");
+               if (mrioc->prepare_for_reset)
+                       return;
+               mrioc->prepare_for_reset = 1;
+               mrioc->prepare_for_reset_timeout_counter = 0;
+       } else if (evtdata->reason_code == MPI3_EVENT_PREPARE_RESET_RC_ABORT) {
+               dprint_event_th(mrioc,
+                   "prepare for reset top half with rc=abort\n");
+               mrioc->prepare_for_reset = 0;
+               mrioc->prepare_for_reset_timeout_counter = 0;
+       }
+       if ((event_reply->msg_flags & MPI3_EVENT_NOTIFY_MSGFLAGS_ACK_MASK)
+           == MPI3_EVENT_NOTIFY_MSGFLAGS_ACK_REQUIRED)
+               mpi3mr_send_event_ack(mrioc, event_reply->event, NULL,
+                   le32_to_cpu(event_reply->event_context));
+}
+
+/**
  * mpi3mr_energypackchg_evt_th - Energy pack change evt tophalf
  * @mrioc: Adapter instance reference
  * @event_reply: event data
@@ -2075,6 +2109,12 @@ void mpi3mr_os_handle_events(struct mpi3mr_ioc *mrioc,
                mpi3mr_pcietopochg_evt_th(mrioc, event_reply);
                break;
        }
+       case MPI3_EVENT_PREPARE_FOR_RESET:
+       {
+               mpi3mr_preparereset_evt_th(mrioc, event_reply);
+               ack_req = 0;
+               break;
+       }
        case MPI3_EVENT_DEVICE_INFO_CHANGED:
        {
                process_evt_bh = 1;