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

Re: GCC:Nested Function Support
by fletcher sandbeck 06 Dec '05

06 Dec '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On 12/6/05 at 12:06 AM by andre-smith(a)speakeasy.net (Andre Smith): >Can nested functions be enabled using some sort of flag in GCC on Mac >OS X. Could someone explain why nested functions were disabled in GCC >on Mac OS X? I am attempting to compile the GRUB2 source code and >have encountered this build error. This thread from another Apple list addresses the issue. <http://lists.apple.com/archives/Xcode-users/2005/Nov/msg00267.html>. [fletcher] _______________________________________________ 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

GCC:Nested Function Support
by Andre Smith 06 Dec '05

06 Dec '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Can nested functions be enabled using some sort of flag in GCC on Mac OS X. Could someone explain why nested functions were disabled in GCC on Mac OS X? I am attempting to compile the GRUB2 source code and have encountered this build error. _______________________________________________ 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: build STLport 5 under 10.4?
by darwindev.mbloore＠xoxy.net 05 Dec '05

05 Dec '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:Cc:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=vMnr19Qa0aYfLSGB5KcJEna+FkM4t3qJ+LGtbzS9GmXQxxT2YFLsVgFVO3rj4fM1OIziiW9eS9WnYD22Qc+s9CRTriwaHZungltEXXMjoO9e2MaxJ8SS9Oc9Qspxvxj5xjLPLN9qLYpCrxqzz9UfRngorXeHwN6GqU5rL+nZefs= ; That was it! Thank you very much, Paul. > Date: Mon, 5 Dec 2005 19:15:46 +0000 > From: Paul Thomas <spongelavapaul(a)mac.com> > Subject: Re: build STLport 5 under 10.4? > > nm libstlport.5.1.0.dylib |grep stlp_std > > - any output means that you need to rebuild your code with > _PTHREADS defined. -- mARK bLOORE <mbloore(a)yahoo.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: build STLport 5 under 10.4?
by Paul Thomas 05 Dec '05

05 Dec '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com i.e. nm libstlport.5.1.0.dylib |grep stlpmtx_std OR nm libstlport.5.1.0.dylib |grep stlp_std Hope this helps, but I am using 5.0 so I might be a little off. Paul. On 5 Dec 2005, at 17:38, darwindev.mbloore(a)xoxy.net wrote: /usr/bin/ld: Undefined symbols: stlpmtx_std::__stl_throw_length_error(char const*) stlpmtx_std::cout stlpmtx_std::ios_base::_M_throw_failure() collect2: ld returned 1 exit status when Xcode runs the command Can you tell me if you did anything special to link with STLport? From: Paul Thomas <spongelavapaul(a)mac.com> Subject: Re: build STLport 5 under 10.4? On 22 Nov 2005, at 17:52, darwindev.mbloore(a)xoxy.net wrote: Has anyone built STLport 5 using gcc4? After setting native include paths in _site_config.h _______________________________________________ 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 did have a few problems with linking, and I think it was because of the thread-safe vs. non-thread-safe library versions. in doc/ stlport_namespaces.txt (in the distribution), it says that stlpmtx_std is the namespace used in the non-debug, non-thread-safe library. I originally had trouble because the libraries built thread safe (stlp_std) but my code was trying to use non-thread-safe (stlpmtx_std). Have a look at the output of nm run on the library libstlport. 5.1.0.dylib and see if the symbols begin with stlpmtx_std or some other namespace. If the output of nm has stlp_std in all the symbols, then it means you have the same problem. It should be solved by adding the _PTHREADS define in the diff I posted before. - any output means the symbols are there and I can't explain your problem - any output means that you need to rebuild your code with _PTHREADS defined. Thank you Paul, that has gotten me started. STLport 5 now compiles and links for me, but I have been unable to link it to a project built with Xcode 2.2. I get /usr/bin/g++-4.0 -o /Users/mark/tests/gentest/build/Debug/gentest -L/Users/mark/tests/gentest/build/Debug -L/Users/mark/misc_code/ metakit-2.4.9.4/builds -L/Users/mark/misc_code/STLport-5.1.0/lib -F/Users/mark/tests/ gentest/build/Debug -filelist /Users/mark/tests/gentest/build/gentest.build/Debug/gentest.build/ Objects-normal/i386/gentest.LinkFileList -lstlport -lstlport.5.1.0 -lstlport.5.1 -lstlport.5 -arch i386 - isysroot /Developer/SDKs/MacOSX10.4u.sdk I have tried setting various symbols, such as LD_LIBRARY and DYNLD_LIBRARY, and putting the stlport.dylib file and the various aliases generated by its makefile into different locations, such as /usr/local/lib and the build folder, without success. I built from CVS (just updated and rebuilt to check) but I only changed _STLP_NATIVE_INCLUDE_PATH, not the others #CXXFLAGS += -fuse-cxa-atexit -- mARK bLOORE <mbloore(a)yahoo.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/spongelavapaul% 40mac.com This email sent to spongelavapaul(a)mac.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: build STLport 5 under 10.4?
by darwindev.mbloore＠xoxy.net 05 Dec '05

