From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Thu, 15 Aug 2002 02:49:19 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Thu, 15 Aug 2002 02:49:19 -0400 Received: from dp.samba.org ([66.70.73.150]:61057 "EHLO lists.samba.org") by vger.kernel.org with ESMTP id ; Thu, 15 Aug 2002 02:49:19 -0400 From: Rusty Russell To: linux-kernel@vger.kernel.org Cc: davidm@hpl.hp.com, torvalds@transmeta.com, ak@suse.de Subject: Getting rid of irq_stat_t Date: Thu, 15 Aug 2002 16:52:52 +1000 Message-Id: <20020815015340.52C2B2C24B@lists.samba.org> Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi all, Doing per-cpu cleanups: everyone except ia64 and x86_64 use the generic irq_stat_t macros. There are two obvious ways to go: one is to simply use per-cpu macros directly (nice and clean), the other is to rework the softirq_pending() etc macros (adding local_softirq_pending()) to use per-cpu everywhere except your two platforms, which means callers need to disable preemption explicitly (get_* and put* versions of the macros seems overkill). The motivation is that accessing irq_stat_t is one of the main reasons for smp_processor_id() etc. Thoughts? Rusty. -- Anyone who quotes me in their sig is an idiot. -- Rusty Russell.