Linus correctly observes that the most important dispatch cases
are now done from kblockd, this isn't ideal for latency reasons.
The original reason for switching dispatches out-of-line was to
avoid too deep a stack, so by _only_ letting the "accidental"
flush directly in schedule() be guarded by offload to kblockd,
we should be able to get the best of both worlds.
So add a blk_schedule_flush_plug() that offloads to kblockd,
and only use that from the schedule() path.
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
Now that we've removed the rq->lock requirement from the first part of
ttwu() and can compute placement without holding any rq->lock, ensure
we execute the second half of ttwu() on the actual cpu we want the
task to run on.
This avoids having to take rq->lock and doing the task enqueue
remotely, saving lots on cacheline transfers.
As measured using: http://oss.oracle.com/~mason/sembench.c
$ for i in /sys/devices/system/cpu/cpu*/cpufreq/scaling_governor ; do echo performance > $i; done
$ echo 4096 32000 64 128 > /proc/sys/kernel/sem
$ ./sembench -t 2048 -w 1900 -o 0
unpatched: run time 30 seconds 647278 worker burns per second
patched: run time 30 seconds 816715 worker burns per second
Reviewed-by: Frank Rowand <frank.rowand@am.sony.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20110405152729.515897185@chello.nl
Currently ttwu() does two rq->lock acquisitions, once on the task's
old rq, holding it over the p->state fiddling and load-balance pass.
Then it drops the old rq->lock to acquire the new rq->lock.
By having serialized ttwu(), p->sched_class, p->cpus_allowed with
p->pi_lock, we can now drop the whole first rq->lock acquisition.
The p->pi_lock serializing concurrent ttwu() calls protects p->state,
which we will set to TASK_WAKING to bridge possible p->pi_lock to
rq->lock gaps and serialize set_task_cpu() calls against
task_rq_lock().
The p->pi_lock serialization of p->sched_class allows us to call
scheduling class methods without holding the rq->lock, and the
serialization of p->cpus_allowed allows us to do the load-balancing
bits without races.
Reviewed-by: Frank Rowand <frank.rowand@am.sony.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/20110405152729.354401150@chello.nl
We really only want to unplug the pending IO when the process actually
goes to sleep. So move the test for flushing the plug up to the place
where we actually deactivate the task - where we have properly checked
for preemption and for the process really sleeping.
Acked-by: Jens Axboe <jaxboe@fusionio.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Instead of relying on static allocations for the sched_domain and
sched_group trees, dynamically allocate and RCU free them.
Allocating this dynamically also allows for some build_sched_groups()
simplification since we can now (like with other simplifications) rely
on the sched_domain tree instead of hard-coded knowledge.
One tricky to note is that detach_destroy_domains() needs to hold
rcu_read_lock() over the entire tear-down, per-cpu is not sufficient
since that can lead to partial sched_group existance (could possibly
be solved by doing the tear-down backwards but this is much more
robust).
A concequence of the above is that we can no longer print the
sched_domain debug stuff from cpu_attach_domain() since that might now
run with preemption disabled (due to classic RCU etc.) and
sched_domain_debug() does some GFP_KERNEL allocations.
Another thing to note is that we now fully rely on normal RCU and not
RCU-sched, this is because with the new and exiting RCU flavours we
grew over the years BH doesn't necessarily hold off RCU-sched grace
periods (-rt is known to break this). This would in fact already cause
us grief since we do sched_domain/sched_group iterations from softirq
context.
This patch is somewhat larger than I would like it to be, but I didn't
find any means of shrinking/splitting this.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20110407122942.245307941@chello.nl
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Again, instead of relying on knowing the possible domains and their
order, simply rely on the sched_domain tree and whatever domains are
present in there to initialize the sched_group cpu_power.
Note: we need to iterate the CPU mask backwards because of the
cpumask_first() condition for iterating up the tree. By iterating the
mask backwards we ensure all groups of a domain are set-up before
starting on the parent groups that rely on its children to be
completely done.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20110407122942.187335414@chello.nl
Signed-off-by: Ingo Molnar <mingo@elte.hu>
The NODE sched_domain is 'special' in that it allocates sched_groups
per CPU, instead of sharing the sched_groups between all CPUs.
While this might have some benefits on large NUMA and avoid remote
memory accesses when iterating the sched_groups, this does break
current code that assumes sched_groups are shared between all
sched_domains (since the dynamic cpu_power patches).
So refactor the NODE groups to behave like all other groups.
(The ALLNODES domain again shared its groups across the CPUs for some
reason).
If someone does measure a performance decrease due to this change we
need to revisit this and come up with another way to have both dynamic
cpu_power and NUMA work nice together.
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Nick Piggin <npiggin@kernel.dk>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: http://lkml.kernel.org/r/20110407122941.978111700@chello.nl
Signed-off-by: Ingo Molnar <mingo@elte.hu>
* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
x86-32, fpu: Fix FPU exception handling on non-SSE systems
x86, hibernate: Initialize mmu_cr4_features during boot
x86-32, NUMA: Fix ACPI NUMA init broken by recent x86-64 change
x86: visws: Fixup irq overhaul fallout
* 'sched-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
sched: Clean up rebalance_domains() load-balance interval calculation
* 'timers-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
x86/mrst/vrtc: Fix boot crash in mrst_rtc_init()
rtc, x86/mrst/vrtc: Fix boot crash in rtc_read_alarm()
* 'irq-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
genirq: Fix cpumask leak in __setup_irq()
* 'perf-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
perf probe: Fix listing incorrect line number with inline function
perf probe: Fix to find recursively inlined function
perf probe: Fix multiple --vars options behavior
perf probe: Fix to remove redundant close
perf probe: Fix to ensure function declared file
Instead of the possible multiple-evaluation of num_online_cpus()
in rebalance_domains() that Linus reported, avoid it altogether
in the normal case since it's implemented with a Hamming weight
function over a cpu bitmask which can be darn expensive for those
with big iron.
This also makes it cleaner, smaller and documents the code.
Reported-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1301991265.2225.12.camel@twins>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
sched_setscheduler() (in sched.c) is called in order of changing the
scheduling policy and/or the real-time priority of a task. Thus,
if we find out that neither of those are actually being modified, it
is possible to return earlier and save the overhead of a full
deactivate+activate cycle of the task in question.
Beside that, if we have more than one SCHED_FIFO task with the same
priority on the same rq (which means they share the same priority queue)
having one of them changing its position in the priority queue because of
a sched_setscheduler (as it happens by means of the deactivate+activate)
that does not actually change the priority violates POSIX which states,
for SCHED_FIFO:
"If a thread whose policy or priority has been modified by
pthread_setschedprio() is a running thread or is runnable, the effect on
its position in the thread list depends on the direction of the
modification, as follows: a. <...> b. If the priority is unchanged, the
thread does not change position in the thread list. c. <...>"
http://pubs.opengroup.org/onlinepubs/009695399/functions/xsh_chap02_08.html
(ed: And the POSIX specification here does, briefly and somewhat unexpectedly,
match what common sense tells us as well. )
Signed-off-by: Dario Faggioli <raistlin@linux.it>
Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1300971618.3960.82.camel@Palantir>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
