[2/6] ubsan: Reintroduce signed and unsigned overflow sanitizers

  Effectively revert commit 6aaa31aeb9cf ("ubsan: remove overflow
checks"), to allow the kernel to be built with the "overflow"
sanitizers again. This gives developers a chance to experiment[1][2][3]
with the instrumentation again, while compilers adjust their sanitizers
to deal with the impact of -fno-strict-oveflow (i.e. moving from
"overflow" checking to "wrap-around" checking).

Notably, the naming of the options is adjusted to use the name "WRAP"
instead of "OVERFLOW". In the strictest sense, arithmetic "overflow"
happens when a result exceeds the storage of the type, and is considered
by the C standard and compilers to be undefined behavior for signed
and pointer types (without -fno-strict-overflow). Unsigned arithmetic
overflow is defined as always wrapping around.

Because the kernel is built with -fno-strict-overflow, signed and pointer
arithmetic is defined to always wrap around instead of "overflowing"
(which could either be elided due to being undefined behavior or would
wrap around, which led to very weird bugs in the kernel).

So, the config options are added back as CONFIG_UBSAN_SIGNED_WRAP and
CONFIG_UBSAN_UNSIGNED_WRAP. Since the kernel has several places that
explicitly depend on wrap-around behavior (e.g. counters, atomics, crypto,
etc), also introduce the __signed_wrap and __unsigned_wrap function
attributes for annotating functions where wrapping is expected and should
not be instrumented. This will allow us to distinguish in the kernel
between intentional and unintentional cases of arithmetic wrap-around.

Additionally keep these disabled under CONFIG_COMPILE_TEST for now.

Link: https://github.com/KSPP/linux/issues/26 [1]
Link: https://github.com/KSPP/linux/issues/27 [2]
Link: https://github.com/KSPP/linux/issues/344 [3]
Cc: Justin Stitt <justinstitt@google.com>
Cc: Miguel Ojeda <ojeda@kernel.org>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Marco Elver <elver@google.com>
Cc: Hao Luo <haoluo@google.com>
Cc: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
---
 include/linux/compiler_types.h | 14 ++++++-
 lib/Kconfig.ubsan              | 19 ++++++++++
 lib/test_ubsan.c               | 49 ++++++++++++++++++++++++
 lib/ubsan.c                    | 68 ++++++++++++++++++++++++++++++++++
 lib/ubsan.h                    |  4 ++
 scripts/Makefile.ubsan         |  2 +
 6 files changed, 155 insertions(+), 1 deletion(-)
  

Hi,

On Mon, Jan 29, 2024 at 10:00:39AM -0800, Kees Cook wrote:
> Effectively revert commit 6aaa31aeb9cf ("ubsan: remove overflow
> checks"), to allow the kernel to be built with the "overflow"
> sanitizers again. This gives developers a chance to experiment[1][2][3]
> with the instrumentation again, while compilers adjust their sanitizers
> to deal with the impact of -fno-strict-oveflow (i.e. moving from
> "overflow" checking to "wrap-around" checking).
>
> Notably, the naming of the options is adjusted to use the name "WRAP"
> instead of "OVERFLOW". In the strictest sense, arithmetic "overflow"
> happens when a result exceeds the storage of the type, and is considered
> by the C standard and compilers to be undefined behavior for signed
> and pointer types (without -fno-strict-overflow). Unsigned arithmetic
> overflow is defined as always wrapping around.
>
> Because the kernel is built with -fno-strict-overflow, signed and pointer
> arithmetic is defined to always wrap around instead of "overflowing"
> (which could either be elided due to being undefined behavior or would
> wrap around, which led to very weird bugs in the kernel).
>
> So, the config options are added back as CONFIG_UBSAN_SIGNED_WRAP and
> CONFIG_UBSAN_UNSIGNED_WRAP. Since the kernel has several places that
> explicitly depend on wrap-around behavior (e.g. counters, atomics, crypto,
> etc), also introduce the __signed_wrap and __unsigned_wrap function
> attributes for annotating functions where wrapping is expected and should
> not be instrumented. This will allow us to distinguish in the kernel
> between intentional and unintentional cases of arithmetic wrap-around.
>
> Additionally keep these disabled under CONFIG_COMPILE_TEST for now.

