I used the C comment style (/* ... */) for the flex and bison files
as in Kconfig (scripts/kconfig/{lexer.l,parser.y})
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Use the __section() shorthand. This avoids escaping double-quotes,
and improves the readability.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
This makes *.mod.c much more readable. I confirmed depmod still
produced the same modules.dep file.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Arnd Bergmann reported false-positive modpost warnings detected by his
randconfig testing of linux-next.
Actually, this happens under the combination of CONFIG_MODVERSIONS
and CONFIG_TRIM_UNUSED_KSYMS since commit 15bfc2348d ("modpost:
check for static EXPORT_SYMBOL* functions").
For example, arch/arm/config/multi_v7_defconfig + CONFIG_MODVERSIONS
+ CONFIG_TRIM_UNUSED_KSYMS produces the following false-positives:
WARNING: "__lshrdi3" [vmlinux] is a static (unknown)
WARNING: "__ashrdi3" [vmlinux] is a static (unknown)
WARNING: "__aeabi_lasr" [vmlinux] is a static (unknown)
WARNING: "__aeabi_llsr" [vmlinux] is a static (unknown)
WARNING: "ftrace_set_clr_event" [vmlinux] is a static (unknown)
WARNING: "__muldi3" [vmlinux] is a static (unknown)
WARNING: "__aeabi_ulcmp" [vmlinux] is a static (unknown)
WARNING: "__ucmpdi2" [vmlinux] is a static (unknown)
WARNING: "__aeabi_lmul" [vmlinux] is a static (unknown)
WARNING: "__bswapsi2" [vmlinux] is a static (unknown)
WARNING: "__bswapdi2" [vmlinux] is a static (unknown)
WARNING: "__ashldi3" [vmlinux] is a static (unknown)
WARNING: "__aeabi_llsl" [vmlinux] is a static (unknown)
The root cause of the problem is not in the modpost, but in the
implementation of CONFIG_TRIM_UNUSED_KSYMS.
If there is at least one untrimmed symbol in the file, genksyms is
invoked to calculate CRC of *all* the exported symbols in that file
even if some of them have been trimmed due to no caller existing.
As a result, .tmp_*.ver files contain CRC of trimmed symbols, thus
unneeded, orphan __crc* symbols are added to objects. It had been
harmless until recently.
With commit 15bfc2348d ("modpost: check for static EXPORT_SYMBOL*
functions"), it is now harmful because the bogus __crc* symbols make
modpost call sym_update_crc() to add the symbols to the hash table,
but there is no one that clears the ->is_static member.
I gave Fixes to the first commit that uncovered the issue, but the
potential problem has long existed since commit f235541699
("export.h: allow for per-symbol configurable EXPORT_SYMBOL()").
Fixes: 15bfc2348d ("modpost: check for static EXPORT_SYMBOL* functions")
Reported-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Tested-by: Arnd Bergmann <arnd@arndb.de>
GCC and Clang have different policy for -Wunused-function; GCC does not
warn unused static inline functions at all whereas Clang does if they
are defined in source files instead of included headers although it has
been suppressed since commit abb2ea7dfd ("compiler, clang: suppress
warning for unused static inline functions").
We often miss to delete unused functions where 'static inline' is used
in *.c files since there is no tool to detect them. Unused code remains
until somebody notices. For example, commit 075ddd7568 ("regulator:
core: remove unused rdev_get_supply()").
Let's remove __maybe_unused from the inline macro to allow Clang to
start finding unused static inline functions. For now, we do this only
for W=1 build since it is not a good idea to sprinkle warnings for the
normal build (e.g. 35 warnings for arch/x86/configs/x86_64_defconfig).
My initial attempt was to add -Wno-unused-function for no W= build
(https://lore.kernel.org/patchwork/patch/1120594/)
Nathan Chancellor pointed out that would weaken Clang's checks since
we would no longer get -Wunused-function without W=1. It is true GCC
would catch unused static non-inline functions, but it would weaken
Clang as a standalone compiler, at least.
Hence, here is a counter implementation. The current problem is, W=...
only controls compiler flags, which are globally effective. There is
no way to address only 'static inline' functions.
This commit defines KBUILD_EXTRA_WARN[123] corresponding to W=[123].
When KBUILD_EXTRA_WARN1 is defined, __maybe_unused is omitted from
the 'inline' macro.
The new macro __inline_maybe_unused makes the code a bit uglier, so I
hope we can remove it entirely after fixing most of the warnings.
If you contribute to code clean-up, please run "make CC=clang W=1"
and check -Wunused-function warnings. You will find lots of unused
functions.
