diff options
author | Martin Peschke <mp3@de.ibm.com> | 2008-03-27 14:22:03 +0100 |
---|---|---|
committer | James Bottomley <James.Bottomley@HansenPartnership.com> | 2008-04-07 12:19:04 -0500 |
commit | 9467a9b3efdd9041202f71cc270bda827a7ec777 (patch) | |
tree | 9e7b8e167da9aa5065b1f7ed1d2a1ab9c2ad47da /drivers | |
parent | 698ec01635819c5ae60090bb4efcbeffc41642fb (diff) |
[SCSI] zfcp: Trace all triggers of error recovery activity
This patch allows any recovery event to be traced back to an exact
cause, e.g. a particular request identified by an id (address).
Signed-off-by: Martin Peschke <mp3@de.ibm.com>
Signed-off-by: Christof Schmitt <christof.schmitt@de.ibm.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Diffstat (limited to 'drivers')
-rw-r--r-- | drivers/s390/scsi/zfcp_aux.c | 31 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_ccw.c | 12 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_dbf.c | 134 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_def.h | 14 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_erp.c | 213 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_ext.h | 25 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_fsf.c | 103 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_qdio.c | 4 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_scsi.c | 2 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_sysfs_adapter.c | 6 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_sysfs_port.c | 6 | ||||
-rw-r--r-- | drivers/s390/scsi/zfcp_sysfs_unit.c | 2 |
12 files changed, 362 insertions, 190 deletions
diff --git a/drivers/s390/scsi/zfcp_aux.c b/drivers/s390/scsi/zfcp_aux.c index f3eff7ebcb6..d2a744200c9 100644 --- a/drivers/s390/scsi/zfcp_aux.c +++ b/drivers/s390/scsi/zfcp_aux.c @@ -1326,10 +1326,10 @@ zfcp_nameserver_enqueue(struct zfcp_adapter *adapter) #define ZFCP_LOG_AREA ZFCP_LOG_AREA_FC -static void -zfcp_fsf_incoming_els_rscn(struct zfcp_adapter *adapter, - struct fsf_status_read_buffer *status_buffer) +static void zfcp_fsf_incoming_els_rscn(struct zfcp_fsf_req *fsf_req) { + struct fsf_status_read_buffer *status_buffer = (void*)fsf_req->data; + struct zfcp_adapter *adapter = fsf_req->adapter; struct fcp_rscn_head *fcp_rscn_head; struct fcp_rscn_element *fcp_rscn_element; struct zfcp_port *port; @@ -1376,7 +1376,8 @@ zfcp_fsf_incoming_els_rscn(struct zfcp_adapter *adapter, ZFCP_LOG_INFO("incoming RSCN, trying to open " "port 0x%016Lx\n", port->wwpn); zfcp_erp_port_reopen(port, - ZFCP_STATUS_COMMON_ERP_FAILED); + ZFCP_STATUS_COMMON_ERP_FAILED, + 82, (u64)fsf_req); continue; } @@ -1407,10 +1408,10 @@ zfcp_fsf_incoming_els_rscn(struct zfcp_adapter *adapter, } } -static void -zfcp_fsf_incoming_els_plogi(struct zfcp_adapter *adapter, - struct fsf_status_read_buffer *status_buffer) +static void zfcp_fsf_incoming_els_plogi(struct zfcp_fsf_req *fsf_req) { + struct fsf_status_read_buffer *status_buffer = (void*)fsf_req->data; + struct zfcp_adapter *adapter = fsf_req->adapter; struct fsf_plogi *els_plogi; struct zfcp_port *port; unsigned long flags; @@ -1429,14 +1430,14 @@ zfcp_fsf_incoming_els_plogi(struct zfcp_adapter *adapter, status_buffer->d_id, zfcp_get_busid_by_adapter(adapter)); } else { - zfcp_erp_port_forced_reopen(port, 0); + zfcp_erp_port_forced_reopen(port, 0, 83, (u64)fsf_req); } } -static void -zfcp_fsf_incoming_els_logo(struct zfcp_adapter *adapter, - struct fsf_status_read_buffer *status_buffer) +static void zfcp_fsf_incoming_els_logo(struct zfcp_fsf_req *fsf_req) { + struct fsf_status_read_buffer *status_buffer = (void*)fsf_req->data; + struct zfcp_adapter *adapter = fsf_req->adapter; struct fcp_logo *els_logo = (struct fcp_logo *) status_buffer->payload; struct zfcp_port *port; unsigned long flags; @@ -1454,7 +1455,7 @@ zfcp_fsf_incoming_els_logo(struct zfcp_adapter *adapter, status_buffer->d_id, zfcp_get_busid_by_adapter(adapter)); } else { - zfcp_erp_port_forced_reopen(port, 0); + zfcp_erp_port_forced_reopen(port, 0, 84, (u64)fsf_req); } } @@ -1481,12 +1482,12 @@ zfcp_fsf_incoming_els(struct zfcp_fsf_req *fsf_req) zfcp_san_dbf_event_incoming_els(fsf_req); if (els_type == LS_PLOGI) - zfcp_fsf_incoming_els_plogi(adapter, status_buffer); + zfcp_fsf_incoming_els_plogi(fsf_req); else if (els_type == LS_LOGO) - zfcp_fsf_incoming_els_logo(adapter, status_buffer); + zfcp_fsf_incoming_els_logo(fsf_req); else if ((els_type & 0xffff0000) == LS_RSCN) /* we are only concerned with the command, not the length */ - zfcp_fsf_incoming_els_rscn(adapter, status_buffer); + zfcp_fsf_incoming_els_rscn(fsf_req); else zfcp_fsf_incoming_els_unknown(adapter, status_buffer); } diff --git a/drivers/s390/scsi/zfcp_ccw.c b/drivers/s390/scsi/zfcp_ccw.c index dd40b1c350b..8edd90583cd 100644 --- a/drivers/s390/scsi/zfcp_ccw.c +++ b/drivers/s390/scsi/zfcp_ccw.c @@ -172,7 +172,7 @@ zfcp_ccw_set_online(struct ccw_device *ccw_device) zfcp_erp_modify_adapter_status(adapter, 10, 0, ZFCP_STATUS_COMMON_RUNNING, ZFCP_SET); - zfcp_erp_adapter_reopen(adapter, ZFCP_STATUS_COMMON_ERP_FAILED); + zfcp_erp_adapter_reopen(adapter, ZFCP_STATUS_COMMON_ERP_FAILED, 85, 0); zfcp_erp_wait(adapter); goto out; @@ -197,7 +197,7 @@ zfcp_ccw_set_offline(struct ccw_device *ccw_device) down(&zfcp_data.config_sema); adapter = dev_get_drvdata(&ccw_device->dev); - zfcp_erp_adapter_shutdown(adapter, 0); + zfcp_erp_adapter_shutdown(adapter, 0, 86, 0); zfcp_erp_wait(adapter); zfcp_erp_thread_kill(adapter); up(&zfcp_data.config_sema); @@ -224,13 +224,13 @@ zfcp_ccw_notify(struct ccw_device *ccw_device, int event) ZFCP_LOG_NORMAL("adapter %s: device gone\n", zfcp_get_busid_by_adapter(adapter)); debug_text_event(adapter->erp_dbf,1,"dev_gone"); - zfcp_erp_adapter_shutdown(adapter, 0); + zfcp_erp_adapter_shutdown(adapter, 0, 87, 0); break; case CIO_NO_PATH: ZFCP_LOG_NORMAL("adapter %s: no path\n", zfcp_get_busid_by_adapter(adapter)); debug_text_event(adapter->erp_dbf,1,"no_path"); - zfcp_erp_adapter_shutdown(adapter, 0); + zfcp_erp_adapter_shutdown(adapter, 0, 88, 0); break; case CIO_OPER: ZFCP_LOG_NORMAL("adapter %s: operational again\n", @@ -240,7 +240,7 @@ zfcp_ccw_notify(struct ccw_device *ccw_device, int event) ZFCP_STATUS_COMMON_RUNNING, ZFCP_SET); zfcp_erp_adapter_reopen(adapter, - ZFCP_STATUS_COMMON_ERP_FAILED); + ZFCP_STATUS_COMMON_ERP_FAILED, 89, 0); break; } zfcp_erp_wait(adapter); @@ -272,7 +272,7 @@ zfcp_ccw_shutdown(struct ccw_device *cdev) down(&zfcp_data.config_sema); adapter = dev_get_drvdata(&cdev->dev); - zfcp_erp_adapter_shutdown(adapter, 0); + zfcp_erp_adapter_shutdown(adapter, 0, 90, 0); zfcp_erp_wait(adapter); up(&zfcp_data.config_sema); } diff --git a/drivers/s390/scsi/zfcp_dbf.c b/drivers/s390/scsi/zfcp_dbf.c index 2fcfe9bec55..f207b0bd0ca 100644 --- a/drivers/s390/scsi/zfcp_dbf.c +++ b/drivers/s390/scsi/zfcp_dbf.c @@ -523,6 +523,7 @@ static struct debug_view zfcp_hba_dbf_view = { static const char *zfcp_rec_dbf_tags[] = { [ZFCP_REC_DBF_ID_THREAD] = "thread", [ZFCP_REC_DBF_ID_TARGET] = "target", + [ZFCP_REC_DBF_ID_TRIGGER] = "trigger", }; static const char *zfcp_rec_dbf_ids[] = { @@ -587,6 +588,89 @@ static const char *zfcp_rec_dbf_ids[] = { [59] = "unit access denied open unit", [60] = "shared unit access denied open unit", [61] = "unit access denied fcp", + [62] = "request timeout", + [63] = "adisc link test reject or timeout", + [64] = "adisc link test d_id changed", + [65] = "adisc link test failed", + [66] = "recovery out of memory", + [67] = "adapter recovery repeated after state change", + [68] = "port recovery repeated after state change", + [69] = "unit recovery repeated after state change", + [70] = "port recovery follow-up after successful adapter recovery", + [71] = "adapter recovery escalation after failed adapter recovery", + [72] = "port recovery follow-up after successful physical port " + "recovery", + [73] = "adapter recovery escalation after failed physical port " + "recovery", + [74] = "unit recovery follow-up after successful port recovery", + [75] = "physical port recovery escalation after failed port " + "recovery", + [76] = "port recovery escalation after failed unit recovery", + [77] = "recovery opening nameserver port", + [78] = "duplicate request