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

Re: IOObjectRelease: passing null? threadsafe?
by Terry Lambert 22 Oct '08

22 Oct '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 10/20/08 8:17 PM, Terry Lambert said: However, it would be better if you didn't depend on there being a safety net there, and only pass in valid object references for which you've previously done a retain. Terry, if (device->X) { IOObjectRelease(device->X); device->X = 0; } -- Terry _______________________________________________ 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 Oct 21, 2008, at 7:56 AM, Sean McBride wrote: 1) How does IOObjectRelease() react to being given null? Does it do nothing like free() or crash like CFRelease()? Yes, I could try, but I'm looking for an API guarantee one way or the other. 2) Is IOObjectRelease() threadsafe? Specifically, if I create some object on the main thread, is it safe to IOObjectRelease() it from an Objective-C garbage collector finalize method (which runs on its own thread)? If you pass it a NULL, that is the same as MACH_PORT_NULL, which makes MACH_PORT_VALID() false, which makes mach_port_deallocate() simply return KERN_SUCCESS, which causes the function to return that. Thanks for this info. Any idea about calling IOObjectRelease() from finalize? I'm not really an IOKit expert, it's just that your question was hanging out there for a while. 8-). We appear to do it in a bunch of places in code we've published, although in the ones I looked at, most of them protected against calling with NULL, e.g.: ...not to happy with the use of 0 instead of MACH_PORT_NULL, but that intentionally plays loose tying games anyway, so it's just 0, so I can't complain too much. I think the reason it does this is because it's possible to tear down the devices from either user space or from kernel space, based on having received a hot-plug event. If these race, then you want to make sure that you've marked the device invalid, so it looks like they use the 0 for that. I also see it happen in the delete routine or in the connection notification event, rather than in the finalize in a number of places, which might avoid the issue of needing to check for the 0 as a marker by delaying things until later, at the cost of holding resources for a bit longer. This email sent to site_archiver(a)lists.apple.com

1 0

Re: IOKit kext Dependencies
by Duane Murphy 21 Oct '08

21 Oct '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256 - --- At Tue, 21 Oct 2008 11:07:23 -0700, Garth Cummings wrote: >Hi Duane, > >From: Duane Murphy <duanemurphy(a)mac.com> >Date: Mon, 20 Oct 2008 15:56:17 -0700 > >Is there a way of expressing dependencies on classes other than the >classes that a kext is a subclass of? > >The kext dependency and loading machinery only knows about kexts; it >knows nothing about individual classes. That's handled by the kernel linker. > >If both classes are in the same kext, they will be available at the same >time. If not, you can use the OSBundleLibraries mechanism to get the >dependency to load first. See the sample code IOKitWithLibrary <http:// >developer.apple.com/sample.code/IOKitWithLibrary> for an example. > >If you actually need to wait for an instance of the class to be created, >waitForService should do the trick. Thanks Garth, It sounds like that will work. I have two drivers in the same kext. One driver is a manager, the other is more like a driver. The driver needs to connect to the manager. The manager connects to IOResources. I was hoping there was a way of expressing that the manager must be in place for the driver to work. It sounds like waitForService will do part of it, and possibly a notification if waitForService times out. The driver can continue some operations if the manager is not available. Thanks, ...Duane -----BEGIN PGP SIGNATURE----- Version: PGP Desktop 9.9.0 (Build 397) Charset: ISO-8859-1 wsBVAwUBSP5GFUrg9acQ4r2CAQgUtQf+P/sRHDRgKiXblrLYxNWxuH5XZYhXOYjm JBI2iVXLA3LwIbw/gIx+g2QLwDwAPVcPFcPlEbEXFwifNAdTA8/Z1Cw0xpsCLxcg 0T1VEBwfKReR0Gp5jKyHJZI3yPXAzBMETkhdUmcNKKrF7oUG0OsgJ2+piozKPlYK Wi4aDiQxpyFaKFjmeqhuPmGBnpVV/p3szvp4zDftz4ujszDAJnXyopLFYDI3NHka icOEdnZbtjD+Fkqpki0KShTkRCi2fHZo0sAsRbVlnFM9VsCFi6R+dei0QxHTGJas 0GKM+GT0//2Flx5Hz0rXRf5j+mcXm7XWoZYy/IWtmoIYl/KybfNQ1A== =hSkB -----END PGP SIGNATURE----- _______________________________________________ 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: IOKit kext Dependencies
by Garth Cummings 21 Oct '08