Some of them are false-positives because the call-sites are disabled
by #ifdef. I do not like to abuse the inline keyword for suppressing
unused-function warnings because it is intended to be a hint for the
compiler optimization. I prefer #ifdef around the definition, or
__maybe_unused if #ifdef would make the code too ugly.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Tested-by: Nathan Chancellor <natechancellor@gmail.com>
KBUILD_ENABLE_EXTRA_GCC_CHECKS started as a switch to add extra warning
options for GCC, but now it is a historical misnomer since we use it
also for Clang, DTC, and even kernel-doc.
Rename it to more sensible, shorter KBUILD_EXTRA_WARN.
For the backward compatibility, KBUILD_ENABLE_EXTRA_GCC_CHECKS is still
supported (but not advertised in the documentation).
I also fixed up 'make help', and updated the documentation.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Reviewed-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Sedat Dilek <sedat.dilek@gmail.com>
Instead of the warning-[123] magic, let's accumulate compiler options
to KBUILD_CFLAGS directly as the top Makefile does. I think this makes
it easier to understand what is going on in this file.
This commit slightly changes the behavior, I think all of which are OK.
[1] Currently, cc-option calls are needlessly evaluated. For example,
warning-3 += $(call cc-option, -Wpacked-bitfield-compat)
needs evaluating only when W=3, but it is actually evaluated for
W=1, W=2 as well. With this commit, only relevant cc-option calls
will be evaluated. This is a slight optimization.
[2] Currently, unsupported level like W=4 is checked by:
$(error W=$(KBUILD_ENABLE_EXTRA_GCC_CHECKS) is unknown)
This will no longer be checked, but I do not think it is a big
deal.
[3] Currently, 4 Clang warnings (Winitializer-overrides, Wformat,
Wsign-compare, Wformat-zero-length) are shown by any of W=1, W=2,
and W=3. With this commit, they will be warned only by W=1. I
think this is a more correct behavior since each warning belongs
to only one group.
For understanding this commit correctly:
We have 3 warning groups, W=1, W=2, and W=3. You may think W=3 has a
higher level than W=1, but they are actually independent. If you like,
you can combine them like W=13. To enable all the warnings, you can
pass W=123. It is shown by 'make help', but not noticed much. Since we
support W= combination, there should not exist intersection among the
three groups. If we enable Winitializer-overrides for W=1, we do not
need to for W=2 or W=3. This is the reason why I think the change [3]
makes sense.
The documentation says -Winitializer-overrides is enabled by default.
(https://clang.llvm.org/docs/DiagnosticsReference.html#winitializer-overrides)
We negate it by passing -Wno-initializer-overrides for the normal
build, but we do not do that for W=1. This means, W=1 effectively
enables -Winitializer-overrides by the clang's default. The same for
the other three.
Add comments in case people are confused with the code.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Tested-by: Sedat Dilek <sedat.dilek@gmail.com>
Acked-by: Nick Desaulniers <ndesaulniers@google.com>
Acked-by: Miguel Ojeda <miguel.ojeda.sandonis@gmail.com>
The merge_config.sh script verifies that all the config options have
their expected value in the resulting file and prints any issues as
warnings. These checks aren't intended to be treated as errors given
the current implementation. However, since "set -e" was added, if the
grep command to look for a config option does not find it the script
will then abort prematurely.
Handle the case where the grep exit status is non-zero by setting
ACTUAL_VAL to an empty string to restore previous functionality.
Fixes: cdfca82157 ("merge_config.sh: Check error codes from make")
Signed-off-by: Guillaume Tucker <guillaume.tucker@collabora.com>
Acked-by: Jon Hunter <jonathanh@nvidia.com>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Kbuild provides per-file compiler flag addition/removal:
CFLAGS_<basetarget>.o
CFLAGS_REMOVE_<basetarget>.o
AFLAGS_<basetarget>.o
AFLAGS_REMOVE_<basetarget>.o
CPPFLAGS_<basetarget>.lds
HOSTCFLAGS_<basetarget>.o
HOSTCXXFLAGS_<basetarget>.o
The <basetarget> is the filename of the target with its directory and
suffix stripped.
This syntax comes into a trouble when two files with the same basename
appear in one Makefile, for example:
obj-y += foo.o
obj-y += dir/foo.o
CFLAGS_foo.o := <some-flags>
Here, the <some-flags> applies to both foo.o and dir/foo.o
The real world problem is:
scripts/kconfig/util.c
scripts/kconfig/lxdialog/util.c
Both files are compiled into scripts/kconfig/mconf, but only the
latter should be given with the ncurses flags.
