sched/isolation: Split housekeeping cpumask
Subsystems (12)
Frederic Weisbecker
maintainercover sched/isolation: Split housekeeping cpumask
Hi,
To prepare for extending cpusets to control CPU isolation features (nohz_full, rcu_nocbs, unbound timers, workqueues, kthreads affinity...), we need to split the global housekeeping_mask to one cpumask per isolation feature.
Doing so is quite a chunk already so I'm working on that as a standalone patchset. Once that get merged, the next step is to finally provide a cpuset interface for one of these isolation features: rcu_nocb could be interesting to handle first as nohz_full depends on it.
git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git
isolation/split
HEAD: 2c07a16ff50d1e722babee28b926d70522e6bd3e
Thanks,
Frederic
---
Frederic Weisbecker (8):
pci: Decouple HK_FLAG_WQ and HK_FLAG_DOMAIN cpumask fetch
workqueue: Decouple HK_FLAG_WQ and HK_FLAG_DOMAIN cpumask fetch
net: Decouple HK_FLAG_WQ and HK_FLAG_DOMAIN cpumask fetch sched/isolation: Use single feature type while referring to housekeeping cpumask sched/isolation: Consolidate check for housekeeping minimum service sched/isolation: Consolidate error handling
sched/isolation: Fix housekeeping_mask memory leak sched/isolation: Split housekeeping cpumask per isolation features
arch/x86/kernel/cpu/aperfmperf.c | 6 +- arch/x86/kvm/x86.c | 2 +- drivers/base/cpu.c | 2 +- drivers/pci/pci-driver.c | 21 +++-- include/linux/sched/isolation.h | 43 ++++++----- kernel/cgroup/cpuset.c | 6 +- kernel/cpu.c | 4 +- kernel/irq/cpuhotplug.c | 4 +- kernel/irq/manage.c | 4 +- kernel/kthread.c | 4 +- kernel/rcu/tasks.h | 2 +- kernel/rcu/tree_plugin.h | 2 +- kernel/sched/core.c | 12 +-- kernel/sched/fair.c | 10 +-- kernel/sched/isolation.c | 162 +++++++++++++++++++++++++-------------- kernel/sched/topology.c | 8 +- kernel/watchdog.c | 2 +- kernel/workqueue.c | 4 +- net/core/net-sysfs.c | 6 +- 19 files changed, 180 insertions(+), 124 deletions(-)
Frederic Weisbecker
maintainerPATCH 1/8 pci: Decouple HK_FLAG_WQ and HK_FLAG_DOMAIN cpumask fetch
To prepare for supporting each feature of the housekeeping cpumask toward cpuset, prepare each of the HK_FLAG_* entries to move to their own cpumask with enforcing to fetch them individually. The new constraint is that multiple HK_FLAG_* entries can't be mixed together anymore in a single call to housekeeping cpumask().
This will later allow, for example, to runtime modify the cpulist passed through "isolcpus=", "nohz_full=" and "rcu_nocbs=" kernel boot parameters.
--- drivers/pci/pci-driver.c | 21 ++++++++++++++++----- 1 file changed, 16 insertions(+), 5 deletions(-)
drivers/pci/pci-driver.c +16 -5
@@ -350,7 +350,6 @@ static int pci_call_probe(struct pci_driver *drv, struct pci_dev *dev, const struct pci_device_id *id) {int error, node, cpu; -int hk_flags =HK_FLAG_DOMAIN |HK_FLAG_WQ ;struct drv_dev_and_id ddi = { drv, dev, id };/*
@@ -368,17 +367,29 @@ static int pci_call_probe(struct pci_driver *drv, struct pci_dev *dev,* device is probed from work_on_cpu() of the Physical device. */ if (node <0 || node >=MAX_NUMNODES || !node_online (node) || -pci_physfn_is_probed (dev)) +pci_physfn_is_probed (dev)) { cpu = nr_cpu_ids; -else + }else { +cpumask_var_t wq_domain_mask; + +if (!zalloc_cpumask_var (&wq_domain_mask,GFP_KERNEL )) { + error = -ENOMEM ; + goto out; + } +cpumask_and (wq_domain_mask, +housekeeping_cpumask (HK_FLAG_WQ ), +housekeeping_cpumask (HK_FLAG_DOMAIN )); + cpu =cpumask_any_and (cpumask_of_node (node), -housekeeping_cpumask (hk_flags)); + wq_domain_mask); +free_cpumask_var (wq_domain_mask); + }if (cpu < nr_cpu_ids) error =work_on_cpu (cpu, local_pci_probe, &ddi);else error =local_pci_probe (&ddi); - +out: dev->is_probed =0 ;cpu_hotplug_enable ();return error;
Signature · 1 line
2.25.1
Frederic Weisbecker
maintainerPATCH 2/8 workqueue: Decouple HK_FLAG_WQ and HK_FLAG_DOMAIN cpumask fetch
In reply to Frederic WeisbeckerTo prepare for supporting each feature of the housekeeping cpumask toward cpuset, prepare each of the HK_FLAG_* entries to move to their own cpumask with enforcing to fetch them individually. The new constraint is that multiple HK_FLAG_* entries can't be mixed together anymore in a single call to housekeeping cpumask().
This will later allow, for example, to runtime modify the cpulist passed through "isolcpus=", "nohz_full=" and "rcu_nocbs=" kernel boot parameters.
--- kernel/workqueue.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
kernel/workqueue.c +2 -2
@@ -6017,13 +6017,13 @@ static void __init wq_numa_init(void)void __init workqueue_init_early (void ) {int std_nice[NR_STD_WORKER_POOLS ] = {0 ,HIGHPRI_NICE_LEVEL }; -int hk_flags =HK_FLAG_DOMAIN |HK_FLAG_WQ ;int i, cpu;BUILD_BUG_ON (__alignof__(struct pool_workqueue ) < __alignof__(long long ));BUG_ON (!alloc_cpumask_var (&wq_unbound_cpumask,GFP_KERNEL )); -cpumask_copy (wq_unbound_cpumask,housekeeping_cpumask (hk_flags)); +cpumask_copy (wq_unbound_cpumask,housekeeping_cpumask (HK_FLAG_WQ )); +cpumask_and (wq_unbound_cpumask, wq_unbound_cpumask,housekeeping_cpumask (HK_FLAG_DOMAIN )); pwq_cache =KMEM_CACHE (pool_workqueue,SLAB_PANIC );
Signature · 1 line
2.25.1
Re: [PATCH 2/8] workqueue: Decouple HK_FLAG_WQ and HK_FLAG_DOMAIN cpumask fetch
Quoted 10 lines · Frederic Weisbecker
To prepare for supporting each feature of the housekeeping cpumask toward cpuset, prepare each of the HK_FLAG_* entries to move to their own cpumask with enforcing to fetch them individually. The new constraint is that multiple HK_FLAG_* entries can't be mixed together anymore in a single call to housekeeping
cpumask().This will later allow, for example, to runtime modify the cpulist passed through "isolcpus=", "nohz_full=" and "rcu_nocbs=" kernel boot parameters.
