Re: [PATCH v2 1/2] Provide READ_ONCE_NOCHECK()

[Andrey Ryabinin <aryabinin@virtuozzo.com>]

On 10/14/2015 04:40 PM, Ingo Molnar wrote:
> 
> * Andrey Ryabinin <aryabinin@virtuozzo.com> wrote:
> 
>>
>>
>> On 10/13/2015 07:02 PM, kbuild test robot wrote:
>>> Hi Andrey,
>>>
>>> [auto build test WARNING on tip/auto-latest -- if it's inappropriate base, please suggest rules for selecting the more suitable base]
>>>
>>> url:    https://github.com/0day-ci/linux/commits/Andrey-Ryabinin/Provide-READ_ONCE_NOCHECK/20151013-204127
>>> config: mn10300-allyesconfig (attached as .config)
>>> reproduce:
>>>         wget https://git.kernel.org/cgit/linux/kernel/git/wfg/lkp-tests.git/plain/sbin/make.cross -O ~/bin/make.cross
>>>         chmod +x ~/bin/make.cross
>>>         # save the attached .config to linux build tree
>>>         make.cross ARCH=mn10300 
>>>
>>> All warnings (new ones prefixed by >>):
>>>
>>>    fs/nfs/filelayout/filelayout.c: In function 'filelayout_read_pagelist':
>>>    fs/nfs/filelayout/filelayout.c:476:2: warning: format '%Zu' expects argument of type 'size_t', but argument 5 has type '__u32' [-Wformat=]
>>>      dprintk("--> %s ino %lu pgbase %u req %Zu@%llu\n",
>>>      ^
>>
>> So the actual warning is here. And it's not new.
>>
>>>    In file included from include/linux/nfs_fs.h:28:0,
>>>                     from fs/nfs/filelayout/filelayout.c:32:
>>>    fs/nfs/filelayout/filelayout.c: In function 'filelayout_write_pagelist':
>>>    include/linux/compiler.h:270:8: warning: format '%Zu' expects argument of type 'size_t', but argument 5 has type '__u32' [-Wformat=]
>>>      union { typeof(x) __val; char __c[1]; } __u;  \
>>>            ^
>>>    include/linux/sunrpc/debug.h:33:24: note: in definition of macro 'dfprintk'
>>>        printk(KERN_DEFAULT args); \
>>>                            ^
>>>>> include/linux/compiler.h:274:22: note: in expansion of macro '__READ_ONCE'
>>>     #define READ_ONCE(x) __READ_ONCE(x, _check)
>>>                          ^
>>
>> And this 'note: in expansion of macro' belongs to the warning above.
>> So it's a false-positive. 
> 
> So if this is a new warning introduced by these patches then the warning needs to 
> be addresses

It's not a new warning. It's a new 'note: in expansion of macro' message which is
belongs to the old warning from above.
I changed the macros, so obviously it caused change in macro expansion backtrace.
The actual warning existed before this patch.


> - regardless of whether it's a false positive or not.

I wasn't very clear. By false positive I meant that it is kbuild robot's false positive.
The warning itself is not a false positive, but it's an old warning, it's not introduced by this patch.

It seems that kbuild robot treats 'note: in expansion of macro' message as a new warning,
but such messages are not warnings by itself. I think, that kbuild robot should just ignore
new 'note: in expansion of macro' messages unless these messages preceded by a new 'warning:' message.


> Thanks,
> 
> 	Ingo
>