catch exceptions as const

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I17001575bc409351f5187b1b269e66968d402c1c

Revert "meson: nlohmann/json has a meson build"

and Revert "meson: Rework valijson subproject dependency"

This reverts commit 9fedd141288d36d41345305532251e4659535751.
and e7948f447eaeb3afaa78d5d29

Revert "meson: nlohmann/json has a meson build"

and Revert "meson: Rework valijson subproject dependency"

This reverts commit 9fedd141288d36d41345305532251e4659535751.
and e7948f447eaeb3afaa78d5d298dc12b0730b8a84.

As they break the yocto build.

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: I8034a362f0866342a1527f468dc84a6ff1c67bd7

show more ...

FruUtils: Avoid truncation in derivation of fruAreaSize

FRU areas can be up to 255 * 8 bytes in length. 2040 can't be
represented in an unsigned 8-bit type, so widen the result type to
match fruBloc

FruUtils: Avoid truncation in derivation of fruAreaSize

FRU areas can be up to 255 * 8 bytes in length. 2040 can't be
represented in an unsigned 8-bit type, so widen the result type to
match fruBlockSize.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: Ie331abc2a50d1d91fced2f03d2ba3275331bdeaa

show more ...

FruUtils: Rework calculateChecksum() implementation

FRU areas are sized by a single-byte count field in units of FRU blocks,
which are 8 bytes in size. As such, it's possible for a FRU area to be
up

FruUtils: Rework calculateChecksum() implementation

FRU areas are sized by a single-byte count field in units of FRU blocks,
which are 8 bytes in size. As such, it's possible for a FRU area to be
up to 255 * 8 or 2040 bytes in size. An accumulation of values over such
an area to calculate the checksum can produce values exceeding 256. In
such a circumstance the current checksum implementation will mask a
negative value, invoking undefined behaviour.

Instead, accumulate into an unsigned, 8-bit type to avoid negative
masking through well-defined wrap behaviour.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: Ie85894004f1048413cddc62d68dd0b7ae8f6f541

show more ...

meson: nlohmann/json has a meson build

Exploit the usual configuration available for subprojects that support
meson directly.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I410cc14ca42

meson: nlohmann/json has a meson build

Exploit the usual configuration available for subprojects that support
meson directly.

Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
Change-Id: I410cc14ca426b1c4c26600f6a72a47aa4d944570

show more ...

Make build on clang

And support clang-tidy rules. The changes are pretty minimal, and were
all done by the clang robot.

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: I29501aa56de1cd63c

Make build on clang

And support clang-tidy rules. The changes are pretty minimal, and were
all done by the clang robot.

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: I29501aa56de1cd63cda233e06a07641458f89345

show more ...

FruDevice: fix possible resource leak

file left open when ioctl call fails.
close it before conitnue.

Signed-off-by: Zhikui Ren <zhikui.ren@intel.com>
Change-Id: I8f6c489dc5340bb8e67cd24f790feb8044

FruDevice: fix possible resource leak

file left open when ioctl call fails.
close it before conitnue.

Signed-off-by: Zhikui Ren <zhikui.ren@intel.com>
Change-Id: I8f6c489dc5340bb8e67cd24f790feb80445a27ed

show more ...

Require all probe interface properties to match

If multiple properties were provided for a probe interface rule, the
match would succeed if the last property (as iterated over by a map)
were a match

Require all probe interface properties to match

If multiple properties were provided for a probe interface rule, the
match would succeed if the last property (as iterated over by a map)
were a match. This commit changes the logic to require all properties to
match.

Tested: Added second (invalid) property value to baseboard.json
FruDevice rule, and verified that the baseboard was no longer presented
by entity-manager after making this change.

Signed-off-by: Jonathan Doman <jonathan.doman@intel.com>
Change-Id: I2b65d0bedb7e0f193c784cdb292653e00cbb34a3

show more ...

Fix dangling reference to scan count

count variable is automatic but was captured by reference in the deepest
nested lambda. This makes all the captures explicit in the post-scan
lambdas.

Tested: e

Fix dangling reference to scan count

count variable is automatic but was captured by reference in the deepest
nested lambda. This makes all the captures explicit in the post-scan
lambdas.

Tested: entity-manager presents the same items before and after this
change.

