When wake_up_new_task(), we use post_init_entity_util_avg() to init
util_avg/runnable_avg based on cpu's util_avg at that time, and
attach task sched_avg to cfs_rq.
Since enqueue_task_fair() -> enqueue_entity() -> update_load_avg()
loop will do attach, we can move this work to update_load_avg().
wake_up_new_task(p)
post_init_entity_util_avg(p)
attach_entity_cfs_rq() --> (1)
activate_task(rq, p)
enqueue_task() := enqueue_task_fair()
enqueue_entity() loop
update_load_avg(cfs_rq, se, UPDATE_TG | DO_ATTACH)
if (!se->avg.last_update_time && (flags & DO_ATTACH))
attach_entity_load_avg() --> (2)
This patch move attach from (1) to (2), update related comments too.
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lore.kernel.org/r/20220818124805.601-9-zhouchengming@bytedance.com
commit 7dc603c902 ("sched/fair: Fix PELT integrity for new tasks")
introduce a TASK_NEW state and an unnessary limitation that would fail
when changing cgroup of new forked task.
Because at that time, we can't handle task_change_group_fair() for new
forked fair task which hasn't been woken up by wake_up_new_task(),
which will cause detach on an unattached task sched_avg problem.
This patch delete this unnessary limitation by adding check before do
detach or attach in task_change_group_fair().
So cpu_cgrp_subsys.can_attach() has nothing to do for fair tasks,
only define it in #ifdef CONFIG_RT_GROUP_SCHED.
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lore.kernel.org/r/20220818124805.601-8-zhouchengming@bytedance.com
commit 7dc603c902 ("sched/fair: Fix PELT integrity for new tasks")
fixed two load tracking problems for new task, including detach on
unattached new task problem.
There still left another detach on unattached task problem for the task
which has been woken up by try_to_wake_up() and waiting for actually
being woken up by sched_ttwu_pending().
try_to_wake_up(p)
cpu = select_task_rq(p)
if (task_cpu(p) != cpu)
set_task_cpu(p, cpu)
migrate_task_rq_fair()
remove_entity_load_avg() --> unattached
se->avg.last_update_time = 0;
__set_task_cpu()
ttwu_queue(p, cpu)
ttwu_queue_wakelist()
__ttwu_queue_wakelist()
task_change_group_fair()
detach_task_cfs_rq()
detach_entity_cfs_rq()
detach_entity_load_avg() --> detach on unattached task
set_task_rq()
attach_task_cfs_rq()
attach_entity_cfs_rq()
attach_entity_load_avg()
The reason of this problem is similar, we should check in detach_entity_cfs_rq()
that se->avg.last_update_time != 0, before do detach_entity_load_avg().
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lore.kernel.org/r/20220818124805.601-7-zhouchengming@bytedance.com
When we are migrating task out of the CPU, we can combine detach and
propagation into dequeue_entity() to save the detach_entity_cfs_rq()
in migrate_task_rq_fair().
This optimization is like combining DO_ATTACH in the enqueue_entity()
when migrating task to the CPU. So we don't have to traverse the CFS tree
extra time to do the detach_entity_cfs_rq() -> propagate_entity_cfs_rq(),
which wouldn't be called anymore with this patch's change.
detach_task()
deactivate_task()
dequeue_task_fair()
for_each_sched_entity(se)
dequeue_entity()
update_load_avg() /* (1) */
detach_entity_load_avg()
set_task_cpu()
migrate_task_rq_fair()
detach_entity_cfs_rq() /* (2) */
update_load_avg();
detach_entity_load_avg();
propagate_entity_cfs_rq();
for_each_sched_entity()
update_load_avg()
This patch save the detach_entity_cfs_rq() called in (2) by doing
the detach_entity_load_avg() for a CPU migrating task inside (1)
(the task being the first se in the loop)
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lore.kernel.org/r/20220818124805.601-6-zhouchengming@bytedance.com
When reading the sched_avg related code, I found the comments in
enqueue/dequeue_entity() are not updated with the current code.
We don't add/subtract entity's runnable_avg from cfs_rq->runnable_avg
during enqueue/dequeue_entity(), those are done only for attach/detach.
This patch updates the comments to reflect the current code working.
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lore.kernel.org/r/20220818124805.601-5-zhouchengming@bytedance.com
We use cpu_cgrp_subsys->fork() to set task group for the new fair task
in cgroup_post_fork().
Since commit b1e8206582 ("sched: Fix yet more sched_fork() races")
has already set_task_rq() for the new fair task in sched_cgroup_fork(),
so cpu_cgrp_subsys->fork() can be removed.
cgroup_can_fork() --> pin parent's sched_task_group
sched_cgroup_fork()
__set_task_cpu()
set_task_rq()
cgroup_post_fork()
ss->fork() := cpu_cgroup_fork()
sched_change_group(..., TASK_SET_GROUP)
task_set_group_fair()
set_task_rq() --> can be removed
After this patch's change, task_change_group_fair() only need to
care about task cgroup migration, make the code much simplier.
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Link: https://lore.kernel.org/r/20220818124805.601-3-zhouchengming@bytedance.com
Previously we only maintain task se depth in task_move_group_fair(),
if a !fair task change task group, its se depth will not be updated,
so commit eb7a59b2c8 ("sched/fair: Reset se-depth when task switched to FAIR")
fix the problem by updating se depth in switched_to_fair() too.
Then commit daa59407b5 ("sched/fair: Unify switched_{from,to}_fair()
and task_move_group_fair()") unified these two functions, moved se.depth
setting to attach_task_cfs_rq(), which further into attach_entity_cfs_rq()
with commit df217913e7 ("sched/fair: Factorize attach/detach entity").
This patch move task se depth maintenance from attach_entity_cfs_rq()
to set_task_rq(), which will be called when CPU/cgroup change, so its
depth will always be correct.
This patch is preparation for the next patch.
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lore.kernel.org/r/20220818124805.601-2-zhouchengming@bytedance.com
The load_balance_mask and select_rq_mask percpu variables are only used in
kernel/sched/fair.c.
Make them static and move their allocation into init_sched_fair_class().
Replace kzalloc_node() with zalloc_cpumask_var_node() to get rid of the
CONFIG_CPUMASK_OFFSTACK #ifdef and to align with per-cpu cpumask
allocation for RT (local_cpu_mask in init_sched_rt_class()) and DL
class (local_cpu_mask_dl in init_sched_dl_class()).
[ mingo: Tidied up changelog & touched up the code. ]
Signed-off-by: Bing Huang <huangbing@kylinos.cn>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Link: https://lore.kernel.org/r/20220722213609.3901-1-huangbing775@126.com
Currently, the values of some fields are printed right-aligned, causing
the field value to be next to the next field name rather than next to its
own field name. So print each field value left-aligned, to make it more
readable.
Before:
stack: 0 pid: 307 ppid: 2 flags:0x00000008
After:
stack:0 pid:308 ppid:2 flags:0x0000000a
This also makes them print in the same style as the other two fields:
task:demo0 state:R running task
Signed-off-by: Zhen Lei <thunder.leizhen@huawei.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Valentin Schneider <vschneid@redhat.com>
Link: https://lore.kernel.org/r/20220727060819.1085-1-thunder.leizhen@huawei.com
Save a multiplication in dl_task_fits_capacity() by using already
maintained per-sched_dl_entity (i.e. per-task) `dl_runtime/dl_deadline`
(dl_density).
cap_scale(dl_deadline, cap) >= dl_runtime
dl_deadline * cap >> SCHED_CAPACITY_SHIFT >= dl_runtime
cap >= dl_runtime << SCHED_CAPACITY_SHIFT / dl_deadline
cap >= (dl_runtime << BW_SHIFT / dl_deadline) >>
BW_SHIFT - SCHED_CAPACITY_SHIFT
cap >= dl_density >> BW_SHIFT - SCHED_CAPACITY_SHIFT
__sched_setscheduler()->__checkparam_dl() ensures that the 2 corner
cases (if conditions) `runtime == RUNTIME_INF (-1)` and `period == 0`
of to_ratio(deadline, runtime) are not met when setting dl_density in
__sched_setscheduler()-> __setscheduler_params()->__setparam_dl().
Signed-off-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20220729111305.1275158-4-dietmar.eggemann@arm.com
dl_cpuset_cpumask_can_shrink() is used to validate whether there is
still enough CPU capacity for DL tasks in the reduced cpuset.
Currently it still operates on `# remaining CPUs in the cpuset` (1).
Change this to use the already capacity-aware DL admission control
__dl_overflow() for the `cpumask can shrink` test.
dl_b->bw = sched_rt_period << BW_SHIFT / sched_rt_period
dl_b->bw * (1) >= currently allocated bandwidth in root_domain (rd)
Replace (1) w/ `\Sum CPU capacity in rd >> SCHED_CAPACITY_SHIFT`
Adapt __dl_bw_capacity() to take a cpumask instead of a CPU number
argument so that `rd->span` and `cpumask of the reduced cpuset` can
be used here.
