From 240c1948491b81cfe40f84ea040a8f2a4966f101 Mon Sep 17 00:00:00 2001 From: Julian Wiedmann Date: Tue, 10 Mar 2020 18:38:01 +0100 Subject: s390/qeth: don't reset default_out_queue When an OSA device in prio-queue setup is reduced to 1 TX queue due to HW restrictions, we reset its the default_out_queue to 0. In the old code this was needed so that qeth_get_priority_queue() gets the queue selection right. But with proper multiqueue support we already reduced dev->real_num_tx_queues to 1, and so the stack puts all traffic on txq 0 without even calling .ndo_select_queue. Thus we can preserve the user's configuration, and apply it if the OSA device later re-gains support for multiple TX queues. Fixes: 73dc2daf110f ("s390/qeth: add TX multiqueue support for OSA devices") Signed-off-by: Julian Wiedmann Signed-off-by: David S. Miller --- drivers/s390/net/qeth_core_main.c | 1 - 1 file changed, 1 deletion(-) (limited to 'drivers/s390') diff --git a/drivers/s390/net/qeth_core_main.c b/drivers/s390/net/qeth_core_main.c index 8ca85c8a01a1..2a05d13a0e79 100644 --- a/drivers/s390/net/qeth_core_main.c +++ b/drivers/s390/net/qeth_core_main.c @@ -1204,7 +1204,6 @@ static int qeth_osa_set_output_queues(struct qeth_card *card, bool single) if (count == 1) dev_info(&card->gdev->dev, "Priority Queueing not supported\n"); - card->qdio.default_out_queue = single ? 0 : QETH_DEFAULT_QUEUE; card->qdio.no_out_queues = count; return 0; } -- cgit v1.2.3 From 17413852804d7e86e6f0576cca32c1541817800e Mon Sep 17 00:00:00 2001 From: Julian Wiedmann Date: Tue, 10 Mar 2020 18:38:02 +0100 Subject: s390/qeth: handle error when backing RX buffer qeth_init_qdio_queues() fills the RX ring with an initial set of RX buffers. If qeth_init_input_buffer() fails to back one of the RX buffers with memory, we need to bail out and report the error. Fixes: 4a71df50047f ("qeth: new qeth device driver") Signed-off-by: Julian Wiedmann Signed-off-by: David S. Miller --- drivers/s390/net/qeth_core_main.c | 13 +++++++++---- 1 file changed, 9 insertions(+), 4 deletions(-) (limited to 'drivers/s390') diff --git a/drivers/s390/net/qeth_core_main.c b/drivers/s390/net/qeth_core_main.c index 2a05d13a0e79..d8f0c610396e 100644 --- a/drivers/s390/net/qeth_core_main.c +++ b/drivers/s390/net/qeth_core_main.c @@ -2624,12 +2624,12 @@ static int qeth_init_input_buffer(struct qeth_card *card, ETH_HLEN + sizeof(struct ipv6hdr)); if (!buf->rx_skb) - return 1; + return -ENOMEM; } pool_entry = qeth_find_free_buffer_pool_entry(card); if (!pool_entry) - return 1; + return -ENOBUFS; /* * since the buffer is accessed only from the input_tasklet @@ -2674,10 +2674,15 @@ static int qeth_init_qdio_queues(struct qeth_card *card) /* inbound queue */ qdio_reset_buffers(card->qdio.in_q->qdio_bufs, QDIO_MAX_BUFFERS_PER_Q); memset(&card->rx, 0, sizeof(struct qeth_rx)); + qeth_initialize_working_pool_list(card); /*give only as many buffers to hardware as we have buffer pool entries*/ - for (i = 0; i < card->qdio.in_buf_pool.buf_count - 1; ++i) - qeth_init_input_buffer(card, &card->qdio.in_q->bufs[i]); + for (i = 0; i < card->qdio.in_buf_pool.buf_count - 1; i++) { + rc = qeth_init_input_buffer(card, &card->qdio.in_q->bufs[i]); + if (rc) + return rc; + } + card->qdio.in_q->next_buf_to_init = card->qdio.in_buf_pool.buf_count - 1; rc = do_QDIO(CARD_DDEV(card), QDIO_FLAG_SYNC_INPUT, 0, 0, -- cgit v1.2.3 From 0e635c2a8713e3e013b4adcb7cb7e28a048b7c42 Mon Sep 17 00:00:00 2001 From: Julian Wiedmann Date: Tue, 10 Mar 2020 18:38:03 +0100 Subject: s390/qeth: cancel RX reclaim work earlier When qeth's napi poll code fails to refill an entirely empty RX ring, it kicks off buffer_reclaim_work to try again later. Make sure that this worker is cancelled when setting the qeth device offline. Otherwise a RX refill action can unexpectedly end up running concurrently to bigger re-configurations (eg. resizing the buffer pool), without any locking. Fixes: b333293058aa ("qeth: add support for af_iucv HiperSockets transport") Signed-off-by: Julian Wiedmann Signed-off-by: David S. Miller --- drivers/s390/net/qeth_core_main.c | 1 - drivers/s390/net/qeth_l2_main.c | 1 + drivers/s390/net/qeth_l3_main.c | 1 + 3 files changed, 2 insertions(+), 1 deletion(-) (limited to 'drivers/s390') diff --git a/drivers/s390/net/qeth_core_main.c b/drivers/s390/net/qeth_core_main.c index d8f0c610396e..a599801d7727 100644 --- a/drivers/s390/net/qeth_core_main.c +++ b/drivers/s390/net/qeth_core_main.c @@ -2392,7 +2392,6 @@ static void qeth_free_qdio_queues(struct qeth_card *card) return; qeth_free_cq(card); - cancel_delayed_work_sync(&card->buffer_reclaim_work); for (j = 0; j < QDIO_MAX_BUFFERS_PER_Q; ++j) { if (card->qdio.in_q->bufs[j].rx_skb) dev_kfree_skb_any(card->qdio.in_q->bufs[j].rx_skb); diff --git a/drivers/s390/net/qeth_l2_main.c b/drivers/s390/net/qeth_l2_main.c index 9972d96820f3..8fb29371788b 100644 --- a/drivers/s390/net/qeth_l2_main.c +++ b/drivers/s390/net/qeth_l2_main.c @@ -284,6 +284,7 @@ static void qeth_l2_stop_card(struct qeth_card *card) if (card->state == CARD_STATE_SOFTSETUP) { qeth_clear_ipacmd_list(card); qeth_drain_output_queues(card); + cancel_delayed_work_sync(&card->buffer_reclaim_work); card->state = CARD_STATE_DOWN; } diff --git a/drivers/s390/net/qeth_l3_main.c b/drivers/s390/net/qeth_l3_main.c index 317d56647a4a..82f800d1d7b3 100644 --- a/drivers/s390/net/qeth_l3_main.c +++ b/drivers/s390/net/qeth_l3_main.c @@ -1178,6 +1178,7 @@ static void qeth_l3_stop_card(struct qeth_card *card) qeth_l3_clear_ip_htable(card, 1); qeth_clear_ipacmd_list(card); qeth_drain_output_queues(card); + cancel_delayed_work_sync(&card->buffer_reclaim_work); card->state = CARD_STATE_DOWN; } -- cgit v1.2.3 From f81649dfa5343eef7e579eb6f8dd8bd6d300ec31 Mon Sep 17 00:00:00 2001 From: Julian Wiedmann Date: Wed, 11 Mar 2020 18:07:09 +0100 Subject: s390/qeth: use page pointers to manage RX buffer pool The RX buffer elements are always backed with full pages, reflect this in the pointer type. Signed-off-by: Julian Wiedmann Signed-off-by: David S. Miller --- drivers/s390/net/qeth_core.h | 2 +- drivers/s390/net/qeth_core_main.c | 35 +++++++++++++++++------------------ 2 files changed, 18 insertions(+), 19 deletions(-) (limited to 'drivers/s390') diff --git a/drivers/s390/net/qeth_core.h b/drivers/s390/net/qeth_core.h index 9575a627a1e1..242b05f644eb 100644 --- a/drivers/s390/net/qeth_core.h +++ b/drivers/s390/net/qeth_core.h @@ -369,7 +369,7 @@ enum qeth_qdio_info_states { struct qeth_buffer_pool_entry { struct list_head list; struct list_head init_list; - void *elements[QDIO_MAX_ELEMENTS_PER_BUFFER]; + struct page *elements[QDIO_MAX_ELEMENTS_PER_BUFFER]; }; struct qeth_qdio_buffer_pool { diff --git a/drivers/s390/net/qeth_core_main.c b/drivers/s390/net/qeth_core_main.c index a599801d7727..8f682fc178a9 100644 --- a/drivers/s390/net/qeth_core_main.c +++ b/drivers/s390/net/qeth_core_main.c @@ -215,7 +215,6 @@ EXPORT_SYMBOL_GPL(qeth_clear_working_pool_list); static int qeth_alloc_buffer_pool(struct qeth_card *card) { struct qeth_buffer_pool_entry *pool_entry; - void *ptr; int i, j; QETH_CARD_TEXT(card, 5, "alocpool"); @@ -225,17 +224,18 @@ static int qeth_alloc_buffer_pool(struct qeth_card *card) qeth_free_buffer_pool(card); return -ENOMEM; } + for (j = 0; j < QETH_MAX_BUFFER_ELEMENTS(card); ++j) { - ptr = (void *) __get_free_page(GFP_KERNEL); - if (!ptr) { + struct page *page = alloc_page(GFP_KERNEL); + + if (!page) { while (j > 0) - free_page((unsigned long) - pool_entry->elements[--j]); + __free_page(pool_entry->elements[--j]); kfree(pool_entry); qeth_free_buffer_pool(card); return -ENOMEM; } - pool_entry->elements[j] = ptr; + pool_entry->elements[j] = page; } list_add(&pool_entry->init_list, &card->qdio.init_pool.entry_list); @@ -1177,7 +1177,7 @@ static void qeth_free_buffer_pool(struct qeth_card *card) list_for_each_entry_safe(pool_entry, tmp, &card->qdio.init_pool.entry_list, init_list){ for (i = 0; i < QETH_MAX_BUFFER_ELEMENTS(card); ++i) - free_page((unsigned long)pool_entry->elements[i]); + __free_page(pool_entry->elements[i]); list_del(&pool_entry->init_list); kfree(pool_entry); } @@ -2573,7 +2573,6 @@ static struct qeth_buffer_pool_entry *qeth_find_free_buffer_pool_entry( struct list_head *plh; struct qeth_buffer_pool_entry *entry; int i, free; - struct page *page; if (list_empty(&card->qdio.in_buf_pool.entry_list)) return NULL; @@ -2582,7 +2581,7 @@ static struct qeth_buffer_pool_entry *qeth_find_free_buffer_pool_entry( entry = list_entry(plh, struct qeth_buffer_pool_entry, list); free = 1; for (i = 0; i < QETH_MAX_BUFFER_ELEMENTS(card); ++i) { - if (page_count(virt_to_page(entry->elements[i])) > 1) { + if (page_count(entry->elements[i]) > 1) { free = 0; break; } @@ -2597,15 +2596,15 @@ static struct qeth_buffer_pool_entry *qeth_find_free_buffer_pool_entry( entry = list_entry(card->qdio.in_buf_pool.entry_list.next, struct qeth_buffer_pool_entry, list); for (i = 0; i < QETH_MAX_BUFFER_ELEMENTS(card); ++i) { - if (page_count(virt_to_page(entry->elements[i])) > 1) { - page = alloc_page(GFP_ATOMIC); - if (!page) { + if (page_count(entry->elements[i]) > 1) { + struct page *page = alloc_page(GFP_ATOMIC); + + if (!page) return NULL; - } else { - free_page((unsigned long)entry->elements[i]); - entry->elements[i] = page_address(page); - QETH_CARD_STAT_INC(card, rx_sg_alloc_page); - } + + __free_page(entry->elements[i]); + entry->elements[i] = page; + QETH_CARD_STAT_INC(card, rx_sg_alloc_page); } } list_del_init(&entry->list); @@ -2641,7 +2640,7 @@ static int qeth_init_input_buffer(struct qeth_card *card, for (i = 0; i < QETH_MAX_BUFFER_ELEMENTS(card); ++i) { buf->buffer->element[i].length = PAGE_SIZE; buf->buffer->element[i].addr = - virt_to_phys(pool_entry->elements[i]); + page_to_phys(pool_entry->elements[i]); if (i == QETH_MAX_BUFFER_ELEMENTS(card) - 1) buf->buffer->element[i].eflags = SBAL_EFLAGS_LAST_ENTRY; else -- cgit v1.2.3 From 0f75e149298bedc48db03259a38a303611d247b1 Mon Sep 17 00:00:00 2001 From: Julian Wiedmann Date: Wed, 11 Mar 2020 18:07:10 +0100 Subject: s390/qeth: refactor buffer pool code In preparation for a subsequent fix, split out helpers to allocate/free individual pool entries. Signed-off-by: Julian Wiedmann Signed-off-by: David S. Miller --- drivers/s390/net/qeth_core_main.c | 83 ++++++++++++++++++++++++--------------- 1 file changed, 51 insertions(+), 32 deletions(-) (limited to 'drivers/s390') diff --git a/drivers/s390/net/qeth_core_main.c b/drivers/s390/net/qeth_core_main.c index 8f682fc178a9..ceab3d0c4dfa 100644 --- a/drivers/s390/net/qeth_core_main.c +++ b/drivers/s390/net/qeth_core_main.c @@ -65,7 +65,6 @@ static struct lock_class_key qdio_out_skb_queue_key; static void qeth_issue_next_read_cb(struct qeth_card *card, struct qeth_cmd_buffer *iob, unsigned int data_length); -static void qeth_free_buffer_pool(struct qeth_card *); static int qeth_qdio_establish(struct qeth_card *); static void qeth_free_qdio_queues(struct qeth_card *card); static void qeth_notify_skbs(struct qeth_qdio_out_q *queue, @@ -212,33 +211,66 @@ void qeth_clear_working_pool_list(struct qeth_card *card) } EXPORT_SYMBOL_GPL(qeth_clear_working_pool_list); +static void qeth_free_pool_entry(struct qeth_buffer_pool_entry *entry) +{ + unsigned int i; + + for (i = 0; i < ARRAY_SIZE(entry->elements); i++) { + if (entry->elements[i]) + __free_page(entry->elements[i]); + } + + kfree(entry); +} + +static void qeth_free_buffer_pool(struct qeth_card *card) +{ + struct qeth_buffer_pool_entry *entry, *tmp; + + list_for_each_entry_safe(entry, tmp, &card->qdio.init_pool.entry_list, + init_list) { + list_del(&entry->init_list); + qeth_free_pool_entry(entry); + } +} + +static struct qeth_buffer_pool_entry *qeth_alloc_pool_entry(unsigned int pages) +{ + struct qeth_buffer_pool_entry *entry; + unsigned int i; + + entry = kzalloc(sizeof(*entry), GFP_KERNEL); + if (!entry) + return NULL; + + for (i = 0; i < pages; i++) { + entry->elements[i] = alloc_page(GFP_KERNEL); + + if (!entry->elements[i]) { + qeth_free_pool_entry(entry); + return NULL; + } + } + + return entry; +} + static int qeth_alloc_buffer_pool(struct qeth_card *card) { - struct qeth_buffer_pool_entry *pool_entry; - int i, j; + unsigned int buf_elements = QETH_MAX_BUFFER_ELEMENTS(card); + unsigned int i; QETH_CARD_TEXT(card, 5, "alocpool"); for (i = 0; i < card->qdio.init_pool.buf_count; ++i) { - pool_entry = kzalloc(sizeof(*pool_entry), GFP_KERNEL); - if (!pool_entry) { + struct qeth_buffer_pool_entry *entry; + + entry = qeth_alloc_pool_entry(buf_elements); + if (!entry) { qeth_free_buffer_pool(card); return -ENOMEM; } - for (j = 0; j < QETH_MAX_BUFFER_ELEMENTS(card); ++j) { - struct page *page = alloc_page(GFP_KERNEL); - - if (!page) { - while (j > 0) - __free_page(pool_entry->elements[--j]); - kfree(pool_entry); - qeth_free_buffer_pool(card); - return -ENOMEM; - } - pool_entry->elements[j] = page; - } - list_add(&pool_entry->init_list, - &card->qdio.init_pool.entry_list); + list_add(&entry->init_list, &card->qdio.init_pool.entry_list); } return 0; } @@ -1170,19 +1202,6 @@ void qeth_drain_output_queues(struct qeth_card *card) } EXPORT_SYMBOL_GPL(qeth_drain_output_queues); -static void qeth_free_buffer_pool(struct qeth_card *card) -{ - struct qeth_buffer_pool_entry *pool_entry, *tmp; - int i = 0; - list_for_each_entry_safe(pool_entry, tmp, - &card->qdio.init_pool.entry_list, init_list){ - for (i = 0; i < QETH_MAX_BUFFER_ELEMENTS(card); ++i) - __free_page(pool_entry->elements[i]); - list_del(&pool_entry->init_list); - kfree(pool_entry); - } -} - static int qeth_osa_set_output_queues(struct qeth_card *card, bool single) { unsigned int count = single ? 1 : card->dev->num_tx_queues; -- cgit v1.2.3 From 5d4f78564c9ca38146102c4f8998edcad5afbc22 Mon Sep 17 00:00:00 2001 From: Julian Wiedmann Date: Wed, 11 Mar 2020 18:07:11 +0100 Subject: s390/qeth: implement smarter resizing of the RX buffer pool The RX buffer pool is allocated in qeth_alloc_qdio_queues(). A subsequent pool resizing is then handled in a very simple way: first free the current pool, then allocate a new pool of the requested size. There's two ways where this can go wrong: 1. if the resize action happens _before_ the initial pool was allocated, then a subsequent initialization will call qeth_alloc_qdio_queues() and fill the pool with a second(!) set of pages. We consume twice the planned amount of memory. This is easy to fix - just skip the resizing if the queues haven't been allocated yet. 2. if the initial pool was created by qeth_alloc_qdio_queues() but a subsequent resizing fails, then the device has no(!) RX buffer pool. The next initialization will _not_ call qeth_alloc_qdio_queues(), and attempting to back the RX buffers with pages in qeth_init_qdio_queues() will fail. Not very difficult to fix either - instead of re-allocating the whole pool, just allocate/free as many entries to match the desired size. Fixes: 4a71df50047f ("qeth: new qeth device driver") Signed-off-by: Julian Wiedmann Signed-off-by: David S. Miller --- drivers/s390/net/qeth_core.h | 2 +- drivers/s390/net/qeth_core_main.c | 55 +++++++++++++++++++++++++++++++++------ drivers/s390/net/qeth_core_sys.c | 9 +++---- drivers/s390/net/qeth_l3_sys.c | 9 +++---- 4 files changed, 56 insertions(+), 19 deletions(-) (limited to 'drivers/s390') diff --git a/drivers/s390/net/qeth_core.h b/drivers/s390/net/qeth_core.h index 242b05f644eb..468cada49e72 100644 --- a/drivers/s390/net/qeth_core.h +++ b/drivers/s390/net/qeth_core.h @@ -983,7 +983,7 @@ extern const struct attribute_group qeth_device_blkt_group; extern const struct device_type qeth_generic_devtype; const char *qeth_get_cardname_short(struct qeth_card *); -int qeth_realloc_buffer_pool(struct qeth_card *, int); +int qeth_resize_buffer_pool(struct qeth_card *card, unsigned int count); int qeth_core_load_discipline(struct qeth_card *, enum qeth_discipline_id); void qeth_core_free_discipline(struct qeth_card *); diff --git a/drivers/s390/net/qeth_core_main.c b/drivers/s390/net/qeth_core_main.c index ceab3d0c4dfa..6d3f2f14b414 100644 --- a/drivers/s390/net/qeth_core_main.c +++ b/drivers/s390/net/qeth_core_main.c @@ -275,18 +275,57 @@ static int