[PATCH] kcov: fix generic Kconfig dependencies if ARCH_WANTS_NO_INSTR

[Marco Elver]

On Thu, 2 Dec 2021 at 18:30, Peter Zijlstra <peterz at infradead.org> wrote:
>
> On Wed, Dec 01, 2021 at 04:26:04PM +0100, Marco Elver wrote:
> > Until recent versions of GCC and Clang, it was not possible to disable
> > KCOV instrumentation via a function attribute. The relevant function
> > attribute was introduced in 540540d06e9d9 ("kcov: add
> > __no_sanitize_coverage to fix noinstr for all architectures").
> >
> > x86 was the first architecture to want a working noinstr, and at the
> > time no compiler support for the attribute existed yet. Therefore,
> > 0f1441b44e823 ("objtool: Fix noinstr vs KCOV") introduced the ability to
> > NOP __sanitizer_cov_*() calls in .noinstr.text.
> >
> > However, this doesn't work for other architectures like arm64 and s390
> > that want a working noinstr per ARCH_WANTS_NO_INSTR.
> >
> > At the time of 0f1441b44e823, we didn't yet have ARCH_WANTS_NO_INSTR,
> > but now we can move the Kconfig dependency checks to the generic KCOV
> > option. KCOV will be available if:
> >
> >       - architecture does not care about noinstr, OR
> >       - we have objtool support (like on x86), OR
> >       - GCC is 12.0 or newer, OR
> >       - Clang is 13.0 or newer.
> >
> > Signed-off-by: Marco Elver <elver at google.com>
> > ---
> >  arch/x86/Kconfig  | 2 +-
> >  lib/Kconfig.debug | 2 ++
> >  2 files changed, 3 insertions(+), 1 deletion(-)
> >
> > diff --git a/arch/x86/Kconfig b/arch/x86/Kconfig
> > index 95dd1ee01546..c030b2ee93b3 100644
> > --- a/arch/x86/Kconfig
> > +++ b/arch/x86/Kconfig
> > @@ -78,7 +78,7 @@ config X86
> >       select ARCH_HAS_FILTER_PGPROT
> >       select ARCH_HAS_FORTIFY_SOURCE
> >       select ARCH_HAS_GCOV_PROFILE_ALL
> > -     select ARCH_HAS_KCOV                    if X86_64 && STACK_VALIDATION
> > +     select ARCH_HAS_KCOV                    if X86_64
> >       select ARCH_HAS_MEM_ENCRYPT
> >       select ARCH_HAS_MEMBARRIER_SYNC_CORE
> >       select ARCH_HAS_NON_OVERLAPPING_ADDRESS_SPACE
> > diff --git a/lib/Kconfig.debug b/lib/Kconfig.debug
> > index 9ef7ce18b4f5..589c8aaa2d5b 100644
> > --- a/lib/Kconfig.debug
> > +++ b/lib/Kconfig.debug
> > @@ -1977,6 +1977,8 @@ config KCOV
> >       bool "Code coverage for fuzzing"
> >       depends on ARCH_HAS_KCOV
> >       depends on CC_HAS_SANCOV_TRACE_PC || GCC_PLUGINS
> > +     depends on !ARCH_WANTS_NO_INSTR || STACK_VALIDATION || \
> > +                GCC_VERSION >= 120000 || CLANG_VERSION >= 130000
>
> Can we write that as something like:
>
>         $(cc-attribute,__no_sanitize_coverage)
>
> instead? Other than that, yes totally.

That'd be nice, but I think we don't have that cc-attribute helper? I
checked how e.g. CC_HAS_NO_PROFILE_FN_ATTR does it, but it won't work
like that because gcc and clang define the attribute differently and
it becomes a mess. That's also what Nathan pointed out here I think:
https://lkml.kernel.org/r/Yaet8x/1WYiADlPh@archlinux-ax161

Let's keep it simple.

> Acked-by: Peter Zijlstra (Intel) <peterz at infradead.org>

Thanks!