From 05f1d8ed03f547054efbc4d29bb7991c958ede95 Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Fri, 21 Jul 2023 21:36:44 +0200 Subject: s390/dasd: fix hanging device after quiesce/resume Quiesce and resume are functions that tell the DASD driver to stop/resume issuing I/Os to a specific DASD. On resume dasd_schedule_block_bh() is called to kick handling of IO requests again. This does unfortunately not cover internal requests which are used for path verification for example. This could lead to a hanging device when a path event or anything else that triggers internal requests occurs on a quiesced device. Fix by also calling dasd_schedule_device_bh() which triggers handling of internal requests on resume. Fixes: 8e09f21574ea ("[S390] dasd: add hyper PAV support to DASD device driver, part 1") Cc: stable@vger.kernel.org Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Link: https://lore.kernel.org/r/20230721193647.3889634-2-sth@linux.ibm.com Signed-off-by: Jens Axboe --- drivers/s390/block/dasd_ioctl.c | 1 + 1 file changed, 1 insertion(+) (limited to 'drivers/s390') diff --git a/drivers/s390/block/dasd_ioctl.c b/drivers/s390/block/dasd_ioctl.c index 513a7e6eee63..d55862605b82 100644 --- a/drivers/s390/block/dasd_ioctl.c +++ b/drivers/s390/block/dasd_ioctl.c @@ -131,6 +131,7 @@ static int dasd_ioctl_resume(struct dasd_block *block) spin_unlock_irqrestore(get_ccwdev_lock(base->cdev), flags); dasd_schedule_block_bh(block); + dasd_schedule_device_bh(base); return 0; } -- cgit From acea28a6b74f458defda7417d2217b051ba7d444 Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Fri, 21 Jul 2023 21:36:45 +0200 Subject: s390/dasd: use correct number of retries for ERP requests If a DASD request fails an error recovery procedure (ERP) request might be built as a copy of the original request to do error recovery. The ERP request gets a number of retries assigned. This number is always 256 no matter what other value might have been set for the original request. This is not what is expected when a user specifies a certain amount of retries for the device via sysfs. Correctly use the number of retries of the original request for ERP requests. Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Link: https://lore.kernel.org/r/20230721193647.3889634-3-sth@linux.ibm.com Signed-off-by: Jens Axboe --- drivers/s390/block/dasd_3990_erp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/s390') diff --git a/drivers/s390/block/dasd_3990_erp.c b/drivers/s390/block/dasd_3990_erp.c index 9fd36c468706..91e9a17b848e 100644 --- a/drivers/s390/block/dasd_3990_erp.c +++ b/drivers/s390/block/dasd_3990_erp.c @@ -2441,7 +2441,7 @@ static struct dasd_ccw_req *dasd_3990_erp_add_erp(struct dasd_ccw_req *cqr) erp->block = cqr->block; erp->magic = cqr->magic; erp->expires = cqr->expires; - erp->retries = 256; + erp->retries = device->default_retries; erp->buildclk = get_tod_clock(); erp->status = DASD_CQR_FILLED; -- cgit From 8a2278ce9c25048d999fe1a3561def75d963f471 Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Fri, 21 Jul 2023 21:36:46 +0200 Subject: s390/dasd: fix hanging device after request requeue The DASD device driver has a function to requeue requests to the blocklayer. This function is used in various cases when basic settings for the device have to be changed like High Performance Ficon related parameters or copy pair settings. The functions iterates over the device->ccw_queue and also removes the requests from the block->ccw_queue. In case the device is started on an alias device instead of the base device it might be removed from the block->ccw_queue without having it canceled properly before. This might lead to a hanging device since the request is no longer on a queue and can not be handled properly. Fix by iterating over the block->ccw_queue instead of the device->ccw_queue. This will take care of all blocklayer related requests and handle them on all associated DASD devices. Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Link: https://lore.kernel.org/r/20230721193647.3889634-4-sth@linux.ibm.com Signed-off-by: Jens Axboe --- drivers/s390/block/dasd.c | 125 ++++++++++++++++++---------------------------- 1 file changed, 48 insertions(+), 77 deletions(-) (limited to 'drivers/s390') diff --git a/drivers/s390/block/dasd.c b/drivers/s390/block/dasd.c index edcbf77852c3..50a5ff70814a 100644 --- a/drivers/s390/block/dasd.c +++ b/drivers/s390/block/dasd.c @@ -2943,41 +2943,32 @@ static void _dasd_wake_block_flush_cb(struct dasd_ccw_req *cqr, void *data) * Requeue a request back to the block request queue * only works for block requests */ -static int _dasd_requeue_request(struct dasd_ccw_req *cqr) +static void _dasd_requeue_request(struct dasd_ccw_req *cqr) { - struct dasd_block *block = cqr->block; struct request *req; - if (!block) - return -EINVAL; /* * If the request is an ERP request there is nothing to requeue. * This will be done with the remaining original request. */ if (cqr->refers) - return 0; + return; spin_lock_irq(&cqr->dq->lock); req = (struct request *) cqr->callback_data; blk_mq_requeue_request(req, true); spin_unlock_irq(&cqr->dq->lock); - return 0; + return; } -/* - * Go through all request on the dasd_block request queue, cancel them - * on the respective dasd_device, and return them to the generic - * block layer. - */ -static int dasd_flush_block_queue(struct dasd_block *block) +static int _dasd_requests_to_flushqueue(struct dasd_block *block, + struct list_head *flush_queue) { struct dasd_ccw_req *cqr, *n; - int rc, i; - struct list_head flush_queue; unsigned long flags; + int rc, i; - INIT_LIST_HEAD(&flush_queue); - spin_lock_bh(&block->queue_lock); + spin_lock_irqsave(&block->queue_lock, flags); rc = 0; restart: list_for_each_entry_safe(cqr, n, &block->ccw_queue, blocklist) { @@ -2992,13 +2983,32 @@ restart: * is returned from the dasd_device layer. */ cqr->callback = _dasd_wake_block_flush_cb; - for (i = 0; cqr != NULL; cqr = cqr->refers, i++) - list_move_tail(&cqr->blocklist, &flush_queue); + for (i = 0; cqr; cqr = cqr->refers, i++) + list_move_tail(&cqr->blocklist, flush_queue); if (i > 1) /* moved more than one request - need to restart */ goto restart; } - spin_unlock_bh(&block->queue_lock); + spin_unlock_irqrestore(&block->queue_lock, flags); + + return rc; +} + +/* + * Go through all request on the dasd_block request queue, cancel them + * on the respective dasd_device, and return them to the generic + * block layer. + */ +static int dasd_flush_block_queue(struct dasd_block *block) +{ + struct dasd_ccw_req *cqr, *n; + struct list_head flush_queue; + unsigned long flags; + int rc; + + INIT_LIST_HEAD(&flush_queue); + rc = _dasd_requests_to_flushqueue(block, &flush_queue); + /* Now call the callback function of flushed requests */ restart_cb: list_for_each_entry_safe(cqr, n, &flush_queue, blocklist) { @@ -3881,75 +3891,36 @@ EXPORT_SYMBOL_GPL(dasd_generic_space_avail); */ int dasd_generic_requeue_all_requests(struct dasd_device *device) { + struct dasd_block *block = device->block; struct list_head requeue_queue; struct dasd_ccw_req *cqr, *n; - struct dasd_ccw_req *refers; int