[PATCH 00/13] Linux 4.14.109-rt58-rc1

[PATCH 00/13] Linux 4.14.109-rt58-rc1

[Tom Zanussi]

From: Tom Zanussi <tom.zanussi@linux.intel.com>

Hello RT Folks!

This is the RT stable review cycle of patch 4.14.109-rt58-rc1.

Please scream at me if I messed something up. Please test the
patches too.

The -rc release will be uploaded to kernel.org and will be
deleted when the final release is out. This is just a review
release (or release candidate).

The pre-releases will not be pushed to the git repository,
only the final release is.

If all goes well, this patch will be converted to the next
main release on 4/12/2019.

To build 4.14.109-rt58-rc1 directly, the following patches should be applied:

  http://www.kernel.org/pub/linux/kernel/v4.x/linux-4.14.tar.xz

  http://www.kernel.org/pub/linux/kernel/v4.x/patch-4.14.109.xz

  http://www.kernel.org/pub/linux/kernel/projects/rt/4.14/patch-4.14.109-rt58-rc1.patch.xz

You can also build from 4.14.109-rt57 by applying the incremental patch:

  http://www.kernel.org/pub/linux/kernel/projects/rt/4.14/incr/patch-4.14.109-rt57-rt58-rc1.patch.xz

Enjoy!
   
   Tom

Changes from v4.14.109-rt57:

Julien Grall (1):
  tty/sysrq: Convert show_lock to raw_spinlock_t

Peter Zijlstra (1):
  sched/fair: Robustify CFS-bandwidth timer locking

Scott Wood (1):
  locking/rt-mutex: Flush block plug on __down_read()

Sebastian Andrzej Siewior (9):
  arm64: fpsimd: use preemp_disable in addition to local_bh_disable()
  sched/fair: Make the hrtimers non-hard again
  rtmutex/rwlock: preserve state like a sleeping lock
  softirq: Avoid "local_softirq_pending" messages if ksoftirqd is
    blocked
  softirq: Avoid "local_softirq_pending" messages if task is in
    cpu_chill()
  hrtimer: Don't lose state in cpu_chill()
  x86: lazy-preempt: properly check against preempt-mask
  hrtimer: cpu_chill(): save task state in ->saved_state()
  powerpc/pseries/iommu: Use a locallock instead local_irq_save()

Tom Zanussi (1):
  Linux 4.14.109-rt58

 arch/arm64/kernel/fpsimd.c             | 10 ++++++
 arch/powerpc/platforms/pseries/iommu.c | 16 +++++----
 arch/x86/include/asm/preempt.h         |  2 +-
 drivers/tty/sysrq.c                    |  6 ++--
 kernel/locking/rtmutex.c               |  2 +-
 kernel/locking/rwlock-rt.c             |  2 +-
 kernel/locking/rwsem-rt.c              |  9 +++++
 kernel/sched/fair.c                    | 34 ++++++++++---------
 kernel/softirq.c                       | 60 +++++++++++++++++++++++++---------
 kernel/time/hrtimer.c                  | 15 +++++++--
 localversion-rt                        |  2 +-
 11 files changed, 111 insertions(+), 47 deletions(-)

-- 
2.14.1

[PATCH 01/13] arm64: fpsimd: use preemp_disable in addition to local_bh_disable()

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[ Upstream commit 3afdfe419c33c1655d0763e070350aad44416594 ]

In v4.16-RT I noticed a number of warnings from task_fpsimd_load(). The
code disables BH and expects that it is not preemptible. On -RT the
task remains preemptible but remains the same CPU. This may corrupt the
content of the SIMD registers if the task is preempted during
saving/restoring those registers.

Add preempt_disable()/enable() to enfore the required semantic on -RT.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
[tom.zanussi@linux.intel.com: ignored sve-related changes]
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>

 Conflicts:
	arch/arm64/kernel/fpsimd.c
---
 arch/arm64/kernel/fpsimd.c | 10 ++++++++++
 1 file changed, 10 insertions(+)

diff --git a/arch/arm64/kernel/fpsimd.c b/arch/arm64/kernel/fpsimd.c
index 5d547deb6996..049641a458f3 100644
--- a/arch/arm64/kernel/fpsimd.c
+++ b/arch/arm64/kernel/fpsimd.c
@@ -172,6 +172,7 @@ void fpsimd_flush_thread(void)
 	if (!system_supports_fpsimd())
 		return;
 
