firefly-linux-kernel-4.4.55.git
9 years agotime: Allow gcc to fold constants when possible
Nicholas Mc Guire [Mon, 18 May 2015 12:19:14 +0000 (14:19 +0200)]
time: Allow gcc to fold constants when possible

To allow constant folding in msecs_to_jiffies() conditionally calls
the HZ dependent _msecs_to_jiffies() helpers or, when gcc can not
figure out constant folding, __msecs_to_jiffies which is the renamed
original msecs_to_jiffies() function.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Cc: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Joe Perches <joe@perches.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Andrew Hunter <ahh@google.com>
Cc: Paul Turner <pjt@google.com>
Cc: Michal Marek <mmarek@suse.cz>
Link: http://lkml.kernel.org/r/1431951554-5563-3-git-send-email-hofrat@osadl.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotime: Refactor msecs_to_jiffies
Nicholas Mc Guire [Mon, 18 May 2015 12:19:13 +0000 (14:19 +0200)]
time: Refactor msecs_to_jiffies

Refactor the msecs_to_jiffies conditional code part in time.c and
jiffies.h putting it into conditional functions rather than #ifdefs
to improve readability.

[ tglx: Verified that there is no binary code change ]

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Cc: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Joe Perches <joe@perches.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Andrew Hunter <ahh@google.com>
Cc: Paul Turner <pjt@google.com>
Cc: Michal Marek <mmarek@suse.cz>
Link: http://lkml.kernel.org/r/1431951554-5563-2-git-send-email-hofrat@osadl.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotime: Move timeconst.h into include/generated
Nicholas Mc Guire [Mon, 18 May 2015 12:19:12 +0000 (14:19 +0200)]
time: Move timeconst.h into include/generated

kernel/time/timeconst.h is moved to include/generated/ and generated
by the top level Kbuild. This allows using timeconst.h in an earlier
build stage.

Signed-off-by: Nicholas Mc Guire <hofrat@osadl.org>
Cc: Masahiro Yamada <yamada.m@jp.panasonic.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Joe Perches <joe@perches.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Andrew Hunter <ahh@google.com>
Cc: Paul Turner <pjt@google.com>
Cc: Michal Marek <mmarek@suse.cz>
Link: http://lkml.kernel.org/r/1431951554-5563-1-git-send-email-hofrat@osadl.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agosched,perf: Fix periodic timers
Peter Zijlstra [Thu, 14 May 2015 10:23:11 +0000 (12:23 +0200)]
sched,perf: Fix periodic timers

In the below two commits (see Fixes) we have periodic timers that can
stop themselves when they're no longer required, but need to be
(re)-started when their idle condition changes.

Further complications is that we want the timer handler to always do
the forward such that it will always correctly deal with the overruns,
and we do not want to race such that the handler has already decided
to stop, but the (external) restart sees the timer still active and we
end up with a 'lost' timer.

The problem with the current code is that the re-start can come before
the callback does the forward, at which point the forward from the
callback will WARN about forwarding an enqueued timer.

Now, conceptually its easy to detect if you're before or after the fwd
by comparing the expiration time against the current time. Of course,
that's expensive (and racy) because we don't have the current time.

Alternatively one could cache this state inside the timer, but then
everybody pays the overhead of maintaining this extra state, and that
is undesired.

The only other option that I could see is the external timer_active
variable, which I tried to kill before. I would love a nicer interface
for this seemingly simple 'problem' but alas.

Fixes: 272325c4821f ("perf: Fix mux_interval hrtimer wreckage")
Fixes: 77a4d1a1b9a1 ("sched: Cleanup bandwidth timers")
Cc: pjt@google.com
Cc: tglx@linutronix.de
Cc: klamm@yandex-team.ru
Cc: mingo@kernel.org
Cc: bsegall@google.com
Cc: hpa@zytor.com
Cc: Sasha Levin <sasha.levin@oracle.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20150514102311.GX21418@twins.programming.kicks-ass.net
9 years agoALSA: drivers: pcsp: Fix printout of resolution
Thomas Gleixner [Tue, 12 May 2015 07:14:51 +0000 (09:14 +0200)]
ALSA: drivers: pcsp: Fix printout of resolution

The recent conversion of the hrtimer resolution failed to convert the
printk format from %li to %u.

Reported-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agonohz: Fix !HIGH_RES_TIMERS hang
Thomas Gleixner [Thu, 7 May 2015 12:35:59 +0000 (14:35 +0200)]
nohz: Fix !HIGH_RES_TIMERS hang

Simon Horman reported this crash on a system with
high-res timers disabled but nohz enabled:

  > ------------[ cut here ]------------
  > kernel BUG at kernel/irq_work.c:135!

    BUG_ON(!irqs_disabled());

So something enabled interrupts in the periodic tick handling machinery,
and that code path indeed has a local_irq_disable()/enable pair in
tick_nohz_switch_to_nohz() which causes havoc. Fix it.

This patch also fixes a +nohz -hrtimers hang reported by Ingo Molnar.

Reported-by: Simon Horman <horms@verge.net.au>
Reported-by: Ingo Molnar <mingo@kernel.org>
Tested-by: Simon Horman <horms@verge.net.au>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Borislav Petkov <bp@alien8.de>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: LAK <linux-arm-kernel@lists.infradead.org>
Cc: Magnus Damm <magnus.damm@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/alpine.DEB.2.11.1505071425520.4225@nanos
Signed-off-by: Ingo Molnar <mingo@kernel.org>
9 years agotick: hrtimer-broadcast: Prevent endless restarting when broadcast device is unused
Andreas Sandberg [Fri, 24 Apr 2015 13:06:05 +0000 (13:06 +0000)]
tick: hrtimer-broadcast: Prevent endless restarting when broadcast device is unused

The hrtimer callback in the hrtimer's tick broadcast code sometimes
incorrectly ends up scheduling events at the current tick causing the
kernel to hang servicing the same hrtimer forever. This typically
happens when a device is swapped out by
tick_install_broadcast_device(), which replaces the event handler with
clock_events_handle_noop() and sets the device mode to
CLOCK_EVT_MODE_UNUSED. If the timer is scheduled when this happens,
the next_event field will not be updated and the hrtimer ends up being
restarted at the current tick. To prevent this from happening, only
try to restart the hrtimer if the broadcast clock event device is in
one of the active modes and try to cancel the timer when entering the
CLOCK_EVT_MODE_UNUSED mode.

Signed-off-by: Andreas Sandberg <andreas.sandberg@arm.com>
Tested-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Link: http://lkml.kernel.org/r/1429880765-5558-1-git-send-email-andreas.sandberg@arm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotimer: Use timer->base for flag checks
Joonwoo Park [Tue, 28 Apr 2015 02:21:49 +0000 (19:21 -0700)]
timer: Use timer->base for flag checks

At present, internal_add_timer() examines flags with 'base' which doesn't
contain flags.  Examine with 'timer->base' to avoid unnecessary waking up
of nohz CPU when timer base has TIMER_DEFERRABLE set.

Signed-off-by: Joonwoo Park <joonwoop@codeaurora.org>
Cc: sboyd@codeaurora.org
Cc: skannan@codeaurora.org
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/1430187709-21087-1-git-send-email-joonwoop@codeaurora.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agoclocksource: sun5i: Fix of_io_request_and_map error check
Maxime Ripard [Sat, 2 May 2015 15:03:25 +0000 (17:03 +0200)]
clocksource: sun5i: Fix of_io_request_and_map error check

of_io_request_and map returns an error pointer, but the current code assumes
that on error the returned pointer will be NULL.

Obviously, that makes the check completely useless. Change the test to actually
check for the proper error code.

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: http://lkml.kernel.org/r/1430579006-32702-6-git-send-email-maxime.ripard@free-electrons.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agoclocksource: integrator: Fix of_io_request_and_map error check
Maxime Ripard [Sat, 2 May 2015 15:03:24 +0000 (17:03 +0200)]
clocksource: integrator: Fix of_io_request_and_map error check

of_io_request_and map returns an error pointer, but the current code assumes
that on error the returned pointer will be NULL.

Obviously, that makes the check completely useless. Change the test to actually
check for the proper error code.

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: http://lkml.kernel.org/r/1430579006-32702-5-git-send-email-maxime.ripard@free-electrons.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agoclocksource: asm9260: Fix of_io_request_and_map error check
Maxime Ripard [Sat, 2 May 2015 15:03:23 +0000 (17:03 +0200)]
clocksource: asm9260: Fix of_io_request_and_map error check

of_io_request_and map returns an error pointer, but the current code assumes
that on error the returned pointer will be NULL.

Obviously, that makes the check completely useless. Change the test to actually
check for the proper error code.

