Move block I/O operations to a separate file. It is because it will
be used later not only by the swap writer.
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
Remove support of reads with offset. This means snapshot_read/write_next
now does not accept count parameter. It allows to clean up the functions
and snapshot handle which no longer needs to care about offsets.
/dev/snapshot handler is converted to simple_{read_from,write_to}_buffer
which take care of offsets.
Signed-off-by: Jiri Slaby <jslaby@suse.cz>
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
The current version of RCU_FAST_NO_HZ reproduces the old CLASSIC_RCU
dyntick-idle bug, as it fails to detect CPUs that have interrupted
or NMIed out of dyntick-idle mode. Fix this by making rcu_needs_cpu()
check the state in the per-CPU rcu_dynticks variables, thus correctly
detecting the dyntick-idle state from an RCU perspective.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Lai Jiangshan noted that up to 10% of the RCU_SOFTIRQ are spurious, and
traced this down to the fact that the current grace-period machinery
will uselessly raise RCU_SOFTIRQ when a given CPU needs to go through
a quiescent state, but has not yet done so. In this situation, there
might well be nothing that RCU_SOFTIRQ can do, and the overhead can be
worth worrying about in the ksoftirqd case. This patch therefore avoids
raising RCU_SOFTIRQ in this situation.
Changes since v1 (http://lkml.org/lkml/2010/3/30/122 from Lai Jiangshan):
o Omit the rcu_qs_pending() prechecks, as they aren't that
much less expensive than the quiescent-state checks.
o Merge with the set_need_resched() patch that reduces IPIs.
o Add the new n_rp_report_qs field to the rcu_pending tracing output.
o Update the tracing documentation accordingly.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
TREE_RCU assumes that CPU numbering is contiguous, but some users need
large holes in the numbering to better map to hardware layout. This patch
makes TREE_RCU (and TREE_PREEMPT_RCU) tolerate large holes in the CPU
numbering. However, NR_CPUS must still be greater than the largest
CPU number.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The existing RCU CPU stall-warning messages can be confusing, especially
in the case where one CPU detects a single other stalled CPU. In addition,
the console messages did not say which flavor of RCU detected the stall,
which can make it difficult to work out exactly what is causing the stall.
This commit improves these messages.
Requested-by: Dhaval Giani <dhaval.giani@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Print boot-time messages if tracing is enabled, if fanout is set
to non-default values, if exact fanout is specified, if accelerated
dyntick-idle grace periods have been enabled, if RCU-lockdep is enabled,
if rcutorture has been boot-time enabled, if the CPU stall detector has
been disabled, or if four-level hierarchy has been enabled.
This is all for TREE_RCU and TREE_PREEMPT_RCU. TINY_RCU will be handled
separately, if at all.
Suggested-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The current RCU CPU stall warnings remain enabled even after a panic
occurs, which some people have found to be a bit counterproductive.
This patch therefore uses a notifier to disable stall warnings once a
panic occurs.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
TINY_RCU does not need rcu_scheduler_active unless CONFIG_DEBUG_LOCK_ALLOC.
So conditionally compile rcu_scheduler_active in order to slim down
rcutiny a bit more. Also gets rid of an EXPORT_SYMBOL_GPL, which is
responsible for most of the slimming.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
The addition of preemptible RCU to treercu resulted in a bit of
confusion and inefficiency surrounding the handling of context switches
for RCU-sched and for RCU-preempt. For RCU-sched, a context switch
is a quiescent state, pure and simple, just like it always has been.
For RCU-preempt, a context switch is in no way a quiescent state, but
special handling is required when a task blocks in an RCU read-side
critical section.
However, the callout from the scheduler and the outer loop in ksoftirqd
still calls something named rcu_sched_qs(), whose name is no longer
accurate. Furthermore, when rcu_check_callbacks() notes an RCU-sched
quiescent state, it ends up unnecessarily (though harmlessly, aside
from the performance hit) enqueuing the current task if it happens to
be running in an RCU-preempt read-side critical section. This not only
increases the maximum latency of scheduler_tick(), it also needlessly
increases the overhead of the next outermost rcu_read_unlock() invocation.
This patch addresses this situation by separating the notion of RCU's
context-switch handling from that of RCU-sched's quiescent states.
The context-switch handling is covered by rcu_note_context_switch() in
general and by rcu_preempt_note_context_switch() for preemptible RCU.
This permits rcu_sched_qs() to handle quiescent states and only quiescent
states. It also reduces the maximum latency of scheduler_tick(), though
probably by much less than a microsecond. Finally, it means that tasks
within preemptible-RCU read-side critical sections avoid incurring the
overhead of queuing unless there really is a context switch.
Suggested-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Acked-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Because synchronize_rcu_bh() is identical to synchronize_sched(),
make the former a static inline invoking the latter, saving the
overhead of an EXPORT_SYMBOL_GPL() and the duplicate code.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Offline CPUs are not in nohz_cpu_mask, but can be ignored when checking
for the last non-dyntick-idle CPU. This patch therefore only checks
online CPUs for not being dyntick idle, allowing fast entry into
full-system dyntick-idle state even when there are some offline CPUs.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Shrink the RCU_INIT_FLAVOR() macro by moving all but the initialization
of the ->rda[] array to rcu_init_one(). The call to rcu_init_one()
can then be moved to the end of the RCU_INIT_FLAVOR() macro, which is
required because rcu_boot_init_percpu_data(), which is now called from
rcu_init_one(), depends on the initialization of the ->rda[] array.
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
This patch adds a check to __rcu_pending() that does a local
set_need_resched() if the current CPU is holding up the current grace
period and if force_quiescent_state() will be called soon. The goal is
to reduce the probability that force_quiescent_state() will need to do
smp_send_reschedule(), which sends an IPI and is therefore more expensive
on most architectures.
Signed-off-by: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
There is no need to disable lockdep after an RCU lockdep splat,
so remove the debug_lockdeps_off() from lockdep_rcu_dereference().
To avoid repeated lockdep splats, use a static variable in the inlined
rcu_dereference_check() and rcu_dereference_protected() macros so that
a given instance splats only once, but so that multiple instances can
be detected per boot.
This is controlled by a new config variable CONFIG_PROVE_RCU_REPEATEDLY,
which is disabled by default. This provides the normal lockdep behavior
by default, but permits people who want to find multiple RCU-lockdep
splats per boot to easily do so.
Requested-by: Eric Paris <eparis@redhat.com>
Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Tested-by: Eric Paris <eparis@redhat.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
How to pick good mult/shift pairs has always been difficult to
describe to folks writing clocksource drivers, since it requires
careful tradeoffs in adjustment accuracy vs overflow limits.
Now, with the clocks_calc_mult_shift function, its much
easier. However, not many clocksources have converted to using that
function, and there is still the issue of the max interval length
assumption being made by each clocksource driver independently.
So this patch simplifies the registration process by having
clocksources be registered with a hz/khz value and the registration
function taking care of setting mult/shift.
This should take most of the confusion out of writing a clocksource
driver.
Additionally it also keeps the shift size tradeoff (more accuracy vs
longer possible nohz times) centralized so the timekeeping core can
keep track of the assumptions being made.
[ tglx: Coding style and comments fixed ]
Signed-off-by: John Stultz <johnstul@us.ibm.com>
LKML-Reference: <1273280858-30143-1-git-send-email-johnstul@us.ibm.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
We can optimize and simplify things taking into account signal->cputimer
is always running when we have configured any process wide cpu timer.
In check_process_timers(), we don't have to check if new updated value of
signal->cputime_expires is smaller, since we maintain new first expiration
time ({prof,virt,sched}_expires) in code flow and all other writes to
expiration cache are protected by sighand->siglock .
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Cc: Balbir Singh <balbir@in.ibm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
This comment should have been removed together with uids_mutex
when removing user sched.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Dhaval Giani <dhaval.giani@gmail.com>
LKML-Reference: <4BE77C6B.5010402@cn.fujitsu.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
For the ondemand cpufreq governor, it is desired that the iowait
time is microaccounted in a similar way as idle time is.
This patch introduces the infrastructure to account and expose
this information via the get_cpu_iowait_time_us() function.
[akpm@linux-foundation.org: fix CONFIG_NO_HZ=n build]
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: davej@redhat.com
LKML-Reference: <20100509082523.284feab6@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Now that the only user of ts->idle_lastupdate is
update_ts_time_stats(), the entire field can be eliminated.
In update_ts_time_stats(), idle_lastupdate is first set to
"now", and a few lines later, the only user is an if() statement
that assigns a variable either to "now" or to
ts->idle_lastupdate, which has the value of "now" at that point.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: davej@redhat.com
LKML-Reference: <20100509082439.2fab0b4f@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This patch folds the updating of the last_update_time into the
update_ts_time_stats() function, and updates the callers.
This allows for further cleanups that are done in the next
patch.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: davej@redhat.com
LKML-Reference: <20100509082403.60072967@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Right now, get_cpu_idle_time_us() only reports the idle
statistics upto the point the CPU entered last idle; not what is
valid right now.
This patch adds an update of the idle statistics to
get_cpu_idle_time_us(), so that calling this function always
returns statistics that are accurate at the point of the call.
