public inbox for linux-i2c@vger.kernel.org
 help / color / mirror / Atom feed
From: "Pali Rohár" <pali@kernel.org>
To: Hans de Goede <hdegoede@redhat.com>
Cc: "Ilpo Järvinen" <ilpo.jarvinen@linux.intel.com>,
	"Andy Shevchenko" <andy@kernel.org>,
	"Paul Menzel" <pmenzel@molgen.mpg.de>,
	"Wolfram Sang" <wsa@kernel.org>,
	eric.piel@tremplin-utc.net, "Marius Hoch" <mail@mariushoch.de>,
	Dell.Client.Kernel@dell.com,
	"Kai Heng Feng" <kai.heng.feng@canonical.com>,
	platform-driver-x86@vger.kernel.org,
	"Jean Delvare" <jdelvare@suse.com>,
	"Andi Shyti" <andi.shyti@kernel.org>,
	linux-i2c@vger.kernel.org
Subject: Re: [PATCH v3 6/6] platform/x86: dell-smo8800: Add support for probing for the accelerometer i2c address
Date: Sat, 22 Jun 2024 15:32:37 +0200	[thread overview]
Message-ID: <20240622133237.b5xsetcxnfu4vu6u@pali> (raw)
In-Reply-To: <20240621122503.10034-7-hdegoede@redhat.com>

On Friday 21 June 2024 14:25:01 Hans de Goede wrote:
> Unfortunately the SMOxxxx ACPI device does not contain the i2c-address
> of the accelerometer. So a DMI product-name to address mapping table
> is used.

This is statement which I got from Dell for 10 years old Dell models.

I have already stated that poking of address in kernel is a big risk
specially for all current and any future dell hardware. Hiding it just
under the kernel parameter is still a risk, specially as neither its
name, nor description say that it is dangerous.

But anyway, why this code is being introduced? Have you communicated
with Dell about this problem?

