runqueue.py: Add task scheduler abstraction and some example schedulers. Improve circular dependency chain debugging code and user feedback.

1 files changed, 237 insertions, 74 deletions

diff --git a/lib/bb/runqueue.py b/lib/bb/runqueue.py
index 7c918a5d8..c970e2531 100644
--- a/lib/bb/runqueue.py
+++ b/lib/bb/runqueue.py
@@ -70,6 +70,88 @@ runQueueCleanUp = 7
 runQueueComplete = 8
 runQueueChildProcess = 9
 
+class RunQueueScheduler:
+    """
+    Control the order tasks are scheduled in.
+    """
+    def __init__(self, runqueue):
+        """
+        The default scheduler just returns the first buildable task (the 
+        priority map is sorted by task numer)
+        """
+        self.rq = runqueue
+        numTasks = len(self.rq.runq_fnid)
+
+        self.prio_map = []
+        self.prio_map.extend(range(numTasks))
+
+    def next(self):
+        """
+        Return the id of the first task we find that is buildable
+        """
+        for task1 in range(len(self.rq.runq_fnid)):
+            task = self.prio_map[task1]
+            if self.rq.runq_running[task] == 1:
+                continue
+            if self.rq.runq_buildable[task] == 1:
+                return task
+
+class RunQueueSchedulerSpeed(RunQueueScheduler):
+    """
+    A scheduler optimised for speed. The priority map is sorted by task weight,
+    heavier weighted tasks (tasks needed by the most other tasks) are run first.
+    """
+    def __init__(self, runqueue):
+        """
+        The priority map is sorted by task weight.
+        """
+        from copy import deepcopy
+
+        self.rq = runqueue
+
+        sortweight = deepcopy(self.rq.runq_weight)
+        sortweight.sort()
+        copyweight = deepcopy(self.rq.runq_weight)
+        self.prio_map = []
+
+        for weight in sortweight:
+            idx = copyweight.index(weight)
+            self.prio_map.append(idx)
+            copyweight[idx] = -1
+
+        self.prio_map.reverse()
+
+class RunQueueSchedulerCompletion(RunQueueSchedulerSpeed):
+    """
+    A scheduler optimised to complete .bb files are quickly as possible. The 
+    priority map is sorted by task weight, but then reordered so once a given 
+    .bb file starts to build, its completed as quickly as possible. This works
+    well where disk space is at a premium and classes like OE's rm_work are in 
+    force.
+    """
+    def __init__(self, runqueue):
+        RunQueueSchedulerSpeed.__init__(self, runqueue)
+        from copy import deepcopy
+
+        #FIXME - whilst this groups all fnids together it does not reorder the
+        #fnid groups optimally.
+ 
+        basemap = deepcopy(self.prio_map)
+        self.prio_map = []
+        while (len(basemap) > 0):
+            entry = basemap.pop(0)
+            self.prio_map.append(entry)
+            fnid = self.rq.runq_fnid[entry]
+            todel = []
+            for entry in basemap:
+                entry_fnid = self.rq.runq_fnid[entry]
+                if entry_fnid == fnid:
+                    todel.append(basemap.index(entry))
+                    self.prio_map.append(entry)
+            todel.reverse()
+            for idx in todel:
+                del basemap[idx]
+
 class RunQueue:
     """
     BitBake Run Queue implementation
@@ -90,8 +172,6 @@ class RunQueue:
         self.runq_task = []
         self.runq_depends = []
         self.runq_revdeps = []
-        self.runq_weight = []
-        self.prio_map = []
 
         self.state = runQueuePrepare
 
@@ -100,6 +180,152 @@ class RunQueue:
         taskname = self.runq_task[task]
         return "%s, %s" % (fn, taskname)
 
