build-unit-test-docker: update to clang-15

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Id5532b96a7067ea831c3fcd67418a85314a3e840

build-unit-test-docker: update ubuntu to kinetic

The clang-15 packages are only going to be available in kinetic
or later. Switch the docker base to that distro version. That
version already has t

build-unit-test-docker: update ubuntu to kinetic

The clang-15 packages are only going to be available in kinetic
or later. Switch the docker base to that distro version. That
version already has the autoconf-archive required for C++20 and
it has gcc-12 by default, so we can remove (or comment for the time
being) some workarounds we had.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I6eb7fd7218027941af949522e043ed5dd6ea4f9b

show more ...

build-unit-test-docker: minor dependency tweaks

Add explicit dependency for objmgr -> phosphor-dbus-interfaces.
Make formatting more consistent.
Remove googletest dependencies when '-Dtests=disabled

build-unit-test-docker: minor dependency tweaks

Add explicit dependency for objmgr -> phosphor-dbus-interfaces.
Make formatting more consistent.
Remove googletest dependencies when '-Dtests=disabled' is passed.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I3a3d7aa290c850181a6c2edaeed1af7df8c44d26

show more ...

build-unit-test-docker: refresh package versions

Update various package versions based on the current
Yocto recipe equivalents for these packages.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz

build-unit-test-docker: refresh package versions

Update various package versions based on the current
Yocto recipe equivalents for these packages.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I12773a06d4d589285bb1477013278a9eae5b5901

show more ...

build-unit-test-docker: use alternative for cpp

We added an 'alternative' for GCC but not for the preprocessor.
It seems potentially problematic to use an older preprocessor,
which came with the dis

build-unit-test-docker: use alternative for cpp

We added an 'alternative' for GCC but not for the preprocessor.
It seems potentially problematic to use an older preprocessor,
which came with the distro, but a newer compiler, so set the
preprocessor as the same version as GCC. The preprocessor is
required to be installed as a separate alternative and not with
the "--slave" option.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I3a3e17b93ab7ea1aad0e5ac3677ec463e0c5b5ed

show more ...

Add python3 packaging module

As part of the effort to standardize our version comparison scheme,
https://gerrit.openbmc.org/c/openbmc/bmcweb/+/57407 needs this package
to be able to sort redfish sch

Add python3 packaging module

As part of the effort to standardize our version comparison scheme,
https://gerrit.openbmc.org/c/openbmc/bmcweb/+/57407 needs this package
to be able to sort redfish schemas at build time. This seems like a
pretty common and well supported package.

Add it to the docker image.

Tested: the aformentioned review passes CI when run locally.

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: Ifdb49cfacf4c6bf27fee882d08bdf469d32cc7a6

show more ...

qemu-docker: Update redfish version

The DMTF/python-redfish-library is updated and it raises
SessionCreationError for login errors with HTTP code 400, and only
raises InvalidCredentialError for HTTP

qemu-docker: Update redfish version

The DMTF/python-redfish-library is updated and it raises
SessionCreationError for login errors with HTTP code 400, and only
raises InvalidCredentialError for HTTP code 401.

The cases when user enter "empty" username or password will return 400
and thus change the expected exception to SessionCreationError.

This commit was released with `redfish` version 3.1.7.
[0]: https://github.com/DMTF/python-redfish-library/commit/9a32d80c1c0858f0360e8d1f5f322278de79a20f

Update the redfish version to trigger docker rebuild.

Tested:
Clean docker image and rebuild docker image, or use this commit
to update docker, make sure the redfish version update.

Use
https://gerrit.openbmc.org/c/openbmc/openbmc-test-automation/+/57366,
the case `Redfish_Login_With_Invalid_Credentials` passed.

Change-Id: I946f6f71c56ccc2af8648ebcd4e7ae42ceacad5f
Signed-off-by: Jian Zhang <zhangjian.3032@bytedance.com>

show more ...

build-unit-test-docker: Add function2 to stdplus

Header only upgrade for uses of std::function that can be converted.

Change-Id: I5f43a461c0446258c5456055569a6ed8c3c4d229
Signed-off-by: William A.

build-unit-test-docker: Add function2 to stdplus

Header only upgrade for uses of std::function that can be converted.

Change-Id: I5f43a461c0446258c5456055569a6ed8c3c4d229
Signed-off-by: William A. Kennington III <wak@google.com>

show more ...

clang-tidy: use TemporaryDirectory

This patch changes clang-tidy build of cmake and meson to use
TemporaryDirectory. So that no need to explicitly delete the build
directory. There is no need to har

clang-tidy: use TemporaryDirectory

This patch changes clang-tidy build of cmake and meson to use
TemporaryDirectory. So that no need to explicitly delete the build
directory. There is no need to hardcode the name as well. I kept the
prefix of the temp folder to be "build". This is to make gitignore of
most repos still work.

