[PATCH] build_bug.h: Also define static_assert() when __CHECKER__ is defined

[Christophe Leroy]

Le 04/05/2023 à 21:55, Tom Rini a écrit :
> On Thu, May 04, 2023 at 05:32:12PM +0000, Christophe Leroy wrote:
>>
>>
>> Le 04/05/2023 à 19:22, Christophe Leroy a écrit :
>>>
>>>
>>> Le 04/05/2023 à 15:54, Tom Rini a écrit :
>>>> On Thu, May 04, 2023 at 06:15:13AM +0000, Christophe Leroy wrote:
>>>>>
>>>>>
>>>>> Le 03/05/2023 à 22:02, Tom Rini a écrit :
>>>>>> On Thu, Jan 26, 2023 at 07:17:48PM +0100, Christophe Leroy wrote:
>>>>>>
>>>>>>> When doing a build with C=2, the following failure is encountered on
>>>>>>> several files:
>>>>>>>
>>>>>>>        CHECK   arch/powerpc/cpu/mpc8xxx/fsl_lbc.c
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c: note: in included file
>>>>>>> (through arch/powerpc/include/asm/global_data.h, include/init.h):
>>>>>>>      include/asm-generic/global_data.h:494:21: error: Expected ) in
>>>>>>> function declarator
>>>>>>>      include/asm-generic/global_data.h:494:21: error: got (
>>>>>>>
>>>>>>> And because of the error, the interesting part which are the
>>>>>>> warnings don't appear. This is because static_assert() is defined
>>>>>>> only when __CHECKER__ is not defined.
>>>>>>>
>>>>>>> Add a stub when __CHECKER__ is defined. With that fix, the expected
>>>>>>> warnings are now seen:
>>>>>>>
>>>>>>>        CHECK   arch/powerpc/cpu/mpc8xxx/fsl_lbc.c
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:32:27: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:32:27:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:32:27:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:32:45: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:32:45:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:32:45:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:35:24: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:35:24:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:35:24:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:35:40: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:35:40:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:35:40:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:67:17: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:67:17:    expected unsigned
>>>>>>> int volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:67:17:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:68:17: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:68:17:    expected unsigned
>>>>>>> int volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:68:17:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:72:17: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:72:17:    expected unsigned
>>>>>>> int volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:72:17:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:73:17: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:73:17:    expected unsigned
>>>>>>> int volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:73:17:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:78:9: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:78:9:    expected unsigned
>>>>>>> int volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:78:9:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:79:9: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:79:9:    expected unsigned
>>>>>>> int volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:79:9:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:131:22: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:131:22:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:131:22:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:132:49: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:132:49:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:132:49:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:144:26: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:144:26:    expected unsigned
>>>>>>> int volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:144:26:    got unsigned int
>>>>>>> [usertype] *mxmr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:144:41: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:144:41:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:144:41:    got unsigned int
>>>>>>> [usertype] *mxmr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:145:27: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:145:27:    expected unsigned
