diff options
author | Jason Gunthorpe <jgg@mellanox.com> | 2019-06-28 21:18:23 -0300 |
---|---|---|
committer | Jason Gunthorpe <jgg@mellanox.com> | 2019-06-28 21:18:23 -0300 |
commit | 371bb62158d53c1fc33e2fb9b6aeb9522caf6cf4 (patch) | |
tree | 9442ae9b22e3cf24fbe8dcefae8862a3e4e05629 /drivers/nvme/host | |
parent | 10dcc7448e9ea49488a38bca7551de1a9da06ad9 (diff) | |
parent | 4b972a01a7da614b4796475f933094751a295a2f (diff) | |
download | linux-371bb62158d53c1fc33e2fb9b6aeb9522caf6cf4.tar.bz2 |
Merge tag 'v5.2-rc6' into rdma.git for-next
For dependencies in next patches.
Resolve conflicts:
- Use uverbs_get_cleared_udata() with new cq allocation flow
- Continue to delete nes despite SPDX conflict
- Resolve list appends in mlx5_command_str()
- Use u16 for vport_rule stuff
- Resolve list appends in struct ib_client
Signed-off-by: Jason Gunthorpe <jgg@mellanox.com>
Diffstat (limited to 'drivers/nvme/host')
-rw-r--r-- | drivers/nvme/host/Kconfig | 1 | ||||
-rw-r--r-- | drivers/nvme/host/core.c | 92 | ||||
-rw-r--r-- | drivers/nvme/host/nvme.h | 1 | ||||
-rw-r--r-- | drivers/nvme/host/pci.c | 33 | ||||
-rw-r--r-- | drivers/nvme/host/rdma.c | 152 | ||||
-rw-r--r-- | drivers/nvme/host/tcp.c | 57 |
6 files changed, 232 insertions, 104 deletions
diff --git a/drivers/nvme/host/Kconfig b/drivers/nvme/host/Kconfig index 0f345e207675..ec43ac9199e2 100644 --- a/drivers/nvme/host/Kconfig +++ b/drivers/nvme/host/Kconfig @@ -1,3 +1,4 @@ +# SPDX-License-Identifier: GPL-2.0-only config NVME_CORE tristate diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c index 7da80f375315..120fb593d1da 100644 --- a/drivers/nvme/host/core.c +++ b/drivers/nvme/host/core.c @@ -1257,9 +1257,9 @@ static u32 nvme_passthru_start(struct nvme_ctrl *ctrl, struct nvme_ns *ns, return 0; } - effects |= nvme_known_admin_effects(opcode); if (ctrl->effects) effects = le32_to_cpu(ctrl->effects->acs[opcode]); + effects |= nvme_known_admin_effects(opcode); /* * For simplicity, IO to all namespaces is quiesced even if the command @@ -1361,9 +1361,14 @@ static struct nvme_ns *nvme_get_ns_from_disk(struct gendisk *disk, { #ifdef CONFIG_NVME_MULTIPATH if (disk->fops == &nvme_ns_head_ops) { + struct nvme_ns *ns; + *head = disk->private_data; *srcu_idx = srcu_read_lock(&(*head)->srcu); - return nvme_find_path(*head); + ns = nvme_find_path(*head); + if (!ns) + srcu_read_unlock(&(*head)->srcu, *srcu_idx); + return ns; } #endif *head = NULL; @@ -1377,42 +1382,56 @@ static void nvme_put_ns_from_disk(struct nvme_ns_head *head, int idx) srcu_read_unlock(&head->srcu, idx); } -static int nvme_ns_ioctl(struct nvme_ns *ns, unsigned cmd, unsigned long arg) +static int nvme_ioctl(struct block_device *bdev, fmode_t mode, + unsigned int cmd, unsigned long arg) { + struct nvme_ns_head *head = NULL; + void __user *argp = (void __user *)arg; + struct nvme_ns *ns; + int srcu_idx, ret; + + ns = nvme_get_ns_from_disk(bdev->bd_disk, &head, &srcu_idx); + if (unlikely(!ns)) + return -EWOULDBLOCK; + + /* + * Handle ioctls that apply to the controller instead of the namespace + * seperately and drop the ns SRCU