| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
The new subtree model brings the subtrees up from the openbmc-machines
layer.
Change-Id: I58a03ae1be374bc79ae1438e65e888375d12d0c0
Signed-off-by: Dave Cobbley <david.j.cobbley@linux.intel.com>
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
This commit adds witherspoon specific json file that specifies
whitelisted URLs to be allowed access by REST server.
Change-Id: Iaa9c77a38ac60748df8d819191596c17c635a18a
Signed-off-by: Nagaraju Goruganti <ngorugan@in.ibm.com>
|
|
Technically bitbake does not care where bbappends live but the
universal norm is to put bbappend files in the same directory
as the base recipe. Update to match meta-witherspoon and oe-core.
Tested: Booted QEMU witherspoon image to shell
Change-Id: I9b06de644ae414f76c2b8f390a6d32fe1e3e55d2
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
The previous phosphor-dbus-monitor rules would only watch
for the functional property to change on a fan, and then
resolve errors for that fan if the property changed to true.
This change will watch for both present and functional property
changes, and resolve fan errors if both properties are true at
that time.
This closes the window of where if a fan is removed long enough for
the present property to change to false but replaced fast enough
that the functional property always stays true then the NotPresent
error that was created will never get resolved so the fault LEDs
won't turn off.
Resolves openbmc/openbmc#3231
Tested: Various combinations of Present/Functional property state
changes and also fan plugs/unplugs.
Change-Id: I3255c0510b95509810b018a2a71d307bd7d4a946
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
OE norms dictate recipes that track HEAD be named:
foo_git.bb
and set PV explicitly to "x.y.z+git${SRCPV}"
Change-Id: Iaeffab1e3bb1df6720ae4e236c3515fb1ec30cc8
Tested: Built the obmc-console recipe
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Change-Id: I71ef26fc33bf9a7f85eaf59bc48139d8b42d0929
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
channel.yaml keeps the info about number of available Ethernet
channels on the given machine.
Resolves openbmc/openbmc#3191
Change-Id: I904e02a96bb6f295c6456b8a8ba26e756c218df0
Signed-off-by: Nagaraju Goruganti <ngorugan@in.ibm.com>
|
|
Change-Id: Ib13ae97aaba4a0af0aefb5b4dbd2d0d6f3c08728
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
Make this service have a .wants dependency to the
obmc-chassis-poweron@.target instead of .requires. This way
if it fails, the power on will still continue, just without
the failing drivers bound to their devices.
The drivers are only used for surfacing hwmon temps and voltages.
The only ones of those actually used by code are the water
cooled witherspoon VDD VRM temperatures which are used by fan
control, and fan control can deal with missing VRM temps.
Tested: Inject a fail into the bind script, and ensure the service
fails but the power on still continues.
Resolves openbmc/openbmc#3205
Change-Id: Id7551b8dd9e36c105f11e91697035d7ca90bdec3
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
Enhanced the ir35221-unbind-bind.sh script to:
1) Retry a few times on failures binding the device driver.
2) Still try to bind other devices even if one device fails.
The script will still return a nonzero exit code if any of
the binds fail.
Tested:
* Do several power cycles.
* Modify script to fail a bind. Check that it returns nonzero.
Change-Id: Ia235557292d792998a1c10260fe2bf8b2e631b2c
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
A change was put in to wait 10 seconds before powering off a system when
too many fan enclosures were non-functional, this reverts that change
back to the original 5 seconds. The original change was done to help
mitigate a performance issue where PDM was held off from being able to
process fan functional state changes.
Reverted change: https://gerrit.openbmc-project.xyz/#/c/8352/
Tested:
System shutdown after 5 seconds with 2 non-functional fan enclosures
Change-Id: If40b54f475a54d151439922fc0893686caeb8ae7
Signed-off-by: Matthew Barth <msbarth@us.ibm.com>
|
|
Add the sync manager which watches for files to be synced
to the alternate BMC chip.
Change-Id: I543dacf389629c024b35d1b855745bbcc0bbd848
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
|
|
Change-Id: I75cd0dcc7f42ae34cab232312f1f453f28cd480d
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
Enable signature verification in the openpower-software-manager code
for witherspoon. This causes an error to be logged if updating to
an unsigned image, or image signed with a different key than the one
on the system, and if field mode is set, it'll stop the activation
process.
Tested: PNOR signature verification is enforced on witherspoon,
verified error is logged with and without field mode enabled, and
activation is prevented with field mode enabled.
