January 2008 - Darwin-dev - lists.apple.com

Re: Cancel Sleep
by Matt Burnett 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=gamma; h=domainkey-signature:received:received:cc:message-id:from:to:in-reply-to:content-type:content-transfer-encoding:mime-version:subject:date:references:x-mailer; bh=edinNko9cVAjsIAP7ns5k1EXRi3bncQNl3CsJdSgPLw=; b=YsspzqJZoK7Yge9nZSryZHR9Wh+teMzVWNUcapPMmxKv2hrkn1ZJdB2PEIYX6052jOhDeX1Q4cJpl4+VDC7fvEJSAlIftWTz8p3aws0jy4ZEvaMdts7PXhyyvo4A6PuBxs0n+oxGHrpt+igN9cvjWFNw4l71wGTxW9YRCAOI88E= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=cc:message-id:from:to:in-reply-to:content-type:content-transfer-encoding:mime-version:subject:date:references:x-mailer; b=LewYweFJlSCVxyzRWAoVBm0uc5/W714adMuumAU1Njl+9XYegqhpX4rwNxrFils8P0PBGrv8PMJmMqVa+i2q4poGK2+pxNoM/DvD7Ler8OZG8qG9Y9vHoKUHwDoblXfTaVHs1LysbufwJVpomEe9Bjecre3pbf+tFR/F1y8F2AY= On Jan 2, 2008, at 9:50 AM, Robert T Wyatt wrote: _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Thats a moot point. If a thief has physical access to a machine, he wont care if its logged in or not. If he wants the data, he can boot off a CD, if they want a machine, then well, they all ready have it. It says the commands are issued by those drivers only. Have by chance tried issuing one yourself, such as kIOPHDisableClamshell. Maybe if you could issue that one you wouldn't need to get notifications. --David Alger Too bad that this may be a mechanism for a thief to use to leave one's machine running and logged in while they move it to another location. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/matt.w.burnett%40gmail.com This email sent to matt.w.burnett(a)gmail.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Cancel Sleep
by Robert T Wyatt 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Organization: The University of Texas at Austin User-agent: Mozilla/5.0 (Macintosh; U; Intel Mac OS X; en-US; rv:1.8.1.11) Gecko/20071128 SeaMonkey/1.1.7 It says the commands are issued by those drivers only. Have by chance tried issuing one yourself, such as kIOPHDisableClamshell. Maybe if you could issue that one you wouldn't need to get notifications. --David Alger Too bad that this may be a mechanism for a thief to use to leave one's machine running and logged in while they move it to another location. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Cancel Sleep
by Amanda Walker 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com --Amanda _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… On Jan 1, 2008, at 11:29 PM, Perry Smith wrote: On Jan 1, 2008, at 10:14 PM, Amanda Walker wrote: Or just don't want to do someone else's homework for them ;-). Ahhh... I see. Being an old fart, I never think about stuff like that. Well, I was using "homework" figuratively. Being an old fart myself, I'd rather help someone research the problem and solve it themselves than solve it for them (for free, anyway). So given the question "how do I intercept low level notifications and/or change power management behavior", suggesting that darwin-dev isn't the best forum, and pointing the original poster at better ones seemed pretty reasonable to me. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Cancel Sleep
by David Elliott 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jan 1, 2008, at 11:15 PM, Jordan K. Hubbard wrote: On Jan 1, 2008, at 7:00 PM, Michael Smith wrote: Convinced? -Dave _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This has wandered quite a bit off the original topic at this point, and become another one of those "I don't care what anyone else thinks, I reckon it should do <blah>" threads. It's not helpful to the OP, who wants something that's not really doable, and it's not helpful to the rest of you because nobody here is in a position to do anything about your opinions. And, I might add, even those who might be in a position to authorize/ effect change in reaction to these various requests to do <blah> can only be far from convinced by the arguments presented thus far. Apple engineering is hardly deaf to feature requests (we receive, appreciate, and act on them all the time) but they have to be argued from some "greater good" perspective to have any chance of success. Personal preference arguments usually don't make it unless it can be demonstrated that there is some large number of persons asking for the same thing (and "asking loud" will not be confused with a strong argument, though folks are always likely to try that approach first). When a laptop has power and an external display connected it should be presumed that the user does not want it to go to sleep when he closes the lid. If you don't believe that and want to continue the Apple dogma then this might work for you: if the machine is capable of safely running closed (e.g. is a PB or MBP) and the user is holding Option while he closes the lid then the laptop should not go to sleep. Since it's impossible to hold option on the internal keyboard while closing the lid, an external keyboard must have been used. That one simple feature will cover 90% of cases people want and will dissuade them from writing their own kernel extension hacks that do really bad things like make the machine ignore lid close entirely. There is almost no chance that a clueless user could invoke this functionality accidently. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Cancel Sleep
by Perry Smith 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Jan 1, 2008, at 10:14 PM, Amanda Walker wrote: Ahhh... I see. Being an old fart, I never think about stuff like that. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Or just don't want to do someone else's homework for them ;-). This email sent to site_archiver(a)lists.apple.com

