Add preempt off timings. A lot of kernel core code is taken from the RT patch
latency trace that was written by Ingo Molnar.
This adds "preemptoff" and "preemptirqsoff" to /debugfs/tracing/available_tracers
Now instead of just tracing irqs off, preemption off can be selected
to be recorded.
When this is selected, it shares the same files as irqs off timings.
One can either trace preemption off, irqs off, or one or the other off.
By echoing "preemptoff" into /debugfs/tracing/current_tracer, recording
of preempt off only is performed. "irqsoff" will only record the time
irqs are disabled, but "preemptirqsoff" will take the total time irqs
or preemption are disabled. Runtime switching of these options is now
supported by simpling echoing in the appropriate trace name into
/debugfs/tracing/current_tracer.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
This patch adds the latency tracer infrastructure. This patch
does not add anything that will select and turn it on, but will
be used by later patches.
If it were to be compiled, it would add the following files
to the debugfs:
The root tracing directory:
/debugfs/tracing/
This patch also adds the following files:
available_tracers
list of available tracers. Currently no tracers are
available. Looking into this file only shows
"none" which is used to unregister all tracers.
current_tracer
The trace that is currently active. Empty on start up.
To switch to a tracer simply echo one of the tracers that
are listed in available_tracers:
example: (used with later patches)
echo function > /debugfs/tracing/current_tracer
To disable the tracer:
echo disable > /debugfs/tracing/current_tracer
tracing_enabled
echoing "1" into this file starts the ftrace function tracing
(if sysctl kernel.ftrace_enabled=1)
echoing "0" turns it off.
latency_trace
This file is readonly and holds the result of the trace.
trace
This file outputs a easier to read version of the trace.
iter_ctrl
Controls the way the output of traces look.
So far there's two controls:
echoing in "symonly" will only show the kallsyms variables
without the addresses (if kallsyms was configured)
echoing in "verbose" will change the output to show
a lot more data, but not very easy to understand by
humans.
echoing in "nosymonly" turns off symonly.
echoing in "noverbose" turns off verbose.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
If CONFIG_FTRACE is selected and /proc/sys/kernel/ftrace_enabled is
set to a non-zero value the ftrace routine will be called everytime
we enter a kernel function that is not marked with the "notrace"
attribute.
The ftrace routine will then call a registered function if a function
happens to be registered.
[ This code has been highly hacked by Steven Rostedt and Ingo Molnar,
so don't blame Arnaldo for all of this ;-) ]
Update:
It is now possible to register more than one ftrace function.
If only one ftrace function is registered, that will be the
function that ftrace calls directly. If more than one function
is registered, then ftrace will call a function that will loop
through the functions to call.
Signed-off-by: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
The tracer wants to be able to convert the state number
into a user visible character. This patch pulls that conversion
string out the scheduler into the header. This way if it were to
ever change, other parts of the kernel will know.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
add 3 lightweight callbacks to the tracer backend.
zero impact if tracing is turned off.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Change references from for_each_cpu_mask to for_each_cpu_mask_nr
where appropriate
Reviewed-by: Paul Jackson <pj@sgi.com>
Reviewed-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Change references from for_each_cpu_mask to for_each_cpu_mask_nr
where appropriate
Reviewed-by: Paul Jackson <pj@sgi.com>
Reviewed-by: Christoph Lameter <clameter@sgi.com>
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
* 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.
For inclusion into sched-devel/latest tree.
Based on:
git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux-2.6.git
+ sched-devel/latest .../mingo/linux-2.6-sched-devel.git
Signed-off-by: Mike Travis <travis@sgi.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
On kvm I have seen some rare hangs in stop_machine when I used more guest
cpus than hosts cpus. e.g. 32 guest cpus on 1 host cpu triggered the
hang quite often. I could also reproduce the problem on a 4 way z/VM host with
a 64 way guest.
It turned out that the guest was consuming all available cpus mostly for
spinning on scheduler locks like rq->lock. This is expected as the threads are
calling yield all the time.
The problem is now, that the host scheduling decisings together with the guest
scheduling decisions and spinlocks not being fair managed to create an
interesting scenario similar to a live lock. (Sometimes the hang resolved
itself after some minutes)
Changing stop_machine to yield the cpu to the hypervisor when yielding inside
the guest fixed the problem for me. While I am not completely happy with this
patch, I think it causes no harm and it really improves the situation for me.
