From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com ([66.187.233.31]:51588 "EHLO mx1.redhat.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1760725AbWLHN5t (ORCPT ); Fri, 8 Dec 2006 08:57:49 -0500 From: David Howells In-Reply-To: <20061208111410.GA31068@flint.arm.linux.org.uk> References: <20061208111410.GA31068@flint.arm.linux.org.uk> <20061207234250.GH1255@flint.arm.linux.org.uk> <20061207085409.228016a2.akpm@osdl.org> <20061207153138.28408.94099.stgit@warthog.cambridge.redhat.com> <20061207153143.28408.7274.stgit@warthog.cambridge.redhat.com> <639.1165521999@redhat.com> <26012.1165535903@redhat.com> Subject: Re: [PATCH 3/3] WorkStruct: Use direct assignment rather than cmpxchg() Date: Fri, 08 Dec 2006 13:57:09 +0000 Message-ID: <4501.1165586229@redhat.com> Sender: linux-arch-owner@vger.kernel.org To: Russell King Cc: David Howells , Andrew Morton , torvalds@osdl.org, davem@davemloft.com, wli@holomorphy.com, matthew@wil.cx, linux-kernel@vger.kernel.org, linux-arch@vger.kernel.org List-ID: Russell King wrote: > These are the constant versions, where the compiler can optimise the > mask and word offset itself. So my inclusion of ARM is correct... Under some circumstances it will write to the target word when it wouldn't actually make a change: static inline int ____atomic_test_and_set_bit(unsigned int bit, volatile unsigned long *p) { unsigned long flags; unsigned int res; unsigned long mask = 1UL << (bit & 31); p += bit >> 5; raw_local_irq_save(flags); res = *p; *p = res | mask; raw_local_irq_restore(flags); return res & mask; } Remember: *p is volatile. David