February 2012 - Darwin-dev - lists.apple.com

Open Source Release
by William Siegrist 22 Feb '12

22 Feb '12

The open source components of Xcode 4.3 have been released. http://www.opensource.apple.com/release/developer-tools-43/ -Bill _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com

1 0

Sudden and hard to reproduce crashes in libcache
by Dragan Milić 21 Feb '12

21 Feb '12

First of all, I apologise if this is not the right topic for this list. Some beta testers of an application I'm working on experience rather strange and sudden crashes, very often when the application is doing nothing, just sitting idle in the background. The crash happens in a thread named "com.apple.root.default-priority" and it's not even controlled by the application. As far as I figure it out, it is triggered by the system to release some memory from caches the application uses. This is the part of the crash report related to that particular thread: Thread 8 Crashed:: Dispatch queue: com.apple.root.default-priority 0 libsystem_kernel.dylib 0x00007fff8d847ce2 __pthread_kill + 10 1 libsystem_c.dylib 0x00007fff9009e7d2 pthread_kill + 95 2 libsystem_c.dylib 0x00007fff9008fa7a abort + 143 3 libc++abi.dylib 0x00007fff8a1e77bc abort_message + 214 4 libc++abi.dylib 0x00007fff8a1e4fcf default_terminate() + 28 5 libobjc.A.dylib 0x00007fff8a69a1b9 _objc_terminate + 94 6 libc++abi.dylib 0x00007fff8a1e5001 safe_handler_caller(void (*)()) + 11 7 libc++abi.dylib 0x00007fff8a1e505c std::terminate() + 16 8 libc++abi.dylib 0x00007fff8a1e6152 __cxa_throw + 114 9 libobjc.A.dylib 0x00007fff8a699e7a objc_exception_throw + 327 10 com.apple.CoreFoundation 0x00007fff8c8b92a9 -[NSException raise] + 9 11 com.apple.ExceptionHandling 0x0000000100332e0a NSExceptionHandlerUncaughtSignalHandler + 37 12 libsystem_c.dylib 0x00007fff900f0cfa _sigtramp + 26 13 com.apple.CoreFoundation 0x00007fff8c793c0b CFRelease + 27 14 libcache.dylib 0x00007fff8fb4d8e4 _value_entry_remove + 142 15 libcache.dylib 0x00007fff8fb4d9cd _entry_unmap + 81 16 libcache.dylib 0x00007fff8fb4da71 _entry_evict + 153 17 libcache.dylib 0x00007fff8fb4db04 _evict_last + 106 18 libcache.dylib 0x00007fff8fb4dcf2 _cache_enforce_limits + 34 19 libcache.dylib 0x00007fff8fb4dea5 cache_handle_memory_pressure + 125 20 libdispatch.dylib 0x00007fff888bf2b6 _dispatch_source_invoke + 635 21 libdispatch.dylib 0x00007fff888bbf77 _dispatch_queue_invoke + 71 22 libdispatch.dylib 0x00007fff888bb760 _dispatch_worker_thread2 + 198 23 libsystem_c.dylib 0x00007fff9009e3da _pthread_wqthread + 316 24 libsystem_c.dylib 0x00007fff9009fb85 start_wqthread + 13 All those crash reports for the crashing thread look the same. Only a few people experience them, but regularly. For others, it practically impossible to reproduce. I'm using higher level NSCache API to cache objects. Cached objects are various image representations and much rarely attributed strings and NSNumber instances. I see the crash happens after CFRelease() call, but I can't figure out why. If I try to do CFRelease(NULL), I get almost the same sequence, similar to: 3 libc++abi.dylib 0x00007fff8a1e77bc abort_message + 214 4 libc++abi.dylib 0x00007fff8a1e4fcf default_terminate() + 28 5 libobjc.A.dylib 0x00007fff8a69a1b9 _objc_terminate + 94 6 libc++abi.dylib 0x00007fff8a1e5001 safe_handler_caller(void (*)()) + 11 7 libc++abi.dylib 0x00007fff8a1e505c std::terminate() + 16 8 libc++abi.dylib 0x00007fff8a1e6152 __cxa_throw + 114 9 libobjc.A.dylib 0x00007fff8a699e7a objc_exception_throw + 327 10 com.apple.CoreFoundation 0x00007fff8c8b92a9 -[NSException raise] + 9 11 com.apple.ExceptionHandling 0x0000000100332e0a NSExceptionHandlerUncaughtSignalHandler + 37 12 libsystem_c.dylib 0x00007fff900f0cfa _sigtramp + 26 But I don't know who and why does (if at all) that NULL pointer release. The documentation for libcache says "Calling cache functions from cache callbacks should be avoided to prevent deadlock." I thought maybe when releasing cached objects libcache calls callback methods to release memory and then image representations are doing some caching on their own, which would effectively be calling cache functions from cache callbacks. But this is just a wild guess, I've no idea what is really going on. So, I'd really appreciate if anyone can help me with this. I've got no idea whether it's a but in the application, or the OS. If necessary, I can send the full crash report and also explain what object are being cached using NSCache. -- Dragan _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com

1 0

Re: RTLD_NEXT does not look into INSERTED_LIBRARIES
by Carlos Sánchez de La Lama 17 Feb '12

17 Feb '12

Hi Peter, Thanks a lot, I have it working now (as I said by linking against original libraries) but this looks much better. Thanks a lot, Carlos El 17/02/2012, a las 00:53, Peter O'Gorman escribió: > On 02/10/2012 02:48 AM, Carlos Sánchez de La Lama wrote: >> Hi all, >> >> I am interposing some functions between an application and a framework, >> >> by using DYLD_INSERT_LIBRARIES="enhancer:profiler", it does not work correctly. When using RTLD_NEXT on dlsym to access next symbol, > > Hi, > > I don't think anybody answered you? > > Don't use dlsym with RTLD_NEXT or DYLD_FORCE_FLAT_NAMESPACE, look at: > > http://opensource.apple.com/source/dyld/dyld-132.13/include/mach-o/dyld-int… > > And use that DYLD_INTERPOSE() macro. You can completely avoid using dlsym, no need for flat namespace, and your replacement functions can even be static. > > Peter _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com

1 0

Re: RTLD_NEXT does not look into INSERTED_LIBRARIES
by Peter O'Gorman 16 Feb '12

16 Feb '12

On 02/10/2012 02:48 AM, Carlos Sánchez de La Lama wrote: Hi all, I am interposing some functions between an application and a framework, by using DYLD_INSERT_LIBRARIES="enhancer:profiler", it does not work correctly. When using RTLD_NEXT on dlsym to access next symbol, Hi, I don't think anybody answered you? Don't use dlsym with RTLD_NEXT or DYLD_FORCE_FLAT_NAMESPACE, look at: http://opensource.apple.com/source/dyld/dyld-132.13/include/mach-o/dyld-int… Peter _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com And use that DYLD_INTERPOSE() macro. You can completely avoid using dlsym, no need for flat namespace, and your replacement functions can even be static.

1 0

Re: SMARTQuery Sample Broken on 10.7
by Alison Cassidy 14 Feb '12

14 Feb '12

Hi Mike, I just downloaded the sources here, and built and ran it without any issues on my 10.7.3 (11D16) machine. The initial build target seemed to be a bit off, but I just kicked it over to 10.7, and then to 10.6, and both builds seemed to run just fine on Lion. Regards, -- Allie On Feb 13, 2012, at 11:37 PM, Mike wrote: > The SMARTQuery sample code project at: > > https://developer.apple.com/library/mac/#samplecode/SMARTQuery/Introduction… > > appears to be broken on 10.7. The VerifyIdentifyData() routine fails. Any ideas on how to fix it so it works on both 10.6 & 10.7? > _______________________________________________ > 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: > https://lists.apple.com/mailman/options/darwin-dev/cooties%40mac.com > > This email sent to cooties(a)mac.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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com

1 0

SMARTQuery Sample Broken on 10.7
by Mike 14 Feb '12

14 Feb '12

The SMARTQuery sample code project at: https://developer.apple.com/library/mac/#samplecode/SMARTQuery/Introduction… _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com appears to be broken on 10.7. The VerifyIdentifyData() routine fails. Any ideas on how to fix it so it works on both 10.6 & 10.7?

1 0

Re: Kernel panic in Lion
by Jean-Daniel Dupas 11 Feb '12

11 Feb '12

Le 11 févr. 2012 à 12:39, Anders Rayner-Karlsson a écrit : > Hi there, > > I've hit a bit of a snag with Lion when using iTerm2 and the patched tmux that the iTerm2 project have provided - it causes the kernel to panic in Lion as you detach from the tmux session. Easiest way to hit it is to install iTerm2 from http://code.google.com/p/iterm2/downloads/list and grab the tmux-for-iTerm2 archive as well, then using homebrew install libevent 2.0.16 (there's instructions around building this at http://code.google.com/p/iterm2/wiki/TmuxIntegration) and then set up two profiles in iTerm2, one that runs "tmux -C" to create a new session, and one that runs "tmux -C attach" to attach to a running tmux session. > > When launching the "tmux -C" session, two windows open, one terminal running in tmux, and one control window with a small menu - in the control window, just hit Escape to detach and the bug's triggered. > > I've tried to attach a debugger to a VM that I use to get some additional information, and this is what I've got: > > Program received signal SIGTRAP, Trace/breakpoint trap. > 0xffffff80002c5e7b in Debugger (message=0xb2f <Address 0xb2f out of bounds>) at model_dep.c:930 > 930 __asm__("int3"); > (gdb) bt > #0 0xffffff80002c5e7b in Debugger (message=0xb2f <Address 0xb2f out of bounds>) at model_dep.c:930 > #1 0xffffff8000220702 in panic (str=0xffffff80006a0bd8 "\"Negative open count?\"@/SourceCache/xnu/xnu-1699.24.23/bsd/miscfs/specfs/spec_vnops.c:1368") at debug.c:306 > #2 0xffffff8000328af2 in spec_close (ap=0xffffff80008b1948) at spec_vnops.c:1368 > #3 0xffffff800032c895 in devfsspec_close (ap=0xffffff80463ebd30) at devfs_vnops.c:665 > #4 0xffffff8000318ef4 in VNOP_CLOSE (vp=0xffffff800aa42170, fflag=48, ctx=0xffffff80008b3d08) at kpi_vfs.c:3417 > #5 0xffffff80002fd1aa in vclean (vp=0xffffff800aa42170, flags=4) at vfs_subr.c:2054 > #6 0xffffff80002fceae in vgone [inlined] () at /SourceCache/xnu/xnu-1699.24.23/bsd/vfs/vfs_subr.c:2255 > #7 0xffffff80002fceae in vnode_reclaim_internal (vp=0xffffff800aa42170, locked=1178516976, reuse=1178516976, flags=1) at vfs_subr.c:4137 > #8 0xffffff80002fe3d1 in vn_revoke (vp=0xffffff800aa42170, flags=0, a_context=0xaa4217000000000) at vfs_subr.c:2197 > #9 0xffffff8000316a1f in VNOP_REVOKE (vp=0xffffff80008b1948, flags=0, ctx=0xaa4217000000000) at kpi_vfs.c:4023 > #10 0xffffff8000540e29 in proc_exit (p=0xffffff8009ac9b80) at kern_exit.c:537 > #11 0xffffff800023ba0b in thread_terminate_self () at thread.c:368 > #12 0xffffff800023bf0a in special_handler (rh=0xffffff8009ab0e20, thread=0xffffff8009772a00) at thread_act.c:844 > #13 0xffffff800023bfa0 in act_execute_returnhandlers () at thread_act.c:754 > #14 0xffffff800021e0ce in ast_taken (reasons=32, enable=1) at ast.c:165 > #15 0xffffff80002c1c8a in i386_astintr (preemption=162205216) at trap.c:1293 > (gdb) > > As I can trigger this at will, if there is anyone that want to get additional information about this to squash this bug, let me know what information you'd want (commands for me to run in gdb, what to look for) and I'll obtain it. I'd like to get this solved, somehow. :) > You should fill a bug report at http://bugreport.apple.com/ > /Anders > _______________________________________________ > 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: > https://lists.apple.com/mailman/options/darwin-dev/devlists%40shadowlab.org > > This email sent to devlists(a)shadowlab.org -- Jean-Daniel _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com

1 0

Kernel panic in Lion
by Anders Rayner-Karlsson 11 Feb '12

11 Feb '12

Hi there, I've hit a bit of a snag with Lion when using iTerm2 and the patched tmux that the iTerm2 project have provided - it causes the kernel to panic in Lion as you detach from the tmux session. Easiest way to hit it is to install iTerm2 from http://code.google.com/p/iterm2/downloads/list and grab the tmux-for-iTerm2 archive as well, then using homebrew install libevent 2.0.16 (there's instructions around building this at http://code.google.com/p/iterm2/wiki/TmuxIntegration) and then set up two profiles in iTerm2, one that runs "tmux -C" to create a new session, and one that runs "tmux -C attach" to attach to a running tmux session. When launching the "tmux -C" session, two windows open, one terminal running in tmux, and one control window with a small menu - in the control window, just hit Escape to detach and the bug's triggered. I've tried to attach a debugger to a VM that I use to get some additional information, and this is what I've got: Program received signal SIGTRAP, Trace/breakpoint trap. 0xffffff80002c5e7b in Debugger (message=0xb2f <Address 0xb2f out of bounds>) at model_dep.c:930 930 __asm__("int3"); (gdb) bt #0 0xffffff80002c5e7b in Debugger (message=0xb2f <Address 0xb2f out of bounds>) at model_dep.c:930 #1 0xffffff8000220702 in panic (str=0xffffff80006a0bd8 "\"Negative open count?\"@/SourceCache/xnu/xnu-1699.24.23/bsd/miscfs/specfs/spec_vnops.c:1368") at debug.c:306 #2 0xffffff8000328af2 in spec_close (ap=0xffffff80008b1948) at spec_vnops.c:1368 #3 0xffffff800032c895 in devfsspec_close (ap=0xffffff80463ebd30) at devfs_vnops.c:665 #4 0xffffff8000318ef4 in VNOP_CLOSE (vp=0xffffff800aa42170, fflag=48, ctx=0xffffff80008b3d08) at kpi_vfs.c:3417 #5 0xffffff80002fd1aa in vclean (vp=0xffffff800aa42170, flags=4) at vfs_subr.c:2054 #6 0xffffff80002fceae in vgone [inlined] () at /SourceCache/xnu/xnu-1699.24.23/bsd/vfs/vfs_subr.c:2255 #7 0xffffff80002fceae in vnode_reclaim_internal (vp=0xffffff800aa42170, locked=1178516976, reuse=1178516976, flags=1) at vfs_subr.c:4137 #8 0xffffff80002fe3d1 in vn_revoke (vp=0xffffff800aa42170, flags=0, a_context=0xaa4217000000000) at vfs_subr.c:2197 #9 0xffffff8000316a1f in VNOP_REVOKE (vp=0xffffff80008b1948, flags=0, ctx=0xaa4217000000000) at kpi_vfs.c:4023 #10 0xffffff8000540e29 in proc_exit (p=0xffffff8009ac9b80) at kern_exit.c:537 #11 0xffffff800023ba0b in thread_terminate_self () at thread.c:368 #12 0xffffff800023bf0a in special_handler (rh=0xffffff8009ab0e20, thread=0xffffff8009772a00) at thread_act.c:844 #13 0xffffff800023bfa0 in act_execute_returnhandlers () at thread_act.c:754 #14 0xffffff800021e0ce in ast_taken (reasons=32, enable=1) at ast.c:165 #15 0xffffff80002c1c8a in i386_astintr (preemption=162205216) at trap.c:1293 (gdb) As I can trigger this at will, if there is anyone that want to get additional information about this to squash this bug, let me know what information you'd want (commands for me to run in gdb, what to look for) and I'll obtain it. I'd like to get this solved, somehow. :) /Anders _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com

1 0

MFS Lives for Lion?
by Mike 11 Feb '12

11 Feb '12

_______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com Any chance I can get MFS Lives sample building for Lion under Xcode 4.2? I noticed the VFS descriptors changed in the file system for VFS'es between pre-10.6 and 10.6+. MFS Lives no longer builds on 10.6+ systems.

1 0

RTLD_NEXT does not look into INSERTED_LIBRARIES
by Carlos Sánchez de La Lama 10 Feb '12

10 Feb '12

Hi all, I am interposing some functions between an application and a framework, having two different interposing dynamic libraries: an enhancer and a profiler, where both define the same functions. They work separately, meaning: application -> framework (normal case, works of course). application -> enhancer -> framework (interposing the enhancer, works). application -> profiler -> framework (interposing the profiler, works). However, when I try to interpose both (enhancing and profiling), like in: application -> enhancer -> profiler -> framework by using DYLD_INSERT_LIBRARIES="enhancer:profiler", it does not work correctly. When using RTLD_NEXT on dlsym to access next symbol, enhancer goes directly for framework symbols even when the profiler was forcedly inserter after it. For clarity, I am also forcing FLAT_NAMESPACE. My guess is RTLD_NEXT looks for linked dynamic libraries only, does not look into runtime inserted ones. I can solve my problem linking the enhancer against the profiler when I want to profile, but (if possible) I would prefer not to need that so I can interpose each "piece" as needed with no need to recompile/relink. Does anybody know if my guess is correct, or am I missing something? Thanks and best regards, Carlos _______________________________________________ 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: https://lists.apple.com/mailman/options/darwin-dev/site_archiver%40lists.ap… This email sent to site_archiver(a)lists.apple.com

1 0