When you compile code with -fobjc-exceptions, you presumably do that to somewhere execute a throw. Lets say, this is my callstack:

abort	// glibc
objc_exception_throw	// libobjc
-[MyClass methodThrowingAnException]	// myclass.so
myClassMethodBouncer	// myclass.so
function_with_a_callback	// thirdparty.so
-[MyClass waitingForAnException]	// myclass.so
main	// main

What has happened ? I compiled all the MyClass code with -fobjc-exceptions and even did not forget to link the resulting shared library with --eh-frame-header. So all my code and libobjc.so has PT_GNU_EH_FRAME information, which is necessary for the "modern" C++ like stack unwinding.

Still not good enough. The third party library code function_with_a_callback, was not linked with --eh-frame-header and the unwinding stops right there. No catch handler can be found and objc_exception_throw treats this as an uncaught exception.

An example for the rule, that with every increase of complexity, there is also an increase in brittleness. ]]> http://www.mulle-kybernetik.com/weblog/2012/01/using_--eh-frame-header_when_u.html http://www.mulle-kybernetik.com/weblog/2012/01/using_--eh-frame-header_when_u.html Fri, 06 Jan 2012 17:41:37 +0100 (27 -> 540) 0.002849s 18859 : 0 19009 : 0 19197 : 0 19306 : 0 19438 : 0 19444 : 0 19444 : 0 19600 : 0 19650 : 0 19765 : 0 19874 : 0 19903 : 0 19912 : 0 19941 : 0 19965 : 0 20441 : 0 20547 : 0 20617 : 0 20629 : 0 20880 : 0 21409 : 0 21862 : 0 23376 : 0 First thing to notice, the program only runs for 0.0028s. That means the frequency of context switches is much higher on CPU 0. 100 times higher in this case. The duration of each context switch is also longer, lets say the average is 20000 cycles making that about 7 times longer.]]> http://www.mulle-kybernetik.com/weblog/2011/12/measuring_context_switches_a_s_3.html http://www.mulle-kybernetik.com/weblog/2011/12/measuring_context_switches_a_s_3.html Fri, 02 Dec 2011 01:33:09 +0100 (24 -> 480) 0.229874s 3356 : 18 3365 : 18 3376 : 18 3382 : 18 3388 : 18 3397 : 18 3403 : 18 3409 : 18 3412 : 18 3412 : 18 3415 : 18 3421 : 18 3429 : 18 3429 : 18 3432 : 18 3456 : 18 3479 : 18 3585 : 18 3647 : 18 3691 : 18 4471 : 18 4756 : 18 8265 : 18 The program took about 0.23s to run, in that time it received 23 context switches. That means a context switch happened every 0.1s i.e. 10Hz. Thats not suprisingly the documented preemption time slice for Mac OS X.

A context switch on CPU 18 took on average about 3.400 cycles. One may assume at these times, that the CPU is otherwise idle.

10 times 3.400 cycles is the approximate number of cycles lost per second: 34.000 cycles. Since my CPU is running at about 3.4GHz with Turboboost, that means - in relation to 3.400.000.000 cycles - context switching lost me about 0.001 % performance.

