PEL: Use lg2 in Manager class

Modernize it a bit and it makes it easier to see debug traces which can
be done by just running the daemon from the command line. There are
quite a few debug traces in

PEL: Use lg2 in Manager class

Modernize it a bit and it makes it easier to see debug traces which can
be done by just running the daemon from the command line. There are
quite a few debug traces in this file.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Id1291e3a3c3882d1ed917d4a414cb6253ebf2644

show more ...

PEL: Use lg2 in DataInterface files

Modernize it a bit and it makes it easier to see debug traces which can
be done by just running the daemon from the command line.

Signed-off-by: Matt Spinler <sp

PEL: Use lg2 in DataInterface files

Modernize it a bit and it makes it easier to see debug traces which can
be done by just running the daemon from the command line.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Ic0416222e29be0c6687b6d31cb6ca4feae2ad619

show more ...

PEL: Handle failing to start a PLDM cmd better

A recent PLDM bug caused the registerReceiveCallback() function, which
is used to setup listening for the PLDM response from the host when
telling the

PEL: Handle failing to start a PLDM cmd better

A recent PLDM bug caused the registerReceiveCallback() function, which
is used to setup listening for the PLDM response from the host when
telling them about a new PEL, to throw an exception.

When this happened, the code got stuck in the 'in progress' state, so it
would never try again when the next PEL came in.

Fix that by having startCommand() throw an exception instead of calling
the failure response function callback. With this change, the code will
continue on to call the cleanupCmd() function so everything is ready
when the next PEL comes in.

Tested:
With the bad PLDM code, after the first PEL ran out of retry attempts,
created another PEL and saw the code attempt again to call PLDM. Also,
wrote a new unit test case for it.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I38034440435d6a86e8dd880eef09499f19dd6e9c

show more ...

PEL: Use lg2 in PLMD related files

There are a lot of debug traces in the code that uses PLDM to send PELs
up to the OS. Convert the files that deal with that to lg2 so that the
debug traces can be

PEL: Use lg2 in PLMD related files

There are a lot of debug traces in the code that uses PLDM to send PELs
up to the OS. Convert the files that deal with that to lg2 so that the
debug traces can be seen by just running phosphor-log-manager from the
command line where lg2 will print to the console as opposed to having to
change the journal priority in an overlay file.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I818bab1bf636fe83d3e7ff64c1bc31ad5e09705e

show more ...

PEL: Remove unnecessary call to restore resolution

The Resolution D-Bus property, which holds the PEL callouts, are saved
along with the other elog properties in the cereal backing file and
don't ne

PEL: Remove unnecessary call to restore resolution

The Resolution D-Bus property, which holds the PEL callouts, are saved
along with the other elog properties in the cereal backing file and
don't need to be manually restored when the daemon starts up.

Tested:
Resolution is still filled in after restarting the daemon.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I18e1b79d40b219a6a37c013355cd965de11ce8cb

show more ...

PEL: Fixed serializing of elog properties

The PEL code modifies the Resolution and EventId properties on the
standard event log objects to fill them in with PEL specific values. It
was originally d

PEL: Fixed serializing of elog properties

The PEL code modifies the Resolution and EventId properties on the
standard event log objects to fill them in with PEL specific values. It
was originally doing this by using the resolution() and eventId()
override functions on the Entry object that would do the elog serialize
after it updated those properties.

Since then, the code was changed to instead call the resolution() and
eventId() functions that also take the bool skipSignal parameter, which
caused it to not call the overridden functions which meant those
properties weren't serialized and so not restored after a restart.

While we could just also override the functions that take that
skipSignal parameter to do the serialize, those get called when the
event log object is restored on startup, so it would cause unnecessary
calls to serialize.

Instead, just call the serialize() function directly after updating the
event ID and resolution when creating a PEL.

Tested:
The Resolution and EventId D-Bus properties are now correct after the
process is restarted.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I8c1822e9c31925983feddab657644c98f37ef079

show more ...

PEL: Error log entries for ECC errors

This commit adds three new error log entries for ECC errors.

The new entries include:
- Correctable ECC memory error collection limit is reached.
- Correctable

PEL: Error log entries for ECC errors

This commit adds three new error log entries for ECC errors.

The new entries include:
- Correctable ECC memory error collection limit is reached.
- Correctable ECC/other correctable memory error.
- Uncorrectable ECC/other uncorrectable memory error.

Change-Id: Ibfa7654bdfbccc5cf7154664a048d931d10b433c
Signed-off-by: Lakshmi Yadlapati <lakshmiy@us.ibm.com>

show more ...

lg2: simplify source_location for clang-16

As of clang-16, `source_location` is fully supported, so we do not need
to use the `experimental::source_location` workarounds anymore.
Eliminate them and

lg2: simplify source_location for clang-16

As of clang-16, `source_location` is fully supported, so we do not need
to use the `experimental::source_location` workarounds anymore.
Eliminate them and use `std::source_location` directly rather than the
`lg2::source_location` alias.

