This build bug:

 In file included from kernel/sched.c:1765:
 kernel/sched_rt.c: In function ‘has_pushable_tasks’:
 kernel/sched_rt.c:1069: error: ‘struct rt_rq’ has no member named ‘pushable_tasks’
 kernel/sched_rt.c: In function ‘pick_next_task_rt’:
 kernel/sched_rt.c:1084: error: ‘struct rq’ has no member named ‘post_schedule’

Triggers because both pushable_tasks and post_schedule are
SMP-only fields.

Move pushable_tasks() to the SMP section and #ifdef the post_schedule use.

Cc: Gregory Haskins <ghaskins@novell.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <20090729150422.17691.55590.stgit@dev.haskins.net>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

A frequent mistake appears to be to call task_of() on a
scheduler entity that is not actually a task, which can result
in a wild pointer.

Add a check to catch these mistakes.
Suggested-by: NIngo Molnar <mingo@elte.hu>
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Reflect "active" cpus in the rq->rd->online field, instead of
the online_map.

The motivation is that things that use the root-domain code
(such as cpupri) only care about cpus classified as "active"
anyway. By synchronizing the root-domain state with the active
map, we allow several optimizations.

For instance, we can remove an extra cpumask_and from the
scheduler hotpath by utilizing rq->rd->online (since it is now
a cached version of cpu_active_map & rq->rd->span).
Signed-off-by: NGregory Haskins <ghaskins@novell.com>
Acked-by: NPeter Zijlstra <peterz@infradead.org>
Acked-by: NMax Krasnyansky <maxk@qualcomm.com>
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <20090730145723.25226.24493.stgit@dev.haskins.net>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

We currently have an explicit "needs_post" vtable method which
returns a stack variable for whether we should later run
post-schedule.  This leads to an awkward exchange of the
variable as it bubbles back up out of the context switch. Peter
Zijlstra observed that this information could be stored in the
run-queue itself instead of handled on the stack.

Therefore, we revert to the method of having context_switch
return void, and update an internal rq->post_schedule variable
when we require further processing.

In addition, we fix a race condition where we try to access
current->sched_class without holding the rq->lock.  This is
technically racy, as the sched-class could change out from
under us.  Instead, we reference the per-rq post_schedule
variable with the runqueue unlocked, but with preemption
disabled to see if we need to reacquire the rq->lock.

Finally, we clean the code up slightly by removing the #ifdef
CONFIG_SMP conditionals from the schedule() call, and implement
some inline helper functions instead.

This patch passes checkpatch, and rt-migrate.
Signed-off-by: NGregory Haskins <ghaskins@novell.com>
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <20090729150422.17691.55590.stgit@dev.haskins.net>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Fixes an easily triggerable BUG() when setting process affinities.

Make sure to count the number of migratable tasks in the same place:
the root rt_rq. Otherwise the number doesn't make sense and we'll hit
the BUG in set_cpus_allowed_rt().

Also, make sure we only count tasks, not groups (this is probably
already taken care of by the fact that rt_se->nr_cpus_allowed will be 0
for groups, but be more explicit)
Tested-by: NThomas Gleixner <tglx@linutronix.de>
CC: stable@kernel.org
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: NGregory Haskins <ghaskins@novell.com>
LKML-Reference: <1247067476.9777.57.camel@twins>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

These are defined as static cpumask_var_t so if MAXSMP is not used,
they are cleared already.  Avoid surprises when MAXSMP is enabled.
Signed-off-by: NYinghai Lu <yinghai.lu@kernel.org>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>

Impact: cleanup

As pointed out by Steven Rostedt.  Since the arg in question is
unused, we simply change cpupri_find() to accept NULL.
Reported-by: NSteven Rostedt <srostedt@redhat.com>
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
LKML-Reference: <200903251501.22664.rusty@rustcorp.com.au>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

cpumask_and() only initializes nr_cpu_ids bits, so the (deprecated)
first_cpu() might find one of those uninitialized bits if nr_cpu_ids
is less than NR_CPUS (as it can be for CONFIG_CPUMASK_OFFSTACK).
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Ingo Molnar wrote:

> here's a new build failure with tip/sched/rt:
>
>   LD      .tmp_vmlinux1
> kernel/built-in.o: In function `set_curr_task_rt':
> sched.c:(.text+0x3675): undefined reference to `plist_del'
> kernel/built-in.o: In function `pick_next_task_rt':
> sched.c:(.text+0x37ce): undefined reference to `plist_del'
> kernel/built-in.o: In function `enqueue_pushable_task':
> sched.c:(.text+0x381c): undefined reference to `plist_del'

Eliminate the plist library kconfig and make it available
unconditionally.
Signed-off-by: NPeter Zijlstra <peterz@infradead.org>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Ingo found a build error in the scheduler when RT_GROUP_SCHED was
enabled, but SMP was not.  This patch rearranges the code such
that it is a little more streamlined and compiles under all permutations
of SMP, UP and RT_GROUP_SCHED.  It was boot tested on my 4-way x86_64
and it still passes preempt-test.
Signed-off-by: NGregory Haskins <ghaskins@novell.com>

Signed-off-by: NGregory Haskins <ghaskins@novell.com>

Impact: reduce stack usage, cleanup

Use a cpumask_var_t in find_lowest_rq() and clean up other old
cpumask_t calls.
Signed-off-by: NMike Travis <travis@sgi.com>

Impact: prevents panic from stack overflow on numa-capable machines.

Some of the "removal of stack hogs" changes in kernel/sched.c by using
node_to_cpumask_ptr were undone by the early cpumask API updates, and
causes a panic due to stack overflow.  This patch undoes those changes
by using cpumask_of_node() which returns a 'const struct cpumask *'.

In addition, cpu_coregoup_map is replaced with cpu_coregroup_mask further
reducing stack usage.  (Both of these updates removed 9 FIXME's!)

Also:
   Pick up some remaining changes from the old 'cpumask_t' functions to
   the new 'struct cpumask *' functions.

   Optimize memory traffic by allocating each percpu local_cpu_mask on the
   same node as the referring cpu.
Signed-off-by: NMike Travis <travis@sgi.com>
Acked-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

A panic was discovered by Chirag Jog where a BUG_ON sanity check
in the new "pushable_task" logic would trigger a panic under
certain circumstances:

http://lkml.org/lkml/2008/9/25/189

Gilles Carry discovered that the root cause was attributed to the
pushable_tasks list getting corrupted in the push_rt_task logic.
This was the result of a dropped rq lock in double_lock_balance
allowing a task in the process of being pushed to potentially migrate
away, and thus corrupt the pushable_tasks() list.

I traced back the problem as introduced by the pushable_tasks patch
that went in recently.   There is a "retry" path in push_rt_task()
that actually had a compound conditional to decide whether to
retry or exit.  I missed the meaning behind the rationale for the
virtual "if(!task) goto out;" portion of the compound statement and
thus did not handle it properly.  The new pushable_tasks logic
actually creates three distinct conditions:

1) an untouched and unpushable task should be dequeued
2) a migrated task where more pushable tasks remain should be retried
3) a migrated task where no more pushable tasks exist should exit

The original logic mushed (1) and (3) together, resulting in the
system dequeuing a migrated task (against an unlocked foreign run-queue
nonetheless).

To fix this, we get rid of the notion of "paranoid" and we support the
three unique conditions properly.  The paranoid feature is no longer
relevant with the new pushable logic (since pushable naturally limits
the loop) anyway, so lets just remove it.
Reported-By: NChirag Jog <chirag@linux.vnet.ibm.com>
Found-by: NGilles Carry <gilles.carry@bull.net>
Signed-off-by: NGregory Haskins <ghaskins@novell.com>

The RT scheduler employs a "push/pull" design to actively balance tasks
within the system (on a per disjoint cpuset basis).  When a task is
awoken, it is immediately determined if there are any lower priority
cpus which should be preempted.  This is opposed to the way normal
SCHED_OTHER tasks behave, which will wait for a periodic rebalancing
operation to occur before spreading out load.

When a particular RQ has more than 1 active RT task, it is said to
be in an "overloaded" state.  Once this occurs, the system enters
the active balancing mode, where it will try to push the task away,
or persuade a different cpu to pull it over.  The system will stay
in this state until the system falls back below the <= 1 queued RT
task per RQ.

However, the current implementation suffers from a limitation in the
push logic.  Once overloaded, all tasks (other than current) on the
RQ are analyzed on every push operation, even if it was previously
unpushable (due to affinity, etc).  Whats more, the operation stops
at the first task that is unpushable and will not look at items
lower in the queue.  This causes two problems:

1) We can have the same tasks analyzed over and over again during each
   push, which extends out the fast path in the scheduler for no
   gain.  Consider a RQ that has dozens of tasks that are bound to a
   core.  Each one of those tasks will be encountered and skipped
   for each push operation while they are queued.

2) There may be lower-priority tasks under the unpushable task that
   could have been successfully pushed, but will never be considered
   until either the unpushable task is cleared, or a pull operation
   succeeds.  The net result is a potential latency source for mid
   priority tasks.

This patch aims to rectify these two conditions by introducing a new
priority sorted list: "pushable_tasks".  A task is added to the list
each time a task is activated or preempted.  It is removed from the
list any time it is deactivated, made current, or fails to push.

This works because a task only needs to be attempted to push once.
After an initial failure to push, the other cpus will eventually try to
pull the task when the conditions are proper.  This also solves the
problem that we don't completely analyze all tasks due to encountering
an unpushable tasks.  Now every task will have a push attempted (when
appropriate).

This reduces latency both by shorting the critical section of the
rq->lock for certain workloads, and by making sure the algorithm
considers all eligible tasks in the system.

[ rostedt: added a couple more BUG_ONs ]
Signed-off-by: NGregory Haskins <ghaskins@novell.com>
Acked-by: NSteven Rostedt <srostedt@redhat.com>

We currently run class->post_schedule() outside of the rq->lock, which
means that we need to test for the need to post_schedule outside of
the lock to avoid a forced reacquistion.  This is currently not a problem
as we only look at rq->rt.overloaded.  However, we want to enhance this
going forward to look at more state to reduce the need to post_schedule to
a bare minimum set.  Therefore, we introduce a new member-func called
needs_post_schedule() which tests for the post_schedule condtion without
actually performing the work.  Therefore it is safe to call this
function before the rq->lock is released, because we are guaranteed not
to drop the lock at an intermediate point (such as what post_schedule()
may do).

We will use this later in the series

[ rostedt: removed paranoid BUG_ON ]
Signed-off-by: NGregory Haskins <ghaskins@novell.com>

There is no sense in wasting time trying to push a task away that
cannot move anywhere else.  We gain no benefit from trying to push
other tasks at this point, so if the task being woken up is non
migratable, just skip the whole operation.  This reduces overhead
in the wakeup path for certain tasks.
Signed-off-by: NGregory Haskins <ghaskins@novell.com>

We currently take the rq->lock for every cpu in an overload state during
pull_rt_tasks().  However, we now have enough information via the
highest_prio.[curr|next] fields to determine if there is any tasks of
interest to warrant the overhead of the rq->lock, before we actually take
it.  So we use this information to reduce lock contention during the
pull for the case where the source-rq doesnt have tasks that preempt
the current task.
Signed-off-by: NGregory Haskins <ghaskins@novell.com>

highest_prio.curr is actually a more accurate way to keep track of
the pull_rt_task() threshold since it is always up to date, even
if the "next" task migrates during double_lock.  Therefore, stop
looking at the "next" task object and simply use the highest_prio.curr.
Signed-off-by: NGregory Haskins <ghaskins@novell.com>

We will use this later in the series to reduce the amount of rq-lock
contention during a pull operation
Signed-off-by: NGregory Haskins <ghaskins@novell.com>

Move some common definitions up to the function prologe to simplify the
body logic.
Signed-off-by: NGregory Haskins <ghaskins@novell.com>

Impact: fix potential of rare crash

for_each_leaf_rt_rq() walks an RCU protected list (rq->leaf_rt_rq_list),
but doesn't use list_for_each_entry_rcu(). Fix this.
Signed-off-by: NBharata B Rao <bharata@linux.vnet.ibm.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Move double_lock_balance()/double_unlock_balance() higher to fix the following
with gcc-3.4.6:

   CC      kernel/sched.o
 In file included from kernel/sched.c:1605:
 kernel/sched_rt.c: In function `find_lock_lowest_rq':
 kernel/sched_rt.c:914: sorry, unimplemented: inlining failed in call to 'double_unlock_balance': function body not available
 kernel/sched_rt.c:1077: sorry, unimplemented: called from here
 make[2]: *** [kernel/sched.o] Error 1
Signed-off-by: NAlexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Impact: build fix for !CONFIG_SMP
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Acked-by: NMike Travis <travis@sgi.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Impact: Trivial API conversion

