]> git.saurik.com Git - apple/xnu.git/blobdiff - tools/lldbmacros/scheduler.py
xnu-7195.50.7.100.1.tar.gz
[apple/xnu.git] / tools / lldbmacros / scheduler.py
old mode 100644 (file)
new mode 100755 (executable)
index cf6baa3..a521b0d
@@ -1,29 +1,58 @@
 from xnu import *
 from utils import *
 from process import *
+from misc import *
+from memory import *
+from ipc import *
 
 # TODO: write scheduler related macros here
 
+# Macro: showallprocrunqcount
+
+@lldb_command('showallprocrunqcount')
+def ShowAllProcRunQCount(cmd_args=None):
+    """ Prints out the runq count for all processors
+    """
+    out_str = "Processor\t# Runnable\n"
+    processor_itr = kern.globals.processor_list
+    while processor_itr:
+        out_str += "{:d}\t\t{:d}\n".format(processor_itr.cpu_id, processor_itr.runq.count)
+        processor_itr = processor_itr.processor_list
+    # out_str += "RT:\t\t{:d}\n".format(kern.globals.rt_runq.count)
+    print out_str
+
+# EndMacro: showallprocrunqcount
+
 # Macro: showinterrupts
+
 @lldb_command('showinterrupts')
 def ShowInterrupts(cmd_args=None):
     """ Prints IRQ, IPI and TMR counts for each CPU
     """
-    bcdata = kern.GetValueFromAddress(kern.GetLoadAddressForSymbol('BootCpuData'), 'cpu_data_t *')
-    print "CPU 0 IRQ: {:d}\n".format(bcdata.cpu_stat.irq_ex_cnt)
-    print "CPU 0 IPI: {:d}\n".format(bcdata.cpu_stat.ipi_cnt)
-    print "CPU 0 TMR: {:d}\n".format(bcdata.cpu_stat.timer_cnt)
-    if (kern.globals.machine_info.physical_cpu == 2):
-        if kern.arch == 'arm':
-            cdentries = kern.GetValueFromAddress(kern.GetLoadAddressForSymbol('CpuDataEntries') + 20, 'uintptr_t *')
-            cpu_data_entry = Cast(dereference(cdentries), 'cpu_data_t *')
-            print "CPU 1 IRQ: {:d}\n".format(cpu_data_entry.cpu_stat.irq_ex_cnt)
-            print "CPU 1 IPI: {:d}\n".format(cpu_data_entry.cpu_stat.ipi_cnt)
-            print "CPU 1 TMR: {:d}\n".format(cpu_data_entry.cpu_stat.timer_cnt)
+
+    if not kern.arch.startswith('arm'):
+        print "showinterrupts is only supported on arm/arm64"
+        return
+
+    base_address = kern.GetLoadAddressForSymbol('CpuDataEntries')
+    struct_size = 16
+    x = 0
+    y = 0
+    while x < unsigned(kern.globals.machine_info.physical_cpu):
+        element = kern.GetValueFromAddress(base_address + (y * struct_size), 'uintptr_t *')[1]
+        if element:
+            cpu_data_entry = Cast(element, 'cpu_data_t *')
+            print "CPU {} IRQ: {:d}\n".format(y, cpu_data_entry.cpu_stat.irq_ex_cnt)
+            print "CPU {} IPI: {:d}\n".format(y, cpu_data_entry.cpu_stat.ipi_cnt)
+            print "CPU {} PMI: {:d}\n".format(y, cpu_data_entry.cpu_monotonic.mtc_npmis)
+            print "CPU {} TMR: {:d}\n".format(y, cpu_data_entry.cpu_stat.timer_cnt)
+            x = x + 1
+        y = y + 1
 
 # EndMacro: showinterrupts
 
 # Macro: showactiveinterrupts
+
 @lldb_command('showactiveinterrupts')
 def ShowActiveInterrupts(cmd_args=None):
     """  Prints the interrupts that are unmasked & active with the Interrupt Controller
@@ -58,6 +87,1232 @@ def ShowActiveInterrupts(cmd_args=None):
             active = dereference(kern.GetValueFromAddress((current_pointer + aic_him_offset) + (4 * group_count), 'uintptr_t *'))
         else:
             mask = mask << 1
-    
 # EndMacro: showactiveinterrupts
 
+# Macro: showirqbyipitimerratio
+@lldb_command('showirqbyipitimerratio')
+def ShowIrqByIpiTimerRatio(cmd_args=None):
+    """ Prints the ratio of IRQ by sum of IPI & TMR counts for each CPU
+    """
+    if kern.arch == "x86_64":
+        print "This macro is not supported on x86_64 architecture"
+        return
+
+    out_str = "IRQ-IT Ratio: "
+    base_address = kern.GetLoadAddressForSymbol('CpuDataEntries')
+    struct_size = 16
+    x = 0
+    y = 0
+    while x < unsigned(kern.globals.machine_info.physical_cpu):
+        element  = kern.GetValueFromAddress(base_address + (y * struct_size), 'uintptr_t *')[1]
+        if element:
+            cpu_data_entry = Cast(element, 'cpu_data_t *')
+            out_str += "   CPU {} [{:.2f}]".format(y, float(cpu_data_entry.cpu_stat.irq_ex_cnt)/(cpu_data_entry.cpu_stat.ipi_cnt + cpu_data_entry.cpu_stat.timer_cnt))
+            x = x + 1
+        y = y + 1
+    print out_str
+
+# EndMacro: showirqbyipitimerratio
+
+#Macro: showinterruptsourceinfo
+@lldb_command('showinterruptsourceinfo')
+def showinterruptsourceinfo(cmd_args = None):
+    """  Extract information of interrupt source causing interrupt storms.
+    """
+    if not cmd_args:
+        print "No arguments passed"
+        return False
+    #Dump IOInterruptVector object
+    print "--- Dumping IOInterruptVector object ---\n"
+    object_info = lldb_run_command("dumpobject {:s} IOInterruptVector".format(cmd_args[0]))
+    print object_info
+    print "--- Dumping IOFilterInterruptEventSource object ---\n"
+    #Dump the IOFilterInterruptEventSource object.
+    target_info=re.search('target =\s+(.*)',object_info)
+    target= target_info.group()
+    target= target.split()
+    #Dump the Object pointer of the source who is triggering the Interrupts.
+    vector_info=lldb_run_command("dumpobject {:s} ".format(target[2]))
+    print vector_info
+    owner_info= re.search('owner =\s+(.*)',vector_info)
+    owner= owner_info.group()
+    owner= owner.split()
+    print "\n\n"
+    out=lldb_run_command(" dumpobject {:s}".format(owner[2]))
+    print out
+
+# EndMacro: showinterruptsourceinfo
+
+@lldb_command('showcurrentabstime')
+def ShowCurremtAbsTime(cmd_args=None):
+    """  Routine to print latest absolute time known to system before being stopped.
