[PATCH] correct disabling of -Wshift-negative-value

[PATCH] correct disabling of -Wshift-negative-value

[Jan Beulich]

The warning is supported only from gcc6 onwards, hence its disabling
needs probing that the (positive) option is actually supported.

Fixes: 1344794a59db ("Kbuild: add -Wno-shift-negative-value where -Wextra is used")
Signed-off-by: Jan Beulich <jbeulich@suse.com>

--- a/scripts/Makefile.extrawarn
+++ b/scripts/Makefile.extrawarn
@@ -154,7 +154,7 @@ else
 # The following turn off the warnings enabled by -Wextra
 KBUILD_CFLAGS += -Wno-missing-field-initializers
 KBUILD_CFLAGS += -Wno-type-limits
-KBUILD_CFLAGS += -Wno-shift-negative-value
+KBUILD_CFLAGS += $(call cc-disable-warning,shift-negative-value)
 
 ifdef CONFIG_CC_IS_CLANG
 KBUILD_CFLAGS += -Wno-enum-enum-conversion

Re: [PATCH] correct disabling of -Wshift-negative-value

[Nathan Chancellor]

Hi Jan,

On Mon, Apr 14, 2025 at 04:23:36PM +0200, Jan Beulich wrote:
> The warning is supported only from gcc6 onwards, hence its disabling
> needs probing that the (positive) option is actually supported.
> 
> Fixes: 1344794a59db ("Kbuild: add -Wno-shift-negative-value where -Wextra is used")
> Signed-off-by: Jan Beulich <jbeulich@suse.com>

The patch itself seems fine (I might add "scripts/Makefile.extrawarn: "
to the title) but if I understand correctly, this is only to support GCC
5? Would this be necessary if GCC 8.1 becomes the minimum supported
version of GCC for building the kernel?

https://lore.kernel.org/20250407094116.1339199-2-arnd@kernel.org/

If people are actively noticing this, perhaps we should send this to
Linus's tree for backports then Arnd could revert it in his series?

> --- a/scripts/Makefile.extrawarn
> +++ b/scripts/Makefile.extrawarn
> @@ -154,7 +154,7 @@ else
>  # The following turn off the warnings enabled by -Wextra
>  KBUILD_CFLAGS += -Wno-missing-field-initializers
>  KBUILD_CFLAGS += -Wno-type-limits
> -KBUILD_CFLAGS += -Wno-shift-negative-value
> +KBUILD_CFLAGS += $(call cc-disable-warning,shift-negative-value)
>  
>  ifdef CONFIG_CC_IS_CLANG
>  KBUILD_CFLAGS += -Wno-enum-enum-conversion

Cheers,
Nathan

Re: [PATCH] correct disabling of -Wshift-negative-value

[Jan Beulich]

On 14.04.2025 21:38, Nathan Chancellor wrote:
> On Mon, Apr 14, 2025 at 04:23:36PM +0200, Jan Beulich wrote:
>> The warning is supported only from gcc6 onwards, hence its disabling
>> needs probing that the (positive) option is actually supported.
>>
>> Fixes: 1344794a59db ("Kbuild: add -Wno-shift-negative-value where -Wextra is used")
>> Signed-off-by: Jan Beulich <jbeulich@suse.com>
> 
> The patch itself seems fine (I might add "scripts/Makefile.extrawarn: "
> to the title) but if I understand correctly, this is only to support GCC
> 5? Would this be necessary if GCC 8.1 becomes the minimum supported
> version of GCC for building the kernel?
> 
> https://lore.kernel.org/20250407094116.1339199-2-arnd@kernel.org/
> 
> If people are actively noticing this, perhaps we should send this to
> Linus's tree for backports then Arnd could revert it in his series?

Actually, aiui that work was meanwhile merged. Hence the patch here is
indeed only applicable to stable trees. No idea how such a situation
is to be handled. (Apparently the same is true for "correct disabling
of -Wstringop-overflow": That option was introduced in gcc7.)

