From 9a5322db46332a4ce42369e86f031b5e963d841c Mon Sep 17 00:00:00 2001 From: Bing Huang Date: Fri, 18 Nov 2022 11:42:08 +0800 Subject: sched/topology: Add __init for init_defrootdomain init_defrootdomain is only used in initialization Signed-off-by: Bing Huang Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Steven Rostedt (Google) Reviewed-by: Randy Dunlap Reviewed-by: Valentin Schneider Link: https://lkml.kernel.org/r/20221118034208.267330-1-huangbing775@126.com --- kernel/sched/topology.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c index 8739c2a5a54e..dea9fa39e7c0 100644 --- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -578,7 +578,7 @@ out: */ struct root_domain def_root_domain; -void init_defrootdomain(void) +void __init init_defrootdomain(void) { init_rootdomain(&def_root_domain); -- cgit v1.2.3 From 8ad075c2eb1f6b4b33436144ea1ef2619f3b6398 Mon Sep 17 00:00:00 2001 From: Josh Don Date: Wed, 16 Nov 2022 16:54:18 -0800 Subject: sched: Async unthrottling for cfs bandwidth CFS bandwidth currently distributes new runtime and unthrottles cfs_rq's inline in an hrtimer callback. Runtime distribution is a per-cpu operation, and unthrottling is a per-cgroup operation, since a tg walk is required. On machines with a large number of cpus and large cgroup hierarchies, this cpus*cgroups work can be too much to do in a single hrtimer callback: since IRQ are disabled, hard lockups may easily occur. Specifically, we've found this scalability issue on configurations with 256 cpus, O(1000) cgroups in the hierarchy being throttled, and high memory bandwidth usage. To fix this, we can instead unthrottle cfs_rq's asynchronously via a CSD. Each cpu is responsible for unthrottling itself, thus sharding the total work more fairly across the system, and avoiding hard lockups. Signed-off-by: Josh Don Signed-off-by: Peter Zijlstra (Intel) Link: https://lkml.kernel.org/r/20221117005418.3499691-1-joshdon@google.com --- kernel/sched/fair.c | 155 ++++++++++++++++++++++++++++++++++++++++++++++----- kernel/sched/sched.h | 8 +++ 2 files changed, 150 insertions(+), 13 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index c36aa54ae071..ea81d481f59b 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -5461,22 +5461,105 @@ unthrottle_throttle: resched_curr(rq); } -static void distribute_cfs_runtime(struct cfs_bandwidth *cfs_b) +#ifdef CONFIG_SMP +static void __cfsb_csd_unthrottle(void *arg) { - struct cfs_rq *cfs_rq; + struct cfs_rq *cursor, *tmp; + struct rq *rq = arg; + struct rq_flags rf; + + rq_lock(rq, &rf); + + /* + * Since we hold rq lock we're safe from concurrent manipulation of + * the CSD list. However, this RCU critical section annotates the + * fact that we pair with sched_free_group_rcu(), so that we cannot + * race with group being freed in the window between removing it + * from the list and advancing to the next entry in the list. + */ + rcu_read_lock(); + + list_for_each_entry_safe(cursor, tmp, &rq->cfsb_csd_list, + throttled_csd_list) { + list_del_init(&cursor->throttled_csd_list); + + if (cfs_rq_throttled(cursor)) + unthrottle_cfs_rq(cursor); + } + + rcu_read_unlock(); + + rq_unlock(rq, &rf); +} + +static inline void __unthrottle_cfs_rq_async(struct cfs_rq *cfs_rq) +{ + struct rq *rq = rq_of(cfs_rq); + bool first; + + if (rq == this_rq()) { + unthrottle_cfs_rq(cfs_rq); + return; + } + + /* Already enqueued */ + if (SCHED_WARN_ON(!list_empty(&cfs_rq->throttled_csd_list))) + return; + + first = list_empty(&rq->cfsb_csd_list); + list_add_tail(&cfs_rq->throttled_csd_list, &rq->cfsb_csd_list); + if (first) + smp_call_function_single_async(cpu_of(rq), &rq->cfsb_csd); +} +#else +static inline void __unthrottle_cfs_rq_async(struct cfs_rq *cfs_rq) +{ + unthrottle_cfs_rq(cfs_rq); +} +#endif + +static void unthrottle_cfs_rq_async(struct cfs_rq *cfs_rq) +{ + lockdep_assert_rq_held(rq_of(cfs_rq)); + + if (SCHED_WARN_ON(!cfs_rq_throttled(cfs_rq) || + cfs_rq->runtime_remaining <= 0)) + return; + + __unthrottle_cfs_rq_async(cfs_rq); +} + +static bool distribute_cfs_runtime(struct cfs_bandwidth *cfs_b) +{ + struct cfs_rq *local_unthrottle = NULL; + int this_cpu = smp_processor_id(); u64 runtime, remaining = 1; + bool throttled = false; + struct cfs_rq *cfs_rq; + struct rq_flags rf; + struct rq *rq; rcu_read_lock(); list_for_each_entry_rcu(cfs_rq, &cfs_b->throttled_cfs_rq, throttled_list) { - struct rq *rq = rq_of(cfs_rq); - struct rq_flags rf; + rq = rq_of(cfs_rq); + + if (!remaining) { + throttled = true; + break; + } rq_lock_irqsave(rq, &rf); if (!cfs_rq_throttled(cfs_rq)) goto next; - /* By the above check, this should never be true */ +#ifdef CONFIG_SMP + /* Already queued for async unthrottle */ + if (!list_empty(&cfs_rq->throttled_csd_list)) + goto next; +#endif + + /* By the above checks, this should never be true */ SCHED_WARN_ON(cfs_rq->runtime_remaining > 0); raw_spin_lock(&cfs_b->lock); @@ -5490,16 +5573,30 @@ static void distribute_cfs_runtime(struct cfs_bandwidth *cfs_b) cfs_rq->runtime_remaining += runtime; /* we check whether we're throttled above */ - if (cfs_rq->runtime_remaining > 0) - unthrottle_cfs_rq(cfs_rq); + if (cfs_rq->runtime_remaining > 0) { + if (cpu_of(rq) != this_cpu || + SCHED_WARN_ON(local_unthrottle)) + unthrottle_cfs_rq_async(cfs_rq); + else + local_unthrottle = cfs_rq; + } else { + throttled = true; + } next: rq_unlock_irqrestore(rq, &rf); - - if (!remaining) - break; } rcu_read_unlock(); + + if (local_unthrottle) { + rq = cpu_rq(this_cpu); + rq_lock_irqsave(rq, &rf); + if (cfs_rq_throttled(local_unthrottle)) + unthrottle_cfs_rq(local_unthrottle); + rq_unlock_irqrestore(rq, &rf); + } + + return throttled; } /* @@ -5544,10 +5641,8 @@ static int do_sched_cfs_period_timer(struct cfs_bandwidth *cfs_b, int overrun, u while (throttled && cfs_b->runtime > 0) { raw_spin_unlock_irqrestore(&cfs_b->lock, flags); /* we can't nest cfs_b->lock while distributing bandwidth */ - distribute_cfs_runtime(cfs_b); + throttled = distribute_cfs_runtime(cfs_b); raw_spin_lock_irqsave(&cfs_b->lock, flags); - - throttled = !list_empty(&cfs_b->throttled_cfs_rq); } /* @@ -5824,6 +5919,9 @@ static void init_cfs_rq_runtime(struct cfs_rq *cfs_rq) { cfs_rq->runtime_enabled = 0; INIT_LIST_HEAD(&cfs_rq->throttled_list); +#ifdef CONFIG_SMP + INIT_LIST_HEAD(&cfs_rq->throttled_csd_list); +#endif } void start_cfs_bandwidth(struct cfs_bandwidth *cfs_b) @@ -5840,12 +5938,38 @@ void start_cfs_bandwidth(struct cfs_bandwidth *cfs_b) static void destroy_cfs_bandwidth(struct cfs_bandwidth *cfs_b) { + int __maybe_unused i; + /* init_cfs_bandwidth() was not called */ if (!cfs_b->throttled_cfs_rq.next) return; hrtimer_cancel(&cfs_b->period_timer); hrtimer_cancel(&cfs_b->slack_timer); + + /* + * It is possible that we still have some cfs_rq's pending on a CSD + * list, though this race is very rare. In order for this to occur, we + * must have raced with the last task leaving the group while there + * exist throttled cfs_rq(s), and the period_timer must have queued the + * CSD item but the remote cpu has not yet processed it. To handle this, + * we can simply flush all pending CSD work inline here. We're + * guaranteed at this point that no additional cfs_rq of this group can + * join a CSD list. + */ +#ifdef CONFIG_SMP + for_each_possible_cpu(i) { + struct rq *rq = cpu_rq(i); + unsigned long flags; + + if (list_empty(&rq->cfsb_csd_list)) + continue; + + local_irq_save(flags); + __cfsb_csd_unthrottle(rq); + local_irq_restore(flags); + } +#endif } /* @@ -12474,6 +12598,11 @@ __init void init_sched_fair_class(void) for_each_possible_cpu(i) { zalloc_cpumask_var_node(&per_cpu(load_balance_mask, i), GFP_KERNEL, cpu_to_node(i)); zalloc_cpumask_var_node(&per_cpu(select_rq_mask, i), GFP_KERNEL, cpu_to_node(i)); + +#ifdef CONFIG_CFS_BANDWIDTH + INIT_CSD(&cpu_rq(i)->cfsb_csd, __cfsb_csd_unthrottle, cpu_rq(i)); + INIT_LIST_HEAD(&cpu_rq(i)->cfsb_csd_list); +#endif } open_softirq(SCHED_SOFTIRQ, run_rebalance_domains); diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 771f8ddb7053..b3d6e819127c 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -645,6 +645,9 @@ struct cfs_rq { int throttled; int throttle_count; struct list_head throttled_list; +#ifdef CONFIG_SMP + struct list_head throttled_csd_list; +#endif #endif /* CONFIG_CFS_BANDWIDTH */ #endif /* CONFIG_FAIR_GROUP_SCHED */ }; @@ -1154,6 +1157,11 @@ struct rq { /* Scratch cpumask to be temporarily used under rq_lock */ cpumask_var_t scratch_mask; + +#if defined(CONFIG_CFS_BANDWIDTH) && defined(CONFIG_SMP) + call_single_data_t cfsb_csd; + struct list_head cfsb_csd_list; +#endif }; #ifdef CONFIG_FAIR_GROUP_SCHED -- cgit v1.2.3 From ee3e3ac05c2631ce1f12d88c9cc9a092f8fe947a Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Tue, 22 Nov 2022 15:39:05 -0500 Subject: rseq: Introduce extensible rseq ABI Introduce the extensible rseq ABI, where the feature size supported by the kernel and the required alignment are communicated to user-space through ELF auxiliary vectors. This allows user-space to call rseq registration with a rseq_len of either 32 bytes for the original struct rseq size (which includes padding), or larger. If rseq_len is larger than 32 bytes, then it must be large enough to contain the feature size communicated to user-space through ELF auxiliary vectors. Signed-off-by: Mathieu Desnoyers Signed-off-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/r/20221122203932.231377-4-mathieu.desnoyers@efficios.com --- include/linux/sched.h | 4 ++++ kernel/ptrace.c | 2 +- kernel/rseq.c | 37 ++++++++++++++++++++++++++++++------- 3 files changed, 35 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/include/linux/sched.h b/include/linux/sched.h index 853d08f7562b..e0bc020a63a9 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1302,6 +1302,7 @@ struct task_struct { #ifdef CONFIG_RSEQ struct rseq __user *rseq; + u32 rseq_len; u32 rseq_sig; /* * RmW on rseq_event_mask must be performed atomically @@ -2352,10 +2353,12 @@ static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) { if (clone_flags & CLONE_VM) { t->rseq = NULL; + t->rseq_len = 0; t->rseq_sig = 0; t->rseq_event_mask = 0; } else { t->rseq = current->rseq; + t->rseq_len = current->rseq_len; t->rseq_sig = current->rseq_sig; t->rseq_event_mask = current->rseq_event_mask; } @@ -2364,6 +2367,7 @@ static inline void rseq_fork(struct task_struct *t, unsigned long clone_flags) static inline void rseq_execve(struct task_struct *t) { t->rseq = NULL; + t->rseq_len = 0; t->rseq_sig = 0; t->rseq_event_mask = 0; } diff --git a/kernel/ptrace.c b/kernel/ptrace.c index 54482193e1ed..0786450074c1 100644 --- a/kernel/ptrace.c +++ b/kernel/ptrace.c @@ -813,7 +813,7 @@ static long ptrace_get_rseq_configuration(struct task_struct *task, { struct ptrace_rseq_configuration conf = { .rseq_abi_pointer = (u64)(uintptr_t)task->rseq, - .rseq_abi_size = sizeof(*task->rseq), + .rseq_abi_size = task->rseq_len, .signature = task->rseq_sig, .flags = 0, }; diff --git a/kernel/rseq.c b/kernel/rseq.c index d38ab944105d..7962738455c9 100644 --- a/kernel/rseq.c +++ b/kernel/rseq.c @@ -18,6 +18,9 @@ #define CREATE_TRACE_POINTS #include +/* The original rseq structure size (including padding) is 32 bytes. */ +#define ORIG_RSEQ_SIZE 32 + #define RSEQ_CS_NO_RESTART_FLAGS (RSEQ_CS_FLAG_NO_RESTART_ON_PREEMPT | \ RSEQ_CS_FLAG_NO_RESTART_ON_SIGNAL | \ RSEQ_CS_FLAG_NO_RESTART_ON_MIGRATE) @@ -87,10 +90,15 @@ static int rseq_update_cpu_id(struct task_struct *t) u32 cpu_id = raw_smp_processor_id(); struct rseq __user *rseq = t->rseq; - if (!user_write_access_begin(rseq, sizeof(*rseq))) + if (!user_write_access_begin(rseq, t->rseq_len)) goto efault; unsafe_put_user(cpu_id, &rseq->cpu_id_start, efault_end); unsafe_put_user(cpu_id, &rseq->cpu_id, efault_end); + /* + * Additional feature fields added after ORIG_RSEQ_SIZE + * need to be conditionally updated only if + * t->rseq_len != ORIG_RSEQ_SIZE. + */ user_write_access_end(); trace_rseq_update(t); return 0; @@ -117,6 +125,11 @@ static int rseq_reset_rseq_cpu_id(struct task_struct *t) */ if (put_user(cpu_id, &t->rseq->cpu_id)) return -EFAULT; + /* + * Additional feature fields added after ORIG_RSEQ_SIZE + * need to be conditionally reset only if + * t->rseq_len != ORIG_RSEQ_SIZE. + */ return 0; } @@ -344,7 +357,7 @@ SYSCALL_DEFINE4(rseq, struct rseq __user *, rseq, u32, rseq_len, /* Unregister rseq for current thread. */ if (current->rseq != rseq || !current->rseq) return -EINVAL; - if (rseq_len != sizeof(*rseq)) + if (rseq_len != current->rseq_len) return -EINVAL; if (current->rseq_sig != sig) return -EPERM; @@ -353,6 +366,7 @@ SYSCALL_DEFINE4(rseq, struct rseq __user *, rseq, u32, rseq_len, return ret; current->rseq = NULL; current->rseq_sig = 0; + current->rseq_len = 0; return 0; } @@ -365,7 +379,7 @@ SYSCALL_DEFINE4(rseq, struct rseq __user *, rseq, u32, rseq_len, * the provided address differs from the prior * one. */ - if (current->rseq != rseq || rseq_len != sizeof(*rseq)) + if (current->rseq != rseq || rseq_len != current->rseq_len) return -EINVAL; if (current->rseq_sig != sig) return -EPERM; @@ -374,15 +388,24 @@ SYSCALL_DEFINE4(rseq, struct rseq __user *, rseq, u32, rseq_len, } /* - * If there was no rseq previously registered, - * ensure the provided rseq is properly aligned and valid. + * If there was no rseq previously registered, ensure the provided rseq + * is properly aligned, as communcated to user-space through the ELF + * auxiliary vector AT_RSEQ_ALIGN. If rseq_len is the original rseq + * size, the required alignment is the original struct rseq alignment. + * + * In order to be valid, rseq_len is either the original rseq size, or + * large enough to contain all supported fields, as communicated to + * user-space through the ELF auxiliary vector AT_RSEQ_FEATURE_SIZE. */ - if (!IS_ALIGNED((unsigned long)rseq, __alignof__(*rseq)) || - rseq_len != sizeof(*rseq)) + if (rseq_len < ORIG_RSEQ_SIZE || + (rseq_len == ORIG_RSEQ_SIZE && !IS_ALIGNED((unsigned long)rseq, ORIG_RSEQ_SIZE)) || + (rseq_len != ORIG_RSEQ_SIZE && (!IS_ALIGNED((unsigned long)rseq, __alignof__(*rseq)) || + rseq_len < offsetof(struct rseq, end)))) return -EINVAL; if (!access_ok(rseq, rseq_len)) return -EFAULT; current->rseq = rseq; + current->rseq_len = rseq_len; current->rseq_sig = sig; /* * If rseq was previously inactive, and has just been -- cgit v1.2.3 From cbae6bac29a8c5cf2f1cb5c6bce35af00cec164b Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Tue, 22 Nov 2022 15:39:06 -0500 Subject: rseq: Extend struct rseq with numa node id Adding the NUMA node id to struct rseq is a straightforward thing to do, and a good way to figure out if anything in the user-space ecosystem prevents extending struct rseq. This NUMA node id field allows memory allocators such as tcmalloc to take advantage of fast access to the current NUMA node id to perform NUMA-aware memory allocation. It can also be useful for implementing fast-paths for NUMA-aware user-space mutexes. It also allows implementing getcpu(2) purely in user-space. Signed-off-by: Mathieu Desnoyers Signed-off-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/r/20221122203932.231377-5-mathieu.desnoyers@efficios.com --- include/trace/events/rseq.h | 4 +++- include/uapi/linux/rseq.h | 8 ++++++++ kernel/rseq.c | 19 +++++++++++++------ 3 files changed, 24 insertions(+), 7 deletions(-) (limited to 'kernel') diff --git a/include/trace/events/rseq.h b/include/trace/events/rseq.h index a04a64bc1a00..dde7a359b4ef 100644 --- a/include/trace/events/rseq.h +++ b/include/trace/events/rseq.h @@ -16,13 +16,15 @@ TRACE_EVENT(rseq_update, TP_STRUCT__entry( __field(s32, cpu_id) + __field(s32, node_id) ), TP_fast_assign( __entry->cpu_id = raw_smp_processor_id(); + __entry->node_id = cpu_to_node(__entry->cpu_id); ), - TP_printk("cpu_id=%d", __entry->cpu_id) + TP_printk("cpu_id=%d node_id=%d", __entry->cpu_id, __entry->node_id) ); TRACE_EVENT(rseq_ip_fixup, diff --git a/include/uapi/linux/rseq.h b/include/uapi/linux/rseq.h index 05d3c4cdeb40..1cb90a435c5c 100644 --- a/include/uapi/linux/rseq.h +++ b/include/uapi/linux/rseq.h @@ -131,6 +131,14 @@ struct rseq { */ __u32 flags; + /* + * Restartable sequences node_id field. Updated by the kernel. Read by + * user-space with single-copy atomicity semantics. This field should + * only be read by the thread which registered this data structure. + * Aligned on 32-bit. Contains the current NUMA node ID. + */ + __u32 node_id; + /* * Flexible array member at end of structure, after last feature field. */ diff --git a/kernel/rseq.c b/kernel/rseq.c index 7962738455c9..193cfcc94ddc 100644 --- a/kernel/rseq.c +++ b/kernel/rseq.c @@ -85,15 +85,17 @@ * F1. */ -static int rseq_update_cpu_id(struct task_struct *t) +static int rseq_update_cpu_node_id(struct task_struct *t) { - u32 cpu_id = raw_smp_processor_id(); struct rseq __user *rseq = t->rseq; + u32 cpu_id = raw_smp_processor_id(); + u32 node_id = cpu_to_node(cpu_id); if (!user_write_access_begin(rseq, t->rseq_len)) goto efault; unsafe_put_user(cpu_id, &rseq->cpu_id_start, efault_end); unsafe_put_user(cpu_id, &rseq->cpu_id, efault_end); + unsafe_put_user(node_id, &rseq->node_id, efault_end); /* * Additional feature fields added after ORIG_RSEQ_SIZE * need to be conditionally updated only if @@ -109,9 +111,9 @@ efault: return -EFAULT; } -static int rseq_reset_rseq_cpu_id(struct task_struct *t) +static int rseq_reset_rseq_cpu_node_id(struct task_struct *t) { - u32 cpu_id_start = 0, cpu_id = RSEQ_CPU_ID_UNINITIALIZED; + u32 cpu_id_start = 0, cpu_id = RSEQ_CPU_ID_UNINITIALIZED, node_id = 0; /* * Reset cpu_id_start to its initial state (0). @@ -125,6 +127,11 @@ static int rseq_reset_rseq_cpu_id(struct task_struct *t) */ if (put_user(cpu_id, &t->rseq->cpu_id)) return -EFAULT; + /* + * Reset node_id to its initial state (0). + */ + if (put_user(node_id, &t->rseq->node_id)) + return -EFAULT; /* * Additional feature fields added after ORIG_RSEQ_SIZE * need to be conditionally reset only if @@ -314,7 +321,7 @@ void __rseq_handle_notify_resume(struct