+         Usage: showcurrentabstime
+    """
+    pset = addressof(kern.globals.pset0)
+    processor_array = kern.globals.processor_array
+    cur_abstime = 0
+
+    while unsigned(pset) != 0:
+        cpu_bitmap = int(pset.cpu_bitmask)
+        for cpuid in IterateBitmap(cpu_bitmap):
+            processor = processor_array[cpuid]
+            if unsigned(processor.last_dispatch) > cur_abstime:
+                cur_abstime = unsigned(processor.last_dispatch)
+
+        pset = pset.pset_list
+
+    print "Last dispatch time known: %d MATUs" % cur_abstime
+
+bucketStr = ["FIXPRI (>UI)", "TIMESHARE_FG", "TIMESHARE_IN", "TIMESHARE_DF", "TIMESHARE_UT", "TIMESHARE_BG"]
+
+@header("{:<18s} | {:>20s} | {:>20s} | {:>10s} | {:>10s}".format('Thread Group', 'Pending (us)', 'Interactivity Score', 'TG Boost', 'Highest Thread Pri'))
+def GetSchedClutchBucketSummary(clutch_bucket):
+    tg_boost = kern.globals.sched_clutch_bucket_group_pri_boost[clutch_bucket.scb_group.scbg_clutch.sc_tg_priority]
+    pending_delta = kern.GetNanotimeFromAbstime(GetRecentTimestamp() - clutch_bucket.scb_group.scbg_pending_data.scct_timestamp) / 1000
+    if (int)(clutch_bucket.scb_group.scbg_pending_data.scct_timestamp) == 18446744073709551615:
+        pending_delta = 0
+    return "0x{:<16x} | {:>20d} | {:>20d} | {:>10d} | {:>10d}".format(clutch_bucket.scb_group.scbg_clutch.sc_tg, pending_delta, clutch_bucket.scb_group.scbg_interactivity_data.scct_count, tg_boost, SchedPriorityStableQueueRootPri(clutch_bucket.scb_thread_runq, 'struct thread', 'th_clutch_runq_link')) 
+
+def ShowSchedClutchForPset(pset):
+    root_clutch = pset.pset_clutch_root
+    print "\n{:s} : {:d}\n\n".format("Current Timestamp", GetRecentTimestamp())
+    print "{:>10s} | {:>20s} | {:>30s} | {:>25s} | {:<18s} | {:>10s} | {:>10s} | {:>15s} | ".format("Root", "Root Buckets", "Clutch Buckets", "Threads", "Address", "Pri (Base)", "Count", "Deadline (us)") + GetSchedClutchBucketSummary.header
+    print "=" * 300
+    print "{:>10s} | {:>20s} | {:>30s} | {:>25s} | 0x{:<16x} | {:>10d} | {:>10d} | {:>15s} | ".format("Root", "*", "*", "*", addressof(root_clutch), (root_clutch.scr_priority if root_clutch.scr_thr_count > 0 else -1), root_clutch.scr_thr_count, "*")
+    print "-" * 300
+
+    for i in range(0, 6):
+        root_bucket = root_clutch.scr_unbound_buckets[i]
+        root_bucket_deadline = 0
+        if root_bucket.scrb_clutch_buckets.scbrq_count != 0 and i != 0:
+            root_bucket_deadline = kern.GetNanotimeFromAbstime(root_bucket.scrb_pqlink.deadline - GetRecentTimestamp()) / 1000
+        print "{:>10s} | {:>20s} | {:>30s} | {:>25s} | 0x{:<16x} | {:>10s} | {:>10s} | {:>15d} | ".format("*", bucketStr[int(root_bucket.scrb_bucket)], "*", "*", addressof(root_bucket), "*", "*", root_bucket_deadline)
+        clutch_bucket_runq = root_bucket.scrb_clutch_buckets
+        clutch_bucket_list = []
+        for pri in range(0,128):
+            clutch_bucket_circleq = clutch_bucket_runq.scbrq_queues[pri]
+            for clutch_bucket in IterateCircleQueue(clutch_bucket_circleq, 'struct sched_clutch_bucket', 'scb_runqlink'):
+                clutch_bucket_list.append(clutch_bucket)
+        if len(clutch_bucket_list) > 0:
+            clutch_bucket_list.sort(key=lambda x: x.scb_priority, reverse=True)
+            for clutch_bucket in clutch_bucket_list:
+                print "{:>10s} | {:>20s} | {:>30s} | {:>25s} | {:<18s} | {:>10s} | {:>10s} | {:>15s} | ".format("", "", "", "", "", "", "", "")
+                print "{:>10s} | {:>20s} | {:>30s} | {:>25s} | 0x{:<16x} | {:>10d} | {:>10d} | {:>15s} | ".format("*", "*", clutch_bucket.scb_group.scbg_clutch.sc_tg.tg_name, "*", clutch_bucket, clutch_bucket.scb_priority, clutch_bucket.scb_thr_count, "*") + GetSchedClutchBucketSummary(clutch_bucket)
+                runq = clutch_bucket.scb_clutchpri_prioq
+                for thread in IterateSchedPriorityQueue(runq, 'struct thread', 'th_clutch_pri_link'):
+                    thread_name = GetThreadName(thread)[-24:]
+                    if len(thread_name) == 0:
+                        thread_name = "<unnamed thread>"
+                    print "{:>10s} | {:>20s} | {:>30s} | {:<25s} | 0x{:<16x} | {:>10d} | {:>10s} | {:>15s} | ".format("*", "*", "*", thread_name, thread, thread.base_pri, "*", "*") 
+        print "-" * 300
+        root_bucket = root_clutch.scr_bound_buckets[i]
+        root_bucket_deadline = 0
+        if root_bucket.scrb_bound_thread_runq.count != 0:
+            root_bucket_deadline = kern.GetNanotimeFromAbstime(root_bucket.scrb_pqlink.deadline - GetRecentTimestamp()) / 1000
+        print "{:>10s} | {:>20s} | {:>30s} | {:>25s} | 0x{:<16x} | {:>10s} | {:>10d} | {:>15d} | ".format("*", bucketStr[int(root_bucket.scrb_bucket)] + " [Bound]", "*", "*", addressof(root_bucket), "*", root_bucket.scrb_bound_thread_runq.count, root_bucket_deadline)
+        if root_bucket.scrb_bound_thread_runq.count == 0:
+            print "-" * 300
+            continue
+        thread_runq = root_bucket.scrb_bound_thread_runq
+        for pri in range(0, 128):
+            thread_circleq = thread_runq.queues[pri]
+            for thread in IterateCircleQueue(thread_circleq, 'struct thread', 'runq_links'):
+                thread_name = GetThreadName(thread)[-24:]
+                if len(thread_name) == 0:
+                    thread_name = "<unnamed thread>"
+                print "{:>10s} | {:>20s} | {:>30s} | {:<25s} | 0x{:<16x} | {:>10d} | {:>10s} | {:>15s} | ".format("*", "*", "*", thread_name, thread, thread.base_pri, "*", "*")
+        print "-" * 300
+
+@lldb_command('showschedclutch')
+def ShowSchedClutch(cmd_args=[]):
+    """ Routine to print the clutch scheduler hierarchy.
+        Usage: showschedclutch <pset>
+    """
+    if not cmd_args:
+        raise ArgumentError("Invalid argument")
+    pset = kern.GetValueFromAddress(cmd_args[0], "processor_set_t")
+    ShowSchedClutchForPset(pset)
+
+@lldb_command('showschedclutchroot')
+def ShowSchedClutchRoot(cmd_args=[]):
+    """ show information about the root of the sched clutch hierarchy
+        Usage: showschedclutchroot <root>
+    """
+    if not cmd_args:
+        raise ArgumentError("Invalid argument")
+    root = kern.GetValueFromAddress(cmd_args[0], "struct sched_clutch_root *")
+    if not root:
+        print "unknown arguments:", str(cmd_args)
+        return False
+    print "{:>30s} : 0x{:<16x}".format("Root", root)
+    print "{:>30s} : 0x{:<16x}".format("Pset", root.scr_pset)
+    print "{:>30s} : {:d}".format("Priority", (root.scr_priority if root.scr_thr_count > 0 else -1))
+    print "{:>30s} : {:d}".format("Urgency", root.scr_urgency)
+    print "{:>30s} : {:d}".format("Threads", root.scr_thr_count)
+    print "{:>30s} : {:d}".format("Current Timestamp", GetRecentTimestamp())
+    print "{:>30s} : {:b} (BG/UT/DF/IN/FG/FIX/NULL)".format("Runnable Root Buckets Bitmap", int(root.scr_runnable_bitmap[0]))
+
+@lldb_command('showschedclutchrootbucket')
+def ShowSchedClutchRootBucket(cmd_args=[]):
+    """ show information about a root bucket in the sched clutch hierarchy
+        Usage: showschedclutchrootbucket <root_bucket>
+    """
+    if not cmd_args:
+        raise ArgumentError("Invalid argument")
+    root_bucket = kern.GetValueFromAddress(cmd_args[0], "struct sched_clutch_root_bucket *")
+    if not root_bucket:
+        print "unknown arguments:", str(cmd_args)
+        return False
+    print "{:<30s} : 0x{:<16x}".format("Root Bucket", root_bucket)
