The Ubuntu runner was not building the yaml support as it's using Ubuntu
22 (jammy) which uses libyaml 0.2.2, but the build requires libyaml
0.2.3. Switch to Ubuntu 23 which has libyaml 0.2.5.
This was not detected by the runner as the Yaml feature defaults to
"auto" which turns off if it fails to find the dependency. In the runner
force yaml to enabled so if it fails to build it will trigger a build
failure.
We also force python support for the same reason.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Instead of running run-tests on all tests, split them down into the 9
separate run-tests test types. This provides better granularity of test
results from the Meson test harness.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
If the tests are run without a full compile they will fail. For example
with the following.
> rm -rf build/
> meson setup build/
> meson test -C build/
This is because the tests rely on the devicetree tools and test
executables.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The python library requires libfdt to build. This would intermittently
fail depending on what order targets compiled.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The test for get_mem_rsv fails on newer versions of Python with the
following error.
> Traceback (most recent call last):
> File "/__w/dtc/dtc/tests/pylibfdt_tests.py", line 421, in testReserveMap
> self.assertEqual([ 0xdeadbeef00000000, 0x100000],
> AssertionError: Lists differ: [16045690981097406464, 1048576] != [0, 16045690981097406464, 1048576]
>
> First differing element 0:
> 16045690981097406464
> 0
>
> Second list contains 1 additional elements.
> First extra element 2:
> 1048576
>
> - [16045690981097406464, 1048576]
> + [0, 16045690981097406464, 1048576]
> ? +++
It appears this is because the PyTuple_GET_SIZE() function that was used
to build the fdt_get_mem_rsv() return value has changed. It now is
returning a non-zero value when it's passed an integer, which causes the
SWIG wrapper to append the returned arguments to the return error rather
then ignore them.
This is valid behaviour per Python's documentation, which says it will
"Return the size of the tuple p, which must be non-NULL and point to a
tuple; no error checking is performed"[1]. As passing an integer is not
a tuple, its return value is undefined.
Fix this issue on older and newer versions by avoiding
PyTuple_GET_SIZE() entirely. Always append the arguments to the list,
and instead use the wrapper python function to check the first argument
and then splice the last two arguments as the return value.
[1] https://docs.python.org/3/c-api/tuple.html#c.PyTuple_GET_SIZE
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
A phandle in an overlay is not supposed to overwrite a phandle that
already exists in the base dtb as this breaks references to the
respective node in the base.
So add another iteration over the fdto that checks for such overwrites
and fixes the fdto phandle's value to match the fdt's.
A test is added that checks that newly added phandles and existing
phandles work as expected.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Message-ID: <20240225175422.156393-2-u.kleine-koenig@pengutronix.de>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The last commit, displaying the meson testlog if we fail tests had the
accidentaly side effect of not propagating the failure to show up properly
in github's dashboard. Fix that.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Currently the way we integrate the existing tests into meson means meson
test itself doesn't show any detailed logs of the failures, those just go
to the meson testlog file. As a hack to see what's failing in CI builds,
display that file if the tests fail.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Setting -D__USE_MINGW_ANSI_STDIO=1 is wrong and should not be used. MinGW
internally uses a macro to select between gnu_printf and printf. Just use
that instead of using a wrong format under clang backends.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
In a couple of places in fdt_overlay.c we need to adjust a phandle value
in a property (either a node's phandle itself or a reference) by some
delta. Currently this is done if a fairly convoluted way, open-coding
loading the value and handling of a non-aligned reference, and then using
fdt_setprop_inplace_partial() to replace the value. This becomes much
simpler if we use fdt_getprop_w() to get a writable pointer to the value
then we can just load/store it with fdt32_{ld,st}().
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Instead of creating 2 libraries manualy, just call both_libraries and
link to the appropriate one as requested.
Fixes compilation when passing -Ddefault_libraries=both as the
static_library name is duplicated.
Signed-off-by: Rosen Penev <rosenp@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
On systems that do not use the FHS, such as NixOS, the `install` program is not
located in `/usr/bin/` as its location is dynamic.
`dtc` can be easily installed on such systems by using the `install` program
available in the `$PATH` with:
make PREFIX=… INSTALL=install
However, this becomes more difficult when `dtc` is being compiled as part of a
larger toolchain, as the toolchain build scripts will not spontaneously pass
such an argument on the command line. This happens for example when `dtc` is
build as a part of the RTEMS build system.