  NR_CPUS -> nr_cpu_ids
  cpumask_t -> struct cpumask
  sizeof(cpumask_t) -> cpumask_size()
  cpumask_a = cpumask_b -> cpumask_copy(&cpumask_a, &cpumask_b)

  cpu_set() -> cpumask_set_cpu()
  first_cpu() -> cpumask_first()
  cpumask_of_cpu() -> cpumask_of()
  cpus_* -> cpumask_*

There are some FIXMEs where we all archs to complete infrastructure
(patches have been sent):

  cpu_coregroup_map -> cpu_coregroup_mask
  node_to_cpumask* -> cpumask_of_node

There is also one FIXME where we pass an array of cpumasks to
partition_sched_domains(): this implies knowing the definition of
'struct cpumask' and the size of a cpumask.  This will be fixed in a
future patch.
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Impact: (future) size reduction for large NR_CPUS.

Dynamically allocating cpumasks (when CONFIG_CPUMASK_OFFSTACK) saves
space for small nr_cpu_ids but big CONFIG_NR_CPUS.  cpumask_var_t
is just a struct cpumask for !CONFIG_CPUMASK_OFFSTACK.
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Impact: stack reduction for large NR_CPUS

Dynamically allocating cpumasks (when CONFIG_CPUMASK_OFFSTACK) saves
stack space.

We simply return if the allocation fails: since we don't use it we
could just pass NULL to cpupri_find and have it handle that.
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Impact: (future) size reduction for large NR_CPUS.

Dynamically allocating cpumasks (when CONFIG_CPUMASK_OFFSTACK) saves
space for small nr_cpu_ids but big CONFIG_NR_CPUS.  cpumask_var_t
is just a struct cpumask for !CONFIG_CPUMASK_OFFSTACK.

def_root_domain is static, and so its masks are initialized with
alloc_bootmem_cpumask_var.  After that, alloc_cpumask_var is used.
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Impact: trivial wrap of member accesses

This eases the transition in the next patch.

We also get rid of a temporary cpumask in find_idlest_cpu() thanks to
for_each_cpu_and, and sched_balance_self() due to getting weight before
setting sd to NULL.
Signed-off-by: NRusty Russell <rusty@rustcorp.com.au>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

We have a test case which measures the variation in the amount of time
needed to perform a fixed amount of work on the preempt_rt kernel. We
started seeing deterioration in it's performance recently. The test
should never take more than 10 microseconds, but we started 5-10%
failure rate.

Using elimination method, we traced the problem to commit
1b12bbc7 (lockdep: re-annotate
scheduler runqueues).

When LOCKDEP is disabled, this patch only adds an additional function
call to double_unlock_balance(). Hence I inlined double_unlock_balance()
and the problem went away. Here is a patch to make this change.
Signed-off-by: NSripathi Kodi <sripathik@in.ibm.com>
Acked-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Impact: micro-optimization to SCHED_FIFO/RR scheduling

A very minor improvement, but might it be better to check sched_rt_runtime(rt_rq)
before taking the rt_runtime_lock?

Peter Zijlstra observes:

> Yes, I think its ok to do so.
>
> Like pointed out in the other thread, there are two races:
>
>  - sched_rt_runtime() going to RUNTIME_INF, and that will be handled
>    properly by sched_rt_runtime_exceeded()
>
>  - sched_rt_runtime() going to !RUNTIME_INF, and here we can miss an
>    accounting cycle, but I don't think that is something to worry too
>    much about.
Signed-off-by: NDimitri Sivanich <sivanich@sgi.com>
Acked-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

--

 kernel/sched_rt.c |    4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

a patch from Henrik Austad did this:

>> Do not declare select_task_rq as part of sched_class when CONFIG_SMP is
>> not set.

Peter observed:

> While a proper cleanup, could you do it by re-arranging the methods so
> as to not create an additional ifdef?

Do not declare select_task_rq and some other methods as part of sched_class
when CONFIG_SMP is not set.

