Re: [PATCH] atomic: cleanup asm-generic atomic*.h inclusion

[Mike Frysinger]

On Fri, Jun 17, 2011 at 13:49, Arun Sharma wrote:
> On 6/17/11 10:28 AM, Mike Frysinger wrote:
>> On Fri, Jun 17, 2011 at 13:11, Arun Sharma wrote:
>>> On 6/16/11 11:36 PM, Mike Frysinger wrote:
>>>>
>>>> i'm not sure this doesnt break things. Âwhat tree exactly should i
>>>> apply this patch to to verify things ?
>>>
>>> I tested 3.0-rc2 + patch 1 + patch 2 + this patch.
>>
>> not sure what those patches are, so i picked at random:
>> Subject: [PATCH 2/3] atomic: move atomic_add_unless to generic code
>> Date: Tue, Â7 Jun 2011 17:07:18 -0700
>> Message-Id:<1307491639-9864-2-git-send-email-asharma@xxxxxx>
>>
>> Subject: [PATCH 1/3] atomic: use<linux/atomic.h>
>> Date: Â Mon, Â6 Jun 2011 10:27:29 -0700
>> Message-Id:<1307381251-15729-1-git-send-email-asharma@xxxxxx>
>
> Yes - these are the right patches.
>
>>
>> but even after just these two, things fail because
>> include/asm-generic/atomic.h provides atomic_add_unless() already.
>> i'm guessing your patch 2/3 should have fixed that somehow since it
>> adds the func to linux/atomic.h.
>
> Thanks for catching this. I missed it because I tested only on i386 and
> x86_64. Looks like only 4 archs are affected:
>
> $ grep -r asm-generic.atomic.h *
> blackfin/include/asm/atomic.h:# include <asm-generic/atomic.h>
> microblaze/include/asm/atomic.h:#include <asm-generic/atomic.h>
> mn10300/include/asm/atomic.h:#include <asm-generic/atomic.h>
> score/include/asm/atomic.h:#include <asm-generic/atomic.h>
>
> I believe the correct fix is:
>
> --- a/include/asm-generic/atomic.h
> +++ b/include/asm-generic/atomic.h
> @@ -129,17 +129,6 @@ static inline void atomic_dec(atomic_t *v)
>
> Â#define cmpxchg64_local(ptr, o, n) __cmpxchg64_local_generic((ptr), (o),
> (n))
>
> -static inline int atomic_add_unless(atomic_t *v, int a, int u)
> -{
> - Âint c, old;
> - Âc = atomic_read(v);
> - Âwhile (c != u && (old = atomic_cmpxchg(v, c, c + a)) != c)
> - Â Âc = old;
> - Âreturn c != u;
> -}
> -
> -#define atomic_inc_not_zero(v) atomic_add_unless((v), 1, 0)
> -
> Âstatic inline void atomic_clear_mask(unsigned long mask, unsigned long
> *addr)
> Â{
> Â Â Â Âunsigned long flags;
>
> Could you help test this? I'll resend the patch series once I get
> confirmation.

fixes one thing while breaking another.  linux/atomic.h includes
asm/atomic.h which includes asm-generic/atomic.h which includes
asm-generic/atomic-long.h which needs atomic_add_unless(), but that
isnt provided until after the asm/atomic.h include in linux/atomic.h.

but linux/atomic.h needs asm/atomic.h before atomic_add_unless()
because it relies on the new __atomic_add_unless().

having linux/atomic.h and asm-generic/atomic.h just strikes me as
wrong.  the point of asm-generic is to unify things, but now we have
two places to unify things without 0 indication as to which is for
which ?  i'm wondering if we shouldnt convert all arches to
asm-generic/atomic.h and then add your new logic there and just skip
this whole linux/atomic.h mess.
-mike
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/