I used cpu_relax for yielding to the hypervisor, does that work on all
architectures?
p.s.: If you want to reproduce the problem, cpu hotplug and kprobes use
stop_machine_run and both triggered the problem after some retries.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
CC: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
The comment was correct -- need to make the code match the comment.
Without this patch, if a CPU goes dynticks idle (and stays there forever)
in just the right phase of preemptible-RCU grace-period processing,
grace periods stall. The offending sequence of events (courtesy
of Promela/spin, at least after I got the liveness criterion coded
correctly...) is as follows:
o CPU 0 is in dynticks-idle mode. Its dynticks_progress_counter
is (say) 10.
o CPU 0 takes an interrupt, so rcu_irq_enter() increments CPU 0's
dynticks_progress_counter to 11.
o CPU 1 is doing RCU grace-period processing in rcu_try_flip_idle(),
sees rcu_pending(), so invokes dyntick_save_progress_counter(),
which in turn takes a snapshot of CPU 0's dynticks_progress_counter
into CPU 0's rcu_dyntick_snapshot -- now set to 11. CPU 1 then
updates the RCU grace-period state to rcu_try_flip_waitack().
o CPU 0 returns from its interrupt, so rcu_irq_exit() increments
CPU 0's dynticks_progress_counter to 12.
o CPU 1 later invokes rcu_try_flip_waitack(), which notices that
CPU 0 has not yet responded, and hence in turn invokes
rcu_try_flip_waitack_needed(). This function examines the
state of CPU 0's dynticks_progress_counter and rcu_dyntick_snapshot
variables, which it copies to curr (== 12) and snap (== 11),
respectively.
Because curr!=snap, the first condition fails.
Because curr-snap is only 1 and snap is odd, the second
condition fails.
rcu_try_flip_waitack_needed() therefore incorrectly concludes
that it must wait for CPU 0 to explicitly acknowledge the
counter flip.
o CPU 0 remains forever in dynticks-idle mode, never taking
any more hardware interrupts or any NMIs, and never running
any more tasks. (Of course, -something- will usually eventually
happen, which might be why we haven't seen this one in the
wild. Still should be fixed!)
Therefore the grace period never ends. Fix is to make the code match
the comment, as shown below. With this fix, the above scenario
would be satisfied with curr being even, and allow the grace period
to proceed.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Josh Triplett <josh@kernel.org>
Cc: Dipankar Sarma <dipankar@in.ibm.com>
Cc: <stable@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Move rcu-protected lists from list.h into a new header file rculist.h.
This is done because list are a very used primitive structure all over the
kernel and it's currently impossible to include other header files in this
list.h without creating some circular dependencies.
For example, list.h implements rcu-protected list and uses rcu_dereference()
without including rcupdate.h. It actually compiles because users of
rcu_dereference() are macros. Others RCU functions could be used too but
aren't probably because of this.
Therefore this patch creates rculist.h which includes rcupdates without to
many changes/troubles.
Signed-off-by: Franck Bui-Huu <fbuihuu@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Josh Triplett <josh@kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add entry to rcu_torture_ops allowing the correct barrier function to
be used upon exit from rcutorture. Also add torture options for the
new call_rcu_sched() API.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Add rcu_barrier_sched() and rcu_barrier_bh(). With these in place,
rcutorture no longer gives the occasional oops when repeatedly starting
and stopping torturing rcu_bh. Also adds the API needed to flush out
pre-existing call_rcu_sched() callbacks.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Add comments to the logic that infers quiescent states when interrupting
from either user mode or the idle loop. Also add a memory barrier: it
appears that James Huang was in fact onto something, as the scheduler
is much less synchronization happy than it once was, so we can no longer
rely on its memory barriers in all cases.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Reported-by: James Huang <jamesclhuang@yahoo.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Fourth cut of patch to provide the call_rcu_sched(). This is again to
synchronize_sched() as call_rcu() is to synchronize_rcu().
Should be fine for experimental and -rt use, but not ready for inclusion.
With some luck, I will be able to tell Andrew to come out of hiding on
the next round.