+    def circular_depchains_handler(self, tasks):
+        """
+        Some tasks aren't buildable, likely due to circular dependency issues.
+        Identify the circular dependencies and print them in a user readable format.
+        """
+        from copy import deepcopy
+
+        valid_chains = []
+        explored_deps = {}
+        msgs = []
+
+        def chain_reorder(chain):
+            """
+            Reorder a dependency chain so the lowest task id is first
+            """
+            lowest = 0
+            new_chain = []
+            for entry in range(len(chain)):
+                if chain[entry] < chain[lowest]:
+                    lowest = entry
+            new_chain.extend(chain[lowest:])
+            new_chain.extend(chain[:lowest])
+            return new_chain
+
+        def chain_compare_equal(chain1, chain2):
+            """
+            Compare two dependency chains and see if they're the same
+            """
+            if len(chain1) != len(chain2):
+                return False
+            for index in range(len(chain1)):
+                if chain1[index] != chain2[index]:
+                    return False
+            return True
+            
+        def chain_array_contains(chain, chain_array):
+            """
+            Return True if chain_array contains chain
+            """
+            for ch in chain_array:
+                if chain_compare_equal(ch, chain):
+                    return True
+            return False
+
+        def find_chains(taskid, prev_chain):
+            prev_chain.append(taskid)
+            total_deps = []
+            total_deps.extend(self.runq_revdeps[taskid])
+            for revdep in self.runq_revdeps[taskid]:
+                if revdep in prev_chain:
+                    idx = prev_chain.index(revdep)
+                    # To prevent duplicates, reorder the chain to start with the lowest taskid
+                    # and search through an array of those we've already printed
+                    chain = prev_chain[idx:]
+                    new_chain = chain_reorder(chain)
+                    if not chain_array_contains(new_chain, valid_chains):
+                        valid_chains.append(new_chain)
+                        msgs.append("Dependency loop #%d found:\n" % len(valid_chains))
+                        for dep in new_chain:
+                            msgs.append("  Task %s (%s) (depends: %s)\n" % (dep, self.get_user_idstring(dep), self.runq_depends[dep]))
+                        msgs.append("\n")
+                    if len(valid_chains) > 10:
+                        msgs.append("Aborted dependency loops search after 10 matches.\n")
+                        return msgs
+                    continue
+                scan = False
+                if revdep not in explored_deps:
+                    scan = True
+                elif revdep in explored_deps[revdep]:
+                    scan = True
+                else:
+                    for dep in prev_chain:
+                        if dep in explored_deps[revdep]:
+                            scan = True
+                if scan:
+                    find_chains(revdep, deepcopy(prev_chain))
+                for dep in explored_deps[revdep]:
+                    if dep not in total_deps:
+                        total_deps.append(dep)
+
+            explored_deps[taskid] = total_deps
+
+        for task in tasks:
+            find_chains(task, [])
+
+        return msgs
+
+    def calculate_task_weights(self, endpoints):
+        """
+        Calculate a number representing the "weight" of each task. Heavier weighted tasks 
+        have more dependencies and hence should be executed sooner for maximum speed.
+
+        This function also sanity checks the task list finding tasks that its not
+        possible to execute due to circular dependencies.
+        """
+
+        numTasks = len(self.runq_fnid)
+        weight = []
+        deps_left = []
+        task_done = []
+
+        for listid in range(numTasks):
+            task_done.append(False)
+            weight.append(0)
+            deps_left.append(len(self.runq_revdeps[listid]))
+
+        for listid in endpoints:
+            weight[listid] = 1
+            task_done[listid] = True
+
+        while 1:
+            next_points = []
+            for listid in endpoints:
+                for revdep in self.runq_depends[listid]:
+                    weight[revdep] = weight[revdep] + weight[listid]
+                    deps_left[revdep] = deps_left[revdep] - 1
+                    if deps_left[revdep] == 0:
+                        next_points.append(revdep)
+                        task_done[revdep] = True
+            endpoints = next_points
+            if len(next_points) == 0:
+                break      
+
+        # Circular dependency sanity check
+        problem_tasks = []
+        for task in range(numTasks):
+            if task_done[task] is False or deps_left[task] != 0:
+                problem_tasks.append(task)
+                bb.msg.debug(2, bb.msg.domain.RunQueue, "Task %s (%s) is not buildable\n" % (task, self.get_user_idstring(task)))
+                bb.msg.debug(2, bb.msg.domain.RunQueue, "(Complete marker was %s and the remaining dependency count was %s)\n\n" % (task_done[task], deps_left[task]))
+
+        if problem_tasks:
+            message = "Unbuildable tasks were found.\n"
+            message = message + "These are usually caused by circular dependencies and any circular dependency chains found will be printed below. Increase the debug level to see a list of unbuildable tasks.\n\n"
+            message = message + "Identifying dependency loops (this may take a short while)...\n"
+            bb.msg.error(bb.msg.domain.RunQueue, message)
+
+            msgs = self.circular_depchains_handler(problem_tasks)
+
+            message = "\n"
+            for msg in msgs:
+                message = message + msg
+            bb.msg.fatal(bb.msg.domain.RunQueue, message)
+
+        return weight
+
     def prepare_runqueue(self):
         """
         Turn a set of taskData into a RunQueue and compute data needed 
@@ -107,9 +333,7 @@ class RunQueue:
         """
 
         depends = []
-        runq_weight1 = []
         runq_build = []
-        runq_done = []
 
         taskData = self.taskData
 
@@ -283,11 +507,8 @@ class RunQueue:
             self.runq_task.append(taskData.tasks_name[task])
             self.runq_depends.append(Set(depends))
             self.runq_revdeps.append(Set())
-            self.runq_weight.append(0)
 
-            runq_weight1.append(0)
             runq_build.append(0)
-            runq_done.append(0)
 
