Commit ec5fbdfb99 ("cgroup/cpuset: Enable update_tasks_cpumask()
on top_cpuset") enabled us to pull CPUs dedicated to child partitions
from tasks in top_cpuset by ignoring per cpu kthreads. However, there
can be other kthreads that are not per cpu but have PF_NO_SETAFFINITY
flag set to indicate that we shouldn't mess with their CPU affinity.
For other kthreads, their affinity will be changed to skip CPUs dedicated
to child partitions whether it is an isolating or a scheduling one.
As all the per cpu kthreads have PF_NO_SETAFFINITY set, the
PF_NO_SETAFFINITY tasks are essentially a superset of per cpu kthreads.
Fix this issue by dropping the kthread_is_per_cpu() check and checking
the PF_NO_SETAFFINITY flag instead.
Fixes: ec5fbdfb99 ("cgroup/cpuset: Enable update_tasks_cpumask() on top_cpuset")
Signed-off-by: Waiman Long <longman@redhat.com>
Acked-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Tejun Heo <tj@kernel.org>
Empty cpumasks can't intersect with any others. Therefore, testing for
non-emptyness is useless.
Signed-off-by: Yury Norov <yury.norov@gmail.com>
Reviewed-by: Waiman Long <longman@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Turn the default 5 second test delay for hibernation into a
configurable module parameter, so users can determine how
long to wait in this pseudo-hibernate state before resuming
the system.
The configurable delay parameter has been added for suspend, so
add an analogous one for hibernation.
Example (wait 30 seconds);
# echo 30 > /sys/module/hibernate/parameters/pm_test_delay
# echo core > /sys/power/pm_test
Signed-off-by: Zihuan Zhang <zhangzihuan@kylinos.cn>
Reviewed-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://patch.msgid.link/20250507063520.419635-1-zhangzihuan@kylinos.cn
[ rjw: Subject and changelog edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
pm_show_wakelocks() is called to generate a string when showing
attributes /sys/power/wake_(lock|unlock), but the string ends
with an unwanted space that was added back by mistake by commit
c9d967b2ce ("PM: wakeup: simplify the output logic of
pm_show_wakelocks()").
Remove the unwanted space.
Fixes: c9d967b2ce ("PM: wakeup: simplify the output logic of pm_show_wakelocks()")
Signed-off-by: Zijun Hu <quic_zijuhu@quicinc.com>
Link: https://patch.msgid.link/20250505-fix_power-v1-1-0f7f2c2f338c@quicinc.com
[ rjw: Changelog edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
In free_nsproxy() and the error path of create_new_namesapces() the
put_*_ns() calls are guarded by unnecessary NULL checks.
put_pid_ns(), put_ipc_ns(), put_uts_ns(), and put_time_ns() will never
receive a NULL argument unless their namespace type is disabled, and in
this case all four become no-ops at compile time anyway. put_mnt_ns()
will never receive a null argument at any time.
This unguarded usage is in line with other call sites of put_*_ns().
Signed-off-by: Joel Savitz <jsavitz@redhat.com>
Link: https://lore.kernel.org/20250508184930.183040-2-jsavitz@redhat.com
Signed-off-by: Christian Brauner <brauner@kernel.org>
Now all the pieces are in place to actually allow the power subsystem
to freeze/thaw filesystems during suspend/resume. Filesystems are only
frozen and thawed if the power subsystem does actually own the freeze.
We could bubble up errors and fail suspend/resume if the error isn't
EBUSY (aka it's already frozen) but I don't think that this is worth it.
Filesystem freezing during suspend/resume is best-effort. If the user
has 500 ext4 filesystems mounted and 4 fail to freeze for whatever
reason then we simply skip them.
What we have now is already a big improvement and let's see how we fare
with it before making our lives even harder (and uglier) than we have
to.
