Re: [PATCHv2] attr: convert to new threadsafe API

From: Stefan Beller <sbeller@google.com>
Date: Oct 12, 2016, 21:47 UTC
Message-ID: <CAGZ79ka5qkDri0gCCLQQcUgqFpOi=np+jm-x046Foc9NOpE6NA@mail.gmail.com>
In-Reply-To: <CA+P7+xqBUT3jUsxciVydO+nRoR+iJygWG=y_ARpiQSs+-kcH2A@mail.gmail.com>

On Wed, Oct 12, 2016 at 2:45 PM, Jacob Keller <jacob.keller@gmail.com> wrote:

Show 22 quoted lines

> On Wed, Oct 12, 2016 at 1:07 PM, Johannes Sixt <j6t@kdbg.org> wrote:
>>
>> Sigh. DCLP, the Double Checked Locking Pattern. These days, it should be
>> common knowledge among professionals that this naïve version _does_not_work_
>> [1]!
>>
>> I suggest you go without it, then measure, and only *then* optimize if it is
>> a bottleneck. Did I read "we do not expect much contention" somewhere?
>>
>> [1] http://www.aristeia.com/Papers/DDJ_Jul_Aug_2004_revised.pdf C++ centric,
>> but applies to C just as well
>>
>> -- Hannes
>>
>
>
> You know, I always wondered why Linux Kernel code needed memory
> barriers but userspace programs didn't seem to use them.. turns out
> they actually *do* need them for the same exact types of problems...
>
> Thanks,
> Jake

In a former job I made use of them, too. So I am kinda embarrassed. (I cannot claim I did not know about these patterns and memory fencing, it just escaped my consciousness).