To: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: William Lee Irwin III <wli@holomorphy.com>,
       Pete Zaitcev <zaitcev@redhat.com>, linux-kernel@vger.kernel.org
Subject: Re: s390(64) per_cpu in modules (ipv6)
References: <20040630130442.GA2440@mschwid3.boeblingen.de.ibm.com>
From: Roland Dreier <roland@topspin.com>
Date: Wed, 30 Jun 2004 09:07:08 -0700
In-Reply-To: <20040630130442.GA2440@mschwid3.boeblingen.de.ibm.com> (Martin
 Schwidefsky's message of "Wed, 30 Jun 2004 15:04:42 +0200")
Message-ID: <52oen1jahf.fsf@topspin.com>
User-Agent: Gnus/5.1006 (Gnus v5.10.6) XEmacs/21.4 (Security Through
 Obscurity, linux)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: linux-kernel-owner@vger.kernel.org
Content-Length: 1082
Lines: 22

    Martin> __attribute_used__ isn't really what we want. If a
    Martin> statically defined per cpu variable isn't used in the C
    Martin> file then gcc should be allowed to remove it. It's not
    Martin> used after all.  What we need is a way to tell the
    Martin> compiler that an inline assembly uses a variable without
    Martin> passing any kind of address of the variable to it.

Actually my understanding is that __attribute_used__ is intended for
exactly that: to let the compiler know that something that is
apparently unused by the C code is actually used by inline assembly.
I'm sure your solution is fine as well but I think this type of
situation is exactly what __attribute_used__ is for.  For example it
is attached to the modversions ____versions array so that it is not
discarded by the compiler.

Best,
  Roland
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/