livetree: Add only new data to fixup nodes instead of complete regeneration

Removing the complete __fixups__ and __local_fixups__ tree might delete
data that should better be retained. See the added test for a situation
that was broken before.

Note that without removing /__fixups__ and /__local_fixups__ in
generate_fixups_tree() and generate_local_fixups_tree() respectively
calling build_and_name_child_node() isn't safe as the nodes might
already exist and then a duplicate would be added. So build_root_node()
has to be used which copes correctly here.

Fixes: 915daadbb62d ("Start with empty __local_fixups__ and __fixups__ nodes")
Closes: https://github.com/dgibson/dtc/issues/170
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@baylibre.com>
Message-ID: <b061ee57157fafbb9d5b9b0b86af760d13a04eda.1755512759.git.u.kleine-koenig@baylibre.com>
[dwg: Use -1 instead of 1 as an error return]
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
3 files changed
tree: b8f6610715f3b15a007b7c847b374bb2c7347a40
  1. .github/
  2. Documentation/
  3. libfdt/
  4. pylibfdt/
  5. scripts/
  6. tests/
  7. .cirrus.yml
  8. .clang-format
  9. .editorconfig
  10. .gitignore
  11. .gitlab-ci.yml
  12. BSD-2-Clause
  13. checks.c
  14. CONTRIBUTING.md
  15. convert-dtsv0-lexer.l
  16. data.c
  17. dtc-lexer.l
  18. dtc-parser.y
  19. dtc.c
  20. dtc.h
  21. dtdiff
  22. fdtdump.c
  23. fdtget.c
  24. fdtoverlay.c
  25. fdtput.c
  26. flattree.c
  27. fstree.c
  28. GPL
  29. livetree.c
  30. Makefile
  31. Makefile.convert-dtsv0
  32. Makefile.dtc
  33. Makefile.utils
  34. meson.build
  35. meson_options.txt
  36. pyproject.toml
  37. README.license
  38. README.md
  39. srcpos.c
  40. srcpos.h
  41. TODO
  42. treesource.c
  43. util.c
  44. util.h
  45. VERSION.txt
  46. version_gen.h.in
  47. yamltree.c
README.md

Device Tree Compiler and libfdt

The source tree contains the Device Tree Compiler (dtc) toolchain for working with device tree source and binary files and also libfdt, a utility library for reading and manipulating the binary format.

dtc and libfdt are maintained by:

Python library

A Python library wrapping libfdt is also available. To build this you will need to install swig and Python development files. On Debian distributions:

$ sudo apt-get install swig python3-dev

The library provides an Fdt class which you can use like this:

$ PYTHONPATH=../pylibfdt python3
>>> import libfdt
>>> fdt = libfdt.Fdt(open('test_tree1.dtb', mode='rb').read())
>>> node = fdt.path_offset('/subnode@1')
>>> print(node)
124
>>> prop_offset = fdt.first_property_offset(node)
>>> prop = fdt.get_property_by_offset(prop_offset)
>>> print('%s=%s' % (prop.name, prop.as_str()))
compatible=subnode1
>>> node2 = fdt.path_offset('/')
>>> print(fdt.getprop(node2, 'compatible').as_str())
test_tree1

You will find tests in tests/pylibfdt_tests.py showing how to use each method. Help is available using the Python help command, e.g.:

$ cd pylibfdt
$ python3 -c "import libfdt; help(libfdt)"

If you add new features, please check code coverage:

$ sudo apt-get install python3-coverage
$ cd tests
# It's just 'coverage' on most other distributions
$ python3-coverage run pylibfdt_tests.py
$ python3-coverage html
# Open 'htmlcov/index.html' in your browser

The library can be installed with pip from a local source tree:

$ pip install . [--user|--prefix=/path/to/install_dir]

Or directly from a remote git repo:

$ pip install git+git://git.kernel.org/pub/scm/utils/dtc/dtc.git@main

The install depends on libfdt shared library being installed on the host system first. Generally, using --user or --prefix is not necessary and pip will use the default location for the Python installation which varies if the user is root or not.

You can also install everything via make if you like, but pip is recommended.

To install both libfdt and pylibfdt you can use:

$ make install [PREFIX=/path/to/install_dir]

To disable building the python library, even if swig and Python are available, use:

$ make NO_PYTHON=1

More work remains to support all of libfdt, including access to numeric values.

Mailing lists