Thinking about this opens up some interesting questions, but lets gather another CPUs results to get some more data... Next time.]]> http://www.mulle-kybernetik.com/weblog/2011/12/measuring_context_switches_a_s_2.html http://www.mulle-kybernetik.com/weblog/2011/12/measuring_context_switches_a_s_2.html Thu, 01 Dec 2011 23:06:25 +0100 #include <assert.h> #include <limits.h> #include <stdio.h> #include <stdlib.h> #include <sys/time.h> #include <time.h> extern unsigned long long _rdtsc(); extern unsigned int _apicID(); typedef struct { unsigned int gap; unsigned int apicID; } context_switch_record; static int compare_context_switch_record( context_switch_record *a, context_switch_record *b) { return( (int) a->gap - (int) b->gap); } int main( int argc, char *argv[]) { context_switch_record infos[ 23]; unsigned long long a, b; unsigned long long normal; unsigned long long diff; unsigned int i; struct timeval start; struct timeval stop; struct timeval elapsed; gettimeofday( &start, NULL); // // establish a "gap" between two rdtsc() calls, that // should be considered as normal (no interrupt) // a = _rdtsc(); b = _rdtsc(); normal = (b - a) * 20; printf( "(%qd -> %qd) ", b - a, normal); for( i = 0; i < sizeof( infos) / sizeof( context_switch_record); i++) { b = _rdtsc(); for(;;) { a = b; b = _rdtsc(); diff = b - a; // // if the delay between a and b is to large // then record it // if( diff > normal) break; } assert( diff <= INT_MAX); // // record cycles spent and current cpuID // infos[ i].gap = diff; infos[ i].apicID = _apicID(); } gettimeofday( &stop, NULL); // // compute time spent waiting for n context switches // elapsed.tv_sec = stop.tv_sec - start.tv_sec; elapsed.tv_usec = stop.tv_usec - start.tv_usec; if( stop.tv_usec < start.tv_usec) { elapsed.tv_sec -= 1; elapsed.tv_usec = 1000000L - start.tv_usec + stop.tv_usec; } printf( "%d.%06lds\n\n", (int) elapsed.tv_sec, (long) elapsed.tv_usec); // // output data and done // qsort( &infos, sizeof( infos) / sizeof( context_switch_record), sizeof( context_switch_record), (void *) compare_context_switch_record); for( i = 0; i < sizeof( infos) / sizeof( context_switch_record); i++) printf( "%u : %u\n", infos[ i].gap, infos[ i].apicID); return( 0); } Here is the Xcode Project.]]> http://www.mulle-kybernetik.com/weblog/2011/12/measuring_context_switches_a_s_1.html http://www.mulle-kybernetik.com/weblog/2011/12/measuring_context_switches_a_s_1.html Thu, 01 Dec 2011 01:44:20 +0100 Intel Application Note 485 for more details)

   .text
   .globl __apicID
__apicID:
   push     %ebx

   movl     $0x1,%eax
   cpuid

   mov      %ebx, %eax
   shr      $24,%eax

   popl     %ebx
   
   ret

	.text
.globl __rdtsc
__rdtsc:
	rdtsc
	ret

It's one of a few experimental projects I put on bitbucket.org. Eventually I will probably migrate it back to Mulle kybernetiK, because the added "social" doesn't really do anything for me. ]]> http://www.mulle-kybernetik.com/weblog/2011/11/unitkit_fixes_updated_on_bitbu.html http://www.mulle-kybernetik.com/weblog/2011/11/unitkit_fixes_updated_on_bitbu.html Thu, 10 Nov 2011 14:53:35 +0100 xterm with an IP address I need to say:

xhost +
xterm -display 127.0.0.1:1 # or some other IP address

Preparation

prepare

sudo mkdir -p /usr/local/src/darwin-10.6.8 sudo chmod 777 /usr/local/src/darwin-10.6.8

Download and Unpack

• Libc-594.9.5.z
• dyld-132.13
• libauto-141.2
• libclosure-38
• libdispatch-84.5.5
• objc4-437.3.
• xnu-1504.15.3

You can also manually download stuff from http://www.opensource.apple.com/release/mac-os-x-1068/.

Installation of Headers

install

sudo mkdir -p /usr/local/include/darwin-10.6.8 sudo chmod 777 /usr/local/include/darwin-10.6.8 cd /usr/local/include/darwin-10.6.8 mkdir libkern mkdir System mkdir System/i386 mkdir System/ppc mkdir System/machine mkdir mach-o cd /usr/local/src/darwin-10.6.8 # find . -name "OSCrossEndian.h" # find . -name "pthread_machdep.h" # find . -name "dyld_priv.h" # find . -name "cpu_capabilities.h" # find . -name Block_private.h # find . -name auto_zone.h cp ./xnu-1504.15.3/libkern/libkern/OSCrossEndian.h /usr/local/include/darwin-10.6.8/libkern/ cp ./Libc-594.9.5/pthreads/pthread_machdep.h /usr/local/include/darwin-10.6.8/System/ cp ./dyld-132.13/include/mach-o/dyld_priv.h /usr/local/include/darwin-10.6.8/mach-o/ cp ./xnu-1504.15.3/osfmk/i386/cpu_capabilities.h /usr/local/include/darwin-10.6.8/System/i386/ cp ./xnu-1504.15.3/osfmk/ppc/cpu_capabilities.h /usr/local/include/darwin-10.6.8/System/ppc/ cp ./xnu-1504.15.3/osfmk/machine/cpu_capabilities.h /usr/local/include/darwin-10.6.8/System/machine/ cp ./libclosure-38/Block_private.h /usr/local/include/darwin-10.6.8/ cp ./libauto-141.2/auto_zone.h /usr/local/include/darwin-10.6.8/

