[PATCH] nvme-fabrics: Use shutdown timeout on disconnect

[Martin Belanger]

From: Daniel Wagner <dwagner at suse.de>
Sent: Friday, October 1, 2021 7:11 AM
To: Martin Belanger <nitram_67 at hotmail.com>
Cc: linux-nvme at lists.infradead.org <linux-nvme at lists.infradead.org>; kbusch at kernel.org <kbusch at kernel.org>; axboe at fb.com <axboe at fb.com>; hch at lst.de <hch at lst.de>; sagi at grimberg.me <sagi at grimberg.me>; Martin Belanger <martin.belanger at dell.com>
Subject: Re: [PATCH] nvme-fabrics: Use shutdown timeout on disconnect 
 
On Thu, Sep 30, 2021 at 10:56:34AM -0400, Martin Belanger wrote:
> @@ -244,8 +257,8 @@ int nvmf_reg_write32(struct nvme_ctrl *ctrl, u32 off, u32 val)
>        cmd.prop_set.offset = cpu_to_le32(off);
>        cmd.prop_set.value = cpu_to_le64(val);
>  
> -     ret = __nvme_submit_sync_cmd(ctrl->fabrics_q, &cmd, NULL, NULL, 0, 0,
> -                     NVME_QID_ANY, 0, 0);
> +     ret = __nvme_submit_sync_cmd(ctrl->fabrics_q, &cmd, NULL, NULL, 0,
> +                                  timeout, NVME_QID_ANY, 0, 0);

If I am not mistaken nvmf_connect_io_queue() suffers from the same
problem.

What I'm trying to fix is the timeout when disconnecting. The disconnect timeout should be short so that we don't get blocked for 1 minute. nvmf_connect_io_queue() is used when establishing a connection, in which case we do want to use the default (longer) timeout.