>>>>>>> int volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:145:27:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:146:32: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:146:32:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:146:32:    got unsigned int *
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:149:39: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:149:39:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:149:39:    got unsigned int
>>>>>>> [usertype] *mxmr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:155:18: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:155:18:    expected unsigned
>>>>>>> int volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:155:18:    got unsigned int
>>>>>>> [usertype] *mxmr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:155:33: warning: incorrect
>>>>>>> type in argument 1 (different address spaces)
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:155:33:    expected unsigned
>>>>>>> int const volatile [noderef] <asn:2> *addr
>>>>>>>      arch/powerpc/cpu/mpc8xxx/fsl_lbc.c:155:33:    got unsigned int
>>>>>>> [usertype] *mxmr
>>>>>>>
>>>>>>> Signed-off-by: Christophe Leroy <christophe.leroy at csgroup.eu>
>>>>>>
>>>>>> Based on my re-read of the thread in patchwork, it seems like the best
>>>>>> answer would be to re-sync this include, along with a few others, from
>>>>>> the latest kernel release, yes?
>>>>>>
>>>>>
>>>>> Yes that's right, a re-sync should work, together with a revert of
>>>>> aa9e891c633 ("include/linux/stddef.h: avoid 'warning: preprocessor token
>>>>> offsetof redefined'") IIRC
>>>>
>>>> Will you be sending a patch, or should I put it on my TODO list (and how
>>>> exactly do I trigger the above problem?).  Thanks.
>>>>
>>>
>>> Well, I was hoping to find some time during winter but I never
>>> succeeded. So if you can put it on your TODO list that's fine. Maybe you
>>> have some automated way to backport stuff from Linux to U-boot ?
>>>
>>> Reproduction is done easily with:
>>>
>>> make CROSS_COMPILE=powerpc-linux- CMPC885_defconfig
>>>
>>> followed by
>>>
>>> make CROSS_COMPILE=powerpc-linux- C=2
>>>
>>>
>>> Then everytime you get a repetitive syntax-based error, it's something
>>> that shouldn't happen. Examples:
>>>
>>>     CHECK   arch/powerpc/cpu/mpc8xx/cpu.c
>>>
>>> arch/powerpc/cpu/mpc8xx/cpu.c: note: in included file (through
>>> arch/powerpc/include/asm/global_data.h):
>>> include/asm-generic/global_data.h:494:21: error: Expected ) in function
>>> declarator
>>> include/asm-generic/global_data.h:494:21: error: got (
>>>
>>>
>>>
>>> arch/powerpc/cpu/mpc8xx/cpu.c: note: in included file (through
>>> include/compiler.h, include/env.h, include/net.h):
>>> /opt/gcc-12.2.0-nolibc/powerpc-linux/bin/../lib/gcc/powerpc-linux/12.2.0/include/stddef.h:433:14: error: expected ; at end of declaration
>>> /opt/gcc-12.2.0-nolibc/powerpc-linux/bin/../lib/gcc/powerpc-linux/12.2.0/include/stddef.h:433:14: error: Expected } at end of specifier
>>> /opt/gcc-12.2.0-nolibc/powerpc-linux/bin/../lib/gcc/powerpc-linux/12.2.0/include/stddef.h:433:14: error: got __max_align_f128
>>> /opt/gcc-12.2.0-nolibc/powerpc-linux/bin/../lib/gcc/powerpc-linux/12.2.0/include/stddef.h:435:1: error: Expected ; at the end of type declaration
>>> /opt/gcc-12.2.0-nolibc/powerpc-linux/bin/../lib/gcc/powerpc-linux/12.2.0/include/stddef.h:435:1: error: got }
>>>
>>
>> There is the same kind of issue with qemu-x86_64_defconfig :
>>
>> make qemu-x86_64_defconfig
>>
>> make C=2
>>
>> ...
>>     CHECK   arch/x86/lib/bdinfo.c
>> arch/x86/lib/bdinfo.c: note: in included file (through
>> include/compiler.h, arch/x86/include/asm/mrccache.h,
>> arch/x86/include/asm/global_data.h, include/init.h):
>> /usr/lib/gcc/x86_64-redhat-linux/12/include/stddef.h:415:9: warning:
>> preprocessor token offsetof redefined
>> arch/x86/lib/bdinfo.c: note: in included file (through
>> include/linux/posix_types.h, include/linux/types.h, include/time.h,
>> include/common.h):
>> include/linux/stddef.h:19:9: this was the original definition
>> arch/x86/lib/bdinfo.c: note: in included file (through
>> arch/x86/include/asm/global_data.h, include/init.h):
>> include/asm-generic/global_data.h:494:21: error: Expected ) in function
>> declarator
>> include/asm-generic/global_data.h:494:21: error: got (
> 
> Thanks, I _think_ we could add a CI test then, once this is resolved, so
> long as C=2 issues still give a zero exit status, or at least a specific
> one we can catch and say OK to that's not the same as a fail to build.
> 

I finally dig into what I did in January found that it was almost ready. 
So let me send a patch as it seems it fixes all issues.

Christophe