linux-toradex.git/scripts/mkcompile_h, branch v5.16-rc3

kbuild: mkcompile_h: consider timestamp if KBUILD_BUILD_TIMESTAMP is set

2021-06-17T01:01:12+00:00

To avoid unnecessary recompilations, mkcompile_h does not regenerate
compile.h if just the timestamp changed.
Though, if KBUILD_BUILD_TIMESTAMP is set, an explicit timestamp for the
build was requested, in which case we should not ignore it.

If a user follows the documentation for reproducible builds [1] and
defines KBUILD_BUILD_TIMESTAMP as the git commit timestamp, a clean
build will have the correct timestamp. A subsequent cherry-pick (or
amend) changes the commit timestamp and if an incremental build is done
with a different KBUILD_BUILD_TIMESTAMP now, that new value is not taken
into consideration. But it should for reproducibility.

Hence, whenever KBUILD_BUILD_TIMESTAMP is explicitly set, do not ignore
UTS_VERSION when making a decision about whether the regenerated version
of compile.h should be moved into place.

[1] https://www.kernel.org/doc/html/latest/kbuild/reproducible-builds.html

Signed-off-by: Matthias Maennich 
Signed-off-by: Masahiro Yamada

kbuild: clean up ${quiet} checks in shell scripts

2021-05-26T19:01:50+00:00

There were efforts to make 'make -s' really silent when it is a
warning-free build.

The conventional way was to let a shell script check ${quiet}, and if
it is 'silent_', suppress the stdout by itself.

With the previous commit, the 'cmd' takes care of it now. The 'cmd' is
also invoked from if_changed, if_changed_dep, and if_changed_rule.

You can omit ${quiet} checks in shell scripts when they are invoked
from the 'cmd' macro.

Signed-off-by: Masahiro Yamada

kbuild: Use uname for LINUX_COMPILE_HOST detection

2020-10-20T15:46:04+00:00

`hostname` may not be present on some systems as it's not mandated by
POSIX/SUSv4. This isn't just a theoretical problem: on Arch Linux,
`hostname` is provided by `inetutils`, which isn't part of the base
distribution.

    ./scripts/mkcompile_h: line 38: hostname: command not found

Use `uname -n` instead, which is more likely to be available (and
mandated by standards).

Signed-off-by: Chris Down 
Signed-off-by: Masahiro Yamada

kbuild: use CONFIG_CC_VERSION_TEXT to construct LINUX_COMPILER macro

2020-05-12T04:28:33+00:00

scripts/mkcompile_h runs $(CC) just for getting the version string.
Reuse CONFIG_CC_VERSION_TEXT for optimization.

For GCC, this slightly changes the version string. I do not think it
is a big deal as we do not have the defined format for LINUX_COMPILER.
In fact, the recent commit 4dcc9a88448a ("kbuild: mkcompile_h:
Include $LD version in /proc/version") added the linker version.

Signed-off-by: Masahiro Yamada

kbuild: mkcompile_h: Include $LD version in /proc/version

2020-04-08T15:13:45+00:00

When doing Clang builds of the kernel, it is possible to link with
either ld.bfd (binutils) or ld.lld (LLVM), but it is not possible to
discover this from a running kernel. Add the "$LD -v" output to
/proc/version.

Signed-off-by: Kees Cook 
Reviewed-by: Nick Desaulniers 
Tested-by: Nick Desaulniers 
Reviewed-by: Nathan Chancellor 
Tested-by: Nathan Chancellor 
Reviewed-by: Fangrui Song 
Reviewed-by: Sedat Dilek 
Tested-by: Sedat Dilek 
Signed-off-by: Masahiro Yamada

kbuild: remove the owner check in mkcompile_h

2020-03-02T12:18:19+00:00

This reverts a very old commit, which dates back to the pre-git era:

