July 2002 - Darwin-kernel - lists.apple.com

RE: thread funnel and sbplio
by George Andre 22 Jul '02

22 Jul '02

> >I'm getting a nasty panic while doing a VOP_GETATTR(). > >It says that "splbio() not under funnel". Traced that > >down to splbio() which checks for the current thread funnel. > >I put some debug prints before the VOP_GETATTR() call and > >the thread has no funnel held. So what I did was add a: > See below. > >thread_funnel_set(kernel_flock,TRUE) > >respective > >thread_funnel_set(kernel_flock,FALSE) around the VOP_GETATTR() > >call, still with the same result... > > This should work. Are you sure you aren't making other calls > requiring the kernel funnel after releasing it? > Hi Brian, Yes, the panic is traced to the VOP_GETATTR call (commenting it out doesn't crash the kernel). > Where in your code are you trying to call this. The kernel usually > makes sure that you are under the correct funnel (which is probably > why your other VOP calls work fine). Places where you don't > automatically have the funnel and must acquire it yourself are: 1) > Your kmod start/stop entry points, 2) Any new thread entry point, and > 3) any timer callback. Well this is funny but I don't have the kernel funnel at all when calling the other stuff and it works perfectly. All this stuff is called from the start() method of a KEXT (actually 2 levels down). I have 4 functions I've implemented (kfread, kfclose, kfopen and kfsize). kfopen,close and kfread work ok (with NO funnel, checked by debug printing the value returned by thread_funnel_get()). These functions use the NDINIT, vn_open, VOP_UNLOCK, vn_rdwr, vn_close, without any failures and I'm able to read from the file what I want.. What I can't do however is call VOP_GETATTR safely in kfsize(). I've even put the whole kfsize() function inbetween thread_funnel_set's, with exactly the same result. What I don't obviously do is have thread_funnel_set called first in my start() and stop(). I'll try that immediately, but logically, it should be the same as making sure the funnel is acquired for one operation, right? I also read somewhere that funnels are released at rescheduling and reacquired after. Can that have something to do with my situation? > PS. Please don't crosspost. Pick one list and send to that. If you > don't get a response, try another list. 'Aight. :) Thanks, George. _______________________________________________ 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: thread funnel and sbplio
by Brian Bergstrand 22 Jul '02

22 Jul '02

At 6:09 PM +0200 7/22/02, George Andre wrote: Kernel ppl, I'm getting a nasty panic while doing a VOP_GETATTR(). It says that "splbio() not under funnel". Traced that down to splbio() which checks for the current thread funnel. I put some debug prints before the VOP_GETATTR() call and the thread has no funnel held. So what I did was add a: See below. thread_funnel_set(kernel_flock,TRUE) respective thread_funnel_set(kernel_flock,FALSE) around the VOP_GETATTR() call, still with the same result... This should work. Are you sure you aren't making other calls requiring the kernel funnel after releasing it? To be noted is that other VOP_ calls, vn_open, vn_rdwr, NDINIT and namei() work perfectly. What I'm not able to do is get the attributes for a open, locked vnode, without getting this panic. Where in your code are you trying to call this. The kernel usually makes sure that you are under the correct funnel (which is probably why your other VOP calls work fine). Places where you don't automatically have the funnel and must acquire it yourself are: 1) Your kmod start/stop entry points, 2) Any new thread entry point, and 3) any timer callback. HTH. PS. Please don't crosspost. Pick one list and send to that. If you don't get a response, try another list. -- Brian Bergstrand <http://www.classicalguitar.net/brian/> PGP Key ID: 0xB6C7B6A2 There's no way to rule innocent men. The only power any government has is the power to crack down on criminals. Well, when there aren't enough criminals, one makes them. One declares so many things to be a crime that it becomes impossible for men to live without breaking laws. - Ayn Rand _______________________________________________ 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

thread funnel and sbplio
by George Andre 22 Jul '02

22 Jul '02

Kernel ppl, I'm getting a nasty panic while doing a VOP_GETATTR(). It says that "splbio() not under funnel". Traced that down to splbio() which checks for the current thread funnel. I put some debug prints before the VOP_GETATTR() call and the thread has no funnel held. So what I did was add a: thread_funnel_set(kernel_flock,TRUE) respective thread_funnel_set(kernel_flock,FALSE) around the VOP_GETATTR() call, still with the same result... To be noted is that other VOP_ calls, vn_open, vn_rdwr, NDINIT and namei() work perfectly. What I'm not able to do is get the attributes for a open, locked vnode, without getting this panic. Anyone know what this is caused by, even though I acquire the kernel funnel? I'm using OSX 10.1.5.. ANY help appreciated, Thanks, George. _______________________________________________ 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: Data link NKE help
by Justin C. Walker 19 Jul '02

