Date:   Mon, 21 May 2018 14:14:04 +0200
From:   Hans de Goede <hdegoede@...hat.com>
To:     Masahiro Yamada <yamada.masahiro@...ionext.com>,
        Martin Blumenstingl <martin.blumenstingl@...glemail.com>
Cc:     Philipp Zabel <p.zabel@...gutronix.de>,
        Rob Herring <robh+dt@...nel.org>,
        Lee Jones <lee.jones@...aro.org>,
        Felipe Balbi <felipe.balbi@...ux.intel.com>,
        DTML <devicetree@...r.kernel.org>, Arnd Bergmann <arnd@...db.de>,
        Linus Walleij <linus.walleij@...aro.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        linux-arm-kernel <linux-arm-kernel@...ts.infradead.org>
Subject: Re: [reset-control] How to initialize hardware state with the shared
 reset line?

Hi,

On 21-05-18 03:27, Masahiro Yamada wrote:
> Hi.
> 
> 
> 2018-05-20 19:57 GMT+09:00 Martin Blumenstingl
> <martin.blumenstingl@...glemail.com>:
>> Hi,
>>
>> On Thu, May 10, 2018 at 11:16 AM, Masahiro Yamada
>> <yamada.masahiro@...ionext.com> wrote:
>> [snip]
>>> I may be missing something, but
>>> one solution might be reset hogging on the
>>> reset provider side.  This allows us to describe
>>> the initial state of reset lines in the reset controller.
>>>
>>> The idea for "reset-hog" is similar to:
>>>   - "gpio-hog" defined in
>>>     Documentation/devicetree/bindings/gpio/gpio.txt
>>>   - "assigned-clocks" defined in
>>>     Documetation/devicetree/bindings/clock/clock-bindings.txt
>>>
>>>
>>>
>>> For example,
>>>
>>>     reset-controller {
>>>              ....
>>>
>>>              line_a {
>>>                    reset-hog;
>>>                    resets = <1>;
>>>                    reset-assert;
>>>              };
>>>     }
>>>
>>>
>>> When the reset controller is registered,
>>> the reset ID '1' is asserted.
>>>
>>>
>>> So, all reset consumers that share the reset line '1'
>>> will start from the asserted state
>>> (i.e. defined state machine state).
>> I wonder if a "reset hog" can be board specific:
>> - GPIO hogs are definitely board specific (meson-gxbb-odroidc2.dts for
>> example uses it to take the USB hub out of reset)
>> - assigned-clock-parents (and the like) can also be board specific (I
>> made up a use-case since I don't know of any actual examples: board A
>> uses an external XTAL while board B uses some other internal
>> clock-source because it doesn't have an external XTAL)
>>
>> however, can reset lines be board specific? or in other words: do we
>> need to describe them in device-tree?
> 
> Indeed.
> 
> I did not come up with board-specific cases.
> 
> The problem we are discussing is SoC-specific,
> and reset-controller drivers are definitely SoC-specific.
> 
> So, I think the initial state can be coded in drivers instead of DT.
> 
> 
>> we could extend struct reset_controller_dev (= reset controller
>> driver) if they are not board specific:
>> - either assert all reset lines by default except if they are listed
>> in a new field (may break backwards compatibility, requires testing of
>> all reset controller drivers)
> 
> This is quite simple, but I am afraid there are some cases where the forcible
> reset-assert is not preferred.
> 
> For example, the earlycon.  When we use earlycon, we would expect it has been
> initialized by a boot-loader, or something.
> If it is reset-asserted on the while, the console output
> will not be good.
> 
> 
> 
>> - specify a list of reset lines and their desired state (or to keep it
>> easy: specify a list of reset lines that should be asserted)
>> (I must admit that this is basically your idea but the definition is
>> moved from device-tree to the reset controller driver)
> 
> Yes, I think the list of "reset line ID" and "init state" pairs
> would be nicer.
> 
> 
>> any "chip" specific differences could be expressed by using a
>> different of_device_id
>>
>> one the other hand: your "reset hog" solution looks fine to me if
>> reset lines can be board specific
>>
>>>  From the discussion with Martin Blumenstingl
>>> (https://lkml.org/lkml/2018/4/28/115),
>>> the problem for Amlogic is that
>>> the reset line is "de-asserted" by default.
>>> If so, the 'reset-hog' would fix the problem,
>>> and DWC3 driver would be able to use
>>> shared, level reset, I think.
>> I think you are right: if we could control the initial state then we
>> should be able to use level resets
> 
> 
> Even further, can we drop the shared reset_control_reset() support, maybe?
> (in other words, revert commit 7da33a37b48f11)

At least one some Allwinner ships shared-reset support accurately
models how the hardware works. It seems that the case you are
trying to fix is actually a special version of shared reset support,
you want shared reset behavior, combined with making sure the reset
is asserted at reset-controller-driver load time.

Regards,

Hans

Powered by blists - more mailing lists