reference early. This avoids a + * deadlock when deleting namespaces using the passthrough interface. + */ + if (cmd == NVME_IOCTL_ADMIN_CMD || is_sed_ioctl(cmd)) { + struct nvme_ctrl *ctrl = ns->ctrl; + + nvme_get_ctrl(ns->ctrl); + nvme_put_ns_from_disk(head, srcu_idx); + + if (cmd == NVME_IOCTL_ADMIN_CMD) + ret = nvme_user_cmd(ctrl, NULL, argp); + else + ret = sed_ioctl(ctrl->opal_dev, cmd, argp); + + nvme_put_ctrl(ctrl); + return ret; + } + switch (cmd) { case NVME_IOCTL_ID: force_successful_syscall_return(); - return ns->head->ns_id; - case NVME_IOCTL_ADMIN_CMD: - return nvme_user_cmd(ns->ctrl, NULL, (void __user *)arg); + ret = ns->head->ns_id; + break; case NVME_IOCTL_IO_CMD: - return nvme_user_cmd(ns->ctrl, ns, (void __user *)arg); + ret = nvme_user_cmd(ns->ctrl, ns, argp); + break; case NVME_IOCTL_SUBMIT_IO: - return nvme_submit_io(ns, (void __user *)arg); + ret = nvme_submit_io(ns, argp); + break; default: -#ifdef CONFIG_NVM if (ns->ndev) - return nvme_nvm_ioctl(ns, cmd, arg); -#endif - if (is_sed_ioctl(cmd)) - return sed_ioctl(ns->ctrl->opal_dev, cmd, - (void __user *) arg); - return -ENOTTY; + ret = nvme_nvm_ioctl(ns, cmd, arg); + else + ret = -ENOTTY; } -} - -static int nvme_ioctl(struct block_device *bdev, fmode_t mode, - unsigned int cmd, unsigned long arg) -{ - struct nvme_ns_head *head = NULL; - struct nvme_ns *ns; - int srcu_idx, ret; - ns = nvme_get_ns_from_disk(bdev->bd_disk, &head, &srcu_idx); - if (unlikely(!ns)) - ret = -EWOULDBLOCK; - else - ret = nvme_ns_ioctl(ns, cmd, arg); nvme_put_ns_from_disk(head, srcu_idx); return ret; } @@ -2557,6 +2576,7 @@ int nvme_init_identify(struct nvme_ctrl *ctrl) ctrl->oacs = le16_to_cpu(id->oacs); ctrl->oncs = le16_to_cpu(id->oncs); + ctrl->mtfa = le16_to_cpu(id->mtfa); ctrl->oaes = le32_to_cpu(id->oaes); atomic_set(&ctrl->abort_limit, id->acl + 1); ctrl->vwc = id->vwc; @@ -3380,7 +3400,8 @@ static int nvme_scan_ns_list(struct nvme_ctrl *ctrl, unsigned nn) { struct nvme_ns *ns; __le32 *ns_list; - unsigned i, j, nsid, prev = 0, num_lists = DIV_ROUND_UP(nn, 1024); + unsigned i, j, nsid, prev = 0; + unsigned num_lists = DIV_ROUND_UP_ULL((u64)nn, 1024); int ret = 0; ns_list = kzalloc(NVME_IDENTIFY_DATA_SIZE, GFP_KERNEL); @@ -3681,6 +3702,7 @@ EXPORT_SYMBOL_GPL(nvme_start_ctrl); void nvme_uninit_ctrl(struct nvme_ctrl *ctrl) { + dev_pm_qos_hide_latency_tolerance(ctrl->device); cdev_device_del(&ctrl->cdev, ctrl->device); } EXPORT_SYMBOL_GPL(nvme_uninit_ctrl); @@ -3880,6 +3902,18 @@ void nvme_start_queues(struct nvme_ctrl *ctrl) } EXPORT_SYMBOL_GPL(nvme_start_queues); + +void nvme_sync_queues(struct nvme_ctrl *ctrl) +{ + struct nvme_ns *ns; + + down_read(&ctrl->namespaces_rwsem); + list_for_each_entry(ns, &ctrl->namespaces, list) + blk_sync_queue(ns->queue); + up_read(&ctrl->namespaces_rwsem); +} +EXPORT_SYMBOL_GPL(nvme_sync_queues); + /* * Check we didn't inadvertently grow the command structure sizes: */ diff --git a/drivers/nvme/host/nvme.h b/drivers/nvme/host/nvme.h index 5ee75b5ff83f..55553d293a98 100644 --- a/drivers/nvme/host/nvme.h +++ b/drivers/nvme/host/nvme.h @@ -441,6 +441,7 @@ void