The warning was intended to spot complete_all() users from hardirq
context on PREEMPT_RT. The warning as-is will also trigger in interrupt
handlers, which are threaded on PREEMPT_RT, which was not intended.
Use lockdep_assert_RT_in_threaded_ctx() which triggers in non-preemptive
context on PREEMPT_RT.
Fixes: a5c6234e1028 ("completion: Use simple wait queues")
Reported-by: kernel test robot <rong.a.chen@intel.com>
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200323152019.4qjwluldohuh3by5@linutronix.de
ARCH_SAVE_PAGE_KEYS has been introduced in order to be able to save
and restore s390 specific storage keys into a hibernation image.
With hibernation support removed from s390 there is no point in
keeping the callbacks.
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: Peter Oberparleiter <oberpar@linux.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>
Commit 3f8fd02b1bf1 ("mm/vmalloc: Sync unmappings in
__purge_vmap_area_lazy()") introduced a call to vmalloc_sync_all() in
the vunmap() code-path. While this change was necessary to maintain
correctness on x86-32-pae kernels, it also adds additional cycles for
architectures that don't need it.
Specifically on x86-64 with CONFIG_VMAP_STACK=y some people reported
severe performance regressions in micro-benchmarks because it now also
calls the x86-64 implementation of vmalloc_sync_all() on vunmap(). But
the vmalloc_sync_all() implementation on x86-64 is only needed for newly
created mappings.
To avoid the unnecessary work on x86-64 and to gain the performance
back, split up vmalloc_sync_all() into two functions:
* vmalloc_sync_mappings(), and
* vmalloc_sync_unmappings()
Most call-sites to vmalloc_sync_all() only care about new mappings being
synchronized. The only exception is the new call-site added in the
above mentioned commit.
Shile Zhang directed us to a report of an 80% regression in reaim
throughput.
Fixes: 3f8fd02b1bf1 ("mm/vmalloc: Sync unmappings in __purge_vmap_area_lazy()")
Reported-by: kernel test robot <oliver.sang@intel.com>
Reported-by: Shile Zhang <shile.zhang@linux.alibaba.com>
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Tested-by: Borislav Petkov <bp@suse.de>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com> [GHES]
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: <stable@vger.kernel.org>
Link: http://lkml.kernel.org/r/20191009124418.8286-1-joro@8bytes.org
Link: https://lists.01.org/hyperkitty/list/lkp@lists.01.org/thread/4D3JPPHBNOSPFK2KEPC6KGKS6J25AIDB/
Link: http://lkml.kernel.org/r/20191113095530.228959-1-shile.zhang@linux.alibaba.com
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Currently, rcu_barrier() ignores offline CPUs, However, it is possible
for an offline no-CBs CPU to have callbacks queued, and rcu_barrier()
must wait for those callbacks. This commit therefore makes rcu_barrier()
directly invoke the rcu_barrier_func() with interrupts disabled for such
CPUs. This requires passing the CPU number into this function so that
it can entrain the rcu_barrier() callback onto the correct CPU's callback
list, given that the code must instead execute on the current CPU.
While in the area, this commit fixes a bug where the first CPU's callback
might have been invoked before rcu_segcblist_entrain() returned, which
would also result in an early wakeup.
Fixes: 5d6742b37727 ("rcu/nocb: Use rcu_segcblist for no-CBs CPUs")
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
[ paulmck: Apply optimization feedback from Boqun Feng. ]
Cc: <stable@vger.kernel.org> # 5.5.x
The rcu_state structure's gp_seq field is only to be modified by the RCU
grace-period kthread, which is single-threaded. This commit therefore
enlists KCSAN's help in enforcing this restriction.
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
The handling of notify->work did not properly maintain notify->kref in two
cases:
1) where the work was already scheduled, another irq_set_affinity_locked()
would get the ref and (no-op-ly) schedule the work. Thus when
irq_affinity_notify() ran, it would drop the original ref but not the
additional one.
2) when cancelling the (old) work in irq_set_affinity_notifier(), if there
was outstanding work a ref had been got for it but was never put.
Fix both by checking the return values of the work handling functions
(schedule_work() for (1) and cancel_work_sync() for (2)) and put the
extra ref if the return value indicates preexisting work.
