From 76e0310c2d5a4fb62f674af7dad16f544950fc13 Mon Sep 17 00:00:00 2001 From: Sachin Kamat Date: Tue, 14 May 2013 08:32:13 +0530 Subject: [PATCH 1/9] NVMe: Remove redundant version.h header include version.h header inclusion is not necessary as detected by checkversion.pl. Signed-off-by: Sachin Kamat Acked-by: Vishal Verma Signed-off-by: Matthew Wilcox --- drivers/block/nvme-scsi.c | 1 - 1 file changed, 1 deletion(-) diff --git a/drivers/block/nvme-scsi.c b/drivers/block/nvme-scsi.c index fed54b039893..64bde6886b43 100644 --- a/drivers/block/nvme-scsi.c +++ b/drivers/block/nvme-scsi.c @@ -44,7 +44,6 @@ #include #include #include -#include #include #include From 710a143dd870cd660b5ae26bb94b6da854376e91 Mon Sep 17 00:00:00 2001 From: Vishal Verma Date: Mon, 13 May 2013 14:55:18 -0600 Subject: [PATCH 2/9] NVMe: Fix a signedness bug in nvme_trans_modesel_get_mp nvme_trans_modesel_get_mp() was defined with a unsigned return type, but can return signed values. Reported-by: Dan Carpenter Signed-off-by: Vishal Verma Signed-off-by: Matthew Wilcox --- drivers/block/nvme-scsi.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/block/nvme-scsi.c b/drivers/block/nvme-scsi.c index 64bde6886b43..102de2f52b5c 100644 --- a/drivers/block/nvme-scsi.c +++ b/drivers/block/nvme-scsi.c @@ -1653,7 +1653,7 @@ static void nvme_trans_modesel_save_bd(struct nvme_ns *ns, u8 *parm_list, } } -static u16 nvme_trans_modesel_get_mp(struct nvme_ns *ns, struct sg_io_hdr *hdr, +static int nvme_trans_modesel_get_mp(struct nvme_ns *ns, struct sg_io_hdr *hdr, u8 *mode_page, u8 page_code) { int res = SNTI_TRANSLATION_SUCCESS; From 5be37bf9c17ffad0590a4044dbb110fe08066923 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Mon, 13 May 2013 18:00:42 +0300 Subject: [PATCH 3/9] MAINTAINERS: update NVM EXPRESS DRIVER file list There isn't an nvme.c file any more. It has been split into multiple files. Signed-off-by: Dan Carpenter Signed-off-by: Matthew Wilcox --- MAINTAINERS | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/MAINTAINERS b/MAINTAINERS index 3a7b07f513da..748a96da4d78 100644 --- a/MAINTAINERS +++ b/MAINTAINERS @@ -5740,7 +5740,7 @@ M: Matthew Wilcox L: linux-nvme@lists.infradead.org T: git git://git.infradead.org/users/willy/linux-nvme.git S: Supported -F: drivers/block/nvme.c +F: drivers/block/nvme* F: include/linux/nvme.h OMAP SUPPORT From 5460fc03105fbed01fe27aa572d9f65bb410a61d Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Mon, 13 May 2013 17:59:50 +0300 Subject: [PATCH 4/9] NVMe: check for integer overflow in nvme_map_user_pages() You need to have CAP_SYS_ADMIN to trigger this overflow but it makes the static checkers complain so we should fix it. The worry is that "length" comes from copy_from_user() so we need to check that "length + offset" can't overflow. I also changed the min_t() cast to be unsigned instead of signed. Now that we cap "length" to INT_MAX it doesn't make a difference, but it's a little easier for reviewers to know that large values aren't cast to negative. Signed-off-by: Dan Carpenter Signed-off-by: Matthew Wilcox --- drivers/block/nvme-core.c | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c index 8efdfaa44a59..437637551d1e 100644 --- a/drivers/block/nvme-core.c +++ b/drivers/block/nvme-core.c @@ -1206,7 +1206,7 @@ struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write, if (addr & 3) return ERR_PTR(-EINVAL); - if (!length) + if (!length || length > INT_MAX - PAGE_SIZE) return ERR_PTR(-EINVAL); offset = offset_in_page(addr); @@ -1227,7 +1227,8 @@ struct nvme_iod *nvme_map_user_pages(struct nvme_dev *dev, int write, sg_init_table(sg, count); for (i = 0; i < count; i++) { sg_set_page(&sg[i], pages[i], - min_t(int, length, PAGE_SIZE - offset), offset); + min_t(unsigned, length, PAGE_SIZE - offset), + offset); length -= (PAGE_SIZE - offset); offset = 0; } From 1287dabd345f447bbe0f7a99fc95ab89bcfc0f5d Mon Sep 17 00:00:00 2001 From: Wei Yongjun Date: Mon, 13 May 2013 22:29:04 +0800 Subject: [PATCH 5/9] NVMe: fix error return code in nvme_submit_bio_queue() nvme_submit_flush_data() might overwrite the initialisation of the return value with 0, so move the -ENOMEM setting close to the usage. Signed-off-by: Wei Yongjun Signed-off-by: Matthew Wilcox --- drivers/block/nvme-core.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c index 437637551d1e..d783f15e0fc5 100644 --- a/drivers/block/nvme-core.c +++ b/drivers/block/nvme-core.c @@ -629,7 +629,7 @@ static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns, struct nvme_command *cmnd; struct nvme_iod *iod; enum dma_data_direction dma_dir; - int cmdid, length, result = -ENOMEM; + int cmdid, length, result; u16 control; u32 dsmgmt; int psegs = bio_phys_segments(ns->queue, bio); @@ -640,6 +640,7 @@ static int nvme_submit_bio_queue(struct nvme_queue *nvmeq, struct nvme_ns *ns, return result; } + result = -ENOMEM; iod = nvme_alloc_iod(psegs, bio->bi_size, GFP_ATOMIC); if (!iod) goto nomem; From 053ab702cc2702f25a97ead087ed344b864785b7 Mon Sep 17 00:00:00 2001 From: Keith Busch Date: Tue, 30 Apr 2013 11:19:38 -0600 Subject: [PATCH 6/9] NVMe: Do not cancel command multiple times Cancelling an already cancelled command does not do anything, so check the command context before cancelling it, continuing if had already been cancelled so we do not log the same problem every second if a device stops responding. Signed-off-by: Keith Busch Signed-off-by: Matthew Wilcox --- drivers/block/nvme-core.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c index d783f15e0fc5..42abf72d3884 100644 --- a/drivers/block/nvme-core.c +++ b/drivers/block/nvme-core.c @@ -978,6 +978,8 @@ static void nvme_cancel_ios(struct nvme_queue *nvmeq, bool timeout) if (timeout && !time_after(now, info[cmdid].timeout)) continue; + if (info[cmdid].ctx == CMD_CTX_CANCELLED) + continue; dev_warn(nvmeq->q_dmadev, "Cancelling I/O %d\n", cmdid); ctx = cancel_cmdid(nvmeq, cmdid, &fn); fn(nvmeq->dev, ctx, &cqe); From cf90bc4830b858487fe4b9b9ecd0031e23ca3e83 Mon Sep 17 00:00:00 2001 From: Chayan Biswas Date: Wed, 22 May 2013 22:34:49 +0000 Subject: [PATCH 7/9] Return the result from user admin command IOCTL even in case of failure We copy the result to user if the command is completed from the controller even if it completes with failure (non-zero) status. A return status of < 0 indicates the command was not completed by the controller. The user application may expect the error code in the result field in case of failure. Signed-off-by: Chayan Biswas Signed-off-by: Matthew Wilcox --- drivers/block/nvme-core.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c index 42abf72d3884..84937089d5db 100644 --- a/drivers/block/nvme-core.c +++ b/drivers/block/nvme-core.c @@ -1439,7 +1439,7 @@ static int nvme_user_admin_cmd(struct nvme_dev *dev, nvme_free_iod(dev, iod); } - if (!status && copy_to_user(&ucmd->result, &cmd.result, + if ((status >= 0) && copy_to_user(&ucmd->result, &cmd.result, sizeof(cmd.result))) status = -EFAULT; From cf9f123b38345b2c2777e642eb9bb561f4b7de91 Mon Sep 17 00:00:00 2001 From: Matthew Wilcox Date: Tue, 28 May 2013 16:46:46 -0400 Subject: [PATCH 8/9] NVMe: Use dma_set_mask() correctly In some circumstances setting a 64-bit DMA mask can fail, as explained in Documentation/DMA-API-HOWTO.txt. Use the recommended code sequence to set a 32-bit DMA mask if setting a 64-bit DMA mask fails. Reported-by: Chayan Biswas Signed-off-by: Matthew Wilcox --- drivers/block/nvme-core.