- fix UP lockup
- another set of UP/SMP cleanups and simplifications
Signed-off-by: NFrank Mayhar <fmayhar@google.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

This is the second resubmission of the posix timer rework patch, posted
a few days ago.

This includes the changes from the previous resubmittion, which addressed
Oleg Nesterov's comments, removing the RCU stuff from the patch and
un-inlining the thread_group_cputime() function for SMP.

In addition, per Ingo Molnar it simplifies the UP code, consolidating much
of it with the SMP version and depending on lower-level SMP/UP handling to
take care of the differences.

It also cleans up some UP compile errors, moves the scheduler stats-related
macros into kernel/sched_stats.h, cleans up a merge error in
kernel/fork.c and has a few other minor fixes and cleanups as suggested
by Oleg and Ingo. Thanks for the review, guys.
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>

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>

What I realized recently is that calling rebuild_sched_domains() in
arch_reinit_sched_domains() by itself is not enough when cpusets are enabled.
partition_sched_domains() code is trying to avoid unnecessary domain rebuilds
and will not actually rebuild anything if new domain masks match the old ones.

What this means is that doing
     echo 1 > /sys/devices/system/cpu/sched_mc_power_savings
on a system with cpusets enabled will not take affect untill something changes
in the cpuset setup (ie new sets created or deleted).

This patch fixes restore correct behaviour where domains must be rebuilt in
order to enable MC powersaving flags.

Test on quad-core Core2 box with both CONFIG_CPUSETS and !CONFIG_CPUSETS.
Also tested on dual-core Core2 laptop. Lockdep is happy and things are working
as expected.
Signed-off-by: NMax Krasnyansky <maxk@qualcomm.com>
Tested-by: NVaidyanathan Srinivasan <svaidy@linux.vnet.ibm.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Spencer reported a problem where utime and stime were going negative despite
the fixes in commit b27f03d4. The suspected
reason for the problem is that signal_struct maintains it's own utime and
stime (of exited tasks), these are not updated using the new task_utime()
routine, hence sig->utime can go backwards and cause the same problem
to occur (sig->utime, adds tsk->utime and not task_utime()). This patch
fixes the problem

TODO: using max(task->prev_utime, derived utime) works for now, but a more
generic solution is to implement cputime_max() and use the cputime_gt()
function for comparison.

Reported-by: spencer@bluehost.com
Signed-off-by: NBalbir Singh <balbir@linux.vnet.ibm.com>
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

David reported that his Niagra spend a little too much time in
tg_shares_up(), which considering he has a large cpu count makes sense.

So scale the ratelimit value with the number of cpus like we do for
other controls as well.
Reported-by: NDavid Miller <davem@davemloft.net>
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

m68k fails to build with these functions inlined in completion.h.  Move
them out of line into sched.c and export them to avoid this problem.
Signed-off-by: NDave Chinner <david@fromorbit.com>
Cc: Geert Uytterhoeven <geert@linux-m68k.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

With 2.6.27-rc3, I hit a kernel panic when running volanoMark on my
new x86_64 machine. I also hit it with other 2.6.27-rc kernels.
See below log.

Basically, function walk_tg_tree and sched_create_group have a race
between accessing and initiating tg->children. Below patch fixes it
by moving tg->children initiation to the front of linking tg->siblings
to parent->children.

