[PATCH] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[PATCH] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Arnd Bergmann]

gcc-8 complains about the prototype for this function:

lib/ubsan.c:432:1: error: ignoring attribute 'noreturn' in declaration of a built-in function '__ubsan_handle_builtin_unreachable' because it conflicts with attribute 'const' [-Werror=attributes]

This removes the noreturn attribute.

Signed-off-by: Arnd Bergmann <arnd@arndb.de>
---
 lib/ubsan.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/ubsan.c b/lib/ubsan.c
index 59fee96c29a0..aa817d79c47c 100644
--- a/lib/ubsan.c
+++ b/lib/ubsan.c
@@ -427,7 +427,7 @@ void __ubsan_handle_shift_out_of_bounds(struct shift_out_of_bounds_data *data,
 EXPORT_SYMBOL(__ubsan_handle_shift_out_of_bounds);
 
 
-void __noreturn
+void
 __ubsan_handle_builtin_unreachable(struct unreachable_data *data)
 {
 	unsigned long flags;
-- 
2.9.0

Re: [PATCH] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Andrey Ryabinin]

On 02/02/2018 06:47 PM, Arnd Bergmann wrote:
> gcc-8 complains about the prototype for this function:
> 
> lib/ubsan.c:432:1: error: ignoring attribute 'noreturn' in declaration of a built-in function '__ubsan_handle_builtin_unreachable' because it conflicts with attribute 'const' [-Werror=attributes]
> 

That's actually a bug in GCC. In GCC __ubsan_handle_builtin_unreachable declared
with 'noreturn' and 'const' attributes:

DEF_SANITIZER_BUILTIN(BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE,                                                                                                                                                    
                      "__ubsan_handle_builtin_unreachable",                                                                                                                                                         
                      BT_FN_VOID_PTR,                                                                                                                                                                               
                      ATTR_COLD_CONST_NORETURN_NOTHROW_LEAF_LIST)


But const attribute doesn't make any sense for function that returns void or doesn't return at all.
Given that gcc-8 has not released yet, it would be better to fix this bug there.

Re: [PATCH] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Arnd Bergmann]

On Mon, Feb 5, 2018 at 10:02 AM, Andrey Ryabinin
<aryabinin@virtuozzo.com> wrote:
>
>
> On 02/02/2018 06:47 PM, Arnd Bergmann wrote:
>> gcc-8 complains about the prototype for this function:
>>
>> lib/ubsan.c:432:1: error: ignoring attribute 'noreturn' in declaration of a built-in function '__ubsan_handle_builtin_unreachable' because it conflicts with attribute 'const' [-Werror=attributes]
>>
>
> That's actually a bug in GCC. In GCC __ubsan_handle_builtin_unreachable declared
> with 'noreturn' and 'const' attributes:
>
> DEF_SANITIZER_BUILTIN(BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE,
>                       "__ubsan_handle_builtin_unreachable",
>                       BT_FN_VOID_PTR,
>                       ATTR_COLD_CONST_NORETURN_NOTHROW_LEAF_LIST)
>
>
> But const attribute doesn't make any sense for function that returns void or doesn't return at all.
> Given that gcc-8 has not released yet, it would be better to fix this bug there.

Ok. Should I open a gcc bug, or will you take care of it?

       Arnd

Re: [PATCH] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Andrey Ryabinin]

On 02/05/2018 01:15 PM, Arnd Bergmann wrote:
> On Mon, Feb 5, 2018 at 10:02 AM, Andrey Ryabinin
> <aryabinin@virtuozzo.com> wrote:
>>
>>
>> On 02/02/2018 06:47 PM, Arnd Bergmann wrote:
>>> gcc-8 complains about the prototype for this function:
>>>
>>> lib/ubsan.c:432:1: error: ignoring attribute 'noreturn' in declaration of a built-in function '__ubsan_handle_builtin_unreachable' because it conflicts with attribute 'const' [-Werror=attributes]
>>>
>>
>> That's actually a bug in GCC. In GCC __ubsan_handle_builtin_unreachable declared
>> with 'noreturn' and 'const' attributes:
>>
>> DEF_SANITIZER_BUILTIN(BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE,
>>                       "__ubsan_handle_builtin_unreachable",
>>                       BT_FN_VOID_PTR,
>>                       ATTR_COLD_CONST_NORETURN_NOTHROW_LEAF_LIST)
>>
>>
>> But const attribute doesn't make any sense for function that returns void or doesn't return at all.
>> Given that gcc-8 has not released yet, it would be better to fix this bug there.
> 
> Ok. Should I open a gcc bug, or will you take care of it?
> 

