Fan PWM monitoring and control
rgrs
rgrs at protonmail.com
Fri Oct 4 23:48:36 AEST 2019
Got it now.
Thank you very much :-)
-------- Original Message --------
On 4 Oct 2019, 19:14, Matthew Barth wrote:
> On 10/4/19 5:38 AM, rgrs wrote:
>> Hi Matt,
>>
>> Thanks for the reply.
>>
>>> floor speed (the lowest the fans are allowed) based on the given average
>>> sensor values for the group of sensors
>>
>> Is this all hwmon values? as in RPM : Average_temp_in_group
> No, as you pulled it from the events.yaml for fan control, its input
> into fan control. You can find this action function in action.hpp/cpp
> within phosphor-fan-presence/control. This is all for configuring how
> you want the fan control algorithm to work.
>>
>> Thanks,
>> Raj
>>
>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>> On Friday, October 4, 2019 2:29 AM, Matthew Barth <msbarth at linux.ibm.com> wrote:
>>
>>> On 10/1/19 3:26 AM, rgrs wrote:
>>>
>>>> Hi Matt,
>>>> Please can you help me with mapping values in events.yaml?
>>>> (https://github.com/openbmc/phosphor-fan-presence/blob/master/control/example/events.yaml)
>>>> How are these values derived?
>>>> How do I make sure, that this is valid for my platform?
>>>> Please can you point to documentation regarding this, i am not able to find from github search.
>>>
>>> Unfortunately in this repo's development there was not alot of time
>>> given to get a fully functional and flexible solution in place for
>>> delivery on the system using this as its fan control solution.
>>> Documentation is limited to the example yaml files and the doxygen
>>> located within the source. Enhancing this documentation is on my list of
>>> todo's when I get the time.
>>>
>>>> actions:
>>>>
>>>> =========
>>>>
>>>> - name: set_floor_from_average_sensor_value
>>>>
>>>> ============================================
>>>>
>>>> map:
>>>>
>>>> =====
>>>>
>>>> value:
>>>>
>>>> =======
>>>>
>>>> - 27000: 3500
>>>>
>>>> ==============
>>>>
>>>> - 32000: 4600
>>>>
>>>> ==============
>>>>
>>>> - 37000: 5200
>>>>
>>>> ==============
>>>>
>>>> - 40000: 5800
>>>>
>>>> ==============
>>>>
>>>> type: std::map<int64_t, uint64_t>
>>>>
>>>> ==================================
>>>>
>>>> - name: set_ceiling_from_average_sensor_value
>>>>
>>>> ==============================================
>>>>
>>>> map:
>>>>
>>>> =====
>>>>
>>>> value:
>>>>
>>>> =======
>>>>
>>>> - 25000: 7200
>>>>
>>>> ==============
>>>>
>>>> - 27000: 10500
>>>>
>>>> ===============
>>>>
>>>> type: std::map<int64_t, uint64_t>
>>>>
>>>> ==================================
>>>
>>> These values are derived by performing thermal characterization. In this
>>> case, it was thermally determined that the values for setting the fan
>>> floor speed (the lowest the fans are allowed) based on the given average
>>> sensor values for the group of sensors defined on the event containing
>>> this action. Similarly, a fan ceiling speed is put in place based on the
>>> same group of sensors' average value so the fans are not allowed to
>>> increase above that speed.
>>>
>>> Validating thermal control parameters for any type of fan control needs
>>> characterization done against the system's components(Proc, dimms, pci
>>> cards, etc..) to enabled the correct fan control parameters.
>>>
>>> Matt
>>>
>>>> Thanks,
>>>> Raj
>>>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>>>> On Thursday, September 26, 2019 8:19 PM, Matthew Barth msbarth at linux.ibm.com wrote:
>>>>
>>>>> On 9/25/19 1:45 AM, rgrs wrote:
>>>>>
>>>>>> Hi Matt,
>>>>>>
>>>>>>> Is the `pwm[#]_enable` sysfs file set to a '1' at any point after the AC cycle (i.e. after a poweron?)?