{----------------panic log------------}

BUG: unable to handle kernel NULL pointer dereference at 0000000000000000
IP: [<ffffffff802292ab>] walk_tg_tree+0x45/0x7f
PGD 1be1c4067 PUD 1bdd8d067 PMD 0
Oops: 0000 [1] SMP
CPU 11
Modules linked in: igb
Pid: 22979, comm: java Not tainted 2.6.27-rc3 #1
RIP: 0010:[<ffffffff802292ab>]  [<ffffffff802292ab>] walk_tg_tree+0x45/0x7f
RSP: 0018:ffff8801bfbbbd18  EFLAGS: 00010083
RAX: 0000000000000000 RBX: ffff8800be0dce40 RCX: ffffffffffffffc0
RDX: ffff880102c43740 RSI: 0000000000000000 RDI: ffff8800be0dce40
RBP: ffff8801bfbbbd48 R08: ffff8800ba437bc8 R09: 0000000000001f40
R10: ffff8801be812100 R11: ffffffff805fdf44 R12: ffff880102c43740
R13: 0000000000000000 R14: ffffffff8022cf0f R15: ffffffff8022749f
FS:  00000000568ac950(0063) GS:ffff8801bfa26d00(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 0000000000000000 CR3: 00000001bd848000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process java (pid: 22979, threadinfo ffff8801b145a000, task ffff8801bf18e450)
Stack:  0000000000000001 ffff8800ba5c8d60 0000000000000001 0000000000000001
 ffff8800bad1ccb8 0000000000000000 ffff8801bfbbbd98 ffffffff8022ed37
 0000000000000001 0000000000000286 ffff8801bd5ee180 ffff8800ba437bc8
Call Trace:
 <IRQ>  [<ffffffff8022ed37>] try_to_wake_up+0x71/0x24c
 [<ffffffff80247177>] autoremove_wake_function+0x9/0x2e
 [<ffffffff80228039>] ? __wake_up_common+0x46/0x76
 [<ffffffff802296d5>] __wake_up+0x38/0x4f
 [<ffffffff806169cc>] tcp_v4_rcv+0x380/0x62e
Signed-off-by: NZhang Yanmin <yanmin_zhang@linux.intel.com>
Acked-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Instead of using a per-rq lock class, use the regular nesting operations.

However, take extra care with double_lock_balance() as it can release the
already held rq->lock (and therefore change its nesting class).

So what can happen is:

 spin_lock(rq->lock);	// this rq subclass 0

 double_lock_balance(rq, other_rq);
   // release rq
   // acquire other_rq->lock subclass 0
   // acquire rq->lock subclass 1

 spin_unlock(other_rq->lock);

leaving you with rq->lock in subclass 1

So a subsequent double_lock_balance() call can try to nest a subclass 1
lock while already holding a subclass 1 lock.

Fix this by introducing double_unlock_balance() which releases the other
rq's lock, but also re-sets the subclass for this rq's lock to 0.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

The "user" parameter to __sched_setscheduler indicates whether the
change is being done on behalf of a user process or not.  If not, we
shouldn't apply any permissions checks, so don't call
security_task_setscheduler().
Signed-off-by: NJeremy Fitzhardinge <jeremy@goop.org>
Tested-by: NSteve Wise <swise@opengridcomputing.com>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: "Rafael J. Wysocki" <rjw@sisk.pl>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

While thinking about David's graph walk lockdep patch it _finally_
dawned on me that there is no reason we have a lock class per cpu ...

Sorry for being dense :-/

The below changes the annotation from a lock class per cpu, to a single
nested lock, as the scheduler never holds more that 2 rq locks at a time
anyway.

If there was code requiring holding all rq locks this would not work and
the original annotation would be the only option, but that not being the
case, this is a much lighter one.

Compiles and boots on a 2-way x86_64.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Cc: David Miller <davem@davemloft.net>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

They are really class devices, but were incorrectly declared.  This
leads to crashes with the recent changes that makes non normal sysdevs
use a different prototype.

[akpm@linux-foundation.org: coding-style fixes]
Signed-off-by: NAndi Kleen <ak@linux.intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Pierre Ossman <drzeus-list@drzeus.cx>
Cc: Greg Kroah-Hartman <gregkh@suse.de>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Test runtime rather than period
Signed-off-by: NRoel Kluin <roel.kluin@gmail.com>
Acked-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

This extends wait_task_inactive() with a new argument so it can be used in
a "soft" mode where it will check for the task changing state unexpectedly
and back off.  There is no change to existing callers.  This lays the
groundwork to allow robust, noninvasive tracing that can try to sample a
blocked thread but back off safely if it wakes up.
Signed-off-by: NRoland McGrath <roland@redhat.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Reviewed-by: NIngo Molnar <mingo@elte.hu>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

A previous patch added the early_initcall(), to allow a cleaner hooking of
pre-SMP initcalls.  Now we remove the older interface, converting all
existing users to the new one.