1 0

Re: Cancel Sleep
by David Elliott 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi Andrew, On Jan 1, 2008, at 10:38 PM, Andrew James wrote: -Dave _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Ok everyone calm down, yes this may not be something everyone will agree on, the people that use this KNOW the risks! [...] How does one get a notification of the lid state changing (When the lid closes, nothing to do with sleep) that does not also mix up with the power state changing, you will notice i dont mention a single thing about forcing the system to stay awake or anything here, so please can we drop the debate and stop flooding the mailing list with the debate that will never end and just get to the point. which just as safe measure ill say once again... How does one get a notification of the lid state changing (When the lid closes, nothing to do with sleep) that does not also mix up with the power state changing This is just a preliminary guess but I think your best bet is to make an IOKit driver that matches on the lid hardware nub. Just look at the Info.plist for AppleACPIButtons.kext, specifically the entry for the AppleACPILid class. I think it should be possible to match on that nub with a non-default match category so that AppleACPILid still attaches. Do whatever it does (disassemble it to find out if you have to) to receive the ACPI events. >From there I assume you already know how you can prevent the system from going to sleep due to AppleACPILid signaling it. By that point you'll have received the lid close event yourself and you can decide whether or not to abort sleep based on your own conditions. Of course, I suppose you could also just write a driver for the ACPI device that matched without using a match category and a higher probe score. That will prevent AppleACPILid from matching at all in which case I would assume closing and opening the lid won't even be noticed by the OS. If you just need a quick hack, that should do it for you. There's basically nothing Apple can do short of special-casing matching for certain hardware devices to keep you from taking over the default match category on the provider nub. If you finish this and come up with a kext that fixes my problem I'll be very grateful. To recap, I _want_ the machine to sleep on lid close like it normally does except when AC power is applied in which case I want it to stay on. This behavior makes the most sense to me and probably to everyone else who isn't drinking the Kool-Aid at the Apple cafeteria. I'd write the damn thing myself but 1) I really think Apple should already do it this way, and 2) I'm not annoyed enough (yet) to spend the time to do it. Another interesting hack would be to check if say "Option" is being held down when the lid is closed. In that case the user obviously has an external keyboard (it's physically impossible to accomplish this feat with an internal one) and the user most definitely wants the machine to stay awake. Best of both worlds. Of course, come to think of it, that's probably possible already. You get the sleep message (whether it was caused by lid close or not is irrelevant) and check the keyboard status. If set, you cancel the sleep. Hmm... This email sent to site_archiver(a)lists.apple.com

1 0

Re: Cancel Sleep
by Jordan K. Hubbard 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Cancel Sleep
by Amanda Walker 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Or just don't want to do someone else's homework for them ;-). --Amanda _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… On Jan 1, 2008, at 10:53 PM, Perry Smith wrote: Sorry for not knowing more. How much of the code is open? Not all of it. Last I checked, the power management kexts are not open source, but there are open source equivalents in XPostFacto (which I just pointed Andrew towards). Judging from the quality of the code snippet you provided along with the dribble "I'm right -- You're wrong" I've watched over the past two days, these guys don't know the answer to your question. Luke: Use the SOURCE... Not all of the drivers are open source. Be prepared to read assembly code in gdb if you take the "step through the code" approach. I recommend 2-machine debugging over firewire. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Cancel Sleep
by Amanda Walker 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com --Amanda _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… On Jan 1, 2008, at 10:38 PM, Andrew James wrote: But seriously can we get back to the point!!!! Some of us have been. I've pointed you at more appropriate mailing lists (darwin-kernel and darwin-drivers). I've outlined an overall strategy for interposing or overriding any kernel behavior that you don't like. I know these messages are getting through to the mailing list. How does one get a notification of the lid state changing (When the lid closes, nothing to do with sleep) that does not also mix up with the power state changing, One doesn't, without writing a kext to implement new kernel-level behavior, for the simple reason that no such notification currently exists. As i've suggested twice now, write a kext that Interposes on (or even replaces) the existing kernel code (at a first guess, the PMU driver). You might not even have to patch running code--you might be able to get by just registering as the recipient for PMU interrupts (since the PMU is where the clamshell sensor is wired to), and queueing up a user-level notification yourself before passing along the interrupt to the system driver. But in any case, you'll have to dig into it at the IOKit driver level, since the system PMU driver is what sends the kIOPMSleepNow notification to the root domain in response to a "clamshell closed" interrupt. It also handles the power button, the AC power state change, etc. There is no notification of the clamshell state that is not "mixed up with the power state changing" for you to register for, so if you need one, you'll have to synthesize one yourself by watching the physical interrupt coming in from the hardware. If you want to see how all this works in excruciating detail, see the OpenPMU driver (part of XPostFacto, which is open source). In particular, take a look at OpenPMUPwrController.cpp. We can't get any closer to the "point" without actually writing the code for you. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Cancel Sleep
by David Alger 02 Jan '08

02 Jan '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0