Age | Commit message (Collapse) | Author | Files | Lines |
|
As of 9a8f30ec5b58 ("obmc-console: Introduce console-id, deprecate
socket-id") in obmc-console `console-id` is preferred over `socket-id`.
Convert all in-tree configurations to `console-id` to enable removal of
support for the deprecated `socket-id` configuration option.
Change-Id: Ifd50603c378cfed5eb793b829b8296b7964e4194
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
Fix compilation error after subtree update:
ERROR: obmc-phosphor-image-1.0-r0 do_rootfs: Postinstall scriptlets of
['ampere-sysfw-hang-handler'] have failed. If the intention is to
defer them to first boot, then please place them into
pkg_postinst_ontarget:${PN} ().
Deferring to first boot via 'exit 1' is no longer supported.
Tested:
1. Compile Mt.Jade and Mt.Mitchell without error
2. Check ampere-boot-progress and ampere-sysfw-hang-handler work.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I5d692cb4078df05ac20ace34429463ea4e9c18ca
|
|
Commit phosphor-state-manager@e960184 corrects the code logic that
relates to only-run-apr-on-power-loss option build. Users have to set
only-run-apr-on-power-loss=true to enable checking AC loss while
recovering Host.
This commit enables this option.
Tested:
1. Set chassis policy to always-on.
$ipmitool chassis policy always-on
2. Power off the HOST
$ipmitool power off
3. AC power.
4. Host is turned on after BMC has been rebooted
5. Set chassis policy to always-off.
$ipmitool chassis policy always-off
6. Power on the HOST
$ipmitool power on
7. AC power.
8. Host is turned off after BMC has been rebooted
9. Set chassis policy to previous
$ipmitool chassis policy previous
10. Power off the HOST
$ipmitool power ff
11. AC power.
12. Host is turned off after BMC has been rebooted
13. Power on the HOST
$ipmitool power on
14. AC power.
15. Host is turned on after BMC has been rebooted
Change-Id: Ic7bdcdfebc0df415232fc2677e464cca3f341366
Signed-off-by: Thang Tran <thuutran@amperecomputing.com>
|
|
Both Mt.Jade and Mt.Mitchell platforms use yaml to declare IPMI
configuration, not dynamically calculated via entity-manager's minValue
and maxValue. So, having dynamic-sensors enabled by default is not
correct.
Instead of enabling dynamic-sensors for ipmitool fru print to use
information from EM, should add more option for this feature instead.
Tested:
1. Verify if ipmitool sdr list can report sensor information correctly
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I3d7d4a692e059367e9edea38a6230b4a2112d602
|
|
This reverts commit ec7cab9378f548e94ee4af83946b2583d1fc7194.
ec7cab9378f5 ("Add socket-id for the first console") was only a partial
fixup of the obmc-console configurations found in the tree. For
historical reasons not all platforms that support multiple consoles
install a client configuration file for all consoles. Instead they relied
on some default behaviour that was removed in obmc-console's 4e7186918599
("Fixed broken dbus interface for multiple consoles"). Applying the
configuration changes in the manner of ec7cab9378f5 ("Add socket-id for
the first console") resulted in an asymmetric configuration between the
client and server, breaking some SSH SOL instances.
As of ae2460d0b8e8 ("obmc-console: Provide a default value for
`console-id`.") in obmc-console the requirement to specify
`socket-id`[^1] is lifted. Instead, the configuration can choose to
override a default value. This restores the original behaviour and
unbreaks SSH SOL.
[^1]: Now an alias of the `console-id` configuration key
Websocket and IPMI consoles remain broken. A fix is under development
that also takes a proper approach to supporting multiple consoles in
bmcweb, and the same solution pattern can be applied in ipmid.
More discussion of the problems involved can be found here:
https://amboar.github.io/notes/2023/05/08/happenings-in-obmc-console.html
Change-Id: I274284c791758f336da6d6301dc523bac2b5dd69
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
The same channel_config.json and power_reading.json are used across
Ampere platforms so move to meta-ampere/meta-common to avoid duplicate.