It is more sensible to use the relative path to the Makefile, like this:
obj-y += foo.o
CFLAGS_foo.o := <some-flags>
obj-y += dir/foo.o
CFLAGS_dir/foo.o := <other-flags>
At first, I attempted to replace $(basetarget) with $*. The $* variable
is replaced with the stem ('%') part in a pattern rule. This works with
most of cases, but does not for explicit rules.
For example, arch/ia64/lib/Makefile reuses rule_as_o_S in its own
explicit rules, so $* will be empty, resulting in ignoring the per-file
AFLAGS.
I introduced a new variable, target-stem, which can be used also from
explicit rules.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Marc Zyngier <maz@kernel.org>
Add NOFAIL check for the strndup call, because the function
allocates memory and can return NULL. All calls to strdup in
modpost are checked with NOFAIL.
Signed-off-by: Denis Efremov <efremov@linux.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Since guid_t is the recommended data type for UUIDs in
kernel (and I guess uuid_le is meant to be ultimately
replaced with it), it should be made available here as
well.
Signed-off-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
CONFIG_SHELL falls back to sh when bash is not installed on the system,
but nobody is testing such a case since bash is usually installed.
So, shell scripts invoked by CONFIG_SHELL are only tested with bash.
It makes it difficult to test whether the hashbang #!/bin/sh is real.
For example, #!/bin/sh in arch/powerpc/kernel/prom_init_check.sh is
false. (I fixed it up)
Besides, some shell scripts invoked by CONFIG_SHELL use bash-extension
and #!/bin/bash is specified as the hashbang, while CONFIG_SHELL may
not always be set to bash.
Probably, the right thing to do is to introduce BASH, which is bash by
default, and always set CONFIG_SHELL to sh. Replace $(CONFIG_SHELL)
with $(BASH) for bash scripts.
If somebody tries to add bash-extension to a #!/bin/sh script, it will
be caught in testing because /bin/sh is a symlink to dash on some major
distributions.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
This '+' was added a long time ago:
| commit c23e6bf05f7802e92fd3da69a1ed35e56f9c85bb (HEAD)
| Author: Kai Germaschewski <kai@tp1.ruhr-uni-bochum.de>
| Date: Mon Oct 28 01:16:34 2002 -0600
|
| kbuild: Fix a "make -j<N>" warning
|
| diff --git a/scripts/Makefile.clean b/scripts/Makefile.clean
| index 2c843e0380bc..e7c392fd5788 100644
| --- a/scripts/Makefile.clean
| +++ b/scripts/Makefile.clean
| @@ -42,7 +42,7 @@ quiet_cmd_clean = CLEAN $(obj)
|
| __clean: $(subdir-ymn)
| ifneq ($(strip $(__clean-files) $(clean-rule)),)
| - $(call cmd,clean)
| + +$(call cmd,clean)
| else
| @:
| endif
At that time, cmd_clean contained $(clean-rule), which was able to
invoke sub-make. That was why cleaning with the -j option showed:
warning: jobserver unavailable: using -j1. Add '+' to parent make rule.
It is not the case any more; cmd_clean now just runs the 'rm' command.
The '+' marker is pointless.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The only the difference between clean-files and clean-dirs is the -r
option passed to the 'rm' command.
You can always pass -r, and then remove the clean-dirs syntax.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Both relative path and absolute path have pros and cons. For example,
we can move the source and objtree around together by using the
relative path to the source tree.
Do not force the absolute path to the source tree. If you prefer the
absolute path, you can specify KBUILD_ABS_SRCTREE=1.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Kbuild descends into scripts/basic/ even before the Kconfig.
I do not expect any other host programs added to this Makefile.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
scripts/package/Makefile does not use $(obj) or $(src) at all.
It actually generates files and directories in the top of $(objtree).
I do not see much sense in descending into scripts/package/.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
I am not a big fan of the $(objtree)/ hack for clean-files/clean-dirs.
These are created in the top of $(objtree), so let's clean them up
from the top Makefile.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
When we execute make after merging the configurations we ignore any
errors it produces causing whatever is running merge_config.sh to be
unaware of any failures. This issue was noticed by Guillaume Tucker
while looking at problems with testing of clang only builds in KernelCI
which caused Kbuild to be unable to find a working host compiler.
This implementation was suggested by Yamada-san.
Suggested-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Reported-by: Guillaume Tucker <guillaume.tucker@collabora.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Makefile.lib is included by Makefile.modfinal as well as Makefile.build.
Move modkern_cflags to Makefile.lib in order to simplify cmd_cc_o_c
in Makefile.modfinal. Move modkern_cflags as well for consistency.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Add CONFIG_ASM_MODVERSIONS. This allows to remove one if-conditional
nesting in scripts/Makefile.build.
scripts/Makefile.build is run every time Kbuild descends into a
sub-directory. So, I want to avoid $(wildcard ...) evaluation
where possible although computing $(wildcard ...) is so cheap that
it may not make measurable performance difference.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org>
I think splitting the modpost and linking modules into separate
Makefiles will be useful especially when more complex build steps
come in. The main motivation of this commit is to integrate the
proposed klp-convert feature cleanly.
I moved the logging 'Building modules, stage 2.' to Makefile.modpost
to avoid the code duplication although I do not know whether or not
this message is needed in the first place.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Currently, the timestamp of module linker scripts are not checked.
Add them to the dependency of modules so they are correctly rebuilt.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Currently, the single target build directly descends into the directory
of the target. For example,
$ make foo/bar/baz.o
... directly descends into foo/bar/.
On the other hand, the normal build usually descends one directory at
a time, i.e. descends into foo/, and then foo/bar/.
This difference causes some problems.
[1] miss subdir-asflags-y, subdir-ccflags-y in upper Makefiles
The options in subdir-{as,cc}flags-y take effect in the current
and its sub-directories. In other words, they are inherited
downward. In the example above, the single target will miss
subdir-{as,cc}flags-y if they are defined in foo/Makefile.
[2] could be built in a different directory
As Documentation/kbuild/modules.rst section 4.3 says, Kbuild can
handle files that are spread over several sub-directories.
The build rule of foo/bar/baz.o may not necessarily be specified in
foo/bar/Makefile. It might be specifies in foo/Makefile as follows:
[foo/Makefile]
obj-y := bar/baz.o
This often happens when a module is so big that its source files
are divided into sub-directories.
In this case, there is no Makefile in the foo/bar/ directory, yet
the single target descends into foo/bar/, then fails due to the
missing Makefile. You can still do 'make foo/bar/' for partial
building, but cannot do 'make foo/bar/baz.s'. I believe the single
target '%.s' is a useful feature for inspecting the compiler output.
Some modules work around this issue by putting an empty Makefile
in every sub-directory.
This commit fixes those problems by making the single target build
descend in the same way as the normal build does.
Another change is the single target build will observe the CONFIG
options. Previously, it allowed users to build the foo.o even when
the corresponding CONFIG_FOO is disabled:
obj-$(CONFIG_FOO) += foo.o
In the new behavior, the single target build will just fail and show
"No rule to make target ..." (or "Nothing to be done for ..." if the
stale object already exists, but cannot be updated).
The disadvantage of this commit is the build speed. Now that the
single target build visits every directory and parses lots of
Makefiles, it is slower than before. (But, I hope it will not be
too slow.)
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
When kallsyms generation happens, temporary vmlinux outputs are linked
but the quiet make output didn't report it, giving the impression that
the prior command is taking longer than expected.
Instead, report the linking step explicitly. While at it, this
consolidates the repeated "kallsyms generation step" into a single
function and removes the existing copy/pasting.
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
scripts/headers_check.pl can detect references to CONFIG options in
exported headers, but it has been disabled for more than a decade.
Reverting commit 7e3fa56141 ("kbuild: drop check for CONFIG_ in
headers_check") would emit the following warnings for headers_check
on x86:
usr/include/mtd/ubi-user.h:283: leaks CONFIG_MTD_UBI_BEB_LIMIT to userspace where it is not valid
usr/include/linux/cm4000_cs.h:26: leaks CONFIG_COMPAT to userspace where it is not valid
usr/include/linux/pkt_cls.h:301: leaks CONFIG_NET_CLS_ACT to userspace where it is not valid
usr/include/linux/videodev2.h:2465: leaks CONFIG_VIDEO_ADV_DEBUG to userspace where it is not valid
usr/include/linux/bpf.h:249: leaks CONFIG_EFFICIENT_UNALIGNED_ACCESS to userspace where it is not valid
usr/include/linux/bpf.h:819: leaks CONFIG_CGROUP_NET_CLASSID to userspace where it is not valid
usr/include/linux/bpf.h:1011: leaks CONFIG_IP_ROUTE_CLASSID to userspace where it is not valid
