From aa877175e7a9982233ed8f10cb4bfddd78d82741 Mon Sep 17 00:00:00 2001 From: Boris Ostrovsky Date: Wed, 3 Aug 2016 13:22:28 -0400 Subject: [PATCH 1/4] cpu/hotplug: Prevent alloc/free of irq descriptors during CPU up/down (again) Now that Xen no longer allocates irqs in _cpu_up() we can restore commit: a89941816726 ("hotplug: Prevent alloc/free of irq descriptors during cpu up/down") Signed-off-by: Boris Ostrovsky Reviewed-by: Juergen Gross Acked-by: Thomas Gleixner Cc: Anna-Maria Gleixner Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Sebastian Andrzej Siewior Cc: david.vrabel@citrix.com Cc: xen-devel@lists.xenproject.org Link: http://lkml.kernel.org/r/1470244948-17674-3-git-send-email-boris.ostrovsky@oracle.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/smpboot.c | 11 ----------- kernel/cpu.c | 8 ++++++++ 2 files changed, 8 insertions(+), 11 deletions(-) diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index 2a6e84a30a54..067de612d3fa 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -1108,17 +1108,8 @@ int native_cpu_up(unsigned int cpu, struct task_struct *tidle) common_cpu_up(cpu, tidle); - /* - * We have to walk the irq descriptors to setup the vector - * space for the cpu which comes online. Prevent irq - * alloc/free across the bringup. - */ - irq_lock_sparse(); - err = do_boot_cpu(apicid, cpu, tidle); - if (err) { - irq_unlock_sparse(); pr_err("do_boot_cpu failed(%d) to wakeup CPU#%u\n", err, cpu); return -EIO; } @@ -1136,8 +1127,6 @@ int native_cpu_up(unsigned int cpu, struct task_struct *tidle) touch_nmi_watchdog(); } - irq_unlock_sparse(); - return 0; } diff --git a/kernel/cpu.c b/kernel/cpu.c index 341bf80f80bd..ec12b726fa6f 100644 --- a/kernel/cpu.c +++ b/kernel/cpu.c @@ -349,8 +349,16 @@ static int bringup_cpu(unsigned int cpu) struct task_struct *idle = idle_thread_get(cpu); int ret; + /* + * Some architectures have to walk the irq descriptors to + * setup the vector space for the cpu which comes online. + * Prevent irq alloc/free across the bringup. + */ + irq_lock_sparse(); + /* Arch-specific enabling code. */ ret = __cpu_up(cpu, idle); + irq_unlock_sparse(); if (ret) { cpu_notify(CPU_UP_CANCELED, cpu); return ret; From a724632ca0c84b494875e9367e07e29472c139ba Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 12 Aug 2016 19:49:38 +0200 Subject: [PATCH 2/4] cpu/hotplug: Rework callback invocation logic This is preparation for the following patch. This rework here changes the arguments of cpuhp_invoke_callback(). It passes now `state' and whether `startup' or `teardown' callback should be invoked. The callback then is looked up by the function. The following is a clanup of callers: - cpuhp_issue_call() has one argument less - struct cpuhp_cpu_state (which is used by the hotplug thread) gets also its callback removed. The decision if it is a single callback invocation moved to the `single' variable. Also a `bringup' variable has been added to distinguish between startup and teardown callback. - take_cpu_down() needs to start one step earlier. We always get here via CPUHP_TEARDOWN_CPU callback. Before that change cpuhp_ap_states + CPUHP_TEARDOWN_CPU pointed to an empty entry because TEARDOWN is saved in bp_states for this reason. Now that we use cpuhp_get_step() to lookup the state we must explicitly skip it in order not to invoke it twice. Signed-off-by: Thomas Gleixner Signed-off-by: Sebastian Andrzej Siewior Cc: Mark Rutland Cc: Peter Zijlstra Cc: Will Deacon Cc: rt@linutronix.de Link: http://lkml.kernel.org/r/1471024183-12666-2-git-send-email-bigeasy@linutronix.de Signed-off-by: Thomas Gleixner --- kernel/cpu.c | 162 +++++++++++++++++++++++++-------------------------- 1 file changed, 80 insertions(+), 82 deletions(-) diff --git a/kernel/cpu.c b/kernel/cpu.c index ec12b726fa6f..d36d8e0abfb8 100644 --- a/kernel/cpu.c +++ b/kernel/cpu.c @@ -37,8 +37,9 @@ * @thread: Pointer to the hotplug thread * @should_run: Thread should execute * @rollback: Perform a rollback - * @cb_stat: The state for a single callback (install/uninstall) - * @cb: Single callback function (install/uninstall) + * @single: Single callback invocation + * @bringup: Single callback bringup or teardown selector + * @cb_state: The state for a single callback (install/uninstall) * @result: Result of the operation * @done: Signal completion to the issuer of the task */ @@ -49,8 +50,9 @@ struct cpuhp_cpu_state { struct task_struct *thread; bool should_run; bool rollback; + bool single; + bool bringup; enum cpuhp_state