Quoted 11 lines
Signed-off-by Frederic Weisbecker <[email protected]>Cc Thomas Gleixner <[email protected]>Cc Juri Lelli <[email protected]>Cc Marcelo Tosatti <[email protected]>Cc Nitesh Lal <[email protected]>Cc Nicolas Saenz <[email protected]>Cc Peter Zijlstra <[email protected]>Cc Christoph Lameter <[email protected]>Cc Tejun Heo <[email protected]>Cc Zefan Li <[email protected]>Cc Alex Belits <[email protected]>
Thanks.
Signature · 1 line
tejun
Frederic Weisbecker
maintainerPATCH 3/8 net: Decouple HK_FLAG_WQ and HK_FLAG_DOMAIN cpumask fetch
In reply to Frederic WeisbeckerTo prepare for supporting each feature of the housekeeping cpumask toward cpuset, prepare each of the HK_FLAG_* entries to move to their own cpumask with enforcing to fetch them individually. The new constraint is that multiple HK_FLAG_* entries can't be mixed together anymore in a single call to housekeeping cpumask().
This will later allow, for example, to runtime modify the cpulist passed through "isolcpus=", "nohz_full=" and "rcu_nocbs=" kernel boot parameters.
--- net/core/net-sysfs.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-)
net/core/net-sysfs.c +3 -3
@@ -831,7 +831,7 @@ static ssize_t store_rps_map(struct netdev_rx_queue *queue, {struct rps_map *old_map, *map;cpumask_var_t mask; -int err, cpu, i, hk_flags; +int err, cpu, i;static DEFINE_MUTEX (rps_map_mutex);if (!capable (CAP_NET_ADMIN ))
@@ -847,8 +847,8 @@ static ssize_t store_rps_map(struct netdev_rx_queue *queue, }if (!cpumask_empty (mask)) { - hk_flags =HK_FLAG_DOMAIN |HK_FLAG_WQ ; -cpumask_and (mask, mask,housekeeping_cpumask (hk_flags)); +cpumask_and (mask, mask,housekeeping_cpumask (HK_FLAG_DOMAIN )); +cpumask_and (mask, mask,housekeeping_cpumask (HK_FLAG_WQ ));if (cpumask_empty (mask)) {free_cpumask_var (mask);return -EINVAL ;
Signature · 1 line
2.25.1
Frederic Weisbecker
maintainerPATCH 4/8 sched/isolation: Use single feature type while referring to housekeeping cpumask
In reply to Frederic WeisbeckerRefer to housekeeping APIs using single feature types instead of flags. This prevents from passing multiple isolation features at once to housekeeping interfaces, which soon won't be possible anymore as each isolation features will have their own cpumask.
--- arch/x86/kernel/cpu/aperfmperf.c | 6 ++--- arch/x86/kvm/x86.c | 2 +- drivers/base/cpu.c | 2 +- drivers/pci/pci-driver.c | 4 +-- include/linux/sched/isolation.h | 43 ++++++++++++++++---------------- kernel/cgroup/cpuset.c | 6 ++--- kernel/cpu.c | 4 +-- kernel/irq/cpuhotplug.c | 4 +-- kernel/irq/manage.c | 4 +-- kernel/kthread.c | 4 +-- kernel/rcu/tasks.h | 2 +- kernel/rcu/tree_plugin.h | 2 +- kernel/sched/core.c | 12 ++++----- kernel/sched/fair.c | 10 ++++---- kernel/sched/isolation.c | 32 ++++++++++++++++-------- kernel/sched/topology.c | 8 +++--- kernel/watchdog.c | 2 +- kernel/workqueue.c | 4 +-- net/core/net-sysfs.c | 4 +-- 19 files changed, 84 insertions(+), 71 deletions(-)
arch/x86/kernel/cpu/aperfmperf.c +3 -3
@@ -91,7 +91,7 @@ unsigned int aperfmperf_get_khz(int cpu)if (!boot_cpu_has (X86_FEATURE_APERFMPERF ))return 0 ; -if (!housekeeping_cpu (cpu,HK_FLAG_MISC )) +if (!housekeeping_cpu (cpu,HK_TYPE_MISC ))return 0 ;if (rcu_is_idle_cpu (cpu))
@@ -114,7 +114,7 @@ void arch_freq_prepare_all(void)return ;for_each_online_cpu (cpu) { -if (!housekeeping_cpu (cpu,HK_FLAG_MISC )) +if (!housekeeping_cpu (cpu,HK_TYPE_MISC ))continue ;if (rcu_is_idle_cpu (cpu))continue ;/* Idle CPUs are completely uninteresting. */
@@ -136,7 +136,7 @@ unsigned int arch_freq_get_on_cpu(int cpu)if (!boot_cpu_has (X86_FEATURE_APERFMPERF ))return 0 ; -if (!housekeeping_cpu (cpu,HK_FLAG_MISC )) +if (!housekeeping_cpu (cpu,HK_TYPE_MISC ))return 0 ;if (aperfmperf_snapshot_cpu (cpu,ktime_get (), true))
arch/x86/kvm/x86.c +1 -1
@@ -8684,7 +8684,7 @@ int kvm_arch_init(void *opaque) }if (pi_inject_timer ==-1 ) - pi_inject_timer =housekeeping_enabled (HK_FLAG_TIMER ); + pi_inject_timer =housekeeping_enabled (HK_TYPE_TIMER );#ifdef CONFIG_X86_64 pvclock_gtod_register_notifier (&pvclock_gtod_notifier);
drivers/base/cpu.c +1 -1
@@ -275,7 +275,7 @@ static ssize_t print_cpus_isolated(struct device *dev,return -ENOMEM ;cpumask_andnot (isolated, cpu_possible_mask, -housekeeping_cpumask (HK_FLAG_DOMAIN )); +housekeeping_cpumask (HK_TYPE_DOMAIN )); len =sysfs_emit (buf,"%*pbl\n" ,cpumask_pr_args (isolated));free_cpumask_var (isolated);
drivers/pci/pci-driver.c +2 -2
@@ -377,8 +377,8 @@ static int pci_call_probe(struct pci_driver *drv, struct pci_dev *dev, goto out; }cpumask_and (wq_domain_mask, -housekeeping_cpumask (HK_FLAG_WQ ), -housekeeping_cpumask (HK_FLAG_DOMAIN )); +housekeeping_cpumask (HK_TYPE_WQ ), +housekeeping_cpumask (HK_TYPE_DOMAIN )); cpu =cpumask_any_and (cpumask_of_node (node), wq_domain_mask);
include/linux/sched/isolation.h +22 -21
76 lines not on this page · raw message
kernel/cgroup/cpuset.c +3 -3