+	preempt_disable();
 	local_bh_disable();
 
 	memset(&current->thread.fpsimd_state, 0, sizeof(struct fpsimd_state));
@@ -179,6 +180,7 @@ void fpsimd_flush_thread(void)
 	set_thread_flag(TIF_FOREIGN_FPSTATE);
 
 	local_bh_enable();
+	preempt_enable();
 }
 
 /*
@@ -190,12 +192,14 @@ void fpsimd_preserve_current_state(void)
 	if (!system_supports_fpsimd())
 		return;
 
+	preempt_disable();
 	local_bh_disable();
 
 	if (!test_thread_flag(TIF_FOREIGN_FPSTATE))
 		fpsimd_save_state(&current->thread.fpsimd_state);
 
 	local_bh_enable();
+	preempt_enable();
 }
 
 /*
@@ -208,6 +212,7 @@ void fpsimd_restore_current_state(void)
 	if (!system_supports_fpsimd())
 		return;
 
+	preempt_disable();
 	local_bh_disable();
 
 	if (test_and_clear_thread_flag(TIF_FOREIGN_FPSTATE)) {
@@ -219,6 +224,7 @@ void fpsimd_restore_current_state(void)
 	}
 
 	local_bh_enable();
+	preempt_enable();
 }
 
 /*
@@ -231,6 +237,7 @@ void fpsimd_update_current_state(struct fpsimd_state *state)
 	if (!system_supports_fpsimd())
 		return;
 
+	preempt_disable();
 	local_bh_disable();
 
 	fpsimd_load_state(state);
@@ -242,6 +249,7 @@ void fpsimd_update_current_state(struct fpsimd_state *state)
 	}
 
 	local_bh_enable();
+	preempt_enable();
 }
 
 /*
@@ -281,6 +289,7 @@ void kernel_neon_begin(void)
 
 	BUG_ON(!may_use_simd());
 
+	preempt_disable();
 	local_bh_disable();
 
 	__this_cpu_write(kernel_neon_busy, true);
@@ -295,6 +304,7 @@ void kernel_neon_begin(void)
 	preempt_disable();
 
 	local_bh_enable();
+	preempt_enable();
 }
 EXPORT_SYMBOL(kernel_neon_begin);
 
-- 
2.14.1

[PATCH 02/13] sched/fair: Robustify CFS-bandwidth timer locking

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Peter Zijlstra <peterz@infradead.org>

[ Upstream commit a44c9da61ad3334a9563102e0194d7197b222ec2 ]

Traditionally hrtimer callbacks were run with IRQs disabled, but with
the introduction of HRTIMER_MODE_SOFT it is possible they run from
SoftIRQ context, which does _NOT_ have IRQs disabled.

Allow for the CFS bandwidth timers (period_timer and slack_timer) to
be ran from SoftIRQ context; this entails removing the assumption that
IRQs are already disabled from the locking.

While mainline doesn't strictly need this, -RT forces all timers not
explicitly marked with MODE_HARD into MODE_SOFT and trips over this.
And marking these timers as MODE_HARD doesn't make sense as they're
not required for RT operation and can potentially be quite expensive.

Cc: Ingo Molnar <mingo@redhat.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Reported-by: Tom Putzeys <tom.putzeys@be.atlascopco.com>
Tested-by: Mike Galbraith <efault@gmx.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20190107125231.GE14122@hirez.programming.kicks-ass.net
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 kernel/sched/fair.c | 30 ++++++++++++++++--------------
 1 file changed, 16 insertions(+), 14 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 9e6ca7d463c0..46403e462404 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -4378,7 +4378,7 @@ static u64 distribute_cfs_runtime(struct cfs_bandwidth *cfs_b,
 		struct rq *rq = rq_of(cfs_rq);
 		struct rq_flags rf;
 
-		rq_lock(rq, &rf);
+		rq_lock_irqsave(rq, &rf);
 		if (!cfs_rq_throttled(cfs_rq))
 			goto next;
 