nvme_complete_async_event(struct nvme_ctrl *ctrl, __le16 status, void nvme_stop_queues(struct nvme_ctrl *ctrl); void nvme_start_queues(struct nvme_ctrl *ctrl); void nvme_kill_queues(struct nvme_ctrl *ctrl); +void nvme_sync_queues(struct nvme_ctrl *ctrl); void nvme_unfreeze(struct nvme_ctrl *ctrl); void nvme_wait_freeze(struct nvme_ctrl *ctrl); void nvme_wait_freeze_timeout(struct nvme_ctrl *ctrl, long timeout); diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c index 2a8708c9ac18..524d6bd6d095 100644 --- a/drivers/nvme/host/pci.c +++ b/drivers/nvme/host/pci.c @@ -464,7 +464,7 @@ static int nvme_pci_map_queues(struct blk_mq_tag_set *set) * affinity), so use the regular blk-mq cpu mapping */ map->queue_offset = qoff; - if (i != HCTX_TYPE_POLL) + if (i != HCTX_TYPE_POLL && offset) blk_mq_pci_map_queues(map, to_pci_dev(dev->dev), offset); else blk_mq_map_queues(map); @@ -1257,7 +1257,6 @@ static enum blk_eh_timer_return nvme_timeout(struct request *req, bool reserved) struct nvme_dev *dev = nvmeq->dev; struct request *abort_req; struct nvme_command cmd; - bool shutdown = false; u32 csts = readl(dev->bar + NVME_REG_CSTS); /* If PCI error recovery process is happening, we cannot reset or @@ -1294,17 +1293,18 @@ static enum blk_eh_timer_return nvme_timeout(struct request *req, bool reserved) * shutdown, so we return BLK_EH_DONE. */ switch (dev->ctrl.state) { - case NVME_CTRL_DELETING: - shutdown = true; - /* fall through */ case NVME_CTRL_CONNECTING: - case NVME_CTRL_RESETTING: + nvme_change_ctrl_state(&dev->ctrl, NVME_CTRL_DELETING); + /* fall through */ + case NVME_CTRL_DELETING: dev_warn_ratelimited(dev->ctrl.device, "I/O %d QID %d timeout, disable controller\n", req->tag, nvmeq->qid); - nvme_dev_disable(dev, shutdown); + nvme_dev_disable(dev, true); nvme_req(req)->flags |= NVME_REQ_CANCELLED; return BLK_EH_DONE; + case NVME_CTRL_RESETTING: + return BLK_EH_RESET_TIMER; default: break; } @@ -2376,7 +2376,7 @@ static void nvme_pci_disable(struct nvme_dev *dev) static void nvme_dev_disable(struct nvme_dev *dev, bool shutdown) { - bool dead = true; + bool dead = true, freeze = false; struct pci_dev *pdev = to_pci_dev(dev->dev); mutex_lock(&dev->shutdown_lock); @@ -2384,8 +2384,10 @@ static void nvme_dev_disable(struct nvme_dev *dev, bool shutdown) u32 csts = readl(dev->bar + NVME_REG_CSTS); if (dev->ctrl.state == NVME_CTRL_LIVE || - dev->ctrl.state == NVME_CTRL_RESETTING) + dev->ctrl.state == NVME_CTRL_RESETTING) { + freeze = true; nvme_start_freeze(&dev->ctrl); + } dead = !!((csts & NVME_CSTS_CFS) || !(csts & NVME_CSTS_RDY) || pdev->error_state != pci_channel_io_normal); } @@ -2394,10 +2396,8 @@ static void nvme_dev_disable(struct nvme_dev *dev, bool shutdown) * Give the controller a chance to complete all entered requests if * doing a safe shutdown. */ - if (!dead) { - if (shutdown) - nvme_wait_freeze_timeout(&dev->ctrl, NVME_IO_TIMEOUT); - } + if (!dead && shutdown && freeze) + nvme_wait_freeze_timeout(&dev->ctrl, NVME_IO_TIMEOUT); nvme_stop_queues(&dev->ctrl); @@ -2492,6 +2492,7 @@ static void nvme_reset_work(struct work_struct *work) */ if (dev->ctrl.ctrl_config & NVME_CC_ENABLE) nvme_dev_disable(dev, false); + nvme_sync_queues(&dev->ctrl); mutex_lock(&dev->shutdown_lock); result = nvme_pci_enable(dev); @@ -2512,6 +2513,12 @@ static void nvme_reset_work(struct work_struct *work) */ dev->ctrl.max_hw_sectors = NVME_MAX_KB_SZ << 1; dev->ctrl.max_segments = NVME_MAX_SEGS; + + /* + * Don't limit the IOMMU merged segment size. + */ + dma_set_max_seg_size(dev->dev, 0xffffffff); + mutex_unlock(&dev->shutdown_lock); /* diff --git a/drivers/nvme/host/rdma.c b/drivers/nvme/host/rdma.c index 0e033b621daf..f3e3a8c861d6 100644 --- a/drivers/nvme/host/rdma.c +++ b/drivers/nvme/host/rdma.c @@ -213,6 +213,11 @@ static struct nvme_rdma_qe *nvme_rdma_alloc_ring(struct ib_device *ibdev, if (!ring) return NULL; + /* + * Bind the CQEs (post recv buffers) DMA mapping to the RDMA queue + * lifetime. It's safe, since any chage in the underlying RDMA device + * will issue error recovery and queue re-creation. + */ for (i = 0; i < ib_queue_size; i++) { if (nvme_rdma_alloc_qe(ibdev, &ring[i], capsule_size, dir)) goto out_free_ring; @@ -274,14 +279,9 @@ static int nvme_rdma_create_qp(struct nvme_rdma_queue *queue, const int factor) static void nvme_rdma_exit_request(struct blk_mq_tag_set *set, struct request *rq, unsigned int hctx_idx) { - struct nvme_rdma_ctrl *ctrl = set->driver_data; struct nvme_rdma_request *req = blk_mq_rq_to_pdu(rq); - int queue_idx = (set == &ctrl->tag_set) ? hctx_idx + 1 : 0; - struct nvme_rdma_queue *queue = &ctrl->queues[queue_idx]; - struct nvme_rdma_device *dev = queue->device; - nvme_rdma_free_qe(dev->dev, &req->sqe, sizeof(struct nvme_command), - DMA_TO_DEVICE); + kfree(req->sqe.data); } static int nvme_rdma_init_request(struct blk_mq_tag_set *set, @@ -292,15 +292,11 @@ static int nvme_rdma_init_request(struct blk_mq_tag_set *set, struct nvme_rdma_request *req = blk_mq_rq_to_pdu(rq); int queue_idx = (set == &ctrl->tag_set) ? hctx_idx + 1 : 0; struct nvme_rdma_queue *queue = &ctrl->queues[queue_idx]; - struct nvme_rdma_device *dev = queue->device; - struct ib_device *ibdev = dev->dev; - int ret; nvme_req(rq)->ctrl = &ctrl->ctrl; - ret = nvme_rdma_alloc_qe(ibdev, &req->sqe, sizeof(struct nvme_command), - DMA_TO_DEVICE); - if (ret) - return ret; + req->sqe.data = kzalloc(sizeof(struct nvme_command), GFP_KERNEL); + if (!req->sqe.data) + return -ENOMEM; req->queue = queue; @@ -641,34 +637,16 @@ static int nvme_rdma_alloc_io_queues(struct nvme_rdma_ctrl *ctrl) { struct nvmf_ctrl_options *opts = ctrl->ctrl.opts; struct ib_device *ibdev = ctrl->device->dev; - unsigned int nr_io_queues; + unsigned int nr_io_queues, nr_default_queues; + unsigned int nr_read_queues, nr_poll_queues; int i, ret; - nr_io_queues = min(opts->nr_io_queues, num_online_cpus()); - - /* - * we map queues according to the device irq vectors for - * optimal locality so we don't need more queues than - * completion vectors. - */ - nr_io_queues = min_t(unsigned int, nr_io_queues, - ibdev->num_comp_vectors); - - if (opts->nr_write_queues) { - ctrl->io_queues[HCTX_TYPE_DEFAULT] = - min(opts->nr_write_queues, nr_io_queues); - nr_io_queues += ctrl->io_queues[HCTX_TYPE_DEFAULT]; - } else { - ctrl->io_queues[HCTX_TYPE_DEFAULT] = nr_io_queues; - } - - ctrl->io_queues[HCTX_TYPE_READ] = nr_io_queues; - - if (opts->nr_poll_queues) { - ctrl->io_queues[HCTX_TYPE_POLL] = - min(opts->nr_poll_queues, num_online_cpus()); - nr_io_queues += ctrl->io_queues[HCTX_TYPE_POLL]; - } + nr_read_queues = min_t(unsigned int, ibdev->num_comp_vectors, + min(opts->nr_io_queues, num_online_cpus())); + nr_default_queues = min_t(unsigned int, ibdev->num_comp_vectors, + min(opts->nr_write_queues, num_online_cpus())); + nr_poll_queues = min(opts->nr_poll_queues, num_online_cpus()); + nr_io_queues = nr_read_queues + nr_default_queues + nr_poll_queues; ret = nvme_set_queue_count(&ctrl->ctrl, &nr_io_queues); if (ret) @@ -681,6 +659,34 @@ static int nvme_rdma_alloc_io_queues(struct nvme_rdma_ctrl *ctrl) dev_info(ctrl->ctrl.device, "creating %d I/O queues.\n", nr_io_queues); + if (opts->nr_write_queues && nr_read_queues < nr_io_queues) { + /* + * separate read/write queues + * hand out dedicated default queues only after we have + * sufficient read queues. + */ + ctrl->io_queues[HCTX_TYPE_READ] = nr_read_queues; + nr_io_queues -= ctrl->io_queues[HCTX_TYPE_READ]; + ctrl->io_queues[HCTX_TYPE_DEFAULT] = + min(nr_default_queues, nr_io_queues); + nr_io_queues -= ctrl->io_queues[HCTX_TYPE_DEFAULT]; + } else { + /* + * shared read/write queues + * either no write queues were requested, or we don't have + * sufficient queue count to have dedicated default queues. + */ + ctrl->io_queues[HCTX_TYPE_DEFAULT] = + min(nr_read_queues, nr_io_queues); + nr_io_queues -= ctrl->io_queues[HCTX_TYPE_DEFAULT]; + } + + if (opts->nr_poll_queues && nr_io_queues) { + /* map dedicated poll queues only if we have queues left */ + ctrl->io_queues[HCTX_TYPE_POLL] = + min(nr_poll_queues, nr_io_queues); + } + for (i = 1; i < ctrl->ctrl.queue_count; i++) { ret = nvme_rdma_alloc_queue(ctrl, i, ctrl->ctrl.sqsize + 1); @@ -769,6 +775,11 @@ static int nvme_rdma_configure_admin_queue(struct nvme_rdma_ctrl *ctrl, ctrl->max_fr_pages = nvme_rdma_get_max_fr_pages(ctrl->device->dev); + /* + * Bind the async event SQE DMA mapping to the admin queue lifetime. + * It's safe, since any chage in the underlying RDMA device will issue + * error recovery and queue re-creation. + */ error = nvme_rdma_alloc_qe(ctrl->device->dev, &ctrl->async_event_sqe, sizeof(struct nvme_command), DMA_TO_DEVICE); if (error) @@ -1709,12 +1720,20 @@ static blk_status_t nvme_rdma_queue_rq(struct blk_mq_hw_ctx *hctx, return nvmf_fail_nonready_command(&queue->ctrl->ctrl, rq); dev = queue->device->dev; + + req->sqe.dma = ib_dma_map_single(dev, req->sqe.data, + sizeof(struct nvme_command), + DMA_TO_DEVICE); + err = ib_dma_mapping_error(dev, req->sqe.dma); + if (unlikely(err)) + return BLK_STS_RESOURCE; + ib_dma_sync_single_for_cpu(dev, sqe->dma, sizeof(struct nvme_command), DMA_TO_DEVICE); ret = nvme_setup_cmd(ns, rq, c); if (ret) - return ret; + goto unmap_qe; blk_mq_start_request(rq); @@ -1739,10 +1758,16 @@ static blk_status_t nvme_rdma_queue_rq(struct