Tested:
1. Check if "ipmitool lan print 1/2/3" works.
2. On Mt.Jade, check if "ipmitool dcmi power reading" works
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I77f401d0b74455c3951fb746f8bea2ddb1c3667b
|
|
obmc-console configurations are missing some service appending, this
commit adds these.
Remove socket connection check in ampere_uartmux_ctrl.sh for meta-jade
because no socket file is used now in obmc-console.
Tested:
$ ampere_uartmux_ctrl <uart_port> <mux>
Can successfully switch between console header and BMC for host uart
ports
$ ssh root@<BMC_IP> -p <client_port>
Can successfully connect to host console uarts
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: Iaa0c4d8f9045fc94f6f50822516aaac0b99f088e
|
|
Fix below warning when starting the ampere-boot-progress.service
service:
[ 12.365381] systemd[1]: /lib/systemd/system/ampere-boot-progress.service:7:
Failed to parse service restart specifier, ignoring: simple
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I0b7aa013b27e23f995745f7a5bebbaa39024eea2
|
|
AmpereOne POR recommends to use power cycle instead of SYS_RESET for
reset actions.
This commit changes REBOOT_ACK handling and warm reboot to use power
cycles.
Tested:
1. "ipmitool power reset" when the host is on power action should
similar "ipmitool power cycle".
2. "ipmitool power reset" when the host is off, power action should be
bypassed.
3. In host os call "reboot", the power action should be power off then
power on (without softoff action before power off).
4. Call "GracefulRestart" and "ForceRestart" from redfish when the
host is on the power action should be similar "power cycle"
5. Call "GracefulRestart" and "ForceRestart" from redfish when the
host is off, power action should be bypassed.
6. Press "reset" button in the chassis will cause power off then power
on the host.
Signed-off-by: Thu Nguyen <thu@os.amperecomputing.com>
Change-Id: I759623b68018b05d758b181936e0d4624eb03d94
|
|
Support to detect GPIO, PSU, FAN, ... faults and turn ON/OFF fault LED.
Tested:
1. Unplug a PSU and check if Fault LED is turned ON.
2. Unplug a FAN and check if Fault LED is turned ON.
3. Stimulate GPIO fault pattern and check if the BMC can detect
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Signed-off-by: Hieu Huynh <hieuh@os.amperecomputing.com>
Signed-off-by: Quang Nguyen <quangn@amperecomputing.com>
Change-Id: Idfcd32953cf811fbe9299a162f604cb8fd028962
|
|
This updates to libpam 1.5.2. This version removes support for
pam_cracklib and pam_tally2. They are replaced by pam_pwquality and
pam_faillock respectively.
Since parameters of pam_cracklb and pam_tally2 are configurable through
Redfish, it's possible that they will remain in the overlay of
/etc/pam.d with the old module names preventing PAM from working
correctly. To avoid this, this commit includes a script that will detect
if the old modules are in the overlay and update the overlay with the
new modules and configuration.
The script will allow updates from libpam 1.3.1 to libpam 1.5.2, but if there
are configured parameters during a downgrade from libpam 1.5.2 to libpam
1.3.1, it will require a factory reset before the downgrade.
pam_pwquality was selected over pam_passwdqc because of better security
and compatibility with pam_cracklib.
Note pam_faillock is necessarily configured into the pam module stack
differently than pam_tally2.
This patchset causes a BMC operational change:
- The pam_tally2 command (invoked from the BMC's command line) is no
longer present. If you used the "pam_tally2 -u USER -r" command
to unlock a user after repeated authentication failures, change to
use: faillock --user USER --reset
Compatibility note / migration issue. If your BMC cannot authenticate
users after installing this change, the cause might be an overlayfs file
hiding the new /etc/pam.d/common-auth file. To find out, use
`grep deny= /etc/pam.d/common-auth` on your BMC. If it shows "tally2"
then your BMC is affected. The recovery is to delete the overlay file,
to factory reset the BMC, or manually-install the changed files.
