diff options
Diffstat (limited to 'arch/powerpc/kernel/smp.c')
| -rw-r--r-- | arch/powerpc/kernel/smp.c | 1564 |
1 files changed, 1300 insertions, 264 deletions
diff --git a/arch/powerpc/kernel/smp.c b/arch/powerpc/kernel/smp.c index 38b0ba65a735..292fee8809bc 100644 --- a/arch/powerpc/kernel/smp.c +++ b/arch/powerpc/kernel/smp.c @@ -1,3 +1,4 @@ +// SPDX-License-Identifier: GPL-2.0-or-later /* * SMP support for ppc. * @@ -8,18 +9,15 @@ * * PowerPC-64 Support added by Dave Engebretsen, Peter Bergner, and * Mike Corrigan {engebret|bergner|mikec}@us.ibm.com - * - * This program is free software; you can redistribute it and/or - * modify it under the terms of the GNU General Public License - * as published by the Free Software Foundation; either version - * 2 of the License, or (at your option) any later version. */ #undef DEBUG #include <linux/kernel.h> #include <linux/export.h> -#include <linux/sched.h> +#include <linux/sched/mm.h> +#include <linux/sched/task_stack.h> +#include <linux/sched/topology.h> #include <linux/smp.h> #include <linux/interrupt.h> #include <linux/delay.h> @@ -31,16 +29,25 @@ #include <linux/cpu.h> #include <linux/notifier.h> #include <linux/topology.h> +#include <linux/profile.h> +#include <linux/processor.h> +#include <linux/random.h> +#include <linux/stackprotector.h> +#include <linux/pgtable.h> +#include <linux/clockchips.h> +#include <linux/kexec.h> #include <asm/ptrace.h> #include <linux/atomic.h> #include <asm/irq.h> +#include <asm/hw_irq.h> +#include <asm/kvm_ppc.h> +#include <asm/dbell.h> #include <asm/page.h> -#include <asm/pgtable.h> -#include <asm/prom.h> #include <asm/smp.h> #include <asm/time.h> #include <asm/machdep.h> +#include <asm/mmu_context.h> #include <asm/cputhreads.h> #include <asm/cputable.h> #include <asm/mpic.h> @@ -50,6 +57,13 @@ #endif #include <asm/vdso.h> #include <asm/debug.h> +#include <asm/cpu_has_feature.h> +#include <asm/ftrace.h> +#include <asm/kup.h> +#include <asm/fadump.h> +#include <asm/systemcfg.h> + +#include <trace/events/ipi.h> #ifdef DEBUG #include <asm/udbg.h> @@ -63,13 +77,60 @@ static DEFINE_PER_CPU(int, cpu_state) = { 0 }; #endif -struct thread_info *secondary_ti; +struct task_struct *secondary_current; +bool has_big_cores __ro_after_init; +bool coregroup_enabled __ro_after_init; +bool thread_group_shares_l2 __ro_after_init; +bool thread_group_shares_l3 __ro_after_init; DEFINE_PER_CPU(cpumask_var_t, cpu_sibling_map); +DEFINE_PER_CPU(cpumask_var_t, cpu_smallcore_map); +DEFINE_PER_CPU(cpumask_var_t, cpu_l2_cache_map); DEFINE_PER_CPU(cpumask_var_t, cpu_core_map); +static DEFINE_PER_CPU(cpumask_var_t, cpu_coregroup_map); EXPORT_PER_CPU_SYMBOL(cpu_sibling_map); +EXPORT_PER_CPU_SYMBOL(cpu_l2_cache_map); EXPORT_PER_CPU_SYMBOL(cpu_core_map); +EXPORT_SYMBOL_GPL(has_big_cores); + +#define MAX_THREAD_LIST_SIZE 8 +#define THREAD_GROUP_SHARE_L1 1 +#define THREAD_GROUP_SHARE_L2_L3 2 +struct thread_groups { + unsigned int property; + unsigned int nr_groups; + unsigned int threads_per_group; + unsigned int thread_list[MAX_THREAD_LIST_SIZE]; +}; + +/* Maximum number of properties that