diff options
Diffstat (limited to 'drivers/s390/scsi/zfcp_erp.c')
-rw-r--r-- | drivers/s390/scsi/zfcp_erp.c | 16 |
1 files changed, 8 insertions, 8 deletions
diff --git a/drivers/s390/scsi/zfcp_erp.c b/drivers/s390/scsi/zfcp_erp.c index 07fa824d179..2dc8110ebf7 100644 --- a/drivers/s390/scsi/zfcp_erp.c +++ b/drivers/s390/scsi/zfcp_erp.c @@ -131,7 +131,7 @@ static void zfcp_close_qdio(struct zfcp_adapter *adapter) debug_text_event(adapter->erp_dbf, 3, "qdio_down2a"); while (qdio_shutdown(adapter->ccw_device, QDIO_FLAG_CLEANUP_USING_CLEAR) == -EINPROGRESS) - msleep(1000); + ssleep(1); debug_text_event(adapter->erp_dbf, 3, "qdio_down2b"); /* cleanup used outbound sbals */ @@ -456,7 +456,7 @@ zfcp_test_link(struct zfcp_port *port) zfcp_port_get(port); retval = zfcp_erp_adisc(port); - if (retval != 0) { + if (retval != 0 && retval != -EBUSY) { zfcp_port_put(port); ZFCP_LOG_NORMAL("reopen needed for port 0x%016Lx " "on adapter %s\n ", port->wwpn, @@ -846,7 +846,8 @@ zfcp_erp_strategy_check_fsfreq(struct zfcp_erp_action *erp_action) if (erp_action->fsf_req) { /* take lock to ensure that request is not deleted meanwhile */ spin_lock(&adapter->req_list_lock); - if (zfcp_reqlist_find(adapter, erp_action->fsf_req->req_id)) { + if (zfcp_reqlist_find_safe(adapter, erp_action->fsf_req) && + erp_action->fsf_req->erp_action == erp_action) { /* fsf_req still exists */ debug_text_event(adapter->erp_dbf, 3, "a_ca_req"); debug_event(adapter->erp_dbf, 3, &erp_action->fsf_req, @@ -1285,7 +1286,7 @@ zfcp_erp_strategy_do_action(struct zfcp_erp_action *erp_action) * note: no lock in subsequent strategy routines * (this allows these routine to call schedule, e.g. * kmalloc with such flags or qdio_initialize & friends) - * Note: in case of timeout, the seperate strategies will fail + * Note: in case of timeout, the separate strategies will fail * anyhow. No need for a special action. Even worse, a nameserver * failure would not wake up waiting ports without the call. */ @@ -1609,7 +1610,6 @@ static void zfcp_erp_scsi_scan(struct work_struct *work) scsi_scan_target(&rport->dev, 0, rport->scsi_target_id, unit->scsi_lun, 0); atomic_clear_mask(ZFCP_STATUS_UNIT_SCSI_WORK_PENDING, &unit->status); - wake_up(&unit->scsi_scan_wq); zfcp_unit_put(unit); kfree(p); } @@ -1900,7 +1900,7 @@ zfcp_erp_adapter_strategy(struct zfcp_erp_action *erp_action) ZFCP_LOG_INFO("Waiting to allow the adapter %s " "to recover itself\n", zfcp_get_busid_by_adapter(adapter)); - msleep(jiffies_to_msecs(ZFCP_TYPE2_RECOVERY_TIME)); + ssleep(ZFCP_TYPE2_RECOVERY_TIME); } return retval; @@ -2080,7 +2080,7 @@ zfcp_erp_adapter_strategy_open_qdio(struct zfcp_erp_action *erp_action) debug_text_event(adapter->erp_dbf, 3, "qdio_down1a"); while (qdio_shutdown(adapter->ccw_device, QDIO_FLAG_CLEANUP_USING_CLEAR) == -EINPROGRESS) - msleep(1000); + ssleep(1); debug_text_event(adapter->erp_dbf, 3, "qdio_down1b"); failed_qdio_establish: @@ -2165,7 +2165,7 @@ zfcp_erp_adapter_strategy_open_fsf_xconfig(struct zfcp_erp_action *erp_action) ZFCP_LOG_DEBUG("host connection still initialising... " "waiting and retrying...\n"); /* sleep a little bit before retry */ - msleep(jiffies_to_msecs(sleep)); + ssleep(sleep); sleep *= 2; } |