We add a new sysctl know /sys/power/freeze_filesystems that will allow
userspace to freeze filesystems during suspend/hibernate. For now it
defaults to off. The thaw logic doesn't require checking whether
freezing is enabled because the power subsystem exclusively owns frozen
filesystems for the duration of suspend/hibernate and is able to skip
filesystems it doesn't need to freeze.
Also it is technically possible that filesystem
filesystem_freeze_enabled is true and power freezes the filesystems but
before freezing all processes another process disables
filesystem_freeze_enabled. If power were to place the filesystems_thaw()
call under filesystems_freeze_enabled it would fail to thaw the
fileystems it frozw. The exclusive holder mechanism makes it possible to
iterate through the list without any concern making sure that no
filesystems are left frozen.
Link: https://lore.kernel.org/r/20250402-work-freeze-v2-3-6719a97b52ac@kernel.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
While an event tears down all links to it as an aux, the iteration
happens on the event's group leader instead of the group itself.
If the event is a group leader, it has no effect because the event is
also its own group leader. But otherwise there would be a risk to detach
all the siblings events from the wrong group leader.
It just happens to work because each sibling's aux link is tested
against the right event before proceeding. Also the ctx lock is the same
for the events and their group leader so the iteration is safe.
Yet the iteration is confusing. Clarify the actual intent.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250424161128.29176-5-frederic@kernel.org
The CPU hotplug handlers are called twice: at prepare and online stage.
Their role is to:
1) Enable/disable a CPU context. This is irrelevant and even buggy at
the prepare stage because the CPU is still offline. On early
secondary CPU up, creating an event attached to that CPU might
silently fail because the CPU context is observed as online but the
context installation's IPI failure is ignored.
2) Update the scope cpumasks and re-migrate the events accordingly in
the CPU down case. This is irrelevant at the prepare stage.
3) Remove the events attached to the context of the offlining CPU. It
even uses an (unnecessary) IPI for it. This is also irrelevant at the
prepare stage.
Also none of the *_PREPARE and *_STARTING architecture perf related CPU
hotplug callbacks rely on CPUHP_PERF_PREPARE.
CPUHP_AP_PERF_ONLINE is enough and the right place to perform the work.
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250424161128.29176-4-frederic@kernel.org
The following commit:
da916e96e2 ("perf: Make perf_pmu_unregister() useable")
has introduced two significant event's parent lifecycle changes:
1) An event that has exited now has EVENT_TOMBSTONE as a parent.
This can result in a situation where the delayed wakeup irq_work can
accidentally dereference EVENT_TOMBSTONE on:
CPU 0 CPU 1
----- -----
__schedule()
local_irq_disable()
rq_lock()
<NMI>
perf_event_overflow()
irq_work_queue(&child->pending_irq)
</NMI>
perf_event_task_sched_out()
raw_spin_lock(&ctx->lock)
ctx_sched_out()
ctx->is_active = 0
event_sched_out(child)
raw_spin_unlock(&ctx->lock)
perf_event_release_kernel(parent)
perf_remove_from_context(child)
raw_spin_lock_irq(&ctx->lock)
// Sees !ctx->is_active
// Removes from context inline
__perf_remove_from_context(child)
perf_child_detach(child)
event->parent = EVENT_TOMBSTONE
raw_spin_rq_unlock_irq(rq);
<IRQ>
perf_pending_irq()
perf_event_wakeup(child)
ring_buffer_wakeup(child)
rcu_dereference(child->parent->rb) <--- CRASH
This also concerns the call to kill_fasync() on parent->fasync.
2) The final parent reference count decrement can now happen before the
the final child reference count decrement. ie: the parent can now
be freed before its child. On PREEMPT_RT, this can result in a
situation where the delayed wakeup irq_work can accidentally
dereference a freed parent:
CPU 0 CPU 1 CPU 2
----- ----- ------
perf_pmu_unregister()
pmu_detach_events()
pmu_get_event()
atomic_long_inc_not_zero(&child->refcount)
<NMI>
perf_event_overflow()
irq_work_queue(&child->pending_irq);
</NMI>
<IRQ>
irq_work_run()
wake_irq_workd()
</IRQ>
preempt_schedule_irq()
=========> SWITCH to workd
irq_work_run_list()
perf_pending_irq()
perf_event_wakeup(child)
ring_buffer_wakeup(child)
event = child->parent
perf_event_release_kernel(parent)
// Not last ref, PMU holds it
put_event(child)
// Last ref
put_event(parent)
free_event()
call_rcu(...)
rcu_core()
free_event_rcu()
rcu_dereference(event->rb) <--- CRASH
This also concerns the call to kill_fasync() on parent->fasync.
The "easy" solution to 1) is to check that event->parent is not
EVENT_TOMBSTONE on perf_event_wakeup() (including both ring buffer
and fasync uses).
The "easy" solution to 2) is to turn perf_event_wakeup() to wholefully
run under rcu_read_lock().
However because of 2), sanity would prescribe to make event::parent
an __rcu pointer and annotate each and every users to prove they are
reliable.
Propose an alternate solution and restore the stable pointer to the
parent until all its children have called _free_event() themselves to
avoid any further accident. Also revert the EVENT_TOMBSTONE design
that is mostly here to determine which caller of perf_event_exit_event()
must perform the refcount decrement on a child event matching the
increment in inherit_event().
Arrange instead for checking the attach state of an event prior to its
removal and decrement the refcount of the child accordingly.
Fixes: da916e96e2 ("perf: Make perf_pmu_unregister() useable")
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
When inherit_event() fails after the child allocation but before the
parent refcount has been incremented, calling put_event() wrongly
decrements the reference to the parent, risking to free it too early.
Also pmu_get_event() can't be holding a reference to the child
concurrently at this point since it is under pmus_srcu critical section.
Fix it with restoring the deleted free_event() function and call it on
the failing child in order to free it directly under the verified
assumption that its refcount is only 1. The refcount to the parent is
then voluntarily omitted.
Fixes: da916e96e2 ("perf: Make perf_pmu_unregister() useable")
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250424161128.29176-2-frederic@kernel.org
The ignore_pid boolean on the per CPU data descriptor is updated at
sched_switch when a new task is scheduled in. If the new task is to be
ignored, it is set to true, otherwise it is set to false. The current task
should always have the correct value as it is updated when the task is
scheduled in.
Instead of breaking up the read of this value, which requires preemption
to be disabled, just use this_cpu_read() which gives a snapshot of the
value. Since the value will always be correct for a given task (because
it's updated at sched switch) it doesn't need preemption disabled.
This will also allow trace events to be called with preemption enabled.
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: https://lore.kernel.org/20250505212235.038958766@goodmis.org
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The per CPU "disabled" value was the original way to disable tracing when
the tracing subsystem was first created. Today, the ring buffer
infrastructure has its own way to disable tracing. In fact, things have
changed so much since 2008 that many things ignore the disable flag.
There's no reason for the function tracer to check it, if tracing is
disabled, the ring buffer will not record the event anyway.
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Link: https://lore.kernel.org/20250505212234.868972758@goodmis.org
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Allocate kernel memory for processing CPU string
(/sys/kernel/tracing/osnoise/cpus) also in osnoise_cpus_write to allow
the writing of a CPU string of an arbitrary length.
This replaces the 256-byte buffer, which is insufficient with the rising
number of CPUs. For example, if I wanted to measure on every even CPU
on a system with 256 CPUs, the string would be 456 characters long.
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/20250425091839.343289-1-tglozar@redhat.com
Signed-off-by: Tomas Glozar <tglozar@redhat.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The structure ftrace_func_mapper only contains a single field and that is
a ftrace_hash. It is used to abstract it out from a normal hash to control
users of how it gets modified.
The freeing of a ftrace_func_mapper structure is:
free_ftrace_hash(&mapper->hash);
Without context, this looks like a bug. It should be commented that it is
not a bug and it is freed this way.
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Mark Rutland <mark.rutland@arm.com>
Link: https://lore.kernel.org/20250416165420.5c717420@gandalf.local.home
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The function graph infrastructure uses subops of the function tracer.
These are not shown in enabled_functions. Add a "subops:" section to the
enabled_functions line to show what functions are attached via subops. If
the subops is from the function_graph infrastructure, then show the entry
and return callbacks that are attached.
Here's an example of the output:
schedule_on_each_cpu (1) tramp: 0xffffffffc03ef000 (ftrace_graph_func+0x0/0x60) ->ftrace_graph_func+0x0/0x60 subops: {ent:trace_graph_entry+0x0/0x20 ret:trace_graph_return+0x0/0x150}
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Link: https://lore.kernel.org/20250410153830.5d97f108@gandalf.local.home
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
The lock guard conversion converted raw_spin_lock_irq() to
scoped_guard(raw_spinlock), which is obviously bogus and makes lockdep
mightily unhappy.
Note to self: Copy and pasta without using brain is a patently bad idea.
Fixes: 88a4df117a ("genirq/cpuhotplug: Convert to lock guards")
Reported-by: Borislav Petkov <bp@alien8.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Borislav Petkov <bp@alien8.de>
Doing cpufreq-specific EAS checks that require accessing policy
internals directly from sched_is_eas_possible() is a bit unfortunate,
so introduce cpufreq_ready_for_eas() in cpufreq, move those checks
into that new function and make sched_is_eas_possible() call it.
While at it, address a possible race between the EAS governor check
and governor change by doing the former under the policy rwsem.
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Reviewed-by: Christian Loehle <christian.loehle@arm.com>
Tested-by: Christian Loehle <christian.loehle@arm.com>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Link: https://patch.msgid.link/2317800.iZASKD2KPV@rjwysocki.net
To receive 428dc9fc08 ("sched_ext: bpf_iter_scx_dsq_new() should always
initialize iterator") which conflicts with cdf5a6faa8 ("sched_ext: Move
dsq_hash into scx_sched"). The conflict is a simple context conflict which
can be resolved by taking changes from both changes in the right order.
BPF programs may call next() and destroy() on BPF iterators even after new()
returns an error value (e.g. bpf_for_each() macro ignores error returns from
new()). bpf_iter_scx_dsq_new() could leave the iterator in an uninitialized
state after an error return causing bpf_iter_scx_dsq_next() to dereference
garbage data. Make bpf_iter_scx_dsq_new() always clear $kit->dsq so that
next() and destroy() become noops.
Signed-off-by: Tejun Heo <tj@kernel.org>
Fixes: 650ba21b13 ("sched_ext: Implement DSQ iterator")
Cc: stable@vger.kernel.org # v6.12+
Acked-by: Andrea Righi <arighi@nvidia.com>
This code pattern trips clang up:
if (fail)
goto undo;
guard(lock)(lock);
do_stuff();
return 0;
undo:
...
as it somehow extends the scope of the guard beyond the return statement.
Replace it with a scoped guard to help it to get its act together.
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Closes: https://lore.kernel.org/oe-kbuild-all/202505071809.ajpPxfoZ-lkp@intel.com/
Split hib_submit_io into a sync and async version. The sync version is
a small wrapper around bdev_rw_virt which implements all the logic to
add a kernel direct mapping range to a bio and synchronously submits it,
while the async version is slightly simplified using the
bio_add_virt_nofail for adding the single range.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Rafael J. Wysocki <rafael@kernel.org>
Reviewed-by: Damien Le Moal <dlemoal@kernel.org>
Link: https://lore.kernel.org/r/20250507120451.4000627-13-hch@lst.de
Signed-off-by: Jens Axboe <axboe@kernel.dk>
