Append node number to dbus names more directly

This way we don't needlessly duplicate the string constants.

Change-Id: Iba0a19ae3a95472fb25c61fa0cf612bd75d4fcab
Signed-off-by: Zev Weiss <zev@bewild

Append node number to dbus names more directly

This way we don't needlessly duplicate the string constants.

Change-Id: Iba0a19ae3a95472fb25c61fa0cf612bd75d4fcab
Signed-off-by: Zev Weiss <zev@bewilderbeest.net>

show more ...

DBUS service name numbering for single and multi hosts

For single host based systems additional busname is added
with "0" at the end of the name. Dbus service name of single hosts is
created with a

DBUS service name numbering for single and multi hosts

For single host based systems additional busname is added
with "0" at the end of the name. Dbus service name of single hosts is
created with and without numbering.
in case of multiple hosts dbus service name will be postfixed with
respective host number.

ex :
single host : xyz.openbmc_project.State.Host and
xyz.openbmc_project.State.Host0

multi host : xyz.openbmc_project.State.Host1, xxx.State.Host2 ... etc

TESTED: Tested and verified in Tiogapass and YosemteV2.
verified using power control busctl commands.

Signed-off-by: Naveen Moses <naveen.mosess@hcl.com>
Change-Id: I51ba1362664dd5582b156a84d31b2403d10411b1

show more ...

Add power/reset button mask checks for RequestedPowerTransition

To fully honor button masking, the RequestedPowerTransition interface
now checks the button mask state prior to actually initiating th

Add power/reset button mask checks for RequestedPowerTransition

To fully honor button masking, the RequestedPowerTransition interface
now checks the button mask state prior to actually initiating the
requested transition. This will make it so the IPMI and Redfish
commands will not be able to subvert the masked buttons.

The last patch set only blocked resets and graceful power transitions.
This patch adds the similar behavior for power transitions.

Tested: disable front panel buttons (ipmitool raw 0 0x0a 0xf0)
and then attempt to reset and power cycle with ipmi
and redfish.

ipmitool chassis power off
redfish /v1/Systems/system/Actions/ComputerSystem.Reset
{ "ResetType": "ForceOff" }

Host power stays on and uninterrupted.

Change-Id: Ic3be2bc7cf364524949c1ee31037fff8d3438d19
Signed-off-by: Vernon Mauery <vernon.mauery@linux.intel.com>

show more ...

Add multiple host Slot AC Power Support

This change provides Slot AC Power feature for the system which has Slot
AC Power support. To enable this feature, the compiler
flag 'SLOT_AC_POWER' must be d

Add multiple host Slot AC Power Support

This change provides Slot AC Power feature for the system which has Slot
AC Power support. To enable this feature, the compiler
flag 'SLOT_AC_POWER' must be defined from the platform side.

The Chassis properites On, Off and PowerCycle will provide the Slot
AC Power On, Off and PowerCycle for the platform with the compiler flag
'SLOT_AC_POWER'

TESTED: Tested and verified in YosemiteV2

Signed-off-by: Naveen Moses <naveen.mosess@hcl.com>
Change-Id: I21b9326962099dfa59ff749d753121f74830db57

show more ...

Dbus Based Gpio Event monitoring

This change provide the implementation for D-bus based Gpio Event monitoring.
For the platform not having direct GPIO access can use dbus based event
monitor by conf

Dbus Based Gpio Event monitoring

This change provide the implementation for D-bus based Gpio Event monitoring.
For the platform not having direct GPIO access can use dbus based event
monitor by configuring dbus names in Json file. x86-power-control will monitor
the property change from the given DbusName and take appropriate action

Updated parsing logic of reading config values from json.

TESTED:Tested and verified in Tiogapass and YosemteV2. The existing functionality
is not broken after adding the JSON parser implementation in Tiogapass.

Signed-off-by: Priyatharshan P <priyatharshanp@hcl.com>
Change-Id: I480c62eccebf6d0a3c4609afaf24d55265d99602

show more ...

Add power/reset button mask checks for RequestedHostTransition

To fully honor button masking, the RequestedHostTransition interface
now checks the button mask state prior to actually initiating the

Add power/reset button mask checks for RequestedHostTransition

To fully honor button masking, the RequestedHostTransition interface
now checks the button mask state prior to actually initiating the
requested transition. This will make it so the IPMI and Redfish
commands will not be able to subvert the masked buttons.

Tested: disable front panel buttons (ipmitool raw 0 0x0a 0xf0)
and then attempt to reset and power cycle with ipmi

ipmitool chassis power reset
ipmitool chassis power cycle
ipmitool chassis power off

Host power stays on and uninterrupted.

Change-Id: Iff66788122e26cd466482bf4a4c3ac443ca4c1ee
Signed-off-by: Vernon Mauery <vernon.mauery@linux.intel.com>

show more ...

Add timing configurations into power-config-host0.json

The timings will be changed according to different hardware design, this change can be used flexibly.

TESTED: the existing functionality is no

Add timing configurations into power-config-host0.json

The timings will be changed according to different hardware design, this change can be used flexibly.

TESTED: the existing functionality is not broken after adding the JSON parser implementation.

Signed-off-by: Charles Hsu <Charles.Hsu@quantatw.com>
Change-Id: I1aada7af79fdb9ed3110e94bee31697b25bb6613

show more ...

Use power_control namespace in main()

There are quite a lot of `power_control::` in main(), which is not
really necessary. Use the namespace in main() so the extra
`power_control::` could be removed

Use power_control namespace in main()

There are quite a lot of `power_control::` in main(), which is not
really necessary. Use the namespace in main() so the extra
`power_control::` could be removed and the code becomes more clean.

Tested: No functional change, verify it builds OK.

Signed-off-by: Lei YU <yulei.sh@bytedance.com>
Change-Id: Ia336c36dbbff53ae673ed0bc5fe25bbe63ecad6d

show more ...

Improve graceful power state handling

The current timeout of 1 minute for an OS to respond to a graceful
shutdown, is not enough for some situations. Extending this to
5 minutes to allow enough tim

Improve graceful power state handling

The current timeout of 1 minute for an OS to respond to a graceful
shutdown, is not enough for some situations. Extending this to
5 minutes to allow enough time for an OS to gracefully shut down.

This causes an issue with the current code which ignores further
power change requests during the graceful timeout, which would now
block forced shutdown and reset commands for 5 minutes. So, this
change also adds support for shutdown and reset commands to be
accepted during a graceful shutdown timeout.

Tested:
Triggered a graceful shutdown and confirmed that it will time out
after 5 minutes and that shutdown and reset commands can be issued
during that time.

Change-Id: Ie88207cbc754a34642b4e1bc9636a257475cdee6
Signed-off-by: Jason M. Bills <jason.m.bills@linux.intel.com>

show more ...

Use phosphor::logging instead of std::cerr

After moving part of the logs to phosphor::logging, it caused
the messages to get logged out of order depending on if they
came from phosphor::logging or s

Use phosphor::logging instead of std::cerr

After moving part of the logs to phosphor::logging, it caused
the messages to get logged out of order depending on if they
came from phosphor::logging or std::cerr.

This change moves all logs to phosphor::logging so they all
appear in the correct order again.

Also fixed a minor typo.

Tested:
Performed a power cycle and confirmed that the logs all look
as expected.

Change-Id: I25f677097d2f6552e09536c3078bb08e1d985986
Signed-off-by: Jason M. Bills <jason.m.bills@linux.intel.com>

show more ...

Add option to use PLT_RST for warm reset detection

To support platforms that provide a PLT_RST eSPI signal,
this change adds an option to detect warm resets from the
eSPI PLT_RST signal instead of P

Add option to use PLT_RST for warm reset detection

To support platforms that provide a PLT_RST eSPI signal,
this change adds an option to detect warm resets from the
eSPI PLT_RST signal instead of POST Complete.

The eSPI PLT_RST signal is provided by the
xyz.openbmc_project.Host.Misc.Manager service.

Tested:
Confirmed that warm reset can be detected based on PLT_RST
changes.