>>>>>>
>>>>>> No
>>>>>>
>>>>>>> If you think the configs are set correctly, is the hwmon service for the pwm devices running after the AC cycle while at BMC standby?
>>>>>>
>>>>>> I'm not sure, I think phosphor-fan starts after chassis power-on.
>>>>>
>>>>> I was asking if the hwmon service is started at BMC standby. It sounds
>>>>> like it still may be config related but I'm not able to determine where
>>>>> it may be. If you believe that hwmon is not setting the pwm[#]_enable
>>>>> file correctly, please debug and provide a fix if that's the case, it'd
>>>>> be much appreciated!
>>>>> Matt
>>>>>
>>>>>> Thanks,
>>>>>> Raj
>>>>>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>>>>>> On Tuesday, September 24, 2019 8:25 PM, Matthew Barth msbarth at linux.ibm.com wrote:
>>>>>>
>>>>>>> On 9/23/19 11:25 PM, rgrs wrote:
>>>>>>>
>>>>>>>> Thanks Matt :) It works now.
>>>>>>>> I have one small issue on configuration of ADT7462.
>>>>>>>> After AC power cycle, it defaults to "automatic" fan control.
>>>>>>>> We need to change this "manual" so that phosphor-fan control register writes are not ignored by ADT7462.
>>>>>>>> I came across https://github.com/openbmc/docs/blob/master/sensor-architecture.md
>>>>>>>>
>>>>>>>>> ENABLE
>>>>>>>>> Will write a value to a pwmN_enable file on startup if present.
>>>>>>>>> ENABLE_fan1 = 2 #Write a 2 to pwm1_enable
>>>>>>>>
>>>>>>>> But this doesn't seem to work. We need it to be initialized to '1'.
>>>>>>>> (It always defaults to 2 after AC cycle)
>>>>>>>
>>>>>>> This sounds like the correct hwmon config to use to set it to a '1' upon
>>>>>>> hwmon starting after an AC cycle. First ensure your hwmon config has
>>>>>>> these `ENABLE_fan[#]` entries set to a corresponding `pwm[#]_enable`
>>>>>>> file in sysfs.
>>>>>>> Is the `pwm[#]_enable` sysfs file set to a '1' at any point after the AC
>>>>>>> cycle (i.e. after a poweron?)? If you think the configs are set
>>>>>>> correctly, is the hwmon service for the pwm devices running after the AC
>>>>>>> cycle while at BMC standby?
>>>>>>>
>>>>>>>> Is host-power-on script a good place to "echo 1 > pwmN_enable"?
>>>>>>>> (planning to add a script to initialize before host power on)
>>>>>>>> Thanks,
>>>>>>>> Raj
>>>>>>>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>>>>>>>> On Friday, September 20, 2019 9:49 PM, Matthew Barth msbarth at linux.ibm.com wrote:
>>>>>>>>
>>>>>>>>> On 9/20/19 4:55 AM, rgrs wrote:
>>>>>>>>>
>>>>>>>>>> Hi Matt,
>>>>>>>>>> I'm able to control fan speed via busctl.
>>>>>>>>>> root at saber:/sys/class/hwmon/hwmon0# busctl set-property xyz.openbmc_project.Hwmon-[2429843755](tel:2429843755).Hwmon1 /xyz/openbmc_project/sensors/fan_tach/FAN0 xyz.openbmc_project.Control.FanPwm Target t '250'
>>>>>>>>>> root at saber:/sys/class/hwmon/hwmon0# cat fan*_input
>>>>>>>>>> 12498
>>>>>>>>>> 0
>>>>>>>>>
>>>>>>>>> Great! that's progress.
>>>>>>>>>
>>>>>>>>>> I need a way to verify if the sensor associations are working properly that fans will ramp up when associated temperature sensor's value increases.