Also gather those methods to avoid CONFIG_SMP mess.
Idea-by: NHenrik Austad <henrik.austad@gmail.com>
Signed-off-by: NLi Zefan <lizf@cn.fujitsu.com>
Acked-by: NPeter Zijlstra <peterz@infradead.org>
Acked-by: NHenrik Austad <henrik@austad.us>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

While working on the new version of the code for SCHED_SPORADIC I
noticed something strange in the present throttling mechanism. More
specifically in the throttling timer handler in sched_rt.c
(do_sched_rt_period_timer()) and in rt_rq_enqueue().

The problem is that, when unthrottling a runqueue, rt_rq_enqueue() only
asks for rescheduling if the runqueue has a sched_entity associated to
it (i.e., rt_rq->rt_se != NULL).
Now, if the runqueue is the root rq (which has a rt_se = NULL)
rescheduling does not take place, and it is delayed to some undefined
instant in the future.

This imply some random bandwidth usage by the RT tasks under throttling.
For instance, setting rt_runtime_us/rt_period_us = 950ms/1000ms an RT
task will get less than 95%. In our tests we got something varying
between 70% to 95%.
Using smaller time values, e.g., 95ms/100ms, things are even worse, and
I can see values also going down to 20-25%!!

The tests we performed are simply running 'yes' as a SCHED_FIFO task,
and checking the CPU usage with top, but we can investigate thoroughly
if you think it is needed.

Things go much better, for us, with the attached patch... Don't know if
it is the best approach, but it solved the issue for us.
Signed-off-by: NDario Faggioli <raistlin@linux.it>
Signed-off-by: NMichael Trimarchi <trimarchimichael@yahoo.it>
Acked-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Cc: <stable@kernel.org>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Hopefully clarify some of this code a little.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Lin Ming reported a 10% OLTP regression against 2.6.27-rc4.

The difference seems to come from different preemption agressiveness,
which affects the cache footprint of the workload and its effective
cache trashing.

Aggresively preempt a task if its avg overlap is very small, this should
avoid the task going to sleep and find it still running when we schedule
back to it - saving a wakeup.
Reported-by: NLin Ming <ming.m.lin@intel.com>
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Overview

This patch reworks the handling of POSIX CPU timers, including the
ITIMER_PROF, ITIMER_VIRT timers and rlimit handling.  It was put together
with the help of Roland McGrath, the owner and original writer of this code.

The problem we ran into, and the reason for this rework, has to do with using
a profiling timer in a process with a large number of threads.  It appears
that the performance of the old implementation of run_posix_cpu_timers() was
at least O(n*3) (where "n" is the number of threads in a process) or worse.
Everything is fine with an increasing number of threads until the time taken
for that routine to run becomes the same as or greater than the tick time, at
which point things degrade rather quickly.

This patch fixes bug 9906, "Weird hang with NPTL and SIGPROF."

Code Changes

This rework corrects the implementation of run_posix_cpu_timers() to make it
run in constant time for a particular machine.  (Performance may vary between
one machine and another depending upon whether the kernel is built as single-
or multiprocessor and, in the latter case, depending upon the number of
running processors.)  To do this, at each tick we now update fields in
signal_struct as well as task_struct.  The run_posix_cpu_timers() function
uses those fields to make its decisions.

We define a new structure, "task_cputime," to contain user, system and
scheduler times and use these in appropriate places:

struct task_cputime {
	cputime_t utime;
	cputime_t stime;
	unsigned long long sum_exec_runtime;
};

This is included in the structure "thread_group_cputime," which is a new
substructure of signal_struct and which varies for uniprocessor versus
multiprocessor kernels.  For uniprocessor kernels, it uses "task_cputime" as
a simple substructure, while for multiprocessor kernels it is a pointer:

struct thread_group_cputime {
	struct task_cputime totals;
};

struct thread_group_cputime {
	struct task_cputime *totals;
};

We also add a new task_cputime substructure directly to signal_struct, to
cache the earliest expiration of process-wide timers, and task_cputime also
replaces the it_*_expires fields of task_struct (used for earliest expiration
of thread timers).  The "thread_group_cputime" structure contains process-wide
timers that are updated via account_user_time() and friends.  In the non-SMP
case the structure is a simple aggregator; unfortunately in the SMP case that
simplicity was not achievable due to cache-line contention between CPUs (in
one measured case performance was actually _worse_ on a 16-cpu system than
the same test on a 4-cpu system, due to this contention).  For SMP, the
thread_group_cputime counters are maintained as a per-cpu structure allocated
using alloc_percpu().  The timer functions update only the timer field in
the structure corresponding to the running CPU, obtained using per_cpu_ptr().

