Gitiles
Code Review Sign In
gerrit.openbmc.org / openbmc / phosphor-pid-control / 8f73ad767e79f7e8db470f55729c29eaa6c3705e
commit8f73ad767e79f7e8db470f55729c29eaa6c3705e[log] [tgz]
authorAlex.Song <zheng.song@intel.com>Thu Oct 07 00:18:27 2021 +0800
committerZheng Song <zheng.song@intel.com>Wed Dec 01 23:17:39 2021 +0800
tree1ab97488f4b65fb653c2c8f7b11d2d8a33d5626a
parent457993f836338aa0c13a32af803fcbc5227c81f3 [diff]
Make specific UNA sensors not trigger failsafe

By convention, sensors at some states like 'not present',
'power state not matching' are marked as 'unavailable' on dbus.
At such states, some specific sensors should not be considered as
failed and trigger pid 'failsafe'.

A typical example is when a system is powered-off, its CPU/DIMM temp
sensors are 'unavailable', these sensors should not be treated as
'failed' and trigger pid 'failsafe'. This is necessary for systems
whose Fans will keep working when the CPU is off.

This feature is configurable per sensor (valid on thermal sensors). It
can be enabled by setting the Pid controller option
"InputUnavailableAsFailed" to 'false' when one configuring the PID module
via entity-manager, or by setting the sensor option "unavailableAsFailed"
to 'false' when one configuring the PID module via JSON. (These options are
optional and default to 'true')

Tested:
1. On a Fan 'always-on' system, enabale this feature on CPU temp sensors,
poweroff the system, 'unavailable' CPU temp sensors do not trigger the
failsafe mode.
2. 'Unavailable' Fans still trigger the failsafe mode.
3. 'Unfunctional' or 'failed' sensors still trigger the failsafe mode.

Signed-off-by: Zheng Song <zheng.song@intel.com>
Change-Id: I1dd1d76466f43e7dcf51c161c96714f1bcfae88d
13 files changed
tree: 1ab97488f4b65fb653c2c8f7b11d2d8a33d5626a
  1. build/
  2. dbus/
  3. errors/
  4. examples/
  5. experiments/
  6. ipmi/
  7. notimpl/
  8. pid/
  9. sensors/
  10. sysfs/
  11. test/
  12. tools/
  13. .clang-format
  14. .gitignore
  15. .lcovrc
  16. .shellcheck
  17. bootstrap.sh
  18. conf.hpp
  19. configure.ac
  20. configure.md
  21. interfaces.hpp
  22. ipmi.md
  23. LICENSE
  24. main.cpp
  25. MAINTAINERS
  26. Makefile.am
  27. OWNERS
  28. phosphor-pid-control.service.in
  29. README.md
  30. setsensor.cpp
  31. swampd_diagram.png
  32. tuning.md
  33. util.cpp
  34. util.hpp
README.md

phosphor-pid-control

This is a daemon running within the OpenBMC environment. It uses a well-defined configuration file to control the temperature of the tray components to keep them within operating conditions. It may require coordination with host-side tooling and OpenBMC.

Overview

The BMC will run a daemon that controls the fans by pre-defined zones. The application will use thermal control, such that each defined zone is kept within a range and adjusted based on thermal information provided from locally readable sensors as well as host-provided information over an IPMI OEM command.

A system (or tray) will be broken out into one or more zones, specified via configuration files or dbus. Each zone will contain at least one fan and at least one temperature sensor and some device margins. The sensor data can be provided via sysfs, dbus, or through IPMI. In either case, default margins should be provided in case of failure or other unknown situation.

The system will run a control loop for each zone with the attempt to maintain the temperature within that zone within the margin for the devices specified.

Configuring

How to configure phosphor-pid-control

Detailed Design

The software will run as a multi-threaded daemon that runs a control loop for each zone, and has a master thread which listens for dbus messages. Each zone will require at least one fan that it exclusively controls, however, zones can share temperature sensors.

Swampd Architecture

In this figure the communications channels between swampd and ipmid and phosphor-hwmon are laid out.