Tested: Compiled the repository with CXX=clang++.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Iba987ed9580d228781acac630681fda172a6cef7

show more ...

PEL : Modify faultlog error to include only serviceable record count

Modified to include only serviceable records count like system
guard record count, unresolved pel's with deconfigured bit set.

N

PEL : Modify faultlog error to include only serviceable record count

Modified to include only serviceable records count like system
guard record count, unresolved pel's with deconfigured bit set.

Not including manual guard and FCO deconfigured record counts
in the PEL as thes are created by the user/servie engineer. These
records will be shown in the nag dump though.

Faultlog PEL will be created only if serviceable records are found.

Tested:
root@xxxx:/tmp# peltool -i 0x500090BF
{
"User Header": {
"Section Version": "1",
"Action Flags": [
"Service Action Required",
"Report Externally",
"HMC Call Home"
],
},
"Primary SRC": {
"Section Version": "1",
"Error Details": {
"Message": "Firmware detected either a FRU deconfigured And/Or guard record",
"GUARD_RECORD_COUNT": [
"0x1",
"Number of system guard records if any"
],
"PEL_WITH_DECONFIG_BIT_COUNT": [
"0x0",
"Number of PEL's having deconfig bit set"
]
},
"Valid Word Count": "0x09",
"Reference Code": "BD50F138",
"Hex Word 6": "00000001",
"Hex Word 7": "00000000",
"Hex Word 8": "00000000",
"Hex Word 9": "00000000",
"Callout Section": {
"Callout Count": "1",
"Callouts": [{
"FRU Type": "Maintenance Procedure Required",
"Priority": "Mandatory, replace all with this type as a unit",
"Procedure": "BMC0008"
}]
}
},
"User Data 1": {
"Section Version": "1",
"Sub-section type": "1",
"Created by": "bmc error logging",
"GUARD_RECORD_COUNT": "1",
"PEL_WITH_DECONFIG_BIT_COUNT": "0"
}
}

Signed-off-by: Marri Devender Rao <devenrao@in.ibm.com>
Change-Id: Id845c3ecc45b7be7c553942658b68d4d928e29b2

show more ...

Don't throw on Settings access fail

The isQuiesceOnErrorEnabled() function makes a call to the Settings
daemon. If that throws an exception, the code just re-throws the
exception. While that excep

Don't throw on Settings access fail

The isQuiesceOnErrorEnabled() function makes a call to the Settings
daemon. If that throws an exception, the code just re-throws the
exception. While that exception is caught by sdbusplus, it causes the
function to exit early and not actually save the event log in the class.

To fix this, just change the code to say quiesceOnError isn't enabled
and continue on.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I814e6dae90483c98e0f24af90baa5f267d22b78d

show more ...

PEL: No PC signals when updating entry props

The PEL code has to update properties on the
xyz.openbmc_project.Logging.Entry interface after the corresponding PEL
has been created. These updates don

PEL: No PC signals when updating entry props

The PEL code has to update properties on the
xyz.openbmc_project.Logging.Entry interface after the corresponding PEL
has been created. These updates don't need to send PropertiesChanged
signals since it is when the event log is being created.

The ipmid daemon is watching for PC signals and it was triggering a
bunch of unnecessary activity.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I6f9a2992abf0454bc95b44bb9897642cecd3200e

show more ...

PEL: Support for CheckstopFlag msg reg field

Similiar to the DeconfigFlag field that was recently added, this one
indicates the PEL is for a hardware checkstop and results in a bit in
SRC hex word 5

PEL: Support for CheckstopFlag msg reg field

Similiar to the DeconfigFlag field that was recently added, this one
indicates the PEL is for a hardware checkstop and results in a bit in
SRC hex word 5 being set.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Ib05de7471ad3e32f48e7f20a5c611abc119fe82a

show more ...

PEL: Fixes for gcc13

* Add the cstdint header file as now required to get the uint* types.
* Fix a move assignment test
* Refactor some nlohmann::json code to avoid:

```
/usr/include/c++/13/valarra

PEL: Fixes for gcc13

* Add the cstdint header file as now required to get the uint* types.
* Fix a move assignment test
* Refactor some nlohmann::json code to avoid:

