| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
By replacing phosphor-hwmon by dbus-sensor, PSU presence detection via
GPIO does not work anymore and it causes confiction with dbus-sensor's
PSUSensor.
Revert the commit to avoid the confliction
Change-Id: Idfd54325c122921a3c5df062bce1b6909b643999
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
|
|
The chassis instance when implementing sensor monitor using
dbus-sensor/entity-manager is /redfish/v1/Chassis/<name> where <name> is
defined from entity-manager configuration.
This commit removes /redfish/v1/Chassis/chassis which is redundant and
make incorrect link from other schema like ComputerSystem and BMC
Manager.
Tested:
1. Check no /redfish/v1/Chassis/chassis available.
2. Check all sensors are available in /redfish/v1/Chassis/Mt_Jade.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I334a94bb65c3caa49d0bcb6ab03a42f7999728d5
|
|
Disable phosphor-hwmon from compilation and clean up sensor inventory
which added before for phosphor-hwmon.
Tested:
1. Check if no phosphor-hwmon compiled and run
2. Check all S0 and S1 ADC sensors are still available in Redfish.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ic6ac13963e755e6ef0d93b3f6486546c61b1abe0
|
|
Enable phosphor-nvme with configuration for nvme_max sensor for FAN
control support later.
Tested:
1. Check NVME Temp:
$ busctl get-property xyz.openbmc_project.nvme.manager \
/xyz/openbmc_project/sensors/temperature/nvme20 \
xyz.openbmc_project.Sensor.Value Value
d 22
2. Check nvme_max has valid value
$ busctl introspect xyz.openbmc_project.VirtualSensor \
/xyz/openbmc_project/sensors/temperature/nvme_max
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ied609ba5b5d39877c09ed04faff5063c7882a689
|
|
Replace sensor monitoring in phosphor-hwmon by
dbus-sensor/entity-manager.
Tested:
1. Check IPMI sensor report for Temp and ADC sensors with correct
threshold
$ ipmitool sdr list
$ ipmitool sensor list all
2. Check Redfish Thermal and Power schema for Temperature and ADC
sensors
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Idbb719ed01a28892bce32adce062c56dd51d3088
|
|
Add support SCP firmware update with checking if the ampere_fru_upgrade
available to execute the update.
This commit also supports 2nd Host SPI-NOR update ability.
Tested:
1. Manually copy ampere_fru_upgrade to the /usr/sbin folder and run SCP
firmware update via Redfish.
2. Flash UEFI firmware update on the secondary Host SPI-NOR.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ic335ee2d21625239c3df1937ed5333708233845b
|
|
Change to use dynamic sensors for FRU support to take advantage of IPMI
fru read/write commands.
Tested:
1. Check FRU reading
$ ipmitool fru print
2. Write fru.bin into FRU EEPROM
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Signed-off-by: Thang Tran <thuutran@amperecomputing.com>
Change-Id: I0d0b52ae7b0fd301a14d0d2894268637d867a5b8
|
|
Configure OCP_AUX_PWREN and/or OCP_MAIN_PWREN GPIOs, even without
changing their values, make the OCP power glitched which consequently
makes the Host OS generates many calltrace message.
When BMC is rebooted, the ampere-gpio-poweron@service service is
triggered that reconfigure these GPIOs.
This comit prevents this case by checking if the Host is currently ON to
skip the reconfiguration.
Tested:
1. Boot the BMC and Host CPU.
2. Reboot BMC. Ensure no call trace for OCP
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I621e025adf72baf84bffd011ea0e04229f739755
|
|
`BBLAYERS_NON_REMOVABLE` is obsolete and no longer required.
As it said by Yocto documentation it can be used by `Hob`
https://www.yoctoproject.org/docs/1.5.2/ref-manual/ref-manual.html#var-BBLAYERS_NON_REMOVABLE
that already removed since Yocto-2.1
https://www.yoctoproject.org/tools-resources/projects/hob
Change-Id: Ibc2d8268a9d837a81e9cf6b0131dba8d0a030a3f
Signed-off-by: Alexander Filippov <a.filippov@yadro.com>
|
|
Remove Mt.Jade u-boot patches as we have other approach for them:
1. Patch 0001 is to set FANs to maximum speed in u-boot to avoid
hardware damage in case of Host ON with high CPU usage. This can be
resolved by setting FAN to maximum speed when the FAN service stop.