Signed-off-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lore.kernel.org/r/20220729111305.1275158-3-dietmar.eggemann@arm.com
Pull irq updates from Thomas Gleixner:
"Updates for interrupt core and drivers:
Core:
- Fix a few inconsistencies between UP and SMP vs interrupt
affinities
- Small updates and cleanups all over the place
New drivers:
- LoongArch interrupt controller
- Renesas RZ/G2L interrupt controller
Updates:
- Hotpath optimization for SiFive PLIC
- Workaround for broken PLIC edge triggered interrupts
- Simall cleanups and improvements as usual"
* tag 'irq-core-2022-08-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (52 commits)
irqchip/mmp: Declare init functions in common header file
irqchip/mips-gic: Check the return value of ioremap() in gic_of_init()
genirq: Use for_each_action_of_desc in actions_show()
irqchip / ACPI: Introduce ACPI_IRQ_MODEL_LPIC for LoongArch
irqchip: Add LoongArch CPU interrupt controller support
irqchip: Add Loongson Extended I/O interrupt controller support
irqchip/loongson-liointc: Add ACPI init support
irqchip/loongson-pch-msi: Add ACPI init support
irqchip/loongson-pch-pic: Add ACPI init support
irqchip: Add Loongson PCH LPC controller support
LoongArch: Prepare to support multiple pch-pic and pch-msi irqdomain
LoongArch: Use ACPI_GENERIC_GSI for gsi handling
genirq/generic_chip: Export irq_unmap_generic_chip
ACPI: irq: Allow acpi_gsi_to_irq() to have an arch-specific fallback
APCI: irq: Add support for multiple GSI domains
LoongArch: Provisionally add ACPICA data structures
irqdomain: Use hwirq_max instead of revmap_size for NOMAP domains
irqdomain: Report irq number for NOMAP domains
irqchip/gic-v3: Fix comment typo
dt-bindings: interrupt-controller: renesas,rzg2l-irqc: Document RZ/V2L SoC
...
Pull perf events updates from Ingo Molnar:
- Fix Intel Alder Lake PEBS memory access latency & data source
profiling info bugs.
- Use Intel large-PEBS hardware feature in more circumstances, to
reduce PMI overhead & reduce sampling data.
- Extend the lost-sample profiling output with the PERF_FORMAT_LOST ABI
variant, which tells tooling the exact number of samples lost.
- Add new IBS register bits definitions.
- AMD uncore events: Add PerfMonV2 DF (Data Fabric) enhancements.
* tag 'perf-core-2022-08-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/x86/ibs: Add new IBS register bits into header
perf/x86/intel: Fix PEBS data source encoding for ADL
perf/x86/intel: Fix PEBS memory access info encoding for ADL
perf/core: Add a new read format to get a number of lost samples
perf/x86/amd/uncore: Add PerfMonV2 RDPMC assignments
perf/x86/amd/uncore: Add PerfMonV2 DF event format
perf/x86/amd/uncore: Detect available DF counters
perf/x86/amd/uncore: Use attr_update for format attributes
perf/x86/amd/uncore: Use dynamic events array
x86/events/intel/ds: Enable large PEBS for PERF_SAMPLE_WEIGHT_TYPE
Pull locking updates from Ingo Molnar:
"This was a fairly quiet cycle for the locking subsystem:
- lockdep: Fix a handful of the more complex lockdep_init_map_*()
primitives that can lose the lock_type & cause false reports. No
such mishap was observed in the wild.
- jump_label improvements: simplify the cross-arch support of initial
NOP patching by making it arch-specific code (used on MIPS only),
and remove the s390 initial NOP patching that was superfluous"
* tag 'locking-core-2022-08-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
locking/lockdep: Fix lockdep_init_map_*() confusion
jump_label: make initial NOP patching the special case
jump_label: mips: move module NOP patching into arch code
jump_label: s390: avoid pointless initial NOP patching
Pull scheduler updates from Ingo Molnar:
"Load-balancing improvements:
- Improve NUMA balancing on AMD Zen systems for affine workloads.
- Improve the handling of reduced-capacity CPUs in load-balancing.
- Energy Model improvements: fix & refine all the energy fairness
metrics (PELT), and remove the conservative threshold requiring 6%
energy savings to migrate a task. Doing this improves power
efficiency for most workloads, and also increases the reliability
of energy-efficiency scheduling.
- Optimize/tweak select_idle_cpu() to spend (much) less time
searching for an idle CPU on overloaded systems. There's reports of
several milliseconds spent there on large systems with large
workloads ...
[ Since the search logic changed, there might be behavioral side
effects. ]
- Improve NUMA imbalance behavior. On certain systems with spare
capacity, initial placement of tasks is non-deterministic, and such
an artificial placement imbalance can persist for a long time,
hurting (and sometimes helping) performance.
The fix is to make fork-time task placement consistent with runtime
NUMA balancing placement.
Note that some performance regressions were reported against this,
caused by workloads that are not memory bandwith limited, which
benefit from the artificial locality of the placement bug(s). Mel
Gorman's conclusion, with which we concur, was that consistency is
better than random workload benefits from non-deterministic bugs:
"Given there is no crystal ball and it's a tradeoff, I think
it's better to be consistent and use similar logic at both fork
time and runtime even if it doesn't have universal benefit."
- Improve core scheduling by fixing a bug in
sched_core_update_cookie() that caused unnecessary forced idling.