>>>>>>>>>
>>>>>>>>> Be sure that your fans.yaml contains the correct sensor name ("FAN0",
>>>>>>>>> etc..) for each fan that contains the Target property on the FanPwm
>>>>>>>>> interface.
>>>>>>>>> You can use busctl to monitor the target pwm sensor changes based on
>>>>>>>>> what you configured within events.yaml for speed increases. This would
>>>>>>>>> show that fan control is requesting updated targets to hwmon. Then hwmon
>>>>>>>>> would write these values out to the pwm sysfs file for your fan
>>>>>>>>> controller device.
>>>>>>>>>
>>>>>>>>>> Thanks,
>>>>>>>>>> Raj
>>>>>>>>>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>>>>>>>>>> On Friday, September 20, 2019 2:57 PM, rgrs rgrs at protonmail.com wrote:
>>>>>>>>>>
>>>>>>>>>>> Hi Matt,
>>>>>>>>>>>
>>>>>>>>>>>>> Not sure I follow what you mean by "no pwm writes", is there an
>>>>>>>>>>>
>>>>>>>>>>> application writing to the fan sensor's pwm interface target property?
>>>>>>>>>>> I thought phoshor-fan control writes to pwm to increase fan speed based on temperature sensors mentioned in events.yaml
>>>>>>>>>>> (recipes-phosphor/fans/phosphor-fan-control-events-config/events.yaml)
>>>>>>>>>>> Is it not the case?
>>>>>>>>>>> Thanks,
>>>>>>>>>>> Raj
>>>>>>>>>>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>>>>>>>>>>> On Thursday, September 19, 2019 8:26 PM, Matthew Barth msbarth at linux.ibm.com wrote:
>>>>>>>>>>>
>>>>>>>>>>>> On 9/19/19 5:16 AM, rgrs wrote:
>>>>>>>>>>>>
>>>>>>>>>>>>> Hi Patrick, Lei,
>>>>>>>>>>>>> Thank you, I made some changes according your comments
>>>>>>>>>>>>>
>>>>>>>>>>>>> - added Mutability::Read to fan yaml.
>>>>>>>>>>>>> - I didn't see ipmi-sensor-inventory.yaml of romulus having fan sensors, so removed them from IPMI.
>>>>>>>>>>>>>
>>>>>>>>>>>>> Now,"busctl monitor" displays Fan tach reads, no pwm writes.
>>>>>>>>>>>>
>>>>>>>>>>>> Not sure I follow what you mean by "no pwm writes", is there an
>>>>>>>>>>>> application writing to the fan sensor's pwm interface target property?
>>>>>>>>>>>>
>>>>>>>>>>>>> Is there any way to manually set PWM value and test this?
>>>>>>>>>>>>> How do I verify PWM values are being written to ADT7462?
>>>>>>>>>>>>
>>>>>>>>>>>> If no other application are writing to the PWM sysfs entries, you should
>>>>>>>>>>>> be able to `echo` a value to each fan's PWM sysfs file to alter the fan
>>>>>>>>>>>> speeds.
>>>>>>>>>>>>
>>>>>>>>>>>>> Another observation is Inventory.Item seems to have invalid(stale) entries.
>>>>>>>>>>>>
>>>>>>>>>>>> I believe inventory is persisted by default so that could explain why
>>>>>>>>>>>> these entries are still showing up. These are persisted under
>>>>>>>>>>>> `/var/lib/phosphor-inventory-manager/..`.
>>>>>>>>>>>>
>>>>>>>>>>>>> IPMI sensor names appear in inventory even after removing from IPMI sensor yaml files (exported JSON in attachment).
>>>>>>>>>>>>> Logs:
>>>>>>>>>>>>> busctl monitor | grep -i fan
>>>>>>>>>>>>> ========================================================================================================================================================
>>>>>>>>>>>>> Monitoring bus message stream.
