Open Radar

 
14-Nov-2012 08:02 PM John Hawkinson:
Summary:
  CrashReporter and gdb disagree on thread numbering (0-based vs. 1-based).
  CrashReporter numbers threads starting from 0.
  gdb numbers threads starting from 1.

  This is medium-annoying if you are trying to discuss a particular thread in
a core dump with another developer -- are you talking about gdb's thread numbers or about CrashReporter's?

  They should be consistent! Fix one. I would imagine gdb is more standard here, and platform-independent.

Steps to Reproduce:

  Crash an app. Compare CrashReporter and gdb output.

Expected Results:

   Consistent numbering.

Actual Results:

CrashReporter reports this for an Adobe InDesign crash:

    Exception Type:  EXC_BAD_ACCESS (SIGBUS)
    Exception Codes: KERN_PROTECTION_FAILURE at 0x0000000000000000
    Crashed Thread:  74
    
    Thread 0:  Dispatch queue: com.apple.main-thread
    ...
    Thread 74 Crashed:
    0   com.adobe.InDesign.Text             0x2207a68f GetPlugIn + 945295
    1   com.adobe.InDesign.Text             0x2207a94c GetPlugIn + 945996
    ...
    Thread 77:
    ...
    Thread 74 crashed with X86 Thread State (32-bit):

And in gdb:

  (gdb) info threads
    78 0x923d7412 in __workq_kernreturn ()
    77 0x923dfaa2 in __semwait_signal ()
    76 0x923b1b42 in semaphore_wait_signal_trap ()
    75 ComposeScannerCache::CalculateParagraph (this=0x1cb19468, position=27) at ComposeScanner.cpp:765
    74 0x923b1b5a in semaphore_timedwait_signal_trap ()
    73 0x923b1b5a in semaphore_timedwait_signal_trap ()
  ...
     3 0x923b1b5a in semaphore_timedwait_signal_trap ()
     2 0x923d8382 in kevent ()
  *  1 0x923b9962 in tiny_free_list_add_ptr ()
  Current language:  auto; currently c++
  (gdb)  


That is, crashreporter's Thread 74 is gdb's Thread 75.

Regression:

Notes: