January 2003 - Darwin-kernel

Re: Building the kernel 6.0.2
by Michael Kress 17 Jan '03

17 Jan '03

Just to be clear, in my previous post the setenv command was broken into two lines in the digest. Realizing this limitation the command should read: setenv CVSROOT \ :pserver:username@anoncvs.opensource.apple.com:/cvs/Darwin mjk _______________________________________________ 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: Disk access management and poor performance
by Jonas Maebe 17 Jan '03

17 Jan '03

On vrijdag, jan 17, 2003, at 11:32 Europe/Brussels, Matthias Kretschmer wrote: Preventing disk access for starting applications if nearly impossible, because the data has to be read out for it. It can only be improved by caching all data that was read or reading out sequentially (if possible) data, that will be used. Darwin 6.0/Mac OS X 10.2 and later keep a record of which data is initially used by an application when it starts up in /var/vm/app_profiles for exactly this reason. Jonas _______________________________________________ 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: Disk access management and poor performance
by Matthias Kretschmer 17 Jan '03

17 Jan '03

Hello! First of all, I don't know what is implemented in the darwin kernel at all, but here are some of my thoughts about this topic: Preventing disk access for starting applications if nearly impossible, because the data has to be read out for it. It can only be improved by caching all data that was read or reading out sequentially (if possible) data, that will be used. In an offline-situation it is possible to calculate what is the best way to read and cache disk-data, but in the online-situation it is somehow hard to tell which data could be read next. It is easy to guess if someone opens a file, that and begins to read it sequentially, that he could read the whole file, so the system could begin to cache it, if memory is available and no other disk-access is forced. This all shouldn't be a problem which is new to the world, so one should find some online-algorithms with a good competivness in the appropriate text-books. One question would be, how swapping is implemented. There has to be some way of telling the system, that the swap files don't have to be cached. Caching of swap-space is somehow stupid :) By using caches, one could also cache disk-write-access, trying to reduce disk-access at all, enhancing spin-down-times and for mobile users this should reduce battery usage. I was thinking of it some while ago, but hadn't had the time to look at the different parts of darwin related to disk-access. Message: 4 > Date: Thu, 16 Jan 2003 17:24:53 +0000 > Subject: Disk access management and poor performance > From: John Hunter <johnhunter(a)netcourrier.com> > To: darwin-kernel(a)lists.apple.com > > Hello, > > I get the impression that many slowdowns in OS X are caused by an > application causing a lot of disk access, which for some reason brings > down the performance of other processes, even if they don't require so > much disk access. For example, contextual menus in the Dock are > terribly unresponsive while an application is lauching, which leads to > stupid Dock behaviour if items are being added or removed (that's my > personal experience on a Titanium Powerbook with loads or RAM); > sometimes it is difficult only to drag a window. > I was wondering why that is? > Also, wouldn't it be possible for the kernel to implement some sort of > throttling policy for disk access (for direct file i/o and for disk > access caused by virtual memory paging)? Isn't that done on other OSes? > I remember an early version of Omniweb had a terrible memory leak. It > was thrashing memory so quickly that it was all but impossible to > force-quit it! It's not good enough that an application can't crash the > whole system, if it can make it so slow as to be unusable... > > I would be grateful for insights from kernel gurus! > John > -- Greetings Matthias Kretschmer _______________________________________________ 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: Building the kernel 6.0.2
by chelhan＠unitel.co.kr 17 Jan '03

17 Jan '03

>Um, sorry for the total ignorance, but what about building on a Mac OS >X on PPC? I am running 10.2.3. > >BTW, are the sources found in xnu the same as those provided in a Mac >OS X install? I mean, if I build mach_kernel will it work if I replace >the mach_kernel that is provided as part of Mac OS X? as of today, macos 10.2.3 has the following xnu tag number 344.23 But apple cvs repository has the most recent(and default) xnu tag number "APPLE-344-12-2, APPLE-10-2, xnu-344-12-2" which is MacOS X 10.2.2 and drwin kernel version 6.2. This version has error that there are no files "vfs_journal.[c, h]" Maybe, It seems that current kernel 6.3 is not yet imported to apple CVS repository. Does anybody know the location macos X 10.2.3 kernel sources are? [Macintosh : ~] shbae% hostinfo Mach kernel version: Darwin Kernel Version 6.3: Sat Dec 14 03:11:25 PST 2002; root:xnu/xnu-344.23.obj~4/RELEASE_PPC Kernel configured for up to 2 processors. 1 processor is physically available. Processor type: ppc7400 (PowerPC 7400) Processor active: 0 Primary memory available: 512.00 megabytes. Default processor set: 50 tasks, 115 threads, 1 processors Load average: 0.52, Mach factor: 0.59 -- Bae, Sok Hon _______________________________________________ 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: Building the kernel 6.0.2
by Jeff Jenkins 17 Jan '03

17 Jan '03

Date: Wed, 15 Jan 2003 18:55:41 +0900 (KST) From: "" <chelhan(a)unitel.co.kr> To: rrp-lists(a)pacbell.net Cc: darwin-kernel(a)lists.apple.com Subject: Building the kernel 6.0.2 Justin Walker has valiantly suggested several possibilities for #1, but I don't see anything for the second item. There is a web page that describes how to build the kernel on a Mac OS X 10.1 system, but it doesn't seem to hold for 10.2. >> Yes Indeed. You install 6.0.2 darwin iso image on x86. and You should >> download CoreFoundations( >> http://web.mit.edu/darwin/www/CF-262-1151.tar.gz >>) >> and gcc3-1208.tar.gz and cctools-440.tar.gz from the >> http://web.mit.edu/darwin/www >> and you retreive xnu source and check your path and the following files >> as listed in 10.1 kernel build howto documents. >>/usr/local/bin/seg_hack >>/usr/local/bin/relpath >>/usr/local/include/streams/streams.h >>/usr/local/include/streams/streamsextra.h >>/usr/local/lib/libkld.a >>/usr/local/lib/system/libstreams.a >> You finished the prerequsite action to build the darwin kernel. >>% source SETUP/setup.csh >>% make exporthdrs >>% make >> That's all. >> Good lucks. Um, sorry for the total ignorance, but what about building on a Mac OS X on PPC? I am running 10.2.3. BTW, are the sources found in xnu the same as those provided in a Mac OS X install? I mean, if I build mach_kernel will it work if I replace the mach_kernel that is provided as part of Mac OS X? -- jrj _______________________________________________ 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

Disk access management and poor performance
by John Hunter 16 Jan '03

16 Jan '03

Hello, I get the impression that many slowdowns in OS X are caused by an application causing a lot of disk access, which for some reason brings down the performance of other processes, even if they don't require so much disk access. For example, contextual menus in the Dock are terribly unresponsive while an application is lauching, which leads to stupid Dock behaviour if items are being added or removed (that's my personal experience on a Titanium Powerbook with loads or RAM); sometimes it is difficult only to drag a window. I was wondering why that is? Also, wouldn't it be possible for the kernel to implement some sort of throttling policy for disk access (for direct file i/o and for disk access caused by virtual memory paging)? Isn't that done on other OSes? I remember an early version of Omniweb had a terrible memory leak. It was thrashing memory so quickly that it was all but impossible to force-quit it! It's not good enough that an application can't crash the whole system, if it can make it so slow as to be unusable... I would be grateful for insights from kernel gurus! John _______________________________________________ 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: Booting alternative kernels from primary disk?
by Josh de Cesare 16 Jan '03

16 Jan '03

Mark, Here's what I do after testing a kernel that did not work... Reboot, and stop at Open Firmware (hold CMD-OPT-O-F). Then I change the boot-file parameter to point at my back up kernel; usually /mach_kenrel.orig. Then boot the machine, fix the kernel and clear boot-file parameter. In OF type: printenv boot-device This will show the path to BootX; usually hd:9,\\:tbxi. The important thing to note here, is the partition number (9) and the path to the disk (hd:). When setting the boot-file use the same path and partition number, but put in the new kernel name at the end. setenv boot-file hd:9,mach_kernel.orig To continue booting type: boot After the machine has booted in terminal or on the console at root, fix /mach_kernel and clear the boot-file with the nvram command: nvram boot-fille="" What do you do if you mess up a boot driver on the disk? The only thing I have found is to boot off of a different partition and replace it with the original one. This can be done from single user off of an install CD. Here's the other way I do kernel or driver development... Net boot everything. It does require two machine and you will need a DHCP server on your network. First enable tftpd on your build machine. Copy /usr/standalone/ppc/bootx.xcoff to your tftpboot directory, usually /private/tftpboot. Make a link in /private/tftpboot to the build results of your xnu build and call is something like mach_kernel.test. You will also need to make an mkext to include all the important drivers. You will need a directory will all the drivers from the target build of the OS, I assume below that they are the same as the build machine. kextcache -lna ppc -m /private/tftpboot/mach_kernel.test.mkext /System/Library/Extensions The mkext name must be <kernel-name>.mkext Now the target machine needs to be configured. Set the boot-device to point over the network to BootX, and the boot-file to point at the kernel. Be sure to fill in the IP address of your build machine. setenv boot-device enet:12.34.56.78,bootx.xcoff setenv boot-file enet:12.34.56.78,mach_kernel.test At this point the machine should be able to boot, by finding BootX and the kernel and mkext, but it will not know what to do about the root disk. Set boot-args to point back at the disk. Normally, if it gets to single user, I'm done, so I include -s, and debug=0x14e in case it doesn't. setenv boot-args -s debug=0x14e rd=*hd:9 Once all the parameters are set type sync-nvram to flush them out. If the boot does not make it all the way through BootX, OF will not flush them out, and you will have to type it all again. As you rebuild new kernel, or change drivers (and rebuild the mkext) on your build machine all you need to do to test the new stuff is reboot the test machine. Good luck. Josh At 6:27 AM -0800 1/16/03, Mark Grimes wrote: "Before you install the new kernel, make a copy of your original kernel. When things fail you can always copy the original kernel back by booting from a different disk or partition. Either X or 9 will work." (from xnu building for mortals - circa 10.1.1) I am rather speechless. This seems like a rather obsolete suggestion for machines that ship with solely OSX now. Is there no such way to boot from the backup kernel you made from the primary disk?. I only have X, I don't have 9. I also only have one Mac. Do I need to keep the original install cd's on hand whenever I am in the mood to dabble? I have been searching the archives for this very question... either this is really a stupid question or everyone has enough methods to get themselves out of binds. Can someone give me a shout out as to where I missed the boat? -- Mark Grimes <mark(a)stateful.net> Stateful Labs _______________________________________________ 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: Building the kernel 6.0.2
by Michael Kress 16 Jan '03

16 Jan '03

I assume you've installed the Developers Kit from the 10.2 Disk. Have you installed the darwintools.pkg? It's on Apple's website, follow the links at the Apple Darwin page. I can't remember the path right now, but it's there somewhere. Get a CVS account. To log in to CVS from the Terminal, setenv CVSROOT :pserver:username@anoncvs.opensource.apple.com:/cvs/Darwin cvs login cvs -z3 checkout -r Apple-344 xnu (Apple-344 is the CVS tag, substitute the one you want) Once the download is complete, cd xnu source SETUP/setup.csh make Once the make is complete sudo cp /mach_kernel /mach_kernel_bck sudo cp Build/obj/RELEASE_PPC/mach_kernel / Restart and if all is well you should have a new kernel running. Good luck finding the kernel you want, though. 10.2.3 hadn't been merged the last time I checked. mjk _______________________________________________ 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

Booting alternative kernels from primary disk?
by Mark Grimes 16 Jan '03

16 Jan '03

"Before you install the new kernel, make a copy of your original kernel. When things fail you can always copy the original kernel back by booting from a different disk or partition. Either X or 9 will work." (from xnu building for mortals - circa 10.1.1) I am rather speechless. This seems like a rather obsolete suggestion for machines that ship with solely OSX now. Is there no such way to boot from the backup kernel you made from the primary disk?. I only have X, I don't have 9. I also only have one Mac. Do I need to keep the original install cd's on hand whenever I am in the mood to dabble? I have been searching the archives for this very question... either this is really a stupid question or everyone has enough methods to get themselves out of binds. Can someone give me a shout out as to where I missed the boat? -- Mark Grimes <mark(a)stateful.net> Stateful Labs _______________________________________________ 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

Building the kernel 6.0.2
by chelhan＠unitel.co.kr 15 Jan '03

15 Jan '03

> Justin Walker has valiantly suggested several possibilities for #1, > but I don't see anything for the second item. There is a web page > that describes how to build the kernel on a Mac OS X 10.1 system, but > it doesn't seem to hold for 10.2. > Yes Indeed. You install 6.0.2 darwin iso image on x86. and You should download CoreFoundations( http://web.mit.edu/darwin/www/CF-262-1151.tar.gz ) and gcc3-1208.tar.gz and cctools-440.tar.gz from the http://web.mit.edu/darwin/www and you retreive xnu source and check your path and the following files as listed in 10.1 kernel build howto documents. /usr/local/bin/seg_hack /usr/local/bin/relpath /usr/local/include/streams/streams.h /usr/local/include/streams/streamsextra.h /usr/local/lib/libkld.a /usr/local/lib/system/libstreams.a You finished the prerequsite action to build the darwin kernel. % source SETUP/setup.csh % make exporthdrs % make That's all. Good lucks. _______________________________________________ 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