Fixes: cd7eab44e994 ("genirq: Add IRQ affinity notifiers")
Fixes: 59c39840f5ab ("genirq: Prevent use-after-free and work list corruption")
Signed-off-by: Edward Cree <ecree@solarflare.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ben Hutchings <ben@decadent.org.uk>
Link: https://lkml.kernel.org/r/24f5983f-2ab5-e83a-44ee-a45b5f9300f5@solarflare.com
Continue what commit:
d820ac4c2fa8 ("locking: rename trace_softirq_[enter|exit] => lockdep_softirq_[enter|exit]")
started, rename these to avoid confusing them with tracepoints.
git grep -l "trace_\(soft\|hard\)\(irq_context\|irqs_enabled\)" | while read file;
do
sed -ie 's/trace_\(soft\|hard\)\(irq_context\|irqs_enabled\)/lockdep_\1\2/g' $file;
done
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Will Deacon <will@kernel.org>
Link: https://lkml.kernel.org/r/20200320115859.178626842@infradead.org
Continue what commit:
d820ac4c2fa8 ("locking: rename trace_softirq_[enter|exit] => lockdep_softirq_[enter|exit]")
started, rename these to avoid confusing them with tracepoints.
git grep -l "trace_softirqs_\(on\|off\)" | while read file;
do
sed -ie 's/trace_softirqs_\(on\|off\)/lockdep_softirqs_\1/g' $file;
done
Reported-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Will Deacon <will@kernel.org>
Link: https://lkml.kernel.org/r/20200320115859.119434738@infradead.org
Continue what commit:
d820ac4c2fa8 ("locking: rename trace_softirq_[enter|exit] => lockdep_softirq_[enter|exit]")
started, rename these to avoid confusing them with tracepoints.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Will Deacon <will@kernel.org>
Link: https://lkml.kernel.org/r/20200320115859.060481361@infradead.org
Splitting run_posix_cpu_timers() into two parts is work in progress which
is stuck on other entry code related problems. The heavy lifting which
involves locking of sighand lock will be moved into task context so the
necessary execution time is burdened on the task and not on interrupt
context.
Until this work completes lockdep with the spinlock nesting rules enabled
would emit warnings for this known context.
Prevent it by setting "->irq_config = 1" for the invocation of
run_posix_cpu_timers() so lockdep does not complain when sighand lock is
acquried. This will be removed once the split is completed.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200321113242.751182723@linutronix.de
Mark irq_work items with IRQ_WORK_HARD_IRQ which should be invoked in
hardirq context even on PREEMPT_RT. IRQ_WORK without this flag will be
invoked in softirq context on PREEMPT_RT.
Set ->irq_config to 1 for the IRQ_WORK items which are invoked in softirq
context so lockdep knows that these can safely acquire a spinlock_t.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200321113242.643576700@linutronix.de
Set current->irq_config = 1 for hrtimers which are not marked to expire in
hard interrupt context during hrtimer_init(). These timers will expire in
softirq context on PREEMPT_RT.
Setting this allows lockdep to differentiate these timers. If a timer is
marked to expire in hard interrupt context then the timer callback is not
supposed to acquire a regular spinlock instead of a raw_spinlock in the
expiry callback.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200321113242.534508206@linutronix.de
Extend lockdep to validate lock wait-type context.
The current wait-types are:
LD_WAIT_FREE, /* wait free, rcu etc.. */
LD_WAIT_SPIN, /* spin loops, raw_spinlock_t etc.. */
LD_WAIT_CONFIG, /* CONFIG_PREEMPT_LOCK, spinlock_t etc.. */
LD_WAIT_SLEEP, /* sleeping locks, mutex_t etc.. */
Where lockdep validates that the current lock (the one being acquired)
fits in the current wait-context (as generated by the held stack).
This ensures that there is no attempt to acquire mutexes while holding
spinlocks, to acquire spinlocks while holding raw_spinlocks and so on. In
other words, its a more fancy might_sleep().
Obviously RCU made the entire ordeal more complex than a simple single
value test because RCU can be acquired in (pretty much) any context and
while it presents a context to nested locks it is not the same as it
got acquired in.
Therefore its necessary to split the wait_type into two values, one
representing the acquire (outer) and one representing the nested context
(inner). For most 'normal' locks these two are the same.
