From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christof Schmitt Subject: [patch 09/13] zfcp: deadlock via shared work queue tasks Date: Wed, 01 Oct 2008 12:42:22 +0200 Message-ID: <20081001104803.099733000@de.ibm.com> References: <20081001104213.585200000@de.ibm.com> Return-path: Received: from mtagate8.de.ibm.com ([195.212.29.157]:36001 "EHLO mtagate8.de.ibm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752476AbYJAKsg (ORCPT ); Wed, 1 Oct 2008 06:48:36 -0400 Content-Disposition: inline; filename=zfcp_workqueue_race.diff Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: James Bottomley Cc: linux-scsi@vger.kernel.org, linux-s390@vger.kernel.org, schwidefsky@de.ibm.com, heiko.carstens@de.ibm.com, Swen Schillig , Christof Schmitt From: Swen Schillig Each adapter reopen trigger automatically a scan_port task which is waiting for the ERP to be finished before further processing. Since the initial device setup enqueues adapter, port and LUN which are individual ERP actions, this process would start after everything is done. Unfortunately the port_reopen requires another scheduled work to be finished which is queued after the automatic scan_port -> deadlock ! This fix creates an own work queue for ERP based nameserver requests. Signed-off-by: Swen Schillig Signed-off-by: Christof Schmitt --- drivers/s390/scsi/zfcp_aux.c | 2 ++ drivers/s390/scsi/zfcp_def.h | 1 + drivers/s390/scsi/zfcp_erp.c | 4 ++-- drivers/s390/scsi/zfcp_fsf.c | 2 +- 4 files changed, 6 insertions(+), 3 deletions(-) --- a/drivers/s390/scsi/zfcp_aux.c 2008-10-01 11:09:06.000000000 +0200 +++ b/drivers/s390/scsi/zfcp_aux.c 2008-10-01 11:09:57.000000000 +0200 @@ -168,6 +168,8 @@ static int __init zfcp_module_init(void) if (!zfcp_data.gid_pn_cache) goto out_gid_cache; + zfcp_data.work_queue = create_singlethread_workqueue("zfcp_wq"); + INIT_LIST_HEAD(&zfcp_data.adapter_list_head); sema_init(&zfcp_data.config_sema, 1); rwlock_init(&zfcp_data.config_lock); --- a/drivers/s390/scsi/zfcp_def.h 2008-10-01 11:09:06.000000000 +0200 +++ b/drivers/s390/scsi/zfcp_def.h 2008-10-01 11:09:57.000000000 +0200 @@ -602,6 +602,7 @@ struct zfcp_data { struct kmem_cache *fsf_req_qtcb_cache; struct kmem_cache *sr_buffer_cache; struct kmem_cache *gid_pn_cache; + struct workqueue_struct *work_queue; }; /* struct used by memory pools for fsf_requests */ --- a/drivers/s390/scsi/zfcp_erp.c 2008-10-01 11:09:06.000000000 +0200 +++ b/drivers/s390/scsi/zfcp_erp.c 2008-10-01 11:09:57.000000000 +0200 @@ -869,7 +869,7 @@ static int zfcp_erp_port_strategy_open_c if (fc_host_port_type(adapter->scsi_host) == FC_PORTTYPE_PTP) return zfcp_erp_open_ptp_port(act); if (!(p_status & ZFCP_STATUS_PORT_DID_DID)) { - schedule_work(&port->gid_pn_work); + queue_work(zfcp_data.work_queue, &port->gid_pn_work); return ZFCP_ERP_CONTINUES; } case ZFCP_ERP_STEP_NAMESERVER_LOOKUP: @@ -1209,7 +1209,7 @@ static void zfcp_erp_schedule_work(struc atomic_set_mask(ZFCP_STATUS_UNIT_SCSI_WORK_PENDING, &unit->status); INIT_WORK(&p->work, zfcp_erp_scsi_scan); p->unit = unit; - schedule_work(&p->work); + queue_work(zfcp_data.work_queue, &p->work); } static void zfcp_erp_rport_register(struct zfcp_port *port) --- a/drivers/s390/scsi/zfcp_fsf.c 2008-10-01 11:09:31.000000000 +0200 +++ b/drivers/s390/scsi/zfcp_fsf.c 2008-10-01 11:09:57.000000000 +0200 @@ -329,7 +329,7 @@ static void zfcp_fsf_status_read_handler zfcp_fsf_req_free(req); atomic_inc(&adapter->stat_miss); - schedule_work(&adapter->stat_work); + queue_work(zfcp_data.work_queue, &adapter->stat_work); } static void zfcp_fsf_fsfstatus_qual_eval(struct zfcp_fsf_req *req) --