Change-Id: I8ac2c260280cf4edec8800e8e2ee12159748f24e
Signed-off-by: Jason M. Bills <jason.m.bills@linux.intel.com>

show more ...

Don't check for warm reset on reset button press

When the reset button is pressed, it checks for warm reset by
waiting 500ms for other signals to indicate a power off. If
power stays on, then the B

Don't check for warm reset on reset button press

When the reset button is pressed, it checks for warm reset by
waiting 500ms for other signals to indicate a power off. If
power stays on, then the BMC returns to the 'on' state.

However, if the POST complete GPIO, which is also triggered by
the reset button, takes longer than 500ms to be detected, it will
trigger another check for warm reset. In addition, it is treated
as a second warm reset and the reset reason from the reset button
gets overwritten.

This changes the reset button to not trigger a warm reset check.
This makes the warm reset check depend on the POST complete GPIO
already being asserted which leaves open a window where if the
reset button is pressed during boot before POST complete, then
the warm reset will not be handled.

Tested: Confirmed that when POST complete de-asserts after the
warm reset check from the reset button press that the reset
reason is correctly set to the reset button.

Change-Id: I837a8b245c25e2a974301f1a58c28c41d63c4be2
Signed-off-by: Jason M. Bills <jason.m.bills@linux.intel.com>

show more ...

Make error-checking in sendPowerControlEvent() more meaningful.

It checks getPowerStateHandler()'s return value against nullptr, but
that had previously been returning a no-op function on error inst

Make error-checking in sendPowerControlEvent() more meaningful.

It checks getPowerStateHandler()'s return value against nullptr, but
that had previously been returning a no-op function on error instead
of null.

Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
Change-Id: I65a33db51769ba579cd7c7f99ee3bac7aab69776

show more ...

Remove James from MAINTAINERS

Will be off the project starting Oct 9th.

Change-Id: I7bd926c6525105d313bc63851416e608c1e3e630
Signed-off-by: James Feist <james.feist@linux.intel.com>

Make the reset-button optional

Make the reset button optional to prevent the program
exit on machines that do not have a reset button

Tested: power-control runs on a machine without a reset

Make the reset-button optional

Make the reset button optional to prevent the program
exit on machines that do not have a reset button

Tested: power-control runs on a machine without a reset
button

Signed-off-by: John Wang <wangzhiqiang.bj@bytedance.com>
Change-Id: Icc82e76ea76e41d33574ef576330a1b5a87ada6a

show more ...

Add SIO Enabled change

Modified the power-control state machine to equip x86-power-conrol to support
for the system which does not have sioPowerGood support.

TESTED: Tested and verified in Tiogapas

Add SIO Enabled change

Modified the power-control state machine to equip x86-power-conrol to support
for the system which does not have sioPowerGood support.

TESTED: Tested and verified in Tiogapass. The existing functionality
is not broken after adding the changes in Tiogapass.

Signed-off-by: Priyatharshan P <priyatharshanp@hcl.com>
Change-Id: I6520c22474124f1dbd903e7522ab2ae5959d4ff5

show more ...

Add compile flag for chassis system reset

Not every platform has complete chassis reset supported, so added
this feature under a compile flag, one can enable this flag to use
this feature.

Tested:

Add compile flag for chassis system reset

Not every platform has complete chassis reset supported, so added
this feature under a compile flag, one can enable this flag to use
this feature.

Tested: Verified this by enabling CHASSIS_SYSTEM_RESET compile
otions via bbappend and checked the system reset interface
/xyz/openbmc_project/state/chassis_system0 created in dbus node

Signed-off-by: Vijay Khemka <vijaykhemka@fb.com>
Change-Id: Ia834f0af309c0cd357f8e988794aa10f9b78248c

show more ...

boost: disable rtti temporarily

There is a bug in the latest version of boost that upstream yocto is
pulling in which causes x86-power-control to not build.

See https://github.com/chriskohlhoff/asi

boost: disable rtti temporarily

There is a bug in the latest version of boost that upstream yocto is
pulling in which causes x86-power-control to not build.

See https://github.com/chriskohlhoff/asio/issues/533 for more
information.

Similar to what OpenBMC did with bmcweb, disable this for now in
x86-power-control.

Resolves openbmc/x86-power-control#7

Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I836a513d86662f06eb146355ba1b0aac92a840b7

show more ...

Fix missing includes

Likey due to
https://github.com/openbmc/sdbusplus/commit/38ab5ecfe3c51501a9380e2c30dbb5fdd88cf4f9

Some Boost features that were missing from being included
made the build stop

Fix missing includes

Likey due to
https://github.com/openbmc/sdbusplus/commit/38ab5ecfe3c51501a9380e2c30dbb5fdd88cf4f9

Some Boost features that were missing from being included
made the build stop working. Fix them.

Change-Id: Ib19702dbb39a007e7badfe87f980b264ffde82f5
Signed-off-by: Jason M. Bills <jason.m.bills@linux.intel.com>

show more ...

Fix config json file path

TESTED : Built image and tested in tiogapass.

Signed-off-by: Kumar Thangavel <thangavel.k@hcl.com>
Change-Id: I941b4c789d39d9f7ee6b00a61d1750a5409c6893

Improve host state monitor

This changes to arg0 in the match to limit to just the State.Host
interface and looks just for the CurrentHostState property to
prevent errors printing for properties we d

Improve host state monitor

This changes to arg0 in the match to limit to just the State.Host
interface and looks just for the CurrentHostState property to
prevent errors printing for properties we don't care about.

It also moves the message.read into a try-catch since it can throw
and changes to get_if instead of get to reduce the number of
possible exceptions.

Tested:
Ran ipmitool power on/off and confirmed that the "Host system DC
power is on/off" message prints correctly without any spurious
error messages.

Change-Id: I7bbe04675db6c8f2a0212529b8ebac9a9879a6b3
Signed-off-by: Jason M. Bills <jason.m.bills@linux.intel.com>

show more ...

Add JSON config support for x86-power-control

Adding a flexibility to configure GPIOs for the power control with a json
configuration file.

TESTED: Tested and verified in Tiogapass. The existing fu

Add JSON config support for x86-power-control

Adding a flexibility to configure GPIOs for the power control with a json
configuration file.

TESTED: Tested and verified in Tiogapass. The existing functionality
is not broken after adding the JSON parser implementation in Tiogapass.

Signed-off-by: Priyatharshan P <priyatharshanp@hcl.com>
Change-Id: I9eb0c845e0924e0bb642d3691fcfaf948b773ce0

show more ...

Remove shut down on VR watchdog timeout

The BMC should not force the system off based on a power good
timeout.

This change removes the shutdown and the "failed transition
to On" state which only ex

Remove shut down on VR watchdog timeout

The BMC should not force the system off based on a power good
timeout.

This change removes the shutdown and the "failed transition
to On" state which only existed to handle the shutdown.
Now on errors during power-on, BMC will log the event and
go to the Off state. The Off state will now also handle
getting a power good event and going straight to On.

Tested:
Forced a power good timeout and confirmed that the BMC and
system both end up in the On state.

Change-Id: If4574a01ae821017604ca8ac0ec54b4d281b4d3b
Signed-off-by: Jason M. Bills <jason.m.bills@linux.intel.com>

show more ...

Use phosphor::logging for the "no action taken" messages

Tested:
Verified that the "no action taken" messages print to the journal.

Change-Id: Ie37228e43232cc1cef133fceed077f589ac7a50f
Signed-off-b

Use phosphor::logging for the "no action taken" messages

Tested:
Verified that the "no action taken" messages print to the journal.

Change-Id: Ie37228e43232cc1cef133fceed077f589ac7a50f
Signed-off-by: Jason M. Bills <jason.m.bills@linux.intel.com>

show more ...

Add host info in log message

Added host info like "Host0" in log message when power state change.
This change is to keep future implementation of multi host support
and currently it only support Hos

Add host info in log message

Added host info like "Host0" in log message when power state change.
This change is to keep future implementation of multi host support
and currently it only support Host0 so this hard code.

Signed-off-by: Vijay Khemka <vijaykhemka@fb.com>
Change-Id: I2d9edc1bf098de954a0ea62dcdfc0858693a1a7a

show more ...