>>>>>>>>>>>>> Sender=:1.45 Path=/xyz/openbmc_project/sensors/fan_tach/FAN0 Interface=org.freedesktop.DBus.Properties Member=PropertiesChanged
>>>>>>>>>>>>> Sender=:1.45 Path=/xyz/openbmc_project/sensors/fan_tach/FAN1 Interface=org.freedesktop.DBus.Properties Member=PropertiesChanged
>>>>>>>>>>>>> Sender=:1.45 Path=/xyz/openbmc_project/sensors/fan_tach/FAN2 Interface=org.freedesktop.DBus.Properties Member=PropertiesChanged
>>>>>>>>>>>>> [...]
>>>>>>>>>>>>> Sender=:1.45 Path=/xyz/openbmc_project/sensors/fan_tach/FAN1 Interface=org.freedesktop.DBus.Properties Member=PropertiesChanged
>>>>>>>>>>>>> Sender=:1.45 Path=/xyz/openbmc_project/sensors/fan_tach/FAN3 Interface=org.freedesktop.DBus.Properties Member=PropertiesChanged
>>>>>>>>>>>>> ^C
>>>>>>>>>>>>> busctl get-property xyz.openbmc_project.Hwmon-[2429843755](tel:2429843755).Hwmon1 /xyz/openbmc_project/sensors/fan_tach/FAN0 xyz.openbmc_project.Control.FanPwm Target
>>>>>>>>>>>>> =============================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================================
>>>>>>>>>>>>> t 0
>>>>>>>>>>>>> busctl set-property xyz.openbmc_project.Hwmon-[2429843755](tel:2429843755).Hwmon1 /xyz/openbmc_project/sensors/fan_tach/FAN0 xyz.openbmc_project.Control.FanPwm Target 8000
>>>>>>>>>>>>> ==============================================================================================================================================================
>>>>>>>>>>>>
>>>>>>>>>>>> This is not a valid call to `busctl set-property...`, its missing the
>>>>>>>>>>>> property's signature type (refer to `busctl help`). Also, a target PWM
>>>>>>>>>>>> of 8000 is invalid, PWM is 0 - 255.
>>>>>>>>>>>>
>>>>>>>>>>>>> Failed to create bus message: Invalid argument
>>>>>>>>>>>>> busctl introspect --no-pager xyz.openbmc_project.Hwmon-[2429843755](tel:2429843755).Hwmon1 /xyz/openbmc_project/sensors/fan_tach/FAN1
>>>>>>>>>>>>> ===================================================================================================================================================================
>>>>>>>>>>>>> NAME TYPE SIGNATURE RESULT/VALUE FLAGS
>>>>>>>>>>>>> org.freedesktop.DBus.Introspectable interface - - -
>>>>>>>>>>>>> .Introspect method - s -
>>>>>>>>>>>>> org.freedesktop.DBus.Peer interface - - -
>>>>>>>>>>>>> .GetMachineId method - s -
>>>>>>>>>>>>> .Ping method - - -
>>>>>>>>>>>>> org.freedesktop.DBus.Properties interface - - -
>>>>>>>>>>>>> .Get method ss v -
>>>>>>>>>>>>> .GetAll method s a{sv} -
>>>>>>>>>>>>> .Set method ssv - -
>>>>>>>>>>>>> .PropertiesChanged signal sa{sv}as - -
>>>>>>>>>>>>> xyz.openbmc_project.Control.FanPwm interface - - -
>>>>>>>>>>>>> .Target property t 0 emits-change writable
>>>>>>>>>>>>> xyz.openbmc_project.Sensor.Value interface - - -
>>>>>>>>>>>>> .MaxValue property x 0 emits-change writable
>>>>>>>>>>>>> .MinValue property x 0 emits-change writable
>>>>>>>>>>>>> .Scale property x 0 emits-change writable
>>>>>>>>>>>>> .Unit property s "xyz.openbmc_project.Sensor.Value.Unit.… emits-change writable
>>>>>>>>>>>>> .Value property x 2983 emits-change writable
>>>>>>>>>>>>> xyz.openbmc_project.State.Decorator.OperationalStatus interface - - -
>>>>>>>>>>>>> .Functional property b true emits-change writable
>>>>>>>>>>>>> Thanks,
>>>>>>>>>>>>> Raj
>>>>>>>>>>>>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>>>>>>>>>>>>> On Wednesday, September 18, 2019 9:35 PM, Patrick Venture venture at google.com wrote:
>>>>>>>>>>>>>
>>>>>>>>>>>>>> On Wed, Sep 18, 2019 at 2:30 AM rgrs rgrs at protonmail.com wrote:
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Hi Patrick, Matt,
>>>>>>>>>>>>>>> I tried instructions from this wiki: https://developer.ibm.com/linuxonpower/2019/01/07/how-to-port-openbmc/
>>>>>>>>>>>>>>> There is no GPIO to detect presence of fans, so tach readings is used instead.
