From 22c17e279a1b03bad7987e4a4192b289b890f293 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 22 Nov 2022 08:27:53 +0100 Subject: blk-mq: fix queue reference leak on blk_mq_alloc_disk_for_queue failure Drop the request queue reference just acquired when __alloc_disk_node failed. Fixes: 6f8191fdf41d ("block: simplify disk shutdown") Reported-by: Al Viro Signed-off-by: Christoph Hellwig Reviewed-by: Damien Le Moal Link: https://lore.kernel.org/r/20221122072753.426077-1-hch@lst.de Signed-off-by: Jens Axboe --- block/blk-mq.c | 7 ++++++- 1 file changed, 6 insertions(+), 1 deletion(-) diff --git a/block/blk-mq.c b/block/blk-mq.c index 6a789cda68a5..228a6696d835 100644 --- a/block/blk-mq.c +++ b/block/blk-mq.c @@ -4045,9 +4045,14 @@ EXPORT_SYMBOL(__blk_mq_alloc_disk); struct gendisk *blk_mq_alloc_disk_for_queue(struct request_queue *q, struct lock_class_key *lkclass) { + struct gendisk *disk; + if (!blk_get_queue(q)) return NULL; - return __alloc_disk_node(q, NUMA_NO_NODE, lkclass); + disk = __alloc_disk_node(q, NUMA_NO_NODE, lkclass); + if (!disk) + blk_put_queue(q); + return disk; } EXPORT_SYMBOL(blk_mq_alloc_disk_for_queue); -- cgit v1.2.3 From 4ba5f0c36cfdda68347269c02961cd90f8443ace Mon Sep 17 00:00:00 2001 From: Colin Ian King Date: Wed, 23 Nov 2022 17:07:16 +0100 Subject: s390/dasd: Fix spelling mistake "Ivalid" -> "Invalid" There is a spelling mistake in a pr_warn message. Fix it. Signed-off-by: Colin Ian King Link: https://lore.kernel.org/r/20220923132103.2486724-1-colin.i.king@gmail.com Signed-off-by: Stefan Haberland Link: https://lore.kernel.org/r/20221123160719.3002694-2-sth@linux.ibm.com Signed-off-by: Jens Axboe --- drivers/s390/block/dasd_ioctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/s390/block/dasd_ioctl.c b/drivers/s390/block/dasd_ioctl.c index d0ddf2cc9786..9327dcdd6e5e 100644 --- a/drivers/s390/block/dasd_ioctl.c +++ b/drivers/s390/block/dasd_ioctl.c @@ -401,7 +401,7 @@ dasd_ioctl_copy_pair_swap(struct block_device *bdev, void __user *argp) return -EFAULT; } if (memchr_inv(data.reserved, 0, sizeof(data.reserved))) { - pr_warn("%s: Ivalid swap data specified.\n", + pr_warn("%s: Invalid swap data specified\n", dev_name(&device->cdev->dev)); dasd_put_device(device); return DASD_COPYPAIRSWAP_INVALID; -- cgit v1.2.3 From b49e648fcca7e420c4ad670a548e19f0e8531c30 Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Wed, 23 Nov 2022 17:07:17 +0100 Subject: s390/dasd: increase printing of debug data payload 32 byte are to less for important data from prefix or other commands. Print up to 128 byte data. This is enough for the largest CCW data we have. Since printk can only print up to 1024 byte at once, print the different parts of the CCW dumps separately. Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Link: https://lore.kernel.org/r/20221123160719.3002694-3-sth@linux.ibm.com Signed-off-by: Jens Axboe --- drivers/s390/block/dasd_eckd.c | 37 +++++++++++++++++++------------------ 1 file changed, 19 insertions(+), 18 deletions(-) diff --git a/drivers/s390/block/dasd_eckd.c b/drivers/s390/block/dasd_eckd.c index 662730f3b027..85bf045c2ff9 100644 --- a/drivers/s390/block/dasd_eckd.c +++ b/drivers/s390/block/dasd_eckd.c @@ -5500,7 +5500,7 @@ dasd_eckd_ioctl(struct dasd_block *block, unsigned int cmd, void __user *argp) * Dump the range of CCWs into 'page' buffer * and return number of printed chars. */ -static int +static void dasd_eckd_dump_ccw_range(struct ccw1 *from, struct ccw1 *to, char *page) { int len, count; @@ -5518,16 +5518,21 @@ dasd_eckd_dump_ccw_range(struct ccw1 *from, struct ccw1 *to, char *page) else datap = (char *) ((addr_t) from->cda); - /* dump data (max 32 bytes) */ - for (count = 0; count < from->count && count < 32; count++) { - if (count % 8 == 0) len += sprintf(page + len, " "); - if (count % 4 == 0) len += sprintf(page + len, " "); + /* dump data (max 128 bytes) */ + for (count = 0; count < from->count && count < 128; count++) { + if (count % 32 == 0) + len += sprintf(page + len, "\n"); + if (count % 8 == 0) + len += sprintf(page + len, " "); + if (count % 4 == 0) + len += sprintf(page + len, " "); len += sprintf(page + len, "%02x", datap[count]); } len += sprintf(page + len, "\n"); from++; } - return len; + if (len > 0) + printk(KERN_ERR "%s", page); } static void @@ -5619,37 +5624,33 @@ static void dasd_eckd_dump_sense_ccw(struct dasd_device *device, if (req) { /* req == NULL for unsolicited interrupts */ /* dump the Channel Program (max 140 Bytes per line) */ - /* Count CCW and print first CCWs (maximum 1024 % 140 = 7) */ + /* Count CCW and print first CCWs (maximum 7) */ first = req->cpaddr; for (last = first; last->flags & (CCW_FLAG_CC | CCW_FLAG_DC); last++); to = min(first + 6, last); - len = sprintf(page, PRINTK_HEADER - " Related CP in req: %p\n", req); - dasd_eckd_dump_ccw_range(first, to, page + len); - printk(KERN_ERR "%s", page); + printk(KERN_ERR PRINTK_HEADER " Related CP in req: %p\n", req); + dasd_eckd_dump_ccw_range(first, to, page); /* print failing CCW area (maximum 4) */ /* scsw->cda is either valid or zero */ - len = 0; from = ++to; fail = (struct ccw1 *)(addr_t) irb->scsw.cmd.cpa; /* failing CCW */ if (from < fail - 2) { from = fail - 2; /* there is a gap - print header */ - len += sprintf(page, PRINTK_HEADER "......\n"); + printk(KERN_ERR PRINTK_HEADER "......\n"); } to = min(fail + 1, last); - len += dasd_eckd_dump_ccw_range(from, to, page + len); + dasd_eckd_dump_ccw_range(from, to, page + len); /* print last CCWs (maximum 2) */ + len = 0; from = max(from, ++to); if (from < last - 1) { from = last - 1; /* there is a gap - print header */ - len += sprintf(page + len, PRINTK_HEADER "......\n"); + printk(KERN_ERR PRINTK_HEADER "......\n"); } - len += dasd_eckd_dump_ccw_range(from, last, page + len); - if (len > 0) - printk(KERN_ERR "%s", page); + dasd_eckd_dump_ccw_range(from, last, page + len); } free_page((unsigned long) page); } -- cgit v1.2.3 From 590ce6d96d6a224b470a3862c33a483d5022bfdb Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Wed, 23 Nov 2022 17:07:18 +0100 Subject: s390/dasd: fix no record found for raw_track_access For DASD devices in raw_track_access mode only full track images are read and written. For this purpose it is not necessary to do search operation in the locate record extended function. The documentation even states that this might fail if the searched record is not found on a track. Currently the driver sets a value of 1 in the search field for the first record after record zero. This is the default for disks not in raw_track_access mode but record 1 might be missing on a completely empty track. There has not been any problem with this on IBM storage servers but it might lead to errors with DASD devices on other vendors storage servers. Fix this by setting the search field to 0. Record zero is always available even on a completely empty track. Fixes: e4dbb0f2b5dd ("[S390] dasd: Add support for raw ECKD access.") Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Link: https://lore.kernel.org/r/20221123160719.3002694-4-sth@linux.ibm.com Signed-off-by: Jens Axboe --- drivers/s390/block/dasd_eckd.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) diff --git a/drivers/s390/block/dasd_eckd.c b/drivers/s390/block/dasd_eckd.c index 85bf045c2ff9..5d0b9991e91a 100644 --- a/drivers/s390/block/dasd_eckd.c +++ b/drivers/s390/block/dasd_eckd.c @@ -4722,7 +4722,6 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_raw(struct dasd_device *startdev, struct dasd_device *basedev; struct req_iterator iter; struct dasd_ccw_req *cqr; - unsigned int first_offs; unsigned int trkcount; unsigned long *idaws; unsigned int size; @@ -4756,7 +4755,6 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_raw(struct dasd_device *startdev, last_trk = (blk_rq_pos(req) + blk_rq_sectors(req) - 1) / DASD_RAW_SECTORS_PER_TRACK; trkcount = last_trk - first_trk + 1; - first_offs = 0; if (rq_data_dir(req) == READ) cmd = DASD_ECKD_CCW_READ_TRACK; @@ -4800,13 +4798,13 @@ static struct dasd_ccw_req *dasd_eckd_build_cp_raw(struct dasd_device *startdev, if (use_prefix) { prefix_LRE(ccw++, data, first_trk, last_trk, cmd, basedev, - startdev, 1, first_offs + 1, trkcount, 0, 0); + startdev, 1, 0, trkcount, 0, 0); } else { define_extent(ccw++, data, first_trk, last_trk, cmd, basedev, 0); ccw[-1].flags |= CCW_FLAG_CC; data += sizeof(struct DE_eckd_data); - locate_record_ext(ccw++, data, first_trk, first_offs + 1, + locate_record_ext(ccw++, data, first_trk, 0, trkcount, cmd, basedev, 0, 0); } -- cgit v1.2.3 From 7e8a05b47ba7200f333eefd19979eeb4d273ceec Mon Sep 17 00:00:00 2001 From: Stefan Haberland Date: Wed, 23 Nov 2022 17:07:19 +0100 Subject: s390/dasd: fix possible buffer overflow in copy_pair_show dasd_copy_relation->entry[] array might be accessed out of bounds if the loop does not break. Fixes: a91ff09d39f9 ("s390/dasd: add copy pair setup") Signed-off-by: Stefan Haberland Reviewed-by: Jan Hoeppner Link: https://lore.kernel.org/r/20221123160719.3002694-5-sth@linux.ibm.com Signed-off-by: Jens Axboe --- drivers/s390/block/dasd_devmap.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/s390/block/dasd_devmap.c b/drivers/s390/block/dasd_devmap.c index cb83f81da416..df17f0f9cb0f 100644 --- a/drivers/s390/block/dasd_devmap.c +++ b/drivers/s390/block/dasd_devmap.c @@ -1954,7 +1954,7 @@ dasd_copy_pair_show(struct device *dev, break; } } - if (!copy->entry[i].primary) + if (i == DASD_CP_ENTRIES) goto out; /* print all secondary */ -- cgit v1.2.3 From 7d4a93176e0142ce16d23c849a47d5b00b856296 Mon Sep 17 00:00:00 2001 From: Ming Lei Date: Mon, 21 Nov 2022 23:56:45 +0800 Subject: ublk_drv: don't forward io commands in reserve order Either ublk_can_use_task_work() is true or not, io commands are forwarded to ublk server in reverse order, since llist_add() is always to add one element to the head of the list. Even though block layer doesn't guarantee request dispatch order, requests should be sent to hardware in the sequence order generated from io scheduler, which usually considers the request's LBA, and order is often important for HDD. So forward io commands in the sequence made from io scheduler by aligning task work with current io_uring command's batch handling, and it has been observed that both can get similar performance data if IORING_SETUP_COOP_TASKRUN is set from ublk server. Reported-by: Andreas Hindborg Cc: Damien Le Moal Signed-off-by: Ming Lei Reviewed-by: Damien Le Moal Reviewed-by: ZiyangZhang Link: https://lore.kernel.org/r/20221121155645.396272-1-ming.lei@redhat.com Signed-off-by: Jens Axboe --- drivers/block/ublk_drv.c | 82 ++++++++++++++++++++++-------------------------- 1 file changed, 38 insertions(+), 44 deletions(-) diff --git a/drivers/block/ublk_drv.c b/drivers/block/ublk_drv.c index f96cb01e9604..e9de9d846b73 100644 --- a/drivers/block/ublk_drv.c +++ b/drivers/block/ublk_drv.c @@ -57,10 +57,8 @@ #define UBLK_PARAM_TYPE_ALL (UBLK_PARAM_TYPE_BASIC | UBLK_PARAM_TYPE_DISCARD) struct ublk_rq_data { - union { - struct callback_head work; - struct llist_node node; - }; + struct llist_node node; + struct callback_head work; }; struct ublk_uring_cmd_pdu { @@ -766,15 +764,31 @@ static inline void __ublk_rq_task_work(struct request *req) ubq_complete_io_cmd(io, UBLK_IO_RES_OK); } +static inline void ublk_forward_io_cmds(struct ublk_queue *ubq) +{ + struct llist_node *io_cmds = llist_del_all(&ubq->io_cmds); + struct ublk_rq_data *data, *tmp; + + io_cmds = llist_reverse_order(io_cmds); + llist_for_each_entry_safe(data, tmp, io_cmds, node) + __ublk_rq_task_work(blk_mq_rq_from_pdu(data)); +} + +static inline void ublk_abort_io_cmds(struct ublk_queue *ubq) +{ + struct