usr/include/linux/bpf.h:1742: leaks CONFIG_BPF_KPROBE_OVERRIDE to userspace where it is not valid
usr/include/linux/bpf.h:1747: leaks CONFIG_FUNCTION_ERROR_INJECTION to userspace where it is not valid
usr/include/linux/bpf.h:1936: leaks CONFIG_XFRM to userspace where it is not valid
usr/include/linux/bpf.h:2184: leaks CONFIG_BPF_LIRC_MODE2 to userspace where it is not valid
usr/include/linux/bpf.h:2210: leaks CONFIG_BPF_LIRC_MODE2 to userspace where it is not valid
usr/include/linux/bpf.h:2227: leaks CONFIG_SOCK_CGROUP_DATA to userspace where it is not valid
usr/include/linux/bpf.h:2311: leaks CONFIG_NET to userspace where it is not valid
usr/include/linux/bpf.h:2348: leaks CONFIG_NET to userspace where it is not valid
usr/include/linux/bpf.h:2422: leaks CONFIG_BPF_LIRC_MODE2 to userspace where it is not valid
usr/include/linux/bpf.h:2528: leaks CONFIG_NET to userspace where it is not valid
usr/include/linux/pktcdvd.h:37: leaks CONFIG_CDROM_PKTCDVD_WCACHE to userspace where it is not valid
usr/include/linux/hw_breakpoint.h:27: leaks CONFIG_HAVE_MIXED_BREAKPOINTS_REGS to userspace where it is not valid
usr/include/linux/raw.h:17: leaks CONFIG_MAX_RAW_DEVS to userspace where it is not valid
usr/include/linux/elfcore.h:62: leaks CONFIG_BINFMT_ELF_FDPIC to userspace where it is not valid
usr/include/linux/eventpoll.h:82: leaks CONFIG_PM_SLEEP to userspace where it is not valid
usr/include/linux/atmdev.h:104: leaks CONFIG_COMPAT to userspace where it is not valid
usr/include/asm-generic/unistd.h:651: leaks CONFIG_MMU to userspace where it is not valid
usr/include/asm-generic/bitsperlong.h:9: leaks CONFIG_64BIT to userspace where it is not valid
usr/include/asm-generic/fcntl.h:119: leaks CONFIG_64BIT to userspace where it is not valid
usr/include/asm/auxvec.h:14: leaks CONFIG_IA32_EMULATION to userspace where it is not valid
usr/include/asm/e820.h:14: leaks CONFIG_NODES_SHIFT to userspace where it is not valid
usr/include/asm/e820.h:39: leaks CONFIG_X86_PMEM_LEGACY to userspace where it is not valid
usr/include/asm/e820.h:49: leaks CONFIG_INTEL_TXT to userspace where it is not valid
usr/include/asm/mman.h:7: leaks CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS to userspace where it is not valid
Most of these are false positives because scripts/headers_check.pl
parses comment lines.
It is also false negative. arch/x86/include/uapi/asm/auxvec.h contains
CONFIG_IA32_EMULATION and CONFIG_X86_64, but the only former is reported.
It would be possible to fix scripts/headers_check.pl, of course.
However, we already have some duplicated checks between headers_check
and CONFIG_UAPI_HEADER_TEST. At this moment of time, there are still
dozens of headers excluded from the header test (usr/include/Makefile),
but we might be able to remove headers_check eventually.
I re-implemented it in scripts/headers_install.sh by using sed because
the most of code in scripts/headers_install.sh is written in sed.
This patch works like this:
[1] Run scripts/unifdef first because we need to drop the code
surrounded by #ifdef __KERNEL__ ... #endif
[2] Remove all C style comments. The sed code is somewhat complicated
since we need to deal with both single and multi line comments.
Precisely speaking, a comment block is replaced with a space just
in case.
CONFIG_FOO/* this is a comment */CONFIG_BAR
should be converted into:
CONFIG_FOO CONFIG_BAR
instead of:
CONFIG_FOOCONFIG_BAR
[3] Match CONFIG_... pattern. It correctly matches to all CONFIG
options that appear in a single line.
After this commit, this would detect the following warnings, all of
which are real ones.
warning: include/uapi/linux/pktcdvd.h: leak CONFIG_CDROM_PKTCDVD_WCACHE to user-space
warning: include/uapi/linux/hw_breakpoint.h: leak CONFIG_HAVE_MIXED_BREAKPOINTS_REGS to user-space
warning: include/uapi/linux/raw.h: leak CONFIG_MAX_RAW_DEVS to user-space
warning: include/uapi/linux/elfcore.h: leak CONFIG_BINFMT_ELF_FDPIC to user-space
warning: include/uapi/linux/eventpoll.h: leak CONFIG_PM_SLEEP to user-space
warning: include/uapi/linux/atmdev.h: leak CONFIG_COMPAT to user-space
warning: include/uapi/asm-generic/fcntl.h: leak CONFIG_64BIT to user-space
warning: arch/x86/include/uapi/asm/auxvec.h: leak CONFIG_IA32_EMULATION to user-space
warning: arch/x86/include/uapi/asm/auxvec.h: leak CONFIG_X86_64 to user-space
warning: arch/x86/include/uapi/asm/mman.h: leak CONFIG_X86_INTEL_MEMORY_PROTECTION_KEYS to user-space
However, it is not nice to show them right now. I created a list of
existing leakages. They are not warned, but a new leakage will be
blocked by the 0-day bot.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Sam Ravnborg <sam@ravnborg.org>
For the single target building %.symtypes from %.S, $(a_flags) is
expanded into the _KERNEL flags even if the object is a part of a
module.