[akpm@linux-foundation.org: cleanups]
[akpm@linux-foundation.org: build fix]
[kosaki.motohiro@jp.fujitsu.com: warning fix]
[kosaki.motohiro@jp.fujitsu.com: warning fix]
Signed-off-by: NEduard - Gabriel Munteanu <eduard.munteanu@linux360.ro>
Cc: Tom Zanussi <tzanussi@gmail.com>
Signed-off-by: NKOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Adapt acct_update_integrals() to include user time when calculating the time
difference.  The units of acct_rss_mem1 and acct_vm_mem1 are also changed from
pages-jiffies to pages-usecs to avoid calling jiffies_to_usecs() in
xacct_add_tsk() which might overflow.
Signed-off-by: NJonathan Lim <jlim@sgi.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

This allow to dynamically generate attributes and share show/store
functions between attributes. Right now most attributes are generated
by special macros and lots of duplicated code. With the attribute
passed it's instead possible to attach some data to the attribute
and then use that in shared low level functions to do different things.

I need this for the dynamically generated bank attributes in the x86
machine check code, but it'll allow some further cleanups.

I converted all users in tree to the new show/store prototype. It's a single
huge patch to avoid unbisectable sections.

Runtime tested: x86-32, x86-64
Compiled only: ia64, powerpc
Not compile tested/only grep converted: sh, arm, avr32
Signed-off-by: NAndi Kleen <ak@linux.intel.com>
Signed-off-by: NGreg Kroah-Hartman <gregkh@suse.de>

Peter pointed out that hrtick_enabled() should use cpu_active().
Signed-off-by: NIngo Molnar <mingo@elte.hu>

random uvesafb failures were reported against Gentoo:

  http://bugs.gentoo.org/show_bug.cgi?id=222799

and Mihai Moldovan bisected it back to:

> 8f4d37ec is first bad commit
> commit 8f4d37ec
> Author: Peter Zijlstra <a.p.zijlstra@chello.nl>
> Date:   Fri Jan 25 21:08:29 2008 +0100
>
>    sched: high-res preemption tick

Linus suspected it to be hrtick + vm86 interaction and observed:

> Btw, Peter, Ingo: I think that commit is doing bad things. They aren't
> _incorrect_ per se, but they are definitely bad.
>
> Why?
>
> Using random _TIF_WORK_MASK flags is really impolite for doing
> "scheduling" work. There's a reason that arch/x86/kernel/entry_32.S
> special-cases the _TIF_NEED_RESCHED flag: we don't want to exit out of
> vm86 mode unnecessarily.
>
> See the "work_notifysig_v86" label, and how it does that
> "save_v86_state()" thing etc etc.

Right, I never liked having to fiddle with those TIF flags. Initially I
needed it because the hrtimer base lock could not nest in the rq lock.
That however is fixed these days.

Currently the only reason left to fiddle with the TIF flags is remote
wakeups. We cannot program a remote cpu's hrtimer. I've been thinking
about using the new and improved IPI function call stuff to implement
hrtimer_start_on().

However that does require that smp_call_function_single(.wait=0) works
from interrupt context - /me looks at the latest series from Jens - Yes
that does seem to be supported, good.

Here's a stab at cleaning this stuff up ...

Mihai reported test success as well.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Tested-by: NMihai Moldovan <ionic@ionic.de>
Cc: Michal Januszewski <spock@gentoo.org>
Cc: Antonino Daplas <adaplas@gmail.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

This is based on Linus' idea of creating cpu_active_map that prevents
scheduler load balancer from migrating tasks to the cpu that is going
down.

It allows us to simplify domain management code and avoid unecessary
domain rebuilds during cpu hotplug event handling.

Please ignore the cpusets part for now. It needs some more work in order
to avoid crazy lock nesting. Although I did simplfy and unify domain
reinitialization logic. We now simply call partition_sched_domains() in
all the cases. This means that we're using exact same code paths as in
cpusets case and hence the test below cover cpusets too.
Cpuset changes to make rebuild_sched_domains() callable from various
contexts are in the separate patch (right next after this one).

