nvme-pci: set doorbell config before unquiescing

[ Upstream commit 06c59d4270 ]

During resets, if queues are unquiesced first, then the host can submit
IOs to the controller using shadow doorbell logic but the controller
won't be aware. This can lead to necessary MMIO doorbells from being
not issued, causing requests to be delayed and timed-out.

Signed-off-by: William Butler <wab@google.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
Stable-dep-of: 26bc0a81f6 ("nvme-pci: fix race condition between reset and nvme_dev_disable()")
Signed-off-by: Sasha Levin <sashal@kernel.org>
This commit is contained in:
William Butler
2024-01-10 18:28:55 +00:00
committed by Greg Kroah-Hartman
parent ed67a5b99e
commit e2c03a2c83
+1 -1
View File
@@ -2757,10 +2757,10 @@ static void nvme_reset_work(struct work_struct *work)
* controller around but remove all namespaces.
*/
if (dev->online_queues > 1) {
nvme_dbbuf_set(dev);
nvme_unquiesce_io_queues(&dev->ctrl);
nvme_wait_freeze(&dev->ctrl);
nvme_pci_update_nr_queues(dev);
nvme_dbbuf_set(dev);
nvme_unfreeze(&dev->ctrl);
} else {
dev_warn(dev->ctrl.device, "IO queues lost\n");