2. Patch 0002 is to enable I2C4 device access. With commit
0b32c1b40@linux, I2C4 enable is set via gpio-hog
3. Patch 0003 will be resolved later by supporting Aspeed watchdog with
boot status.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I247f99ebac7c3ca10a2dc77f1909f5c0c519401d
|
|
Change to implement virtual USB Ethernet using phosphor-misc's usb-ctrl.
Tested:
1. Boot BMC to Linux. Login and check if usb0 ethernet interface exists.
2. Power ON the Host. Check if usb0 automatically sets its IP to
192.168.0.10
3. Restart the ampere-usbnet.service service. Check if no error happens.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I4de722a691a112db851b6dda0a7821b9f78b4d72
|
|
Fix all issues from the shellcheck checked on bash shells under
meta-ampere.
Tested: Verify the following features:
1. Power control (on,off, cycle, graceful shutdown, hard reset).
2. UART switching.
3. UEFI firmware update.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Idabf839b7521ecadb642230cc8bb3472c787002e
|
|
The obmc-system-mgmt feature is currently used in the image to trigger
inclusion of a virtual-provider which provides a number of packages many
systems need. Partially revert the removal of this feature so that
the outcome is:
1. The empty obmc-phosphor-sysd package is still removed.
2. By default the 'obmc-system-mgmt' feature is included, unless
specifically exempted.
3. All EVB platforms remove the 'obmc-system-mgmt' feature since
they have no system they are managing.
This partially reverts commit 060ad3ff7fcc30aff78a9e504efee9d8fa0d4526.
Tested:
* Built `bletchley` and confirmed `packagegroup-fb-apps-system` and
`entity-manager` are present.
```
entity-manager armv7ahf-vfpv4d16 0.1+git0+6bf41588ab-r0
packagegroup-fb-apps-system all 1.0-r1
```
* Built `witherspoon` and confirmed `packagegroup-op-apps-system` and
`pdbg` are present.
```
packagegroup-op-apps-system noarch 1.0
pdbg arm1176jzs 3.3
```
* Ran `bitbake -p` on `evb-ast2600` to confirm the undefined
`virtual-obmc-system-mgmt` is not being included in the image.
Change-Id: I8b7804d5101cc84a2c57473b3f85672bf7767c67
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
|
|
Every machine layer treats 'system-management' as either part of a
package-group or removes the feature. The sample implementation in
meta-phosphor is a do-nothing shell script (and up until recently was a
Python script). There appears to be no useful purpose to this feature
as a stand-alone concept, so remove it.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I20ca1fa8ff3cb01cac2d07d4ded84e0769e4514b
|
|
Modify the machine meta-layer configs to remove the use of the
OBMC_MACHINE_FEATURES indirection and favor the Yocto MACHINE_FEATURES
variable instead.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Ifafb79e4e4c010e9476b9547cd4982f5b645060e
|
|
Manojkiran Eda (1):
Add OWNERS file
Change-Id: I60701ca35fc2027fa31db97029d846f9efacc31c
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
Enable phosphor-aspeed-spi-master-mode.inc and remove the u-boot
patch that used for this feature.
Also remove GPIO setting patches which the setting already implemented
in the platform init package.
Tested: on Mt.Jade platform with OCP plugged
1. Bind/unbind Host SPI-NOR and ensure no problem
2. Turn off Host, reboot BMC and check if Host SPI-NOR is probe
successfully.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ie85166edca93b882868e1373696fbf664bee2ee4
|
|
The ampere_update_mac service gets BMC MAC Address from FruDevice but
currently uses Inventory.Manager as its dependency. This makes the
service sometime not work correctly.
This commit updates the dependency to EntityManager to fix the issue
Tested:
1. Clear eth1addr variable. Do A/C power the BMC and ensure the BMC MAC
is set to eth1addr
2. Clear eth1addr variable. Reboot BMC and ensure the BMC MAC is set to eth1addr
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I00c78222d487c72db5e6e06f8cc8b45cfde2f253
|
|
Ampere BMC specification requires the chassis power policy to always-on
by default. However, it is set by default to always-off in OpenBMC.