+    print "{:<30s} : {:s}".format("Bucket Name", bucketStr[int(root_bucket.scrb_bucket)])
+    print "{:<30s} : {:d}".format("Deadline", (root_bucket.scrb_pqlink.deadline if root_bucket.scrb_clutch_buckets.scbrq_count != 0 else 0))
+    print "{:<30s} : {:d}".format("Current Timestamp", GetRecentTimestamp())
+    print "\n"
+    clutch_bucket_runq = root_bucket.scrb_clutch_buckets
+    clutch_bucket_list = []
+    for pri in range(0,128):
+        clutch_bucket_circleq = clutch_bucket_runq.scbrq_queues[pri]
+        for clutch_bucket in IterateCircleQueue(clutch_bucket_circleq, 'struct sched_clutch_bucket', 'scb_runqlink'):
+            clutch_bucket_list.append(clutch_bucket)
+    if len(clutch_bucket_list) > 0:
+        print "=" * 240
+        print "{:>30s} | {:>18s} | {:>20s} | {:>20s} | ".format("Name", "Clutch Bucket", "Priority", "Count") + GetSchedClutchBucketSummary.header
+        print "=" * 240
+        clutch_bucket_list.sort(key=lambda x: x.scb_priority, reverse=True)
+        for clutch_bucket in clutch_bucket_list:
+            print "{:>30s} | 0x{:<16x} | {:>20d} | {:>20d} | ".format(clutch_bucket.scb_group.scbg_clutch.sc_tg.tg_name, clutch_bucket, clutch_bucket.scb_priority, clutch_bucket.scb_thr_count) + GetSchedClutchBucketSummary(clutch_bucket)
+
+def SchedClutchBucketDetails(clutch_bucket):
+    print "{:<30s} : 0x{:<16x}".format("Clutch Bucket", clutch_bucket)
+    print "{:<30s} : {:s}".format("Scheduling Bucket", bucketStr[(int)(clutch_bucket.scb_bucket)])
+    print "{:<30s} : 0x{:<16x}".format("Clutch Bucket Group", clutch_bucket.scb_group)
+    print "{:<30s} : {:s}".format("TG Name", clutch_bucket.scb_group.scbg_clutch.sc_tg.tg_name)
+    print "{:<30s} : {:d}".format("Priority", clutch_bucket.scb_priority)
+    print "{:<30s} : {:d}".format("Thread Count", clutch_bucket.scb_thr_count)
+    print "{:<30s} : 0x{:<16x}".format("Thread Group", clutch_bucket.scb_group.scbg_clutch.sc_tg)
+    print "{:<30s} : {:6d} (inherited from clutch bucket group)".format("Interactivity Score", clutch_bucket.scb_group.scbg_interactivity_data.scct_count)
+    print "{:<30s} : {:6d} (inherited from clutch bucket group)".format("Last Timeshare Update Tick", clutch_bucket.scb_group.scbg_timeshare_tick)
+    print "{:<30s} : {:6d} (inherited from clutch bucket group)".format("Priority Shift", clutch_bucket.scb_group.scbg_pri_shift) 
+    print "\n"
+    runq = clutch_bucket.scb_clutchpri_prioq
+    thread_list = []
+    for thread in IterateSchedPriorityQueue(runq, 'struct thread', 'th_clutch_pri_link'):
+        thread_list.append(thread)
+    if len(thread_list) > 0:
+        print "=" * 240
+        print GetThreadSummary.header + "{:s}".format("Process Name")
+        print "=" * 240
+        for thread in thread_list:
+            proc = Cast(thread.task.bsd_info, 'proc *')
+            print GetThreadSummary(thread) + "{:s}".format(GetProcName(proc))
+
+@lldb_command('showschedclutchbucket')
+def ShowSchedClutchBucket(cmd_args=[]):
+    """ show information about a clutch bucket in the sched clutch hierarchy
+        Usage: showschedclutchbucket <clutch_bucket>
+    """
+    if not cmd_args:
+        raise ArgumentError("Invalid argument")
+    clutch_bucket = kern.GetValueFromAddress(cmd_args[0], "struct sched_clutch_bucket *")
+    if not clutch_bucket:
+        print "unknown arguments:", str(cmd_args)
+        return False
+    SchedClutchBucketDetails(clutch_bucket)
+
+@lldb_command('abs2nano')
+def ShowAbstimeToNanoTime(cmd_args=[]):
+    """ convert mach_absolute_time units to nano seconds
+        Usage: (lldb) abs2nano <timestamp in MATUs>
+    """
+    if not cmd_args:
+        raise ArgumentError("Invalid argument")
+    timedata = ArgumentStringToInt(cmd_args[0])
+    ns = kern.GetNanotimeFromAbstime(timedata)
+    us = float(ns) / 1000 
+    ms = us / 1000 
+    s = ms / 1000 
+    
+    if s > 60 :
+        m = s / 60
+        h = m / 60
+        d = h / 24
+        
+        print "{:d} ns, {:f} us, {:f} ms, {:f} s, {:f} m, {:f} h, {:f} d".format(ns, us, ms, s, m, h, d)
+    else:
+        print "{:d} ns, {:f} us, {:f} ms, {:f} s".format(ns, us, ms, s)
+
+ # Macro: showschedhistory
+
+def GetRecentTimestamp():
+    """
+    Return a recent timestamp.
+    TODO: on x86, if not in the debugger, then look at the scheduler
+    """
+    if kern.arch == 'x86_64':
+        most_recent_dispatch = GetSchedMostRecentDispatch(False)
+        if most_recent_dispatch > kern.globals.debugger_entry_time :
+            return most_recent_dispatch
+        else :
+            return kern.globals.debugger_entry_time
+    else :
+        return GetSchedMostRecentDispatch(False)
+
+def GetSchedMostRecentDispatch(show_processor_details=False):
+    """ Return the most recent dispatch on the system, printing processor
+        details if argument is true.
+    """
+    processor_list = kern.globals.processor_list
+
+    most_recent_dispatch = 0
+    current_processor = processor_list
+
+    while unsigned(current_processor) > 0:
+        active_thread = current_processor.active_thread
+        if unsigned(active_thread) != 0 :
+            task_val = active_thread.task
+            proc_val = Cast(task_val.bsd_info, 'proc *')
+            proc_name = "<unknown>" if unsigned(proc_val) == 0 else GetProcName(proc_val)
+
+        last_dispatch = unsigned(current_processor.last_dispatch)
+
+        if kern.arch == 'x86_64':
+            cpu_data = kern.globals.cpu_data_ptr[current_processor.cpu_id]
+            if (cpu_data != 0) :
+                cpu_debugger_time = max(cpu_data.debugger_entry_time, cpu_data.debugger_ipi_time)
+            time_since_dispatch = unsigned(cpu_debugger_time - last_dispatch)
+            time_since_dispatch_us = kern.GetNanotimeFromAbstime(time_since_dispatch) / 1000.0
+            time_since_debugger = unsigned(cpu_debugger_time - kern.globals.debugger_entry_time)
+            time_since_debugger_us = kern.GetNanotimeFromAbstime(time_since_debugger) / 1000.0
+
+            if show_processor_details:
+                print "Processor last dispatch: {:16d} Entered debugger: {:16d} ({:8.3f} us after dispatch, {:8.3f} us after debugger) Active thread: 0x{t:<16x} 0x{t.thread_id:<8x} {proc_name:s}".format(last_dispatch, cpu_debugger_time,
+                        time_since_dispatch_us, time_since_debugger_us, t=active_thread, proc_name=proc_name)
+        else:
+            if show_processor_details:
+                print "Processor last dispatch: {:16d} Active thread: 0x{t:<16x} 0x{t.thread_id:<8x} {proc_name:s}".format(last_dispatch, t=active_thread, proc_name=proc_name)
+
+        if last_dispatch > most_recent_dispatch:
+            most_recent_dispatch = last_dispatch
+
+        current_processor = current_processor.processor_list
+
+    return most_recent_dispatch
+
+@header("{:<18s} {:<10s} {:>16s} {:>16s} {:>16s} {:>16s} {:>18s} {:>16s} {:>16s} {:>16s} {:>16s} {:2s} {:2s} {:2s} {:>2s} {:<19s} {:<9s} {:>10s} {:>10s} {:>10s} {:>10s} {:>10s} {:>11s} {:>8s}".format("thread", "id", "on-core", "off-core", "runnable", "prichange", "last-duration (us)", "since-off (us)", "since-on (us)", "pending (us)", "pri-change (us)", "BP", "SP", "TP", "MP", "sched-mode", "state", "cpu-usage", "delta", "sch-usage", "stamp", "shift", "task", "thread-name"))
+def ShowThreadSchedHistory(thread, most_recent_dispatch):
+    """ Given a thread and the most recent dispatch time of a thread on the
+        system, print out details about scheduler history for the thread.
