October 2007 - Darwin-kernel

Re: Socket protocol which does not use Unsupported.kext
by Gi youl Kim 19 Oct '07

19 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; bh=3+knzzb+in6sLDP0TgAmV43QFu1Ak5PClrz8s2+9duw=; b=p/K4S7qrXLVug7110umlXfX13/+uGSIjotLoCMlYjd/ZHWtcaiAF0GAsws3Qm2/TXudgmmwczo6QQBHDk54/Gg8vRRKjBqAgReF+grxCLaPpO+yJubFofp+IrIhnpg+7hr9kZXU3DVpBD5cS1ocsnXMXCmbIuJAviHOvobgEG9A= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references; b=gDJgI8mkHryEabsYBsJAGSAYJC0uapoGxFQp7X2LXw82l/SxUPtYurm0pG5AaPW25m0aQrR80qsS6T3Pbe/HacJytAna8NHsPlFqD5fCKAtrDdClQSZrrGDEoNQhkW+RHTBgBhuAW4EjjgL1nvq6idl6Wee25x4Dot1mSoPDEjg= Hi Micheal: The drawbacks you posted are the reason why I want to code without unsupported APIs. If you implemented SCTP with socket support, let me know the functions to register sockets which are supported by Apple. By the way, what is kernel sources and headers and the headers provided by Apple? I have xnu-792.18.15 sources and MacOSX10.*.sdk. Do you mean that I should use MacOSX10.*.sdk? FYI My protocol is 'LPX' which has protocol type 0x88ad. LPX is used for our product which is kind of storage over the ethernet. Here is plist which I use <dict> <key>com.apple.kpi.bsd</key> <string>8.0.0</string> <key>com.apple.kpi.iokit</key> <string>8.0.0</string> <key>com.apple.kpi.libkern</key> <string>8.0.0</string> <key>com.apple.kpi.mach</key> <string>8.0.0</string> <key>com.apple.kpi.unsupported</key> <string>8.0.0</string> </dict> On 10/19/07, Michael Tuexen <Michael.Tuexen(a)lurchi.franken.de> wrote: > Hi Aingoppa, > > what protocol do you want to implement? We have implemented SCTP, a > transport protocol. Basically, it is the same code as in the FreeBSD 7.0 > kernel, but with some changes for Mac OS X (locking, timers). > > You need to use > > <dict> > <key>com.apple.kernel</key> > <string>8.10.1</string> > </dict> > > in your Info.plist file and also use some headers from the > kernel sources instead of the headers provided by Apple. > > This all and the usage of net_add_proto() and net_del_proto() > is an unsupported API and has several drawbacks: > 1. Apple can change it any time or make it impossible to > implement transport protocols via NKEs. Fortunately, > we could use the API from 10.3.9 to 10.4.10. Not sure > about 10.5. > 2. You can build a NKE only for one specific Mac OS X version. > Although, you can build a universal binary for the NKE, it > does only load on Intel or PPC, because the kernel has different > versions. > 3. You really depend on Apple, providing the source for the kernel > to get the header files you need (you do not need the source files). > This means you can not delevop/test a Leopard version until Apple > releases the sources for a Leopard kernel. Not sure when this > will happen. Currently not even the 10.4.10 sources are available. > > Best regards > Michael > > On Oct 19, 2007, at 5:24 PM, Gi youl Kim wrote: > > > I want to create a protocol which support socket interface. I could > > create one using net_add_proto() and net_del_proto(). > > But those functions are included by Unsupported.kext. And worse, > > some functions like sbfree() was not even exported. > > So I had to use codes in open xnu source to the protocol work. > > > > Can you create a socket protocol which does not use Unsupported.kext? > > If you can't, how can you create a protocol? > > > > Aingoppa. > > _______________________________________________ > > Do not post admin requests to the list. They will be ignored. > > Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) > > Help/Unsubscribe/Update your Subscription: > > http://lists.apple.com/mailman/options/darwin-kernel/michael.tuexen% > > 40lurchi.franken.de > > > > This email sent to michael.tuexen(a)lurchi.franken.de > > > > _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Socket protocol which does not use Unsupported.kext
by Michael Tuexen 19 Oct '07