19 Jul '02

On Thursday, July 18, 2002, at 06:52 PM, Mike Cashwell wrote: Greetings all! I have a new project in front of me involving network port authentication. From what I can tell, I'll have a data link NKE and a user-land daemon. My plan is to be a good citizen and keep the in-kernel portion as small and tightly focused as possible. I expect to use a startupitem to load the NKE and start the daemon. After a trip to Apple's documentation and this list's archives I keep finding references to a pdf file that discusses NKE development. The URL is http://developer.apple.com/techpubs/macosx/Darwin/Extensions/NKE.pdf but the file I get from it is only 1K in size and neither Acrobat Reader nor Preview can make sense of it. Does anyone know of or have a good copy of this file? I realize it's out of date but it must be better than nothing. There have been references to file:///Developer/Documentation/Darwin/Extensions/NKE.pdf, which is much bigger than 1K; check that out. While I'd like to read that pdf before peppering the list with a lot of questions, here's what I'm looking for: 1: Page 123 of the KernelProgramming.pdf shows that Data Link NKEs can be both above and below the DLIL. What's the difference between these two and why would I choose one over the other? And once that's decided, how do I in the code choose where my NKE plugs in? This is discussed briefly in the doc; the deal is: - above DLIL, it's a "protocol filter" NKE, and sits between a specific protocol stack and a specific device - below DLIL, it's an "interface filter" NKE, and sits above a specific device so the difference is in the former case, you get a look at all traffic to and from a specified device, of a specified protocol type; while in the latter, you get a look at all traffic to and from a specified device, period. The distinction in the code is made by the registration call you make. 2: Assuming my NKE/daemon approach is valid, how does the NKE find network interfaces (of interest) and insert itself so that traffic cannot bypass it? It will need to do this when the NKE first loads and deal with interfaces appearing and disappearing thereafter. Anyone have pointers for this? Is there only one "instance" of the NKE or is it one per interface? Check out the SharedIP NKE for an indication of how this is done. This NKE handles the support for sharing an IP address with the Classic (Streams) stack, so that a single IP address is needed for both environments. The Classic app gets a list of the devices of interest from the kernel, and sends the name of the device it wants to the NKE via an IOCTL. There are other ways to communicate with a "filter-type" NKE than IOCTL; this is used for SharedIP since the Classic app already has a socket open to the device (a PF_NDRV socket, giving it 'raw' access). See below. 3: And lastly, I will need the daemon and NKE to be able to communicate with each other. How does the daemon find the NKE (or NKEs if there must be one per interface)? How do they establish bidirectional communication? (I only need a small amount of data transferred infrequently. I'm not sending all network traffic out to user land and back!) Look for the NKEMgr NKE in the darwin repository; this is the (current) mechanism used to communicate with NKEs, although in the longer term, this will be incorporated into the base kernel. There's an NKE called TCPLogger, and a command called tcplog, in the repository which show how to use this scheme. My plan for this (following what documentation I can find) is to use syscall() for the daemon to send data to the NKE. But I will need to allow the NKE to send data or events back to the daemon independently (ie: no polling!). With the PF_NKE (as it's now called) scheme implemented by NKEMgr, you get what you want. This is currently a "hack", awaiting integration into the base system, but it will work. In the IOKit world with a userclient I'd have a daemon thread wait on a queue that it had made known to the NKE. But NKEs don't have workloops or userclients, right? Any guidance would be appreciated. Of course data link NKE sample code would probably answer most of this. Anyone ever seen any? SharedIP and VLan are two that are in the repository. Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | If you're not confused, | You're not paying attention *--------------------------------------*-------------------------------* _______________________________________________ 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

Data link NKE help
by Mike Cashwell 19 Jul '02

19 Jul '02

Greetings all! I have a new project in front of me involving network port authentication. From what I can tell, I'll have a data link NKE and a user-land daemon. My plan is to be a good citizen and keep the in-kernel portion as small and tightly focused as possible. I expect to use a startupitem to load the NKE and start the daemon. After a trip to Apple's documentation and this list's archives I keep finding references to a pdf file that discusses NKE development. The URL is http://developer.apple.com/techpubs/macosx/Darwin/Extensions/NKE.pdf but the file I get from it is only 1K in size and neither Acrobat Reader nor Preview can make sense of it. Does anyone know of or have a good copy of this file? I realize it's out of date but it must be better than nothing. While I'd like to read that pdf before peppering the list with a lot of questions, here's what I'm looking for: 1: Page 123 of the KernelProgramming.pdf shows that Data Link NKEs can be both above and below the DLIL. What's the difference between these two and why would I choose one over the other? And once that's decided, how do I in the code choose where my NKE plugs in? 2: Assuming my NKE/daemon approach is valid, how does the NKE find network interfaces (of interest) and insert itself so that traffic cannot bypass it? It will need to do this when the NKE first loads and deal with interfaces appearing and disappearing thereafter. Anyone have pointers for this? Is there only one "instance" of the NKE or is it one per interface? 3: And lastly, I will need the daemon and NKE to be able to communicate with each other. How does the daemon find the NKE (or NKEs if there must be one per interface)? How do they establish bidirectional communication? (I only need a small amount of data transferred infrequently. I'm not sending all network traffic out to user land and back!) My plan for this (following what documentation I can find) is to use syscall() for the daemon to send data to the NKE. But I will need to allow the NKE to send data or events back to the daemon independently (ie: no polling!). In the IOKit world with a userclient I'd have a daemon thread wait on a queue that it had made known to the NKE. But NKEs don't have workloops or userclients, right? Any guidance would be appreciated. Of course data link NKE sample code would probably answer most of this. Anyone ever seen any? Thanks! Mike _______________________________________________ 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: stripping symbols from NKE (kext)
by Timothy Standing 18 Jul '02

18 Jul '02

Message: 3 Date: Wed, 17 Jul 2002 14:22:05 -0500 To: Peter Lovell <vpndev(a)mac.com>, darwin-kernel(a)lists.apple.com From: Brian Bergstrand <brian(a)classicalguitar.net> Subject: Re: stripping symbols from NKE (kext) At 3:07 PM -0400 7/17/02, Peter Lovell wrote: What's the right way to strip extra symbols from an NKE-style kext >(not IOKit one). > >I need to keep just the start and stop entry points, but can't seem >to get it right. Or can it not be done within PB, but requires a separate "strip" post-process step? Thanks.....Peter Lovell PB never seems to do the strip. I always use strip -x from the CLI. -- Brian Bergstrand <http://www.classicalguitar.net/brian/> PGP Key ID: 0xB6C7B6A2 Unix is user-friendly -- it's just selective about it's friends. --__--__-- Message: 4 Date: Wed, 17 Jul 2002 13:46:11 -0700 Subject: Re: stripping symbols from NKE (kext) Cc: Peter Lovell <vpndev(a)mac.com>, darwin-kernel(a)lists.apple.com To: Brian Bergstrand <brian(a)classicalguitar.net> From: Dean Reece <dean.reece(a)apple.com> PB will strip the kext if you set the "Build Style" (in the "Targets" tab) to "Deployment". Cheers, - Dean If I build all four executables in my application (including the kext and helper tools) in deployment mode (with 4/02 Dev Tools), I can strip more than 1.5 MB from the application bundle by running strip -x on each of the executables. It doesn't seem like PB is actually doing the strip. Is it PB supposed to be stripping symbols when you use deployment mode? If so, it looks like it is broken. Tim Standing SoftRAID, LLC _______________________________________________ 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: stripping symbols from NKE (kext)
by Dean Reece 17 Jul '02

17 Jul '02

PB will strip the kext if you set the "Build Style" (in the "Targets" tab) to "Deployment". Cheers, - Dean On Wednesday, July 17, 2002, at 12:22 PM, Brian Bergstrand wrote: At 3:07 PM -0400 7/17/02, Peter Lovell wrote: What's the right way to strip extra symbols from an NKE-style kext (not IOKit one). I need to keep just the start and stop entry points, but can't seem to get it right. Or can it not be done within PB, but requires a separate "strip" post process step? PB never seems to do the strip. I always use strip -x from the CLI. _______________________________________________ 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: stripping symbols from NKE (kext)
by Brian Bergstrand 17 Jul '02

17 Jul '02

At 3:07 PM -0400 7/17/02, Peter Lovell wrote: What's the right way to strip extra symbols from an NKE-style kext (not IOKit one). I need to keep just the start and stop entry points, but can't seem to get it right. Or can it not be done within PB, but requires a separate "strip" post-process step? Thanks.....Peter Lovell PB never seems to do the strip. I always use strip -x from the CLI. -- Brian Bergstrand <http://www.classicalguitar.net/brian/> PGP Key ID: 0xB6C7B6A2 Unix is user-friendly -- it's just selective about it's friends. _______________________________________________ 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

stripping symbols from NKE (kext)
by Peter Lovell 17 Jul '02

17 Jul '02

What's the right way to strip extra symbols from an NKE-style kext (not IOKit one). I need to keep just the start and stop entry points, but can't seem to get it right. Or can it not be done within PB, but requires a separate "strip" post-process step? Thanks.....Peter Lovell _______________________________________________ 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: File System Filter Driver for MAC OS X ?
by Krause, Axel 3805 S-WF-RD6 17 Jul '02

17 Jul '02

Is there some programming API that can filter (e.g.log) I/O operations for one or more file systems or file system volumes ( in order to work in conjunction with applications like antivirus or encryption programs ; as it exists for MicrosoftB.Windows NTB.) ? Axel Krause _______________________________________________ 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