Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Fri, 26 Aug 2022 12:46:53 +0200
From:   Sebastian Andrzej Siewior <bigeasy@linutronix.de>
To:     Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Cc:     Felipe Balbi <balbi@kernel.org>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        linux-usb@vger.kernel.org, linux-kernel@vger.kernel.org,
        Dmitry Bogdanov <d.bogdanov@yadro.com>,
        linux-scsi@vger.kernel.org, target-devel@vger.kernel.org
Subject: Re: [PATCH v2 24/25] usb: gadget: f_tcm: Check overlapped command
Message-ID: <YwikndRzQM+ToQV5@linutronix.de>
References: <cover.1658192351.git.Thinh.Nguyen@synopsys.com>
 <0150d7b669ad80e94596b371cbce0460a327ce7c.1658192351.git.Thinh.Nguyen@synopsys.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <0150d7b669ad80e94596b371cbce0460a327ce7c.1658192351.git.Thinh.Nguyen@synopsys.com>
Precedence: bulk

On 2022-07-18 18:28:34 [-0700], Thinh Nguyen wrote:
> diff --git a/drivers/usb/gadget/function/f_tcm.c b/drivers/usb/gadget/function/f_tcm.c
> index 8b99ee07df87..dfa3e7c043a3 100644
> --- a/drivers/usb/gadget/function/f_tcm.c
> +++ b/drivers/usb/gadget/function/f_tcm.c
> @@ -1214,6 +1233,51 @@ static void usbg_cmd_work(struct work_struct *work)
>  	return;
>  
>  skip:
> +	if (cmd->tmr_rsp == RC_OVERLAPPED_TAG) {
> +		struct se_session *se_sess;
> +		struct uas_stream *stream;
> +
> +		se_sess = cmd->fu->tpg->tpg_nexus->tvn_se_sess;
> +		stream = uasp_get_stream_by_tag(cmd->fu, cmd->tag);
> +
> +		/*
> +		 * There's no guarantee of a matching completion order between
> +		 * different endpoints. i.e. The device may receive a new (CDB)
> +		 * command request completion of the command endpoint before it
> +		 * gets notified of the previous command status completion from
> +		 * a status endpoint. The driver still needs to detect
> +		 * misbehaving host and respond with an overlap command tag. To
> +		 * prevent false overlapped tag failure, give the active and
> +		 * matching stream id a short time (1ms) to complete before
> +		 * respond with overlapped command failure.
> +		 */
> +		msleep(1);

How likely is it for this to happen? Is there maybe some synchronisation
missing? If I see this right, in order to get here, you will already
spill the message "Command tag %d overlapped" which does not look good.
Why should the host re-use a tag which did not yet complete?

Sebastian