Change-Id: I6b8b74f146066da058e137779faf9af157f7131b
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
|
|
The OCC related sensor fault events should be subscribed after the OCCs
active precondition state passes. This keeps from registering for OCC
provided sensor fault signals prior to the OCC driver being bound and
active.
Tested:
OCC fault sensor signals only registered for after being active
Change-Id: I032ebb0145b5d16e4d54e4873429916e05d3cf48
Signed-off-by: Matthew Barth <msbarth@us.ibm.com>
|
|
Set the default value of the TPMEnable sensor to true on
Witherspoon systems to catch systems that have a missing
TPM card populated.
Tested: Verified the settings value for TPMEnable is set
to true (1) when the BMC boots up.
Change-Id: I445e67dc53d3bac5dc7f6f613ac03202aab567c4
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
|
|
The power values being displayed were basically instantaneous
readings as measured by the APSS monitor device, and as such
could show a lot of peaks and valleys that would be undesirable
to surface.
In the future they may get added back in with some averaging done.
In the meantime, the user can use the nvidia-smi command line tool
on the host to get the proper power values.
Tested: No GPU power sensors in D-Bus.
Resolves openbmc/openbmc#3183
Change-Id: Ie2d92560867f29dbf852c53c55b8d371f5fef06f
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
The GPU power hwmon sensors recently changed names to match
the GPU temperature sensor naming conventions. Update this
YAML to reflect that.
Change-Id: I5100ee514bed3ec7901deddcfd3bd9d619d8cd51
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
The code will now copy the correct air or water cooled
config file to the correct name so phosphor-hwmon can find it.
It was just a link before, but that would cause problems with
code updates as this directory is preserved, so if older
code was put back on the link would point to nothing and
phosphor-hwmon would never start for the MAX31785.
Now with the copy the old code can still pick up this file.
The only hole this has is if someone goes from the new code
back to the old code, and then swaps air/water cooling, as the
old code will then pick up the config file generated by the
newer code, which now won't match the actual config. This
would be a very unlikely scenario.
Change-Id: I45db3c3ac8265caae30163a4e90035189e861e82
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
Pick up the latest RAS policy table, which is used to
add the EventID and Description fields to error logs on
Witherspoon.
Change-Id: I18c0e855372d438a5c5171fcf6fbf56bd7376aed
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
Change to using an indirect lookup with the *_function_id file
to determine what the OCC's hwmon power values represent, similar
to how the OCC temps used the *_label files.
The current code was hardcoding them, assuming hwmon would number
them the same every time, but that is not the case.
The function IDs are documented here:
https://github.com/open-power/docs/blob/master/occ/OCC_P9_FW_Interfaces.pdf
Resolves openbmc/openbmc#3152
Change-Id: Id4c50c7175a8108c317d9c301ad4a682a4f4c89f
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
-commit #f24faab provides basic infra for FRUs which are accessible
through host but host is not sending the details for those FRUs.
-GPU FRUs are of type mentioned at commit #f24faab.This commit
hardcodes the WITHERSPOON GPU FRU info into YAML.
Change-Id: I783bf77d8e95abf687d2605e2c97a5bb10504c70
Signed-off-by: Nagaraju Goruganti <ngorugan@in.ibm.com>
|
|
Witherspoon requires an dev_id.json file whose content is partially
derived from data provided by the os-release package. os-release is
updated for each commit, as some of its content (VERSION and
VERSION_ID) can be derived from `git describe`. As dev_id.json was
provided by the phosphor-ipmi-host package, every commit transitively
triggered a rebuild of phosphor-ipmi-host in order to satisfy
Witherspoon's requirements.
Always rebuilding phosphor-ipmi-host is unhelpful for several reasons:
* It needlessly reduces CI throughput, as it is likely the commits in
question do not modify the phosphor-ipmi-host package.
* GCC suffers from what appears to be an unfixable[1] bug[2] that causes
phoshor-ipmi-host to consume large (>5GiB) amounts of RAM when
compiling some (at least Witherspoon) sensor configurations.
To avoid this, separate the configuration files out into
virtual/phosphor-ipmi-config and phosphor-ipmi-config packages that
phosphor-ipmi-host RDEPENDS on. Witherspoon provides an alternative
implementation in witherspoon-ipmi-config to mangle dev_id.json to its
particular requirements.