@@ -4395,7 +4395,7 @@ static u64 distribute_cfs_runtime(struct cfs_bandwidth *cfs_b,
 			unthrottle_cfs_rq(cfs_rq);
 
 next:
-		rq_unlock(rq, &rf);
+		rq_unlock_irqrestore(rq, &rf);
 
 		if (!remaining)
 			break;
@@ -4411,7 +4411,7 @@ static u64 distribute_cfs_runtime(struct cfs_bandwidth *cfs_b,
  * period the timer is deactivated until scheduling resumes; cfs_b->idle is
  * used to track this state.
  */
-static int do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b, int overrun)
+static int do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b, int overrun, unsigned long flags)
 {
 	u64 runtime, runtime_expires;
 	int throttled;
@@ -4453,11 +4453,11 @@ static int do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b, int overrun)
 	while (throttled && cfs_b->runtime > 0 && !cfs_b->distribute_running) {
 		runtime = cfs_b->runtime;
 		cfs_b->distribute_running = 1;
-		raw_spin_unlock(&cfs_b->lock);
+		raw_spin_unlock_irqrestore(&cfs_b->lock, flags);
 		/* we can't nest cfs_b->lock while distributing bandwidth */
 		runtime = distribute_cfs_runtime(cfs_b, runtime,
 						 runtime_expires);
-		raw_spin_lock(&cfs_b->lock);
+		raw_spin_lock_irqsave(&cfs_b->lock, flags);
 
 		cfs_b->distribute_running = 0;
 		throttled = !list_empty(&cfs_b->throttled_cfs_rq);
@@ -4566,17 +4566,18 @@ static __always_inline void return_cfs_rq_runtime(struct cfs_rq *cfs_rq)
 static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b)
 {
 	u64 runtime = 0, slice = sched_cfs_bandwidth_slice();
+	unsigned long flags;
 	u64 expires;
 
 	/* confirm we're still not at a refresh boundary */
-	raw_spin_lock(&cfs_b->lock);
+	raw_spin_lock_irqsave(&cfs_b->lock, flags);
 	if (cfs_b->distribute_running) {
-		raw_spin_unlock(&cfs_b->lock);
+		raw_spin_unlock_irqrestore(&cfs_b->lock, flags);
 		return;
 	}
 
 	if (runtime_refresh_within(cfs_b, min_bandwidth_expiration)) {
-		raw_spin_unlock(&cfs_b->lock);
+		raw_spin_unlock_irqrestore(&cfs_b->lock, flags);
 		return;
 	}
 
@@ -4587,18 +4588,18 @@ static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b)
 	if (runtime)
 		cfs_b->distribute_running = 1;
 
-	raw_spin_unlock(&cfs_b->lock);
+	raw_spin_unlock_irqrestore(&cfs_b->lock, flags);
 
 	if (!runtime)
 		return;
 
 	runtime = distribute_cfs_runtime(cfs_b, runtime, expires);
 
-	raw_spin_lock(&cfs_b->lock);
+	raw_spin_lock_irqsave(&cfs_b->lock, flags);
 	if (expires == cfs_b->runtime_expires)
 		cfs_b->runtime -= min(runtime, cfs_b->runtime);
 	cfs_b->distribute_running = 0;
-	raw_spin_unlock(&cfs_b->lock);
+	raw_spin_unlock_irqrestore(&cfs_b->lock, flags);
 }
 
 /*
@@ -4676,20 +4677,21 @@ static enum hrtimer_restart sched_cfs_period_timer(struct hrtimer *timer)
 {
 	struct cfs_bandwidth *cfs_b =
 		container_of(timer, struct cfs_bandwidth, period_timer);
+	unsigned long flags;
 	int overrun;
 	int idle = 0;
 
-	raw_spin_lock(&cfs_b->lock);
+	raw_spin_lock_irqsave(&cfs_b->lock, flags);
 	for (;;) {
 		overrun = hrtimer_forward_now(timer, cfs_b->period);
 		if (!overrun)
 			break;
 
-		idle = do_sched_cfs_period_timer(cfs_b, overrun);
+		idle = do_sched_cfs_period_timer(cfs_b, overrun, flags);
 	}
 	if (idle)
 		cfs_b->period_active = 0;
-	raw_spin_unlock(&cfs_b->lock);
+	raw_spin_unlock_irqrestore(&cfs_b->lock, flags);
 
 	return idle ? HRTIMER_NORESTART : HRTIMER_RESTART;
 }
-- 
2.14.1

[PATCH 03/13] sched/fair: Make the hrtimers non-hard again

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[ Upstream commit 73d5c2697339118cd33dc1ff1ffcb21d59cafe1a ]

Since commit "sched/fair: Robustify CFS-bandwidth timer locking" both
hrtimer can run in softirq context because now interrupts are disabled
as part of the locking procedure.

Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 kernel/sched/fair.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 46403e462404..7d2a8fa9a6ed 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -4704,9 +4704,9 @@ void init_cfs_bandwidth(struct cfs_bandwidth *cfs_b)
 	cfs_b->period = ns_to_ktime(default_cfs_period());
 
 	INIT_LIST_HEAD(&cfs_b->throttled_cfs_rq);
-	hrtimer_init(&cfs_b->period_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED_HARD);
+	hrtimer_init(&cfs_b->period_timer, CLOCK_MONOTONIC, HRTIMER_MODE_ABS_PINNED);
 	cfs_b->period_timer.function = sched_cfs_period_timer;
-	hrtimer_init(&cfs_b->slack_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL_HARD);
+	hrtimer_init(&cfs_b->slack_timer, CLOCK_MONOTONIC, HRTIMER_MODE_REL);
 	cfs_b->slack_timer.function = sched_cfs_slack_timer;
 	cfs_b->distribute_running = 0;
 }
-- 
2.14.1

[PATCH 04/13] locking/rt-mutex: Flush block plug on __down_read()

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Scott Wood <swood@redhat.com>

[ Upstream commit 0d5d32425a0cab4a282dd79fc953d76a3ab17072 ]

__down_read() bypasses the rtmutex frontend to call
rt_mutex_slowlock_locked() directly, and thus it needs to call
blk_schedule_flush_flug() itself.

Cc: stable-rt@vger.kernel.org
Signed-off-by: Scott Wood <swood@redhat.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>

 Conflicts:
	kernel/locking/rwsem-rt.c
---
 kernel/locking/rtmutex.c  | 2 +-
 kernel/locking/rwsem-rt.c | 9 +++++++++
 2 files changed, 10 insertions(+), 1 deletion(-)

diff --git a/kernel/locking/rtmutex.c b/kernel/locking/rtmutex.c
index 2cf515877b8e..1177f2815040 100644
--- a/kernel/locking/rtmutex.c
+++ b/kernel/locking/rtmutex.c
@@ -1946,7 +1946,7 @@ rt_mutex_fastlock(struct rt_mutex *lock, int state,
 	 * If rt_mutex blocks, the function sched_submit_work will not call
 	 * blk_schedule_flush_plug (because tsk_is_pi_blocked would be true).
 	 * We must call blk_schedule_flush_plug here, if we don't call it,
-	 * a deadlock in device mapper may happen.
+	 * a deadlock in I/O may happen.
 	 */
 	if (unlikely(blk_needs_flush_plug(current)))
 		blk_schedule_flush_plug(current);
diff --git a/kernel/locking/rwsem-rt.c b/kernel/locking/rwsem-rt.c
index 26991ddb6c5a..dbdde2476b1c 100644
--- a/kernel/locking/rwsem-rt.c
+++ b/kernel/locking/rwsem-rt.c
@@ -1,5 +1,6 @@
 /*
  */
+#include <linux/blkdev.h>
 #include <linux/rwsem.h>
 #include <linux/sched/debug.h>
 #include <linux/sched/signal.h>
@@ -86,6 +87,14 @@ void __sched __down_read(struct rw_semaphore *sem)
 
 	if (__down_read_trylock(sem))
 		return;
+	/*
+	 * If rt_mutex blocks, the function sched_submit_work will not call
+	 * blk_schedule_flush_plug (because tsk_is_pi_blocked would be true).
+	 * We must call blk_schedule_flush_plug here, if we don't call it,
+	 * a deadlock in I/O may happen.
+	 */
+	if (unlikely(blk_needs_flush_plug(current)))
+		blk_schedule_flush_plug(current);
 
 	might_sleep();
 	raw_spin_lock_irq(&m->wait_lock);
-- 
2.14.1

[PATCH 05/13] rtmutex/rwlock: preserve state like a sleeping lock

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[ Upstream commit 4a1bf9e54a013d55ab156cb462c68054cb52ee2d ]

The rwlock is spinning while acquiring a lock. Therefore it must become
a sleeping lock on RT and preserve its task state while sleeping and
waiting for the lock to become available.

Reported-by: Joe Korty <Joe.Korty@concurrent-rt.com>
Cc: stable-rt@vger.kernel.org
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 kernel/locking/rwlock-rt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/locking/rwlock-rt.c b/kernel/locking/rwlock-rt.c
index f2e155b2c4a8..c3b91205161c 100644
--- a/kernel/locking/rwlock-rt.c
+++ b/kernel/locking/rwlock-rt.c
@@ -128,7 +128,7 @@ void __sched __read_rt_lock(struct rt_rw_lock *lock)
 	 * That would put Reader1 behind the writer waiting on
 	 * Reader2 to call read_unlock() which might be unbound.
 	 */
-	rt_mutex_init_waiter(&waiter, false);
+	rt_mutex_init_waiter(&waiter, true);
 	rt_spin_lock_slowlock_locked(m, &waiter, flags);
 	/*
 	 * The slowlock() above is guaranteed to return with the rtmutex is
-- 
2.14.1

[PATCH 06/13] softirq: Avoid "local_softirq_pending" messages if ksoftirqd is blocked

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[ Upstream commit 2cf32c1a3d9352df8017dbf84a1462c4a60a1826 ]

If the ksoftirqd thread has a softirq pending and is blocked on the
`local_softirq_locks' lock then softirq_check_pending_idle() won't
complain because the "lock owner" will mask away this softirq from the
mask of pending softirqs.
If ksoftirqd has an additional softirq pending then it won't be masked
out because we never look at ksoftirqd's mask.

If there are still pending softirqs while going to idle check
ksoftirqd's and ktimersfotd's mask before complaining about unhandled
softirqs.

Cc: stable-rt@vger.kernel.org
Tested-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 kernel/softirq.c | 57 ++++++++++++++++++++++++++++++++++++++++----------------
 1 file changed, 41 insertions(+), 16 deletions(-)

diff --git a/kernel/softirq.c b/kernel/softirq.c
index 583c9ecf04e3..be752eadfc68 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -92,6 +92,31 @@ static inline void softirq_clr_runner(unsigned int sirq)
 	sr->runner[sirq] = NULL;
 }
 
+static bool softirq_check_runner_tsk(struct task_struct *tsk,
+				     unsigned int *pending)
+{
+	bool ret = false;
+
+	if (!tsk)
+		return ret;
+
+	/*
+	 * The wakeup code in rtmutex.c wakes up the task
+	 * _before_ it sets pi_blocked_on to NULL under
+	 * tsk->pi_lock. So we need to check for both: state
+	 * and pi_blocked_on.
+	 */
+	raw_spin_lock(&tsk->pi_lock);
+	if (tsk->pi_blocked_on || tsk->state == TASK_RUNNING) {
+		/* Clear all bits pending in that task */
+		*pending &= ~(tsk->softirqs_raised);
+		ret = true;
+	}
+	raw_spin_unlock(&tsk->pi_lock);
+
+	return ret;
+}
+
 /*
  * On preempt-rt a softirq running context might be blocked on a
  * lock. There might be no other runnable task on this CPU because the
@@ -104,6 +129,7 @@ static inline void softirq_clr_runner(unsigned int sirq)
  */
 void softirq_check_pending_idle(void)
 {
+	struct task_struct *tsk;
 	static int rate_limit;
 	struct softirq_runner *sr = this_cpu_ptr(&softirq_runners);
 	u32 warnpending;
@@ -113,24 +139,23 @@ void softirq_check_pending_idle(void)
 		return;
 
 	warnpending = local_softirq_pending() & SOFTIRQ_STOP_IDLE_MASK;
+	if (!warnpending)
+		return;
 	for (i = 0; i < NR_SOFTIRQS; i++) {
-		struct task_struct *tsk = sr->runner[i];
+		tsk = sr->runner[i];
 
-		/*
-		 * The wakeup code in rtmutex.c wakes up the task
-		 * _before_ it sets pi_blocked_on to NULL under
-		 * tsk->pi_lock. So we need to check for both: state
-		 * and pi_blocked_on.
-		 */
-		if (tsk) {
-			raw_spin_lock(&tsk->pi_lock);
-			if (tsk->pi_blocked_on || tsk->state == TASK_RUNNING) {
-				/* Clear all bits pending in that task */
-				warnpending &= ~(tsk->softirqs_raised);
-				warnpending &= ~(1 << i);
-			}
-			raw_spin_unlock(&tsk->pi_lock);
-		}
+		if (softirq_check_runner_tsk(tsk, &warnpending))
+			warnpending &= ~(1 << i);
+	}
+
+	if (warnpending) {
+		tsk = __this_cpu_read(ksoftirqd);
+		softirq_check_runner_tsk(tsk, &warnpending);
+	}
+
+	if (warnpending) {
+		tsk = __this_cpu_read(ktimer_softirqd);
+		softirq_check_runner_tsk(tsk, &warnpending);
 	}
 
 	if (warnpending) {
-- 
2.14.1

[PATCH 07/13] softirq: Avoid "local_softirq_pending" messages if task is in cpu_chill()

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[ Upstream commit a3429e0147dfd3fc585f989aa49a76579c2b6565 ]

If the softirq thread enters cpu_chill() then ->state is UNINTERRUPTIBLE
and has no ->pi_blocked_on set and so its mask is not taken into account.