This is present in the patch but perhaps its worth noting here that x86
has trouble booting with the unsigned-integer-overflow sanitizer on.

>
> Link: https://github.com/KSPP/linux/issues/26 [1]
> Link: https://github.com/KSPP/linux/issues/27 [2]
> Link: https://github.com/KSPP/linux/issues/344 [3]
> Cc: Justin Stitt <justinstitt@google.com>
> Cc: Miguel Ojeda <ojeda@kernel.org>
> Cc: Nathan Chancellor <nathan@kernel.org>
> Cc: Nick Desaulniers <ndesaulniers@google.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Marco Elver <elver@google.com>
> Cc: Hao Luo <haoluo@google.com>
> Cc: Przemek Kitszel <przemyslaw.kitszel@intel.com>
> Signed-off-by: Kees Cook <keescook@chromium.org>


This patch adheres to the language semantics as I understand them.
Moreover, we would've had to send a patch similar to this once we land
some better sanitizer + -fno-strict-oveflow support in the compilers.

Currently, though, -fsanitize=signed-integer-overflow instruments very
little (if anything at all) due to compiler optimizations in conjunction
with -fno-strict-oveflow. I am working on a new
-fsanitize=signed-integer-wrap in Clang which will instrument more
arithmetic even under -fno-strict-oveflow.