groups of threads within a core can share */ +#define MAX_THREAD_GROUP_PROPERTIES 2 + +struct thread_groups_list { + unsigned int nr_properties; + struct thread_groups property_tgs[MAX_THREAD_GROUP_PROPERTIES]; +}; + +static struct thread_groups_list tgl[NR_CPUS] __initdata; +/* + * On big-cores system, thread_group_l1_cache_map for each CPU corresponds to + * the set its siblings that share the L1-cache. + */ +DEFINE_PER_CPU(cpumask_var_t, thread_group_l1_cache_map); + +/* + * On some big-cores system, thread_group_l2_cache_map for each CPU + * corresponds to the set its siblings within the core that share the + * L2-cache. + */ +DEFINE_PER_CPU(cpumask_var_t, thread_group_l2_cache_map); + +/* + * On P10, thread_group_l3_cache_map for each CPU is equal to the + * thread_group_l2_cache_map + */ +DEFINE_PER_CPU(cpumask_var_t, thread_group_l3_cache_map); /* SMP operations for this machine */ struct smp_ops_t *smp_ops; @@ -79,20 +140,41 @@ volatile unsigned int cpu_callin_map[NR_CPUS]; int smt_enabled_at_boot = 1; -static void (*crash_ipi_function_ptr)(struct pt_regs *) = NULL; +/* + * Returns 1 if the specified cpu should be brought up during boot. + * Used to inhibit booting threads if they've been disabled or + * limited on the command line + */ +int smp_generic_cpu_bootable(unsigned int nr) +{ + /* Special case - we inhibit secondary thread startup + * during boot if the user requests it. + */ + if (system_state < SYSTEM_RUNNING && cpu_has_feature(CPU_FTR_SMT)) { + if (!smt_enabled_at_boot && cpu_thread_in_core(nr) != 0) + return 0; + if (smt_enabled_at_boot + && cpu_thread_in_core(nr) >= smt_enabled_at_boot) + return 0; + } + + return 1; +} + #ifdef CONFIG_PPC64 int smp_generic_kick_cpu(int nr) { - BUG_ON(nr < 0 || nr >= NR_CPUS); + if (nr < 0 || nr >= nr_cpu_ids) + return -EINVAL; /* * The processor is currently spinning, waiting for the * cpu_start field to become non-zero After we set cpu_start, * the processor will continue on to secondary_start */ - if (!paca[nr].cpu_start) { - paca[nr].cpu_start = 1; + if (!paca_ptrs[nr]->cpu_start) { + paca_ptrs[nr]->cpu_start = 1; smp_mb(); return 0; } @@ -123,38 +205,47 @@ static irqreturn_t reschedule_action(int irq, void *data) return IRQ_HANDLED; } -static irqreturn_t call_function_single_action(int irq, void *data) +#ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST +static irqreturn_t tick_broadcast_ipi_action(int irq, void *data) { - generic_smp_call_function_single_interrupt(); + timer_broadcast_interrupt(); return IRQ_HANDLED; } +#endif -static irqreturn_t debug_ipi_action(int irq, void *data) +#ifdef CONFIG_NMI_IPI +static irqreturn_t nmi_ipi_action(int irq, void *data) { - if (crash_ipi_function_ptr) { - crash_ipi_function_ptr(get_irq_regs()); - return IRQ_HANDLED; - } - -#ifdef CONFIG_DEBUGGER - debugger_ipi(get_irq_regs()); -#endif /* CONFIG_DEBUGGER */ - + smp_handle_nmi_ipi(get_irq_regs()); return IRQ_HANDLED; } +#endif static irq_handler_t