We define a set of inline functions in sched.h that we use to maintain the
thread_group_cputime structure and hide the differences between UP and SMP
implementations from the rest of the kernel.  The thread_group_cputime_init()
function initializes the thread_group_cputime structure for the given task.
The thread_group_cputime_alloc() is a no-op for UP; for SMP it calls the
out-of-line function thread_group_cputime_alloc_smp() to allocate and fill
in the per-cpu structures and fields.  The thread_group_cputime_free()
function, also a no-op for UP, in SMP frees the per-cpu structures.  The
thread_group_cputime_clone_thread() function (also a UP no-op) for SMP calls
thread_group_cputime_alloc() if the per-cpu structures haven't yet been
allocated.  The thread_group_cputime() function fills the task_cputime
structure it is passed with the contents of the thread_group_cputime fields;
in UP it's that simple but in SMP it must also safely check that tsk->signal
is non-NULL (if it is it just uses the appropriate fields of task_struct) and,
if so, sums the per-cpu values for each online CPU.  Finally, the three
functions account_group_user_time(), account_group_system_time() and
account_group_exec_runtime() are used by timer functions to update the
respective fields of the thread_group_cputime structure.

Non-SMP operation is trivial and will not be mentioned further.

The per-cpu structure is always allocated when a task creates its first new
thread, via a call to thread_group_cputime_clone_thread() from copy_signal().
It is freed at process exit via a call to thread_group_cputime_free() from
cleanup_signal().

All functions that formerly summed utime/stime/sum_sched_runtime values from
from all threads in the thread group now use thread_group_cputime() to
snapshot the values in the thread_group_cputime structure or the values in
the task structure itself if the per-cpu structure hasn't been allocated.

Finally, the code in kernel/posix-cpu-timers.c has changed quite a bit.
The run_posix_cpu_timers() function has been split into a fast path and a
slow path; the former safely checks whether there are any expired thread
timers and, if not, just returns, while the slow path does the heavy lifting.
With the dedicated thread group fields, timers are no longer "rebalanced" and
the process_timer_rebalance() function and related code has gone away.  All
summing loops are gone and all code that used them now uses the
thread_group_cputime() inline.  When process-wide timers are set, the new
task_cputime structure in signal_struct is used to cache the earliest
expiration; this is checked in the fast path.

Performance

The fix appears not to add significant overhead to existing operations.  It
generally performs the same as the current code except in two cases, one in
which it performs slightly worse (Case 5 below) and one in which it performs
very significantly better (Case 2 below).  Overall it's a wash except in those
two cases.

I've since done somewhat more involved testing on a dual-core Opteron system.

Case 1: With no itimer running, for a test with 100,000 threads, the fixed
	kernel took 1428.5 seconds, 513 seconds more than the unfixed system,
	all of which was spent in the system.  There were twice as many
	voluntary context switches with the fix as without it.

Case 2: With an itimer running at .01 second ticks and 4000 threads (the most
	an unmodified kernel can handle), the fixed kernel ran the test in
	eight percent of the time (5.8 seconds as opposed to 70 seconds) and
	had better tick accuracy (.012 seconds per tick as opposed to .023
	seconds per tick).

Case 3: A 4000-thread test with an initial timer tick of .01 second and an
	interval of 10,000 seconds (i.e. a timer that ticks only once) had
	very nearly the same performance in both cases:  6.3 seconds elapsed
	for the fixed kernel versus 5.5 seconds for the unfixed kernel.

With fewer threads (eight in these tests), the Case 1 test ran in essentially
the same time on both the modified and unmodified kernels (5.2 seconds versus
5.8 seconds).  The Case 2 test ran in about the same time as well, 5.9 seconds
versus 5.4 seconds but again with much better tick accuracy, .013 seconds per
tick versus .025 seconds per tick for the unmodified kernel.

Since the fix affected the rlimit code, I also tested soft and hard CPU limits.