The convert-pam-configs service is intended to handle this problem.
Tested: as follows, for local users only (not tested with LDAP)
Note OpenBMC configuration defaults to an AccountLockoutThreshold
value of 0 which does not lock account passwords no matter how many
consecutive failed authentication attempts. To configure this on
the BMC, for example, use:
curl -X PATCH https://${bmc}/redfish/v1/AccountService
-d '{"AccountLockoutThreshold": 3, "AccountLockoutDuration": 60}'
Tested update scenarios:
1. Install from scratch. Success.
2. Install over firmware which had old PAM configs. Success.
Tested update scenarios for the convert-pam-configs service.
Tested changing the password via various interfaces:
- the passwd command
- the PATCH Refish AccountService {Password: NEW}
- SSH (accessible only when the password is expired)
- IPMI user set password (accessible for unexpired password)
Tested both good and bad (unacceptable) passwords.
Tested account lockout after N bad passwords
Tested unlock via Redfish.
Also, because its implementation changed, ensure reading and writing the
D-Bus User AccountPolicy RememberOldPasswordTimes property continues to
work. There is no Redfish API for this.
Signed-off-by: Joseph Reynolds <joseph-reynolds@charter.net>
Signed-off-by: Jason M. Bills <jason.m.bills@linux.intel.com>
Change-Id: I7b712cf7cfbf7b0bc79da42f822540baee66ca4f
|
|
This drop adds socket-id for first console. This was left out before
but with the new design this parameter is required and enforced by
obmc-console code. It is decided to set "socket-id = console0"
initially because that is the current path used by bmcweb but in future
you can change this value to whatever is appropriate.
The console information is now available through the DBUS interface.
Each console server register it's object path on DBUS. The leaf node of
the object path is used to create a console path. For example in the
following console0 the leaf node "console0" is used to make up the
console path and the GUI will use console path "/console/console0" to
connect to the console.
busctl tree xyz.openbmc_project.Console.console0
`-/xyz
`-/xyz/openbmc_project
`-/xyz/openbmc_project/console
`-/xyz/openbmc_project/console/console0
Here is a discord thread where new design was discussed.
https://discord.com/channels/775381525260664832/1083551792094249051
Tested:
Tested on the rainier system.
Related commits:
1) phosphor-dbus-interface: https://gerrit.openbmc.org/c/openbmc/phosphor-dbus-interfaces/+/61486
2) obmc-console: https://gerrit.openbmc.org/c/openbmc/obmc-console/+/62496
3) bmcweb: https://gerrit.openbmc.org/c/openbmc/bmcweb/+/62525
Change-Id: I5fd06f3aac69872a3e5032d592a9b59711795cfe
Signed-off-by: Ninad Palsule <ninadpalsule@us.ibm.com>
|
|
Upstream obmc-console now supports the meson build system. I consider
the autotools build system deprecated and intend to remove it.
Change-Id: Ic8eea195bee0b3bea1cb9418593596ce1aaceeae
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
usb0 is a virtual USB Ethernet which is configured to static IP
192.168.0.10 in /etc/systemd/network/00-bmc-usb0.network without DHCP
parameter with expecting that DHCP=no.
However, IPMI and Redfish report IP source as DHCP, not static.
This commit explicitly specifies in network configuration as DHCP=no so
that IPMI and Redfish report correctly.
Tested:
1. Check address source for usb0 via IPMI
# ipmitool lan print 3 | grep "IP Address Source"
IP Address Source : Static Address
2. Check address source for usb0 via Redfish
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Id14bfc289da6d36552bd37448ec1e4c30ff32a0a
|
|
This commit fixes shelllcheck source in ampere_uartmux_ctrl.sh
to point to the exact source path.
