As reported by Cody Haas [1], when there is concurrent map lookup and
map update operation in an existing element for htab of maps, the map
lookup procedure may return -ENOENT unexpectedly.
The root cause is twofold:
1) the update of existing element involves two separated list operation
In htab_map_update_elem(), it first inserts the new element at the head
of list, then it deletes the old element. Therefore, it is possible a
lookup operation has already iterated to the middle of the list when a
concurrent update operation begins, and the lookup operation will fail
to find the target element.
2) the immediate reuse of htab element.
It is more subtle. Even through the lookup operation finds the old
element, it is possible that the target element has been removed by a
concurrent update operation, and the element has been reused immediately
by other update operation which runs on the same CPU as the previous
update operation, and the element is inserted into the same bucket list.
After these steps above, when the lookup operation tries to compare the
key in the old element with the expected key, the match will fail
because the key in the old element have been overwritten by other update
operation.
The two-step update process is relatively straightforward to address.
The more challenging aspect is the immediate reuse. As Alexei pointed
out:
So since 2022 both prealloc and no_prealloc reuse elements.
We can consider a new flag for the hash map like F_REUSE_AFTER_RCU_GP
that will use _rcu() flavor of freeing into bpf_ma,
but it has to have a strong reason.
Given that htab of maps doesn't support special field in value and
directly stores the inner map pointer in htab_element, just do in-place
update for htab of maps instead of attempting to address the immediate
reuse issue.
[1]: https://lore.kernel.org/xdp-newbies/CAH7f-ULFTwKdoH_t2SFc5rWCVYLEg-14d1fBYWH2eekudsnTRg@mail.gmail.com/
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Hou Tao <houtao1@huawei.com>
Link: https://lore.kernel.org/r/20250401062250.543403-4-houtao@huaweicloud.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
All hash maps store map key and map value together. The relative offset
of the map value compared to the map key is round_up(key_size, 8).
Therefore, factor out a common helper htab_elem_value() to calculate the
address of the map value instead of duplicating the logic.
Acked-by: Andrii Nakryiko <andrii@kernel.org>
Signed-off-by: Hou Tao <houtao1@huawei.com>
Link: https://lore.kernel.org/r/20250401062250.543403-2-houtao@huaweicloud.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
The link_create.flags are currently not used for multi-kprobes, so return
-EINVAL if it is set, same as for other attach APIs.
We allow target_fd, on the other hand, to have an arbitrary value for
multi-kprobe, as there are existing users (libbpf) relying on this.
Fixes: 0dcac27254 ("bpf: Add multi kprobe link")
Signed-off-by: Tao Chen <chen.dylane@linux.dev>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20250407035752.1108927-1-chen.dylane@linux.dev
Move posix_timers_cache initialization to posixtimer_init(). At that point
the memory subsystem is already up and running.
Also move the cache pointer to the __timer_data variable to avoid
potential false sharing, since it never was marked as __ro_after_init.
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/all/20250402133114.253901-1-edumazet@google.com
scx_has_op is used to encode which ops are implemented by the BPF scheduler
into an array of static_keys. While this saves a bit of branching overhead,
that is unlikely to be noticeable compared to the overall cost. As the
global static_keys can't work with the planned hierarchical multiple
scheduler support, replace the static_key array with a bitmap.
In repeated hackbench runs before and after static_keys removal on an AMD
Ryzen 3900X, I couldn't tell any measurable performance difference.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Changwoo Min <changwoo@igalia.com>
Acked-by: Andrea Righi <arighi@nvidia.com>
scx_ops_allow_queued_wakeup is used to encode SCX_OPS_ALLOW_QUEUED_WAKEUP
into a static_key. The test is gated behind scx_enabled(), and, even when
sched_ext is enabled, is unlikely for the static_key usage to make any
meaningful difference. It is made to use a static_key mostly because there
was no reason not to. However, global static_keys can't work with the
planned hierarchical multiple scheduler support. Remove the static_key and
instead test SCX_OPS_ALLOW_QUEUED_WAKEUP directly.
In repeated hackbench runs before and after static_keys removal on an AMD
Ryzen 3900X, I couldn't tell any measurable performance difference.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Changwoo Min <changwoo@igalia.com>
Acked-by: Andrea Righi <arighi@nvidia.com>
scx_ops_cpu_preempt is used to encode whether ops.cpu_acquire/release() are
implemented into a static_key. These tests aren't hot enough for static_key
usage to make any meaningful difference and are made to use a static_key
mostly because there was no reason not to. However, global static_keys can't
work with the planned hierarchical multiple scheduler support. Remove the
static_key and instead use an internal ops flag SCX_OPS_HAS_CPU_PREEMPT to
record and test whether ops.cpu_acquire/release() are implemented.