This commit updates the setting to always-on by default.
Tested:
1. Do factory reset the BMC.
2. Check the power status
$ root@mtjade:~# ipmitool chassis status
Power Restore Policy : always-on
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I7ec478dc60faa5fe51a091599265ec807bab22e9
|
|
Adds the ampere-ipmi-oem.bb file to support ampere-ipmi-oem features.
Signed-off-by: Hieu Huynh <hieuh@os.amperecomputing.com>
Change-Id: I869a3e73573dd43103b52640ef3ac2d1912280dc
|
|
There are some redundant codes inside meta-mtjade:
- ampere-reset-host-check@.service: no longer be used.
- redundant :mtjade in phosphor-software-manager_%.bbappend
- Incorrect GPIO function call in ampere_platform_init.sh
This commit fixes the above issues.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Iadde88ba6472815556a2c0839c1e3e127c2a04d4
|
|
Several GPIOs should be configured properly when booting BMC. Currently,
these GPIOs are configured in u-boot. However, some must be configured
in Linux. For example, the BMC_READY GPIO should be set only after BMC
is ready to work.
This commit introduce platform-init service which starts after BMC boots
to configure GPIOs pins required by hardware design.
Tested:
1. Do A/C power, check GPIO227 is set to 0.
2. Reboot BMC, check GPIO227 is set to 0.
3. Run gpioinfo and check GPIO directions
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ie0cfe171659ad740c6a25297f2fcce259870d43d
|
|
The BMC_READY pin is used to notify CPLD that BMC is running and it is
ready to control the Host power. This pin should be always high after
the BMC boots successful.
This commit removes BMC_READY setting in the skeleton config so that is
is not set low when powering off the Host.
Tested:
1. Power off the host. Check if BMC_READY is still HIGH.
Signed-off-by: Thu B Nguyen <tbnguyen@amperecomputing.com>
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Idc3a2b7ef92e6f8881d346f29920e184585911c3
|
|
FruDevice service is always running no matter if entity-manager is used
or not. It scans for all FRU EEPROMs, parse and store information into
dbus.
This commit changes to get BMC MAC Address from FruDevice data instead
of ipmi-fru-parser.
Tested:
1. Clear environment in u-boot, boot BMC to Linux and check if the
eth1addr variable is set with value from FRU.
2. Change BMC MAC address from FRU's Board Extra. Reboot BMC and check
if the eth1addr variable is set with new BMC MAC Address
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I99b5d46c8a22b98402f749cdebd1c0aad3d6daf0
|
|
Enable support for smbios-mdr using IPMI BLOB transport protocol,
and disable the cpuinfo feature.
Signed-off-by: Hieu Huynh <hieuh@os.amperecomputing.com>
Change-Id: I0ed7ff78ba3910ab60018e0b8f55b14a22fcbb2a
|
|
Support a wrapper script to detect the Host firmware image by checking
the ExtendedVersion field from the MANIFEST file to flash into
the appropriate device components. The supported image type include:
- ExtendedVersion=secondary: flash to secondary Host SPI-NOR
- ExtendedVersion=scp-primary: flash to primary Boot EEPROM
- ExtendedVersion=scp-secondary: flash to alternative Boot EEPROM
- ExtendedVersion=fru: write to the FRU EEPROM.
Tested:
1. Flash UEFI firmware into the primary Host SPI-NOR via WebUI
2. Flash UEFI firmware into the secondary Host SPI-NOR via WebUI
Signed-off-by: Chanh Nguyen <chanh@os.amperecomputing.com>
Change-Id: I7dff2ca0f8b9de85c2cbac1a49d04b6e14721a03
|
|
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Ia05891430c6c97a89b7bc8ba4558ae496866bf7d
|
|
bitbake offers a choice of DEB, RPM or IPK packaging. To a degree the
choice is functionally arbitrary for image generation but control over
the package format becomes important if we want to:
1. Include runtime package management functionality in the firmware image
2. Mess about with the packages on the build system
With respect to 1 the IPK format and opkg (an ipk package manager) are
designed for embedded systems[1] - by contrast to RPMs have heavier
dependencies and a greater impact on the size and complexity of the
firmware image.
Regarding 2, the embedded nature and the need for opkg to work without
much fuss leads to a lower configuration barrier by comparison to RPMs.
