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;
Message-ID: <cfd35493a293cf6f7ddd68fe3cc665989ea36015.camel@mediatek.com>
Subject: Re: [PATCH 1/2] usb: mtu3: fix ep0's stall of out data stage
From:   Chunfeng Yun <chunfeng.yun@mediatek.com>
To:     Alan Stern <stern@rowland.harvard.edu>
CC:     Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        Matthias Brugger <matthias.bgg@gmail.com>,
        <linux-usb@vger.kernel.org>,
        <linux-arm-kernel@lists.infradead.org>,
        <linux-mediatek@lists.infradead.org>,
        <linux-kernel@vger.kernel.org>,
        Eddie Hung <eddie.hung@mediatek.com>,
        Min Guo <min.guo@mediatek.com>,
        Tianping Fang <tianping.fang@mediatek.com>,
        <Stable@vger.kernel.org>
Date:   Thu, 29 Sep 2022 14:30:45 +0800
In-Reply-To: <YzRofTAx+3pPCbrL@rowland.harvard.edu>
References: <20220928091721.26112-1-chunfeng.yun@mediatek.com>
         <YzRofTAx+3pPCbrL@rowland.harvard.edu>
Content-Type: text/plain; charset="UTF-8"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Precedence: bulk

On Wed, 2022-09-28 at 11:30 -0400, Alan Stern wrote:
> On Wed, Sep 28, 2022 at 05:17:20PM +0800, Chunfeng Yun wrote:
> > It happens when enable uvc function, the flow as below:
> > the controller switch to data stage, then call
> >     -> foward_to_driver() -> composite_setup() ->
> > uvc_function_setup(),
> > it send out an event to user layer to notify it call
> >     -> ioctl() -> uvc_send_response() -> usb_ep_queue(),
> > but before the user call ioctl to queue ep0's buffer, the host
> > already send
> > out data, but the controller find that no buffer is queued to
> > receive data,
> > it send out STALL handshake.
> > 
> > To fix the issue, don't send out ACK of setup stage to switch to
> > out data
> > stage until the buffer is available.
> 
> You might find it is better to use the delayed_status routines
> already 
> present in the Gadget core.  Instead of delaying the response to the 
> Setup packet of the second control transfer, delay the status
> response 
> to the first control transfer.
Ok, I'll try to use delayed_status to handle this issue.

Thanks a lot.

> 
> This approach has the advantage of working even when the second
> transfer 
> is not control but something else, such as bulk.
> 
> Also it agrees better with the way the USB spec intends control 
> transfers to work.  The UDC is not supposed to complete the status
> stage 
> of a control transfer until the gadget has fully processed the 
> transfer's information and is ready to go forward.
> 
> Alan Stern