+    """
+
+    thread_name = ""
+
+    if unsigned(thread.uthread) != 0:
+        uthread = Cast(thread.uthread, 'uthread *')
+        # Doing the straightforward thing blows up weirdly, so use some indirections to get back on track
+        if unsigned(uthread.pth_name) != 0 :
+            thread_name = str(kern.GetValueFromAddress(unsigned(uthread.pth_name), 'char*'))
+
+    task = thread.task
+    task_name = "unknown"
+    if task and unsigned(task.bsd_info):
+        p = Cast(task.bsd_info, 'proc *')
+        task_name = GetProcName(p)
+
+    sched_mode = ""
+
+    mode = str(thread.sched_mode)
+    if "TIMESHARE" in mode:
+        sched_mode+="timeshare"
+    elif "FIXED" in mode:
+        sched_mode+="fixed"
+    elif "REALTIME" in mode:
+        sched_mode+="realtime"
+
+    if (unsigned(thread.bound_processor) != 0):
+        sched_mode+="-bound"
+
+    # TH_SFLAG_THROTTLED
+    if (unsigned(thread.sched_flags) & 0x0004):
+        sched_mode+="-BG"
+
+    state = thread.state
+
+    thread_state_chars = {0x0:'', 0x1:'W', 0x2:'S', 0x4:'R', 0x8:'U', 0x10:'H', 0x20:'A', 0x40:'P', 0x80:'I'}
+    state_str = ''
+    mask = 0x1
+    while mask <= 0x80 :
+        state_str += thread_state_chars[int(state & mask)]
+        mask = mask << 1
+
+    last_on = thread.computation_epoch
+    last_off = thread.last_run_time
+    last_runnable = thread.last_made_runnable_time
+    last_prichange = thread.last_basepri_change_time
+
+    if int(last_runnable) == 18446744073709551615 :
+        last_runnable = 0
+
+    if int(last_prichange) == 18446744073709551615 :
+        last_prichange = 0
+
+    time_on_abs = unsigned(last_off - last_on)
+    time_on_us = kern.GetNanotimeFromAbstime(time_on_abs) / 1000.0
+
+    time_pending_abs = unsigned(most_recent_dispatch - last_runnable)
+    time_pending_us = kern.GetNanotimeFromAbstime(time_pending_abs) / 1000.0
+
+    if int(last_runnable) == 0 :
+        time_pending_us = 0
+
+    last_prichange_abs = unsigned(most_recent_dispatch - last_prichange)
+    last_prichange_us = kern.GetNanotimeFromAbstime(last_prichange_abs) / 1000.0
+
+    if int(last_prichange) == 0 :
+        last_prichange_us = 0
+
+    time_since_off_abs = unsigned(most_recent_dispatch - last_off)
+    time_since_off_us = kern.GetNanotimeFromAbstime(time_since_off_abs) / 1000.0
+    time_since_on_abs = unsigned(most_recent_dispatch - last_on)
+    time_since_on_us = kern.GetNanotimeFromAbstime(time_since_on_abs) / 1000.0
+
+    fmt  = "0x{t:<16x} 0x{t.thread_id:<8x} {t.computation_epoch:16d} {t.last_run_time:16d} {last_runnable:16d} {last_prichange:16d} {time_on_us:18.3f} {time_since_off_us:16.3f} {time_since_on_us:16.3f} {time_pending_us:16.3f} {last_prichange_us:16.3f}"
+    fmt2 = " {t.base_pri:2d} {t.sched_pri:2d} {t.task_priority:2d} {t.max_priority:2d} {sched_mode:19s}"
+    fmt3 = " {state:9s} {t.cpu_usage:10d} {t.cpu_delta:10d} {t.sched_usage:10d} {t.sched_stamp:10d} {t.pri_shift:10d} {name:s} {thread_name:s}"
+
+    out_str = fmt.format(t=thread, time_on_us=time_on_us, time_since_off_us=time_since_off_us, time_since_on_us=time_since_on_us, last_runnable=last_runnable, time_pending_us=time_pending_us, last_prichange=last_prichange, last_prichange_us=last_prichange_us)
+    out_str += fmt2.format(t=thread, sched_mode=sched_mode)
+    out_str += fmt3.format(t=thread, state=state_str, name=task_name, thread_name=thread_name)
+
+    print out_str
+
+def SortThreads(threads, column):
+        if column != 'on-core' and column != 'off-core' and column != 'last-duration':
+            raise ArgumentError("unsupported sort column")
+        if column == 'on-core':
+            threads.sort(key=lambda t: t.computation_epoch)
+        elif column == 'off-core':
+            threads.sort(key=lambda t: t.last_run_time)
+        else:
+            threads.sort(key=lambda t: t.last_run_time - t.computation_epoch)
+
+@lldb_command('showschedhistory', 'S:')
+def ShowSchedHistory(cmd_args=None, cmd_options=None):
+    """ Routine to print out thread scheduling history, optionally sorted by a
+        column.
+
+        Usage: showschedhistory [-S on-core|off-core|last-duration] [<thread-ptr> ...]
+    """
+
+    sort_column = None
+    if '-S' in cmd_options:
+        sort_column = cmd_options['-S']
+
+    if cmd_args:
+        most_recent_dispatch = GetSchedMostRecentDispatch(False)
+
+        print ShowThreadSchedHistory.header
+
+        if sort_column:
+            threads = []
+            for thread_ptr in cmd_args:
+                threads.append(kern.GetValueFromAddress(ArgumentStringToInt(thread_ptr), 'thread *'))
+
+            SortThreads(threads, sort_column)
+
+            for thread in threads:
+                ShowThreadSchedHistory(thread, most_recent_dispatch)
+        else:
+            for thread_ptr in cmd_args:
+                thread = kern.GetValueFromAddress(ArgumentStringToInt(thread_ptr), 'thread *')
+                ShowThreadSchedHistory(thread, most_recent_dispatch)
+
+        return
+
+    run_buckets = kern.globals.sched_run_buckets
+
+    run_count      = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_RUN')]
+    fixpri_count   = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_FIXPRI')]
+    share_fg_count = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_FG')]
+    share_df_count = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_DF')]
+    share_ut_count = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_UT')]
+    share_bg_count = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_BG')]
+
+    sched_pri_shifts = kern.globals.sched_run_buckets
+
+    share_fg_shift = sched_pri_shifts[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_FG')]
+    share_df_shift = sched_pri_shifts[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_DF')]
+    share_ut_shift = sched_pri_shifts[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_UT')]
+    share_bg_shift = sched_pri_shifts[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_BG')]
+
+
+    print "Processors: {g.processor_avail_count:d} Runnable threads: {:d} Fixpri threads: {:d}\n".format(run_count, fixpri_count, g=kern.globals)
+    print "FG Timeshare threads: {:d} DF Timeshare threads: {:d} UT Timeshare threads: {:d} BG Timeshare threads: {:d}\n".format(share_fg_count, share_df_count, share_ut_count, share_bg_count)
+    print "Mach factor: {g.sched_mach_factor:d} Load factor: {g.sched_load_average:d} Sched tick: {g.sched_tick:d} timestamp: {g.sched_tick_last_abstime:d} interval:{g.sched_tick_interval:d}\n".format(g=kern.globals)
+    print "Fixed shift: {g.sched_fixed_shift:d} FG shift: {:d} DF shift: {:d} UT shift: {:d} BG shift: {:d}\n".format(share_fg_shift, share_df_shift, share_ut_shift, share_bg_shift, g=kern.globals)
+    print "sched_pri_decay_band_limit: {g.sched_pri_decay_band_limit:d} sched_decay_usage_age_factor: {g.sched_decay_usage_age_factor:d}\n".format(g=kern.globals)
+
+    if kern.arch == 'x86_64':
+        print "debugger_entry_time: {g.debugger_entry_time:d}\n".format(g=kern.globals)
+
+    most_recent_dispatch = GetSchedMostRecentDispatch(True)
+    print "Most recent dispatch: " + str(most_recent_dispatch)
+
+    print ShowThreadSchedHistory.header
+
+    if sort_column:
+        threads = [t for t in IterateQueue(kern.globals.threads, 'thread *', 'threads')]
+
+        SortThreads(threads, sort_column)
+
+        for thread in threads:
+            ShowThreadSchedHistory(thread, most_recent_dispatch)
+    else:
+        for thread in IterateQueue(kern.globals.threads, 'thread *', 'threads'):
+            ShowThreadSchedHistory(thread, most_recent_dispatch)
+
+
+# EndMacro: showschedhistory
+
+def int32(n):
+    n = n & 0xffffffff
+    return (n ^ 0x80000000) - 0x80000000
+
+# Macro: showallprocessors
+
+def ShowGroupSetSummary(runq, task_map):
+    """ Internal function to print summary of group run queue
+        params: runq - value representing struct run_queue *
+    """
+
+    print "    runq: count {: <10d} highq: {: <10d} urgency {: <10d}\n".format(runq.count, int32(runq.highq), runq.urgency)
+
+    runq_queue_i = 0
+    runq_queue_count = sizeof(runq.queues)/sizeof(runq.queues[0])
+
+    for runq_queue_i in xrange(runq_queue_count) :
+        runq_queue_head = addressof(runq.queues[runq_queue_i])
+        runq_queue_p = runq_queue_head.next
+
+        if unsigned(runq_queue_p) != unsigned(runq_queue_head):
+            runq_queue_this_count = 0
+
+            for entry in ParanoidIterateLinkageChain(runq_queue_head, "sched_entry_t", "entry_links", circleQueue=True):
+                runq_queue_this_count += 1
+
+            print "      Queue [{: <#012x}] Priority {: <3d} count {:d}\n".format(runq_queue_head, runq_queue_i, runq_queue_this_count)
+            for entry in ParanoidIterateLinkageChain(runq_queue_head, "sched_entry_t", "entry_links", circleQueue=True):