cb_state; - int (*cb)(unsigned int cpu); int result; struct completion done; #endif @@ -79,24 +81,43 @@ static DEFINE_MUTEX(cpuhp_state_mutex); static struct cpuhp_step cpuhp_bp_states[]; static struct cpuhp_step cpuhp_ap_states[]; +static bool cpuhp_is_ap_state(enum cpuhp_state state) +{ + /* + * The extra check for CPUHP_TEARDOWN_CPU is only for documentation + * purposes as that state is handled explicitly in cpu_down. + */ + return state > CPUHP_BRINGUP_CPU && state != CPUHP_TEARDOWN_CPU; +} + +static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state) +{ + struct cpuhp_step *sp; + + sp = cpuhp_is_ap_state(state) ? cpuhp_ap_states : cpuhp_bp_states; + return sp + state; +} + /** * cpuhp_invoke_callback _ Invoke the callbacks for a given state * @cpu: The cpu for which the callback should be invoked * @step: The step in the state machine - * @cb: The callback function to invoke + * @bringup: True if the bringup callback should be invoked * * Called from cpu hotplug and from the state register machinery */ -static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state step, - int (*cb)(unsigned int)) +static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state state, + bool bringup) { struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu); + struct cpuhp_step *step = cpuhp_get_step(state); + int (*cb)(unsigned int cpu) = bringup ? step->startup : step->teardown; int ret = 0; if (cb) { - trace_cpuhp_enter(cpu, st->target, step, cb); + trace_cpuhp_enter(cpu, st->target, state, cb); ret = cb(cpu); - trace_cpuhp_exit(cpu, st->state, step, ret); + trace_cpuhp_exit(cpu, st->state, state, ret); } return ret; } @@ -371,62 +392,55 @@ static int bringup_cpu(unsigned int cpu) /* * Hotplug state machine related functions */ -static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st, - struct cpuhp_step *steps) +static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st) { for (st->state++; st->state < st->target; st->state++) { - struct cpuhp_step *step = steps + st->state; + struct cpuhp_step *step = cpuhp_get_step(st->state); if (!step->skip_onerr) - cpuhp_invoke_callback(cpu, st->state, step->startup); + cpuhp_invoke_callback(cpu, st->state, true); } } static int cpuhp_down_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st, - struct cpuhp_step *steps, enum cpuhp_state target) + enum cpuhp_state target) { enum cpuhp_state prev_state = st->state; int ret = 0; for (; st->state > target; st->state--) { - struct cpuhp_step *step = steps + st->state; - - ret = cpuhp_invoke_callback(cpu, st->state, step->teardown); + ret = cpuhp_invoke_callback(cpu, st->state, false); if (ret) { st->target = prev_state; - undo_cpu_down(cpu, st, steps); + undo_cpu_down(cpu, st); break; } } return ret; } -static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st, - struct cpuhp_step *steps) +static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st) { for (st->state--; st->state > st->target; st->state--) { - struct cpuhp_step *step = steps + st->state; + struct cpuhp_step *step = cpuhp_get_step(st->state); if (!step->skip_onerr) - cpuhp_invoke_callback(cpu, st->state, step->teardown); + cpuhp_invoke_callback(cpu, st->state, false); } } static int cpuhp_up_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st, - struct cpuhp_step *steps, enum cpuhp_state target) + enum cpuhp_state target) { enum cpuhp_state prev_state = st->state; int ret = 0; while (st->state < target) { - struct cpuhp_step *step; - st->state++; - step = steps + st->state; - ret = cpuhp_invoke_callback(cpu, st->state, step->startup); + ret = cpuhp_invoke_callback(cpu, st->state, true); if (ret) { st->target = prev_state; - undo_cpu_up(cpu, st, steps); + undo_cpu_up(cpu, st); break; } } @@ -455,13 +469,13 @@ static int cpuhp_ap_offline(unsigned int cpu, struct cpuhp_cpu_state *st) { enum cpuhp_state target = max((int)st->target, CPUHP_TEARDOWN_CPU); - return cpuhp_down_callbacks(cpu, st, cpuhp_ap_states, target); + return cpuhp_down_callbacks(cpu, st, target); } /* Execute the online startup callbacks. Used to be CPU_ONLINE */ static int cpuhp_ap_online(unsigned int cpu, struct cpuhp_cpu_state *st) { - return cpuhp_up_callbacks(cpu, st, cpuhp_ap_states, st->target); + return cpuhp_up_callbacks(cpu, st, st->target); } /* @@ -484,18 +498,20 @@ static void cpuhp_thread_fun(unsigned int cpu) st->should_run = false; /* Single callback invocation for [un]install ? */ - if (st->cb) { + if (st->single) { if (st->cb_state < CPUHP_AP_ONLINE) { local_irq_disable(); - ret = cpuhp_invoke_callback(cpu, st->cb_state, st->cb); + ret = cpuhp_invoke_callback(cpu, st->cb_state, + st->bringup); local_irq_enable(); } else { - ret = cpuhp_invoke_callback(cpu, st->cb_state, st->cb); + ret = cpuhp_invoke_callback(cpu, st->cb_state, + st->bringup); } } else if (st->rollback) { BUG_ON(st->state < CPUHP_AP_ONLINE_IDLE); - undo_cpu_down(cpu, st, cpuhp_ap_states); + undo_cpu_down(cpu, st); /* * This is a momentary workaround to keep the notifier users * happy. Will go away once we got rid of the notifiers. @@ -517,8 +533,8 @@ static void cpuhp_thread_fun(unsigned int cpu) } /* Invoke a single callback on a remote cpu */ -static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, - int (*cb)(unsigned int)) +static int +cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup) { struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu); @@ -530,10 +546,12 @@ static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, * we invoke the thread function directly. */ if (!st->thread) - return cpuhp_invoke_callback(cpu, state, cb); + return cpuhp_invoke_callback(cpu, state, bringup); st->cb_state = state; - st->cb = cb; + st->single = true; + st->bringup = bringup; + /* * Make sure the above stores are visible before should_run becomes * true. Paired with the mb() above in cpuhp_thread_fun() @@ -549,7 +567,7 @@ static int cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, static void __cpuhp_kick_ap_work(struct cpuhp_cpu_state *st) { st->result = 0; - st->cb = NULL; + st->single = false; /* * Make sure the above stores are visible before should_run becomes * true. Paired with the mb() above in cpuhp_thread_fun() @@ -700,12 +718,16 @@ static int take_cpu_down(void *_param) if (err < 0) return err; + /* + * We get here while we are in CPUHP_TEARDOWN_CPU state and we must not + * do this step again. + */ + WARN_ON(st->state != CPUHP_TEARDOWN_CPU); + st->state--; /* Invoke the former CPU_DYING callbacks */ - for (; st->state > target; st->state--) { - struct cpuhp_step *step = cpuhp_ap_states + st->state; + for (; st->state > target; st->state--) + cpuhp_invoke_callback(cpu, st->state, false); - cpuhp_invoke_callback(cpu, st->state, step->teardown); - } /* Give up timekeeping duties */ tick_handover_do_timer(); /* Park the stopper thread */ @@ -844,7 +866,7 @@ static int __ref _cpu_down(unsigned int cpu, int tasks_frozen, * The AP brought itself down to CPUHP_TEARDOWN_CPU. So we need * to do the further cleanups. */ - ret = cpuhp_down_callbacks(cpu, st, cpuhp_bp_states, target); + ret = cpuhp_down_callbacks(cpu, st, target); if (ret && st->state > CPUHP_TEARDOWN_CPU && st->state < prev_state) { st->target = prev_state; st->rollback = true; @@ -898,11 +920,8 @@ void notify_cpu_starting(unsigned int cpu) enum cpuhp_state target = min((int)st->target, CPUHP_AP_ONLINE); while (st->state < target) { - struct cpuhp_step *step; - st->state++; - step = cpuhp_ap_states + st->state; - cpuhp_invoke_callback(cpu, st->state, step->startup); + cpuhp_invoke_callback(cpu, st->state, true); } } @@ -987,7 +1006,7 @@ static int _cpu_up(unsigned int cpu, int tasks_frozen, enum cpuhp_state target) * responsible for bringing it up to the target state. */ target = min((int)target, CPUHP_BRINGUP_CPU); - ret = cpuhp_up_callbacks(cpu, st, cpuhp_bp_states, target); + ret = cpuhp_up_callbacks(cpu, st, target); out: cpu_hotplug_done(); return ret; @@ -1364,23 +1383,6 @@ static int cpuhp_cb_check(enum cpuhp_state state) return 0; } -static bool cpuhp_is_ap_state(enum cpuhp_state state) -{ - /* - * The extra check for CPUHP_TEARDOWN_CPU is only for documentation - * purposes as that state is handled explicitely in cpu_down. - */ - return state > CPUHP_BRINGUP_CPU && state != CPUHP_TEARDOWN_CPU; -} - -static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state) -{ - struct cpuhp_step *sp; - - sp = cpuhp_is_ap_state(state) ? cpuhp_ap_states : cpuhp_bp_states; - return sp + state; -} - static void cpuhp_store_callbacks(enum cpuhp_state state, const char *name, int (*startup)(unsigned int cpu), @@ -1406,12 +1408,12 @@ static void *cpuhp_get_teardown_cb(enum cpuhp_state state) * Call the startup/teardown function for a step either on the AP or * on the current CPU. */ -static int cpuhp_issue_call(int cpu, enum cpuhp_state state, - int (*cb)(unsigned int), bool bringup) +static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup) { + struct cpuhp_step *sp = cpuhp_get_step(state); int ret; - if (!cb) + if ((bringup && !sp->startup) || (!bringup && !sp->teardown)) return 0; /* * The non AP bound callbacks can fail on bringup. On teardown @@ -1419,11 +1421,11 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, */ #ifdef CONFIG_SMP if (cpuhp_is_ap_state(state)) - ret = cpuhp_invoke_ap_callback(cpu, state, cb); + ret = cpuhp_invoke_ap_callback(cpu, state, bringup); else - ret = cpuhp_invoke_callback(cpu, state, cb); + ret = cpuhp_invoke_callback(cpu, state, bringup); #else - ret = cpuhp_invoke_callback(cpu, state, cb); + ret = cpuhp_invoke_callback(cpu, state, bringup); #endif BUG_ON(ret && !bringup); return ret; @@ -1434,14 +1436,10 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, * * Note: The teardown callbacks for rollback are not allowed to fail! */ -static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state, - int (*teardown)(unsigned int cpu)) +static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state) { int cpu; - if (!teardown) - return; - /* Roll back the already executed steps on the other cpus */ for_each_present_cpu(cpu) { struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu); @@ -1452,7 +1450,7 @@ static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state, /* Did we invoke the startup call on that cpu ? */ if (cpustate >= state) - cpuhp_issue_call(cpu, state, teardown, false); + cpuhp_issue_call(cpu, state, false); } } @@ -1527,9 +1525,10 @@ int __cpuhp_setup_state(enum cpuhp_state state, if (cpustate < state) continue; - ret = cpuhp_issue_call(cpu, state, startup, true); + ret = cpuhp_issue_call(cpu, state, true); if (ret) { - cpuhp_rollback_install(cpu, state, teardown); + if (teardown) + cpuhp_rollback_install(cpu, state); cpuhp_store_callbacks(state, NULL, NULL, NULL); goto out; } @@ -1553,14 +1552,13 @@ EXPORT_SYMBOL(__cpuhp_setup_state); */ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke) { - int (*teardown)(unsigned int cpu) = cpuhp_get_teardown_cb(state); int cpu; BUG_ON(cpuhp_cb_check(state)); get_online_cpus(); - if (!invoke || !teardown) + if (!invoke || !cpuhp_get_teardown_cb(state)) goto remove; /* @@ -1573,7 +1571,7 @@ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke) int cpustate = st->state; if (cpustate >= state) - cpuhp_issue_call(cpu, state, teardown, false); + cpuhp_issue_call(cpu, state, false); } remove: cpuhp_store_callbacks(state, NULL, NULL, NULL); From cf392d10b69e6e6c57ceea48b347a2ab1a4b75b2 Mon Sep 17 00:00:00 2001 From: Thomas Gleixner Date: Fri, 12 Aug 2016 19:49:39 +0200 Subject: [PATCH 3/4] cpu/hotplug: Add multi instance support This patch adds the ability for a given state to have multiple instances. Until now all states have a single instance and the startup / teardown callback use global variables. A few drivers need to perform a the same callbacks on multiple "instances". Currently we have three drivers in tree which all have a global list which they iterate over. With multi instance they support don't need their private list and the functionality has been moved into core code. Plus we hold the hotplug lock in core so no cpus comes/goes while instances are registered and we do rollback in error case :) Signed-off-by: Thomas Gleixner Signed-off-by: Sebastian Andrzej Siewior Cc: Mark Rutland Cc: Peter Zijlstra Cc: Will Deacon Cc: rt@linutronix.de Link: http://lkml.kernel.org/r/1471024183-12666-3-git-send-email-bigeasy@linutronix.de Signed-off-by: Thomas Gleixner --- include/linux/cpuhotplug.h | 110 +++++++++++++++++- include/trace/events/cpuhp.h | 28 +++++ kernel/cpu.c | 216 +++++++++++++++++++++++++++++------ 3 files changed, 317 insertions(+), 37 deletions(-) diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h index 242bf530edfc..dcfe619171b4 100644 --- a/include/linux/cpuhotplug.h +++ b/include/linux/cpuhotplug.h @@ -99,7 +99,7 @@ enum cpuhp_state { int __cpuhp_setup_state(enum cpuhp_state state, const char *name, bool invoke, int (*startup)(unsigned int cpu), - int (*teardown)(unsigned int cpu)); + int (*teardown)(unsigned int cpu), bool multi_instance); /** * cpuhp_setup_state - Setup hotplug state callbacks with calling the callbacks @@ -116,7 +116,7 @@ static inline int cpuhp_setup_state(enum cpuhp_state state, int (*startup)(unsigned int cpu), int (*teardown)(unsigned int cpu)) { - return __cpuhp_setup_state(state, name, true, startup, teardown); + return __cpuhp_setup_state(state, name, true, startup, teardown, false); } /** @@ -135,7 +135,66 @@ static inline int cpuhp_setup_state_nocalls(enum cpuhp_state state, int (*startup)(unsigned int cpu), int (*teardown)(unsigned int cpu)) { - return __cpuhp_setup_state(state, name, false, startup, teardown); + return __cpuhp_setup_state(state, name, false, startup, teardown, + false); +} + +/** + * cpuhp_setup_state_multi - Add callbacks for multi state + * @state: The state for which the calls are installed + * @name: Name of the callback. + * @startup: startup callback function + * @teardown: teardown callback function + * + * Sets the internal multi_instance flag and prepares a state to work as a multi + * instance callback. No callbacks are invoked at this point. The callbacks are + * invoked once an instance for this state are registered via + * @cpuhp_state_add_instance or @cpuhp_state_add_instance_nocalls. + */ +static inline int cpuhp_setup_state_multi(enum cpuhp_state state, + const char *name, + int (*startup)(unsigned int cpu, + struct hlist_node *node), + int (*teardown)(unsigned int cpu, + struct hlist_node *node)) +{ + return __cpuhp_setup_state(state, name, false, + (void *) startup, + (void *) teardown, true); +} + +int __cpuhp_state_add_instance(enum cpuhp_state state, struct hlist_node *node, + bool invoke); + +/** + * cpuhp_state_add_instance - Add an instance for a state and invoke startup + * callback. + * @state: The state for which the instance is installed + * @node: The node for this individual state. + * + * Installs the instance for the @state and invokes the startup callback on + * the present cpus which have already reached the @state. The @state must have + * been earlier marked as multi-instance by @cpuhp_setup_state_multi. + */ +static inline int cpuhp_state_add_instance(enum cpuhp_state state, + struct hlist_node *node) +{ + return __cpuhp_state_add_instance(state, node, true); +} + +/** + * cpuhp_state_add_instance_nocalls - Add an instance for a state without + * invoking the startup callback. + * @state: The state for which the instance is installed + * @node: The node for this individual state. + * + * Installs the instance for the @state The @state must have been earlier + * marked as multi-instance by @cpuhp_setup_state_multi. + */ +static inline int cpuhp_state_add_instance_nocalls(enum cpuhp_state state, + struct hlist_node *node) +{ + return __cpuhp_state_add_instance(state, node, false); } void __cpuhp_remove_state(enum cpuhp_state state, bool invoke); @@ -162,6 +221,51 @@ static inline void cpuhp_remove_state_nocalls(enum cpuhp_state state) __cpuhp_remove_state(state, false); } +/** + * cpuhp_remove_multi_state - Remove hotplug multi state callback + * @state: The state for which the calls are removed + * + * Removes the callback functions from a multi state. This is the reverse of + * cpuhp_setup_state_multi(). All instances should have been removed before + * invoking this function. + */ +static inline void cpuhp_remove_multi_state(enum cpuhp_state state) +{ + __cpuhp_remove_state(state, false); +} + +int __cpuhp_state_remove_instance(enum cpuhp_state state, + struct hlist_node *node, bool invoke); + +/** + * cpuhp_state_remove_instance - Remove hotplug instance from state and invoke + * the teardown callback + * @state: The state from which the instance is removed + * @node: The node for this individual state. + * + * Removes the instance and invokes the teardown callback on the present cpus + * which have already reached the @state. + */ +static inline int cpuhp_state_remove_instance(enum cpuhp_state state, + struct hlist_node *node) +{ + return __cpuhp_state_remove_instance(state, node, true); +} + +/** + * cpuhp_state_remove_instance_nocalls - Remove hotplug instance from state + * without invoking the reatdown callback + * @state: The state from which the instance is removed + * @node: The node for this individual state. + * + * Removes the instance without invoking the teardown callback. + */ +static inline int cpuhp_state_remove_instance_nocalls(enum cpuhp_state state, + struct hlist_node *node) +{ + return __cpuhp_state_remove_instance(state, node, false); +} + #ifdef CONFIG_SMP void cpuhp_online_idle(enum cpuhp_state state); #else diff --git a/include/trace/events/cpuhp.h b/include/trace/events/cpuhp.h index a72bd93ec7e5..996953db91d7 100644 --- a/include/trace/events/cpuhp.h +++ b/include/trace/events/cpuhp.h @@ -33,6 +33,34 @@ TRACE_EVENT(cpuhp_enter, __entry->cpu, __entry->target, __entry->idx, __entry->fun) ); +TRACE_EVENT(cpuhp_multi_enter, + + TP_PROTO(unsigned int cpu, + int target, + int idx, + int (*fun)(unsigned int, struct