Signed-off-by: Maxime Ripard <maxime.ripard@free-electrons.com>
Cc: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: http://lkml.kernel.org/r/1430579006-32702-4-git-send-email-maxime.ripard@free-electrons.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotick-broadcast: Fix the printing of broadcast masks
Preeti U Murthy [Tue, 28 Apr 2015 08:45:20 +0000 (14:15 +0530)]
tick-broadcast: Fix the printing of broadcast masks

Today the number of bits of the broadcast masks that is output into
/proc/timer_list is sizeof(unsigned long). This means that on machines
with a larger number of CPUs, the bitmasks of CPUs beyond this range do
not appear.

Fix this by using bitmap printing through "%*pb" instead, so as to
output the broadcast masks for the range of nr_cpu_ids into
/proc/timer_list.

Signed-off-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: peterz@infradead.org
Cc: linuxppc-dev@ozlabs.org
Cc: john.stultz@linaro.org
Link: http://lkml.kernel.org/r/20150428084520.3314.62668.stgit@preeti.in.ibm.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotick: broadcast: Simplify oneshot logic and shorten lock region
Thomas Gleixner [Tue, 5 May 2015 07:44:24 +0000 (09:44 +0200)]
tick: broadcast: Simplify oneshot logic and shorten lock region

Simplify the oneshot logic by avoiding the reprogramming loops. That
also allows to call the cpu local handler outside of the
broadcast_lock held region.

Tested-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotick: broadcast: Prevent livelock from event handler
Thomas Gleixner [Tue, 5 May 2015 08:00:13 +0000 (10:00 +0200)]
tick: broadcast: Prevent livelock from event handler

With the removal of the hrtimer softirq the switch to highres/nohz
mode happens in the tick interrupt. That leads to a livelock when the
per cpu event handler is directly called from the broadcast handler
under broadcast lock because broadcast lock needs to be taken for the
highres/nohz switch as well.

Solve this by calling the cpu local handler outside the broadcast_lock
held region.

Fixes: c6eb3f70d448 "hrtimer: Get rid of hrtimer softirq"
Reported-and-tested-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agoperf: Remove unused function perf_mux_hrtimer_cancel()
Thomas Gleixner [Mon, 4 May 2015 11:51:12 +0000 (13:51 +0200)]
perf: Remove unused function perf_mux_hrtimer_cancel()

Reported-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agopower: reset: ltc2952: Remove bogus hrtimer_start() return value checks
Thomas Gleixner [Mon, 13 Apr 2015 14:43:35 +0000 (16:43 +0200)]
power: reset: ltc2952: Remove bogus hrtimer_start() return value checks

The return value of hrtimer_start() tells whether the timer was
inactive or active already when hrtimer_start() was called.

The code emits a bogus warning if the timer was active already
claiming that the timer could not be started.

Remove it along with the bogus comment in the else path.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Frans Klaver <frans.klaver@xsens.com>
Cc: Sebastian Reichel <sre@kernel.org>
Cc: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Wolfram Sang <wsa@the-dreams.de>
Cc: linux-pm@vger.kernel.org
9 years agosched: debug: Remove the cfs bandwidth timer_active printout
Thomas Gleixner [Thu, 23 Apr 2015 11:58:09 +0000 (13:58 +0200)]
sched: debug: Remove the cfs bandwidth timer_active printout

The struct member is gone.

Reported-by: fengguang.wu@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
9 years agoperf: perf_mux_hrtimer_cancel() can be static
kbuild test robot [Wed, 22 Apr 2015 20:00:00 +0000 (04:00 +0800)]
perf: perf_mux_hrtimer_cancel() can be static

Signed-off-by: Fengguang Wu <fengguang.wu@intel.com>
Cc: kbuild-all@01.org
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20150422200000.GA122603@lkp-sb04
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agoperf: Fix mux_interval hrtimer wreckage
Peter Zijlstra [Wed, 15 Apr 2015 09:41:58 +0000 (11:41 +0200)]
perf: Fix mux_interval hrtimer wreckage

Thomas stumbled over the hrtimer_forward_now() in
perf_event_mux_interval_ms_store() and noticed its broken-ness.

You cannot just change the expiry time of an active timer, it will
destroy the red-black tree order and cause havoc.

Change it to (re)start the timer instead, (re)starting a timer will
dequeue and enqueue a timer and therefore preserve rb-tree order.

Since we cannot enqueue remotely, wrap the thing in
cpu_function_call(), this however mandates that we restrict ourselves
to online cpus. Also serialize the entire setting so we don't get
multiple concurrent threads trying to update to different values.

Also fix a problem in perf_mux_hrtimer_restart(), checking against
hrtimer_active() can actually loose us the timer when timer->state ==
HRTIMER_STATE_CALLBACK and the callback has already decided NORESTART.

Furthermore it doesn't make any sense to test
hrtimer_callback_running() when we already tested hrtimer_active(),
but with the above change, we explicitly must call it when
callback_running.

Lastly, rename a few functions:

  s/perf_cpu_hrtimer_/perf_mux_hrtimer_/ -- because I could not find
                                            the mux timer function

  s/\<hr\>/timer/ -- because that's the normal way of calling things.

Fixes: 62b856397927 ("perf: Add sysfs entry to adjust multiplexing interval per PMU")
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20150415095011.863052571@infradead.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agosched: Cleanup bandwidth timers
Peter Zijlstra [Wed, 15 Apr 2015 09:41:57 +0000 (11:41 +0200)]
sched: Cleanup bandwidth timers

Roman reported a 3 cpu lockup scenario involving __start_cfs_bandwidth().

The more I look at that code the more I'm convinced its crack, that
entire __start_cfs_bandwidth() thing is brain melting, we don't need to
cancel a timer before starting it, *hrtimer_start*() will happily remove
the timer for you if its still enqueued.

Removing that, removes a big part of the problem, no more ugly cancel
loop to get stuck in.

So now, if I understand things right, the entire reason you have this
cfs_b->lock guarded ->timer_active nonsense is to make sure we don't
accidentally lose the timer.

It appears to me that it should be possible to guarantee that same by
unconditionally (re)starting the timer when !queued. Because regardless
what hrtimer::function will return, if we beat it to (re)enqueue the
timer, it doesn't matter.

Now, because hrtimers don't come with any serialization guarantees we
must ensure both handler and (re)start loop serialize their access to
the hrtimer to avoid both trying to forward the timer at the same
time.

Update the rt bandwidth timer to match.

This effectively reverts: 09dc4ab03936 ("sched/fair: Fix
tg_set_cfs_bandwidth() deadlock on rq->lock").

Reported-by: Roman Gushchin <klamm@yandex-team.ru>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Ben Segall <bsegall@google.com>
Cc: Paul Turner <pjt@google.com>
Link: http://lkml.kernel.org/r/20150415095011.804589208@infradead.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Allow concurrent hrtimer_start() for self restarting timers
Peter Zijlstra [Tue, 20 May 2014 13:49:48 +0000 (15:49 +0200)]
hrtimer: Allow concurrent hrtimer_start() for self restarting timers

Because we drop cpu_base->lock around calling hrtimer::function, it is
possible for hrtimer_start() to come in between and enqueue the timer.

If hrtimer::function then returns HRTIMER_RESTART we'll hit the BUG_ON
because HRTIMER_STATE_ENQUEUED will be set.

Since the above is a perfectly valid scenario, remove the BUG_ON and
make the enqueue_hrtimer() call conditional on the timer not being
enqueued already.

NOTE: in that concurrent scenario its entirely common for both sites
to want to modify the hrtimer, since hrtimers don't provide
serialization themselves be sure to provide some such that the
hrtimer::function and the hrtimer_start() caller don't both try and
fudge the expiration state at the same time.

To that effect, add a WARN when someone tries to forward an already
enqueued timer, the most common way to change the expiry of self
restarting timers. Ideally we'd put the WARN in everything modifying
the expiry but most of that is inlines and we don't need the bloat.

Fixes: 2d44ae4d7135 ("hrtimer: clean up cpu->base locking tricks")
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Ben Segall <bsegall@google.com>
Cc: Roman Gushchin <klamm@yandex-team.ru>
Cc: Paul Turner <pjt@google.com>
Link: http://lkml.kernel.org/r/20150415113105.GT5029@twins.programming.kicks-ass.net
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotimer: Put usleep_range into the __sched section
Thomas Gleixner [Tue, 14 Apr 2015 21:09:30 +0000 (21:09 +0000)]
timer: Put usleep_range into the __sched section

do_usleep_range() and schedule_hrtimeout_range() are __sched as
well. So it makes no sense to have the exported function in a
different section.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203503.833709502@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotimer: Remove pointless return value of do_usleep_range()
Thomas Gleixner [Tue, 14 Apr 2015 21:09:28 +0000 (21:09 +0000)]
timer: Remove pointless return value of do_usleep_range()

The only user ignores it anyway and rightfully so.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203503.756060258@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Avoid locking in hrtimer_cancel() if timer not active
Thomas Gleixner [Tue, 14 Apr 2015 21:09:25 +0000 (21:09 +0000)]
hrtimer: Avoid locking in hrtimer_cancel() if timer not active

We can do a lockless check for hrtimer_active before actually taking
the lock in hrtimer[_try_to]_cancel. This is useful for hotpath users
like nanosleep as they avoid the lock dance when the timer has
expired.