[ To make static initialization easier we have the rule that:
.outer == INV means .outer == .inner; because INV == 0. ]
It further means that its required to find the minimal .inner of the held
stack to compare against the outer of the new lock; because while 'normal'
RCU presents a CONFIG type to nested locks, if it is taken while already
holding a SPIN type it obviously doesn't relax the rules.
Below is an example output generated by the trivial test code:
raw_spin_lock(&foo);
spin_lock(&bar);
spin_unlock(&bar);
raw_spin_unlock(&foo);
[ BUG: Invalid wait context ]
-----------------------------
swapper/0/1 is trying to lock:
ffffc90000013f20 (&bar){....}-{3:3}, at: kernel_init+0xdb/0x187
other info that might help us debug this:
1 lock held by swapper/0/1:
#0: ffffc90000013ee0 (&foo){+.+.}-{2:2}, at: kernel_init+0xd1/0x187
The way to read it is to look at the new -{n,m} part in the lock
description; -{3:3} for the attempted lock, and try and match that up to
the held locks, which in this case is the one: -{2,2}.
This tells that the acquiring lock requires a more relaxed environment than
presented by the lock stack.
Currently only the normal locks and RCU are converted, the rest of the
lockdep users defaults to .inner = INV which is ignored. More conversions
can be done when desired.
The check for spinlock_t nesting is not enabled by default. It's a separate
config option for now as there are known problems which are currently
addressed. The config option allows to identify these problems and to
verify that the solutions found are indeed solving them.
The config switch will be removed and the checks will permanently enabled
once the vast majority of issues has been addressed.
[ bigeasy: Move LD_WAIT_FREE,… out of CONFIG_LOCKDEP to avoid compile
failure with CONFIG_DEBUG_SPINLOCK + !CONFIG_LOCKDEP]
[ tglx: Add the config option ]
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200321113242.427089655@linutronix.de
completion uses a wait_queue_head_t to enqueue waiters.
wait_queue_head_t contains a spinlock_t to protect the list of waiters
which excludes it from being used in truly atomic context on a PREEMPT_RT
enabled kernel.
The spinlock in the wait queue head cannot be replaced by a raw_spinlock
because:
- wait queues can have custom wakeup callbacks, which acquire other
spinlock_t locks and have potentially long execution times
- wake_up() walks an unbounded number of list entries during the wake up
and may wake an unbounded number of waiters.
For simplicity and performance reasons complete() should be usable on
PREEMPT_RT enabled kernels.
completions do not use custom wakeup callbacks and are usually single
waiter, except for a few corner cases.
Replace the wait queue in the completion with a simple wait queue (swait),
which uses a raw_spinlock_t for protecting the waiter list and therefore is
safe to use inside truly atomic regions on PREEMPT_RT.
There is no semantical or functional change:
- completions use the exclusive wait mode which is what swait provides
- complete() wakes one exclusive waiter
- complete_all() wakes all waiters while holding the lock which protects
the wait queue against newly incoming waiters. The conversion to swait
preserves this behaviour.
complete_all() might cause unbound latencies with a large number of waiters
being woken at once, but most complete_all() usage sites are either in
testing or initialization code or have only a really small number of
concurrent waiters which for now does not cause a latency problem. Keep it
simple for now.
The fixup of the warning check in the USB gadget driver is just a straight
forward conversion of the lockless waiter check from one waitqueue type to
the other.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Reviewed-by: Davidlohr Bueso <dbueso@suse.de>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Acked-by: Linus Torvalds <torvalds@linux-foundation.org>
Link: https://lkml.kernel.org/r/20200321113242.317954042@linutronix.de
As a preparation to use simple wait queues for completions:
- Provide swake_up_all_locked() to support complete_all()
- Make __prepare_to_swait() public available
This is done to enable the usage of complete() within truly atomic contexts
on a PREEMPT_RT enabled kernel.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200321113242.228481202@linutronix.de
seqlock consists of a sequence counter and a spinlock_t which is used to
serialize the writers. spinlock_t is substituted by a "sleeping" spinlock
on PREEMPT_RT enabled kernels which breaks the usage in the timekeeping
code as the writers are executed in hard interrupt and therefore
non-preemptible context even on PREEMPT_RT.
The spinlock in seqlock cannot be unconditionally replaced by a
raw_spinlock_t as many seqlock users have nesting spinlock sections or
other code which is not suitable to run in truly atomic context on RT.