id", + [79] = "link down", + [80] = "exclusive read-only unit access unsupported", + [81] = "shared read-write unit access unsupported", + [82] = "incoming rscn", + [83] = "incoming plogi", + [84] = "incoming logo", + [85] = "online", + [86] = "offline", + [87] = "ccw device gone", + [88] = "ccw device no path", + [89] = "ccw device operational", + [90] = "ccw device shutdown", + [91] = "sysfs port addition", + [92] = "sysfs port removal", + [93] = "sysfs adapter recovery", + [94] = "sysfs unit addition", + [95] = "sysfs unit removal", + [96] = "sysfs port recovery", + [97] = "sysfs unit recovery", + [98] = "sequence number mismatch", + [99] = "link up", + [100] = "error state", + [101] = "status read physical port closed", + [102] = "link up status read", + [103] = "too many failed status read buffers", + [104] = "port handle not valid abort", + [105] = "lun handle not valid abort", + [106] = "port handle not valid ct", + [107] = "port handle not valid close port", + [108] = "port handle not valid close physical port", + [109] = "port handle not valid open unit", + [110] = "port handle not valid close unit", + [111] = "lun handle not valid close unit", + [112] = "port handle not valid fcp", + [113] = "lun handle not valid fcp", + [114] = "handle mismatch fcp", + [115] = "lun not valid fcp", + [116] = "qdio send failed", + [117] = "version mismatch", + [118] = "incompatible qtcb type", + [119] = "unknown protocol status", + [120] = "unknown fsf command", + [121] = "no recommendation for status qualifier", + [122] = "status read physical port closed in error", + [123] = "fc service class not supported ct", + [124] = "fc service class not supported els", + [125] = "need newer zfcp", + [126] = "need newer microcode", + [127] = "arbitrated loop not supported", + [128] = "unknown topology", + [129] = "qtcb size mismatch", + [130] = "unknown fsf status ecd", + [131] = "fcp request too big", + [132] = "fc service class not supported fcp", + [133] = "data direction not valid fcp", + [134] = "command length not valid fcp", + [135] = "status read act update", + [136] = "status read cfdc update", + [137] = "hbaapi port open", + [138] = "hbaapi unit open", + [139] = "hbaapi unit shutdown", + [140] = "qdio error", + [141] = "scsi host reset", }; static int zfcp_rec_dbf_view_format(debug_info_t *id, struct debug_view *view, @@ -613,6 +697,17 @@ static int zfcp_rec_dbf_view_format(debug_info_t *id, struct debug_view *view, zfcp_dbf_out(&p, "wwpn", "0x%016Lx", r->u.target.wwpn); zfcp_dbf_out(&p, "fcp_lun", "0x%016Lx", r->u.target.fcp_lun); break; + case ZFCP_REC_DBF_ID_TRIGGER: + zfcp_dbf_out(&p, "reference", "0x%016Lx", r->u.trigger.ref); + zfcp_dbf_out(&p, "erp_action", "0x%016Lx", r->u.trigger.action); + zfcp_dbf_out(&p, "requested", "%d", r->u.trigger.want); + zfcp_dbf_out(&p, "executed", "%d", r->u.trigger.need); + zfcp_dbf_out(&p, "wwpn", "0x%016Lx", r->u.trigger.wwpn); + zfcp_dbf_out(&p, "fcp_lun", "0x%016Lx", r->u.trigger.fcp_lun); + zfcp_dbf_out(&p, "adapter_status", "0x%08x", r->u.trigger.as); + zfcp_dbf_out(&p, "port_status", "0x%08x", r->u.trigger.ps); + zfcp_dbf_out(&p, "unit_status", "0x%08x", r->u.trigger.us); + break; } sprintf(p, "\n"); return (p - buf) + 1; @@ -727,6 +822,45 @@ void zfcp_rec_dbf_event_unit(u8 id, u64 ref, struct zfcp_unit *unit) unit->fcp_lun); } +/** + * zfcp_rec_dbf_event_trigger - trace event for triggered error recovery + * @id2: identifier for error recovery trigger + * @ref: additional reference (e.g. request) + * @want: originally requested error recovery action + * @need: error recovery action actually initiated + * @action: address of error recovery action struct + * @adapter: adapter + * @port: port + * @unit: unit + */ +void zfcp_rec_dbf_event_trigger(u8 id2, u64 ref, u8 want, u8 need, u64 action, + struct zfcp_adapter *adapter, + struct zfcp_port *port, struct zfcp_unit *unit) +{ + struct zfcp_rec_dbf_record *r = &adapter->rec_dbf_buf; + unsigned long flags; + + spin_lock_irqsave(&adapter->rec_dbf_lock, flags); + memset(r, 0, sizeof(*r)); + r->id = ZFCP_REC_DBF_ID_TRIGGER; + r->id2 = id2; + r->u.trigger.ref = ref; + r->u.trigger.want = want; + r->u.trigger.need = need; + r->u.trigger.action = action; + r->u.trigger.as = atomic_read(&adapter->status); + if (port) { + r->u.trigger.ps = atomic_read(&port->status); + r->u.trigger.wwpn = port->wwpn; + } + if (unit) { + r->u.trigger.us = atomic_read(&unit->status); + r->u.trigger.fcp_lun = unit->fcp_lun; + } + debug_event(adapter->rec_dbf, action ? 