Case 4: With a hard CPU limit of 20 seconds and eight threads (and an itimer
	running), the modified kernel was very slightly favored in that while
	it killed the process in 19.997 seconds of CPU time (5.002 seconds of
	wall time), only .003 seconds of that was system time, the rest was
	user time.  The unmodified kernel killed the process in 20.001 seconds
	of CPU (5.014 seconds of wall time) of which .016 seconds was system
	time.  Really, though, the results were too close to call.  The results
	were essentially the same with no itimer running.

Case 5: With a soft limit of 20 seconds and a hard limit of 2000 seconds
	(where the hard limit would never be reached) and an itimer running,
	the modified kernel exhibited worse tick accuracy than the unmodified
	kernel: .050 seconds/tick versus .028 seconds/tick.  Otherwise,
	performance was almost indistinguishable.  With no itimer running this
	test exhibited virtually identical behavior and times in both cases.

In times past I did some limited performance testing.  those results are below.

On a four-cpu Opteron system without this fix, a sixteen-thread test executed
in 3569.991 seconds, of which user was 3568.435s and system was 1.556s.  On
the same system with the fix, user and elapsed time were about the same, but
system time dropped to 0.007 seconds.  Performance with eight, four and one
thread were comparable.  Interestingly, the timer ticks with the fix seemed
more accurate:  The sixteen-thread test with the fix received 149543 ticks
for 0.024 seconds per tick, while the same test without the fix received 58720
for 0.061 seconds per tick.  Both cases were configured for an interval of
0.01 seconds.  Again, the other tests were comparable.  Each thread in this
test computed the primes up to 25,000,000.

I also did a test with a large number of threads, 100,000 threads, which is
impossible without the fix.  In this case each thread computed the primes only
up to 10,000 (to make the runtime manageable).  System time dominated, at
1546.968 seconds out of a total 2176.906 seconds (giving a user time of
629.938s).  It received 147651 ticks for 0.015 seconds per tick, still quite
accurate.  There is obviously no comparable test without the fix.
Signed-off-by: NFrank Mayhar <fmayhar@google.com>
Cc: Roland McGrath <roland@redhat.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

On my tulsa x86-64 machine, kernel 2.6.25-rc5 couldn't boot randomly.

Basically, function __enable_runtime forgets to reset rt_rq->rt_throttled
to 0. When every cpu is up, per-cpu migration_thread is created and it runs
very fast, sometimes to mark the corresponding rt_rq->rt_throttled to 1 very
quickly. After all cpus are up, with below calling chain:

   sched_init_smp => arch_init_sched_domains => build_sched_domains => ...
=> cpu_attach_domain => rq_attach_root => set_rq_online => ...
=> _enable_runtime

_enable_runtime is called against every rt_rq again, so rt_rq->rt_time is
reset to 0, but rt_rq->rt_throttled might be still 1. Later on function
do_sched_rt_period_timer couldn't reset it, and all RT tasks couldn't be
scheduled to run on that cpu. here is RT task migration_thread which is
woken up when a task is migrated to another cpu.

Below patch fixes it against 2.6.27-rc5.
Signed-off-by: NZhang Yanmin <yanmin_zhang@linux.intel.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

It fixes an accounting bug where we would continue accumulating runtime
even though the bandwidth control is disabled. This would lead to very long
throttle periods once bandwidth control gets turned on again.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

When sysctl_sched_rt_runtime is set to something other than -1 and the
CONFIG_RT_GROUP_SCHED kernel parameter is NOT enabled, we get into a state
where we see one or more CPUs idling forvever even though there are
real-time
tasks in their rt runqueue that are able to run (no longer throttled).

The sequence is:

- A real-time task is running when the timer sets the rt runqueue
    to throttled, and the rt task is resched_task()ed and switched
    out, and idle is switched in since there are no non-rt tasks to
    run on that cpu.

- Eventually the do_sched_rt_period_timer() runs and un-throttles
    the rt runqueue, but we just exit the timer interrupt and go back
    to executing the idle task in the idle loop forever.

If we change the sched_rt_rq_enqueue() routine to use some of the code
from the CONFIG_RT_GROUP_SCHED enabled version of this same routine and
resched_task() the currently executing task (idle in our case) if it is
a lower priority task than the higher rt task in the now un-throttled
runqueue, the problem is no longer observed.
Signed-off-by: NJohn Blackwood <john.blackwood@ccur.com>
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

More extensive disable of bandwidth control. It allows sysctl_sched_rt_runtime
to disable full group bandwidth control.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>