- Improve wakeup-balancing by allowing same-LLC wakeup of idle CPUs
for newly woken tasks.
- Fix a newidle balancing bug that introduced unnecessary wakeup
latencies.
ABI improvements/fixes:
- Do not check capabilities and do not issue capability check denial
messages when a scheduler syscall doesn't require privileges. (Such
as increasing niceness.)
- Add forced-idle accounting to cgroups too.
- Fix/improve the RSEQ ABI to not just silently accept unknown flags.
(No existing tooling is known to have learned to rely on the
previous behavior.)
- Depreciate the (unused) RSEQ_CS_FLAG_NO_RESTART_ON_* flags.
Optimizations:
- Optimize & simplify leaf_cfs_rq_list()
- Micro-optimize set_nr_{and_not,if}_polling() via try_cmpxchg().
Misc fixes & cleanups:
- Fix the RSEQ self-tests on RISC-V and Glibc 2.35 systems.
- Fix a full-NOHZ bug that can in some cases result in the tick not
being re-enabled when the last SCHED_RT task is gone from a
runqueue but there's still SCHED_OTHER tasks around.
- Various PREEMPT_RT related fixes.
- Misc cleanups & smaller fixes"
* tag 'sched-core-2022-08-01' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (32 commits)
rseq: Kill process when unknown flags are encountered in ABI structures
rseq: Deprecate RSEQ_CS_FLAG_NO_RESTART_ON_* flags
sched/core: Fix the bug that task won't enqueue into core tree when update cookie
nohz/full, sched/rt: Fix missed tick-reenabling bug in dequeue_task_rt()
sched/core: Always flush pending blk_plug
sched/fair: fix case with reduced capacity CPU
sched/core: Use try_cmpxchg in set_nr_{and_not,if}_polling
sched/core: add forced idle accounting for cgroups
sched/fair: Remove the energy margin in feec()
sched/fair: Remove task_util from effective utilization in feec()
sched/fair: Use the same cpumask per-PD throughout find_energy_efficient_cpu()
sched/fair: Rename select_idle_mask to select_rq_mask
sched, drivers: Remove max param from effective_cpu_util()/sched_cpu_util()
sched/fair: Decay task PELT values during wakeup migration
sched/fair: Provide u64 read for 32-bits arch helper
sched/fair: Introduce SIS_UTIL to search idle CPU based on sum of util_avg
sched: only perform capability check on privileged operation
sched: Remove unused function group_first_cpu()
sched/fair: Remove redundant word " *"
selftests/rseq: check if libc rseq support is registered
...
rseq_abi()->flags and rseq_abi()->rseq_cs->flags 29 upper bits are
currently unused.
The current behavior when those bits are set is to ignore them. This is
not an ideal behavior, because when future features will start using
those flags, if user-space fails to correctly validate that the kernel
indeed supports those flags (e.g. with a new sys_rseq flags bit) before
using them, it may incorrectly assume that the kernel will handle those
flags way when in fact those will be silently ignored on older kernels.
Validating that unused flags bits are cleared will allow a smoother
transition when those flags will start to be used by allowing
applications to fail early, and obviously, when they attempt to use the
new flags on an older kernel that does not support them.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Link: https://lkml.kernel.org/r/20220622194617.1155957-2-mathieu.desnoyers@efficios.com
Pull x86 fixes from Borislav Petkov:
- Update the 'mitigations=' kernel param documentation
- Check the IBPB feature flag before enabling IBPB in firmware calls
because cloud vendors' fantasy when it comes to creating guest
configurations is unlimited
- Unexport sev_es_ghcb_hv_call() before 5.19 releases now that HyperV
doesn't need it anymore
- Remove dead CONFIG_* items
* tag 'x86_urgent_for_v5.19' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
docs/kernel-parameters: Update descriptions for "mitigations=" param with retbleed
x86/bugs: Do not enable IBPB at firmware entry when IBPB is not available
Revert "x86/sev: Expose sev_es_ghcb_hv_call() for use by HyperV"
x86/configs: Update configs in x86_debug.config
Pull locking fix from Borislav Petkov:
- Avoid rwsem lockups in certain situations when handling the handoff
bit
* tag 'locking_urgent_for_v5.19' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
locking/rwsem: Allow slowpath writer to ignore handoff bit if not set by first waiter
With commit d257cc8cb8 ("locking/rwsem: Make handoff bit handling more
consistent"), the writer that sets the handoff bit can be interrupted
out without clearing the bit if the wait queue isn't empty. This disables
reader and writer optimistic lock spinning and stealing.
Now if a non-first writer in the queue is somehow woken up or a new
waiter enters the slowpath, it can't acquire the lock. This is not the
case before commit d257cc8cb8 as the writer that set the handoff bit
will clear it when exiting out via the out_nolock path. This is less
efficient as the busy rwsem stays in an unlock state for a longer time.
