Backport the fix for deadlock in CFS-bandwidth timer locking

Low-latency profile of StarlingX is affected by a deadlock in
CFS scheduler. spin_lock is used in IRQ handler there instead of
spin_lock_irqsave. This leads to an attempt to lock the same
spinlock twice and inevitable system freeze. Backporting c0ad4aa4d8
commit from upstream kernel to cure the issue.

Change-Id: I5416c0e0886f42d2bcec8e3e5da063e6af6916f8
Closes-bug: 1832854
Signed-off-by: Alex Kozyrev <alex.kozyrev@windriver.com>

kernel/kernel-rt/centos/build_srpm.data

@@ -1,4 +1,4 @@
 COPY_LIST="files/*"
-TIS_PATCH_VER=2
+TIS_PATCH_VER=3
 BUILD_IS_BIG=11
 BUILD_IS_SLOW=12

kernel/kernel-rt/centos/meta_patches/Compile-issues.patch

@@ -1,9 +1,9 @@
-From 8fbd0edbcf6ed51b9e58d267d0ce7b40f00118a1 Mon Sep 17 00:00:00 2001
+From 6fe892d415b3d728d223069eacb6f291fc38d86d Mon Sep 17 00:00:00 2001
-From: Bart Wensley <barton.wensley@windriver.com>
+From: Alex Kozyrev <alex.kozyrev@windriver.com>
-Date: Tue, 9 Jul 2019 07:18:00 -0500
+Date: Mon, 29 Jul 2019 11:48:51 -0400
 Subject: [PATCH 1/1] Compile issues
 
-Signed-off-by: Bart Wensley <barton.wensley@windriver.com>
+Signed-off-by: Alex Kozyrev <alex.kozyrev@windriver.com>
 ---
  SPECS/kernel-rt.spec | 8 ++++++++
  1 file changed, 8 insertions(+)
@@ -12,22 +12,22 @@ diff --git a/SPECS/kernel-rt.spec b/SPECS/kernel-rt.spec
 index 3b7985c..5025db7 100644
 --- a/SPECS/kernel-rt.spec
 +++ b/SPECS/kernel-rt.spec
-@@ -418,6 +418,11 @@ Patch1027: dpt_i2o-fix-build-warning.patch
+@@ -418,6 +418,11 @@ # DRBD was choking on write same
- # DRBD was choking on write same
  Patch1028: turn-off-write-same-in-smartqpi-driver.patch
  Patch1029: restrict-iSCSI-kthreads-to-CPUs-in-cpu_kthread_mask.patch
-+Patch1030: fix-compilation-issues.patch
+ Patch1030: robustify-CFS-bandwidth-timer-locking.patch
++Patch1031: fix-compilation-issues.patch
 +# Fix CentOS 7.6 upgrade compile error
-+Patch1031: fix-CentOS-7.6-upgrade-compile-error.patch
++Patch1032: fix-CentOS-7.6-upgrade-compile-error.patch
 +# Compile fix for disabling CONFIG_MEMCG_KMEM
-+Patch1032: compile-fix-for-disabling-CONFIG_MEMCG_KMEM.patch
++Patch1033: compile-fix-for-disabling-CONFIG_MEMCG_KMEM.patch
  
  BuildRoot: %{_tmppath}/kernel-%{KVERREL}-root
  
-@@ -780,6 +785,9 @@ ApplyPatch aic94xx-Skip-reading-user-settings-if-flash-is-not-f.patch
+@@ -781,6 +786,9 @@ ApplyPatch dpt_i2o-fix-build-warning.patch
- ApplyPatch dpt_i2o-fix-build-warning.patch
  ApplyPatch turn-off-write-same-in-smartqpi-driver.patch
  ApplyPatch restrict-iSCSI-kthreads-to-CPUs-in-cpu_kthread_mask.patch
+ ApplyPatch robustify-CFS-bandwidth-timer-locking.patch
 +ApplyPatch fix-compilation-issues.patch
 +ApplyPatch fix-CentOS-7.6-upgrade-compile-error.patch
 +ApplyPatch compile-fix-for-disabling-CONFIG_MEMCG_KMEM.patch