This is safe because active is true when the timer is enqueued or the
callback is running. Taking the hrtimer base lock does not protect
against concurrent hrtimer_start calls, the callsite has to do the
proper serialization itself.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203503.580273114@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Remove hrtimer_start() return value
Thomas Gleixner [Tue, 14 Apr 2015 21:09:23 +0000 (21:09 +0000)]
hrtimer: Remove hrtimer_start() return value

No user was ever interested whether the timer was active or not when
it was started. All abusers of the return value are gone, so get rid
of it.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203503.483556394@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotick: broadcast-hrtimer: Remove overly clever return value abuse
Thomas Gleixner [Tue, 14 Apr 2015 21:09:22 +0000 (21:09 +0000)]
tick: broadcast-hrtimer: Remove overly clever return value abuse

The assignment of bc_moved in the conditional construct relies on the
fact that in the case of hrtimer_start() invocation the return value
is always 0. It took me a while to understand it.

We want to get rid of the hrtimer_start() return value. Open code the
logic which makes it readable as well.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203503.404751457@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agoalarmtimer: Get rid of unused return value
Thomas Gleixner [Tue, 14 Apr 2015 21:09:18 +0000 (21:09 +0000)]
alarmtimer: Get rid of unused return value

We want to get rid of the hrtimer_start() return value and the alarm
timer return value is nowhere used. Remove it.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/20150414203503.243910615@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agonet: core: pktgen: Remove bogus hrtimer_active() check
Thomas Gleixner [Tue, 14 Apr 2015 21:09:16 +0000 (21:09 +0000)]
net: core: pktgen: Remove bogus hrtimer_active() check

The check for hrtimer_active() after starting the timer is
pointless. If the timer is inactive it has expired already and
therefor the task pointer is already NULL.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: David S. Miller <davem@davemloft.net>
Link: http://lkml.kernel.org/r/20150414203503.165258315@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agortmutex: Remove bogus hrtimer_active() check
Thomas Gleixner [Tue, 14 Apr 2015 21:09:15 +0000 (21:09 +0000)]
rtmutex: Remove bogus hrtimer_active() check

The check for hrtimer_active() after starting the timer is
pointless. If the timer is inactive it has expired already and
therefor the task pointer is already NULL.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203503.081830481@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agofutex: Remove bogus hrtimer_active() check
Thomas Gleixner [Tue, 14 Apr 2015 21:09:13 +0000 (21:09 +0000)]
futex: Remove bogus hrtimer_active() check

The check for hrtimer_active() after starting the timer is
pointless. If the timer is inactive it has expired already and
therefor the task pointer is already NULL.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203502.985825453@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Remove bogus hrtimer_active() check
Thomas Gleixner [Tue, 14 Apr 2015 21:09:11 +0000 (21:09 +0000)]
hrtimer: Remove bogus hrtimer_active() check

The check for hrtimer_active() after starting the timer is
pointless. If the timer is inactive it has expired already and
therefor the task pointer is already NULL.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203502.907149271@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Make hrtimer_start() a inline wrapper
Thomas Gleixner [Tue, 14 Apr 2015 21:09:10 +0000 (21:09 +0000)]
hrtimer: Make hrtimer_start() a inline wrapper

No point for an extra export just to set the extra argument of
hrtimer_start_range_ns() to 0.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203502.808544539@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Get rid of __hrtimer_start_range_ns()
Thomas Gleixner [Tue, 14 Apr 2015 21:09:08 +0000 (21:09 +0000)]
hrtimer: Get rid of __hrtimer_start_range_ns()

No more callers. Remove the leftovers.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203502.707871492@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agosched: deadline: Use hrtimer_start()
Thomas Gleixner [Tue, 14 Apr 2015 21:09:06 +0000 (21:09 +0000)]
sched: deadline: Use hrtimer_start()

hrtimer_start() does not longer defer already expired timers to the
softirq. Get rid of the __hrtimer_start_range_ns() invocation.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203502.627353666@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agosched: core: Use hrtimer_start[_expires]()
Thomas Gleixner [Tue, 14 Apr 2015 21:09:05 +0000 (21:09 +0000)]
sched: core: Use hrtimer_start[_expires]()

hrtimer_start() now enforces a timer interrupt when an already expired
timer is enqueued.

Get rid of the __hrtimer_start_range_ns() invocations and the loops
around it.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203502.531131739@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agoperf: core: Use hrtimer_start()
Thomas Gleixner [Tue, 14 Apr 2015 21:09:03 +0000 (21:09 +0000)]
perf: core: Use hrtimer_start()

hrtimer_start() does not longer defer already expired timers to the
softirq. Get rid of the __hrtimer_start_range_ns() invocation.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203502.452104213@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agox86: perf: uncore: Use hrtimer_start()
Thomas Gleixner [Tue, 14 Apr 2015 21:09:01 +0000 (21:09 +0000)]
x86: perf: uncore: Use hrtimer_start()

hrtimer_start() does not longer defer already expired timers to the
softirq. Get rid of the __hrtimer_start_range_ns() invocation.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: x86@kernel.org
Link: http://lkml.kernel.org/r/20150414203502.360555157@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agox86: perf: Use hrtimer_start()
Thomas Gleixner [Tue, 14 Apr 2015 21:09:00 +0000 (21:09 +0000)]
x86: perf: Use hrtimer_start()

hrtimer_start() does not longer defer already expired timers to the
softirq. Get rid of the __hrtimer_start_range_ns() invocation.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: x86@kernel.org
Link: http://lkml.kernel.org/r/20150414203502.260487331@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotick: Nohz: Rework next timer evaluation
Thomas Gleixner [Tue, 14 Apr 2015 21:08:58 +0000 (21:08 +0000)]
tick: Nohz: Rework next timer evaluation

The evaluation of the next timer in the nohz code is based on jiffies
while all the tick internals are nano seconds based. We have also to
convert hrtimer nanoseconds to jiffies in the !highres case. That's
just wrong and introduces interesting corner cases.

Turn it around and convert the next timer wheel timer expiry and the
rcu event to clock monotonic and base all calculations on
nanoseconds. That identifies the case where no timer is pending
clearly with an absolute expiry value of KTIME_MAX.

Makes the code more readable and gets rid of the jiffies magic in the
nohz code.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Lai Jiangshan <laijs@cn.fujitsu.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Link: http://lkml.kernel.org/r/20150414203502.184198593@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotick: Sched: Restructure code
Thomas Gleixner [Tue, 14 Apr 2015 21:08:56 +0000 (21:08 +0000)]
tick: Sched: Restructure code

Get rid of one indentation level. Preparatory patch for a major
rework. No functional change.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Link: http://lkml.kernel.org/r/20150414203502.101563235@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotick: sched: Force tick interrupt and get rid of softirq magic
Thomas Gleixner [Tue, 14 Apr 2015 21:08:54 +0000 (21:08 +0000)]
tick: sched: Force tick interrupt and get rid of softirq magic

We already got rid of the hrtimer reprogramming loops and hoops as
hrtimer now enforces an interrupt if the enqueued time is in the past.

Do the same for the nohz non highres mode. That gets rid of the need
to raise the softirq which only serves the purpose of getting the
machine out of the inner idle loop.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Link: http://lkml.kernel.org/r/20150414203502.023464878@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotick: sched: Remove hrtimer_active() checks
Thomas Gleixner [Tue, 14 Apr 2015 21:08:52 +0000 (21:08 +0000)]
tick: sched: Remove hrtimer_active() checks

hrtimer_start() enforces a timer interrupt if the timer is already
expired. Get rid of the checks and the forward loop.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Frederic Weisbecker <fweisbec@gmail.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: John Stultz <john.stultz@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Link: http://lkml.kernel.org/r/20150414203501.943658239@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Get rid of hrtimer softirq
Thomas Gleixner [Tue, 14 Apr 2015 21:08:51 +0000 (21:08 +0000)]
hrtimer: Get rid of hrtimer softirq

hrtimer softirq is a leftover from the initial implementation and
serves only the purpose to handle the enqueueing of already expired
timers in the high resolution timer mode. We discussed whether we
change the return value and force all start sites to handle that the
timer is already expired, but that would be a Herculean task and I'm
not sure whether its a good idea to enforce that handling on
everyone.

A simpler solution is to enforce a timer interrupt instead of raising
and scheduling a softirq. Just use the existing infrastructure to do
so and remove all the softirq leftovers.

The HRTIMER softirq enum is now unused, but kept around because trace
parsers rely on the existing numbering.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203501.840834708@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Keep pointer to first timer and simplify __remove_hrtimer()
Thomas Gleixner [Tue, 14 Apr 2015 21:08:49 +0000 (21:08 +0000)]
hrtimer: Keep pointer to first timer and simplify __remove_hrtimer()

__remove_hrtimer() needs to evaluate the expiry time to figure out
whether the timer which is removed is eventually the first expiring
timer on the cpu. Keep a pointer to it, which is lazily updated, so we
can avoid the evaluation dance and retrieve the information from there.

Generates slightly better code.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203501.752838019@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Make use of timerqueue_add/del return values
Thomas Gleixner [Tue, 14 Apr 2015 21:08:47 +0000 (21:08 +0000)]
hrtimer: Make use of timerqueue_add/del return values

Use the return value instead of reevaluating the information.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203501.658152945@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotimerqueue: Let timerqueue_add/del return information
Thomas Gleixner [Tue, 14 Apr 2015 21:08:46 +0000 (21:08 +0000)]
timerqueue: Let timerqueue_add/del return information

The hrtimer code is interested whether the added timer is the first
one to expire and whether the removed timer was the last one in the
tree. The add/del routines have that information already. So we can
return it right away instead of reevaluating it at the call site.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/20150414203501.579063647@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Align the hrtimer clock bases as well
Thomas Gleixner [Tue, 14 Apr 2015 21:08:44 +0000 (21:08 +0000)]
hrtimer: Align the hrtimer clock bases as well

We don't use cacheline_align here because that might waste lot of
space on 32bit machine with 64 bytes cachelines and on 64bit machines
with 128 bytes cachelines.

The size of struct hrtimer_clock_base is 64byte on 64bit and 32byte on
32bit machines. So we utilize the cache lines proper.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203501.498165771@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Cache line align the hrtimer cpu base
Thomas Gleixner [Tue, 14 Apr 2015 21:08:42 +0000 (21:08 +0000)]
hrtimer: Cache line align the hrtimer cpu base

We really want that data structure to start at a cache line boundary.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203501.417597627@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Use cpu_base->active_base for hotpath iterators
Thomas Gleixner [Tue, 14 Apr 2015 21:08:41 +0000 (21:08 +0000)]
hrtimer: Use cpu_base->active_base for hotpath iterators

The active_bases field is guaranteed to be in sync with the timerqueue
of the corresponding clock base. So we can use it for iterating over
the clock bases. This allows to break out early if no more active
clock bases are available and avoids touching the cache lines of
inactive clock bases.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203501.322887675@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Use bits for various boolean indicators
Thomas Gleixner [Tue, 14 Apr 2015 21:08:39 +0000 (21:08 +0000)]
hrtimer: Use bits for various boolean indicators

No point in wasting 12 byte storage space. Generates better code as well.

Text size reduction:
       x8664 -64, i386 -16, ARM -132, ARM64 -0, power64 -48

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203501.227955358@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Make offset update smarter
Thomas Gleixner [Tue, 14 Apr 2015 21:08:37 +0000 (21:08 +0000)]
hrtimer: Make offset update smarter

On every tick/hrtimer interrupt we update the offset variables of the
clock bases. That's silly because these offsets change very seldom.

Add a sequence counter to the time keeping code which keeps track of
the offset updates (clock_was_set()). Have a sequence cache in the
hrtimer cpu bases to evaluate whether the offsets must be updated or
not. This allows us later to avoid pointless cacheline pollution.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/20150414203501.132820245@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: John Stultz <john.stultz@linaro.org>
9 years agohrtimer: Get rid of softirq time
Thomas Gleixner [Tue, 14 Apr 2015 21:08:35 +0000 (21:08 +0000)]
hrtimer: Get rid of softirq time

The softirq time field in the clock bases is an optimization from the
early days of hrtimers. It provides a coarse "jiffies" like time
mostly for self rearming timers.

But that comes with a price:
    - Larger code size
    - Extra storage space
    - Duplicated functions with really small differences

The benefit of this is optimization is marginal for contemporary
systems.

Consolidate everything on the high resolution timer
implementation. This makes further optimizations possible.

Text size reduction:
       x8664 -95, i386 -356, ARM -148, ARM64 -40, power64 -16

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203501.039977424@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Make the statistics fields smaller
Thomas Gleixner [Tue, 14 Apr 2015 21:08:34 +0000 (21:08 +0000)]
hrtimer: Make the statistics fields smaller

No point in having usigned long for /proc/timer_list statistics. Make
them unsigned int.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203500.959773467@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Get rid of hrtimer_get_res()
Thomas Gleixner [Tue, 14 Apr 2015 21:08:32 +0000 (21:08 +0000)]
hrtimer: Get rid of hrtimer_get_res()

The resolution is directly accessible now. So its simpler just to fill
in the values of the timespec and be done with it.

Text size reduction (combined with "hrtimer: Get rid of the resolution
field in hrtimer_clock_base"):
       x8664 -61, i386 -221, ARM -60, power64 -48

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203500.879888080@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agosound: Use hrtimer_resolution instead of hrtimer_get_res()
Thomas Gleixner [Tue, 14 Apr 2015 21:08:30 +0000 (21:08 +0000)]
sound: Use hrtimer_resolution instead of hrtimer_get_res()

No point in converting a timespec now that the value is directly
accessible. Get rid of the null check while at it. Resolution is
guaranteed to be > 0.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Takashi Iwai <tiwai@suse.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jaroslav Kysela <perex@perex.cz>
Cc: alsa-devel@alsa-project.org
Link: http://lkml.kernel.org/r/20150414203500.799133359@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agonet: sched: Use hrtimer_resolution instead of hrtimer_get_res()
Thomas Gleixner [Tue, 14 Apr 2015 21:08:28 +0000 (21:08 +0000)]
net: sched: Use hrtimer_resolution instead of hrtimer_get_res()

No point in converting a timespec now that the value is directly
accessible.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: David S. Miller <davem@davemloft.net>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jamal Hadi Salim <jhs@mojatatu.com>
Link: http://lkml.kernel.org/r/20150414203500.720623028@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Get rid of the resolution field in hrtimer_clock_base
Thomas Gleixner [Tue, 14 Apr 2015 21:08:27 +0000 (21:08 +0000)]
hrtimer: Get rid of the resolution field in hrtimer_clock_base

The field has no value because all clock bases have the same
resolution. The resolution only changes when we switch to high
resolution timer mode. We can evaluate that from a single static
variable as well. In the !HIGHRES case its simply a constant.

Export the variable, so we can simplify the usage sites.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Viresh Kumar <viresh.kumar@linaro.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/r/20150414203500.645454122@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Update active_bases before calling hrtimer_force_reprogram()
Viresh Kumar [Tue, 14 Apr 2015 21:08:25 +0000 (21:08 +0000)]
hrtimer: Update active_bases before calling hrtimer_force_reprogram()

'active_bases' indicates which clock-base have active timer. The
intention of this bit field was to avoid evaluating inactive bases. It
was introduced with the introduction of the BOOTTIME and TAI clock
bases, but it was never brought into full use.

We want to use it now, but in __remove_hrtimer() the update happens
after the calling hrtimer_force_reprogram() which has to evaluate all
clock bases for the next expiring timer. So in case the last timer of
a clock base got removed we still see the active bit and therefor
evaluate the clock base for no value. There are further optimizations
possible when active_bases is updated in the right place.

Move the update before the call to hrtimer_force_reprogram()

[ tglx: Massaged changelog ]

Signed-off-by: Viresh Kumar <viresh.kumar@linaro.org>
Reviewed-by: Preeti U Murthy <preeti@linux.vnet.ibm.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: linaro-kernel@lists.linaro.org
Link: http://lkml.kernel.org/r/20150414203500.533438642@linutronix.de
Link: http://lkml.kernel.org/r/c7c8ebcd9ed88bb09d76059c745a1fafb48314e7.1428039899.git.viresh.kumar@linaro.org
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agohrtimer: Document hrtimer_forward[_now]() proper
Thomas Gleixner [Mon, 13 Apr 2015 21:02:22 +0000 (21:02 +0000)]
hrtimer: Document hrtimer_forward[_now]() proper

Document the calling context conditions.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20150413210035.178751779@linutronix.de
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotimekeeping: Remove stale function prototype
Thomas Gleixner [Wed, 22 Apr 2015 09:44:15 +0000 (11:44 +0200)]
timekeeping: Remove stale function prototype

commit 61edec81d260 "timekeeping: Simplify timekeeping_clocktai()"
implemented timekeeping_clocktai() as an inline function, but left the
old extern prototype in the header file. Remove it.

Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotime: Remove nonexistent function prototype
Yingjoe Chen [Fri, 10 Apr 2015 13:55:50 +0000 (21:55 +0800)]
time: Remove nonexistent function prototype

The function clocksource_get_next() was removed in commit 75c5158f70
(timekeeping: Update clocksource with stop_machine), but the
prototype was not removed with it. Remove the prototype.

Signed-off-by: Yingjoe Chen <yingjoe.chen@mediatek.com>
Cc: <linux-arm-kernel@lists.infradead.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: <srv_heupstream@mediatek.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/1428674150-1780-1-git-send-email-yingjoe.chen@mediatek.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agotimer_list: Reduce SEQ_printf footprint
Joe Perches [Fri, 17 Apr 2015 18:39:18 +0000 (11:39 -0700)]
timer_list: Reduce SEQ_printf footprint

This macro can be converted to a static function to reduce
object size.

(x86-64 defconfig)
$ size kernel/time/timer_list.o*
   text    data     bss     dec     hex filename
   6583       8       0    6591    19bf kernel/time/timer_list.o.old
   4647       8       0    4655    122f kernel/time/timer_list.o.new

Signed-off-by: Joe Perches <joe@perches.com>
Cc: John Stultz <john.stultz@linaro.org>
Link: http://lkml.kernel.org/r/1429295958.2850.104.camel@perches.com
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
9 years agoMerge tag 'remoteproc-4.1-next' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 20 Apr 2015 22:40:10 +0000 (15:40 -0700)]
Merge tag 'remoteproc-4.1-next' of git://git./linux/kernel/git/ohad/remoteproc

Pull remoteproc update from Ohad Ben-Cohen:
 "Suman Anna is adding remoteproc support for processors not behind
  IOMMUs"

* tag 'remoteproc-4.1-next' of git://git.kernel.org/pub/scm/linux/kernel/git/ohad/remoteproc:
  remoteproc: add IOMMU hardware capability flag

9 years agoMerge branch 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild
Linus Torvalds [Mon, 20 Apr 2015 22:31:49 +0000 (15:31 -0700)]
Merge branch 'misc' of git://git./linux/kernel/git/mmarek/kbuild

Pull misc kbuild updates:
 "This is the remaining part of kbuild stuff for v4.1-rc1:

   - One wew coccinelle script and a clarification of the proposed fix
     in bugon.coccinelle

   - CONFIG_KERNEL_LZ4 support for extract-ikconfig"

* 'misc' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild:
  scripts/coccinelle/misc/bugon.cocci: update bug_on conversion warning
  scripts/extract-ikconfig: Support LZ4-compressed images.
  irqf_oneshot.cocci: add check of devm_request_threaded_irq()

9 years agoMerge tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Mon, 20 Apr 2015 22:25:19 +0000 (15:25 -0700)]
Merge tag 'scsi-misc' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "I'd like to say these were a set of regressions for the recent merge
  window code.  Unfortunately, they all predate the merge window code
  (stable cc'd).

  There are two fixes for data integrity (mostly only showing up on
  module removal), an mvsas crash with expander attached SATA devices
  which goes back to the dawn of the driver but is only just being
  picked up as sas expanders become a standard item in low end server
  hardware, an am53c974 one because the interrupt data isn't fully
  initialised before the line is and a megaraid_sas one because it uses
  smp_processor_id() to select MSI-X queues and that now triggers a
  WARN_ON()"

* tag 'scsi-misc' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  mvsas: fix panic on expander attached SATA devices
  am53c974: Fix crash during modprobe
  megaraid_sas: use raw_smp_processor_id()
  sd: Fix missing ATO tag check
  sd: Unregister integrity profile

9 years agoMerge tag 'fbdev-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux
Linus Torvalds [Mon, 20 Apr 2015 22:16:25 +0000 (15:16 -0700)]
Merge tag 'fbdev-4.1' of git://git./linux/kernel/git/tomba/linux

Pull fbdev updates from Tomi Valkeinen:
 "Small fixes and improvements to various fbdev drivers"

* tag 'fbdev-4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/tomba/linux: (24 commits)
  omapdss: extend pm notifier to handle hibernation
  OMAPDSS: Correct video ports description file path in DT binding doc
  OMAPDSS: disable VT switch
  fbdev: sh_mobile_lcdc: Fix destruction of uninitialized mutex
  video: fbdev: sh_mobile_lcdcfb: Fix ROP3 sysfs attribute parsing
  fbdev: pm3fb: cleanup some confusing indenting
  hyperv: hyperv_fb: match wait_for_completion_timeout return type
  video: fbdev: use msecs_to_jiffies for time conversions
  fbdev: via/via_clock: fix sparse warning
  video: fbdev: make of_device_id array const
  fbdev: sm501fb: use memset_io
  OMAPDSS: workaround for MFLAG + NV12 issue
  OMAPDSS: Add support for MFLAG
  OMAPDSS: setup default fifo thresholds
  OMAPDSS: DISPC: lock access to DISPC_CONTROL & DISPC_CONFIG
  OMAPDSS: DISPC: fix div by zero issue in overlay scaling
  OMAPDSS: DISPC: change sync_pclk_edge default value
  OMAPDSS: change signal_level & signal_edge enum values
  OMAPDSS: DISPC: explicit handling for sync and de levels
  OMAPDSS: DISPC: remove OMAPDSS_DRIVE_SIG_OPPOSITE_EDGES
  ...

9 years agoMerge branch 'drm-next-merged' of git://people.freedesktop.org/~airlied/linux
Linus Torvalds [Mon, 20 Apr 2015 21:06:06 +0000 (14:06 -0700)]
Merge branch 'drm-next-merged' of git://people.freedesktop.org/~airlied/linux

Pull drm updates from Dave Airlie:
 "Highlights:

  Core:
   - Virtual GEM layer merged, this has been around for a long time, and
     it provides a software backed device that allows userspace to use
     it as a GEM shared memory handler.  This makes it a lot easier to
     do certain things when you have no GPU but still have to deal with
     DRI expectations.
   - atomic helper updates.
   - framebuffer modifier interface added.
   - i2c over auxch displayport fixes.
   - fb width/height confusion fixes.
   - new driver for ps8622/ps8625 bridge chips
   - lots of new panels

  i915:
   - more plane atomic conversion
   - vGPU guest support for XenGT
   - Skylake workarounds and fixes
   - Y-tiling support
   - work on dynamic pagetable allocation
   - EU count report param for gen9+
   - CHV fixes (no longer prelim)
   - remove ilk rc6
   - frontbuffer tracking for fbc
   - Displayport link rate refactoring
   - sprite colorkey refactor

  radeon:
   - Displayport MST support (not enabled by default)
   - non-ATOM native hw auxch support (DCE5+)
   - output csc support
   - new queries for userspace debug support
   - new VCE packet

  nouveau:
   - gk20a iommu support
   - gm107 graphics support
   - more gm20x bringup (waiting on signed nvidia fw).

  amdkfd:
   - multiple kgd instance support
   - use 64-bit time accessors

  msm:
   - stolen memory support
   - DSI and dual-DSI support
   - snapdragon 410 support

  exynos:
   - cleanups for atomic and pageflip

  imx-drm:
   - more media-bus formats
   - TV output prep
   - drm panel support

  tegra:
   - hw vblank counter using host1x syncpoints

  omap:
   - universal plane support
   - prep work for atomic modesetting

  rcar-du:
   - ported to atomic modesetting

  atmel-hlcdc:
   - ported to atomic modesetting
   - added suspend/resume support

  sti:
   - ported to atomic modesetting

  dwhdmi:
   - more compliant audio support
   - update rockchip phy support

  tda998x:
   - DT probing for attached crtcs
   - simplified EDID reading

  rockchip:
   - fixes

  adv7511:
   - fixes"

* 'drm-next-merged' of git://people.freedesktop.org/~airlied/linux: (689 commits)
  media-bus: Fixup RGB444_1X12, RGB565_1X16, and YUV8_1X24 media bus format
  drm/i915: Dont enable CS_PARSER_ERROR interrupts at all
  drm/i915: Move drm_framebuffer_unreference out of struct_mutex for takeover
  drm: fix trivial typo mistake
  drm: Make integer overflow checking cover universal cursor updates (v2)
  drm/nouveau/bios: fix fetching from acpi on certain systems
  drm/nouveau/gr/gm206: initial init+ctx code
  drm/nouveau/ce/gm206: enable support via gm204 code
  drm/nouveau/fifo/gm206: enable support via gm204 code
  drm/nouveau/gr/gm204: initial init+ctx code
  drm/nouveau: support for buffer moves via MaxwellDmaCopyA
  drm/nouveau/ce/gm204: initial support
  drm/nouveau: add support for gm20x fifo channels
  drm/nouveau/fifo/gm204: initial support
  drm/nouveau/gr/gk104-: prevent reading non-existent regs in intr handler
  drm/nouveau/gr/gm107: very slightly demagic part of attrib cb setup
  drm/nouveau/gr/gk104-: correct crop/zrop num_active_fbps setting
  drm/nouveau/gr/gf100-: add symbolic names for classes
  drm/nouveau/gr/gm107: support tpc "strand" ctxsw in gpccs ucode
  drm/nouveau/gr/gf100-: support mmio access with gpc offset from gpccs ucode
  ...

