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: <626564ae-8dd6-d271-7a4f-11071df2d7e8@samsung.com>
Date:   Wed, 5 Dec 2018 15:34:11 +0100
From:   Marek Szyprowski <m.szyprowski@...sung.com>
To:     Linus Walleij <linus.walleij@...aro.org>,
        Liam Girdwood <lgirdwood@...il.com>,
        Mark Brown <broonie@...nel.org>
Cc:     linux-kernel@...r.kernel.org,
        Charles Keepax <ckeepax@...nsource.cirrus.com>,
        Bartosz Golaszewski <bgolaszewski@...libre.com>
Subject: Re: [PATCH 10/15 v3] gpio: devres: Handle nonexclusive GPIOs

Hi Linus,

On 2018-12-05 13:47, Linus Walleij wrote:
> When we get a nonexeclusive GPIO descriptor using managed
> resources, we should only add it to the list of managed
> resources once: on the first user. Augment the
> devm_gpiod_get_index() and devm_gpiod_get_from_of_node()
> calls to account for this by checking if the descriptor
> is already resource managed before we proceed to allocate
> a new resource management struct.
>
> Cc: Mark Brown <broonie@...nel.org>
> Cc: Bartosz Golaszewski <bgolaszewski@...libre.com>
> Cc: Marek Szyprowski <m.szyprowski@...sung.com>
> Fixes: b0ce7b29bfcd ("regulator/gpio: Allow nonexclusive GPIO access")
> Reported-by: Marek Szyprowski <m.szyprowski@...sung.com>
> Signed-off-by: Linus Walleij <linus.walleij@...aro.org>
> ---
> ChangeLog v1->v3:
> - New patch.
> - This fix is in response to an issue Marek saw in the fixups
>   for resource-managed GPIO descriptors used with ena_gpiod
>   in the regulator subsystem.
> - I first thought to apply it to the GPIO tree directly, but
>   since it is not a regression it is better to put it into
>   the regulator tree with the rest of the patches.
> ---
>  drivers/gpio/gpiolib-devres.c | 50 ++++++++++++++++++++++++++---------
>  1 file changed, 38 insertions(+), 12 deletions(-)
>
> diff --git a/drivers/gpio/gpiolib-devres.c b/drivers/gpio/gpiolib-devres.c
> index 01959369360b..a57e968025a8 100644
> --- a/drivers/gpio/gpiolib-devres.c
> +++ b/drivers/gpio/gpiolib-devres.c
> @@ -98,15 +98,28 @@ struct gpio_desc *__must_check devm_gpiod_get_index(struct device *dev,
>  	struct gpio_desc **dr;
>  	struct gpio_desc *desc;
>  
> +	desc = gpiod_get_index(dev, con_id, idx, flags);
> +	if (IS_ERR(desc))
> +		return desc;
> +
> +	/*
> +	 * For non-exclusive GPIO descriptors, check if this descriptor is
> +	 * already under resource management by this device.
> +	 */
> +	if (flags & GPIOD_FLAGS_BIT_NONEXCLUSIVE) {
> +		struct devres *dres;
> +
> +		dres = devres_find(dev, devm_gpiod_release,
> +				   devm_gpiod_match, desc);

    devres_find(dev, devm_gpiod_release, devm_gpiod_match, &desc);

> +		if (dres)
> +			return desc;
> +	}
> +
>  	dr = devres_alloc(devm_gpiod_release, sizeof(struct gpio_desc *),
>  			  GFP_KERNEL);
> -	if (!dr)
> +	if (!dr) {
> +		gpiod_put(desc);
>  		return ERR_PTR(-ENOMEM);
> -
> -	desc = gpiod_get_index(dev, con_id, idx, flags);
> -	if (IS_ERR(desc)) {
> -		devres_free(dr);
> -		return desc;
>  	}
>  
>  	*dr = desc;
> @@ -140,15 +153,28 @@ struct gpio_desc *devm_gpiod_get_from_of_node(struct device *dev,
>  	struct gpio_desc **dr;
>  	struct gpio_desc *desc;
>  
> +	desc = gpiod_get_from_of_node(node, propname, index, dflags, label);
> +	if (IS_ERR(desc))
> +		return desc;
> +
> +	/*
> +	 * For non-exclusive GPIO descriptors, check if this descriptor is
> +	 * already under resource management by this device.
> +	 */
> +	if (dflags & GPIOD_FLAGS_BIT_NONEXCLUSIVE) {
> +		struct devres *dres;
> +
> +		dres = devres_find(dev, devm_gpiod_release,
> +				   devm_gpiod_match, desc);

 	devres_find(dev, devm_gpiod_release, devm_gpiod_match, &desc);

> +		if (dres)
> +			return desc;
> +	}
> +
>  	dr = devres_alloc(devm_gpiod_release, sizeof(struct gpio_desc *),
>  			  GFP_KERNEL);
> -	if (!dr)
> +	if (!dr) {
> +		gpiod_put(desc);
>  		return ERR_PTR(-ENOMEM);
> -
> -	desc = gpiod_get_from_of_node(node, propname, index, dflags, label);
> -	if (IS_ERR(desc)) {
> -		devres_free(dr);
> -		return desc;
>  	}
>  
>  	*dr = desc;

Best regards
-- 
Marek Szyprowski, PhD
Samsung R&D Institute Poland

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