05 Dec '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=yahoo.com; h=Message-ID:Received:Date:From:Subject:To:In-Reply-To:MIME-Version:Content-Type:Content-Transfer-Encoding; b=oUwCQyxlcnrgnDlXUss8rzVVRa5Ua5793eK1qHHX8oHXBVYHnIDiPp09t2OxK8v+QEMU0ybKgQMVEBIOwCEdA/HAGZYPQCzmszqZk/E0+i5rCPch+5Y6ewVkqzkUhImxh0jmGGVq2NZVJPHTnsogyQE947UlhMnlU7sJL7BvuSY= ; Thank you Paul, that has gotten me started. STLport 5 now compiles and links for me, but I have been unable to link it to a project built with Xcode 2.2. I get /usr/bin/ld: Undefined symbols: stlpmtx_std::__stl_throw_length_error(char const*) stlpmtx_std::cout stlpmtx_std::ios_base::_M_throw_failure() collect2: ld returned 1 exit status when Xcode runs the command /usr/bin/g++-4.0 -o /Users/mark/tests/gentest/build/Debug/gentest -L/Users/mark/tests/gentest/build/Debug -L/Users/mark/misc_code/metakit-2.4.9.4/builds -L/Users/mark/misc_code/STLport-5.1.0/lib -F/Users/mark/tests/gentest/build/Debug -filelist /Users/mark/tests/gentest/build/gentest.build/Debug/gentest.build/Objects-normal/i386/gentest.LinkFileList -lstlport -lstlport.5.1.0 -lstlport.5.1 -lstlport.5 -arch i386 -isysroot /Developer/SDKs/MacOSX10.4u.sdk I have tried setting various symbols, such as LD_LIBRARY and DYNLD_LIBRARY, and putting the stlport.dylib file and the various aliases generated by its makefile into different locations, such as /usr/local/lib and the build folder, without success. Can you tell me if you did anything special to link with STLport? > From: Paul Thomas <spongelavapaul(a)mac.com> > Subject: Re: build STLport 5 under 10.4? > On 22 Nov 2005, at 17:52, darwindev.mbloore(a)xoxy.net wrote: > > > Has anyone built STLport 5 using gcc4? After setting native > > include paths in _site_config.h > > > > I built from CVS (just updated and rebuilt to check) but I only changed > _STLP_NATIVE_INCLUDE_PATH, not the others > > #CXXFLAGS += -fuse-cxa-atexit -- mARK bLOORE <mbloore(a)yahoo.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: weird IOKit crash in IOIteratorNext()
by Chris Thomas 05 Dec '05