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c index 84937089d5db..a57d3bcec803 100644 --- a/drivers/block/nvme-core.c +++ b/drivers/block/nvme-core.c @@ -1927,8 +1927,14 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id) INIT_LIST_HEAD(&dev->namespaces); dev->pci_dev = pdev; pci_set_drvdata(pdev, dev); - dma_set_mask(&pdev->dev, DMA_BIT_MASK(64)); - dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(64)); + + if (!dma_set_mask(&pdev->dev, DMA_BIT_MASK(64))) + dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(64)); + else if (!dma_set_mask(&pdev->dev, DMA_BIT_MASK(32))) + dma_set_coherent_mask(&pdev->dev, DMA_BIT_MASK(32)); + else + goto disable; + result = nvme_set_instance(dev); if (result) goto disable; From fa08a396647767abd24a9e7015cb177121d0cf15 Mon Sep 17 00:00:00 2001 From: Ramachandra Rao Gajula Date: Sat, 11 May 2013 15:19:31 -0700 Subject: [PATCH 9/9] NVMe: Add MSI support Some devices only have support for MSI, not MSI-X. While MSI is more limited, it still provides better performance than line-based interrupts. Signed-off-by: Ramachandra Gajula Signed-off-by: Matthew Wilcox --- drivers/block/nvme-core.c | 40 +++++++++++++++++++++++++++++++-------- 1 file changed, 32 insertions(+), 8 deletions(-) diff --git a/drivers/block/nvme-core.c b/drivers/block/nvme-core.c index a57d3bcec803..ce79a590b45b 100644 --- a/drivers/block/nvme-core.c +++ b/drivers/block/nvme-core.c @@ -1637,7 +1637,8 @@ static int set_queue_count(struct nvme_dev *dev, int count) static int nvme_setup_io_queues(struct nvme_dev *dev) { - int result, cpu, i, nr_io_queues, db_bar_size, q_depth; + struct pci_dev *pdev = dev->pci_dev; + int result, cpu, i, nr_io_queues, db_bar_size, q_depth, q_count; nr_io_queues = num_online_cpus(); result = set_queue_count(dev, nr_io_queues); @@ -1646,14 +1647,14 @@ static int nvme_setup_io_queues(struct nvme_dev *dev) if (result < nr_io_queues) nr_io_queues = result; + q_count = nr_io_queues; /* Deregister the admin queue's interrupt */ free_irq(dev->entry[0].vector, dev->queues[0]); db_bar_size = 4096 + ((nr_io_queues + 1) << (dev->db_stride + 3)); if (db_bar_size > 8192) { iounmap(dev->bar); - dev->bar = ioremap(pci_resource_start(dev->pci_dev, 0), - db_bar_size); + dev->bar = ioremap(pci_resource_start(pdev, 0), db_bar_size); dev->dbs = ((void __iomem *)dev->bar) + 4096; dev->queues[0]->q_db = dev->dbs; } @@ -1661,19 +1662,36 @@ static int nvme_setup_io_queues(struct nvme_dev *dev) for (i = 0; i < nr_io_queues; i++) dev->entry[i].entry = i; for (;;) { - result = pci_enable_msix(dev->pci_dev, dev->entry, - nr_io_queues); + result = pci_enable_msix(pdev, dev->entry, nr_io_queues); if (result == 0) { break; } else if (result > 0) { nr_io_queues = result; continue; } else { - nr_io_queues = 1; + nr_io_queues = 0; break; } } + if (nr_io_queues == 0) { + nr_io_queues = q_count; + for (;;) { + result = pci_enable_msi_block(pdev, nr_io_queues); + if (result == 0) { + for (i = 0; i < nr_io_queues; i++) + dev->entry[i].vector = i + pdev->irq; + break; + } else if (result > 0) { + nr_io_queues = result; + continue; + } else { + nr_io_queues = 1; + break; + } + } + } + result = queue_request_irq(dev, dev->queues[0], "nvme admin"); /* XXX: handle failure here */ @@ -1854,7 +1872,10 @@ static void nvme_free_dev(struct kref *kref) { struct nvme_dev *dev = container_of(kref, struct nvme_dev, kref); nvme_dev_remove(dev); - pci_disable_msix(dev->pci_dev); + if (dev->pci_dev->msi_enabled) + pci_disable_msi(dev->pci_dev); + else if (dev->pci_dev->msix_enabled) + pci_disable_msix(dev->pci_dev); iounmap(dev->bar); nvme_release_instance(dev); nvme_release_prp_pools(dev); @@ -1987,7 +2008,10 @@ static int nvme_probe(struct pci_dev *pdev, const struct pci_device_id *id) unmap: iounmap(dev->bar); disable_msix: - pci_disable_msix(pdev); + if (dev->pci_dev->msi_enabled) + pci_disable_msi(dev->pci_dev); + else if (dev->pci_dev->msix_enabled) + pci_disable_msix(dev->pci_dev); nvme_release_instance(dev); nvme_release_prp_pools(dev); disable: