From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mx1.redhat.com ([66.187.233.31]:52357 "EHLO mx1.redhat.com") by vger.kernel.org with ESMTP id S936250AbWLDMX7 (ORCPT ); Mon, 4 Dec 2006 07:23:59 -0500 From: David Howells In-Reply-To: <20061204114851.GA25859@elte.hu> References: <20061204114851.GA25859@elte.hu> <20061201172149.GC3078@ftp.linux.org.uk> <1165064370.24604.36.camel@localhost.localdomain> <20061202140521.GJ3078@ftp.linux.org.uk> <1165070713.24604.50.camel@localhost.localdomain> <20061202160252.GQ14076@parisc-linux.org> <1165082803.24604.54.camel@localhost.localdomain> <20061202181957.GK3078@ftp.linux.org.uk> Subject: Re: [RFC] timers, pointers to functions and type safety Date: Mon, 04 Dec 2006 12:22:44 +0000 Message-ID: <28665.1165234964@redhat.com> Sender: linux-arch-owner@vger.kernel.org To: Ingo Molnar Cc: Al Viro , Thomas Gleixner , Matthew Wilcox , Linus Torvalds , linux-arch@vger.kernel.org, linux-kernel@vger.kernel.org List-ID: Ingo Molnar wrote: > the question is: which is more important, the type safety of a > container_of() [or type cast], which if we get it wrong produces a > /very/ trivial crash that is trivial to fix - or embedded timers data > structure size all around the kernel? I believe the latter is more > important. Indeed yes. Using container_of() and ditching the data value, you generally have to have one extra instruction per timer handler, if that, but you are able to discard one instruction or more from __run_timers() and struct timer_list discards a word. You will almost certainly have far more timer_list structs in the kernel than timer handler functions, therefore it's a space win, and possibly also a time win (if the reduction of __run_timers() is greater than the increase in the timer handler). And that extra instruction in the timer handler is usually going to be an addition or subtraction of a small immediate value - which may be zero (in which case the insn is dropped) or which may be folded directly into memory access instruction offsets. David