Zone Specification

A configuration file will need to exist for each board.

Each zone must have at least one fan that it exclusively controls. Each zone must have at least one temperature sensor, but they may be shared.

The internal thermometers specified can be read via sysfs or dbus.

Chassis Delta

Due to data center requirements, the delta between the outgoing air temperature and the environmental air temperature must be no greater than 15C.

IPMI Access to Phosphor-pid-control

OEM-IPMI Definitions

Set Sensor Value

Tools needs to update the thermal controller with information not necessarily available to the BMC. This will comprise of a list of temperature (or margin?) sensors that are updated by the set sensor command. Because they don't represent real sensors in the system, the set sensor handler can simply broadcast the update as a properties update on dbus when it receives the command over IPMI.

Set Fan PWM

A tool can override a specific fan's PWM when we implement the set sensor IPMI command pathway.

Get Fan Tach

A tool can read fan_tach through the normal IPMI interface presently exported for sensors.

Sensor Update Loop

The plan is to listen for fan_tach updates for each fan in a background thread. This will receive an update from phosphor-hwmon each time it updates any sensor it cares about.

By default phosphor-hwmon reads each sensor in turn and then sleeps for 1 second. We'll be updating phosphor-hwmon to sleep for a shorter period -- how short though is still TBD. We'll also be updating phosphor-hwmon to support pwm as a target.

Thermal Control Loops

Each zone will require a control loop that monitors the associated thermals and controls the fan(s). The EC PID loop is designed to hit the fans 10 times per second to drive them to the desired value and read the sensors once per second. We'll be receiving sensor updates with such regularly, however, at present it takes ~0.13s to read all 8 fans. Which can't be read constantly without bringing the system to its knees -- in that all CPU cycles would be spent reading the fans. TBD on how frequently we'll be reading the fan sensors and the impact this will have.

Main Thread

The main thread will manage the other threads, and process the initial configuration files. It will also register a dbus handler for the OEM message.

Enabling Logging & Tuning

By default, swampd won't log information. To enable logging pass "-l" on the command line with a parameter that is the folder into which to write the logs.

The log files will be named {folderpath}/zone_{zoneid}.log.

To enable tuning, pass "-t" on the command line.

See Logging & Tuning for more information.

Code Layout

The code is broken out into modules as follows:

  • dbus - Any read or write interface that uses dbus primarily.
  • experiments - Small execution paths that allow for fan examination including how quickly fans respond to changes.
  • ipmi - Manual control for any zone is handled by receiving an IPMI message. This holds the ipmid provider for receiving those messages and sending them onto swampd.
  • notimpl - These are read-only and write-only interface implementations that can be dropped into a pluggable sensor to make it complete.
  • pid - This contains all the PID associated code, including the zone definition, controller definition, and the PID computational code.
  • scripts - This contains the scripts that convert YAML into C++.
  • sensors - This contains a couple of sensor types including the pluggable sensor's definition. It also holds the sensor manager.
  • sysfs - This contains code that reads from or writes to sysfs.
  • threads - Most of swampd's threads run in this method where there's just a dbus bus that we manage.

Example System Configurations

Two Margin Sensors Into Three Fans (Non-Step PID)

A single zone system where multiple margin thermal sensors are fed into one PID
that generates the output RPM for a set of fans controlled by one PID.

margin sensors as input to thermal pid

fleeting0+---->+-------+    +-------+     Thermal PID sampled
               |  min()+--->+  PID  |     slower rate.
fleeting1+---->+-------+    +---+---+
                                |
                                |
                                | RPM setpoint
                    Current RPM v
                             +--+-----+
  The Fan PID        fan0+--->        |  New PWM  +-->fan0
  samples at a               |        |           |
  faster rate        fan1+--->  PID   +---------->--->fan1
  speeding up the            |        |           |
  fans.              fan2+--->        |           +-->fan2
                       ^     +--------+                +
                       |                               |
                       +-------------------------------+
                              RPM updated by PWM.