smp_ipi_action[] = { [PPC_MSG_CALL_FUNCTION] = call_function_action, [PPC_MSG_RESCHEDULE] = reschedule_action, - [PPC_MSG_CALL_FUNC_SINGLE] = call_function_single_action, - [PPC_MSG_DEBUGGER_BREAK] = debug_ipi_action, +#ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST + [PPC_MSG_TICK_BROADCAST] = tick_broadcast_ipi_action, +#endif +#ifdef CONFIG_NMI_IPI + [PPC_MSG_NMI_IPI] = nmi_ipi_action, +#endif }; +/* + * The NMI IPI is a fallback and not truly non-maskable. It is simpler + * than going through the call function infrastructure, and strongly + * serialized, so it is more appropriate for debugging. + */ const char *smp_ipi_name[] = { [PPC_MSG_CALL_FUNCTION] = "ipi call function", [PPC_MSG_RESCHEDULE] = "ipi reschedule", - [PPC_MSG_CALL_FUNC_SINGLE] = "ipi call function single", - [PPC_MSG_DEBUGGER_BREAK] = "ipi debugger", +#ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST + [PPC_MSG_TICK_BROADCAST] = "ipi tick-broadcast", +#endif +#ifdef CONFIG_NMI_IPI + [PPC_MSG_NMI_IPI] = "nmi ipi", +#endif }; /* optional function to request ipi, for controllers with >= 4 ipis */ @@ -162,17 +253,16 @@ int smp_request_message_ipi(int virq, int msg) { int err; - if (msg < 0 || msg > PPC_MSG_DEBUGGER_BREAK) { + if (msg < 0 || msg > PPC_MSG_NMI_IPI) return -EINVAL; - } -#if !defined(CONFIG_DEBUGGER) && !defined(CONFIG_KEXEC) - if (msg == PPC_MSG_DEBUGGER_BREAK) { +#ifndef CONFIG_NMI_IPI + if (msg == PPC_MSG_NMI_IPI) return 1; - } #endif + err = request_irq(virq, smp_ipi_action[msg], IRQF_PERCPU | IRQF_NO_THREAD | IRQF_NO_SUSPEND, - smp_ipi_name[msg], 0); + smp_ipi_name[msg], NULL); WARN(err < 0, "unable to request_irq %d for %s (rc %d)\n", virq, smp_ipi_name[msg], err); @@ -181,19 +271,11 @@ int smp_request_message_ipi(int virq, int msg) #ifdef CONFIG_PPC_SMP_MUXED_IPI struct cpu_messages { - int messages; /* current messages */ - unsigned long data; /* data for cause ipi */ + long messages; /* current messages */ }; static DEFINE_PER_CPU_SHARED_ALIGNED(struct cpu_messages, ipi_message); -void smp_muxed_ipi_set_data(int cpu, unsigned long data) -{ - struct cpu_messages *info = &per_cpu(ipi_message, cpu); - - info->data = data; -} - -void smp_muxed_ipi_message_pass(int cpu, int msg) +void smp_muxed_ipi_set_message(int cpu, int msg) { struct cpu_messages *info = &per_cpu(ipi_message, cpu); char *message = (char *)&info->messages; @@ -202,37 +284,66 @@ void smp_muxed_ipi_message_pass(int cpu, int msg) * Order previous accesses before accesses in the IPI handler. */ smp_mb(); - message[msg] = 1; + WRITE_ONCE(message[msg], 1); +} + +void smp_muxed_ipi_message_pass(int cpu, int msg) +{ + smp_muxed_ipi_set_message(cpu, msg); + /* * cause_ipi functions are required to include a full barrier * before doing whatever causes the IPI. */ - smp_ops->cause_ipi(cpu, info->data); + smp_ops->cause_ipi(cpu); } +#ifdef __BIG_ENDIAN__ +#define IPI_MESSAGE(A) (1uL << ((BITS_PER_LONG - 8) - 8 * (A))) +#else +#define IPI_MESSAGE(A) (1uL << (8 * (A))) +#endif + irqreturn_t smp_ipi_demux(void) { - struct cpu_messages *info = &__get_cpu_var(ipi_message); - unsigned int all; - mb(); /* order any irq clear */ + return smp_ipi_demux_relaxed(); +} + +/* sync-free variant. Callers should ensure