mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
To: Elson Serrao <quic_eserrao@quicinc.com>
Cc: Thinh Nguyen <Thinh.Nguyen@synopsys.com>,
	"gregkh@linuxfoundation.org" <gregkh@linuxfoundation.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"linux-usb@vger.kernel.org" <linux-usb@vger.kernel.org>,
	"quic_wcheng@quicinc.com" <quic_wcheng@quicinc.com>,
	"quic_jackp@quicinc.com" <quic_jackp@quicinc.com>
Subject: Re: [PATCH] usb: dwc3: Skip TRBs while removing requests in disconnect path
Date: Thu, 1 Jun 2023 01:30:02 +0000	[thread overview]
Message-ID: <20230601012953.47xh7meyr2woowpc@synopsys.com> (raw)
In-Reply-To: <38255534-f242-dc06-9216-1568da9b0285@quicinc.com>

On Wed, May 31, 2023, Elson Serrao wrote:
> 
> 
> On 5/31/2023 4:20 PM, Thinh Nguyen wrote:
> > On Wed, May 31, 2023, Elson Roy Serrao wrote:
> > > Consider a scenario where cable disconnect happens when there is an active
> > > usb reqest queued to the UDC. As part of the disconnect we would issue an
> > > end transfer with no interrupt-on-completion before giving back this
> > > request. Since we are giving back the request without skipping TRBs the
> > > num_trbs field of dwc3_request still holds the stale value previously used.
> > > Function drivers re-use same request for a given bind-unbind session and
> > > hence their dwc3_request context gets preserved across cable
> > > disconnect/connect. When such a request gets re-queued after cable connect,
> > 
> > Why would we preserve the request after a disconnect? The request is
> > associated with an endpoint, and after disconnect, the endpoint is no
> > longer valid. Shouldn't the request be freed then?
> > 
> 
> 
> Function drivers generally allocate usb requests during bind when an
> endpoint is allocated to it (through usb_ep_autoconfig). These requests are
> freed when an unbind is called as the function is no longer associated with
> any end point. The function driver is free to re-use these requests
> throughout this bind-unbind session. The only restriction is that the
> function drivers wont be able to queue any requests as long as the endpoint

> is disabled. But that doesn't enforce function drivers to free the requests
> with ep_disable(). Even though the endpoint is disabled with cable
> disconnect, that endpoint is still associated with that particular function
> driver until that function is unbound.
> 
> As an example below is how f_ncm driver allocates and frees the requests
> during bind/unbind
> 
> Bind()
> ...
> ep = usb_ep_autoconfig(cdev->gadget, &fs_ncm_notify_desc);
> if (!ep)
> 	goto fail;
> ncm->notify = ep;
> 
> status = -ENOMEM;
> 
> /* allocate notification request and buffer */
> ncm->notify_req = usb_ep_alloc_request(ep, GFP_KERNEL);
> ...
> 
> The endpoint is enabled later when set_alt is received and disabled in
> ncm_disable when the connection goes down (cable disconnect scenario)
> 
> 
> Unbind()
> ....
> kfree(ncm->notify_req->buf);
> usb_ep_free_request(ncm->notify, ncm->notify_req);
> 
> I see similar implementation in other function drivers as well. That is,
> keep the usb requests allocated throughout the bind-unbind session and
> independent of ep_enable/ep_disable .
> 
> Thanks
> Elson
> 

Thanks for the clarification. Then you just need to reset the num_trbs
count when giving back the request. Can we do that in
dwc3_gadget_del_and_unmap_request()?

Please add a fix tag.

Thanks,
Thinh

  reply	other threads:[~2023-06-01  1:30 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <1685562871-17024-1-git-send-email-quic_eserrao@quicinc.com>
2023-05-31 23:20 ` Thinh Nguyen
2023-06-01  0:57   ` Elson Serrao
2023-06-01  1:30     ` Thinh Nguyen [this message]
2023-06-01 20:23       ` Elson Serrao

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230601012953.47xh7meyr2woowpc@synopsys.com \
    --to=thinh.nguyen@synopsys.com \
    --cc=gregkh@linuxfoundation.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-usb@vger.kernel.org \
    --cc=quic_eserrao@quicinc.com \
    --cc=quic_jackp@quicinc.com \
    --cc=quic_wcheng@quicinc.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®