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: I38d92582bb7783b18ff19fb8ae5e5abe0c1e3d44
|
|
This commit refactors obmc-console setup for mtjade
Tested:
1. Successfully connects to host consoles via ssh sol
$ ssh root@${bmc} -p 220*
2. Successfully connects host consoles via ipmi sol
$ ipmitool -I lanplus -H ${bmc} -U root -P 0penBmc \
-C 17 sol activate
3. All the host console logs are in /var/log/
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: Ie01361139644f5836646d1d98934cda52828ec70
|
|
This commit refactors obmc-console setup for mtmitchell
Tested:
1. Successfully connects to host consoles via ssh sol
$ ssh root@${bmc} -p 220*
2. Successfully connects host consoles via ipmi sol
$ ipmitool -I lanplus -H ${bmc} -U root -P 0penBmc \
-C 17 sol activate
3. All the host console logs are in /var/log/
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: Ia1e6faaa4757e8f478352fd03ed977beeac15200
|
|
Many scripts in meta-ampere had `shellcheck source=/dev/null`, which
isn't as useful as putting the real sourced script. Fix up all of
these in the meta-layer to point to the right script, which as a
side-effect allow a few global-disables to be removed.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Icb43dc76a19edbdada8b4caa6ced57ff07f42d27
|
|
Issue: When users execute power cycle via out-of-band IPMI command, they
can not execute other out-of-band IPMI commands. The issue is reproduced
as below.
Step 1:
ipmitool -H <BMCIP> -U root -P 0penBmc -C 17 -I lanplus chassis \
power cycle
> Chassis Power Control: Cycle
Step 2:
ipmitool -H <BMCIP> -U root -P 0penBmc -C 17 -I lanplus chassis status
> Error: no response from RAKP 1 message
> Error: Received an Unexpected RAKP 2 message
> Error: no response from RAKP 1 message
> Error: Received an Unexpected RAKP 2 message
> Error: no response from RAKP 1 message
> Error: Received an Unexpected RAKP 2 message
> Error: no response from RAKP 1 message
> Get HPM.x Capabilities request failed, compcode = d4
> Get Device ID command failed
> No valid response received
> Unable to get Chassis Power Status
Root cause: When users execute out-of-band IPMI commands, OpenBMC has to
execute PAM authentication. BMC is using SHA512 algorithm to calculate
hashed value with default number of rounds (5000 times). When users
request "power cycle", many tasks have to be executed, therefore, CPU
does not have enough resource to complete the calculation SHA512 for
other requests on time.
Solution: Update the salted of default password (0penBmc) to specific
number of rounds. The number of rounds is the minimum value of SHA512
algorithmi (1000 times) to reduce SHA512 calculation time. The value is
generated from command 'openssl passwd -6 -salt rounds=1000\$UGMqyqdG
0penBmc'.
Tested:
1. Request power cycle
"ipmitool -H <BMCIP> -U root -P 0penBmc -C 17 -I
lanplus chassis power cycle"
> Chassis Power Control: Cycle
2. Request read power status immediately
"ipmitool -H <BMCIP> -U root -P 0penBmc -C 17 -I
lanplus chassis status"
> Chassis Power is on
Change-Id: I260f52413d9ad0ae4ef73fb8466c4460c0365ea6
Signed-off-by: Thang Tran <thuutran@amperecomputing.com>
|
|
Instead of overwriting ssifbridge.service to set bmc-ok, use systemd
drop-in units instead.
Tested:
1. Stop ssifbridge.service service, check if bmc-ok is 0.
2. Start ssifbridge.service service, check if bmc-ok is 1.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I36e065ed84c4c8aa154996d9a31cfc21d0ab5657
|
|
This commit is to log DIMM training fail information in case DDR
initialized is fail.
Tested: Inject DDR error and verify with Redfish and/or sel log:
1. IPMI chassis power off
2. tweak nvparam to inject DDR error
3. IPMI chassis power on
4. Read sel log and/or Redfish to confirm correct DDR event is
logged
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I2a87c9b775faf17b8112d08eff4df44286a444bc
|
|
When BMC is rebooted, because PSON_L has pull up to P3V3_STB, it changes its
value to HIGH which causes the Host is turned OFF.