synchronization */ +irqreturn_t smp_ipi_demux_relaxed(void) +{ + struct cpu_messages *info; + unsigned long all; + + info = this_cpu_ptr(&ipi_message); do { all = xchg(&info->messages, 0); - -#ifdef __BIG_ENDIAN - if (all & (1 << (24 - 8 * PPC_MSG_CALL_FUNCTION))) +#if defined(CONFIG_KVM_XICS) && defined(CONFIG_KVM_BOOK3S_HV_POSSIBLE) + /* + * Must check for PPC_MSG_RM_HOST_ACTION messages + * before PPC_MSG_CALL_FUNCTION messages because when + * a VM is destroyed, we call kick_all_cpus_sync() + * to ensure that any pending PPC_MSG_RM_HOST_ACTION + * messages have completed before we free any VCPUs. + */ + if (all & IPI_MESSAGE(PPC_MSG_RM_HOST_ACTION)) + kvmppc_xics_ipi_action(); +#endif + if (all & IPI_MESSAGE(PPC_MSG_CALL_FUNCTION)) generic_smp_call_function_interrupt(); - if (all & (1 << (24 - 8 * PPC_MSG_RESCHEDULE))) + if (all & IPI_MESSAGE(PPC_MSG_RESCHEDULE)) scheduler_ipi(); - if (all & (1 << (24 - 8 * PPC_MSG_CALL_FUNC_SINGLE))) - generic_smp_call_function_single_interrupt(); - if (all & (1 << (24 - 8 * PPC_MSG_DEBUGGER_BREAK))) - debug_ipi_action(0, NULL); -#else -#error Unsupported ENDIAN +#ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST + if (all & IPI_MESSAGE(PPC_MSG_TICK_BROADCAST)) + timer_broadcast_interrupt(); #endif - } while (info->messages); +#ifdef CONFIG_NMI_IPI + if (all & IPI_MESSAGE(PPC_MSG_NMI_IPI)) + nmi_ipi_action(0, NULL); +#endif + } while (READ_ONCE(info->messages)); return IRQ_HANDLED; } @@ -248,16 +359,16 @@ static inline void do_message_pass(int cpu, int msg) #endif } -void smp_send_reschedule(int cpu) +void arch_smp_send_reschedule(int cpu) { if (likely(smp_ops)) do_message_pass(cpu, PPC_MSG_RESCHEDULE); } -EXPORT_SYMBOL_GPL(smp_send_reschedule); +EXPORT_SYMBOL_GPL(arch_smp_send_reschedule); void arch_send_call_function_single_ipi(int cpu) { - do_message_pass(cpu, PPC_MSG_CALL_FUNC_SINGLE); + do_message_pass(cpu, PPC_MSG_CALL_FUNCTION); } void arch_send_call_function_ipi_mask(const struct cpumask *mask) @@ -268,66 +379,743 @@ void arch_send_call_function_ipi_mask(const struct cpumask *mask) do_message_pass(cpu, PPC_MSG_CALL_FUNCTION); } -#if defined(CONFIG_DEBUGGER) || defined(CONFIG_KEXEC) -void smp_send_debugger_break(void) +#ifdef CONFIG_NMI_IPI + +/* + * "NMI IPI" system. + * + * NMI IPIs may not be recoverable, so should not be used as ongoing part of + * a running system. They can be used for crash, debug, halt/reboot, etc. + * + * The IPI call waits with interrupts disabled until all targets enter the + * NMI handler, then returns. Subsequent IPIs can be issued before targets + * have returned from their handlers, so there is no guarantee about + * concurrency or re-entrancy. + * + * A new NMI can be issued before all targets exit the handler. + * + * The IPI call may time out without all targets entering the NMI handler. + * In that case, there is some logic to recover (and ignore subsequent + * NMI interrupts that may eventually be raised), but the platform interrupt + * handler may not be able to distinguish this from other exception causes, + * which may cause a crash. + */ + +static