From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753424AbbIGPmg (ORCPT ); Mon, 7 Sep 2015 11:42:36 -0400 Received: from mail-pa0-f48.google.com ([209.85.220.48]:34172 "EHLO mail-pa0-f48.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750818AbbIGPmd convert rfc822-to-8bit (ORCPT ); Mon, 7 Sep 2015 11:42:33 -0400 Subject: Re: [RFC PATCH 0/3] Implement IRQ stack on ARM64 Mime-Version: 1.0 (Apple Message framework v1283) Content-Type: text/plain; charset=windows-1252 From: Jungseok Lee In-Reply-To: <55EDA040.90208@arm.com> Date: Tue, 8 Sep 2015 00:42:47 +0900 Cc: Catalin Marinas , Will Deacon , "linux-arm-kernel@lists.infradead.org" , "linux-kernel@vger.kernel.org" Content-Transfer-Encoding: 8BIT Message-Id: References: <1441376587-12979-1-git-send-email-jungseoklee85@gmail.com> <55EDA040.90208@arm.com> To: James Morse X-Mailer: Apple Mail (2.1283) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Sep 7, 2015, at 11:33 PM, James Morse wrote: > On 04/09/15 15:23, Jungseok Lee wrote: >> ARM64 kernel allocates 16KB kernel stack when creating a process. In case >> of low memory platforms with tough workloads on userland, this order-2 >> allocation request reaches to memory pressure and performance degradation >> simultaenously since VM page allocator falls into slowpath frequently, >> which triggers page reclaim and compaction. >> >> I believe that one of the best solutions is to reduce kernel stack size. >> According to the following data from stack tracer with some fixes, [1], >> a separate IRQ stack would greatly help to decrease a kernel stack depth. >> > > Hi Jungseok Lee, Hi James Morse, > I was working on a similar patch for irq stack, (patch as a follow up email). > > I suggest we work together on a single implementation. I think the only > major difference is that you're using sp_el0 as a temporary register to > store a copy of the stack-pointer to find struct thread_info, whereas I was > copying it between stacks (ends up as 2x ldp/stps), which keeps the change > restricted to irq_stack setup code. > > We should get some feedback as to which approach is preferred. Great idea! I'd really like to figure out the most ideal implementation of this feature. Best Regards Jungseok Lee