By not hardcoding a predefined path for `install`, as is done for other
executables, `dtc` will allow the one in the `$PATH` to be used.
Signed-off-by: Samuel Tardieu <sam@rfc1149.net>
Message-ID: <20240208191405.1597654-1-sam@rfc1149.net>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
If default_library is set to static, the libfdt target (which just
uses library()) is already static, so we should just use that.
This fixes this Meson error:
libfdt/meson.build:37:11: ERROR: Tried to create target "fdt", but a target of that name already exists.
Signed-off-by: Alyssa Ross <hi@alyssa.is>
Message-ID: <20240123130742.185409-1-hi@alyssa.is>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
By convention, the PKG_CONFIG environment variable is used to tell
build systems which pkg-config executable should be used. This is
often used when cross compiling, where it might be set to something
like "aarch64-unknown-linux-gnu-pkg-config".
Signed-off-by: Alyssa Ross <hi@alyssa.is>
Message-ID: <20240123130409.181128-2-hi@alyssa.is>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
When building from shallow clone, tag is not available
and version defaults to git hash.
Problem is that some builds check DTC version and fail the comparison.
Example is https://git.trustedfirmware.org/TF-A/trusted-firmware-a.git
Which fails to build with following error:
dtc version too old (039a994), you need at least version 1.4.4
Drop --always from git describe command, see
https://github.com/mesonbuild/meson/blob/1.3.0/mesonbuild/utils/universal.py#L773
This will make it more closer to build via Makefile.
Signed-off-by: Peter Marko <peter.marko@siemens.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
When the DTS output has no type markers, we have to guess the type. Prior
to commit 32b9c61307 ("Preserve datatype markers when emitting dts
format"), instances of string lists would be delimited. Since then, a
single string with embedded "\0"s are emitted. An embedded "\0" is valid
for DTS files, but that's a rare exception and lists of strings are the
overwhelming majority. Restore the prior behavior.
stringlist.dts is reused for testing this, but needs a couple of tweaks
in order to match the dts output.
Reported-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Rob Herring <robh@kernel.org>
Message-ID: <20231027142901.2536622-1-robh@kernel.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Reject empty paths and negative lengths, according to the DT spec v0.4:
The convention for specifying a device path is:
/node-name-1/node-name-2/node-name-N
The path to the root node is /.
This prevents the access to path[0] from ever being out-of-bounds.
Signed-off-by: Pierre-Clément Tosi <ptosi@google.com>
Message-ID: <20231010092822.qo2nxc3g47t26dqs@google.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Ensure that the alias found matches the device tree specification v0.4:
Each property of the /aliases node defines an alias. The property
name specifies the alias name. The property value specifies the full
path to a node in the devicetree.
This protects against a stack overflow caused by
fdt_path_offset_namelen(fdt, path, namelen)
calling
fdt_path_offset(fdt, fdt_get_alias_namelen(fdt, path, namelen))
leading to infinite recursion on DTs with "circular" aliases.
This fix was originally written by Mike McTernan for Android in [1].
[1]: 9308e7f977
Signed-off-by: Pierre-Clément Tosi <ptosi@google.com>
Acked-by: Mike McTernan <mikemcternan@google.com>
Message-ID: <20231010092725.63h7c45p2fnmj577@google.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Boolean properties are unusual in that their presense or absence
indicates the value of the property. This makes them a little painful to
support using the existing getprop() support.
Add new methods to deal with booleans specifically.
Signed-off-by: Simon Glass <sjg@chromium.org>
Message-ID: <20230912182716.248253-1-sjg@chromium.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This fixes the output of the `dtc --help` command as the last few
entries were offset by one.
Signed-off-by: Charles Perry <charles.perry@savoirfairelinux.com>
Message-ID: <20230904143104.1941715-1-charles.perry@savoirfairelinux.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
According to the documentation, the function should default to the very
common property name <reg> when none is "specified". However, neither
passing NULL (ends up calling strlen(NULL) and segfaults) nor ""
(appends a property with an empty name) implements this behavior.
