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

if_set_bpf_tap
by Samuel Lellouche 10 Mar '02

10 Mar '02

Hi.. Has anyone managed to use the if_set_bpf_tap function documented in Network Kernel Extension ? Here is what I got when i try to load a network kext i'm developping. (I'm on OSX 10.1.3) [g4:~sam/OSX-PF] root# more OSX_PF.c #include <gen.h> #include <sys/systm.h> #include <mach/mach_types.h> kern_return_t OSX_PF_start (kmod_info_t * ki, void * d) { int test; printf("Module has loaded!\n"); test = 0; if_set_bpf_tap(test,test,test); return KERN_SUCCESS; } kern_return_t OSX_PF_stop (kmod_info_t * ki, void * d) { printf("Module will be unloaded\n"); return KERN_SUCCESS; } [g4:~sam/OSX-PF] root# kextload build/OSX-PF.kext kmodload: kld(): kmodload: Undefined symbols:kmodload: kmodload: _if_set_bpf_tap kmodload: kld_load() failed.Error loading module: kmod load error. [g4:~sam/OSX-PF] root# Thanx for any help. -- Samuel Lellouche - Omatis Expert Systhmes, riseaux et securiti E-mail: sam(a)omatis.com Tel: 01 53 14 59 14 / 01 53 14 59 31 _______________________________________________ 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: updating HFS+ inode info. on open() call
by Shawn Erickson 09 Mar '02

09 Mar '02

Why does the data need to be stored with the file (not that it isn't a sensible goal). Why not have a database of files and their usage be maintained by your logic that watches open/close. This would be better at supporting your predictive pre-fetching if the database was designed to allow for the types of searches you need to do. Having to grok the file system for usage information could be slow because it is not tuned for it. Basically you could do this by having a file system "filter" that you could lay over _any_ file system. This filter would maintain the database of usage directly. Basically the moment you try to store this information in the file system you become file system dependent (which may be fine in your solution). -Shawn On Wednesday, March 6, 2002, at 08:56 PM, Ian Brown wrote: I am trying to store statistical information with files that persists across reboots. Theoretically it looks like an attribute record (HFSPlusAttrInlineData) would be the best place to put this information, however it does not look like those work yet (correct me if I am wrong.) FIRST: I want to store a counter with each file (or file metadata) counting open()'s. I can then set them to zero and run filesystem benchmarks and get file access statistics. "But," they say, "this can be done by taking a trace." I know, but it would not show me how to store statistical information with files. Why do I want to do this then? NEXT: I will be able to store predictive information with files. Specifically I would like to store a reference to the file that gets accessed next. I plan to do this at the time of the next open(). This will give me file grouping information and predictive pre-fetching information. The situation with mmap() had not crossed my mind, however I am not sure if it is an issue with what I am ultimately trying to do. Ian What exactly are you trying to do? Is this total number of open() calls for a file counter? Does it persist across reboot? Are you planning on decrementing the counter when close() is called? If so, What do you intend to do for the case open() mmap() close() Now you counter goes to 0, but the inode is still in use... --Umesh -- Umesh Vaishampayan Apple Computer, Inc. Mac OS X Kernel Ph: (408) 974 0229 _______________________________________________ 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. _______________________________________________ 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: Need quick and reliable thread or timer
by Phil Montoya 08 Mar '02

08 Mar '02

With regard to your last quesiton... look at clock.h. I believe most of these calls are safe in the kernel or even in an interrupt task. As far as a periodic call back you may want to attach yourself to an interrupt and install an interrupt service routine. I doubt anything else would give you the reliability you are looking for. -Phil -----Original Message----- From: Sean Morrell [mailto:smorrell@artlogic.com] Sent: Friday, March 08, 2002 10:19 AM To: darwin-kernel(a)lists.apple.com Subject: Need quick and reliable thread or timer I am writing an audio driver and would like to perform a very small amount of work at least every millisecond, with no more variance than a tenth of a millsecond. What mechanisms can be employed here? Can a thread be used? Can the priority be raised high enough to ensure it is running once every millsecond? Is there a timer as accurate as a Time Manager Task in OS 9? Thanks, Sean _______________________________________________ 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. _______________________________________________ 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

Need quick and reliable thread or timer
by Sean Morrell 08 Mar '02

08 Mar '02

I am writing an audio driver and would like to perform a very small amount of work at least every millisecond, with no more variance than a tenth of a millsecond. What mechanisms can be employed here? Can a thread be used? Can the priority be raised high enough to ensure it is running once every millsecond? Is there a timer as accurate as a Time Manager Task in OS 9? Thanks, Sean _______________________________________________ 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: updating HFS+ inode info. on open() call
by Ian Brown 07 Mar '02

07 Mar '02

I am trying to store statistical information with files that persists across reboots. Theoretically it looks like an attribute record (HFSPlusAttrInlineData) would be the best place to put this information, however it does not look like those work yet (correct me if I am wrong.) FIRST: I want to store a counter with each file (or file metadata) counting open()'s. I can then set them to zero and run filesystem benchmarks and get file access statistics. "But," they say, "this can be done by taking a trace." I know, but it would not show me how to store statistical information with files. Why do I want to do this then? NEXT: I will be able to store predictive information with files. Specifically I would like to store a reference to the file that gets accessed next. I plan to do this at the time of the next open(). This will give me file grouping information and predictive pre-fetching information. The situation with mmap() had not crossed my mind, however I am not sure if it is an issue with what I am ultimately trying to do. Ian What exactly are you trying to do? Is this total number of open() calls for a file counter? Does it persist across reboot? Are you planning on decrementing the counter when close() is called? If so, What do you intend to do for the case open() mmap() close() Now you counter goes to 0, but the inode is still in use... --Umesh -- Umesh Vaishampayan Apple Computer, Inc. Mac OS X Kernel Ph: (408) 974 0229 _______________________________________________ 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: updating HFS+ inode info. on open() call
by Umesh Vaishampayan 07 Mar '02

07 Mar '02

On Monday, March 4, 2002, at 04:56 PM, Ian Brown wrote: I am looking for a way to update inode information whenever an open() call is made to an HFS+ file. Specifically I would like to keep a counter in the inode that counts the number of open() calls made to that file. Does anyone have any ideas on how to go about this? Regards, Ian What exactly are you trying to do? Is this total number of open() calls for a file counter? Does it persist across reboot? Are you planning on decrementing the counter when close() is called? If so, What do you intend to do for the case open() mmap() close() Now you counter goes to 0, but the inode is still in use... --Umesh -- Umesh Vaishampayan Apple Computer, Inc. Mac OS X Kernel Ph: (408) 974 0229 _______________________________________________ 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: IPSec Question
by Isao SEKI 05 Mar '02

05 Mar '02

At 16:47 03/05/2002 +0100, Emiel Kollof wrote: > Although I'd like to use ipv6 on Mac OS X, I doubt if Carbon or Cocoa > apps will be able to do ipv6. But having a ipv6 capable ircII on OSX > would be fine for me :) Btw, I am developing Cocoa application, irc client. UsoChat which is supporting both ipv4 and v6 connection. If you have interest, please visit here at http://gongon.com/persons/iseki/UsoChat/UsoChat.html Isao _______________________________________________ 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: IPSec Question
by Emiel Kollof 05 Mar '02

05 Mar '02

Jun-ichiro itojun Hagino heeft op dinsdag 5 maart 2002 om 12:56 het volgende geschreven: if you apply the following patch to Darwin kernel source (xnu) Apple-201-15 tag, kernel portion will become ready for IPsec. ftp://ftp.itojun.org/pub/misc/darwin-v6-patch.Apple-201-15.diff there are problems in compiling userland tools (like setkey(8) for setting up ipsec secrets) so it is still useless... I know... Even though I have these: http://www.hackerheaven.org/files/kame-puma-darwin-122101.tgz They won't build on Mac OS X. Mostly makefile problems. If someone manages to compile this on OS X, I want the binaries :-) Although I'd like to use ipv6 on Mac OS X, I doubt if Carbon or Cocoa apps will be able to do ipv6. But having a ipv6 capable ircII on OSX would be fine for me :) Cheers, Emiel _______________________________________________ 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: IPSec Question
by Jun-ichiro itojun Hagino 05 Mar '02

05 Mar '02

>I noticed on NetBSD.org that IPsec support is built in since 1999 but you >must rebuild the kernel to use it. I poked around under the kernel framework >of Mac OS X and found a header file related directly to IPSec along with >several other header files that are related to the way IPSec is implemented >on NetBSD. Darwin is based on NetBSD so this makes sense. > >I know that there are plenty of people out there looking for a IPSec support >for Mac OS X. At least I know I am chopping at the bit for it! > >I am not a Unix guru but am learning (slowly) SO I was wondering if anyone >out there, that really has a good grasp of BSD networking, has rebuilt >Darwin/Mac OS X with IPSec turned on in the kernel and been able to get >IPSec tunneling to work? if you apply the following patch to Darwin kernel source (xnu) Apple-201-15 tag, kernel portion will become ready for IPsec. ftp://ftp.itojun.org/pub/misc/darwin-v6-patch.Apple-201-15.diff there are problems in compiling userland tools (like setkey(8) for setting up ipsec secrets) so it is still useless... itojun _______________________________________________ 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: IPSec Question
by Josh Graessley 05 Mar '02

05 Mar '02

On 3/4/02 6:22 PM, "Guy Gardner" <guygardner(a)attbi.com> wrote: > All, > > I noticed on NetBSD.org that IPsec support is built in since 1999 but you > must rebuild the kernel to use it. I poked around under the kernel framework > of Mac OS X and found a header file related directly to IPSec along with > several other header files that are related to the way IPSec is implemented > on NetBSD. Darwin is based on NetBSD so this makes sense. My understanding is that the networking pieces of our kernel are loosely based on FreeBSD 3.2. FreeBSD didn't have the Kame IPSec and IPv6 integrated in to it like FreeBSD 4.4 does. A version of the Kame IPv6 and IPSec support was merged in to our kernel a long long time ago. Our IPSec and IPv6 support is out of date, which is probably one of the reasons it was never turned on. > I know that there are plenty of people out there looking for a IPSec support > for Mac OS X. At least I know I am chopping at the bit for it! > > I am not a Unix guru but am learning (slowly) SO I was wondering if anyone > out there, that really has a good grasp of BSD networking, has rebuilt > Darwin/Mac OS X with IPSec turned on in the kernel and been able to get > IPSec tunneling to work? I believe a few people have worked on compiling the Darwin sources with IPSec and IPv6 turned on. There are a few problems you will run in to with AirPort and something else declaring symbols that are declared in the kernel. Check out the archives for the details. > And: Does anyone have an idea if Apple is considering the next Mac OS X and > Darwin refreshes with IPSec options built into the Kernel and providing a > nice UI/Command line way to set this all up to work? The networking sources in the kernel have been updated to be closer to FreeBSD 4.4. I believe the merged sources haven't been pushed out to the anoncvs server yet. I think they should appear in the tag Apple-240 or later. It should be possible to build xnu with the ipsec and inet6 options. The IPv6 and IPSec command line tools as well as libinet6 and libipsec are not in Darwin yet, so you'll have no good way to test IPv6 and IPSec short of porting the libraries and tools yourself. -josh _______________________________________________ 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