This includes resetting the start of the idle time for
accounting purposes to avoid double accounting.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: davej@redhat.com
LKML-Reference: <20100509082323.2d2f1945@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Currently, two places update the idle statistics (and more to
come later in this series).
This patch creates a helper function for updating these
statistics.
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: davej@redhat.com
LKML-Reference: <20100509082245.163e67ed@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The exported function get_cpu_idle_time_us() has no comment
describing it; add a kerneldoc comment
Signed-off-by: Arjan van de Ven <arjan@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Reviewed-by: Rik van Riel <riel@redhat.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: davej@redhat.com
LKML-Reference: <20100509082208.7cb721f0@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Drop the nested field as we don't use it. Every nested state can
be computed from a state machine on post processing already.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp>
Cc: Steven Rostedt <rostedt@goodmis.org>
Drop the waittime field from the lock_acquired event, we can
calculate it by substracting the lock_acquired event timestamp
with the matching lock_acquire one.
It is not needed and takes useless space in the traces.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Hitoshi Mitake <mitake@dcl.info.waseda.ac.jp>
Cc: Steven Rostedt <rostedt@goodmis.org>
Move enable/disable_kprobe() API out from debugfs related code,
because these interfaces are not related to debugfs interface.
This fixes a compiler warning.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Acked-by: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Acked-by: Tony Luck <tony.luck@intel.com>
Cc: systemtap <systemtap@sources.redhat.com>
Cc: DLE <dle-develop@lists.sourceforge.net>
LKML-Reference: <20100427223312.2322.60512.stgit@localhost6.localdomain6>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When !CONFIG_SMP, cpu_stop functions weren't defined at all which
could lead to build failures if UP code uses cpu_stop facility. Add
dummy cpu_stop implementation for UP. The waiting variants execute
the work function directly with preempt disabled and
stop_one_cpu_nowait() schedules a workqueue work.
Makefile and ifdefs around stop_machine implementation are updated to
accomodate CONFIG_SMP && !CONFIG_STOP_MACHINE case.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Ingo Molnar <mingo@elte.hu>
Commit 6bde9b6ce0 ("perf: Add
group scheduling transactional APIs") added code to allow a
group to be scheduled in a single transaction. However, it
introduced a bug in handling events whose pmu does not implement
transactions -- at the end of scheduling in the events in the
group, in the non-transactional case the code now falls through
to the group_error label, and proceeds to unschedule all the
events in the group and return failure.
This fixes it by returning 0 (success) in the non-transactional
case.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: eranian@gmail.com
LKML-Reference: <20100508105800.GB10650@brick.ozlabs.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
A while back there was a discussion regarding the rt_secret_interval timer.
Given that we've had the ability to do emergency route cache rebuilds for awhile
now, based on a statistical analysis of the various hash chain lengths in the
cache, the use of the flush timer is somewhat redundant. This patch removes the
rt_secret_interval sysctl, allowing us to rely solely on the statistical
analysis mechanism to determine the need for route cache flushes.
Signed-off-by: Neil Horman <nhorman@tuxdriver.com>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
rcu: create rcu_my_thread_group_empty() wrapper
memcg: css_id() must be called under rcu_read_lock()
cgroup: Check task_lock in task_subsys_state()
sched: Fix an RCU warning in print_task()
cgroup: Fix an RCU warning in alloc_css_id()
cgroup: Fix an RCU warning in cgroup_path()
KEYS: Fix an RCU warning in the reading of user keys
KEYS: Fix an RCU warning
Add group scheduling transactional APIs to struct pmu.
These APIs will be implemented in arch code, based on Peter's idea as
below.
> the idea behind hw_perf_group_sched_in() is to not perform
> schedulability tests on each event in the group, but to add the group
> as a whole and then perform one test.
>
> Of course, when that test fails, you'll have to roll-back the whole
> group again.
>
> So start_txn (or a better name) would simply toggle a flag in the pmu
> implementation that will make pmu::enable() not perform the
> schedulablilty test.
>
> Then commit_txn() will perform the schedulability test (so note the
> method has to have a !void return value.
>
> This will allow us to use the regular
> kernel/perf_event.c::group_sched_in() and all the rollback code.
> Currently each hw_perf_group_sched_in() implementation duplicates all
> the rolllback code (with various bugs).
->start_txn:
Start group events scheduling transaction, set a flag to make
pmu::enable() not perform the schedulability test, it will be performed
at commit time.
->commit_txn:
Commit group events scheduling transaction, perform the group
schedulability as a whole
->cancel_txn:
Stop group events scheduling transaction, clear the flag so
pmu::enable() will perform the schedulability test.