$(real-obj-m:.o=.symtypes): modkern_aflags := $(KBUILD_AFLAGS_MODULE) $(AFLAGS_MODULE)
... would fix the issue, but it is not nice to duplicate similar code
for every suffix.
Implement modkern_aflags in the same way as modkern_cflags.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Currently, Kbuild treats an object as multi-used when any of
$(foo-objs), $(foo-y), $(foo-m) is set. It makes more sense to
check $(foo-) as well.
In the context of foo-$(CONFIG_FOO_FEATURE1), CONFIG_FOO_FEATURE1
could be unset.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Update the build scripts and the version magic to reflect when
CONFIG_PREEMPT_RT is enabled in the same way as CONFIG_PREEMPT is treated.
The resulting version strings:
Linux m 5.3.0-rc1+ #100 SMP Fri Jul 26 ...
Linux m 5.3.0-rc1+ #101 SMP PREEMPT Fri Jul 26 ...
Linux m 5.3.0-rc1+ #102 SMP PREEMPT_RT Fri Jul 26 ...
The module vermagic:
5.3.0-rc1+ SMP mod_unload modversions
5.3.0-rc1+ SMP preempt mod_unload modversions
5.3.0-rc1+ SMP preempt_rt mod_unload modversions
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Flex and bison are used for kconfig, dtc, genksyms, all of which are
host programs. I never imagine the kernel embeds a parser or a lexer.
Move the flex and bison rules to scripts/Makefile.host. This file is
included only when hostprogs-y etc. is present in the Makefile in the
directory. So, parsing these rules are skipped in most of directories.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
We generally expect bison to create not only a C file, but also a
header, which will be included from the lexer.
Currently, Kbuild generates them in separate rules. So, for instance,
when building Kconfig, you will notice bison is invoked twice:
HOSTCC scripts/kconfig/conf.o
HOSTCC scripts/kconfig/confdata.o
HOSTCC scripts/kconfig/expr.o
LEX scripts/kconfig/lexer.lex.c
YACC scripts/kconfig/parser.tab.h
HOSTCC scripts/kconfig/lexer.lex.o
YACC scripts/kconfig/parser.tab.c
HOSTCC scripts/kconfig/parser.tab.o
HOSTCC scripts/kconfig/preprocess.o
HOSTCC scripts/kconfig/symbol.o
HOSTLD scripts/kconfig/conf
Make handles such cases nicely in pattern rules [1]. Merge the two
rules so that one invokcation of bison can generate both of them.
HOSTCC scripts/kconfig/conf.o
HOSTCC scripts/kconfig/confdata.o
HOSTCC scripts/kconfig/expr.o
LEX scripts/kconfig/lexer.lex.c
YACC scripts/kconfig/parser.tab.[ch]
HOSTCC scripts/kconfig/lexer.lex.o
HOSTCC scripts/kconfig/parser.tab.o
HOSTCC scripts/kconfig/preprocess.o
HOSTCC scripts/kconfig/symbol.o
HOSTLD scripts/kconfig/conf
[1] Pattern rule
GNU Make manual says:
"Pattern rules may have more than one target. Unlike normal rules,
this does not act as many different rules with the same prerequisites
and recipe. If a pattern rule has multiple targets, make knows that
the rule's recipe is responsible for making all of the targets. The
recipe is executed only once to make all the targets. When searching
for a pattern rule to match a target, the target patterns of a rule
other than the one that matches the target in need of a rule are
incidental: make worries only about giving a recipe and prerequisites
to the file presently in question. However, when this file's recipe is
run, the other targets are marked as having been updated themselves."
https://www.gnu.org/software/make/manual/html_node/Pattern-Intro.html
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
$(basename ...) trims the last suffix. Using it is more intuitive in
my opinion.
This pattern rule makes %.asn1.c and %.asn1.h at the same time.
Previously, the short log showed only either of them, depending on
the target file in question.
