Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1764955AbYBUABX (ORCPT ); Wed, 20 Feb 2008 19:01:23 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1758460AbYBUABC (ORCPT ); Wed, 20 Feb 2008 19:01:02 -0500 Received: from smtp2.linux-foundation.org ([207.189.120.14]:34966 "EHLO smtp2.linux-foundation.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1758388AbYBUABA (ORCPT ); Wed, 20 Feb 2008 19:01:00 -0500 Date: Wed, 20 Feb 2008 16:00:34 -0800 (PST) From: Linus Torvalds To: "Rafael J. Wysocki" cc: Jesse Barnes , Jeff Chua , lkml , Dave Airlie , linux-acpi@vger.kernel.org, suspend-devel List , Greg KH , Alexey Starikovskiy Subject: Re: 2.6.25-rc2 System no longer powers off after suspend-to-disk. Screen becomes green. In-Reply-To: <200802210035.14058.rjw@sisk.pl> Message-ID: References: <200802202336.45492.rjw@sisk.pl> <200802210035.14058.rjw@sisk.pl> User-Agent: Alpine 1.00 (LFD 882 2007-12-20) 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 Content-Length: 1900 Lines: 42 On Thu, 21 Feb 2008, Rafael J. Wysocki wrote: > > In fact we have acpi_pci_choose_state() that tells the driver which power > state to put the device into in ->suspend(). If that is used, the device ends > up in the state expected by to BIOS for S4. First off, nobody should *ever* use that directly anyway. Secondly, the one that people should use ("pci_choose_state()") doesn't actually do what you claim it does. It does all kinds of wrong things, and doesn't even take the target state into account at all. So look again. > No. Again, if there are devices that wake us up from S4, but not from S5, > they need to be handled differently in the *enter S4* case (hibernation) and > in the *enter S5* case (powering off the system). And again, what does this have to do with (the example I used) the graphics hardware? Answer: nothing. The example I gave you we simply DO THE WRONG THING FOR. Same thing for things like USB devices - where pci_choose_state() doesn't work to begin with. Why do we call "suspend()" on such a thing when we don't want to suspend it? We shouldn't. We should call "freeze/unfreeze" (which are no-ops) and then finally perhaps "poweroff", and that final stage might want to spin things down or similar. But *none* of it has anything to do with suspend, and none of it has anything to do with pci_choose_state() (much less acpi_pci_choose_state) The fact is, we should let the driver decide, and we should make it clear to the driver writer what he is deciding about - rather than basically lie and say "suspend the device and put it into D3" even when that's the last thing it should ever do. Linus -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/