+                group_addr = unsigned(entry) - (sizeof(dereference(entry)) * unsigned(entry.sched_pri))
+                group = kern.GetValueFromAddress(unsigned(group_addr), 'sched_group_t')
+                task = task_map.get(unsigned(group), 0x0)
+                if task == 0x0 :
+                    print "Cannot find task for group: {: <#012x}".format(group)
+                print "\tEntry [{: <#012x}] Priority {: <3d} Group {: <#012x} Task {: <#012x}\n".format(unsigned(entry), entry.sched_pri, unsigned(group), unsigned(task))
+
+@lldb_command('showrunq')
+def ShowRunq(cmd_args=None):
+    """  Routine to print information of a runq
+         Usage: showrunq <runq>
+    """
+
+    if not cmd_args:
+        print "No arguments passed"
+        print ShowRunq.__doc__
+        return False
+
+    runq = kern.GetValueFromAddress(cmd_args[0], 'struct run_queue *')
+    ShowRunQSummary(runq)
+
+def ShowRunQSummary(runq):
+    """ Internal function to print summary of run_queue
+        params: runq - value representing struct run_queue *
+    """
+
+    print "    runq: count {: <10d} highq: {: <10d} urgency {: <10d}\n".format(runq.count, int32(runq.highq), runq.urgency)
+
+    runq_queue_i = 0
+    runq_queue_count = sizeof(runq.queues)/sizeof(runq.queues[0])
+
+    for runq_queue_i in xrange(runq_queue_count) :
+        runq_queue_head = addressof(runq.queues[runq_queue_i])
+        runq_queue_p = runq_queue_head.head
+
+        if unsigned(runq_queue_p):
+            runq_queue_this_count = 0
+
+            for thread in ParanoidIterateLinkageChain(runq_queue_head, "thread_t", "runq_links", circleQueue=True):
+                runq_queue_this_count += 1
+
+            print "      Queue [{: <#012x}] Priority {: <3d} count {:d}\n".format(runq_queue_head, runq_queue_i, runq_queue_this_count)
+            print "\t" + GetThreadSummary.header + "\n"
+            for thread in ParanoidIterateLinkageChain(runq_queue_head, "thread_t", "runq_links", circleQueue=True):
+                print "\t" + GetThreadSummary(thread) + "\n"
+                if config['verbosity'] > vHUMAN :
+                    print "\t" + GetThreadBackTrace(thread, prefix="\t\t") + "\n"
+
+def ShowRTRunQSummary(rt_runq):
+    if (hex(rt_runq.count) == hex(0xfdfdfdfd)) :
+        print "    Realtime Queue ({:<#012x}) uninitialized\n".format(addressof(rt_runq.queue))
+        return
+    print "    Realtime Queue ({:<#012x}) Count {:d}\n".format(addressof(rt_runq.queue), rt_runq.count)
+    if rt_runq.count != 0:
+        print "\t" + GetThreadSummary.header + "\n"
+        for rt_runq_thread in ParanoidIterateLinkageChain(rt_runq.queue, "thread_t", "runq_links", circleQueue=True):
+            print "\t" + GetThreadSummary(rt_runq_thread) + "\n"
+
+def ShowGrrrSummary(grrr_runq):
+    """ Internal function to print summary of grrr_run_queue
+        params: grrr_runq - value representing struct grrr_run_queue *
+    """
+    print "    GRRR Info: Count {: <10d} Weight {: <10d} Current Group {: <#012x}\n".format(grrr_runq.count,
+        grrr_runq.weight, grrr_runq.current_group)
+    grrr_group_i = 0
+    grrr_group_count = sizeof(grrr_runq.groups)/sizeof(grrr_runq.groups[0])
+    for grrr_group_i in xrange(grrr_group_count) :
+        grrr_group = addressof(grrr_runq.groups[grrr_group_i])
+        if grrr_group.count > 0:
+            print "      Group {: <3d} [{: <#012x}] ".format(grrr_group.index, grrr_group)
+            print "Count {:d} Weight {:d}\n".format(grrr_group.count, grrr_group.weight)
+            grrr_group_client_head = addressof(grrr_group.clients)
+            print GetThreadSummary.header
+            for thread in ParanoidIterateLinkageChain(grrr_group_client_head, "thread_t", "runq_links", circleQueue=True):
+                print "\t" + GetThreadSummary(thread) + "\n"
+                if config['verbosity'] > vHUMAN :
+                    print "\t" + GetThreadBackTrace(thread, prefix="\t\t") + "\n"
+
+def ShowActiveThread(processor):
+    if (processor.active_thread != 0) :
+        print "\t" + GetThreadSummary.header + "\n"
+        print "\t" + GetThreadSummary(processor.active_thread) + "\n"
+
+@lldb_command('showallprocessors')
+@lldb_command('showscheduler')
+def ShowScheduler(cmd_args=None):
+    """  Routine to print information of all psets and processors
+         Usage: showscheduler
+    """
+    node = addressof(kern.globals.pset_node0)
+    show_grrr = 0
+    show_priority_runq = 0
+    show_priority_pset_runq = 0
+    show_group_pset_runq = 0
+    show_clutch = 0
+    show_edge = 0
+    sched_string = str(kern.globals.sched_string)
+
+    if sched_string == "traditional":
+        show_priority_runq = 1
+    elif sched_string == "traditional_with_pset_runqueue":
+        show_priority_pset_runq = 1
+    elif sched_string == "grrr":
+        show_grrr = 1
+    elif sched_string == "multiq":
+        show_priority_runq = 1
+        show_group_pset_runq = 1
+    elif sched_string == "dualq":
+        show_priority_pset_runq = 1
+        show_priority_runq = 1
+    elif sched_string == "amp":
+        show_priority_pset_runq = 1
+        show_priority_runq = 1
+    elif sched_string == "clutch":
+        show_clutch = 1
+    elif sched_string == "edge":
+        show_edge = 1
+    else :
+        print "Unknown sched_string {:s}".format(sched_string)
+
+    print "Scheduler: {:s}\n".format(sched_string)
+
+    if show_clutch == 0 and show_edge == 0:
+        run_buckets = kern.globals.sched_run_buckets
+        run_count      = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_RUN')]
+        fixpri_count   = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_FIXPRI')]
+        share_fg_count = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_FG')]
+        share_df_count = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_DF')]
+        share_ut_count = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_UT')]
+        share_bg_count = run_buckets[GetEnumValue('sched_bucket_t::TH_BUCKET_SHARE_BG')]
+        print "Processors: {g.processor_avail_count:d} Runnable threads: {:d} Fixpri threads: {:d}\n".format(run_count, fixpri_count, g=kern.globals)
+        print "FG Timeshare threads: {:d} DF Timeshare threads: {:d} UT Timeshare threads: {:d} BG Timeshare threads: {:d}\n".format(share_fg_count, share_df_count, share_ut_count, share_bg_count)
+    
+    processor_offline     = GetEnumValue('processor_state_t::PROCESSOR_OFF_LINE')
+    processor_idle        = GetEnumValue('processor_state_t::PROCESSOR_IDLE')
+    processor_dispatching = GetEnumValue('processor_state_t::PROCESSOR_DISPATCHING')
+    processor_running     = GetEnumValue('processor_state_t::PROCESSOR_RUNNING')
+
+    if show_group_pset_runq:
+        if hasattr(kern.globals, "multiq_sanity_check"):
+            print "multiq scheduler config: deep-drain {g.deep_drain:d}, ceiling {g.drain_ceiling:d}, depth limit {g.drain_depth_limit:d}, band limit {g.drain_band_limit:d}, sanity check {g.multiq_sanity_check:d}\n".format(g=kern.globals)
+        else:
+            print "multiq scheduler config: deep-drain {g.deep_drain:d}, ceiling {g.drain_ceiling:d}, depth limit {g.drain_depth_limit:d}, band limit {g.drain_band_limit:d}\n".format(g=kern.globals)
+
+        # Create a group->task mapping
+        task_map = {}
+        for task in kern.tasks:
+            task_map[unsigned(task.sched_group)] = task
+        for task in kern.terminated_tasks:
+            task_map[unsigned(task.sched_group)] = task
+
+    print " \n"
+
+    while node != 0:
+        pset = node.psets
+        pset = kern.GetValueFromAddress(unsigned(pset), 'struct processor_set *')
+
+        while pset != 0:
+            print "Processor Set  {: <#012x} Count {:d} (cpu_id {:<#x}-{:<#x})\n".format(pset,
+                unsigned(pset.cpu_set_count), pset.cpu_set_low, pset.cpu_set_hi)
+
+            rt_runq = kern.GetValueFromAddress(unsigned(addressof(pset.rt_runq)), 'struct rt_queue *')
+            ShowRTRunQSummary(rt_runq)
+
+            if show_priority_pset_runq:
+                runq = kern.GetValueFromAddress(unsigned(addressof(pset.pset_runq)), 'struct run_queue *')
+                ShowRunQSummary(runq)
+
+            if show_group_pset_runq:
+                print "Main Runq:\n"
+                runq = kern.GetValueFromAddress(unsigned(addressof(pset.pset_runq)), 'struct run_queue *')
+                ShowGroupSetSummary(runq, task_map)
+                print "All Groups:\n"
+                # TODO: Possibly output task header for each group
+                for group in IterateQueue(kern.globals.sched_groups, "sched_group_t", "sched_groups"):
+                    if (group.runq.count != 0) :
+                        task = task_map.get(unsigned(group), "Unknown task!")