Tested:
1. used phosphor-user-manager as a typical meson repo. Tested
with/without clang-tidy. Both worked
2. used peci-pcie as a typical CMakeList repo. Tested with/without
clang-tidy. Both worked (but the repo itself doesn't pass clang build
bacause of its own reason)

Signed-off-by: Nan Zhou <nanzhoumails@gmail.com>
Change-Id: Ic355e390db7349ded73ad3d17a03a40b83460249

show more ...

clang-tidy: try to remove old directories

Just like the existing meson build codes, this patch tries to delete old
clang-tidy build directories before making a new one.

Tested:
Tested both cmake an

clang-tidy: try to remove old directories

Just like the existing meson build codes, this patch tries to delete old
clang-tidy build directories before making a new one.

Tested:
Tested both cmake and meson repos, old directories are deleted as
expected before making a new one.

Signed-off-by: Nan Zhou <nanzhoumails@gmail.com>
Change-Id: I36a484d5832a1f427ef0e1b9b1f28183ccdc387e

show more ...

build-unit-test-docker: add cli11 to mapper

One of the applications in the mapper repository is adding a dependency
on cli11.

Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Change-Id: Id5

build-unit-test-docker: add cli11 to mapper

One of the applications in the mapper repository is adding a dependency
on cli11.

Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
Change-Id: Id50b701bf94f46b08927beb0fc9d7e187e1f0c47

show more ...

build-unit-test-docker: update GCC to 12

Yocto is already using GCC 12. Upgrade for compatibility.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I0ab6e39efba280e4bf60305c9909ad35e

build-unit-test-docker: update GCC to 12

Yocto is already using GCC 12. Upgrade for compatibility.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I0ab6e39efba280e4bf60305c9909ad35e71caec1

show more ...

build-unit-test-docker: upgrade googletest to 1.12.1

This is the same version used in Yocto now.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I843a9da46b87ecc9363c2a9bbddfa22daab6

build-unit-test-docker: upgrade googletest to 1.12.1

This is the same version used in Yocto now.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I843a9da46b87ecc9363c2a9bbddfa22daab60b99

show more ...

build-unit-test-docker: remove libunifex

Most of the original authors of libunifex have moved in another
direction, so sdbusplus will not be using it anymore.

Signed-off-by: Patrick Williams <patri

build-unit-test-docker: remove libunifex

Most of the original authors of libunifex have moved in another
direction, so sdbusplus will not be using it anymore.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I22df03eb6ced5bfbd481aa2eb1eb76b750a06d9c

show more ...

Change eslint output to human readable

Today in the build logs, eslint prints a json payload of every file it
checked, regardless of whether there are errors present, which isn't
really human readab

Change eslint output to human readable

Today in the build logs, eslint prints a json payload of every file it
checked, regardless of whether there are errors present, which isn't
really human readable and clogs up log space.

This commit changes to the "stylish" output which is human readable.
Examples of which are in the eslint docs[1].

Tested: ran a build, and observed the json gone, and a much more
readable log present.

```
eslint . --ignore-path .gitignore --no-eslintrc -c /build/eslint-global-config.json --ext .json --format=stylish --resolve-plugins-relative-to /usr/local/lib/node_modules --no-error-on-unmatched-pattern
```

[1]: https://eslint.org/docs/latest/user-guide/formatters/#stylish

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: Ib13630bdce65351215324add4c9c3233fff06c61

show more ...

build-unit-test-docker: remove catch2

Based on searching the org in Github[1], there do not appear to be
anymore usages of Catch2. Remove it from our Docker image.

1. https://github.com/search?q=o

build-unit-test-docker: remove catch2

Based on searching the org in Github[1], there do not appear to be
anymore usages of Catch2. Remove it from our Docker image.

1. https://github.com/search?q=org%3Aopenbmc+catch2&type=code

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I2cf89c533149266aa908a15bd825bfdbcd450c55

show more ...

unit-test: cppcheck: use temporary directory

The previous code would create a directory named `cppcheck-temp`
in the root of the repository under test as the cppcheck cache but
never cleaned it up.

unit-test: cppcheck: use temporary directory

The previous code would create a directory named `cppcheck-temp`
in the root of the repository under test as the cppcheck cache but
never cleaned it up. This left stray directories around when using
the unit test locally.

Switch the mechanism to use tempfile.TemporaryDirectory so that
the directory is automatically cleaned up when the testing is
complete.

