4.14-stable review patch. If anyone has any objections, please let me know.
------------------
[ Upstream commit f6c8e432cb0479255322c5d0335b9f1699a0270c ]
nvme_stop_ctrl can be called also for reset flow and there is no need to
flush the scan_work as namespaces are not being removed. This can cause
deadlock in rdma, fc and loop drivers since nvme_stop_ctrl barriers
before controller teardown (and specifically I/O cancellation of the
scan_work itself) takes place, but the scan_work will be blocked anyways
so there is no need to flush it.
Instead, move scan_work flush to nvme_remove_namespaces() where it really
needs to flush.
Reported-by: Ming Lei <ming.lei@xxxxxxxxxx>
Signed-off-by: Sagi Grimberg <sagi@xxxxxxxxxxx>
Reviewed-by: Keith Busch <keith.busch@xxxxxxxxx>
Reviewed by: James Smart <jsmart2021@xxxxxxxxx>
Tested-by: Ewan D. Milne <emilne@xxxxxxxxxx>
Signed-off-by: Christoph Hellwig <hch@xxxxxx>
Signed-off-by: Sasha Levin <sashal@xxxxxxxxxx>
---
drivers/nvme/host/core.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/nvme/host/core.c b/drivers/nvme/host/core.c
index 3a63d58d2ca9..65f3f1a34b6b 100644
--- a/drivers/nvme/host/core.c
+++ b/drivers/nvme/host/core.c
@@ -2572,6 +2572,9 @@ void nvme_remove_namespaces(struct nvme_ctrl *ctrl)
{
struct nvme_ns *ns, *next;
+ /* prevent racing with ns scanning */
+ flush_work(&ctrl->scan_work);
+
/*
* The dead states indicates the controller was not gracefully
* disconnected. In that case, we won't be able to flush any data while
@@ -2743,7 +2746,6 @@ void nvme_stop_ctrl(struct nvme_ctrl *ctrl)
{
nvme_stop_keep_alive(ctrl);
flush_work(&ctrl->async_event_work);
- flush_work(&ctrl->scan_work);
cancel_work_sync(&ctrl->fw_act_work);
}
EXPORT_SYMBOL_GPL(nvme_stop_ctrl);
--
2.19.1