+                        print "Group {: <#012x} Task {: <#012x}\n".format(unsigned(group), unsigned(task))
+                        ShowRunQSummary(group.runq)
+            print " \n"
+            
+            processor_array = kern.globals.processor_array
+
+            print "Active Processors:\n"
+            active_bitmap = int(pset.cpu_state_map[processor_dispatching]) | int(pset.cpu_state_map[processor_running])
+            for cpuid in IterateBitmap(active_bitmap):
+                processor = processor_array[cpuid]
+                if processor != 0:
+                    print "    " + GetProcessorSummary(processor)
+                    ShowActiveThread(processor)
+
+                    if show_priority_runq:
+                        runq = processor.runq
+                        ShowRunQSummary(runq)
+                    if show_grrr:
+                        grrr_runq = processor.grrr_runq
+                        ShowGrrrSummary(grrr_runq)
+            print " \n"
+
+
+            print "Idle Processors:\n"
+            idle_bitmap = int(pset.cpu_state_map[processor_idle]) & int(pset.primary_map)
+            for cpuid in IterateBitmap(idle_bitmap):
+                processor = processor_array[cpuid]
+                if processor != 0:
+                    print "    " + GetProcessorSummary(processor)
+                    ShowActiveThread(processor)
+
+                    if show_priority_runq:
+                        ShowRunQSummary(processor.runq)
+            print " \n"
+
+
+            print "Idle Secondary Processors:\n"
+            idle_bitmap = int(pset.cpu_state_map[processor_idle]) & ~(int(pset.primary_map))
+            for cpuid in IterateBitmap(idle_bitmap):
+                processor = processor_array[cpuid]
+                if processor != 0:
+                    print "    " + GetProcessorSummary(processor)
+                    ShowActiveThread(processor)
+
+                    if show_priority_runq:
+                        print ShowRunQSummary(processor.runq)
+            print " \n"
+
+
+            print "Other Processors:\n"
+            other_bitmap = 0
+            for i in range(processor_offline, processor_idle):
+                other_bitmap |= int(pset.cpu_state_map[i])
+            other_bitmap &= int(pset.cpu_bitmask)
+            for cpuid in IterateBitmap(other_bitmap):
+                processor = processor_array[cpuid]
+                if processor != 0:
+                    print "    " + GetProcessorSummary(processor)
+                    ShowActiveThread(processor)
+
+                    if show_priority_runq:
+                        ShowRunQSummary(processor.runq)
+            print " \n"
+
+            if show_clutch or show_edge:
+                cluster_type = "SMP"
+                if pset.pset_type == 1:
+                    cluster_type = "E"
+                elif pset.pset_type == 2:
+                    cluster_type = "P"
+                print "=== Clutch Scheduler Hierarchy Pset{:d} (Type: {:s}) ] ===\n\n".format(pset.pset_cluster_id, cluster_type)
+                ShowSchedClutchForPset(pset)
+
+            pset = pset.pset_list
+
+        node = node.node_list
+
+    print "\nCrashed Threads Queue: ({:<#012x})\n".format(addressof(kern.globals.crashed_threads_queue))
+    first = True
+    for thread in ParanoidIterateLinkageChain(kern.globals.crashed_threads_queue, "thread_t", "runq_links"):
+        if first:
+            print "\t" + GetThreadSummary.header
+            first = False
+        print "\t" + GetThreadSummary(thread)
+
+    def dump_mpsc_thread_queue(name, head):
+        head = addressof(head)
+        print "\n{:s}: ({:<#012x})\n".format(name, head)
+        first = True
+        for thread in IterateMPSCQueue(head.mpd_queue, 'struct thread', 'mpsc_links'):
+            if first:
+                print "\t" + GetThreadSummary.header
+                first = False
+            print "\t" + GetThreadSummary(thread)
+
+    dump_mpsc_thread_queue("Terminate Queue", kern.globals.thread_terminate_queue)
+    dump_mpsc_thread_queue("Waiting For Kernel Stacks Queue", kern.globals.thread_stack_queue)
+    dump_mpsc_thread_queue("Thread Exception Queue", kern.globals.thread_exception_queue)
+    dump_mpsc_thread_queue("Thread Deallocate Queue", kern.globals.thread_deallocate_queue)
+
+    print "\n"
+
+    print "\n"
+
+# EndMacro: showallprocessors
+
+
+def ParanoidIterateLinkageChain(queue_head, element_type, field_name, field_ofst=0, circleQueue=False):
+    """ Iterate over a Linkage Chain queue in kernel of type queue_head_t or circle_queue_head_t. (osfmk/kern/queue.h method 1 or circle_queue.h)
+        This is equivalent to the qe_foreach_element() macro
+        Blows up aggressively and descriptively when something goes wrong iterating a queue.
+        Prints correctness errors, and throws exceptions on 'cannot proceed' errors
+        If this is annoying, set the global 'enable_paranoia' to false.
+
+        params:
+            queue_head   - value       : Value object for queue_head.
+            element_type - lldb.SBType : pointer type of the element which contains the queue_chain_t. Typically its structs like thread, task etc..
+                         - str         : OR a string describing the type. ex. 'task *'
+            field_name   - str         : Name of the field (in element) which holds a queue_chain_t
+            field_ofst   - int         : offset from the 'field_name' (in element) which holds a queue_chain_t
+                                         This is mostly useful if a particular element contains an array of queue_chain_t
+        returns:
+            A generator does not return. It is used for iterating.