Signed-off-by: Jonathan Doman <jonathan.doman@intel.com>
Change-Id: I34bce907e43adb67067797c74b2feaf489d742dc

show more ...

Make a copy of the probe string to fix warning

The newest version of gcc uncovered a subtle bug where we were relying
on a reference to an rvalue in a loop. How this manifested was that
probeComman

Make a copy of the probe string to fix warning

The newest version of gcc uncovered a subtle bug where we were relying
on a reference to an rvalue in a loop. How this manifested was that
probeCommand was of type nlohmann::json, which has begin() and end()
overloads for std::string, but would return by value instead of by
reference to the nlohmann internals. Because of this, technically the
reference we were looking at is destroyed each iteration of the loop.

This never caused an issue in practice, as the lifetime of probeCommand
was always greater than probe, but it's good to not rely on undefined
behavior.

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: Ie89df5d69236331de68788a7f3c5af862160c069

show more ...

Check for errors before trying to open a bus

This is a fix to a regression introduced as part of
9b86787adea3f8f29fac2acbb9fa0f48fbcf244a
namely, the recursive iterator doesn't check for the existen

Check for errors before trying to open a bus

This is a fix to a regression introduced as part of
9b86787adea3f8f29fac2acbb9fa0f48fbcf244a
namely, the recursive iterator doesn't check for the existence of a bus
path before attempting to open it, which results in a crash when
std::filesystem throws an exception.

This was reported as part of bug #8

Tested:
Launched entity-manager in qemu, which was previously crashing and
observed "Unable to open path /sys/bus/i2c/devices/i2c-4" in the journal
log, with entity-manager staying running.

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: I638a66e271df0041d4df75be22eb03c064d9cf68

show more ...

entity-manager: allow non-numeric substitution for name

Currently EntityManager crashes if it can't parse template substitution
as numeric variable. This prevent to use test string substitutions the

entity-manager: allow non-numeric substitution for name

Currently EntityManager crashes if it can't parse template substitution
as numeric variable. This prevent to use test string substitutions there.

Tested: use string for template substitution in configuration name:
{
"Name": "$name",
"Probe": [
"com.yadro.HWManager.Fan({'part_number': 'ASMFAN781101A'})"
],
.....
}
Where name is string:
com.yadro.HWManager.Fan interface - - -
.name property s "Sys_Fan1" emits-change writable
.part_number property s "ASMFAN781102A" emits-change writable

Change-Id: I7139a51e9644dba857ff1c542cf127e723182375
Signed-off-by: Igor Kononenko <i.kononenko@yadro.com>
Signed-off-by: Andrei Kartashev <a.kartashev@yadro.com>

show more ...

Fix early exit for non D-Bus probes

The existing code had a bug where it would require a D-Bus interface to
be present for a probe in order to continue on to create any
configuration objects. This

Fix early exit for non D-Bus probes

The existing code had a bug where it would require a D-Bus interface to
be present for a probe in order to continue on to create any
configuration objects. This doesn't always work because some probe
statements, like "TRUE", don't need any D-bus interface checks to
succeed.

The fix is to continue on, and just pass an empty map of D-Bus
interfaces into the template substitution function.

This bug was introduced in my 'Do tmpl substitutions using all
properties on path" commit.

Tested:
Tested with a Probe: "TRUE" config and saw the configuration object for
it created on D-Bus.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: I672a5fe055801d8bd674778350ce8f434cc0c430

show more ...

Improve initialization of I2C sensors

After an AC cycle validation has witnessed some systems sensors are
missing. As Entity Manager begins the process of scanning for
sesnsors, and creating the har

Improve initialization of I2C sensors

After an AC cycle validation has witnessed some systems sensors are
missing. As Entity Manager begins the process of scanning for
sesnsors, and creating the hardware monitoring nodes, there are
occassionally some failures to correctly create the node. This
manifests itself by the 'dd' kernel driver emitting an -EBUSY error
message. Unfortunately the 'dd' driver also eats the error code, and
continues. This is by design.

This commit modifies how the nodes are initialized. The steps taken:
1. Determine if the node is already present
2. Create the node if it is not
3. Set a timer, to give the kernel time to create the node
4. For those sensors that create a "hwmon" subdir, search for that
directory after the timer elapses.
5. If the subdir is not present, delete the device, and try again by
initiating another timer.
6. Continue until the subdir exists, or a retry count expires.