         # Step B - Mark all active tasks
         #
@@ -349,10 +570,7 @@ class RunQueue:
                 del self.runq_fnid[listid-delcount]
                 del self.runq_task[listid-delcount]
                 del self.runq_depends[listid-delcount]
-                del self.runq_weight[listid-delcount]
-                del runq_weight1[listid-delcount]
                 del runq_build[listid-delcount]
-                del runq_done[listid-delcount]
                 del self.runq_revdeps[listid-delcount]
                 delcount = delcount + 1
                 maps.append(-1)
@@ -394,51 +612,23 @@ class RunQueue:
         for listid in range(len(self.runq_fnid)):
             revdeps = self.runq_revdeps[listid]
             if len(revdeps) == 0:
-                runq_done[listid] = 1
-                self.runq_weight[listid] = 1
                 endpoints.append(listid)
             for dep in revdeps:
                 if dep in self.runq_depends[listid]:
                     #self.dump_data(taskData)
                     bb.msg.fatal(bb.msg.domain.RunQueue, "Task %s (%s) has circular dependency on %s (%s)" % (taskData.fn_index[self.runq_fnid[dep]], self.runq_task[dep] , taskData.fn_index[self.runq_fnid[listid]], self.runq_task[listid]))
-            runq_weight1[listid] = len(revdeps)
 
         bb.msg.note(2, bb.msg.domain.RunQueue, "Compute totals (have %s endpoint(s))" % len(endpoints))
 
-        while 1:
-            next_points = []
-            for listid in endpoints:
-                for revdep in self.runq_depends[listid]:
-                    self.runq_weight[revdep] = self.runq_weight[revdep] + self.runq_weight[listid]
-                    runq_weight1[revdep] = runq_weight1[revdep] - 1
-                    if runq_weight1[revdep] == 0:
-                        next_points.append(revdep)
-                        runq_done[revdep] = 1
-            endpoints = next_points
-            if len(next_points) == 0:
-                break           
+        # Calculate task weights 
+        # Check of higher length circular dependencies
+        self.runq_weight = self.calculate_task_weights(endpoints)
 
-        # Sanity Checks
-        for task in range(len(self.runq_fnid)):
-            if runq_done[task] == 0:
-                seen = []
-                deps_seen = []
-                def print_chain(taskid, finish):
-                    seen.append(taskid)
-                    for revdep in self.runq_revdeps[taskid]:
-                        if runq_done[revdep] == 0 and revdep not in seen and not finish:
-                            bb.msg.error(bb.msg.domain.RunQueue, "Task %s (%s) (depends: %s)" % (revdep, self.get_user_idstring(revdep), self.runq_depends[revdep]))
-                            if revdep in deps_seen:
-                                bb.msg.error(bb.msg.domain.RunQueue, "Chain ends at Task %s (%s)" % (revdep, self.get_user_idstring(revdep)))
-                                finish = True
-                                return
-                            for dep in self.runq_depends[revdep]:
-                                deps_seen.append(dep)
-                            print_chain(revdep, finish)
-                print_chain(task, False)
-                bb.msg.fatal(bb.msg.domain.RunQueue, "Task %s (%s) not processed!\nThis is probably a circular dependency (the chain might be printed above)." % (task, self.get_user_idstring(task)))
-            if runq_weight1[task] != 0:
-                bb.msg.fatal(bb.msg.domain.RunQueue, "Task %s (%s) count not zero!" % (task, self.get_user_idstring(task)))
+        # Decide what order to execute the tasks in, pick a scheduler
+        # FIXME - Allow user selection
+        #self.sched = RunQueueScheduler(self)
+        self.sched = RunQueueSchedulerSpeed(self)
+        #self.sched = RunQueueSchedulerCompletion(self)
 
         # Sanity Check - Check for multiple tasks building the same provider
         prov_list = {}
@@ -461,21 +651,6 @@ class RunQueue:
         #if error:
         #    bb.msg.fatal(bb.msg.domain.RunQueue, "Corrupted metadata configuration detected, aborting...")
 
-
-        # Make a weight sorted map
-        from copy import deepcopy
-
-        sortweight = deepcopy(self.runq_weight)
-        sortweight.sort()
-        copyweight = deepcopy(self.runq_weight)
-        self.prio_map = []
-
-        for weight in sortweight:
-            idx = copyweight.index(weight)
-            self.prio_map.append(idx)
-            copyweight[idx] = -1
-        self.prio_map.reverse()
-
         #self.dump_data(taskData)
 
         self.state = runQueueRunInit
@@ -591,18 +766,6 @@ class RunQueue:
         self.state = runQueueFailedCleanUp
         bb.event.fire(runQueueTaskFailed(task, self.stats, self, self.cfgData))
 
-    def get_next_task(self):
-        """
-        Return the id of the highest priority task that is buildable
-        """
-        for task1 in range(self.stats.total):
-            task = self.prio_map[task1]
-            if self.runq_running[task] == 1:
-                continue
-            if self.runq_buildable[task] == 1:
-                return task
-        return None
-
     def execute_runqueue_internal(self):
         """
         Run the tasks in a queue prepared by prepare_runqueue
@@ -615,7 +778,7 @@ class RunQueue:
         while True:
             task = None
             if self.stats.active < self.number_tasks:
-                task = self.get_next_task()
+                task = self.sched.next()
             if task is not None:
                 fn = self.taskData.fn_index[self.runq_fnid[task]]