March 2005 - Darwin-dev - lists.apple.com

Symbol table, atos, nlist
by Antonio Ferraioli 04 Mar '05

04 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com /** According to gcc documentation: called upon function entry */ void __attribute__((__no_instrument_function__)) __cyg_profile_func_enter(void *this_fn, void *call_site) { callDeep += 2; cygDebugMessage(deepPrefix(@"enter", this_fn, call_site)); } /** According to gcc documentation: called upon function exit */ void __attribute__((__no_instrument_function__)) __cyg_profile_func_exit(void *this_fn, void *call_site) { cygDebugMessage(deepPrefix(@"exit", this_fn, call_site)); callDeep -= 2; }

1 0

applesingle for > 2GB ; ._ files ; rsync
by Dave Yost 03 Mar '05

03 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com At 4:49 PM -0700 2001-04-26, Byron Han wrote: >Subject: Re: Resource Forks aware rsync? (fwd) > >as a matter of fact we DO need a new file format since AppleSingle is >limited to <2GB file sizes... Well? Is there a new format yet? http://developer.apple.com/cgi-bin/search.pl?&q=applesingle doesn't illuminate even the old format. And while we're at it: is there documentation somewhere as to the format of a ._ file, as referred to here: http://developer.apple.com/documentation/MacOSX/Conceptual/SystemOverview/F… and is there an official name for these ._ files? Are they simply the other half of appledouble? I'm putting together a spec for an extended rsync, which I will float here. I wonder if the Tiger rsync will satisfy it. Thanks Dave > > >On Thu, 26 Apr 2001, Peter Bierman wrote: > >> At 10:55 AM -0400 4/26/01, James E. Quick wrote: >> >I am working on a tar port right now. >> > >> >So far it looks like a small amount of information will be >> >sufficient. The whole FInfo+FXInfo (or DInfo + DXInfo structures) >> >along with backup and creation times. If anyone else knows of other >> >structures which are relevant, please drop a note. >> >> >> Please don't invent yet another Macintosh file+metadata encoding method. >> >> AppleSingle properly encodes the data fork, resource fork, and all of the metadata, both currently used fields and reserved fields. If you use AppleSingle instead of inventing your own encoding, then you'll know that your files will continue to work even if Apple starts using metadata you weren't saving. >> >> Besides, the world doesn't need yet another file format. >> >> -pmb >> _______________________________________________ >> darwin-development mailing list >> darwin-development(a)lists.apple.com >> http://www.lists.apple.com/mailman/listinfo/darwin-development >> > >-- >Byron Han - Localization and Release Engineering - Apple Computer, Inc. >_______________________________________________ >darwin-development mailing list >darwin-development(a)lists.apple.com >http://www.lists.apple.com/mailman/listinfo/darwin-development _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: kernel panics and nvram issues
by Shantonu Sen 03 Mar '05

03 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Thu, 3 Mar 2005, John Francini wrote: Kevin, Here are some possible options... This is your best bet. Shantonu _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… a) Boot into Open Firmware and set it there. To do so: b) If you've got Panther on another disk or partition (possibly on a FireWire external disk), boot that, run the nvram command, then shut down and reboot into Tiger. This will unfortunately not quite work. When you use Startup Disk to switch your boot device, the "boot-args" variable is erased (in order to prevent stray things like NetBoot rp= arguments from messing up your next boot). If you want to do this, use Startup Disk, quit System Preferences, when open Terminal and use nvram, and then reboot. Don't do it in the opposite order. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Call stack in signal handler
by Geoffrey Schmit 03 Mar '05

03 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 03/03/2005 at 11:59 AM +0100, Jonas Maebe <jonas(a)zeus.ugent.be> wrote: > Is there a standard way to crawl the stack safely when you are in a > procedure called by signal handler? I believe that MoreBacktrace (part of the MoreIsBetter sample code), handles this: <http://developer.apple.com/samplecode/MoreIsBetter/listing41.html> geoff -- Geoffrey Schmit Senior Software Engineer National Instruments geoffrey.schmit(a)ni.com www.ni.com _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: kernel panics and nvram issues
by John Francini 03 Mar '05

03 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Kevin, Here are some possible options... Hope this helps, John Francini On Mar 3, 2005, at 4:24 PM, Kevin Elliott wrote: Sorry about replying to an old message... Is their any way around this? _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… a) Boot into Open Firmware and set it there. To do so: 0. Reboot your system. 1. Hold down Command-Option-O-F as the computer chimes. Hold this chord down until you get the Open Firmware prompt (ok). 2. Enter the command: setenv boot-args -v 3. Boot the machine by entering the command: mac-boot This will stick until you either zap the PRAM or enter a "reset-all" Open Firmware command. b) If you've got Panther on another disk or partition (possibly on a FireWire external disk), boot that, run the nvram command, then shut down and reboot into Tiger. At 15:12 -0800 on 2/8/05, Shantonu Sen wrote: You need to reboot successfully at least once. The kernel only syncs the nvram settings at shutdown since it has a >finite number of stores. You are most likely running the nvram(8) commandline tool and not rebooting. The core dumping settings will not >be honored (they are only parsed at startup) and the settings will not be preserved. I'm debug a tiger issue by turning on verbose booting using nvram boot-args="-v". Unfortunately, for unknown reasons, the latest tiger releases have not been shutting down cleanly on my test machine. -- ______________________________________________________ Arguing with an engineer is like wrestling with a pig in mud. After a while, you realize the pig is enjoying it. ______________________________________________________ Kevin Elliott <mailto:kelliott@mac.com> AIM/iChatAV: kelliott(a)mac.com (video chat available) ______________________________________________________ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/francini%40mac.com This email sent to francini(a)mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: kernel panics and nvram issues
by Kevin Elliott 03 Mar '05

03 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Sorry about replying to an old message... Is their any way around this? At 15:12 -0800 on 2/8/05, Shantonu Sen wrote: You need to reboot successfully at least once. The kernel only syncs the nvram settings at shutdown since it has a >finite number of stores. You are most likely running the nvram(8) commandline tool and not rebooting. The core dumping settings will not >be honored (they are only parsed at startup) and the settings will not be preserved. I'm debug a tiger issue by turning on verbose booting using nvram boot-args="-v". Unfortunately, for unknown reasons, the latest tiger releases have not been shutting down cleanly on my test machine. -- ______________________________________________________ Arguing with an engineer is like wrestling with a pig in mud. After a while, you realize the pig is enjoying it. ______________________________________________________ Kevin Elliott <mailto:kelliott@mac.com> AIM/iChatAV: kelliott(a)mac.com (video chat available) ______________________________________________________ _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… This email sent to site_archiver(a)lists.apple.com

1 0

Re: Virtual memory leaks...
by Mike Smith 03 Mar '05

03 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com We have a very large application that uses a couple of dozen libraries from various sources. Our application is leaking virtual memory. Periodically, there is a 1 Mb bump in virtual memory, which is never reclaimed. Over time, several days, the accumulative affects of this leak are that we run out of virtual memory. We've tried all the standard tools such as leaks, and MallocDebug, but our application is too large for these. They either crash or slow our application down to the point that it is useless. Have you tried introspecting your application with 'heap'? I'm wondering if there is someway for me to get my hooks into the system allocator. If so, everytime I see a 1 Mb bump in virtual memory I could log the stack to a file. The stack may lead me to the leak. = Mike _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… It's unlikely; the 1MB bump is probably the allocator grabbing a new heap for tiny (< 512B) objects. You can interpose at several levels. For the ultimate in dirty tricks (I do not recommend this *except* for debugging purposes) look at malloc_zone_t in <malloc/malloc.h>. You can fetch the default malloc zone (used to back generic malloc() calls) with malloc_default_zone(). You might also look at the malloc stack logging options. You're not going to survive "several days" running with it on, but it might help you locate the sorts of callers allocating your leaked size, which would be less painful than catching a lot of breakpoints. This email sent to site_archiver(a)lists.apple.com

1 0

How do I find the cause when OSX locks up?
by Anton Kuzmin 03 Mar '05

03 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com User-agent: Mozilla Thunderbird 1.0 (Windows/20041206) Thank you, Anton Kuzmin _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I'm experiencing a very strange problem on 10.3.8 with one of our own proxies (smtp/pop3 proxy) - from time to time (once or twice a day), the whole system locks up - I can move the mouse cursor, but everything else becomes completely unresponsive, all applications(even daemons) freeze, but I can ping this computer while it's locked up, and if I try to connect to any standard service (like ftp or ssh) - the tcp connection can be established instantly but no data is ever received by the client (i.e. ftp server prompt). I've tried to run a simple script on this computer that logs top and zprint output every second into a file, and the data from the script shows no apparent anomalies in memory usage just before the computer locks up. Sometimes it unlocks itself after a random amount of time (can be from 20 seconds to a few hours) in which case the top output shows crazy cpu usage percentages for different processes, which total at up to 700-1000% (which could be just a problem with top getting confused by the lock up). Are there any other monitoring utilities except top and zprint that I can use to figure out what's going on? I tried the same configuration on another computer and it's the same problem again, which, I think, rules out possible hardware problems. Any insight on what other resources(or lack thereof) can cause this behavior and how to monitor them would be very helpful. This email sent to site_archiver(a)lists.apple.com

1 0

Call stack in signal handler
by Jonas Maebe 03 Mar '05

03 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hello, Thanks, Jonas _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… Is there a standard way to crawl the stack safely when you are in a procedure called by signal handler? Just following the saved stack pointers causes a crash when you arrive at the signal frame. Inside the signal handler itself it's of course possible to obtain the original stack pointer from the signal's context, but when you're inside a function called from a signal handler, this function may not know there's a signal frame somewhere higher up on the stack. I looked at the gdb code, but that one simply tries to peek at some values and catches any SIGSEGVS that may occur as a result. Is this really the only way? Or is there a default implementation in one of the system libraries which I can use? (I don't need symbolic resolution, just the addresses themselves). This email sent to site_archiver(a)lists.apple.com

1 0

Re: Standard way to communicate between NKEs and user space?
by Peter Sichel 03 Mar '05

03 Mar '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Mar 3, 2005, at 2:27 AM, OL&L Lists wrote: - Peter Sichel Sustainable Softworks <http://www.sustworks.com> _______________________________________________ Do not post admin requests to the list. They will be ignored. Darwin-dev mailing list (Darwin-dev(a)lists.apple.com) Help/Unsubscribe/Update your Subscription: http://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.app… I am sure this question has been asked many times but can someone point me to the docs that explain the standard way on OS X to enable two-way communication between a NKE KEXT and a user-space app? The standard way is through a socket. There are a few variations on this technique, but the basic idea is to make your NKE a socket filter and then intercept calls to set socket options and/or read and write data in the sockets buffer. To connect from your application, you open a UDP (or PF_NKE) socket and then set the option to add your socket filter. One wrinkle is that the documentation for this API is a little obscure, and the API itself is changing for Mac OS X Tiger. You'll want to dig through Apple's examples and find the file "NKE.pdf" on Jaguar or earlier to see how it all hangs together. This email sent to site_archiver(a)lists.apple.com

1 0