21 Oct '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: IOKit Kext Dependencies
by Joe van Tunen 21 Oct '08

21 Oct '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: AckzpdAODm7MbJ+ZEd2mGgAwZbwyQg== Thread-topic: IOKit Kext Dependencies User-agent: Microsoft-Entourage/11.4.0.080122 I think waitForService will do what you need. It will wait for the matching IOService to call registerService and returns a pointer to that IOService. google: waitforservice site:developer.apple.com Or search all of Apple (including mailing lists and darwin source for examples and discussion) google: waitforservice site:apple.com > From: Duane Murphy <duanemurphy(a)mac.com> > Date: Mon, 20 Oct 2008 15:56:17 -0700 > > Is there a way of expressing dependencies on classes other than the > classes that a kext is a subclass of? > > For example, I'm developing a kext that obviously has dependencies on > it's base class, but it also needs to find another class in the > registry. It would be best if an instance of the dependent class was > already loaded when the second class loads. > > Is it possible to express such a dependency in the Info.plist? > Info.plist has OSBundleLibraries but that seems like it is too high. > > Our implementation has two classes that coordinate with each other. I > would like to have one load before the other. Is that possible? Both > classes are in the same kext (and the same library). > > Thank you for your ideas, > > > ...Duane _______________________________________________ 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: IOObjectRelease: passing null? threadsafe?
by Sean McBride 21 Oct '08

21 Oct '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Organization: Rogue Research On 10/20/08 8:17 PM, Terry Lambert said: >> 1) How does IOObjectRelease() react to being given null? Does it do >> nothing like free() or crash like CFRelease()? Yes, I could try, but >> I'm looking for an API guarantee one way or the other. >> >> 2) Is IOObjectRelease() threadsafe? Specifically, if I create some >> object on the main thread, is it safe to IOObjectRelease() it from an >> Objective-C garbage collector finalize method (which runs on its own >> thread)? > >If you pass it a NULL, that is the same as MACH_PORT_NULL, which makes >MACH_PORT_VALID() false, which makes mach_port_deallocate() simply >return KERN_SUCCESS, which causes the function to return that. > >However, it would be better if you didn't depend on there being a >safety net there, and only pass in valid object references for which >you've previously done a retain. Terry, Thanks for this info. Any idea about calling IOObjectRelease() from finalize? Cheers, -- ____________________________________________________________ Sean McBride, B. Eng sean(a)rogue-research.com Rogue Research www.rogue-research.com Mac Software Developer Montréal, Québec, Canada _______________________________________________ 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: lanuchd questions
by Jason Coco 21 Oct '08

21 Oct '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:mime-version:subject:date:references :x-mailer; bh=KAks1CJc9O2TIZeiEtX7E5rBCC9YYDCd5nLGumvLt6s=; b=I+2Z7g2au2NNJWDL6MYUVOT0Zo6u0c1DTRFjbmwLBNU+57F/7/y6XAk8ZkT863SQDB kB47riCnjbjHRQUjAqnIyK8Oz5TT12q8LyM0Cf9/bOGLFEv4WJ7um6jlrPdC69fzTYuI a9xEj7m/WLPhx9wh19/QD9GidpeFB9w4hz2YU= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=gamma; h=cc:message-id:from:to:in-reply-to:content-type:mime-version:subject :date:references:x-mailer; b=pMNLnuXVP5NrlwG+TWVi19F1p8tbz/1mbtkbCsRsf2fQV6aulm7FmjPxCCaZKBoXaR WMCUOblo4V0OoN2sCKGagBHX72b6/5G+b/AVWEbdGmp4Khg5NrehqMywnmnLFh0Pz7Js R13CpCcDr/jCjh3/4YU6/EgbK2MCdlk57irC0= On Oct 21, 2008, at 07:24 , Karan, Cem (Civ, ARL/CISD) wrote: On Oct 20, 2008, at 16:34 , Karan, Cem (Civ, ARL/CISD) wrote: The moment we have dependencies, then we have to worry about cycles (deadlock), deciding what to do if a process that we're cheesy method to handle the problem. Write a script that launchd launches that runs your first program. When that program script submits the other programs to launchd via 'launchctl submit'. Since we haven't put anything into /Library/LaunchDaemons launchd places, we get the same effect as using /var/run, so everything is always launched in the correct order each time we reboot. That's kinda what I did... I ended up writing a daemon which was connected to launchd, starts up and does all the initialization stuff. It then actually manages some of the other processes and they all communicate over a unix socket. _______________________________________________ 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 On Monday, October 20, 2008 4:43 PM Jason Coco wrote: Truthfully, I'd rather petition Apple to make it a part of launchd. dependent on goes down, etc. And, now that I'm thinking about it, there is another returns, the or the other Do you mean you call daemon() or fork() with exit()? No, it's a proper launchd daemon. It communicates via its unix domain socket and is pretty much on-demand for that. However, it manages 2 other daemons as well so if they're set to launch-on-start it starts and launches and monitors them. smime.p7s