To clarify that two files are being generated by the single recipe,
I changed the log as follows:
Before:
ASN.1 crypto/asymmetric_keys/x509.asn1.c
After:
ASN.1 crypto/asymmetric_keys/x509.asn1.[ch]
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The gold linker has known issues of failing the build both in random and in
predictible ways:
- The x86/X32 VDSO build fails with:
arch/x86/entry/vdso/vclock_gettime-x32.o:vclock_gettime.c:function do_hres:
error: relocation overflow: reference to 'hvclock_page'
That's a known issue for years and the usual workaround is to disable
CONFIG_X86_32
- A recent build failure is caused by turning a relocation into an
absolute one for unknown reasons. See link below.
- There are a couple of gold workarounds applied already, but reports
about broken builds with ld.gold keep coming in on a regular base and in
most cases the root cause is unclear.
In context of the most recent fail H.J. stated:
"Since building a workable kernel for different kernel configurations
isn't a requirement for gold, I don't recommend gold for kernel."
So instead of dealing with attempts to duct tape gold support without
understanding the root cause and without support from the gold folks, fail
the build when gold is detected.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/CAMe9rOqMqkQ0LNpm25yE_Yt0FKp05WmHOrwc0aRDb53miFKM+w@mail.gmail.com
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Tested-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
This patch adds a check to warn about static EXPORT_SYMBOL* functions
during the modpost. In most of the cases, a static symbol marked for
exporting is an odd combination that should be fixed either by deleting
the exporting mark or by removing the static attribute and adding the
appropriate declaration to headers.
This check could help to detect the following problems:
1. 550113d4e9 ("i2c: add newly exported functions to the header, too")
2. 54638c6eaf ("net: phy: make exported variables non-static")
3. 98ef2046f2 ("mm: remove the exporting of totalram_pages")
4. 73df167c81 ("s390/zcrypt: remove the exporting of ap_query_configuration")
5. a57caf8c52 ("sunrpc/cache: remove the exporting of cache_seq_next")
6. e4e4730698 ("crypto: skcipher - remove the exporting of skcipher_walk_next")
7. 14b4c48bb1 ("gve: Remove the exporting of gve_probe")
8. 9b79ee9773 ("scsi: libsas: remove the exporting of sas_wait_eh")
9. ...
The build time impact is very limited and is almost at the unnoticeable
level (< 1 sec).
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>
Signed-off-by: Denis Efremov <efremov@linux.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Since commit ff9b45c55b ("kbuild: modpost: read modules.order instead
of $(MODVERDIR)/*.mod"), a module is no longer built in the following
pattern:
[Makefile]
subdir-y := some-module
[some-module/Makefile]
obj-m := some-module.o
You cannot write Makefile this way in upstream because modules.order is
not correctly generated. subdir-y is used to descend to a sub-directory
that builds tools, device trees, etc.
For external modules, the modules order does not matter. So, the
Makefile above was known to work.
I believe the Makefile should be re-written as follows:
[Makefile]
obj-m := some-module/
[some-module/Makefile]
obj-m := some-module.o
However, people will have no idea if their Makefile suddenly stops
working. In fact, I received questions from multiple people.
Show a warning for a while if obj-m is specified in a Makefile visited
by subdir-y or subdir-m.
I touched the %/ rule to avoid false-positive warnings for the single
target.
Cc: Jan Kiszka <jan.kiszka@siemens.com>
Cc: Tom Stonecypher <thomas.edwardx.stonecypher@intel.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Tested-by: Jan Kiszka <jan.kiszka@siemens.com>
The modules.order files in directories visited by the chain of obj-y
or obj-m are merged to the upper-level ones, and become parts of the
top-level modules.order. On the other hand, there is no need to
generate modules.order in directories visited by subdir-y or subdir-m
since they would become orphan anyway.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
The current implementation of need-builtin is false-positive,
for example, in the following Makefile:
obj-m := foo/
obj-y := foo/bar/
..., where foo/built-in.a is not required.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
I removed the single target %.ko in commit ff9b45c55b ("kbuild:
modpost: read modules.order instead of $(MODVERDIR)/*.mod") because
the modpost stage does not work reliably. For instance, the module
dependency, modversion, etc. do not work if we lack symbol information
from the other modules.
Yet, some people still want to build only one module in their interest,
and it may be still useful if it is used within those limitations.
Fixes: ff9b45c55b ("kbuild: modpost: read modules.order instead of $(MODVERDIR)/*.mod")
Reported-by: Don Brace <don.brace@microsemi.com>
Reported-by: Arend Van Spriel <arend.vanspriel@broadcom.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Prior to this commit, starting nconfig, xconfig or gconfig, and saving
the .config file more than once caused data loss, where a .config file
that contained only comments would be written to disk starting from the
second save operation.
