Path: csiph.com!news.swapon.de!eternal-september.org!feeder.eternal-september.org!reader01.eternal-september.org!.POSTED!not-for-mail
From: Keith Thompson <kst-u@mib.org>
Newsgroups: comp.lang.c++
Subject: Re: Performance of unaligned memory-accesses
Date: Thu, 08 Aug 2019 13:33:09 -0700
Organization: None to speak of
Lines: 47
Message-ID: <lntvar8kje.fsf@kst-u.example.com>
References: <qidmlj$62b$1@news.albasani.net> <slrnqklhgn.cdo.grahn+nntp@frailea.sa.invalid> <qieh9b$r7m$1@news.albasani.net> <qiemk4$1o5$1@dont-email.me> <qieolb$79o$1@news.albasani.net> <qieqqb$5cu$1@dont-email.me> <slrnqkmasn.1th.ccollins476ad@pseudoephedrine.nightmare-heaven.no-ip.biz> <qifbmh$edd$1@dont-email.me> <lnd0hga4da.fsf@kst-u.example.com> <qignba$d1a$1@dont-email.me>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Injection-Info: reader02.eternal-september.org; posting-host="c2e3c649afb59b0d1215e9c099fa96fa"; logging-data="17775"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1+NsCoD++AnCTsjMdLxmMCT"
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/23.1 (gnu/linux)
Cancel-Lock: sha1:32AQEMKCtgvBvxAEYVUXl8SCfeo= sha1:Bt8p39Sa2LbUJbE5IxSYHHoubUU=
Xref: csiph.com comp.lang.c++:67465

David Brown <david.brown@hesbynett.no> writes:
> On 08/08/2019 02:27, Keith Thompson wrote:
>> David Brown <david.brown@hesbynett.no> writes:
>> [...]
>>> Compiler extensions like "packed" are fine in my book.  Use them if they 
>>> make the code clearer and more efficient - assuming, of course, that the 
>>> lack of portability is not a problem.  Compilers that don't support such 
>>> features will complain, so you don't get silent problems.
>> 
>> Compiler extension like "packed" can cause problems.  For example,
>> suppose you have something like this (using gcc syntax):
>> 
>>     struct foo {
>>         char c;
>>         int i;
>>     } __attribute__((packed));
>>     foo obj;
>>     some_func(&obj.i);
>> 
>> some_func() takes an argument of type int*, but there's no indication
>> that it's misaligned, so it can't take any special steps to avoid
>> blowing up when it dereferences its argument.
>> 
>> Recent versions of gcc and clang warn about taking the address of a
>> misaligned member of a packed structure.
>> 
>> https://stackoverflow.com/q/8568432/827263
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=51628
>> 
>
> Agreed.
>
> I did not mean to imply that "packed" is something you should use widely
> - IMHO, it is usually a poor choice, and I rarely use it myself.  I
> merely mean that I think it is fine to use compiler-specific features
> like that if they give you the code you need.  They are defined
> behaviour for the compiler you are using, and give hard compile-time
> errors on compilers that don't support them.  The code will not be as
> portable, but you will avoid surprises with silent errors.

The bug report I cited shows a case where "packed" gives *undefined*
behavior by generating an invalid pointer without using pointer casts.

-- 
Keith Thompson (The_Other_Keith) kst-u@mib.org  <http://www.ghoti.net/~kst>
Will write code for food.
void Void(void) { Void(); } /* The recursive call of the void */