1 0

RE: lanuchd questions
by Karan, Cem (Civ, ARL/CISD) 21 Oct '08

21 Oct '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Thread-index: Acky9H2LkrDbrJFsR063kJ3aTRWW8QAeLdOw Thread-topic: lanuchd questions On Monday, October 20, 2008 4:43 PM Jason Coco wrote: > On Oct 20, 2008, at 16:34 , Karan, Cem (Civ, ARL/CISD) wrote: > > > Truthfully, I'd rather petition Apple to make it a part of > launchd. > > The > > moment we have dependencies, then we have to worry about cycles > > (deadlock), deciding what to do if a process that we're > dependent on > > goes down, etc. And, now that I'm thinking about it, there > is another > > cheesy method to handle the problem. Write a script that launchd > > launches that runs your first program. When that program > returns, the > > script submits the other programs to launchd via 'launchctl submit'. > > Since we haven't put anything into /Library/LaunchDaemons > or the other > > launchd places, we get the same effect as using /var/run, so > > everything is always launched in the correct order each time we > > reboot. > > That's kinda what I did... I ended up writing a daemon which > was connected to launchd, starts up and does all the > initialization stuff. It then actually manages some of the > other processes and they all communicate over a unix socket. Do you mean you call daemon() or fork() with exit()? Thanks, Cem Karan _______________________________________________ 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: IOObjectRelease: passing null? threadsafe?
by Terry Lambert 21 Oct '08

21 Oct '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi all, The docs for IOObjectRelease() are terse. I'm wondering: 1) How does IOObjectRelease() react to being given null? Does it do nothing like free() or crash like CFRelease()? Yes, I could try, but I'm looking for an API guarantee one way or the other. -- Terry _______________________________________________ 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 Oct 20, 2008, at 3:41 PM, Sean McBride wrote: 2) Is IOObjectRelease() threadsafe? Specifically, if I create some object on the main thread, is it safe to IOObjectRelease() it from an Objective-C garbage collector finalize method (which runs on its own thread)? If you pass it a NULL, that is the same as MACH_PORT_NULL, which makes MACH_PORT_VALID() false, which makes mach_port_deallocate() simply return KERN_SUCCESS, which causes the function to return that. However, it would be better if you didn't depend on there being a safety net there, and only pass in valid object references for which you've previously done a retain. This email sent to site_archiver(a)lists.apple.com

1 0

IOKit Kext Dependencies
by Duane Murphy 20 Oct '08

20 Oct '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Is there a way of expressing dependencies on classes other than the classes that a kext is a subclass of? For example, I'm developing a kext that obviously has dependencies on it's base class, but it also needs to find another class in the registry. It would be best if an instance of the dependent class was already loaded when the second class loads. Is it possible to express such a dependency in the Info.plist? Info.plist has OSBundleLibraries but that seems like it is too high. Our implementation has two classes that coordinate with each other. I would like to have one load before the other. Is that possible? Both classes are in the same kext (and the same library). Thank you for your ideas, ...Duane _______________________________________________ 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

IOObjectRelease: passing null? threadsafe?
by Sean McBride 20 Oct '08

20 Oct '08

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Organization: Rogue Research Hi all, The docs for IOObjectRelease() are terse. I'm wondering: 1) How does IOObjectRelease() react to being given null? Does it do nothing like free() or crash like CFRelease()? Yes, I could try, but I'm looking for an API guarantee one way or the other. 2) Is IOObjectRelease() threadsafe? Specifically, if I create some object on the main thread, is it safe to IOObjectRelease() it from an Objective-C garbage collector finalize method (which runs on its own thread)? Thanks, -- ____________________________________________________________ Sean McBride, B. Eng sean(a)rogue-research.com Rogue Research www.rogue-research.com Mac Software Developer Montréal, Québec, Canada _______________________________________________ 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