Instead of providing a raw_seqlock API for a single use case, open code the
seqlock for the jiffies use case and implement it with a raw_spinlock_t and
a sequence counter.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200321113242.120587764@linutronix.de
Extend rcuwait_wait_event() with a state variable so that it is not
restricted to UNINTERRUPTIBLE waits.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200321113241.824030968@linutronix.de
Trying to initialize a structure with "= {};" will not always clean out
all padding locations in a structure. So be explicit and call memset to
initialize everything for a number of bpf information structures that
are then copied from userspace, sometimes from smaller memory locations
than the size of the structure.
Reported-by: Daniel Borkmann <daniel@iogearbox.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Link: https://lore.kernel.org/bpf/20200320162258.GA794295@kroah.com
For the bpf syscall, we are relying on the compiler to properly zero out
the bpf_attr union that we copy userspace data into. Unfortunately that
doesn't always work properly, padding and other oddities might not be
correctly zeroed, and in some tests odd things have been found when the
stack is pre-initialized to other values.
Fix this by explicitly memsetting the structure to 0 before using it.
Reported-by: Maciej Żenczykowski <maze@google.com>
Reported-by: John Stultz <john.stultz@linaro.org>
Reported-by: Alexander Potapenko <glider@google.com>
Reported-by: Alistair Delva <adelva@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Link: https://android-review.googlesource.com/c/kernel/common/+/1235490
Link: https://lore.kernel.org/bpf/20200320094813.GA421650@kroah.com
nmi_enter() does lockdep_off() and hence lockdep ignores everything.
And NMI context makes it impossible to do full IN-NMI tracking like we
do IN-HARDIRQ, that could result in graph_lock recursion.
However, since look_up_lock_class() is lockless, we can find the class
of a lock that has prior use and detect IN-NMI after USED, just not
USED after IN-NMI.
NOTE: By shifting the lockdep_off() recursion count to bit-16, we can
easily differentiate between actual recursion and off.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Link: https://lkml.kernel.org/r/20200221134215.090538203@infradead.org
There were two patterns for lockdep_recursion:
Pattern-A:
if (current->lockdep_recursion)
return
current->lockdep_recursion = 1;
/* do stuff */
current->lockdep_recursion = 0;
Pattern-B:
current->lockdep_recursion++;
/* do stuff */
current->lockdep_recursion--;
But a third pattern has emerged:
Pattern-C:
current->lockdep_recursion = 1;
/* do stuff */
current->lockdep_recursion = 0;
And while this isn't broken per-se, it is highly dangerous because it
doesn't nest properly.
Get rid of all Pattern-C instances and shore up Pattern-A with a
warning.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200313093325.GW12561@hirez.programming.kicks-ass.net
Qian Cai reported a bug when PROVE_RCU_LIST=y, and read on /proc/lockdep
triggered a warning:
[ ] DEBUG_LOCKS_WARN_ON(current->hardirqs_enabled)
...
[ ] Call Trace:
[ ] lock_is_held_type+0x5d/0x150
[ ] ? rcu_lockdep_current_cpu_online+0x64/0x80
[ ] rcu_read_lock_any_held+0xac/0x100
[ ] ? rcu_read_lock_held+0xc0/0xc0
[ ] ? __slab_free+0x421/0x540
[ ] ? kasan_kmalloc+0x9/0x10
[ ] ? __kmalloc_node+0x1d7/0x320
[ ] ? kvmalloc_node+0x6f/0x80
[ ] __bfs+0x28a/0x3c0
[ ] ? class_equal+0x30/0x30
[ ] lockdep_count_forward_deps+0x11a/0x1a0
The warning got triggered because lockdep_count_forward_deps() call
__bfs() without current->lockdep_recursion being set, as a result
a lockdep internal function (__bfs()) is checked by lockdep, which is
unexpected, and the inconsistency between the irq-off state and the
state traced by lockdep caused the warning.