A virtual is used rather than a simple bbappends for Witherspoon, as the
bbappend approach breaks builds of machines other than Witherspoon if
Witherspoon is built first: The Witherspoon-specific dev_id.json file is
deployed in its mangled form into e.g. a Zaius image. Specifically, the
following sequence will trigger the issue:
$ TEMPLATECONF=.../witherspoon.conf . openbmc-env
$ bitbake obmc-phosphor-image
$ rm -rf conf
$ TEMPLATECONF=.../zaius.conf . openbmc-env
$ bitbake obmc-phosphor-image
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80290#c26
[2] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80290
Change-Id: Ib9629fc77b29e2deeab3f1c3a145d9e966c14ec4
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
Add a condition check for fan1 that will only create fan1 associated
functional properties and actively monitor fan1 if the condition passes.
The condition checks the CoolingType is not WaterCooled.
Tested:
Fan1 functional properties are created when condition passes
Fan1 is monitored when condition passes
Change-Id: I0d9a91ceea500f82e4ccad2dbdc887a14286cf95
Signed-off-by: Matthew Barth <msbarth@us.ibm.com>
|
|
The phoshor-fan-monitor-init and phosphor-fan-control-init
package names that the SYSTEMD_OVERRIDE directives were given
don't actually exist, as those apps are just part of the
phosphor-fan-control and phosphor-fan-monitor packages.
Fixup these directives so the fan watchdog configuration is
used for these services which means if they fail the fan
watchdog will trip as desired.
Tested: Check that phosphor-fan-control-init@0.service.d and
phosphor-fan-monitor-init@0.service.d now show up, and killed
fan-control-init and watched that the watchdog triggered.
Change-Id: I1786891ec5da7045b4f0a9f72d5ae29f454fa9a0
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
These packages are now part of meta-phosphor, so they can be removed
from the meta-witherspoon directory.
The feature is still enabled by their inclusion in
meta-witherspoon/recipes-phosphor/images/obmc-phosphor-image.bbappend.
Change-Id: I653e2b3c96bf9e9ab4d607f60355ff68432e855a
Signed-off-by: Joel Stanley <joel@jms.id.au>
|
|
When a temp sensor provided by the OCC is faulted, set the fans to full
speed(10500 RPMs)
Tested:
Set Functional property of an OCC provided sensor to `true` and
fans are set to 10500.
Fan speeds drop when the same Functional property is set to `false`
Resolves: openbmc/openbmc#2222
Change-Id: Iddd9049fc9677ca587422a5bf4104c0d3787af1a
Signed-off-by: Matthew Barth <msbarth@us.ibm.com>
|
|
The values are just outside of the allowed margins of the
fan floor and max speed values.
Tested: Test that the thresholds show up in the web UI
and that the web UI shows when they have been
crossed.
Resolves openbmc/openbmc#3114
Change-Id: I6420252d98b974921d6858dbecb3b52f5a8c646a
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
Add a new udev rule to start the phosphor-hwmon instance for
the MAX31785 fan controller differently than other hwmon devices.
Also introduce different hwmon config files for air and water cooled
systems where the water cooled system doesn't use fan 1.
The new udev rule starts a service that will check the cooling type
and then link in the appropriate config file before starting the
actual phosphor-hwmon service.
Note that this phosphor-hwmon service is still stopped the same as
any other phosphor-hwmon service.
Tested: Test that phosphor-hwmon is started using the correct
config files on air and water cooled systems.
Change-Id: Ia93d97df024e7b6bff5eb379ddea4730fe34b55b
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
This service will be started in the udev rule for the MAX31785
hotplug add event. It will wait for the phosphor-cooling-type
service to be done, and then run the script to start the
phosphor-hwmon instance for the MAX31785 fan controller after it
chooses the correct config file to use.
Tested: Test that this service correctly runs the script and that
phosphor-hwmon does get started.
Change-Id: I4495f7073d63004d4dd8333df8964d69beaae4ea
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
The phosphor-hwmon instance for the MAX31785 fan controller
chip needs to use different config files based on if the system
is air or water cooled, since different numbers of fans are used.
This script will check the cooling type, link in the correct file
based on that cooling type, and then start the actual hwmon
service.
Tested: Tested as part of end to end testing of this feature.
Change-Id: Ie3efee5d3c6671911b505223152d32f3f5c5c259
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
Change the type of the phosphor-cooling-type service to
a oneshot, and also tell it to remain after exit.
This is so that any services that depend on this won't start
until the the phosphor-cooling-type application is done so
they can rely on the cooling type property being set in the
inventory.