@@ -811,7 +811,7 @@ static int generate_sched_domains(cpumask_var_t **domains,update_domain_attr_tree (dattr, &top_cpuset); }cpumask_and (doms[0 ], top_cpuset.effective_cpus, -housekeeping_cpumask (HK_FLAG_DOMAIN )); +housekeeping_cpumask (HK_TYPE_DOMAIN )); goto done; }
@@ -841,7 +841,7 @@ static int generate_sched_domains(cpumask_var_t **domains,if (!cpumask_empty (cp->cpus_allowed) && !(is_sched_load_balance (cp) &&cpumask_intersects (cp->cpus_allowed, -housekeeping_cpumask (HK_FLAG_DOMAIN )))) +housekeeping_cpumask (HK_TYPE_DOMAIN ))))continue ; if (root_load_balance &&
@@ -930,7 +930,7 @@ static int generate_sched_domains(cpumask_var_t **domains,if (apn == b->pn) {cpumask_or (dp, dp, b->effective_cpus); -cpumask_and (dp, dp,housekeeping_cpumask (HK_FLAG_DOMAIN )); +cpumask_and (dp, dp,housekeeping_cpumask (HK_TYPE_DOMAIN ));if (dattr)update_domain_attr_tree (dattr + nslot, b);
kernel/cpu.c +2 -2
@@ -1488,8 +1488,8 @@ int freeze_secondary_cpus(int primary)cpu_maps_update_begin ();if (primary ==-1 ) { primary =cpumask_first (cpu_online_mask); -if (!housekeeping_cpu (primary,HK_FLAG_TIMER )) - primary =housekeeping_any_cpu (HK_FLAG_TIMER ); +if (!housekeeping_cpu (primary,HK_TYPE_TIMER )) + primary =housekeeping_any_cpu (HK_TYPE_TIMER ); }else {if (!cpu_online (primary)) primary =cpumask_first (cpu_online_mask);
kernel/irq/cpuhotplug.c +2 -2
@@ -176,10 +176,10 @@ static bool hk_should_isolate(struct irq_data *data, unsigned int cpu) {const struct cpumask *hk_mask; -if (!housekeeping_enabled (HK_FLAG_MANAGED_IRQ )) +if (!housekeeping_enabled (HK_TYPE_MANAGED_IRQ ))return false; - hk_mask =housekeeping_cpumask (HK_FLAG_MANAGED_IRQ ); + hk_mask =housekeeping_cpumask (HK_TYPE_MANAGED_IRQ );if (cpumask_subset (irq_data_get_effective_affinity_mask (data), hk_mask))return false;
kernel/irq/manage.c +2 -2
@@ -247,13 +247,13 @@ int irq_do_set_affinity(struct irq_data *data, const struct cpumask *mask,* online. */ if (irqd_affinity_is_managed (data) && -housekeeping_enabled (HK_FLAG_MANAGED_IRQ )) { +housekeeping_enabled (HK_TYPE_MANAGED_IRQ )) {const struct cpumask *hk_mask, *prog_mask;static DEFINE_RAW_SPINLOCK (tmp_mask_lock);static struct cpumask tmp_mask; - hk_mask =housekeeping_cpumask (HK_FLAG_MANAGED_IRQ ); + hk_mask =housekeeping_cpumask (HK_TYPE_MANAGED_IRQ );raw_spin_lock (&tmp_mask_lock);cpumask_and (&tmp_mask, mask, hk_mask);
kernel/kthread.c +2 -2
@@ -306,7 +306,7 @@ static int kthread(void *_create)* back to default in case they have been changed. */ sched_setscheduler_nocheck (current,SCHED_NORMAL , ¶m); -set_cpus_allowed_ptr (current,housekeeping_cpumask (HK_FLAG_KTHREAD )); +set_cpus_allowed_ptr (current,housekeeping_cpumask (HK_TYPE_KTHREAD ));/* OK, tell user we're spawned, wait for stop or wakeup */ __set_current_state (TASK_UNINTERRUPTIBLE );
@@ -661,7 +661,7 @@ int kthreadd(void *unused)/* Setup a clean context for our children to inherit. */ set_task_comm (tsk,"kthreadd" );ignore_signals (tsk); -set_cpus_allowed_ptr (tsk,housekeeping_cpumask (HK_FLAG_KTHREAD )); +set_cpus_allowed_ptr (tsk,housekeeping_cpumask (HK_TYPE_KTHREAD ));set_mems_allowed (node_states[N_MEMORY ]); current->flags |=PF_NOFREEZE ;
kernel/rcu/tasks.h +1 -1
@@ -187,7 +187,7 @@ static int __noreturn rcu_tasks_kthread(void *arg)struct rcu_tasks *rtp = arg;/* Run on housekeeping CPUs by default. Sysadm can move if desired. */ -housekeeping_affine (current,HK_FLAG_RCU ); +housekeeping_affine (current,HK_TYPE_RCU );WRITE_ONCE (rtp->kthread_ptr, current);// Let GPs start! /*
kernel/rcu/tree_plugin.h +1 -1
@@ -1475,7 +1475,7 @@ static void rcu_bind_gp_kthread(void) {if (!tick_nohz_full_enabled ())return ; -housekeeping_affine (current,HK_FLAG_RCU ); +housekeeping_affine (current,HK_TYPE_RCU ); }/* Record the current task on dyntick-idle entry. */
kernel/sched/core.c +6 -6
@@ -1013,13 +1013,13 @@ int get_nohz_timer_target(void)struct sched_domain *sd;const struct cpumask *hk_mask; -if (housekeeping_cpu (cpu,HK_FLAG_TIMER )) { +if (housekeeping_cpu (cpu,HK_TYPE_TIMER )) {if (!idle_cpu (cpu))return cpu; default_cpu = cpu; } - hk_mask =housekeeping_cpumask (HK_FLAG_TIMER ); + hk_mask =housekeeping_cpumask (HK_TYPE_TIMER );rcu_read_lock ();for_each_domain (cpu, sd) {
@@ -1035,7 +1035,7 @@ int get_nohz_timer_target(void) }if (default_cpu ==-1 ) - default_cpu =housekeeping_any_cpu (HK_FLAG_TIMER ); + default_cpu =housekeeping_any_cpu (HK_TYPE_TIMER ); cpu = default_cpu; unlock:rcu_read_unlock ();
@@ -5355,7 +5355,7 @@ static void sched_tick_start(int cpu)int os;struct tick_work *twork; -if (housekeeping_cpu (cpu,HK_FLAG_TICK )) +if (housekeeping_cpu (cpu,HK_TYPE_TICK ))return ;WARN_ON_ONCE (!tick_work_cpu);
@@ -5376,7 +5376,7 @@ static void sched_tick_stop(int cpu)struct tick_work *twork;int os; -if (housekeeping_cpu (cpu,HK_FLAG_TICK )) +if (housekeeping_cpu (cpu,HK_TYPE_TICK ))return ;WARN_ON_ONCE (!tick_work_cpu);
