linux-kernel - Re: [RFC][PATCH 1/3 v2] usb: dwc2: Add extcon support to dwc2 driver

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 6 Dec 2016 16:35:06 -0800
From:   John Stultz <john.stultz@...aro.org>
To:     John Youn <John.Youn@...opsys.com>
Cc:     lkml <linux-kernel@...r.kernel.org>, Wei Xu <xuwei5@...ilicon.com>,
        Guodong Xu <guodong.xu@...aro.org>,
        Amit Pundir <amit.pundir@...aro.org>,
        Rob Herring <robh+dt@...nel.org>,
        Douglas Anderson <dianders@...omium.org>,
        Chen Yu <chenyu56@...wei.com>,
        Kishon Vijay Abraham I <kishon@...com>,
        Felipe Balbi <felipe.balbi@...ux.intel.com>,
        Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
        "linux-usb@...r.kernel.org" <linux-usb@...r.kernel.org>
Subject: Re: [RFC][PATCH 1/3 v2] usb: dwc2: Add extcon support to dwc2 driver

On Tue, Dec 6, 2016 at 4:26 PM, John Youn <John.Youn@...opsys.com> wrote:
> On 12/6/2016 4:05 PM, John Stultz wrote:
>> On Tue, Dec 6, 2016 at 3:17 PM, John Youn <John.Youn@...opsys.com> wrote:
>>> Also, do you really need this at all? Wasn't your system previously
>>> able to detect the ID pin change correctly via the connection id
>>> status change interrupt? This would only be needed if that were not
>>> the case.
>>
>> So it can be made work w/o this, but we needed other hacks because the
>> usb-gadget disconnect logic never triggered when the cable was
>> unplugged. The controller would jump over to host mode, then when we
>> re-plugged in the usb-gadget cable, it would fail often as we never
>> got a disconnect signal.  That's why earlier I was using this hack to
>> force gadget disconnect before the reset was called:
>> https://lkml.org/lkml/2016/10/20/26
>
> Other than the triggering WARN_ON() in fifo init, is there any other
> negative effects?

Well, when we see the WARN_ON, it doesn't connect into usb-gadget
mode. I had to unplug and re-plug the cable.
(The hack I linked to above avoids this, but I suspect its not correct).

Also Amit Pundir had mentioned earlier that the UDC sysfs state
doesn't get reported correctly since it doesn't register the
usb-gadget as unplugged until the cable is re-inserted.

> We are revisiting this fifo init code and I think the fifo init is not
> necessary for USB_RESET purposes. This should get rid of a race
> condition where the EP's are not disabled before attempting to
> initialize their FIFO's. Which should get rid of the WARN_ON().
>
> If this is the only issue, then this will probably resolve it.

(Basically that and the two suspend fixes I sent along in this patchset :).

I'd be happy to test anything you're playing with.

thanks
-john