| 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
| ||
|
Message-ID: <7bc8ec9a-f559-96c8-36fd-6e11a1420626@roeck-us.net>
Date: Sat, 16 Nov 2019 07:53:10 -0800
From: Guenter Roeck <linux@...ck-us.net>
To: Pali Rohár <pali.rohar@...il.com>,
Giovanni Mascellani <gio@...ian.org>
Cc: Jean Delvare <jdelvare@...e.com>, linux-hwmon@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] dell-smm-hwmon: Add support for disabling automatic
BIOS fan control
On 11/16/19 6:58 AM, Pali Rohár wrote:
> On Saturday 16 November 2019 15:06:49 Giovanni Mascellani wrote:
>> This patch exports standard hwmon pwmX_enable sysfs attribute for
>> enabling or disabling automatic fan control by BIOS. Standard value
>> "1" is for disabling automatic BIOS fan control and value "2" for
>> enabling.
>>
>> By default BIOS auto mode is enabled by laptop firmware.
>>
>> When BIOS auto mode is enabled, custom fan speed value (set via hwmon
>> pwmX sysfs attribute) is overwritten by SMM in few seconds and
>> therefore any custom settings are without effect. So this is reason
>> why implementing option for disabling BIOS auto mode is needed.
>>
>> So finally this patch allows kernel to set and control fan speed on
>> laptops, but it can be dangerous (like setting speed of other fans).
>>
>> The SMM commands to enable or disable automatic fan control are not
>> documented and are not the same on all Dell laptops. Therefore a
>> whitelist is used to send the correct codes only on laptopts for which
>> they are known.
>>
>> This patch was originally developed by Pali Rohár; later Giovanni
>> Mascellani implemented the whitelist.
>>
>> Signed-off-by: Giovanni Mascellani <gio@...ian.org>
>> Co-Developer-by: Pali Rohár <pali.rohar@...il.com>
>
> Hello! Patch looks good, see small suggestions below. The only important
> change for this patch is to hide those new pwmX_enable entries on
> unsupported machines.
>
>> ---
>> drivers/hwmon/dell-smm-hwmon.c | 111 ++++++++++++++++++++++++++++++---
>> 1 file changed, 101 insertions(+), 10 deletions(-)
>>
>> diff --git a/drivers/hwmon/dell-smm-hwmon.c b/drivers/hwmon/dell-smm-hwmon.c
>> index 4212d022d253..67b8c0adede8 100644
>> --- a/drivers/hwmon/dell-smm-hwmon.c
>> +++ b/drivers/hwmon/dell-smm-hwmon.c
>> @@ -68,6 +68,8 @@ static uint i8k_pwm_mult;
>> static uint i8k_fan_max = I8K_FAN_HIGH;
>> static bool disallow_fan_type_call;
>> static bool disallow_fan_support;
>> +static unsigned int smm_manual_fan;
>> +static unsigned int smm_auto_fan;
The "smm_" prefix does not have any value here.
>>
>> #define I8K_HWMON_HAVE_TEMP1 (1 << 0)
>> #define I8K_HWMON_HAVE_TEMP2 (1 << 1)
>> @@ -300,6 +302,17 @@ static int i8k_get_fan_nominal_speed(int fan, int speed)
>> return i8k_smm(®s) ? : (regs.eax & 0xffff) * i8k_fan_mult;
>> }
>>
>> +/*
>> + * Enable or disable automatic BIOS fan control support
>> + */
>> +static int i8k_enable_fan_auto_mode(bool enable)
>> +{
>> + struct smm_regs regs = { };
>> +
>
> For safely reasons I would suggest to add:
>
> if (disallow_fan_support)
> return -EINVAL;
>
>> + regs.eax = enable ? smm_auto_fan : smm_manual_fan;
>> + return i8k_smm(®s);
>> +}
>> +
>> /*
>> * Set the fan speed (off, low, high). Returns the new fan status.
>> */
>> @@ -726,6 +739,35 @@ static ssize_t i8k_hwmon_pwm_store(struct device *dev,
>> return err < 0 ? -EIO : count;
>> }
>>
>> +static ssize_t i8k_hwmon_pwm_enable_store(struct device *dev,
>> + struct device_attribute *attr,
>> + const char *buf, size_t count)
>> +{
>> + int err;
>> + bool enable;
>> + unsigned long val;
>> +
>> + if (!smm_auto_fan)
>> + return -ENODEV;
>> +
>> + err = kstrtoul(buf, 10, &val);
>> + if (err)
>> + return err;
>> +
>
> ====
>
>> + if (val == 0)
>> + return -EINVAL;
>> + if (val > 2)
>> + return -EINVAL;
>> +
>> + enable = (val != 1);
>
> ====
>
> Just small suggestion: I would write it more-opencoded to immediately
> see what values are valid and what are they meaning. It look me quite
> more time to check that above logic is correct (according to hwmon
> documentation).