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84210

>        Arnd
> 

Re: [PATCH] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Olof Johansson]

On Mon, Feb 5, 2018 at 2:31 AM Andrey Ryabinin <aryabinin@virtuozzo.com> wrote:
>
> On 02/05/2018 01:15 PM, Arnd Bergmann wrote:
> > On Mon, Feb 5, 2018 at 10:02 AM, Andrey Ryabinin
> > <aryabinin@virtuozzo.com> wrote:
> >>
> >>
> >> On 02/02/2018 06:47 PM, Arnd Bergmann wrote:
> >>> gcc-8 complains about the prototype for this function:
> >>>
> >>> lib/ubsan.c:432:1: error: ignoring attribute 'noreturn' in declaration of a built-in function '__ubsan_handle_builtin_unreachable' because it conflicts with attribute 'const' [-Werror=attributes]
> >>>
> >>
> >> That's actually a bug in GCC. In GCC __ubsan_handle_builtin_unreachable declared
> >> with 'noreturn' and 'const' attributes:
> >>
> >> DEF_SANITIZER_BUILTIN(BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE,
> >>                       "__ubsan_handle_builtin_unreachable",
> >>                       BT_FN_VOID_PTR,
> >>                       ATTR_COLD_CONST_NORETURN_NOTHROW_LEAF_LIST)
> >>
> >>
> >> But const attribute doesn't make any sense for function that returns void or doesn't return at all.
> >> Given that gcc-8 has not released yet, it would be better to fix this bug there.
> >
> > Ok. Should I open a gcc bug, or will you take care of it?
> >
>
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84210

8.2.0 still shows this, and there's been no movement on that bug. How
about we pick up this patch until it's been resolved?


-Olof

Re: [PATCH] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Andrey Ryabinin]

On 11/03/2018 12:27 AM, Olof Johansson wrote:
> On Mon, Feb 5, 2018 at 2:31 AM Andrey Ryabinin <aryabinin@virtuozzo.com> wrote:
>>
>> On 02/05/2018 01:15 PM, Arnd Bergmann wrote:
>>> On Mon, Feb 5, 2018 at 10:02 AM, Andrey Ryabinin
>>> <aryabinin@virtuozzo.com> wrote:
>>>>
>>>>
>>>> On 02/02/2018 06:47 PM, Arnd Bergmann wrote:
>>>>> gcc-8 complains about the prototype for this function:
>>>>>
>>>>> lib/ubsan.c:432:1: error: ignoring attribute 'noreturn' in declaration of a built-in function '__ubsan_handle_builtin_unreachable' because it conflicts with attribute 'const' [-Werror=attributes]
>>>>>
>>>>
>>>> That's actually a bug in GCC. In GCC __ubsan_handle_builtin_unreachable declared
>>>> with 'noreturn' and 'const' attributes:
>>>>
>>>> DEF_SANITIZER_BUILTIN(BUILT_IN_UBSAN_HANDLE_BUILTIN_UNREACHABLE,
>>>>                       "__ubsan_handle_builtin_unreachable",
>>>>                       BT_FN_VOID_PTR,
>>>>                       ATTR_COLD_CONST_NORETURN_NOTHROW_LEAF_LIST)
>>>>
>>>>
>>>> But const attribute doesn't make any sense for function that returns void or doesn't return at all.
>>>> Given that gcc-8 has not released yet, it would be better to fix this bug there.
>>>
>>> Ok. Should I open a gcc bug, or will you take care of it?
>>>
>>
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84210
> 
> 8.2.0 still shows this, and there's been no movement on that bug. How
> about we pick up this patch until it's been resolved?
> 

Sure, I'll add info about GCC bug in changelog and resend the patch.

> 
> -Olof
> 

[PATCH v2] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Andrey Ryabinin]

From: Arnd Bergmann <arnd@arndb.de>

gcc-8 complains about the prototype for this function:

lib/ubsan.c:432:1: error: ignoring attribute 'noreturn' in declaration of a built-in function '__ubsan_handle_builtin_unreachable' because it conflicts with attribute 'const' [-Werror=attributes]

This is actually a GCC's bug. In GCC internals
__ubsan_handle_builtin_unreachable() declared with both 'noreturn' and
'const' attributes instead of only 'noreturn':
   https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84210

Workaround this by removing the noreturn attribute.

