Devicetree
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Svyatoslav Ryhel" <clamor95@gmail.com>
Cc: devicetree@vger.kernel.org, robh@kernel.org, conor+dt@kernel.org
Subject: Re: [PATCH v1 2/2] platform/asus: Add driver for ASUS Transformer's Dock of Tegra based devices
Date: Wed, 18 Nov 2026 14:16:25 +0000	[thread overview]
Message-ID: <sashiko-outbox-221177@kernel.org> (raw)
In-Reply-To: <20261118134555.195479-3-clamor95@gmail.com>

Thank you for your contribution! Sashiko AI review found 6 potential
issues to consider:
- [High] platform/asus: missing I2C dependency for ASUS_TRANSFORMER_DOCK
  in Kconfig
- [High] platform/asus: NULL dereference in asus_tf_dock_probe() from
  IS_ERR() checks
- [High] platform/asus: module refcount underflow in
  asus_tf_dock_probe()
- [Medium] platform/asus: unchecked gpiod_get_value_cansleep() in
  asus_tf_dock_interrupt()
- [Medium] platform/asus: unchecked gpiod_to_irq() error return in
  asus_tf_dock_probe()
- [Medium] platform/asus: asus-transformer-dock: bypass of
  __i2c_transfer() on parent
--

commit e7bf89be9d43a55643ec7570e3aed3724b7cdbf9
Author: Michał Mirosław <mirq-linux@rere.qmqm.pl>

platform/asus: Add driver for ASUS Transformer's Dock of Tegra based devices

This commit adds a platform driver for the hot-pluggable I2C bus in the
ASUS Transformer dock on Tegra-based devices, registering or removing a
forwarding I2C adapter based on a detect GPIO interrupt.

> diff --git a/drivers/platform/asus/Kconfig b/drivers/platform/asus/Kconfig
> index 0219418ce3523..af670a3c0d014 100644
> --- a/drivers/platform/asus/Kconfig
> +++ b/drivers/platform/asus/Kconfig
> @@ -113,6 +113,17 @@ config ASUS_ARMOURY
>  	  To compile this driver as a module, choose M here: the module will
>  	  be called asus-armoury.
>  
> +config ASUS_TRANSFORMER_DOCK
> +	tristate "ASUS Transformer Dock for Tegra based devices"
> +	depends on GPIOLIB
> +	depends on OF

[Severity: High]
Should ASUS_TRANSFORMER_DOCK also depend on I2C?  The driver calls I2C core
helpers such as i2c_add_adapter(), i2c_del_adapter() and i2c_put_adapter(),
which fails to compile when CONFIG_I2C=n and fails to link when
CONFIG_I2C=m and CONFIG_ASUS_TRANSFORMER_DOCK=y.

[ ... ]

> diff --git a/drivers/platform/asus/asus-transformer-dock.c b/drivers/platform/asus/asus-transformer-dock.c
> new file mode 100644
> index 0000000000000..7c9346af5e1f3
> --- /dev/null
> +++ b/drivers/platform/asus/asus-transformer-dock.c
> @@ -0,0 +1,260 @@

[ ... ]

> +static int asus_tf_dock_master_xfer(struct i2c_adapter *adap,
> +				    struct i2c_msg msgs[], int num)
> +{
> +	struct i2c_adapter *parent = asus_tf_dock_parent(adap);
> +
> +	return parent->algo->master_xfer(parent, msgs, num);
> +}
> +
> +static int asus_tf_dock_smbus_xfer(struct i2c_adapter *adap, u16 addr,
> +				   unsigned short flags, char read_write,
> +				   u8 command, int protocol,
> +				   union i2c_smbus_data *data)
> +{
> +	struct i2c_adapter *parent = asus_tf_dock_parent(adap);
> +
> +	return parent->algo->smbus_xfer(parent, addr, flags, read_write,
> +					command, protocol, data);
> +}

[Severity: Medium]
Should these callbacks use __i2c_transfer(parent, ...) and
__i2c_smbus_xfer(parent, ...) instead of calling parent->algo directly?

Calling parent->algo directly skips __i2c_check_suspended(parent), atomic
transfer dispatch (master_xfer_atomic and smbus_xfer_atomic) and SMBus
emulation fallback on -EOPNOTSUPP.  Because priv->adap.dev.parent is set to
the platform device rather than &parent->dev, a transfer through priv->adap
after tegra_i2c_suspend_noirq() has marked parent suspended and disabled
controller clocks can enter tegra_i2c_xfer() while clocks are off.

