[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20201004070847.GA1650@Ryzen-9-3900X.localdomain>
Date: Sun, 4 Oct 2020 00:08:47 -0700
From: Nathan Chancellor <natechancellor@...il.com>
To: Kees Cook <keescook@...omium.org>
Cc: Andrew Morton <akpm@...ux-foundation.org>,
Linus Torvalds <torvalds@...ux-foundation.org>,
Ard Biesheuvel <ardb@...nel.org>,
Arnd Bergmann <arnd@...db.de>,
Masahiro Yamada <masahiroy@...nel.org>,
Michal Marek <michal.lkml@...kovi.net>,
Nick Desaulniers <ndesaulniers@...gle.com>,
Marco Elver <elver@...gle.com>,
Randy Dunlap <rdunlap@...radead.org>,
Dmitry Vyukov <dvyukov@...gle.com>,
George Popescu <georgepope@...roid.com>,
Herbert Xu <herbert@...dor.apana.org.au>,
Peter Oberparleiter <oberpar@...ux.ibm.com>,
Andrey Ryabinin <aryabinin@...tuozzo.com>,
clang-built-linux@...glegroups.com, linux-kbuild@...r.kernel.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH 1/4] ubsan: Move cc-option tests into Kconfig
On Fri, Oct 02, 2020 at 03:15:24PM -0700, Kees Cook wrote:
> Instead of doing if/endif blocks with cc-option calls in the UBSAN
> Makefile, move all the tests into Kconfig and use the Makefile to
> collect the results.
>
> Suggested-by: Linus Torvalds <torvalds@...ux-foundation.org>
> Signed-off-by: Kees Cook <keescook@...omium.org>
> Link: https://lore.kernel.org/lkml/CAHk-=wjPasyJrDuwDnpHJS2TuQfExwe=px-SzLeN8GFMAQJPmQ@mail.gmail.com/
I tested menuconfig to make sure all the flags when CONFIG_UBSAN_MISC is
flipped.
Reviewed-by: Nathan Chancellor <natechancellor@...il.com>
Tested-by: Nathan Chancellor <natechancellor@...il.com>
One comment below.
> ---
> lib/Kconfig.ubsan | 48 +++++++++++++++++++++++++++++++++++++++-
> scripts/Makefile.ubsan | 50 ++++++++++++++----------------------------
> 2 files changed, 64 insertions(+), 34 deletions(-)
>
> diff --git a/lib/Kconfig.ubsan b/lib/Kconfig.ubsan
> index 58f8d03d037b..c0b801871e0b 100644
> --- a/lib/Kconfig.ubsan
> +++ b/lib/Kconfig.ubsan
> @@ -36,10 +36,17 @@ config UBSAN_KCOV_BROKEN
> See https://bugs.llvm.org/show_bug.cgi?id=45831 for the status
> in newer releases.
>
> +config CC_HAS_UBSAN_BOUNDS
> + def_bool $(cc-option,-fsanitize=bounds)
> +
> +config CC_HAS_UBSAN_ARRAY_BOUNDS
> + def_bool $(cc-option,-fsanitize=array-bounds)
> +
> config UBSAN_BOUNDS
> bool "Perform array index bounds checking"
> default UBSAN
> depends on !UBSAN_KCOV_BROKEN
> + depends on CC_HAS_UBSAN_ARRAY_BOUNDS || CC_HAS_UBSAN_BOUNDS
> help
> This option enables detection of directly indexed out of bounds
> array accesses, where the array size is known at compile time.
> @@ -47,11 +54,17 @@ config UBSAN_BOUNDS
> to the {str,mem}*cpy() family of functions (that is addressed
> by CONFIG_FORTIFY_SOURCE).
>
> +config CC_ARG_UBSAN_BOUNDS
> + string
> + default "-fsanitize=array-bounds" if CC_HAS_UBSAN_ARRAY_BOUNDS
> + default "-fsanitize=bounds"
> + depends on UBSAN_BOUNDS
> +
> config UBSAN_LOCAL_BOUNDS
> bool "Perform array local bounds checking"
> depends on UBSAN_TRAP
> - depends on CC_IS_CLANG
> depends on !UBSAN_KCOV_BROKEN
> + depends on $(cc-option,-fsanitize=local-bounds)
> help
> This option enables -fsanitize=local-bounds which traps when an
> exception/error is detected. Therefore, it should be enabled only
> @@ -69,6 +82,38 @@ config UBSAN_MISC
> own Kconfig options. Disable this if you only want to have
> individually selected checks.