ksignal *ksig, struct pt_regs *regs) if (unlikely(ret < 0)) goto error; } - if (unlikely(rseq_update_cpu_id(t))) + if (unlikely(rseq_update_cpu_node_id(t))) goto error; return; @@ -361,7 +368,7 @@ SYSCALL_DEFINE4(rseq, struct rseq __user *, rseq, u32, rseq_len, return -EINVAL; if (current->rseq_sig != sig) return -EPERM; - ret = rseq_reset_rseq_cpu_id(current); + ret = rseq_reset_rseq_cpu_node_id(current); if (ret) return ret; current->rseq = NULL; -- cgit v1.2.3 From af7f588d8f7355bc4298dd1962d7826358fc95f0 Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Tue, 22 Nov 2022 15:39:09 -0500 Subject: sched: Introduce per-memory-map concurrency ID MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit This feature allows the scheduler to expose a per-memory map concurrency ID to user-space. This concurrency ID is within the possible cpus range, and is temporarily (and uniquely) assigned while threads are actively running within a memory map. If a memory map has fewer threads than cores, or is limited to run on few cores concurrently through sched affinity or cgroup cpusets, the concurrency IDs will be values close to 0, thus allowing efficient use of user-space memory for per-cpu data structures. This feature is meant to be exposed by a new rseq thread area field. The primary purpose of this feature is to do the heavy-lifting needed by memory allocators to allow them to use per-cpu data structures efficiently in the following situations: - Single-threaded applications, - Multi-threaded applications on large systems (many cores) with limited cpu affinity mask, - Multi-threaded applications on large systems (many cores) with restricted cgroup cpuset per container. One of the key concern from scheduler maintainers is the overhead associated with additional spin locks or atomic operations in the scheduler fast-path. This is why the following optimization is implemented. On context switch between threads belonging to the same memory map, transfer the mm_cid from prev to next without any atomic ops. This takes care of use-cases involving frequent context switch between threads belonging to the same memory map. Additional optimizations can be done if the spin locks added when context switching between threads belonging to different memory maps end up being a performance bottleneck. Those are left out of this patch though. A performance impact would have to be clearly demonstrated to justify the added complexity. The credit goes to Paul Turner (Google) for the original virtual cpu id idea. This feature is implemented based on the discussions with Paul Turner and Peter Oskolkov (Google), but I took the liberty to implement scheduler fast-path optimizations and my own NUMA-awareness scheme. The rumor has it that Google have been running a rseq vcpu_id extension internally in production for a year. The tcmalloc source code indeed has comments hinting at a vcpu_id prototype extension to the rseq system call [1]. The following benchmarks do not show any significant overhead added to the scheduler context switch by this feature: * perf bench sched messaging (process) Baseline: 86.5±0.3 ms With mm_cid: 86.7±2.6 ms * perf bench sched messaging (threaded) Baseline: 84.3±3.0 ms With mm_cid: 84.7±2.6 ms * hackbench (process) Baseline: 82.9±2.7 ms With mm_cid: 82.9±2.9 ms * hackbench (threaded) Baseline: 85.2±2.6 ms With mm_cid: 84.4±2.9 ms [1] https://github.com/google/tcmalloc/blob/master/tcmalloc/internal/linux_syscall_support.h#L26 Signed-off-by: Mathieu Desnoyers Signed-off-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/r/20221122203932.231377-8-mathieu.desnoyers@efficios.com --- fs/exec.c | 4 ++++ include/linux/mm.h | 25 +++++++++++++++++++++ include/linux/mm_types.h | 43 ++++++++++++++++++++++++++++++++++- include/linux/sched.h | 5 +++++ init/Kconfig | 4 ++++ kernel/fork.c | 8 ++++++- kernel/sched/core.c | 51 ++++++++++++++++++++++++++++++++++++++++++ kernel/sched/sched.h | 58 ++++++++++++++++++++++++++++++++++++++++++++++++ kernel/signal.c | 2 ++ 9 files changed, 198 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/fs/exec.c b/fs/exec.c index ab913243a367..58f16312b983 100644 --- a/fs/exec.c +++ b/fs/exec.c @@ -1010,6 +1010,7 @@ static int exec_mmap(struct mm_struct *mm) active_mm = tsk->active_mm; tsk->active_mm = mm; tsk->mm = mm; + mm_init_cid(mm); /* * This prevents preemption while active_mm is being loaded and * it and mm are being updated, which could cause problems for @@ -1822,6 +1823,7 @@ static int bprm_execve(struct linux_binprm *bprm, */ check_unsafe_exec(bprm); current->in_execve = 1; + sched_mm_cid_before_execve(current); file = do_open_execat(fd, filename, flags); retval = PTR_ERR(file); @@ -1852,6 +1854,7 @@ static int bprm_execve(struct linux_binprm *bprm, if (retval < 0) goto out; + sched_mm_cid_after_execve(current); /* execve succeeded */ current->fs->in_exec = 0; current->in_execve = 0; @@ -1871,6 +1874,7 @@ out: force_fatal_sig(SIGSEGV); out_unmark: + sched_mm_cid_after_execve(current); current->fs->in_exec = 0; current->in_execve = 0; diff --git a/include/linux/mm.h b/include/linux/mm.h index f3f196e4d66d..cf008c26a883 100644 --- a/include/linux/mm.h +++ b/include/linux/mm.h @@ -1976,6 +1976,31 @@ struct zap_details { /* Set in unmap_vmas() to indicate a final unmap call. Only used by hugetlb */ #define ZAP_FLAG_UNMAP ((__force zap_flags_t) BIT(1)) +#ifdef CONFIG_SCHED_MM_CID +void sched_mm_cid_before_execve(struct task_struct *t); +void sched_mm_cid_after_execve(struct task_struct *t); +void sched_mm_cid_fork(struct task_struct *t); +void sched_mm_cid_exit_signals(struct task_struct *t); +static inline int task_mm_cid(struct task_struct *t) +{ + return t->mm_cid; +} +#else +static inline void sched_mm_cid_before_execve(struct task_struct *t) { } +static inline void sched_mm_cid_after_execve(struct task_struct *t) { } +static inline void sched_mm_cid_fork(struct task_struct *t) { } +static inline void sched_mm_cid_exit_signals(struct task_struct *t) { } +static inline int task_mm_cid(struct task_struct *t) +{ + /* + * Use the processor id as a fall-back when the mm cid feature is + * disabled. This provides functional per-cpu data structure accesses + * in user-space, althrough it won't provide the memory usage benefits. + */ + return raw_smp_processor_id(); +} +#endif + #ifdef CONFIG_MMU extern bool can_do_mlock(void); #else diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h index 3b8475007734..1c3bf76063d2 100644 --- a/include/linux/mm_types.h +++ b/include/linux/mm_types.h @@ -645,7 +645,18 @@ struct mm_struct { * &struct mm_struct is freed. */ atomic_t mm_count; - +#ifdef CONFIG_SCHED_MM_CID + /** + * @cid_lock: Protect cid bitmap updates vs lookups. + * + * Prevent situations where updates to the cid bitmap happen + * concurrently with lookups. Those can lead to situations + * where a lookup cannot find a free bit simply because it was + * unlucky enough to load, non-atomically, bitmap words as they + * were being concurrently updated by the updaters. + */ + raw_spinlock_t cid_lock; +#endif #ifdef CONFIG_MMU atomic_long_t pgtables_bytes; /* PTE page table pages */ #endif @@ -909,6 +920,36 @@ static inline void vma_iter_init(struct vma_iterator *vmi, vmi->mas.node = MAS_START; } +#ifdef CONFIG_SCHED_MM_CID +/* Accessor for struct mm_struct's cidmask. */ +static inline cpumask_t *mm_cidmask(struct mm_struct *mm) +{ + unsigned long cid_bitmap = (unsigned long)mm; + + cid_bitmap += offsetof(struct mm_struct, cpu_bitmap); + /* Skip cpu_bitmap */ + cid_bitmap += cpumask_size(); + return (struct cpumask *)cid_bitmap; +} + +static inline void mm_init_cid(struct mm_struct *mm) +{ + raw_spin_lock_init(&mm->cid_lock); + cpumask_clear(mm_cidmask(mm)); +} + +static inline unsigned int mm_cid_size(void) +{ + return cpumask_size(); +} +#else /* CONFIG_SCHED_MM_CID */ +static inline void mm_init_cid(struct mm_struct *mm) { } +static inline unsigned int mm_cid_size(void) +{ + return 0; +} +#endif /* CONFIG_SCHED_MM_CID */ + struct mmu_gather; extern void tlb_gather_mmu(struct mmu_gather *tlb, struct mm_struct *mm); extern void tlb_gather_mmu_fullmm(struct mmu_gather *tlb, struct mm_struct *mm); diff --git a/include/linux/sched.h b/include/linux/sched.h index e0bc020a63a9..4df2b3e76b30 100644 --- a/include/linux/sched.h +++ b/include/linux/sched.h @@ -1311,6 +1311,11 @@ struct task_struct { unsigned long rseq_event_mask; #endif +#ifdef CONFIG_SCHED_MM_CID + int mm_cid; /* Current cid in mm */ + int mm_cid_active; /* Whether cid bitmap is active */ +#endif + struct tlbflush_unmap_batch tlb_ubc; union { diff --git a/init/Kconfig b/init/Kconfig index 7e5c3ddc341d..1ce960aa453e 100644 --- a/init/Kconfig +++ b/init/Kconfig @@ -1041,6 +1041,10 @@ config RT_GROUP_SCHED endif #CGROUP_SCHED +config SCHED_MM_CID + def_bool y + depends on SMP && RSEQ + config UCLAMP_TASK_GROUP bool "Utilization clamping per group of tasks" depends on CGROUP_SCHED diff --git a/kernel/fork.c b/kernel/fork.c index 9f7fe3541897..82b2b5846aae 100644 --- a/kernel/fork.c +++ b/kernel/fork.c @@ -1060,6 +1060,10 @@ static struct task_struct *dup_task_struct(struct task_struct *orig, int node) tsk->reported_split_lock = 0; #endif +#ifdef CONFIG_SCHED_MM_CID + tsk->mm_cid = -1; + tsk->mm_cid_active = 0; +#endif return tsk; free_stack: @@ -1169,6 +1173,7 @@ static struct mm_struct *mm_init(struct mm_struct *mm, struct task_struct *p, mm->user_ns = get_user_ns(user_ns); lru_gen_init_mm(mm); + mm_init_cid(mm); return mm; fail_pcpu: @@ -1601,6 +1606,7 @@ static int copy_mm(unsigned long clone_flags, struct task_struct *tsk) tsk->mm = mm; tsk->active_mm = mm; + sched_mm_cid_fork(tsk); return 0; } @@ -3034,7 +3040,7 @@ void __init mm_cache_init(void) * dynamically sized based on the maximum CPU number this system * can have, taking hotplug into account (nr_cpu_ids). */ - mm_size = sizeof(struct mm_struct) + cpumask_size(); + mm_size = sizeof(struct mm_struct) + cpumask_size() + mm_cid_size(); mm_cachep = kmem_cache_create_usercopy("mm_struct", mm_size, ARCH_MIN_MMSTRUCT_ALIGN, diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 25b582b6ee5f..75830b7dee8f 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -5052,6 +5052,7 @@ prepare_task_switch(struct rq *rq, struct task_struct *prev, sched_info_switch(rq, prev, next); perf_event_task_sched_out(prev, next); rseq_preempt(prev); + switch_mm_cid(prev, next); fire_sched_out_preempt_notifiers(prev, next); kmap_local_sched_out(); prepare_task(next); @@ -11305,3 +11306,53 @@ void call_trace_sched_update_nr_running(struct rq *rq, int count) { trace_sched_update_nr_running_tp(rq, count); } + +#ifdef CONFIG_SCHED_MM_CID +void sched_mm_cid_exit_signals(struct task_struct *t) +{ + struct mm_struct *mm = t->mm; + unsigned long flags; + + if (!mm) + return; + local_irq_save(flags); + mm_cid_put(mm, t->mm_cid); + t->mm_cid = -1; + t->mm_cid_active = 0; + local_irq_restore(flags); +} + +void sched_mm_cid_before_execve(struct task_struct *t) +{ + struct mm_struct *mm = t->mm; + unsigned long flags; + + if (!mm) + return; + local_irq_save(flags); + mm_cid_put(mm, t->mm_cid); + t->mm_cid = -1; + t->mm_cid_active = 0; + local_irq_restore(flags); +} + +void sched_mm_cid_after_execve(struct task_struct *t) +{ + struct mm_struct *mm = t->mm; + unsigned long flags; + + WARN_ON_ONCE((t->flags & PF_KTHREAD) || !t->mm); + + local_irq_save(flags); + t->mm_cid = mm_cid_get(mm); + t->mm_cid_active = 1; + local_irq_restore(flags); + rseq_set_notify_resume(t); +} + +void sched_mm_cid_fork(struct task_struct *t) +{ + WARN_ON_ONCE((t->flags & PF_KTHREAD) || !t->mm || t->mm_cid != -1); + t->mm_cid_active = 1; +} +#endif diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index b3d6e819127c..c2d7467fdde1 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -3269,4 +3269,62 @@ static inline void update_current_exec_runtime(struct task_struct *curr, cgroup_account_cputime(curr, delta_exec); } +#ifdef CONFIG_SCHED_MM_CID +static inline int __mm_cid_get(struct mm_struct *mm) +{ + struct cpumask *cpumask; + int cid; + + cpumask = mm_cidmask(mm); + cid = cpumask_first_zero(cpumask); + if (cid >= nr_cpu_ids) + return -1; + __cpumask_set_cpu(cid, cpumask); + return cid; +} + +static inline void mm_cid_put(struct mm_struct *mm, int cid) +{ + lockdep_assert_irqs_disabled(); + if (cid < 0) + return; + raw_spin_lock(&mm->cid_lock); + __cpumask_clear_cpu(cid, mm_cidmask(mm)); + raw_spin_unlock(&mm->cid_lock); +} + +static inline int mm_cid_get(struct mm_struct *mm) +{ + int ret; + + lockdep_assert_irqs_disabled(); + raw_spin_lock(&mm->cid_lock); + ret = __mm_cid_get(mm); + raw_spin_unlock(&mm->cid_lock); + return ret; +} + +static inline void switch_mm_cid(struct task_struct *prev, struct task_struct *next) +{ + if (prev->mm_cid_active) { + if (next->mm_cid_active && next->mm == prev->mm) { + /* + * Context switch between threads in same mm, hand over + * the mm_cid from prev to next. + */ + next->mm_cid = prev->mm_cid; + prev->mm_cid = -1; + return; + } + mm_cid_put(prev->mm, prev->mm_cid); + prev->mm_cid = -1; + } + if (next->mm_cid_active) + next->mm_cid = mm_cid_get(next->mm); +} + +#else +static inline void switch_mm_cid(struct task_struct *prev, struct task_struct *next) { } +#endif + #endif /* _KERNEL_SCHED_SCHED_H */ diff --git a/kernel/signal.c b/kernel/signal.c index ae26da61c4d9..8cb28f1df294 100644 --- a/kernel/signal.c +++ b/kernel/signal.c @@ -2951,6 +2951,7 @@ void exit_signals(struct task_struct *tsk) cgroup_threadgroup_change_begin(tsk); if (thread_group_empty(tsk) || (tsk->signal->flags & SIGNAL_GROUP_EXIT)) { + sched_mm_cid_exit_signals(tsk); tsk->flags |= PF_EXITING; cgroup_threadgroup_change_end(tsk); return; @@ -2961,6 +2962,7 @@ void exit_signals(struct task_struct *tsk) * From now this task is not visible for group-wide signals, * see wants_signal(), do_signal_stop(). */ + sched_mm_cid_exit_signals(tsk); tsk->flags |= PF_EXITING; cgroup_threadgroup_change_end(tsk); -- cgit v1.2.3 From f7b01bb0b57f994a44ea6368536b59062b796381 Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Tue, 22 Nov 2022 15:39:10 -0500 Subject: rseq: Extend struct rseq with per-memory-map concurrency ID If a memory map has fewer threads than there are cores on the system, or is limited to run on few cores concurrently through sched affinity or cgroup cpusets, the concurrency IDs will be values close to 0, thus allowing efficient use of user-space memory for per-cpu data structures. Signed-off-by: Mathieu Desnoyers Signed-off-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/r/20221122203932.231377-9-mathieu.desnoyers@efficios.com --- include/uapi/linux/rseq.h | 9 +++++++++ kernel/rseq.c | 11 ++++++++++- 2 files changed, 19 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/include/uapi/linux/rseq.h b/include/uapi/linux/rseq.h index 1cb90a435c5c..c233aae5eac9 100644 --- a/include/uapi/linux/rseq.h +++ b/include/uapi/linux/rseq.h @@ -139,6 +139,15 @@ struct rseq { */ __u32 node_id; + /* + * Restartable sequences mm_cid field. Updated by the kernel. Read by + * user-space with single-copy atomicity semantics. This field should + * only be read by the thread which registered this data structure. + * Aligned on 32-bit. Contains the current thread's concurrency ID + * (allocated uniquely within a memory map). + */ + __u32 mm_cid; + /* * Flexible array member at end of structure, after last feature field. */ diff --git a/kernel/rseq.c b/kernel/rseq.c index 193cfcc94ddc..9de6e35fe679 100644 --- a/kernel/rseq.c +++ b/kernel/rseq.c @@ -90,12 +90,15 @@ static int rseq_update_cpu_node_id(struct task_struct *t) struct rseq __user *rseq = t->rseq; u32 cpu_id = raw_smp_processor_id(); u32 node_id = cpu_to_node(cpu_id); + u32 mm_cid = task_mm_cid(t); + WARN_ON_ONCE((int) mm_cid < 0); if (!user_write_access_begin(rseq, t->rseq_len)) goto efault; unsafe_put_user(cpu_id, &rseq->cpu_id_start, efault_end); unsafe_put_user(cpu_id, &rseq->cpu_id, efault_end); unsafe_put_user(node_id, &rseq->node_id, efault_end); + unsafe_put_user(mm_cid, &rseq->mm_cid, efault_end); /* * Additional feature fields added after ORIG_RSEQ_SIZE * need to be conditionally updated only if @@ -113,7 +116,8 @@ efault: static int rseq_reset_rseq_cpu_node_id(struct task_struct *t) { - u32 cpu_id_start = 0, cpu_id = RSEQ_CPU_ID_UNINITIALIZED, node_id = 0; + u32 cpu_id_start = 0, cpu_id = RSEQ_CPU_ID_UNINITIALIZED, node_id = 0, + mm_cid = 0; /* * Reset cpu_id_start to its initial state (0). @@ -132,6 +136,11 @@ static int rseq_reset_rseq_cpu_node_id(struct task_struct *t) */ if (put_user(node_id, &t->rseq->node_id)) return -EFAULT; + /* + * Reset mm_cid to its initial state (0). + */ + if (put_user(mm_cid, &t->rseq->mm_cid)) + return -EFAULT; /* * Additional feature fields added after ORIG_RSEQ_SIZE * need to be conditionally reset only if -- cgit v1.2.3 From 904cbab71dda1689d41a240541179f21ff433c40 Mon Sep 17 00:00:00 2001 From: "Matthew Wilcox (Oracle)" Date: Mon, 12 Dec 2022 14:49:46 +0000 Subject: sched: Make const-safe With a modified container_of() that preserves constness, the compiler finds some pointers which should have been marked as const. task_of() also needs to become const-preserving for the !FAIR_GROUP_SCHED case so that cfs_rq_of() can take a const argument. No change to generated code. Signed-off-by: Matthew Wilcox (Oracle) Signed-off-by: Peter Zijlstra (Intel) Link: https://lkml.kernel.org/r/20221212144946.2657785-1-willy@infradead.org --- kernel/sched/core.c | 8 +++++--- kernel/sched/fair.c | 16 +++++++++------- kernel/sched/sched.h | 22 ++++++++++------------ 3 files changed, 24 insertions(+), 22 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 75830b7dee8f..1f3259c5f1c2 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -152,7 +152,7 @@ __read_mostly int scheduler_running; DEFINE_STATIC_KEY_FALSE(__sched_core_enabled); /* kernel prio, less is more */ -static inline int __task_prio(struct task_struct *p) +static inline int __task_prio(const struct task_struct *p) { if (p->sched_class == &stop_sched_class) /* trumps deadline */ return -2; @@ -174,7 +174,8 @@ static inline int __task_prio(struct task_struct *p) */ /* real prio, less is less */ -static inline bool prio_less(struct task_struct *a, struct task_struct *b, bool in_fi) +static inline bool prio_less(const struct task_struct *a, + const struct task_struct *b, bool in_fi) { int pa = __task_prio(a), pb = __task_prio(b); @@ -194,7 +195,8 @@ static inline bool prio_less(struct task_struct *a, struct task_struct *b, bool return false; } -static inline bool __sched_core_less(struct task_struct *a, struct task_struct *b) +static inline bool __sched_core_less(const struct task_struct *a, + const struct task_struct *b) { if (a->core_cookie < b->core_cookie) return true; diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index ea81d481f59b..d0e2a48a882b 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -468,7 +468,7 @@ is_same_group(struct sched_entity *se, struct sched_entity *pse) return NULL; } -static inline struct sched_entity *parent_entity(struct sched_entity *se) +static inline struct sched_entity *parent_entity(const struct sched_entity *se) { return se->parent; } @@ -595,8 +595,8 @@ static inline u64 min_vruntime(u64 min_vruntime, u64 vruntime) return min_vruntime; } -static inline bool entity_before(struct sched_entity *a, - struct sched_entity *b) +static inline bool entity_before(const struct sched_entity *a, + const struct sched_entity *b) { return (s64)(a->vruntime - b->vruntime) < 0; } @@ -11852,7 +11852,8 @@ static inline void task_tick_core(struct rq *rq, struct task_struct *curr) /* * se_fi_update - Update the cfs_rq->min_vruntime_fi in a CFS hierarchy if needed. */ -static void se_fi_update(struct sched_entity *se, unsigned int fi_seq, bool forceidle) +static void se_fi_update(const struct sched_entity *se, unsigned int fi_seq, + bool forceidle) { for_each_sched_entity(se) { struct cfs_rq *cfs_rq = cfs_rq_of(se); @@ -11877,11 +11878,12 @@ void task_vruntime_update(struct rq *rq, struct task_struct *p, bool in_fi) se_fi_update(se, rq->core->core_forceidle_seq, in_fi); } -bool cfs_prio_less(struct task_struct *a, struct task_struct *b, bool in_fi) +bool cfs_prio_less(const struct task_struct *a, const struct task_struct *b, + bool in_fi) { struct rq *rq = task_rq(a); - struct sched_entity *sea = &a->se; - struct sched_entity *seb = &b->se; + const struct sched_entity *sea = &a->se; + const struct sched_entity *seb = &b->se; struct cfs_rq *cfs_rqa; struct cfs_rq *cfs_rqb; s64 delta; diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index c2d7467fdde1..1072502976df 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -248,7 +248,7 @@ static inline void update_avg(u64 *avg, u64 sample) #define SCHED_DL_FLAGS (SCHED_FLAG_RECLAIM | SCHED_FLAG_DL_OVERRUN | SCHED_FLAG_SUGOV) -static inline bool dl_entity_is_special(struct sched_dl_entity *dl_se) +static inline bool dl_entity_is_special(const struct sched_dl_entity *dl_se) { #ifdef CONFIG_CPU_FREQ_GOV_SCHEDUTIL return unlikely(dl_se->flags & SCHED_FLAG_SUGOV); @@ -260,8 +260,8 @@ static inline bool dl_entity_is_special(struct sched_dl_entity *dl_se) /* * Tells if entity @a should preempt entity @b. */ -static inline bool -dl_entity_preempt(struct sched_dl_entity *a, struct sched_dl_entity *b) +static inline bool dl_entity_preempt(const struct sched_dl_entity *a, + const struct sched_dl_entity *b) { return dl_entity_is_special(a) || dl_time_before(a->deadline, b->deadline); @@ -1244,7 +1244,8 @@ static inline raw_spinlock_t *__rq_lockp(struct rq *rq) return &rq->__lock; } -bool cfs_prio_less(struct task_struct *a, struct task_struct *b, bool fi); +bool cfs_prio_less(const struct task_struct *a, const struct task_struct *b, + bool fi); /* * Helpers to check if the CPU's core cookie matches with the task's cookie @@ -1423,7 +1424,7 @@ static inline struct cfs_rq *task_cfs_rq(struct task_struct *p) } /* runqueue on which this entity is (to be) queued */ -static inline struct cfs_rq *cfs_rq_of(struct sched_entity *se) +static inline struct cfs_rq *cfs_rq_of(const struct sched_entity *se) { return se->cfs_rq; } @@ -1436,19 +1437,16 @@ static inline struct cfs_rq *group_cfs_rq(struct sched_entity *grp) #else -static inline struct task_struct *task_of(struct sched_entity *se) -{ - return container_of(se, struct task_struct, se); -} +#define task_of(_se) container_of(_se, struct task_struct, se) -static inline struct cfs_rq *task_cfs_rq(struct task_struct *p) +static inline struct cfs_rq *task_cfs_rq(const struct task_struct *p) { return &task_rq(p)->cfs; } -static inline struct cfs_rq *cfs_rq_of(struct sched_entity *se) +static inline struct cfs_rq *cfs_rq_of(const struct sched_entity *se) { - struct task_struct *p = task_of(se); + const struct task_struct *p = task_of(se); struct rq *rq = task_rq(p); return &rq->cfs; -- cgit v1.2.3 From feaed76376fd612b9450ad1251aec20a63ccfe92 Mon Sep 17 00:00:00 2001 From: Hao Jia Date: Fri, 16 Dec 2022 14:24:05 +0800 Subject: sched/numa: Stop an exhastive search if an idle core is found In update_numa_stats() we try to find an idle cpu on the NUMA node, preferably an idle core. we can stop looking for the next idle core or idle cpu after finding an idle core. But we can't stop the whole loop of scanning the CPU, because we need to calculate approximate NUMA stats at a point in time. For example, the src and dst nr_running is needed by task_numa_find_cpu(). Signed-off-by: Hao Jia Signed-off-by: Peter Zijlstra (Intel) Acked-by: Mel Gorman Link: https://lore.kernel.org/r/20221216062406.7812-2-jiahao.os@bytedance.com --- kernel/sched/fair.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index d0e2a48a882b..aaff2092ffdf 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -1804,7 +1804,7 @@ static void update_numa_stats(struct task_numa_env *env, ns->nr_running += rq->cfs.h_nr_running; ns->compute_capacity += capacity_of(cpu); - if (find_idle && !rq->nr_running && idle_cpu(cpu)) { + if (find_idle && idle_core < 0 && !rq->nr_running && idle_cpu(cpu)) { if (READ_ONCE(rq->numa_migrate_on) || !cpumask_test_cpu(cpu, env->p->cpus_ptr)) continue; -- cgit v1.2.3 From 8589018acc65e5ddfd111f0a7ee85f9afde3a830 Mon Sep 17 00:00:00 2001 From: Hao Jia Date: Fri, 16 Dec 2022 14:24:06 +0800 Subject: sched/core: Adjusting the order of scanning CPU When select_idle_capacity() starts scanning for an idle CPU, it starts with target CPU that has already been checked in select_idle_sibling(). So we start checking from the next CPU and try the target CPU at the end. Similarly for task_numa_assign(), we have just checked numa_migrate_on of dst_cpu, so start from the next CPU. This also works for steal_cookie_task(), the first scan must fail and start directly from the next one. Signed-off-by: Hao Jia Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Vincent Guittot Acked-by: Mel Gorman Link: https://lore.kernel.org/r/20221216062406.7812-3-jiahao.os@bytedance.com --- kernel/sched/core.c | 2 +- kernel/sched/fair.c | 4 ++-- 2 files changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 1f3259c5f1c2..048ec2417990 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -6209,7 +6209,7 @@ static bool steal_cookie_task(int cpu, struct sched_domain *sd) { int i; - for_each_cpu_wrap(i, sched_domain_span(sd), cpu) { + for_each_cpu_wrap(i, sched_domain_span(sd), cpu + 1) { if (i == cpu) continue; diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index aaff2092ffdf..e9d906a9bba9 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -1836,7 +1836,7 @@ static void task_numa_assign(struct task_numa_env *env, int start = env->dst_cpu; /* Find alternative idle CPU. */ - for_each_cpu_wrap(cpu, cpumask_of_node(env->dst_nid), start) { + for_each_cpu_wrap(cpu, cpumask_of_node(env->dst_nid), start + 1) { if (cpu == env->best_cpu || !idle_cpu(cpu) || !cpumask_test_cpu(cpu, env->p->cpus_ptr)) { continue; @@ -6935,7 +6935,7 @@ select_idle_capacity(struct task_struct *p, struct sched_domain *sd, int target) util_min = uclamp_eff_value(p, UCLAMP_MIN); util_max = uclamp_eff_value(p, UCLAMP_MAX); - for_each_cpu_wrap(cpu, cpus, target) { + for_each_cpu_wrap(cpu, cpus, target + 1) { unsigned long cpu_cap = capacity_of(cpu); if (!available_idle_cpu(cpu) && !sched_idle_cpu(cpu)) -- cgit v1.2.3 From c89970202a1153b2fc230e89f90c180bd5bcbcef Mon Sep 17 00:00:00 2001 From: Nicholas Piggin Date: Tue, 20 Dec 2022 17:07:05 +1000 Subject: cputime: remove cputime_to_nsecs fallback The archs that use cputime_to_nsecs() internally provide their own definition and don't need the fallback. cputime_to_usecs() unused except in this fallback, and is not defined anywhere. This removes the final remnant of the cputime_t code from the kernel. Signed-off-by: Nicholas Piggin Signed-off-by: Peter Zijlstra (Intel) Acked-by: Alexander Gordeev Link: https://lore.kernel.org/r/20221220070705.2958959-1-npiggin@gmail.com --- arch/s390/kernel/idle.c | 2 +- arch/s390/kernel/vtime.c | 2 +- include/linux/sched/cputime.h | 9 --------- kernel/sched/cputime.c | 4 ++++ 4 files changed, 6 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/arch/s390/kernel/idle.c b/arch/s390/kernel/idle.c index 4bf1ee293f2b..a6bbceaf7616 100644 --- a/arch/s390/kernel/idle.c +++ b/arch/s390/kernel/idle.c @@ -12,9 +12,9 @@ #include #include #include -#include #include #include +#include #include #include #include "entry.h" diff --git a/arch/s390/kernel/vtime.c b/arch/s390/kernel/vtime.c index 9436f3053b88..e0a88dcaf5cb 100644 --- a/arch/s390/kernel/vtime.c +++ b/arch/s390/kernel/vtime.c @@ -7,13 +7,13 @@ */ #include -#include #include #include #include #include #include #include +#include #include #include #include diff --git a/include/linux/sched/cputime.h b/include/linux/sched/cputime.h index ce3c58286062..5f8fd5b24a2e 100644 --- a/include/linux/sched/cputime.h +++ b/include/linux/sched/cputime.h @@ -8,15 +8,6 @@ * cputime accounting APIs: */ -#ifdef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE -#include - -#ifndef cputime_to_nsecs -# define cputime_to_nsecs(__ct) \ - (cputime_to_usecs(__ct) * NSEC_PER_USEC) -#endif -#endif /* CONFIG_VIRT_CPU_ACCOUNTING_NATIVE */ - #ifdef CONFIG_VIRT_CPU_ACCOUNTING_GEN extern bool task_cputime(struct task_struct *t, u64 *utime, u64 *stime); diff --git a/kernel/sched/cputime.c b/kernel/sched/cputime.c index 95fc77853743..af7952f12e6c 100644 --- a/kernel/sched/cputime.c +++ b/kernel/sched/cputime.c @@ -3,6 +3,10 @@ * Simple CPU accounting cgroup controller */ +#ifdef CONFIG_VIRT_CPU_ACCOUNTING_NATIVE + #include +#endif + #ifdef CONFIG_IRQ_TIME_ACCOUNTING /* -- cgit v1.2.3 From bbd0b031509b880b4e9a880bb27ca2a30ad081ab Mon Sep 17 00:00:00 2001 From: Mathieu Desnoyers Date: Mon, 2 Jan 2023 10:12:16 -0500 Subject: sched/rseq: Fix concurrency ID handling of usermodehelper kthreads sched_mm_cid_after_execve() does not expect NULL t->mm, but it may happen if a usermodehelper kthread fails when attempting to execute a binary. sched_mm_cid_fork() can be issued from a usermodehelper kthread, which has t->flags PF_KTHREAD set. Fixes: af7f588d8f73 ("sched: Introduce per-memory-map concurrency ID") Reported-by: kernel test robot Reported-by: Borislav Petkov Signed-off-by: Mathieu Desnoyers Signed-off-by: Borislav Petkov (AMD) Link: https://lore.kernel.org/oe-lkp/202212301353.5c959d72-yujie.liu@intel.com --- kernel/sched/core.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 048ec2417990..f99ee69867e3 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -11343,8 +11343,8 @@ void sched_mm_cid_after_execve(struct task_struct *t) struct mm_struct *mm = t->mm; unsigned long flags; - WARN_ON_ONCE((t->flags & PF_KTHREAD) || !t->mm); - + if (!mm) + return; local_irq_save(flags); t->mm_cid = mm_cid_get(mm); t->mm_cid_active = 1; @@ -11354,7 +11354,7 @@ void sched_mm_cid_after_execve(struct task_struct *t) void sched_mm_cid_fork(struct task_struct *t) { - WARN_ON_ONCE((t->flags & PF_KTHREAD) || !t->mm || t->mm_cid != -1); + WARN_ON_ONCE(!t->mm || t->mm_cid != -1); t->mm_cid_active = 1; } #endif -- cgit v1.2.3 From ef90cf2281a013d359d24d51732af990badf6e03 Mon Sep 17 00:00:00 2001 From: Bing Huang Date: Thu, 5 Jan 2023 09:49:43 +0800 Subject: sched/topology: Add __init for sched_init_domains() sched_init_domains() is only used in initialization Signed-off-by: Bing Huang Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20230105014943.9857-1-huangbing775@126.com --- kernel/sched/topology.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c index dea9fa39e7c0..d93c3379e901 100644 --- a/kernel/sched/topology.c +++ b/kernel/sched/topology.c @@ -2451,7 +2451,7 @@ void free_sched_domains(cpumask_var_t doms[], unsigned int ndoms) * Set up scheduler domains and groups. For now this just excludes isolated * CPUs, but could be used to exclude other special cases in the future. */ -int sched_init_domains(const struct cpumask *cpu_map) +int __init sched_init_domains(const struct cpumask *cpu_map) { int err; -- cgit v1.2.3 From efe09385864f3441c71711f91e621992f9423c01 Mon Sep 17 00:00:00 2001 From: Chengming Zhou Date: Fri, 23 Dec 2022 18:32:56 +0800 Subject: sched/core: Micro-optimize ttwu_runnable() ttwu_runnable() is used as a fast wakeup path when the wakee task is running on CPU or runnable on RQ, in both cases we can just set its state to TASK_RUNNING to prevent a sleep. If the wakee task is on_cpu running, we don't need to update_rq_clock() or check_preempt_curr(). But if the wakee task is on_rq && !on_cpu (e.g. an IRQ hit before the task got to schedule() and the task been preempted), we should check_preempt_curr() to see if it can preempt the current running. This also removes the class->task_woken() callback from ttwu_runnable(), which wasn't required per the RT/DL implementations: any required push operation would have been queued during class->set_next_task() when p got preempted. ttwu_runnable() also loses the update to rq->idle_stamp, as by definition the rq cannot be idle in this scenario. Suggested-by: Valentin Schneider Suggested-by: Peter Zijlstra Signed-off-by: Chengming Zhou Signed-off-by: Ingo Molnar Reviewed-by: Valentin Schneider Reviewed-by: Steven Rostedt (Google) Link: https://lore.kernel.org/r/20221223103257.4962-1-zhouchengming@bytedance.com --- kernel/sched/core.c | 13 ++++++++++--- 1 file changed, 10 insertions(+), 3 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index f99ee69867e3..255a318cf24e 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -3720,9 +3720,16 @@ static int ttwu_runnable(struct task_struct *p, int wake_flags) rq = __task_rq_lock(p, &rf); if (task_on_rq_queued(p)) { - /* check_preempt_curr() may use rq clock */ - update_rq_clock(rq); - ttwu_do_wakeup(rq, p, wake_flags, &rf); + if (!task_on_cpu(rq, p)) { + /* + * When on_rq && !on_cpu the task is preempted, see if + * it should preempt the task that is current now. + */ + update_rq_clock(rq); + check_preempt_curr(rq, p, wake_flags); + } + WRITE_ONCE(p->__state, TASK_RUNNING); + trace_sched_wakeup(p); ret = 1; } __task_rq_unlock(rq, &rf); -- cgit v1.2.3 From 160fb0d83f206b3429fc495864a022110f9e4978 Mon Sep 17 00:00:00 2001 From: Chengming Zhou Date: Fri, 23 Dec 2022 18:32:57 +0800 Subject: sched/core: Reorganize ttwu_do_wakeup() and ttwu_do_activate() ttwu_do_activate() is used for a complete wakeup, in which we will activate_task() and use ttwu_do_wakeup() to mark the task runnable and perform wakeup-preemption, also call class->task_woken() callback and update the rq->idle_stamp. Since ttwu_runnable() is not a complete wakeup, don't need all those done in ttwu_do_wakeup(), so we can move those to ttwu_do_activate() to simplify ttwu_do_wakeup(), making it only mark the task runnable to be reused in ttwu_runnable() and try_to_wake_up(). This patch should not have any functional changes. Suggested-by: Peter Zijlstra Signed-off-by: Chengming Zhou Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20221223103257.4962-2-zhouchengming@bytedance.com --- kernel/sched/core.c | 64 ++++++++++++++++++++++++++--------------------------- 1 file changed, 31 insertions(+), 33 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/core.c b/kernel/sched/core.c index 255a318cf24e..03b8529db73f 100644 --- a/kernel/sched/core.c +++ b/kernel/sched/core.c @@ -3625,14 +3625,39 @@ ttwu_stat(struct task_struct *p, int cpu, int wake_flags) } /* - * Mark the task runnable and perform wakeup-preemption. + * Mark the task runnable. */ -static void ttwu_do_wakeup(struct rq *rq, struct task_struct *p, int wake_flags, - struct rq_flags *rf) +static inline void ttwu_do_wakeup(struct task_struct *p) { - check_preempt_curr(rq, p, wake_flags); WRITE_ONCE(p->__state, TASK_RUNNING); trace_sched_wakeup(p); +} + +static void +ttwu_do_activate(struct rq *rq, struct task_struct *p, int wake_flags, + struct rq_flags *rf) +{ + int en_flags = ENQUEUE_WAKEUP | ENQUEUE_NOCLOCK; + + lockdep_assert_rq_held(rq); + + if (p->sched_contributes_to_load) + rq->nr_uninterruptible--; + +#ifdef CONFIG_SMP + if (wake_flags & WF_MIGRATED) + en_flags |= ENQUEUE_MIGRATED; + else +#endif + if (p->in_iowait) { + delayacct_blkio_end(p); + atomic_dec(&task_rq(p)->nr_iowait); + } + + activate_task(rq, p, en_flags); + check_preempt_curr(rq, p, wake_flags); + + ttwu_do_wakeup(p); #ifdef CONFIG_SMP if (p->sched_class->task_woken) { @@ -3662,31 +3687,6 @@ static void ttwu_do_wakeup(struct rq *rq, struct task_struct *p, int wake_flags, #endif } -static void -ttwu_do_activate(struct rq *rq, struct task_struct *p, int wake_flags, - struct rq_flags *rf) -{ - int en_flags = ENQUEUE_WAKEUP | ENQUEUE_NOCLOCK; - - lockdep_assert_rq_held(rq); - - if (p->sched_contributes_to_load) - rq->nr_uninterruptible--; - -#ifdef CONFIG_SMP - if (wake_flags & WF_MIGRATED) - en_flags |= ENQUEUE_MIGRATED; - else -#endif - if (p->in_iowait) { - delayacct_blkio_end(p); - atomic_dec(&task_rq(p)->nr_iowait); - } - - activate_task(rq, p, en_flags); - ttwu_do_wakeup(rq, p, wake_flags, rf); -} - /* * Consider @p being inside a