Change the Xcode Project of objc

Open the project settings in Xcode and set:

• HEADER_SEARCH_PATHS=/usr/local/include/darwin-10.6.8 $(inherited) for all Configurations
• GCC_PREPROCESSOR_DEFINITIONS=$(inherited) PRIVATE for Debug and
• GCC_PREPROCESSOR_DEFINITIONS=$(inherited) NDEBUG PRIVATE for Release

• HEADER_SEARCH_PATHS = $(inherited) $(DSTROOT)/usr/include/** $(DSTROOT)/usr/local/include/** $(CONFIGURATION_BUILD_DIR)/usr/include/** $<(CONFIGURATION_BUILD_DIR)/usr/local/include/**

And now it should build and link without errors.

Go ahead and trash your system with the debug linked version :) ]]> http://www.mulle-kybernetik.com/weblog/2011/10/how_to_build_libobjc_for_os_x.html http://www.mulle-kybernetik.com/weblog/2011/10/how_to_build_libobjc_for_os_x.html Sat, 15 Oct 2011 13:50:33 +0100 484 gnutar cfz MyProject.tgz MyProject .... 493 rm -rf MyProject 494 gnutar xfz MyProject.tgz 495 cd MyProject 496 open MyProject.xcodeproj 497 cd .. then I typed

$ rm -rf MyProject ; !gn

rm -rf MyProject ; gnutar cfz MyProject.tgz MyProject

$ gnutar xfz MyProject.tgz MyProject
gnutar: MyProject: Not found in archive
gnutar: Error exit delayed from previous errors

In a way it's a little old fashioned text adventure with 3D graphics. But because you do not get any points, you don't know when you have exhausted the game or not.

Here is a little map that helps. It's not very pretty. Be aware that in the beginning and during parts of the game, you have the choice of two rabbits and so you can actually walk two paths in parallel for a while. ]]> http://www.mulle-kybernetik.com/weblog/2011/09/a_map_for_ruins_a_computer_gam.html http://www.mulle-kybernetik.com/weblog/2011/09/a_map_for_ruins_a_computer_gam.html Tue, 27 Sep 2011 22:13:21 +0100 script.sh

#!/bin/sh 
#
# Mulle kybernetiK 2011, coded by Nat!

cat <<EOF > /tmp/x.m  

@interface X 
{
}

+ (void *) z;

@end


@interface Y : X
{
}

+ (void *) y;

@end


@implementation Y

+ (void *) y
{
   return( [self z]);
}

@end
EOF

gcc -Wselector -c /tmp/x.m 

it will produce the following output:

x.m:25: warning: creating selector for nonexistent method 'z'

I go to an internet cafe, where they have a computer with USB and internet, which is good. I can download putty (SSH), though forbidden by the cafe rules, from there and run it, which is good. From there I could access the server and get the phone number. But now comes the technology idiocy. Can I use it ?

A machine in an internet cafe is certainly compromised with a key logger, so whatever account I access, is gonna get cracked. A solution would be to change the password later from a safe terminal, but that leaves a window of compromisabilty.

I figure out, that using hushmail is probably the best bet to send out an email for help. (I used to use googlemail for that, before they became all nosey and evil). So I create a new email account, but email isn't quite as often checked by my office-mate. SMS would be better.

I could use a free SMS service, but that probably puts him on a SPAM list and I don't have the number (yet) anyway. Seems the best option is to wait it out, while my mobile phone is SLOWLY charging. The problem is, that while its in its low power status, it won't turn on at all.

