linux-kernel - Re: [PATCH RFC 0/9] gpio: improve support for shared GPIOs

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: <aPetHXE0zuWANRNd@smile.fi.intel.com>
Date: Tue, 21 Oct 2025 18:56:13 +0300
From: Andy Shevchenko <andriy.shevchenko@...el.com>
To: Bartosz Golaszewski <brgl@...ev.pl>
Cc: Manivannan Sadhasivam <mani@...nel.org>,
	Dmitry Torokhov <dmitry.torokhov@...il.com>,
	Kees Cook <kees@...nel.org>, Mika Westerberg <westeri@...nel.org>,
	Andrew Morton <akpm@...ux-foundation.org>,
	Linus Walleij <linus.walleij@...aro.org>,
	Rob Herring <robh@...nel.org>,
	Krzysztof Kozlowski <krzk+dt@...nel.org>,
	Conor Dooley <conor+dt@...nel.org>,
	Saravana Kannan <saravanak@...gle.com>,
	Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
	Andy Shevchenko <andy@...nel.org>,
	Catalin Marinas <catalin.marinas@....com>,
	Will Deacon <will@...nel.org>,
	Srinivas Kandagatla <srini@...nel.org>,
	Liam Girdwood <lgirdwood@...il.com>,
	Mark Brown <broonie@...nel.org>, Jaroslav Kysela <perex@...ex.cz>,
	Takashi Iwai <tiwai@...e.com>, linux-hardening@...r.kernel.org,
	linux-kernel@...r.kernel.org, linux-gpio@...r.kernel.org,
	linux-arm-kernel@...ts.infradead.org, linux-sound@...r.kernel.org,
	linux-arm-msm@...r.kernel.org,
	Bartosz Golaszewski <bartosz.golaszewski@...aro.org>
Subject: Re: [PATCH RFC 0/9] gpio: improve support for shared GPIOs

On Tue, Oct 21, 2025 at 02:06:30PM +0200, Bartosz Golaszewski wrote:
> On Tue, Oct 21, 2025 at 3:53 AM Manivannan Sadhasivam <mani@...nel.org> wrote:

...

> > > That only happens if the driver uses the reset API. If you go with the
> > > GPIOLIB then none of this matters. I definitely don't want to change
> > > the existing DT sources either but I want to find out if the code in
> > > this series is suitable (with some modifications) for supporting the
> > > PERST# line or if the logic behind it is more complex and possibly
> > > requires separate, more fine-grained handling.
> >
> > All PCI controllers relied on '{reset/perst}-gpios' property for handling the
> > PERST# signal. Now if we change it to a reset line, then the drivers have to
> > first detect it as a reset line and use the reset APIs, if not fallback to gpiod
> > APIs (for DT backwards compatibility), which will add unncessary churn IMO.

Can't the reset check be added into a common place once for all the PCI controllers?
Also one can make a helper that does all necessary checks and if required falls back
to the GPIO.

int pcie_controller_get_perst(... dev, const chat *con_id /* fallback */);

(name is from top of my mind without following any patterns).


> Ok so some platforms define perst-gpios while others use reset-gpios,
> I see now. Yeah, it's better to go with explicit GPIOs then.
> 
> > But if there is no way the GPIO subsystem is going to support shared GPIOs, then
> > we have to live with it.
> 
> Well, there is going to be. We already de-facto have it but it doesn't
> work very well and is fragile (I'm talking about the non-exclusive
> flag). I very much intend to bring this upstream.
> 
> My question wrt PCI PERST# was whether this is useful for it because
> IIRC all endpoints sharing the signal will assert it (or rather their
> pwrctl drivers will) and then only deassert it once all endpoints are
> powered up. This would translate to the pwrctl driver doing the
> following for each endpoint:
> 
> perst = gpiod_get(dev, "perst");
> gpiod_set_value_cansleep(perst, 1);
> 
> Do the power up.
> 
> gpiod_set_value_cansleep(perst, 0);
> 
> And with the implementation this series proposes it would mean that
> the perst signal will go high after the first endpoint pwrctl driver
> sets it to high and only go down once the last driver sets it to low.
> The only thing I'm not sure about is the synchronization between the
> endpoints - how do we wait for all of them to be powered-up before
> calling the last gpiod_set_value()?

The refcount inside GPIO descriptor? Then gpiod_set_value() will do the magic.

-- 
With Best Regards,
Andy Shevchenko