speed improvement with kernel 4.2.5 / 4.3.3 / 4.6.6

[fvdw]

Thx for this feedback probably we mis the fan driver, need to check my notes on fan if 2big2 and 5big2. Do you see any difference regarding a fan driver in dmesg when comparing with 3.14.2 on the 2big2

[uelpenich]

Sorry, no differences found.

PS: Power LED brightness is not set at boot time. But it is possible to set brightness by the GUI.

[fvdw]

:scratch
in 4.3.3 the gpio fan driver is present and it is initialized

Code: Select all

[   13.638877] gpio-fan gpio-fan: GPIO fan initialized

the code in the kernel setup file is unchanged, it looks like this

Code: Select all

/* GPIO FAN only 2big2 */
#define NET2BIG_V2_GPIO_FAN_FAIL      25
static struct gpio_fan_alarm net2big_v2_fan_alarm = {
   .gpio      = 25,
   .active_low   = 1,
};

static struct gpio_fan_platform_data net2big_v2_fan_data = {
   .alarm   = &net2big_v2_fan_alarm,
};

static struct platform_device net2big_v2_gpio_fan = {
   .name   = "gpio-fan",
   .id   = -1,
   .dev   = {
      .platform_data   = &net2big_v2_fan_data,
   },
};

Do you have in /sys/devices/platform an directory with name gpio-fan ? and if yes what is the content ?

ps was the fan working ok in 4.2.5 ?

[andjohn]

Hey FVDW,
I'm still getting failure to liniarize, but i think its just over gigabit.

[uelpenich]

Hello,

Do you have in /sys/devices/platform an directory with name gpio-fan ? and if yes what is the content ?

Yes, have a look to the appended file.

was the fan working ok in 4.2.5 ?

yes, at full speed, it is very noisy.

With the 3.14.2 kernel the fan starts running at power-on. It stops (or runs very slowly) when the kernel is booted and rcS starts its actions.
Questions:
What is the desired action?
How works the fan control? Is it a continous fan speed controller or a simple on-off controller?
The kernel setup file doesn't show a gpio channel number? I had a look for a class comparable to the led definitions (/sys/class/leds) and couldn't find anything (/sys/class/gpio ?).

PS: maybe kernel 3.14.2 is wrong and 4.2.5 is o.k. ???

[fvdw]

afaik the 2big2 has an independent fan control unit with a temperature sensor, the behavior in kernel 3.14.2 is the "normal" one. I must check if the driver has been changed in the newer kernels to see if that is the cause of the problem.
The kernel setup file merely only defines a gpio pin for alarm purpose, which is set when the temperature measured by the fan goes above a threshold. The fan on the 5big2 is controlled differently there we have a temperature measurement of the disk and we set the speed of the fan by our own drivers

ps what is inside the sub directories "driver" and "hwmon" ? (in sys/devices/platform/gpio-fan)
Can you compare what you have in 4.3.3 with what is present in 3.1.14

[fvdw]

Yes gpio-fan driver is modified, however I can not see why the changes they made would affect the operation on the 2big2.
Anyhow added a recompiled kernel with the gpio-fan.c file of 4.3.3 replaced by the one in 3.14.2 let see if the fan has now the same behavior as in 3.14.2.
If it is still running at full speed I guess the alarm gpio setting might be a reason


--edit file removed, not a solution

[uelpenich]

Hello,
thank you for the fast reply.

what is inside the sub directories "driver" and "hwmon" ? (in sys/devices/platform/gpio-fan)

Here it is. Not easy to understand due to all the links :-)

root@2big2:/sys/devices/platform/gpio-fan # ls -l
total 0
lrwxrwxrwx 1 root root 0 2016-01-03 21:48 driver -> ../../../bus/platform/drivers/gpio-fan
-rw-r--r-- 1 root root 4096 2016-01-03 21:48 driver_override
drwxr-xr-x 3 root root 0 2016-01-03 21:37 hwmon
-r--r--r-- 1 root root 4096 2016-01-03 21:48 modalias
drwxr-xr-x 2 root root 0 2016-01-03 21:48 power
lrwxrwxrwx 1 root root 0 2016-01-03 21:48 subsystem -> ../../../bus/platform
-rw-r--r-- 1 root root 4096 2016-01-03 21:37 uevent
root@2big2:/sys/devices/platform/gpio-fan #

root@2big2:/sys/devices/platform/gpio-fan # cd driver
root@2big2:/sys/devices/platform/gpio-fan/driver # ls -l
total 0
--w------- 1 root root 4096 2016-01-03 21:53 bind
lrwxrwxrwx 1 root root 0 2016-01-03 21:53 gpio-fan -> ../../../../devices/platform/gpio-fan
--w------- 1 root root 4096 2016-01-03 21:53 uevent
--w------- 1 root root 4096 2016-01-03 21:53 unbind
root@2big2:/sys/devices/platform/gpio-fan/driver #