Apart from this warning, lockdep internal functions like __bfs() should
always be protected by current->lockdep_recursion to avoid potential
deadlocks and data inconsistency, therefore add the
current->lockdep_recursion on-and-off section to protect __bfs() in both
lockdep_count_forward_deps() and lockdep_count_backward_deps()
Reported-by: Qian Cai <cai@lca.pw>
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200312151258.128036-1-boqun.feng@gmail.com
This NULL check is reversed so it leads to a Smatch warning and
presumably a NULL dereference.
kernel/events/core.c:1598 perf_event_groups_less()
error: we previously assumed 'right->cgrp->css.cgroup' could be null
(see line 1590)
Fixes: 95ed6c707f26 ("perf/cgroup: Order events in RB tree by cgroup id")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200312105637.GA8960@mwanda
Kan and Andi reported that we fail to kill rotation when the flexible
events go empty, but the context does not. XXX moar
Fixes: fd7d55172d1e ("perf/cgroups: Don't rotate events for cgroups unnecessarily")
Reported-by: Andi Kleen <ak@linux.intel.com>
Reported-by: Kan Liang <kan.liang@linux.intel.com>
Tested-by: Kan Liang <kan.liang@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200305123851.GX2596@hirez.programming.kicks-ass.net
In update_sg_wakeup_stats(), the comment says:
Computing avg_load makes sense only when group is fully
busy or overloaded.
But, the code below this comment does not check like this.
From reading the code about avg_load in other functions, I
confirm that avg_load should be calculated in fully busy or
overloaded case. The comment is correct and the checking
condition is wrong. So, change that condition.
Fixes: 57abff067a08 ("sched/fair: Rework find_idlest_group()")
Signed-off-by: Tao Zhou <ouwen210@hotmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Acked-by: Mel Gorman <mgorman@suse.de>
Link: https://lkml.kernel.org/r/Message-ID:
If we failed to find a fitting CPU, in cpupri_find(), we only fallback
to the level we found a hit at.
But Steve suggested to fallback to a second full scan instead as this
could be a better effort.
https://lore.kernel.org/lkml/20200304135404.146c56eb@gandalf.local.home/
We trigger the 2nd search unconditionally since the argument about
triggering a full search is that the recorded fall back level might have
become empty by then. Which means storing any data about what happened
would be meaningless and stale.
I had a humble try at timing it and it seemed okay for the small 6 CPUs
system I was running on
https://lore.kernel.org/lkml/20200305124324.42x6ehjxbnjkklnh@e107158-lin.cambridge.arm.com/
On large system this second full scan could be expensive. But there are
no users outside capacity awareness for this fitness function at the
moment. Heterogeneous systems tend to be small with 8cores in total.
Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Qais Yousef <qais.yousef@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Link: https://lkml.kernel.org/r/20200310142219.syxzn5ljpdxqtbgx@e107158-lin.cambridge.arm.com
when we create a kthread with ktrhead_create_on_cpu(),the child thread
entry is ktread.c:ktrhead() which will be preempted by the parent after
call complete(done) while schedule() is not called yet,then the parent
will call wait_task_inactive(child) but the child is still on the runqueue,
so the parent will schedule_hrtimeout() for 1 jiffy,it will waste a lot of
time,especially on startup.
parent child
ktrhead_create_on_cpu()
wait_fo_completion(&done) -----> ktread.c:ktrhead()
|----- complete(done);--wakeup and preempted by parent
kthread_bind() <------------| |-> schedule();--dequeue here
wait_task_inactive(child) |
schedule_hrtimeout(1 jiffy) -|
So we hope the child just wakeup parent but not preempted by parent, and the
child is going to call schedule() soon,then the parent will not call
schedule_hrtimeout(1 jiffy) as the child is already dequeue.
The same issue for ktrhead_park()&&kthread_parkme().
This patch can save 120ms on rk312x startup with CONFIG_HZ=300.
Signed-off-by: Liang Chen <cl@rock-chips.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Link: https://lkml.kernel.org/r/20200306070133.18335-2-cl@rock-chips.com
During load_balancing, a group with spare capacity will try to pull some
utilizations from an overloaded group. In such case, the load balance
looks for the runqueue with the highest utilization. Nevertheless, it
should also ensure that there are some pending tasks to pull otherwise
the load balance will fail to pull a task and the spread of the load will
be delayed.
This situation is quite transient but it's possible to highlight the
effect with a short run of sysbench test so the time to spread task impacts
the global result significantly.
Below are the average results for 15 iterations on an arm64 octo core:
sysbench --test=cpu --num-threads=8 --max-requests=1000 run
tip/sched/core +patchset
total time: 172ms 158ms
per-request statistics:
avg: 1.337ms 1.244ms
max: 21.191ms 10.753ms
The average max doesn't fully reflect the wide spread of the value which
ranges from 1.350ms to more than 41ms for the tip/sched/core and from
1.350ms to 21ms with the patch.