> At support to have the kernel probe for the i2c-address for modesl
> which are not on the list.
> 
> The new probing code sits behind a new probe_i2c_addr module parameter,
> which is disabled by default because probing might be dangerous.
> 
> Signed-off-by: Hans de Goede <hdegoede@redhat.com>
> ---
>  drivers/platform/x86/dell/dell-smo8800.c | 152 ++++++++++++++++++++++-
>  1 file changed, 147 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/platform/x86/dell/dell-smo8800.c b/drivers/platform/x86/dell/dell-smo8800.c
> index 4c79b2599d96..d64d200e927a 100644
> --- a/drivers/platform/x86/dell/dell-smo8800.c
> +++ b/drivers/platform/x86/dell/dell-smo8800.c
> @@ -10,6 +10,7 @@
>   */
>  
>  #define DRIVER_NAME "smo8800"
> +#define LIS3_WHO_AM_I 0x0f
>  
>  #include <linux/device/bus.h>
>  #include <linux/dmi.h>
> @@ -25,6 +26,10 @@
>  #include <linux/uaccess.h>
>  #include <linux/workqueue.h>
>  
> +static bool probe_i2c_addr;
> +module_param(probe_i2c_addr, bool, 0444);
> +MODULE_PARM_DESC(probe_i2c_addr, "Probe the i801 I2C bus for the accelerometer on models where the address is unknown");
> +
>  struct smo8800_device {
>  	u32 irq;                     /* acpi device irq */
>  	atomic_t counter;            /* count after last read */
> @@ -225,6 +230,130 @@ static const struct dmi_system_id smo8800_lis3lv02d_devices[] = {
>  	{ }
>  };
>  
> +/*
> + * This is the kernel version of the single register device sanity checks from
> + * the i2c_safety_check function from lm_sensors sensor-detect script:
> + * This is meant to prevent access to 1-register-only devices,
> + * which are designed to be accessed with SMBus receive byte and SMBus send
> + * byte transactions (i.e. short reads and short writes) and treat SMBus
> + * read byte as a real write followed by a read. The device detection
> + * routines would write random values to the chip with possibly very nasty
> + * results for the hardware. Note that this function won't catch all such
> + * chips, as it assumes that reads and writes relate to the same register,
> + * but that's the best we can do.
> + */
> +static int i2c_safety_check(struct device *dev, struct i2c_adapter *adap, u8 addr)
> +{
> +	union i2c_smbus_data smbus_data;
> +	int err;
> +	u8 data;
> +
> +	/*
> +	 * First receive a byte from the chip, and remember it. This
> +	 * also checks if there is a device at the address at all.
> +	 */
> +	err = i2c_smbus_xfer(adap, addr, 0, I2C_SMBUS_READ, 0,
> +			     I2C_SMBUS_BYTE, &smbus_data);
> +	if (err < 0)
> +		return err;
> +
> +	data = smbus_data.byte;
> +
> +	/*
> +	 * Receive a byte again; very likely to be the same for
> +	 * 1-register-only devices.
> +	 */
> +	err = i2c_smbus_xfer(adap, addr, 0, I2C_SMBUS_READ, 0,
> +			     I2C_SMBUS_BYTE, &smbus_data);
> +	if (err < 0)
> +		return err;
> +
> +	if (smbus_data.byte != data)
> +		return 0; /* Not a 1-register-only device. */
> +
> +	/*
> +	 * Then try a standard byte read, with a register offset equal to
> +	 * the read byte; for 1-register-only device this should read
> +	 * the same byte value in return.
> +	 */
> +	err = i2c_smbus_xfer(adap, addr, 0, I2C_SMBUS_READ, data,
> +			     I2C_SMBUS_BYTE_DATA, &smbus_data);
> +	if (err < 0)
> +		return err;
> +
> +	if (smbus_data.byte != data)
> +		return 0; /* Not a 1-register-only device. */
> +
> +	/*
> +	 * Then try a standard byte read, with a slightly different register
> +	 * offset; this should again read the register offset in return.
> +	 */
> +	err = i2c_smbus_xfer(adap, addr, 0, I2C_SMBUS_READ, data ^ 0x01,
> +			     I2C_SMBUS_BYTE_DATA, &smbus_data);
> +	if (err < 0)
> +		return err;
> +
> +	if (smbus_data.byte != (data ^ 0x01))
> +		return 0; /* Not a 1-register-only device. */
> +
> +	/*
> +	 * Apparently this is a 1-register-only device, restore the original
> +	 * register value and leave it alone.