atomic_t __nmi_ipi_lock = ATOMIC_INIT(0); +static struct cpumask nmi_ipi_pending_mask; +static bool nmi_ipi_busy = false; +static void (*nmi_ipi_function)(struct pt_regs *) = NULL; + +noinstr static void nmi_ipi_lock_start(unsigned long *flags) { - int cpu; + raw_local_irq_save(*flags); + hard_irq_disable(); + while (raw_atomic_cmpxchg(&__nmi_ipi_lock, 0, 1) == 1) { + raw_local_irq_restore(*flags); + spin_until_cond(raw_atomic_read(&__nmi_ipi_lock) == 0); + raw_local_irq_save(*flags); + hard_irq_disable(); + } +} + +noinstr static void nmi_ipi_lock(void) +{ + while (raw_atomic_cmpxchg(&__nmi_ipi_lock, 0, 1) == 1) + spin_until_cond(raw_atomic_read(&__nmi_ipi_lock) == 0); +} + +noinstr static void nmi_ipi_unlock(void) +{ + smp_mb(); + WARN_ON(raw_atomic_read(&__nmi_ipi_lock) != 1); + raw_atomic_set(&__nmi_ipi_lock, 0); +} + +noinstr static void nmi_ipi_unlock_end(unsigned long *flags) +{ + nmi_ipi_unlock(); + raw_local_irq_restore(*flags); +} + +/* + * Platform NMI handler calls this to ack + */ +noinstr int smp_handle_nmi_ipi(struct pt_regs *regs) +{ + void (*fn)(struct pt_regs *) = NULL; + unsigned long flags; int me = raw_smp_processor_id(); + int ret = 0; - if (unlikely(!smp_ops)) + /* + * Unexpected NMIs are possible here because the interrupt may not + * be able to distinguish NMI IPIs from other types of NMIs, or + * because the caller may have timed out. + */ + nmi_ipi_lock_start(&flags); + if (cpumask_test_cpu(me, &nmi_ipi_pending_mask)) { + cpumask_clear_cpu(me, &nmi_ipi_pending_mask); + fn = READ_ONCE(nmi_ipi_function); + WARN_ON_ONCE(!fn); + ret = 1; + } + nmi_ipi_unlock_end(&flags); + + if (fn) + fn(regs); + + return ret; +} + +static void do_smp_send_nmi_ipi(int cpu, bool safe) +{ + if (!safe && smp_ops->cause_nmi_ipi && smp_ops->cause_nmi_ipi(cpu)) return; - for_each_online_cpu(cpu) - if (cpu != me) - do_message_pass(cpu, PPC_MSG_DEBUGGER_BREAK); + if (cpu >= 0) { + do_message_pass(cpu, PPC_MSG_NMI_IPI); + } else { + int c; + + for_each_online_cpu(c) { + if (c == raw_smp_processor_id()) + continue; + do_message_pass(c, PPC_MSG_NMI_IPI); + } + } +} + +/* + * - cpu is the target CPU (must not be this CPU), or NMI_IPI_ALL_OTHERS. + * - fn is the target callback function. + * - delay_us > 0 is the delay before giving up waiting for targets to + * begin executing the handler, == 0 specifies indefinite delay. + */ +static int __smp_send_nmi_ipi(int cpu, void (*fn)(struct pt_regs *), + u64 delay_us, bool safe) +{ + unsigned long flags; + int me = raw_smp_processor_id(); + int ret = 1; + + BUG_ON(cpu == me); + BUG_ON(cpu < 0 && cpu != NMI_IPI_ALL_OTHERS); + + if (unlikely(!smp_ops)) + return 0; + + nmi_ipi_lock_start(&flags); + while (nmi_ipi_busy) { + nmi_ipi_unlock_end(&flags); + spin_until_cond(!nmi_ipi_busy); + nmi_ipi_lock_start(&flags); + } + nmi_ipi_busy = true; + nmi_ipi_function = fn; + + WARN_ON_ONCE(!cpumask_empty(&nmi_ipi_pending_mask)); + + if (cpu < 0) { + /* ALL_OTHERS */ + cpumask_copy(&nmi_ipi_pending_mask, cpu_online_mask); + cpumask_clear_cpu(me, &nmi_ipi_pending_mask); + } else { + cpumask_set_cpu(cpu, &nmi_ipi_pending_mask); + } + + nmi_ipi_unlock(); + + /* Interrupts remain hard disabled */ + + do_smp_send_nmi_ipi(cpu, safe); + + nmi_ipi_lock(); + /* nmi_ipi_busy is set here, so unlock/lock is okay */ + while (!cpumask_empty(&nmi_ipi_pending_mask)) { + nmi_ipi_unlock(); + udelay(1); + nmi_ipi_lock(); + if (delay_us) { + delay_us--; + if (!delay_us) + break; + } + } + + if (!cpumask_empty(&nmi_ipi_pending_mask)) { + /* Timeout waiting for CPUs to call smp_handle_nmi_ipi */ + ret = 0; + cpumask_clear(&nmi_ipi_pending_mask); + } + + nmi_ipi_function = NULL; + nmi_ipi_busy = false; + + nmi_ipi_unlock_end(&flags); + + return ret; +} + +int smp_send_nmi_ipi(int cpu, void (*fn)(struct pt_regs *), u64 delay_us) +{ + return __smp_send_nmi_ipi(cpu, fn, delay_us, false); +} + +int smp_send_safe_nmi_ipi(int cpu, void (*fn)(struct pt_regs *), u64 delay_us) +{ + return __smp_send_nmi_ipi(cpu, fn, delay_us, true); +} +#endif /* CONFIG_NMI_IPI */ + +#ifdef CONFIG_GENERIC_CLOCKEVENTS_BROADCAST +void tick_broadcast(const struct cpumask *mask) +{ + unsigned int cpu; + + for_each_cpu(cpu, mask) + do_message_pass(cpu, PPC_MSG_TICK_BROADCAST); +} +#endif + +#ifdef CONFIG_DEBUGGER +static void debugger_ipi_callback(struct pt_regs *regs) +{ + debugger_ipi(regs); +} + +void smp_send_debugger_break(void) +{ + smp_send_nmi_ipi(NMI_IPI_ALL_OTHERS, debugger_ipi_callback, 1000000); } #endif -#ifdef CONFIG_KEXEC +#ifdef CONFIG_CRASH_DUMP void crash_send_ipi(void (*crash_ipi_callback)(struct pt_regs *)) { - crash_ipi_function_ptr = crash_ipi_callback; - if (crash_ipi_callback) { - mb(); - smp_send_debugger_break(); + int cpu; + + smp_send_nmi_ipi(NMI_IPI_ALL_OTHERS, crash_ipi_callback, 1000000); + if (kdump_in_progress() && crash_wake_offline) { + for_each_present_cpu(cpu) { + if (cpu_online(cpu)) + continue; + /* + * crash_ipi_callback will wait for + * all cpus, including offline CPUs. + * We don't care about nmi_ipi_function. + * Offline cpus will jump straight into + * crash_ipi_callback, we can skip the + * entire NMI dance and waiting for + * cpus to clear pending mask, etc. + */ + do_smp_send_nmi_ipi(cpu, false); + } } } #endif +void crash_smp_send_stop(void) +{ + static bool stopped = false; + + /* + * In case of fadump, register data for all CPUs is captured by f/w + * on ibm,os-term rtas call. Skip IPI callbacks to other CPUs before + * this rtas call to avoid tricky post processing of those CPUs' + * backtraces. + */ + if (should_fadump_crash()) + return; + + if (stopped) + return; + + stopped = true; + +#ifdef CONFIG_CRASH_DUMP + if (kexec_crash_image) { + crash_kexec_prepare(); + return; + } +#endif + + smp_send_stop(); +} + +#ifdef CONFIG_NMI_IPI +static void nmi_stop_this_cpu(struct pt_regs *regs) +{ + /* + * IRQs are already hard disabled by the smp_handle_nmi_ipi. + */ + set_cpu_online(smp_processor_id(), false); + + spin_begin(); + while (1) + spin_cpu_relax(); +} + +void smp_send_stop(void) +{ + smp_send_nmi_ipi(NMI_IPI_ALL_OTHERS, nmi_stop_this_cpu, 1000000); +} + +#else /* CONFIG_NMI_IPI */ + static void stop_this_cpu(void *dummy) { - /* Remove