Other factors like waiting for an idle load balance or cache hotness
can delay the spreading of the tasks which explains why we can still
have up to 21ms with the patch.
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200312165429.990-1-vincent.guittot@linaro.org
During our testing, we found a case that shares no longer
working correctly, the cgroup topology is like:
/sys/fs/cgroup/cpu/A (shares=102400)
/sys/fs/cgroup/cpu/A/B (shares=2)
/sys/fs/cgroup/cpu/A/B/C (shares=1024)
/sys/fs/cgroup/cpu/D (shares=1024)
/sys/fs/cgroup/cpu/D/E (shares=1024)
/sys/fs/cgroup/cpu/D/E/F (shares=1024)
The same benchmark is running in group C & F, no other tasks are
running, the benchmark is capable to consumed all the CPUs.
We suppose the group C will win more CPU resources since it could
enjoy all the shares of group A, but it's F who wins much more.
The reason is because we have group B with shares as 2, since
A->cfs_rq.load.weight == B->se.load.weight == B->shares/nr_cpus,
so A->cfs_rq.load.weight become very small.
And in calc_group_shares() we calculate shares as:
load = max(scale_load_down(cfs_rq->load.weight), cfs_rq->avg.load_avg);
shares = (tg_shares * load) / tg_weight;
Since the 'cfs_rq->load.weight' is too small, the load become 0
after scale down, although 'tg_shares' is 102400, shares of the se
which stand for group A on root cfs_rq become 2.
While the se of D on root cfs_rq is far more bigger than 2, so it
wins the battle.
Thus when scale_load_down() scale real weight down to 0, it's no
longer telling the real story, the caller will have the wrong
information and the calculation will be buggy.
This patch add check in scale_load_down(), so the real weight will
be >= MIN_SHARES after scale, after applied the group C wins as
expected.
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Michael Wang <yun.wang@linux.alibaba.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lkml.kernel.org/r/38e8e212-59a1-64b2-b247-b6d0b52d8dc1@linux.alibaba.com
The task->flags is a 32-bits flag, in which 31 bits have already been
consumed. So it is hardly to introduce other new per process flag.
Currently there're still enough spaces in the bit-field section of
task_struct, so we can define the memstall state as a single bit in
task_struct instead.
This patch also removes an out-of-date comment pointed by Matthew.
Suggested-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Link: https://lkml.kernel.org/r/1584408485-1921-1-git-send-email-laoar.shao@gmail.com
When switching tasks running on a CPU, the psi state of a cgroup
containing both of these tasks does not change. Right now, we don't
exploit that, and can perform many unnecessary state changes in nested
hierarchies, especially when most activity comes from one leaf cgroup.
This patch implements an optimization where we only update cgroups
whose state actually changes during a task switch. These are all
cgroups that contain one task but not the other, up to the first
shared ancestor. When both tasks are in the same group, we don't need
to update anything at all.
We can identify the first shared ancestor by walking the groups of the
incoming task until we see TSK_ONCPU set on the local CPU; that's the
first group that also contains the outgoing task.
The new psi_task_switch() is similar to psi_task_change(). To allow
code reuse, move the task flag maintenance code into a new function
and the poll/avg worker wakeups into the shared psi_group_change().
Suggested-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200316191333.115523-3-hannes@cmpxchg.org
For simplicity, cpu pressure is defined as having more than one
runnable task on a given CPU. This works on the system-level, but it
has limitations in a cgrouped reality: When cpu.max is in use, it
doesn't capture the time in which a task is not executing on the CPU
due to throttling. Likewise, it doesn't capture the time in which a
competing cgroup is occupying the CPU - meaning it only reflects
cgroup-internal competitive pressure, not outside pressure.
Enable tracking of currently executing tasks, and then change the
definition of cpu pressure in a cgroup from
NR_RUNNING > 1
to
NR_RUNNING > ON_CPU
which will capture the effects of cpu.max as well as competition from
outside the cgroup.
After this patch, a cgroup running `stress -c 1` with a cpu.max
setting of 5000 10000 shows ~50% continuous CPU pressure.
Signed-off-by: Johannes Weiner <hannes@cmpxchg.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20200316191333.115523-2-hannes@cmpxchg.org
Currently, when updating the affinity of tasks via either cpusets.cpus,
or, sched_setaffinity(); tasks not currently running within the newly
specified mask will be arbitrarily assigned to the first CPU within the
mask.
This (particularly in the case that we are restricting masks) can
result in many tasks being assigned to the first CPUs of their new
masks.
This:
1) Can induce scheduling delays while the load-balancer has a chance to
spread them between their new CPUs.
2) Can antogonize a poor load-balancer behavior where it has a
difficult time recognizing that a cross-socket imbalance has been
forced by an affinity mask.
This change adds a new cpumask interface to allow iterated calls to
distribute within the intersection of the provided masks.
The cases that this mainly affects are:
- modifying cpuset.cpus
- when tasks join a cpuset
- when modifying a task's affinity via sched_setaffinity(2)
Signed-off-by: Paul Turner <pjt@google.com>
Signed-off-by: Josh Don <joshdon@google.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Qais Yousef <qais.yousef@arm.com>
Tested-by: Qais Yousef <qais.yousef@arm.com>
Link: https://lkml.kernel.org/r/20200311010113.136465-1-joshdon@google.com
When a cfs rq is throttled, the latter and its child are removed from the
leaf list but their nr_running is not changed which includes staying higher
than 1. When a task is enqueued in this throttled branch, the cfs rqs must
be added back in order to ensure correct ordering in the list but this can
only happens if nr_running == 1.
When cfs bandwidth is used, we call unconditionnaly list_add_leaf_cfs_rq()
when enqueuing an entity to make sure that the complete branch will be
added.
Similarly unthrottle_cfs_rq() can stop adding cfs in the list when a parent
is throttled. Iterate the remaining entity to ensure that the complete
branch will be added in the list.
Reported-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Vincent Guittot <vincent.guittot@linaro.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Tested-by: Christian Borntraeger <borntraeger@de.ibm.com>
Tested-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Cc: stable@vger.kernel.org
Cc: stable@vger.kernel.org #v5.1+
Link: https://lkml.kernel.org/r/20200306135257.25044-1-vincent.guittot@linaro.org
As it is fine to perform several "peeks" of event data in the ring buffer
via the iterator before moving it forward, do not re-read the event, just
return what was read before. Otherwise, it can cause inconsistent results,
especially when testing multiple CPU buffers to interleave them.
Link: http://lkml.kernel.org/r/20200317213416.592032170@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
As the iterator will be reading a live buffer, and if the event being read
is on a page that a writer crosses, it will fail and try again, the
condition in rb_iter_peek() that only allows a retry to happen three times
is no longer valid. Allow rb_iter_peek() to retry more than three times
without killing the ring buffer, but only if rb_iter_head_event() had failed
at least once.
Link: http://lkml.kernel.org/r/20200317213416.452888193@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Have the ring_buffer_iter structure have a place to store an event, such
that it can not be overwritten by a writer, and load it in such a way via
rb_iter_head_event() that it will return NULL and reset the iter to the
start of the current page if a writer updated the page.
Link: http://lkml.kernel.org/r/20200317213416.306959216@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Have the ring_buffer_iter structure contain a page_stamp, such that it can
be used to see if the writer entered the page the iterator is on. When going
to a new page, the iterator will record the time stamp of that page. When
reading events, it can copy the event to an internal buffer on the iterator
(to be implemented later), then check the page's time stamp with its own to
see if the writer entered the page. If so, it will need to try to read the
event again.
Link: http://lkml.kernel.org/r/20200317213416.163549674@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
When the ring buffer was first created, the iterator followed the normal
producer/consumer operations where it had both a peek() operation, that just
returned the event at the current location, and a read(), that would return
the event at the current location and also increment the iterator such that
the next peek() or read() will return the next event.
The only use of the ring_buffer_read() is currently to move the iterator to
the next location and nothing now actually reads the event it returns.
Rename this function to its actual use case to ring_buffer_iter_advance(),
which also adds the "iter" part to the name, which is more meaningful. As
the timestamp returned by ring_buffer_read() was never used, there's no
reason that this new version should bother having returning it. It will also
become a void function.
Link: http://lkml.kernel.org/r/20200317213416.018928618@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
It was complained about that when the trace file is read, that the tracing
is disabled, as the iterator expects writing to the buffer it reads is not
updated. Several steps are needed to make the iterator handle a writer,
by testing if things have changed as it reads.
This step is to make ring_buffer_empty() expect the buffer to be changing.
Note if the current location of the iterator is overwritten, then it will
return false as new data is being added. Note, that this means that data
will be skipped.
Link: http://lkml.kernel.org/r/20200317213415.870741809@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
In order to have the iterator read the buffer even when it's still updating,
it requires that the ring buffer iterator saves each event in a separate
location outside the ring buffer such that its use is immutable.
There's one use case that saves off the event returned from the ring buffer
interator and calls it again to look at the next event, before going back to
use the first event. As the ring buffer iterator will only have a single
copy, this use case will no longer be supported.
Instead, have the one use case create its own buffer to store the first
event when looking at the next event. This way, when looking at the first
event again, it wont be corrupted by the second read.
Link: http://lkml.kernel.org/r/20200317213415.722539921@goodmis.org
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
Clang warns:
../kernel/trace/trace.c:9335:33: warning: array comparison always
evaluates to true [-Wtautological-compare]
if (__stop___trace_bprintk_fmt != __start___trace_bprintk_fmt)
^
1 warning generated.
These are not true arrays, they are linker defined symbols, which are
just addresses. Using the address of operator silences the warning and
does not change the runtime result of the check (tested with some print
statements compiled in with clang + ld.lld and gcc + ld.bfd in QEMU).
Link: http://lkml.kernel.org/r/20200220051011.26113-1-natechancellor@gmail.com
Link: https://github.com/ClangBuiltLinux/linux/issues/893
Suggested-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
This reverts commit d441dceb5dce71150f28add80d36d91bbfccba99 due to
boot failures.
Reported-by: Qian Cai <cai@lca.pw>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Waiman Long <longman@redhat.com>
Simplify gen_btf logic to make it work with llvm-objcopy. The existing
'file format' and 'architecture' parsing logic is brittle and does not
work with llvm-objcopy/llvm-objdump.
'file format' output of llvm-objdump>=11 will match GNU objdump, but
'architecture' (bfdarch) may not.
.BTF in .tmp_vmlinux.btf is non-SHF_ALLOC. Add the SHF_ALLOC flag
because it is part of vmlinux image used for introspection. C code
can reference the section via linker script defined __start_BTF and
__stop_BTF. This fixes a small problem that previous .BTF had the
SHF_WRITE flag (objcopy -I binary -O elf* synthesized .data).
Additionally, `objcopy -I binary` synthesized symbols
_binary__btf_vmlinux_bin_start and _binary__btf_vmlinux_bin_stop (not
used elsewhere) are replaced with more commonplace __start_BTF and
__stop_BTF.
Add 2>/dev/null because GNU objcopy (but not llvm-objcopy) warns
"empty loadable segment detected at vaddr=0xffffffff81000000, is this intentional?"
We use a dd command to change the e_type field in the ELF header from
ET_EXEC to ET_REL so that lld will accept .btf.vmlinux.bin.o. Accepting
ET_EXEC as an input file is an extremely rare GNU ld feature that lld
does not intend to support, because this is error-prone.
The output section description .BTF in include/asm-generic/vmlinux.lds.h
avoids potential subtle orphan section placement issues and suppresses
--orphan-handling=warn warnings.
Fixes: df786c9b9476 ("bpf: Force .BTF section start to zero when dumping from vmlinux")
Fixes: cb0cc635c7a9 ("powerpc: Include .BTF section")
Reported-by: Nathan Chancellor <natechancellor@gmail.com>
Signed-off-by: Fangrui Song <maskray@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Tested-by: Stanislav Fomichev <sdf@google.com>
Tested-by: Andrii Nakryiko <andriin@fb.com>
Reviewed-by: Stanislav Fomichev <sdf@google.com>
Reviewed-by: Kees Cook <keescook@chromium.org>
Acked-by: Andrii Nakryiko <andriin@fb.com>
Acked-by: Michael Ellerman <mpe@ellerman.id.au> (powerpc)
Link: https://github.com/ClangBuiltLinux/linux/issues/871
Link: https://lore.kernel.org/bpf/20200318222746.173648-1-maskray@google.com