kernel/kernel-rt/centos/meta_patches/Kernel-source-patches-for-TiC.patch

@@ -1,21 +1,18 @@
-From 4412b4d092c2f38feed1d0f8ea1a69c799315663 Mon Sep 17 00:00:00 2001
+From 6a04eb3881ccb3c592b4b47d36bde90f1e33c598 Mon Sep 17 00:00:00 2001
-Message-Id: <4412b4d092c2f38feed1d0f8ea1a69c799315663.1528227675.git.Jim.Somerville@windriver.com>
+From: Alex Kozyrev <alex.kozyrev@windriver.com>
-In-Reply-To: <dc53b79fa7cf0faa28e58d72ef4b34d405b469c0.1528227674.git.Jim.Somerville@windriver.com>
+Date: Mon, 29 Jul 2019 11:48:49 -0400
-References: <dc53b79fa7cf0faa28e58d72ef4b34d405b469c0.1528227674.git.Jim.Somerville@windriver.com>
-From: Jim Somerville <Jim.Somerville@windriver.com>
-Date: Mon, 23 Apr 2018 15:19:36 -0400
 Subject: [PATCH 2/3] Kernel source patches for TiC
 
-Signed-off-by: Jim Somerville <Jim.Somerville@windriver.com>
+Signed-off-by: Alex Kozyrev <alex.kozyrev@windriver.com>
 ---
- SPECS/kernel-rt.spec | 62 ++++++++++++++++++++++++++++++++++++++++++++++++++++
+ SPECS/kernel-rt.spec | 64 ++++++++++++++++++++++++++++++++++++++++++++++++++++
- 1 file changed, 62 insertions(+)
+ 1 file changed, 64 insertions(+)
 
 diff --git a/SPECS/kernel-rt.spec b/SPECS/kernel-rt.spec
 index 905ae52..15114e6 100644
 --- a/SPECS/kernel-rt.spec
 +++ b/SPECS/kernel-rt.spec
-@@ -386,6 +386,38 @@ Source1000: modprobe-dccp-blacklist.conf
+@@ -386,6 +386,39 @@ Source1000: modprobe-dccp-blacklist.conf
  
  # Empty final patch file to facilitate testing of kernel patches
  Patch999999: linux-kernel-test.patch
@@ -51,10 +48,11 @@ index 905ae52..15114e6 100644
 +# DRBD was choking on write same
 +Patch1028: turn-off-write-same-in-smartqpi-driver.patch
 +Patch1029: restrict-iSCSI-kthreads-to-CPUs-in-cpu_kthread_mask.patch
++Patch1030: robustify-CFS-bandwidth-timer-locking.patch
  
  BuildRoot: %{_tmppath}/kernel-%{KVERREL}-root
  
-@@ -718,6 +750,36 @@ cp %{SOURCE38} .
+@@ -718,6 +751,37 @@ cp %{SOURCE38} .
  
  ## Apply Patches here
  ApplyPatch linux-kernel-test.patch
@@ -88,6 +86,7 @@ index 905ae52..15114e6 100644
 +ApplyPatch dpt_i2o-fix-build-warning.patch
 +ApplyPatch turn-off-write-same-in-smartqpi-driver.patch
 +ApplyPatch restrict-iSCSI-kthreads-to-CPUs-in-cpu_kthread_mask.patch
++ApplyPatch robustify-CFS-bandwidth-timer-locking.patch
  