blk_mq_hw_ctx *hctx, } return BLK_STS_OK; + err: if (err == -ENOMEM || err == -EAGAIN) - return BLK_STS_RESOURCE; - return BLK_STS_IOERR; + ret = BLK_STS_RESOURCE; + else + ret = BLK_STS_IOERR; +unmap_qe: + ib_dma_unmap_single(dev, req->sqe.dma, sizeof(struct nvme_command), + DMA_TO_DEVICE); + return ret; } static int nvme_rdma_poll(struct blk_mq_hw_ctx *hctx) @@ -1755,25 +1780,36 @@ static int nvme_rdma_poll(struct blk_mq_hw_ctx *hctx) static void nvme_rdma_complete_rq(struct request *rq) { struct nvme_rdma_request *req = blk_mq_rq_to_pdu(rq); + struct nvme_rdma_queue *queue = req->queue; + struct ib_device *ibdev = queue->device->dev; - nvme_rdma_unmap_data(req->queue, rq); + nvme_rdma_unmap_data(queue, rq); + ib_dma_unmap_single(ibdev, req->sqe.dma, sizeof(struct nvme_command), + DMA_TO_DEVICE); nvme_complete_rq(rq); } static int nvme_rdma_map_queues(struct blk_mq_tag_set *set) { struct nvme_rdma_ctrl *ctrl = set->driver_data; + struct nvmf_ctrl_options *opts = ctrl->ctrl.opts; - set->map[HCTX_TYPE_DEFAULT].queue_offset = 0; - set->map[HCTX_TYPE_DEFAULT].nr_queues = - ctrl->io_queues[HCTX_TYPE_DEFAULT]; - set->map[HCTX_TYPE_READ].nr_queues = ctrl->io_queues[HCTX_TYPE_READ]; - if (ctrl->ctrl.opts->nr_write_queues) { + if (opts->nr_write_queues && ctrl->io_queues[HCTX_TYPE_READ]) { /* separate read/write queues */ + set->map[HCTX_TYPE_DEFAULT].nr_queues = + ctrl->io_queues[HCTX_TYPE_DEFAULT]; + set->map[HCTX_TYPE_DEFAULT].queue_offset = 0; + set->map[HCTX_TYPE_READ].nr_queues = + ctrl->io_queues[HCTX_TYPE_READ]; set->map[HCTX_TYPE_READ].queue_offset = - ctrl->io_queues[HCTX_TYPE_DEFAULT]; + ctrl->io_queues[HCTX_TYPE_DEFAULT]; } else { - /* mixed read/write queues */ + /* shared read/write queues */ + set->map[HCTX_TYPE_DEFAULT].nr_queues = + ctrl->io_queues[HCTX_TYPE_DEFAULT]; + set->map[HCTX_TYPE_DEFAULT].queue_offset = 0; + set->map[HCTX_TYPE_READ].nr_queues = + ctrl->io_queues[HCTX_TYPE_DEFAULT]; set->map[HCTX_TYPE_READ].queue_offset = 0; } blk_mq_rdma_map_queues(&set->map[HCTX_TYPE_DEFAULT], @@ -1781,16 +1817,22 @@ static int nvme_rdma_map_queues(struct blk_mq_tag_set *set) blk_mq_rdma_map_queues(&set->map[HCTX_TYPE_READ], ctrl->device->dev, 0); - if (ctrl->ctrl.opts->nr_poll_queues) { + if (opts->nr_poll_queues && ctrl->io_queues[HCTX_TYPE_POLL]) { + /* map dedicated poll queues only if we have queues left */ set->map[HCTX_TYPE_POLL].nr_queues = ctrl->io_queues[HCTX_TYPE_POLL]; set->map[HCTX_TYPE_POLL].queue_offset = - ctrl->io_queues[HCTX_TYPE_DEFAULT]; - if (ctrl->ctrl.opts->nr_write_queues) - set->map[HCTX_TYPE_POLL].queue_offset += - ctrl->io_queues[HCTX_TYPE_READ]; + ctrl->io_queues[HCTX_TYPE_DEFAULT] + + ctrl->io_queues[HCTX_TYPE_READ]; blk_mq_map_queues(&set->map[HCTX_TYPE_POLL]); } + + dev_info(ctrl->ctrl.device, + "mapped %d/%d/%d default/read/poll queues.