9 years agoMerge tag 'iommu-updates-v4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro...
Linus Torvalds [Mon, 20 Apr 2015 17:50:05 +0000 (10:50 -0700)]
Merge tag 'iommu-updates-v4.1' of git://git./linux/kernel/git/joro/iommu

Pull IOMMU updates from Joerg Roedel:
 "Not much this time, but the changes include:

   - moving domain allocation into the iommu drivers to prepare for the
     introduction of default domains for devices

   - fixing the IO page-table code in the AMD IOMMU driver to correctly
     encode large page sizes

   - extension of the PCI support in the ARM-SMMU driver

   - various fixes and cleanups"

* tag 'iommu-updates-v4.1' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (34 commits)
  iommu/amd: Correctly encode huge pages in iommu page tables
  iommu/amd: Optimize amd_iommu_iova_to_phys for new fetch_pte interface
  iommu/amd: Optimize alloc_new_range for new fetch_pte interface
  iommu/amd: Optimize iommu_unmap_page for new fetch_pte interface
  iommu/amd: Return the pte page-size in fetch_pte
  iommu/amd: Add support for contiguous dma allocator
  iommu/amd: Don't allocate with __GFP_ZERO in alloc_coherent
  iommu/amd: Ignore BUS_NOTIFY_UNBOUND_DRIVER event
  iommu/amd: Use BUS_NOTIFY_REMOVED_DEVICE
  iommu/tegra: smmu: Compute PFN mask at runtime
  iommu/tegra: gart: Set aperture at domain initialization time
  iommu/tegra: Setup aperture
  iommu: Remove domain_init and domain_free iommu_ops
  iommu/fsl: Make use of domain_alloc and domain_free
  iommu/rockchip: Make use of domain_alloc and domain_free
  iommu/ipmmu-vmsa: Make use of domain_alloc and domain_free
  iommu/shmobile: Make use of domain_alloc and domain_free
  iommu/msm: Make use of domain_alloc and domain_free
  iommu/tegra-gart: Make use of domain_alloc and domain_free
  iommu/tegra-smmu: Make use of domain_alloc and domain_free
  ...

9 years agoMerge tag 'cpumask-next-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Mon, 20 Apr 2015 17:19:03 +0000 (10:19 -0700)]
Merge tag 'cpumask-next-for-linus' of git://git./linux/kernel/git/rusty/linux

Pull final removal of deprecated cpus_* cpumask functions from Rusty Russell:
 "This is the final removal (after several years!) of the obsolete
  cpus_* functions, prompted by their mis-use in staging.

  With these function removed, all cpu functions should only iterate to
  nr_cpu_ids, so we finally only allocate that many bits when cpumasks
  are allocated offstack"

* tag 'cpumask-next-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rusty/linux: (25 commits)
  cpumask: remove __first_cpu / __next_cpu
  cpumask: resurrect CPU_MASK_CPU0
  linux/cpumask.h: add typechecking to cpumask_test_cpu
  cpumask: only allocate nr_cpumask_bits.
  Fix weird uses of num_online_cpus().
  cpumask: remove deprecated functions.
  mips: fix obsolete cpumask_of_cpu usage.
  x86: fix more deprecated cpu function usage.
  ia64: remove deprecated cpus_ usage.
  powerpc: fix deprecated CPU_MASK_CPU0 usage.
  CPU_MASK_ALL/CPU_MASK_NONE: remove from deprecated region.
  staging/lustre/o2iblnd: Don't use cpus_weight
  staging/lustre/libcfs: replace deprecated cpus_ calls with cpumask_
  staging/lustre/ptlrpc: Do not use deprecated cpus_* functions
  blackfin: fix up obsolete cpu function usage.
  parisc: fix up obsolete cpu function usage.
  tile: fix up obsolete cpu function usage.
  arm64: fix up obsolete cpu function usage.
  mips: fix up obsolete cpu function usage.
  x86: fix up obsolete cpu function usage.
  ...

9 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Mon, 20 Apr 2015 17:15:33 +0000 (10:15 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/s390/linux

Pull more s390 updates from Martin Schwidefsky:
 "The big thing in this second merge for s390 is the new eBPF JIT from
  Michael which replaces the old 32-bit backend.

  The remaining commits are bug fixes"

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/pci: add locking for fmb access
  s390/pci: extract software counters from fmb
  s390/dasd: Fix unresumed device after suspend/resume having no paths
  s390/dasd: fix unresumed device after suspend/resume
  s390/dasd: fix inability to set a DASD device offline
  s390/mm: Fix memory hotplug for unaligned standby memory
  s390/bpf: Add s390x eBPF JIT compiler backend
  s390: Use bool function return values of true/false not 1/0

9 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu
Linus Torvalds [Mon, 20 Apr 2015 17:12:29 +0000 (10:12 -0700)]
Merge branch 'for-next' of git://git./linux/kernel/git/gerg/m68knommu

Pull m68k fixes from Greg Ungerer:
 "Nothing big, spelling fixes and fix/cleanup for ColdFire eth device setup"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu:
  m68knommu: fix fec setup warning for ColdFire 5271 builds
  m68knommu: ColdFire 5271 only has a single FEC controller
  m68k: Fix trivial typos in comments

9 years agosmp: don't use 16-bit words for atomic accesses
Linus Torvalds [Mon, 20 Apr 2015 16:08:49 +0000 (09:08 -0700)]
smp: don't use 16-bit words for atomic accesses

Yes, it should work, but it's a bad idea.  Not only did ARM64 not have
the 16-bit access code (there's a separate patch to add it), it's just
not a good atomic type.  Some architectures fundamentally don't do
atomic accesses in them (alpha), and it's not like it saves any space
here anyway because of structure packing issues.

We normally should aim for flags to be "unsigned int" or "unsigned
long".  And if space is at a premium, use a single byte (although that
causes problems on alpha again).  There might be very special cases
where a 16-byte entity is really wanted, but this is not one of them.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agoMerge omapdss topic branch for fbdev 4.1
Tomi Valkeinen [Mon, 20 Apr 2015 09:09:31 +0000 (12:09 +0300)]
Merge omapdss topic branch for fbdev 4.1

9 years agoomapdss: extend pm notifier to handle hibernation
Grygorii Strashko [Wed, 25 Feb 2015 17:03:56 +0000 (19:03 +0200)]
omapdss: extend pm notifier to handle hibernation

Add handling of missed events in omap_dss_pm_notif which are
needed to support hibernation (suspend to disk).

Signed-off-by: Grygorii Strashko <Grygorii.Strashko@linaro.org>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
9 years agoOMAPDSS: Correct video ports description file path in DT binding doc
Javier Martinez Canillas [Sat, 7 Mar 2015 16:08:03 +0000 (17:08 +0100)]
OMAPDSS: Correct video ports description file path in DT binding doc

The doc refers to Documentation/devicetree/bindings/video/video-ports.txt
which does not exist. The documentation seems to be outdated and wants to
refer to Documentation/devicetree/bindings/graph.txt instead.

Signed-off-by: Javier Martinez Canillas <javier.martinez@collabora.co.uk>
Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
9 years agoOMAPDSS: disable VT switch
Tomi Valkeinen [Wed, 25 Feb 2015 10:08:14 +0000 (12:08 +0200)]
OMAPDSS: disable VT switch

We don't need VT switch when suspending/resuming, so disable it. This
speeds up suspend/resume.

Signed-off-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
Cc: NeilBrown <neil@brown.name>
9 years agoMerge Linus master into drm-next
Dave Airlie [Mon, 20 Apr 2015 01:32:26 +0000 (11:32 +1000)]
Merge Linus master into drm-next

The merge is clean, but the arm build fails afterwards,
due to API changes in the regulator tree.

I've included the patch into the merge to fix the build.

Signed-off-by: Dave Airlie <airlied@redhat.com>
9 years agomedia-bus: Fixup RGB444_1X12, RGB565_1X16, and YUV8_1X24 media bus format
Philipp Zabel [Fri, 17 Apr 2015 17:12:41 +0000 (19:12 +0200)]
media-bus: Fixup RGB444_1X12, RGB565_1X16, and YUV8_1X24 media bus format

Change the constant values for RGB444_1X12, RGB565_1X16, and YUV8_1X24 media
bus formats in anticipation of a merge conflict with the media tree, where
the old values are already taken by RBG888_1X24, RGB888_1X32_PADHI, and
VUY8_1X24, respectively.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Signed-off-by: Dave Airlie <airlied@redhat.com>
9 years agoMerge branch 'turbostat' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux
Linus Torvalds [Sun, 19 Apr 2015 21:31:41 +0000 (14:31 -0700)]
Merge branch 'turbostat' of git://git./linux/kernel/git/lenb/linux

Pull turbostat update from Len Brown:
 "Updates to the turbostat utility.

