Change the output during CMake configure from
Devicetree configuration written to .../devicetree.conf
Parsing /home/ulf/z/z/Kconfig
Loaded configuration '.../frdm_kw41z_defconfig'
Merged configuration '.../prj.conf'
Configuration saved to '.../.config'
to
Devicetree header saved to '.../devicetree_unfixed.h'
Parsing /home/ulf/z/z/Kconfig
Loaded configuration '.../frdm_kw41z_defconfig'
Merged configuration '.../prj.conf'
Configuration saved to '.../.config'
Kconfig header saved to '.../autoconf.h'
devicetree_unfixed.h is more useful to be aware of than devicetree.conf
(still hoping we can get rid of it at some point), and seeing the
Kconfig header clarifies things to.
"Saved" instead of "written" for consistency. Maybe it could say
"Kconfig configuration" instead of "configuration" too, though it gets a
bit spammy in other contexts where the message shows up.
Updates Kconfiglib (and menuconfig/guiconfig, just to sync) to upstream
revision 061e71f7d7, to get this commit in, which is used to print the
header path:
Return a message from Kconfig.write_autoconf()
Like for Kconfig.write_config() and Kconfig.write_min_config(),
return a string from Kconfig.write_autoconf() with a message saying
that the header got saved, or that there were no changes to it. Can
be handy in tools.
Also make the "no change" message for the various files more
specific, by mentioning what type of file it is (configuration,
header, etc.)
Return True/False from Kconfig._write_if_changed() to indicate if
the file was updated. This also allows it to be reused in
Kconfig.write_min_config().
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update menuconfig (and kconfiglib and guiconfig, just to sync) to
upstream revision 424d0d38e7, to get this commit in, which works around
a crash on some macOS Python installations.
menuconfig: Work around crash on resize on some macOS systems
get_wch() has started raising curses.error when resizing the
terminal on some macOS Python installations. Work around for now by
falling back on getch(), which still works.
See https://github.com/ulfalizer/Kconfiglib/issues/84. Needs more
investigation, but I don't have a Mac handy.
Based on https://github.com/ulfalizer/Kconfiglib/pull/85, but with
some more comments.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib to upstream revision 9c0b562c94 to get this commit in:
Add Kconfig.__init__() helper flag for suppressing tracebacks
Tools that don't use standard_kconfig() currently generate spammy
tracebacks for e.g. syntax errors.
Add a suppress_traceback flag to Kconfig.__init__() for catching
"expected" exceptions and printing them to stderr and exiting with
status 1. Use it to make all tools consistently hide tracebacks.
Use the new flag to hide tracebacks for expected exceptions in
kconfig.py, lint.py, and genrest.py.
Some menuconfig robustness tweaks for wonky terminals are included as
well, and a new feature for customizing .config and autoconf.h header
comments via environment variables.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib, menuconfig, and guiconfig to upstream revision
faa1d21998, mostly to get this commit in:
Add public helpers for generating "<name> (defined at ...)" strings
Have Symbol/Choice.name_and_loc return strings like
"MY_SYM (defined at foo:1, bar:2)"
"<choice> (defined at foo:4)"
I've added a function like that in at least four different scripts
now, so that's probably a sign that it's a worthwhile helper.
Clean up the tests/Klocation tests a bit while adding tests.
Use the new helper to simplify kconfig.py a bit. Also clean it up a bit
by removing some unused stuff.
Some other minor improvements are included as well, e.g. to make
menuconfig/guiconfig give more helpful errors on invalid arguments.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Prompted by an upstream bug report. Nothing in Zephyr triggers this at
the moment, but might as well fix it.
Update Kconfiglib to upstream revision 7d05084b7e, to get this commit
in:
Fix handling of parentheses in macro argument values
As an oversight, there was no check for nested parentheses in macro
arguments, making the preprocessor think the call ended after
'void)' in
def_bool $(success,echo 'void foo(void) { asm inline (""); }' \
| $(CC) -x c - -c -o /dev/null)
This broke the latest linux-next kernels (with a Kconfig error),
starting with commit eb111869301e1 ("compiler-types.h: add
asm_inline definition").
I remember seeing this when going through the C code, but somehow
forgot to put it in. Fix it, and clean up _expand_macro() a bit at
the same time.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib (and menuconfig/guiconfig, just to sync) to upstream
revision f2ce282eca, to get this commit in:
Allow preprocessor user functions to access the parsing location
Just requires making Kconfig.filename/linenr public.
'lineno' would be a more standard name, but be consistent with
MenuNode.linenr.
This can be used to give friendly errors in
scripts/kconfig/kconfigfunctions.py, e.g. for
https://github.com/zephyrproject-rtos/zephyr/pull/18752.
Some minor optimizations are included too.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Similar deal to commit cc14c40a2d ("kconfiglib: Unclutter symbol
strings, avoid redundant writes, misc.").
Hide the direct dependencies in the defaults, selects, and implies
sections. Do the same in menuconfig/guiconfig as well.
This uses a new Kconfiglib API, so update Kconfiglib to upstream
revision 164ef007a8. This also includes some minor optimizations and
cleanups.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update kconfiglib, menuconfig, and guiconfig to upstream revision
5c904f4549 to get various improvements and fixes in:
- Marc Herbert found an issue involving symlinks, absolute paths,
and rsource that could lead to files not being found. The root cause
was relpath() assuming that symlink/../bar is the same as bar/, which
isn't guaranteed.
Fix it by handling paths in a simpler, more textual way.
- Propagated dependencies from 'depends on' are now stripped from
properties when symbols are printed (e.g. in information dialogs and
generated documentation).
The printed representation now also uses shorthands.
Before:
config A
bool
prompt "foo" if C && D
default A if B && C && D
depends on C && D
After:
config A
bool "foo"
default A if B
depends on C && D
- Before writing a configuration file or header, Kconfiglib now
compares the previous contents of the file against the new contents,
and skips the write if there's no change. This avoids updating the
modification time, and can save work.
A message like "No change to '.config'" is shown when there's no
change.
- .config now has '# end of <menu>' comments to make it easier to see
where a menu ends. This was taken from a change to the C tools.
- load_config() and write_(min_)config() now return a message that can
be printed with print(kconf.load_config()). This allows messages to
be reused in e.g. the configuration interfaces (nice now that there's
also a "No change to..." string).
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update menuconfig (and Kconfiglib, just to sync) to upstream revision
7e0b9f7ae1, to get these commits in:
menuconfig: Improve space/enter behavior slightly
Space toggles value is possible, and enters menus otherwise. Enter
works the other way around.
Make this explicit in the code, which also fixes some corner cases,
like space doing nothing on a y-selected menuconfig symbol.
-------------------------------------------------------------------
menuconfig: Fix display issue for unsatisfied-deps selected symbol
with children
A symbol with unsatisfied direct dependencies can end up with visible
children in an implicit submenu if it is selected (though that
generates a warning), so the optimization in _shown_nodes() isn't
safe, and causes the child nodes to not be shown outside show-all
mode.
Just remove the optimization. Trying things out some more,
everything's plenty fast enough anyway.
Checking the direct dependencies of the parent instead would be safe.
The menu path now says "(Top)" instead of "(top menu)" too, which is a
bit more compact. Make the same change in genrest.py.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib and menuconfig to upstream revision 90c5573c19, to get
these commits in:
Warn for unquoted argument to 'source', etc.
Print a warning suggesting to add quotes for things like
source foo/bar/Kconfig
menu title
prompt unquoted
Example warning:
Kconfig:32: warning: style: quotes recommended around
'lib/Kconfig.debug' in 'source lib/Kconfig.debug'
That quoteless syntax is supported for compatibility with old
versions of the C tools. It only works for a single word.
==================================================================
menuconfig: Include all parents in menu paths
Previously, symbols not defined with 'menuconfig' with children
weren't listed in the children's menu paths. It was deliberate, but
it's probably an anti-feature in retrospect, because it can make it
harder to find stuff by following the menu path.
Don't try to be clever and just list all the parent nodes in the
menu path.
==================================================================
menuconfig: Fix display issue for optional-prompt menuconfigs
_shown_nodes() needs to check whether invisible 'menuconfig' symbols
with optional prompts have visible children, so that they can be
shown outside show-all mode. Previously, only 'config' symbols were
checked.
==================================================================
menuconfig: Remember last saved/loaded path and improve
_conf_changed
Remember the last path that was manually saved/loaded instead of
reverting back to standard_config_filename() (e.g. .config).
Remember the path to the last saved minimal configuration separately
as well.
Also improve the _conf_changed behavior when loading a .config
within the interface. Instead of always treating it as needing to be
saved, check if it's outdated, like for the .config file loaded on
startup.
Also make the exit message ("No changes to save", etc.) always
include the target .config file, which is helpful. Previously, only
the save message did.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Sigvart pointed out that the error message for missing Kconfig files
(which rambles on a lot about $srctree) can easily kick people off in
the wrong direction.
Update Kconfiglib to upstream revision 9f26eb3ae3 to add the following
commit, which makes the error message point to stuff that's more likely
to be the problem:
Give more helpful error messages when files are missing
The current error message talks a lot about $srctree, but $srctree
is seldom the culprit in practice. More common is 'source
"$(SOME_ENV_VAR)/foo"`, where SOME_ENV_VAR hasn't been set.
Include the complete 'source ...' line for missing Kconfig files,
and mention unset environment variables as a hint. Only mention
$srctree briefly.
Also shorten the message when a .config can't be a found a bit. This
message would usually only be seen when working directly with the
library.
Example error:
Kconfig:7: '/Kconfig' not found (in 'source
"$(SOME_ENV_VAR)/Kconfig"'). Check that environment variables are
set correctly (e.g. $srctree, which is unset or blank). Also note
that unset environment variables expand to the empty string.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib (and menuconfig, just to sync) to upstream revision
99a7af769352b, to add the commit below, for a doc issue reported by
Sebastian Bøe:
Document that kconfig_filenames keeps absolute paths as-is
Came up in https://github.com/ulfalizer/Kconfiglib/issues/67.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update menuconfig and Kconfiglib to upstream revision 16539b4f223fa, to
add the commit below.
menuconfig: Prompt for save if a different .config would be saved
Previously, menuconfig.py only prompted for saving the configuration
if .config didn't exist or the user changed symbol values within the
interface.
Also make it prompt for save if Kconfig symbols have been added,
removed, or have had their defaults changed, provided it would make
the saved .config differ from the loaded one.
This usually won't matter for correctness, because loading an
outdated configuration performs an implicit olddefconfig, but it's
less confusing.
Also add a Kconfig.missing_syms attribute that records all
assignments to undefined symbols in the most recently loaded .config
file. This is needed to implement the check for whether the saved
.config would be different.
As an unrelated change, always prompt for saving if a .config has
been loaded from within the menuconfig interface. The intention is
probably often to save the configuration somewhere else, even if it
isn't modified.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update menuconfig (and Kconfiglib, just to sync) to upstream revision
e629c33d31e22, to add the commit below.
menuconfig: Add show-help mode
Pressing F toggles show-help mode, where the help window at the bottom
displays the help text of the currently selected item, if any. Can be
handy when browsing through symbols.
Also mention the different modes that are available in the module
docstring.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib (and menuconfig, just to sync) to upstream revision
094f4a9622046, to add the commit below.
Save existing configuration to .<filename>.old in write_config()
Add a default-True 'save_old' flag to write_config(). If 'save_old' is
True and an existing configuration file is being overwritten, a copy
of the old configuration file is saved to .<filename>.old (e.g.
.config.old) in the same directory.
Errors are ignored, as the old configuration would usually just be a
nice-to-have, and not essential.
The same functionality could be added for minimal configuration files
and headers, but it's probably most useful for configuration files.
Other changes:
- Parsing performance is improved a bit
- scripts/kconfig/kconfig.py now prints the path to the merged
configuration in zephyr/.config, to make it a bit easier to
discover.
Fixes: #2907
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib and menuconfig to upstream revision e47d7eff1012e, to
add some small fixes/improvements/changes:
- Raise errors for extra trailing tokens anywhere. They were silently
ignored in some places:
* end{if,menu,choice} <extra tokens>
* {default,select,...} FOO <extra tokens> (though e.g.
'default FOO if' raised an error)
- Rephrase the warning when selecting a choice symbol to make it
clearer that select never has any effect on choice symbols.
- Display empty menus with '----' instead of '---> (empty)'. This
matches the C tools. It might be less confusing for symbols defined
with 'menuconfig', which is where you most often get empty menus
(when the symbol is n).
- Speed up parsing performance
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib and menuconfig to upstream revision d3866958c7685, to
add various improvements:
- Support HOME and END in the jump-to dialog in the menuconfig. END can
be handy as choices, menus, and comments appear at the end.
- Add more fine-grained warning controls for multiple assignments to a
symbol in configuration files. Use it to simplify kconfig.py a bit.
Clean up kconfig.py a bit in other ways too, e.g. by removing unused
imports.
- Improve Kconfig parsing performance slightly
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update menuconfig and Kconfiglib to upstream revision 68426efeae6aa, to
add two minor menuconfig improvements and a small bugfix:
- Expected errors, like syntax errors in Kconfig files, no longer
generate a Python backtrace. Just the error message is shown.
- Entering a choice now places the cursor on the selected symbol, if
any.
- Having two consecutive empty 'if's (which could also appear e.g. due
to osource) crashed the menuconfig when entering show-all mode, due
to the 'if' removal logic in Kconfiglib failing to remove one of
them. All configurations of 'if's are now correctly removed.
This error was found while reading the code.
Some minor optimizations and some internal cleanup is included as well.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update menuconfig (and Kconfiglib, just to sync) to upstream revision
bf1701b36634b, to add this commit:
menuconfig: Add jump-to for choices, menus, and comments
For choices, search the name and the prompt. This is the same as for
symbols, except names are optional (and rare) for choices.
For menus and comments, search the prompt (title text).
When jumping to a non-empty choice or menu, jump into it instead of
jumping to its menu node. If show-all mode is off and there are
visible items in the choice/menu, then jump to the first visible
node. Otherwise, enable show-all and jump to the first node.
Previously, only symbols could be jumped to.
Various other small fixes/improvements are included too:
- The "no range constraints" text was dropped from the input dialog
when settings int/hex symbols without an active 'range'. It might be
more confusing than helpful.
- A crash when pressing Ctrl-F (the view-help shortcut) with no matches
in the jump-to dialog was fixed
- Some gnome-terminal shoddiness was worked around to remove minor
jumpiness when reducing the height of the terminal
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update menuconfig (and Kconfiglib, just to sync) to upstream revision
256e5b3e38e92 to get the fix below in, for an issue in an external
project. Also update genrest.py with a similar fix (the genrest issue
was what originally prompted it).
menuconfig: Improve/fix promptless choice handling
The code assumed that all parent (interface) menus always have a
prompt, which is false for items in promptless choices. This led to
a crash e.g. when viewing the symbol information for a symbol within
a promptless choice.
Promptless choices with children can show up "legitimately" when
people define choices in multiple locations to add symbols, though
this is broken in the C tools.
Use standard_sc_expr_str(node.item) instead of the non-existing
prompt for promptless choices. That way they show up as
'<choice (name if any>)>', which is consistent with how they're
shown elsewhere.
This commit also changes how choice names are displayed in
show-name/show-all mode, to the standard_sc_expr_str() format.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib (and menuconfig, just to sync) to upstream revision
8087311cd91be to get the following fix in, for an issue reported by
pfalcon:
Give clearer errors for bad endchoice/endif/endmenu nesting
An endchoice/endif/endmenu with no corresponding choice/if/menu
generated a cryptic 'unrecognized construct' parse error. Improve
the error message so that the problem is pointed out explicitly:
kconfiglib.KconfigError: Kconfig:37: couldn't parse 'endmenu': no
corresponding 'menu'
Reported in https://github.com/ulfalizer/Kconfiglib/issues/56.
This update also adds support for user-defined preprocessor functions in
Python, which could potentially be handy for DTS/Kconfig integration.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib to upstream revision 2be02fac78527 to add support for
expanding macros within symbol names, extending the existing Kconfig
preprocessor
(https://github.com/torvalds/linux/blob/master/Documentation/kbuild/
kconfig-macro-language.txt). Some minor optimizations are included as
well.
This makes it possible to add Kconfig templates to avoid code
repetition, e.g. like below:
Kconfig.log_template:
choice
prompt "Max compiled-in log level for $(module-str)"
config $(module)_LOG_LEVEL_OFF
bool "Off"
config $(module)_LOG_LEVEL_ERR
bool "Error"
config $(module)_LOG_LEVEL_WRN
bool "Warning"
config $(module)_LOG_LEVEL_INF
bool "Info"
config $(module)_LOG_LEVEL_DBG
bool "Debug"
endchoice
config $(module)_LOG_LEVEL
int
default 0 if $(module)_LOG_LEVEL_OFF
default 1 if $(module)_LOG_LEVEL_ERR
default 2 if $(module)_LOG_LEVEL_WRN
default 3 if $(module)_LOG_LEVEL_INF
default 4 if $(module)_LOG_LEVEL_DBG
Using the template:
module = FOO
module-str = foo
source "Kconfig.log_template"
...
module = BAR
module-str = bar
source "Kconfig.log_template"
This feature might create harder-to-read Kconfig files if abused, so it
should probably be used sparingly.
Fixes: #9761
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update menuconfig (and Kconfiglib, just to sync) to upstream revision
78682a8e3c4fe to get the following fix in, which could cause certain
visible symbols to not show up in the menuconfig interface:
menuconfig: Always show implicit submenus with visible nodes
Currently, the symbol BAR below (which ends up indented in an
implicit submenu) is shown only if DEP is non-n (or if show-all mode
is enabled):
config FOO
bool "foo" if DEP
default y
config BAR
bool "bar" if FOO
This is bad, because it hides visible symbols from the interface.
The assumption was that an implicit submenu (which is only created
if the parent has a prompt) would never have visible items when the
parent item is invisible, but prompt-specific conditions and
select/imply can break that assumption.
Fix it by always showing implicit submenus with visible nodes, along
with the parent node. If the parent node is invisible, show it in
red, like in show-all mode (which happens automatically). That's
probably better than having mysteriously indented nodes when the
parent is invisible.
(Some other tweaks were made later to avoid showing red outside show-all
mode, because it might look confusing/broken.)
A new color scheme for the menuconfig has been added as well,
contributed by Mitja Horvat (pinkfluid). It can be enabled by setting
the environment variable MENUCONFIG_THEME to 'aquatic'.
The default theme has been tweaked to use bold text for the selected
item too.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib and menuconfig to upstream revision a28bc4da9762e,
which adds include path information to menuconfig, showing how the
Kconfig file of the symbol got 'source'd in addition to showing the
definition location.
Add include path information to the auto-generated Kconfig documentation
too.
Some small Kconfiglib bugs are fixed to, like error reporting for
recursive 'source's being broken (crashing instead of printing the
error), a minor file descriptor leak, and syntax checking not catching
extra trailing tokens after 'if <expr>' and 'depends on <expr>'.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Commit f6bf897780 ("kconfiglib: Add preprocessor and two warnings")
accidentally made the ordering of #define's in generated header files
random (dependent on Python's set() implementation). Minimal
configuration files (which can be saved from the menuconfig) were
affected too.
The intent is for the order to match .config files, which in turn match
symbol definition order. This doesn't affect behavior in any way, but
makes things more readable (and is handier if files are checked in).
Update Kconfiglib to upstream revision a3252f620fad9 to make the
ordering match again. Tests have been added upstream.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Due a design goof, the top-level Kconfig file passed to
Kconfig.__init__() wasn't looked up relative to $srctree. This broke a
planned fix for an issue with the CI check for references to undefined
Kconfig files in external projects.
Update Kconfiglib to upstream revision 953cc12464 to fix it.
Also update doc/CMakeLists.txt to pass 'Kconfig' instead of '../Kconfig'
to genrest.py. $srctree is already set to $ZEPHYR_BASE.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib and menuconfig to upstream revision 6aea8d06b637e.
Kconfiglib changes:
- Kconfig files are now looked up just relative to $srctree (if set).
Previously, each source'd Kconfig file was also looked up in the
current directory, and Kconfig files there could override Kconfig
files in $srctree. This is what the C tools do.
I'm pretty sure that behavior was a bug in the C tools all along, and
only meant for .config files. It caused problems (and an ugly
workaround) for the undefined Kconfig symbol CI check in an external
project.
The new behavior also saves a bunch of open()'s, though it's probably
not noticeable.
- Setting the KCONFIG_STRICT environment variable to 'y' now makes
Kconfiglib itself warn for references to undefined symbols. This
isn't safe in general, as some projects use multiple Kconfig trees
with shared Kconfig files (e.g. the Linux kernel).
This will be used to simplify the undefined Kconfig symbol CI check.
- It's now possible to customize how symbols and choices are printed
within expressions.
This will be used to make the RST link generation in genrest.py less
hacky.
- Instead of having 'gsource', a plain 'source' is now globbing, and
requires at least one match. There's also 'osource', for when it's
okay for a glob pattern to match no files.
'gsource' had the design flaw that there was no way to require at
least one file to match. I plan on replacing all 'gsource' statements
with plain 'source's later, but 'gsource' is still supported for
backwards compatibility.
- def_int, def_hex, and def_string are now available as a Kconfig
extensions, analogous to def_bool (set type and add default).
- Misc. internal cleanup.
menuconfig changes:
- Boolean value hints are no longer shown to the right of defaults in
the symbol information for int/hex symbols. Stuff like
'- 74 (value: "n")' wasn't helpful, and looked confusing.
- Symbol information has been made more compact in general, e.g. by
skipping value hints where they aren't helpful
('FOO(=y)' instead of 'FOO(=y) (=y)'), and by shortening stuff like
'(value: "y")' to just '(=y)'.
- Misc. internal cleanup.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Up until now, Zephyr has patched Kconfig to use the last 'default' with
a satisfied condition, instead of the first one. I'm not sure why the
patch was added (it predates Kconfiglib), but I suspect it's related to
Kconfig.defconfig files.
There are at least three problems with the patch:
1. It's inconsistent with how Kconfig works in other projects, which
might confuse newcomers.
2. Due to oversights, earlier 'range' properties are still preferred,
as well as earlier 'default' properties on choices.
In addition to being inconsistent, this makes it impossible to
override 'range' properties and choice 'default' properties if the
base definition of the symbol/choice already has 'range'/'default'
properties.
I've seen errors caused by the inconsistency, and I suspect there
are more.
3. A fork of Kconfiglib that adds the patch needs to be maintained.
Get rid of the patch and go back to standard Kconfig behavior, as
follows:
1. Include the Kconfig.defconfig files first instead of last in
Kconfig.zephyr.
2. Include boards/Kconfig and arch/<arch>/Kconfig first instead of
last in arch/Kconfig.
3. Include arch/<arch>/soc/*/Kconfig first instead of last in
arch/<arch>/Kconfig.
4. Swap a few other 'source's to preserve behavior for some scattered
symbols with multiple definitions.
Swap 'source's in some no-op cases too, where it might match the
intent.
5. Reverse the defaults on symbol definitions that have more than one
default.
Skip defaults that are mutually exclusive, e.g. where each default
has an 'if <some board>' condition. They are already safe.
6. Remove the prefer-later-defaults patch from Kconfiglib.
Testing was done with a Python script that lists all Kconfig
symbols/choices with multiple defaults, along with a whitelist of fixed
symbols. The script also verifies that there are no "unreachable"
defaults hidden by defaults without conditions
As an additional test, zephyr/.config was generated before and after the
change for several samples and checked to be identical (after sorting).
This commit includes some default-related cleanups as well:
- Simplify some symbol definitions, e.g. where a default has 'if FOO'
when the symbol already has 'depends on FOO'.
- Remove some redundant 'default ""' for string symbols. This is the
implicit default.
Piggyback fixes for swapped ranges on BT_L2CAP_RX_MTU and
BT_L2CAP_TX_MTU (caused by confusing inconsistency).
Piggyback some fixes for style nits too, e.g. unindented help texts.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
$srctree was changed to an absolute path when the documentation building
was switched over to CMake, which uncovered a bug in Kconfiglib that
caused symbols in gsource'd files to show up with absolute paths in the
auto-generated Kconfig documentation.
This commit adds upstream commit ac692af07a123 ("Fix absolute $srctree
prefixes showing up on gsource'd files"), which fixes it.
Upstream commit message:
When using gsource with $srctree set to an absolute path, the $srctree
prefix would show up in MenuNode.filename, trickling its way into e.g.
generated documentation.
This was due to a broken test: os.path.isabs() was checked after
joining the pattern with $srctree, making it mistake an absolute
$srctree for an absolute path in the Kconfig file.
Fix the test.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib to upstream revision 547fced630611 to get a new
Kconfig preprocessor in, documented in
https://github.com/torvalds/linux/blob/master/Documentation/kbuild/kconfig-macro-language.txt.
The preprocessor allows shell functions to be run directly from Kconfig.
Things like 'default "prefix-$(shell,some-cmd)' and
'depends on (success,some-cmd)' are supported.
The preprocessor might come in handy for Kconfig/DTS integration. I'm
thinking of extending it so that Python functions can be called as well.
There's also two new warnings:
- Trying to use an int/hex symbol like a bool symbol in an expression
(where it will always evaluate to n)
- Having a 'default' on an int/hex symbol that lies outside an active
'range'.
The parser is more strict now as well (due to dropping some hacks for 3+
year old kernel versions).
A related fix for scripts/kconfig.py is also included:
The comment above the whitelist lies. I accidentally changed the warning
text for the select-with-unsatisfied-dependencies warning while
generalizing it (for m-valued dependencies, which you'd never get in
Zephyr).
Update the whitelist to detect the new warning text.
Last-minute piggyback:
Include a change that improved the parse time for U-Boot from 4 seconds
to about 0.6 seconds, related to symbols defined in multiple locations.
It might be helpful for Zephyr as well, as it also uses a lot of symbols
defined in multiple locations.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib to get upstream commit eb6c21a9b33a2 ("Turn
MenuNode/Symbol/Choice.referenced() into a @property") in. It converts
the MenuNode.referenced() function into a property, which makes the API
more consistent (read-only stuff uses properties).
Also update scripts/ci/list_undef_kconfig_refs.py to access .referenced
as a property.
Piggyback a small is_num() simplification.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib to get upstream commit ca89ca0c0c420 ("Add dependency
loop detection") in.
Upstream commit message
=======================
Pretty long overdue.
Until now, dependency loops have raised a hard-to-debug Python
RecursionError during evaluation. A Kconfiglib exception is raised now
instead, with a message that lists all the items in the loop.
See the comment at the start of _check_dep_loop_sym() for an overview of
the algorithm. At a high level, it's loop detection in a directed graph
by keeping track of unvisited/visited nodes during depth-first search.
(A third "visited, known to not be in a dependency loop" state is used
as well.)
Choices complicate things, as they're inherently loopy: The choice
depends on the choice symbols and vice versa, and the choice symbols in
a sense all depend on each other.
Add the choice-to-choice-symbol dependencies separately after dependency
loop detection, so that there's just the choice-symbol-to-choice
dependencies to deal with. It simplifies things, as it makes it possible
to tell dependencies from 'prompt' and 'default' conditions on the
choice from choice symbol dependencies.
Do some flag shenanigans to prevent the choice from being "re-entered"
while looping through the choice symbols. Maybe this could be cleaned up
a bit somehow...
Example exception message:
Dependency loop
===============
A (defined at tests/Kdeploop10:1), with definition...
config A
bool
depends on B
...depends on B (defined at tests/Kdeploop10:5), with definition...
config B
bool
depends on C = 7
...depends on C (defined at tests/Kdeploop10:9), with definition...
config C
int
range D 8
...depends on D (defined at tests/Kdeploop10:13), with definition...
config D
int
default 3 if E
default 8
...depends on E (defined at tests/Kdeploop10:18), with definition...
config E
bool
(select-related dependencies: F && G)
...depends on G (defined at tests/Kdeploop10:25), with definition...
config G
bool
depends on H
...depends on the choice symbol H (defined at tests/Kdeploop10:32), with
definition...
config H
bool
prompt "H" if I && <choice>
depends on I && <choice>
...depends on the choice symbol I (defined at tests/Kdeploop10:41), with
definition...
config I
bool
prompt "I" if <choice>
depends on <choice>
...depends on <choice> (defined at tests/Kdeploop10:38), with
definition...
choice
bool
prompt "choice" if J
...depends on J (defined at tests/Kdeploop10:46), with definition...
config J
bool
depends on A
...depends again on A (defined at tests/Kdeploop10:1)
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
This commit gets upstream commit dc0b022247b85 ("Correctly report choice
locations in some warnings") in, which fixes some warnings that would
previously report the location of a choice as being "undefined" (which
is impossible).
Upstream commit message:
Menu nodes were added to choices after parsing their properties,
making some warnings generated during parsing (as opposed to in
_check_choice_sanity()) incorrectly give the choice as '<choice>
(undefined)'.