> +	 */
> +	i2c_smbus_xfer(adap, addr, 0, I2C_SMBUS_WRITE, data,
> +		       I2C_SMBUS_BYTE, NULL);
> +	dev_warn(dev, "I2C safety check for address 0x%02x failed, skipping\n", addr);
> +	return -ENODEV;
> +}
> +
> +static int smo8800_detect_accel(struct smo8800_device *smo8800,
> +				struct i2c_adapter *adap, u8 addr,
> +				struct i2c_board_info *info, bool probe)
> +{
> +	union i2c_smbus_data smbus_data;
> +	const char *type;
> +	int err;
> +
> +	if (probe) {
> +		dev_info(smo8800->dev, "Probing for accelerometer on address 0x%02x\n", addr);
> +		err = i2c_safety_check(smo8800->dev, adap, addr);
> +		if (err < 0)
> +			return err;
> +	}
> +
> +	err = i2c_smbus_xfer(adap, addr, 0, I2C_SMBUS_READ, LIS3_WHO_AM_I,
> +			     I2C_SMBUS_BYTE_DATA, &smbus_data);
> +	if (err < 0) {
> +		dev_warn(smo8800->dev, "Failed to read who-am-i register: %d\n", err);
> +		return err;
> +	}
> +
> +	/* who-am-i register mappings from drivers/misc/lis3lv02d/lis3lv02d.c */
> +	switch (smbus_data.byte) {
> +	case 0x32:
> +		type = "lis331dlh";
> +		break;
> +	case 0x33:
> +		type = "lis2de12"; /* LIS3DC / HP3DC in drivers/misc/lis3lv02d/lis3lv02d.c */
> +		break;
> +	case 0x3a:
> +		type = "lis3lv02dl_accel";
> +		break;
> +	case 0x3b:
> +		type = "lis302dl";
> +		break;
> +	default:
> +		dev_warn(smo8800->dev, "Unknown who-am-i register value 0x%02x\n",
> +			 smbus_data.byte);
> +		return -ENODEV;
> +	}
> +
> +	dev_dbg(smo8800->dev, "Detected %s accelerometer on address 0x%02x\n", type, addr);
> +	strscpy(info->type, "lis3lv02d", I2C_NAME_SIZE);
> +	info->addr = addr;
> +	return 0;
> +}
> +
>  static int smo8800_find_i801(struct device *dev, void *data)
>  {
>  	struct i2c_adapter *adap, **adap_ret = data;
> @@ -247,6 +376,7 @@ static void smo8800_instantiate_i2c_client(struct work_struct *work)
>  	const struct dmi_system_id *lis3lv02d_dmi_id;
>  	struct i2c_board_info info = { };
>  	struct i2c_adapter *adap = NULL;
> +	int err;
>  
>  	if (smo8800->i2c_dev)
>  		return;
> @@ -256,11 +386,22 @@ static void smo8800_instantiate_i2c_client(struct work_struct *work)
>  		return;
>  
>  	lis3lv02d_dmi_id = dmi_first_match(smo8800_lis3lv02d_devices);
> -	if (!lis3lv02d_dmi_id)
> +	if (lis3lv02d_dmi_id) {
> +		info.addr = (long)lis3lv02d_dmi_id->driver_data;
> +		/* Always detect the accel-type, this also checks the accel is actually there */
> +		err = smo8800_detect_accel(smo8800, adap, info.addr, &info, false);
> +		if (err)
> +			goto out_put_adapter;
> +	} else if (probe_i2c_addr) {
> +		/* First try address 0x29 (most used) and then try 0x1d */
> +		if (smo8800_detect_accel(smo8800, adap, 0x29, &info, true) != 0 &&
> +		    smo8800_detect_accel(smo8800, adap, 0x1d, &info, true) != 0) {
> +			dev_warn(smo8800->dev, "failed to probe for lis3lv02d I2C address\n");
> +			goto out_put_adapter;
> +		}
> +	} else {
>  		goto out_put_adapter;
> -
> -	info.addr = (long)lis3lv02d_dmi_id->driver_data;
> -	strscpy(info.type, "lis3lv02d", I2C_NAME_SIZE);
> +	}
>  
>  	smo8800->i2c_dev = i2c_new_client_device(adap, &info);
>  	if (IS_ERR(smo8800->i2c_dev)) {
> @@ -357,7 +498,7 @@ static int smo8800_probe(struct platform_device *device)
>  			 smo8800->irq);
>  	}
>  
> -	if (dmi_check_system(smo8800_lis3lv02d_devices)) {
> +	if (dmi_check_system(smo8800_lis3lv02d_devices) || probe_i2c_addr) {
>  		/*
>  		 * Register i2c-bus notifier + queue initial scan for lis3lv02d
>  		 * i2c_client instantiation.
> @@ -370,6 +511,7 @@ static int smo8800_probe(struct platform_device *device)
>  	} else {
>  		dev_warn(&device->dev,
>  			 "lis3lv02d accelerometer is present on SMBus but its address is unknown, skipping registration\n");
> +		dev_info(&device->dev, "Pass dell_smo8800.probe_i2c_addr=1 on the kernel commandline to probe, this may be dangerous!\n");
>  		if (!smo8800->irq)
>  			return -ENODEV;
>  	}
> -- 
> 2.45.1
> 