Reviewed-by: Stephane Eranian <eranian@google.com>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Signed-off-by: Lin Ming <ming.m.lin@intel.com>
Cc: David Miller <davem@davemloft.net>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1272002160.5707.60.camel@minggr.sh.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Stephane reported a lockdep warning while using PERF_FORMAT_GROUP.
The issue is that perf_event_read_group() takes faults while holding
the ctx->mutex, while perf_event_release_kernel() can be called from
munmap(). Which makes for an AB-BA deadlock.
Except we can never establish the deadlock because we'll only ever
call perf_event_release_kernel() after all file descriptors are dead
so there is no concurrency possible.
Reported-by: Stephane Eranian <eranian@google.com>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Both Stephane and Corey reported that PERF_FORMAT_GROUP didn't work
as expected if the task the counters were attached to quit before
the read() call.
The cause is that we unconditionally destroy the grouping when we
remove counters from their context. Fix this by only doing this when
we free the counter itself.
Reported-by: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Reported-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1273160566.5605.404.camel@twins>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
struct rq isn't visible outside of sched.o so its near useless to
expose the pointer, also there are no users of it, so remove it.
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1272997616.1642.207.camel@laptop>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When calling check_prevs_add(), if all validations passed
add_lock_to_list() will add new lock to dependency tree and
alloc stack_trace for each list_entry.
But at this time, we are always on the same stack, so stack_trace
for each list_entry has the same value. This is redundant and eats
up lots of memory which could lead to warning on low
MAX_STACK_TRACE_ENTRIES.
Use one copy of stack_trace instead.
V2: As suggested by Peter Zijlstra, move save_trace() from
check_prevs_add() to check_prev_add().
Add tracking for trylock dependence which is also redundant.
Signed-off-by: Yong Zhang <yong.zhang0@windriver.com>
Cc: David S. Miller <davem@davemloft.net>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <20100504065711.GC10784@windriver.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
If synchronize_sched_expedited() is ever to be called from within
kernel/sched.c in a !SMP PREEMPT kernel, the !SMP implementation needs
a barrier().
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
The original code doesn't work because "call" is never NULL there.
Signed-off-by: Dan Carpenter <error27@gmail.com>
LKML-Reference: <20100320143911.GF5331@bicker>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
The memory barriers must be in the SMP case, not in the !SMP case.
Also add a barrier after the atomic_inc() in order to ensure that
other CPUs see post-synchronize_sched_expedited() actions as following
the expedited grace period.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
The paranoid check which verifies that the cpu_stop callback is
actually called on all online cpus is completely superflous. It's
guaranteed by cpu_stop facility and if it didn't work as advertised
other things would go horribly wrong and trying to recover using
synchronize_sched() wouldn't be very meaningful.
Kill the paranoid check. Removal of this feature is done as a
separate step so that it can serve as a bisection point if something
actually goes wrong.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: Josh Triplett <josh@freedesktop.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Dimitri Sivanich <sivanich@sgi.com>
Currently migration_thread is serving three purposes - migration
pusher, context to execute active_load_balance() and forced context
switcher for expedited RCU synchronize_sched. All three roles are
hardcoded into migration_thread() and determining which job is
scheduled is slightly messy.
This patch kills migration_thread and replaces all three uses with
cpu_stop. The three different roles of migration_thread() are
splitted into three separate cpu_stop callbacks -
migration_cpu_stop(), active_load_balance_cpu_stop() and
synchronize_sched_expedited_cpu_stop() - and each use case now simply
asks cpu_stop to execute the callback as necessary.
synchronize_sched_expedited() was implemented with private
preallocated resources and custom multi-cpu queueing and waiting
logic, both of which are provided by cpu_stop.
synchronize_sched_expedited_count is made atomic and all other shared
resources along with the mutex are dropped.
synchronize_sched_expedited() also implemented a check to detect cases
where not all the callback got executed on their assigned cpus and
fall back to synchronize_sched(). If called with cpu hotplug blocked,
cpu_stop already guarantees that and the condition cannot happen;
otherwise, stop_machine() would break. However, this patch preserves
the paranoid check using a cpumask to record on which cpus the stopper
ran so that it can serve as a bisection point if something actually
goes wrong theree.
Because the internal execution state is no longer visible,
rcu_expedited_torture_stats() is removed.
This patch also renames cpu_stop threads to from "stopper/%d" to
"migration/%d". The names of these threads ultimately don't matter
and there's no reason to make unnecessary userland visible changes.
With this patch applied, stop_machine() and sched now share the same
resources. stop_machine() is faster without wasting any resources and
sched migration users are much cleaner.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: Josh Triplett <josh@freedesktop.org>
Cc: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Dimitri Sivanich <sivanich@sgi.com>
Reimplement stop_machine using cpu_stop. As cpu stoppers are
guaranteed to be available for all online cpus,
stop_machine_create/destroy() are no longer necessary and removed.
With resource management and synchronization handled by cpu_stop, the
new implementation is much simpler. Asking the cpu_stop to execute
the stop_cpu() state machine on all online cpus with cpu hotplug
disabled is enough.
stop_machine itself doesn't need to manage any global resources
anymore, so all per-instance information is rolled into struct
stop_machine_data and the mutex and all static data variables are
removed.
The previous implementation created and destroyed RT workqueues as
necessary which made stop_machine() calls highly expensive on very
large machines. According to Dimitri Sivanich, preventing the dynamic
creation/destruction makes booting faster more than twice on very
large machines. cpu_stop resources are preallocated for all online
cpus and should have the same effect.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Dimitri Sivanich <sivanich@sgi.com>
Implement a simplistic per-cpu maximum priority cpu monopolization
mechanism. A non-sleeping callback can be scheduled to run on one or
multiple cpus with maximum priority monopolozing those cpus. This is
primarily to replace and unify RT workqueue usage in stop_machine and
scheduler migration_thread which currently is serving multiple
purposes.
Four functions are provided - stop_one_cpu(), stop_one_cpu_nowait(),
stop_cpus() and try_stop_cpus().
This is to allow clean sharing of resources among stop_cpu and all the
migration thread users. One stopper thread per cpu is created which
is currently named "stopper/CPU". This will eventually replace the
migration thread and take on its name.
* This facility was originally named cpuhog and lived in separate
files but Peter Zijlstra nacked the name and thus got renamed to
cpu_stop and moved into stop_machine.c.
* Better reporting of preemption leak as per Peter's suggestion.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Dimitri Sivanich <sivanich@sgi.com>
Some RCU-lockdep splat repairs need to know whether they are running
in a single-threaded process. Unfortunately, the thread_group_empty()
primitive is defined in sched.h, and can induce #include hell. This
commit therefore introduces a rcu_my_thread_group_empty() wrapper that
is defined in rcupdate.c, thus avoiding the need to include sched.h
everywhere.
Signed-off-by: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
perf: Fix resource leak in failure path of perf_event_open()
* 'core-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
rcu: Fix RCU lockdep splat on freezer_fork path
rcu: Fix RCU lockdep splat in set_task_cpu on fork path
mutex: Don't spin when the owner CPU is offline or other weird cases
With CONFIG_PROVE_RCU=y, a warning can be triggered:
$ cat /proc/sched_debug
...
kernel/cgroup.c:1649 invoked rcu_dereference_check() without protection!
...
Both cgroup_path() and task_group() should be called with either
rcu_read_lock or cgroup_mutex held.
The rcu_dereference_check() does include cgroup_lock_is_held(), so we
know that this lock is not held. Therefore, in a CONFIG_PREEMPT kernel,
to say nothing of a CONFIG_PREEMPT_RT kernel, the original code could
have ended up copying a string out of the freelist.
This patch inserts RCU read-side primitives needed to prevent this
scenario.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
With CONFIG_PROVE_RCU=y, a warning can be triggered:
# mount -t cgroup -o memory xxx /mnt
# mkdir /mnt/0
...
kernel/cgroup.c:4442 invoked rcu_dereference_check() without protection!
...
This is a false-positive. It's safe to directly access parent_css->id.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
with CONFIG_PROVE_RCU=y, a warning can be triggered:
# mount -t cgroup -o debug xxx /mnt
# cat /proc/$$/cgroup
...
kernel/cgroup.c:1649 invoked rcu_dereference_check() without protection!
...
This is a false-positive, because cgroup_path() can be called
with either rcu_read_lock() held or cgroup_mutex held.
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Fix this build error:
kernel/hw_breakpoint.c:58:1: error: pasting "__pcpu_scope_" and "*" does not give a valid preprocessing token
It happens if CONFIG_DEBUG_FORCE_WEAK_PER_CPU, because we concatenate
someting with the name and we have the "*" in the name.
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Cc: K. Prasad <prasad@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Jason Wessel <jason.wessel@windriver.com>
LKML-Reference: <20100503133942.GA5497@nowhere>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
No need to disable preemption in the debug_atomic_* ops, as
we ensure interrupts are disabled already.
So let's use the __this_cpu_ops() rather than this_cpu_ops() that
enclose the ops in a preempt disabled section.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Fix a silly copy-paste mistake that was making debug_atomic_dec use
this_cpu_inc instead of this_cpu_dec.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
We forgot to provide a !CONFIG_DEBUG_LOCKDEP case for the
redundant_hardirqs_on stat handling.
Manage that in the headers with a new __debug_atomic_inc() helper.