llist_node *io_cmds = llist_del_all(&ubq->io_cmds); + struct ublk_rq_data *data, *tmp; + + llist_for_each_entry_safe(data, tmp, io_cmds, node) + __ublk_abort_rq(ubq, blk_mq_rq_from_pdu(data)); +} + static void ublk_rq_task_work_cb(struct io_uring_cmd *cmd) { struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd); struct ublk_queue *ubq = pdu->ubq; - struct llist_node *io_cmds = llist_del_all(&ubq->io_cmds); - struct ublk_rq_data *data; - llist_for_each_entry(data, io_cmds, node) - __ublk_rq_task_work(blk_mq_rq_from_pdu(data)); + ublk_forward_io_cmds(ubq); } static void ublk_rq_task_work_fn(struct callback_head *work) @@ -782,14 +796,20 @@ static void ublk_rq_task_work_fn(struct callback_head *work) struct ublk_rq_data *data = container_of(work, struct ublk_rq_data, work); struct request *req = blk_mq_rq_from_pdu(data); + struct ublk_queue *ubq = req->mq_hctx->driver_data; - __ublk_rq_task_work(req); + ublk_forward_io_cmds(ubq); } -static void ublk_submit_cmd(struct ublk_queue *ubq, const struct request *rq) +static void ublk_queue_cmd(struct ublk_queue *ubq, struct request *rq) { - struct ublk_io *io = &ubq->ios[rq->tag]; + struct ublk_rq_data *data = blk_mq_rq_to_pdu(rq); + struct ublk_io *io; + if (!llist_add(&data->node, &ubq->io_cmds)) + return; + + io = &ubq->ios[rq->tag]; /* * If the check pass, we know that this is a re-issued request aborted * previously in monitor_work because the ubq_daemon(cmd's task) is @@ -803,11 +823,11 @@ static void ublk_submit_cmd(struct ublk_queue *ubq, const struct request *rq) * guarantees that here is a re-issued request aborted previously. */ if (unlikely(io->flags & UBLK_IO_FLAG_ABORTED)) { - struct llist_node *io_cmds = llist_del_all(&ubq->io_cmds); - struct ublk_rq_data *data; - - llist_for_each_entry(data, io_cmds, node) - __ublk_abort_rq(ubq, blk_mq_rq_from_pdu(data)); + ublk_abort_io_cmds(ubq); + } else if (ublk_can_use_task_work(ubq)) { + if (task_work_add(ubq->ubq_daemon, &data->work, + TWA_SIGNAL_NO_IPI)) + ublk_abort_io_cmds(ubq); } else { struct io_uring_cmd *cmd = io->cmd; struct ublk_uring_cmd_pdu *pdu = ublk_get_uring_cmd_pdu(cmd); @@ -817,23 +837,6 @@ static void ublk_submit_cmd(struct ublk_queue *ubq, const struct request *rq) } } -static void ublk_queue_cmd(struct ublk_queue *ubq, struct request *rq, - bool last) -{ - struct ublk_rq_data *data = blk_mq_rq_to_pdu(rq); - - if (ublk_can_use_task_work(ubq)) { - enum task_work_notify_mode notify_mode = last ? - TWA_SIGNAL_NO_IPI : TWA_NONE; - - if (task_work_add(ubq->ubq_daemon, &data->work, notify_mode)) - __ublk_abort_rq(ubq, rq); - } else { - if (llist_add(&data->node, &ubq->io_cmds)) - ublk_submit_cmd(ubq, rq); - } -} - static blk_status_t ublk_queue_rq(struct blk_mq_hw_ctx *hctx, const struct blk_mq_queue_data *bd) { @@ -865,19 +868,11 @@ static blk_status_t ublk_queue_rq(struct blk_mq_hw_ctx *hctx, return BLK_STS_OK; } - ublk_queue_cmd(ubq, rq, bd->last); + ublk_queue_cmd(ubq, rq); return BLK_STS_OK; } -static void ublk_commit_rqs(struct blk_mq_hw_ctx *hctx) -{ - struct ublk_queue *ubq = hctx->driver_data; - - if (ublk_can_use_task_work(ubq)) - __set_notify_signal(ubq->ubq_daemon); -} - static int ublk_init_hctx(struct blk_mq_hw_ctx *hctx, void *driver_data, unsigned int hctx_idx) { @@ -899,7 +894,6 @@ static int ublk_init_rq(struct blk_mq_tag_set *set, struct request *req, static const struct blk_mq_ops ublk_mq_ops = { .queue_rq = ublk_queue_rq, - .commit_rqs = ublk_commit_rqs, .init_hctx = ublk_init_hctx, .init_request = ublk_init_rq, }; @@ -1197,7 +1191,7 @@ static void ublk_handle_need_get_data(struct ublk_device *ub, int q_id, struct ublk_queue *ubq = ublk_get_queue(ub, q_id); struct request *req = blk_mq_tag_to_rq(ub->tag_set.tags[q_id], tag); - ublk_queue_cmd(ubq, req, true); + ublk_queue_cmd(ubq, req); } static int ublk_ch_uring_cmd(struct io_uring_cmd *cmd, unsigned int issue_flags) -- cgit v1.2.3