>>>>>>>>>>>>>>> Here's what I did while trying to enable Fan control:
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> 1. Added Fan tach sensor in hwmon (adt7462.conf) is in the form LABEL=FANx_RPM
>>>>>>>>>>>>>>> 2. "ipmi-sensor-inventory.yaml" has below config:
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> 60: &fan_tach
>>>>>>>>>>>>>>> sensorType: 0x04
>>>>>>>>>>>>>>> path: /xyz/openbmc_project/sensors/fan_tach/FAN0_RPM
>>>>>>>>>>>>>>> sensorReadingType: 0x01
>>>>>>>>>>>>>>> multiplierM: 100
>>>>>>>>>>>>>>> offsetB: 0
>>>>>>>>>>>>>>> bExp: 0
>>>>>>>>>>>>>>> unit: xyz.openbmc_project.Sensor.Value.Unit.RPMS
>>>>>>>>>>>>>>> scale: 0
>>>>>>>>>>>>>>> rExp: 0
>>>>>>>>>>>>>>> serviceInterface: org.freedesktop.DBus.Properties
>>>>>>>>>>>>>>> readingType: readingData
>>>>>>>>>>>>>>> interfaces:
>>>>>>>>>>>>>>> xyz.openbmc_project.Sensor.Value:
>>>>>>>>>>>>>>> Value:
>>>>>>>>>>>>>>> Offsets:
>>>>>>>>>>>>>>> 0xFF:
>>>>>>>>>>>>>>> type: int64_t
>>>>>>>>>>>>>>> and FAN PWM as
>>>>>>>>>>>>>>> 89: &fanpwm
>>>>>>>>>>>>>>> sensorType: 0x04
>>>>>>>>>>>>>>> path: /xyz/openbmc_project/sensors/fan_tach/FAN0_RPM
>>>>>>>>>>>>>>> sensorReadingType: 0x01
>>>>>>>>>>>>>>> multiplierM: 1
>>>>>>>>>>>>>>> offsetB: 0
>>>>>>>>>>>>>>> bExp: 0
>>>>>>>>>>>>>>> mutability: Mutability::Write
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> You may need to add ::Read here too, I'm not sure anymore, but you can
>>>>>>>>>>>>>> check the script that builds this:
>>>>>>>>>>>>>> https://github.com/openbmc/phosphor-host-ipmid/blob/master/scripts/writesensor.mako.cpp#L69
>>>>>>>>>>>>>> It'll default to read, but if you've only set write, I don't know.