05 Dec '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com On Dec 1, 2005, at 10:08 PM, Julian Mayer wrote: - its prebound, incorrectly. - its been stomped by a memory smasher. You might turn off prebinding and see if that fixes the issue. Chris _______________________________________________ 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… IOIteratorNext does not call CFRelease at + 0x10, it calls _io_iterator_next by way of a dyld_stub, which loads a function pointer and jumps to it. There are only two reasonable ways that dyld's function pointer could be wrong. My bet, from having witnessed this happen before, is the first. it is still not clear to me why the binary in question crashes on one machine and runs properly on all other...is there a real problem and the prebinding problem only accounts for an incorrect crash log, or is prebinding completely responsible for the problem? if so, what do i have to do to remedy this problem? does this also mean that the rest of the crash log is also wrong? You may find a difference between 10.3.9 vs 10.4.x. Also, if prebinding is redone by an installer, the problem may (or may not) be magically fixed (or magically broken). In one case -- in a dylib I dynamically loaded using NSAddImage -- not only did I have to disable prebinding, I had to set the "bind everything at load time" flag. It "just worked" on 10.4.x, but on 10.3.9, it would bind incorrectly at runtime -- but not on all 10.3.9 systems, just some of them... This email sent to site_archiver(a)lists.apple.com

1 0

Re: Determining the canonical case for a file name on HFS+
by Adam Nohejl 03 Dec '05

03 Dec '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hello Camillo, You wrote: "I need to know this for the purpose of comparison, and I cannot simply do a case-insensitive comparison because I do not know if / folder/ is on a case-insensitive or case-sensitive volume." This is not a good solution, because: 1. Other file systems can be case-insensitive too (FAT for instance). -- Adam Nohejl Loki Software mailto:adam@lokisw.com http://lokisw.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… 2005/12/02 v 1:59, Rehan Rizvi: I was just wondering, what is stopping you from finding out what kind of volume it is? The information is there in the volume header. If you want to find out more about volume headers you can do a search on the developer.apple.com for "tn 1150" or HFSPlusVolumeHeader. 2. There can be other equivalent names with the same case, in HFS for instance composed vs decomposed Unicode - "a\314\201" and "\303\241" is the same file name (represents aacute). And of course if you're comparing paths you've got from the user, "a/.///./../a/./b" and "a/b" is also the same file. This email sent to site_archiver(a)lists.apple.com

1 0

Re: Catching and preventing hard sleep
by Ethan Bold 02 Dec '05

02 Dec '05

site_archiver(a)lists.apple.com Delivered-To: darwin-dev(a)lists.apple.com Hi Eric- Sorry. Not preventable at all. -ethan On Dec 2, 2005, at 1:08 PM, apple_lists(a)mobileneurons.com wrote: Greetings, // Include <IOKit/pwr_mgt/RootDomain.h> // Prevents hard sleep when closing lid. found in Insomnia::init() IOPMrootDomain *root = getPMRootDomain(); root->receivePowerNotification(kIOPMDisableClamshell); // Allows hard sleep when closing lid. found in Insomnia::free() IOPMrootDomain *root = getPMRootDomain(); root->receivePowerNotification(kIOPMEnableClamshell); * http://binaervarianz.de/projekte/programmieren/meltmac/ I appreciate any and all commentary. Most humble regards, Eric Coffin mobileneurons _______________________________________________ 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/ebold%40apple.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… Sorry, catching and preventing hard sleep is impossible, even from the kernel. I have not seen Insomnia.kext before, but that kext has a funny, if completely unsupported technique for disabling clamshell sleep. When you attempted to kextload Insomnia.kext at kIOMessageSystemWillSleep time, the kernel had already run through the clamshell checks; Insomnia.kext would have loaded far too late. I have been reading through documentation and mailing list archives for sometime in search of a particular bit of functionality specific to apple laptops. I need to catch and veto hard sleep caused by the closing of a laptop lid on system running OS 10.4. >From my research I understand that this may be achieved through the Kernel, likely through a KEXT. I have looked at the example RegisterForSleep, which shows how a program can register to listen for and handle system sleep calls. This allows an application to prevent idle sleep and gives them the oppourtunity to do some housekeeping before an impending hard sleep, caused by closing the laptop, or clicking sleep from the apple menu. I have also looked at the KEXT called Insomnia* which when installed prevents the system from going into hard sleep when the laptop lid is closed. This is achieved by communicating with the root power domain: Though preventing sleep is part of the solution I need to implement, more importantly I must be able to catch it, prevent it. Attempting a combination of RegisterForSleep and Insomnia did not do it. I loaded the KEXT when the event kIOMessageSystemWillSleep was sent to my program but to no avail. The system will sleep. I have read a few brief snippets within the mailing list that said it is possible to catch a hard sleep at the kernel level and veto it, but I have yet to discover how to do this. In an older version of <IOKIT/pwr_mgt/IOPM.h> I came across the following enumeration definition: // Power events enum { kClamshellClosedEventMask = (1<<0), // User closed lid kDockingBarEventMask = (1<<1), // OBSOLETE kACPlugEventMask = (1<<2), // User plugged or unplugged adapter kFrontPanelButtonEventMask = (1<<3), // User hit the front panel button kBatteryStatusEventMask = (1<<4) // Battery status has changed }; kClamshellClosedEventMask looks very interesting. I don't have a sense of how I could setup a event listener for this though. As well, the current IOPM.h file on my system is actually without this definition. This email sent to ebold(a)apple.com This email sent to site_archiver(a)lists.apple.com