With ipk it becomes possible to reuse the packages built during image
preparation for core analysis without needing to generate an SDK:
```
$ export LD_LIBRARY_PATH=./tmp/work/x86_64-linux/opkg-native/*/recipe-sysroot-native/usr/lib
$ MY_DEBUG_ROOT=tmp/rootfs-debug
$ ./tmp/sysroots-components/x86_64/opkg-native/usr/bin/opkg \
-f ./tmp/work/p10bmc-openbmc-linux-gnueabi/obmc-phosphor-image/*/opkg.conf \
-o $MY_DEBUG_ROOT \
update
$ fakeroot ./tmp/sysroots-components/x86_64/opkg-native/usr/bin/opkg \
-f ./tmp/work/p10bmc-openbmc-linux-gnueabi/obmc-phosphor-image/1.0-r0/opkg.conf \
-o $MY_DEBUG_ROOT \
install dbus-sensors dbus-sensors-dbg
$ gdb-multiarch
(gdb) set solib-absolute-prefix .../tmp/rootfs-debug
(gdb) add-auto-load-safe-path .../tmp/rootfs-debug
(gdb) file tmp/rootfs-debug/usr/bin/nvmesensor
(gdb) core-file obmcdump_17_9597/core.nvmesensor.0.aae91b519d0e4e0e8bbe746e3f6cd25f.2779.9594000000
Core was generated by `/usr/bin/nvmesensor'.
Program terminated with signal SIGABRT, Aborted.
pthread_kill.c:45
45 pthread_kill.c: No such file or directory.
(gdb) bt
pthread_kill.c:45
../sysdeps/posix/raise.c:26
/home/andrew/src/openbmc/openbmc/build/p10bmc/tmp/rootfs-debug/usr/lib/libstdc++.so.6
/home/andrew/src/openbmc/openbmc/build/p10bmc/tmp/rootfs-debug/usr/lib/libstdc++.so.6
/home/andrew/src/openbmc/openbmc/build/p10bmc/tmp/rootfs-debug/usr/lib/libstdc++.so.6
/home/andrew/src/openbmc/openbmc/build/p10bmc/tmp/rootfs-debug/usr/lib/libstdc++.so.6
"xyz.openbmc_project.NVMeSensor", this=0x488f04) at
/usr/include/sdbusplus/bus.hpp:234
../../../../../../workspace/sources/dbus-sensors/src/NVMeSensorMain.cpp:159
(gdb)
```
This approach documented in the Poky Reference Manual:
https://www.yoctoproject.org/docs/1.0/poky-ref-manual/poky-ref-manual.html#platdev-gdb-remotedebug-launch-gdb-inferiorbins
Switch all machines to IPK to align the debugging experience with
upstream's documentation and to facilitate efficient use of packaged
software at runtime.
[1] https://openwrt.org/docs/guide-user/additional-software/opkg
Change-Id: I8ef526add2d7a6790de1b3eb3fb85cd39b864f23
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
|
|
Disable this option to does not support TFTP firmware update as it is
an unsecured protocol.
Tested:
1. On Webui, Configuration -> Firmware. No more option TFTP server on
Update firmware.
Signed-off-by: Hieu Huynh <hieuh@os.amperecomputing.com>
Change-Id: I8cb3285154b62ac2ee026485ad3f1d9d9586a016
|
|
The patch to support flashcp offset parameter is fail to apply because
of new mtd-utils revision. This commit updates the patch to resolve the
compilation warning on apply failure.
Tested:
1. Compile Mt.Jade OpenBMC without warning about mtd-utils.
2. Flash EDKII firmware using ampere_flash_bios.sh and check Host boots
successfully.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I9026ac5bafc07ee5eff7efdf7890d0eef2161c52
|
|
Upstream bbclasses changed to typically use the `:${PN}` override
syntax, including the SYSTEMD_ variables. Change our systemd.bbclass
to do the same for consistency and perform a tree-wide variable
replacement.
Spot checked by building bletchley and witherspoon and checking
some of the SYSTEMD_LINK directives on installed packages under qemu.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I20a9dd809bff8af8759488734f80486c7228c6eb
|
|
Enable rsyslog to save IPMI SEL event into /var/log/ipmi_sel.