Tested: Checked that a service that depends on this one
and gets the cooling type will read the correct value.
Change-Id: Id4852983a31787b19bee8a2106a46128c294b2ce
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
For IPMI v2.0/RMCP+ sessions, authentication, integrity and
confidentiality algorithms are involved. The algorithms used in
combination are called Cipher Suites. Each cipher suite has the
id as the key and a collection of algorithms.
Each algorithm has a number assigned in the IPMI specification.
Cipher data information can be read by the Get Channel Cipher
suites command, and the RMCP+ messaging Cipher Suite entries
support and the RMCP+ messaging Cipher Suite Entries parameter
of the Get LAN configuration parameters command.
Change-Id: I3460046a0c72acd5cb2f9066f264142a2378120c
Signed-off-by: Tom Joseph <tomjoseph@in.ibm.com>
|
|
Match up the GPU power sensor labels with the temperature ones,
so they are both GPU0 through GPU5 instead of P0_GPU0 through
P1_GPU2.
Resolves openbmc/openbmc#3111
Change-Id: I509349049b8528415084e45c09200c16d2628a31
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
Add the fbterm, uart-render-controller and liberation font packages to
the witherspoon build.
Change-Id: Ifa20e9455f06f2ae4d7e9547f62e70ea102b4c89
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
|
|
This change adds a small daemon to monitor the state of the host (power
and video device status), and start/stop the fbterm UART renderer as
appropriate.
Change-Id: I6c7cf7a88e9a19cd87d6b781a4a1519597882345
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
|
|
For text-to-video rendering, we need a font. This change adds a small
overlay to the existing liberation fonts package, to only use the
monospace font.
Change-Id: I7e0bc307dceb95732f94616d9249df34b484b629
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
|
|
This change adds a definition for fbterm, customised for rendering UART
data to a local framebuffer device.
fbterm upstream isn't very active - last release was in 2008. So, use a
newer git tree, which includes a couple of fixes and customisations for
our usage in uart rendering.
Change-Id: Ib19e879283548dbe403401b6fecfaaeb8cf74c1c
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
|
|
This is used by the ibm-logging code, and contains
additional information about the error log entries
that can show up on a Witherspoon. This file is maintained
by the IBM service team, and checked in as-is. A script
runs during the build to pull out the fields used by
the BMC code.
It also enables the use of the com.ibm.Logging.Policy
D-Bus interface.
Resolves openbmc/openbmc#2998
Change-Id: Idf08e397f4e93112ece5cbfe9db67f5682637c32
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
When a sensor provided from an OCC contains a value of
0x00(unavailable), the sensor will be removed from dbus by hwmon. An
InterfacesRemoved signal is triggered when this occurs and this enables
fan control to remove that sensor's interface(and associated properties)
from its internal cache used across set speed events.
Tested:
An InterfacesRemoved signal is received for an unavailable sensor
That sensor's interface and properties are removed from the cache
Change-Id: Ia6343688c2824a984d23a9b16ea9e709d5ed2f9e
Signed-off-by: Matthew Barth <msbarth@us.ibm.com>
|
|
The witherspoon power engineer has requested the lowering of the
over-current warning values
Testing: Going to let CI validate basic boot. The power engineer has
done his own more intense testing of these values.
Resolves openbmc/openbmc#3099
Change-Id: I4a2b2bb132dd43e05828e4c0ed3448dfd4237301
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
Added "mutability" flag to indicate if a sensor is READ/WRITE/RW.
Change-Id: I3dcc27526adc2a5d6b1206e8beee66093220763a
Signed-off-by: Jayanth Othayoth <ojayanth@in.ibm.com>
|
|
Enable signature verification in the phosphor-software-manager code
for witherspoon. This causes an error to be logged if updating to
an unsigned image, or image signed with a different key than the one
on the system, and if field mode is set, it'll stop the activation
process.
Tested: Signature verification is enforced on witherspoon,
verified error is logged with and without field mode enabled, and
activation is prevented with field mode enabled.
Change-Id: Ifc8f8054f8d852cc16942af9cbf58d60aff3fc33
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
|
|
Change-Id: I7a678b64290c99a53dbecb326be767b00e3b7aeb
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
When an EAGAIN return code(11) is returned for any sensor from an OCC,
remove that sensor from dbus.
