summaryrefslogtreecommitdiffstats
path: root/bitbake
diff options
context:
space:
mode:
authorPatrick Ohly <patrick.ohly@intel.com>2017-01-19 20:35:16 +0100
committerRichard Purdie <richard.purdie@linuxfoundation.org>2017-01-20 11:52:52 +0000
commitbf0ea57e4a5dc5f47b2fb7623c3cea9a3ae7e4cb (patch)
tree586665013b75330b9638d242ba0b4b4d39ef7fb6 /bitbake
parent633697cce651ddb22c18083087531dce333c752c (diff)
downloadpoky-bf0ea57e4a5dc5f47b2fb7623c3cea9a3ae7e4cb.tar.gz
bitbake: runqueue.py: revised completion scheduler
The idea is that tasks which complete building a recipe (like do_package_qa) are more important than tasks which start building new recipes (do_fetch) or those which increase disk usage (do_compile). Therefore tasks get ordered like this (most important first, do_rm_work before do_build because the enhanced rm_work.bbclass was used): 1. ID /work/poky/meta/recipes-support/popt/popt_1.16.bb:do_build 2. ID /work/poky/meta/recipes-core/readline/readline_6.3.bb:do_build 3. ID /work/poky/meta/recipes-connectivity/libnss-mdns/libnss-mdns_0.10.bb:do_build ... 464. ID /work/poky/meta/recipes-sato/images/core-image-sato.bb:do_build 465. ID /work/poky/meta/recipes-graphics/xorg-proto/inputproto_2.3.2.bb:do_rm_work 466. ID /work/poky/meta/recipes-devtools/python/python3_3.5.2.bb:do_rm_work 467. ID /work/poky/meta/recipes-core/packagegroups/packagegroup-base.bb:do_rm_work ... 3620. ID virtual:native:/work/poky/meta/recipes-extended/pbzip2/pbzip2_1.1.13.bb:do_install 3621. ID /work/poky/meta/recipes-devtools/qemu/qemu-helper-native_1.0.bb:do_install 3622. ID /work/poky/meta/recipes-core/zlib/zlib_1.2.8.bb:do_compile_ptest_base 3623. ID /work/poky/meta/recipes-extended/bzip2/bzip2_1.0.6.bb:do_compile_ptest_base ... 3645. ID /work/poky/meta/recipes-support/libevent/libevent_2.0.22.bb:do_compile_ptest_base 3646. ID /work/poky/meta/recipes-core/busybox/busybox_1.24.1.bb:do_compile_ptest_base 3647. ID /work/poky/meta/recipes-kernel/linux/linux-yocto_4.8.bb:do_uboot_mkimage 3648. ID /work/poky/meta/recipes-kernel/linux/linux-yocto_4.8.bb:do_sizecheck 3649. ID /work/poky/meta/recipes-kernel/linux/linux-yocto_4.8.bb:do_strip 3650. ID /work/poky/meta/recipes-kernel/linux/linux-yocto_4.8.bb:do_compile_kernelmodules 3651. ID /work/poky/meta/recipes-kernel/linux/linux-yocto_4.8.bb:do_shared_workdir 3652. ID /work/poky/meta/recipes-kernel/linux/linux-yocto_4.8.bb:do_kernel_link_images 3653. ID /work/poky/meta/recipes-devtools/quilt/quilt-native_0.64.bb:do_compile 3654. ID /work/poky/meta/recipes-extended/texinfo-dummy-native/texinfo-dummy-native.bb:do_compile ... The order of the same task between different recipes is the same as with the speed scheduler, i.e. more important recipes come first. (Bitbake rev: 70e297e5c285ce0a02e9efd3117ff62cdc77ec12) Signed-off-by: Patrick Ohly <patrick.ohly@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Diffstat (limited to 'bitbake')
-rw-r--r--bitbake/lib/bb/runqueue.py125
1 files changed, 101 insertions, 24 deletions
diff --git a/bitbake/lib/bb/runqueue.py b/bitbake/lib/bb/runqueue.py
index 7e651a45d7..728b5fbb28 100644
--- a/bitbake/lib/bb/runqueue.py
+++ b/bitbake/lib/bb/runqueue.py
@@ -184,6 +184,18 @@ class RunQueueScheduler(object):
184 def newbuilable(self, task): 184 def newbuilable(self, task):
185 self.buildable.append(task) 185 self.buildable.append(task)
186 186
187 def describe_task(self, taskid):
188 result = 'ID %s' % taskid
189 if self.rev_prio_map:
190 result = result + (' pri %d' % self.rev_prio_map[taskid])
191 return result
192
193 def dump_prio(self, comment):
194 bb.debug(3, '%s (most important first):\n%s' %
195 (comment,
196 '\n'.join(['%d. %s' % (index + 1, self.describe_task(taskid)) for
197 index, taskid in enumerate(self.prio_map)])))
198
187class RunQueueSchedulerSpeed(RunQueueScheduler): 199class RunQueueSchedulerSpeed(RunQueueScheduler):
188 """ 200 """
189 A scheduler optimised for speed. The priority map is sorted by task weight, 201 A scheduler optimised for speed. The priority map is sorted by task weight,
@@ -213,35 +225,100 @@ class RunQueueSchedulerSpeed(RunQueueScheduler):
213 225
214class RunQueueSchedulerCompletion(RunQueueSchedulerSpeed): 226class RunQueueSchedulerCompletion(RunQueueSchedulerSpeed):
215 """ 227 """
216 A scheduler optimised to complete .bb files are quickly as possible. The 228 A scheduler optimised to complete .bb files as quickly as possible. The
217 priority map is sorted by task weight, but then reordered so once a given 229 priority map is sorted by task weight, but then reordered so once a given
218 .bb file starts to build, it's completed as quickly as possible. This works 230 .bb file starts to build, it's completed as quickly as possible by
219 well where disk space is at a premium and classes like OE's rm_work are in 231 running all tasks related to the same .bb file one after the after.