```
/usr/include/c++/13/valarray:1201:1: note: template argument deduction/substitution failed:
../extensions/openpower-pels/registry.cpp:665:43: note: ‘const nlohmann::json_abi_v3_11_2::basic_json<>::value_type’ {aka ‘const nlohmann::json_abi_v3_11_2::basic_json<>’} is not derived from ‘const std::valarray<_Tp>’
665 | (name == j["SRC"]["ReasonCode"] && type == LookupType::reasonCode));
```

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Ia3e733602134a60008d0d47934f95a217d2a0eb1

show more ...

PEL: Trace less when watching Present changes

The fans and power supplies in IO expansion drawers are also in the
inventory, hosted by PLDM, and have their Present property set to true
at least once

PEL: Trace less when watching Present changes

The fans and power supplies in IO expansion drawers are also in the
inventory, hosted by PLDM, and have their Present property set to true
at least once every boot after the hypervisor starts and tells PLDM
they are present. This causes multiple traces every boot from the PEL
code watching for fan and power supply hot plugs. Also, one of the
times the items become present, the call to get its location code from
PLDM fails which causes another trace.

Since the PEL code doesn't care about fans and PSs in IO drawers, just
change these traces to debug ones so that no traces show up on normal
boots. There will still be a trace if the code does end up clearing the
deconfig flag from a PEL with a plugged local fan/PS as a callout.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I18cf95e6adfc6aa5700fe33a74c18014b36d0b9d

show more ...

PEL: Up the state PEL journal lines to 100

Change the number of journal lines captured to the max of 100 for state
related PELs, since it seems like there's always interesting things
going on when t

PEL: Up the state PEL journal lines to 100

Change the number of journal lines captured to the max of 100 for state
related PELs, since it seems like there's always interesting things
going on when those are created. The resulting PEL was about 11KB,
which is still well under the max of 16KB.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I8344431386d8eb555814b64ea1b929aa4ec6b925

show more ...

PEL: Add reg entry for master FSI detection fail

This error is created when something goes wrong during the scan the FSI
device driver does.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-

PEL: Add reg entry for master FSI detection fail

This error is created when something goes wrong during the scan the FSI
device driver does.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Ifd166ad3395602cc6d1f36356fc121379b3d2765

show more ...

PEL: Use uppercase hex in message registry

The code in the Registry class requires the ReasonCode field in the
message registry to use upper case hex digits (A-F) to be able to find a
match to print

PEL: Use uppercase hex in message registry

The code in the Registry class requires the ReasonCode field in the
message registry to use upper case hex digits (A-F) to be able to find a
match to print the description in peltool, so change the message
registry schema to require that. Also fix an entry that had lower case.

Alternatively, the code could be fixed, but with the schema checking
that runs in CI to catch errors this method seems less impactful.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I4f26c876575be58232f1408f9357465a3340002a

show more ...

PEL: Update README.md with latest functionality

Describe how the code will clear the deconfig flag for replaced fans and
power supplies.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id:

PEL: Update README.md with latest functionality

Describe how the code will clear the deconfig flag for replaced fans and
power supplies.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I8a78748a8093fb764c890433e9e3d4d3248843bb

show more ...

PEL: Clear deconfig flag after callout replaced

Making use of the previous commit's framework to call a function when a
fan or power supply becomes present, add code to the Manager class to
register

PEL: Clear deconfig flag after callout replaced

Making use of the previous commit's framework to call a function when a
fan or power supply becomes present, add code to the Manager class to
register a callback that will clear the deconfig flag for all PELs
created with the power-thermal or fan component ID that have the
location code of the replaced fan/PS as a callout.

This way, the degraded mode reporting code will no longer pick up those
PELs in its report as since the hardware was replaced those PELs are no
longer relevant.

This is necessary only for fans or power supplies because they're the
only N+1 hardware that can be hot plugged at runtime. And also because
this is what the IBM service team wants.

Tested:
Simulated missing hardware (changed present D-Bus property for fans,
toggled PSU presence GPIO in the simulator for PSs). Saw errors get
created for it, then simulated replacing it and saw those errors have
their deconfig flag cleared, verifying before and after with peltool:

```
// Remove and replace fan
phosphor-fan-monitor: Fan /system/chassis/motherboard/fan0 presence state change to false
phosphor-log-manager: Created PEL 0x50000002 (BMC ID 2) with SRC 110076F1
phosphor-fan-monitor: Fan /system/chassis/motherboard/fan0 presence state change to true
phosphor-log-manager: Detected FRU /xyz/openbmc_project/inventory/system/chassis/motherboard/fan0 (U78DB.ND0.1234567-A0) present
phosphor-log-manager: Clearing deconfig flag in PEL 0x50000002 with SRC 110076F1 because U78DB.ND0.1234567-A0 was replaced