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> serviceInterface: org.freedesktop.DBus.Properties
>>>>>>>>>>>>>>> readingType: readingData
>>>>>>>>>>>>>>> interfaces:
>>>>>>>>>>>>>>> xyz.openbmc_project.Control.FanPwm:
>>>>>>>>>>>>>>> Target:
>>>>>>>>>>>>>>> Offsets:
>>>>>>>>>>>>>>> 0xFF:
>>>>>>>>>>>>>>> type: uint64_t
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> 3. I copied meta-romulus/recipes-phosphor/fans to my platform and changed FAN names from "fanx" to "FANx_RPM" in yaml files.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> This is what get in busctl tree related to Fans:
>>>>>>>>>>>>>>> (I expected to see FanPwm entries in busctl and I'm not sure why there are multiple entries in inventory).
>>>>>>>>>>>>>>> Service xyz.openbmc_project.Hwmon-[2429843755](tel:2429843755).Hwmon1:
>>>>>>>>>>>>>>> `-/xyz`-/xyz/openbmc_project
>>>>>>>>>>>>>>> `-/xyz/openbmc_project/sensors |-/xyz/openbmc_project/sensors/fan_tach | |-/xyz/openbmc_project/sensors/fan_tach/FAN0_RPM | |-/xyz/openbmc_project/sensors/fan_tach/FAN1_RPM | |-/xyz/openbmc_project/sensors/fan_tach/FAN2_RPM |`-/xyz/openbmc_project/sensors/fan_tach/FAN3_RPM
>>>>>>>>>>>>>>> `-/xyz/openbmc_project/sensors/temperature`-/xyz/openbmc_project/sensors/temperature/BACKPLANE_TEMP
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> You don't get a separate entry for the PWM from phosphor-hwmon.
>>>>>>>>>>>>>> It's a separate interface on the same service and path.
>>>>>>>>>>>>>> I recommend you introspect on the path and see what interfaces are
>>>>>>>>>>>>>> available (as well as adding "| read" above):
>>>>>>>>>>>>>> busctl introspect --no-pager
>>>>>>>>>>>>>> xyz.openbmc_project.Hwmon-[2429843755](tel:2429843755).Hwmon1
>>>>>>>>>>>>>> /xyz/openbmc_project/sensors/fan_tach/FAN0_RPM
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> Service xyz.openbmc_project.Inventory.Manager:
>>>>>>>>>>>>>>> `-/xyz`-/xyz/openbmc_project
>>>>>>>>>>>>>>> `-/xyz/openbmc_project/inventory`-/xyz/openbmc_project/inventory/system
>>>>>>>>>>>>>>> `-/xyz/openbmc_project/inventory/system/chassis`-/xyz/openbmc_project/inventory/system/chassis/motherboard
>>>>>>>>>>>>>>> |-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN0
>>>>>>>>>>>>>>> | `-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN0/FAN0 |-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN0_RPM |`-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN0_RPM/FAN0_RPM
>>>>>>>>>>>>>>> |-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN1
>>>>>>>>>>>>>>> | `-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN1/FAN1 |-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN1_RPM |`-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN1_RPM/FAN1_RPM
>>>>>>>>>>>>>>> |-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN2
>>>>>>>>>>>>>>> | `-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN2/FAN2 |-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN2_RPM |`-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN2_RPM/FAN2_RPM
>>>>>>>>>>>>>>> |-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN3
>>>>>>>>>>>>>>> | `-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN3/FAN3 |-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN3_RPM |`-/xyz/openbmc_project/inventory/system/chassis/motherboard/FAN3_RPM/FAN3_RPM
>>>>>>>>>>>>>>> |-/xyz/openbmc_project/inventory/system/chassis/motherboard/fan0
>>>>>>>>>>>>>>> |-/xyz/openbmc_project/inventory/system/chassis/motherboard/fan1
>>>>>>>>>>>>>>> |-/xyz/openbmc_project/inventory/system/chassis/motherboard/fan2
>>>>>>>>>>>>>>> `-/xyz/openbmc_project/inventory/system/chassis/motherboard/fan3
>>>>>>>>>>>>>>> Thanks,
>>>>>>>>>>>>>>> Raj
>>>>>>>>>>>>>>> ‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
>>>>>>>>>>>>>>> On Tuesday, September 17, 2019 9:18 PM, Patrick Venture venture at google.com wrote:
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>> On Tue, Sep 17, 2019 at 8:32 AM Matthew Barth msbarth at linux.ibm.com wrote:
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> On 9/17/19 5:40 AM, rgrs wrote:
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> Hi,
>>>>>>>>>>>>>>>>>> I need some help with phosphor hwmon related to fan sensors.