wait loop: * @@ -3728,8 +3728,7 @@ static int ttwu_runnable(struct task_struct *p, int wake_flags) update_rq_clock(rq); check_preempt_curr(rq, p, wake_flags); } - WRITE_ONCE(p->__state, TASK_RUNNING); - trace_sched_wakeup(p); + ttwu_do_wakeup(p); ret = 1; } __task_rq_unlock(rq, &rf); @@ -4095,8 +4094,7 @@ try_to_wake_up(struct task_struct *p, unsigned int state, int wake_flags) goto out; trace_sched_waking(p); - WRITE_ONCE(p->__state, TASK_RUNNING); - trace_sched_wakeup(p); + ttwu_do_wakeup(p); goto out; } -- cgit v1.2.3 From 948fb4c4e9cb2ad15b87e6e7cf0dd3b4ec17b8a4 Mon Sep 17 00:00:00 2001 From: Lukasz Luba Date: Thu, 8 Dec 2022 16:02:56 +0000 Subject: cpufreq, sched/util: Optimize operations with single CPU capacity lookup The max CPU capacity is the same for all CPUs sharing frequency domain. There is a way to avoid heavy operations in a loop for each CPU by leveraging this knowledge. Thus, simplify the looping code in the sugov_next_freq_shared() and drop heavy multiplications. Instead, use simple max() to get the highest utilization from these CPUs. This is useful for platforms with many (4 or 6) little CPUs. We avoid heavy 2*PD_CPU_NUM multiplications in that loop, which is called billions of times, since it's not limited by the schedutil time delta filter in sugov_should_update_freq(). When there was no need to change frequency the code bailed out, not updating the sg_policy::last_freq_update_time. Then every visit after delta_ns time longer than the sg_policy::freq_update_delay_ns goes through and triggers the next frequency calculation code. Although, if the next frequency, as outcome of that, would be the same as current frequency, we won't update the sg_policy::last_freq_update_time and the story will be repeated (in a very short period, sometimes a few microseconds). The max CPU capacity must be fetched every time we are called, due to difficulties during the policy setup, where we are not able to get the normalized CPU capacity at the right time. The fetched CPU capacity value is than used in sugov_iowait_apply() to calculate the right boost. This required a few changes in the local functions and arguments. The capacity value should hopefully be fetched once when needed and then passed over CPU registers to those functions. Signed-off-by: Lukasz Luba Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20221208160256.859-2-lukasz.luba@arm.com Cc: Peter Zijlstra Cc: Patrick Bellasi Cc: Vincent Guittot Cc: Rafael J. Wysocki Cc: Viresh Kumar --- kernel/sched/cpufreq_schedutil.c | 43 +++++++++++++++++++++------------------- 1 file changed, 23 insertions(+), 20 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/cpufreq_schedutil.c b/kernel/sched/cpufreq_schedutil.c index 1207c78f85c1..5c840151f3bb 100644 --- a/kernel/sched/cpufreq_schedutil.c +++ b/kernel/sched/cpufreq_schedutil.c @@ -48,7 +48,6 @@ struct sugov_cpu { unsigned long util; unsigned long bw_dl; - unsigned long max; /* The field below is for single-CPU policies only: */ #ifdef CONFIG_NO_HZ_COMMON @@ -158,7 +157,6 @@ static void sugov_get_util(struct sugov_cpu *sg_cpu) { struct rq *rq = cpu_rq(sg_cpu->cpu); - sg_cpu->max = arch_scale_cpu_capacity(sg_cpu->cpu); sg_cpu->bw_dl = cpu_bw_dl(rq); sg_cpu->util = effective_cpu_util(sg_cpu->cpu, cpu_util_cfs(sg_cpu->cpu), FREQUENCY_UTIL, NULL); @@ -238,6 +236,7 @@ static void sugov_iowait_boost(struct sugov_cpu *sg_cpu, u64 time, * sugov_iowait_apply() - Apply the IO boost to a CPU. * @sg_cpu: the sugov data for the cpu to boost * @time: the update time from the caller + * @max_cap: the max CPU capacity * * A CPU running a task which woken up after an IO operation can have its * utilization boosted to speed up the completion of those IO operations. @@ -251,7 +250,8 @@ static void sugov_iowait_boost(struct sugov_cpu *sg_cpu, u64 time, * This mechanism is designed to boost high frequently IO waiting tasks, while * being more conservative on tasks which does sporadic IO operations. */ -static void sugov_iowait_apply(struct sugov_cpu *sg_cpu, u64 time) +static void sugov_iowait_apply(struct sugov_cpu *sg_cpu, u64 time, + unsigned long max_cap) { unsigned long boost; @@ -280,7 +280,7 @@ static void sugov_iowait_apply(struct sugov_cpu *sg_cpu, u64 time) * sg_cpu->util is already in capacity scale; convert iowait_boost * into the same scale so we can compare. */ - boost = (sg_cpu->iowait_boost * sg_cpu->max) >> SCHED_CAPACITY_SHIFT; + boost = (sg_cpu->iowait_boost * max_cap) >> SCHED_CAPACITY_SHIFT; boost = uclamp_rq_util_with(cpu_rq(sg_cpu->cpu), boost, NULL); if (sg_cpu->util < boost) sg_cpu->util = boost; @@ -310,7 +310,8 @@ static inline void ignore_dl_rate_limit(struct sugov_cpu *sg_cpu) } static inline bool sugov_update_single_common(struct sugov_cpu *sg_cpu, - u64 time, unsigned int flags) + u64 time, unsigned long max_cap, + unsigned int flags) { sugov_iowait_boost(sg_cpu, time, flags); sg_cpu->last_update = time; @@ -321,7 +322,7 @@ static inline bool sugov_update_single_common(struct sugov_cpu *sg_cpu, return false; sugov_get_util(sg_cpu); - sugov_iowait_apply(sg_cpu, time); + sugov_iowait_apply(sg_cpu, time, max_cap); return true; } @@ -332,12 +333,15 @@ static void sugov_update_single_freq(struct update_util_data *hook, u64 time, struct sugov_cpu *sg_cpu = container_of(hook, struct sugov_cpu, update_util); struct sugov_policy *sg_policy = sg_cpu->sg_policy; unsigned int cached_freq = sg_policy->cached_raw_freq; + unsigned long max_cap; unsigned int next_f; - if (!sugov_update_single_common(sg_cpu, time, flags)) + max_cap = arch_scale_cpu_capacity(sg_cpu->cpu); + + if (!sugov_update_single_common(sg_cpu, time, max_cap, flags)) return; - next_f = get_next_freq(sg_policy, sg_cpu->util, sg_cpu->max); + next_f = get_next_freq(sg_policy, sg_cpu->util, max_cap); /* * Do not reduce the frequency if the CPU has not been idle * recently, as the reduction is likely to be premature then. @@ -374,6 +378,7 @@ static void sugov_update_single_perf(struct update_util_data *hook, u64 time, { struct sugov_cpu *sg_cpu = container_of(hook, struct sugov_cpu, update_util); unsigned long prev_util = sg_cpu->util; + unsigned long max_cap; /* * Fall back to the "frequency" path if frequency invariance is not @@ -385,7 +390,9 @@ static void sugov_update_single_perf(struct update_util_data *hook, u64 time, return; } - if (!sugov_update_single_common(sg_cpu, time, flags)) + max_cap = arch_scale_cpu_capacity(sg_cpu->cpu); + + if (!sugov_update_single_common(sg_cpu, time, max_cap, flags)) return; /* @@ -399,7 +406,7 @@ static void sugov_update_single_perf(struct update_util_data *hook, u64 time, sg_cpu->util = prev_util; cpufreq_driver_adjust_perf(sg_cpu->cpu, map_util_perf(sg_cpu->bw_dl), - map_util_perf(sg_cpu->util), sg_cpu->max); + map_util_perf(sg_cpu->util), max_cap); sg_cpu->sg_policy->last_freq_update_time = time; } @@ -408,25 +415,21 @@ static unsigned int sugov_next_freq_shared(struct sugov_cpu *sg_cpu, u64 time) { struct sugov_policy *sg_policy = sg_cpu->sg_policy; struct cpufreq_policy *policy = sg_policy->policy; - unsigned long util = 0, max = 1; + unsigned long util = 0, max_cap; unsigned int j; + max_cap = arch_scale_cpu_capacity(sg_cpu->cpu); + for_each_cpu(j, policy->cpus) { struct sugov_cpu *j_sg_cpu = &per_cpu(sugov_cpu, j); - unsigned long j_util, j_max; sugov_get_util(j_sg_cpu); - sugov_iowait_apply(j_sg_cpu, time); - j_util = j_sg_cpu->util; - j_max = j_sg_cpu->max; + sugov_iowait_apply(j_sg_cpu, time, max_cap); - if (j_util * max > j_max * util) { - util = j_util; - max = j_max; - } + util = max(j_sg_cpu->util, util); } - return get_next_freq(sg_policy, util, max); + return get_next_freq(sg_policy, util, max_cap); } static void -- cgit v1.2.3 From 544a4f2ecd45f9d6ed78d207583f39130ad40349 Mon Sep 17 00:00:00 2001 From: Michal Clapinski Date: Wed, 7 Dec 2022 17:43:37 +0100 Subject: sched/membarrier: Introduce MEMBARRIER_CMD_GET_REGISTRATIONS Provide a method to query previously issued registrations. Signed-off-by: Michal Clapinski Signed-off-by: Ingo Molnar Acked-by: Mathieu Desnoyers Acked-by: Paul E. McKenney Link: https://lore.kernel.org/r/20221207164338.1535591-2-mclapinski@google.com --- include/uapi/linux/membarrier.h | 4 ++++ kernel/sched/membarrier.c | 39 ++++++++++++++++++++++++++++++++++++++- 2 files changed, 42 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/include/uapi/linux/membarrier.h b/include/uapi/linux/membarrier.h index 737605897f36..5f3ad6d5be6f 100644 --- a/include/uapi/linux/membarrier.h +++ b/include/uapi/linux/membarrier.h @@ -137,6 +137,9 @@ * @MEMBARRIER_CMD_SHARED: * Alias to MEMBARRIER_CMD_GLOBAL. Provided for * header backward compatibility. + * @MEMBARRIER_CMD_GET_REGISTRATIONS: + * Returns a bitmask of previously issued + * registration commands. * * Command to be passed to the membarrier system call. The commands need to * be a single bit each, except for MEMBARRIER_CMD_QUERY which is assigned to @@ -153,6 +156,7 @@ enum membarrier_cmd { MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_SYNC_CORE = (1 << 6), MEMBARRIER_CMD_PRIVATE_EXPEDITED_RSEQ = (1 << 7), MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_RSEQ = (1 << 8), + MEMBARRIER_CMD_GET_REGISTRATIONS = (1 << 9), /* Alias for header backward compatibility. */ MEMBARRIER_CMD_SHARED = MEMBARRIER_CMD_GLOBAL, diff --git a/kernel/sched/membarrier.c b/kernel/sched/membarrier.c index 0c5be7ebb1dc..2ad881d07752 100644 --- a/kernel/sched/membarrier.c +++ b/kernel/sched/membarrier.c @@ -159,7 +159,8 @@ | MEMBARRIER_CMD_PRIVATE_EXPEDITED \ | MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED \ | MEMBARRIER_PRIVATE_EXPEDITED_SYNC_CORE_BITMASK \ - | MEMBARRIER_PRIVATE_EXPEDITED_RSEQ_BITMASK) + | MEMBARRIER_PRIVATE_EXPEDITED_RSEQ_BITMASK \ + | MEMBARRIER_CMD_GET_REGISTRATIONS) static void ipi_mb(void *info) { @@ -540,6 +541,40 @@ static int membarrier_register_private_expedited(int flags) return 0; } +static int membarrier_get_registrations(void) +{ + struct task_struct *p = current; + struct mm_struct *mm = p->mm; + int registrations_mask = 0, membarrier_state, i; + static const int states[] = { + MEMBARRIER_STATE_GLOBAL_EXPEDITED | + MEMBARRIER_STATE_GLOBAL_EXPEDITED_READY, + MEMBARRIER_STATE_PRIVATE_EXPEDITED | + MEMBARRIER_STATE_PRIVATE_EXPEDITED_READY, + MEMBARRIER_STATE_PRIVATE_EXPEDITED_SYNC_CORE | + MEMBARRIER_STATE_PRIVATE_EXPEDITED_SYNC_CORE_READY, + MEMBARRIER_STATE_PRIVATE_EXPEDITED_RSEQ | + MEMBARRIER_STATE_PRIVATE_EXPEDITED_RSEQ_READY + }; + static const int registration_cmds[] = { + MEMBARRIER_CMD_REGISTER_GLOBAL_EXPEDITED, + MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED, + MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_SYNC_CORE, + MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_RSEQ + }; + BUILD_BUG_ON(ARRAY_SIZE(states) != ARRAY_SIZE(registration_cmds)); + + membarrier_state = atomic_read(&mm->membarrier_state); + for (i = 0; i < ARRAY_SIZE(states); ++i) { + if (membarrier_state & states[i]) { + registrations_mask |= registration_cmds[i]; + membarrier_state &= ~states[i]; + } + } + WARN_ON_ONCE(membarrier_state != 0); + return registrations_mask; +} + /** * sys_membarrier - issue memory barriers on a set of threads * @cmd: Takes command values defined in enum membarrier_cmd. @@ -623,6 +658,8 @@ SYSCALL_DEFINE3(membarrier, int, cmd, unsigned int, flags, int, cpu_id) return membarrier_private_expedited(MEMBARRIER_FLAG_RSEQ, cpu_id); case MEMBARRIER_CMD_REGISTER_PRIVATE_EXPEDITED_RSEQ: return membarrier_register_private_expedited(MEMBARRIER_FLAG_RSEQ); + case MEMBARRIER_CMD_GET_REGISTRATIONS: + return membarrier_get_registrations(); default: return -EINVAL; } -- cgit v1.2.3 From a01353cf1896ea5b8a7bbc5e2b2d38feed8b7aaa Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 12 Jan 2023 20:43:27 +0100 Subject: cpuidle: Fix ct_idle_*() usage The whole disable-RCU, enable-IRQS dance is very intricate since changing IRQ state is traced, which depends on RCU. Add two helpers for the cpuidle case that mirror the entry code: ct_cpuidle_enter() ct_cpuidle_exit() And fix all the cases where the enter/exit dance was buggy. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Tested-by: Tony Lindgren Tested-by: Ulf Hansson Acked-by: Rafael J. Wysocki Acked-by: Frederic Weisbecker Link: https://lore.kernel.org/r/20230112195540.130014793@infradead.org --- arch/arm/mach-imx/cpuidle-imx6q.c | 4 ++-- arch/arm/mach-imx/cpuidle-imx6sx.c | 4 ++-- arch/arm/mach-omap2/cpuidle34xx.c | 4 ++-- arch/arm/mach-omap2/cpuidle44xx.c | 8 +++---- drivers/acpi/processor_idle.c | 8 +++++-- drivers/cpuidle/cpuidle-big_little.c | 4 ++-- drivers/cpuidle/cpuidle-mvebu-v7.c | 4 ++-- drivers/cpuidle/cpuidle-psci.c | 4 ++-- drivers/cpuidle/cpuidle-riscv-sbi.c | 4 ++-- drivers/cpuidle/cpuidle-tegra.c | 8 +++---- drivers/cpuidle/cpuidle.c | 11 +++++---- include/linux/clockchips.h | 4 ++-- include/linux/cpuidle.h | 34 +++++++++++++++++++++++++-- kernel/sched/idle.c | 45 +++++++++++------------------------- kernel/time/tick-broadcast.c | 6 ++++- 15 files changed, 86 insertions(+), 66 deletions(-) (limited to 'kernel') diff --git a/arch/arm/mach-imx/cpuidle-imx6q.c b/arch/arm/mach-imx/cpuidle-imx6q.c index d086cbae09c3..c24c78a67cc1 100644 --- a/arch/arm/mach-imx/cpuidle-imx6q.c +++ b/arch/arm/mach-imx/cpuidle-imx6q.c @@ -25,9 +25,9 @@ static int imx6q_enter_wait(struct cpuidle_device *dev, imx6_set_lpm(WAIT_UNCLOCKED); raw_spin_unlock(&cpuidle_lock); - ct_idle_enter(); + ct_cpuidle_enter(); cpu_do_idle(); - ct_idle_exit(); + ct_cpuidle_exit(); raw_spin_lock(&cpuidle_lock); if (num_idle_cpus-- == num_online_cpus()) diff --git a/arch/arm/mach-imx/cpuidle-imx6sx.c b/arch/arm/mach-imx/cpuidle-imx6sx.c index 1dc01f6b0f36..479f06286b50 100644 --- a/arch/arm/mach-imx/cpuidle-imx6sx.c +++ b/arch/arm/mach-imx/cpuidle-imx6sx.c @@ -47,9 +47,9 @@ static int imx6sx_enter_wait(struct cpuidle_device *dev, cpu_pm_enter(); cpu_cluster_pm_enter(); - ct_idle_enter(); + ct_cpuidle_enter(); cpu_suspend(0, imx6sx_idle_finish); - ct_idle_exit(); + ct_cpuidle_exit(); cpu_cluster_pm_exit(); cpu_pm_exit(); diff --git a/arch/arm/mach-omap2/cpuidle34xx.c b/arch/arm/mach-omap2/cpuidle34xx.c index cedf5cbe451f..6d63769cef0f 100644 --- a/arch/arm/mach-omap2/cpuidle34xx.c +++ b/arch/arm/mach-omap2/cpuidle34xx.c @@ -133,9 +133,9 @@ static int omap3_enter_idle(struct cpuidle_device *dev, } /* Execute ARM wfi */ - ct_idle_enter(); + ct_cpuidle_enter(); omap_sram_idle(); - ct_idle_exit(); + ct_cpuidle_exit(); /* * Call idle CPU PM enter notifier chain to restore diff --git a/arch/arm/mach-omap2/cpuidle44xx.c b/arch/arm/mach-omap2/cpuidle44xx.c index 953ad888737c..3c97d5676e81 100644 --- a/arch/arm/mach-omap2/cpuidle44xx.c +++ b/arch/arm/mach-omap2/cpuidle44xx.c @@ -105,9 +105,9 @@ static int omap_enter_idle_smp(struct cpuidle_device *dev, } raw_spin_unlock_irqrestore(&mpu_lock, flag); - ct_idle_enter(); + ct_cpuidle_enter(); omap4_enter_lowpower(dev->cpu, cx->cpu_state); - ct_idle_exit(); + ct_cpuidle_exit(); raw_spin_lock_irqsave(&mpu_lock, flag); if (cx->mpu_state_vote == num_online_cpus()) @@ -186,10 +186,10 @@ static int omap_enter_idle_coupled(struct cpuidle_device *dev, } } - ct_idle_enter(); + ct_cpuidle_enter(); omap4_enter_lowpower(dev->cpu, cx->cpu_state); cpu_done[dev->cpu] = true; - ct_idle_exit(); + ct_cpuidle_exit(); /* Wakeup CPU1 only if it is not offlined */ if (dev->cpu == 0 && cpumask_test_cpu(1, cpu_online_mask)) { diff --git a/drivers/acpi/processor_idle.c b/drivers/acpi/processor_idle.c index 566f7dbf7ab3..a589cfac8c0f 100644 --- a/drivers/acpi/processor_idle.c +++ b/drivers/acpi/processor_idle.c @@ -642,6 +642,8 @@ static int __cpuidle acpi_idle_enter_bm(struct cpuidle_driver *drv, */ bool dis_bm = pr->flags.bm_control; + instrumentation_begin(); + /* If we can skip BM, demote to a safe state. */ if (!cx->bm_sts_skip && acpi_idle_bm_check()) { dis_bm = false; @@ -663,11 +665,11 @@ static int __cpuidle acpi_idle_enter_bm(struct cpuidle_driver *drv, raw_spin_unlock(&c3_lock); } - ct_idle_enter(); + ct_cpuidle_enter(); acpi_idle_do_entry(cx); - ct_idle_exit(); + ct_cpuidle_exit(); /* Re-enable bus master arbitration */ if (dis_bm) { @@ -677,6 +679,8 @@ static int __cpuidle acpi_idle_enter_bm(struct cpuidle_driver *drv, raw_spin_unlock(&c3_lock); } + instrumentation_end(); + return index; } diff --git a/drivers/cpuidle/cpuidle-big_little.c b/drivers/cpuidle/cpuidle-big_little.c index fffd4ed0c4d1..5858db21e08c 100644 --- a/drivers/cpuidle/cpuidle-big_little.c +++ b/drivers/cpuidle/cpuidle-big_little.c @@ -126,13 +126,13 @@ static int bl_enter_powerdown(struct cpuidle_device *dev, struct cpuidle_driver *drv, int idx) { cpu_pm_enter(); - ct_idle_enter(); + ct_cpuidle_enter(); cpu_suspend(0, bl_powerdown_finisher); /* signals the MCPM core that CPU is out of low power state */ mcpm_cpu_powered_up(); - ct_idle_exit(); + ct_cpuidle_exit(); cpu_pm_exit(); diff --git a/drivers/cpuidle/cpuidle-mvebu-v7.c b/drivers/cpuidle/cpuidle-mvebu-v7.c index c9568aa9410c..20bfb26d5a88 100644 --- a/drivers/cpuidle/cpuidle-mvebu-v7.c +++ b/drivers/cpuidle/cpuidle-mvebu-v7.c @@ -36,9 +36,9 @@ static int mvebu_v7_enter_idle(struct cpuidle_device *dev, if (drv->states[index].flags & MVEBU_V7_FLAG_DEEP_IDLE) deepidle = true; - ct_idle_enter(); + ct_cpuidle_enter(); ret = mvebu_v7_cpu_suspend(deepidle); - ct_idle_exit(); + ct_cpuidle_exit(); cpu_pm_exit(); diff --git a/drivers/cpuidle/cpuidle-psci.c b/drivers/cpuidle/cpuidle-psci.c index 969808cef520..58b2cbba98c8 100644 --- a/drivers/cpuidle/cpuidle-psci.c +++ b/drivers/cpuidle/cpuidle-psci.c @@ -74,7 +74,7 @@ static int __psci_enter_domain_idle_state(struct cpuidle_device *dev, else pm_runtime_put_sync_suspend(pd_dev); - ct_idle_enter(); + ct_cpuidle_enter(); state = psci_get_domain_state(); if (!state) @@ -82,7 +82,7 @@ static int __psci_enter_domain_idle_state(struct cpuidle_device *dev, ret = psci_cpu_suspend_enter(state) ? -1 : idx; - ct_idle_exit(); + ct_cpuidle_exit(); if (s2idle) dev_pm_genpd_resume(pd_dev); diff --git a/drivers/cpuidle/cpuidle-riscv-sbi.c b/drivers/cpuidle/cpuidle-riscv-sbi.c index cbdbb11b972b..0a480f5799a7 100644 --- a/drivers/cpuidle/cpuidle-riscv-sbi.c +++ b/drivers/cpuidle/cpuidle-riscv-sbi.c @@ -126,7 +126,7 @@ static int __sbi_enter_domain_idle_state(struct cpuidle_device *dev, else pm_runtime_put_sync_suspend(pd_dev); - ct_idle_enter(); + ct_cpuidle_enter(); if (sbi_is_domain_state_available()) state = sbi_get_domain_state(); @@ -135,7 +135,7 @@ static int __sbi_enter_domain_idle_state(struct cpuidle_device *dev, ret = sbi_suspend(state) ? -1 : idx; - ct_idle_exit(); + ct_cpuidle_exit(); if (s2idle) dev_pm_genpd_resume(pd_dev); diff --git a/drivers/cpuidle/cpuidle-tegra.c b/drivers/cpuidle/cpuidle-tegra.c index 3ca5cfb9d322..9c2903c1b1c0 100644 --- a/drivers/cpuidle/cpuidle-tegra.c +++ b/drivers/cpuidle/cpuidle-tegra.c @@ -183,7 +183,7 @@ static int tegra_cpuidle_state_enter(struct cpuidle_device *dev, tegra_pm_set_cpu_in_lp2(); cpu_pm_enter(); - ct_idle_enter(); + ct_cpuidle_enter(); switch (index) { case TEGRA_C7: @@ -199,7 +199,7 @@ static int tegra_cpuidle_state_enter(struct cpuidle_device *dev, break; } - ct_idle_exit(); + ct_cpuidle_exit(); cpu_pm_exit(); tegra_pm_clear_cpu_in_lp2(); @@ -240,10 +240,10 @@ static int tegra_cpuidle_enter(struct cpuidle_device *dev, if (index == TEGRA_C1) { if (do_rcu) - ct_idle_enter(); + ct_cpuidle_enter(); ret = arm_cpuidle_simple_enter(dev, drv, index); if (do_rcu) - ct_idle_exit(); + ct_cpuidle_exit(); } else ret = tegra_cpuidle_state_enter(dev, index, cpu); diff --git a/drivers/cpuidle/cpuidle.c b/drivers/cpuidle/cpuidle.c index 95c801f0b9a8..08374c7935ce 100644 --- a/drivers/cpuidle/cpuidle.c +++ b/drivers/cpuidle/cpuidle.c @@ -14,6 +14,7 @@ #include #include #include +#include #include #include #include @@ -152,12 +153,12 @@ static void enter_s2idle_proper(struct cpuidle_driver *drv, */ stop_critical_timings(); if (!