|commit 5d1cfb5b12f72145d30ba0f53c9f238144b122b8
|Author: Kai Germaschewski 
|Date:   Sat Jul 27 02:53:19 2002 -0500
|
|    kbuild: Fix compiling/installing as different users
|
|    "make bzImage && sudo make install" had the problem that during
|    the "sudo make install" the build system would notice that the information
|    in include/linux/compile.h is not accurate (it says "compiled by ",
|    but we are root), thus causing compile.h to be updated and leading to
|    some recompiles.
|
|    We now only update "compile.h" if the current user is the owner of
|    include/linux/autoconf.h, i.e. the user who did the "make *config". So the
|    above sequence will correctly state "compiled by ".
|
|diff --git a/scripts/mkcompile_h b/scripts/mkcompile_h
|index 6313db96172..cd956380978 100755
|--- a/scripts/mkcompile_h
|+++ b/scripts/mkcompile_h
|@@ -3,6 +3,17 @@ ARCH=$2
| SMP=$3
| CC=$4
|
|+# If compile.h exists already and we don't own autoconf.h
|+# (i.e. we're not the same user who did make *config), don't
|+# modify compile.h
|+# So "sudo make install" won't change the "compiled by "
|+# do "compiled by root"
|+
|+if [ -r $TARGET -a ! -O ../include/linux/autoconf.h ]; then
|+  echo ' (not modified)'
|+  exit 0
|+fi
|+
| if [ -r ../.version ]; then
|   VERSION=`cat ../.version`
| else

The 'make bzImage && sudo make install' problem no longer happens
because commit 1648e4f80506 ("x86, kbuild: make "make install" not
depend on vmlinux") fixed the root cause.

Commit 19514fc665ff ("arm, kbuild: make "make install" not depend on
vmlinux") fixed the similar issue on ARM, with detailed explanation.

So, the rule is that the installation targets should never trigger
the builds of any build artifact. By following it, this check is
unneeded.

Signed-off-by: Masahiro Yamada

mkcompile_h: use printf for LINUX_COMPILE_BY

2019-12-14T06:53:04+00:00

Commit 858805b336be ("kbuild: add $(BASH) to run scripts with
bash-extension") shed light on portability issues. Here is another one.

Since commit f07726048d59 ("Fix handling of backlash character in
LINUX_COMPILE_BY name"), we must escape a backslash contained in
LINUX_COMPILE_BY. This is not working on such distros as Ubuntu.

As the POSIX spec [1] says, if any of the operands contain a backslash
( '\' ) character, the results are implementation-defined.

The actual shell of /bin/sh could be bash, dash, etc. depending on
distros, and the behavior of builtin echo command is different among
them.

The bash builtin echo, unless -e is given, copies the arguments to
stdout without expanding escape sequences (BSD-like behavior).

The dash builtin echo, in contrast, adopts System V behavior, which
does expand escape sequences without any option given.

Even non-builtin /bin/echo behaves differently depending on the system.
Due to these variations, echo is considered as a non-portable command.
Using printf is the common solution to avoid the portability issue.

[1] https://pubs.opengroup.org/onlinepubs/009695399/utilities/echo.html

Fixes: 858805b336be ("kbuild: add $(BASH) to run scripts with bash-extension")
Reported-by: XXing Wei 
Signed-off-by: Masahiro Yamada

mkcompile_h: git rid of UTS_TRUNCATE from LINUX_COMPILE_{BY,HOST}

2019-12-14T06:53:04+00:00

UTS_VERSION is set to struct uts_namespace, hence a too long string
should be truncated so it fits in 64 characters.

On the other hand, LINUX_COMPILE_BY/HOST are not set to uts_namespace.
They are just used in the banners, which do not have specific length
limitation.

I dug into the git history, but I could not find the reason why
these two strings must fit in 64 characters. Remove them.

Now that UTS_VERSION is the only user of UTS_TRUNCATE, I squashed it.

Signed-off-by: Masahiro Yamada

Kbuild: Handle PREEMPT_RT for version string and magic

2019-08-13T16:10:42+00:00

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 
Signed-off-by: Masahiro Yamada

kbuild: remove unnecessary in-subshell execution

2019-01-28T00:11:17+00:00

The commands surrounded by ( ) are executed in a subshell, but in
most cases, we do not need to spawn an extra subshell.

Signed-off-by: Masahiro Yamada