Add code to check P3V3_STB and recover PSON_L to correct state before setting
BMC_RDY.
Tested:
1. Turn ON the Host
2. Run BMC reboot stress test and check if the Host is still ON
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ic1e3a403d2833b3549fa4f08dcce21a27d8e2308
|
|
New docker container is flagging some issues in this shell script:
```
Command appears to be unreachable. Check usage (or ignore if invoked indirectly)
```
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I1a64d3d6663812bfde2404fded05cfcae7b81a1c
|
|
We are suppose to use direct executable calls in service files
rather than indirections through /bin/env (See Anti-patterns).
Tested:
- Check related services to work like before
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Idb494ce781fa1bdcac176cd89612e6016e6e04d9
|
|
Handle HIGHTEMP assert/deassert GPIO events for both socket 0 and
socket 1.
Tested:
1. Stop fan monitor service
2. Stress CPU HOST
stress-ng.sh 600 5 max 100
4. Wait for S0_SOC_Pkg_Temp exceeds 105 degree and check for event.
5. Turn OFF CPU stress and wait for S0_SOC_Pkg_Temp goes below 105C.
Check for deassert event
Signed-off-by: Chanh Nguyen <chanh@os.amperecomputing.com>
Change-Id: I7f2528b50c9a04e7c5ccbf56e9d2e04555351bba
|
|
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I8ea79d284356629f5b7b24ad3a99336e9e2e297e
|
|
Change to handle HIGHTEMP event via multi-gpio-monitor application,
replacing ampere-gpio-handling.
Tested:
1. Stop fan monitor service
2. Stress CPU HOST
stress-ng.sh 600 5 max 100
3. Wait SOC Temp goes over 105 degree.
5. Check for HIGHTEMP event log.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ia6b34c7d1d7b458e6db60ca79a7a091e19ecaf06
|
|
This commit enables package configurations delay-host-control=20
and monitor-use-host-state, in order to delay and rescan before
turning host off when fan sensors are not detected after power state
changes from Off to On; and use CurrentHostState to decide host state
for phosphor-fan-monitor instead of pgood state.
Tested:
1. Host is ON, reboot BMC
=> Host is not turned off unexpectedly during BMC boot up
2. Power cycle host
=> fan sensors are not marked as non-functional
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: Iea6ed195a42e41bee23066e670b67c19bcfda722
|
|
Update Mt.Jade IPMI sensor configuration via the latest SDR Definition.
Tested:
1. Run 'ipmitool sensor list' and check if reported sensor information
is correct.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Id41c1a40070edf5c09f843a5d9b93e0b9d4f35d2
|
|
Ampere account lockout policy is defined as:
'Five consecutive failed login attempts will lock the account for 30 minutes".
While the login attempt was set to 5 already, the 30-minute lock period
(auto-unlock) has not been set.
This commit adds the missing 30-minute lockout by default.
Tested:
1. Enter WebUI -> User management
2. Click on Account policy settings and check if:
- Max failed login attempts: 5
- User unlock method: Automatic after timeout
- Timeout duration (seconds): 1800
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ieed9a719c424a289687e66314f4c68de5d108b6e
|
|
Remove power-button and reset-button from platform init to avoid GPIO
access conflict with phosphor-buttons.
Tested:
1. Check power button works.
2. Check reset button works
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I940e62a9c32a7abe232e02843e1b7b6078481b46
|
|
Enable PostCode from the phosphor-host-postd to support OEM Boot
Progress.
Tested:
1. Get Boot.Raw tree:
root@mtmitchell:~# busctl tree xyz.openbmc_project.State.Boot.Raw
`-/xyz
`-/xyz/openbmc_project
`-/xyz/openbmc_project/state
`-/xyz/openbmc_project/state/boot
`-/xyz/openbmc_project/state/boot/raw0
Signed-off-by: Hieu Huynh <hieuh@os.amperecomputing.com>
Change-Id: Idfd28d2d6512d7aa63339d5567628797538930b3
|
|
When the IP Address is configured to static IP, changing BMC MAC Address
in FRU does not change the MAC address in eth0. This commit fixes the
issue by adding IPMI MAC address update and some other configuration via
ipmitool lan set commands to ampere_update_mac.sh to ensure the update
is taking effect.
