[PATCH 5.10] ubsan: remove CONFIG_UBSAN_OBJECT_SIZE

[PATCH 5.10] ubsan: remove CONFIG_UBSAN_OBJECT_SIZE

[Tadeusz Struk]

From: Kees Cook <keescook@chromium.org>

Upstream commit: 69d0db01e210 ("ubsan: remove CONFIG_UBSAN_OBJECT_SIZE")

The object-size sanitizer is redundant to -Warray-bounds, and
inappropriately performs its checks at run-time when all information
needed for the evaluation is available at compile-time, making it quite
difficult to use:

https://bugzilla.kernel.org/show_bug.cgi?id=214861

This run-time object-size checks also trigger false-positive errors,
like the below, that make it quite difficult to test stable kernels in
test automations like syzkaller:

https://syzkaller.appspot.com/text?tag=Error&x=12b3aac3700000

With -Warray-bounds almost enabled globally, it doesn't make sense to
keep this around.

Link: https://lkml.kernel.org/r/20211203235346.110809-1-keescook@chromium.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Reviewed-by: Marco Elver <elver@google.com>
Cc: Masahiro Yamada <masahiroy@kernel.org>
Cc: Michal Marek <michal.lkml@markovi.net>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: "Peter Zijlstra (Intel)" <peterz@infradead.org>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Tadeusz Struk <tadeusz.struk@linaro.org>
---
 lib/test_ubsan.c       | 11 -----------
 scripts/Makefile.ubsan |  1 -
 2 files changed, 12 deletions(-)

diff --git a/lib/test_ubsan.c b/lib/test_ubsan.c
index 9ea10adf7a66..b1d0a6ecfe1b 100644
--- a/lib/test_ubsan.c
+++ b/lib/test_ubsan.c
@@ -89,16 +89,6 @@ static void test_ubsan_misaligned_access(void)
 	*ptr = val;
 }
 
-static void test_ubsan_object_size_mismatch(void)
-{
-	/* "((aligned(8)))" helps this not into be misaligned for ptr-access. */
-	volatile int val __aligned(8) = 4;
-	volatile long long *ptr, val2;
-
-	ptr = (long long *)&val;
-	val2 = *ptr;
-}
-
 static const test_ubsan_fp test_ubsan_array[] = {
 	test_ubsan_add_overflow,
 	test_ubsan_sub_overflow,
@@ -110,7 +100,6 @@ static const test_ubsan_fp test_ubsan_array[] = {
 	test_ubsan_load_invalid_value,
 	//test_ubsan_null_ptr_deref, /* exclude it because there is a crash */
 	test_ubsan_misaligned_access,
-	test_ubsan_object_size_mismatch,
 };
 
 static int __init test_ubsan_init(void)
diff --git a/scripts/Makefile.ubsan b/scripts/Makefile.ubsan
index 9716dab06bc7..2156e18391a3 100644
--- a/scripts/Makefile.ubsan
+++ b/scripts/Makefile.ubsan
@@ -23,7 +23,6 @@ ifdef CONFIG_UBSAN_MISC
       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
-- 
2.35.1

Re: [PATCH 5.10] ubsan: remove CONFIG_UBSAN_OBJECT_SIZE

[Tadeusz Struk]

On 3/31/22 09:02, Tadeusz Struk wrote:
> Upstream commit: 69d0db01e210 ("ubsan: remove CONFIG_UBSAN_OBJECT_SIZE")
> 
> The object-size sanitizer is redundant to -Warray-bounds, and
> inappropriately performs its checks at run-time when all information
> needed for the evaluation is available at compile-time, making it quite
> difficult to use:
> 
> https://bugzilla.kernel.org/show_bug.cgi?id=214861
> 
> This run-time object-size checks also trigger false-positive errors,
> like the below, that make it quite difficult to test stable kernels in
> test automations like syzkaller:
> 
> https://syzkaller.appspot.com/text?tag=Error&x=12b3aac3700000
> 
> With -Warray-bounds almost enabled globally, it doesn't make sense to
> keep this around.

Hi,
This back-port is for 5.10 only. Please also cherry-pick the original
commit 69d0db01e210 ("ubsan: remove CONFIG_UBSAN_OBJECT_SIZE")
to 5.15.y and 5.16.y. There is no back-port required for these kernels.

-- 
Thanks,
Tadeusz

Re: [PATCH 5.10] ubsan: remove CONFIG_UBSAN_OBJECT_SIZE

[Greg Kroah-Hartman]

On Thu, Mar 31, 2022 at 09:07:02AM -0700, Tadeusz Struk wrote:
> On 3/31/22 09:02, Tadeusz Struk wrote:
> > Upstream commit: 69d0db01e210 ("ubsan: remove CONFIG_UBSAN_OBJECT_SIZE")
> > 
> > The object-size sanitizer is redundant to -Warray-bounds, and
> > inappropriately performs its checks at run-time when all information
> > needed for the evaluation is available at compile-time, making it quite
> > difficult to use:
> > 
> > https://bugzilla.kernel.org/show_bug.cgi?id=214861
> > 
> > This run-time object-size checks also trigger false-positive errors,
> > like the below, that make it quite difficult to test stable kernels in
> > test automations like syzkaller:
> > 
> > https://syzkaller.appspot.com/text?tag=Error&x=12b3aac3700000
> > 
> > With -Warray-bounds almost enabled globally, it doesn't make sense to
> > keep this around.
> 
> Hi,
> This back-port is for 5.10 only. Please also cherry-pick the original
> commit 69d0db01e210 ("ubsan: remove CONFIG_UBSAN_OBJECT_SIZE")
> to 5.15.y and 5.16.y. There is no back-port required for these kernels.

Now queued up, thanks.

greg k-h