19 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Hi Aingoppa, what protocol do you want to implement? We have implemented SCTP, a transport protocol. Basically, it is the same code as in the FreeBSD 7.0 kernel, but with some changes for Mac OS X (locking, timers). You need to use <dict> <key>com.apple.kernel</key> <string>8.10.1</string> </dict> in your Info.plist file and also use some headers from the kernel sources instead of the headers provided by Apple. This all and the usage of net_add_proto() and net_del_proto() is an unsupported API and has several drawbacks: 1. Apple can change it any time or make it impossible to implement transport protocols via NKEs. Fortunately, we could use the API from 10.3.9 to 10.4.10. Not sure about 10.5. 2. You can build a NKE only for one specific Mac OS X version. Although, you can build a universal binary for the NKE, it does only load on Intel or PPC, because the kernel has different versions. 3. You really depend on Apple, providing the source for the kernel to get the header files you need (you do not need the source files). This means you can not delevop/test a Leopard version until Apple releases the sources for a Leopard kernel. Not sure when this will happen. Currently not even the 10.4.10 sources are available. Best regards Michael On Oct 19, 2007, at 5:24 PM, Gi youl Kim wrote: Can you create a socket protocol which does not use Unsupported.kext? If you can't, how can you create a protocol? This email sent to michael.tuexen(a)lurchi.franken.de _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I want to create a protocol which support socket interface. I could create one using net_add_proto() and net_del_proto(). But those functions are included by Unsupported.kext. And worse, some functions like sbfree() was not even exported. So I had to use codes in open xnu source to the protocol work. Aingoppa. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/michael.tuexen% 40lurchi.franken.de This email sent to site_archiver(a)lists.apple.com

1 0

Socket protocol which does not use Unsupported.kext
by Gi youl Kim 19 Oct '07

19 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Dkim-signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=beta; h=domainkey-signature:received:received:mime-version:content-transfer-encoding:message-id:content-type:to:from:subject:date:x-mailer; bh=7WzUdWC2NneplAwSzkjukGm8wzX/ZHYtp6EjUupDlbQ=; b=uL664w1lF42B6w1xKoHyGkQmGyFmSsxnBexBjW5Cfw9wnwmJegVElSl1zwPnhS/4t88aN3C6wYS8eJN3FTLaELHDEis2rqreDLuqd0D8d0n0AoU6ZH84RXJTYOyUw8GDdMNz0M4FNX/2YTFLDZDXt3FhfoILOfBiDuF2EsSsiN4= Domainkey-signature: a=rsa-sha1; c=nofws; d=gmail.com; s=beta; h=received:mime-version:content-transfer-encoding:message-id:content-type:to:from:subject:date:x-mailer; b=b8mO9TjIJPSwBbCXY3+OgyfPnXmO2GLzEI3+Q7C8xDzdgcPqK29R5gtKD6CkzUDsgkxxatkXIMTpJrgrMDFMZxqvf6ffCnTdin41a+2uXSI/29Eg4/uTP+mCAu00F5TCSHXN+KtGEnlBu3LQ5bJJKTLUlAdeijOEnzNbgE9ea44= Can you create a socket protocol which does not use Unsupported.kext? If you can't, how can you create a protocol? Aingoppa. _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… I want to create a protocol which support socket interface. I could create one using net_add_proto() and net_del_proto(). But those functions are included by Unsupported.kext. And worse, some functions like sbfree() was not even exported. So I had to use codes in open xnu source to the protocol work. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kext Load on demand and daemon communication
by Michael Smith 19 Oct '07

19 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com On Oct 18, 2007, at 8:44 AM, Eric Long wrote: I looked at IOServiceAddMatchingNotification(). That seems pretty straight-forward. What I'm not clear on is what stage my driver will be in when I get a notice in my callback. The notification will be posted as a consequence of your driver calling registerService(). Right. Assuming you mean "registerService()", then yes, that's possible. = Mike _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… But does that mean my daemon will receive that notice before the driver exits registerToService()? If my daemon connects to the kext, sends config data, and begins listening for messages from the kext, all from within the notification callback before returning, can I count on my driver's validateNewMedia() method not being called until after I return from the callback, in the normal sequence of events? Without seeing what you've implemented and having more time than I do to analyse it, I can't answer this. You should not, as a general rule, call registerService() until your driver is in a state where it is ready to be called into immediately. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kext Load on demand and daemon communication
by Eric Long 18 Oct '07

18 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcgRncVvBBPPwX2REdy1wQAX8si2KA== Thread-topic: Kext Load on demand and daemon communication User-agent: Microsoft-Entourage/11.3.6.070618 >> I looked at IOServiceAddMatchingNotification(). That seems pretty >> straight-forward. What I'm not clear on is what stage my driver >> will be in >> when I get a notice in my callback. > > The notification will be posted as a consequence of your driver > calling registerService(). > Right. But does that mean my daemon will receive that notice before the driver exits registerToService()? If my daemon connects to the kext, sends config data, and begins listening for messages from the kext, all from within the notification callback before returning, can I count on my driver's validateNewMedia() method not being called until after I return from the callback, in the normal sequence of events? Eric _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kext Load on demand and daemon communication
by Michael Smith 18 Oct '07

18 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com = Mike _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… On Oct 17, 2007, at 11:44 AM, Eric Long wrote: I looked at IOServiceAddMatchingNotification(). That seems pretty straight-forward. What I'm not clear on is what stage my driver will be in when I get a notice in my callback. The notification will be posted as a consequence of your driver calling registerService(). This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kext Load on demand and daemon communication
by Eric Long 17 Oct '07

