Re: [PATCH 0/7] preempt_count rework -v2

[Linus Torvalds <torvalds@linux-foundation.org>]

On Wed, Sep 11, 2013 at 6:13 AM, Peter Zijlstra <peterz@infradead.org> wrote:
> On Tue, Sep 10, 2013 at 02:43:06PM -0700, Linus Torvalds wrote:
>> That said, looking at your patch, I get the *very* strong feeling that
>> we could make a macro that does all the repetitions for us, and then
>> have a
>>
>>   GENERATE_RMW(atomic_sub_and_test, LOCK_PREFIX "subl", "e", "")
>
> The below seems to compile..

You have that horrible duplication in the macro itself now, though.

The only difference between the 4-byte and the 8-byte one is the "l"
vs "q". And the counter increment/decrement argument (by the caller)
defines whether it's add/dec/inc. Why not just add those as parameters
to the macro, and suddenly the macro becomes 10x smaller.

An excessively complex macro that makes the arguments trivially
simpler is not worth it.

Especially since that excessive macro complexity now means that your
macro is useless for things that the *simpler* macro could have done,
like the bit-test-and-modify cases.

So your complexity actually makes things worse.

So just pass in the operation name and size. Then somebody will use it
for test_and_set_bit() and friends too.

              Linus