(target_state->flags & CPUIDLE_FLAG_RCU_IDLE)) - ct_idle_enter(); + ct_cpuidle_enter(); target_state->enter_s2idle(dev, drv, index); if (WARN_ON_ONCE(!irqs_disabled())) - local_irq_disable(); + raw_local_irq_disable(); if (!(target_state->flags & CPUIDLE_FLAG_RCU_IDLE)) - ct_idle_exit(); + ct_cpuidle_exit(); tick_unfreeze(); start_critical_timings(); @@ -235,14 +236,14 @@ int cpuidle_enter_state(struct cpuidle_device *dev, struct cpuidle_driver *drv, stop_critical_timings(); if (!(target_state->flags & CPUIDLE_FLAG_RCU_IDLE)) - ct_idle_enter(); + ct_cpuidle_enter(); entered_state = target_state->enter(dev, drv, index); if (WARN_ONCE(!irqs_disabled(), "%ps leaked IRQ state", target_state->enter)) raw_local_irq_disable(); if (!(target_state->flags & CPUIDLE_FLAG_RCU_IDLE)) - ct_idle_exit(); + ct_cpuidle_exit(); start_critical_timings(); sched_clock_idle_wakeup_event(); diff --git a/include/linux/clockchips.h b/include/linux/clockchips.h index 8ae9a95ebf5b..9aac31d856f3 100644 --- a/include/linux/clockchips.h +++ b/include/linux/clockchips.h @@ -211,7 +211,7 @@ extern int tick_receive_broadcast(void); extern void tick_setup_hrtimer_broadcast(void); extern int tick_check_broadcast_expired(void); # else -static inline int tick_check_broadcast_expired(void) { return 0; } +static __always_inline int tick_check_broadcast_expired(void) { return 0; } static inline void tick_setup_hrtimer_broadcast(void) { } # endif @@ -219,7 +219,7 @@ static inline void tick_setup_hrtimer_broadcast(void) { } static inline void clockevents_suspend(void) { } static inline void clockevents_resume(void) { } -static inline int tick_check_broadcast_expired(void) { return 0; } +static __always_inline int tick_check_broadcast_expired(void) { return 0; } static inline void tick_setup_hrtimer_broadcast(void) { } #endif /* !CONFIG_GENERIC_CLOCKEVENTS */ diff --git a/include/linux/cpuidle.h b/include/linux/cpuidle.h index 0ddc11e44302..630c879143c7 100644 --- a/include/linux/cpuidle.h +++ b/include/linux/cpuidle.h @@ -14,6 +14,7 @@ #include #include #include +#include #define CPUIDLE_STATE_MAX 10 #define CPUIDLE_NAME_LEN 16 @@ -115,6 +116,35 @@ struct cpuidle_device { DECLARE_PER_CPU(struct cpuidle_device *, cpuidle_devices); DECLARE_PER_CPU(struct cpuidle_device, cpuidle_dev); +static __always_inline void ct_cpuidle_enter(void) +{ + lockdep_assert_irqs_disabled(); + /* + * Idle is allowed to (temporary) enable IRQs. It + * will return with IRQs disabled. + * + * Trace IRQs enable here, then switch off RCU, and have + * arch_cpu_idle() use raw_local_irq_enable(). Note that + * ct_idle_enter() relies on lockdep IRQ state, so switch that + * last -- this is very similar to the entry code. + */ + trace_hardirqs_on_prepare(); + lockdep_hardirqs_on_prepare(); + instrumentation_end(); + ct_idle_enter(); + lockdep_hardirqs_on(_RET_IP_); +} + +static __always_inline void ct_cpuidle_exit(void) +{ + /* + * Carefully undo the above. + */ + lockdep_hardirqs_off(_RET_IP_); + ct_idle_exit(); + instrumentation_begin(); +} + /**************************** * CPUIDLE DRIVER INTERFACE * ****************************/ @@ -289,9 +319,9 @@ extern s64 cpuidle_governor_latency_req(unsigned int cpu); if (!is_retention) \ __ret = cpu_pm_enter(); \ if (!__ret) { \ - ct_idle_enter(); \ + ct_cpuidle_enter(); \ __ret = low_level_idle_enter(state); \ - ct_idle_exit(); \ + ct_cpuidle_exit(); \ if (!is_retention) \ cpu_pm_exit(); \ } \ diff --git a/kernel/sched/idle.c b/kernel/sched/idle.c index f26ab2675f7d..e924602ec43b 100644 --- a/kernel/sched/idle.c +++ b/kernel/sched/idle.c @@ -51,18 +51,22 @@ __setup("hlt", cpu_idle_nopoll_setup); static noinline int __cpuidle cpu_idle_poll(void) { + instrumentation_begin(); trace_cpu_idle(0, smp_processor_id()); stop_critical_timings(); - ct_idle_enter(); - local_irq_enable(); + ct_cpuidle_enter(); + raw_local_irq_enable(); while (!tif_need_resched() && (cpu_idle_force_poll || tick_check_broadcast_expired())) cpu_relax(); + raw_local_irq_disable(); - ct_idle_exit(); + ct_cpuidle_exit(); start_critical_timings(); trace_cpu_idle(PWR_EVENT_EXIT, smp_processor_id()); + local_irq_enable(); + instrumentation_end(); return 1; } @@ -85,44 +89,21 @@ void __weak arch_cpu_idle(void) */ void __cpuidle default_idle_call(void) { - if (current_clr_polling_and_test()) { - local_irq_enable(); - } else { - + instrumentation_begin(); + if (!current_clr_polling_and_test()) { trace_cpu_idle(1, smp_processor_id()); stop_critical_timings(); - /* - * arch_cpu_idle() is supposed to enable IRQs, however - * we can't do that because of RCU and tracing. - * - * Trace IRQs enable here, then switch off RCU, and have - * arch_cpu_idle() use raw_local_irq_enable(). Note that - * ct_idle_enter() relies on lockdep IRQ state, so switch that - * last -- this is very similar to the entry code. - */ - trace_hardirqs_on_prepare(); - lockdep_hardirqs_on_prepare(); - ct_idle_enter(); - lockdep_hardirqs_on(_THIS_IP_); - + ct_cpuidle_enter(); arch_cpu_idle(); - - /* - * OK, so IRQs are enabled here, but RCU needs them disabled to - * turn itself back on.. funny thing is that disabling IRQs - * will cause tracing, which needs RCU. Jump through hoops to - * make it 'work'. - */ raw_local_irq_disable(); - lockdep_hardirqs_off(_THIS_IP_); - ct_idle_exit(); - lockdep_hardirqs_on(_THIS_IP_); - raw_local_irq_enable(); + ct_cpuidle_exit(); start_critical_timings(); trace_cpu_idle(PWR_EVENT_EXIT, smp_processor_id()); } + local_irq_enable(); + instrumentation_end(); } static int call_cpuidle_s2idle(struct cpuidle_driver *drv, diff --git a/kernel/time/tick-broadcast.c b/kernel/time/tick-broadcast.c index f7fe6fe36173..93bf2b4e47e5 100644 --- a/kernel/time/tick-broadcast.c +++ b/kernel/time/tick-broadcast.c @@ -622,9 +622,13 @@ struct cpumask *tick_get_broadcast_oneshot_mask(void) * to avoid a deep idle transition as we are about to get the * broadcast IPI right away. */ -int tick_check_broadcast_expired(void) +noinstr int tick_check_broadcast_expired(void) { +#ifdef _ASM_GENERIC_BITOPS_INSTRUMENTED_NON_ATOMIC_H + return arch_test_bit(smp_processor_id(), cpumask_bits(tick_broadcast_force_mask)); +#else return cpumask_test_cpu(smp_processor_id(), tick_broadcast_force_mask); +#endif } /* -- cgit v1.2.3 From 924aed1646bf2859726f7e5dd9265ba14358fbb8 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 12 Jan 2023 20:43:28 +0100 Subject: cpuidle, cpu_pm: Remove RCU fiddling from cpu_pm_{enter,exit}() All callers should still have RCU enabled. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Tested-by: Tony Lindgren Tested-by: Ulf Hansson Reviewed-by: Ulf Hansson Acked-by: Mark Rutland Acked-by: Rafael J. Wysocki Acked-by: Frederic Weisbecker Link: https://lore.kernel.org/r/20230112195540.190860672@infradead.org --- kernel/cpu_pm.c | 9 --------- 1 file changed, 9 deletions(-) (limited to 'kernel') diff --git a/kernel/cpu_pm.c b/kernel/cpu_pm.c index ba4ba71facf9..b0f0d15085db 100644 --- a/kernel/cpu_pm.c +++ b/kernel/cpu_pm.c @@ -30,16 +30,9 @@ static int cpu_pm_notify(enum cpu_pm_event event) { int ret; - /* - * This introduces a RCU read critical section, which could be - * disfunctional in cpu idle. Copy RCU_NONIDLE code to let RCU know - * this. - */ - ct_irq_enter_irqson(); rcu_read_lock(); ret = raw_notifier_call_chain(&cpu_pm_notifier.chain, event, NULL); rcu_read_unlock(); - ct_irq_exit_irqson(); return notifier_to_errno(ret); } @@ -49,11 +42,9 @@ static int cpu_pm_notify_robust(enum cpu_pm_event event_up, enum cpu_pm_event ev unsigned long flags; int ret; - ct_irq_enter_irqson(); raw_spin_lock_irqsave(&cpu_pm_notifier.lock, flags); ret = raw_notifier_call_chain_robust(&cpu_pm_notifier.chain, event_up, event_down, NULL); raw_spin_unlock_irqrestore(&cpu_pm_notifier.lock, flags); - ct_irq_exit_irqson(); return notifier_to_errno(ret); } -- cgit v1.2.3 From 89b3098703bd2aa3237ef10a704e6a5838e6ea69 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 12 Jan 2023 20:43:35 +0100 Subject: arch/idle: Change arch_cpu_idle() behavior: always exit with IRQs disabled Current arch_cpu_idle() is called with IRQs disabled, but will return with IRQs enabled. However, the very first thing the generic code does after calling arch_cpu_idle() is raw_local_irq_disable(). This means that architectures that can idle with IRQs disabled end up doing a pointless 'enable-disable' dance. Therefore, push this IRQ disabling into the idle function, meaning that those architectures can avoid the pointless IRQ state flipping. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Tested-by: Tony Lindgren Tested-by: Ulf Hansson Reviewed-by: Gautham R. Shenoy Acked-by: Mark Rutland [arm64] Acked-by: Rafael J. Wysocki Acked-by: Guo Ren Acked-by: Frederic Weisbecker Link: https://lore.kernel.org/r/20230112195540.618076436@infradead.org --- arch/alpha/kernel/process.c | 1 - arch/arc/kernel/process.c | 3 +++ arch/arm/kernel/process.c | 1 - arch/arm/mach-gemini/board-dt.c | 3 ++- arch/arm64/kernel/idle.c | 1 - arch/csky/kernel/process.c | 1 - arch/csky/kernel/smp.c | 2 +- arch/hexagon/kernel/process.c | 1 - arch/ia64/kernel/process.c | 1 + arch/loongarch/kernel/idle.c | 1 + arch/microblaze/kernel/process.c | 1 - arch/mips/kernel/idle.c | 8 +++----- arch/nios2/kernel/process.c | 1 - arch/openrisc/kernel/process.c | 1 + arch/parisc/kernel/process.c | 2 -- arch/powerpc/kernel/idle.c | 5 ++--- arch/riscv/kernel/process.c | 1 - arch/s390/kernel/idle.c | 1 - arch/sh/kernel/idle.c | 1 + arch/sparc/kernel/leon_pmc.c | 4 ++++ arch/sparc/kernel/process_32.c | 1 - arch/sparc/kernel/process_64.c | 3 ++- arch/um/kernel/process.c | 1 - arch/x86/coco/tdx/tdx.c | 3 +++ arch/x86/kernel/process.c | 15 ++++----------- arch/xtensa/kernel/process.c | 1 + kernel/sched/idle.c | 2 -- 27 files changed, 29 insertions(+), 37 deletions(-) (limited to 'kernel') diff --git a/arch/alpha/kernel/process.c b/arch/alpha/kernel/process.c index 65fdae9e48f3..ce20c31828a0 100644 --- a/arch/alpha/kernel/process.c +++ b/arch/alpha/kernel/process.c @@ -57,7 +57,6 @@ EXPORT_SYMBOL(pm_power_off); void arch_cpu_idle(void) { wtint(0); - raw_local_irq_enable(); } void arch_cpu_idle_dead(void) diff --git a/arch/arc/kernel/process.c b/arch/arc/kernel/process.c index 3369f0700702..980b71da2f61 100644 --- a/arch/arc/kernel/process.c +++ b/arch/arc/kernel/process.c @@ -114,6 +114,8 @@ void arch_cpu_idle(void) "sleep %0 \n" : :"I"(arg)); /* can't be "r" has to be embedded const */ + + raw_local_irq_disable(); } #else /* ARC700 */ @@ -122,6 +124,7 @@ void arch_cpu_idle(void) { /* sleep, but enable both set E1/E2 (levels of interrupts) before committing */ __asm__ __volatile__("sleep 0x3 \n"); + raw_local_irq_disable(); } #endif diff --git a/arch/arm/kernel/process.c b/arch/arm/kernel/process.c index f811733a8fc5..c81e7be2b4ea 100644 --- a/arch/arm/kernel/process.c +++ b/arch/arm/kernel/process.c @@ -78,7 +78,6 @@ void arch_cpu_idle(void) arm_pm_idle(); else cpu_do_idle(); - raw_local_irq_enable(); } void arch_cpu_idle_prepare(void) diff --git a/arch/arm/mach-gemini/board-dt.c b/arch/arm/mach-gemini/board-dt.c index de0afcc8d94a..fbafe7475c02 100644 --- a/arch/arm/mach-gemini/board-dt.c +++ b/arch/arm/mach-gemini/board-dt.c @@ -42,8 +42,9 @@ static void gemini_idle(void) */ /* FIXME: Enabling interrupts here is racy! */ - local_irq_enable(); + raw_local_irq_enable(); cpu_do_idle(); + raw_local_irq_disable(); } static void __init gemini_init_machine(void) diff --git a/arch/arm64/kernel/idle.c b/arch/arm64/kernel/idle.c index a2cfbacec2bb..c1125753fe9b 100644 --- a/arch/arm64/kernel/idle.c +++ b/arch/arm64/kernel/idle.c @@ -42,5 +42,4 @@ void noinstr arch_cpu_idle(void) * tricks */ cpu_do_idle(); - raw_local_irq_enable(); } diff --git a/arch/csky/kernel/process.c b/arch/csky/kernel/process.c index 2b0ed515a88e..0c6e4b17fe00 100644 --- a/arch/csky/kernel/process.c +++ b/arch/csky/kernel/process.c @@ -100,6 +100,5 @@ void arch_cpu_idle(void) #ifdef CONFIG_CPU_PM_STOP asm volatile("stop\n"); #endif - raw_local_irq_enable(); } #endif diff --git a/arch/csky/kernel/smp.c b/arch/csky/kernel/smp.c index 4b605aa2e1d6..b45d1073307f 100644 --- a/arch/csky/kernel/smp.c +++ b/arch/csky/kernel/smp.c @@ -309,7 +309,7 @@ void arch_cpu_idle_dead(void) while (!secondary_stack) arch_cpu_idle(); - local_irq_disable(); + raw_local_irq_disable(); asm volatile( "mov sp, %0\n" diff --git a/arch/hexagon/kernel/process.c b/arch/hexagon/kernel/process.c index e15eeaebd785..dd7f74ea2c20 100644 --- a/arch/hexagon/kernel/process.c +++ b/arch/hexagon/kernel/process.c @@ -44,7 +44,6 @@ void arch_cpu_idle(void) { __vmwait(); /* interrupts wake us up, but irqs are still disabled */ - raw_local_irq_enable(); } /* diff --git a/arch/ia64/kernel/process.c b/arch/ia64/kernel/process.c index 416305e550e2..f6195a0a00ae 100644 --- a/arch/ia64/kernel/process.c +++ b/arch/ia64/kernel/process.c @@ -242,6 +242,7 @@ void arch_cpu_idle(void) (*mark_idle)(1); raw_safe_halt(); + raw_local_irq_disable(); if (mark_idle) (*mark_idle)(0); diff --git a/arch/loongarch/kernel/idle.c b/arch/loongarch/kernel/idle.c index 1a65d0527d25..0b5dd2faeb90 100644 --- a/arch/loongarch/kernel/idle.c +++ b/arch/loongarch/kernel/idle.c @@ -13,4 +13,5 @@ void __cpuidle arch_cpu_idle(void) { raw_local_irq_enable(); __arch_cpu_idle(); /* idle instruction needs irq enabled */ + raw_local_irq_disable(); } diff --git a/arch/microblaze/kernel/process.c b/arch/microblaze/kernel/process.c index 1f802aab2b96..56342e11442d 100644 --- a/arch/microblaze/kernel/process.c +++ b/arch/microblaze/kernel/process.c @@ -140,5 +140,4 @@ int elf_core_copy_task_fpregs(struct task_struct *t, elf_fpregset_t *fpu) void arch_cpu_idle(void) { - raw_local_irq_enable(); } diff --git a/arch/mips/kernel/idle.c b/arch/mips/kernel/idle.c index 53adcc1b2ed5..3e7ccdd9a797 100644 --- a/arch/mips/kernel/idle.c +++ b/arch/mips/kernel/idle.c @@ -33,13 +33,13 @@ static void __cpuidle r3081_wait(void) { unsigned long cfg = read_c0_conf(); write_c0_conf(cfg | R30XX_CONF_HALT); - raw_local_irq_enable(); } void __cpuidle r4k_wait(void) { raw_local_irq_enable(); __r4k_wait(); + raw_local_irq_disable(); } /* @@ -57,7 +57,6 @@ void __cpuidle r4k_wait_irqoff(void) " .set arch=r4000 \n" " wait \n" " .set pop \n"); - raw_local_irq_enable(); } /* @@ -77,7 +76,6 @@ static void __cpuidle rm7k_wait_irqoff(void) " wait \n" " mtc0 $1, $12 # stalls until W stage \n" " .set pop \n"); - raw_local_irq_enable(); } /* @@ -103,6 +101,8 @@ static void __cpuidle au1k_wait(void) " nop \n" " .set pop \n" : : "r" (au1k_wait), "r" (c0status)); + + raw_local_irq_disable(); } static int __initdata nowait; @@ -245,8 +245,6 @@ void arch_cpu_idle(void) { if (cpu_wait) cpu_wait(); - else - raw_local_irq_enable(); } #ifdef CONFIG_CPU_IDLE diff --git a/arch/nios2/kernel/process.c b/arch/nios2/kernel/process.c index 29593b98567d..f84021303f6a 100644 --- a/arch/nios2/kernel/process.c +++ b/arch/nios2/kernel/process.c @@ -33,7 +33,6 @@ EXPORT_SYMBOL(pm_power_off); void arch_cpu_idle(void) { - raw_local_irq_enable(); } /* diff --git a/arch/openrisc/kernel/process.c b/arch/openrisc/kernel/process.c index f94b5ec06786..dfa558f98ed8 100644 --- a/arch/openrisc/kernel/process.c +++ b/arch/openrisc/kernel/process.c @@ -102,6 +102,7 @@ void arch_cpu_idle(void) raw_local_irq_enable(); if (mfspr(SPR_UPR) & SPR_UPR_PMP) mtspr(SPR_PMR, mfspr(SPR_PMR) | SPR_PMR_DME); + raw_local_irq_disable(); } void (*pm_power_off)(void) = NULL; diff --git a/arch/parisc/kernel/process.c b/arch/parisc/kernel/process.c index c4f8374c7018..c064719b49b0 100644 --- a/arch/parisc/kernel/process.c +++ b/arch/parisc/kernel/process.c @@ -183,8 +183,6 @@ void arch_cpu_idle_dead(void) void __cpuidle arch_cpu_idle(void) { - raw_local_irq_enable(); - /* nop on real hardware, qemu will idle sleep. */ asm volatile("or %%r10,%%r10,%%r10\n":::); } diff --git a/arch/powerpc/kernel/idle.c b/arch/powerpc/kernel/idle.c index 77cd4c5a2d63..b9a725abc596 100644 --- a/arch/powerpc/kernel/idle.c +++ b/arch/powerpc/kernel/idle.c @@ -51,10 +51,9 @@ void arch_cpu_idle(void) * Some power_save functions return with * interrupts enabled, some