Fixes:
kernel/lockdep.c:2306: error: 'lockdep_stats' undeclared (first use in this function)
kernel/lockdep.c:2306: error: (Each undeclared identifier is reported only once
kernel/lockdep.c:2306: error: for each function it appears in.)
Reported-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
This patch adds a cpumask affinity hint to the irq_desc structure,
along with a registration function and a read-only proc entry for each
interrupt.
This affinity_hint handle for each interrupt can be used by underlying
drivers that need a better mechanism to control interrupt affinity.
The underlying driver can register a cpumask for the interrupt, which
will allow the driver to provide the CPU mask for the interrupt to
anything that requests it. The intent is to extend the userspace
daemon, irqbalance, to help hint to it a preferred CPU mask to balance
the interrupt into.
[ tglx: Fixed compile warnings, added WARN_ON, made SMP only ]
Signed-off-by: Peter P Waskiewicz Jr <peter.p.waskiewicz.jr@intel.com>
Cc: davem@davemloft.net
Cc: arjan@linux.jf.intel.com
Cc: bhutchings@solarflare.com
LKML-Reference: <20100430214445.3992.41647.stgit@ppwaskie-hc2.jf.intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
The breakpoint generic layer assumes that archs always know in advance
the static number of address registers available to host breakpoints
through the HBP_NUM macro.
However this is not true for every archs. For example Arm needs to get
this information dynamically to handle the compatiblity between
different versions.
To solve this, this patch proposes to drop the static HBP_NUM macro
and let the arch provide the number of available slots through a
new hw_breakpoint_slots() function. For archs that have
CONFIG_HAVE_MIXED_BREAKPOINTS_REGS selected, it will be called once
as the number of registers fits for instruction and data breakpoints
together.
For the others it will be called first to get the number of
instruction breakpoint registers and another time to get the
data breakpoint registers, the targeted type is given as a
parameter of hw_breakpoint_slots().
Reported-by: Will Deacon <will.deacon@arm.com>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Cc: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Cc: K. Prasad <prasad@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Ingo Molnar <mingo@elte.hu>
Depending on their nature and on what an arch supports, breakpoints
may consume more than one address register. For example a simple
absolute address match usually only requires one address register.
But an address range match may consume two registers.
Currently our slot allocation constraints, that tend to reflect the
limited arch's resources, always consider that a breakpoint consumes
one slot.
Then provide a way for archs to tell us the weight of a breakpoint
through a new hw_breakpoint_weight() helper. This weight will be
computed against the generic allocation constraints instead of
a constant value.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Cc: K. Prasad <prasad@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
There are two outstanding fashions for archs to implement hardware
breakpoints.
The first is to separate breakpoint address pattern definition
space between data and instruction breakpoints. We then have
typically distinct instruction address breakpoint registers
and data address breakpoint registers, delivered with
separate control registers for data and instruction breakpoints
as well. This is the case of PowerPc and ARM for example.
The second consists in having merged breakpoint address space
definition between data and instruction breakpoint. Address
registers can host either instruction or data address and
the access mode for the breakpoint is defined in a control
register. This is the case of x86 and Super H.
This patch adds a new CONFIG_HAVE_MIXED_BREAKPOINTS_REGS config
that archs can select if they belong to the second case. Those
will have their slot allocation merged for instructions and
data breakpoints.
The others will have a separate slot tracking between data and
instruction breakpoints.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Cc: K. Prasad <prasad@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@elte.hu>
The current policies of breakpoints in x86 and SH are the following:
- task bound breakpoints can only break on userspace addresses
- cpu wide breakpoints can only break on kernel addresses
The former rule prevents ptrace breakpoints to be set to trigger on
kernel addresses, which is good. But as a side effect, we can't
breakpoint on kernel addresses for task bound breakpoints.
The latter rule simply makes no sense, there is no reason why we
can't set breakpoints on userspace while performing cpu bound
profiles.
We want the following new policies:
- task bound breakpoint can set userspace address breakpoints, with
no particular privilege required.
- task bound breakpoints can set kernelspace address breakpoints but
must be privileged to do that.
- cpu bound breakpoints can do what they want as they are privileged
already.
To implement these new policies, this patch checks if we are dealing
with a kernel address breakpoint, if so and if the exclude_kernel
parameter is set, we tell the user that the breakpoint is invalid,
which makes a good generic ptrace protection.
If we don't have exclude_kernel, ensure the user has the right
privileges as kernel breakpoints are quite sensitive (risk of
trap recursion attacks and global performance impacts).
[ Paul Mundt: keep addr space check for sh signal delivery and fix
double function declaration]
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Cc: K. Prasad <prasad@linux.vnet.ibm.com>
Cc: Paul Mundt <lethal@linux-sh.org>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
We stopped checking disabled breakpoints because we weren't
allowing breakpoints on NULL addresses. And gdb tends to set
NULL addresses on inactive breakpoints.