rc; - INIT_LIST_HEAD(&requeue_queue); - spin_lock_irq(get_ccwdev_lock(device->cdev)); - rc = 0; - list_for_each_entry_safe(cqr, n, &device->ccw_queue, devlist) { - /* Check status and move request to flush_queue */ - if (cqr->status == DASD_CQR_IN_IO) { - rc = device->discipline->term_IO(cqr); - if (rc) { - /* unable to terminate requeust */ - dev_err(&device->cdev->dev, - "Unable to terminate request %p " - "on suspend\n", cqr); - spin_unlock_irq(get_ccwdev_lock(device->cdev)); - dasd_put_device(device); - return rc; - } - } - list_move_tail(&cqr->devlist, &requeue_queue); - } - spin_unlock_irq(get_ccwdev_lock(device->cdev)); - - list_for_each_entry_safe(cqr, n, &requeue_queue, devlist) { - wait_event(dasd_flush_wq, - (cqr->status != DASD_CQR_CLEAR_PENDING)); + if (!block) + return 0; - /* - * requeue requests to blocklayer will only work - * for block device requests - */ - if (_dasd_requeue_request(cqr)) - continue; + INIT_LIST_HEAD(&requeue_queue); + rc = _dasd_requests_to_flushqueue(block, &requeue_queue); - /* remove requests from device and block queue */ - list_del_init(&cqr->devlist); - while (cqr->refers != NULL) { - refers = cqr->refers; - /* remove the request from the block queue */ - list_del(&cqr->blocklist); - /* free the finished erp request */ - dasd_free_erp_request(cqr, cqr->memdev); - cqr = refers; + /* Now call the callback function of flushed requests */ +restart_cb: + list_for_each_entry_safe(cqr, n, &requeue_queue, blocklist) { + wait_event(dasd_flush_wq, (cqr->status < DASD_CQR_QUEUED)); + /* Process finished ERP request. */ + if (cqr->refers) { + spin_lock_bh(&block->queue_lock); + __dasd_process_erp(block->base, cqr); + spin_unlock_bh(&block->queue_lock); + /* restart list_for_xx loop since dasd_process_erp + * might remove multiple elements + */ + goto restart_cb; } - - /* - * _dasd_requeue_request already checked for a valid - * blockdevice, no need to check again - * all erp requests (cqr->refers) have a cqr->block - * pointer copy from the original cqr - */ + _dasd_requeue_request(cqr); list_del_init(&cqr->blocklist); cqr->block->base->discipline->free_cp( cqr, (struct request *) cqr->callback_data); } - - /* - * if requests remain then they are internal request - * and go back to the device queue - */ - if (!list_empty(&requeue_queue)) { - /* move freeze_queue to start of the ccw_queue */ - spin_lock_irq(get_ccwdev_lock(device->cdev)); - list_splice_tail(&requeue_queue, &device->ccw_queue); - spin_unlock_irq(get_ccwdev_lock(device->cdev)); - } dasd_schedule_device_bh(device); return rc; } -- cgit From 856d8e3c633b183df23549ce760ae84478a7098d Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Fri, 21 Jul 2023 21:36:47 +0200 Subject: s390/dasd: print copy pair message only for the correct error The DASD driver has certain types of requests that might be rejected by the storage server or z/VM because they are not supported. Since the missing support of the command is not a real issue there is no user visible kernel error message for this. For copy pair setups there is a specific error that IO is not allowed on secondary devices. This error case is explicitly handled and an error message is printed. The code checking for the error did use a bitwise 'and' that is used to check for specific bits. But in this case the whole sense byte has to match. This leads to the problem that the copy pair related error message is erroneously printed for other error cases that are usually not reported. This might heavily confuse users and lead to follow on actions that might disrupt application processing. Fix by checking the sense byte for the exact value and not single bits. Cc: stable@vger.kernel.org # 6.1+ Fixes: 1fca631a1185 ("s390/dasd: suppress generic error messages for PPRC secondary devices") Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Link: https://lore.kernel.org/r/20230721193647.3889634-5-sth@linux.ibm.com Signed-off-by: Jens Axboe --- drivers/s390/block/dasd_3990_erp.