+            value  : An object thats of type (element_type). Always a pointer object
+        example usage:
+            for thread in IterateQueue(kern.globals.threads, 'thread *', 'threads'):
+                print thread.thread_id
+    """
+
+    if type(element_type) is str:
+        element_type = gettype(element_type)
+
+    # Some ways of constructing a queue head seem to end up with the
+    # struct object as the value and not a pointer to the struct head
+    # In that case, addressof will give us a pointer to the struct, which is what we need
+    if not queue_head.GetSBValue().GetType().IsPointerType() :
+        queue_head = addressof(queue_head)
+
+    if circleQueue:
+        # Mosh the value into a brand new value, to really get rid of its old cvalue history
+        queue_head = kern.GetValueFromAddress(unsigned(queue_head), 'struct circle_queue_head *').head
+    else:
+        # Mosh the value into a brand new value, to really get rid of its old cvalue history
+        queue_head = kern.GetValueFromAddress(unsigned(queue_head), 'struct queue_entry *')
+
+    if unsigned(queue_head) == 0:
+        if not circleQueue and ParanoidIterateLinkageChain.enable_paranoia:
+            print "bad queue_head_t: {:s}".format(queue_head)
+        return
+
+    if element_type.IsPointerType():
+        struct_type = element_type.GetPointeeType()
+    else:
+        struct_type = element_type
+
+    elem_ofst = getfieldoffset(struct_type, field_name) + field_ofst
+
+    try:
+        link = queue_head.next
+        last_link = queue_head
+        try_read_next = unsigned(queue_head.next)
+    except:
+        print "Exception while looking at queue_head: {:>#18x}".format(unsigned(queue_head))
+        raise
+
+    if ParanoidIterateLinkageChain.enable_paranoia:
+        if unsigned(queue_head.next) == 0:
+            raise ValueError("NULL next pointer on head: queue_head {:>#18x} next: {:>#18x} prev: {:>#18x}".format(queue_head, queue_head.next, queue_head.prev))
+        if unsigned(queue_head.prev) == 0:
+            print "NULL prev pointer on head: queue_head {:>#18x} next: {:>#18x} prev: {:>#18x}".format(queue_head, queue_head.next, queue_head.prev)
+        if unsigned(queue_head.next) == unsigned(queue_head) and unsigned(queue_head.prev) != unsigned(queue_head):
+            print "corrupt queue_head {:>#18x} next: {:>#18x} prev: {:>#18x}".format(queue_head, queue_head.next, queue_head.prev)
+
+    if ParanoidIterateLinkageChain.enable_debug :
+        print "starting at queue_head {:>#18x} next: {:>#18x} prev: {:>#18x}".format(queue_head, queue_head.next, queue_head.prev)
+
+    addr = 0
+    obj = 0
+
+    try:
+        while True:
+            if not circleQueue and unsigned(queue_head) == unsigned(link):
+                break;
+            if ParanoidIterateLinkageChain.enable_paranoia:
+                if unsigned(link.next) == 0:
+                    raise ValueError("NULL next pointer: queue_head {:>#18x} link: {:>#18x} next: {:>#18x} prev: {:>#18x}".format(queue_head, link, link.next, link.prev))
+                if unsigned(link.prev) == 0:
+                    print "NULL prev pointer: queue_head {:>#18x} link: {:>#18x} next: {:>#18x} prev: {:>#18x}".format(queue_head, link, link.next, link.prev)
+                if unsigned(last_link) != unsigned(link.prev):
+                    print "Corrupt prev pointer: queue_head {:>#18x} link: {:>#18x} next: {:>#18x} prev: {:>#18x} prev link: {:>#18x} ".format(
+                            queue_head, link, link.next, link.prev, last_link)
+
+            addr = unsigned(link) - unsigned(elem_ofst);
+            obj = kern.GetValueFromAddress(addr, element_type)
+            if ParanoidIterateLinkageChain.enable_debug :
+                print "yielding link: {:>#18x} next: {:>#18x} prev: {:>#18x} addr: {:>#18x} obj: {:>#18x}".format(link, link.next, link.prev, addr, obj)
+            yield obj
+            last_link = link
+            link = link.next
+            if circleQueue and unsigned(queue_head) == unsigned(link):
+                break;
+    except:
+        exc_info = sys.exc_info()
+        try:
+            print "Exception while iterating queue: {:>#18x} link: {:>#18x} addr: {:>#18x} obj: {:>#18x} last link: {:>#18x}".format(queue_head, link, addr, obj, last_link)
+        except:
+            import traceback
+            traceback.print_exc()
+        raise exc_info[0], exc_info[1], exc_info[2]
+
+ParanoidIterateLinkageChain.enable_paranoia = True
+ParanoidIterateLinkageChain.enable_debug = False
+
+def LinkageChainEmpty(queue_head):
+    if not queue_head.GetSBValue().GetType().IsPointerType() :
+        queue_head = addressof(queue_head)
+
+    # Mosh the value into a brand new value, to really get rid of its old cvalue history
+    # avoid using GetValueFromAddress
+    queue_head = value(queue_head.GetSBValue().CreateValueFromExpression(None,'(void *)'+str(unsigned(queue_head))))
+    queue_head = cast(queue_head, 'struct queue_entry *')
+
+    link = queue_head.next
+
+    return unsigned(queue_head) == unsigned(link)
+
+def bit_first(bitmap):
+    return bitmap.bit_length() - 1
+
+def lsb_first(bitmap):
+    bitmap = bitmap & -bitmap
+    return bit_first(bitmap)
+
+def IterateBitmap(bitmap):
+    """ Iterate over a bitmap, returning the index of set bits starting from 0
+
+        params:
+            bitmap       - value       : bitmap
+        returns:
+            A generator does not return. It is used for iterating.
+            value  : index of a set bit
+        example usage:
+            for cpuid in IterateBitmap(running_bitmap):
+                print processor_array[cpuid]
+    """
+    i = lsb_first(bitmap)
+    while (i >= 0):
+        yield i
+        bitmap = bitmap & ~((1 << (i + 1)) - 1)
+        i = lsb_first(bitmap)
+
+
+# Macro: showallcallouts
+
+from kevent import GetKnoteKqueue
+
+def ShowThreadCall(prefix, call, recent_timestamp, is_pending=False):
+    """
+    Print a description of a thread_call_t and its relationship to its expected fire time
+    """
+    func = call.tc_func
+    param0 = call.tc_param0
+    param1 = call.tc_param1
+
+    is_iotes = False
+
+    func_name = kern.Symbolicate(func)
+
+    extra_string = ""
+
+    strip_func = kern.StripKernelPAC(unsigned(func))
+
+    func_syms = kern.SymbolicateFromAddress(strip_func)
+    # returns an array of SBSymbol
+
+    if func_syms and func_syms[0] :
+        func_name = func_syms[0].GetName()
+
+        try :
+            if ("IOTimerEventSource::timeoutAndRelease" in func_name or
+                "IOTimerEventSource::timeoutSignaled" in func_name) :
+                iotes = Cast(call.tc_param0, 'IOTimerEventSource*')
+                try:
+                    func = iotes.action
+                    param0 = iotes.owner
+                    param1 = unsigned(iotes)
+                except AttributeError:
+                    # This is horrible, horrible, horrible.  But it works.  Needed because IOEventSource hides the action member in an
+                    # anonymous union when XNU_PRIVATE_SOURCE is set.  To grab it, we work backwards from the enabled member.
+                    func = dereference(kern.GetValueFromAddress(addressof(iotes.enabled) - sizeof('IOEventSource::Action'), 'uint64_t *'))
+                    param0 = iotes.owner
+                    param1 = unsigned(iotes)
+
+                workloop = iotes.workLoop
+                thread = workloop.workThread
+
+                is_iotes = True
+
+                # re-symbolicate the func we found inside the IOTES
+                strip_func = kern.StripKernelPAC(unsigned(func))
+                func_syms = kern.SymbolicateFromAddress(strip_func)
+                if func_syms and func_syms[0] :
+                    func_name = func_syms[0].GetName()
+                else :
+                    func_name = str(FindKmodNameForAddr(func))
+
+                # cast from IOThread to thread_t, because IOThread is sometimes opaque
+                thread = Cast(thread, 'thread_t')
+                thread_id = thread.thread_id
+                thread_name = GetThreadName(thread)
+
+                extra_string += "workloop thread: {:#x} ({:#x}) {:s}".format(thread, thread_id, thread_name)
+
+            if "filt_timerexpire" in func_name :
+                knote = Cast(call.tc_param0, 'struct knote *')
+                kqueue = GetKnoteKqueue(knote)
+                proc = kqueue.kq_p
+                proc_name = GetProcName(proc)
+                proc_pid = proc.p_pid
+
+                extra_string += "kq: {:#018x} {:s}[{:d}]".format(kqueue, proc_name, proc_pid)
+
+            if "mk_timer_expire" in func_name :
+                timer = Cast(call.tc_param0, 'struct mk_timer *')
+                port = timer.port
+
+                extra_string += "port: {:#018x} {:s}".format(port, GetPortDestinationSummary(port))
+
+            if "workq_kill_old_threads_call" in func_name :
+                workq = Cast(call.tc_param0, 'struct workqueue *')
+                proc = workq.wq_proc
+                proc_name = GetProcName(proc)
+                proc_pid = proc.p_pid
+
+                extra_string += "{:s}[{:d}]".format(proc_name, proc_pid)
+
+            if ("workq_add_new_threads_call" in func_name or
+                "realitexpire" in func_name):