[ ... ]

> +static irqreturn_t asus_tf_dock_interrupt(int irq, void *dev_id)
> +{
> +	struct asus_tf_dock_priv *priv = dev_id;
> +
> +	/* debounce */
> +	msleep(20);
> +
> +	if (gpiod_get_value_cansleep(priv->gpio))
> +		asus_tf_dock_activate(priv);
> +	else
> +		asus_tf_dock_deactivate(priv);
> +
> +	return IRQ_HANDLED;
> +}

[Severity: Medium]
Can a negative error code from gpiod_get_value_cansleep() cause
asus_tf_dock_interrupt() to falsely activate the dock adapter?
gpiod_get_value_cansleep() returns a negative errno on failure, which
evaluates as true in the if condition.

> +
> +static void wrap_i2c_put_adapter(void *adapter)
> +{
> +	i2c_put_adapter(adapter);
> +}
> +
> +static void wrap_asus_tf_dock_deactivate(void *priv)
> +{
> +	asus_tf_dock_deactivate(priv);
> +}
> +
> +static int asus_tf_dock_probe(struct platform_device *pdev)
> +{
> +	struct device *dev = &pdev->dev;
> +	struct device_node *parent_np;
> +	struct i2c_adapter *parent;
> +	struct asus_tf_dock_priv *priv;
> +	bool is_i2c, is_smbus;
> +	int err;
> +
> +	priv = devm_kzalloc_obj(dev, struct asus_tf_dock_priv);
> +	if (!priv)
> +		return -ENOMEM;
> +
> +	platform_set_drvdata(pdev, priv);
> +	priv->dev = dev;
> +
> +	parent_np = of_parse_phandle(dev->of_node, "i2c-parent", 0);
> +	if (IS_ERR(parent_np))
> +		return dev_err_probe(dev, PTR_ERR(parent_np),
> +				     "Failed to parse i2c-parent\n");
> +
> +	parent = of_find_i2c_adapter_by_node(parent_np);
> +	of_node_put(parent_np);
> +	if (IS_ERR(parent))
> +		return dev_err_probe(dev, PTR_ERR(parent),
> +				     "Failed to get parent I2C adapter\n");

[Severity: High]
Can this cause a NULL pointer dereference at parent->algo->master_xfer when
the i2c-parent phandle is missing or the parent adapter has not probed yet?
Both of_parse_phandle() and of_find_i2c_adapter_by_node() return NULL on
failure rather than an ERR_PTR, so IS_ERR() evaluates to false and
execution continues with parent == NULL.

> +	priv->parent = parent;
> +
> +	err = devm_add_action_or_reset(dev, wrap_i2c_put_adapter, parent);
> +	if (err)
> +		return err;

[Severity: High]
Does pairing of_find_i2c_adapter_by_node() with i2c_put_adapter() underflow
parent->owner's module reference count on probe failure or unbind?
of_find_i2c_adapter_by_node() only takes a reference on &parent->dev,
whereas i2c_put_adapter() calls both module_put(adap->owner) and
put_device(&adap->dev).  Should this use of_get_i2c_adapter_by_node()
instead?

[ ... ]

> +	err = devm_request_threaded_irq(dev, gpiod_to_irq(priv->gpio), NULL,
> +					asus_tf_dock_interrupt,
> +					IRQF_TRIGGER_RISING |
> +					IRQF_TRIGGER_FALLING |
> +					IRQF_ONESHOT | IRQF_SHARED,
> +					"asus-transformer-dock", priv);
> +	if (err)
> +		return dev_err_probe(dev, err, "Failed to register IRQ %d\n",
> +				     gpiod_to_irq(priv->gpio));
> +
> +	irq_wake_thread(gpiod_to_irq(priv->gpio), priv);

[Severity: Medium]
Should the return value of gpiod_to_irq(priv->gpio) be stored in a signed
int and checked for < 0 before calling devm_request_threaded_irq()?  Passing
a negative errno to the unsigned int irq parameter converts it into a bogus
IRQ number and replaces the actual error code with -EINVAL.

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261118134555.195479-1-clamor95@gmail.com?part=2

      reply	other threads:[~2026-11-18 14:16 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-18 13:45 [PATCH v1 0/2] platform/asus: Add driver for ASUS Transformer's Dock Svyatoslav Ryhel
2026-11-18 13:45 ` [PATCH v1 1/2] dt-bindings: i2c: Document " Svyatoslav Ryhel
2026-11-18 14:40   ` Wolfram Sang
2026-11-18 14:42     ` Svyatoslav Ryhel
2026-11-18 13:45 ` [PATCH v1 2/2] platform/asus: Add driver for ASUS Transformer's Dock of Tegra based devices Svyatoslav Ryhel
2026-11-18 14:16   ` sashiko-bot [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=sashiko-outbox-221177@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=clamor95@gmail.com \
    --cc=conor+dt@kernel.org \
    --cc=devicetree@vger.kernel.org \
    --cc=robh@kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox