kconfig: do not reparent the menu inside a choice block

The boolean 'choice' is used to list exclusively selected config
options.

You must not add a dependency between choice members, because such a
dependency would create an invisible entry.

In the following test case, it is impossible to choose 'C'.

[Test Case 1]

choice
prompt "Choose one, but how to choose C?"

config A
bool "A"

config B
bool "B"

config C
bool "C"
depends on A

endchoice

Hence, Kconfig shows the following error message:

Kconfig:1:error: recursive dependency detected!
Kconfig:1: choice <choice> contains symbol C
Kconfig:10: symbol C is part of choice A
Kconfig:4: symbol A is part of choice <choice>
For a resolution refer to Documentation/kbuild/kconfig-language.rst
subsection "Kconfig recursive dependency limitations"

However, Kconfig does not report anything for the following similar code:

[Test Case 2]

choice
prompt "Choose one, but how to choose B?"

config A
bool "A"

config B
bool "B"
depends on A

config C
bool "C"

endchoice

This is because menu_finalize() reparents the menu tree when an entry
depends on the preceding one.

With reparenting, the menu tree:

choice
|- A
|- B
\- C

... will be transformed into the following structure:

choice
|- A
| \- B
\- C

Consequently, Kconfig considers only 'A' and 'C' as choice members.
This behavior is awkward. The second test case should be an error too.

This commit stops reparenting inside a choice.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: check prompt for choice while parsing

This can be checked on-the-fly.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: remove named choice support

Commit 5a1aa8a1aff6 ("kconfig: add named choice group") did not provide
enough explanation regarding its benefits. A use case was found in
another project [1] sometime later, this feature has never been used in
the kernel.

[1]: https://lore.kernel.org/all/201012150034.01356.yann.morin.1998@anciens.enib.fr/

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>

kconfig: use generic macros to implement symbol hashtable

Use helper macros in hashtable.h for generic hashtable implementation.

We can git rid of the hash head index of for_all_symbols().

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: print recursive dependency errors in the parsed order

for_all_symbols() iterates in the symbol hash table. The order of
iteration depends on the hash table implementation.

If you use it for printing errors, they are shown in random order.

For example, the order of following test input and the corresponding
error do not match:
- scripts/kconfig/tests/err_recursive_dep/Kconfig
- scripts/kconfig/tests/err_recursive_dep/expected_stderr

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: do not delay the cur_filename update

Currently, cur_filename is updated at the first token of each statement.
However, this seems unnecessary based on my understanding; the parser
can use the same variable as the lexer tracks.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: associate struct menu with file name directly

struct menu is linked to struct file for diagnostic purposes.
It is always used to retrieve the file name through menu->file->name.

Associate struct menu with the file name directly.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: remove zconf_curname() and zconf_lineno()

Now zconf_curname() and zconf_lineno() are so simple that they just
return cur_filename, cur_lineno, respectively.

Remove these functions, and then use cur_filename and cur_lineno
directly.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: split preprocessor prototypes into preprocess.h

These are needed only for the parse stage. Move the prototypes into
a separate header to make sure they are not used after that.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: call env_write_dep() right after yyparse()

This allows preprocess.c to free up all of its resources when the parse
stage is finished. It also ensures conf_write_autoconf_cmd() produces
consistent results even if called multiple times for any reason.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: write Kconfig files to autoconf.cmd in order

Currently, include/config/autoconf.cmd saves included Kconfig files in
reverse order. While this is not a big deal, it is inconsistent with
other *.cmd files generated by fixdep.

Output the included Kconfig files in the included order.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: remove unneeded sym_find() call in conf_parse()

sym_find("n") is equivalent to &symbol_no.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: fix off-by-one in zconf_error()

yyerror() reports the line number of the next line.

This +1 adjustment was introduced more than 20 years ago [1]. At that
time, the line number was decremented then incremented back and forth.

The line number management was refactored in a more maintainable way.
Such compensation is no longer needed.

[1]: https://git.kernel.org/pub/scm/linux/kernel/git/history/history.git/commit/?id=d4f8a4530eb07a1385fd17b0e62a7dce97486f49

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: split menu.c out of parser.y

Compile menu.c as an independent compilation unit.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: change sym_change_count to a boolean flag

sym_change_count has no good reason to be 'int' type.

sym_set_change_count() compares the old and new values after casting
both of them to (bool). I do not see any practical diffrence between
sym_set_change_count(1) and sym_add_change_count(1).

Use the boolean flag, conf_changed.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: change "modules" from sub-option to first-level attribute

Now "modules" is the only member of the "option" property.

Remove "option", and move "modules" to the top level property.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: remove allnoconfig_y option

Now that the only user, CONFIG_EMBEDDED has stopped using this option,
remove it entirely.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: change defconfig_list option to environment variable

"defconfig_list" is a weird option that defines a static symbol that
declares the list of base config files in case the .config does not
exist yet.

This is quite different from other normal symbols; we just abused the
"string" type and the "default" properties to list out the input files.
They must be fixed values since these are searched for and loaded in
the parse stage.

It is an ugly hack, and should not exist in the first place. Providing
this feature as an environment variable is a saner approach.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: allow only 'config', 'comment', and 'if' inside 'choice'

The code block surrounded by 'if' ... 'endif' is reduced into if_stmt,
which is accepted in the 'choice' context. Therefore, you can write any
statements within a choice block by wrapping 'if y' ... 'end'.

For example, you can create a menu inside a choice, like follows:

---------------->8----------------
choice
prompt "choice"

config A
bool "A"

config B
bool "B"

if y

menu "strange menu"

config C
bool "C"

endmenu

endif

endchoice
---------------->8----------------

I want to change such a weird structure into a syntax error.

In fact, the USB gadget Kconfig had used nested 'choice' for no good
reason until commit df8df5e4bc37 ("usb: get rid of 'choice' for
legacy gadget drivers") killed it.

I think the 'source' inside 'choice' is on the fence. It is at least
gramatically sensible as long as the included file contains only
bool/tristate configs. However, it makes the code unreadable, and people
tend to forget the fact that the file is included from the choice
block. Commit 10e5e6c24963 ("usb: gadget: move choice ... endchoice to
legacy/Kconfig") got rid of the only usecase.

Going forward, you can only use 'config', 'comment', and 'if' inside
'choice'. This also recursively applies to 'if' blocks inside 'choice'.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: remove sym from struct property

struct property can reference to the symbol that it is associated with
by prop->menu->sym.

Fix up the one usage of prop->sym, and remove sym from struct property.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: remove 'prompt' symbol

Now that 'prompt' is only reduced from T_WORD_QUOTE without any action,
use T_WORD_QUOTE directly.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: drop T_WORD from the RHS of 'prompt' symbol

Commit 8636a1f9677d ("treewide: surround Kconfig file paths with double
quotes") killed use-cases to reduce an unquoted string into the 'prompt'
symbol.

Kconfig still allows to use an unquoted string in the context of menu,
source, or prompt.

So, you can omit quoting if the prompt is a single word:

bool foo

..., but I do not think this is so useful.

Let's require quoting:

bool "foo"

All the Kconfig files in the kernel are written in this way.

Remove the T_WORD from the right-hand side of the symbol 'prompt'.

Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>

kconfig: split util.c out of parser.y

util.c exists both in scripts/kconfig/ and scripts/kconfig/lxdialog.

Prior to commit 54b8ae66ae1a ("kbuild: change *FLAGS_<basetarget>.o
to take the path relative to $(obj)"), Kbuild could not pass different
flags to source files with the same basename. Now that this issue
was solved, you can split util.c out of parser.y and compile them
independently of each other.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>

kconfig: rename zconf.y to parser.y

Use a more logical name.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>