Xcode 1.5 and zerolink resulting in EXC_BAD_ACCESS

• Christopher Hunt Xcode 1.5 and zerolink resulting in EXC_BAD_ACCESS Aug 25 2004, 11:15
  Hi there,
  
  I have a program using Objective C and C++ with multiple pthreads. I'm
  often getting an EXC_BAD_ACCESS when launching via the debugger. My
  session tells me that I've got a number of threads started and that
  their traces go something like this:
  
  Thread 1:
  #0    0x8fe26b08 in __dyld_syscall_thread_switch
  #1    0x8fe16004 in __dyld_yield
  #2    0x8fe15d60 in __dyld_set_lock
  #3    0x8fe0a39c in __dyld_check_and_report_undefineds
  #4    0x8fe11690 in __dyld_link_in_need_modules
  #5    0x8fe1124c in __dyld_bind_lazy_symbol_reference
  #6    0x8fe01340 in __dyld_stub_binding_helper_interface
  
  Thread 4:
  #0    0x8fe26b08 in __dyld_syscall_thread_switch
  #1    0x8fe16004 in __dyld_yield
  #2    0x8fe15d60 in __dyld_set_lock
  #3    0x8fe13278 in __dyld__dyld_link_module
  #4    0x9003f4a8 in NSLinkModule
  #5    0x7fdf0c00 in findSymbolInSet
  #6    0x7fdf0c84 in undefinedHandler
  #7    0x8fe0a398 in __dyld_check_and_report_undefineds
  #8    0x8fe11690 in __dyld_link_in_need_modules
  #9    0x8fe1124c in __dyld_bind_lazy_symbol_reference
  #10    0x8fe01340 in __dyld_stub_binding_helper_interface
  
  Thread 5:
  #0    0x8fe0cc10 in __dyld_link_library_module
  #1    0x8fe0c6ac in __dyld_resolve_undefineds
  #2    0x8fe115b8 in __dyld_link_in_need_modules
  #3    0x8fe1124c in __dyld_bind_lazy_symbol_reference
  #4    0x8fe01340 in __dyld_stub_binding_helper_interface
  
  If I turn the zerolink option off then I don't get the EXC_BAD_ACCESS.
  If I simply run the application (i.e. without the debugger) all appears
  well, but I suspect that this could just be a race condition style
  situation.
  
  The above stack traces are not the same every time... and sometimes I
  even get a SIGABRT instead...
  
  Any ideas on why I'm getting a EXC_BAD_ACCESS when debugging with
  multiple threads running and zerolink? Is zerolink something to be
  avoided with multiple threads?
  
  Cheers,
  -C
  
  _______________________________________________
  MacOSX-dev mailing list
  <MacOSX-dev...>
  http://www.omnigroup.com/mailman/listinfo/macosx-dev
• David Dunham Re: Xcode 1.5 and zerolink resulting in EXC_BAD_ACCESS Aug 25 2004, 22:42
  On 25 Aug 2004, at 02:15, Christopher Hunt wrote:
  
  

  > Any ideas on why I'm getting a EXC_BAD_ACCESS when debugging with
  > multiple threads running and zerolink? Is zerolink something to be
  > avoided with multiple threads?
  

  
  I've done all my thread coding with CodeWarrior and normal linking, but
  my first guess is that you have a threading error. All ZeroLink may do
  is change the timing of things and expose the bug faster.
  
  David Dunham        Macintosh Game Developer        GameHouse Studios
  <david...>    206 442 5881 x22    http://www.gamehouse.com
  
  _______________________________________________
  MacOSX-dev mailing list
  <MacOSX-dev...>
  http://www.omnigroup.com/mailman/listinfo/macosx-dev
• Christopher Hunt Re: Xcode 1.5 and zerolink resulting in EXC_BAD_ACCESS Aug 26 2004, 11:06
  On 26/08/2004, at 5:36 PM, <macosx-dev-request...> wrote:
  
  

  > Any ideas on why I'm getting a EXC_BAD_ACCESS when debugging with
  >> multiple threads running and zerolink? Is zerolink something to be
  >> avoided with multiple threads?
  >
  > I've done all my thread coding with CodeWarrior and normal linking, but
  > my first guess is that you have a threading error. All ZeroLink may do
  > is change the timing of things and expose the bug faster.
  

  Actually having visited the Xcode users forum, I noticed that libstdc++
  isn't compatible with zerolink. Apparently there are many things
  declared as _private_extern_ which is something zerolink doesn't
  tolerate.
  
  So, the rule seems to be, if you're using C++ don't use zerolink.
  
  Cheers,
  -C
  
  _______________________________________________
  MacOSX-dev mailing list
  <MacOSX-dev...>
  http://www.omnigroup.com/mailman/listinfo/macosx-dev