nvme-pci: fix race condition between reset and nvme_dev_disable()
PATCH 1/1 nvme-pci: fix race condition between reset and nvme_dev_disable()
nvme_dev_disable() modifies the dev->online_queues field, therefore nvme_pci_update_nr_queues() should avoid racing against it, otherwise we could end up passing invalid values to blk_mq_update_nr_hw_queues().
WARNING: CPU: 39 PID: 61303 at drivers/pci/msi/api.c:347
pci_irq_get_affinity+0x187/0x210
Workqueue: nvme-reset-wq nvme_reset_work [nvme]
RIP: 0010:pci_irq_get_affinity+0x187/0x210
Call Trace:
<TASK>
? blk_mq_pci_map_queues+0x87/0x3c0
? pci_irq_get_affinity+0x187/0x210
blk_mq_pci_map_queues+0x87/0x3c0
nvme_pci_map_queues+0x189/0x460 [nvme]
blk_mq_update_nr_hw_queues+0x2a/0x40
nvme_reset_work+0x1be/0x2a0 [nvme]Fix the bug by locking the shutdown_lock mutex before using dev->online_queues. Give up if nvme_dev_disable() is running or if it has been executed already.
Fixes 949928c1c731 ("NVMe: Fix possible queue use after freed")
Tested-by Yi Zhang <[email protected]>
Signed-off-by Maurizio Lombardi <[email protected]>
--- drivers/nvme/host/pci.c | 19 ++++++++++++++++--- 1 file changed, 16 insertions(+), 3 deletions(-)
drivers/nvme/host/pci.c +16 -3
@@ -2506,17 +2506,29 @@ static unsigned int nvme_pci_nr_maps(struct nvme_dev *dev)return 1 ; } -static void nvme_pci_update_nr_queues (struct nvme_dev *dev) +static int nvme_pci_update_nr_queues (struct nvme_dev *dev) {if (!dev->ctrl.tagset) {nvme_alloc_io_tag_set (&dev->ctrl, &dev->tagset, &nvme_mq_ops,nvme_pci_nr_maps (dev),sizeof (struct nvme_iod )); -return ; +return 0 ; + } + +/* Give up if we are racing with nvme_dev_disable() */ +if (!mutex_trylock (&dev->shutdown_lock)) +return -ENODEV ; + +/* Check if nvme_dev_disable() has been executed already */ +if (!dev->online_queues) { +mutex_unlock (&dev->shutdown_lock); +return -ENODEV ; }blk_mq_update_nr_hw_queues (&dev->tagset, dev->online_queues -1 );/* free previously allocated queues that are no longer usable */ nvme_free_queues (dev, dev->online_queues); +mutex_unlock (&dev->shutdown_lock); +return 0 ; }static int nvme_pci_enable (struct nvme_dev *dev)
@@ -2797,7 +2809,8 @@ static void nvme_reset_work(struct work_struct *work)nvme_dbbuf_set (dev);nvme_unquiesce_io_queues (&dev->ctrl);nvme_wait_freeze (&dev->ctrl); -nvme_pci_update_nr_queues (dev); +if (nvme_pci_update_nr_queues (dev)) + goto out;nvme_unfreeze (&dev->ctrl); }else {dev_warn (dev->ctrl.device,"IO queues lost\n" );
Signature · 1 line
2.43.5
drivers/nvme/host/pci.c static unsigned int nvme_pci_nr_maps(struct nvme_dev *dev) · line 2509
Quoted 2 lines · Maurizio Lombardi
drivers/nvme/host/pci.c +1
-1-static void nvme_pci_update_nr_queues (struct nvme_dev *dev) +static int nvme_pci_update_nr_queues (struct nvme_dev *dev)
It would probably be nicer to return a bool here instead of a negative error that is always -ENODEV.
Otherwise looks good:
Reviewed-by Christoph Hellwig <[email protected]>