Passes multi-day rcutorture sessions with concurrent CPU hotplugging.
Fixes since the first version include a bug that could result in
indefinite blocking (spotted by Gautham Shenoy), better resiliency
against CPU-hotplug operations, and other minor fixes.
Fixes since the second version include reworking grace-period detection
to avoid deadlocks that could happen when running concurrently with
CPU hotplug, adding Mathieu's fix to avoid the softlockup messages,
as well as Mathieu's fix to allow use earlier in boot.
Fixes since the third version include a wrong-CPU bug spotted by
Andrew, getting rid of the obsolete synchronize_kernel API that somehow
snuck back in, merging spin_unlock() and local_irq_restore() in a
few places, commenting the code that checks for quiescent states based
on interrupting from user-mode execution or the idle loop, removing
some inline attributes, and some code-style changes.
Known/suspected shortcomings:
o I still do not entirely trust the sleep/wakeup logic. Next step
will be to use a private snapshot of the CPU online mask in
rcu_sched_grace_period() -- if the CPU wasn't there at the start
of the grace period, we don't need to hear from it. And the
bit about accounting for changes in online CPUs inside of
rcu_sched_grace_period() is ugly anyway.
o It might be good for rcu_sched_grace_period() to invoke
resched_cpu() when a given CPU wasn't responding quickly,
but resched_cpu() is declared static...
This patch also fixes a long-standing bug in the earlier preemptable-RCU
implementation of synchronize_rcu() that could result in loss of
concurrent external changes to a task's CPU affinity mask. I still cannot
remember who reported this...
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
rcu_batches_completed and rcu_patches_completed_bh are both declared
in rcuclassic.h and rcupreempt.h. This patch removes the extra
prototypes for them from rcupdate.h.
rcu_batches_completed_bh is defined as a static inline in the rcupreempt.h
header file. Trying to export this as EXPORT_SYMBOL_GPL causes linking problems
with the powerpc linker. There's no need to export a static inlined function.
Modules must be compiled with the same type of RCU implementation as the
kernel they are for.
Signed-off-by: Steven Rostedt <srostedt@redhat.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
All uses of list_for_each_rcu() can be profitably replaced by the
easier-to-use list_for_each_entry_rcu(). This patch makes this change
for the Audit system, in preparation for removing the list_for_each_rcu()
API entirely. This time with well-formed SOB.
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Limit sysctl_nr_open - we don't want ->max_fds to exceed MAX_INT and
we don't want size calculation for ->fd[] to overflow.
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Currently, the mutex debug code checks the lock->owner before lock->magic, so
a corrupt mutex will most likely result in failing the owner check, rather
than the magic check.
This change to debug_mutex_unlock does the magic check first, so
we have a better idea of what breaks.
Signed-off-by: Jeremy Kerr <jk@ozlabs.org>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add a common hex array in hexdump.c so everyone can use it.
Add a common hi/lo helper to avoid the shifting masking that is
done to get the upper and lower nibbles of a byte value.
Pull the pack_hex_byte helper from kgdb as it is opencoded many
places in the tree that will be consolidated.
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com>
Acked-by: Paul Mundt <lethal@linux-sh.org>
Cc: Jason Wessel <jason.wessel@windriver.com>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
It acts exactly like a regular 'cond_resched()', but will not get
optimized away when CONFIG_PREEMPT is set.
Normal kernel code is already preemptable in the presense of
CONFIG_PREEMPT, so cond_resched() is optimized away (see commit
02b67cc3ba "sched: do not do
cond_resched() when CONFIG_PREEMPT").
But when wanting to conditionally reschedule while holding a lock, you
need to use "cond_sched_lock(lock)", and the new function is the BKL
equivalent of that.
Also make fs/locks.c use it.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The generic semaphore rewrite had a huge performance regression on AIM7
(and potentially other BKL-heavy benchmarks) because the generic
semaphores had been rewritten to be simple to understand and fair. The
latter, in particular, turns a semaphore-based BKL implementation into a
mess of scheduling.
The attempt to fix the performance regression failed miserably (see the
previous commit 00b41ec261 'Revert
"semaphore: fix"'), and so for now the simple and sane approach is to
instead just go back to the old spinlock-based BKL implementation that
never had any issues like this.