This bug manifests itself because the SYMBOL_WRITTEN flag is never
cleared after the first call to conf_write, and subsequent calls to
conf_write then skip all of the configuration symbols due to the
SYMBOL_WRITTEN flag being set.
This commit resolves this issue by clearing the SYMBOL_WRITTEN flag
from all symbols before conf_write returns.
Fixes: 8e2442a5f8 ("kconfig: fix missing choice values in auto.conf")
Cc: linux-stable <stable@vger.kernel.org> # 4.19+
Signed-off-by: M. Vefa Bicakci <m.v.b@runbox.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
If the particular version of clang a user has doesn't enable
-Werror=unknown-warning-option by default, even though it is the
default[1], then make sure to pass the option to the Kconfig cc-option
command so that testing options from Kconfig files works properly.
Otherwise, depending on the default values setup in the clang toolchain
we will silently assume options such as -Wmaybe-uninitialized are
supported by clang, when they really aren't.
A compilation issue only started happening for me once commit
589834b3a0 ("kbuild: Add -Werror=unknown-warning-option to
CLANG_FLAGS") was applied on top of commit b303c6df80 ("kbuild:
compute false-positive -Wmaybe-uninitialized cases in Kconfig"). This
leads kbuild to try and test for the existence of the
-Wmaybe-uninitialized flag with the cc-option command in
scripts/Kconfig.include, and it doesn't see an error returned from the
option test so it sets the config value to Y. Then the Makefile tries to
pass the unknown option on the command line and
-Werror=unknown-warning-option catches the invalid option and breaks the
build. Before commit 589834b3a0 ("kbuild: Add
-Werror=unknown-warning-option to CLANG_FLAGS") the build works fine,
but any cc-option test of a warning option in Kconfig files silently
evaluates to true, even if the warning option flag isn't supported on
clang.
Note: This doesn't change cc-option usages in Makefiles because those
use a different rule that includes KBUILD_CFLAGS by default (see the
__cc-option command in scripts/Kbuild.incluide). The KBUILD_CFLAGS
variable already has the -Werror=unknown-warning-option flag set. Thanks
to Doug for pointing out the different rule.
[1] https://clang.llvm.org/docs/DiagnosticsReference.html#wunknown-warning-option
Cc: Peter Smith <peter.smith@linaro.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Douglas Anderson <dianders@chromium.org>
Signed-off-by: Stephen Boyd <swboyd@chromium.org>
Reviewed-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Since commit ff9b45c55b ("kbuild: modpost: read modules.order instead
of $(MODVERDIR)/*.mod"), 'make vmlinux' emits a warning, like this:
$ make defconfig vmlinux
[ snip ]
LD vmlinux.o
cat: modules.order: No such file or directory
MODPOST vmlinux.o
MODINFO modules.builtin.modinfo
KSYM .tmp_kallsyms1.o
KSYM .tmp_kallsyms2.o
LD vmlinux
SORTEX vmlinux
SYSMAP System.map
When building only vmlinux, KBUILD_MODULES is not set. Hence, the
modules.order is not generated. For the vmlinux modpost, it is not
necessary at all.
Separate scripts/Makefile.modpost for the vmlinux/modules stages.
This works more efficiently because the vmlinux modpost does not
need to include .*.cmd files.
Fixes: ff9b45c55b ("kbuild: modpost: read modules.order instead of $(MODVERDIR)/*.mod")
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
__modpost is a phony target. The dependency on FORCE is pointless.
All the objects have been built in the previous stage, so the
dependency on the objects are not necessary either.
Count the number of modules in a more straightforward way.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
KBUILD_EXTRA_SYMBOLS makes sense only when building external modules.
Moreover, the modpost sets 'external_module' if the -e option is given.
I replaced $(patsubst %, -e %,...) with simpler $(addprefix -e,...)
while I was here.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
If a build rule fails, the .DELETE_ON_ERROR special target removes the
target, but does nothing for the .*.cmd file, which might be corrupted.
So, .*.cmd files should be included only when the corresponding targets
exist.
Commit 392885ee82 ("kbuild: let fixdep directly write to .*.cmd
files") missed to fix up this file.
Fixes: 392885ee82 ("kbuild: let fixdep directly write to .*.cmd")
Cc: <stable@vger.kernel.org> # v5.0+
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
UAPI headers licensed under GPL are supposed to have exception
"WITH Linux-syscall-note" so that they can be included into non-GPL
user space application code.
Unfortunately, people often miss to add it. Break 'make headers'
when any of exported headers lacks the exception note so that the
0-day bot can easily catch it.
Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