->sleeping_lock is increased by cpu_chill() since it is also requried to
avoid a splat by RCU in case cpu_chill() is used while a RCU-read lock
is held. Use the same mechanism for the softirq-pending check.

Cc: stable-rt@vger.kernel.org
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 kernel/softirq.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/kernel/softirq.c b/kernel/softirq.c
index be752eadfc68..e42ee95fccda 100644
--- a/kernel/softirq.c
+++ b/kernel/softirq.c
@@ -105,9 +105,12 @@ static bool softirq_check_runner_tsk(struct task_struct *tsk,
 	 * _before_ it sets pi_blocked_on to NULL under
 	 * tsk->pi_lock. So we need to check for both: state
 	 * and pi_blocked_on.
+	 * The test against UNINTERRUPTIBLE + ->sleeping_lock is in case the
+	 * task does cpu_chill().
 	 */
 	raw_spin_lock(&tsk->pi_lock);
-	if (tsk->pi_blocked_on || tsk->state == TASK_RUNNING) {
+	if (tsk->pi_blocked_on || tsk->state == TASK_RUNNING ||
+	    (tsk->state == TASK_UNINTERRUPTIBLE && tsk->sleeping_lock)) {
 		/* Clear all bits pending in that task */
 		*pending &= ~(tsk->softirqs_raised);
 		ret = true;
-- 
2.14.1

[PATCH 08/13] hrtimer: Don't lose state in cpu_chill()

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[ Upstream commit 585456236766349f692f9fbcc089d71411831f88 ]

In cpu_chill() the state is set to TASK_UNINTERRUPTIBLE and a timer is
programmed. On return the state is always TASK_RUNNING which means we
lose the state if it was something other than RUNNING. Also
set_current_state() sets ->task_state_change to within cpu_chill() which
is not expected.

Save the task state on entry and restore it on return. Simply set the
state in order to avoid updating ->task_state_change.

Cc: stable-rt@vger.kernel.org
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 kernel/time/hrtimer.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index c8d806126381..6da6b5dded3c 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -1866,15 +1866,18 @@ void cpu_chill(void)
 {
 	ktime_t chill_time;
 	unsigned int freeze_flag = current->flags & PF_NOFREEZE;
+	long saved_state;
 
+	saved_state = current->state;
 	chill_time = ktime_set(0, NSEC_PER_MSEC);
-	set_current_state(TASK_UNINTERRUPTIBLE);
+	__set_current_state_no_track(TASK_UNINTERRUPTIBLE);
 	current->flags |= PF_NOFREEZE;
 	sleeping_lock_inc();
 	schedule_hrtimeout(&chill_time, HRTIMER_MODE_REL_HARD);
 	sleeping_lock_dec();
 	if (!freeze_flag)
 		current->flags &= ~PF_NOFREEZE;
+	__set_current_state_no_track(saved_state);
 }
 EXPORT_SYMBOL(cpu_chill);
 #endif
-- 
2.14.1

[PATCH 09/13] x86: lazy-preempt: properly check against preempt-mask

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[ Upstream commit 0f8a27bb5f45cc8a66f488549cecc4ce4d4533c2 ]

should_resched() should check against preempt_offset after unmasking the
need-resched-bit. Otherwise should_resched() won't work for
preempt_offset != 0 and lazy-preempt set.

Cc: stable-rt@vger.kernel.org
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 arch/x86/include/asm/preempt.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/x86/include/asm/preempt.h b/arch/x86/include/asm/preempt.h
index 22992c837795..f66708779274 100644
--- a/arch/x86/include/asm/preempt.h
+++ b/arch/x86/include/asm/preempt.h
@@ -118,7 +118,7 @@ static __always_inline bool should_resched(int preempt_offset)
 
 	/* preempt count == 0 ? */
 	tmp &= ~PREEMPT_NEED_RESCHED;
-	if (tmp)
+	if (tmp != preempt_offset)
 		return false;
 	if (current_thread_info()->preempt_lazy_count)
 		return false;
-- 
2.14.1

[PATCH 10/13] hrtimer: cpu_chill(): save task state in ->saved_state()

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[ Upstream commit d70f218ee02d6fc3832274653aa6c6e98dd9aa43 ]

In the previous change I saved the current task state on stack. This was
bad because while the task is scheduled-out it might receive a wake-up.
The wake up changes the task state and we must not destroy it.