Tested: Ran unit-test locally and observed cppcheck-temp directory
no longer remains[1].

1. https://github.com/williamspatrick/dotfiles/blob/840250f0c56b299a8de47cc2ea7c558a14eeb517/env/30_linux/lfopenbmc.zsh#L76

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Idde8feb28bef663d156c5d1d1f21a3d8c629c72d

show more ...

Update meson to 0.63.0

Reason: meson 0.63.0 provides a new feature which allows user to apply
diff file on a subproject. This gives user a flexibility to have code
changes in subprojects while runni

Update meson to 0.63.0

Reason: meson 0.63.0 provides a new feature which allows user to apply
diff file on a subproject. This gives user a flexibility to have code
changes in subprojects while running the unit test.

Reference: https://mesonbuild.com/Wrap-dependency-system-manual.html#diff-files

Tested: Locally ran unit test and confirmed that the subproject applied
the diff files correctly.

Signed-off-by: Michael Shen <gpgpgp@google.com>
Change-Id: Icb1ad03d28496120f5afe8010cfa3a3b855975c8

show more ...

Enforce commit message style

Throughout the project, it is a manual human process to enforce the idea
of commit message formatting, and leads to more conflict than would
ideally be required for some

Enforce commit message style

Throughout the project, it is a manual human process to enforce the idea
of commit message formatting, and leads to more conflict than would
ideally be required for something that's relatively algorithmic, and
able to be enforced by CI. Jenkins is able to give faster response
times to users, thus ensuring that committers are more likely to be able
to resolve their commit message issues in a timely manner.

This commit adds the gitlint[1] application to our builds, and
integrates its checks with CI in the format-code.sh script. Gitlint
appears to be a relatively active project with a number of releases,
relatively up to date commits on its github, and by the documentation as
well as this authors testing, appears to do exactly what the project
needs in terms of checks.

gitlint has a number of configuration options[2], of which the defaults
appear to be ok for OpenBMCs style requirements. This commit checks in a
.gitlint file that was generated via 'gitlint generate-config' to use as
a starting point.

To avoid impacting the entire project at this time, this commit checks
for the existence of a .openbmc-enforce-gitlint file in the root of the
directory, and uses that to gate its scanning. At some point in the
future, once we've determined this meets our needs, this check will be
removed so that we can enforce this project-wide.

This commit makes use of the gitlint plugin system to support one
important feature that OpenBMC requires for block line length. The
custom line length rule allows:
1. Block comments to exceed the line length limit
2. Signed-Off-By sections to exceed the line length limit
3. Tabbed in sections to exceed the line length limit

Presumably this same mechanism could be used to handle openbmc/openbmc
commits, to allow meta-<name> to precede the title and go over the
allowed limit, but for the moment, format-code.sh does not apply to
openbmc/openbmc, so this would be for a future change to repotest

When these fails, it attempts to give the user a message that conveys
these allowals to let them fix their commit message quickly.

Tested:
Created a commit with a title that was too long, as well as added a
.openbmc-enforce-gitlint file in bmcweb. Ran openbmc-build-scripts and observed.

'''
-: UC1 Line exceeds max length (101>72).
It's possible you intended to use one of the following exceptions:
1. Put logs or bash lines in a quoted section with triple quotes (''') before and after the section
2. Put a long link at the bottom in a footnote. example: [1] https://my_long_link.com
Line that was too long:
: "VERY LONG LOG LINE THAT TAKES WAY TOO MUCH SPACE AND GOES OVER 72 CHARACTERS, which is a problem"
'''

[1] https://jorisroovers.com/gitlint/
[2] https://jorisroovers.com/gitlint/configuration/
[3] https://jorisroovers.com/gitlint/user_defined_rules/
[4] https://github.com/jorisroovers/gitlint/issues/255#issuecomment-1063494186

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: If42a22bfeca223fd5bc8f35ed937aa5f60713f2a

show more ...

Update robotframework-requests package

Changes:
- Requires this build script changes and code from test code
https://gerrit.openbmc.org/c/openbmc/openbmc-test-automation/+/56052
- Github

Update robotframework-requests package

Changes:
- Requires this build script changes and code from test code
https://gerrit.openbmc.org/c/openbmc/openbmc-test-automation/+/56052
- Github task openbmc/openbmc-test-automation#2105
- add robotframework-jsonlibrary package to replace deprecated
keyword To JSON in test suites and resource files.

Tested: HW Jenkins test CI, code update, ffdc and few other test suites
which are impacted due to this changes.