Reviewed-by: Justin Stitt <justinstitt@google.com>

> ---
>  include/linux/compiler_types.h | 14 ++++++-
>  lib/Kconfig.ubsan              | 19 ++++++++++
>  lib/test_ubsan.c               | 49 ++++++++++++++++++++++++
>  lib/ubsan.c                    | 68 ++++++++++++++++++++++++++++++++++
>  lib/ubsan.h                    |  4 ++
>  scripts/Makefile.ubsan         |  2 +
>  6 files changed, 155 insertions(+), 1 deletion(-)
>
> diff --git a/include/linux/compiler_types.h b/include/linux/compiler_types.h
> index 6f1ca49306d2..e585614f3152 100644
> --- a/include/linux/compiler_types.h
> +++ b/include/linux/compiler_types.h
> @@ -282,11 +282,23 @@ struct ftrace_likely_data {
>  #define __no_sanitize_or_inline __always_inline
>  #endif
>
> +/* Allow wrapping arithmetic within an annotated function. */
> +#ifdef CONFIG_UBSAN_SIGNED_WRAP
> +# define __signed_wrap __attribute__((no_sanitize("signed-integer-overflow")))
> +#else
> +# define __signed_wrap
> +#endif
> +#ifdef CONFIG_UBSAN_UNSIGNED_WRAP
> +# define __unsigned_wrap __attribute__((no_sanitize("unsigned-integer-overflow")))
> +#else
> +# define __unsigned_wrap
> +#endif
> +
>  /* Section for code which can't be instrumented at all */
>  #define __noinstr_section(section)					\
>  	noinline notrace __attribute((__section__(section)))		\
>  	__no_kcsan __no_sanitize_address __no_profile __no_sanitize_coverage \
> -	__no_sanitize_memory
> +	__no_sanitize_memory __signed_wrap __unsigned_wrap
>
>  #define noinstr __noinstr_section(".noinstr.text")
>
> diff --git a/lib/Kconfig.ubsan b/lib/Kconfig.ubsan
> index 59e21bfec188..a7003e5bd2a1 100644
> --- a/lib/Kconfig.ubsan
> +++ b/lib/Kconfig.ubsan
> @@ -116,6 +116,25 @@ config UBSAN_UNREACHABLE
>  	  This option enables -fsanitize=unreachable which checks for control
>  	  flow reaching an expected-to-be-unreachable position.
>
> +config UBSAN_SIGNED_WRAP
> +	bool "Perform checking for signed arithmetic wrap-around"
> +	default UBSAN
> +	depends on !COMPILE_TEST
> +	depends on $(cc-option,-fsanitize=signed-integer-overflow)
> +	help
> +	  This option enables -fsanitize=signed-integer-overflow which checks
> +	  for wrap-around of any arithmetic operations with signed integers.
> +
> +config UBSAN_UNSIGNED_WRAP
> +	bool "Perform checking for unsigned arithmetic wrap-around"
> +	depends on $(cc-option,-fsanitize=unsigned-integer-overflow)
> +	depends on !X86_32 # avoid excessive stack usage on x86-32/clang
> +	depends on !COMPILE_TEST
> +	help
> +	  This option enables -fsanitize=unsigned-integer-overflow which checks
> +	  for wrap-around of any arithmetic operations with unsigned integers. This
> +	  currently causes x86 to fail to boot.
> +
>  config UBSAN_BOOL
>  	bool "Perform checking for non-boolean values used as boolean"
>  	default UBSAN
> diff --git a/lib/test_ubsan.c b/lib/test_ubsan.c
> index 2062be1f2e80..84d8092d6c32 100644
> --- a/lib/test_ubsan.c
> +++ b/lib/test_ubsan.c
> @@ -11,6 +11,51 @@ typedef void(*test_ubsan_fp)(void);
>  			#config, IS_ENABLED(config) ? "y" : "n");	\
>  	} while (0)
>
> +static void test_ubsan_add_overflow(void)
> +{
> +	volatile int val = INT_MAX;
> +	volatile unsigned int uval = UINT_MAX;
> +
> +	UBSAN_TEST(CONFIG_UBSAN_SIGNED_WRAP);
> +	val += 2;
> +
> +	UBSAN_TEST(CONFIG_UBSAN_UNSIGNED_WRAP);
> +	uval += 2;
> +}
> +
> +static void test_ubsan_sub_overflow(void)
> +{
> +	volatile int val = INT_MIN;
> +	volatile unsigned int uval = 0;
> +	volatile int val2 = 2;
> +
> +	UBSAN_TEST(CONFIG_UBSAN_SIGNED_WRAP);
> +	val -= val2;
> +
> +	UBSAN_TEST(CONFIG_UBSAN_UNSIGNED_WRAP);
> +	uval -= val2;
> +}
> +
> +static void test_ubsan_mul_overflow(void)
> +{
> +	volatile int val = INT_MAX / 2;
> +	volatile unsigned int uval = UINT_MAX / 2;
> +
> +	UBSAN_TEST(CONFIG_UBSAN_SIGNED_WRAP);
> +	val *= 3;
> +
> +	UBSAN_TEST(CONFIG_UBSAN_UNSIGNED_WRAP);
> +	uval *= 3;
> +}
> +
> +static void test_ubsan_negate_overflow(void)
> +{
> +	volatile int val = INT_MIN;
> +