qeth_alloc_buffer_pool(struct qeth_card *card) return 0; } -int qeth_realloc_buffer_pool(struct qeth_card *card, int bufcnt) +int qeth_resize_buffer_pool(struct qeth_card *card, unsigned int count) { + unsigned int buf_elements = QETH_MAX_BUFFER_ELEMENTS(card); + struct qeth_qdio_buffer_pool *pool = &card->qdio.init_pool; + struct qeth_buffer_pool_entry *entry, *tmp; + int delta = count - pool->buf_count; + LIST_HEAD(entries); + QETH_CARD_TEXT(card, 2, "realcbp"); - /* TODO: steel/add buffers from/to a running card's buffer pool (?) */ - qeth_clear_working_pool_list(card); - qeth_free_buffer_pool(card); - card->qdio.in_buf_pool.buf_count = bufcnt; - card->qdio.init_pool.buf_count = bufcnt; - return qeth_alloc_buffer_pool(card); + /* Defer until queue is allocated: */ + if (!card->qdio.in_q) + goto out; + + /* Remove entries from the pool: */ + while (delta < 0) { + entry = list_first_entry(&pool->entry_list, + struct qeth_buffer_pool_entry, + init_list); + list_del(&entry->init_list); + qeth_free_pool_entry(entry); + + delta++; + } + + /* Allocate additional entries: */ + while (delta > 0) { + entry = qeth_alloc_pool_entry(buf_elements); + if (!entry) { + list_for_each_entry_safe(entry, tmp, &entries, + init_list) { + list_del(&entry->init_list); + qeth_free_pool_entry(entry); + } + + return -ENOMEM; + } + + list_add(&entry->init_list, &entries); + + delta--; + } + + list_splice(&entries, &pool->entry_list); + +out: + card->qdio.in_buf_pool.buf_count = count; + pool->buf_count = count; + return 0; } -EXPORT_SYMBOL_GPL(qeth_realloc_buffer_pool); +EXPORT_SYMBOL_GPL(qeth_resize_buffer_pool); static void qeth_free_qdio_queue(struct qeth_qdio_q *q) { diff --git a/drivers/s390/net/qeth_core_sys.c b/drivers/s390/net/qeth_core_sys.c index 2bd9993aa60b..78cae61bc924 100644 --- a/drivers/s390/net/qeth_core_sys.c +++ b/drivers/s390/net/qeth_core_sys.c @@ -247,8 +247,8 @@ static ssize_t qeth_dev_bufcnt_store(struct device *dev, struct device_attribute *attr, const char *buf, size_t count) { struct qeth_card *card = dev_get_drvdata(dev); + unsigned int cnt; char *tmp; - int cnt, old_cnt; int rc = 0; mutex_lock(&card->conf_mutex); @@ -257,13 +257,12 @@ static ssize_t qeth_dev_bufcnt_store(struct device *dev, goto out; } - old_cnt = card->qdio.in_buf_pool.buf_count; cnt = simple_strtoul(buf, &tmp, 10); cnt = (cnt < QETH_IN_BUF_COUNT_MIN) ? QETH_IN_BUF_COUNT_MIN : ((cnt > QETH_IN_BUF_COUNT_MAX) ? QETH_IN_BUF_COUNT_MAX : cnt); - if (old_cnt != cnt) { - rc = qeth_realloc_buffer_pool(card, cnt); - } + + rc = qeth_resize_buffer_pool(card, cnt); + out: mutex_unlock(&card->conf_mutex); return rc ? rc : count; diff --git a/drivers/s390/net/qeth_l3_sys.c b/drivers/s390/net/qeth_l3_sys.c index 29f2517d2a31..a3d1c3bdfadb 100644 --- a/drivers/s390/net/qeth_l3_sys.c +++ b/drivers/s390/net/qeth_l3_sys.c @@ -206,12 +206,11 @@ static ssize_t qeth_l3_dev_sniffer_store(struct device *dev, qdio_get_ssqd_desc(CARD_DDEV(card), &card->ssqd); if (card->ssqd.qdioac2 & CHSC_AC2_SNIFFER_AVAILABLE) { card->options.sniffer = i; - if (card->qdio.init_pool.buf_count != - QETH_IN_BUF_COUNT_MAX) - qeth_realloc_buffer_pool(card, - QETH_IN_BUF_COUNT_MAX); - } else + qeth_resize_buffer_pool(card, QETH_IN_BUF_COUNT_MAX); + } else { rc = -EPERM; + } + break; default: rc = -EINVAL; -- cgit v1.2.3 From 5e6bdd37c5526ef01326df5dabb93011ee89237e Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Thu, 12 Mar 2020 14:17:15 +0100 Subject: s390/dasd: fix data corruption for thin provisioned devices Devices are formatted in multiple of tracks. For an Extent Space Efficient (ESE) volume we get errors when accessing unformatted tracks. In this case the driver either formats the track on the flight for write requests or returns zero data for read requests. In case a request spans multiple tracks, the indication of an unformatted track presented for the first track is incorrectly applied to all tracks covered by the request. As a result, tracks containing data will be handled as empty, resulting in zero data being returned on read, or