[aryabinin: Add information about GCC bug in changelog]
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
---
 lib/ubsan.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

diff --git a/lib/ubsan.c b/lib/ubsan.c
index 59fee96c29a0..e4162f59a81c 100644
--- a/lib/ubsan.c
+++ b/lib/ubsan.c
@@ -427,8 +427,7 @@ void __ubsan_handle_shift_out_of_bounds(struct shift_out_of_bounds_data *data,
 EXPORT_SYMBOL(__ubsan_handle_shift_out_of_bounds);
 
 
-void __noreturn
-__ubsan_handle_builtin_unreachable(struct unreachable_data *data)
+void __ubsan_handle_builtin_unreachable(struct unreachable_data *data)
 {
 	unsigned long flags;
 
-- 
2.18.1

Re: [PATCH v2] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Olof Johansson]

On Wed, Nov 7, 2018 at 6:45 AM Andrey Ryabinin <aryabinin@virtuozzo.com> wrote:
>
> From: Arnd Bergmann <arnd@arndb.de>
>
> gcc-8 complains about the prototype for this function:
>
> lib/ubsan.c:432:1: error: ignoring attribute 'noreturn' in declaration of a built-in function '__ubsan_handle_builtin_unreachable' because it conflicts with attribute 'const' [-Werror=attributes]
>
> This is actually a GCC's bug. In GCC internals
> __ubsan_handle_builtin_unreachable() declared with both 'noreturn' and
> 'const' attributes instead of only 'noreturn':
>    https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84210
>
> Workaround this by removing the noreturn attribute.
>
> [aryabinin: Add information about GCC bug in changelog]
> Signed-off-by: Arnd Bergmann <arnd@arndb.de>
> Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>

Acked-by: Olof Johansson <olof@lixom.net>

While it's a GCC bug, it's there in official releases now and it makes
warnings noisy so we'd rather silence it.


-Olof

Re: [PATCH v2] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Andrew Morton]

On Wed,  7 Nov 2018 17:45:16 +0300 Andrey Ryabinin <aryabinin@virtuozzo.com> wrote:

> From: Arnd Bergmann <arnd@arndb.de>
> 
> gcc-8 complains about the prototype for this function:
> 
> lib/ubsan.c:432:1: error: ignoring attribute 'noreturn' in declaration of a built-in function '__ubsan_handle_builtin_unreachable' because it conflicts with attribute 'const' [-Werror=attributes]
> 
> This is actually a GCC's bug. In GCC internals
> __ubsan_handle_builtin_unreachable() declared with both 'noreturn' and
> 'const' attributes instead of only 'noreturn':
>    https://gcc.gnu.org/bugzilla/show_bug.cgi?id=84210
> 
> Workaround this by removing the noreturn attribute.
> 
> ...
>
> --- a/lib/ubsan.c
> +++ b/lib/ubsan.c
> @@ -427,8 +427,7 @@ void __ubsan_handle_shift_out_of_bounds(struct shift_out_of_bounds_data *data,
>  EXPORT_SYMBOL(__ubsan_handle_shift_out_of_bounds);
>  
>  
> -void __noreturn
> -__ubsan_handle_builtin_unreachable(struct unreachable_data *data)
> +void __ubsan_handle_builtin_unreachable(struct unreachable_data *data)
>  {
>  	unsigned long flags;

This code has been here since 2016 and presumably people will want to
build older kernels with newer gcc's.  So I'll add cc:stable, OK?

Re: [PATCH v2] ubsan: don't mark __ubsan_handle_builtin_unreachable as noreturn

[Arnd Bergmann]

On Wed, Nov 7, 2018 at 9:55 PM Andrew Morton <akpm@linux-foundation.org> wrote:
>
> On Wed,  7 Nov 2018 17:45:16 +0300 Andrey Ryabinin <aryabinin@virtuozzo.com> wrote:
>
> > From: Arnd Bergmann <arnd@arndb.de>

> > --- a/lib/ubsan.c
> > +++ b/lib/ubsan.c
> > @@ -427,8 +427,7 @@ void __ubsan_handle_shift_out_of_bounds(struct shift_out_of_bounds_data *data,
> >  EXPORT_SYMBOL(__ubsan_handle_shift_out_of_bounds);
> >
> >
> > -void __noreturn
> > -__ubsan_handle_builtin_unreachable(struct unreachable_data *data)
> > +void __ubsan_handle_builtin_unreachable(struct unreachable_data *data)
> >  {
> >       unsigned long flags;
>
> This code has been here since 2016 and presumably people will want to
> build older kernels with newer gcc's.  So I'll add cc:stable, OK?
>

Yes, sounds good. Thanks,

      arnd