[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <f5c53785-077e-0e2a-ae3d-178b3e73b80b@gmail.com>
Date: Sun, 27 Jun 2021 06:47:23 -0700
From: James Smart <jsmart2021@...il.com>
To: Daniel Wagner <dwagner@...e.de>, linux-nvme@...ts.infradead.org
Cc: linux-kernel@...r.kernel.org,
James Smart <james.smart@...adcom.com>,
Keith Busch <kbusch@...nel.org>, Jens Axboe <axboe@...com>,
Ming Lei <ming.lei@...hat.com>,
Sagi Grimberg <sagi@...mberg.me>
Subject: Re: [PATCH 1/2] nvme-fc: Update hardware queues before using them
On 6/25/2021 3:16 AM, Daniel Wagner wrote:
> In case the number of hardware queues changes, do the update the
> tagset and ctx to hctx first before using the mapping to recreate and
> connnect the IO queues.
>
> Signed-off-by: Daniel Wagner <dwagner@...e.de>
> ---
> drivers/nvme/host/fc.c | 16 ++++++++--------
> 1 file changed, 8 insertions(+), 8 deletions(-)
>
> diff --git a/drivers/nvme/host/fc.c b/drivers/nvme/host/fc.c
> index 8a3c4814d21b..a9645cd89eca 100644
Makes sense. Thanks. Although it does bring up that perhaps, if the
hw queue count changes, thus it no longer matches what was set on the
target, the new value should be set on the target to release resources
on the target.
Note: the same behavior exists in the other transports as we all started
from the same lineage. So those should be updated as well. Granted
you'll need to break out the queue count set and checking which was done
on fc but not on the other transports.
Reviewed-by: James Smart <jsmart2021@...il.com>
-- james
Powered by blists - more mailing lists