This patch also has the advantage of being reported to fix the
regression completely according to Yanmin Zhang, unlike the semaphore
hack which still left a couple percentage point regression.
As a spinlock, the BKL obviously has the potential to be a latency
issue, but it's not really any different from any other spinlock in that
respect. We do want to get rid of the BKL asap, but that has been the
plan for several years.
These days, the biggest users are in the tty layer (open/release in
particular) and Alan holds out some hope:
"tty release is probably a few months away from getting cured - I'm
afraid it will almost certainly be the very last user of the BKL in
tty to get fixed as it depends on everything else being sanely locked."
so while we're not there yet, we do have a plan of action.
Tested-by: Yanmin Zhang <yanmin_zhang@linux.intel.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Matthew Wilcox <matthew@wil.cx>
Cc: Alexander Viro <viro@ftp.linux.org.uk>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This reverts commit bf726eab37, as it has
been reported to cause a regression with processes stuck in __down(),
apparently because some missing wakeup.
Quoth Sven Wegener:
"I'm currently investigating a regression that has showed up with my
last git pull yesterday. Bisecting the commits showed bf726e
"semaphore: fix" to be the culprit, reverting it fixed the issue.
Symptoms: During heavy filesystem usage (e.g. a kernel compile) I get
several compiler processes in uninterruptible sleep, blocking all i/o
on the filesystem. System is an Intel Core 2 Quad running a 64bit
kernel and userspace. Filesystem is xfs on top of lvm. See below for
the output of sysrq-w."
See
http://lkml.org/lkml/2008/5/10/45
for full report.
In the meantime, we can just fix the BKL performance regression by
reverting back to the good old BKL spinlock implementation instead,
since any sleeping lock will generally perform badly, especially if it
tries to be fair.
Reported-by: Sven Wegener <sven.wegener@stealer.net>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Linus found a logic bug: we ignore the version number in a module's
vermagic string if we have CONFIG_MODVERSIONS set, but modversions
also lets through a module with no __versions section for modprobe
--force (with tainting, but still).
We should only ignore the start of the vermagic string if the module
actually *has* crcs to check. Rather than (say) having an
entertaining hissy fit and creating a config option to work around the
buggy code.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
We allow missing __versions sections, because modprobe --force strips
it. It makes less sense to allow sections where there's no version
for a specific symbol the module uses, so disallow that.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* 'for-linus' of git://git.kernel.dk/linux-2.6-block:
Revert "relay: fix splice problem"
docbook: fix bio missing parameter
block: use unitialized_var() in bio_alloc_bioset()
block: avoid duplicate calls to get_part() in disk stat code
cfq-iosched: make io priorities inherit CPU scheduling class as well as nice
block: optimize generic_unplug_device()
block: get rid of likely/unlikely predictions in merge logic
vfs: splice remove_suid() cleanup
cfq-iosched: fix RCU race in the cfq io_context destructor handling
block: adjust tagging function queue bit locking
block: sysfs store function needs to grab queue_lock and use queue_flag_*()
Due to a merge conflict, the sched_relax_domain_level control file was marked
as being handled by cpuset_read/write_u64, but the code to handle it was
actually in cpuset_common_file_read/write.
Since the value being written/read is in fact a signed integer, it should be
treated as such; this patch adds cpuset_read/write_s64 functions, and uses
them to handle the sched_relax_domain_level file.
With this patch, the sched_relax_domain_level can be read and written, and the
correct contents seen/updated.
Signed-off-by: Paul Menage <menage@google.com>
Cc: Hidetoshi Seto <seto.hidetoshi@jp.fujitsu.com>
Cc: Paul Jackson <pj@sgi.com>
Cc: Ingo Molnar <mingo@elte.hu>
Reviewed-by: Li Zefan <lizf@cn.fujitsu.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The conversion between virtual and real time is as follows:
dvt = rw/w * dt <=> dt = w/rw * dvt
Since we want the fair sleeper granularity to be in real time, we actually
need to do:
dvt = - rw/w * l
This bug could be related to the regression reported by Yanmin Zhang:
| Comparing with kernel 2.6.25, sysbench+mysql(oltp, readonly) has lots
| of regressions with 2.6.26-rc1:
|
| 1) 8-core stoakley: 28%;
| 2) 16-core tigerton: 20%;
| 3) Itanium Montvale: 50%.