But refusing NULL addresses was actually a regression that has
been fixed now. There is no reason anymore to not validate
inactive breakpoint settings.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Cc: Will Deacon <will.deacon@arm.com>
Cc: Mahesh Salgaonkar <mahesh@linux.vnet.ibm.com>
Cc: K. Prasad <prasad@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Ingo Molnar <mingo@elte.hu>
__print_hex() prints values in an array in hex (w/o '0x') (space separated)
EX) 92 33 32 f3 ee 4d
Signed-off-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Tomohiro Kusumi <kusumi.tomohiro@jp.fujitsu.com>
Signed-off-by: Kei Tokunaga <tokunaga.keiich@jp.fujitsu.com>
Acked-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: James Bottomley <James.Bottomley@suse.de>
Simplify debug_atomic_inc/dec by using this_cpu_inc/dec() instead
of doing it through an indirect get_cpu_var() and a manual
incrementation.
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
When a path restore the flags while irqs are already enabled, we
update the per cpu var redundant_hardirqs_on in a racy fashion
and debug_atomic_inc() warns about this situation.
In this particular case, loosing a few hits in a stat is not a big
deal, so increment it without protection.
v2: Don't bother with disabling irq, we can miss one count in
rare situations
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
In short: change cancel_work_sync(work) to mark this work as "never
queued" upon return.
When cancel_work_sync(work) succeeds, we know that this work can't be
queued or running, and since we own WORK_STRUCT_PENDING nobody can change
the bits in work->data under us. This means we can also clear the "cwq"
part along with _PENDING bit lockless before return, unless the work is
queued nobody can assume get_wq_data() is stable even under cwq->lock.
This change can speedup the subsequent cancel/flush requests, and as
Dmitry pointed out this simplifies the usage of work_struct's which
can be queued on different workqueues. Consider this pseudo code from
the input subsystem:
struct workqueue_struct *WQ;
struct work_struct *WORK;
for (;;) {
WQ = create_workqueue();
...
if (condition())
queue_work(WQ, WORK);
...
cancel_work_sync(WORK);
destroy_workqueue(WQ);
}
If condition() returns T and then F, cancel_work_sync() will crash the
kernel because WORK->data still points to the already destroyed workqueue.
With this patch the code like above becomes correct.
Suggested-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
This patch (as1319) adds kerneldoc and a pointed warning to
flush_scheduled_work().
Signed-off-by: Alan Stern <stern@rowland.harvard.edu>
Signed-off-by: Tejun Heo <tj@kernel.org>
flush_delayed_work() always uses keventd_wq for re-queueing,
but it should use the workqueue this dwork was queued on.
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
When sleep_time is off the function profiler ignores the time that a task
is scheduled out. When the task is scheduled out a timestamp is taken.
When the task is scheduled back in, the timestamp is compared to the
current time and the saved calltimes are adjusted accordingly.
But when stopping the function profiler, the sched switch hook that
does this adjustment was stopped before shutting down the tracer.
This allowed some tasks to not get their timestamps set when they
scheduled out. When the function profiler started again, this would
skew the times of the scheduler functions.
This patch moves the stopping of the sched switch to after the function
profiler is stopped. It also ignores zero set calltimes, which may
happen on start up.
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
When combined with function graph tracing the ftrace function profiler
also prints the average run time of functions. While this gives us some
good information, it doesn't tell us anything about the variance of the
run times of the function. This change prints out the s^2 sample
standard deviation alongside the average.
This change adds one entry to the profile record structure. This
increases the memory footprint of the function profiler by 1/3 on a
32-bit system, and by 1/5 on a 64-bit system when function graphing is
enabled, though the memory is only allocated when the profiler is turned
on. During the profiling, one extra line of code adds the squared
calltime to the new record entry, so this should not adversly affect
performance.
Note that the square of the sample standard deviation is printed because
there is no sqrt implementation for unsigned long long in the kernel.
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
LKML-Reference: <1272304925-2436-1-git-send-email-chase.douglas@canonical.com>
[ fixed comment about ns^2 -> us^2 conversion ]
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
With the addition of the "missed events" flags that is stored in the
commit field of the ring buffer page, the ring_buffer_benchmark
was not updated to handle this. If events are missed, then the
missed events flag is set in the ring buffer page, the benchmark
will count that flag as part of the size of the page and will hit the BUG()
when it tries to read beyond the page.
The solution is simply to have the ring buffer benchmark mask off
the extra bits.
Reported-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
When performing a non-consuming read, a synchronize_sched() is
performed once for every cpu which is actively tracing.
