Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:88066
Return-Path: <ajf@ajf.me>
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Delivered-To: mailing list internals@lists.php.net
Received: (qmail 21499 invoked from network); 6 Sep 2015 18:05:34 -0000
Received: from unknown (HELO lists.php.net) (127.0.0.1)
  by localhost with SMTP; 6 Sep 2015 18:05:34 -0000
X-Host-Fingerprint: 2.122.81.212 unknown  
Received: from [2.122.81.212] ([2.122.81.212:13543] helo=localhost.localdomain)
	by pb1.pair.com (ecelerity 2.1.1.9-wez r(12769M)) with ESMTP
	id E1/06-59944-D608CE55 for <internals@lists.php.net>; Sun, 06 Sep 2015 14:05:34 -0400
Message-ID: <E1.06.59944.D608CE55@pb1.pair.com>
To: internals@lists.php.net
References: <02a601d0dbed$2c828df0$8587a9d0$@belski.net>
 <CAL+t6cHdJNNySG0x76a1k_903XrFmfcCDHHH8LbmiTRNoQ3c4Q@mail.gmail.com>
 <02ae01d0dbf5$01c17330$05445990$@belski.net>
 <02b001d0dbf5$607ab7b0$21702710$@belski.net>
 <CAL+t6cEsfH6m6HPbr9KoenSA3N35spucpoE=norTWS9dKZJJNA@mail.gmail.com>
 <02ba01d0dbfb$3de2c390$b9a84ab0$@belski.net>
 <CAL+t6cFXSZ5WYCZ00=uBkiZqVNWcHh9SBu+Kgtnr-v0iP8sRyA@mail.gmail.com>
Date: Sun, 6 Sep 2015 19:05:29 +0100
User-Agent: Mozilla/5.0 (Macintosh; Intel Mac OS X 10.10; rv:38.0)
 Gecko/20100101 Firefox/38.0 SeaMonkey/2.35
MIME-Version: 1.0
In-Reply-To: <CAL+t6cFXSZ5WYCZ00=uBkiZqVNWcHh9SBu+Kgtnr-v0iP8sRyA@mail.gmail.com>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
X-Posted-By: 2.122.81.212
Subject: Re: [PHP-DEV] Overflow checks and integral vars comparison
From: ajf@ajf.me (Andrea Faulds)

Hi Sherif,

Sherif Ramadan wrote:
>
> Of course, but the ASM can also be ported to other architectures and
> wrapped in #ifdef for non x86 and MSVC or other compilers, for example.
> It's not impossible to achieve some sane degree of portability there. I see
> Andrea already worked on some of this in zend_operators.h for example
> https://github.com/php/php-src/blob/ee2e1691080dad2a3110107dd8bd02ee23b41fa0/Zend/zend_operators.h#L437
>
> Of course, making integer overflow checks safe and efficient is by no means
> easy on every given architecture, but surely we could aim to support at
> least the broadest current architectures at first and fall back to C
> overflow checks if necessary.
>

<minor off-topic note>

The email I'm replying to here is more than a month old now.

Just to set the record straight, though, PHP's assembly and pure C 
overflow checks on arithmetic operations aren't my work, they've been 
there for a long time. If you were to `git blame` them you might find 
the original author eventually. The only overflow checks I added to PHP 
were for float-to-int conversions, and that doesn't require inline asm.

One thing I *did* do, however, was rip out that inline assembly in my 
bigint branch (which wasn't merged in the end, alas), since it was 
out-of-date (didn't handle bigints) and I wouldn't dare mess with it. 
Instead, I replaced it it with clang and GCC-compatible intrinsics, 
which require no assembly knowledge to understand:

https://github.com/php/php-src/commit/bdcd9bbd29fac886192fbc410ccb96bd4cb93375

At some point, I might make a pull request to add this to PHP master, if 
it provides a performance boost.

</off-topic>

Thanks.

-- 
Andrea Faulds
http://ajf.me/