This not only boots but also easily handles
	while true; do make clean; make -j 8; done
and
	while true; do on-off-cpu 1; done
at the same time.
(on-off-cpu 1 simple does echo 0/1 > /sys/.../cpu1/online thing).

Suprisingly the box (dual-core Core2) is quite usable. In fact I'm typing
this on right now in gnome-terminal and things are moving just fine.

Also this is running with most of the debug features enabled (lockdep,
mutex, etc) no BUG_ONs or lockdep complaints so far.

I believe I addressed all of the Dmitry's comments for original Linus'
version. I changed both fair and rt balancer to mask out non-active cpus.
And replaced cpu_is_offline() with !cpu_active() in the main scheduler
code where it made sense (to me).
Signed-off-by: NMax Krasnyanskiy <maxk@qualcomm.com>
Acked-by: NLinus Torvalds <torvalds@linux-foundation.org>
Acked-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: NGregory Haskins <ghaskins@novell.com>
Cc: dmitry.adamushko@gmail.com
Cc: pj@sgi.com
Signed-off-by: NIngo Molnar <mingo@elte.hu>

  * Remove 16k stack requirements in isolated_cpu_setup when NR_CPUS=4096.
Signed-off-by: NMike Travis <travis@sgi.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Clean up __migrate_task(): to just have separate "done" and "fail"
cases, instead of that "out" case with random error behavior.
Signed-off-by: NIngo Molnar <mingo@elte.hu>

I think we may have a race between try_to_wake_up() and
migrate_live_tasks() -> move_task_off_dead_cpu() when the later one
may end up looping endlessly.

Interrupts are enabled on other CPUs when migration_call(CPU_DEAD, ...) is
called so we may get a race between try_to_wake_up() and
migrate_live_tasks() -> move_task_off_dead_cpu(). The former one may push
a task out of a dead CPU causing the later one to loop endlessly.

Heiko Carstens observed:

| That's exactly what explains a dump I got yesterday. Thanks for fixing! :)
Signed-off-by: NDmitry Adamushko <dmitry.adamushko@gmail.com>
Cc: miaox@cn.fujitsu.com
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: Heiko Carstens <heiko.carstens@de.ibm.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Avi Kivity <avi@qumranet.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

  * Replace usages of MAX_NUMNODES with nr_node_ids in kernel/sched.c,
    where appropriate.  This saves some allocated space as well as many
    wasted cycles going through node entries that are non-existent.
Signed-off-by: NMike Travis <travis@sgi.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

On Thu, Jun 19, 2008 at 12:27:14PM +0200, Peter Zijlstra wrote:
> On Thu, 2008-06-05 at 10:50 +0530, Ankita Garg wrote:
>
> > Thanks Peter for the explanation...
> >
> > I agree with the above and that is the reason why I did not see weird
> > values with cpu_time. But, run_delay still would suffer skews as the end
> > points for delta could be taken on different cpus due to migration (more
> > so on RT kernel due to the push-pull operations). With the below patch,
> > I could not reproduce the issue I had seen earlier. After every dequeue,
> > we take the delta and start wait measurements from zero when moved to a
> > different rq.
>
> OK, so task delay delay accounting is broken because it doesn't take
> migration into account.
>
> What you've done is make it symmetric wrt enqueue, and account it like
>
>   cpu0      cpu1
>
> enqueue
>  <wait-d1>
> dequeue
>             enqueue
>              <wait-d2>
>             run
>
> Where you add both d1 and d2 to the run_delay,.. right?
>

Thanks for reviewing the patch. The above is exactly what I have done.

> This seems like a good fix, however it looks like the patch will break
> compilation in !CONFIG_SCHEDSTATS && !CONFIG_TASK_DELAY_ACCT, of it
> failing to provide a stub for sched_info_dequeue() in that case.

Fixed. Pl. find the new patch below.
Signed-off-by: NAnkita Garg <ankita@in.ibm.com>
Acked-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Gregory Haskins <ghaskins@novell.com>
Cc: rostedt@goodmis.org
Cc: suresh.b.siddha@intel.com
Cc: aneesh.kumar@linux.vnet.ibm.com
Cc: dhaval@linux.vnet.ibm.com
Cc: vatsa@linux.vnet.ibm.com
Cc: David Bahi <DBahi@novell.com>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

We have the notion of tracking process-coupling (a.k.a. buddy-wake) via
the p->se.last_wake / p->se.avg_overlap facilities, but it is only used
for cfs to cfs interactions.  There is no reason why an rt to cfs
interaction cannot share in establishing a relationhip in a similar
manner.

Because PREEMPT_RT runs many kernel threads as FIFO priority, we often
times have heavy interaction between RT threads waking CFS applications.
This patch offers a substantial boost (50-60%+) in perfomance under those
circumstances.
Signed-off-by: NGregory Haskins <ghaskins@novell.com>
Cc: npiggin@suse.de
Cc: rostedt@goodmis.org
Acked-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Inspired by Peter Zijlstra.
Signed-off-by: NGregory Haskins <ghaskins@novell.com>
Cc: npiggin@suse.de
Cc: rostedt@goodmis.org
Acked-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Here it is another little Oops we found while configuring invalid values
via cgroups:

echo 0 > /dev/cgroups/0/cpu.rt_period_us
or
echo 4294967296 > /dev/cgroups/0/cpu.rt_period_us

[  205.509825] divide error: 0000 [#1]
[  205.510151] Modules linked in:
[  205.510151]
[  205.510151] Pid: 2339, comm: bash Not tainted (2.6.26-rc8 #33)
[  205.510151] EIP: 0060:[<c030c6ef>] EFLAGS: 00000293 CPU: 0
[  205.510151] EIP is at div64_u64+0x5f/0x70
[  205.510151] EAX: 0000389f EBX: 00000000 ECX: 00000000 EDX: 00000000
[  205.510151] ESI: d9800000 EDI: 00000000 EBP: c6cede60 ESP: c6cede50
[  205.510151]  DS: 007b ES: 007b FS: 0000 GS: 0033 SS: 0068
[  205.510151] Process bash (pid: 2339, ti=c6cec000 task=c79be370 task.ti=c6cec000)
[  205.510151] Stack: d9800000 0000389f c05971a0 d9800000 c6cedeb4 c0214dbd 00000000 00000000
[  205.510151]        c6cede88 c0242bd8 c05377c0 c7a41b40 00000000 00000000 00000000 c05971a0
[  205.510151]        c780ed20 c7508494 c7a41b40 00000000 00000002 c6cedebc c05971a0 ffffffea
[  205.510151] Call Trace:
[  205.510151]  [<c0214dbd>] ? __rt_schedulable+0x1cd/0x240
[  205.510151]  [<c0242bd8>] ? cgroup_file_open+0x18/0xe0
[  205.510151]  [<c0214fe4>] ? tg_set_bandwidth+0xa4/0xf0
[  205.510151]  [<c0215066>] ? sched_group_set_rt_period+0x36/0x50
[  205.510151]  [<c021508e>] ? cpu_rt_period_write_uint+0xe/0x10
[  205.510151]  [<c0242dc5>] ? cgroup_file_write+0x125/0x160
[  205.510151]  [<c0232c15>] ? hrtimer_interrupt+0x155/0x190
[  205.510151]  [<c02f047f>] ? security_file_permission+0xf/0x20
[  205.510151]  [<c0277ad8>] ? rw_verify_area+0x48/0xc0
[  205.510151]  [<c0283744>] ? dupfd+0x104/0x130
[  205.510151]  [<c027838c>] ? vfs_write+0x9c/0x160
[  205.510151]  [<c0242ca0>] ? cgroup_file_write+0x0/0x160
[  205.510151]  [<c027850d>] ? sys_write+0x3d/0x70
[  205.510151]  [<c0203019>] ? sysenter_past_esp+0x6a/0x91
[  205.510151]  =======================
[  205.510151] Code: 0f 45 de 31 f6 0f ad d0 d3 ea f6 c1 20 0f 45 c2 0f 45 d6 89 45 f0 89 55 f4 8b 55 f4 31 c9 8b 45 f0 39 d3 89 c6 77 08 89 d0 31 d2 <f7> f3 89 c1 83 c4 08 89 f0 f7 f3 89 ca 5b 5e 5d c3 55 89 e5 56
[  205.510151] EIP: [<c030c6ef>] div64_u64+0x5f/0x70 SS:ESP 0068:c6cede50

The attached patch solves the issue for me.

I'm checking as soon as possible for the period not being zero since, if
it is, going ahead is useless. This way we also save a mutex_lock() and
a read_lock() wrt doing it inside tg_set_bandwidth() or
__rt_schedulable().
Signed-off-by: NDario Faggioli <raistlin@linux.it>
Signed-off-by: NMichael Trimarchi <trimarchimichael@yahoo.it>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

This patch fixes the following warning:

kernel/sched.c:1667: warning: 'cfs_rq_set_shares' defined but not used

This seems the correct way to fix this; cfs_rq_set_shares() is only used
in a single place, which is also inside #ifdef CONFIG_FAIR_GROUP_SCHED.
Signed-off-by: NVegard Nossum <vegard.nossum@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

fix:

kernel/sched.c: In function ‘sched_group_set_shares':
kernel/sched.c:8635: error: implicit declaration of function ‘cfs_rq_set_shares'
Signed-off-by: NIngo Molnar <mingo@elte.hu>

the CPU hotplug problems (crashes under high-volume unplug+replug
tests) seem to be related to migrate_dead_tasks().

Firstly I added traces to see all tasks being migrated with
migrate_live_tasks() and migrate_dead_tasks(). On my setup the problem
pops up (the one with "se == NULL" in the loop of
pick_next_task_fair()) shortly after the traces indicate that some has
been migrated with migrate_dead_tasks()). btw., I can reproduce it
much faster now with just a plain cpu down/up loop.

[disclaimer] Well, unless I'm really missing something important in
this late hour [/desclaimer] pick_next_task() is not something
appropriate for migrate_dead_tasks() :-)

the following change seems to eliminate the problem on my setup
(although, I kept it running only for a few minutes to get a few
messages indicating migrate_dead_tasks() does move tasks and the
system is still ok)
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Increase the accuracy of the effective_load values.

Not only consider the current increment (as per the attempted wakeup), but
also consider the delta between when we last adjusted the shares and the
current situation.
Signed-off-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

rw_i = {2, 4, 1, 0}
s_i = {2/7, 4/7, 1/7, 0}

wakeup on cpu0, weight=1

rw'_i = {3, 4, 1, 0}
s'_i = {3/8, 4/8, 1/8, 0}

s_0 = S * rw_0 / \Sum rw_j ->
  \Sum rw_j = S*rw_0/s_0 = 1*2*7/2 = 7 (correct)

s'_0 = S * (rw_0 + 1) / (\Sum rw_j + 1) =
       1 * (2+1) / (7+1) = 3/8 (correct

so we find that adding 1 to cpu0 gains 5/56 in weight
if say the other cpu were, cpu1, we'd also have to calculate its 4/56 loss
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

We found that the affine wakeup code needs rather accurate load figures
to be effective. The trouble is that updating the load figures is fairly
expensive with group scheduling. Therefore ratelimit the updating.
Signed-off-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

In case the domain is empty, pretend there is a single task on each cpu, so
that together with the boost logic we end up giving 1/n shares to each
cpu.
Signed-off-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

The bias given by source/target_load functions can be very large, disable
it by default to get faster convergence.
Signed-off-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Priority looses much of its meaning in a hierarchical context. So don't
use it in balance decisions.
Signed-off-by: NPeter Zijlstra <peterz@infradead.org>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

find_busiest_group() has some assumptions about task weight being in the
NICE_0_LOAD range. Hierarchical task groups break this assumption - fix this
by replacing it with the average task weight, which will adapt the situation.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Remove the fall-back to SCHED_LOAD_SCALE by remembering the previous value of
cpu_avg_load_per_task() - this is useful because of the hierarchical group
model in which task weight can be much smaller.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: NIngo Molnar <mingo@elte.hu>

Finding the least idle cpu is more accurate when done with updated shares.
Signed-off-by: NPeter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Srivatsa Vaddagiri <vatsa@linux.vnet.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Signed-off-by: NIngo Molnar <mingo@elte.hu>