  Just one kernel dependency in this batch -- added a #define to
  msr-index.h"

* 'turbostat' of git://git.kernel.org/pub/scm/linux/kernel/git/lenb/linux:
  tools/power turbostat: correct dumped pkg-cstate-limit value
  tools/power turbostat: calculate TSC frequency from CPUID(0x15) on SKL
  tools/power turbostat: correct DRAM RAPL units on recent Xeon processors
  tools/power turbostat: Initial Skylake support
  tools/power turbostat: Use $(CURDIR) instead of $(PWD) and add support for O= option in Makefile
  tools/power turbostat: modprobe msr, if needed
  tools/power turbostat: dump MSR_TURBO_RATIO_LIMIT2
  tools/power turbostat: use new MSR_TURBO_RATIO_LIMIT names
  x86 msr-index: define MSR_TURBO_RATIO_LIMIT,1,2
  tools/power turbostat: label base frequency
  tools/power turbostat: update PERF_LIMIT_REASONS decoding
  tools/power turbostat: simplify default output

9 years agoMerge tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso...
Linus Torvalds [Sun, 19 Apr 2015 21:26:31 +0000 (14:26 -0700)]
Merge tag 'ext4_for_linus' of git://git./linux/kernel/git/tytso/ext4

Pull ext4 updates from Ted Ts'o:
 "A few bug fixes and add support for file-system level encryption in
  ext4"

* tag 'ext4_for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4: (31 commits)
  ext4 crypto: enable encryption feature flag
  ext4 crypto: add symlink encryption
  ext4 crypto: enable filename encryption
  ext4 crypto: filename encryption modifications
  ext4 crypto: partial update to namei.c for fname crypto
  ext4 crypto: insert encrypted filenames into a leaf directory block
  ext4 crypto: teach ext4_htree_store_dirent() to store decrypted filenames
  ext4 crypto: filename encryption facilities
  ext4 crypto: implement the ext4 decryption read path
  ext4 crypto: implement the ext4 encryption write path
  ext4 crypto: inherit encryption policies on inode and directory create
  ext4 crypto: enforce context consistency
  ext4 crypto: add encryption key management facilities
  ext4 crypto: add ext4 encryption facilities
  ext4 crypto: add encryption policy and password salt support
  ext4 crypto: add encryption xattr support
  ext4 crypto: export ext4_empty_dir()
  ext4 crypto: add ext4 encryption Kconfig
  ext4 crypto: reserve codepoints used by the ext4 encryption feature
  ext4 crypto: add ext4_mpage_readpages()
  ...

9 years agohexdump: avoid warning in test function
Linus Torvalds [Sun, 19 Apr 2015 20:48:40 +0000 (13:48 -0700)]
hexdump: avoid warning in test function

The test_data_1_le[] array is a const array of const char *.  To avoid
dropping any const information, we need to use "const char * const *",
not just "const char **".

I'm not sure why the different test arrays end up having different
const'ness, but let's make the pointer we use to traverse them as const
as possible, since we modify neither the array of pointers _or_ the
pointers we find in the array.

Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agofs: take i_mutex during prepare_binprm for set[ug]id executables
Jann Horn [Sun, 19 Apr 2015 00:48:39 +0000 (02:48 +0200)]
fs: take i_mutex during prepare_binprm for set[ug]id executables

This prevents a race between chown() and execve(), where chowning a
setuid-user binary to root would momentarily make the binary setuid
root.

This patch was mostly written by Linus Torvalds.

Signed-off-by: Jann Horn <jann@thejh.net>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agosmp: Fix error case handling in smp_call_function_*()
Linus Torvalds [Sun, 19 Apr 2015 08:56:03 +0000 (04:56 -0400)]
smp: Fix error case handling in smp_call_function_*()

Commit 8053871d0f7f ("smp: Fix smp_call_function_single_async()
locking") fixed the locking for the asynchronous smp-call case, but in
the process of moving the lock handling around, one of the error cases
ended up not unlocking the call data at all.

This went unnoticed on x86, because this is a "caller is buggy" case,
where the caller is trying to call a non-existent CPU.  But apparently
ARM does that (at least under qemu-arm).  Bindly doing cross-cpu calls
to random CPU's that aren't even online seems a bit fishy, but the error
handling was clearly not correct.

Simply add the missing "csd_unlock()" to the error path.

Reported-and-tested-by: Guenter Roeck <linux@roeck-us.net>
Analyzed-by: Rabin Vincent <rabin@rab.in>
Acked-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
9 years agocpumask: remove __first_cpu / __next_cpu
Rusty Russell [Sat, 18 Apr 2015 01:48:27 +0000 (11:18 +0930)]
cpumask: remove __first_cpu / __next_cpu

They were for use by the deprecated first_cpu() and next_cpu() wrappers,
but sparc used them directly.

They're now replaced by cpumask_first / cpumask_next.  And __next_cpu_nr
is completely obsolete.

Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: David S. Miller <davem@davemloft.net>
9 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc
Linus Torvalds [Sat, 18 Apr 2015 22:01:29 +0000 (18:01 -0400)]
Merge git://git./linux/kernel/git/davem/sparc

Pull sparc fixes from David Miller
 "Unfortunately, I brown paper bagged the generic iommu pool allocator
  by applying the wrong revision of the patch series.

  This reverts the bad one, and puts the right one in"

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/sparc:
  iommu-common: Fix PARISC compile-time warnings
  sparc: Make LDC use common iommu poll management functions
  sparc: Make sparc64 use scalable lib/iommu-common.c functions
  Break up monolithic iommu table/lock into finer graularity pools and lock
  sparc: Revert generic IOMMU allocator.

9 years agoMerge tag 'for-linus-4.1-merge-window' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sat, 18 Apr 2015 21:45:30 +0000 (17:45 -0400)]
Merge tag 'for-linus-4.1-merge-window' of git://git./linux/kernel/git/ericvh/v9fs

Pull 9pfs updates from Eric Van Hensbergen:
 "Some accumulated cleanup patches for kerneldoc and unused variables as
  well as some lock bug fixes and adding privateport option for RDMA"

* tag 'for-linus-4.1-merge-window' of git://git.kernel.org/pub/scm/linux/kernel/git/ericvh/v9fs:
  net/9p: add a privport option for RDMA transport.
  fs/9p: Initialize status in v9fs_file_do_lock.
  net/9p: Initialize opts->privport as it should be.
  net/9p: use memcpy() instead of snprintf() in p9_mount_tag_show()
  9p: use unsigned integers for nwqid/count
  9p: do not crash on unknown lock status code
  9p: fix error handling in v9fs_file_do_lock
  9p: remove unused variable in p9_fd_create()
  9p: kerneldoc warning fixes

9 years agoMerge branch 'iommu-generic-allocator'
David S. Miller [Sat, 18 Apr 2015 19:35:09 +0000 (12:35 -0700)]
Merge branch 'iommu-generic-allocator'

Sowmini Varadhan says:

====================
Generic IOMMU pooled allocator

Investigation of network performance on Sparc shows a high
degree of locking contention in the IOMMU allocator, and it
was noticed that the PowerPC code has a better locking model.

This patch series tries to extract the generic parts of the
PowerPC code so that it can be shared across multiple PCI
devices and architectures.

v10: resend patchv9 without RFC tag, and a new mail Message-Id,
(previous non-RFC attempt did not show up on the patchwork queue?)

Full revision history below:
v2 changes:
  - incorporate David Miller editorial comments: sparc specific
    fields moved from iommu-common into sparc's iommu_64.h
  - make the npools value an input parameter, for the case when
    the iommu map size is not very large
  - cookie_to_index mapping, and optimizations for span-boundary
    check, for use case such as LDC.

v3: eliminate iommu_sparc, rearrange the ->demap indirection to
    be invoked under the pool lock.

v4: David Miller review changes:
  - s/IOMMU_ERROR_CODE/DMA_ERROR_CODE
  - page_table_map_base and page_table_shift are unsigned long, not u32.

v5: removed ->cookie_to_index and ->demap indirection from the
    iommu_tbl_ops The caller needs to call these functions as needed,
    before invoking the generic arena allocator functions.
    Added the "skip_span_boundary" argument to iommu_tbl_pool_init() for
    those callers like LDC which do no care about span boundary checks.

v6: removed iommu_tbl_ops, and instead pass the ->flush_all as
    an indirection to iommu_tbl_pool_init(); only invoke ->flush_all
    when there is no large_pool, based on the assumption that large-pool
    usage is infrequently encountered

v7: moved pool_hash initialization to lib/iommu-common.c and cleaned up
    code duplication from sun4v/sun4u/ldc.

v8: Addresses BenH comments with one exception: I've left the
    IOMMU_POOL_HASH as is, so that powerpc can tailor it to their
    convenience.  Discard trylock for simple spin_lock to acquire pool

v9: Addresses latest BenH comments: need_flush checks, add support
    for dma mask and align_order.

v10: resend without RFC tag, and new mail Message-Id.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoiommu-common: Fix PARISC compile-time warnings
Sowmini Varadhan [Sat, 18 Apr 2015 19:33:55 +0000 (12:33 -0700)]
iommu-common: Fix PARISC compile-time warnings

Fixes warnings due to
- no DMA_ERROR_CODE on PARISC,
- sizeof (unsigned long) == 4 bytes on PARISC.

Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agosparc: Make LDC use common iommu poll management functions
Sowmini Varadhan [Thu, 9 Apr 2015 19:33:32 +0000 (15:33 -0400)]
sparc: Make LDC use common iommu poll management functions

Note that this conversion is only being done to consolidate the
code and ensure that the common code provides the sufficient
abstraction. It is not expected to result in any noticeable
performance improvement, as there is typically one ldc_iommu
per vnet_port, and each one has 8k entries, with a typical
request for 1-4 pages.  Thus LDC uses npools == 1.

Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agosparc: Make sparc64 use scalable lib/iommu-common.c functions
Sowmini Varadhan [Thu, 9 Apr 2015 19:33:31 +0000 (15:33 -0400)]
sparc: Make sparc64 use scalable lib/iommu-common.c functions

In iperf experiments running linux as the Tx side (TCP client) with
10 threads results in a severe performance drop when TSO is disabled,
indicating a weakness in the software that can be avoided by using
the scalable IOMMU arena DMA allocation.

Baseline numbers before this patch:
   with default settings (TSO enabled) :    9-9.5 Gbps
   Disable TSO using ethtool- drops badly:  2-3 Gbps.

After this patch, iperf client with 10 threads, can give a
throughput of at least 8.5 Gbps, even when TSO is disabled.

Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agoBreak up monolithic iommu table/lock into finer graularity pools and lock
Sowmini Varadhan [Thu, 9 Apr 2015 19:33:30 +0000 (15:33 -0400)]
Break up monolithic iommu table/lock into finer graularity pools and lock

Investigation of multithreaded iperf experiments on an ethernet
interface show the iommu->lock as the hottest lock identified by
lockstat, with something of the order of  21M contentions out of
27M acquisitions, and an average wait time of 26 us for the lock.
This is not efficient. A more scalable design is to follow the ppc
model, where the iommu_map_table has multiple pools, each stretching
over a segment of the map, and with a separate lock for each pool.
This model allows for better parallelization of the iommu map search.

This patch adds the iommu range alloc/free function infrastructure.

Signed-off-by: Sowmini Varadhan <sowmini.varadhan@oracle.com>
Acked-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agosparc: Revert generic IOMMU allocator.
David S. Miller [Sat, 18 Apr 2015 19:31:25 +0000 (12:31 -0700)]
sparc: Revert generic IOMMU allocator.

I applied the wrong version of this patch series, V4 instead
of V10, due to a patchwork bundling snafu.

Signed-off-by: David S. Miller <davem@davemloft.net>
9 years agotools/power turbostat: correct dumped pkg-cstate-limit value
Len Brown [Thu, 2 Apr 2015 01:02:57 +0000 (21:02 -0400)]
tools/power turbostat: correct dumped pkg-cstate-limit value

HSW expanded MSR_PKG_CST_CONFIG_CONTROL.Package-C-State-Limit,
from bits[2:0] used by previous implementations, to [3:0].
The value 1000b is unlimited, and is used by BDW and SKL too.

Signed-off-by: Len Brown <len.brown@intel.com>
9 years agotools/power turbostat: calculate TSC frequency from CPUID(0x15) on SKL
Len Brown [Thu, 2 Apr 2015 01:02:57 +0000 (21:02 -0400)]
tools/power turbostat: calculate TSC frequency from CPUID(0x15) on SKL

turbostat --debug
...
CPUID(0x15): eax_crystal: 2 ebx_tsc: 100 ecx_crystal_hz: 0
TSC: 1200 MHz (24000000 Hz * 100 / 2 / 1000000)

Signed-off-by: Len Brown <len.brown@intel.com>
9 years agotools/power turbostat: correct DRAM RAPL units on recent Xeon processors
Andrey Semin [Fri, 5 Dec 2014 05:07:00 +0000 (00:07 -0500)]
tools/power turbostat: correct DRAM RAPL units on recent Xeon processors

While not yet documented in the Software Developer's Manual,
the data-sheet for modern Xeon states that DRAM RAPL ENERGY units
are fixed at 15.3 uJ, rather than being discovered via MSR.

Before this patch, DRAM energy on these products is over-stated by turbostat
because the RAPL units are 4x larger.

ref: "Xeon E5-2600 v3/E5-1600 v3 Datasheet Volume 2"
http://www.intel.com/content/dam/www/public/us/en/documents/datasheets/xeon-e5-v3-datasheet-vol-2.pdf

Signed-off-by: Andrey Semin <andrey.semin@intel.com>
Signed-off-by: Len Brown <len.brown@intel.com>
9 years agotools/power turbostat: Initial Skylake support
Len Brown [Thu, 26 Mar 2015 04:50:30 +0000 (00:50 -0400)]
tools/power turbostat: Initial Skylake support

Skylake adds some additional residency counters.

Skylake supports a different mix of RAPL registers
from any previous product.

In most other ways, Skylake is like Broadwell.

Signed-off-by: Len Brown <len.brown@intel.com>
9 years agotools/power turbostat: Use $(CURDIR) instead of $(PWD) and add support for O= option...
Thomas D [Mon, 5 Jan 2015 20:37:23 +0000 (21:37 +0100)]
tools/power turbostat: Use $(CURDIR) instead of $(PWD) and add support for O= option in Makefile

Since commit ee0778a30153
("tools/power: turbostat: make Makefile a bit more capable")
turbostat's Makefile is using

  [...]
  BUILD_OUTPUT    := $(PWD)
  [...]

which obviously causes trouble when building "turbostat" with

  make -C /usr/src/linux/tools/power/x86/turbostat ARCH=x86 turbostat

because GNU make does not update nor guarantee that $PWD is set.

This patch changes the Makefile to use $CURDIR instead, which GNU make
guarantees to set and update (i.e. when using "make -C ...") and also
adds support for the O= option (see "make help" in your root of your
kernel source tree for more details).

Link: https://bugs.gentoo.org/show_bug.cgi?id=533918
Fixes: ee0778a30153 ("tools/power: turbostat: make Makefile a bit more capable")
Signed-off-by: Thomas D. <whissi@whissi.de>
Cc: Mark Asselstine <mark.asselstine@windriver.com>
Signed-off-by: Len Brown <len.brown@intel.com>
9 years agotools/power turbostat: modprobe msr, if needed
Len Brown [Tue, 24 Mar 2015 20:37:35 +0000 (16:37 -0400)]
tools/power turbostat: modprobe msr, if needed

Some distros (Ubuntu) ship the msr driver as a module.
If turbosat is run as root on those systems, and discovers
that there is no /dev/cpu/cpu0/msr, it will now "modprobe msr"
for the user.

If not root, the modprobe attempt will fail, and turbostat will exit as before:

turbostat: no /dev/cpu/0/msr, Try "# modprobe msr" : No such file or directory

Signed-off-by: Len Brown <len.brown@intel.com>
9 years agotools/power turbostat: dump MSR_TURBO_RATIO_LIMIT2
Len Brown [Tue, 24 Mar 2015 00:29:09 +0000 (20:29 -0400)]
tools/power turbostat: dump MSR_TURBO_RATIO_LIMIT2

and up to 18 cores of turbo ratio limit
when using the turbostat --debug option.

Signed-off-by: Len Brown <len.brown@intel.com>
9 years agotools/power turbostat: use new MSR_TURBO_RATIO_LIMIT names
Len Brown [Mon, 13 Apr 2015 20:08:18 +0000 (16:08 -0400)]
tools/power turbostat: use new MSR_TURBO_RATIO_LIMIT names

s/MSR_NHM_TURBO_RATIO_LIMIT/MSR_TURBO_RATIO_LIMIT/
s/MSR_IVT_TURBO_RATIO_LIMIT/MSR_TURBO_RATIO_LIMIT1/

syntax only -- use the documented strings describing these registers.

Signed-off-by: Len Brown <len.brown@intel.com>