>
> if (val == 1)
> enable = false;
> else if (val == 2)
> enable = true;
> else
> return -EINVAL;
>
> or via switch:
>
> switch (val) {
> case 1:
> enable = false;
> break;
> case 2:
> enable = true;
> break;
> default:
> return -EINVAL;
> }
>
>> +
>> + mutex_lock(&i8k_mutex);
>> + err = i8k_enable_fan_auto_mode(enable);
>> + mutex_unlock(&i8k_mutex);
>> +
>> + return err ? -EIO : count;
>> +}
>> +
>> static SENSOR_DEVICE_ATTR_RO(temp1_input, i8k_hwmon_temp, 0);
>> static SENSOR_DEVICE_ATTR_RO(temp1_label, i8k_hwmon_temp_label, 0);
>> static SENSOR_DEVICE_ATTR_RO(temp2_input, i8k_hwmon_temp, 1);
>> @@ -749,12 +791,15 @@ static SENSOR_DEVICE_ATTR_RO(temp10_label, i8k_hwmon_temp_label, 9);
>> static SENSOR_DEVICE_ATTR_RO(fan1_input, i8k_hwmon_fan, 0);
>> static SENSOR_DEVICE_ATTR_RO(fan1_label, i8k_hwmon_fan_label, 0);
>> static SENSOR_DEVICE_ATTR_RW(pwm1, i8k_hwmon_pwm, 0);
>> +static SENSOR_DEVICE_ATTR_WO(pwm1_enable, i8k_hwmon_pwm_enable, 0);
>> static SENSOR_DEVICE_ATTR_RO(fan2_input, i8k_hwmon_fan, 1);
>> static SENSOR_DEVICE_ATTR_RO(fan2_label, i8k_hwmon_fan_label, 1);
>> static SENSOR_DEVICE_ATTR_RW(pwm2, i8k_hwmon_pwm, 1);
>> +static SENSOR_DEVICE_ATTR_WO(pwm2_enable, i8k_hwmon_pwm_enable, 0);
>> static SENSOR_DEVICE_ATTR_RO(fan3_input, i8k_hwmon_fan, 2);
>> static SENSOR_DEVICE_ATTR_RO(fan3_label, i8k_hwmon_fan_label, 2);
>> static SENSOR_DEVICE_ATTR_RW(pwm3, i8k_hwmon_pwm, 2);
>> +static SENSOR_DEVICE_ATTR_WO(pwm3_enable, i8k_hwmon_pwm_enable, 0);
>>
>> static struct attribute *i8k_attrs[] = {
>> &sensor_dev_attr_temp1_input.dev_attr.attr, /* 0 */
>> @@ -780,12 +825,15 @@ static struct attribute *i8k_attrs[] = {
>> &sensor_dev_attr_fan1_input.dev_attr.attr, /* 20 */
>> &sensor_dev_attr_fan1_label.dev_attr.attr, /* 21 */
>> &sensor_dev_attr_pwm1.dev_attr.attr, /* 22 */
>> - &sensor_dev_attr_fan2_input.dev_attr.attr, /* 23 */
>> - &sensor_dev_attr_fan2_label.dev_attr.attr, /* 24 */
>> - &sensor_dev_attr_pwm2.dev_attr.attr, /* 25 */
>> - &sensor_dev_attr_fan3_input.dev_attr.attr, /* 26 */
>> - &sensor_dev_attr_fan3_label.dev_attr.attr, /* 27 */
>> - &sensor_dev_attr_pwm3.dev_attr.attr, /* 28 */
>> + &sensor_dev_attr_pwm1_enable.dev_attr.attr, /* 23 */
>> + &sensor_dev_attr_fan2_input.dev_attr.attr, /* 24 */
>> + &sensor_dev_attr_fan2_label.dev_attr.attr, /* 25 */
>> + &sensor_dev_attr_pwm2.dev_attr.attr, /* 26 */
>> + &sensor_dev_attr_pwm2_enable.dev_attr.attr, /* 27 */
>> + &sensor_dev_attr_fan3_input.dev_attr.attr, /* 28 */
>> + &sensor_dev_attr_fan3_label.dev_attr.attr, /* 29 */
>> + &sensor_dev_attr_pwm3.dev_attr.attr, /* 30 */
>> + &sensor_dev_attr_pwm3_enable.dev_attr.attr, /* 31 */
>> NULL
>> };
>>
>> @@ -828,13 +876,13 @@ static umode_t i8k_is_visible(struct kobject *kobj, struct attribute *attr,
>> !(i8k_hwmon_flags & I8K_HWMON_HAVE_TEMP10))
>> return 0;
>>
>> - if (index >= 20 && index <= 22 &&
>> + if (index >= 20 && index <= 23 &&
>> !(i8k_hwmon_flags & I8K_HWMON_HAVE_FAN1))
>> return 0;
>> - if (index >= 23 && index <= 25 &&
>> + if (index >= 24 && index <= 27 &&
>> !(i8k_hwmon_flags & I8K_HWMON_HAVE_FAN2))
>> return 0;
>> - if (index >= 26 && index <= 28 &&
>> + if (index >= 28 && index <= 31 &&
>> !(i8k_hwmon_flags & I8K_HWMON_HAVE_FAN3))
>> return 0;
>
>
> Indexes 23, 27 and 31 should not be visible when "smm_auto_fan" is not
> available. Please add check for this.