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'drivers/s390') diff --git a/drivers/s390/block/dasd_3990_erp.c b/drivers/s390/block/dasd_3990_erp.c index 91e9a17b848e..89957bb7244d 100644 --- a/drivers/s390/block/dasd_3990_erp.c +++ b/drivers/s390/block/dasd_3990_erp.c @@ -1050,7 +1050,7 @@ dasd_3990_erp_com_rej(struct dasd_ccw_req * erp, char *sense) dev_err(&device->cdev->dev, "An I/O request was rejected" " because writing is inhibited\n"); erp = dasd_3990_erp_cleanup(erp, DASD_CQR_FAILED); - } else if (sense[7] & SNS7_INVALID_ON_SEC) { + } else if (sense[7] == SNS7_INVALID_ON_SEC) { dev_err(&device->cdev->dev, "An I/O request was rejected on a copy pair secondary device\n"); /* suppress dump of sense data for this error */ set_bit(DASD_CQR_SUPPRESS_CR, &erp->refers->flags); -- cgit From e65851989001c0c9ba9177564b13b38201c0854c Mon Sep 17 00:00:00 2001 From: Steffen Maier Date: Mon, 24 Jul 2023 16:51:56 +0200 Subject: scsi: zfcp: Defer fc_rport blocking until after ADISC response Storage devices are free to send RSCNs, e.g. for internal state changes. If this happens on all connected paths, zfcp risks temporarily losing all paths at the same time. This has strong requirements on multipath configuration such as "no_path_retry queue". Avoid such situations by deferring fc_rport blocking until after the ADISC response, when any actual state change of the remote port became clear. The already existing port recovery triggers explicitly block the fc_rport. The triggers are: on ADISC reject or timeout (typical cable pull case), and on ADISC indicating that the remote port has changed its WWPN or the port is meanwhile no longer open. As a side effect, this also removes a confusing direct function call to another work item function zfcp_scsi_rport_work() instead of scheduling that other work item. It was probably done that way to have the rport block side effect immediate and synchronous to the caller. Fixes: a2fa0aede07c ("[SCSI] zfcp: Block FC transport rports early on errors") Cc: stable@vger.kernel.org #v2.6.30+ Reviewed-by: Benjamin Block Reviewed-by: Fedor Loshakov Signed-off-by: Steffen Maier Link: https://lore.kernel.org/r/20230724145156.3920244-1-maier@linux.ibm.com Signed-off-by: Martin K. Petersen --- drivers/s390/scsi/zfcp_fc.c | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) (limited to 'drivers/s390') diff --git a/drivers/s390/scsi/zfcp_fc.c b/drivers/s390/scsi/zfcp_fc.c index f21307537829..4f0d0e55f0d4 100644 --- a/drivers/s390/scsi/zfcp_fc.c +++ b/drivers/s390/scsi/zfcp_fc.c @@ -534,8 +534,7 @@ static void zfcp_fc_adisc_handler(void *data) /* re-init to undo drop from zfcp_fc_adisc() */ port->d_id = ntoh24(adisc_resp->adisc_port_id); - /* port is good, unblock rport without going through erp */ - zfcp_scsi_schedule_rport_register(port); + /* port is still good, nothing to do */ out: atomic_andnot(ZFCP_STATUS_PORT_LINK_TEST, &port->status); put_device(&port->dev); @@ -595,9 +594,6 @@ void zfcp_fc_link_test_work(struct work_struct *work) int retval; set_worker_desc("zadisc%16llx", port->wwpn); /* < WORKER_DESC_LEN=24 */ - get_device(&port->dev); - port->rport_task = RPORT_DEL; - zfcp_scsi_rport_work(&port->rport_work); /* only issue one test