CFS-devel, latest code

The latest sched-devel.git tree can be pulled from:

git://git.kernel.org/pub/scm/linux/kernel/git/mingo/linux-2.6-sched-devel.git

This is a quick iteration after yesterday's: a couple of group scheduling bugs were found/debugged and fixed by Srivatsa Vaddagiri and Mike Galbraith. There's also a yield fix from Dmitry Adamushko, a build fix from S.Ceglar Onur and Andrew Morton, a cleanup from Hiroshi Shimamoto and the usual stream of goodies from Peter Zijlstra. Rebased it to -rc8 as well.

there are no known regressions at the moment in the sched-devel.git codebase. (yay :)

Ingo

----------------------------------------->
the shortlog relative to 2.6.23-rc8:

Dmitry Adamushko (9):
sched: clean up struct load_stat
sched: clean up schedstat block in dequeue_entity()
sched: sched_setscheduler() fix
sched: add set_curr_task() calls
sched: do not keep current in the tree and get rid of sched_entity::fair_key
sched: optimize task_new_fair()
sched: simplify sched_class::yield_task()
sched: rework enqueue/dequeue_entity() to get rid of set_curr_task()
sched: yield fix

Hiroshi Shimamoto (1):
sched: clean up sched_fork()

Ingo Molnar (44):
sched: fix new-task method
sched: resched task in task_new_fair()
sched: small sched_debug cleanup
sched: debug: track maximum 'slice'
sched: uniform tunings
sched: use constants if !CONFIG_SCHED_DEBUG
sched: remove stat_gran
sched: remove precise CPU load
sched: remove precise CPU load calculations #2
sched: track cfs_rq->curr on !group-scheduling too
sched: cleanup: simplify cfs_rq_curr() methods
sched: uninline __enqueue_entity()/__dequeue_entity()
sched: speed up update_load_add/_sub()
sched: clean up calc_weighted()
sched: introduce se->vruntime
sched: move sched_feat() definitions
sched: optimize vruntime based scheduling
sched: simplify check_preempt() methods
sched: wakeup granularity fix
sched: add se->vruntime debugging
sched: add more vruntime statistics
sched: debug: update exec_clock only when SCHED_DEBUG
sched: remove wait_runtime limit
sched: remove wait_runtime fields and features
sched: x86: allow single-depth wchan output
sched: fix delay accounting performance regression
sched: prettify /proc/sched_debug output
sched: enhance debug output
sched: kernel/sched_fair.c whitespace cleanups
sched: fair-group sched, cleanups
sched: enable CONFIG_FAIR_GROUP_SCHED=y by default sched debug: BKL usage statistics
sched: remove unneeded tunables
sched debug: print settings
sched debug: more width for parameter printouts
sched: entity_key() fix
sched: remove condition from set_task_cpu()
sched: remove last_min_vruntime effect
sched: undo some of the recent changes
sched: fix place_entity()
sched: fix sched_fork()
sched: remove set_leftmost()
sched: clean up schedstats, cnt -> count
sched: cleanup, remove stale comment

Matthias Kaehlcke (1):
sched: use list_for_each_entry_safe() in __wake_up_common()

Mike Galbraith (2):
sched: fix SMP migration latencies
sched: fix formatting of /proc/sched_debug

Peter Zijlstra (12):
sched: simplify SCHED_FEAT_* code
sched: new task placement for vruntime
sched: simplify adaptive latency
sched: clean up new task placement
sched: add tree based averages
sched: handle vruntime overflow
sched: better min_vruntime tracking
sched: add vslice
sched debug: check spread
sched: max_vruntime() simplification
sched: clean up min_vruntime use
sched: speed up and simplify vslice calculations

S.Ceglar Onur (1):
sched debug: BKL usage statistics, fix

Srivatsa Vaddagiri (9):
sched: group-scheduler core
sched: revert recent removal of set_curr_task()
sched: fix minor bug in yield
sched: print nr_running and load in /proc/sched_debug
sched: print &rq->cfs stats
sched: clean up code under CONFIG_FAIR_GROUP_SCHED
sched: add fair-user scheduler
sched: group scheduler wakeup latency fix
sched: group scheduler SMP migration fix

 arch/i386/Kconfig       |   11 
 fs/proc/base.c          |    2 
 include/linux/sched.h   |   55 ++-
 init/Kconfig            |   21 +
 kernel/delayacct.c      |    2 
 kernel/sched.c          |  577 +++++++++++++++++++++++++-------------
 kernel/sched_debug.c    |  250 +++++++++++-----
 kernel/sched_fair.c     |  718 +++++++++++++++++-------------------------------
 kernel/sched_idletask.c |    5 
 kernel/sched_rt.c       |   12 
 kernel/sched_stats.h    |   28 -
 kernel/sysctl.c         |   31 --
 kernel/user.c           |   43 ++
 13 files changed, 954 insertions(+), 801 deletions(-)

This is required for it to compile.