> +	UBSAN_TEST(CONFIG_UBSAN_SIGNED_WRAP);
> +	val = -val;
> +}
> +
>  static void test_ubsan_divrem_overflow(void)
>  {
>  	volatile int val = 16;
> @@ -90,6 +135,10 @@ static void test_ubsan_misaligned_access(void)
>  }
>
>  static const test_ubsan_fp test_ubsan_array[] = {
> +	test_ubsan_add_overflow,
> +	test_ubsan_sub_overflow,
> +	test_ubsan_mul_overflow,
> +	test_ubsan_negate_overflow,
>  	test_ubsan_shift_out_of_bounds,
>  	test_ubsan_out_of_bounds,
>  	test_ubsan_load_invalid_value,
> diff --git a/lib/ubsan.c b/lib/ubsan.c
> index df4f8d1354bb..5fc107f61934 100644
> --- a/lib/ubsan.c
> +++ b/lib/ubsan.c
> @@ -222,6 +222,74 @@ static void ubsan_epilogue(void)
>  	check_panic_on_warn("UBSAN");
>  }
>
> +static void handle_overflow(struct overflow_data *data, void *lhs,
> +			void *rhs, char op)
> +{
> +
> +	struct type_descriptor *type = data->type;
> +	char lhs_val_str[VALUE_LENGTH];
> +	char rhs_val_str[VALUE_LENGTH];
> +
> +	if (suppress_report(&data->location))
> +		return;
> +
> +	ubsan_prologue(&data->location, type_is_signed(type) ?
> +			"signed-integer-overflow" :
> +			"unsigned-integer-overflow");
> +
> +	val_to_string(lhs_val_str, sizeof(lhs_val_str), type, lhs);
> +	val_to_string(rhs_val_str, sizeof(rhs_val_str), type, rhs);
> +	pr_err("%s %c %s cannot be represented in type %s\n",
> +		lhs_val_str,
> +		op,
> +		rhs_val_str,
> +		type->type_name);
> +
> +	ubsan_epilogue();
> +}
> +
> +void __ubsan_handle_add_overflow(void *data,
> +				void *lhs, void *rhs)
> +{
> +
> +	handle_overflow(data, lhs, rhs, '+');
> +}
> +EXPORT_SYMBOL(__ubsan_handle_add_overflow);
> +
> +void __ubsan_handle_sub_overflow(void *data,
> +				void *lhs, void *rhs)
> +{
> +	handle_overflow(data, lhs, rhs, '-');
> +}
> +EXPORT_SYMBOL(__ubsan_handle_sub_overflow);
> +
> +void __ubsan_handle_mul_overflow(void *data,
> +				void *lhs, void *rhs)
> +{
> +	handle_overflow(data, lhs, rhs, '*');
> +}
> +EXPORT_SYMBOL(__ubsan_handle_mul_overflow);
> +
> +void __ubsan_handle_negate_overflow(void *_data, void *old_val)
> +{
> +	struct overflow_data *data = _data;
> +	char old_val_str[VALUE_LENGTH];
> +
> +	if (suppress_report(&data->location))
> +		return;
> +
> +	ubsan_prologue(&data->location, "negation-overflow");
> +
> +	val_to_string(old_val_str, sizeof(old_val_str), data->type, old_val);
> +
> +	pr_err("negation of %s cannot be represented in type %s:\n",
> +		old_val_str, data->type->type_name);
> +
> +	ubsan_epilogue();
> +}
> +EXPORT_SYMBOL(__ubsan_handle_negate_overflow);
> +
> +
>  void __ubsan_handle_divrem_overflow(void *_data, void *lhs, void *rhs)
>  {
>  	struct overflow_data *data = _data;
> diff --git a/lib/ubsan.h b/lib/ubsan.h
> index 5d99ab81913b..0abbbac8700d 100644
> --- a/lib/ubsan.h
> +++ b/lib/ubsan.h
> @@ -124,6 +124,10 @@ typedef s64 s_max;
>  typedef u64 u_max;
>  #endif
>
> +void __ubsan_handle_add_overflow(void *data, void *lhs, void *rhs);
> +void __ubsan_handle_sub_overflow(void *data, void *lhs, void *rhs);
> +void __ubsan_handle_mul_overflow(void *data, void *lhs, void *rhs);
> +void __ubsan_handle_negate_overflow(void *_data, void *old_val);
>  void __ubsan_handle_divrem_overflow(void *_data, void *lhs, void *rhs);
>  void __ubsan_handle_type_mismatch(struct type_mismatch_data *data, void *ptr);
>  void __ubsan_handle_type_mismatch_v1(void *_data, void *ptr);
> diff --git a/scripts/Makefile.ubsan b/scripts/Makefile.ubsan
> index 7cf42231042b..7b2f3d554c59 100644
> --- a/scripts/Makefile.ubsan
> +++ b/scripts/Makefile.ubsan
> @@ -8,6 +8,8 @@ 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_WRAP)	+= -fsanitize=signed-integer-overflow
> +ubsan-cflags-$(CONFIG_UBSAN_UNSIGNED_WRAP)	+= -fsanitize=unsigned-integer-overflow
>  ubsan-cflags-$(CONFIG_UBSAN_BOOL)		+= -fsanitize=bool
>  ubsan-cflags-$(CONFIG_UBSAN_ENUM)		+= -fsanitize=enum
>  ubsan-cflags-$(CONFIG_UBSAN_TRAP)		+= $(call cc-option,-fsanitize-trap=undefined,-fsanitize-undefined-trap-on-error)
> --
> 2.34.1
>

