From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id ; Tue, 15 Oct 2002 04:16:14 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id ; Tue, 15 Oct 2002 04:16:14 -0400 Received: from [195.39.17.254] ([195.39.17.254]:14340 "EHLO Elf.ucw.cz") by vger.kernel.org with ESMTP id ; Tue, 15 Oct 2002 04:16:13 -0400 Date: Tue, 15 Oct 2002 10:20:50 +0200 From: Pavel Machek To: Ralf Gerbig Cc: "Grover, Andrew" , linux-kernel@vger.kernel.org Subject: Re: 2.5.42 ACPI/Suspend with an Acer Travelmate 350 Message-ID: <20021015102049.C277@elf.ucw.cz> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.3.23i X-Warning: Reading this can be dangerous to your mental health. Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Hi! > Hi Andrew, Pavel, lkml > > I just poked around with /proc/acpi/sleep without much hope as I was > never able to get this otherwise nice lapdog to sleep _and_ resume > before. The swsusp patches on 2.4.1[89] worked exept the touchpad > would not respond after the resume. > > But wow, this time at least the kernel survived! > > Most of the following is hand copied, beware the typos > > echo 1 > /proc/acpi/sleep > =pdflush: bogus wakeup! > ==| We should not need to stop processes (etc) on S1, etc. > Same here: > > echo 5 > /proc/acpi/sleep > > Suspending devices > Shutting down devices Oopses on shutdown... We should not suspend devices here, either... Pavel -- When do you have heart between your knees?