don't. */ - if (irqs_disabled()) - raw_local_irq_enable(); + if (!irqs_disabled()) + raw_local_irq_disable(); } else { - raw_local_irq_enable(); /* * Go into low thread priority and possibly * low power mode. diff --git a/arch/riscv/kernel/process.c b/arch/riscv/kernel/process.c index 8955f2432c2d..774ffde386ab 100644 --- a/arch/riscv/kernel/process.c +++ b/arch/riscv/kernel/process.c @@ -39,7 +39,6 @@ extern asmlinkage void ret_from_kernel_thread(void); void arch_cpu_idle(void) { cpu_do_idle(); - raw_local_irq_enable(); } void __show_regs(struct pt_regs *regs) diff --git a/arch/s390/kernel/idle.c b/arch/s390/kernel/idle.c index a6bbceaf7616..b04fb418307c 100644 --- a/arch/s390/kernel/idle.c +++ b/arch/s390/kernel/idle.c @@ -66,7 +66,6 @@ void arch_cpu_idle(void) idle->idle_count++; account_idle_time(cputime_to_nsecs(idle_time)); raw_write_seqcount_end(&idle->seqcount); - raw_local_irq_enable(); } static ssize_t show_idle_count(struct device *dev, diff --git a/arch/sh/kernel/idle.c b/arch/sh/kernel/idle.c index f59814983bd5..3418c40f0099 100644 --- a/arch/sh/kernel/idle.c +++ b/arch/sh/kernel/idle.c @@ -25,6 +25,7 @@ void default_idle(void) raw_local_irq_enable(); /* Isn't this racy ? */ cpu_sleep(); + raw_local_irq_disable(); clear_bl_bit(); } diff --git a/arch/sparc/kernel/leon_pmc.c b/arch/sparc/kernel/leon_pmc.c index 396f46bca52e..6c00cbad7fb5 100644 --- a/arch/sparc/kernel/leon_pmc.c +++ b/arch/sparc/kernel/leon_pmc.c @@ -57,6 +57,8 @@ static void pmc_leon_idle_fixup(void) "lda [%0] %1, %%g0\n" : : "r"(address), "i"(ASI_LEON_BYPASS)); + + raw_local_irq_disable(); } /* @@ -70,6 +72,8 @@ static void pmc_leon_idle(void) /* For systems without power-down, this will be no-op */ __asm__ __volatile__ ("wr %g0, %asr19\n\t"); + + raw_local_irq_disable(); } /* Install LEON Power Down function */ diff --git a/arch/sparc/kernel/process_32.c b/arch/sparc/kernel/process_32.c index 33b0215a4182..9c7c662cb565 100644 --- a/arch/sparc/kernel/process_32.c +++ b/arch/sparc/kernel/process_32.c @@ -71,7 +71,6 @@ void arch_cpu_idle(void) { if (sparc_idle) (*sparc_idle)(); - raw_local_irq_enable(); } /* XXX cli/sti -> local_irq_xxx here, check this works once SMP is fixed. */ diff --git a/arch/sparc/kernel/process_64.c b/arch/sparc/kernel/process_64.c index 6335b698a4b4..91c2b8124527 100644 --- a/arch/sparc/kernel/process_64.c +++ b/arch/sparc/kernel/process_64.c @@ -59,7 +59,6 @@ void arch_cpu_idle(void) { if (tlb_type != hypervisor) { touch_nmi_watchdog(); - raw_local_irq_enable(); } else { unsigned long pstate; @@ -90,6 +89,8 @@ void arch_cpu_idle(void) "wrpr %0, %%g0, %%pstate" : "=&r" (pstate) : "i" (PSTATE_IE)); + + raw_local_irq_disable(); } } diff --git a/arch/um/kernel/process.c b/arch/um/kernel/process.c index 47830ade35ed..106b7da2f8d6 100644 --- a/arch/um/kernel/process.c +++ b/arch/um/kernel/process.c @@ -218,7 +218,6 @@ void arch_cpu_idle(void) { cpu_tasks[current_thread_info()->cpu].pid = os_getpid(); um_idle_sleep(); - raw_local_irq_enable(); } int __cant_sleep(void) { diff --git a/arch/x86/coco/tdx/tdx.c b/arch/x86/coco/tdx/tdx.c index cfd4c95b9f04..f4a59c365321 100644 --- a/arch/x86/coco/tdx/tdx.c +++ b/arch/x86/coco/tdx/tdx.c @@ -274,6 +274,9 @@ void __cpuidle tdx_safe_halt(void) */ if (__halt(irq_disabled, do_sti)) WARN_ONCE(1, "HLT instruction emulation failed\n"); + + /* XXX I can't make sense of what @do_sti actually does */ + raw_local_irq_disable(); } static int read_msr(struct pt_regs *regs, struct ve_info *ve) diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c index 8481fbefe282..e57cd31bfec4 100644 --- a/arch/x86/kernel/process.c +++ b/arch/x86/kernel/process.c @@ -701,6 +701,7 @@ EXPORT_SYMBOL(boot_option_idle_override); void __cpuidle default_idle(void) { raw_safe_halt(); + raw_local_irq_disable(); } #if defined(CONFIG_APM_MODULE) || defined(CONFIG_HALTPOLL_CPUIDLE_MODULE) EXPORT_SYMBOL(default_idle); @@ -806,13 +807,7 @@ static void amd_e400_idle(void) default_idle(); - /* - * The switch back from broadcast mode needs to be called with - * interrupts disabled. - */ - raw_local_irq_disable(); tick_broadcast_exit(); - raw_local_irq_enable(); } /* @@ -870,12 +865,10 @@ static __cpuidle void mwait_idle(void) } __monitor((void *)¤t_thread_info()->flags, 0, 0); - if (!need_resched()) + if (!need_resched()) { __sti_mwait(0, 0); - else - raw_local_irq_enable(); - } else { - raw_local_irq_enable(); + raw_local_irq_disable(); + } } __current_clr_polling(); } diff --git a/arch/xtensa/kernel/process.c b/arch/xtensa/kernel/process.c index 68e0e2f06d66..a815577d25fd 100644 --- a/arch/xtensa/kernel/process.c +++ b/arch/xtensa/kernel/process.c @@ -183,6 +183,7 @@ void coprocessor_flush_release_all(struct thread_info *ti) void arch_cpu_idle(void) { platform_idle(); + raw_local_irq_disable(); } /* diff --git a/kernel/sched/idle.c b/kernel/sched/idle.c index e924602ec43b..e9ef66be2870 100644 --- a/kernel/sched/idle.c +++ b/kernel/sched/idle.c @@ -79,7 +79,6 @@ void __weak arch_cpu_idle_dead(void) { } void __weak arch_cpu_idle(void) { cpu_idle_force_poll = 1; - raw_local_irq_enable(); } /** @@ -96,7 +95,6 @@ void __cpuidle default_idle_call(void) ct_cpuidle_enter(); arch_cpu_idle(); - raw_local_irq_disable(); ct_cpuidle_exit(); start_critical_timings(); -- cgit v1.2.3 From 880970b56b732b158385991b232592144256f330 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 12 Jan 2023 20:43:39 +0100 Subject: printk: Remove trace_.*_rcuidle() usage The problem, per commit fc98c3c8c9dc ("printk: use rcuidle console tracepoint"), was printk usage from the cpuidle path where RCU was already disabled. Per the patches earlier in this series, this is no longer the case. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Tested-by: Tony Lindgren Tested-by: Ulf Hansson Reviewed-by: Sergey Senozhatsky Acked-by: Petr Mladek Acked-by: Rafael J. Wysocki Acked-by: Frederic Weisbecker Link: https://lore.kernel.org/r/20230112195540.865735001@infradead.org --- kernel/printk/printk.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/printk/printk.c b/kernel/printk/printk.c index 7decf1e9c486..1ab53dda5002 100644 --- a/kernel/printk/printk.c +++ b/kernel/printk/printk.c @@ -2194,7 +2194,7 @@ static u16 printk_sprint(char *text, u16 size, int facility, } } - trace_console_rcuidle(text, text_len); + trace_console(text, text_len); return text_len; } -- cgit v1.2.3 From e3ee5e66f78e9950b9ada276700abd95e9b144d7 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 12 Jan 2023 20:43:40 +0100 Subject: time/tick-broadcast: Remove RCU_NONIDLE() usage No callers left that have already disabled RCU. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Tested-by: Tony Lindgren Tested-by: Ulf Hansson Acked-by: Mark Rutland Acked-by: Rafael J. Wysocki Acked-by: Frederic Weisbecker Link: https://lore.kernel.org/r/20230112195540.927904612@infradead.org --- kernel/time/tick-broadcast-hrtimer.c | 29 ++++++++++++----------------- 1 file changed, 12 insertions(+), 17 deletions(-) (limited to 'kernel') diff --git a/kernel/time/tick-broadcast-hrtimer.c b/kernel/time/tick-broadcast-hrtimer.c index 797eb93103ad..e28f9210f8a1 100644 --- a/kernel/time/tick-broadcast-hrtimer.c +++ b/kernel/time/tick-broadcast-hrtimer.c @@ -56,25 +56,20 @@ static int bc_set_next(ktime_t expires, struct clock_event_device *bc) * hrtimer callback function is currently running, then * hrtimer_start() cannot move it and the timer stays on the CPU on * which it is assigned at the moment. + */ + hrtimer_start(&bctimer, expires, HRTIMER_MODE_ABS_PINNED_HARD); + /* + * The core tick broadcast mode expects bc->bound_on to be set + * correctly to prevent a CPU which has the broadcast hrtimer + * armed from going deep idle. * - * As this can be called from idle code, the hrtimer_start() - * invocation has to be wrapped with RCU_NONIDLE() as - * hrtimer_start() can call into tracing. + * As tick_broadcast_lock is held, nothing can change the cpu + * base which was just established in hrtimer_start() above. So + * the below access is safe even without holding the hrtimer + * base lock. */ - RCU_NONIDLE( { - hrtimer_start(&bctimer, expires, HRTIMER_MODE_ABS_PINNED_HARD); - /* - * The core tick broadcast mode expects bc->bound_on to be set - * correctly to prevent a CPU which has the broadcast hrtimer - * armed from going deep idle. - * - * As tick_broadcast_lock is held, nothing can change the cpu - * base which was just established in hrtimer_start() above. So - * the below access is safe even without holding the hrtimer - * base lock. - */ - bc->bound_on = bctimer.base->cpu_base->cpu; - } ); + bc->bound_on = bctimer.base->cpu_base->cpu; + return 0; } -- cgit v1.2.3 From dc7305606d480f34d9ed960fcaceac81d0f38436 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 12 Jan 2023 20:43:47 +0100 Subject: tracing: Remove trace_hardirqs_{on,off}_caller() Per commit 56e62a737028 ("s390: convert to generic entry") the last and only callers of trace_hardirqs_{on,off}_caller() went away, clean up. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20230112195541.355283994@infradead.org --- kernel/trace/trace_preemptirq.c | 29 ----------------------------- 1 file changed, 29 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_preemptirq.c b/kernel/trace/trace_preemptirq.c index 1e130da1b742..629f2854e12b 100644 --- a/kernel/trace/trace_preemptirq.c +++ b/kernel/trace/trace_preemptirq.c @@ -84,35 +84,6 @@ void trace_hardirqs_off(void) } EXPORT_SYMBOL(trace_hardirqs_off); NOKPROBE_SYMBOL(trace_hardirqs_off); - -__visible void trace_hardirqs_on_caller(unsigned long caller_addr) -{ - if (this_cpu_read(tracing_irq_cpu)) { - if (!in_nmi()) - trace_irq_enable_rcuidle(CALLER_ADDR0, caller_addr); - tracer_hardirqs_on(CALLER_ADDR0, caller_addr); - this_cpu_write(tracing_irq_cpu, 0); - } - - lockdep_hardirqs_on_prepare(); - lockdep_hardirqs_on(caller_addr); -} -EXPORT_SYMBOL(trace_hardirqs_on_caller); -NOKPROBE_SYMBOL(trace_hardirqs_on_caller); - -__visible void trace_hardirqs_off_caller(unsigned long caller_addr) -{ - lockdep_hardirqs_off(caller_addr); - - if (!this_cpu_read(tracing_irq_cpu)) { - this_cpu_write(tracing_irq_cpu, 1); - tracer_hardirqs_off(CALLER_ADDR0, caller_addr); - if (!in_nmi()) - trace_irq_disable_rcuidle(CALLER_ADDR0, caller_addr); - } -} -EXPORT_SYMBOL(trace_hardirqs_off_caller); -NOKPROBE_SYMBOL(trace_hardirqs_off_caller); #endif /* CONFIG_TRACE_IRQFLAGS */ #ifdef CONFIG_TRACE_PREEMPT_TOGGLE -- cgit v1.2.3 From 408b961146be4c1a776ce285c3c289afab15298a Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 12 Jan 2023 20:43:48 +0100 Subject: tracing: WARN on rcuidle ARCH_WANTS_NO_INSTR (a superset of CONFIG_GENERIC_ENTRY) disallows any and all tracing when RCU isn't enabled. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Tested-by: Tony Lindgren Tested-by: Ulf Hansson Acked-by: Rafael J. Wysocki Acked-by: Frederic Weisbecker Link: https://lore.kernel.org/r/20230112195541.416110581@infradead.org --- include/linux/tracepoint.h | 15 +++++++++++++-- kernel/trace/trace.c | 3 +++ 2 files changed, 16 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/include/linux/tracepoint.h b/include/linux/tracepoint.h index 4b33b95eb8be..552f80b8362f 100644 --- a/include/linux/tracepoint.h +++ b/include/linux/tracepoint.h @@ -177,6 +177,17 @@ static inline struct tracepoint *tracepoint_ptr_deref(tracepoint_ptr_t *p) #define __DO_TRACE_CALL(name, args) __traceiter_##name(NULL, args) #endif /* CONFIG_HAVE_STATIC_CALL */ +/* + * ARCH_WANTS_NO_INSTR archs are expected to have sanitized entry and idle + * code that disallow any/all tracing/instrumentation when RCU isn't watching. + */ +#ifdef CONFIG_ARCH_WANTS_NO_INSTR +#define RCUIDLE_COND(rcuidle) (rcuidle) +#else +/* srcu can't be used from NMI */ +#define RCUIDLE_COND(rcuidle) (rcuidle && in_nmi()) +#endif + /* * it_func[0] is never NULL because there is at least one element in the array * when the array itself is non NULL. @@ -188,8 +199,8 @@ static inline struct tracepoint *tracepoint_ptr_deref(tracepoint_ptr_t *p) if (!(cond)) \ return; \ \ - /* srcu can't be used from NMI */ \ - WARN_ON_ONCE(rcuidle && in_nmi()); \ + if (WARN_ON_ONCE(RCUIDLE_COND(rcuidle))) \ + return; \ \ /* keep srcu and sched-rcu usage consistent */ \ preempt_disable_notrace(); \ diff --git a/kernel/trace/trace.c b/kernel/trace/trace.c index a555a861b978..54ec1592ac79 100644 --- a/kernel/trace/trace.c +++ b/kernel/trace/trace.c @@ -3128,6 +3128,9 @@ void __trace_stack(struct trace_array *tr, unsigned int trace_ctx, return; } + if (WARN_ON_ONCE(IS_ENABLED(CONFIG_GENERIC_ENTRY))) + return; + /* * When an NMI triggers, RCU is enabled via ct_nmi_enter(), * but if the above rcu_is_watching() failed, then the NMI -- cgit v1.2.3 From 9aedeaed6fc6fe8452b9b8225e95cc2b8631ff91 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 12 Jan 2023 20:43:49 +0100 Subject: tracing, hardirq: No moar _rcuidle() tracing Robot reported that trace_hardirqs_{on,off}() tickle the forbidden _rcuidle() tracepoint through local_irq_{en,dis}able(). For 'sane' configs, these calls will only happen with RCU enabled and as such can use the regular tracepoint. This also means it's possible to trace them from NMI context again. Reported-by: kernel test robot Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20230112195541.477416709@infradead.org --- kernel/trace/trace_preemptirq.c | 26 ++++++++++++++++++-------- 1 file changed, 18 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_preemptirq.c b/kernel/trace/trace_preemptirq.c index 629f2854e12b..f992444a0b1f 100644 --- a/kernel/trace/trace_preemptirq.c +++ b/kernel/trace/trace_preemptirq.c @@ -19,6 +19,20 @@ /* Per-cpu variable to prevent redundant calls when IRQs already off */ static DEFINE_PER_CPU(int, tracing_irq_cpu); +/* + * Use regular trace points on architectures that implement noinstr + * tooling: these calls will only happen with RCU enabled, which can + * use a regular tracepoint. + * + * On older architectures, use the rcuidle tracing methods (which + * aren't NMI-safe - so exclude NMI contexts): + */ +#ifdef CONFIG_ARCH_WANTS_NO_INSTR +#define trace(point) trace_##point +#else +#define trace(point) if (!in_nmi()) trace_##point##_rcuidle +#endif + /* * Like trace_hardirqs_on() but without the lockdep invocation. This is * used in the low level entry code where the ordering vs. RCU is important @@ -28,8 +42,7 @@ static DEFINE_PER_CPU(int, tracing_irq_cpu); void trace_hardirqs_on_prepare(void) { if (this_cpu_read(tracing_irq_cpu)) { - if (!in_nmi()) - trace_irq_enable(CALLER_ADDR0, CALLER_ADDR1); + trace(irq_enable)(CALLER_ADDR0, CALLER_ADDR1); tracer_hardirqs_on(CALLER_ADDR0, CALLER_ADDR1); this_cpu_write(tracing_irq_cpu, 0); } @@ -40,8 +53,7 @@ NOKPROBE_SYMBOL(trace_hardirqs_on_prepare); void trace_hardirqs_on(void) { if (this_cpu_read(tracing_irq_cpu)) { - if (!in_nmi()) - trace_irq_enable_rcuidle(CALLER_ADDR0, CALLER_ADDR1); + trace(irq_enable)(CALLER_ADDR0, CALLER_ADDR1); tracer_hardirqs_on(CALLER_ADDR0, CALLER_ADDR1); this_cpu_write(tracing_irq_cpu, 0); } @@ -63,8 +75,7 @@ void trace_hardirqs_off_finish(void) if (!this_cpu_read(tracing_irq_cpu)) { this_cpu_write(tracing_irq_cpu, 1); tracer_hardirqs_off(CALLER_ADDR0, CALLER_ADDR1); - if (!in_nmi()) - trace_irq_disable(CALLER_ADDR0, CALLER_ADDR1); + trace(irq_disable)(CALLER_ADDR0, CALLER_ADDR1); } } @@ -78,8 +89,7 @@ void trace_hardirqs_off(void) if (!this_cpu_read(tracing_irq_cpu)) { this_cpu_write(tracing_irq_cpu, 1); tracer_hardirqs_off(CALLER_ADDR0, CALLER_ADDR1); - if (!in_nmi()) - trace_irq_disable_rcuidle(CALLER_ADDR0, CALLER_ADDR1); + trace(irq_disable)(CALLER_ADDR0, CALLER_ADDR1); } } EXPORT_SYMBOL(trace_hardirqs_off); -- cgit v1.2.3 From 0e26e1de0032779e43929174339429c16307a299 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 12 Jan 2023 20:44:05 +0100 Subject: context_tracking: Fix noinstr vs KASAN Low level noinstr context-tracking code is calling out to instrumented code on KASAN: vmlinux.o: warning: objtool: __ct_user_enter+0x72: call to __kasan_check_write() leaves .noinstr.text section vmlinux.o: warning: objtool: __ct_user_exit+0x47: call to __kasan_check_write() leaves .noinstr.text section Use even lower level atomic methods to avoid the instrumentation. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20230112195542.458034262@infradead.org --- kernel/context_tracking.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'kernel') diff --git a/kernel/context_tracking.c b/kernel/context_tracking.c index 77978e372377..a09f1c19336a 100644 --- a/kernel/context_tracking.c +++ b/kernel/context_tracking.c @@ -510,7 +510,7 @@ void noinstr __ct_user_enter(enum ctx_state state) * In this we case we don't care about any concurrency/ordering. */ if (!IS_ENABLED(CONFIG_CONTEXT_TRACKING_IDLE)) - atomic_set(&ct->state, state); + arch_atomic_set(&ct->state, state); } else { /* * Even if context tracking is disabled on this CPU, because it's outside @@ -527,7 +527,7 @@ void noinstr __ct_user_enter(enum ctx_state state) */ if (!IS_ENABLED(CONFIG_CONTEXT_TRACKING_IDLE)) { /* Tracking for vtime only, no concurrent RCU EQS accounting */ - atomic_set(&ct->state, state); + arch_atomic_set(&ct->state, state); } else { /* * Tracking for vtime and RCU EQS. Make sure we don't race @@ -535,7 +535,7 @@ void noinstr __ct_user_enter(enum ctx_state state) * RCU only requires RCU_DYNTICKS_IDX increments to be fully * ordered. */ - atomic_add(state, &ct->state); + arch_atomic_add(state, &ct->state); } } } @@ -630,12 +630,12 @@ void noinstr __ct_user_exit(enum ctx_state state) * In this we case we don't care about any concurrency/ordering. */ if (!IS_ENABLED(CONFIG_CONTEXT_TRACKING_IDLE)) - atomic_set(&ct->state, CONTEXT_KERNEL); + arch_atomic_set(&ct->state, CONTEXT_KERNEL); } else { if (!IS_ENABLED(CONFIG_CONTEXT_TRACKING_IDLE)) { /* Tracking for vtime only, no concurrent RCU EQS accounting */ - atomic_set(&ct->state, CONTEXT_KERNEL); + arch_atomic_set(&ct->state, CONTEXT_KERNEL); } else { /* * Tracking for vtime and RCU EQS. Make sure we don't race @@ -643,7 +643,7 @@ void noinstr __ct_user_exit(enum ctx_state state) * RCU only requires RCU_DYNTICKS_IDX increments to be fully * ordered. */ - atomic_sub(state, &ct->state); + arch_atomic_sub(state, &ct->state); } } } -- cgit v1.2.3 From 79ba1e607d68178db7d3fe4f6a4aa38f06805e7b Mon Sep 17 00:00:00 2001 From: Vincent Guittot Date: Fri, 13 Jan 2023 14:36:13 +0100 Subject: sched/fair: Limit sched slice duration In presence of a lot of small weight tasks like sched_idle tasks, normal or high weight tasks can see their ideal runtime (sched_slice) to increase to hundreds ms whereas it normally stays below sysctl_sched_latency. 2 normal tasks running on a CPU will have a max sched_slice of 12ms (half of the sched_period). This means that they will make progress every sysctl_sched_latency period. If we now add 1000 idle tasks on the CPU, the sched_period becomes 3006 ms and the ideal runtime of the normal tasks becomes 609 ms. It will even become 1500ms if the idle tasks belongs to an idle cgroup. This means that the scheduler will look for picking another waiting task after 609ms running time (1500ms respectively). The idle tasks change significantly the way the 2 normal tasks interleave their running time slot whereas they should have a small impact. Such long sched_slice can delay significantly the release of resources as the tasks can wait hundreds of ms before the next running slot just because of idle tasks queued on the rq. Cap the ideal_runtime to sysctl_sched_latency to make sure that tasks will regularly make progress and will not be significantly impacted by idle/background tasks queued on the rq. Signed-off-by: Vincent Guittot Signed-off-by: Ingo Molnar Tested-by: Dietmar Eggemann Link: https://lore.kernel.org/r/20230113133613.257342-1-vincent.guittot@linaro.org --- kernel/sched/fair.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index e9d906a9bba9..d4db72f8f84e 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -4896,7 +4896,13 @@ check_preempt_tick(struct cfs_rq *cfs_rq, struct sched_entity *curr) struct sched_entity *se; s64 delta; - ideal_runtime = sched_slice(cfs_rq, curr); + /* + * When many tasks blow up the sched_period; it is possible that + * sched_slice() reports unusually large results (when many tasks are + * very light for example). Therefore impose a maximum. + */ + ideal_runtime = min_t(u64, sched_slice(cfs_rq, curr), sysctl_sched_latency); + delta_exec = curr->sum_exec_runtime - curr->prev_sum_exec_runtime; if (delta_exec > ideal_runtime) { resched_curr(rq_of(cfs_rq)); -- cgit v1.2.3 From 001c28e57187570e4b5aa4492c7a957fb6d65d7b Mon Sep 17 00:00:00 2001 From: Nicholas Piggin Date: Fri, 20 Jan 2023 11:18:20 +1000 Subject: exit: Detect and fix irq disabled state in oops If a task oopses with irqs disabled, this can cause various cascading problems in the oops path such as sleep-from-invalid warnings, and potentially worse. Since commit 0258b5fd7c712 ("coredump: Limit coredumps to a single thread group"), the unconditional irq enable in coredump_task_exit() will "fix" the irq state to be enabled early in do_exit(), so currently this may not be triggerable, but that is coincidental and fragile. Detect and fix the irqs_disabled() condition in the oops path before calling do_exit(), similarly to the way in_atomic() is handled. Reported-by: Michael Ellerman Signed-off-by: Nicholas Piggin Signed-off-by: Peter Zijlstra (Intel) Acked-by: "Eric W. Biederman" Link: https://lore.kernel.org/lkml/20221004094401.708299-1-npiggin@gmail.com/ --- kernel/exit.c | 7 +++++++ 1 file changed, 7 insertions(+) (limited to 'kernel') diff --git a/kernel/exit.c b/kernel/exit.c index 15dc2ec80c46..bccfa4218356 100644 --- a/kernel/exit.c +++ b/kernel/exit.c @@ -807,6 +807,8 @@ void __noreturn do_exit(long code) struct task_struct *tsk = current; int group_dead; + WARN_ON(irqs_disabled()); + synchronize_group_exit(tsk, code); WARN_ON(tsk->plug); @@ -938,6 +940,11 @@ void __noreturn make_task_dead(int signr) if (unlikely(!tsk->pid)) panic("Attempted to kill the idle task!"); + if (unlikely(irqs_disabled())) { + pr_info("note: %s[%d] exited with irqs disabled\n", + current->comm, task_pid_nr(current)); + local_irq_enable(); + } if (unlikely(in_atomic())) { pr_info("note: %s[%d] exited with preempt_count %d\n", current->comm, task_pid_nr(current), -- cgit v1.2.3 From 5a5d7e9badd2cb8065db171961bd30bd3595e4b6 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 26 Jan 2023 16:08:31 +0100 Subject: cpuidle: lib/bug: Disable rcu_is_watching() during WARN/BUG In order to avoid WARN/BUG from generating nested or even recursive warnings, force rcu_is_watching() true during WARN/lockdep_rcu_suspicious(). Notably things like unwinding the stack can trigger rcu_dereference() warnings, which then triggers more unwinding which then triggers more warnings etc.. Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20230126151323.408156109@infradead.org --- include/linux/context_tracking.h | 27 +++++++++++++++++++++++++++ kernel/locking/lockdep.c | 3 +++ kernel/panic.c | 5 +++++ lib/bug.c | 15 ++++++++++++++- 4 files changed, 49 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/include/linux/context_tracking.h b/include/linux/context_tracking.h index dcef4a9e4d63..d4afa8508a80 100644 --- a/include/linux/context_tracking.h +++ b/include/linux/context_tracking.h @@ -130,9 +130,36 @@ static __always_inline unsigned long ct_state_inc(int incby) return arch_atomic_add_return(incby, this_cpu_ptr(&context_tracking.state)); } +static __always_inline bool warn_rcu_enter(void) +{ + bool ret = false; + + /* + * Horrible hack to shut up recursive RCU isn't watching fail since + * lots of the actual reporting also relies on RCU. + */ + preempt_disable_notrace(); + if (rcu_dynticks_curr_cpu_in_eqs()) { + ret = true; + ct_state_inc(RCU_DYNTICKS_IDX); + } + + return ret; +} + +static __always_inline void warn_rcu_exit(bool rcu) +{ + if (rcu) + ct_state_inc(RCU_DYNTICKS_IDX); + preempt_enable_notrace(); +} + #else static inline void ct_idle_enter(void) { } static inline void ct_idle_exit(void) { } + +static __always_inline bool warn_rcu_enter(void) { return false; } +static __always_inline void warn_rcu_exit(bool rcu) { } #endif /* !CONFIG_CONTEXT_TRACKING_IDLE */ #endif diff --git a/kernel/locking/lockdep.c b/kernel/locking/lockdep.c index e3375bc40dad..50d4863974e7 100644 --- a/kernel/locking/lockdep.c +++ b/kernel/locking/lockdep.c @@ -55,6 +55,7 @@ #include #include #include +#include #include @@ -6555,6 +6556,7 @@ void lockdep_rcu_suspicious(const char *file, const int line, const char *s) { struct task_struct *curr = current; int dl = READ_ONCE(debug_locks); + bool rcu = warn_rcu_enter(); /* Note: the following can be executed concurrently, so be careful. */ pr_warn("\n"); @@ -6595,5 +6597,6 @@ void lockdep_rcu_suspicious(const char *file, const int line, const char *s) lockdep_print_held_locks(curr); pr_warn("\nstack backtrace:\n"); dump_stack(); + warn_rcu_exit(rcu); } EXPORT_SYMBOL_GPL(lockdep_rcu_suspicious); diff --git a/kernel/panic.c b/kernel/panic.c index 463c9295bc28..487f5b03bf83 100644 --- a/kernel/panic.c +++ b/kernel/panic.c @@ -34,6 +34,7 @@ #include #include #include +#include #include #include @@ -679,6 +680,7 @@ void __warn(const char *file, int line, void *caller, unsigned taint, void warn_slowpath_fmt(const char *file, int line, unsigned taint, const char *fmt, ...) { + bool rcu = warn_rcu_enter(); struct warn_args args; pr_warn(CUT_HERE); @@ -693,11 +695,13 @@ void warn_slowpath_fmt(const char *file, int line, unsigned taint, va_start(args.args, fmt); __warn(file, line, __builtin_return_address(0), taint, NULL, &args); va_end(args.args); + warn_rcu_exit(rcu); } EXPORT_SYMBOL(warn_slowpath_fmt); #else void __warn_printk(const char *fmt, ...) { + bool rcu = warn_rcu_enter(); va_list args; pr_warn(CUT_HERE); @@ -705,6 +709,7 @@ void __warn_printk(const char *fmt, ...) va_start(args, fmt); vprintk(fmt, args); va_end(args); + warn_rcu_exit(rcu); } EXPORT_SYMBOL(__warn_printk); #endif diff --git a/lib/bug.c b/lib/bug.c index c223a2575b72..e0ff21989990 100644 --- a/lib/bug.c +++ b/lib/bug.c @@ -47,6 +47,7 @@ #include #include #include +#include extern struct bug_entry __start___bug_table[], __stop___bug_table[]; @@ -153,7 +154,7 @@ struct bug_entry *find_bug(unsigned long bugaddr) return module_find_bug(bugaddr); } -enum bug_trap_type report_bug(unsigned long bugaddr, struct pt_regs *regs) +static enum bug_trap_type __report_bug(unsigned long bugaddr, struct pt_regs *regs) { struct bug_entry *bug; const char *file; @@ -209,6 +210,18 @@ enum bug_trap_type report_bug(unsigned long bugaddr, struct pt_regs *regs) return BUG_TRAP_TYPE_BUG; } +enum bug_trap_type report_bug(unsigned long bugaddr, struct pt_regs *regs) +{ + enum bug_trap_type ret; + bool rcu = false; + + rcu = warn_rcu_enter(); + ret = __report_bug(bugaddr, regs); + warn_rcu_exit(rcu); + + return ret; +} + static void clear_once_table(struct bug_entry *start, struct bug_entry *end) { struct bug_entry *bug; -- cgit v1.2.3 From 3017ba4b831bc7fd67cc82e744116b6e45e259a4 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Tue, 31 Jan 2023 09:50:36 +0100 Subject: cpuidle: tracing, preempt: Squash _rcuidle tracing Extend/fix commit: 9aedeaed6fc6 ("tracing, hardirq: No moar _rcuidle() tracing") ... to also cover trace_preempt_{on,off}() which were mysteriously untouched. Fixes: 9aedeaed6fc6 ("tracing, hardirq: No moar _rcuidle() tracing") Reported-by: Mark Rutland Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Tested-by: Mark Rutland Link: https://lkml.kernel.org/r/Y9D5AfnOukWNOZ5q@hirez.programming.kicks-ass.net Link: https://lore.kernel.org/r/Y9jWXKgkxY5EZVwW@hirez.programming.kicks-ass.net --- kernel/trace/trace_preemptirq.c | 14 ++++++-------- 1 file changed, 6 insertions(+), 8 deletions(-) (limited to 'kernel') diff --git a/kernel/trace/trace_preemptirq.c b/kernel/trace/trace_preemptirq.c index f992444a0b1f..e37446f7916e 100644 --- a/kernel/trace/trace_preemptirq.c +++ b/kernel/trace/trace_preemptirq.c @@ -15,10 +15,6 @@ #define CREATE_TRACE_POINTS #include -#ifdef CONFIG_TRACE_IRQFLAGS -/* Per-cpu variable to prevent redundant calls when IRQs already off */ -static DEFINE_PER_CPU(int, tracing_irq_cpu); - /* * Use regular trace points on architectures that implement noinstr * tooling: these calls will only happen with RCU enabled, which can @@ -33,6 +29,10 @@ static DEFINE_PER_CPU(int, tracing_irq_cpu); #define trace(point) if (!in_nmi()) trace_##point##_rcuidle #endif +#ifdef CONFIG_TRACE_IRQFLAGS +/* Per-cpu variable to prevent redundant calls when IRQs already off */ +static DEFINE_PER_CPU(int, tracing_irq_cpu); + /* * Like trace_hardirqs_on() but without the lockdep invocation. This is * used in the low level entry code where the ordering vs. RCU is important @@ -100,15 +100,13 @@ NOKPROBE_SYMBOL(trace_hardirqs_off); void trace_preempt_on(unsigned long a0, unsigned long a1) { - if (!in_nmi()) - trace_preempt_enable_rcuidle(a0, a1); + trace(preempt_enable)(a0, a1); tracer_preempt_on(a0, a1); } void trace_preempt_off(unsigned long a0, unsigned long a1) { - if (!in_nmi()) - trace_preempt_disable_rcuidle(a0, a1); + trace(preempt_disable)(a0, a1); tracer_preempt_off(a0, a1); } #endif -- cgit v1.2.3 From 776f22913b8e50011004c6ae43004711dab7efa5 Mon Sep 17 00:00:00 2001 From: Peter Zijlstra Date: Thu, 26 Jan 2023 16:08:37 +0100 Subject: sched/clock: Make local_clock() noinstr With sched_clock() noinstr, provide a noinstr implementation of local_clock(). Signed-off-by: Peter Zijlstra (Intel) Signed-off-by: Ingo Molnar Link: https://lore.kernel.org/r/20230126151323.760767043@infradead.org --- include/linux/sched/clock.h | 8 +++----- kernel/sched/clock.c | 27 +++++++++++++++++++++------ 2 files changed, 24 insertions(+), 11 deletions(-) (limited to 'kernel') diff --git a/include/linux/sched/clock.h b/include/linux/sched/clock.h index 867d588314e0..ca008f7d3615 100644 --- a/include/linux/sched/clock.h +++ b/include/linux/sched/clock.h @@ -45,7 +45,7 @@ static inline u64 cpu_clock(int cpu) return sched_clock(); } -static inline u64 local_clock(void) +static __always_inline u64 local_clock(void) { return sched_clock(); } @@ -79,10 +79,8 @@ static inline u64 cpu_clock(int cpu) return sched_clock_cpu(cpu); } -static inline u64 local_clock(void) -{ - return sched_clock_cpu(raw_smp_processor_id()); -} +extern u64 local_clock(void); + #endif #ifdef CONFIG_IRQ_TIME_ACCOUNTING diff --git a/kernel/sched/clock.c b/kernel/sched/clock.c index e374c0c923da..5732fa75ebab 100644 --- a/kernel/sched/clock.c +++ b/kernel/sched/clock.c @@ -93,7 +93,7 @@ struct sched_clock_data { static DEFINE_PER_CPU_SHARED_ALIGNED(struct sched_clock_data, sched_clock_data); -notrace static inline struct sched_clock_data *this_scd(void) +static __always_inline struct sched_clock_data *this_scd(void) { return this_cpu_ptr(&sched_clock_data); } @@ -244,12 +244,12 @@ late_initcall(sched_clock_init_late); * min, max except they take wrapping into account */ -notrace static inline u64 wrap_min(u64 x, u64 y) +static __always_inline u64 wrap_min(u64 x, u64 y) { return (s64)(x - y) < 0 ? x : y; } -notrace static inline u64 wrap_max(u64 x, u64 y) +static __always_inline u64 wrap_max(u64 x, u64 y) { return (s64)(x - y) > 0 ? x : y; } @@ -260,7 +260,7 @@ notrace static inline u64 wrap_max(u64 x, u64 y) * - filter out backward motion * - use the GTOD tick value to create a window to filter crazy TSC values */ -notrace static u64 sched_clock_local(struct sched_clock_data *scd) +static __always_inline u64 sched_clock_local(struct sched_clock_data *scd) { u64 now, clock, old_clock, min_clock, max_clock, gtod; s64 delta; @@ -287,13 +287,28 @@ again: clock = wrap_max(clock, min_clock); clock = wrap_min(clock, max_clock); - if (!try_cmpxchg64(&scd->clock, &old_clock, clock)) + if (!arch_try_cmpxchg64(&scd->clock, &old_clock, clock)) goto again; return clock; } -notrace static u64 sched_clock_remote(struct sched_clock_data *scd) +noinstr u64 local_clock(void) +{ + u64 clock; + + if (static_branch_likely(&__sched_clock_stable)) + return sched_clock() + __sched_clock_offset; + + preempt_disable_notrace(); + clock = sched_clock_local(this_scd()); + preempt_enable_notrace(); + + return clock; +} +EXPORT_SYMBOL_GPL(local_clock); + +static notrace u64 sched_clock_remote(struct sched_clock_data *scd) { struct sched_clock_data *my_scd = this_scd(); u64 this_clock, remote_clock; -- cgit v1.2.3 From e5ed0550c04c5469ecdc1634d8aa18c8609590f0 Mon Sep 17 00:00:00 2001 From: Vincent Guittot Date: Wed, 1 Feb 2023 15:36:27 +0100 Subject: sched/fair: unlink misfit task from cpu overutilized By taking into account uclamp_min, the 1:1 relation between task misfit and cpu overutilized is no more true as a task with a small util_avg may not fit a high capacity cpu because of uclamp_min constraint. Add a new state in util_fits_cpu() to reflect the case that task would fit a CPU except for the uclamp_min hint which is a performance requirement. Use -1 to reflect that a CPU doesn't fit only because of uclamp_min so we can use this new value to take additional action to select the best CPU that doesn't match uclamp_min hint. When util_fits_cpu() returns -1, we will continue to look for a possible CPU with better performance, which replaces Capacity Inversion detection with capacity_orig_of() - thermal_load_avg to detect a capacity inversion. Signed-off-by: Vincent Guittot Reviewed-and-tested-by: Qais Yousef Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Dietmar Eggemann Tested-by: Kajetan Puchalski Link: https://lore.kernel.org/r/20230201143628.270912-2-vincent.guittot@linaro.org --- kernel/sched/fair.c | 105 ++++++++++++++++++++++++++++++++++++++++------------ 1 file changed, 82 insertions(+), 23 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 7c46485d65d7..074742f107c0 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -4561,8 +4561,8 @@ static inline int util_fits_cpu(unsigned long util, * handle the case uclamp_min > uclamp_max. */ uclamp_min = min(uclamp_min, uclamp_max); - if (util < uclamp_min && capacity_orig != SCHED_CAPACITY_SCALE) - fits = fits && (uclamp_min <= capacity_orig_thermal); + if (fits && (util < uclamp_min) && (uclamp_min > capacity_orig_thermal)) + return -1; return fits; } @@ -4572,7 +4572,11 @@ static inline int task_fits_cpu(struct task_struct *p, int cpu) unsigned long uclamp_min = uclamp_eff_value(p, UCLAMP_MIN); unsigned long uclamp_max = uclamp_eff_value(p, UCLAMP_MAX); unsigned long util = task_util_est(p); - return util_fits_cpu(util, uclamp_min, uclamp_max, cpu); + /* + * Return true only if the cpu fully fits the task requirements, which + * include the utilization but also the performance hints. + */ + return (util_fits_cpu(util, uclamp_min, uclamp_max, cpu) > 0); } static inline void update_misfit_status(struct task_struct *p, struct rq *rq) @@ -6138,6 +6142,7 @@ static inline bool cpu_overutilized(int cpu) unsigned long rq_util_min = uclamp_rq_get(cpu_rq(cpu), UCLAMP_MIN); unsigned long rq_util_max = uclamp_rq_get(cpu_rq(cpu), UCLAMP_MAX); + /* Return true only if the utilization doesn't fit CPU's capacity */ return !util_fits_cpu(cpu_util_cfs(cpu), rq_util_min, rq_util_max, cpu); } @@ -6931,6 +6936,7 @@ static int select_idle_capacity(struct task_struct *p, struct sched_domain *sd, int target) { unsigned long task_util, util_min, util_max, best_cap = 0; + int fits, best_fits = 0; int cpu, best_cpu = -1; struct cpumask *cpus; @@ -6946,12 +6952,28 @@ select_idle_capacity(struct task_struct *p, struct sched_domain *sd, int target) if (!available_idle_cpu(cpu) && !sched_idle_cpu(cpu)) continue; - if (util_fits_cpu(task_util, util_min, util_max, cpu)) + + fits = util_fits_cpu(task_util, util_min, util_max, cpu); + + /* This CPU fits with all requirements */ + if (fits > 0) return cpu; + /* + * Only the min performance hint (i.e. uclamp_min) doesn't fit. + * Look for the CPU with best capacity. + */ + else if (fits < 0) + cpu_cap = capacity_orig_of(cpu) - thermal_load_avg(cpu_rq(cpu)); - if (cpu_cap > best_cap) { + /* + * First, select CPU which fits better (-1 being better than 0). + * Then, select the one with best capacity at same level. + */ + if ((fits < best_fits) || + ((fits == best_fits) && (cpu_cap > best_cap))) { best_cap = cpu_cap; best_cpu = cpu; + best_fits = fits; } } @@ -6964,7 +6986,11 @@ static inline bool asym_fits_cpu(unsigned long util, int cpu) { if (sched_asym_cpucap_active()) - return util_fits_cpu(util, util_min, util_max, cpu); + /* + * Return true only if the cpu fully fits the task requirements + * which include the utilization and the performance hints. + */ + return (util_fits_cpu(util, util_min, util_max, cpu) > 0); return true; } @@ -7331,6 +7357,9 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) unsigned long p_util_max = uclamp_is_used() ? uclamp_eff_value(p, UCLAMP_MAX) : 1024; struct root_domain *rd = this_rq()->rd; int cpu, best_energy_cpu, target = -1; + int prev_fits = -1, best_fits = -1; + unsigned long best_thermal_cap = 0; + unsigned long prev_thermal_cap = 0; struct sched_domain *sd; struct perf_domain *pd; struct energy_env eenv; @@ -7366,6 +7395,7 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) unsigned long prev_spare_cap = 0; int max_spare_cap_cpu = -1; unsigned long base_energy; + int fits, max_fits = -1; cpumask_and(cpus, perf_domain_span(pd), cpu_online_mask); @@ -7415,7 +7445,9 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) util_min = max(rq_util_min, p_util_min); util_max = max(rq_util_max, p_util_max); } - if (!util_fits_cpu(util, util_min, util_max, cpu)) + + fits = util_fits_cpu(util, util_min, util_max, cpu); + if (!fits) continue; lsub_positive(&cpu_cap, util); @@ -7423,7 +7455,9 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) if (cpu == prev_cpu) { /* Always use prev_cpu as a candidate. */ prev_spare_cap = cpu_cap; - } else if (cpu_cap > max_spare_cap) { + prev_fits = fits; + } else if ((fits > max_fits) || + ((fits == max_fits) && (cpu_cap > max_spare_cap))) { /* * Find the CPU with the maximum spare capacity * among the remaining CPUs in the performance @@ -7431,6 +7465,7 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) */ max_spare_cap = cpu_cap; max_spare_cap_cpu = cpu; + max_fits = fits; } } @@ -7449,26 +7484,50 @@ static int find_energy_efficient_cpu(struct task_struct *p, int prev_cpu) if (prev_delta < base_energy) goto unlock; prev_delta -= base_energy; + prev_thermal_cap = cpu_thermal_cap; best_delta = min(best_delta, prev_delta); } /* Evaluate the energy impact of using max_spare_cap_cpu. */ if (max_spare_cap_cpu >= 0 && max_spare_cap > prev_spare_cap) { + /* Current best energy cpu fits better */ + if (max_fits < best_fits) + continue; + + /* + * Both don't fit performance hint (i.e. uclamp_min) + * but best energy cpu has better capacity. + */ + if ((max_fits < 0) && + (cpu_thermal_cap <= best_thermal_cap)) + continue; + cur_delta = compute_energy(&eenv, pd, cpus, p, max_spare_cap_cpu); /* CPU utilization has changed */ if (cur_delta < base_energy) goto unlock; cur_delta -= base_energy; - if (cur_delta < best_delta) { - best_delta = cur_delta; - best_energy_cpu = max_spare_cap_cpu; - } + + /* + * Both fit for the task but best energy cpu has lower + * energy impact. + */ + if ((max_fits > 0) && (best_fits > 0) && + (cur_delta >= best_delta)) + continue; + + best_delta = cur_delta; + best_energy_cpu = max_spare_cap_cpu; + best_fits = max_fits; + best_thermal_cap = cpu_thermal_cap; } } rcu_read_unlock(); - if (best_delta < prev_delta) + if ((best_fits > prev_fits) || + ((best_fits > 0) && (best_delta < prev_delta)) || + ((best_fits < 0) && (best_thermal_cap > prev_thermal_cap))) target = best_energy_cpu; return target; @@ -10271,24 +10330,23 @@ static struct sched_group *find_busiest_group(struct lb_env *env) */ update_sd_lb_stats(env, &sds); - if (sched_energy_enabled()) { - struct root_domain *rd = env->dst_rq->rd; - - if (rcu_dereference(rd->pd) && !READ_ONCE(rd->overutilized)) - goto out_balanced; - } - - local = &sds.local_stat; - busiest = &sds.busiest_stat; - /* There is no busy sibling group to pull tasks from */ if (!sds.busiest) goto out_balanced; + busiest = &sds.busiest_stat; + /* Misfit tasks should be dealt with regardless of the avg load */ if (busiest->group_type == group_misfit_task) goto force_balance; + if (sched_energy_enabled()) { + struct root_domain *rd = env->dst_rq->rd; + + if (rcu_dereference(rd->pd) && !READ_ONCE(rd->overutilized)) + goto out_balanced; + } + /* ASYM feature bypasses nice load balance check */ if (busiest->group_type == group_asym_packing) goto force_balance; @@ -10301,6 +10359,7 @@ static struct sched_group *find_busiest_group(struct lb_env *env) if (busiest->group_type == group_imbalanced) goto force_balance; + local = &sds.local_stat; /* * If the local group is busier than the selected busiest group * don't try and pull any tasks. -- cgit v1.2.3 From a2e90611b9f425adbbfcdaa5b5e49958ddf6f61b Mon Sep 17 00:00:00 2001 From: Vincent Guittot Date: Wed, 1 Feb 2023 15:36:28 +0100 Subject: sched/fair: Remove capacity inversion detection Remove the capacity inversion detection which is now handled by util_fits_cpu() returning -1 when we need to continue to look for a potential CPU with better performance. This ends up almost reverting patches below except for some comments: commit da07d2f9c153 ("sched/fair: Fixes for capacity inversion detection") commit aa69c36f31aa ("sched/fair: Consider capacity inversion in util_fits_cpu()") commit 44c7b80bffc3 ("sched/fair: Detect capacity inversion") Signed-off-by: Vincent Guittot Signed-off-by: Peter Zijlstra (Intel) Link: https://lore.kernel.org/r/20230201143628.270912-3-vincent.guittot@linaro.org --- kernel/sched/fair.c | 84 ++++------------------------------------------------ kernel/sched/sched.h | 19 ------------ 2 files changed, 5 insertions(+), 98 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index 074742f107c0..c6c8e7f52935 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -4476,17 +4476,9 @@ static inline int util_fits_cpu(unsigned long util, * * For uclamp_max, we can tolerate a drop in performance level as the * goal is to cap the task. So it's okay if it's getting less. - * - * In case of capacity inversion we should honour the inverted capacity - * for both uclamp_min and uclamp_max all the time. */ - capacity_orig = cpu_in_capacity_inversion(cpu); - if (capacity_orig) { - capacity_orig_thermal = capacity_orig; - } else { - capacity_orig = capacity_orig_of(cpu); - capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu); - } + capacity_orig = capacity_orig_of(cpu); + capacity_orig_thermal = capacity_orig - arch_scale_thermal_pressure(cpu); /* * We want to force a task to fit a cpu as implied by uclamp_max. @@ -9027,82 +9019,16 @@ static unsigned long scale_rt_capacity(int cpu) static void update_cpu_capacity(struct sched_domain *sd, int cpu) { - unsigned long capacity_orig = arch_scale_cpu_capacity(cpu); unsigned long capacity = scale_rt_capacity(cpu); struct sched_group *sdg = sd->groups; - struct rq *rq = cpu_rq(cpu); - rq->cpu_capacity_orig = capacity_orig; + cpu_rq(cpu)->cpu_capacity_orig = arch_scale_cpu_capacity(cpu); if (!capacity) capacity = 1; - rq->cpu_capacity = capacity; - - /* - * Detect if the performance domain is in capacity inversion state. - * - * Capacity inversion happens when another perf domain with equal or - * lower capacity_orig_of() ends up having higher capacity than this - * domain after subtracting thermal pressure. - * - * We only take into account thermal pressure in this detection as it's - * the only metric that actually results in *real* reduction of - * capacity due to performance points (OPPs) being dropped/become - * unreachable due to thermal throttling. - * - * We assume: - * * That all cpus in a perf domain have the same capacity_orig - * (same uArch). - * * Thermal pressure will impact all cpus in this perf domain - * equally. - */ - if (sched_energy_enabled()) { - unsigned long inv_cap = capacity_orig - thermal_load_avg(rq); - struct perf_domain *pd; - - rcu_read_lock(); - - pd = rcu_dereference(rq->rd->pd); - rq->cpu_capacity_inverted = 0; - - for (; pd; pd = pd->next) { - struct cpumask *pd_span = perf_domain_span(pd); - unsigned long pd_cap_orig, pd_cap; - - /* We can't be inverted against our own pd */ - if (cpumask_test_cpu(cpu_of(rq), pd_span)) - continue; - - cpu = cpumask_any(pd_span); - pd_cap_orig = arch_scale_cpu_capacity(cpu); - - if (capacity_orig < pd_cap_orig) - continue; - - /* - * handle the case of multiple perf domains have the - * same capacity_orig but one of them is under higher - * thermal pressure. We record it as capacity - * inversion. - */ - if (capacity_orig == pd_cap_orig) { - pd_cap = pd_cap_orig - thermal_load_avg(cpu_rq(cpu)); - - if (pd_cap > inv_cap) { - rq->cpu_capacity_inverted = inv_cap; - break; - } - } else if (pd_cap_orig > inv_cap) { - rq->cpu_capacity_inverted = inv_cap; - break; - } - } - - rcu_read_unlock(); - } - - trace_sched_cpu_capacity_tp(rq); + cpu_rq(cpu)->cpu_capacity = capacity; + trace_sched_cpu_capacity_tp(cpu_rq(cpu)); sdg->sgc->capacity = capacity; sdg->sgc->min_capacity = capacity; diff --git a/kernel/sched/sched.h b/kernel/sched/sched.h index 1072502976df..3e8df6d31c1e 100644 --- a/kernel/sched/sched.h +++ b/kernel/sched/sched.h @@ -1044,7 +1044,6 @@ struct rq { unsigned long cpu_capacity; unsigned long cpu_capacity_orig; - unsigned long cpu_capacity_inverted; struct balance_callback *balance_callback; @@ -2899,24 +2898,6 @@ static inline unsigned long capacity_orig_of(int cpu) return cpu_rq(cpu)->cpu_capacity_orig; } -/* - * Returns inverted capacity if the CPU is in capacity inversion state. - * 0 otherwise. - * - * Capacity inversion detection only considers thermal impact where actual - * performance points (OPPs) gets dropped. - * - * Capacity inversion state happens when another performance domain that has - * equal or lower capacity_orig_of() becomes effectively larger than the perf - * domain this CPU belongs to due to thermal pressure throttling it hard. - * - * See comment in update_cpu_capacity(). - */ -static inline unsigned long cpu_in_capacity_inversion(int cpu) -{ - return cpu_rq(cpu)->cpu_capacity_inverted; -} - /** * enum cpu_util_type - CPU utilization type * @FREQUENCY_UTIL: Utilization used to select frequency -- cgit v1.2.3 From 829c1651e9c4a6f78398d3e67651cef9bb6b42cc Mon Sep 17 00:00:00 2001 From: Zhang Qiao Date: Mon, 30 Jan 2023 13:22:16 +0100 Subject: sched/fair: sanitize vruntime of entity being placed When a scheduling entity is placed onto cfs_rq, its vruntime is pulled to the base level (around cfs_rq->min_vruntime), so that the entity doesn't gain extra boost when placed backwards. However, if the entity being placed wasn't executed for a long time, its vruntime may get too far behind (e.g. while cfs_rq was executing a low-weight hog), which can inverse the vruntime comparison due to s64 overflow. This results in the entity being placed with its original vruntime way forwards, so that it will effectively never get to the cpu. To prevent that, ignore the vruntime of the entity being placed if it didn't execute for much longer than the characteristic sheduler time scale. [rkagan: formatted, adjusted commit log, comments, cutoff value] Signed-off-by: Zhang Qiao Co-developed-by: Roman Kagan Signed-off-by: Roman Kagan Signed-off-by: Peter Zijlstra (Intel) Link: https://lkml.kernel.org/r/20230130122216.3555094-1-rkagan@amazon.de --- kernel/sched/fair.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c index c6c8e7f52935..ff4dbbae3b10 100644 --- a/kernel/sched/fair.c +++ b/kernel/sched/fair.c @@ -4652,6 +4652,7 @@ static void place_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int initial) { u64 vruntime = cfs_rq->min_vruntime; + u64 sleep_time; /* * The 'current' period is already promised to the current tasks, @@ -4681,8 +4682,18 @@ place_entity(struct cfs_rq *cfs_rq, struct sched_entity *se, int initial) vruntime -= thresh; } - /* ensure we never gain time by being placed backwards. */ - se->vruntime = max_vruntime(se->vruntime, vruntime); + /* + * Pull vruntime of the entity being placed to the base level of + * cfs_rq, to prevent boosting it if placed backwards. If the entity + * slept for a long time, don't even try to compare its vruntime with + * the base as it may be too far off and the comparison may get + * inversed due to s64 overflow. + */ + sleep_time = rq_clock_task(rq_of(cfs_rq)) - se->exec_start; + if ((s64)sleep_time > 60LL * NSEC_PER_SEC) + se->vruntime = vruntime; + else + se->vruntime = max_vruntime(se->vruntime, vruntime); } static void check_enqueue_throttle(struct cfs_rq *cfs_rq); -- cgit v1.2.3 From 7ea98dfa44917a201e76d4fe96bf61d76e60f524 Mon Sep 17 00:00:00 2001 From: Valentin Schneider Date: Mon, 6 Feb 2023 14:06:12 +0000 Subject: sched/deadline: Add more reschedule cases to prio_changed_dl() I've been tracking down an issue on a ~5.17ish kernel where: CPUx CPUy
[idle task keeps running here until *something* accidentally sets TIF_NEED_RESCHED] On that kernel, it is quite easy to trigger using rt-tests's deadline_test [1] with the test running on isolated CPUs (this reduces the chance of something unrelated setting TIF_NEED_RESCHED on the idle tasks, making the issue even more obvious as the hung task detector chimes in). I haven't been able to reproduce this using a mainline kernel, even if I revert 2972e3050e35 ("tracing: Make trace_marker{,_raw} stream-like") which gets rid of the lock involved in the above test, *but* I cannot convince myself the issue isn't there from looking at the code. Make prio_changed_dl() issue a reschedule if the current task isn't a deadline one. While at it, ensure a reschedule is emitted when a queued-but-not-current task gets boosted with an earlier deadline that current's. [1]: https://git.kernel.org/pub/scm/utils/rt-tests/rt-tests.git Signed-off-by: Valentin Schneider Signed-off-by: Peter Zijlstra (Intel) Acked-by: Juri Lelli Link: https://lore.kernel.org/r/20230206140612.701871-1-vschneid@redhat.com --- kernel/sched/deadline.c | 42 +++++++++++++++++++++++++++--------------- 1 file changed, 27 insertions(+), 15 deletions(-) (limited to 'kernel') diff --git a/kernel/sched/deadline.c b/kernel/sched/deadline.c index 0d97d54276cc..71b24371a6f7 100644 --- a/kernel/sched/deadline.c +++ b/kernel/sched/deadline.c @@ -2663,17 +2663,20 @@ static void switched_to_dl(struct rq *rq, struct task_struct *p) static void prio_changed_dl(struct rq *rq, struct task_struct *p, int oldprio) { - if (task_on_rq_queued(p) || task_current(rq, p)) { + if (!task_on_rq_queued(p)) + return; + #ifdef CONFIG_SMP - /* - * This might be too much, but unfortunately - * we don't have the old deadline value, and - * we can't argue if the task is increasing - * or lowering its prio, so... - */ - if (!rq->dl.overloaded) - deadline_queue_pull_task(rq); + /* + * This might be too much, but unfortunately + * we don't have the old deadline value, and + * we can't argue if the task is increasing + * or lowering its prio, so... + */ + if (!rq->dl.overloaded) + deadline_queue_pull_task(rq); + if (task_current(rq, p)) { /* * If we now have a earlier deadline task than p, * then reschedule, provided p is still on this @@ -2681,15 +2684,24 @@ static void prio_changed_dl(struct rq *rq, struct task_struct *p, */ if (dl_time_before(rq->dl.earliest_dl.curr, p->dl.deadline)) resched_curr(rq); -#else + } else { /* - * Again, we don't know if p has a earlier - * or later deadline, so let's blindly set a - * (maybe not needed) rescheduling point. + * Current may not be deadline in case p was throttled but we + * have just replenished it (e.g. rt_mutex_setprio()). + * + * Otherwise, if p was given an earlier deadline, reschedule. */ - resched_curr(rq); -#endif /* CONFIG_SMP */ + if (!dl_task(rq->curr) || + dl_time_before(p->dl.deadline, rq->curr->dl.deadline)) + resched_curr(rq); } +#else + /* + * We don't know if p has a earlier or later deadline, so let's blindly + * set a (maybe not needed) rescheduling point. + */ + resched_curr(rq); +#endif } DEFINE_SCHED_CLASS(dl) = { -- cgit v1.2.3 From 7c4a5b89a0b5a57a64b601775b296abf77a9fe97 Mon Sep 17 00:00:00 2001 From: Pietro Borrello Date: Mon, 6 Feb 2023 22:33:54 +0000 Subject: sched/rt: pick_next_rt_entity(): check list_entry Commit 326587b84078 ("sched: fix goto retry in pick_next_task_rt()") removed any path which could make pick_next_rt_entity() return NULL. However, BUG_ON(!rt_se) in _pick_next_task_rt() (the only caller of pick_next_rt_entity()) still checks the error condition, which can never happen, since list_entry() never returns NULL. Remove the BUG_ON check, and instead emit a warning in the only possible error condition here: the queue being empty which should never happen. Fixes: 326587b84078 ("sched: fix goto retry in pick_next_task_rt()") Signed-off-by: Pietro Borrello Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Phil Auld Reviewed-by: Steven Rostedt (Google) Link: https://lore.kernel.org/r/20230128-list-entry-null-check-sched-v3-1-b1a71bd1ac6b@diag.uniroma1.it --- kernel/sched/rt.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'kernel') diff --git a/kernel/sched/rt.c b/kernel/sched/rt.c index ed2a47e4ddae..0a11f44adee5 100644 --- a/kernel/sched/rt.c +++ b/kernel/sched/rt.c @@ -1777,6 +1777,8 @@ static struct sched_rt_entity *pick_next_rt_entity(struct rt_rq *rt_rq) BUG_ON(idx >= MAX_RT_PRIO); queue = array->queue + idx; + if (SCHED_WARN_ON(list_empty(queue))) + return NULL; next = list_entry(queue->next, struct sched_rt_entity, run_list); return next; @@ -1789,7 +1791,8 @@ static struct task_struct *_pick_next_task_rt(struct rq *rq) do { rt_se = pick_next_rt_entity(rt_rq); - BUG_ON(!rt_se); + if (unlikely(!rt_se)) + return NULL; rt_rq = group_rt_rq(rt_se); } while (rt_rq); -- cgit v1.2.3