In repeated hackbench runs before and after static_keys removal on an AMD
Ryzen 3900X, I couldn't tell any measurable performance difference.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Changwoo Min <changwoo@igalia.com>
Acked-by: Andrea Righi <arighi@nvidia.com>
scx_ops_enq_last/exiting/migration_disabled are used to encode the
corresponding SCX_OPS_ flags into static_keys. These flags aren't hot enough
for static_key usage to make any meaningful difference and are made
static_keys mostly because there was no reason not to. However, global
static_keys can't work with the planned hierarchical multiple scheduler
support. Remove the static_keys and test the ops flags directly.
In repeated hackbench runs before and after static_keys removal on an AMD
Ryzen 3900X, I couldn't tell any measurable performance difference.
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Changwoo Min <changwoo@igalia.com>
Acked-by: Andrea Righi <arighi@nvidia.com>
The following causes a vsnprintf fault:
# echo 's:wake_lat char[] wakee; u64 delta;' >> /sys/kernel/tracing/dynamic_events
# echo 'hist:keys=pid:ts=common_timestamp.usecs if !(common_flags & 0x18)' > /sys/kernel/tracing/events/sched/sched_waking/trigger
# echo 'hist:keys=next_pid:delta=common_timestamp.usecs-$ts:onmatch(sched.sched_waking).trace(wake_lat,next_comm,$delta)' > /sys/kernel/tracing/events/sched/sched_switch/trigger
Because the synthetic event's "wakee" field is created as a dynamic string
(even though the string copied is not). The print format to print the
dynamic string changed from "%*s" to "%s" because another location
(__set_synth_event_print_fmt()) exported this to user space, and user
space did not need that. But it is still used in print_synth_event(), and
the output looks like:
<idle>-0 [001] d..5. 193.428167: wake_lat: wakee=(efault)sshd-sessiondelta=155
sshd-session-879 [001] d..5. 193.811080: wake_lat: wakee=(efault)kworker/u34:5delta=58
<idle>-0 [002] d..5. 193.811198: wake_lat: wakee=(efault)bashdelta=91
bash-880 [002] d..5. 193.811371: wake_lat: wakee=(efault)kworker/u35:2delta=21
<idle>-0 [001] d..5. 193.811516: wake_lat: wakee=(efault)sshd-sessiondelta=129
sshd-session-879 [001] d..5. 193.967576: wake_lat: wakee=(efault)kworker/u34:5delta=50
The length isn't needed as the string is always nul terminated. Just print
the string and not add the length (which was hard coded to the max string
length anyway).
Cc: stable@vger.kernel.org
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Tom Zanussi <zanussi@kernel.org>
Cc: Douglas Raillard <douglas.raillard@arm.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Link: https://lore.kernel.org/20250407154139.69955768@gandalf.local.home
Fixes: 4d38328eb4 ("tracing: Fix synth event printk format for str fields");
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
When building with W=1, this variable is unused for configs with
CONFIG_CMA_SIZE_SEL_PERCENTAGE=y:
kernel/dma/contiguous.c:67:26: error: 'size_bytes' defined but not used [-Werror=unused-const-variable=]
Change this to a macro to avoid the warning.
Fixes: c64be2bb1c ("drivers: add Contiguous Memory Allocator")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Marek Szyprowski <m.szyprowski@samsung.com>
Link: https://lore.kernel.org/r/20250409151557.3890443-1-arnd@kernel.org
Move sparc sysctls (reboot-cmd, stop-a, scons-poweroff and tsb-ratio)
into a new file (arch/sparc/kernel/setup.c). This file will be included
for both 32 and 64 bit sparc. Leave "tsb-ratio" under SPARC64 ifdef as
it was in kernel/sysctl.c. The sysctl table register is called with
arch_initcall placing it after its original place in proc_root_init.
This is part of a greater effort to move ctl tables into their
respective subsystems which will reduce the merge conflicts in
kernel/sysctl.c.
Signed-off-by: Joel Granados <joel.granados@kernel.org>
Move stack_tracer_enabled into trace_stack_sysctl_table. This is part of
a greater effort to move ctl tables into their respective subsystems
which will reduce the merge conflicts in kernel/sysctl.c.
Acked-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Joel Granados <joel.granados@kernel.org>
Move trace ctl tables into their own const array in
kernel/trace/trace.c. The sysctl table register is called with
subsys_initcall placing if after its original place in proc_root_init.
This is part of a greater effort to move ctl tables into their
respective subsystems which will reduce the merge conflicts in
kernel/sysctl.c.