Signed-off-by: HuyLe <hule@amperecomputing.com>
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: Ib0466ee130af6d294fed03587c4d96328221d017
|
|
OpenBMC uses os-release.bbappend to control the FW version, which is
the output of 'git describe --dirty --long'. The format of this command's
output is: {annotated-tag-name}-{num-commits-to-tag}-{current-commit-hash}.
'-dirty' will be appended to the output if the working space is not clean.
In addition, git describe --dirty --long only find an annotated tag, not
lightweight tag which will miss internal release.
This commit updates the rule to extract the right revision from Ampere
build and release tag.
Tested: tag with v1.02.100-ampere and check WebUI display revision
1.02.100.
Signed-off-by: Thu Cao <thcao@os.amperecomputing.com>
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: I6d5e6d2dcdb4988809f71d48a5b0e2710e7810d2
|
|
Mapper ships with systemd unit and dbus configuration files now, so
remove unnecessary bitbake classes.
Mapper is dbus-activated, so obmc-mapper.target dependencies are
redundant and no longer necessary.
obmc-mapper.target is the last target in obmc-targets, so remove that
recipe and any dependencies on it.
Change-Id: I59ff8dcd480b4449d0ef59422ba80c12848dade9
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
The default login attempt threshold is currently 0. This patch changes
the value to 5 to match with Ampere's functional specification.
Test:
1. Login via web and check Account policy settings
2. Default value at "Max failed login attempts Value must be between 0
– 65535" is 5
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: I3b46d23f02b8c64ce27fd68cda335fe4550f926f
|
|
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I8e54833ac78e540e9dd5011533d53ff9a3af6763
|
|
This commit supports phosphor-hostlogger in STREAM mode, to forward
the byte stream from obmc-console to rsyslog via the imuxsock module.
The logs can be obtained in /var/log/. The logs are rotated by
logrotate. The size threshold of cpu console is 256KB and the others are
64KB. Each log is polled every 1 minute to check if its file size
exceeds the threshold to rotate it to the new file. Each console can be
saved in up to 2 log files.
Tested (CPU console):
1. Change the mux of CPU console to BMC
$ ampere_uartmux_ctrl.sh 1 2
2. Connect to CPU SOL console
$ ssh root@<BMC_IP> -p 2200
3. Execute commands in the CPU SOL console
=> The console output on CPU SOL console will all be logged to
/var/log/obmc-console-cpu.log
=> The log will be rotated to /var/log/obmc-console-cpu.log.1 if the
original log file size exceeds 256KB every 1min poll
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: I336bde09c4ef61ea68e05bef2c0b4e4298ea7959
|
|
This commit supports phosphor-hostlogger in STREAM mode, to forward
the byte stream from obmc-console to rsyslog via the imuxsock module.
The logs can be obtained in /var/log/. The logs are rotated by
logrotate. The size threshold of cpu console is 256KB and the others are
64KB. Each log is polled every 1 minute to check if its file size
exceeds the threshold to rotate it to the new file. Each console can be
saved in up to 2 log files.
Tested (CPU console):
1. Change the mux of CPU console to BMC
$ ampere_uartmux_ctrl.sh 1 2
2. Connect to CPU SOL console
$ ssh root@<BMC_IP> -p 2200
3. Execute commands in the CPU SOL console
=> The console output on CPU SOL console will all be logged to
/var/log/obmc-console-cpu.log
=> The log will be rotated to /var/log/obmc-console-cpu.log.1 if the
original log file size exceeds 256KB every 1min poll
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: I5586d1c634ec0f244069b868bd48b75560520dc1
|
|
Before flashing UEFI firmware, the BMC calls "obmcutil chassisoff" to
turn OFF the Host and wait 10s before checking if the Host is actually
OFF or not.