hlist_node *), + struct hlist_node *node), + + TP_ARGS(cpu, target, idx, fun, node), + + TP_STRUCT__entry( + __field( unsigned int, cpu ) + __field( int, target ) + __field( int, idx ) + __field( void *, fun ) + ), + + TP_fast_assign( + __entry->cpu = cpu; + __entry->target = target; + __entry->idx = idx; + __entry->fun = fun; + ), + + TP_printk("cpu: %04u target: %3d step: %3d (%pf)", + __entry->cpu, __entry->target, __entry->idx, __entry->fun) +); + TRACE_EVENT(cpuhp_exit, TP_PROTO(unsigned int cpu, diff --git a/kernel/cpu.c b/kernel/cpu.c index d36d8e0abfb8..c506485eaa75 100644 --- a/kernel/cpu.c +++ b/kernel/cpu.c @@ -52,6 +52,7 @@ struct cpuhp_cpu_state { bool rollback; bool single; bool bringup; + struct hlist_node *node; enum cpuhp_state cb_state; int result; struct completion done; @@ -70,11 +71,21 @@ static DEFINE_PER_CPU(struct cpuhp_cpu_state, cpuhp_state); * @cant_stop: Bringup/teardown can't be stopped at this step */ struct cpuhp_step { - const char *name; - int (*startup)(unsigned int cpu); - int (*teardown)(unsigned int cpu); - bool skip_onerr; - bool cant_stop; + const char *name; + union { + int (*startup)(unsigned int cpu); + int (*startup_multi)(unsigned int cpu, + struct hlist_node *node); + }; + union { + int (*teardown)(unsigned int cpu); + int (*teardown_multi)(unsigned int cpu, + struct hlist_node *node); + }; + struct hlist_head list; + bool skip_onerr; + bool cant_stop; + bool multi_instance; }; static DEFINE_MUTEX(cpuhp_state_mutex); @@ -104,20 +115,59 @@ static struct cpuhp_step *cpuhp_get_step(enum cpuhp_state state) * @step: The step in the state machine * @bringup: True if the bringup callback should be invoked * - * Called from cpu hotplug and from the state register machinery + * Called from cpu hotplug and from the state register machinery. */ static int cpuhp_invoke_callback(unsigned int cpu, enum cpuhp_state state, - bool bringup) + bool bringup, struct hlist_node *node) { struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu); struct cpuhp_step *step = cpuhp_get_step(state); - int (*cb)(unsigned int cpu) = bringup ? step->startup : step->teardown; - int ret = 0; + int (*cbm)(unsigned int cpu, struct hlist_node *node); + int (*cb)(unsigned int cpu); + int ret, cnt; - if (cb) { + if (!step->multi_instance) { + cb = bringup ? step->startup : step->teardown; + if (!cb) + return 0; trace_cpuhp_enter(cpu, st->target, state, cb); ret = cb(cpu); trace_cpuhp_exit(cpu, st->state, state, ret); + return ret; + } + cbm = bringup ? step->startup_multi : step->teardown_multi; + if (!cbm) + return 0; + + /* Single invocation for instance add/remove */ + if (node) { + trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node); + ret = cbm(cpu, node); + trace_cpuhp_exit(cpu, st->state, state, ret); + return ret; + } + + /* State transition. Invoke on all instances */ + cnt = 0; + hlist_for_each(node, &step->list) { + trace_cpuhp_multi_enter(cpu, st->target, state, cbm, node); + ret = cbm(cpu, node); + trace_cpuhp_exit(cpu, st->state, state, ret); + if (ret) + goto err; + cnt++; + } + return 0; +err: + /* Rollback the instances if one failed */ + cbm = !bringup ? step->startup_multi : step->teardown_multi; + if (!cbm) + return ret; + + hlist_for_each(node, &step->list) { + if (!cnt--) + break; + cbm(cpu, node); } return ret; } @@ -398,7 +448,7 @@ static void undo_cpu_down(unsigned int cpu, struct cpuhp_cpu_state *st) struct cpuhp_step *step = cpuhp_get_step(st->state); if (!step->skip_onerr) - cpuhp_invoke_callback(cpu, st->state, true); + cpuhp_invoke_callback(cpu, st->state, true, NULL); } } @@ -409,7 +459,7 @@ static int cpuhp_down_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st, int ret = 0; for (; st->state > target; st->state--) { - ret = cpuhp_invoke_callback(cpu, st->state, false); + ret = cpuhp_invoke_callback(cpu, st->state, false, NULL); if (ret) { st->target = prev_state; undo_cpu_down(cpu, st); @@ -425,7 +475,7 @@ static void undo_cpu_up(unsigned int cpu, struct cpuhp_cpu_state *st) struct cpuhp_step *step = cpuhp_get_step(st->state); if (!step->skip_onerr) - cpuhp_invoke_callback(cpu, st->state, false); + cpuhp_invoke_callback(cpu, st->state, false, NULL); } } @@ -437,7 +487,7 @@ static int cpuhp_up_callbacks(unsigned int cpu, struct cpuhp_cpu_state *st, while (st->state < target) { st->state++; - ret = cpuhp_invoke_callback(cpu, st->state, true); + ret = cpuhp_invoke_callback(cpu, st->state, true, NULL); if (ret) { st->target = prev_state; undo_cpu_up(cpu, st); @@ -502,11 +552,11 @@ static void cpuhp_thread_fun(unsigned int cpu) if (st->cb_state < CPUHP_AP_ONLINE) { local_irq_disable(); ret = cpuhp_invoke_callback(cpu, st->cb_state, - st->bringup); + st->bringup, st->node); local_irq_enable(); } else { ret = cpuhp_invoke_callback(cpu, st->cb_state, - st->bringup); + st->bringup, st->node); } } else if (st->rollback) { BUG_ON(st->state < CPUHP_AP_ONLINE_IDLE); @@ -534,7 +584,8 @@ static void cpuhp_thread_fun(unsigned int cpu) /* Invoke a single callback on a remote cpu */ static int -cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup) +cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup, + struct hlist_node *node) { struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu); @@ -546,11 +597,12 @@ cpuhp_invoke_ap_callback(int cpu, enum cpuhp_state state, bool bringup) * we invoke the thread function directly. */ if (!st->thread) - return cpuhp_invoke_callback(cpu, state, bringup); + return cpuhp_invoke_callback(cpu, state, bringup, node); st->cb_state = state; st->single = true; st->bringup = bringup; + st->node = node; /* * Make sure the above stores are visible before should_run becomes @@ -726,7 +778,7 @@ static int take_cpu_down(void *_param) st->state--; /* Invoke the former CPU_DYING callbacks */ for (; st->state > target; st->state--) - cpuhp_invoke_callback(cpu, st->state, false); + cpuhp_invoke_callback(cpu, st->state, false, NULL); /* Give up timekeeping duties */ tick_handover_do_timer(); @@ -921,7 +973,7 @@ void notify_cpu_starting(unsigned int cpu) while (st->state < target) { st->state++; - cpuhp_invoke_callback(cpu, st->state, true); + cpuhp_invoke_callback(cpu, st->state, true, NULL); } } @@ -1386,7 +1438,8 @@ static int cpuhp_cb_check(enum cpuhp_state state) static void cpuhp_store_callbacks(enum cpuhp_state state, const char *name, int (*startup)(unsigned int cpu), - int (*teardown)(unsigned int cpu)) + int (*teardown)(unsigned int cpu), + bool multi_instance) { /* (Un)Install the callbacks for further cpu hotplug operations */ struct cpuhp_step *sp; @@ -1396,6 +1449,8 @@ static void cpuhp_store_callbacks(enum cpuhp_state state, sp->startup = startup; sp->teardown = teardown; sp->name = name; + sp->multi_instance = multi_instance; + INIT_HLIST_HEAD(&sp->list); mutex_unlock(&cpuhp_state_mutex); } @@ -1408,7 +1463,8 @@ static void *cpuhp_get_teardown_cb(enum cpuhp_state state) * Call the startup/teardown function for a step either on the AP or * on the current CPU. */ -static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup) +static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup, + struct hlist_node *node) { struct cpuhp_step *sp = cpuhp_get_step(state); int ret; @@ -1421,11 +1477,11 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup) */ #ifdef CONFIG_SMP if (cpuhp_is_ap_state(state)) - ret = cpuhp_invoke_ap_callback(cpu, state, bringup); + ret = cpuhp_invoke_ap_callback(cpu, state, bringup, node); else - ret = cpuhp_invoke_callback(cpu, state, bringup); + ret = cpuhp_invoke_callback(cpu, state, bringup, node); #else - ret = cpuhp_invoke_callback(cpu, state, bringup); + ret = cpuhp_invoke_callback(cpu, state, bringup, node); #endif BUG_ON(ret && !bringup); return ret; @@ -1436,7 +1492,8 @@ static int cpuhp_issue_call(int cpu, enum cpuhp_state state, bool bringup) * * Note: The teardown callbacks for rollback are not allowed to fail! */ -static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state) +static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state, + struct hlist_node *node) { int cpu; @@ -1450,7 +1507,7 @@ static void cpuhp_rollback_install(int failedcpu, enum cpuhp_state state) /* Did we invoke the startup call on that cpu ? */ if (cpustate >= state) - cpuhp_issue_call(cpu, state, false); + cpuhp_issue_call(cpu, state, false, node); } } @@ -1477,6 +1534,52 @@ static int cpuhp_reserve_state(enum cpuhp_state state) return -ENOSPC; } +int __cpuhp_state_add_instance(enum cpuhp_state state, struct hlist_node *node, + bool invoke) +{ + struct cpuhp_step *sp; + int cpu; + int ret; + + sp = cpuhp_get_step(state); + if (sp->multi_instance == false) + return -EINVAL; + + get_online_cpus(); + + if (!invoke || !sp->startup_multi) + goto add_node; + + /* + * Try to call the startup callback for each present cpu + * depending on the hotplug state of the cpu. + */ + for_each_present_cpu(cpu) { + struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu); + int cpustate = st->state; + + if (cpustate < state) + continue; + + ret = cpuhp_issue_call(cpu, state, true, node); + if (ret) { + if (sp->teardown_multi) + cpuhp_rollback_install(cpu, state, node); + goto err; + } + } +add_node: + ret = 0; + mutex_lock(&cpuhp_state_mutex); + hlist_add_head(node, &sp->list); + mutex_unlock(&cpuhp_state_mutex); + +err: + put_online_cpus(); + return ret; +} +EXPORT_SYMBOL_GPL(__cpuhp_state_add_instance); + /** * __cpuhp_setup_state - Setup the callbacks for an hotplug machine state * @state: The state to setup @@ -1490,7 +1593,8 @@ static int cpuhp_reserve_state(enum cpuhp_state state) int __cpuhp_setup_state(enum cpuhp_state state, const char *name, bool invoke, int (*startup)(unsigned int cpu), - int (*teardown)(unsigned int cpu)) + int (*teardown)(unsigned int cpu), + bool multi_instance) { int cpu, ret = 0; int dyn_state = 0; @@ -1509,7 +1613,7 @@ int __cpuhp_setup_state(enum cpuhp_state state, state = ret; } - cpuhp_store_callbacks(state, name, startup, teardown); + cpuhp_store_callbacks(state, name, startup, teardown, multi_instance); if (!invoke || !startup) goto out; @@ -1525,11 +1629,11 @@ int __cpuhp_setup_state(enum cpuhp_state state, if (cpustate < state) continue; - ret = cpuhp_issue_call(cpu, state, true); + ret = cpuhp_issue_call(cpu, state, true, NULL); if (ret) { if (teardown) - cpuhp_rollback_install(cpu, state); - cpuhp_store_callbacks(state, NULL, NULL, NULL); + cpuhp_rollback_install(cpu, state, NULL); + cpuhp_store_callbacks(state, NULL, NULL, NULL, false); goto out; } } @@ -1541,6 +1645,42 @@ out: } EXPORT_SYMBOL(__cpuhp_setup_state); +int __cpuhp_state_remove_instance(enum cpuhp_state state, + struct hlist_node *node, bool invoke) +{ + struct cpuhp_step *sp = cpuhp_get_step(state); + int cpu; + + BUG_ON(cpuhp_cb_check(state)); + + if (!sp->multi_instance) + return -EINVAL; + + get_online_cpus(); + if (!invoke || !cpuhp_get_teardown_cb(state)) + goto remove; + /* + * Call the teardown callback for each present cpu depending + * on the hotplug state of the cpu. This function is not + * allowed to fail currently! + */ + for_each_present_cpu(cpu) { + struct cpuhp_cpu_state *st = per_cpu_ptr(&cpuhp_state, cpu); + int cpustate = st->state; + + if (cpustate >= state) + cpuhp_issue_call(cpu, state, false, node); + } + +remove: + mutex_lock(&cpuhp_state_mutex); + hlist_del(node); + mutex_unlock(&cpuhp_state_mutex); + put_online_cpus(); + + return 0; +} +EXPORT_SYMBOL_GPL(__cpuhp_state_remove_instance); /** * __cpuhp_remove_state - Remove the callbacks for an hotplug machine state * @state: The state to remove @@ -1552,12 +1692,20 @@ EXPORT_SYMBOL(__cpuhp_setup_state); */ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke) { + struct cpuhp_step *sp = cpuhp_get_step(state); int cpu; BUG_ON(cpuhp_cb_check(state)); get_online_cpus(); + if (sp->multi_instance) { + WARN(!hlist_empty(&sp->list), + "Error: Removing state %d which has instances left.\n", + state); + goto remove; + } + if (!invoke || !cpuhp_get_teardown_cb(state)) goto remove; @@ -1571,10 +1719,10 @@ void __cpuhp_remove_state(enum cpuhp_state state, bool invoke) int cpustate = st->state; if (cpustate >= state) - cpuhp_issue_call(cpu, state, false); + cpuhp_issue_call(cpu, state, false, NULL); } remove: - cpuhp_store_callbacks(state, NULL, NULL, NULL); + cpuhp_store_callbacks(state, NULL, NULL, NULL, false); put_online_cpus(); } EXPORT_SYMBOL(__cpuhp_remove_state); From e2a738f7a88f32622684d972d654a9fed026555f Mon Sep 17 00:00:00 2001 From: Sebastian Andrzej Siewior Date: Tue, 6 Sep 2016 19:04:55 +0200 Subject: [PATCH 4/4] blk/mq: Reserve hotplug states for block multiqueue This patch only reserves two CPU hotplug states for block/mq so the block tree can apply the conversion patches. Signed-off-by: Sebastian Andrzej Siewior Cc: Peter Zijlstra Cc: Jens Axboe Cc: rt@linutronix.de Link: http://lkml.kernel.org/r/20160906170457.32393-20-bigeasy@linutronix.de Signed-off-by: Thomas Gleixner --- include/linux/cpuhotplug.h | 2 ++ 1 file changed, 2 insertions(+) diff --git a/include/linux/cpuhotplug.h b/include/linux/cpuhotplug.h index dcfe619171b4..2ac07d01bdb5 100644 --- a/include/linux/cpuhotplug.h +++ b/include/linux/cpuhotplug.h @@ -14,6 +14,7 @@ enum cpuhp_state { CPUHP_PERF_SUPERH, CPUHP_X86_HPET_DEAD, CPUHP_X86_APB_DEAD, + CPUHP_BLK_MQ_DEAD, CPUHP_WORKQUEUE_PREP, CPUHP_POWER_NUMA_PREPARE, CPUHP_HRTIMERS_PREPARE, @@ -22,6 +23,7 @@ enum cpuhp_state { CPUHP_SMPCFD_PREPARE, CPUHP_RCUTREE_PREP, CPUHP_NOTIFY_PREPARE, + CPUHP_BLK_MQ_PREPARE, CPUHP_TIMERS_DEAD, CPUHP_BRINGUP_CPU, CPUHP_AP_IDLE_DEAD,