August 2003 - Darwin-kernel

Re: Kernel Extension Activation without Reboot?
by Fernando Urbina 25 Aug '03

25 Aug '03

On 8/25/03 2:34 PM, "Lance Drake" <lance(a)drake.org> wrote: > The object that should have over-ridden is sitting in the IOreg listing > there where my KEXT should be loaded. If I reboot, myKext is in place > and works fine. If the object you are matching to already has a kext loaded to it, your loading of a new kext, even with a higher matching criteria, will not load against it. On reboot, your kext will match to it with a higher matching critera and hence will load against that object. That's my guess on what's going on. - Nano _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Kernel Extension Activation without Reboot?
by Lance Drake 25 Aug '03

25 Aug '03

Step 1) Copy into place (in the /SystemLibrary/Extensions/' folder) myKext which is a subclass of an existing KEXT. Step 2) Run 'kextload... myKext...' which succeeds. 'kexstat' lists it as running, albeit with far fewer listed dependencies. Step 3) Create an ioreg dump to a textfile. (ioreg -l > ~/Desktop/ioreg.txt) Observed: The object that should have over-ridden is sitting in the IOreg listing there where my KEXT should be loaded. If I reboot, myKext is in place and works fine. Without having to sustain a reboot - is there any preferred way to get my KEXT to become active and located in the place in the IORegistry it where it normally shows up after reboot? BTW - at the time I am trying to get myKext activated, there is no 'client' attached to the class I am trying to override. Thanks, Lance Drake _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: question about communication between KEXTs
by Quinn 25 Aug '03

25 Aug '03

At 12:57 -0600 24/8/03, Lance Drake wrote: Regarding this inter-kext communication situation, is there any reason why the KEXTs cannot discover themselves via the IORegistry and legally send messages via calls like 'setProperties( OSObject* properties)' ? No. It would seem that there already exists the ability to use OS services intended for such chores and that there is no need to invent some new, circuitous path. Indeed. Although it's hard to be 100% sure without analysing the specific situation in more detail. S+E -- Quinn "The Eskimo!" <http://www.apple.com/developer/> Apple Developer Technical Support * Networking, Communications, Hardware _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: question about communication between KEXTs
by Lance Drake 24 Aug '03

24 Aug '03

Regarding this inter-kext communication situation, is there any reason why the KEXTs cannot discover themselves via the IORegistry and legally send messages via calls like 'setProperties( OSObject* properties)' ? It would seem that there already exists the ability to use OS services intended for such chores and that there is no need to invent some new, circuitous path. Please correct me if this too is ill-advised. Lance Drake _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: question about communication between KEXTs
by Michael George 22 Aug '03

22 Aug '03

On Thursday, August 21, 2003, at 05:41 PM, rimas avizienis wrote: here's my situation. i've written a coreaudio driver to support a piece of custom hardware that transmits and receives digital audio over ethernet. my current driver communicates with the network interface by using dlil_attach_interface_filter to attach a callback function. this works fine for receiving packets and i transmit packets using dlil_inject_if_output. the problem with this driver is that when the machine is under a heavy load (running an audio synthesis program which does its computation in the high priority coreaudio callback thread) the outgoing packets don't get transmitted until the high priority thread relinquishes control (and then they get sent out in a big burst). this results in latency which i'm trying to minimize. i am presently trying to modify my coreaudio driver so that it speaks directly to the ethernet driver (bypassing the network stack entirely). Is this going to be for your own personal use, or to be distributed with your customer hardware that you hope to sell? If the latter, I would warn against anything that bypasses the network stack. -Michael _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: question about communication between KEXTs
by Brian Bergstrand 22 Aug '03

22 Aug '03

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Sounds like a hack to me, but if it works... Anyway, you need to declare a dependancy on the AppleGMac driver in your audio kext's OSBundleLibraries PLIST dictionary. This consists of the drivers reverse DNS id and the version. The kernel linker will then know to link in the AppleGMac driver when loading your kext. HTH. On Thursday, August 21, 2003, at 04:41 PM, rimas avizienis wrote: > greetings, > > I have acquired and read and compiled the > AppleGMacEthernet source but I'm not sure how to go > about doing what I want to do. I first tried > declaring a global variable in the AppleGMacEthernet > KEXT, thinking that i could then reference that > variable from my audio driver KEXT. however when i > try to load the audio driver KEXT it complains about > an undefined symbol for the variable. obviously i'm > missing something about the way the kernel namespace > works. > > anyways i would be very grateful for any advice or > direction as to how best establish communication > between these two KEXTs, like how i could get a > reference to the ethernet driver object from my > coreaudio driver or something of the sort. > > thanks! > > -rimas > Brian Bergstrand <http://www.classicalguitar.net/brian/> PGP Key: <http://www.classicalguitar.net/brian/misc/public_key.txt> Blessed is he who expects nothing, for he shall never be disappointed. - - Jonathan Swift -----BEGIN PGP SIGNATURE----- Version: PGP 8.0.2 iQA/AwUBP0VgHnnR2Fu2x7aiEQI88ACggllWAVXa4PX2ojfRq67EqYqARB8AnRm1 d2cYIT3poRFrgEXejTqoLAXk =Veu/ -----END PGP SIGNATURE----- _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