1 : 4, r, sizeof(*r)); + spin_unlock_irqrestore(&adapter->rec_dbf_lock, flags); +} + static void _zfcp_san_dbf_event_common_ct(const char *tag, struct zfcp_fsf_req *fsf_req, u32 s_id, u32 d_id, void *buffer, int buflen) diff --git a/drivers/s390/scsi/zfcp_def.h b/drivers/s390/scsi/zfcp_def.h index 659ec739706..aad75cfa1c6 100644 --- a/drivers/s390/scsi/zfcp_def.h +++ b/drivers/s390/scsi/zfcp_def.h @@ -295,18 +295,32 @@ struct zfcp_rec_dbf_record_target { u32 erp_count; } __attribute__ ((packed)); +struct zfcp_rec_dbf_record_trigger { + u8 want; + u8 need; + u32 as; + u32 ps; + u32 us; + u64 ref; + u64 action; + u64 wwpn; + u64 fcp_lun; +} __attribute__ ((packed)); + struct zfcp_rec_dbf_record { u8 id; u8 id2; union { struct zfcp_rec_dbf_record_thread thread; struct zfcp_rec_dbf_record_target target; + struct zfcp_rec_dbf_record_trigger trigger; } u; } __attribute__ ((packed)); enum { ZFCP_REC_DBF_ID_THREAD, ZFCP_REC_DBF_ID_TARGET, + ZFCP_REC_DBF_ID_TRIGGER, }; struct zfcp_hba_dbf_record_response { diff --git a/drivers/s390/scsi/zfcp_erp.c b/drivers/s390/scsi/zfcp_erp.c index bee03443efd..55e034b10de 100644 --- a/drivers/s390/scsi/zfcp_erp.c +++ b/drivers/s390/scsi/zfcp_erp.c @@ -26,13 +26,16 @@ static int zfcp_erp_adisc(struct zfcp_port *); static void zfcp_erp_adisc_handler(unsigned long); -static int zfcp_erp_adapter_reopen_internal(struct zfcp_adapter *, int); -static int zfcp_erp_port_forced_reopen_internal(struct zfcp_port *, int); -static int zfcp_erp_port_reopen_internal(struct zfcp_port *, int); -static int zfcp_erp_unit_reopen_internal(struct zfcp_unit *, int); +static int zfcp_erp_adapter_reopen_internal(struct zfcp_adapter *, int, u8, + u64); +static int zfcp_erp_port_forced_reopen_internal(struct zfcp_port *, int, u8, + u64); +static int zfcp_erp_port_reopen_internal(struct zfcp_port *, int, u8, u64); +static int zfcp_erp_unit_reopen_internal(struct zfcp_unit *, int, u8, u64); -static int zfcp_erp_port_reopen_all_internal(struct zfcp_adapter *, int); -static int zfcp_erp_unit_reopen_all_internal(struct zfcp_port *, int); +static int zfcp_erp_port_reopen_all_internal(struct zfcp_adapter *, int, u8, + u64); +static int zfcp_erp_unit_reopen_all_internal(struct zfcp_port *, int, u8, u64); static void zfcp_erp_adapter_block(struct zfcp_adapter *, int); static void zfcp_erp_adapter_unblock(struct zfcp_adapter *); @@ -97,7 +100,8 @@ static void zfcp_erp_action_dismiss_unit(struct zfcp_unit *); static void zfcp_erp_action_dismiss(struct zfcp_erp_action *); static int zfcp_erp_action_enqueue(int, struct zfcp_adapter *, - struct zfcp_port *, struct zfcp_unit *); + struct zfcp_port *, struct zfcp_unit *, + u8 id, u64 ref); static int zfcp_erp_action_dequeue(struct zfcp_erp_action *); static void zfcp_erp_action_cleanup(int, struct zfcp_adapter *, struct zfcp_port *, struct zfcp_unit *, @@ -179,7 +183,7 @@ static void zfcp_close_fsf(struct zfcp_adapter *adapter) static void zfcp_fsf_request_timeout_handler(unsigned long data) { struct zfcp_adapter *adapter = (struct zfcp_adapter *) data; - zfcp_erp_adapter_reopen(adapter, ZFCP_STATUS_COMMON_ERP_FAILED); + zfcp_erp_adapter_reopen(adapter, ZFCP_STATUS_COMMON_ERP_FAILED, 62, 0); } void zfcp_fsf_start_timer(struct zfcp_fsf_req *fsf_req, unsigned long timeout) @@ -200,8 +204,8 @@ void zfcp_fsf_start_timer(struct zfcp_fsf_req *fsf_req, unsigned long timeout) * returns: 0 - initiated action successfully * <0 - failed to initiate action */ -static int -zfcp_erp_adapter_reopen_internal(struct zfcp_adapter *adapter, int clear_mask) +static int zfcp_erp_adapter_reopen_internal(struct zfcp_adapter *adapter, + int clear_mask, u8 id, u64 ref) { int