[PATCH v1] nvme: avoid race in shutdown namespace removal

[Daniel Wagner]

When we remove the siblings entry, we update ns->head->list, hence we
can't separate the removal and test for being empty. They have to be
in the same critical section to avoid a race.

Fixes: 5396fdac56d8 ("nvme: fix refcounting imbalance when all paths are down")
Cc: Hannes Reinecke <hare at suse.de>
Cc: Keith Busch <kbusch at kernel.org>
Signed-off-by: Daniel Wagner <dwagner at suse.de>
---

I am able to hit this race window when I try to remove two paths
at the same time by making delete_controller asynchronous.

[   93.977701] nvme nvme0: Removing ctrl: NQN "nqn.2014-08.org.nvmexpress:NVMf:uuid:de63429f-50a4-4e03-ade6-0be27b75be77"
[   93.994213] nvme nvme1: Removing ctrl: NQN "nqn.2014-08.org.nvmexpress:NVMf:uuid:de63429f-50a4-4e03-ade6-0be27b75be77"
[   94.009093] del cdev ffff991a00b3c388 minor 0
[   94.009102] CPU: 2 PID: 13239 Comm: nvme Not tainted 5.14.0-rc4+ #29
[   94.009109] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.12.0-59-gc9ba527-rebuilt.opensuse.org 04/01/2014
[   94.009112] Call Trace:
[   94.009119]  dump_stack_lvl+0x33/0x42
[   94.009133]  nvme_cdev_del+0x2d/0x60 [nvme_core]
[   94.009148]  nvme_mpath_shutdown_disk+0x41/0x50 [nvme_core]
[   94.009157]  nvme_ns_remove+0x199/0x1c0 [nvme_core]
[   94.009166]  nvme_remove_namespaces+0xac/0xf0 [nvme_core]
[   94.009175]  nvme_do_delete_ctrl+0x43/0x60 [nvme_core]
[   94.009182]  nvme_sysfs_delete+0x42/0x60 [nvme_core]
[   94.009190]  kernfs_fop_write_iter+0x12c/0x1a0
[   94.009219]  new_sync_write+0x11c/0x1b0
[   94.009229]  vfs_write+0x1ea/0x250
[   94.009236]  ksys_write+0xa1/0xe0
[   94.009242]  do_syscall_64+0x37/0x80
[   94.009256]  entry_SYSCALL_64_after_hwframe+0x44/0xae

With the patch only one of the nvme_do_delete_ctrl() will see
last_path = true and I can't observe any crash.

Though one thing I am not really sure how it interacts with
nvme_init_ns_head() as we could be in running nvme_init_ns_head()
after we have set last_path = true. I haven't really figured
out yet what this would mean. Is this a real problem?

 drivers/nvme/host/core.c | 12 +++++-------
 1 file changed, 5 insertions(+), 7 deletions(-)

diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 42b69f3c6e20..953d07d6a29d 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -3809,8 +3809,13 @@ static void nvme_ns_remove(struct nvme_ns *ns)
 	set_capacity(ns->disk, 0);
 	nvme_fault_inject_fini(&ns->fault_inject);
 
+	/* Synchronize with nvme_init_ns_head() */
 	mutex_lock(&ns->ctrl->subsys->lock);
 	list_del_rcu(&ns->siblings);
+	if (list_empty(&ns->head->list)) {
+		list_del_init(&ns->head->entry);
+		last_path = true;
+	}
 	mutex_unlock(&ns->ctrl->subsys->lock);
 
 	synchronize_rcu(); /* guarantee not available in head->list */
@@ -3830,13 +3835,6 @@ static void nvme_ns_remove(struct nvme_ns *ns)
 	list_del_init(&ns->list);
 	up_write(&ns->ctrl->namespaces_rwsem);
 
-	/* Synchronize with nvme_init_ns_head() */
-	mutex_lock(&ns->head->subsys->lock);
-	if (list_empty(&ns->head->list)) {
-		list_del_init(&ns->head->entry);
-		last_path = true;
-	}
-	mutex_unlock(&ns->head->subsys->lock);
 	if (last_path)
 		nvme_mpath_shutdown_disk(ns->head);
 	nvme_put_ns(ns);
-- 
2.29.2