// Remove and replace PS
phosphor-log-manager: Created PEL 0x50000003 (BMC ID 3) with SRC 110015F6
...
phosphor-psu-monitor: Updating inventory present property. present:true invpath:/system/chassis/motherboard/powersupply0 name:powersupply0
phosphor-log-manager: Detected FRU /xyz/openbmc_project/inventory/system/chassis/motherboard/powersupply0 (U78DB.ND0.1234567-E0) present
phosphor-log-manager: Clearing deconfig flag in PEL 0x50000003 with SRC 110015F6 because U78DB.ND0.1234567-E0 was replaced
```

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Iee05b4a612ca8f438f8c89f37b4e7b529a131a9f

show more ...

PEL: Watch for fan/PS hotplugs

Code is going to need to know when a fan or power supply (the only
hotpluggable redundant FRUs) are added or replaced so that it can clear
a flag in PELs that have tha

PEL: Watch for fan/PS hotplugs

Code is going to need to know when a fan or power supply (the only
hotpluggable redundant FRUs) are added or replaced so that it can clear
a flag in PELs that have that HW as a callout. This is so other code
that is doing degraded mode notifications will not do any notifications
for PELs calling out HW that has been replaced.

To enable this functionality, add support to the DataInterface class to
tell subscribers via a function callback when a fan or power supply
becomes present, as indicated by the Present property.

Code will watch the Present property of fan and power supplies to change
via the PropertiesChanged signal, as well as watch for InterfacesAdded
to catch when they show up on D-Bus in the first place.

It won't start any of these watches until the BMC gets to ready state so
that the inventory has a chance to be populated first.

On the first boot when the inventory was previously empty there will be
a round of inventory InterfacesAdded signals that will wake up the
daemon when PLDM receives the processor core presence information from
hostboot, but it just checks and sees it's the wrong interface and
returns, and I'm not sure how it can be avoided.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I93d0727c3082677826db4a4a02c1a30986f6099b

show more ...

PEL: Modify Repo::updatePEL behavior

The Repository class's updatePEL function takes a PEL and then
calls a passed in function to update it.

Change the behavior slightly so the callback returns a b

PEL: Modify Repo::updatePEL behavior

The Repository class's updatePEL function takes a PEL and then
calls a passed in function to update it.

Change the behavior slightly so the callback returns a bool - true if
the PEL was actually updated, and false else. That way the code knows
if the PEL needs to be written back out or not.

There is also a minor change to refresh the PEL attributes map for the
PEL inside the updatePEL function so it doesn't need to be done outside
of it.

This is all to support upcoming functionality where an updatePEL call
won't know if the PEL needs to be updated until the PEL fields can be
checked.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Ic1eabd2fcd8dfc7f559be24142b3e147d4b65062

show more ...

PEL:test: Refactor mocked checkDumpStatus usage

When creating an instance of the SRC class in a testcase, it needs a
filled in mock of the checkDumpStatus() function. Instead of manually
doing that

PEL:test: Refactor mocked checkDumpStatus usage

When creating an instance of the SRC class in a testcase, it needs a
filled in mock of the checkDumpStatus() function. Instead of manually
doing that everywhere a PEL or SRC class is created, just do it in the
constructor of the mock DataInterface class.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I74790c67251465aae87d318ea37891d9eabab5e5

show more ...

PEL:pel_manager_test: Refactor temp dir cleanup

Several testcases were creating a temporary directory and then removing
it at the end, except for one which missed the cleanup so the dir would
stick

PEL:pel_manager_test: Refactor temp dir cleanup

Several testcases were creating a temporary directory and then removing
it at the end, except for one which missed the cleanup so the dir would
stick around in /tmp after the run was done.

To fix that and to make sure it doesn't happen again, just add the
creating and deleting of the temp directory to the test fixture class so
the deleting happens automatically.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I2abb720b4c0aeb9dec6117d6399c3e4692709c68

show more ...

PEL: peltool: show hostboot deconfig/guard flags

The deconfig and guard flags in the SRC section are also valid for
hostboot SRCs, so display them in peltool for hostboot PELs.

Signed-off-by: Matt

PEL: peltool: show hostboot deconfig/guard flags

The deconfig and guard flags in the SRC section are also valid for
hostboot SRCs, so display them in peltool for hostboot PELs.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I3a36d61ced592e0718e016665587444b7e14f964

show more ...

PEL: New D-Bus properties on PEL entry iface

Fill in the 4 newly added properties on the PEL entry D-Bus interface:
- Platform log ID (PLID)
- Deconfig flag from the SRC section
- Guard flag from th

PEL: New D-Bus properties on PEL entry iface

Fill in the 4 newly added properties on the PEL entry D-Bus interface:
- Platform log ID (PLID)
- Deconfig flag from the SRC section
- Guard flag from the SRC section
- Creation timestamp

These were also added to the PELAttributes map in the Repository class
so that each PEL wouldn't have to be reconstructed from a file again
when creating the D-Bus objects.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I7878645f56c634e6111fcecc22ab27673d0c0f5d

show more ...