@@ -9212,7 +9212,7 @@ void __init sched_init_smp(void)mutex_unlock (&sched_domains_mutex);/* Move init over to a non-isolated CPU */ -if (set_cpus_allowed_ptr (current,housekeeping_cpumask (HK_FLAG_DOMAIN )) <0 ) +if (set_cpus_allowed_ptr (current,housekeeping_cpumask (HK_TYPE_DOMAIN )) <0 )BUG (); current->flags &= ~PF_NO_SETAFFINITY ;sched_init_granularity ();
kernel/sched/fair.c +5 -5
@@ -10362,7 +10362,7 @@ static inline int on_null_domain(struct rq *rq)* - When one of the busy CPUs notice that there may be an idle rebalancing * needed, they will kick the idle load balancer, which then does idle * load balancing for all the idle CPUs. -* - HK_FLAG_MISC CPUs are used for this task, because HK_FLAG_SCHED not set +* - HK_TYPE_MISC CPUs are used for this task, because HK_TYPE_SCHED not set * anywhere yet. */
@@ -10371,7 +10371,7 @@ static inline int find_new_ilb(void)int ilb;const struct cpumask *hk_mask; - hk_mask =housekeeping_cpumask (HK_FLAG_MISC ); + hk_mask =housekeeping_cpumask (HK_TYPE_MISC );for_each_cpu_and (ilb, nohz.idle_cpus_mask, hk_mask) {
@@ -10387,7 +10387,7 @@ static inline int find_new_ilb(void)/* * Kick a CPU to do the nohz balancing, if it is time for it. We pick any -* idle CPU in the HK_FLAG_MISC housekeeping set (if there is one). +* idle CPU in the HK_TYPE_MISC housekeeping set (if there is one). */ static void kick_ilb (unsigned int flags) {
@@ -10600,7 +10600,7 @@ void nohz_balance_enter_idle(int cpu)return ;/* Spare idle load balancing on CPUs that don't want to be disturbed: */ -if (!housekeeping_cpu (cpu,HK_FLAG_SCHED )) +if (!housekeeping_cpu (cpu,HK_TYPE_SCHED ))return ;/*
@@ -10816,7 +10816,7 @@ static void nohz_newidle_balance(struct rq *this_rq)* This CPU doesn't want to be disturbed by scheduler * housekeeping */ -if (!housekeeping_cpu (this_cpu,HK_FLAG_SCHED )) +if (!housekeeping_cpu (this_cpu,HK_TYPE_SCHED ))return ;/* Will wake up very soon. No time for doing anything else*/
kernel/sched/isolation.c +22 -10
@@ -9,23 +9,35 @@*/ #include "sched.h" +enum hk_flags { +HK_FLAG_TIMER =BIT (HK_TYPE_TIMER ), +HK_FLAG_RCU =BIT (HK_TYPE_RCU ), +HK_FLAG_MISC =BIT (HK_TYPE_MISC ), +HK_FLAG_SCHED =BIT (HK_TYPE_SCHED ), +HK_FLAG_TICK =BIT (HK_TYPE_TICK ), +HK_FLAG_DOMAIN =BIT (HK_TYPE_DOMAIN ), +HK_FLAG_WQ =BIT (HK_TYPE_WQ ), +HK_FLAG_MANAGED_IRQ =BIT (HK_TYPE_MANAGED_IRQ ), +HK_FLAG_KTHREAD =BIT (HK_TYPE_KTHREAD ), +}; +DEFINE_STATIC_KEY_FALSE (housekeeping_overridden);EXPORT_SYMBOL_GPL (housekeeping_overridden);static cpumask_var_t housekeeping_mask;static unsigned int housekeeping_flags; -bool housekeeping_enabled (enum hk_flags flags) +bool housekeeping_enabled (enum hk_type type) { -return !!(housekeeping_flags & flags); +return !!(housekeeping_flags &BIT (type)); }EXPORT_SYMBOL_GPL (housekeeping_enabled); -int housekeeping_any_cpu (enum hk_flags flags) +int housekeeping_any_cpu (enum hk_type type) {int cpu;if (static_branch_unlikely (&housekeeping_overridden)) { -if (housekeeping_flags & flags) { +if (housekeeping_flags &BIT (type)) { cpu =sched_numa_find_closest (housekeeping_mask,smp_processor_id ());if (cpu < nr_cpu_ids)return cpu;
@@ -37,27 +49,27 @@ int housekeeping_any_cpu(enum hk_flags flags) }EXPORT_SYMBOL_GPL (housekeeping_any_cpu); -const struct cpumask *housekeeping_cpumask (enum hk_flags flags) +const struct cpumask *housekeeping_cpumask (enum hk_type type) {if (static_branch_unlikely (&housekeeping_overridden)) -if (housekeeping_flags & flags) +if (housekeeping_flags &BIT (type))return housekeeping_mask;return cpu_possible_mask; }EXPORT_SYMBOL_GPL (housekeeping_cpumask); -void housekeeping_affine (struct task_struct *t,enum hk_flags flags) +void housekeeping_affine (struct task_struct *t,enum hk_type type) {if (static_branch_unlikely (&housekeeping_overridden)) -if (housekeeping_flags & flags) +if (housekeeping_flags &BIT (type))set_cpus_allowed_ptr (t, housekeeping_mask); }EXPORT_SYMBOL_GPL (housekeeping_affine); -bool housekeeping_test_cpu (int cpu,enum hk_flags flags) +bool housekeeping_test_cpu (int cpu,enum hk_type type) {if (static_branch_unlikely (&housekeeping_overridden)) -if (housekeeping_flags & flags) +if (housekeeping_flags &BIT (type))return cpumask_test_cpu (cpu, housekeeping_mask);return true; }
kernel/sched/topology.c +4 -4
@@ -1366,7 +1366,7 @@ static void asym_cpu_capacity_scan(void)list_for_each_entry (entry, &asym_cap_list, link)cpumask_clear (cpu_capacity_span (entry)); -for_each_cpu_and (cpu, cpu_possible_mask,housekeeping_cpumask (HK_FLAG_DOMAIN )) +for_each_cpu_and (cpu, cpu_possible_mask,housekeeping_cpumask (HK_TYPE_DOMAIN ))asym_cpu_capacity_update_data (cpu);list_for_each_entry_safe (entry, next, &asym_cap_list, link) {
@@ -2351,7 +2351,7 @@ int sched_init_domains(const struct cpumask *cpu_map) doms_cur =alloc_sched_domains (ndoms_cur);if (!doms_cur) doms_cur = &fallback_doms; -cpumask_and (doms_cur[0 ], cpu_map,housekeeping_cpumask (HK_FLAG_DOMAIN )); +cpumask_and (doms_cur[0 ], cpu_map,housekeeping_cpumask (HK_TYPE_DOMAIN )); err =build_sched_domains (doms_cur[0 ],NULL );return err;