Tested:
1. Log IPMI SEL event using xyz.openbmc_project.Logging.IPMI
2. Check /var/log/ipmi_sel for the new entry
root@mtjade:~# cat /var/log/ipmi_sel
2021-09-01T09:23:49.022943+00:00 1,c0,3ACD000FEB00000000000000,,,
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ic02a6166498251ed1e84a56e4847f465a069c872
|
|
Add DCMI sensor configuration for the Mt.Jade platform.
Tested:
1. root@mtjade:~# ipmitool dcmi sensors
Inlet: 2 temperature sensors found:
Record ID 0x00f3: PSU0_TEMP | 26 degrees C | ok
Record ID 0x00f4: PSU1_TEMP | 31 degrees C | ok
Baseboard: 6 temperature sensors found:
Record ID 0x0001: TS1_Temp | 31 degrees C | ok
Record ID 0x0002: TS2_Temp | 32 degrees C | ok
Record ID 0x0003: TS3_Temp | 31 degrees C | ok
Record ID 0x0004: TS4_Temp | 30 degrees C | ok
Record ID 0x0005: TS5_Temp | 35 degrees C | ok
Record ID 0x0006: TS6_Temp | 32 degrees C | ok
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Iff8a94e523346347b7e449522ab5a12b841aee9e
|
|
Since 29086950, phosphor-net-ipmid has updated the SOL parameters
by obtaining the attribute values of the
xyz.openbmc_project.Ipmi.SOL interface.
It is necessary to enable SOL by default on Mt.Jade with correct
parameters.
Tested:
1. ipmitool -I lanplus -H x -U x -P x -C 17 raw sol info
2. ipmitool -I lanplus -H x -U x -P x -C 17 sol activate
ipmitool -I lanplus -H x -U x -P x -C 17 sol deactivate
Use the above commands to successfully activate and deactivate SOL
3. Activate IPMI SOL and boot Host to Linux 5 times without
disconnected.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I8e4d0e06eaa9c8995cde27f17f0d0d274b553f0d
|
|
Fix missing Yocto syntax change in phosphor-hwmon append file which make
sensor configured not work correctly.
Tested:
1. Run "ipmitool sdr list" and check all sensors reported correctly
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ia4653180ac1a66dac13ee6d7edf5c34a106967e9
|
|
We've typically kept these LAYERSERIES_COMPAT to 2 releases: the current
and the upcoming. Remove 'gatesgarth' is it is now 2 releases back.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I5e812a94fed1738898af75c0fdee81996a5bbf20
|
|
Make the following changes to fix compilation errors:
1. Add honister to LAYERSERIES_COMPAT
2. Increase CONF_VERSION to 2.
3. Fix mistake on previous change on Yocto syntax update.
Tested:
1. Compile OpenBMC for Mt.Jade without error.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I5ac3f568d776109955759fab78844b1d38a51af1
|
|
Run convert-overrides.py meta-ampere and update missing changes as
below:
- Change all _mtjade to :mtjade
- Change all _${PN} to :${PN}
Tested:
1. Check compilation without problem.
2. Check basic functions like switch UART mux, power control.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I42038b2a98ed1951d6c770bdf49f5fc440f8163e
|
|
Using IMAGE_INSTALL in mtjade.conf makes other Ampere recipes not
included in the build. This commit fixes the issue by enabling the
mtd-utils in OBMC_IMAGE_EXTRA_INSTALL_append instead.
Tested:
1. Check if all Ampere services and scripts available.
2. Check power control operations
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I17293ba520c05a7b5a438767c790f00220156d3f
|
|
The OCP module in Mt.Jade platform supports the GPIO pins to set the
power state. The SPI0 module also supports the GPIO pins to config SPI
mode. These gpio pins have to be set to the right states before power
off/on. The mtjade-gpio-config package will support that.
Tested:
1. Power off/on the host.
2. The OCP and SPI0 gpio pins should be in the corrected states.
Signed-off-by: ThuBaNguyen <thu@os.amperecomputing.com>
Change-Id: I46bca50222b0428f73aa9c05003f650f6ec0cc0b
|
|
The power functions are failed to work. There is one mistake in
merging ampere_power_utils.sh script. This commit fixes the mistake.