overwriting existing data with zero on write. Fix by determining the track that gets the NRF error. For write requests only format the track that is surely not formatted. For Read requests all tracks before have returned valid data and should not be touched. All tracks after the unformatted track might be formatted or not. Those are returned to the blocklayer to build a new request. When using alias devices there is a chance that multiple write requests trigger a format of the same track which might lead to data loss. Ensure that a track is formatted only once by maintaining a list of currently processed tracks. Fixes: 5e2b17e712cf ("s390/dasd: Add dynamic formatting support for ESE volumes") Cc: stable@vger.kernel.org # 5.3+ Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Reviewed-by: Peter Oberparleiter Signed-off-by: Jens Axboe --- drivers/s390/block/dasd.c | 27 ++++++- drivers/s390/block/dasd_eckd.c | 163 +++++++++++++++++++++++++++++++++++++++-- drivers/s390/block/dasd_int.h | 15 +++- 3 files changed, 193 insertions(+), 12 deletions(-) (limited to 'drivers/s390') diff --git a/drivers/s390/block/dasd.c b/drivers/s390/block/dasd.c index 6cca72782af6..cf87eb27879f 100644 --- a/drivers/s390/block/dasd.c +++ b/drivers/s390/block/dasd.c @@ -178,6 +178,8 @@ struct dasd_block *dasd_alloc_block(void) (unsigned long) block); INIT_LIST_HEAD(&block->ccw_queue); spin_lock_init(&block->queue_lock); + INIT_LIST_HEAD(&block->format_list); + spin_lock_init(&block->format_lock); timer_setup(&block->timer, dasd_block_timeout, 0); spin_lock_init(&block->profile.lock); @@ -1779,20 +1781,26 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm, if (dasd_ese_needs_format(cqr->block, irb)) { if (rq_data_dir((struct request *)cqr->callback_data) == READ) { - device->discipline->ese_read(cqr); + device->discipline->ese_read(cqr, irb); cqr->status = DASD_CQR_SUCCESS; cqr->stopclk = now; dasd_device_clear_timer(device); dasd_schedule_device_bh(device); return; } - fcqr = device->discipline->ese_format(device, cqr); + fcqr = device->discipline->ese_format(device, cqr, irb); if (IS_ERR(fcqr)) { + if (PTR_ERR(fcqr) == -EINVAL) { + cqr->status = DASD_CQR_ERROR; + return; + } /* * If we can't format now, let the request go * one extra round. Maybe we can format later. */ cqr->status = DASD_CQR_QUEUED; + dasd_schedule_device_bh(device); + return; } else { fcqr->status = DASD_CQR_QUEUED; cqr->status = DASD_CQR_QUEUED; @@ -2748,11 +2756,13 @@ static void __dasd_cleanup_cqr(struct dasd_ccw_req *cqr) { struct request *req; blk_status_t error = BLK_STS_OK; + unsigned int proc_bytes; int status; req = (struct request *) cqr->callback_data; dasd_profile_end(cqr->block, cqr, req); + proc_bytes = cqr->proc_bytes; status = cqr->block->base->discipline->free_cp(cqr, req); if (status < 0) error = errno_to_blk_status(status); @@ -2783,7 +2793,18 @@ static void __dasd_cleanup_cqr(struct dasd_ccw_req *cqr) blk_mq_end_request(req, error); blk_mq_run_hw_queues(req->q, true); } else { - blk_mq_complete_request(req); + /* + * Partial completed requests can happen with ESE devices. + * During read we might have gotten a NRF error and have to + * complete a request partially. + */ + if (proc_bytes) { + blk_update_request(req, BLK_STS_OK, + blk_rq_bytes(req) - proc_bytes); + blk_mq_requeue_request(req, true); + } else { + blk_mq_complete_request(req); + } } } diff --git a/drivers/s390/block/dasd_eckd.c b/drivers/s390/block/dasd_eckd.c index a28b9ff82378..ad44d22e8859 100644 --- a/drivers/s390/block/dasd_eckd.c +++ b/drivers/s390/block/dasd_eckd.c @@ -207,6 +207,45 @@ static void set_ch_t(struct ch_t *geo, __u32 cyl, __u8 head) geo->head |= head; } +/* + * calculate failing track from sense data depending if + * it is an EAV device or not + */ +static int dasd_eckd_track_from_irb(struct irb *irb, struct dasd_device *device, + sector_t *track) +{ + struct dasd_eckd_private *private = device->private; + u8 *sense = NULL; + u32 cyl; + u8 head; + + sense = dasd_get_sense(irb); + if (!sense) { + DBF_DEV_EVENT(DBF_WARNING, device, "%s", + "ESE error no sense data\n"); + return -EINVAL; + } + if (!