This is very expensive, and can make it take several seconds to open
up the 'trace' file with lots of cpus.
Only one synchronize_sched() call is actually necessary. What is
desired is for all cpus to see the disabling state change. So we
transform the existing sequence:
for_each_cpu() {
ring_buffer_read_start();
}
where each ring_buffer_start() call performs a synchronize_sched(),
into the following:
for_each_cpu() {
ring_buffer_read_prepare();
}
ring_buffer_read_prepare_sync();
for_each_cpu() {
ring_buffer_read_start();
}
wherein only the single ring_buffer_read_prepare_sync() call needs to
do the synchronize_sched().
The first phase, via ring_buffer_read_prepare(), allocates the 'iter'
memory and increments ->record_disabled.
In the second phase, ring_buffer_read_prepare_sync() makes sure this
->record_disabled state is visible fully to all cpus.
And in the final third phase, the ring_buffer_read_start() calls reset
the 'iter' objects allocated in the first phase since we now know that
none of the cpus are adding trace entries any more.
This makes openning the 'trace' file nearly instantaneous on a
sparc64 Niagara2 box with 128 cpus tracing.
Signed-off-by: David S. Miller <davem@davemloft.net>
LKML-Reference: <20100420.154711.11246950.davem@davemloft.net>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Add function graph output to irqsoff tracer.
The graph output is enabled by setting new 'display-graph' trace option.
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
LKML-Reference: <1270227683-14631-4-git-send-email-jolsa@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Let the function graph tracer have custom flags passed to its
output functions.
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
LKML-Reference: <1270227683-14631-3-git-send-email-jolsa@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
Add ftrace events for graph tracer, so the graph output could be shared
with other tracers.
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
LKML-Reference: <1270227683-14631-2-git-send-email-jolsa@redhat.com>
Signed-off-by: Steven Rostedt <rostedt@goodmis.org>
On ppc64 you get this error:
$ setarch ppc -R true
setarch: ppc: Unrecognized architecture
because uname still reports ppc64 as the machine.
So mask off the personality flags when checking for PER_LINUX32.
Signed-off-by: Andreas Schwab <schwab@linux-m68k.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Issues in the current select_idle_sibling() logic in select_task_rq_fair()
in the context of a task wake-up:
a) Once we select the idle sibling, we use that domain (spanning the cpu that
the task is currently woken-up and the idle sibling that we found) in our
wake_affine() decisions. This domain is completely different from the
domain(we are supposed to use) that spans the cpu that the task currently
woken-up and the cpu where the task previously ran.
b) We do select_idle_sibling() check only for the cpu that the task is
currently woken-up on. If select_task_rq_fair() selects the previously run
cpu for waking the task, doing a select_idle_sibling() check
for that cpu also helps and we don't do this currently.
c) In the scenarios where the cpu that the task is woken-up is busy but
with its HT siblings are idle, we are selecting the task be woken-up
on the idle HT sibling instead of a core that it previously ran
and currently completely idle. i.e., we are not taking decisions based on
wake_affine() but directly selecting an idle sibling that can cause
an imbalance at the SMT/MC level which will be later corrected by the
periodic load balancer.
Fix this by first going through the load imbalance calculations using
wake_affine() and once we make a decision of woken-up cpu vs previously-ran cpu,
then choose a possible idle sibling for waking up the task on.
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1270079265.7835.8.camel@sbs-t61.sc.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Dave reported that his large SPARC machines spend lots of time in
hweight64(), try and optimize some of those needless cpumask_weight()
invocations (esp. with the large offstack cpumasks these are very
expensive indeed).
Reported-by: David Miller <davem@davemloft.net>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <new-submission>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Chase reported that due to us decrementing calc_load_task prematurely
(before the next LOAD_FREQ sample), the load average could be scewed
by as much as the number of CPUs in the machine.
This patch, based on Chase's patch, cures the problem by keeping the
delta of the CPU going into NO_HZ idle separately and folding that in
on the next LOAD_FREQ update.
This restores the balance and we get strict LOAD_FREQ period samples.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Acked-by: Chase Douglas <chase.douglas@canonical.com>
LKML-Reference: <1271934490.1776.343.camel@laptop>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Due to recent load-balancer changes that delay the task migration to
the next wakeup, the adaptive mutex spinning ends up in a live lock
when the owner's CPU gets offlined because the cpu_online() check
lives before the owner running check.
This patch changes mutex_spin_on_owner() to return 0 (don't spin) in
any case where we aren't sure about the owner struct validity or CPU
number, and if the said CPU is offline. There is no point going back &
re-evaluate spinning in corner cases like that, let's just go to
sleep.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1271212509.13059.135.camel@pasglop>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
