Re: [PATCH 1/3] kbuild: provide reasonable defaults for tool coverage

[Masahiro Yamada]

On Tue, May 28, 2024 at 8:36 PM Arnd Bergmann <arnd@xxxxxxxx> wrote:
>
> On Mon, May 6, 2024, at 15:35, Masahiro Yamada wrote:
> > The objtool, sanitizers (KASAN, UBSAN, etc.), and profilers (GCOV, etc.)
> > are intended for kernel space objects. To exclude objects from their
> > coverage, you need to set variables such as OBJECT_FILES_NON_STNDARD=y,
> > KASAN_SANITIZE=n, etc.
> >
> > For instance, the following are not kernel objects, and therefore should
> > opt out of coverage:
> >
> >   - vDSO
> >   - purgatory
> >   - bootloader (arch/*/boot/)
> >
> > Kbuild can detect these cases without relying on such variables because
> > objects not directly linked to vmlinux or modules are considered
> > "non-standard objects".
> >
> > Detecting objects linked to vmlinux or modules is straightforward:
> >
> >   - objects added to obj-y are linked to vmlinux
> >   - objects added to lib-y are linked to vmlinux
> >   - objects added to obj-m are linked to modules
> >
>
> I noticed new randconfig build warnings and bisected them
> down to this patch:
>
> warning: unsafe memchr_inv() usage lacked '__read_overflow' symbol in lib/test_fortify/read_overflow-memchr_inv.c
> warning: unsafe memchr() usage lacked '__read_overflow' warning in lib/test_fortify/read_overflow-memchr.c
> warning: unsafe memscan() usage lacked '__read_overflow' symbol in lib/test_fortify/read_overflow-memscan.c
> warning: unsafe memcmp() usage lacked '__read_overflow' warning in lib/test_fortify/read_overflow-memcmp.c
> warning: unsafe memcpy() usage lacked '__read_overflow2' symbol in lib/test_fortify/read_overflow2-memcpy.c
> warning: unsafe memcmp() usage lacked '__read_overflow2' warning in lib/test_fortify/read_overflow2-memcmp.c
> warning: unsafe memmove() usage lacked '__read_overflow2' symbol in lib/test_fortify/read_overflow2-memmove.c
> warning: unsafe memcpy() usage lacked '__read_overflow2_field' symbol in lib/test_fortify/read_overflow2_field-memcpy.c
> warning: unsafe memmove() usage lacked '__read_overflow2_field' symbol in lib/test_fortify/read_overflow2_field-memmove.c
> warning: unsafe memcpy() usage lacked '__write_overflow' symbol in lib/test_fortify/write_overflow-memcpy.c
> warning: unsafe memmove() usage lacked '__write_overflow' symbol in lib/test_fortify/write_overflow-memmove.c
> warning: unsafe memset() usage lacked '__write_overflow' symbol in lib/test_fortify/write_overflow-memset.c
> warning: unsafe strcpy() usage lacked '__write_overflow' symbol in lib/test_fortify/write_overflow-strcpy-lit.c
> warning: unsafe strcpy() usage lacked '__write_overflow' symbol in lib/test_fortify/write_overflow-strcpy.c
> warning: unsafe strncpy() usage lacked '__write_overflow' symbol in lib/test_fortify/write_overflow-strncpy-src.c
> warning: unsafe strncpy() usage lacked '__write_overflow' symbol in lib/test_fortify/write_overflow-strncpy.c
> warning: unsafe strscpy() usage lacked '__write_overflow' symbol in lib/test_fortify/write_overflow-strscpy.c
> warning: unsafe memcpy() usage lacked '__write_overflow_field' symbol in lib/test_fortify/write_overflow_field-memcpy.c
> warning: unsafe memmove() usage lacked '__write_overflow_field' symbol in lib/test_fortify/write_overflow_field-memmove.c
> warning: unsafe memset() usage lacked '__write_overflow_field' symbol in lib/test_fortify/write_overflow_field-memset.c
>
> I don't understand the nature of this warning, but I see
> that your patch ended up dropping -fsanitize=kernel-address
> from the compiler flags because the lib/test_fortify/*.c files
> don't match the $(is-kernel-object) rule. Adding back
> -fsanitize=kernel-address shuts up these warnings.


In my understanding, fortify-string is independent of KASAN.

I do not understand why -fsanitize=kernel-address matters.



> I've applied a local workaround in my randconfig tree
>
> diff --git a/lib/Makefile b/lib/Makefile
> index ddcb76b294b5..d7b8fab64068 100644
> --- a/lib/Makefile
> +++ b/lib/Makefile
> @@ -425,5 +425,7 @@ $(obj)/$(TEST_FORTIFY_LOG): $(addprefix $(obj)/, $(TEST_FORTIFY_LOGS)) FORCE
>
>  # Fake dependency to trigger the fortify tests.
>  ifeq ($(CONFIG_FORTIFY_SOURCE),y)
> +ifndef CONFIG_KASAN
>  $(obj)/string.o: $(obj)/$(TEST_FORTIFY_LOG)
> +endif
>  endif
>
>
> which I don't think we want upstream. Can you and Kees come
> up with a proper fix instead?
>


I set CONFIG_FORTIFY_SOURCE=y and CONFIG_KASAN=y,
but I did not observe such warnings.
Is this arch or compiler-specific?


Could you provide me with the steps to reproduce it?





--
Best Regards
Masahiro Yamada