Date: Thu, 13 Jan 2000 22:01:27 -0500 From: "Mikhail Evstiounin" <evstiounin@adelphia.net> To: <keramida@ceid.upatras.gr> Cc: <freebsd-questions@freebsd.org> Subject: Re: Volatile variables Message-ID: <010701bf5e3b$a6dac5e0$a9353018@evstiouninadelphia.net.pit.adelphia.net>
next in thread | raw e-mail | index | archive | help
-----Original Message-----
From: Giorgos Keramidas <charon@hades.hell.gr>
To: Mikhail Evstiounin <evstiounin@adelphia.net>
>
>As I said in a previous posting a few minutes ago, Oliver did not use
>that example of mine for supporting his notion of `volatile'. Not only
I know, I am bad, but, please don't kill me:-( I appologized acouple hours
ago. It's my fault.
>because it was not his example, but because the example was ill
>fortuned since I did not use `cc -O2 -S hello.c' when converting the C
>source to assembler. When I tried this a few seconds ago it gave:
>
> % cat hello.c
> volatile int k;
Could you try to use atomic instead of int - I don't have FreeBSD under
my hand and just want to see a generated code.
> int main (void)
> {
> k = 0;
> k = 1;
> return 0;
> }
> % cc -O2 -S hello.c
> % cat hello.s
> .text
> .p2align 2
> .globl main
> .type main,@function
> main:
> pushl %ebp
> movl %esp,%ebp
> movl $0,k
> movl $1,k
> xorl %eax,%eax
> leave
> ret
> .Lfe1:
> .size main,.Lfe1-main
> .comm k,4,4
>
>which does in fact use what you suggested in a previous posting, the
>instructions that set `k' to 1 are converted to `movl $1,k' which I can
>assume that executes atomically as far as signals are concerned.
>
>It seems that `volatile' is effective in GNU C only when optimizations
>are enabled, because without optimizations the code is, well...,
>unoptimal, and can cause great troubles when used with signals ;)
We should ask about this gcc team. It's possible, that this is a different
understanding of vilotile. Because, I still do not agree with your
usage/understanding - I am still thinking that it should be used, if
value of a vriable could be changed in ways not defined by the language.
And sighandler will change it within a way defined by the language -
assigments. But usage of a global resource from different async
processes requires synchronization and this is a totally different
problem. Take a look at two threads running simultineosly on two
CPU, sharing an access to the same variable - vilotile is not going
to help you. In my mind, sighandler is async process, but it accesses
variable in a way defined by the language. From my point of view,
two different problems were mixed and, as a result, totally wrong
way to solve them were chosen. I was trying to emphasize only this
part, but didn't myself clear enough. Sorry.
>
>--
>Giorgos Keramidas, < keramida @ ceid . upatras . gr >
>"What we have to learn to do, we learn by doing." [Aristotle]
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?010701bf5e3b$a6dac5e0$a9353018>