1 0

Re: Notification handling without run loop of a primary thread
by Vladimir Magaziy 02 Dec '05

02 Dec '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com Hi Brian, Thanks for so quick reply. > The run loop needs an event source or it will just exit (as you > witnessed). So, before calling CFRunLoopRun() you should register > your notification handler with the run loop. I noticed that when I add a notification observer in a primary thread using the CFNotificationCenterAddObserver the run loop works correctly without any source and CFRunLoopRun does not return a control, this in turn adds some entries to a run loop internal representation -- modes and current modes set entries, this set does not contain any source!: CFShow(CFRunLoopGetCurrent()), but if I do the same in a separate thread an internal run loop set contains appropriate entries, but its behavior are not the same. As I get, this is the one mechanism to register notification handler with a run loop. Would you please suggest me how should I initialize a run loop source you mentioned? Moreover, I have no idea about an infill of its context (I mean CFRunLoopSourceContext structure). Thanks again. Best regards, Vladimir Magaziy. _______________________________________________ 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: Notification handling without run loop of a primary thread
by Brian Bergstrand 02 Dec '05

02 Dec '05

site_archiver(a)lists.apple.com Delivered-To: Darwin-dev(a)lists.apple.com -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 On Dec 2, 2005, at 3:08 PM, Vladimir Magaziy wrote: Hi all, HTH. Brian Bergstrand <http://www.bergstrand.org/brian/> PGP Key ID: 0xB6C7B6A2 -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (Darwin) iD8DBQFDkL6qedHYW7bHtqIRAkNvAJ9vdNCt63GsLk0eTNzReiXsWXeIrgCgygkb wup2rTUyHJZTcVkF2xMkyXA= =p17F -----END PGP SIGNATURE----- _______________________________________________ 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 trying to handle distributed notifications in a CoreFoundation Tool that does not run a run loop of its primary thread (actually, it is the requirement). I can not find any solution that works, so I ask you about possibility to do in general. I thought that it is possible to track notifications using a run loop of a separate thread (see code snippet below), but CFRunLoopRun function does not run a thread run loop, it simply returns a control and a thread finishes. I guess that I do not initialize a thread run loop correctly, so please suggest me the appropriate way to run it and force to handle notifications. The run loop needs an event source or it will just exit (as you witnessed). So, before calling CFRunLoopRun() you should register your notification handler with the run loop. This email sent to site_archiver(a)lists.apple.com

1 0