17 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcgQOfr7OZEnw3wtEdyk5AAX8si2KAAs8Jex Thread-topic: Kext Load on demand and daemon communication User-agent: Microsoft-Entourage/11.3.6.070618 > In the meantime, there are some less-than-ideal workarounds: > > o As Shaun mentioned, there's KUNC. > > o Another option is for your launchd property list to specify some > on-demand criteria that your KEXT satisfies. For example, your > launchd property list could register a UNIX domain socket with a well > known path. When your KEXT comes up, it connects to that socket, > thereby triggering the launch of your daemon. That's good information for the list. In my case, I have a daemon that is always up for other reasons. It just needs to know when my kext is loaded. I looked at IOServiceAddMatchingNotification(). That seems pretty straight-forward. What I'm not clear on is what stage my driver will be in when I get a notice in my callback. What I want my daemon to do is to configure my kext, then listen for messages from it that it must respond to. I expect at the time my daemon is launched there may be driver instances already published before my daemon came up. The kext will be using default behaviors at this point, because it can't read any settings from disk itself. The daemon will configure the kext to handle future events, then start listening to the it for messages. That's as much as I can do for existing instances. In the case of notices I receive in my callback... So long as there is any published driver I want to listen for messages from my kext. I can track that by counting publications vs. terminations. If I receive a first published notice for my driver at a time when there are no other published services, my daemon need to configure the kext via is user client and start listening for messages from it. In this case, however, I don't want my driver's validateNewMedia() method to execute before I have sent configuration info to the kext's user client and I am listening for messages. Is there a way I can count on doing this in terms of the order of operations? I mean, if I set this all up when I receive the publish notice in the daemon, before exiting the callback, am I ok? Is there something else I should do? I'm just trying to figure out how to make this work. Thanks, Eric _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Launchd API documentation
by Quinn 17 Oct '07

17 Oct '07

site_archiver(a)lists.apple.com Delivered-To: Darwin-kernel(a)lists.apple.com <http://developer.apple.com/samplecode/SampleD/index.html> <http://lists.macosforge.org/mailman/listinfo/launchd-dev> S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Relations, Developer Technical Support, Core OS/Hardware _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… At 12:49 -0700 16/10/07, Todd Heberlein wrote: Is there *any* documentation for the launchd APIs? Not yet. APIs at this level are typically documented in man pages, and there is no <x-man-page://3/launch>. This is being tracked as <rdar://problem/3886256>. There is, however, reasonably good sample code showing how to create a launchd-aware daemon (URL below), and there's more to come soon. The reason I ask is that I wrote a simple network service written in C++ that reads in input from cin, writes the information back out cout, and appends the same information in a file. It makes no assumptions about how it is started (i.e., via command line or as a network service). Everything works fine. It sounds like your service will be running in inetd-compatibility mode (that is, you have the "inetdCompatibility" property in your launchd property list), which means you shouldn't have to use the APIs in "launch.h" at all. BUT when I added a getauid() or setauid() to get/set Audit IDs I get very strange results. I need to figure out if I am doing something wrong by *not* calling the launchd APIs, or is there something wrong with the audit APIs. I don't think the API-level documentation you'd get from a "launch.h" man page will help you with subtle problems like this. You might try posting to the launchd-dev mailing list. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kext Load on demand and daemon communication
by Quinn 17 Oct '07

17 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com In the meantime, there are some less-than-ideal workarounds: o As Shaun mentioned, there's KUNC. S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Relations, Developer Technical Support, Core OS/Hardware _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… At 10:53 -0700 16/10/07, Eric Long wrote: What's the secret to making this work? It's not what you're missing, it's what the system is missing (-: Both I/O Kit and launchd are strongly oriented towards launch-on-demand semantics. What's missing, however, is a connection between the two. For example, you could imagine a world where your daemon's launchd property list file contains an I/O Kit matching dictionary, and launchd launches it whenever a matching I/O Kit service exists. Alas, implementing this is tricky, because it would introduce a layering violation between launchd and I/O Kit. We're hoping to address this <rdar://problem/4551362>, but you won't see it for a while (you definitely won't see it on 26 Oct :-). o Another option is for your launchd property list to specify some on-demand criteria that your KEXT satisfies. For example, your launchd property list could register a UNIX domain socket with a well known path. When your KEXT comes up, it connects to that socket, thereby triggering the launch of your daemon. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Kext Load on demand and daemon communication
by Eric Long 16 Oct '07

16 Oct '07

site_archiver(a)lists.apple.com Delivered-To: darwin-kernel(a)lists.apple.com Thread-index: AcgQOfr7OZEnw3wtEdyk5AAX8si2KA== Thread-topic: Kext Load on demand and daemon communication User-agent: Microsoft-Entourage/11.3.6.070618 > listen for service matching notifications Ah. That led me to find this: <http://developer.apple.com/documentation/DeviceDrivers/Conceptual/Accessing Hardware/AH_Finding_Devices/chapter_4_section_2.html#//apple_ref/doc/uid/TP3 0000379-BABEACCJ> Thanks, Eric _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-kernel mailing list (Darwin-kernel(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-kernel/site_archiver%40lists.… This email sent to site_archiver(a)lists.apple.com

1 0