  parent reply	other threads:[~2024-06-22 13:32 UTC|newest]

Thread overview: 48+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-06-21 12:24 [PATCH v3 0/6] i2c-i801 / dell-smo8800: Move instantiation of lis3lv02d i2c_client from i2c-i801 to dell-smo8800 Hans de Goede
2024-06-21 12:24 ` [PATCH v3 1/6] i2c: core: Setup i2c_adapter runtime-pm before calling device_add() Hans de Goede
2024-06-21 15:08   ` Andy Shevchenko
2024-06-21 12:24 ` [PATCH v3 2/6] i2c: i801: Use a different adapter-name for IDF adapters Hans de Goede
2024-06-21 15:13   ` Andy Shevchenko
2024-06-22 12:46   ` Pali Rohár
2024-06-22 13:56     ` Hans de Goede
2024-06-22 14:08       ` Pali Rohár
2024-06-22 14:14         ` Hans de Goede
2024-06-22 14:23           ` Pali Rohár
2024-06-22 14:29             ` Hans de Goede
2024-06-22 15:07               ` Pali Rohár
2024-06-23 13:58                 ` Hans de Goede
2024-06-21 12:24 ` [PATCH v3 3/6] platform/x86: dell-smo8800: Move instantiation of lis3lv02d i2c_client from i2c-i801 to dell-smo8800 Hans de Goede
2024-06-21 15:24   ` Andy Shevchenko
2024-06-22 13:59     ` Hans de Goede
2024-06-22 13:16   ` Pali Rohár
2024-06-22 14:06     ` Hans de Goede
2024-06-22 14:20       ` Pali Rohár
2024-06-22 14:26         ` Hans de Goede
2024-06-22 15:12           ` Pali Rohár
2024-06-22 16:35             ` Pali Rohár
2024-06-23 13:56               ` Hans de Goede
2024-06-23 14:09                 ` Hans de Goede
2024-06-22 22:36           ` Andy Shevchenko
2024-06-22 22:41             ` Pali Rohár
2024-06-22 16:26         ` Pali Rohár
2024-06-23 13:46           ` Hans de Goede
2024-06-22 16:43         ` Pali Rohár
2024-06-22 22:43           ` Andy Shevchenko
2024-06-22 22:50             ` Pali Rohár
2024-06-22 22:53               ` Andy Shevchenko
2024-06-23 14:00           ` Hans de Goede
2024-06-22 15:35   ` Pali Rohár
2024-06-23 13:45     ` Hans de Goede
2024-06-23 14:30   ` Pali Rohár
2024-06-21 12:24 ` [PATCH v3 4/6] platform/x86: dell-smo8800: Allow lis3lv02d i2c_client instantiation without IRQ Hans de Goede
2024-06-21 15:30   ` Andy Shevchenko
2024-06-22 13:20   ` Pali Rohár
2024-06-22 14:07     ` Hans de Goede
2024-06-22 15:14       ` Pali Rohár
2024-06-21 12:25 ` [PATCH v3 5/6] platform/x86: dell-smo8800: Add a couple more models to dell_lis3lv02d_devices[] Hans de Goede
2024-06-21 12:25 ` [PATCH v3 6/6] platform/x86: dell-smo8800: Add support for probing for the accelerometer i2c address Hans de Goede
2024-06-21 15:37   ` Andy Shevchenko
2024-06-22 13:32   ` Pali Rohár [this message]
2024-06-22 14:21     ` Hans de Goede
2024-06-22 14:50       ` Pali Rohár
2024-06-22 22:50         ` Andy Shevchenko

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=20240622133237.b5xsetcxnfu4vu6u@pali \
    --to=pali@kernel.org \
    --cc=Dell.Client.Kernel@dell.com \
    --cc=andi.shyti@kernel.org \
    --cc=andy@kernel.org \
    --cc=eric.piel@tremplin-utc.net \
    --cc=hdegoede@redhat.com \
    --cc=ilpo.jarvinen@linux.intel.com \
    --cc=jdelvare@suse.com \
    --cc=kai.heng.feng@canonical.com \
    --cc=linux-i2c@vger.kernel.org \
    --cc=mail@mariushoch.de \
    --cc=platform-driver-x86@vger.kernel.org \
    --cc=pmenzel@molgen.mpg.de \
    --cc=wsa@kernel.org \
    /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