Furthermore, the test case supposed to cover this default value actually
passes the value to the function, somewhat defeating its own purpose:
/* 2. default property name */
// ...
err = fdt_appendprop_addrrange(fdt, 0, offset, "reg", addr, size);
if (err)
FAIL("Failed to set \"reg\": %s", fdt_strerror(err));
check_getprop_addrrange(fdt, 0, offset, "reg", 1);
Finally, nothing in the implementation of the function seems to attempt
to cover that use-case.
As the feature can't ever have been used by clients and as the resulting
reduced readability of the caller seems (IMO) to outweigh any potential
benefit this API would bring, remove the erroneous documentation instead
of trying to fix the function.
Reported-by: Mostafa Saleh <smostafa@google.com>
Signed-off-by: Pierre-Clément Tosi <ptosi@google.com>
Message-ID: <20230831123918.rf54emwkzgtcb7aw@google.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The Meson build system requires that the generated shared library uses
the libfdt.so.X.Y.Z naming scheme. But the Makefile is generating
libfdt-X.Y.Z.so. We want to keep the output of both systems the same to
avoid issues, so we rename the Makefile to match Meson.
Additionally, Meson generates the base "libfdt.so -> libfdt.so.1"
symlink which the Makefile hasn't been doing, add that as well.
This shouldn't impact existing users as the linker should be looking for
libfdt.so.1 which won't change and will still point to the correct file.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The SETUPTOOLS_SCM_PRETEND_VERSION was needed as ./setup.py would fail
without it. As setuptools_scm will fail if there is not git repo, and
the github workflow container does not include the source code git repo.
A previous commit added "fallback_version" to setuptools_scm which
instructs it to use the version from VERSION.txt when the git repo is
missing. So this hack is no longer needed.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
When building pylibfdt from the released tarballs[1] setup.py will fail
with the following.
> LookupError: setuptools-scm was unable to detect version for dtc.
> Make sure you're either building from a fully intact git repository or
> PyPI tarballs. Most other sources (such as GitHub's tarballs, a git
> checkout without the .git folder) don't contain the necessary metadata
> and will not work.
seutptools_scm supports a 'fallback_version' that will use the provided
string as the version when the git repo is not available.
[1] https://www.kernel.org/pub/software/utils/dtc/dtc-1.7.0.tar.xz
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
To synchronize the release version of the Makefile and Meson build
systems, pull the version info from a shared file.
Meson requires that the shared library version follow the X.Y.Z
numbering scheme. But the Makefile supported building shared libraries
with an EXTRAVERSION appended, for example X.Y.Z-rc1. We want to keep
the output of both systems the same to avoid issues, so we drop support
for the Makefile EXTRAVERSION.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
These symbols were not added to the version script when they were added
to libfdt.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This file is indented with tabs, but editorconfig defaults all files to
spaces.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
When running under Meson, check_tests() is generating dtb build files in
the source directory. This is because dtb is named by appending
".test.dtb" to the full source file name.
Use basename to extract just the source filename and write it to the
working directory which is the build directory.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Fixes the following warnings
> tests/meson.build:123: WARNING: Project targets '>=0.56.0' but uses
feature deprecated since '0.55.0': ExternalProgram.path. use
ExternalProgram.full_path() instead
> tests/meson.build:124: WARNING: Project targets '>=0.56.0' but uses
feature deprecated since '0.56.0': meson.source_root. use
meson.project_source_root() or meson.global_source_root() instead.
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Fixes the following warning
> pylibfdt/meson.build:2: WARNING: Project targets '>=0.56.0' but uses
feature deprecated since '0.55.0': ExternalProgram.path. use
ExternalProgram.full_path() instead
Do not use full_path() as suggested. setup_py is being called as a
command by custom_target() which understands how to properly inherit the
object returned by find_program().
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Set the minimum required version of Meson based on the highest version
feature used, as detected by meson-setup.
* 0.56.0: {'meson.project_build_root'}
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Build the libfdt with the correct version number by pulling the version
from the top-level project.
Change as suggested from https://github.com/dgibson/dtc/pull/95#issuecomment-1546933095
Signed-off-by: Brandon Maier <brandon.maier@collins.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The new meson build system support diverges from former build system
where the tests were not built until required.
This has caused an issue in NixOS[1] due to broken build of tests in
Darwin platform, so this patch allows the control if tests should be
build or not.