Signed-off-by: Joel Granados <joel.granados@kernel.org>
Acked-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Move print-fatal-signals into its own const ctl table array in
kernel/signal.c. This is part of a greater effort to move ctl tables
into their respective subsystems which will reduce the merge conflicts
in kernel/sysctl.c.
Signed-off-by: Joel Granados <joel.granados@kernel.org>
Move panic, panic_on_oops, panic_print, panic_on_warn into
kerne/panic.c. This is part of a greater effort to move ctl tables into
their respective subsystems which will reduce the merge conflicts in
kernel/sysctl.c.
Signed-off-by: Joel Granados <joel.granados@kernel.org>
When booting BeagleBone Black:
------------[ cut here ]------------
WARNING: CPU: 0 PID: 0 at kernel/locking/lockdep.c:4398 lockdep_hardirqs_on_prepare+0x23c/0x280
DEBUG_LOCKS_WARN_ON(early_boot_irqs_disabled)
CPU: 0 UID: 0 PID: 0 Comm: swapper/0 Not tainted 6.15.0-rc1-boneblack-00004-g195298c3b116 #209 NONE
Hardware name: Generic AM33XX (Flattened Device Tree)
Call trace:
_raw_spin_unlock_irq from irq_map_generic_chip+0x144/0x190
irq_map_generic_chip from irq_domain_associate_locked+0x68/0x164
irq_domain_associate_locked from irq_create_fwspec_mapping+0x34c/0x43c
irq_create_fwspec_mapping from irq_create_of_mapping+0x64/0x8c
irq_create_of_mapping from irq_of_parse_and_map+0x54/0x7c
irq_of_parse_and_map from dmtimer_clkevt_init_common+0x54/0x15c
dmtimer_clkevt_init_common from dmtimer_systimer_init+0x41c/0x5b8
dmtimer_systimer_init from timer_probe+0x68/0xf0
timer_probe from start_kernel+0x4a4/0x6bc
start_kernel from 0x0
irq event stamp: 0
hardirqs last enabled at (0): [<00000000>] 0x0
hardirqs last disabled at (0): [<00000000>] 0x0
softirqs last enabled at (0): [<00000000>] 0x0
softirqs last disabled at (0): [<00000000>] 0x0
---[ end trace 0000000000000000 ]---
and:
------------[ cut here ]------------
WARNING: CPU: 0 PID: 0 at init/main.c:1022 start_kernel+0x4e8/0x6bc
Interrupts were enabled early
CPU: 0 UID: 0 PID: 0 Comm: swapper/0 Tainted: G W 6.15.0-rc1-boneblack-00004-g195298c3b116 #209 NONE
Tainted: [W]=WARN
Hardware name: Generic AM33XX (Flattened Device Tree)
Call trace:
unwind_backtrace from show_stack+0x10/0x14
show_stack from dump_stack_lvl+0x6c/0x90
dump_stack_lvl from __warn+0x70/0x1b0
__warn from warn_slowpath_fmt+0x1d4/0x1ec
warn_slowpath_fmt from start_kernel+0x4e8/0x6bc
start_kernel from 0x0
irq event stamp: 0
hardirqs last enabled at (0): [<00000000>] 0x0
hardirqs last disabled at (0): [<00000000>] 0x0
softirqs last enabled at (0): [<00000000>] 0x0
softirqs last disabled at (0): [<00000000>] 0x0
---[ end trace 0000000000000000 ]---
Fix this by correcting two misconversions of
raw_spin_{,un}lock_irq{save,restore}() to lock guards.
Fixes: 195298c3b1 ("genirq/generic-chip: Convert core code to lock guards")
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/all/514f94c5891c61ac0a4a7fdad113e75db1eea367.1744135467.git.geert+renesas@glider.be
Pull probes fixes from Masami Hiramatsu:
- fprobe: remove fprobe_hlist_node when module unloading
When a fprobe target module is removed, the fprobe_hlist_node should
be removed from the fprobe's hash table to prevent reusing
accidentally if another module is loaded at the same address.
- fprobe: lock module while registering fprobe
The module containing the function to be probeed is locked using a
reference counter until the fprobe registration is complete, which
prevents use after free.
- fprobe-events: fix possible UAF on modules
Basically as same as above, but in the fprobe-events layer we also
need to get module reference counter when we find the tracepoint in
the module.
* tag 'probes-fixes-v6.14' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
tracing: fprobe: Cleanup fprobe hash when module unloading
tracing: fprobe events: Fix possible UAF on modules
tracing: fprobe: Fix to lock module while registering fprobe
Pull cgroup fixes from Tejun Heo:
- A number of cpuset remote partition related fixes and cleanups along
with selftest updates.
- A change from this merge window made cgroup_rstat_updated_list()
called outside cgroup_rstat_lock leading to list corruptions. Fix it
by relocating the call inside the lock.