  # move off upstream version mechanism
  if [ -e localversion-rt ]; then

kernel/kernel-rt/centos/patches/robustify-CFS-bandwidth-timer-locking.patch

@@ -0,0 +1,166 @@
+From c0ad4aa4d8416a39ad262a2bd68b30acd951bf0e Mon Sep 17 00:00:00 2001
+From: Peter Zijlstra <peterz@infradead.org>
+Date: Mon, 7 Jan 2019 13:52:31 +0100
+Subject: sched/fair: Robustify CFS-bandwidth timer locking
+
+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.
+
+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>
+Cc: Linus Torvalds <torvalds@linux-foundation.org>
+Cc: Peter Zijlstra <peterz@infradead.org>
+Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
+Cc: Thomas Gleixner <tglx@linutronix.de>
+Link: https://lkml.kernel.org/r/20190107125231.GE14122@hirez.programming.kicks-ass.net
+Signed-off-by: Ingo Molnar <mingo@kernel.org>
+Signed-off-by: Alex Kozyrev <alex.kozyrev@windriver.com>
+
+---
+ kernel/sched/fair.c | 33 ++++++++++++++++++---------------
+ 1 file changed, 18 insertions(+), 15 deletions(-)
+
+diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
+index d3d746b..e9a8d95 100644
+--- a/kernel/sched/fair.c
++++ b/kernel/sched/fair.c
+@@ -3488,13 +3488,14 @@ static u64 distribute_cfs_runtime(struct cfs_bandwidth *cfs_b,
+ 	struct cfs_rq *cfs_rq;
+ 	u64 runtime;
+ 	u64 starting_runtime = remaining;
++	unsigned long flags;
+ 
+ 	rcu_read_lock();
+ 	list_for_each_entry_rcu(cfs_rq, &cfs_b->throttled_cfs_rq,
+ 				throttled_list) {
+ 		struct rq *rq = rq_of(cfs_rq);
+ 
+-		raw_spin_lock(&rq->lock);
++		raw_spin_lock_irqsave(&rq->lock, flags);
+ 		if (!cfs_rq_throttled(cfs_rq))
+ 			goto next;
+ 
+@@ -3511,7 +3512,7 @@ static u64 distribute_cfs_runtime(struct cfs_bandwidth *cfs_b,
+ 			unthrottle_cfs_rq(cfs_rq);
+ 
+ next:
+-		raw_spin_unlock(&rq->lock);
++		raw_spin_unlock_irqrestore(&rq->lock, flags);
+ 
+ 		if (!remaining)
+ 			break;
+@@ -3527,7 +3528,7 @@ next:
+  * 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;
+@@ -3576,11 +3577,11 @@ 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);
+@@ -3689,17 +3690,18 @@ 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;
+ 	}
+ 
+@@ -3710,18 +3712,18 @@ 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);
+ }
+ 
+ /*
+@@ -3785,7 +3787,7 @@ static void check_cfs_rq_runtime(struct cfs_rq *cfs_rq)
+ }
+ 
+ static inline u64 default_cfs_period(void);
+-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 overrunn, unsigned long flags);
+ static void do_sched_cfs_slack_timer(struct cfs_bandwidth *cfs_b);
+ 
+ static enum hrtimer_restart sched_cfs_slack_timer(struct hrtimer *timer)
+@@ -3802,10 +3804,11 @@ static enum hrtimer_restart sched_cfs_period_timer(struct hrtimer *timer)
+ 	struct cfs_bandwidth *cfs_b =
+ 		container_of(timer, struct cfs_bandwidth, period_timer);
+ 	ktime_t now;
++	unsigned long flags;
+ 	int overrun;
+ 	int idle = 0;
+ 
+-	raw_spin_lock(&cfs_b->lock);
++	raw_spin_lock_irqsave(&cfs_b->lock, flags);
+ 	for (;;) {
+ 		now = hrtimer_cb_get_time(timer);
+ 		overrun = hrtimer_forward(timer, now, cfs_b->period);
+@@ -3813,9 +3816,9 @@ static enum hrtimer_restart sched_cfs_period_timer(struct hrtimer *timer)
+ 		if (!overrun)
+ 			break;
+ 
+-		idle = do_sched_cfs_period_timer(cfs_b, overrun);
++		idle = do_sched_cfs_period_timer(cfs_b, overrun, flags);
+ 	}
+-	raw_spin_unlock(&cfs_b->lock);
++	raw_spin_unlock_irqrestore(&cfs_b->lock, flags);
+ 
+ 	return idle ? HRTIMER_NORESTART : HRTIMER_RESTART;
+ }
+-- 
+1.8.3.1
+