Save the task-state in ->saved_state under a PI-lock to unsure that
state changes during are not missed while the task temporary scheduled
out.

Reported-by: Mike Galbraith <efault@gmx.de>
Tested-by: Mike Galbraith <efault@gmx.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 kernel/time/hrtimer.c | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)

diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
index 6da6b5dded3c..9990b567b089 100644
--- a/kernel/time/hrtimer.c
+++ b/kernel/time/hrtimer.c
@@ -1864,20 +1864,28 @@ COMPAT_SYSCALL_DEFINE2(nanosleep, struct compat_timespec __user *, rqtp,
  */
 void cpu_chill(void)
 {
-	ktime_t chill_time;
 	unsigned int freeze_flag = current->flags & PF_NOFREEZE;
-	long saved_state;
+	struct task_struct *self = current;
+	ktime_t chill_time;
 
-	saved_state = current->state;
-	chill_time = ktime_set(0, NSEC_PER_MSEC);
+	raw_spin_lock_irq(&self->pi_lock);
+	self->saved_state = self->state;
 	__set_current_state_no_track(TASK_UNINTERRUPTIBLE);
+	raw_spin_unlock_irq(&self->pi_lock);
+
+	chill_time = ktime_set(0, NSEC_PER_MSEC);
+
 	current->flags |= PF_NOFREEZE;
 	sleeping_lock_inc();
 	schedule_hrtimeout(&chill_time, HRTIMER_MODE_REL_HARD);
 	sleeping_lock_dec();
 	if (!freeze_flag)
 		current->flags &= ~PF_NOFREEZE;
-	__set_current_state_no_track(saved_state);
+
+	raw_spin_lock_irq(&self->pi_lock);
+	__set_current_state_no_track(self->saved_state);
+	self->saved_state = TASK_RUNNING;
+	raw_spin_unlock_irq(&self->pi_lock);
 }
 EXPORT_SYMBOL(cpu_chill);
 #endif
-- 
2.14.1

[PATCH 11/13] tty/sysrq: Convert show_lock to raw_spinlock_t

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Julien Grall <julien.grall@arm.com>

[ Upstream commit db80c207bffd0f49e984e9889ce62279bc3abd6c ]

Systems which don't provide arch_trigger_cpumask_backtrace() will
invoke showacpu() from a smp_call_function() function which is invoked
with disabled interrupts even on -RT systems.

The function acquires the show_lock lock which only purpose is to
ensure that the CPUs don't print simultaneously. Otherwise the
output would clash and it would be hard to tell the output from CPUx
apart from CPUy.

On -RT the spin_lock() can not be acquired from this context. A
raw_spin_lock() is required. It will introduce the system's latency
by performing the sysrq request and other CPUs will block on the lock
until the request is done. This is okay because the user asked for a
backtrace of all active CPUs and under "normal circumstances in
production" this path should not be triggered.

Signed-off-by: Julien Grall <julien.grall@arm.com>
[bigeasy@linuxtronix.de: commit description]
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Acked-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: stable-rt@vger.kernel.org
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 drivers/tty/sysrq.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/tty/sysrq.c b/drivers/tty/sysrq.c
index 377b3592384e..5d1c6d496e2e 100644
--- a/drivers/tty/sysrq.c
+++ b/drivers/tty/sysrq.c
@@ -215,7 +215,7 @@ static struct sysrq_key_op sysrq_showlocks_op = {
 #endif
 
 #ifdef CONFIG_SMP
-static DEFINE_SPINLOCK(show_lock);
+static DEFINE_RAW_SPINLOCK(show_lock);
 
 static void showacpu(void *dummy)
 {
@@ -225,10 +225,10 @@ static void showacpu(void *dummy)
 	if (idle_cpu(smp_processor_id()))
 		return;
 
-	spin_lock_irqsave(&show_lock, flags);
+	raw_spin_lock_irqsave(&show_lock, flags);
 	pr_info("CPU%d:\n", smp_processor_id());
 	show_stack(NULL, NULL);
-	spin_unlock_irqrestore(&show_lock, flags);
+	raw_spin_unlock_irqrestore(&show_lock, flags);
 }
 
 static void sysrq_showregs_othercpus(struct work_struct *dummy)
-- 
2.14.1

[PATCH 12/13] powerpc/pseries/iommu: Use a locallock instead local_irq_save()

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Sebastian Andrzej Siewior <bigeasy@linutronix.de>

[ Upstream commit 5c1b4cd70e2ca0c81038b65babe6dc66086322e0 ]

The locallock protects the per-CPU variable tce_page. The function
attempts to allocate memory while tce_page is protected (by disabling
interrupts).