In some cases, 10s is not enough. This will make the flashing script
sees the Host is still ON and bypasses the flashing.
This commit changes to check Host state each 2s until maximum 60s to
avoid above failure.
Tested: check below steps 10 times and ensure all done successful:
1. Request to flash SCP firmware via Redfish.
2. Sleep 120s, when CPU is booting (did not jump to OS)
3. Request to flash UEFI firmware via Redfish.
4. Check if the flashing successes
Signed-off-by: Thang Tran <thuutran@amperecomputing.com>
Change-Id: I403f4b432e53b3b091315612488c50dbb595ea76
|
|
Some targets were changed in phosphor-state-manager at commit 86a1ebf9
"move chassis power services to phosphor-state-manager" which makes our
dependency incorrect.
This commit fixes the target dependency to work with new change
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I2c2b321161db367a86b5c1dc86071abab04f03b9
|
|
This commit supports phosphor-fan application for mtjade platform.
phosphor-fan-control service controls fan speed following the mapping
tables with TS4_Temp, OCP_Temp, nvme_bp_max and nvme_m2_max sensor.
phosphor-fan-presence-tach and phosphor-fan-monitor services are fully
supported.
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: I11a598f627d5e261689c2bd9d85ce65f4fc330ea
|
|
This commit removes phosphor-nvme implementation in mtjade
platform and configures to use dbus-sensors's nvmesensor.
Signed-off-by: Chau Ly <chaul@amperecomputing.com>
Change-Id: Ib9ce787e34265382a222697fb102fe63b4ba8a03
|
|
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: If4608c0520c5cc7f04f82193d7fc2f5841fdbd85
|
|
This power control implementation is not specific to openpower systems
(as seen by the variety of systems including it) so rename it to
something more appropriate.
This is using code from skeleton which is not ideal but that's a change
for another day.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Iffe6427b5494558b53cd6f5716a08cce82cb90d8
|
|
Add script to bind smpro driver when the Host is available but it is not
bound before. This helps applications that use smpro driver works.
Tested:
1. Do A/C the system.
2. Turn ON the Host
3. Check if smpro nodes are available
# ls /sys/bus/i2c/drivers/smpro-core/
2-004e 2-004f bind uevent unbind
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I6f96dc1f78ab77dae29bdc266949dcf7845de68c
|
|
Handle Altra Boot Progress reported from SCP regiter map by the use of
smpro-misc driver.
Tested: check boot progress is updated in Redfish for:
1. Reboot Host.
2. IPMI chassis power reset
3. IPMI chassis power cycle
4. IPMI chassis power off, then on
5. WebUI Orderly – OS shuts down, then server reboots
6. WebUI Immediate – Server reboots without OS shutting down
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: I9483328ccf5731904d0e6e71074561f7dd743bed
|
|
Enable smpro-hwmon/errmon/misc support for Mt.Jade.
This commit also enables IPMI SSIF and XFS filesystem support
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: Ic9bfb2cbdaa8b873b9bae44da35c7869bc456a3c
|
|
The meta-ampere/meta-jade folder contains features for Mt.Jade platform
which might support various chassis. Having :mtjade is redundant and
will limit features to work only with "standard mtjade".
This commit removes this string so that additional MACHINEs based on
Mt.Jade can work.
Signed-off-by: Thang Q. Nguyen <thang@os.amperecomputing.com>
Change-Id: If57acd557d678b8834f0c77f98855277e4eee844
|
|
Monitors GPIOF4 for 1P system and GPION5 for 2P system to detect if
system firmware hang. If these GPIOs stop asserting for more than three
seconds, BMC will trigger events and reset the system.
Tested:
1. Simulate the GPIO that deasserted for more than three seconds.
2. Fail to boot to Host, check the event log is created
and system will be reset.
Signed-off-by: Hieu Huynh <hieuh@os.amperecomputing.com>
Change-Id: I2c2d5d50e5ebd6d9b9e3435ab1c17393dac1d9cf
|