this CPU */ + hard_irq_disable(); + + /* + * Offlining CPUs in stop_this_cpu can result in scheduler warnings, + * (see commit de6e5d38417e), but printk_safe_flush_on_panic() wants + * to know other CPUs are offline before it breaks locks to flush + * printk buffers, in case we panic()ed while holding the lock. + */ set_cpu_online(smp_processor_id(), false); - local_irq_disable(); + spin_begin(); while (1) - ; + spin_cpu_relax(); } void smp_send_stop(void) { + static bool stopped = false; + + /* + * Prevent waiting on csd lock from a previous smp_send_stop. + * This is racy, but in general callers try to do the right + * thing and only fire off one smp_send_stop (e.g., see + * kernel/panic.c) + */ + if (stopped) + return; + + stopped = true; + smp_call_function(stop_this_cpu, NULL, 0); } +#endif /* CONFIG_NMI_IPI */ -struct thread_info *current_set[NR_CPUS]; +static struct task_struct *current_set[NR_CPUS]; static void smp_store_cpu_info(int id) { per_cpu(cpu_pvr, id) = mfspr(SPRN_PVR); -#ifdef CONFIG_PPC_FSL_BOOK3E +#ifdef CONFIG_PPC_E500 per_cpu(next_tlbcam_idx, id) = (mfspr(SPRN_TLB1CFG) & TLBnCFG_N_ENTRY) - 1; #endif } +/* + * Relationships between CPUs are maintained in a set of per-cpu cpumasks so + * rather than just passing around the cpumask we pass around a function that + * returns the that cpumask for the given CPU. + */ +static void set_cpus_related(int i, int j, struct cpumask *(*get_cpumask)(int)) +{ + cpumask_set_cpu(i, get_cpumask(j)); + cpumask_set_cpu(j, get_cpumask(i)); +} + +#ifdef CONFIG_HOTPLUG_CPU +static void set_cpus_unrelated(int i, int j, + struct cpumask *(*get_cpumask)(int)) +{ + cpumask_clear_cpu(i, get_cpumask(j)); + cpumask_clear_cpu(j, get_cpumask(i)); +} +#endif + +/* + * Extends set_cpus_related. Instead of setting one CPU at a time in + * dstmask, set srcmask at oneshot. dstmask should be super set of srcmask. + */ +static void or_cpumasks_related(int i, int j, struct cpumask *(*srcmask)(int), + struct cpumask *(*dstmask)(int)) +{ + struct cpumask *mask; + int k; + + mask = srcmask(j); + for_each_cpu(k, srcmask(i)) + cpumask_or(dstmask(k), dstmask(k), mask); + + if (i == j) + return; + + mask = srcmask(i); + for_each_cpu(k, srcmask(j)) + cpumask_or(dstmask(k), dstmask(k), mask); +} + +/* + * parse_thread_groups: Parses the "ibm,thread-groups" device tree + * property for the CPU device node @dn and stores + * the parsed output in the thread_groups_list + * structure @tglp. + * + * @dn: The device node of the CPU device. + * @tglp: Pointer to a thread group list structure into which the parsed + * output of "ibm,thread-groups" is stored. + * + * ibm,thread-groups[0..N-1] array defines which group of threads in + * the CPU-device node can be grouped together based on the property. + * + * This array can represent thread groupings for multiple properties. + * + * ibm,thread-groups[i + 0] tells us the property based on which the + * threads are being grouped together. If this value is 1, it implies + * that the threads in the same group share L1, translation cache. If + * the value is 2, it implies that