retval; @@ -221,7 +225,7 @@ zfcp_erp_adapter_reopen_internal(struct zfcp_adapter *adapter, int clear_mask) goto out; } retval = zfcp_erp_action_enqueue(ZFCP_ERP_ACTION_REOPEN_ADAPTER, - adapter, NULL, NULL); + adapter, NULL, NULL, id, ref); out: return retval; @@ -236,56 +240,56 @@ zfcp_erp_adapter_reopen_internal(struct zfcp_adapter *adapter, int clear_mask) * returns: 0 - initiated action successfully * <0 - failed to initiate action */ -int -zfcp_erp_adapter_reopen(struct zfcp_adapter *adapter, int clear_mask) +int zfcp_erp_adapter_reopen(struct zfcp_adapter *adapter, int clear_mask, + u8 id, u64 ref) { int retval; unsigned long flags; read_lock_irqsave(&zfcp_data.config_lock, flags); write_lock(&adapter->erp_lock); - retval = zfcp_erp_adapter_reopen_internal(adapter, clear_mask); + retval = zfcp_erp_adapter_reopen_internal(adapter, clear_mask, id, ref); write_unlock(&adapter->erp_lock); read_unlock_irqrestore(&zfcp_data.config_lock, flags); return retval; } -int -zfcp_erp_adapter_shutdown(struct zfcp_adapter *adapter, int clear_mask) +int zfcp_erp_adapter_shutdown(struct zfcp_adapter *adapter, int clear_mask, + u8 id, u64 ref) { int retval; retval = zfcp_erp_adapter_reopen(adapter, ZFCP_STATUS_COMMON_RUNNING | ZFCP_STATUS_COMMON_ERP_FAILED | - clear_mask); + clear_mask, id, ref); return retval; } -int -zfcp_erp_port_shutdown(struct zfcp_port *port, int clear_mask) +int zfcp_erp_port_shutdown(struct zfcp_port *port, int clear_mask, u8 id, + u64 ref) { int retval; retval = zfcp_erp_port_reopen(port, ZFCP_STATUS_COMMON_RUNNING | ZFCP_STATUS_COMMON_ERP_FAILED | - clear_mask); + clear_mask, id, ref); return retval; } -int -zfcp_erp_unit_shutdown(struct zfcp_unit *unit, int clear_mask) +int zfcp_erp_unit_shutdown(struct zfcp_unit *unit, int clear_mask, u8 id, + u64 ref) { int retval; retval = zfcp_erp_unit_reopen(unit, ZFCP_STATUS_COMMON_RUNNING | ZFCP_STATUS_COMMON_ERP_FAILED | - clear_mask); + clear_mask, id, ref); return retval; } @@ -400,7 +404,7 @@ zfcp_erp_adisc_handler(unsigned long data) "(adapter %s, port d_id=0x%06x)\n", zfcp_get_busid_by_adapter(adapter), d_id); debug_text_event(adapter->erp_dbf, 3, "forcreop"); - if (zfcp_erp_port_forced_reopen(port, 0)) + if (zfcp_erp_port_forced_reopen(port, 0, 63, 0)) ZFCP_LOG_NORMAL("failed reopen of port " "(adapter %s, wwpn=0x%016Lx)\n", zfcp_get_busid_by_port(port), @@ -427,7 +431,7 @@ zfcp_erp_adisc_handler(unsigned long data) "adisc_resp_wwpn=0x%016Lx)\n", zfcp_get_busid_by_port(port), port->wwpn, (wwn_t) adisc->wwpn); - if (zfcp_erp_port_reopen(port, 0)) + if (zfcp_erp_port_reopen(port, 0, 64, 0)) ZFCP_LOG_NORMAL("failed reopen of port " "(adapter %s, wwpn=0x%016Lx)\n", zfcp_get_busid_by_port(port), @@ -461,7 +465,7 @@ zfcp_test_link(struct zfcp_port *port) ZFCP_LOG_NORMAL("reopen needed for port 0x%016Lx " "on adapter %s\n ", port->wwpn, zfcp_get_busid_by_port(port)); - retval = zfcp_erp_port_forced_reopen(port, 0); + retval = zfcp_erp_port_forced_reopen(port, 0, 65, 0); if (retval != 0) { ZFCP_LOG_NORMAL("reopen of remote port 0x%016Lx " "on adapter %s failed\n", port->wwpn, @@ -484,8 +488,8 @@ zfcp_test_link(struct zfcp_port *port) * returns: 0 - initiated action successfully * <0 - failed to initiate action */ -static int -zfcp_erp_port_forced_reopen_internal(struct zfcp_port *port, int clear_mask) +static int zfcp_erp_port_forced_reopen_internal(struct zfcp_port *port, + int clear_mask, u8 id, u64 ref) { int retval; struct zfcp_adapter *adapter = port->adapter; @@ -509,7 +513,7 @@ zfcp_erp_port_forced_reopen_internal(struct zfcp_port *port, int clear_mask) } retval = zfcp_erp_action_enqueue(ZFCP_ERP_ACTION_REOPEN_PORT_FORCED, - port->adapter, port, NULL); + port->adapter, port, NULL, id, ref); out: return retval; @@ -524,8 +528,8 @@ zfcp_erp_port_forced_reopen_internal(struct zfcp_port *port, int clear_mask) * returns: 0 - initiated action successfully * <0 - failed to initiate action */ -int -zfcp_erp_port_forced_reopen(struct zfcp_port *port, int clear_mask) +int zfcp_erp_port_forced_reopen(struct zfcp_port *port, int clear_mask, u8 id, + u64 