question about communication between KEXTs
by rimas avizienis 21 Aug '03

21 Aug '03

greetings, here's my situation. i've written a coreaudio driver to support a piece of custom hardware that transmits and receives digital audio over ethernet. my current driver communicates with the network interface by using dlil_attach_interface_filter to attach a callback function. this works fine for receiving packets and i transmit packets using dlil_inject_if_output. the problem with this driver is that when the machine is under a heavy load (running an audio synthesis program which does its computation in the high priority coreaudio callback thread) the outgoing packets don't get transmitted until the high priority thread relinquishes control (and then they get sent out in a big burst). this results in latency which i'm trying to minimize. i am presently trying to modify my coreaudio driver so that it speaks directly to the ethernet driver (bypassing the network stack entirely). my plan was to modify the AppleGMacEthernet driver in such a way as to enable my coreaudio driver to pass a callback function to the ethernet driver which would get called when a packet of the proper type was received. I have acquired and read and compiled the AppleGMacEthernet source but I'm not sure how to go about doing what I want to do. I first tried declaring a global variable in the AppleGMacEthernet KEXT, thinking that i could then reference that variable from my audio driver KEXT. however when i try to load the audio driver KEXT it complains about an undefined symbol for the variable. obviously i'm missing something about the way the kernel namespace works. anyways i would be very grateful for any advice or direction as to how best establish communication between these two KEXTs, like how i could get a reference to the ethernet driver object from my coreaudio driver or something of the sort. thanks! -rimas __________________________________ Do you Yahoo!? Yahoo! SiteBuilder - Free, easy-to-use web site design software http://sitebuilder.yahoo.com _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Out of kernel stack space
by Jim Magee 20 Aug '03

20 Aug '03

On Tuesday, August 19, 2003, at 8:02 PM, raldrich(a)mac.com wrote: My biggest question is, what thread APIs should I be using. Keeping in mind that I'd like this file system to eventually port to other flavors of BSD Unix, I don't think I should use IOKit threads, and I'm not sure about Mach threads either - Would Posix threads be appropriate? Nope. There is no pthread support in the kernel. And it really doesn't matter which of the other mechanisms you use to allocate your thread - they all assume a fixed size kernel stack (16KB currently). Also, it occurs to me that what I'm trying to do here doesn't really need the overhead of threads - an API similar to Windoze's _Call_On_My_Stack kernel function would work just as well. Does anything like that exist in Darwin? Nope. We do the opposite (call this when I wake up - forget about my stack until then). That's why a fixed-sized kernel stack is important on our platform. It sounds like you might have some stack allocations to go hunt down after all. --Jim _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Out of kernel stack space
by raldrich＠mac.com 20 Aug '03

20 Aug '03

Folks, I'm currently porting a UDF 2.x file system to MacOS X, and I've run into a bit of a problem. It seems that the original writers of the file system library didn't pay a whole lot of attention to stack usage, and I occasionally run the I/O kit out of stack space. Unfortunately, I don't have the leisure to rewrite the library, and I don't think I'd ever find all of the places where stack usage is too high anyway. I'm thinking that the simplest method for dealing with this would be to 'bottleneck' my file system's I/O by creating a separate thread for the I/O, hand the I/O off to that other thread, and block the calling thread until the I/O thread completes. My biggest question is, what thread APIs should I be using. Keeping in mind that I'd like this file system to eventually port to other flavors of BSD Unix, I don't think I should use IOKit threads, and I'm not sure about Mach threads either - Would Posix threads be appropriate? Also, it occurs to me that what I'm trying to do here doesn't really need the overhead of threads - an API similar to Windoze's _Call_On_My_Stack kernel function would work just as well. Does anything like that exist in Darwin? Thanks for your time, Ron Aldrich Software Architects, Inc. _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0

Re: Mail and the Kernel
by Justin C. Walker 18 Aug '03

18 Aug '03

This surely doesn't belong on the kernel list (the kernel never feels itself moved to send mail to anyone). Try darwinos-users, or darwin-developement. Regards, Justin On Monday, August 18, 2003, at 02:03 PM, David Blanton wrote: I have been trying to understand how applications send mail i.e. What system functions are likely to be called to send mail. Can anyone give me to a starting point? Thanks. David Blanton _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored. -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | Some people have a mental | horizon of radius zero, and | call it their point of view. | -- David Hilbert *--------------------------------------*-------------------------------* _______________________________________________ darwin-kernel mailing list | darwin-kernel(a)lists.apple.com Help/Unsubscribe/Archives: http://www.lists.apple.com/mailman/listinfo/darwin-kernel Do not post admin requests to the list. They will be ignored.

1 0