Change-Id: I73634fc3fd12ddd817d6016ebf156118d407fb24
Signed-off-by: George Keishing <gkeishin@in.ibm.com>

show more ...

spelling: add ignore list

Thu Ba reported that their name is getting rejected as a spelling
mistake in their Signed-off-by. Add capability to ignore words
from the primary dictionaries and add 'ba'

spelling: add ignore list

Thu Ba reported that their name is getting rejected as a spelling
mistake in their Signed-off-by. Add capability to ignore words
from the primary dictionaries and add 'ba' to the ignore list.

Tested:

Before change, manually running codespell results in:
```
./config/openbmc-spelling-ignore.txt:1: Ba ==> By, be
```
Manually running codespell with ignore-list option removes this.

Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I33134e519e28024926b6da00340d2d52e2e884bd

show more ...

scripts/unit-test.py: Fix an indent issue

The indent is incorrect and is fixed by autopep8

Signed-off-by: Lei YU <yulei.sh@bytedance.com>
Change-Id: I2d91c2bd3c32e2fc495eca137406fe234726a105

stdplus: Depend on googletest for support lib

This now provides some additional test case routines that require
googletest to build

Change-Id: I191705fa1b65e731308946c1eb72edc09b189bee
Signed-off-b

stdplus: Depend on googletest for support lib

This now provides some additional test case routines that require
googletest to build

Change-Id: I191705fa1b65e731308946c1eb72edc09b189bee
Signed-off-by: William A. Kennington III <wak@google.com>

show more ...

Improve cppcheck results and calling code

In the ideal case, cppcheck would have an option where it could be
enforcing in the future. This commit works toward that goal, but
doesn't quite get there

Improve cppcheck results and calling code

In the ideal case, cppcheck would have an option where it could be
enforcing in the future. This commit works toward that goal, but
doesn't quite get there.

First, it updates cppcheck to the latest copy from github, instead of
pulling through apt sources. This is done because the current cppcheck
has an issue with std::string_view, and erroneously believes that it
should be passed by reference. The master branch has a fix for this.

Next, it modifies the call of cppcheck in a couple ways, first, to rely
on compile_commands.json from the build step, rather than attempting to
build a list of files itself. This is less error prone, and allows
"whole binary" scans with cppcheck, which leads to fewer false
positives. On the negative side, this means that non-meson/non-cmake
projects will no longer get scanned, which, considering that meson is
what we're converging on, and for any given project a meson port should
probably take priority over cppcheck improvements, this seems
reasonable.

Because of the previous change, the use of popen and communicate() can
be replaced with our more normal methods for calling commands, which
cleans up the code quite a bit.

Adds a cppcheck-temp folder as recommended by the cppcheck docs for
storing intermediate results. This decreases the amount of time
cppcheck takes to run when scanning the same code multiple times.

After that, some argument changes are done to suppress checks that are
noisy, and the project doesn't currently have a way to make pass.

useStlAlgorithm this is just a suggestion, and not always correct, so
it's difficult to make it enforcing. Also, as a project, we tend to not
optimize for std::algorithms as much as we could. At some point in the
future we could turn this back on and make the project have a stronger
opinion on this style, but for the moment, this seems like something to
tackle later.
unusedStructMember this appears to either have bugs, or not understand
when a public struct member is used outside of the compile unit. It
leads to significant noise for things that are actually used, but get
marked unused because they're not used in every compile unit.
postfixOperator use of prefix increment indead of postfix on iterators
is not generally something we enforce on the project, and doesn't seem
like it improves in either the readability or the code, as well as leads
to a lot of noise. Disable for the moment.
unreadVariable Similar to unusedStructMember, this check seems to not be
able to see all possible ifdefs in a section of code, and leads to many
false positives.
knownConditionTrueFalse is supressed because apparently cppcheck isn't
able to read macros, so this flags significant portions of code behind
ifdefs. More investigation is warranted here, but for the moment,
disabling the check gets us closer to the goal

A patch series getting bmcweb code to the point where it's 100% passing
is here, if you want to see the types of changes that cppcheck proposes.
For all the non-disabled checks, cppcheck seems accurate, with no false
positives that I could find, although some slightly pedantic
suggestions.

https://gerrit.openbmc.org/c/openbmc/bmcweb/+/55001

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: I7730cc69ad1817cb7d75c04be3b7d1da6c1030d4

show more ...

Update boost to 1.79

Same as yocto is using.

Tested: Built dbus-sensors uring patches (which require 1.79) and they
built correctly.

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: Ic83a

Update boost to 1.79

Same as yocto is using.

Tested: Built dbus-sensors uring patches (which require 1.79) and they
built correctly.

Signed-off-by: Ed Tanous <edtanous@google.com>
Change-Id: Ic83a5ccb940fbc3bce4f8abedce1087e5cdfc102

show more ...