root@2big2:/sys/devices/platform/gpio-fan # cd hwmon
root@2big2:/sys/devices/platform/gpio-fan/hwmon # ls -l
total 0
drwxr-xr-x 3 root root 0 2016-01-03 21:37 hwmon1
root@2big2:/sys/devices/platform/gpio-fan/hwmon # cd hwmon1
root@2big2:/sys/devices/platform/gpio-fan/hwmon/hwmon1 # ls -l
total 0
lrwxrwxrwx 1 root root 0 2016-01-03 21:50 device -> ../../../gpio-fan
-r--r--r-- 1 root root 4096 2016-01-03 21:50 fan1_alarm
-r--r--r-- 1 root root 4096 2016-01-03 21:50 name
drwxr-xr-x 2 root root 0 2016-01-03 21:50 power
lrwxrwxrwx 1 root root 0 2016-01-03 21:50 subsystem -> ../../../../../class/hwmon
-rw-r--r-- 1 root root 4096 2016-01-03 21:37 uevent
root@2big2:/sys/devices/platform/gpio-fan/hwmon/hwmon1 # cat fan1_alarm
0
root@2big2:/sys/devices/platform/gpio-fan/hwmon/hwmon1 # cat name
gpio_fan
root@2big2:/sys/devices/platform/gpio-fan/hwmon/hwmon1 # cat uevent
root@2big2:/sys/devices/platform/gpio-fan/hwmon/hwmon1 # cd power
root@2big2:/sys/devices/platform/gpio-fan/hwmon/hwmon1/power # ls -l
total 0
-rw-r--r-- 1 root root 4096 2016-01-03 21:51 autosuspend_delay_ms
-rw-r--r-- 1 root root 4096 2016-01-03 21:51 control
-r--r--r-- 1 root root 4096 2016-01-03 21:51 runtime_active_time
-r--r--r-- 1 root root 4096 2016-01-03 21:51 runtime_status
-r--r--r-- 1 root root 4096 2016-01-03 21:51 runtime_suspended_time
root@2big2:/sys/devices/platform/gpio-fan/hwmon/hwmon1/power #

Yes gpio-fan driver is modified, however I can not see why the changes they made would affect the operation on the 2big2.
Anyhow added a recompiled kernel with the gpio-fan.c file of 4.3.3 replaced by the one in 3.14.2 let see if the fan has now the same behavior as in 3.14.2.
If it is still running at full speed I guess the alarm gpio setting might be a reason

It is the same behavior as the 4.3.3 gpio-fan.c version: the fan runs at full speed. Can I do some debugging? With the led debugging I used the nano editor to switch the leds.

You told me the fan control reads the disk temperature to control the fan speed. Is it possibly a bug or different behaviour with the disk driver?

PS: now the content of /sys/devices/platform/gpio-fan looks a little bit different:

root@2big2:/sys/devices/platform/gpio-fan # ls -l
total 0
-rw-r--r-- 1 root root 4096 2016-01-03 22:02 driver_override
-r--r--r-- 1 root root 4096 2016-01-03 22:02 modalias
drwxr-xr-x 2 root root 0 2016-01-03 22:02 power
lrwxrwxrwx 1 root root 0 2016-01-03 22:02 subsystem -> ../../../bus/platform
-rw-r--r-- 1 root root 4096 2016-01-03 21:58 uevent
root@2big2:/sys/devices/platform/gpio-fan #

[fvdw]

oot@2big2:/sys/devices/platform/gpio-fan # ls -l
total 0
-rw-r--r-- 1 root root 4096 2016-01-03 22:02 driver_override
-r--r--r-- 1 root root 4096 2016-01-03 22:02 modalias
drwxr-xr-x 2 root root 0 2016-01-03 22:02 power
lrwxrwxrwx 1 root root 0 2016-01-03 22:02 subsystem -> ../../../bus/platform
-rw-r--r-- 1 root root 4096 2016-01-03 21:58 uevent
root@2big2:/sys/devices/platform/gpio-fan #

This not good, seems driver is not loaded, I mis hwmon folder.
So using the old gpio-fan code is not the way to go, forget kernel 5.

If you run kernel 3 (4.3.3.) and compare with 3.14.2 do you see any difference in in the /sys/devices/platform/gpio-fan files and folders ?

Where you could experiment is with /sys/devices/platform/gpio-fan/hwmon/hwmon1/control
That is write-able. You can use cat to see current content, compare 3.14.2 with 4.3.3To change the value you can cd to the directory and then for example use this command to put the value "1" inside

Code: Select all

echo 1 > control

The same with other files that have a rw attribute like /sys/devices/platform/gpio-fan/driver-override

ps about temperature, seems you misread my post, no this driver doesn't read disk temperature, it has its own temperature sensor and control board that controls the fan (afaik) so different from 5big2. The question now is why does it behave different :scratch

[fvdw]

Hey FVDW,
I'm still getting failure to liniarize, but i think its just over gigabit.

I don't understand what you actually mean with only over gigabit.
I use at home a gigabit lan and see no problems (cable ?)