Two hours later of idle documentation reading, my office mate and dog appear in the internet cafe with my keys from the office. I pay 2 € for internet access fees, for two hours. Which kind of makes me wonder how a business like that survives. It's even cheaper than in 3rd world countries.

Later: another solution could have been to buy/rent/lend another phone and switch SIM cards. But I figured the internet cafe thing would be easier and quicker.

]]> http://www.mulle-kybernetik.com/weblog/2011/08/the_idiocy_of_technology_and_m.html http://www.mulle-kybernetik.com/weblog/2011/08/the_idiocy_of_technology_and_m.html Mon, 15 Aug 2011 14:54:53 +0100 ]]> http://www.mulle-kybernetik.com/weblog/2011/07/lion_ohne_worte.html http://www.mulle-kybernetik.com/weblog/2011/07/lion_ohne_worte.html Tue, 26 Jul 2011 22:15:51 +0100 automatic versioning of documents in Lion. Eventually it's gonna be great (I hope) but currently it has this half-baked feel to it.

1. I have some of my projects in my VMware Lion mounted as a shared folder. For some reason, Xcode had the worst time getting these to open. So it is seemingly stuck forever opening these projects, so I have to "kill" Xcode.

   When I restart Xcode, it tries to read these projects again, being stuck again... There is no way to break this vicious circle. What is needed is an option to launch with clearing the applications memory.

   This helplessness is an amazing feeling I haven't had on OS X since quite a few years actually.

2. I use TextEdit to write shell scripts. I often refine them in the shell with vi for small changes, but for larger edits I use TextEdit, which has the file still open.

   This gave me a slew of errors: Jul 22 01:25:30 Lion TextEdit[770]: open on /Users/test/Desktop/patcher.sh: Operation not permitted Jul 22 01:25:30 Lion [0x0-0x53053].com.apple.TextEdit[770]: [ERROR] genstore_storage.c:_CopyfileAddWithArchiveAndSizingCallback:665 copyfile callback got an error what=1 stage=3 src=/Users/test/Desktop/patcher.sh dst=/Users/test/Library/Containers/com.apple.TextEdit/Data/Library/Autosave Information/66455544-1790-463C-BDD9-0E408B2A2E83.genstore.noindex/.genstore_staging/5RyonsL/staged, errno 1 (Operation not permitted) Jul 22 01:25:30 Lion [0x0-0x53053].com.apple.TextEdit[770]: [ERROR] genstore_storage.c:GSAddPathAsGeneration:980 copyfile "/Users/test/Desktop/patcher.sh" to "/Users/test/Library/Containers/com.apple.TextEdit/Data/Library/Autosave Information/66455544-1790-463C-BDD9-0E408B2A2E83.genstore.noindex/.genstore_staging/5RyonsL/staged" failed with error 1 (Operation not permitted) Jul 22 01:25:30 Lion TextEdit[770]: NSFileVersion tried to tried to add a new generation and failed. Versioned file URL: file://localhost/Users/test/Desktop/patcher.sh, contents URL: file://localhost/Users/test/Desktop/patcher.sh, error: Error Domain=GSLibraryErrorDomain Code=1 "Der Vorgang konnte nicht abgeschlossen werden. (GSLibraryErrorDomain-Fehler 1.)" UserInfo=0x7f8038624e30 {} Jul 22 01:25:30 Lion TextEdit[770]: NSDocument failed to preserve the old version of a document. Here's the error: Error Domain=GSLibraryErrorDomain Code=1 "Der Vorgang konnte nicht abgeschlossen werden. (GSLibraryErrorDomain-Fehler 1.)" UserInfo=0x7f8038624e30 {}

   Also in TextEdit there is no "Save as" or "Save to" option anymore... In the words of John McEnroe: "You can not be serious". This right there makes TextEdit now unusable to me. ]]> http://www.mulle-kybernetik.com/weblog/2011/07/textedit_and_lion_file_version.html http://www.mulle-kybernetik.com/weblog/2011/07/textedit_and_lion_file_version.html Fri, 22 Jul 2011 12:37:34 +0100