* tag 'cgroup-for-6.15-rc1-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/cgroup:
cgroup/cpuset: Fix race between newly created partition and dying one
cgroup: rstat: call cgroup_rstat_updated_list with cgroup_rstat_lock
selftest/cgroup: Add a remote partition transition test to test_cpuset_prs.sh
selftest/cgroup: Clean up and restructure test_cpuset_prs.sh
selftest/cgroup: Update test_cpuset_prs.sh to use | as effective CPUs and state separator
cgroup/cpuset: Remove unneeded goto in sched_partition_write() and rename it
cgroup/cpuset: Code cleanup and comment update
cgroup/cpuset: Don't allow creation of local partition over a remote one
cgroup/cpuset: Remove remote_partition_check() & make update_cpumasks_hier() handle remote partition
cgroup/cpuset: Fix error handling in remote_partition_disable()
cgroup/cpuset: Fix incorrect isolated_cpus update in update_parent_effective_cpumask()
Pull for-6.15-fixes to receive:
e776b26e37 ("sched_ext: Remove cpu.weight / cpu.idle unimplemented warnings")
which conflicts with:
1a7ff7216c ("sched_ext: Drop "ops" from scx_ops_enable_state and friends")
The former removes code updated by the latter. Resolved by removing the
updated section.
Signed-off-by: Tejun Heo <tj@kernel.org>
poll_state_synchronize_srcu() uses rcu_seq_done() unlike
poll_state_synchronize_rcu() which uses rcu_seq_done_exact().
The rcu_seq_done_exact() makes more sense for polling API, as with
this API, there is a higher chance that there is a significant delay
between the get_state..() and poll_state..() calls since a cookie
can be stored and reused at a later time. During such a delay, if
the gp_seq counter progresses more than ULONG_MAX/2 distance, then
poll_state..() may return false for a long time unwantedly.
Fix by using the more accurate rcu_seq_done_exact() API which is
exactly what straight RCU's polling does.
It may make sense, as future work, to add debug code here as well, where
we compare a physical timestamp between get_state..() and poll_state()
calls and yell if significant time has past but the grace period has
still not progressed.
Reviewed-by: Neeraj Upadhyay <Neeraj.Upadhyay@amd.com>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Reviewed-by: Kent Overstreet <kent.overstreet@linux.dev>
Cc: Kent Overstreet <kent.overstreet@linux.dev>
Signed-off-by: Joel Fernandes <joelagnelf@nvidia.com>
The exising code uses housekeeping_any_cpu() to select a cpu for
a given housekeeping task. However, this often ends up calling
cpumask_any_and() which is defined as cpumask_first_and() which has
the effect of alyways using the first cpu among those available.
The same applies when multiple NUMA nodes are involved. In that
case the first cpu in the local node is chosen which does provide
a bit of spreading but with multiple HK cpus per node the same
issues arise.
We have numerous cases where a single HK cpu just cannot keep up
and the remote_tick warning fires. It also can lead to the other
things (orchastration sw, HA keepalives etc) on the HK cpus getting
starved which leads to other issues. In these cases we recommend
increasing the number of HK cpus. But... that only helps the
userspace tasks somewhat. It does not help the actual housekeeping
part.
Spread the HK work out by having housekeeping_any_cpu() and
sched_numa_find_closest() use cpumask_any_and_distribute()
instead of cpumask_any_and().
Signed-off-by: Phil Auld <pauld@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Waiman Long <longman@redhat.com>
Reviewed-by: Vishal Chourasia <vishalc@linux.ibm.com>
Acked-by: Frederic Weisbecker <frederic@kernel.org>
Link: https://lore.kernel.org/r/20250218184618.1331715-1-pauld@redhat.com
Overview
========
When a CPU chooses to call push_rt_task and picks a task to push to
another CPU's runqueue then it will call find_lock_lowest_rq method
which would take a double lock on both CPUs' runqueues. If one of the
locks aren't readily available, it may lead to dropping the current
runqueue lock and reacquiring both the locks at once. During this window
it is possible that the task is already migrated and is running on some
other CPU. These cases are already handled. However, if the task is
migrated and has already been executed and another CPU is now trying to
wake it up (ttwu) such that it is queued again on the runqeue
(on_rq is 1) and also if the task was run by the same CPU, then the
current checks will pass even though the task was migrated out and is no
longer in the pushable tasks list.
Crashes
=======
This bug resulted in quite a few flavors of crashes triggering kernel
panics with various crash signatures such as assert failures, page
faults, null pointer dereferences, and queue corruption errors all
coming from scheduler itself.