220 force. 232 This works well where disk space is at a premium and classes like OE's
233 rm_work are in force.
221 """ 234 """
222 name = "completion" 235 name = "completion"
223 236
224 def __init__(self, runqueue, rqdata): 237 def __init__(self, runqueue, rqdata):
225 RunQueueSchedulerSpeed.__init__(self, runqueue, rqdata) 238 super(RunQueueSchedulerCompletion, self).__init__(runqueue, rqdata)
226 239
227 #FIXME - whilst this groups all fns together it does not reorder the 240 # Extract list of tasks for each recipe, with tasks sorted
228 #fn groups optimally. 241 # ascending from "must run first" (typically do_fetch) to
229 242 # "runs last" (do_build). The speed scheduler prioritizes
230 basemap = copy.deepcopy(self.prio_map) 243 # tasks that must run first before the ones that run later;
231 self.prio_map = [] 244 # this is what we depend on here.
232 while (len(basemap) > 0): 245 task_lists = {}
233 entry = basemap.pop(0) 246 for taskid in self.prio_map:
234 self.prio_map.append(entry) 247 fn, taskname = taskid.rsplit(':', 1)
235 fn = fn_from_tid(entry) 248 task_lists.setdefault(fn, []).append(taskname)
236 todel = [] 249
237 for entry in basemap: 250 # Now unify the different task lists. The strategy is that
238 entry_fn = fn_from_tid(entry) 251 # common tasks get skipped and new ones get inserted after the
239 if entry_fn == fn: 252 # preceeding common one(s) as they are found. Because task
240 todel.append(basemap.index(entry)) 253 # lists should differ only by their number of tasks, but not
241 self.prio_map.append(entry) 254 # the ordering of the common tasks, this should result in a
242 todel.reverse() 255 # deterministic result that is a superset of the individual
243 for idx in todel: 256 # task ordering.
244 del basemap[idx] 257 all_tasks = []
258 for recipe, new_tasks in task_lists.items():
259 index = 0
260 old_task = all_tasks[index] if index < len(all_tasks) else None
261 for new_task in new_tasks:
262 if old_task == new_task:
263 # Common task, skip it. This is the fast-path which
264 # avoids a full search.
265 index += 1
266 old_task = all_tasks[index] if index < len(all_tasks) else None
267 else:
268 try:
269 index = all_tasks.index(new_task)
270 # Already present, just not at the current
271 # place. We re-synchronized by changing the
272 # index so that it matches again. Now
273 # move on to the next existing task.
274 index += 1
275 old_task = all_tasks[index] if index < len(all_tasks) else None
276 except ValueError:
277 # Not present. Insert before old_task, which
278 # remains the same (but gets shifted back).
279 all_tasks.insert(index, new_task)
280 index += 1
281 bb.debug(3, 'merged task list: %s' % all_tasks)
282
283 # Now reverse the order so that tasks that finish the work on one
284 # recipe are considered more imporant (= come first). The ordering
285 # is now so that do_build is most important.
286 all_tasks.reverse()
287
288 # Group tasks of the same kind before tasks of less important
289 # kinds at the head of the queue (because earlier = lower
290 # priority number = runs earlier), while preserving the
291 # ordering by recipe. If recipe foo is more important than
292 # bar, then the goal is to work on foo's do_populate_sysroot
293 # before bar's do_populate_sysroot and on the more important
294 # tasks of foo before any of the less important tasks in any
295 # other recipe (if those other recipes are more important than
296 # foo).
297 #
298 # All of this only applies when tasks are runable. Explicit
299 # dependencies still override this ordering by priority.
300 #
301 # Here's an example why this priority re-ordering helps with
302 # minimizing disk usage. Consider a recipe foo with a higher
303 # priority than bar where foo DEPENDS on bar. Then the
304 # implicit rule (from base.bbclass) is that foo's do_configure
305 # depends on bar's do_populate_sysroot. This ensures that
306 # bar's do_populate_sysroot gets done first. Normally the
307 # tasks from foo would continue to run once that is done, and
308 # bar only gets completed and cleaned up later. By ordering
309 # bar's task that depend on bar's do_populate_sysroot before foo's
310 # do_configure, that problem gets avoided.
311 task_index = 0
312 self.dump_prio('original priorities')
313 for task in all_tasks:
314 for index in range(task_index, self.numTasks):
315 taskid = self.prio_map[index]
316 taskname = taskid.rsplit(':', 1)[1]
317 if taskname == task:
318 del self.prio_map[index]
319 self.prio_map.insert(task_index, taskid)
320 task_index += 1
321 self.dump_prio('completion priorities')
245 322
246class RunTaskEntry(object): 323class RunTaskEntry(object):
247 def __init__(self): 324 def __init__(self):