Tested:
Ran AC cycles via a script.
After each cycle, wait for the SUT to DC on, and arrive at the EFI
Shell> prompt.
Issue "ipmitool sensor list", capturing the results
Search the list for all of the sensors that have been reported as
missing after AC cycles.

Change-Id: I118df674162677d66e7d211b089430fce384086b
Signed-off-by: Johnathan Mantey <johnathanx.mantey@intel.com>

show more ...

FruUtils: Add back the else in formatFRU.

There is no "else" in this if-statement and that causes the name very
wrong. The dbus path always be "/xyz/openbmc_project/FruDevice/UNKNOWN",
and the prope

FruUtils: Add back the else in formatFRU.

There is no "else" in this if-statement and that causes the name very
wrong. The dbus path always be "/xyz/openbmc_project/FruDevice/UNKNOWN",
and the properties of the dbus-object are FRU_CUSTOM_FIELD_NAME which is
"INFO_AM".

Tested:
After applying this patch, information of all FRU becomes normal.

Signed-off-by: Scron-Chang <Scron.Chang@quantatw.com>
Change-Id: I1e9ee88bbd1d02a1b96cf52a02250efb407cec58

show more ...

FruDevice: Define the day 1/1/96.

FruDevice should decide the first day of the month. Otherwise, the date on its
own dbus-object gets different from the result of ipmitool.

Tested:
After applying t

FruDevice: Define the day 1/1/96.

FruDevice should decide the first day of the month. Otherwise, the date on its
own dbus-object gets different from the result of ipmitool.

Tested:
After applying this patch, the date on the dbus-object is the same as what
ipmitool shows.

For example:
root@xxx:/usr/bin# busctl get-property xyz.openbmc_project.FruDevice \
> /xyz/openbmc_project/FruDevice/xxx xyz.openbmc_project.FruDevice \
> BOARD_MANUFACTURE_DATE
s "2020-09-24 - 06:25:00"
root@xxx:/usr/bin# ipmitool fru print 132 | grep "Date"
Board Mfg Date : Thu Sep 24 06:25:00 2020 UTC

Signed-off-by: Scron-Chang <Scron.Chang@quantatw.com>
Change-Id: Ic0ed37f18661bafff26676027dc5e16743968221

show more ...

Moved the common code and functions between
fru deamons in entitymanager.

Store the common functions/variable declarations
between multiple fru related deamons in the
FruUtils.cpp and FruUtils.hpp f

Moved the common code and functions between
fru deamons in entitymanager.

Store the common functions/variable declarations
between multiple fru related deamons in the
FruUtils.cpp and FruUtils.hpp files in entitymanager.

TESTED : Built Facebook YosemiteV2 images and loaded
on the target hardware. Verified all the fru's read and write.

Signed-off-by: Kumar Thangavel <thangavel.k@hcl.com>
Change-Id: I42129998603915af7a81f2bb8dff39dc198e8264

show more ...

FruDevice: Move verifyOffset to util

Moving verifyOffset module to fruUtils.cpp

Signed-off-by: Vijay Khemka <vijaykhemkalinux@gmail.com>
Change-Id: I8d8450783d421c1db384ca1332fa5aee90a41ced

Do tmpl substitutions using all properties on path

The current code would substitute the template arguments in the JSON
files, such as $bus, by only looking at the properties of the interface
that p

Do tmpl substitutions using all properties on path

The current code would substitute the template arguments in the JSON
files, such as $bus, by only looking at the properties of the interface
that passed the probe.

This commit will try to find a substitution using the properties of all
of the interfaces on the same object path as the interface that passed
the probe. It does this by reading and storing the properties of all
interfaces that came back from the existing GetSubTree call that finds
the paths of the probed interfaces, instead of just limiting it to the
interfaces that were probed. The template substitutions are then tried
on the properties of all interfaces until one is successful.

This change is being made so that the interface used in the probe
doesn't also need to contain properties for the device details, such as
the bus property, that otherwise wouldn't belong there. For example,
the com.ibm.ipzvpd.VINI interface, which models EEPROM contents, is
currently used in probes on IBM systems, and with this change the bus
and address properties can now be on a separate interface, such as the
recently proposed xyz.openbmc_project.Inventory.Decorator.I2CDevice.

