Re: [PATCH 2.6.31.1] include/linux/kmemcheck.h: fix a sparse warning

[Bart Van Assche <bart.vanassche@gmail.com>]

On Wed, Oct 14, 2009 at 1:26 PM, Vegard Nossum <vegard.nossum@gmail.com> wrote:
> 2009/10/14 Bart Van Assche <bart.vanassche@gmail.com>:
>> On Fri, Oct 2, 2009 at 9:51 AM, Vegard Nossum <vegard.nossum@gmail.com> wrote:
>>>
>>> 2009/10/2 Bart Van Assche <bart.vanassche@gmail.com>:
>>> > On Sat, Sep 26, 2009 at 2:28 PM, Bart Van Assche
>>> > <bart.vanassche@gmail.com> wrote:
>>> >>
>>> >> Checking a 2.6.31.1 kernel configured with allyesconfig/allmodconfig
>>> >> with sparse (make C=2) triggers a sparse warning on code that uses the
>>> >> kmemcheck_annotate_bitfield() macro. An example of such a warning:
>>> >>
>>> >> include/net/inet_sock.h:208:17: warning: do-while statement is not a compound statement
>>> >>
>>> >> Signed-off-by: Bart Van Assche <bart.vanassche@gmail.com>
>>> >> Cc: Vegard Nossum <vegardno@ifi.uio.no>
>>> >> Cc: Andrew Morton <akpm@linux-foundation.org>
>>> >>
>>> >> --- linux-2.6.31.1/include/linux/kmemcheck-orig.h       2009-09-26 13:53:44.000000000 +0200
>>> >> +++ linux-2.6.31.1/include/linux/kmemcheck.h    2009-09-26 13:53:56.000000000 +0200
>>> >> @@ -137,13 +137,13 @@ static inline void kmemcheck_mark_initia
>>> >>        int name##_end[0];
>>> >>
>>> >>  #define kmemcheck_annotate_bitfield(ptr, name)                         \
>>> >> -       do if (ptr) {                                                   \
>>> >> +       do { if (ptr) {                                                 \
>>> >>                int _n = (long) &((ptr)->name##_end)                    \
>>> >>                        - (long) &((ptr)->name##_begin);                \
>>> >>                BUILD_BUG_ON(_n < 0);                                   \
>>> >>                                                                        \
>>> >>                kmemcheck_mark_initialized(&((ptr)->name##_begin), _n); \
>>> >> -       } while (0)
>>> >> +       } } while (0)
>>> >>
>>> >>  #define kmemcheck_annotate_variable(var)                               \
>>> >>        do {                                                            \
>>> >
>>> > (ping)
>>> >
>>> > Did anyone already have the time to review the patch above ?
>>>
>>> A patch for this problem has already been applied in latest mainline.
>>
>> Unfortunately this issue is still present in 2.6.31.4, which has been
>> released on October 12 (yesterday). Is the patch that has been applied
>> in the mainline kernel suitable for backporting ?
>
> I assumed that this wouldn't be suitable for stable kernels, as the
> change is purely syntactic. Is there a good reason for applying this
> patch to the stable series? Perhaps Greg can answer this (Cced)?

Which patch are you referring to (commit ID) ?

Bart.