Tested:
1. Power off the host when it is on.
2. The host state should be changed to off.
Signed-off-by: ThuBaNguyen <thu@os.amperecomputing.com>
Change-Id: Iaa253e4b8d0f645a2b7097501e8c602d693cb37a
|
|
When BMC boots, phosphor-host-check uses the software interfaces(IPMI
inband Hearbeat command or PLDM interface) to verify the host state
and creates the host on indicator file. Ampere Altra SoCs do not
support those software interfaces. They use the GPIO interface. GPIO
pin S0_FW_BOOT_OK go high when the host is on. This commit appends
ampere-reset-host-check@.service to phosphor-state-manager This
service will replaces the role of phosphor-reset-host-check@.service.
Tested:
1. Boot up BMC with the host is off/on.
2. Verify the CurrentHostState.
Signed-off-by: ThuBaNguyen <thu@os.amperecomputing.com>
Change-Id: Ie4f73538266fff7d8c5fd40b437ffebc82cb12c5
|
|
When phosphor-state-manager handles the power action to turn on the
host, phosphor-state-manager does not use the software interfaces to
verify the host state. It set the host state to running after all of
required services of the power action target are done. The power on
watchdog is used to verify the host state and handle the failure in
the host state changing. But Mt.Jade platforms do not handle power on
watchdog. This causes the CurrentHostState dbus property is Running
before the host is ready.
In Mt.Jade platform, the SCP drives S0_FW_BOOT_OK to high when the
host is already on. This GPIO can be used to identify the host state.
One service should be added to the required list. This service checks
the host state in power on and triggers obmc-host-quiesce@%i.target
when the host is failed to boot. ampere-host-check package adds
ampere-host-on-host-check@.service for that purpose.
Tested:
1. Boot up BMC with the host is off/on. Check CurrentHostState.
2. Call Ipmitool power off/on. Check the CurrentHostState.
3. Call Ipmitool chassis power off/on. Check the CurrentHostState.
4. Call Ipmitool power cycle. Check the CurrentHostState.
5. Call Ipmitool chassis power cycle. Check the CurrentHostState.
6. Power off the host. Set power policty to Always-on. AC power.
The host should be on. And CurrentHostState should be Running.
7. Power on the host. Set power policty to Always-off. AC power.
The host should be off. And CurrentHostState should be Off.
Change-Id: I9b0b346883a5f0203bf078453065c18ae4603311
Signed-off-by: ThuBaNguyen <thu@os.amperecomputing.com>
|
|
ampere-hostctrl package includes the generic codes and the platform
specific codes. The generic sources are applied for all of the Ampere
platforms, they are the service files and the scripts to setting the
GPIO pins. The platform specific sources are the GPIO pin config file
and the gpio monitor config files.
This commit moves the platform specific files to the platform code.
Tested:
1. Do the power control actions use ipmitool, BMC web and Redfish.
2. Make sure the server states are changed as expected.
Signed-off-by: ThuBaNguyen <thu@os.amperecomputing.com>
Change-Id: Ic0e261109f7a5dfaf38c883bff0afedab763a0b6
|
|
Refactor the Mt.Jade Host power control operations to support Altra
CPU following the Altra SoC BMC Interface specification which still
satisfying the OpenBMC power control framework.
Tested: check the below cases and ensure Host state report correctly:
1. Run "reboot" from Host Linux. Check the Host is graceful
shutdown, then SYS_RESET is asserted and the Host is booted.
2. Run "shutdown" from Host Linux. Check the Host is graceful
shutdown then the chassis power is turned OFF.
3. Run "ipmitool chassis power soft". Check the Host is graceful
shutdown
4. Execute Redfish GracefulShutdown. Check the Host is graceful
shutdown then the chassis power is turned OFF.
5. Execute Redfish GracefulRestart. Check the Host is graceful
shutdown then turned ON.
6. Check other IPMI chassis power and Redfish Reset actions work.
Signed-off-by: ThuBaNguyen <thu@os.amperecomputing.com>
Change-Id: I3a9716f37d7e80bdc1a1d0422e561c8fc51ad040
|
|
UEFI firmware for Ampere Altra-based platform starts at offset 0x400000
but currently the firmware image is flashed start at offset 0. This
makes the Host fail to boot.