\n", + ctrl->io_queues[HCTX_TYPE_DEFAULT], + ctrl->io_queues[HCTX_TYPE_READ], + ctrl->io_queues[HCTX_TYPE_POLL]); + return 0; } diff --git a/drivers/nvme/host/tcp.c b/drivers/nvme/host/tcp.c index 2b107a1d152b..08a2501b9357 100644 --- a/drivers/nvme/host/tcp.c +++ b/drivers/nvme/host/tcp.c @@ -111,6 +111,7 @@ struct nvme_tcp_ctrl { struct work_struct err_work; struct delayed_work connect_work; struct nvme_tcp_request async_req; + u32 io_queues[HCTX_MAX_TYPES]; }; static LIST_HEAD(nvme_tcp_ctrl_list); @@ -1564,6 +1565,35 @@ static unsigned int nvme_tcp_nr_io_queues(struct nvme_ctrl *ctrl) return nr_io_queues; } +static void nvme_tcp_set_io_queues(struct nvme_ctrl *nctrl, + unsigned int nr_io_queues) +{ + struct nvme_tcp_ctrl *ctrl = to_tcp_ctrl(nctrl); + struct nvmf_ctrl_options *opts = nctrl->opts; + + if (opts->nr_write_queues && opts->nr_io_queues < nr_io_queues) { + /* + * separate read/write queues + * hand out dedicated default queues only after we have + * sufficient read queues. + */ + ctrl->io_queues[HCTX_TYPE_READ] = opts->nr_io_queues; + nr_io_queues -= ctrl->io_queues[HCTX_TYPE_READ]; + ctrl->io_queues[HCTX_TYPE_DEFAULT] = + min(opts->nr_write_queues, nr_io_queues); + nr_io_queues -= ctrl->io_queues[HCTX_TYPE_DEFAULT]; + } else { + /* + * shared read/write queues + * either no write queues were requested, or we don't have + * sufficient queue count to have dedicated default queues. + */ + ctrl->io_queues[HCTX_TYPE_DEFAULT] = + min(opts->nr_io_queues, nr_io_queues); + nr_io_queues -= ctrl->io_queues[HCTX_TYPE_DEFAULT]; + } +} + static int nvme_tcp_alloc_io_queues(struct nvme_ctrl *ctrl) { unsigned int nr_io_queues; @@ -1581,6 +1611,8 @@ static int nvme_tcp_alloc_io_queues(struct nvme_ctrl *ctrl) dev_info(ctrl->device, "creating %d I/O queues.\n", nr_io_queues); + nvme_tcp_set_io_queues(ctrl, nr_io_queues); + return __nvme_tcp_alloc_io_queues(ctrl); } @@ -2089,23 +2121,34 @@ static blk_status_t nvme_tcp_queue_rq(struct blk_mq_hw_ctx *hctx, static int nvme_tcp_map_queues(struct blk_mq_tag_set *set) { struct nvme_tcp_ctrl *ctrl = set->driver_data; + struct nvmf_ctrl_options *opts = ctrl->ctrl.opts; - set->map[HCTX_TYPE_DEFAULT].queue_offset = 0; - set->map[HCTX_TYPE_READ].nr_queues = ctrl->ctrl.opts->nr_io_queues; - if (ctrl->ctrl.opts->nr_write_queues) { + if (opts->nr_write_queues && ctrl->io_queues[HCTX_TYPE_READ]) { /* separate read/write queues */ set->map[HCTX_TYPE_DEFAULT].nr_queues = - ctrl->ctrl.opts->nr_write_queues; + ctrl->io_queues[HCTX_TYPE_DEFAULT]; + set->map[HCTX_TYPE_DEFAULT].queue_offset = 0; + set->map[HCTX_TYPE_READ].nr_queues = + ctrl->io_queues[HCTX_TYPE_READ]; set->map[HCTX_TYPE_READ].queue_offset = - ctrl->ctrl.opts->nr_write_queues; + ctrl->io_queues[HCTX_TYPE_DEFAULT]; } else { - /* mixed read/write queues */ + /* shared read/write queues */ set->map[HCTX_TYPE_DEFAULT].nr_queues = - ctrl->ctrl.opts->nr_io_queues; + ctrl->io_queues[HCTX_TYPE_DEFAULT]; + set->map[HCTX_TYPE_DEFAULT].queue_offset = 0; + set->map[HCTX_TYPE_READ].nr_queues = + ctrl->io_queues[HCTX_TYPE_DEFAULT]; set->map[HCTX_TYPE_READ].queue_offset = 0; } blk_mq_map_queues(&set->map[HCTX_TYPE_DEFAULT]); blk_mq_map_queues(&set->map[HCTX_TYPE_READ]); + + dev_info(ctrl->ctrl.device, + "mapped %d/%d default/read queues.\n", + ctrl->io_queues[HCTX_TYPE_DEFAULT], + ctrl->io_queues[HCTX_TYPE_READ]); + return 0; } |