August 2002 - Darwin-kernel

Re[2]: TCPLogger Actual PB Project
by Quinn 26 Aug '02

26 Aug '02

At 7:26 -0700 24/8/02, devdrvr.net Multi-Platform SW wrote: Went to the URL below, logged on, where is TCPLogger in this directory path? I see 50 or so folders and went in half of them that had names even close to a net-kext-socket-thing and no TCPLogger. Is that the right path? And yes, the projects there are complete and buildable, thanks. Stuff has move around a little on the CVS repository. Here's the exact URL for TCPLogger. <http://www.opensource.apple.com/cgi-bin/registered/cvs/src/live/TCPLogger/> As a rule, most Darwin stuff is now either at the top level or in "src/live", except the kernel (xnu), which is in "src/notlive". 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

Darwin Kernel Tidbits
by Chris Edgington 24 Aug '02

24 Aug '02

Just a few things I learned today while running my build of the kernel: * You can use the latest toolset (bootstrap_cmds, etc.) to build a previous version of the kernel successfully. I built 201.42.3 (OS X 10.1.5) without problems. * You can add IOLog statements to virtually any piece of kernel code to get nice info in system.log. * If you've got a panic that's stopping you before the system.log output is recorded, you can comment out the panic, replace it with an IOLog of "Kernel would have paniced here because..." and then you can see all of the system.log that led up to the panic. * I'm SO glad the source for this OS is available! Later, -Chris _______________________________________________ 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

Matching tools to xnu
by Chris Edgington 24 Aug '02

24 Aug '02

How do I determine which version of the various tools to checkout in order to build a version of the kernel? For example, I'm running 10.1.5 - which has xnu 201.42.3. I can checkout xnu with tag Apple-201-42-3, but how does this relate to the tags on bootstrap_cmds for example? Thanks, -Chris _______________________________________________ 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[2]: TCPLogger Actual PB Project
by devdrvr.net Multi-Platform SW 24 Aug '02

24 Aug '02

Went to the URL below, logged on, where is TCPLogger in this directory path? I see 50 or so folders and went in half of them that had names even close to a net-kext-socket-thing and no TCPLogger. Is that the right path? And yes, the projects there are complete and buildable, thanks. --Perry > See http://www.opensource.apple.com/cgi-bin/registered/cvs. The > projects there should be complete and buildable. _______________________________________________ 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: TCPLogger Actual PB Project
by Justin C. Walker 24 Aug '02

24 Aug '02

On Thursday, August 22, 2002, at 02:14 , devdrvr.net Multi-Platform SW wrote: OK. I see an MIT site that has http://web.mit.edu/darwin/src/modules/TCPLogger/ some of the files that would make a building project for the NKE sample on Darwin. But where on Apple's site is the NKE proj I can actually build? Justin was good enough to mention that it is there. He's such a nice guy :-}. See http://www.opensource.apple.com/cgi-bin/registered/cvs. The projects there should be complete and buildable. Regards, Justin -- Justin C. Walker, Curmudgeon-At-Large * Institute for General Semantics | It's not whether you win or lose... | It's whether *I* win or lose. *--------------------------------------*-------------------------------* _______________________________________________ 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

TCPLogger Actual PB Project
by devdrvr.net Multi-Platform SW 22 Aug '02

22 Aug '02

OK. I see an MIT site that has http://web.mit.edu/darwin/src/modules/TCPLogger/ some of the files that would make a building project for the NKE sample on Darwin. But where on Apple's site is the NKE proj I can actually build? Justin was good enough to mention that it is there. Thanks, --Perry ======== 01 (530) 400-5692 [Davis, CA USA] mailto:perry_gregg@post.harvard.edu http://www.devdrvr.net/PerryCard.html ftp://www.devdrvr.net/ _______________________________________________ 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: Loading KEXTs dynamically
by Peter Lovell 21 Aug '02

21 Aug '02

On Wednesday, August 21, 2002, at 01:24 PM, Dean Reece wrote: You still use kextload to load, but it's quite a different kextload. We were careful to preserve the options that were likely to be used for programmatic driver loading. The biggest changes are to options that are only useful for development (symbol generation, etc). True. the new one is quite nice. The one thing which caught me off-guard (and took a while to figure out) is that the new one generates output on stdout, whereas the old one did not. That causes a problem if you have a pipe setup in place and the parent wasn't expecting output from kextload on stdout. I solved the problem by changing the way kextload was invoked and redirecting stdout to null. Regards.....Peter _______________________________________________ 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: Loading KEXTs dynamically
by Dean Reece 21 Aug '02

21 Aug '02

On Tuesday, August 20, 2002, at 07:21 AM, Peter Lovell wrote: On Monday, August 19, 2002, at 12:25 PM, Sean McBride wrote: Can I load and use a KEXT dynamically? If so, what is the API and where would I find a source example? man kextload You really should read the various docs about IOKit and kexts, etc. Also note that the rules have changed quite substantially in Jaguar. Actually, the rules haven't really changed; they are just being enforced now. The "Fixer" panel is our concession to the fact that such drivers are in circulation and should continue to be supported. You still use kextload to load, but it's quite a different kextload. We were careful to preserve the options that were likely to be used for programmatic driver loading. The biggest changes are to options that are only useful for development (symbol generation, etc). And access permissions and ownership rules are applied to kexts now. Yes - we should have been enforcing them from day one, but that somehow slipped by. kmodstat and kmodunload are replaced by kext-named commands. Yes, though the kmod* commands should not have been used for programmatic loading of kexts; they were present just for developers, and were subsumed by the various kext* commands. Cheers, - Dean _______________________________________________ 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: No space in available paging segments
by Philip Parker 21 Aug '02

21 Aug '02

On Wednesday, August 21, 2002, at 09:12 AM, Olivier Cinquin wrote: I have not seen this message with MacOS X, however when MacOS X is adding swap spaces in /var/vm, it runs like a slug. I experienced this feature early in MacOS X, so I modified /etc/rc and changed the arguments to dynamic_pager to dynamic_pager -H 40000000 -L 576870920 -S 536870912 -F ${swapdir}/swapfile So now it creates 524MB swap spaces instead on 80MB, and its performance was much more predictable. Yes, this was extremely helpful. I don't have an problem anymore. I wonder why your modification isn't the default setting? Olivier Glad it helped, did you notice how many swapfiles MacOS X created before making this modification ? Default System Configuration is always a compromise, so Apple were going for the 90%, and more importantly stability for the 90%. Personally, I'd rather see a primary swap partition so that any kernel panics can be collected (option) with savecore, instead of having to either copy down a screen of info, or taking a picture of it. On the dynamic_pager side, it might be better to use real memory as a factor for its arguments. Philip -- email: philip_parker(a)mac.com #include <std_disclaimer> _______________________________________________ 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: No space in available paging segments
by Philip Parker 21 Aug '02

21 Aug '02

On Tuesday, August 20, 2002, at 10:33 AM, Olivier Cinquin wrote: On Tuesday, August 20, 2002, at 04:34 , David A. Gatwood wrote: On Tue, 20 Aug 2002, Olivier Cinquin wrote: Hello, I was just wondering what the following message means? Could only find an obscure reference to KDE on the Web... Aug 20 11:14:55 localhost mach_kernel: (default pager): [KERNEL]: no space in available paging segments; swapon suggested I get the message after playing animated gifs I created, which causes an incredible amount of memory thrashing when played from Web browsers, for some reason... Wow. I haven't seen that message since MkLinux DR3. Does your system crash when this happens? No, it just gets very unresponsive and spends its time swapping... The message was repeated more than 100 times within a few minutes. Olivier I have not seen this message with MacOS X, however when MacOS X is adding swap spaces in /var/vm, it runs like a slug. I experienced this feature early in MacOS X, so I modified /etc/rc and changed the arguments to dynamic_pager to dynamic_pager -H 40000000 -L 576870920 -S 536870912 -F ${swapdir}/swapfile So now it creates 524MB swap spaces instead on 80MB, and its performance was much more predictable. Now as to your message, in other VM UNIX's I've run across, they have been limited to a number of swap devices (16 comes to mind), so perhaps the system is looking for more swap space to map, but dynamic_pager cannot add any more. 16 X 80 = 1280, only ~ 1GB If you can repeat this situation with the default setup, do a ls -al /var/ vm and see how many swapfiles there are. Also note that when it does not require extra swap spaces, the dynamic_pager will deallocate swapfiles except for the primary one, and this also seems to take time, as active pages seem to be consolidated from one swapfile to another ... a lot of IO occurs. I wonder if KMAP space is also consolidated, I'd expect serious pregnant pauses in a non Monolithic Kernel. Hope this helps Philip _______________________________________________ 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