Use local_irq_save() instead of local_irq_disable().

Cc: stable-rt@vger.kernel.org
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Tom Zanussi <tom.zanussi@linux.intel.com>
---
 arch/powerpc/platforms/pseries/iommu.c | 16 ++++++++++------
 1 file changed, 10 insertions(+), 6 deletions(-)

diff --git a/arch/powerpc/platforms/pseries/iommu.c b/arch/powerpc/platforms/pseries/iommu.c
index 7c181467d0ad..4bd046492c60 100644
--- a/arch/powerpc/platforms/pseries/iommu.c
+++ b/arch/powerpc/platforms/pseries/iommu.c
@@ -38,6 +38,7 @@
 #include <linux/of.h>
 #include <linux/iommu.h>
 #include <linux/rculist.h>
+#include <linux/locallock.h>
 #include <asm/io.h>
 #include <asm/prom.h>
 #include <asm/rtas.h>
@@ -213,6 +214,7 @@ static int tce_build_pSeriesLP(struct iommu_table *tbl, long tcenum,
 }
 
 static DEFINE_PER_CPU(__be64 *, tce_page);
+static DEFINE_LOCAL_IRQ_LOCK(tcp_page_lock);
 
 static int tce_buildmulti_pSeriesLP(struct iommu_table *tbl, long tcenum,
 				     long npages, unsigned long uaddr,
@@ -233,7 +235,8 @@ static int tce_buildmulti_pSeriesLP(struct iommu_table *tbl, long tcenum,
 		                           direction, attrs);
 	}
 
-	local_irq_save(flags);	/* to protect tcep and the page behind it */
+	/* to protect tcep and the page behind it */
+	local_lock_irqsave(tcp_page_lock, flags);
 
 	tcep = __this_cpu_read(tce_page);
 
@@ -244,7 +247,7 @@ static int tce_buildmulti_pSeriesLP(struct iommu_table *tbl, long tcenum,
 		tcep = (__be64 *)__get_free_page(GFP_ATOMIC);
 		/* If allocation fails, fall back to the loop implementation */
 		if (!tcep) {
-			local_irq_restore(flags);
+			local_unlock_irqrestore(tcp_page_lock, flags);
 			return tce_build_pSeriesLP(tbl, tcenum, npages, uaddr,
 					    direction, attrs);
 		}
@@ -278,7 +281,7 @@ static int tce_buildmulti_pSeriesLP(struct iommu_table *tbl, long tcenum,
 		tcenum += limit;
 	} while (npages > 0 && !rc);
 
-	local_irq_restore(flags);
+	local_unlock_irqrestore(tcp_page_lock, flags);
 
 	if (unlikely(rc == H_NOT_ENOUGH_RESOURCES)) {
 		ret = (int)rc;
@@ -436,13 +439,14 @@ static int tce_setrange_multi_pSeriesLP(unsigned long start_pfn,
 	u64 rc = 0;
 	long l, limit;
 
-	local_irq_disable();	/* to protect tcep and the page behind it */
+	/* to protect tcep and the page behind it */
+	local_lock_irq(tcp_page_lock);
 	tcep = __this_cpu_read(tce_page);
 
 	if (!tcep) {
 		tcep = (__be64 *)__get_free_page(GFP_ATOMIC);
 		if (!tcep) {
-			local_irq_enable();
+			local_unlock_irq(tcp_page_lock);
 			return -ENOMEM;
 		}
 		__this_cpu_write(tce_page, tcep);
@@ -488,7 +492,7 @@ static int tce_setrange_multi_pSeriesLP(unsigned long start_pfn,
 
 	/* error cleanup: caller will clear whole range */
 
-	local_irq_enable();
+	local_unlock_irq(tcp_page_lock);
 	return rc;
 }
 
-- 
2.14.1

[PATCH 13/13] Linux 4.14.109-rt58

[Tom Zanussi]

4.14.109-rt58-rc1 stable review patch.
If anyone has any objections, please let me know.

------------------

From: Tom Zanussi <tom.zanussi@linux.intel.com>

---
 localversion-rt | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/localversion-rt b/localversion-rt
index c06cc4356292..f9df2cf089cf 100644
--- a/localversion-rt
+++ b/localversion-rt
@@ -1 +1 @@
--rt57
+-rt58
-- 
2.14.1