Reported-by: "Zhang, Yanmin" <yanmin_zhang@linux.intel.com>
Signed-off-by: Mike Galbraith <efault@gmx.de>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Yanmin Zhang reported:
| Comparing with kernel 2.6.25, AIM7 (use tmpfs) has more th
| regression under 2.6.26-rc1 on my 8-core stoakley, 16-core tigerton,
| and Itanium Montecito. Bisect located the patch below:
|
| 64ac24e738 is first bad commit
| commit 64ac24e738
| Author: Matthew Wilcox <matthew@wil.cx>
| Date: Fri Mar 7 21:55:58 2008 -0500
|
| Generic semaphore implementation
|
| After I manually reverted the patch against 2.6.26-rc1 while fixing
| lots of conflicts/errors, aim7 regression became less than 2%.
i reproduced the AIM7 workload and can confirm Yanmin's findings that
-.26-rc1 regresses over .25 - by over 67% here.
Looking at the workload i found and fixed what i believe to be the real
bug causing the AIM7 regression: it was inefficient wakeup / scheduling
/ locking behavior of the new generic semaphore code, causing suboptimal
performance.
The problem comes from the following code. The new semaphore code does
this on down():
spin_lock_irqsave(&sem->lock, flags);
if (likely(sem->count > 0))
sem->count--;
else
__down(sem);
spin_unlock_irqrestore(&sem->lock, flags);
and this on up():
spin_lock_irqsave(&sem->lock, flags);
if (likely(list_empty(&sem->wait_list)))
sem->count++;
else
__up(sem);
spin_unlock_irqrestore(&sem->lock, flags);
where __up() does:
list_del(&waiter->list);
waiter->up = 1;
wake_up_process(waiter->task);
and where __down() does this in essence:
list_add_tail(&waiter.list, &sem->wait_list);
waiter.task = task;
waiter.up = 0;
for (;;) {
[...]
spin_unlock_irq(&sem->lock);
timeout = schedule_timeout(timeout);
spin_lock_irq(&sem->lock);
if (waiter.up)
return 0;
}
the fastpath looks good and obvious, but note the following property of
the contended path: if there's a task on the ->wait_list, the up() of
the current owner will "pass over" ownership to that waiting task, in a
wake-one manner, via the waiter->up flag and by removing the waiter from
the wait list.
That is all and fine in principle, but as implemented in
kernel/semaphore.c it also creates a nasty, hidden source of contention!
The contention comes from the following property of the new semaphore
code: the new owner owns the semaphore exclusively, even if it is not
running yet.
So if the old owner, even if just a few instructions later, does a
down() [lock_kernel()] again, it will be blocked and will have to wait
on the new owner to eventually be scheduled (possibly on another CPU)!
Or if another task gets to lock_kernel() sooner than the "new owner"
scheduled, it will be blocked unnecessarily and for a very long time
when there are 2000 tasks running.
I.e. the implementation of the new semaphores code does wake-one and
lock ownership in a very restrictive way - it does not allow
opportunistic re-locking of the lock at all and keeps the scheduler from
picking task order intelligently.
This kind of scheduling, with 2000 AIM7 processes running, creates awful
cross-scheduling between those 2000 tasks, causes reduced parallelism, a
throttled runqueue length and a lot of idle time. With increasing number
of CPUs it causes an exponentially worse behavior in AIM7, as the chance
for a newly woken new-owner task to actually run anytime soon is less
and less likely.
Note that it takes just a tiny bit of contention for the 'new-semaphore
catastrophy' to happen: the wakeup latencies get added to whatever small
contention there is, and quickly snowball out of control!
I believe Yanmin's findings and numbers support this analysis too.
The best fix for this problem is to use the same scheduling logic that
the kernel/mutex.c code uses: keep the wake-one behavior (that is OK and
wanted because we do not want to over-schedule), but also allow
opportunistic locking of the lock even if a wakee is already "in
flight".
The patch below implements this new logic. With this patch applied the
AIM7 regression is largely fixed on my quad testbox:
# v2.6.25 vanilla:
..................