ref) { int retval; unsigned long flags; @@ -534,7 +538,8 @@ zfcp_erp_port_forced_reopen(struct zfcp_port *port, int clear_mask) adapter = port->adapter; read_lock_irqsave(&zfcp_data.config_lock, flags); write_lock(&adapter->erp_lock); - retval = zfcp_erp_port_forced_reopen_internal(port, clear_mask); + retval = zfcp_erp_port_forced_reopen_internal(port, clear_mask, id, + ref); write_unlock(&adapter->erp_lock); read_unlock_irqrestore(&zfcp_data.config_lock, flags); @@ -551,8 +556,8 @@ zfcp_erp_port_forced_reopen(struct zfcp_port *port, int clear_mask) * returns: 0 - initiated action successfully * <0 - failed to initiate action */ -static int -zfcp_erp_port_reopen_internal(struct zfcp_port *port, int clear_mask) +static int zfcp_erp_port_reopen_internal(struct zfcp_port *port, int clear_mask, + u8 id, u64 ref) { int retval; struct zfcp_adapter *adapter = port->adapter; @@ -578,7 +583,7 @@ zfcp_erp_port_reopen_internal(struct zfcp_port *port, int clear_mask) } retval = zfcp_erp_action_enqueue(ZFCP_ERP_ACTION_REOPEN_PORT, - port->adapter, port, NULL); + port->adapter, port, NULL, id, ref); out: return retval; @@ -594,8 +599,7 @@ zfcp_erp_port_reopen_internal(struct zfcp_port *port, int clear_mask) * correct locking. An error recovery task is initiated to do the reopen. * To wait for the completion of the reopen zfcp_erp_wait should be used. */ -int -zfcp_erp_port_reopen(struct zfcp_port *port, int clear_mask) +int zfcp_erp_port_reopen(struct zfcp_port *port, int clear_mask, u8 id, u64 ref) { int retval; unsigned long flags; @@ -603,7 +607,7 @@ zfcp_erp_port_reopen(struct zfcp_port *port, int clear_mask) read_lock_irqsave(&zfcp_data.config_lock, flags); write_lock(&adapter->erp_lock); - retval = zfcp_erp_port_reopen_internal(port, clear_mask); + retval = zfcp_erp_port_reopen_internal(port, clear_mask, id, ref); write_unlock(&adapter->erp_lock); read_unlock_irqrestore(&zfcp_data.config_lock, flags); @@ -620,8 +624,8 @@ zfcp_erp_port_reopen(struct zfcp_port *port, int clear_mask) * returns: 0 - initiated action successfully * <0 - failed to initiate action */ -static int -zfcp_erp_unit_reopen_internal(struct zfcp_unit *unit, int clear_mask) +static int zfcp_erp_unit_reopen_internal(struct zfcp_unit *unit, int clear_mask, + u8 id, u64 ref) { int retval; struct zfcp_adapter *adapter = unit->port->adapter; @@ -647,7 +651,7 @@ zfcp_erp_unit_reopen_internal(struct zfcp_unit *unit, int clear_mask) } retval = zfcp_erp_action_enqueue(ZFCP_ERP_ACTION_REOPEN_UNIT, - unit->port->adapter, unit->port, unit); + adapter, unit->port, unit, id, ref); out: return retval; } @@ -662,8 +666,7 @@ zfcp_erp_unit_reopen_internal(struct zfcp_unit *unit, int clear_mask) * locking. An error recovery task is initiated to do the reopen. * To wait for the completion of the reopen zfcp_erp_wait should be used. */ -int -zfcp_erp_unit_reopen(struct zfcp_unit *unit, int clear_mask) +int zfcp_erp_unit_reopen(struct zfcp_unit *unit, int clear_mask, u8 id, u64 ref) { int retval; unsigned long flags; @@ -675,7 +678,7 @@ zfcp_erp_unit_reopen(struct zfcp_unit *unit, int clear_mask) read_lock_irqsave(&zfcp_data.config_lock, flags); write_lock(&adapter->erp_lock); - retval = zfcp_erp_unit_reopen_internal(unit, clear_mask); + retval = zfcp_erp_unit_reopen_internal(unit, clear_mask, id, ref); write_unlock(&adapter->erp_lock); read_unlock_irqrestore(&zfcp_data.config_lock, flags); @@ -1215,7 +1218,7 @@ zfcp_erp_strategy(struct zfcp_erp_action *erp_action) "restarting I/O on adapter %s " "to free mempool\n", zfcp_get_busid_by_adapter(adapter)); - zfcp_erp_adapter_reopen_internal(adapter, 0); + zfcp_erp_adapter_reopen_internal(adapter, 0, 66, 0); } else { debug_text_event(adapter->erp_dbf, 2, "a_st_memw"); retval = zfcp_erp_strategy_memwait(erp_action); @@ -1499,7 +1502,9 @@ zfcp_erp_strategy_statechange(int action, case ZFCP_ERP_ACTION_REOPEN_ADAPTER: if (zfcp_erp_strategy_statechange_detected(&adapter->status, status)) { - zfcp_erp_adapter_reopen_internal(adapter, ZFCP_STATUS_COMMON_ERP_FAILED); + zfcp_erp_adapter_reopen_internal(adapter, + ZFCP_STATUS_COMMON_ERP_FAILED, + 