@@ -2440,7 +2440,7 @@ void partition_sched_domains_locked(int ndoms_new, cpumask_var_t doms_new[],if (doms_new) { n =1 ;cpumask_and (doms_new[0 ], cpu_active_mask, -housekeeping_cpumask (HK_FLAG_DOMAIN )); +housekeeping_cpumask (HK_TYPE_DOMAIN )); } }else { n = ndoms_new;
@@ -2475,7 +2475,7 @@ void partition_sched_domains_locked(int ndoms_new, cpumask_var_t doms_new[], n =0 ; doms_new = &fallback_doms;cpumask_and (doms_new[0 ], cpu_active_mask, -housekeeping_cpumask (HK_FLAG_DOMAIN )); +housekeeping_cpumask (HK_TYPE_DOMAIN )); }/* Build new domains: */
kernel/watchdog.c +1 -1
@@ -748,7 +748,7 @@ void __init lockup_detector_init(void)pr_info ("Disabling watchdog on nohz_full cores by default\n" );cpumask_copy (&watchdog_cpumask, -housekeeping_cpumask (HK_FLAG_TIMER )); +housekeeping_cpumask (HK_TYPE_TIMER ));if (!watchdog_nmi_probe ()) nmi_watchdog_available = true;
kernel/workqueue.c +2 -2
@@ -6022,8 +6022,8 @@ void __init workqueue_init_early(void)BUILD_BUG_ON (__alignof__(struct pool_workqueue ) < __alignof__(long long ));BUG_ON (!alloc_cpumask_var (&wq_unbound_cpumask,GFP_KERNEL )); -cpumask_copy (wq_unbound_cpumask,housekeeping_cpumask (HK_FLAG_WQ )); -cpumask_and (wq_unbound_cpumask, wq_unbound_cpumask,housekeeping_cpumask (HK_FLAG_DOMAIN )); +cpumask_copy (wq_unbound_cpumask,housekeeping_cpumask (HK_TYPE_WQ )); +cpumask_and (wq_unbound_cpumask, wq_unbound_cpumask,housekeeping_cpumask (HK_TYPE_DOMAIN )); pwq_cache =KMEM_CACHE (pool_workqueue,SLAB_PANIC );
net/core/net-sysfs.c +2 -2
@@ -847,8 +847,8 @@ static ssize_t store_rps_map(struct netdev_rx_queue *queue, }if (!cpumask_empty (mask)) { -cpumask_and (mask, mask,housekeeping_cpumask (HK_FLAG_DOMAIN )); -cpumask_and (mask, mask,housekeeping_cpumask (HK_FLAG_WQ )); +cpumask_and (mask, mask,housekeeping_cpumask (HK_TYPE_DOMAIN )); +cpumask_and (mask, mask,housekeeping_cpumask (HK_TYPE_WQ ));if (cpumask_empty (mask)) {free_cpumask_var (mask);return -EINVAL ;
Signature · 1 line
2.25.1
Frederic Weisbecker
maintainerPATCH 5/8 sched/isolation: Consolidate check for housekeeping minimum service
In reply to Frederic WeisbeckerThere can be two subsequent calls to housekeeping_setup() due to "nohz_full=" and "isolcpus=" that can mix up. The two passes each have their own way to deal with an empty housekeeping set of CPUs. Consolidate this part and remove the awful "tmp" based naming.
--- kernel/sched/isolation.c | 35 +++++++++++++++++------------------ 1 file changed, 17 insertions(+), 18 deletions(-)
kernel/sched/isolation.c +17 -18
@@ -91,8 +91,7 @@ void __init housekeeping_init(void)static int __init housekeeping_setup (char *str,enum hk_flags flags) { -cpumask_var_t non_housekeeping_mask; -cpumask_var_t tmp; +cpumask_var_t non_housekeeping_mask, housekeeping_staging;alloc_bootmem_cpumask_var (&non_housekeeping_mask);if (cpulist_parse (str, non_housekeeping_mask) <0 ) {
@@ -101,32 +100,32 @@ static int __init housekeeping_setup(char *str, enum hk_flags flags)return 0 ; } -alloc_bootmem_cpumask_var (&tmp); -if (!housekeeping_flags) { -alloc_bootmem_cpumask_var (&housekeeping_mask); -cpumask_andnot (housekeeping_mask, - cpu_possible_mask, non_housekeeping_mask); +alloc_bootmem_cpumask_var (&housekeeping_staging); +cpumask_andnot (housekeeping_staging, + cpu_possible_mask, non_housekeeping_mask); -cpumask_andnot (tmp, cpu_present_mask, non_housekeeping_mask); -if (cpumask_empty (tmp)) { +if (!cpumask_intersects (cpu_present_mask, housekeeping_staging)) { +__cpumask_set_cpu (smp_processor_id (), housekeeping_staging); +__cpumask_clear_cpu (smp_processor_id (), non_housekeeping_mask); +if (!housekeeping_flags) {pr_warn ("Housekeeping: must include one present CPU, " "using boot CPU:%d\n" ,smp_processor_id ()); -__cpumask_set_cpu (smp_processor_id (), housekeeping_mask); -__cpumask_clear_cpu (smp_processor_id (), non_housekeeping_mask); } + } + +if (!housekeeping_flags) { +alloc_bootmem_cpumask_var (&housekeeping_mask); +cpumask_copy (housekeeping_mask, housekeeping_staging); }else { -cpumask_andnot (tmp, cpu_present_mask, non_housekeeping_mask); -if (cpumask_empty (tmp)) -__cpumask_clear_cpu (smp_processor_id (), non_housekeeping_mask); -cpumask_andnot (tmp, cpu_possible_mask, non_housekeeping_mask); -if (!cpumask_equal (tmp, housekeeping_mask)) { +if (!cpumask_equal (housekeeping_staging, housekeeping_mask)) {pr_warn ("Housekeeping: nohz_full= must match isolcpus=\n" ); -free_bootmem_cpumask_var (tmp); +free_bootmem_cpumask_var (housekeeping_staging);free_bootmem_cpumask_var (non_housekeeping_mask);return 0 ; } } -free_bootmem_cpumask_var (tmp); + +free_bootmem_cpumask_var (housekeeping_staging);if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) {if (IS_ENABLED (CONFIG_NO_HZ_FULL )) {
Signature · 1 line
2.25.1
Frederic Weisbecker
maintainerPATCH 6/8 sched/isolation: Consolidate error handling
In reply to Frederic WeisbeckerCentralize the mask freeing and return value for the error path. This makes potential leaks more visible.
--- kernel/sched/isolation.c | 19 +++++++++---------- 1 file changed, 9 insertions(+), 10 deletions(-)
kernel/sched/isolation.c +9 -10
@@ -92,12 +92,12 @@ void __init housekeeping_init(void)static int __init housekeeping_setup (char *str,enum hk_flags flags) {cpumask_var_t non_housekeeping_mask, housekeeping_staging; +int err =0 ;alloc_bootmem_cpumask_var (&non_housekeeping_mask);if (cpulist_parse (str, non_housekeeping_mask) <0 ) {pr_warn ("Housekeeping: nohz_full= or isolcpus= incorrect CPU range\n" ); -free_bootmem_cpumask_var (non_housekeeping_mask); -return 0 ; + goto free_non_housekeeping_mask; }alloc_bootmem_cpumask_var (&housekeeping_staging);
@@ -119,30 +119,29 @@ static int __init housekeeping_setup(char *str, enum hk_flags flags) }else {if (!cpumask_equal (housekeeping_staging, housekeeping_mask)) {pr_warn ("Housekeeping: nohz_full= must match isolcpus=\n" ); -free_bootmem_cpumask_var (housekeeping_staging); -free_bootmem_cpumask_var (non_housekeeping_mask); -return 0 ; + goto free_housekeeping_staging; } } -free_bootmem_cpumask_var (housekeeping_staging); -if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) {if (IS_ENABLED (CONFIG_NO_HZ_FULL )) {tick_nohz_full_setup (non_housekeeping_mask); }else {pr_warn ("Housekeeping: nohz unsupported." " Build with CONFIG_NO_HZ_FULL\n" ); -free_bootmem_cpumask_var (non_housekeeping_mask); -return 0 ; + goto free_housekeeping_staging; } } housekeeping_flags |= flags; + err =1 ; +free_housekeeping_staging: +free_bootmem_cpumask_var (housekeeping_staging); +free_non_housekeeping_mask:free_bootmem_cpumask_var (non_housekeeping_mask); -return 1 ; +return err; }static int __init housekeeping_nohz_full_setup (char *str)
Signature · 1 line
2.25.1
Frederic Weisbecker
maintainerPATCH 7/8 sched/isolation: Fix housekeeping_mask memory leak
In reply to Frederic WeisbeckerIf "nohz_full=" or "isolcpus=nohz" are called with CONFIG_NO_HZ_FULL=n, housekeeping_mask doesn't get freed despite it being unused if housekeeping_setup() is called for the first time.
Check this scenario first to fix this, so that no useless allocation is performed.
--- kernel/sched/isolation.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-)
kernel/sched/isolation.c +10 -9
@@ -94,6 +94,14 @@ static int __init housekeeping_setup(char *str, enum hk_flags flags)cpumask_var_t non_housekeeping_mask, housekeeping_staging;int err =0 ; +if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) { +if (!IS_ENABLED (CONFIG_NO_HZ_FULL )) { +pr_warn ("Housekeeping: nohz unsupported." +" Build with CONFIG_NO_HZ_FULL\n" ); +return 0 ; + } + } +alloc_bootmem_cpumask_var (&non_housekeeping_mask);if (cpulist_parse (str, non_housekeeping_mask) <0 ) {pr_warn ("Housekeeping: nohz_full= or isolcpus= incorrect CPU range\n" );
@@ -123,15 +131,8 @@ static int __init housekeeping_setup(char *str, enum hk_flags flags) } } -if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) { -if (IS_ENABLED (CONFIG_NO_HZ_FULL )) { -tick_nohz_full_setup (non_housekeeping_mask); - }else { -pr_warn ("Housekeeping: nohz unsupported." -" Build with CONFIG_NO_HZ_FULL\n" ); - goto free_housekeeping_staging; - } - } +if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) +tick_nohz_full_setup (non_housekeeping_mask); housekeeping_flags |= flags; err =1 ;
Signature · 1 line
2.25.1
Re: [PATCH 7/8] sched/isolation: Fix housekeeping_mask memory leak
Quoted 8 lines · Frederic Weisbecker
If "nohz_full=" or "isolcpus=nohz" are called with
CONFIG_NO_HZ_FULL=n,housekeeping_maskdoesn't get freed despite it being unused ifhousekeeping_setup()is called for the first time.Check this scenario first to fix this, so that no useless allocation is performed.
Signed-off-by Frederic Weisbecker <[email protected]>
Quoted 54 lines
Cc Thomas Gleixner <[email protected]>Cc Juri Lelli <[email protected]>Cc Marcelo Tosatti <[email protected]>Cc Nitesh Lal <[email protected]>Cc Nicolas Saenz <[email protected]>Cc Peter Zijlstra <[email protected]>Cc Christoph Lameter <[email protected]>Cc Tejun Heo <[email protected]>Cc Zefan Li <[email protected]>Cc Alex Belits <[email protected]>--- kernel/sched/isolation.c | 19 ++++++++++--------- 1 file changed, 10 insertions(+), 9 deletions(-)kernel/sched/isolation.c +10
-9@@ -94,6 +94,14 @@ static int __init housekeeping_setup(char *str, enum hk_flags flags)cpumask_var_t non_housekeeping_mask, housekeeping_staging;int err =0 ; +if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) { +if (!IS_ENABLED (CONFIG_NO_HZ_FULL )) { +pr_warn ("Housekeeping: nohz unsupported." +" Build with CONFIG_NO_HZ_FULL\n" ); +return 0 ; + } + } +alloc_bootmem_cpumask_var (&non_housekeeping_mask);if (cpulist_parse (str, non_housekeeping_mask) <0 ) {pr_warn ("Housekeeping: nohz_full= or isolcpus= incorrect CPU range\n" );@@ -123,15 +131,8 @@ static int __init housekeeping_setup(char *str, enum hk_flags flags) } } -if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) { -if (IS_ENABLED (CONFIG_NO_HZ_FULL )) { -tick_nohz_full_setup (non_housekeeping_mask); - }else { -pr_warn ("Housekeeping: nohz unsupported." -" Build with CONFIG_NO_HZ_FULL\n" ); - goto free_housekeeping_staging; - } - } +if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) +tick_nohz_full_setup (non_housekeeping_mask); housekeeping_flags |= flags; err =1 ;Signature · 1 line
2.25.1
Frederic Weisbecker
maintainerPATCH 8/8 sched/isolation: Split housekeeping cpumask per isolation features
In reply to Frederic WeisbeckerTo prepare for supporting each housekeeping feature toward cpuset, split the global housekeeping cpumask per HK_TYPE_* entry.
This will later allow, for example, to runtime modify the cpulist passed through "isolcpus=", "nohz_full=" and "rcu_nocbs=" kernel boot parameters.
--- kernel/sched/isolation.c | 91 +++++++++++++++++++++++++++------------- 1 file changed, 62 insertions(+), 29 deletions(-)
kernel/sched/isolation.c +62 -29
@@ -23,12 +23,17 @@ enum hk_flags {DEFINE_STATIC_KEY_FALSE (housekeeping_overridden);EXPORT_SYMBOL_GPL (housekeeping_overridden); -static cpumask_var_t housekeeping_mask; -static unsigned int housekeeping_flags; + +struct housekeeping { +cpumask_var_t cpumasks[HK_TYPE_MAX ]; +unsigned long flags; +}; + +static struct housekeeping housekeeping;bool housekeeping_enabled (enum hk_type type) { -return !!(housekeeping_flags &BIT (type)); +return !!(housekeeping.flags &BIT (type)); }EXPORT_SYMBOL_GPL (housekeeping_enabled);
@@ -37,12 +42,12 @@ int housekeeping_any_cpu(enum hk_type type)int cpu;if (static_branch_unlikely (&housekeeping_overridden)) { -if (housekeeping_flags &BIT (type)) { - cpu =sched_numa_find_closest (housekeeping_mask,smp_processor_id ()); +if (housekeeping.flags &BIT (type)) { + cpu =sched_numa_find_closest (housekeeping.cpumasks[type],smp_processor_id ());if (cpu < nr_cpu_ids)return cpu; -return cpumask_any_and (housekeeping_mask, cpu_online_mask); +return cpumask_any_and (housekeeping.cpumasks[type], cpu_online_mask); } }return smp_processor_id ();
@@ -52,8 +57,8 @@ EXPORT_SYMBOL_GPL(housekeeping_any_cpu);const struct cpumask *housekeeping_cpumask (enum hk_type type) {if (static_branch_unlikely (&housekeeping_overridden)) -if (housekeeping_flags &BIT (type)) -return housekeeping_mask; +if (housekeeping.flags &BIT (type)) +return housekeeping.cpumasks[type];return cpu_possible_mask; }EXPORT_SYMBOL_GPL (housekeeping_cpumask);
@@ -61,40 +66,53 @@ EXPORT_SYMBOL_GPL(housekeeping_cpumask);void housekeeping_affine (struct task_struct *t,enum hk_type type) {if (static_branch_unlikely (&housekeeping_overridden)) -if (housekeeping_flags &BIT (type)) -set_cpus_allowed_ptr (t, housekeeping_mask); +if (housekeeping.flags &BIT (type)) +set_cpus_allowed_ptr (t, housekeeping.cpumasks[type]); }EXPORT_SYMBOL_GPL (housekeeping_affine);bool housekeeping_test_cpu (int cpu,enum hk_type type) {if (static_branch_unlikely (&housekeeping_overridden)) -if (housekeeping_flags &BIT (type)) -return cpumask_test_cpu (cpu, housekeeping_mask); +if (housekeeping.flags &BIT (type)) +return cpumask_test_cpu (cpu, housekeeping.cpumasks[type]);return true; }EXPORT_SYMBOL_GPL (housekeeping_test_cpu);void __init housekeeping_init (void ) { -if (!housekeeping_flags) +enum hk_type type; + +if (!housekeeping.flags)return ;static_branch_enable (&housekeeping_overridden); -if (housekeeping_flags &HK_FLAG_TICK ) +if (housekeeping.flags &HK_FLAG_TICK )sched_tick_offload_init (); -/* We need at least one CPU to handle housekeeping work */ -WARN_ON_ONCE (cpumask_empty (housekeeping_mask)); +for_each_set_bit (type, &housekeeping.flags,HK_TYPE_MAX ) { +/* We need at least one CPU to handle housekeeping work */ +WARN_ON_ONCE (cpumask_empty (housekeeping.cpumasks[type])); + } } -static int __init housekeeping_setup (char *str,enum hk_flags flags) +static void __init housekeeping_setup_type (enum hk_type type, +cpumask_var_t housekeeping_staging) +{ + +alloc_bootmem_cpumask_var (&housekeeping.cpumasks[type]); +cpumask_copy (housekeeping.cpumasks[type], + housekeeping_staging); +} + +static int __init housekeeping_setup (char *str,unsigned long flags) {cpumask_var_t non_housekeeping_mask, housekeeping_staging;int err =0 ; -if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) { +if ((flags &HK_FLAG_TICK ) && !(housekeeping.flags &HK_FLAG_TICK )) {if (!IS_ENABLED (CONFIG_NO_HZ_FULL )) {pr_warn ("Housekeeping: nohz unsupported." " Build with CONFIG_NO_HZ_FULL\n" );
@@ -115,26 +133,41 @@ static int __init housekeeping_setup(char *str, enum hk_flags flags)if (!cpumask_intersects (cpu_present_mask, housekeeping_staging)) {__cpumask_set_cpu (smp_processor_id (), housekeeping_staging);__cpumask_clear_cpu (smp_processor_id (), non_housekeeping_mask); -if (!housekeeping_flags) { +if (!housekeeping.flags) {pr_warn ("Housekeeping: must include one present CPU, " "using boot CPU:%d\n" ,smp_processor_id ()); } } -if (!housekeeping_flags) { -alloc_bootmem_cpumask_var (&housekeeping_mask); -cpumask_copy (housekeeping_mask, housekeeping_staging); +if (!housekeeping.flags) { +/* First setup call ("nohz_full=" or "isolcpus=") */ +enum hk_type type; + +for_each_set_bit (type, &flags,HK_TYPE_MAX ) +housekeeping_setup_type (type, housekeeping_staging); }else { -if (!cpumask_equal (housekeeping_staging, housekeeping_mask)) { -pr_warn ("Housekeeping: nohz_full= must match isolcpus=\n" ); - goto free_housekeeping_staging; +/* Second setup call ("nohz_full=" after "isolcpus=" or the reverse) */ +enum hk_type type; +unsigned long iter_flags = flags & housekeeping.flags; + +for_each_set_bit (type, &iter_flags,HK_TYPE_MAX ) { +if (!cpumask_equal (housekeeping_staging, + housekeeping.cpumasks[type])) { +pr_warn ("Housekeeping: nohz_full= must match isolcpus=\n" ); + goto free_housekeeping_staging; + } } + + iter_flags = flags & ~housekeeping.flags; + +for_each_set_bit (type, &iter_flags,HK_TYPE_MAX ) +housekeeping_setup_type (type, housekeeping_staging); } -if ((flags &HK_FLAG_TICK ) && !(housekeeping_flags &HK_FLAG_TICK )) +if ((flags &HK_FLAG_TICK ) && !(housekeeping.flags &HK_FLAG_TICK ))tick_nohz_full_setup (non_housekeeping_mask); - housekeeping_flags |= flags; + housekeeping.flags |= flags; err =1 ; free_housekeeping_staging:
@@ -147,7 +180,7 @@ static int __init housekeeping_setup(char *str, enum hk_flags flags)static int __init housekeeping_nohz_full_setup (char *str) { -unsigned int flags; +unsigned long flags; flags =HK_FLAG_TICK |HK_FLAG_WQ |HK_FLAG_TIMER |HK_FLAG_RCU |HK_FLAG_MISC |HK_FLAG_KTHREAD ;
@@ -158,7 +191,7 @@ __setup("nohz_full=", housekeeping_nohz_full_setup);static int __init housekeeping_isolcpus_setup (char *str) { -unsigned int flags =0 ; +unsigned long flags =0 ;bool illegal = false;char *par;int len;
Signature · 1 line
2.25.1
Quoted 12 lines · Frederic Weisbecker
Hi,
To prepare for extending cpusets to control CPU isolation features (
nohz_full,rcu_nocbs, unbound timers, workqueues, kthreads affinity...), we need to split the globalhousekeeping_maskto one cpumask per isolation feature.Doing so is quite a chunk already so I'm working on that as a standalone patchset. Once that get merged, the next step is to finally provide a cpuset interface for one of these isolation features:
rcu_nocbcould be interesting to handle first asnohz_fulldepends on it.
Quoted 7 lines
git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git
isolation/splitHEAD:
2c07a16ff50d1e722babee28b926d70522e6bd3eThanks,
Frederic
Looks good.
Quoted 12 lines · On Tue, Jan 04, 2022 at 03:49:36PM +0100 Frederic Weisbecker wrote:
Hi,
To prepare for extending cpusets to control CPU isolation features (
nohz_full,rcu_nocbs, unbound timers, workqueues, kthreads affinity...), we need to split the globalhousekeeping_maskto one cpumask per isolation feature.Doing so is quite a chunk already so I'm working on that as a standalone patchset. Once that get merged, the next step is to finally provide a cpuset interface for one of these isolation features:
rcu_nocbcould be interesting to handle first asnohz_fulldepends on it.
Quoted 8 lines
git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git
isolation/splitHEAD:
2c07a16ff50d1e722babee28b926d70522e6bd3eThanks,
Frederic
---
This is excellent, thanks!
Quoted 33 lines
Frederic Weisbecker (8):
pci: DecoupleHK_FLAG_WQandHK_FLAG_DOMAINcpumask fetch
workqueue: DecoupleHK_FLAG_WQandHK_FLAG_DOMAINcpumask fetch
net: DecoupleHK_FLAG_WQandHK_FLAG_DOMAINcpumask fetch sched/isolation: Use single feature type while referring to housekeeping cpumask sched/isolation: Consolidate check for housekeeping minimum service sched/isolation: Consolidate error handling
sched/isolation: Fixhousekeeping_maskmemory leak sched/isolation: Split housekeeping cpumask per isolation featuresarch/x86/kernel/cpu/aperfmperf.c | 6 +- arch/x86/kvm/x86.c | 2 +- drivers/base/cpu.c | 2 +- drivers/pci/pci-driver.c | 21 +++-- include/linux/sched/isolation.h | 43 ++++++----- kernel/cgroup/cpuset.c | 6 +- kernel/cpu.c | 4 +- kernel/irq/cpuhotplug.c | 4 +- kernel/irq/manage.c | 4 +- kernel/kthread.c | 4 +- kernel/rcu/tasks.h | 2 +- kernel/rcu/tree_plugin.h | 2 +- kernel/sched/core.c | 12 +-- kernel/sched/fair.c | 10 +-- kernel/sched/isolation.c | 162 +++++++++++++++++++++++++-------------- kernel/sched/topology.c | 8 +- kernel/watchdog.c | 2 +- kernel/workqueue.c | 4 +- net/core/net-sysfs.c | 6 +- 19 files changed, 180 insertions(+), 124 deletions(-)
Hi,
Quoted 44 lines · Frederic Weisbecker
Hi,
To prepare for extending cpusets to control CPU isolation features (
nohz_full,rcu_nocbs, unbound timers, workqueues, kthreads affinity...), we need to split the globalhousekeeping_maskto one cpumask per isolation feature.Doing so is quite a chunk already so I'm working on that as a standalone patchset. Once that get merged, the next step is to finally provide a cpuset interface for one of these isolation features:
rcu_nocbcould be interesting to handle first asnohz_fulldepends on it.
git://git.kernel.org/pub/scm/linux/kernel/git/frederic/linux-dynticks.git
isolation/splitHEAD:
2c07a16ff50d1e722babee28b926d70522e6bd3eThanks,
Frederic
---Frederic Weisbecker (8):
pci: DecoupleHK_FLAG_WQandHK_FLAG_DOMAINcpumask fetch
workqueue: DecoupleHK_FLAG_WQandHK_FLAG_DOMAINcpumask fetch
net: DecoupleHK_FLAG_WQandHK_FLAG_DOMAINcpumask fetch sched/isolation: Use single feature type while referring to housekeeping cpumask sched/isolation: Consolidate check for housekeeping minimum service sched/isolation: Consolidate error handling
sched/isolation: Fixhousekeeping_maskmemory leak sched/isolation: Split housekeeping cpumask per isolation featuresarch/x86/kernel/cpu/aperfmperf.c | 6 +- arch/x86/kvm/x86.c | 2 +- drivers/base/cpu.c | 2 +- drivers/pci/pci-driver.c | 21 +++-- include/linux/sched/isolation.h | 43 ++++++----- kernel/cgroup/cpuset.c | 6 +- kernel/cpu.c | 4 +- kernel/irq/cpuhotplug.c | 4 +- kernel/irq/manage.c | 4 +- kernel/kthread.c | 4 +- kernel/rcu/tasks.h | 2 +- kernel/rcu/tree_plugin.h | 2 +-
Quoted 8 lines
kernel/sched/core.c | 12 +-- kernel/sched/fair.c | 10 +-- kernel/sched/isolation.c | 162 +++++++++++++++++++++++++-------------- kernel/sched/topology.c | 8 +- kernel/watchdog.c | 2 +- kernel/workqueue.c | 4 +- net/core/net-sysfs.c | 6 +- 19 files changed, 180 insertions(+), 124 deletions(-)
Thanks for this!
Best,
Juri