svn commit: r334702 - head/sys/sys

[Ravi Pokala]

> I believe the theory is that the compiler (remember, this is __builtin_memset) can optimize away portions of the zeroing, or can optimize zeroing for small sizes.

Ahhh! I didn't consider that the compiler would be doing analysis of the larger context, and potentially skipping zeroing parts that are set immediately after the call.

Thanks!

-Ravi (rpokala@)

-----Original Message-----
From: "Jonathan T. Looney" <jtl at freebsd.org>
Date: 2018-06-06, Wednesday at 22:58
To: Ravi Pokala <rpokala at freebsd.org>
Cc: Mateusz Guzik <mjguzik at gmail.com>, Mateusz Guzik <mjg at freebsd.org>, src-committers <src-committers at freebsd.org>, <svn-src-all at freebsd.org>, <svn-src-head at freebsd.org>
Subject: Re: svn commit: r334702 - head/sys/sys

> On Wed, Jun 6, 2018 at 10:14 PM, Ravi Pokala <rpokala at freebsd.org> wrote:
>>
>> -----Original Message-----
>> From: <owner-src-committers at freebsd.org> on behalf of Mateusz Guzik <mjguzik at gmail.com>
>> Date: 2018-06-06, Wednesday at 09:01
>> To: Ravi Pokala <rpokala at freebsd.org>
>> Cc: Mateusz Guzik <mjg at freebsd.org>, src-committers <src-committers at freebsd.org>, <svn-src-all at freebsd.org>, <svn-src-head at freebsd.org>
>> Subject: Re: svn commit: r334702 - head/sys/sys
>>
>>> On Wed, Jun 6, 2018 at 1:35 PM, Ravi Pokala <rpokala at freebsd.org> wrote:
>>>
>>>>> + * Passing the flag down requires malloc to blindly zero the entire object.
>>>>> + * In practice a lot of the zeroing can be avoided if most of the object
>>>>> + * gets explicitly initialized after the allocation. Letting the compiler
>>>>> + * zero in place gives it the opportunity to take advantage of this state.
>>>>
>>>> This part, I still don't understand. :-(
>>>
>>> The call to bzero() is still for the full length passed in, so how does this help?
>>>
>>> bzero is:
>>> #define bzero(buf, len) __builtin_memset((buf), 0, (len))
>> 
>> I'm afraid that doesn't answer my question; you're passing the full length to __builtin_memset() too.
> 
> I believe the theory is that the compiler (remember, this is __builtin_memset) can optimize away portions of the zeroing, or can optimize zeroing for small sizes.
> 
> For example, imagine you do this:
> 
>     struct foo {
>         uint32_t a;
>         uint32_t b;
>     };
> 
>     struct foo *
>     alloc_foo(void)
>     {
>         struct foo *rv;
> 
>         rv = malloc(sizeof(*rv), M_TMP, M_WAITOK|M_ZERO);
>         rv->a = 1;
>         rv->b = 2;
>         return (rv);
>     }
> 
> In theory, the compiler can be smart enough to know that the entire structure is initialized, so it is not necessary to zero it.
> 
> (I personally have not tested how well this works in practice. However, this change theoretically lets the compiler be smarter and optimize away unneeded work.)
> 
> At minimum, it should let the compiler replace calls to memset() (and the loops there) with optimal instructions to zero the exact amount of memory that needs to be initialized. (Again, I haven't personally tested how smart the compilers we use are about producing optimal code in this situation.)
> 
> Jonathan