>>>>>>>>>>>>>>>>>> I'm trying to control Fans in a server motherboard using ADT7462
>>>>>>>>>>>>>>>>>> controller. Hw mon porting is done and the sysfs interfaces are working
>>>>>>>>>>>>>>>>>> as expected.
>>>>>>>>>>>>>>>>>> While I am able to read RPM, i don't find pwm/target to control the fans
>>>>>>>>>>>>>>>>>> exposed as dbus objects.
>>>>>>>>>>>>>>>>>> Do you have the hwmon config set similar to this to denote the fan
>>>>>>>>>>>>>>>>>> targets are of PWM type?
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> https://github.com/openbmc/openbmc/blob/master/meta-ibm/meta-romulus/recipes-phosphor/sensors/phosphor-hwmon/obmc/hwmon/ahb/apb/pwm-tacho-controller%401e786000.conf
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> busctl --no-page tree xyz.openbmc_project.FanSensor
>>>>>>>>>>>>>>>>>> ====================================================
>>>>>>>>>>>>>>>>>> Failed to introspect object / of service xyz.openbmc_project.FanSensor:
>>>>>>>>>>>>>>>>>> The name is not activatable
>>>>>>>>>>>>>>>>>> No objects discovered.
>>>>>>>>>>>>>>>>>> That does not look like a hwmon service name. If you are using
>>>>>>>>>>>>>>>>>> phosphor-objmgr, you can use `# mapper get-service /xyz/openbmc_project/sensors/fan_tach` otherwise you'll need to find the
>>>>>>>>>>>>>>>>>> hwmon service name associated with your fan controller device.
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>> running`busctl tree --no-pager` will just list everything, and you
>>>>>>>>>>>>>>>> should see your fans in there. You can use the mapper too. I was
>>>>>>>>>>>>>>>> curious, are you using entity-manager/dbus-sensors? When you mention
>>>>>>>>>>>>>>>> FanSensor as the service, that's what it sounds like.
>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> I came across this post, I think it is similar to my issue.
>>>>>>>>>>>>>>>>>> (https://gerrit.openbmc-project.xyz/c/openbmc/phosphor-hwmon/+/8353)
>>>>>>>>>>>>>>>>>> Please someone point to latest documentation or reference Fan porting
>>>>>>>>>>>>>>>>>> from openBMC repo?
>>>>>>>>>>>>>>>>>> What are the exact fields in YAML that are used create FanPWM objects.
>>>>>>>>>>>>>>>>>> The Romulus machine should be a good reference for PWM based fan control
>>>>>>>>>>>>>>>>>> using phosphor-hwmon. It uses the
>>>>>>>>>>>>>>>>>> phosphor-fan-presence/[control|monitor] apps for controlling and
>>>>>>>>>>>>>>>>>> monitoring fans
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> https://github.com/openbmc/openbmc/tree/master/meta-ibm/meta-romulus
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>>> (I'm using Feb release in my sandbox,
>>>>>>>>>>>>>>>>>> https://github.com/openbmc/docs/blob/master/release/release-notes.md#26-feb-4-2019)
>>>>>>>>>>>>>>>>>> Thanks,
>>>>>>>>>>>>>>>>>> Raj
>>>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>>> Matt
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/openbmc/attachments/20191004/943872f3/attachment-0001.htm>
More information about the openbmc
mailing list