>
>>
>> @@ -1135,12 +1183,48 @@ static struct dmi_system_id i8k_blacklist_fan_support_dmi_table[] __initdata = {
>> { }
>> };
>>
>> +struct i8k_manual_fan_data {
>> + unsigned int smm_manual_fan;
>> + unsigned int smm_auto_fan;
>> +};
>
> Just cosmetic suggestion: As this structure contains data for both
> manual and automatic mode I would not use "manual" in name. But e.g.
> something like "i8k_bios_fan_control_data"...
>
Or i8k_fan_control_data. Also, "manual" and "auto" for the variable
names would be sufficient.
>> +
>> +enum i8k_manual_fans {
>> + I8K_FAN_34A3_35A3,
>> +};
>> +
>> +static const struct i8k_manual_fan_data i8k_manual_fan_data[] = {
>> + [I8K_FAN_34A3_35A3] = {
>> + .smm_manual_fan = 0x34a3,
>> + .smm_auto_fan = 0x35a3,
>> + },
>> +};
>> +
>> +static struct dmi_system_id i8k_whitelist_manual_fan[] __initdata = {
>> + {
>> + .ident = "Dell Precision 5530",
>> + .matches = {
>> + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
>> + DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Precision 5530"),
>> + },
>> + .driver_data = (void *)&i8k_manual_fan_data[I8K_FAN_34A3_35A3],
>> + },
>> + {
>> + .ident = "Dell Latitude E6440",
>> + .matches = {
>> + DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
>> + DMI_EXACT_MATCH(DMI_PRODUCT_NAME, "Latitude E6440"),
>> + },
>> + .driver_data = (void *)&i8k_manual_fan_data[I8K_FAN_34A3_35A3],
>> + },
>> + { }
>> +};
>> +
Drop the "manual" from the variable names. maybe use "fan_control" instead of
"manual_fan".
>> /*
>> * Probe for the presence of a supported laptop.
>> */
>> static int __init i8k_probe(void)
>> {
>> - const struct dmi_system_id *id;
>> + const struct dmi_system_id *id, *manual_fan;
>> int fan, ret;
>>
>> /*
>> @@ -1200,6 +1284,13 @@ static int __init i8k_probe(void)
>> i8k_fan_max = fan_max ? : I8K_FAN_HIGH; /* Must not be 0 */
>> i8k_pwm_mult = DIV_ROUND_UP(255, i8k_fan_max);
>>
>> + manual_fan = dmi_first_match(i8k_whitelist_manual_fan);
>> + if (manual_fan && manual_fan->driver_data) {
>> + const struct i8k_manual_fan_data *manual_fan_data = manual_fan->driver_data;
>> + smm_manual_fan = manual_fan_data->smm_manual_fan;
>> + smm_auto_fan = manual_fan_data->smm_auto_fan;
>
> As this feature is experimental, I would suggest to add some pr_info
> message that experimental BIOS fan control support is exported via hwmon
> interface. It may happen that it would not work for some people and for
> debugging purposes it would be easier to spot problem if dmesg provides
> such message.
>
>> + }
>> +
>> if (!fan_mult) {
>> /*
>> * Autodetect fan multiplier based on nominal rpm
>
Powered by blists - more mailing lists