In some cases, this new behavior may cause lockups as shown in [1] and
[2].
This patch allows a non-first writer to ignore the handoff bit if it
is not originally set or initiated by the first waiter. This patch is
shown to be effective in fixing the lockup problem reported in [1].
[1] https://lore.kernel.org/lkml/20220617134325.GC30825@techsingularity.net/
[2] https://lore.kernel.org/lkml/3f02975c-1a9d-be20-32cf-f1d8e3dfafcc@oracle.com/
Fixes: d257cc8cb8 ("locking/rwsem: Make handoff bit handling more consistent")
Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: John Donnelly <john.p.donnelly@oracle.com>
Tested-by: Mel Gorman <mgorman@techsingularity.net>
Link: https://lore.kernel.org/r/20220622200419.778799-1-longman@redhat.com
Pull workqueue fix from Tejun Heo:
"Just one commit to suppress a spurious warning added during the 5.19
cycle"
* tag 'wq-for-5.19-rc8-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/tj/wq:
workqueue: Avoid a false warning in unbind_workers()
Doing set_cpus_allowed_ptr() with wq_unbound_cpumask can be possible
fails and trigger the false warning.
Use cpu_possible_mask instead when wq_unbound_cpumask has no active CPUs.
It is very easy to trigger the warning:
Set wq_unbound_cpumask to a small set of CPUs.
Offline all the CPUs of wq_unbound_cpumask.
Offline an extra CPU and trigger the warning.
Fixes: 10a5a651e3 ("workqueue: Restrict kworker in the offline CPU pool running on housekeeping CPUs")
Signed-off-by: Lai Jiangshan <jiangshan.ljs@antgroup.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
If a watch is being added to a queue, it needs to guard against
interference from addition of a new watch, manual removal of a watch and
removal of a watch due to some other queue being destroyed.
KEYCTL_WATCH_KEY guards against this for the same {key,queue} pair by
holding the key->sem writelocked and by holding refs on both the key and
the queue - but that doesn't prevent interaction from other {key,queue}
pairs.
While add_watch_to_object() does take the spinlock on the event queue,
it doesn't take the lock on the source's watch list. The assumption was
that the caller would prevent that (say by taking key->sem) - but that
doesn't prevent interference from the destruction of another queue.
Fix this by locking the watcher list in add_watch_to_object().
Fixes: c73be61ced ("pipe: Add general notification queue support")
Reported-by: syzbot+03d7b43290037d1f87ca@syzkaller.appspotmail.com
Signed-off-by: David Howells <dhowells@redhat.com>
cc: keyrings@vger.kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Since __post_watch_notification() walks wlist->watchers with only the
RCU read lock held, we need to use RCU methods to add to the list (we
already use RCU methods to remove from the list).
Fix add_watch_to_object() to use hlist_add_head_rcu() instead of
hlist_add_head() for that list.
Fixes: c73be61ced ("pipe: Add general notification queue support")
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull irqchip/genirq updates from Marc Zyngier:
* Core code update:
- Non-SMP IRQ affinity fixes, allowing UP kernel to behave similarly
to SMP ones for the purpose of interrupt affinity
- Let irq_set_chip_handler_name_locked() take a const struct irq_chip *
- Tidy-up the NOMAP irqdomain API variant
- Teach action_show() to use for_each_action_of_desc()
- Make irq_chip_request_resources_parent() allow the parent callback
to be optional
- Remove dynamic allocations from populate_parent_alloc_arg()
* New drivers:
- Merge the long awaited IRQ support for the LoongArch architecture,
with the provisional ACPICA update (to be reverted once the official
support lands)
- New Renesas RZ/G2L IRQC driver, equipped with its companion GPIO
driver
* Driver updates
- Optimise the hot path operations for the SiFive PLIC, trading the
locking for per-CPU priority masking masking operations which are
apparently faster
- Work around broken PLIC implementations that deal pretty badly with
edge-triggered interrupts. Flag two implementations as affected.
- Simplify the irq-stm32-exti driver, particularly the table that
remaps the interrupts from exti to the GIC, reducing the memory usage
- Convert the ocelot irq_chip to being immutable
- Check ioremap() return value in the MIPS GIC driver
- Move MMP driver init function declarations into the common .h
- The obligatory typo fixes
Link: https://lore.kernel.org/all/20220727192356.1860546-1-maz@kernel.org
* irq/misc-5.20:
: .
: Misc IRQ changes for 5.20:
:
: - Let irq_set_chip_handler_name_locked() take a const struct irq_chip *
:
: - Convert the ocelot irq_chip to being immutable (depends on the above)
:
: - Tidy-up the NOMAP irqdomain API variant
:
: - Teach action_show() to use for_each_action_of_desc()
:
: - Check ioremap() return value in the MIPS GIC driver
:
: - Move MMP driver init function declarations into the common .h
:
: - The obligatory typo fixes
: .
irqchip/mmp: Declare init functions in common header file
irqchip/mips-gic: Check the return value of ioremap() in gic_of_init()
genirq: Use for_each_action_of_desc in actions_show()
irqdomain: Use hwirq_max instead of revmap_size for NOMAP domains
irqdomain: Report irq number for NOMAP domains
irqchip/gic-v3: Fix comment typo
pinctrl: ocelot: Make irq_chip immutable
genirq: Allow irq_set_chip_handler_name_locked() to take a const irq_chip
Signed-off-by: Marc Zyngier <maz@kernel.org>
Pull scheduler fix from Borislav Petkov:
"A single fix to correct a wrong BUG_ON() condition for deboosted
tasks"
* tag 'sched_urgent_for_v5.19_rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
sched/deadline: Fix BUG_ON condition for deboosted tasks
Pull RCU fix from Paul McKenney:
"This contains a pair of commits that fix 282d8998e9 ("srcu: Prevent
expedited GPs and blocking readers from consuming CPU"), which was
itself a fix to an SRCU expedited grace-period problem that could
prevent kernel live patching (KLP) from completing.
That SRCU fix for KLP introduced large (as in minutes) boot-time
delays to embedded Linux kernels running on qemu/KVM. These delays
were due to the emulation of certain MMIO operations controlling
memory layout, which were emulated with one expedited grace period per
access. Common configurations required thousands of boot-time MMIO
accesses, and thus thousands of boot-time expedited SRCU grace
periods.
In these configurations, the occasional sleeps that allowed KLP to
proceed caused excessive boot delays. These commits preserve enough
sleeps to permit KLP to proceed, but few enough that the virtual
embedded kernels still boot reasonably quickly.
This represents a regression introduced in the v5.19 merge window, and
the bug is causing significant inconvenience"
* tag 'rcu-urgent.2022.07.21a' of git://git.kernel.org/pub/scm/linux/kernel/git/paulmck/linux-rcu:
srcu: Make expedited RCU grace periods block even less frequently
srcu: Block less aggressively for expedited grace periods
Sedat Dilek noticed that I had an extraneous semicolon at the end of a
line in the previous patch.
It's harmless, but unintentional, and while compilers just treat it as
an extra empty statement, for all I know some other tooling might warn
about it. So clean it up before other people notice too ;)
Fixes: 353f7988dd ("watchqueue: make sure to serialize 'wqueue->defunct' properly")
Reported-by: Sedat Dilek <sedat.dilek@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Reported-by: Sedat Dilek <sedat.dilek@gmail.com>
In function sched_core_update_cookie(), a task will enqueue into the
core tree only when it enqueued before, that is, if an uncookied task
is cookied, it will not enqueue into the core tree until it enqueue
again, which will result in unnecessary force idle.
Here follows the scenario:
CPU x and CPU y are a pair of SMT siblings.
1. Start task a running on CPU x without sleeping, and task b and
task c running on CPU y without sleeping.
2. We create a cookie and share it to task a and task b, and then
we create another cookie and share it to task c.
3. Simpling core_forceidle_sum of task a and b from /proc/PID/sched
And we will find out that core_forceidle_sum of task a takes 30%
time of the sampling period, which shouldn't happen as task a and b
have the same cookie.
Then we migrate task a to CPU x', migrate task b and c to CPU y', where
CPU x' and CPU y' are a pair of SMT siblings, and sampling again, we
will found out that core_forceidle_sum of task a and b are almost zero.
To solve this problem, we enqueue the task into the core tree if it's
on rq.
Fixes: 6e33cad0af49("sched: Trivial core scheduling cookie management")
Signed-off-by: Cruz Zhao <CruzZhao@linux.alibaba.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/1656403045-100840-2-git-send-email-CruzZhao@linux.alibaba.com
dequeue_task_rt() only decrements 'rt_rq->rt_nr_running' after having
called sched_update_tick_dependency() preventing it from re-enabling the
tick on systems that no longer have pending SCHED_RT tasks but have
multiple runnable SCHED_OTHER tasks:
dequeue_task_rt()
dequeue_rt_entity()
dequeue_rt_stack()
dequeue_top_rt_rq()
sub_nr_running() // decrements rq->nr_running
sched_update_tick_dependency()
sched_can_stop_tick() // checks rq->rt.rt_nr_running,
...
__dequeue_rt_entity()
dec_rt_tasks() // decrements rq->rt.rt_nr_running
...
Every other scheduler class performs the operation in the opposite
order, and sched_update_tick_dependency() expects the values to be
updated as such. So avoid the misbehaviour by inverting the order in
which the above operations are performed in the RT scheduler.
Fixes: 76d92ac305 ("sched: Migrate sched to use new tick dependency mask model")
Signed-off-by: Nicolas Saenz Julienne <nsaenzju@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Valentin Schneider <vschneid@redhat.com>
Reviewed-by: Phil Auld <pauld@redhat.com>
Link: https://lore.kernel.org/r/20220628092259.330171-1-nsaenzju@redhat.com
Tasks the are being deboosted from SCHED_DEADLINE might enter
enqueue_task_dl() one last time and hit an erroneous BUG_ON condition:
since they are not boosted anymore, the if (is_dl_boosted()) branch is
not taken, but the else if (!dl_prio) is and inside this one we
BUG_ON(!is_dl_boosted), which is of course false (BUG_ON triggered)
otherwise we had entered the if branch above. Long story short, the
current condition doesn't make sense and always leads to triggering of a
BUG.
Fix this by only checking enqueue flags, properly: ENQUEUE_REPLENISH has
to be present, but additional flags are not a problem.
Fixes: 64be6f1f5f ("sched/deadline: Don't replenish from a !SCHED_DEADLINE entity")
Signed-off-by: Juri Lelli <juri.lelli@redhat.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20220714151908.533052-1-juri.lelli@redhat.com
When the pipe is closed, we mark the associated watchqueue defunct by
calling watch_queue_clear(). However, while that is protected by the
watchqueue lock, new watchqueue entries aren't actually added under that
lock at all: they use the pipe->rd_wait.lock instead, and looking up
that pipe happens without any locking.
The watchqueue code uses the RCU read-side section to make sure that the
wqueue entry itself hasn't disappeared, but that does not protect the
pipe_info in any way.
So make sure to actually hold the wqueue lock when posting watch events,
properly serializing against the pipe being torn down.
Reported-by: Noam Rathaus <noamr@ssd-disclosure.com>
Cc: Greg KH <gregkh@linuxfoundation.org>
Cc: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* irq/loongarch:
: .
: Merge the long awaited IRQ support for the LoongArch architecture.
:
: From the cover letter:
:
: "Currently, LoongArch based processors (e.g. Loongson-3A5000)
: can only work together with LS7A chipsets. The irq chips in
: LoongArch computers include CPUINTC (CPU Core Interrupt
: Controller), LIOINTC (Legacy I/O Interrupt Controller),
: EIOINTC (Extended I/O Interrupt Controller), PCH-PIC (Main
: Interrupt Controller in LS7A chipset), PCH-LPC (LPC Interrupt
: Controller in LS7A chipset) and PCH-MSI (MSI Interrupt Controller)."
:
: Note that this comes with non-official, arch private ACPICA
: definitions until the official ACPICA update is realeased.
: .
irqchip / ACPI: Introduce ACPI_IRQ_MODEL_LPIC for LoongArch
irqchip: Add LoongArch CPU interrupt controller support
irqchip: Add Loongson Extended I/O interrupt controller support
irqchip/loongson-liointc: Add ACPI init support
irqchip/loongson-pch-msi: Add ACPI init support
irqchip/loongson-pch-pic: Add ACPI init support
irqchip: Add Loongson PCH LPC controller support
LoongArch: Prepare to support multiple pch-pic and pch-msi irqdomain
LoongArch: Use ACPI_GENERIC_GSI for gsi handling
genirq/generic_chip: Export irq_unmap_generic_chip
ACPI: irq: Allow acpi_gsi_to_irq() to have an arch-specific fallback
APCI: irq: Add support for multiple GSI domains
LoongArch: Provisionally add ACPICA data structures
Signed-off-by: Marc Zyngier <maz@kernel.org>
The purpose of commit 282d8998e9 ("srcu: Prevent expedited GPs
and blocking readers from consuming CPU") was to prevent a long
series of never-blocking expedited SRCU grace periods from blocking
kernel-live-patching (KLP) progress. Although it was successful, it also
resulted in excessive boot times on certain embedded workloads running
under qemu with the "-bios QEMU_EFI.fd" command line. Here "excessive"
means increasing the boot time up into the three-to-four minute range.
This increase in boot time was due to the more than 6000 back-to-back
invocations of synchronize_rcu_expedited() within the KVM host OS, which
in turn resulted from qemu's emulation of a long series of MMIO accesses.
Commit 640a7d37c3f4 ("srcu: Block less aggressively for expedited grace
periods") did not significantly help this particular use case.
Zhangfei Gao and Shameerali Kolothum Thodi did experiments varying the
value of SRCU_MAX_NODELAY_PHASE with HZ=250 and with various values
of non-sleeping per phase counts on a system with preemption enabled,
and observed the following boot times:
+──────────────────────────+────────────────+
| SRCU_MAX_NODELAY_PHASE | Boot time (s) |
+──────────────────────────+────────────────+
| 100 | 30.053 |
| 150 | 25.151 |
| 200 | 20.704 |
| 250 | 15.748 |
| 500 | 11.401 |
| 1000 | 11.443 |
| 10000 | 11.258 |
| 1000000 | 11.154 |
+──────────────────────────+────────────────+
Analysis on the experiment results show additional improvements with
CPU-bound delays approaching one jiffy in duration. This improvement was
also seen when number of per-phase iterations were scaled to one jiffy.
This commit therefore scales per-grace-period phase number of non-sleeping
polls so that non-sleeping polls extend for about one jiffy. In addition,
the delay-calculation call to srcu_get_delay() in srcu_gp_end() is
replaced with a simple check for an expedited grace period. This change
schedules callback invocation immediately after expedited grace periods
complete, which results in greatly improved boot times. Testing done
by Marc and Zhangfei confirms that this change recovers most of the
performance degradation in boottime; for CONFIG_HZ_250 configuration,
specifically, boot times improve from 3m50s to 41s on Marc's setup;
and from 2m40s to ~9.7s on Zhangfei's setup.
In addition to the changes to default per phase delays, this
change adds 3 new kernel parameters - srcutree.srcu_max_nodelay,
srcutree.srcu_max_nodelay_phase, and srcutree.srcu_retry_check_delay.
This allows users to configure the srcu grace period scanning delays in
order to more quickly react to additional use cases.
Fixes: 640a7d37c3f4 ("srcu: Block less aggressively for expedited grace periods")
Fixes: 282d8998e9 ("srcu: Prevent expedited GPs and blocking readers from consuming CPU")
Reported-by: Zhangfei Gao <zhangfei.gao@linaro.org>
Reported-by: yueluck <yueluck@163.com>
Signed-off-by: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Tested-by: Marc Zyngier <maz@kernel.org>
Tested-by: Zhangfei Gao <zhangfei.gao@linaro.org>
Link: https://lore.kernel.org/all/20615615-0013-5adc-584f-2b1d5c03ebfc@linaro.org/
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Commit 282d8998e9 ("srcu: Prevent expedited GPs and blocking readers
from consuming CPU") fixed a problem where a long-running expedited SRCU
grace period could block kernel live patching. It did so by giving up
on expediting once a given SRCU expedited grace period grew too old.
Unfortunately, this added excessive delays to boots of virtual embedded
systems specifying "-bios QEMU_EFI.fd" to qemu. This commit therefore
makes the transition away from expediting less aggressive, increasing
the per-grace-period phase number of non-sleeping polls of readers from
one to three and increasing the required grace-period age from one jiffy
(actually from zero to one jiffies) to two jiffies (actually from one
to two jiffies).
Fixes: 282d8998e9 ("srcu: Prevent expedited GPs and blocking readers from consuming CPU")
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Reported-by: Zhangfei Gao <zhangfei.gao@linaro.org>
Reported-by: chenxiang (M)" <chenxiang66@hisilicon.com>
Cc: Shameerali Kolothum Thodi <shameerali.kolothum.thodi@huawei.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Link: https://lore.kernel.org/all/20615615-0013-5adc-584f-2b1d5c03ebfc@linaro.org/
NOMAP irq domains use the revmap_size field to indicate the maximum
hwirq number the domain accepts. This is a bit confusing as
revmap_size is usually used to indicate the size of the revmap array,
which a NOMAP domain doesn't have.
Instead, use the hwirq_max field which has the correct semantics, and
keep revmap_size to 0 for a NOMAP domain.
Signed-off-by: Xu Qiang <xuqiang36@huawei.com>
[maz: commit message]
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20220719063641.56541-3-xuqiang36@huawei.com
When using a NOMAP domain, __irq_resolve_mapping() doesn't store
the Linux IRQ number at the address optionally provided by the caller.
While this isn't a huge deal (the returned value is guaranteed
to the hwirq that was passed as a parameter), let's honour the letter
of the API by writing the expected value.
Fixes: d22558dd0a (“irqdomain: Introduce irq_resolve_mapping()”)
Signed-off-by: Xu Qiang <xuqiang36@huawei.com>
[maz: commit message]
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20220719063641.56541-2-xuqiang36@huawei.com
Pull perf fix from Borislav Petkov:
- A single data race fix on the perf event cleanup path to avoid
endless loops due to insufficient locking
* tag 'perf_urgent_for_v5.19_rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
perf/core: Fix data race between perf_event_set_output() and perf_mmap_close()
Pull printk fix from Petr Mladek:
- Make pr_flush() fast when consoles are suspended.
* tag 'printk-for-5.19-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/printk/linux:
printk: do not wait for consoles when suspended
Pyll sysctl fix from Luis Chamberlain:
"Only one fix for sysctl"
* tag 'sysctl-fixes-5.19-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/mcgrof/linux:
mm: sysctl: fix missing numa_stat when !CONFIG_HUGETLB_PAGE
