linux-kernel - Re: [PATCH] pinctrl: Really force states during suspend/resume

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]

Message-ID: <CACRpkdaUMGy1s24hsrs0wCr8s7Di_9WSmrsM8is784deChnr9g@mail.gmail.com>
Date:   Thu, 23 Feb 2017 11:30:06 +0100
From:   Linus Walleij <linus.walleij@...aro.org>
To:     Andy Shevchenko <andy.shevchenko@...il.com>,
        Ulf Hansson <ulf.hansson@...aro.org>
Cc:     Florian Fainelli <f.fainelli@...il.com>,
        Al Cooper <alcooperx@...il.com>,
        Stephen Warren <swarren@...dia.com>,
        "open list:PIN CONTROL SUBSYSTEM" <linux-gpio@...r.kernel.org>,
        open list <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] pinctrl: Really force states during suspend/resume

On Wed, Feb 8, 2017 at 10:46 PM, Andy Shevchenko
<andy.shevchenko@...il.com> wrote:

> Btw, I have got similar issue and thinking about those states they are
> quite orthogonal to the pin states. Wouldn't be better to actually
> differentiate PM related states and pin states?

I don't fully understand what you mean here, but I like the sound
of it.

"sleep" and "default" were traditionally related to the system
suspend/resume states. It was suggested that the core handle this
automatically, but it doesn't work because of things like that
userspace can disable a TTY/UART and then it should sleep,
regardless of the state of the system.

Runtime PM "sleep" and "resume" is closer to what we want to
achieve here, and might be a good integration point.
(CC:ing Ulf, he's looking into things like this.)

> In my case I have a ->probe() function where device is requested GPIO
> in order to make it wake capable source without using anywhere else.
> So, this requires to have "init" state to be defined which is kinda
> inconvenient.
>
> On resume/suspend it calls pinctrl_pm_state*() and requires "default"
> and "sleep" states to be defined.
>
> I think GPIO case is quite generic and pin control framework lacks of
> something like switching some pins of the group to GPIO state and back
> whenever they defined as wake capable sources.

I guess by "GPIO state" you are referring to what is discussed in
Documentation/pinctrl.txt as "GPIO mode pitfalls", i.e. it is not really
used as a GPIO, but as part of a device functionality it just happens
that the TRM calls the asynchronous (low power mode) edge detector
mode "GPIO".

> I would work towards fixing this issue anyway (to get UART runtime PM
> working on serial consoles).

Everyone would be grateful for that.

Yours,
Linus Walleij