linux-kernel - Re: [PATCH v2 5/7] platform/x86: ideapad-laptop: Expose camera

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 15 Nov 2022 21:43:35 +0100
From:   Hans de Goede <hdegoede@...hat.com>
To:     Eray Orçunus <erayorcunus@...il.com>,
        platform-driver-x86@...r.kernel.org
Cc:     linux-kernel@...r.kernel.org, linux-input@...r.kernel.org,
        ike.pan@...onical.com, jikos@...nel.org,
        benjamin.tissoires@...hat.com, dmitry.torokhov@...il.com,
        mgross@...ux.intel.com, pobrn@...tonmail.com
Subject: Re: [PATCH v2 5/7] platform/x86: ideapad-laptop: Expose camera_power
 only if supported

Hi Eray,

On 10/29/22 14:03, Eray Orçunus wrote:
> IdeaPads dropped support for VPCCMD_W_CAMERA somewhere between 2014-2016,
> none of the IdeaPads produced after that I tested supports it. Fortunately
> I found a way to check it; if the DSDT has camera device(s) defined, it
> shouldn't have working VPCCMD_W_CAMERA, thus camera_power shouldn't be
> exposed to sysfs. To accomplish this, walk the ACPI namespace in
> ideapad_check_features and check the devices starting with "CAM".
> Tested on 520-15IKB and Legion Y520, which successfully didn't expose
> the camera_power attribute.
> 
> Link: https://www.spinics.net/lists/platform-driver-x86/msg26147.html
> Signed-off-by: Eray Orçunus <erayorcunus@...il.com>
> ---
>  drivers/platform/x86/ideapad-laptop.c | 52 ++++++++++++++++++++++++++-
>  1 file changed, 51 insertions(+), 1 deletion(-)
> 
> diff --git a/drivers/platform/x86/ideapad-laptop.c b/drivers/platform/x86/ideapad-laptop.c
> index f3d4f2beda07..e8c088e7a53d 100644
> --- a/drivers/platform/x86/ideapad-laptop.c
> +++ b/drivers/platform/x86/ideapad-laptop.c
> @@ -149,6 +149,7 @@ struct ideapad_private {
>  		bool fn_lock              : 1;
>  		bool hw_rfkill_switch     : 1;
>  		bool kbd_bl               : 1;
> +		bool cam_ctrl_via_ec      : 1;
>  		bool touchpad_ctrl_via_ec : 1;
>  		bool usb_charging         : 1;
>  	} features;
> @@ -163,6 +164,24 @@ static bool no_bt_rfkill;
>  module_param(no_bt_rfkill, bool, 0444);
>  MODULE_PARM_DESC(no_bt_rfkill, "No rfkill for bluetooth.");
>  
> +static acpi_status acpi_find_device_callback(acpi_handle handle, u32 level,
> +					     void *context, void **return_value)
> +{
> +	struct acpi_buffer ret_buf;
> +	char buffer[8];
> +
> +	ret_buf.length = sizeof(buffer);
> +	ret_buf.pointer = buffer;
> +
> +	if (ACPI_SUCCESS(acpi_get_name(handle, ACPI_SINGLE_NAME, &ret_buf)))
> +		if (strstarts(ret_buf.pointer, context)) {
> +			*return_value = handle;
> +			return AE_CTRL_TERMINATE;
> +		}
> +
> +	return AE_OK;
> +}
> +
>  /*
>   * ACPI Helpers
>   */
> @@ -675,7 +694,7 @@ static umode_t ideapad_is_visible(struct kobject *kobj,
>  	bool supported = true;
>  
>  	if (attr == &dev_attr_camera_power.attr)
> -		supported = test_bit(CFG_CAP_CAM_BIT, &priv->cfg);
> +		supported = priv->features.cam_ctrl_via_ec;
>  	else if (attr == &dev_attr_conservation_mode.attr)
>  		supported = priv->features.conservation_mode;
>  	else if (attr == &dev_attr_fan_mode.attr)
> @@ -1527,6 +1546,37 @@ static void ideapad_check_features(struct ideapad_private *priv)
>  
>  	priv->features.hw_rfkill_switch = dmi_check_system(hw_rfkill_list);
>  
> +	/*
> +	 * Some IdeaPads have camera switch via EC (mostly older ones),
> +	 * some don't. Fortunately we know that if DSDT contains device
> +	 * object for the camera, camera isn't switchable via EC.
> +	 * So, let's walk the namespace and try to find CAM* object.
> +	 * If we can't find it, set cam_ctrl_via_ec to true.
> +	 */
> +
> +	priv->features.cam_ctrl_via_ec = false;

There is no need to explicitly set this to false since the entire
struct is allocated with kzalloc() and a bunch of other features
flags are also not explicitly set to false. Please drop this line.

> +
> +	if (test_bit(CFG_CAP_CAM_BIT, &priv->cfg)) {
> +		acpi_handle temp_handle = NULL;
> +		acpi_handle pci_handle;
> +		acpi_status status;
> +
> +		status = acpi_get_handle(handle, "^^^", &pci_handle);
> +		if (ACPI_SUCCESS(status)) {
> +			status = acpi_walk_namespace(ACPI_TYPE_DEVICE, pci_handle,
> +						     ACPI_UINT32_MAX,
> +						     acpi_find_device_callback,
> +						     NULL, "CAM",
> +						     &temp_handle);

Why not just use acpi_walk_namespace(ACPI_TYPE_DEVICE, ACPI_ROOT_OBJECT, ... ?

The PCI root is usually pretty much the only object under the root anyways
and this way you can avoid the acpi_get_handle() call + its error handling,
so using ACPI_ROOT_OBJECT would lead to a nice cleanup.

> +
> +			if (ACPI_SUCCESS(status) && temp_handle == NULL)
> +				priv->features.cam_ctrl_via_ec = true;
> +
> +		} else
> +			dev_warn(&priv->platform_device->dev,
> +				"Could not find PCI* node in the namespace\n");
> +	}
> +
>  	/* Most ideapads with ELAN0634 touchpad don't use EC touchpad switch */
>  	priv->features.touchpad_ctrl_via_ec = !acpi_dev_present("ELAN0634", NULL, -1);
>  

Regards,

Hans