+                proc = Cast(call.tc_param0, 'struct proc *')
+                proc_name = GetProcName(proc)
+                proc_pid = proc.p_pid
+
+                extra_string += "{:s}[{:d}]".format(proc_name, proc_pid)
+
+        except:
+            print "exception generating extra_string for call: {:#018x}".format(call)
+            if ShowThreadCall.enable_debug :
+                raise
+
+    if (func_name == "") :
+        func_name = FindKmodNameForAddr(func)
+
+    if (call.tc_flags & GetEnumValue('thread_call_flags_t::THREAD_CALL_FLAG_CONTINUOUS')) :
+        timer_fire = call.tc_pqlink.deadline - (recent_timestamp + kern.globals.mach_absolutetime_asleep)
+    else :
+        timer_fire = call.tc_pqlink.deadline - recent_timestamp
+
+    timer_fire_s = kern.GetNanotimeFromAbstime(timer_fire) / 1000000000.0
+
+    ttd_s = kern.GetNanotimeFromAbstime(call.tc_ttd) / 1000000000.0
+
+    if (is_pending) :
+        pending_time = call.tc_pending_timestamp - recent_timestamp
+        pending_time = kern.GetNanotimeFromAbstime(pending_time) / 1000000000.0
+
+    flags = int(call.tc_flags)
+    # TODO: extract this out of the thread_call_flags_t enum
+    thread_call_flags = {0x0:'', 0x1:'A', 0x2:'W', 0x4:'D', 0x8:'R', 0x10:'S', 0x20:'O',
+            0x40:'P', 0x80:'L', 0x100:'C'}
+
+    flags_str = ''
+    mask = 0x1
+    while mask <= 0x100 :
+        flags_str += thread_call_flags[int(flags & mask)]
+        mask = mask << 1
+
+    if is_iotes :
+        flags_str += 'I'
+
+    if (is_pending) :
+        print ("{:s}{:#018x}: {:18d} {:18d} {:16.06f} {:16.06f} {:16.06f} {:9s} " +
+                "{:#018x} ({:#018x}, {:#018x}) ({:s}) {:s}").format(prefix,
+                unsigned(call), call.tc_pqlink.deadline, call.tc_soft_deadline, ttd_s,
+                timer_fire_s, pending_time, flags_str,
+                func, param0, param1, func_name, extra_string)
+    else :
+        print ("{:s}{:#018x}: {:18d} {:18d} {:16.06f} {:16.06f} {:9s} " +
+                "{:#018x} ({:#018x}, {:#018x}) ({:s}) {:s}").format(prefix,
+                unsigned(call), call.tc_pqlink.deadline, call.tc_soft_deadline, ttd_s,
+                timer_fire_s, flags_str,
+                func, param0, param1, func_name, extra_string)
+
+ShowThreadCall.enable_debug = False
+
+@header("{:>18s}  {:>18s} {:>18s} {:>16s} {:>16s} {:9s} {:>18s}".format(
+            "entry", "deadline", "soft_deadline",
+            "duration (s)", "to go (s)", "flags", "(*func) (param0, param1)"))
+def PrintThreadGroup(group):
+    header = PrintThreadGroup.header
+    pending_header = "{:>18s}  {:>18s} {:>18s} {:>16s} {:>16s} {:>16s} {:9s} {:>18s}".format(
+            "entry", "deadline", "soft_deadline",
+            "duration (s)", "to go (s)", "pending", "flags", "(*func) (param0, param1)")
+
+    recent_timestamp = GetRecentTimestamp()
+
+    idle_timestamp_distance = group.idle_timestamp - recent_timestamp
+    idle_timestamp_distance_s = kern.GetNanotimeFromAbstime(idle_timestamp_distance) / 1000000000.0
+
+    is_parallel = ""
+
+    if (group.tcg_flags & GetEnumValue('thread_call_group_flags_t::TCG_PARALLEL')) :
+        is_parallel = " (parallel)"
+
+    print "Group: {g.tcg_name:s} ({:#18x}){:s}".format(unsigned(group), is_parallel, g=group)
+    print "\t" +"Thread Priority: {g.tcg_thread_pri:d}\n".format(g=group)
+    print ("\t" +"Active: {g.active_count:<3d} Idle: {g.idle_count:<3d}" +
+        "Blocked: {g.blocked_count:<3d} Pending: {g.pending_count:<3d}" +
+        "Target: {g.target_thread_count:<3d}\n").format(g=group)
+
+    if unsigned(group.idle_timestamp) is not 0 :
+        print "\t" +"Idle Timestamp: {g.idle_timestamp:d} ({:03.06f})\n".format(idle_timestamp_distance_s,
+            g=group)
+
+    print "\t" +"Pending Queue: ({:>#18x})\n".format(addressof(group.pending_queue))
+    if not LinkageChainEmpty(group.pending_queue) :
+        print "\t\t" + pending_header
+    for call in ParanoidIterateLinkageChain(group.pending_queue, "thread_call_t", "tc_qlink"):
+        ShowThreadCall("\t\t", call, recent_timestamp, is_pending=True)
+
+    print "\t" +"Delayed Queue (Absolute Time): ({:>#18x}) timer: ({:>#18x})\n".format(
+            addressof(group.delayed_queues[0]), addressof(group.delayed_timers[0]))
+    if not LinkageChainEmpty(group.delayed_queues[0]) :
+        print "\t\t" + header
+    for call in ParanoidIterateLinkageChain(group.delayed_queues[0], "thread_call_t", "tc_qlink"):
+        ShowThreadCall("\t\t", call, recent_timestamp)
+
+    print "\t" +"Delayed Queue (Continuous Time): ({:>#18x}) timer: ({:>#18x})\n".format(
+            addressof(group.delayed_queues[1]), addressof(group.delayed_timers[1]))
+    if not LinkageChainEmpty(group.delayed_queues[1]) :
+        print "\t\t" + header
+    for call in ParanoidIterateLinkageChain(group.delayed_queues[1], "thread_call_t", "tc_qlink"):
+        ShowThreadCall("\t\t", call, recent_timestamp)
+
+def PrintThreadCallThreads() :
+    callout_flag = GetEnumValue('thread_tag_t::THREAD_TAG_CALLOUT')
+    recent_timestamp = GetRecentTimestamp()
+
+    for thread in IterateQueue(kern.globals.kernel_task.threads, 'thread *', 'task_threads'):
+        if (thread.thread_tag & callout_flag) :
+            print " {:#20x} {:#12x} {:s}".format(thread, thread.thread_id, GetThreadName(thread))
+            state = thread.thc_state
+            if state and state.thc_call :
+                print "\t" + PrintThreadGroup.header
+                ShowThreadCall("\t", state.thc_call, recent_timestamp)
+                soft_deadline = state.thc_call_soft_deadline
+                slop_time = state.thc_call_hard_deadline - soft_deadline
+                slop_time = kern.GetNanotimeFromAbstime(slop_time) / 1000000000.0
+                print "\t original soft deadline {:d}, hard deadline {:d} (leeway {:.06f}s)".format(
+                        soft_deadline, state.thc_call_hard_deadline, slop_time)
+                enqueue_time = state.thc_call_pending_timestamp - soft_deadline
+                enqueue_time = kern.GetNanotimeFromAbstime(enqueue_time) / 1000000000.0
+                print "\t time to enqueue after deadline: {:.06f}s (enqueued at: {:d})".format(
+                        enqueue_time, state.thc_call_pending_timestamp)
+                wait_time = state.thc_call_start - state.thc_call_pending_timestamp
+                wait_time = kern.GetNanotimeFromAbstime(wait_time) / 1000000000.0
+                print "\t time to start executing after enqueue: {:.06f}s (executing at: {:d})".format(
+                        wait_time, state.thc_call_start)
+
+                if (state.thc_IOTES_invocation_timestamp) :
+                    iotes_acquire_time = state.thc_IOTES_invocation_timestamp - state.thc_call_start
+                    iotes_acquire_time = kern.GetNanotimeFromAbstime(iotes_acquire_time) / 1000000000.0
+                    print "\t IOTES acquire time: {:.06f}s (acquired at: {:d})".format(
+                            iotes_acquire_time, state.thc_IOTES_invocation_timestamp)
+
+
+@lldb_command('showcalloutgroup')
+def ShowCalloutGroup(cmd_args=None):
+    """ Prints out the pending and delayed thread calls for a specific group
+
+        Pass 'threads' to show the thread call threads themselves.
+
+        Callout flags:
+
+        A - Allocated memory owned by thread_call.c
+        W - Wait - thread waiting for call to finish running
+        D - Delayed - deadline based
+        R - Running - currently executing on a thread
+        S - Signal - call from timer interrupt instead of thread
+        O - Once - pend the enqueue if re-armed while running
+        P - Reschedule pending - enqueue is pending due to re-arm while running
+        L - Rate-limited - (App Nap)
+        C - Continuous time - Timeout is in mach_continuous_time
+        I - Callout is an IOTimerEventSource
+    """
+    if not cmd_args:
+        print "No arguments passed"
+        print ShowCalloutGroup.__doc__
+        return False
+
+    if "threads" in cmd_args[0] :
+        PrintThreadCallThreads()
+        return
+
+    group = kern.GetValueFromAddress(cmd_args[0], 'struct thread_call_group *')
+    if not group:
+        print "unknown arguments:", str(cmd_args)
+        return False
+
+    PrintThreadGroup(group)
+
+@lldb_command('showallcallouts')
+def ShowAllCallouts(cmd_args=None):
+    """ Prints out the pending and delayed thread calls for the thread call groups
+
+        Callout flags:
+
+        A - Allocated memory owned by thread_call.c
+        W - Wait - thread waiting for call to finish running
+        D - Delayed - deadline based
+        R - Running - currently executing on a thread
+        S - Signal - call from timer interrupt instead of thread
+        O - Once - pend the enqueue if re-armed while running
+        P - Reschedule pending - enqueue is pending due to re-arm while running
+        L - Rate-limited - (App Nap)
+        C - Continuous time - Timeout is in mach_continuous_time
+        I - Callout is an IOTimerEventSource
+    """
+    index_max = GetEnumValue('thread_call_index_t::THREAD_CALL_INDEX_MAX')
+
+    for i in range (0, index_max) :
+        group = addressof(kern.globals.thread_call_groups[i])
+        PrintThreadGroup(group)
+
+    print "Thread Call Threads:"
+    PrintThreadCallThreads()
+
+# EndMacro: showallcallouts
+