Functionally requires: https://gerrit.openbmc-project.xyz/#/c/9893/
Tested:
OCC sensors are removed from dbus when EAGAIN is returned
Change-Id: I2fee78ff7876b8750555454436dd7b53eb3a0e78
Signed-off-by: Matthew Barth <msbarth@us.ibm.com>
|
|
This service file needs to be started only for
witherspoon.
This is used to stop the ethernet port failover
on the system.
Resolves openbmc/openbmc#2931
Change-Id: I0f144942df1d77de4d990a4aad8d9764d8455251
Signed-off-by: Ratan Gupta <ratagupt@in.ibm.com>
|
|
IPMI command get device id supports auxiliary firmware version field
in the response. This field is 4 bytes and the first 2 bytes would
contain the count from the version tag in BCD format, the next 2 bytes
would represent the release version in BCD format. This change can
support the VERSION_ID format of IBM enterprise build and a build
from the master.
Resolves openbmc/openbmc#2972
Change-Id: I4f4e23a3be687b814cef51b1b96c7b97778bcb16
Signed-off-by: Tom Joseph <tomjoseph@in.ibm.com>
|
|
We have 236 commits, 127 files changed, 17766 insertions(+), 2128
deletions(-). Some of these are backports from upstream. This list
does not include patches from the 4.13 stable releases, but we do
include those in the dev-4.13 branch.
1 Alexey Khoroshilov
34 Andrew Jeffery
1 Arnd Bergmann
1 Benjamin Herrenschmidt
1 Bhumika Goyal
1 Brad Bishop
1 Brendan Higgins
11 Christopher Bostic
1 Cyril Bur
14 Cédric Le Goater
49 Edward A. James
3 Gavin Shan
1 Guenter Roeck
8 Ivan Mikhaylov
1 Jacek Anaszewski
1 James Feist
7 Jeremy Kerr
72 Joel Stanley
2 Julia Lawall
1 Ken Chen
6 Lei YU
3 Milton Miller
1 Mykola Kostenok
1 Patrick Venture
2 Philipp Zabel
1 Rick Altherr
11 Samuel Mendoza-Jonas
2 Wei Yongjun
1 Xo Wang
1 Yong Li
Note that the 4.13 branch is EOL'd by the Linux community, and as such
should not be used for any products beyond development.
React to removal of occ hwmon instances from device trees with a
new startup/shutdown mechanism for phosphor-hwmon.
To fix this, a helper script will be used to start the service that
will pass the service the device tree name if it is present, or the
udev device path if it isn't. This script will still run from the
udev rule as before, but it will stop and start the service itself
without using the SYSTEMD_WANTS attribute.
As the path to the hwmon environment file matches the service
template argument, the paths for the OCC .conf files need to change
to match the device path instead of the previous device tree path.
Note that the pure device path would have the hwmon instance number
in it, but since that can't be known ahead of time it is stripped
off by the script that starts the service.
In addition, the pure device path for the OCCs contain several
':'s, meaning the associated environment files would also need to.
However, Yocto/Bitbake cannot handle a ':' in a file path, so they
are converted to '--'s by the script that starts the service and
phosphor-hwmon will convert them back internally when it starts.
The service file also needed some changes now that the service
lifetime is no longer controlled by systemd via SYSTEMD_WANTS.
This script will be called by a udev rule to start and
stop phosphor-hwmon when the hwmon device driver is started
and stopped.
It is passed both the device path and the OF_FULLNAME device
tree attribute. If OF_FULLNAME is present, it will start the
service with that as its template argument, otherwise it will
use the device path. This is to handle devices that aren't in
the device tree so they won't have OF_FULLNAME.
If a '/hwmon/hwmonN' is in the path it is removed, as this path
is also used as a path to an environment file and so must be
known ahead of time, which the hwmon instance N is not.
If there is a ':' in the path name, it is converted to a '--'.
Yocto/Bitbake cannot handle a ':' in file paths.
Resolves openbmc/openbmc#2953
Change-Id: I815be4d6d9e1cbea8428bb1bb8c332776ee71ece
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Missing the ${D} on the install command - this still
managed to work on the older versions of yocto but
not with the latest one.
Also removed the PROVIDES=<recipe-name> line since
it's redundant.
Change-Id: I83127f0c457ec31bd99311371ec76abaf6fbc76a
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
|
|
Adding power supply to BMC accessible FRUs.
Partially Resolves openbmc/openbmc#2984
Change-Id: Ib4af12f281f433ecac727324ec75e95fd62b8714
Signed-off-by: Nagaraju Goruganti <ngorugan@in.ibm.com>
|
