Re: [PATCH] selftests/livepatch: build test modules against the selected kernel

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 2 potential issue(s) to consider:
- [Medium] kbuild: target-specific MAKEOVERRIDES assignment is ignored by GNU Make
- [Medium] selftests: livepatch: tests falsely skip on target machines without KDIR
--

commit 52a5776aaff15a8419816c34023de1b05a2d6d7f
Author: Sasha Levin <sashal@kernel.org>

selftests/livepatch: build test modules against the selected kernel

This commit attempts to fix building livepatch test modules against an output
directory by selecting KDIR from O or KBUILD_OUTPUT. It also aims to stop
forwarding O and KBUILD_OUTPUT to the kernel sub-make so that the build works
correctly against distribution headers or output directories.

> diff --git a/tools/testing/selftests/livepatch/functions.sh b/tools/testing/selftests/livepatch/functions.sh
> index 429fce1f31431..8b26b70f92d8c 100644
> --- a/tools/testing/selftests/livepatch/functions.sh
> +++ b/tools/testing/selftests/livepatch/functions.sh
> @@ -150,6 +150,7 @@ function cleanup() {
>  #		 the ftrace_enabled sysctl.
>  function setup_config() {
>  	is_root
> +	[[ -d "$SYSFS_KLP_DIR" ]] || skip "livepatch is not available"
>  	has_kdir
>  	push_config

[Severity: Medium]
Is the call to has_kdir inside setup_config still appropriate here?

When running tests via the kselftest_install workflow on a target system
without a kernel build tree, has_kdir fails and unconditionally skips the
tests. This causes the tests to falsely skip on installed target machines,
even if the test modules were already successfully compiled and are available
to load.

> diff --git a/tools/testing/selftests/livepatch/test_modules/Makefile b/tools/testing/selftests/livepatch/test_modules/Makefile
> index a13d398585dcb..1fbeaac9c2b2c 100644
> --- a/tools/testing/selftests/livepatch/test_modules/Makefile
> +++ b/tools/testing/selftests/livepatch/test_modules/Makefile
> @@ -16,14 +16,16 @@ obj-m += test_klp_atomic_replace.o \
>  	test_klp_state3.o \
>  	test_klp_syscall.o
>  
> +modules clean: MAKEOVERRIDES := $(filter-out O=% KBUILD_OUTPUT=%,$(MAKEOVERRIDES))

[Severity: Medium]
Will this target-specific assignment of MAKEOVERRIDES be ignored by GNU Make?

GNU Make does not support target-specific assignments for special variables
like MAKEOVERRIDES and MAKEFLAGS. As a result, the global O= and
KBUILD_OUTPUT= overrides will still be passed to the sub-make via MAKEFLAGS.

Could this leak the un-filtered variables to Kbuild and cause the build to
fail against output directories when make O=... is used, defeating the purpose
of the patch?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261014200154.1365200-1-sashal@kernel.org?part=1