(sense[27] & DASD_SENSE_BIT_2)) { + DBF_DEV_EVENT(DBF_WARNING, device, "%s", + "ESE error no valid track data\n"); + return -EINVAL; + } + + if (sense[27] & DASD_SENSE_BIT_3) { + /* enhanced addressing */ + cyl = sense[30] << 20; + cyl |= (sense[31] & 0xF0) << 12; + cyl |= sense[28] << 8; + cyl |= sense[29]; + } else { + cyl = sense[29] << 8; + cyl |= sense[30]; + } + head = sense[31] & 0x0F; + *track = cyl * private->rdc_data.trk_per_cyl + head; + return 0; +} + static int set_timestamp(struct ccw1 *ccw, struct DE_eckd_data *data, struct dasd_device *device) { @@ -2986,6 +3025,37 @@ static int dasd_eckd_format_device(struct dasd_device *base, 0, NULL); } +static bool test_and_set_format_track(struct dasd_format_entry *to_format, + struct dasd_block *block) +{ + struct dasd_format_entry *format; + unsigned long flags; + bool rc = false; + + spin_lock_irqsave(&block->format_lock, flags); + list_for_each_entry(format, &block->format_list, list) { + if (format->track == to_format->track) { + rc = true; + goto out; + } + } + list_add_tail(&to_format->list, &block->format_list); + +out: + spin_unlock_irqrestore(&block->format_lock, flags); + return rc; +} + +static void clear_format_track(struct dasd_format_entry *format, + struct dasd_block *block) +{ + unsigned long flags; + + spin_lock_irqsave(&block->format_lock, flags); + list_del_init(&format->list); + spin_unlock_irqrestore(&block->format_lock, flags); +} + /* * Callback function to free ESE format requests. */ @@ -2993,15 +3063,19 @@ static void dasd_eckd_ese_format_cb(struct dasd_ccw_req *cqr, void *data) { struct dasd_device *device = cqr->startdev; struct dasd_eckd_private *private = device->private; + struct dasd_format_entry *format = data; + clear_format_track(format, cqr->basedev->block); private->count--; dasd_ffree_request(cqr, device); } static struct dasd_ccw_req * -dasd_eckd_ese_format(struct dasd_device *startdev, struct dasd_ccw_req *cqr) +dasd_eckd_ese_format(struct dasd_device *startdev, struct dasd_ccw_req *cqr, + struct irb *irb) { struct dasd_eckd_private *private; + struct dasd_format_entry *format; struct format_data_t fdata; unsigned int recs_per_trk; struct dasd_ccw_req *fcqr; @@ -3011,23 +3085,39 @@ dasd_eckd_ese_format(struct dasd_device *startdev, struct dasd_ccw_req *cqr) struct request *req; sector_t first_trk; sector_t last_trk; + sector_t curr_trk; int rc; req = cqr->callback_data; - base = cqr->block->base; + block = cqr->block; + base = block->base; private = base->private; - block = base->block; blksize = block->bp_block; recs_per_trk = recs_per_track(&private->rdc_data, 0, blksize); + format = &startdev->format_entry; first_trk = blk_rq_pos(req) >> block->s2b_shift; sector_div(first_trk, recs_per_trk); last_trk = (blk_rq_pos(req) + blk_rq_sectors(req) - 1) >> block->s2b_shift; sector_div(last_trk, recs_per_trk); + rc = dasd_eckd_track_from_irb(irb, base, &curr_trk); + if (rc) + return ERR_PTR(rc); - fdata.start_unit = first_trk; - fdata.stop_unit = last_trk; + if (curr_trk < first_trk || curr_trk > last_trk) { + DBF_DEV_EVENT(DBF_WARNING, startdev, + "ESE error track %llu not within range %llu - %llu\n", + curr_trk, first_trk, last_trk); + return ERR_PTR(-EINVAL); + } + format->track = curr_trk; + /* test if track is already in formatting by another thread */ + if (test_and_set_format_track(format, block)) + return ERR_PTR(-EEXIST); + + fdata.start_unit = curr_trk; + fdata.stop_unit = curr_trk; fdata.blksize = blksize; fdata.intensity = private->uses_cdl ? DASD_FMT_INT_COMPAT : 0; @@ -3044,6 +3134,7 @@ dasd_eckd_ese_format(struct dasd_device *startdev, struct dasd_ccw_req *cqr) return fcqr; fcqr->callback = dasd_eckd_ese_format_cb; + fcqr->callback_data = (void *) format; return fcqr; } @@ -3051,29 +3142,87 @@ dasd_eckd_ese_format(struct dasd_device *startdev, struct dasd_ccw_req *cqr) /* * When data is read from an unformatted area of an ESE volume, this function * returns zeroed data