Jan

Re: [PATCH] correct disabling of -Wshift-negative-value

[Nathan Chancellor]

On Tue, Apr 15, 2025 at 07:53:12AM +0200, Jan Beulich wrote:
> On 14.04.2025 21:38, Nathan Chancellor wrote:
> > On Mon, Apr 14, 2025 at 04:23:36PM +0200, Jan Beulich wrote:
> >> The warning is supported only from gcc6 onwards, hence its disabling
> >> needs probing that the (positive) option is actually supported.
> >>
> >> Fixes: 1344794a59db ("Kbuild: add -Wno-shift-negative-value where -Wextra is used")
> >> Signed-off-by: Jan Beulich <jbeulich@suse.com>
> > 
> > The patch itself seems fine (I might add "scripts/Makefile.extrawarn: "
> > to the title) but if I understand correctly, this is only to support GCC
> > 5? Would this be necessary if GCC 8.1 becomes the minimum supported
> > version of GCC for building the kernel?
> > 
> > https://lore.kernel.org/20250407094116.1339199-2-arnd@kernel.org/
> > 
> > If people are actively noticing this, perhaps we should send this to
> > Linus's tree for backports then Arnd could revert it in his series?
> 
> Actually, aiui that work was meanwhile merged. Hence the patch here is

I think only x86 has been bumped to a minimum of GCC 8.1 so far, done in
commit a3e8fe814ad1 ("x86/build: Raise the minimum GCC version to 8.1").
That series is still being reviewed/worked on as far as I can tell, so I
think these patches would still be needed for other architectures.

> indeed only applicable to stable trees. No idea how such a situation
> is to be handled. (Apparently the same is true for "correct disabling
> of -Wstringop-overflow": That option was introduced in gcc7.)

Normally, you would just send a patch to the stable maintainers and
mailing list just like this but the commit message would say something
like "<xyz problem> was fixed in mainline in such a way that cannot be
easily backported due to <xyz reason> so do <xyz thing> to resolve said
problem in the stable trees".

Cheers,
Nathan

Re: [PATCH] correct disabling of -Wshift-negative-value

[Jan Beulich]

On 15.04.2025 23:15, Nathan Chancellor wrote:
> On Tue, Apr 15, 2025 at 07:53:12AM +0200, Jan Beulich wrote:
>> On 14.04.2025 21:38, Nathan Chancellor wrote:
>>> On Mon, Apr 14, 2025 at 04:23:36PM +0200, Jan Beulich wrote:
>>>> The warning is supported only from gcc6 onwards, hence its disabling
>>>> needs probing that the (positive) option is actually supported.
>>>>
>>>> Fixes: 1344794a59db ("Kbuild: add -Wno-shift-negative-value where -Wextra is used")
>>>> Signed-off-by: Jan Beulich <jbeulich@suse.com>
>>>
>>> The patch itself seems fine (I might add "scripts/Makefile.extrawarn: "
>>> to the title) but if I understand correctly, this is only to support GCC
>>> 5? Would this be necessary if GCC 8.1 becomes the minimum supported
>>> version of GCC for building the kernel?
>>>
>>> https://lore.kernel.org/20250407094116.1339199-2-arnd@kernel.org/
>>>
>>> If people are actively noticing this, perhaps we should send this to
>>> Linus's tree for backports then Arnd could revert it in his series?
>>
>> Actually, aiui that work was meanwhile merged. Hence the patch here is
> 
> I think only x86 has been bumped to a minimum of GCC 8.1 so far, done in
> commit a3e8fe814ad1 ("x86/build: Raise the minimum GCC version to 8.1").
> That series is still being reviewed/worked on as far as I can tell, so I
> think these patches would still be needed for other architectures.

Oh, I see. I merely noticed the bump (after sending) on x86. Then what you
said in your earlier reply is probably the best that can be done for now
(and then also for the other patch).

Jan

>> indeed only applicable to stable trees. No idea how such a situation
>> is to be handled. (Apparently the same is true for "correct disabling
>> of -Wstringop-overflow": That option was introduced in gcc7.)
> 
> Normally, you would just send a patch to the stable maintainers and
> mailing list just like this but the commit message would say something
> like "<xyz problem> was fixed in mainline in such a way that cannot be
> easily backported due to <xyz reason> so do <xyz thing> to resolve said
> problem in the stable trees".
> 
> Cheers,
> Nathan