1. https://github.com/NixOS/nixpkgs/pull/235210
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
When we only need the libfdt calling the target install-lib also builds the
executables listed in $(BINS) because this target depends on all
Instead lets make install-lib only depend on libfdt.
Signed-off-by: Nicolas Escande <nico.escande@gmail.com>
Message-ID: <20230726144336.677135-1-nico.escande@gmail.com>
Reviewed-by: Simon Glass <sjg@chromium.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The value passed to the <ctype.h> functions shall be the value of an unsigned
char or EOF. It is implementation-defined if the char type is signed or
unsigned. Cast to unsigned char to avoid undefined behaviour on systems where
char is signed.
This cast is already present in other parts of the code base.
Signed-off-by: Sebastian Huber <sebastian.huber@embedded-brains.de>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The function mentioned in the comment, fdt_finished(), should be
changed to fdt_finish().
Signed-off-by: Yan-Jie Wang <yanjiewtw@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
-undefined error is the equivalent of --no-undefined for the macOS
linker, but -undefined would also be understood as a valid argument for
GNU ld so we use the supported linker variant.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
This records detailed usage of labels in a dtb. This is needed in
overlays (and enabled implicitly for these). For ordinary device trees
it can be used to restore labels when compiling back to dts format.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Message-Id: <20230523080941.419330-1-u.kleine-koenig@pengutronix.de>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
set SETUPTOOLS_SCM_PRETEND_VERSION="0" variable because GitHub Actions
does not copy the .git directory into the container. Without that, the
build fails with the following error
LookupError: setuptools-scm was unable to detect version for /__w/dtc/dtc.
Signed-off-by: Biswapriyo Nath <nathbappai@gmail.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
The handling of "type preservation" dts output is based on the idea of
"phandles with arguments" in properties, which isn't really a thing, other
than a fairly common convention about how bindings are written. There's
nothing preventing a binding which freely mixes phandles and other integers
in an array of cells.
Currently write_propval() handles this incorrectly: specifically the case
of a phandle which follows a regular integer in a 32-bit cell array, but
without a new '< >' delimited causing an extra TYPE_UINT32 marker to be
inserted. In this case it omits the necessary space between the integer
and the phandle reference, leading to output which can't be sent back into
dtc and parsed.
Correct this, and update tests to match. I think this is more or less
correct for now, but really write_propval() is a big mess :(.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
* bad-graph-child-address.dts: additional child address test since the
one in bad-graph.dts is now shadowed by its prerequisites also
failing.
* bad-graph-reg-cells.dts: test warnings produced by check_graph_reg().
Signed-off-by: Johannes Beisswenger <johannes.beisswenger@cetitec.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
If an endpoint node has a 'reg' property which consists of more than
one cell (4 bytes) and given that matching '#address-cells' and '#size-cells'
properties are specified on the port node an assertion is triggered in
check_graph_child_address() before the relevant diagnostic checks in
check_graph_reg() (called by check_graph_port() and check_graph_endpoint()) are executed.
The issue is fixed by making graph_child_address depend on the
graph_port and graph_endpoint checks.
Additionally the assertion can also be triggered if the length of the
'reg' property is less than 4 bytes e.g. by specifying
'reg = "a";'. In that case however other warnings are produced
highlighting the malformed property before dtc crashes.
Example dts file triggering the issue:
/dts-v1/;
/ {
bar: bar {
port {
bar_con: endpoint {
remote-endpoint = <&foo_con>;
};
};
};
foo {
port {
#address-cells = <1>;
#size-cells = <1>; // should always be 0
foo_con: endpoint@1 {
reg = <1 2>; // causes assertion failure instead of diagnostic
remote-endpoint = <&bar_con>;
};
};
};
};
Signed-off-by: Johannes Beisswenger <johannes.beisswenger@cetitec.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
(Free) Travis-ci is gone since November 2020, cirrus seems to be
unreliable showing build breakages unrelated to changes.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
libyaml before 0.2.3 expects non-const string parameters. Supporting
both variants would require either cpp magic or ignoring
"discarded-qualifiers" compiler warnings. For the sake of simplicity
just support libyaml 0.2.3 and newer.
Note that NO_YAML can be overwritten on the make command line.
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Brandon Maier