67, 0); retval = ZFCP_ERP_EXIT; } break; @@ -1508,7 +1513,9 @@ zfcp_erp_strategy_statechange(int action, case ZFCP_ERP_ACTION_REOPEN_PORT: if (zfcp_erp_strategy_statechange_detected(&port->status, status)) { - zfcp_erp_port_reopen_internal(port, ZFCP_STATUS_COMMON_ERP_FAILED); + zfcp_erp_port_reopen_internal(port, + ZFCP_STATUS_COMMON_ERP_FAILED, + 68, 0); retval = ZFCP_ERP_EXIT; } break; @@ -1516,7 +1523,9 @@ zfcp_erp_strategy_statechange(int action, case ZFCP_ERP_ACTION_REOPEN_UNIT: if (zfcp_erp_strategy_statechange_detected(&unit->status, status)) { - zfcp_erp_unit_reopen_internal(unit, ZFCP_STATUS_COMMON_ERP_FAILED); + zfcp_erp_unit_reopen_internal(unit, + ZFCP_STATUS_COMMON_ERP_FAILED, + 69, 0); retval = ZFCP_ERP_EXIT; } break; @@ -1700,29 +1709,29 @@ zfcp_erp_strategy_followup_actions(int action, case ZFCP_ERP_ACTION_REOPEN_ADAPTER: if (status == ZFCP_ERP_SUCCEEDED) - zfcp_erp_port_reopen_all_internal(adapter, 0); + zfcp_erp_port_reopen_all_internal(adapter, 0, 70, 0); else - zfcp_erp_adapter_reopen_internal(adapter, 0); + zfcp_erp_adapter_reopen_internal(adapter, 0, 71, 0); break; case ZFCP_ERP_ACTION_REOPEN_PORT_FORCED: if (status == ZFCP_ERP_SUCCEEDED) - zfcp_erp_port_reopen_internal(port, 0); + zfcp_erp_port_reopen_internal(port, 0, 72, 0); else - zfcp_erp_adapter_reopen_internal(adapter, 0); + zfcp_erp_adapter_reopen_internal(adapter, 0, 73, 0); break; case ZFCP_ERP_ACTION_REOPEN_PORT: if (status == ZFCP_ERP_SUCCEEDED) - zfcp_erp_unit_reopen_all_internal(port, 0); + zfcp_erp_unit_reopen_all_internal(port, 0, 74, 0); else - zfcp_erp_port_forced_reopen_internal(port, 0); + zfcp_erp_port_forced_reopen_internal(port, 0, 75, 0); break; case ZFCP_ERP_ACTION_REOPEN_UNIT: /* Nothing to do if status == ZFCP_ERP_SUCCEEDED */ if (status != ZFCP_ERP_SUCCEEDED) - zfcp_erp_port_reopen_internal(unit->port, 0); + zfcp_erp_port_reopen_internal(unit->port, 0, 76, 0); break; } @@ -1863,30 +1872,32 @@ void zfcp_erp_modify_unit_status(struct zfcp_unit *unit, u8 id, u64 ref, * returns: 0 - initiated action successfully * <0 - failed to initiate action */ -int -zfcp_erp_port_reopen_all(struct zfcp_adapter *adapter, int clear_mask) +int zfcp_erp_port_reopen_all(struct zfcp_adapter *adapter, int clear_mask, + u8 id, u64 ref) { int retval; unsigned long flags; read_lock_irqsave(&zfcp_data.config_lock, flags); write_lock(&adapter->erp_lock); - retval = zfcp_erp_port_reopen_all_internal(adapter, clear_mask); + retval = zfcp_erp_port_reopen_all_internal(adapter, clear_mask, id, + ref); write_unlock(&adapter->erp_lock); read_unlock_irqrestore(&zfcp_data.config_lock, flags); return retval; } -static int -zfcp_erp_port_reopen_all_internal(struct zfcp_adapter *adapter, int clear_mask) +static int zfcp_erp_port_reopen_all_internal(struct zfcp_adapter *adapter, + int clear_mask, u8 id, u64 ref) { int retval = 0; struct zfcp_port *port; list_for_each_entry(port, &adapter->port_list_head, list) if (!atomic_test_mask(ZFCP_STATUS_PORT_WKA, &port->status)) - zfcp_erp_port_reopen_internal(port, clear_mask); + zfcp_erp_port_reopen_internal(port, clear_mask, id, + ref); return retval; } @@ -1898,14 +1909,14 @@ zfcp_erp_port_reopen_all_internal(struct zfcp_adapter *adapter, int clear_mask) * * returns: FIXME */ -static int -zfcp_erp_unit_reopen_all_internal(struct zfcp_port *port, int clear_mask) +static int zfcp_erp_unit_reopen_all_internal(struct zfcp_port *port, + int clear_mask, u8 id, u64 ref) { int retval = 0; struct zfcp_unit *unit; list_for_each_entry(unit, &port->unit_list_head, list) - zfcp_erp_unit_reopen_internal(unit, clear_mask); + zfcp_erp_unit_reopen_internal(unit, clear_mask, id, ref); return retval; } @@ -2466,8 +2477,8 @@ zfcp_erp_port_strategy_open_common(struct zfcp_erp_action *erp_action) /* nameserver port may live again */ atomic_set_mask(ZFCP_STATUS_COMMON_RUNNING, &adapter->nameserver_port->status); - if (zfcp_erp_port_reopen(adapter->nameserver_port, 0) - >= 0) { + if (zfcp_erp_port_reopen(adapter->nameserver_port, 0, + 77, (u64)erp_action) >= 0) { erp_action->step = ZFCP_ERP_STEP_NAMESERVER_OPEN; retval = ZFCP_ERP_CONTINUES; @@ -2963,14 +2974,12 @@ void zfcp_erp_start_timer(struct zfcp_fsf_req *fsf_req) * * returns: */ -static int -zfcp_erp_action_enqueue(int action, - struct zfcp_adapter *adapter, - struct zfcp_port *port, struct zfcp_unit *unit) +static int zfcp_erp_action_enqueue(int want, struct zfcp_adapter *adapter, + struct zfcp_port *port, + struct zfcp_unit *unit, u8 id, u64 ref) { - int retval = 1; + int retval = 1, need = want; struct zfcp_erp_action *erp_action = NULL; - int stronger_action = 0; u32 status = 0; /* @@ -2989,9 +2998,9 @@ zfcp_erp_action_enqueue(int action, &adapter->status)) return -EIO; - debug_event(adapter->erp_dbf, 4, &action, sizeof (int)); + debug_event(adapter->erp_dbf, 4, &want, sizeof (int)); /* check whether we really need this */ - switch (action) { + switch (want) { case ZFCP_ERP_ACTION_REOPEN_UNIT: if (atomic_test_mask (ZFCP_STATUS_COMMON_ERP_INUSE, &unit->status)) { @@ -3009,10 +3018,8 @@ zfcp_erp_action_enqueue(int action, goto out; } if (!atomic_test_mask - (ZFCP_STATUS_COMMON_UNBLOCKED, &port->status)) { - stronger_action = ZFCP_ERP_ACTION_REOPEN_PORT; - unit = NULL; - } + (ZFCP_STATUS_COMMON_UNBLOCKED, &port->status)) + need = ZFCP_ERP_ACTION_REOPEN_PORT; /* fall through !!! */ case ZFCP_ERP_ACTION_REOPEN_PORT: @@ -3032,7 +3039,7 @@ zfcp_erp_action_enqueue(int action, ZFCP_ERP_ACTION_REOPEN_PORT_FORCED) { ZFCP_LOG_INFO("dropped erp action %i (port " "0x%016Lx, action in use: %i)\n", - action, port->wwpn, + want, port->wwpn, port->erp_action.action); debug_text_event(adapter->erp_dbf, 4, "pf_actenq_drp"); @@ -3050,10 +3057,8 @@ zfcp_erp_action_enqueue(int action, goto out; } if (!atomic_test_mask - (ZFCP_STATUS_COMMON_UNBLOCKED, &adapter->status)) { - stronger_action = ZFCP_ERP_ACTION_REOPEN_ADAPTER; - port = NULL; - } + (ZFCP_STATUS_COMMON_UNBLOCKED, &adapter->status)) + need = ZFCP_ERP_ACTION_REOPEN_ADAPTER; /* fall through !!! */ case ZFCP_ERP_ACTION_REOPEN_ADAPTER: @@ -3066,30 +3071,28 @@ zfcp_erp_action_enqueue(int action, default: debug_text_exception(adapter->erp_dbf, 1, "a_actenq_bug"); - debug_event(adapter->erp_dbf, 1, &action, sizeof (int)); + debug_event(adapter->erp_dbf, 1, &want, sizeof (int)); ZFCP_LOG_NORMAL("bug: unknown erp action requested " "on adapter %s (action=%d)\n", - zfcp_get_busid_by_adapter(adapter), action); + zfcp_get_busid_by_adapter(adapter), want); goto out; } /* check whether we need something stronger first */ - if (stronger_action) { + if (need) { debug_text_event(adapter->erp_dbf, 4, "a_actenq_str"); - debug_event(adapter->erp_dbf, 4, &stronger_action, + debug_event(adapter->erp_dbf, 4, &need, sizeof (int)); ZFCP_LOG_DEBUG("stronger erp action %d needed before " "erp action %d on adapter %s\n", - stronger_action, action, - zfcp_get_busid_by_adapter(adapter)); - action = stronger_action; + need, want, zfcp_get_busid_by_adapter(adapter)); } /* mark adapter to have some error recovery pending */ atomic_set_mask(ZFCP_STATUS_ADAPTER_ERP_PENDING, &adapter->status); /* setup error recovery action */ - switch (action) { + switch (need) { case ZFCP_ERP_ACTION_REOPEN_UNIT: zfcp_unit_get(unit); @@ -3128,7 +3131,7 @@ zfcp_erp_action_enqueue(int action, erp_action->adapter = adapter; erp_action->port = port; erp_action->unit = unit; - erp_action->action = action; + erp_action->action = need; erp_action->status = status; ++adapter->erp_total_count; @@ -3139,6 +3142,8 @@ zfcp_erp_action_enqueue(int action, zfcp_rec_dbf_event_thread(1, adapter, 0); retval = 0; out: + zfcp_rec_dbf_event_trigger(id, ref, want, need, (u64)erp_action, + adapter, port, unit); return retval; } @@ -3322,7 +3327,7 @@ void zfcp_erp_port_boxed(struct zfcp_port *port, u8 id, u64 ref) zfcp_erp_modify_port_status(port, id, ref, ZFCP_STATUS_COMMON_ACCESS_BOXED, ZFCP_SET); read_unlock_irqrestore(&zfcp_data.config_lock, flags); - zfcp_erp_port_reopen(port, ZFCP_STATUS_COMMON_ERP_FAILED); + zfcp_erp_port_reopen(port, ZFCP_STATUS_COMMON_ERP_FAILED, id, ref); } void zfcp_erp_unit_boxed(struct zfcp_unit *unit, u8 id, u64 ref) @@ -3333,7 +3338,7 @@ v |