Tasks Jobs/Min JTI Real CPU Jobs/sec/task
2000 56096.4 91 207.5 789.7 0.4675
2000 55894.4 94 208.2 792.7 0.4658
# v2.6.26-rc1-166-gc0a1811 vanilla:
...................................
Tasks Jobs/Min JTI Real CPU Jobs/sec/task
2000 33230.6 83 350.3 784.5 0.2769
2000 31778.1 86 366.3 783.6 0.2648
# v2.6.26-rc1-166-gc0a1811 + semaphore-speedup:
...............................................
Tasks Jobs/Min JTI Real CPU Jobs/sec/task
2000 55707.1 92 209.0 795.6 0.4642
2000 55704.4 96 209.0 796.0 0.4642
i.e. a 67% speedup. We are now back to within 1% of the v2.6.25
performance levels and have zero idle time during the test, as expected.
Btw., interactivity also improved dramatically with the fix - for
example console-switching became almost instantaneous during this
workload (which after all is running 2000 tasks at once!), without the
patch it was stuck for a minute at times.
There's another nice side-effect of this speedup patch, the new generic
semaphore code got even smaller:
text data bss dec hex filename
1241 0 0 1241 4d9 semaphore.o.before
1207 0 0 1207 4b7 semaphore.o.after
(because the waiter.up complication got removed.)
Longer-term we should look into using the mutex code for the generic
semaphore code as well - but i's not easy due to legacies and it's
outside of the scope of v2.6.26 and outside the scope of this patch as
well.
Bisected-by: "Zhang, Yanmin" <yanmin_zhang@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
this replaces the rq->clock stuff (and possibly cpu_clock()).
- architectures that have an 'imperfect' hardware clock can set
CONFIG_HAVE_UNSTABLE_SCHED_CLOCK
- the 'jiffie' window might be superfulous when we update tick_gtod
before the __update_sched_clock() call in sched_clock_tick()
- cpu_clock() might be implemented as:
sched_clock_cpu(smp_processor_id())
if the accuracy proves good enough - how far can TSC drift in a
single jiffie when considering the filtering and idle hooks?
[ mingo@elte.hu: various fixes and cleanups ]
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
David Miller pointed it out that nothing in cpu_clock() sets
prev_cpu_time. This caused __sync_cpu_clock() to be called
all the time - against the intention of this code.
The result was that in practice we hit a global spinlock every
time cpu_clock() is called - which - even though cpu_clock()
is used for tracing and debugging, is suboptimal.
While at it, also:
- move the irq disabling to the outest layer,
this should make cpu_clock() warp-free when called with irqs
enabled.
- use long long instead of cycles_t - for platforms where cycles_t
is 32-bit.
Reported-by: David Miller <davem@davemloft.net>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
When I echoed 0 into the "cpu.shares" file, a Div0 error occured.
We found it is caused by the following calling.
sched_group_set_shares(tg, shares)
set_se_shares(tg->se[i], shares/nr_cpu_ids)
__set_se_shares(se, shares)
div64_64((1ULL<<32), shares)
When the echoed value was less than the number of processores, the result of the
sentence "shares/nr_cpu_ids" was 0, and then the system called div64() to divide
the result, the Div0 error occured.
It is unnecessary that the shares value is divided by nr_cpu_ids, I think.
Because in the function __update_group_shares_cpu() and init_tg_cfs_entry(),
the shares value isn't divided by nr_cpu_ids when setting shares of the sched
entity.
This patch fixes this bug. And echoing ULONG_MAX value into cpu.shares also
causes Div0 error, so we set a macro MAX_SHARES to limit the max value of
shares.
Signed-off-by: Miao Xie <miaox@cn.fujitsu.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Concurrent calls to detach_destroy_domains and arch_init_sched_domains
were prevented by the old scheduler subsystem cpu hotplug mutex. When
this got converted to get_online_cpus() the locking got broken.
Unlike before now several processes can concurrently enter the critical
sections that were protected by the old lock.
So use the already present doms_cur_mutex to protect these sections again.
Cc: Gautham R Shenoy <ego@in.ibm.com>
Cc: Paul Jackson <pj@sgi.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Revert debugging commit 7ba2e74ab5.
print_cfs_rq_tasks() can induce live-lock if a task is dequeued
during list traversal.
Signed-off-by: Mike Galbraith <efault@gmx.de>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