Some of the crashes:
-> kernel BUG at kernel/sched/rt.c:1616! BUG_ON(idx >= MAX_RT_PRIO)
Call Trace:
? __die_body+0x1a/0x60
? die+0x2a/0x50
? do_trap+0x85/0x100
? pick_next_task_rt+0x6e/0x1d0
? do_error_trap+0x64/0xa0
? pick_next_task_rt+0x6e/0x1d0
? exc_invalid_op+0x4c/0x60
? pick_next_task_rt+0x6e/0x1d0
? asm_exc_invalid_op+0x12/0x20
? pick_next_task_rt+0x6e/0x1d0
__schedule+0x5cb/0x790
? update_ts_time_stats+0x55/0x70
schedule_idle+0x1e/0x40
do_idle+0x15e/0x200
cpu_startup_entry+0x19/0x20
start_secondary+0x117/0x160
secondary_startup_64_no_verify+0xb0/0xbb
-> BUG: kernel NULL pointer dereference, address: 00000000000000c0
Call Trace:
? __die_body+0x1a/0x60
? no_context+0x183/0x350
? __warn+0x8a/0xe0
? exc_page_fault+0x3d6/0x520
? asm_exc_page_fault+0x1e/0x30
? pick_next_task_rt+0xb5/0x1d0
? pick_next_task_rt+0x8c/0x1d0
__schedule+0x583/0x7e0
? update_ts_time_stats+0x55/0x70
schedule_idle+0x1e/0x40
do_idle+0x15e/0x200
cpu_startup_entry+0x19/0x20
start_secondary+0x117/0x160
secondary_startup_64_no_verify+0xb0/0xbb
-> BUG: unable to handle page fault for address: ffff9464daea5900
kernel BUG at kernel/sched/rt.c:1861! BUG_ON(rq->cpu != task_cpu(p))
-> kernel BUG at kernel/sched/rt.c:1055! BUG_ON(!rq->nr_running)
Call Trace:
? __die_body+0x1a/0x60
? die+0x2a/0x50
? do_trap+0x85/0x100
? dequeue_top_rt_rq+0xa2/0xb0
? do_error_trap+0x64/0xa0
? dequeue_top_rt_rq+0xa2/0xb0
? exc_invalid_op+0x4c/0x60
? dequeue_top_rt_rq+0xa2/0xb0
? asm_exc_invalid_op+0x12/0x20
? dequeue_top_rt_rq+0xa2/0xb0
dequeue_rt_entity+0x1f/0x70
dequeue_task_rt+0x2d/0x70
__schedule+0x1a8/0x7e0
? blk_finish_plug+0x25/0x40
schedule+0x3c/0xb0
futex_wait_queue_me+0xb6/0x120
futex_wait+0xd9/0x240
do_futex+0x344/0xa90
? get_mm_exe_file+0x30/0x60
? audit_exe_compare+0x58/0x70
? audit_filter_rules.constprop.26+0x65e/0x1220
__x64_sys_futex+0x148/0x1f0
do_syscall_64+0x30/0x80
entry_SYSCALL_64_after_hwframe+0x62/0xc7
-> BUG: unable to handle page fault for address: ffff8cf3608bc2c0
Call Trace:
? __die_body+0x1a/0x60
? no_context+0x183/0x350
? spurious_kernel_fault+0x171/0x1c0
? exc_page_fault+0x3b6/0x520
? plist_check_list+0x15/0x40
? plist_check_list+0x2e/0x40
? asm_exc_page_fault+0x1e/0x30
? _cond_resched+0x15/0x30
? futex_wait_queue_me+0xc8/0x120
? futex_wait+0xd9/0x240
? try_to_wake_up+0x1b8/0x490
? futex_wake+0x78/0x160
? do_futex+0xcd/0xa90
? plist_check_list+0x15/0x40
? plist_check_list+0x2e/0x40
? plist_del+0x6a/0xd0
? plist_check_list+0x15/0x40
? plist_check_list+0x2e/0x40
? dequeue_pushable_task+0x20/0x70
? __schedule+0x382/0x7e0
? asm_sysvec_reschedule_ipi+0xa/0x20
? schedule+0x3c/0xb0
? exit_to_user_mode_prepare+0x9e/0x150
? irqentry_exit_to_user_mode+0x5/0x30
? asm_sysvec_reschedule_ipi+0x12/0x20
Above are some of the common examples of the crashes that were observed
due to this issue.
Details
=======
Let's look at the following scenario to understand this race.
1) CPU A enters push_rt_task
a) CPU A has chosen next_task = task p.
b) CPU A calls find_lock_lowest_rq(Task p, CPU Z’s rq).
c) CPU A identifies CPU X as a destination CPU (X < Z).
d) CPU A enters double_lock_balance(CPU Z’s rq, CPU X’s rq).
e) Since X is lower than Z, CPU A unlocks CPU Z’s rq. Someone else has
locked CPU X’s rq, and thus, CPU A must wait.
2) At CPU Z
a) Previous task has completed execution and thus, CPU Z enters
schedule, locks its own rq after CPU A releases it.
b) CPU Z dequeues previous task and begins executing task p.
c) CPU Z unlocks its rq.
d) Task p yields the CPU (ex. by doing IO or waiting to acquire a
lock) which triggers the schedule function on CPU Z.
e) CPU Z enters schedule again, locks its own rq, and dequeues task p.
f) As part of dequeue, it sets p.on_rq = 0 and unlocks its rq.
3) At CPU B
a) CPU B enters try_to_wake_up with input task p.
b) Since CPU Z dequeued task p, p.on_rq = 0, and CPU B updates
B.state = WAKING.
c) CPU B via select_task_rq determines CPU Y as the target CPU.
4) The race
a) CPU A acquires CPU X’s lock and relocks CPU Z.
b) CPU A reads task p.cpu = Z and incorrectly concludes task p is
still on CPU Z.
c) CPU A failed to notice task p had been dequeued from CPU Z while
CPU A was waiting for locks in double_lock_balance. If CPU A knew
that task p had been dequeued, it would return NULL forcing
push_rt_task to give up the task p's migration.
d) CPU B updates task p.cpu = Y and calls ttwu_queue.
e) CPU B locks Ys rq. CPU B enqueues task p onto Y and sets task
p.on_rq = 1.
f) CPU B unlocks CPU Y, triggering memory synchronization.
g) CPU A reads task p.on_rq = 1, cementing its assumption that task p
has not migrated.
h) CPU A decides to migrate p to CPU X.
This leads to A dequeuing p from Y's queue and various crashes down the
line.
Solution
========
The solution here is fairly simple. After obtaining the lock (at 4a),
the check is enhanced to make sure that the task is still at the head of
the pushable tasks list. If not, then it is anyway not suitable for
being pushed out.
Testing
=======
The fix is tested on a cluster of 3 nodes, where the panics due to this
are hit every couple of days. A fix similar to this was deployed on such
cluster and was stable for more than 30 days.
Co-developed-by: Jon Kohler <jon@nutanix.com>
Signed-off-by: Jon Kohler <jon@nutanix.com>
Co-developed-by: Gauri Patwardhan <gauri.patwardhan@nutanix.com>
Signed-off-by: Gauri Patwardhan <gauri.patwardhan@nutanix.com>
Co-developed-by: Rahul Chunduru <rahul.chunduru@nutanix.com>
Signed-off-by: Rahul Chunduru <rahul.chunduru@nutanix.com>
Signed-off-by: Harshit Agarwal <harshit@nutanix.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: "Steven Rostedt (Google)" <rostedt@goodmis.org>
Reviewed-by: Phil Auld <pauld@redhat.com>
Tested-by: Will Ton <william.ton@nutanix.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20250225180553.167995-1-harshit@nutanix.com
The numbers used in rcu_seq_done_exact() lack some explanation behind
their magic. Especially after the commit:
85aad7cc41 ("rcu: Fix get_state_synchronize_rcu_full() GP-start detection")
which reported a subtle issue where a new GP sequence snapshot was taken
on the root node state while a grace period had already been started and
reflected on the global state sequence but not yet on the root node
sequence, making a polling user waiting on a wrong already started grace
period that would ignore freshly online CPUs.
The fix involved taking the snaphot on the global state sequence and
waiting on the root node sequence. And since a grace period is first
started on the global state and only afterward reflected on the root
node, a snapshot taken on the global state sequence might be two full
grace periods ahead of the root node as in the following example:
rnp->gp_seq = rcu_state.gp_seq = 0
CPU 0 CPU 1
----- -----
// rcu_state.gp_seq = 1
rcu_seq_start(&rcu_state.gp_seq)
// snap = 8
snap = rcu_seq_snap(&rcu_state.gp_seq)
// Two full GP differences
rcu_seq_done_exact(&rnp->gp_seq, snap)
// rnp->gp_seq = 1
WRITE_ONCE(rnp->gp_seq, rcu_state.gp_seq);
Add a comment about those expectations and to clarify the magic within
the relevant function.
Note that the issue arises mainly with the use of rcu_seq_done_exact()
which has a much tigher guardband (of 2 GPs) to ensure the false-negative
window of the API during wraparound is limited to just 2 GPs.
rcu_seq_done() does not have such strict requirements, however its large
false-negative window of ULONG_MAX/2 is not ideal for the polling API.
However, this also means care is needed to ensure the guardband is as
large as needed to avoid the example scenario describe above which a
warning added in an earlier patch does.
[ Comment wordsmithing by Joel ]
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Joel Fernandes <joelagnelf@nvidia.com>
With CONFIG_RT_GROUP_SCHED but runtime disabling of RT_GROUPs we expect
the existence of the root task_group only and all rt_sched_entity'ies
should be queued on root's rt_rq.
If we get a non-root RT_GROUP something went wrong.
Signed-off-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250310170442.504716-9-mkoutny@suse.com
The previous patch improved the rcu_seq_done_exact() function by adding
a meaningful constant for the guardband.
Ensure that this is working for the future by a quick check during
rcu_gp_init().
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Joel Fernandes <joelagnelf@nvidia.com>
Thanks to kernel cmdline being available early, before any
cgroup hierarchy exists, we can achieve the RT_GROUP_SCHED boottime
disabling goal by simply skipping any creation (and destruction) of
RT_GROUP data and its exposure via RT attributes.
We can do this thanks to previously placed runtime guards that would
redirect all operations to root_task_group's data when RT_GROUP_SCHED
disabled.
Signed-off-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250310170442.504716-8-mkoutny@suse.com
The rcu_seq_done_exact() function checks if a grace period has completed by
comparing sequence numbers. It includes a guard band to handle sequence number
wraparound, which was previously expressed using the magic number calculation
'3 * RCU_SEQ_STATE_MASK + 1'.
This magic number is not immediately obvious in terms of what it represents.
Instead, the reason we need this tiny guardband is because of the lag between
the setting of rcu_state.gp_seq_polled and root rnp's gp_seq in rcu_gp_init().
This guardband needs to be at least 2 GPs worth of counts, to avoid recognizing
the newly started GP as completed immediately, due to the following sequence
which arises due to the delay between update of rcu_state.gp_seq_polled and
root rnp's gp_seq:
rnp->gp_seq = rcu_state.gp_seq = 0
CPU 0 CPU 1
----- -----
// rcu_state.gp_seq = 1
rcu_seq_start(&rcu_state.gp_seq)
// snap = 8
snap = rcu_seq_snap(&rcu_state.gp_seq)
// Two full GP differences
rcu_seq_done_exact(&rnp->gp_seq, snap)
// rnp->gp_seq = 1
WRITE_ONCE(rnp->gp_seq, rcu_state.gp_seq);
This can happen due to get_state_synchronize_rcu_full() sampling
rcu_state.gp_seq_polled, however the poll_state_synchronize_rcu_full()
sampling the root rnp's gp_seq. The delay between the update of the 2
counters occurs in rcu_gp_init() during which the counters briefly go
out of sync.
Make the guardband explictly 2 GPs. This improves code readability and
maintainability by making the intent clearer as well.
Suggested-by: Frederic Weisbecker <frederic@kernel.org>
Reviewed-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Joel Fernandes <joelagnelf@nvidia.com>
When RT_GROUPs are compiled but not exposed, their bandwidth cannot
be configured (and it is not initialized for non-root task_groups neither).
Therefore bypass any checks of task vs task_group bandwidth.
This will achieve behavior very similar to setups that have
!CONFIG_RT_GROUP_SCHED and attach cpu controller to cgroup v2 hierarchy.
(On a related note, this may allow having RT tasks with
CONFIG_RT_GROUP_SCHED and cgroup v2 hierarchy.)
Signed-off-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250310170442.504716-7-mkoutny@suse.com
First, we want to prevent placement of RT tasks on non-root rt_rqs which
we achieve in the task migration code that'd fall back to
root_task_group's rt_rq.
Second, we want to work with only root_task_group's rt_rq when iterating
all "real" rt_rqs when RT_GROUP is disabled. To achieve this we keep
root_task_group as the first one on the task_groups and break out
quickly.
Signed-off-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250310170442.504716-6-mkoutny@suse.com
rt_rq->tg may be NULL which denotes the root task_group.
Store the pointer to root_task_group directly so that callers may use
rt_rq->tg homogenously.
root_task_group exists always with CONFIG_CGROUPS_SCHED,
CONFIG_RT_GROUP_SCHED depends on that.
This changes root level rt_rq's default limit from infinity to the
value of (originally) global RT throttling.
Signed-off-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250310170442.504716-4-mkoutny@suse.com
Convert the blocks guarded by macros to regular code so that the RT
group code gets more compile validation. Reasoning is in
Documentation/process/coding-style.rst 21) Conditional Compilation.
With that, no functional change is expected.
Signed-off-by: Michal Koutný <mkoutny@suse.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20250310170442.504716-2-mkoutny@suse.com
commit 10a35e6812 ("sched/pelt: Skip updating util_est when
utilization is higher than CPU's capacity")
prevents util_est from being updated if util_avg is higher than the
underlying CPU capacity to avoid overestimating the task when the CPU
is capped (due to thermal issue for instance). In this scenario, the
task will miss its deadlines and start overlapping its wake-up events
for instance. The task will appear as always running when the CPU is
just not powerful enough to allow having a good estimation of the
task.
commit b8c9636140 ("sched/fair/util_est: Implement faster ramp-up
EWMA on utilization increases")
sets ewma to util_avg when ewma > util_avg, allowing ewma to quickly
grow instead of slowly converge to the new util_avg value when a task
profile changes from small to big.
However, the 2 conditions:
- Check util_avg against max CPU capacity
- Check whether util_est > util_avg
are placed in an order such as it is possible to set util_est to a
value higher than the CPU capacity if util_est > util_avg, but
util_est is prevented to decay as long as:
CPU capacity < util_avg < util_est.
Just remove the check as either:
1.
There is idle time on the CPU. In that case the util_avg value of the
task is actually correct. It is possible that the task missed a
deadline and appears bigger, but this is also the case when the
util_avg of the task is lower than the maximum CPU capacity.
2.
There is no idle time. In that case, the util_avg value might aswell
be an under estimation of the size of the task.
It is possible that undesired frequency spikes will appear when the
task is later enqueued with an inflated util_est value, but the
frequency spike might aswell be deserved. The absence of idle time
prevents from drawing any conclusion.
Signed-off-by: Pierre Gondois <pierre.gondois@arm.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.rog>
Link: https://lore.kernel.org/r/20250325150542.1077344-1-pierre.gondois@arm.com
Use a different approach to topology_span_sane(), that checks for the
same constraint of no partial overlaps for any two CPU sets for
non-NUMA topology levels, but does so in a way that is O(N) rather
than O(N^2).
Instead of comparing with all other masks to detect collisions, keep
one mask that includes all CPUs seen so far and detect collisions with
a single cpumask_intersects test.
If the current mask has no collisions with previously seen masks, it
should be a new mask, which can be uniquely identified by the lowest
bit set in this mask. Keep a pointer to this mask for future
reference (in an array indexed by the lowest bit set), and add the
CPUs in this mask to the list of those seen.
If the current mask does collide with previously seen masks, it should
be exactly equal to a mask seen before, looked up in the same array
indexed by the lowest bit set in the mask, a single comparison.
Move the topology_span_sane() check out of the existing topology level
loop, let it use its own loop so that the array allocation can be done
only once, shared across levels.
On a system with 1920 processors (16 sockets, 60 cores, 2 threads),
the average time to take one processor offline is reduced from 2.18
seconds to 1.01 seconds. (Off-lining 959 of 1920 processors took
34m49.765s without this change, 16m10.038s with this change in place.)
Signed-off-by: Steve Wahl <steve.wahl@hpe.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Valentin Schneider <vschneid@redhat.com>
Reviewed-by: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Tested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Tested-by: Valentin Schneider <vschneid@redhat.com>
Tested-by: Madadi Vineeth Reddy <vineethr@linux.ibm.com>
Link: https://lore.kernel.org/r/20250304160844.75373-2-steve.wahl@hpe.com
Gabriele noted that in case of signal_pending_state(), the tracepoint
sees a stale task-state.
Fixes: fa2c3254d7 ("sched/tracing: Don't re-read p->state when emitting sched_switch event")
Reported-by: Gabriele Monaco <gmonaco@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: Valentin Schneider <vschneid@redhat.com>
Previously it was only safe to call perf_pmu_unregister() if there
were no active events of that pmu around -- which was impossible to
guarantee since it races all sorts against perf_init_event().
Rework the whole thing by:
- keeping track of all events for a given pmu
- 'hiding' the pmu from perf_init_event()
- waiting for the appropriate (s)rcu grace periods such that all
prior references to the PMU will be completed
- detaching all still existing events of that pmu (see first point)
and moving them to a new REVOKED state.
- actually freeing the pmu data.
Where notably the new REVOKED state must inhibit all event actions
from reaching code that wants to use event->pmu.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Ravi Bangoria <ravi.bangoria@amd.com>
Link: https://lkml.kernel.org/r/20250307193723.525402029@infradead.org
The task passed to perf_event_exit_task() is not a child, it is
current. Fix this confusing naming, since much of the rest of the code
also relies on it being current.
Specifically, both exec() and exit() callers use it with current as
the argument.
Notably, task_ctx_sched_out() doesn't make much sense outside of
current.
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Ravi Bangoria <ravi.bangoria@amd.com>
Link: https://lkml.kernel.org/r/20250307193305.486326750@infradead.org