and thereby mimics a read of zero data. + * + * The first unformatted track is the one that got the NRF error, the address is + * encoded in the sense data. + * + * All tracks before have returned valid data and should not be touched. + * All tracks after the unformatted track might be formatted or not. This is + * currently not known, remember the processed data and return the remainder of + * the request to the blocklayer in __dasd_cleanup_cqr(). */ -static void dasd_eckd_ese_read(struct dasd_ccw_req *cqr) +static int dasd_eckd_ese_read(struct dasd_ccw_req *cqr, struct irb *irb) { + struct dasd_eckd_private *private; + sector_t first_trk, last_trk; + sector_t first_blk, last_blk; unsigned int blksize, off; + unsigned int recs_per_trk; struct dasd_device *base; struct req_iterator iter; + struct dasd_block *block; + unsigned int skip_block; + unsigned int blk_count; struct request *req; struct bio_vec bv; + sector_t curr_trk; + sector_t end_blk; char *dst; + int rc; req = (struct request *) cqr->callback_data; base = cqr->block->base; blksize = base->block->bp_block; + block = cqr->block; + private = base->private; + skip_block = 0; + blk_count = 0; + + recs_per_trk = recs_per_track(&private->rdc_data, 0, blksize); + first_trk = first_blk = blk_rq_pos(req) >> block->s2b_shift; + sector_div(first_trk, recs_per_trk); + last_trk = last_blk = + (blk_rq_pos(req) + blk_rq_sectors(req) - 1) >> block->s2b_shift; + sector_div(last_trk, recs_per_trk); + rc = dasd_eckd_track_from_irb(irb, base, &curr_trk); + if (rc) + return rc; + + /* sanity check if the current track from sense data is valid */ + if (curr_trk < first_trk || curr_trk > last_trk) { + DBF_DEV_EVENT(DBF_WARNING, base, + "ESE error track %llu not within range %llu - %llu\n", + curr_trk, first_trk, last_trk); + return -EINVAL; + } + + /* + * if not the first track got the NRF error we have to skip over valid + * blocks + */ + if (curr_trk != first_trk) + skip_block = curr_trk * recs_per_trk - first_blk; + + /* we have no information beyond the current track */ + end_blk = (curr_trk + 1) * recs_per_trk; rq_for_each_segment(bv, req, iter) { dst = page_address(bv.bv_page) + bv.bv_offset; for (off = 0; off < bv.bv_len; off += blksize) { - if (dst && rq_data_dir(req) == READ) { + if (first_blk + blk_count >= end_blk) { + cqr->proc_bytes = blk_count * blksize; + return 0; + } + if (dst && !skip_block) { dst += off; memset(dst, 0, blksize); + } else { + skip_block--; } + blk_count++; } } + return 0; } /* diff --git a/drivers/s390/block/dasd_int.h b/drivers/s390/block/dasd_int.h index 91c9f9586e0f..fa552f9f1666 100644 --- a/drivers/s390/block/dasd_int.h +++ b/drivers/s390/block/dasd_int.h @@ -187,6 +187,7 @@ struct dasd_ccw_req { void (*callback)(struct dasd_ccw_req *, void *data); void *callback_data; + unsigned int proc_bytes; /* bytes for partial completion */ }; /* @@ -387,8 +388,9 @@ struct dasd_discipline { int (*ext_pool_warn_thrshld)(struct dasd_device *); int (*ext_pool_oos)(struct dasd_device *); int (*ext_pool_exhaust)(struct dasd_device *, struct dasd_ccw_req *); - struct dasd_ccw_req *(*ese_format)(struct dasd_device *, struct dasd_ccw_req *); - void (*ese_read)(struct dasd_ccw_req *); + struct dasd_ccw_req *(*ese_format)(struct dasd_device *, + struct dasd_ccw_req *, struct irb *); + int (*ese_read)(struct dasd_ccw_req *, struct irb *); }; extern struct dasd_discipline *dasd_diag_discipline_pointer; @@ -474,6 +476,11 @@ struct dasd_profile { spinlock_t lock; }; +struct dasd_format_entry { + struct list_head list; + sector_t track; +}; + struct dasd_device { /* Block device stuff. */ struct dasd_block *block; @@ -539,6 +546,7 @@ struct dasd_device { struct dentry *debugfs_dentry; struct dentry *hosts_dentry; struct dasd_profile profile; + struct dasd_format_entry format_entry; }; struct dasd_block { @@ -564,6 +572,9 @@ struct dasd_block { struct dentry *debugfs_dentry; struct dasd_profile profile; + + struct list_head format_list; + spinlock_t format_lock; }; struct dasd_attention_data { -- cgit v1.2.3