---
 include/linux/sched.h |    1 +
 1 files changed, 1 insertion(+)

Index: current/include/linux/sched.h
===================================================================
@@ -1404,6 +1404,7 @@ extern unsigned int sysctl_sched_wakeup_
 extern unsigned int sysctl_sched_batch_wakeup_granularity;
 extern unsigned int sysctl_sched_child_runs_first;
 extern unsigned int sysctl_sched_features;
+extern unsigned int sysctl_sched_nr_latency;
 #endif
 
 extern unsigned int sysctl_sched_compat_yield;

Regards,
vatsa

Ingo,
Few more patches follow on top of this latest sched-devel tree.

Pls consider for inclusion.

--
Regards,
vatsa

Fix coding style issues reported by Randy Dunlap and others

---
 init/Kconfig         |   14 +++++++-------
 kernel/sched_debug.c |    8 ++------
 2 files changed, 9 insertions(+), 13 deletions(-)

Index: current/init/Kconfig
===================================================================
@@ -282,11 +282,11 @@ config CPUSETS
 	  Say N if unsure.
 
 config FAIR_GROUP_SCHED
-	bool "Fair group cpu scheduler"
+	bool "Fair group CPU scheduler"
 	default y
 	depends on EXPERIMENTAL
 	help
-	  This feature lets cpu scheduler recognize task groups and control cpu
+	  This feature lets CPU scheduler recognize task groups and control CPU
 	  bandwidth allocation to such task groups.
 
 choice

@@ -294,11 +294,11 @@ choice
 	prompt "Basis for grouping tasks"
 	default FAIR_USER_SCHED
 
-	config FAIR_USER_SCHED
-		bool "user id"
-		help
-		  This option will choose userid as the basis for grouping
-		  tasks, thus providing equal cpu bandwidth to each user.
+config FAIR_USER_SCHED
+	bool "user id"
+	help
+	  This option will choose userid as the basis for grouping
+	  tasks, thus providing equal CPU bandwidth to each user.
 
 endchoice
 

Index: current/kernel/sched_debug.c
===================================================================
@@ -239,11 +239,7 @@ static int
 root_user_share_read_proc(char *page, char **start, off_t off, int count,
 				 int *eof, void *data)
 {
-	int len;
-
-	len = sprintf(page, "%d\n", init_task_grp_load);
-
-	return len;
+	return sprintf(page, "%d\n", init_task_grp_load);
 }
 
 static int

@@ -297,7 +293,7 @@ static int __init init_sched_debug_procf
 	pe->proc_fops = &sched_debug_fops;
 
 #ifdef CONFIG_FAIR_USER_SCHED
-	pe = create_proc_entry("root_user_share", 0644, NULL);
+	pe = create_proc_entry("root_user_cpu_share", 0644, NULL);
 	if (!pe)
 		return -ENOMEM;
 

Regards,
vatsa

Umm, that's not a debugging thing. It appears to be a tunable allowing you to configure what percentage of the total CPU that UID 0 gets which is likely to be useful to configure on production systems; at least until better group-scheduling tools are produced.

Cheers,
Kyle Moffett

Hi Ingo,

This is not debug stuff. It is a tunable to give the root user more weight with respect to the other users.

regards,
Dhaval

Recent fix to check_preempt_wakeup() to check for preemption at higher levels caused a size bloat for !CONFIG_FAIR_GROUP_SCHED.

Fix the problem.

  42277   10598     320   53195    cfcb kernel/sched.o-before_this_patch
  42216   10598     320   53134    cf8e kernel/sched.o-after_this_patch

---
 kernel/sched_fair.c |   43 +++++++++++++++++++++++++------------------
 1 files changed, 25 insertions(+), 18 deletions(-)

Index: current/kernel/sched_fair.c
===================================================================
@@ -640,15 +640,21 @@ static inline struct cfs_rq *cpu_cfs_rq(
 #define for_each_leaf_cfs_rq(rq, cfs_rq) \
 	list_for_each_entry(cfs_rq, &rq->leaf_cfs_rq_list, leaf_cfs_rq_list)
 
-/* Do the two (enqueued) tasks belong to the same group ? */
-static inline int is_same_group(struct task_struct *curr, struct task_struct *p)
+/* Do the two (enqueued) entities belong to the same group ? */
+static inline int
+is_same_group(struct sched_entity *se, struct sched_entity *pse)
 {
-	if (curr->se.cfs_rq == p->se.cfs_rq)
+	if (se->cfs_rq == pse->cfs_rq)
 		return 1;
 
 	return 0;
 }
 
+static inline struct sched_entity *parent_entity(struct sched_entity *se)
+{
+	return se->parent;
+}
+
 #else	/* CONFIG_FAIR_GROUP_SCHED */
 
 #define for_each_sched_entity(se) \

@@ -681,11 +687,17 @@ static inline struct cfs_rq *cpu_cfs_rq(
 #define for_each_leaf_cfs_rq(rq, cfs_rq) \
 		for (cfs_rq = &rq->cfs; cfs_rq; cfs_rq = NULL)
 
-static inline int is_same_group(struct task_struct *curr, struct task_struct *p)
+static inline int
+is_same_group(struct sched_entity *se, struct sched_entity *pse)
 {
 	return 1;
 }
 
+static inline struct sched_entity *parent_entity(struct sched_entity *se)
+{
+	return NULL;
+}
+
 #endif	/* CONFIG_FAIR_GROUP_SCHED */
 
 /*

@@ -775,8 +787,9 @@ static void yield_task_fair(struct rq *r
 static void check_preempt_wakeup(struct rq *rq, struct task_struct *p)
 {
 	struct task_struct *curr = rq->curr;
-	struct cfs_rq *cfs_rq = task_cfs_rq(curr), *pcfs_rq;
+	struct cfs_rq *cfs_rq = task_cfs_rq(curr);
 	struct sched_entity *se = &curr->se, *pse = &p->se;
+	s64 delta;
 
 	if (unlikely(rt_prio(p->prio))) {
 		update_rq_clock(rq);

@@ -785,21 +798,15 @@ static void check_preempt_wakeup(struct 
 		return;
 	}
 
-	for_each_sched_entity(se) {
-		cfs_rq = cfs_rq_of(se);
-		pcfs_rq = cfs_rq_of(pse);
+	while (!is_same_group(se, pse)) {
+		se = parent_entity(se);
+		pse = parent_entity(pse);
+	}
 
-		if (cfs_rq == pcfs_rq) {
-			s64 delta = se->vruntime - pse->vruntime;
+	delta = se->vruntime - pse->vruntime;
 
-			if (delta > (s64)sysctl_sched_wakeup_granularity)
-				resched_task(curr);
-			break;
-		}
-#ifdef CONFIG_FAIR_GROUP_SCHED
-		pse = pse->parent;
-#endif
-	}
+	if (delta > (s64)sysctl_sched_wakeup_granularity)
+		resched_task(curr);
 }
 
 static struct task_struct *pick_next_task_fair(struct rq *rq)

Regards,
vatsa

There is a possibility that because of task of a group moving from one cpu to another, it may gain more cpu time that desired. See
http://marc.info/?l=linux-kernel&m=119073197730334 for details.

This is an attempt to fix that problem. Basically it simulates dequeue of higher level entities as if they are going to sleep. Similarly it simulate wakeup of higher level entities as if they are waking up from sleep.

---
 kernel/sched_fair.c |    2 ++
 1 files changed, 2 insertions(+)

Index: current/kernel/sched_fair.c
===================================================================
@@ -715,6 +715,7 @@ static void enqueue_task_fair(struct rq 
 			break;
 		cfs_rq = cfs_rq_of(se);
 		enqueue_entity(cfs_rq, se, wakeup);
+		wakeup = 1;
 	}
 }
 

@@ -734,6 +735,7 @@ static void dequeue_task_fair(struct rq 
 		/* Don't dequeue parent if it has other entities besides us */
 		if (cfs_rq->load.weight)
 			break;
+		sleep = 1;
 	}
 }
 

Regards,
vatsa

thanks, applied.

Ingo