Add the node before parsing choice properties to fix those warnings.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
This commit fixes a pretty nasty bug that could cause properties on
symbols and choices defined in multiple locations to end up in the wrong
order, potentially affecting evaluation.
Alexander Wachter ran into this for an out-of-tree build.
Multi.def. symbols are rare in the Linux kernel, which is what the
Kconfiglib test suite uses for compatibility testing, so this managed to
slip through. Comprehensive selftests have been added for property
ordering on nested multi.def. symbols/choices.
This bug was introduced by commit e307ba340c ("kconfiglib: Record
which MenuNode has each property").
Commit message from Kconfiglib (c8801514d63aa)
==============================================
Fix incorrectly ordered properties for some nested multi.def. symbols
_propagate_deps() visits menu nodes roughly breadth-first, meaning
properties on symbols and choices defined in multiple locations could
end up in the wrong order when copied from the menu node for some
unlucky if/menu nestings.
Fix it by moving the menu-node-to-symbol/choice property copying in
_finalize_tree() so that it's guaranteed to happen in definition order.
This bug was introduced by commit 63a4418 ("Record which MenuNode has
each property").
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Update Kconfiglib to upstream revision 9fba375c65341 (+ local Zephyr
modifications) to get commit 94020beb311eb ("Make Kconfig._choices
public") in. It will be used to generate Kconfig reference pages for
choices.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Unquoted string defaults work through a quirk of Kconfig (undefined
symbols get their name as their string value), but look confusing. It's
done inconsistently now too.
Suggested by Kumar Gala.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
- Show the value of each symbol in the jump-to dialog.
- Search names and prompts separately in the jump-to dialog.
Previously, '_BAR$' wouldn't match FOO_BAR if it had a prompt,
because '_BAR$' was matched against the string 'FOO_BAR "prompt"'.
- Show the working directory in save/load dialog boxes. Paths will be
relative to it, and Carles Cufi pointed out that it might not be
obvious.
Also allow ~ to be used to refer to the home directory.
- Implement scroll offset for edit boxes. This makes it clearer when
they're scrolled horizontally.
Piggyback an update of Kconfiglib to the latest version (no functional
changes).
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Leading/trailing whitespace in prompts requires ugly workarounds in
genrest.py, as e.g. *prompt * is invalid RST. strip() all prompts in
Kconfiglib and get rid of the genrest.py workarounds. Add a warning too.
The Kconfiglib update has some unrelated cleanups and fixes (that won't
affect Zephyr).
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
This commit does some major surgery to Kconfiglib so that properties
(defaults, selects, etc.) can be shown on the menu node that actually
has the propetrty for symbols and choices defined in multiple locations.
This will be used to improve the output of genrest.py and the symbol
information display in the menuconfig.
The parsing code is a bit simpler now too as a side effect.
Commit message from Kconfiglib (63a44186137e2)
==============================================
This allows accurate documentation to be generated for symbols and
choices defined in multiple locations. There are now MenuNode.defaults,
MenuNode.selects, etc., lists that mirror the corresponding
Symbol/Choice lists.
Symbol/Choice.__str__() now correctly show property locations as well,
by simply concatenating the strings returned by MenuNode.__str__() for
each node.
_parse_properties() was modified to add all properties directly to the
menu node instead of adding them to the contained symbol or choice. The
properties are then copied up to symbols and choices in
_finalize_tree(). Dependency propagation is handled at the same time.
As a side effect, this cleans up the code a bit and de-bloats
_parse_properties().
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
This commit gets rid of the 'option env="ENV_VAR"' bounce symbols.
"$FOO" now expands directly to the value of the environment variable
FOO, instead of to the value of the Kconfig symbol FOO.
This change is likely to soon appear in the C tools as well. Those
'option env' symbols always seemed kinda pointless, and have broken
dependency handling due to forcing symbol evaluation during parsing,
before all the symbols have even been seen.
Compatibility with the C tools could be retained by naming all
'option env' symbols the same as the environment variable they
reference.
This commit also updated the Zephyr documentation to explain the new
behavior. It's relevant for $ZEPHYR_BASE and out-of-tree Kconfig
extensions.
Commit message from Kconfiglib (cbf32e29a130d)
==============================================
Make "$FOO" directly reference the environment variable $FOO in e.g.
'source' statements, instead of the symbol FOO. Use os.path.expandvars()
to expand strings (which preserves "$FOO" as-is if no environment
variable FOO exists).
This gets rid of the 'option env' "bounce" symbols, which are mostly
just spam and are buggy in the C tools (dependencies aren't always
respected, due to parsing and evaluation getting mixed up). The same
change will probably appear soon in the C tools as well.
Keep accepting 'option env' to preserve some backwards compatibility,
but ignore it when expanding strings. For compatibility with the C
tools, bounce symbols will need to be named the same as the environment
variables they reference (which is the case for the Linux kernel).
This is a compatibility break, so the major version will be bumped to 6
at the next release.
The main motivation for adding this now is to allow recording properties
on each MenuNode in a clean way. 'option env' symbols interact badly
with delayed dependency propagation.
Side note: I have a feeling that recording environment variable values
might be redundant to trigger rebuilds if sync_deps() is run at each
compile. It should detect all changes to symbol values due to
environment variables changing value.
Signed-off-by: Ulf Magnusson <Ulf.Magnusson@nordicsemi.no>
Until now, Zephyr has used a patched Kconfiglib that turns 'source' into
a globbing source (by replacing 'source' with 'gsource' at the token
level). There's two problems with this:
- The patch needs to be maintained separately
- Misspelled filenames are silently ignored, as they look like glob
patterns that don't match anything
Fix it as follows:
1. Replace all 'source' statements that use wildcards with 'gsource'
2. Remove the custom Kconfiglib patch so that 'source' no longer globs
The sed pattern '/source.*[*?]/s/source/gsource/' was run over all
Kconfig* files to do the replacement.
source's that use environment variables that might contain glob patterns
were manually changed to gsource.
Building the docs in doc/ is a good test, as doc/Makefile deliberately
sets the environment variables to glob up as many Kconfig files as
possible.
Signed-off-by: Ulf Magnusson <ulfalizer@gmail.com>
Update Kconfiglib to upstream revision da40c014398f3 (+ local Zephyr
modifications) to get commit da40c014398f3 ("Force encoding to UTF-8 by
default on Python 3") in. It sets a (configurable) UTF-8 default for
Python 3, overriding the encoding specified in the current locale.
I've decided that this is a good idea after some problem reports
unrelated to Zephyr. Running with the C locale breaks things horribly
otherwise, and the fix isn't obvious.
Plain strings aren't decoded on Python 2, so no changes are needed
there.
Related PEP: https://www.python.org/dev/peps/pep-0538/
Signed-off-by: Ulf Magnusson <ulfalizer@gmail.com>
Update Kconfiglib to upstream revision ed3ceaa056262 (+ local Zephyr
modifications) to get commits c1c5ef2eb1009 ("Print a warning for
malformed .config lines") and ed3ceaa05626f ("Make warnings available in
a list") in.
This warning for malformed .config lines will be turned into an error in
kconfig.py, which is made easier by making the warnings available in a
list in Kconfiglib.
It's now configurable whether warnings are printed to stderr or not. In
this case, it still makes sense to print them to stderr as well.
Suggested by Marti Bolivar.
Signed-off-by: Ulf Magnusson <ulfalizer@gmail.com>
Update Kconfiglib to upstream revision 509e374dfcadb (+ local Zephyr
modifications) to get commit 509e374dfcadb ("Add Choice.direct_dep
field") in. It is used by the upcoming Python menuconfig implementation
when displaying information about choices.
Origin: https://github.com/zephyrproject-rtos/Kconfiglib/tree/zephyr
Signed-off-by: Ulf Magnusson <ulfalizer@gmail.com>
Update Kconfiglib to upstream revision 105c835e70a5b (+ local Zephyr
modifications) to get commit 105c835e70a5b ("Add helper for splitting
expressions") in. It will simplify the 'select' logic in genrest.py and
the upcoming menuconfig implementation, and also simplifies some
Kconfiglib internals.
split_expr() will also be helpful if genrest.py ever needs to split
other expressions, e.g. over multiple lines.
Origin: https://github.com/zephyrproject-rtos/Kconfiglib/tree/zephyr
Signed-off-by: Ulf Magnusson <ulfalizer@gmail.com>
Update Kconfiglib to upstream revision 981d24aff7654 (+ local Zephyr
modifications) to get commit 981d24aff7654 ("Set is_menuconfig True on
the top menu") in. It will be needed by the menuconfig implementation.
Origin: https://github.com/zephyrproject-rtos/Kconfiglib/tree/zephyr
Signed-off-by: Ulf Magnusson <ulfalizer@gmail.com>
Update Kconfiglib to upstream revision 2259d353426f1 (+ local Zephyr
modifications) to get commit 2259d353426f1 ("Generalize is_menuconfig to
non-symbol items") in. It will be used by the menuconfig implementation.
Origin: https://github.com/zephyrproject-rtos/Kconfiglib/tree/zephyr
Signed-off-by: Ulf Magnusson <ulfalizer@gmail.com>
Update Kconfiglib to upstream revision 7245bad9ebb58 (+ local Zephyr
modifications), to get the following improvements/fixes in:
- Print a warning if an int or hex symbol is assigned a value that lies
outside an active 'range' constraint.
- Turn 'FOO' into a link in 'select FOO' when generating the Kconfig
reference documentation.
- Parenthesize '&&' expressions within '||' expressions --
(A && B) || (C && D) is more readable than A && B || C && D.
The final two fixes will only be visible once the fix for #5622 gets in.
Fixes#6749Fixes#6844
Origin: https://github.com/zephyrproject-rtos/Kconfiglib/tree/zephyr
Signed-off-by: Ulf Magnusson <ulfalizer@gmail.com>
