linux-kernel - Re: [PATCH v10 0/6] Introduced new Cadence USBSS DRD 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:   Fri, 9 Aug 2019 13:36:13 +0300
From:   Roger Quadros <rogerq@...com>
To:     Pawel Laszczak <pawell@...ence.com>, Pavel Machek <pavel@...x.de>
CC:     "gregkh@...uxfoundation.org" <gregkh@...uxfoundation.org>,
        "felipe.balbi@...ux.intel.com" <felipe.balbi@...ux.intel.com>,
        "linux-usb@...r.kernel.org" <linux-usb@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "jbergsagel@...com" <jbergsagel@...com>,
        "nsekhar@...com" <nsekhar@...com>, "nm@...com" <nm@...com>,
        Suresh Punnoose <sureshp@...ence.com>,
        Jayshri Dajiram Pawar <jpawar@...ence.com>,
        Rahul Kumar <kurahul@...ence.com>,
        Anil Joy Varughese <aniljoy@...ence.com>
Subject: Re: [PATCH v10 0/6] Introduced new Cadence USBSS DRD Driver.

Pawel,

On 08/08/2019 07:12, Pawel Laszczak wrote:
> Hi Roger,
> 
>>
>>
>> On 23/07/2019 07:32, Pawel Laszczak wrote:
>>
>>> Hi,
>>
>>>
>>
>>>> On Mon 2019-07-22 13:56:44, Pavel Machek wrote:
>>
>>>>> Hi!
>>
>>>>>
>>
>>>>>>>>> This patch introduce new Cadence USBSS DRD driver to linux kernel.
>>
>>>>>>>>>
>>
>>>>>>>>> The Cadence USBSS DRD Controller is a highly configurable IP Core which
>>
>>>>>>>>> can be instantiated as Dual-Role Device (DRD), Peripheral Only and
>>
>>>>>>>>> Host Only (XHCI)configurations.
>>
>>>>>>>>
>>
>>>>>>>> I see you are using debugfs to select between DRD, peripheral-onlyh and XHCI...
>>
>>>>>>>>
>>
>>>>>>>> Is that good idea?
>>
>>>>>>>
>>
>>>>>>> Yes driver allows selecting dr_mode by debugfs. Controller also support such functionality
>>
>>>>>>> so I don't understand why would it not be a good idea.
>>
>>>>>>>
>>
>>>>>>> I personally use this for testing but it can be used to limit controller functionality without
>>
>>>>>>> recompiling kernel.
>>
>>>>>>
>>
>>>>>> debugfs is ONLY for debugging, never rely on it being enabled, or
>>
>>>>>> mounted, on a system in order to have any normal operation happen.
>>
>>>>>>
>>
>>>>>> So for testing, yes, this is fine.  If this is going to be the normal
>>
>>>>>> api/interface for how to control this driver, no, that is not acceptable
>>
>>>>>> at all.
>>
>>>>>
>>
>>>>> It makes a lot of sense for end-user to toggle this... for example
>>
>>>>> when he is lacking right cable for proper otg detection. As it is
>>
>>>>> third driver offering this functionality, I believe we should stop
>>
>>>>> treating it as debugging.
>>
>>>>
>>
>>>> At least renesas usb controller seems to have variables in sysfs:
>>
>>>> drivers/phy/renesas/phy-rcar-gen3-usb2.c : functions role_show and
>>
>>>> role_store. See also
>>
>>>> Documentation/ABI/testing/sysfs-platform-phy-rcar-gen3-usb2 .
>>
>>>>
>>
>>>> I believe this driver should do same.
>>
>>>>
>>
>>>
>>
>>> CDNS3 driver use the role framework and also has such variable defined
>>
>>> in role switch framework.
>>
>>>
>>
>>> https://urldefense.proofpoint.com/v2/url?u=https-
>> 3A__elixir.bootlin.com_linux_latest_source_drivers_usb_roles_class.c&d=DwICaQ&c=aUq983L2pue2FqKFoP6PGHMJQyoJ7kl3s3GZ-
>> _haXqY&r=e1OgxfvkL0qo9XO6fX1gscva-w03uSYC1nIyxl89-rI&m=_jBsEOB3gtoQVvsVk8k2Pz8dp9zhzZbbL4M0tINJLR8&s=mq5ce-d4Td-
>> lc3OvcvektfSHhXPAL2Go2gWP-q9QwTY&e=
>>
> 
> The meaning is little different. Role switch framework allow to changing role [ host -> device, device -> host ]
> 
> The debugfs.c allows to limit dr_mode. 
> 
>>
>> Can we get rid of the debugfs interface for user initiated role change and just
>>
>> rely on role switch framework via sysfs?
>>
>>
>>
>> We do need user initiated role changes in production systems. So we can't
>>
>> rely on debugfs for this.
> 
> But I assume that in production systems this will be disabled. 
> cdns3-$(CONFIG_DEBUG_FS)		+= debugfs.o
> 

But we do want user space based role switching in production systems.

> I think that I understand your concerns. My idea was not to expand the supported 
> dr_mode. Rather I wanted to have possibility to limit this (only for testing). 
> 
> Eg. 
> If  cdns->dr_mode = USB_DR_MODE_OTG
> then we can limit mode to HOST or DEVICE or DRD

In this case we register with role switch framework.

> if cdns->dr_mode == USB_DR_MODE_HOST || 
>    cdns->dr_mode == USB_DR_MODE_PERIPHERAL) 
> then driver can't change anything

In this case we don't register to role switch framework.

> 
> It allows me for testing some functionality using only single board 
> and even with lacking right cable for proper otg detection. 
> 
> So, removing this can cause that testing some functionality 
> will  be limited on my boards.
> 
> If you rely want to remove this, maybe we could do this 
> after putting this driver to kernel ?.  

I don't want you to remove the user based role change functionality.
I'm just asking to rely on role switch framework for that and not debugfs.

> 
> Maintaining this as my internal code before putting this driver 
> to kernel will be problematic.
> 
> Regards,
> Pawell
> 
> 

-- 
cheers,
-roger
Texas Instruments Finland Oy, Porkkalankatu 22, 00180 Helsinki.
Y-tunnus/Business ID: 0615521-4. Kotipaikka/Domicile: Helsinki