>
> +config UBSAN_SHIFT
> + def_bool UBSAN_MISC
> + depends on $(cc-option,-fsanitize=shift)
> +
> +config UBSAN_DIV_ZERO
> + def_bool UBSAN_MISC
> + depends on $(cc-option,-fsanitize=integer-divide-by-zero)
> +
> +config UBSAN_UNREACHABLE
> + def_bool UBSAN_MISC
> + depends on $(cc-option,-fsanitize=unreachable)
> +
> +config UBSAN_SIGNED_OVERFLOW
> + def_bool UBSAN_MISC
> + depends on $(cc-option,-fsanitize=signed-integer-overflow)
> +
> +config UBSAN_UNSIGNED_OVERFLOW
> + def_bool UBSAN_MISC
> + depends on $(cc-option,-fsanitize=unsigned-integer-overflow)
> +
> +config UBSAN_OBJECT_SIZE
> + def_bool UBSAN_MISC
> + depends on $(cc-option,-fsanitize=object-size)
> +
> +config UBSAN_BOOL
> + def_bool UBSAN_MISC
> + depends on $(cc-option,-fsanitize=bool)
> +
> +config UBSAN_ENUM
> + def_bool UBSAN_MISC
> + depends on $(cc-option,-fsanitize=enum)
> +
> config UBSAN_SANITIZE_ALL
> bool "Enable instrumentation for the entire kernel"
> depends on ARCH_HAS_UBSAN_SANITIZE_ALL
> @@ -89,6 +134,7 @@ config UBSAN_ALIGNMENT
> bool "Enable checks for pointers alignment"
> default !HAVE_EFFICIENT_UNALIGNED_ACCESS
> depends on !UBSAN_TRAP
> + depends on $(cc-option,-fsanitize=alignment)
> help
> This option enables the check of unaligned memory accesses.
> Enabling this option on architectures that support unaligned
> diff --git a/scripts/Makefile.ubsan b/scripts/Makefile.ubsan
> index 9716dab06bc7..72862da47baf 100644
> --- a/scripts/Makefile.ubsan
> +++ b/scripts/Makefile.ubsan
> @@ -1,37 +1,21 @@
> # SPDX-License-Identifier: GPL-2.0
>
> -export CFLAGS_UBSAN :=
> +# -fsanitize=* options makes GCC less smart than usual and
> +# increases the number of 'maybe-uninitialized' false-positives.
> +ubsan-cflags-$(CONFIG_UBSAN) += $(call cc-disable-warning, maybe-uninitialized)
Is this just to force -Wno-maybe-uninitialized even when W=2?
-Wmaybe-uninitialized is already disabled globally after
commit 78a5255ffb6a ("Stop the ad-hoc games with
-Wno-maybe-initialized"). I feel like it might be worth a comment in
case that changes in the future but maybe that is a bit much.
> -ifdef CONFIG_UBSAN_ALIGNMENT
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize=alignment)
> -endif
> +# Enable available and selected UBSAN features.
> +ubsan-cflags-$(CONFIG_UBSAN_ALIGNMENT) += -fsanitize=alignment
> +ubsan-cflags-$(CONFIG_UBSAN_BOUNDS) += $(CONFIG_CC_ARG_UBSAN_BOUNDS)
> +ubsan-cflags-$(CONFIG_UBSAN_LOCAL_BOUNDS) += -fsanitize=local-bounds
> +ubsan-cflags-$(CONFIG_UBSAN_SHIFT) += -fsanitize=shift
> +ubsan-cflags-$(CONFIG_UBSAN_DIV_ZERO) += -fsanitize=integer-divide-by-zero
> +ubsan-cflags-$(CONFIG_UBSAN_UNREACHABLE) += -fsanitize=unreachable
> +ubsan-cflags-$(CONFIG_UBSAN_SIGNED_OVERFLOW) += -fsanitize=signed-integer-overflow
> +ubsan-cflags-$(CONFIG_UBSAN_UNSIGNED_OVERFLOW) += -fsanitize=unsigned-integer-overflow
> +ubsan-cflags-$(CONFIG_UBSAN_OBJECT_SIZE) += -fsanitize=object-size
> +ubsan-cflags-$(CONFIG_UBSAN_BOOL) += -fsanitize=bool
> +ubsan-cflags-$(CONFIG_UBSAN_ENUM) += -fsanitize=enum
> +ubsan-cflags-$(CONFIG_UBSAN_TRAP) += -fsanitize-undefined-trap-on-error
>
> -ifdef CONFIG_UBSAN_BOUNDS
> - ifdef CONFIG_CC_IS_CLANG
> - CFLAGS_UBSAN += -fsanitize=array-bounds
> - else
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize=bounds)
> - endif
> -endif
> -
> -ifdef CONFIG_UBSAN_LOCAL_BOUNDS
> - CFLAGS_UBSAN += -fsanitize=local-bounds
> -endif
> -
> -ifdef CONFIG_UBSAN_MISC
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize=shift)
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize=integer-divide-by-zero)
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize=unreachable)
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize=signed-integer-overflow)
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize=object-size)
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize=bool)
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize=enum)
> -endif
> -
> -ifdef CONFIG_UBSAN_TRAP
> - CFLAGS_UBSAN += $(call cc-option, -fsanitize-undefined-trap-on-error)
> -endif
> -
> - # -fsanitize=* options makes GCC less smart than usual and
> - # increase number of 'maybe-uninitialized false-positives
> - CFLAGS_UBSAN += $(call cc-option, -Wno-maybe-uninitialized)
> +export CFLAGS_UBSAN := $(ubsan-cflags-y)
> --
> 2.25.1
Powered by blists - more mailing lists