linux-kernel - Re: [PATCH v2 0/3] usb: musb: da8xx: Fix few issues

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:   Fri, 28 Oct 2016 12:50:36 -0500
From:   David Lechner <david@...hnology.com>
To:     Alexandre Bailon <abailon@...libre.com>, khilman@...libre.com,
        b-liu@...com, balbi@...nel.org
Cc:     linux-kernel@...r.kernel.org, linux-usb@...r.kernel.org
Subject: Re: [PATCH v2 0/3] usb: musb: da8xx: Fix few issues

On 10/28/2016 04:31 AM, Alexandre Bailon wrote:
> On 10/27/2016 08:44 PM, David Lechner wrote:
>> On 10/27/2016 12:16 PM, David Lechner wrote:
>>> On 10/26/2016 05:58 AM, Alexandre Bailon wrote:
>>>> Currently, the USB OTG of the da8xx doesn't work.
>>>> This series intend to fix them.
>>>>
>>>> Change in v2:
>>>> * Fix the error path da8xx_musb_init()
>>>>
>>>> Alexandre Bailon (3):
>>>>   usb: musb: da8xx: Call earlier clk_prepare_enable()
>>>>   phy: da8xx-usb: Configure CFGCHIP2 to support OTG workaround
>>>>   usb: musb: da8xx: Only execute the OTG workaround when phy in OTG mode
>>>>
>>>>  drivers/phy/phy-da8xx-usb.c | 17 ++++++++++++-----
>>>>  drivers/usb/musb/da8xx.c    | 28 +++++++++++++++++++---------
>>>>  2 files changed, 31 insertions(+), 14 deletions(-)
>>>>
>>>
>>> I have found another problem with peripheral mode. When we force
>>> peripheral mode, the glue layer currently uses CFGCHIP2 to override the
>>> VBUS and ID. This causes it to not be able to detect disconnection
>>> because the VBUS is overridden.
> How have you found it ? Does it cause any issues ?
> I mean I had to enable traces to see that disconnect was not happening.

I am also using the device tree patch series. I specified dr_mode = 
"peripheral" in my device tree because the device is only wired for use 
as a peripheral port.

I have actually known about this issue for a long time. See
<https://github.com/ev3dev/ev3dev/issues/244> and 
<http://e2e.ti.com/support/embedded/linux/f/354/p/65332/237094#2370942>


>>>
>>> Here is a patch to fix the problem. I have tested this on LEGO
>>> MINDSTORMS EV3 (AM1808). This works because the ID pin is internally
>>> pulled up on the SoC, so we don't need to override it.
> Actually, I'm wonder if that if not related to VBUS sensing.
> May be we should set CFGCHIP2_VBDTCTEN in device mode.

Yes, as we discussed in another thread, we should set CFGCHIP2_VBDTCTEN 
and CFGCHIP2_SESNDEN *always* regardless of mode. I still have the 
problem described above with these two enabled if and only if we are 
also setting CFGCHIP2_OTG_FORCE_PERIPHERAL.

>>>
>>> ---
>>>
>>> diff --git a/drivers/usb/musb/da8xx.c b/drivers/usb/musb/da8xx.c
>>> index 2bc12a2..33daa3b 100644
>>> --- a/drivers/usb/musb/da8xx.c
>>> +++ b/drivers/usb/musb/da8xx.c
>>> @@ -374,9 +374,7 @@ static int da8xx_musb_set_mode(struct musb *musb, u8
>>> musb_mode)
>>>         case MUSB_HOST:         /* Force VBUS valid, ID = 0 */
>>>                 phy_mode = PHY_MODE_USB_HOST;
>>>                 break;
>>> -       case MUSB_PERIPHERAL:   /* Force VBUS valid, ID = 1 */
>>> -               phy_mode = PHY_MODE_USB_DEVICE;
>>> -               break;
>>> +       case MUSB_PERIPHERAL:
>>>         case MUSB_OTG:          /* Don't override the VBUS/ID
>>> comparators */
>>>                 phy_mode = PHY_MODE_USB_OTG;
>>>                 break;
>>>
>>> ---
>>>
>>> If this works for other SoCs/boards, I think we should make this change.
>>> If it doesn't work, we could work around the VBUS problem by polling
>>> VBUSSENSE in CFGCHIP2. But, I like the simple solution above better.
>>
>> I have realized that due to the way my device is wired, I can actually
>> use OTG mode and it will behave exactly as peripheral mode because the
>> ID pin is not connected. So, maybe this patch is not needed after all.
>>
> Actually, I'm sure that is related to ID

The ID pin on my device is not connected, so I don't think this has to 
do with ID. It is always high (internally pulled up).

> I have the same issue except the disconnect is called when I use the
> OTG mode.
>

This is exactly the behavior I am seeing.