command at one time per port */ if (atomic_read(&port->status) & ZFCP_STATUS_PORT_LINK_TEST) -- cgit From 1cfef80d4c2b2c599189f36f36320b205d9447d9 Mon Sep 17 00:00:00 2001 From: Alexandra Winter Date: Tue, 1 Aug 2023 10:00:16 +0200 Subject: s390/qeth: Don't call dev_close/dev_open (DOWN/UP) dev_close() and dev_open() are issued to change the interface state to DOWN or UP (dev->flags IFF_UP). When the netdev is set DOWN it loses e.g its Ipv6 addresses and routes. We don't want this in cases of device recovery (triggered by hardware or software) or when the qeth device is set offline. Setting a qeth device offline or online and device recovery actions call netif_device_detach() and/or netif_device_attach(). That will reset or set the LOWER_UP indication i.e. change the dev->state Bit __LINK_STATE_PRESENT. That is enough to e.g. cause bond failovers, and still preserves the interface settings that are handled by the network stack. Don't call dev_open() nor dev_close() from the qeth device driver. Let the network stack handle this. Fixes: d4560150cb47 ("s390/qeth: call dev_close() during recovery") Signed-off-by: Alexandra Winter Reviewed-by: Wenjia Zhang Signed-off-by: David S. Miller --- drivers/s390/net/qeth_core.h | 1 - drivers/s390/net/qeth_core_main.c | 2 -- drivers/s390/net/qeth_l2_main.c | 9 ++++++--- drivers/s390/net/qeth_l3_main.c | 8 +++++--- 4 files changed, 11 insertions(+), 9 deletions(-) (limited to 'drivers/s390') diff --git a/drivers/s390/net/qeth_core.h b/drivers/s390/net/qeth_core.h index 1d195429753d..613eab729704 100644 --- a/drivers/s390/net/qeth_core.h +++ b/drivers/s390/net/qeth_core.h @@ -716,7 +716,6 @@ struct qeth_card_info { u16 chid; u8 ids_valid:1; /* cssid,iid,chid */ u8 dev_addr_is_registered:1; - u8 open_when_online:1; u8 promisc_mode:1; u8 use_v1_blkt:1; u8 is_vm_nic:1; diff --git a/drivers/s390/net/qeth_core_main.c b/drivers/s390/net/qeth_core_main.c index 1d5b207c2b9e..cd783290bde5 100644 --- a/drivers/s390/net/qeth_core_main.c +++ b/drivers/s390/net/qeth_core_main.c @@ -5373,8 +5373,6 @@ int qeth_set_offline(struct qeth_card *card, const struct qeth_discipline *disc, qeth_clear_ipacmd_list(card); rtnl_lock(); - card->info.open_when_online = card->dev->flags & IFF_UP; - dev_close(card->dev); netif_device_detach(card->dev); netif_carrier_off(card->dev); rtnl_unlock(); diff --git a/drivers/s390/net/qeth_l2_main.c b/drivers/s390/net/qeth_l2_main.c index 9f13ed170a43..75910c0bcc2b 100644 --- a/drivers/s390/net/qeth_l2_main.c +++ b/drivers/s390/net/qeth_l2_main.c @@ -2388,9 +2388,12 @@ static int qeth_l2_set_online(struct qeth_card *card, bool carrier_ok) qeth_enable_hw_features(dev); qeth_l2_enable_brport_features(card); - if (card->info.open_when_online) { - card->info.open_when_online = 0; - dev_open(dev, NULL); + if (netif_running(dev)) { + local_bh_disable(); + napi_schedule(&card->napi); + /* kick-start the NAPI softirq: */ + local_bh_enable(); + qeth_l2_set_rx_mode(dev); } rtnl_unlock(); } diff --git a/drivers/s390/net/qeth_l3_main.c b/drivers/s390/net/qeth_l3_main.c index af4e60d2917e..b92a32b4b114 100644 --- a/drivers/s390/net/qeth_l3_main.c +++ b/drivers/s390/net/qeth_l3_main.c @@ -2018,9 +2018,11 @@ static int qeth_l3_set_online(struct qeth_card *card, bool carrier_ok) netif_device_attach(dev); qeth_enable_hw_features(dev); - if (card->info.open_when_online) { - card->info.open_when_online = 0; - dev_open(dev, NULL); + if (netif_running(dev)) { + local_bh_disable(); + napi_schedule(&card->napi); + /* kick-start the NAPI softirq: */ + local_bh_enable(); } rtnl_unlock(); } -- cgit