From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752782Ab0CYQWR (ORCPT ); Thu, 25 Mar 2010 12:22:17 -0400 Received: from smtp1.linux-foundation.org ([140.211.169.13]:35371 "EHLO smtp1.linux-foundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752588Ab0CYQWP (ORCPT ); Thu, 25 Mar 2010 12:22:15 -0400 Date: Thu, 25 Mar 2010 09:17:13 -0700 (PDT) From: Linus Torvalds To: Thomas Gleixner cc: Andi Kleen , x86@kernel.org, LKML , jesse.brandeburg@intel.com Subject: Re: [PATCH] Prevent nested interrupts when the IRQ stack is near overflowing v2 In-Reply-To: Message-ID: References: <20100324190150.GA18803@basil.fritz.box> <20100325003652.GG20695@one.firstfloor.org> <20100325093744.GH20695@one.firstfloor.org> User-Agent: Alpine 2.00 (LFD 1167 2008-08-23) MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Thu, 25 Mar 2010, Linus Torvalds wrote: > > NOTE! Historically, the "fast" handlers also had a much faster irq > response because they didn't do that whole MASK/ACK/END thing. So they'd > just keep the CPU interrupts disabled, and ACK at the end, and I think > we've even used AUTOEIO so that they didn't need any ACK at all, and we > never touched the interrupt controller itself for them. Btw, it was even more extreme than that. The fast irq handlers got a totally separate kernel entry point, and wouldn't save all registers, only the compiler-clobbered ones. Which is why they then had no "struct pt_regs" etc. And yes, it really mattered. Then later we got so bloated that it wasn't much of an issue - and just made everything more complicated. Linus