Tested: Tested that the $bus template can be successfully filled in
with the Bus property from an interface other than the one the probe was
matched on.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Ic6f1539b38f6a4098f131d7f14cad6b6ddff041f

show more ...

Fix lambda capture in findDbusObjects

In findDbusObjects(), the interfaces variable was getting moved as part
of the async method callback creation, but was then used after that as
part of the async

Fix lambda capture in findDbusObjects

In findDbusObjects(), the interfaces variable was getting moved as part
of the async method callback creation, but was then used after that as
part of the async method call arguments. Change the lambda to capture
it by value so that it is still usable as an argument to the GetSubTree
D-Bus method call.

Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Ia81df828759292fac3157d37bf797c9d7c03c8f0

show more ...

FruDevice: Enforce FRU layout order

As per Section 17 of FRU specification, FRU information areas are
required to be in order in FRU data layout which means all offset
value should be in increasing

FruDevice: Enforce FRU layout order

As per Section 17 of FRU specification, FRU information areas are
required to be in order in FRU data layout which means all offset
value should be in increasing order or can be 0 if that area is
not present. Verifying these Fru layout order in early header
parsing.

Signed-off-by: Vijay Khemka <vijaykhemkalinux@gmail.com>
Change-Id: I612297df73ad3661074f16b2b6c1551dd4b055b6

show more ...

Checking nullptr of busmap[0]

If no device in i2cbus2 under /dev, busmap[0] would be nullptr.
So initialize busmap[0] for this case.

Change-Id: I019bd850474af6b1763b8b71c14d73a233fe881a
Signed-off

Checking nullptr of busmap[0]

If no device in i2cbus2 under /dev, busmap[0] would be nullptr.
So initialize busmap[0] for this case.

Change-Id: I019bd850474af6b1763b8b71c14d73a233fe881a
Signed-off-by: Helen Huang <he.huang@linux.intel.com>

show more ...

FruDevice: Check for language code

Currently we are supporting only english language for 8 bit ascii
string parsing. If language code detected as non english, we still
allows parsing for language in

FruDevice: Check for language code

Currently we are supporting only english language for 8 bit ascii
string parsing. If language code detected as non english, we still
allows parsing for language independent data like binary, bcd type
etc but for language dependent data with type code (11) mentioned
in section 13 of FRU specification, we are reporting this as an
error for non english language code. As per section 10, chassis
area info will always be in english that's why it doesn't have
any language bytes in it's field header.

Signed-off-by: Vijay Khemka <vijaykhemkalinux@gmail.com>
Change-Id: I3dec0d37110f877da48e484104281389d54047a8

show more ...

FruDevice: Verify offset overlap

Checking each field area offset if they are overlapping with
other field area offset.

Signed-off-by: Vijay Khemka <vijaykhemkalinux@gmail.com>
Change-Id: Id0fcb704a

FruDevice: Verify offset overlap

Checking each field area offset if they are overlapping with
other field area offset.

Signed-off-by: Vijay Khemka <vijaykhemkalinux@gmail.com>
Change-Id: Id0fcb704ac20083fae1760ec7d6e6cb559442526

show more ...

Emplace baseboardDev to I2C bus0

Fix issue:
FRU devices on I2C bus0 is replaced by baseboardDev,
So the FRU device on I2C bus0 could not be registed on entity
manager

Solution:
Use emplace() to add

Emplace baseboardDev to I2C bus0

Fix issue:
FRU devices on I2C bus0 is replaced by baseboardDev,
So the FRU device on I2C bus0 could not be registed on entity
manager

Solution:
Use emplace() to add the baseboardDev to I2C bus0.

Tested:
Tested on the platform with FRU device on I2C bus0,
Use 'busctl tree xyz.openbmc_project.FruDevice' and 'ipmitool fru'
to check FRU device,both baseboardDev and FRU device on I2C bus0
could be shown.

Signed-off-by: Helen Huang <he.huang@linux.intel.com>
Change-Id: I49d37212a1c044a59c8d81694534ab6a747c7d3a

show more ...