This commit supports offset option for the mtd-util's flashcp utility
and update the script to flash UEFI firmware starting at offset
0x400000.
Tested:
1. Turn on the Host.
2. Flash UEFI firmware. Wait for the flash complete and check if Host
can boot successfully.
4. Turn off the Host.
5. Flash UEFI firmware. Wait for the flash complete.
6. Turn on the Host and check if Host can boot successfully.
Signed-off-by: Tung Nguyen <tung.nguyen@amperecomputing.com>
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I0d48bfd0a7b7350b46a98bf10fe4de3887f09f02
|
|
The current ampere-hostctl package removes the power off/on services
of phosphor-state-manager and uses Ampere services. This solution is
not correct. Ampere platform should use power control functions of
phosphor-state-manager. And only appends or overides the default
services by Ampere's services if need.
By default, to handle power soft action, phosphor-state-manager will
trigger obmc-host-shutdown@0.target. This target then call
xyz.openbmc_project.Ipmi.Internal.SoftPowerOff.service to request OS
shutdown. When the host OS shutdown is already done, the target will
trigger obmc-chassis-poweroff@0.target to turn off the chassis. The
default xyz.openbmc_project.Ipmi.Internal.SoftPowerOff.service uses
Ipmi inband to communicate with the host. It is different with Ampere
Mt.Jade platform. We use the GPIO pin SHD_REQ to request shutdown the
host OS. The host will trigger SHD_ACK when shutdown is done. So
*.Ipmi.Internal.SoftPowerOff.service will be overide by Ampere service
named ampere.*.Ipmi.Internal.SoftPowerOff.service. This service will
trigger SHD_REQ pin and wait for SHD_ACK before start
obmc-chassis-poweroff@0.target.
This commit removes ampere-chassis-poweroff, ampere-host-shutdown
serives which handle power off/soft actions and restore
phosphor-state-manager's services. It also supports
ampere.xyz.openbmc_project.Ipmi.Internal.SoftPowerOff.service which
will overide xyz.openbmc_project.Ipmi.Internal.SoftPowerOff.service.
Tested:
1. Create "power soft" actions use ipmitool, redfish and BMC web.
2. Make sure the power order will be shutdown the host then power
of the chassis.
3. Create "power off" actions use ipmitool, redfish and BMC web.
4. Make sure the power action will be power off the chassis.
Signed-off-by: ThuBaNguyen <thu@os.amperecomputing.com>
Change-Id: Ibc0dc8c62408e8282520c9b70e41ab75c10137f6
|
|
Ampere Mt.Jade power control involves different action during the
course of action which does not involve rebooting the Host. Thus
disable the use of Host Power-On watchdog.
Tested:
1. Power on host.
2. In bmc console call "journalctl -f".
3. "ipmitool power reset" make sure there are no
phosphor-watchdog messages.
Signed-off-by: ThuBaNguyen <thu@os.amperecomputing.com>
Change-Id: Iaedd120f203b4222bf57e5fbdc7be84cd22d2e2e
|
|
This makes it possible for us to change out build specific settings.
Change-Id: I862366d8efa97c4c3181b842662bf28dd5bec270
Signed-off-by: William A. Kennington III <wak@google.com>
|
|
The SpeedMbds of ethernet interface is implemented in getInterfaceInfo()
from the phosphor-networkd repo. It is disabled by NIC_SUPPORTS_ETHTOOL
conditional compilation. The default compile mode is to omit the
getInterfaceInfo() which causes the SpeedMbps not set.
This commit supports enable the NIC_SUPPORTS_ETHTOOL to report correct
speed.
Tested:
1.Get SpeedMbps via rest:
$ curl -XGET --user root:0penBmc -H "Content-Type: application/json" \
-H "If-match: *" --insecure \
https://10.38.14.2/redfish/v1/Managers/bmc/EthernetInterfaces/eth1 \
grep | "SpeedMbps"
2.Get speed of eth1 interface on BMC console
$ cat /sys/class/net/eth1/speed
The value getting from openbmc#1 and openbmc#2 match together.
Signed-off-by: Hieu Huynh <hieuh@os.amperecomputing.com>
Change-Id: I57047151fc4eea08c857b95c0770d89ba95f9bc7
|