Thanks
Justin
  

On Mon, Jan 29, 2024 at 07:54:18PM +0000, Justin Stitt wrote:
> Hi,
> 
> On Mon, Jan 29, 2024 at 10:00:39AM -0800, Kees Cook wrote:
> > Effectively revert commit 6aaa31aeb9cf ("ubsan: remove overflow
> > checks"), to allow the kernel to be built with the "overflow"
> > sanitizers again. This gives developers a chance to experiment[1][2][3]
> > with the instrumentation again, while compilers adjust their sanitizers
> > to deal with the impact of -fno-strict-oveflow (i.e. moving from
> > "overflow" checking to "wrap-around" checking).
> >
> > Notably, the naming of the options is adjusted to use the name "WRAP"
> > instead of "OVERFLOW". In the strictest sense, arithmetic "overflow"
> > happens when a result exceeds the storage of the type, and is considered
> > by the C standard and compilers to be undefined behavior for signed
> > and pointer types (without -fno-strict-overflow). Unsigned arithmetic
> > overflow is defined as always wrapping around.
> >
> > Because the kernel is built with -fno-strict-overflow, signed and pointer
> > arithmetic is defined to always wrap around instead of "overflowing"
> > (which could either be elided due to being undefined behavior or would
> > wrap around, which led to very weird bugs in the kernel).
> >
> > So, the config options are added back as CONFIG_UBSAN_SIGNED_WRAP and
> > CONFIG_UBSAN_UNSIGNED_WRAP. Since the kernel has several places that
> > explicitly depend on wrap-around behavior (e.g. counters, atomics, crypto,
> > etc), also introduce the __signed_wrap and __unsigned_wrap function
> > attributes for annotating functions where wrapping is expected and should
> > not be instrumented. This will allow us to distinguish in the kernel
> > between intentional and unintentional cases of arithmetic wrap-around.
> >
> > Additionally keep these disabled under CONFIG_COMPILE_TEST for now.
> 
> This is present in the patch but perhaps its worth noting here that x86
> has trouble booting with the unsigned-integer-overflow sanitizer on.

Yeah, though this is fixed later in the series.

> 
> >
> > Link: https://github.com/KSPP/linux/issues/26 [1]
> > Link: https://github.com/KSPP/linux/issues/27 [2]
> > Link: https://github.com/KSPP/linux/issues/344 [3]
> > Cc: Justin Stitt <justinstitt@google.com>
> > Cc: Miguel Ojeda <ojeda@kernel.org>
> > Cc: Nathan Chancellor <nathan@kernel.org>
> > Cc: Nick Desaulniers <ndesaulniers@google.com>
> > Cc: Peter Zijlstra <peterz@infradead.org>
> > Cc: Marco Elver <elver@google.com>
> > Cc: Hao Luo <haoluo@google.com>
> > Cc: Przemek Kitszel <przemyslaw.kitszel@intel.com>
> > Signed-off-by: Kees Cook <keescook@chromium.org>
> 
> 
> This patch adheres to the language semantics as I understand them.
> Moreover, we would've had to send a patch similar to this once we land
> some better sanitizer + -fno-strict-oveflow support in the compilers.
> 
> Currently, though, -fsanitize=signed-integer-overflow instruments very
> little (if anything at all) due to compiler optimizations in conjunction
> with -fno-strict-oveflow. I am working on a new
> -fsanitize=signed-integer-wrap in Clang which will instrument more
> arithmetic even under -fno-strict-oveflow.

Agreed -- I'm mainly getting these back into the kernel so folks working
on this have a common base to work from.

> Reviewed-by: Justin Stitt <justinstitt@google.com>

Thanks!

-Kees