the threads in the same group share + * the same L2 cache. + * + * ibm,thread-groups[i+1] tells us how many such thread groups exist for the + * property ibm,thread-groups[i] + * + * ibm,thread-groups[i+2] tells us the number of threads in each such + * group. + * Suppose k = (ibm,thread-groups[i+1] * ibm,thread-groups[i+2]), then, + * + * ibm,thread-groups[i+3..i+k+2] (is the list of threads identified by + * "ibm,ppc-interrupt-server#s" arranged as per their membership in + * the grouping. + * + * Example: + * If "ibm,thread-groups" = [1,2,4,8,10,12,14,9,11,13,15,2,2,4,8,10,12,14,9,11,13,15] + * This can be decomposed up into two consecutive arrays: + * a) [1,2,4,8,10,12,14,9,11,13,15] + * b) [2,2,4,8,10,12,14,9,11,13,15] + * + * where in, + * + * a) provides information of Property "1" being shared by "2" groups, + * each with "4" threads each. The "ibm,ppc-interrupt-server#s" of + * the first group is {8,10,12,14} and the + * "ibm,ppc-interrupt-server#s" of the second group is + * {9,11,13,15}. Property "1" is indicative of the thread in the + * group sharing L1 cache, translation cache and Instruction Data + * flow. + * + * b) provides information of Property "2" being shared by "2" groups, + * each group with "4" threads. The "ibm,ppc-interrupt-server#s" of + * the first group is {8,10,12,14} and the + * "ibm,ppc-interrupt-server#s" of the second group is + * {9,11,13,15}. Property "2" indicates that the threads in each + * group share the L2-cache. + * + * Returns 0 on success, -EINVAL if the property does not exist, + * -ENODATA if property does not have a value, and -EOVERFLOW if the + * property data isn't large enough. + */ +static int parse_thread_groups(struct device_node *dn, + struct thread_groups_list *tglp) +{ + unsigned int property_idx = 0; + u32 *thread_group_array; + size_t total_threads; + int ret = 0, count; + u32 *thread_list; + int i = 0; + + count = of_property_count_u32_elems(dn, "ibm,thread-groups"); + thread_group_array = kcalloc(count, sizeof(u32), GFP_KERNEL); + ret = of_property_read_u32_array(dn, "ibm,thread-groups", + thread_group_array, count); + if (ret) + goto out_free; + + while (i < count && property_idx < MAX_THREAD_GROUP_PROPERTIES) { + int j; + struct thread_groups *tg = &tglp->property_tgs[property_idx++]; + + tg->property = thread_group_array[i]; + tg->nr_groups = thread_group_array[i + 1]; + tg->threads_per_group = thread_group_array[i + 2]; + total_threads = tg->nr_groups * tg->threads_per_group; + + thread_list = &thread_group_array[i + 3]; + + for (j = 0; j < total_threads; j++) + tg->thread_list[j] = thread_list[j]; + i = i + 3 + total_threads; + } + + tglp->nr_properties = property_idx; + +out_free: + kfree(thread_group_array); + return ret; +} + +/* + * get_cpu_thread_group_start : Searches the thread group in tg->thread_list + * that @cpu belongs to. + * + * @cpu : The logical CPU whose thread group is being searched. + * @tg : The thread-group structure of the CPU node which @cpu belongs + * to. + * + * Returns the index to tg->thread_list that points to the start + * of the thread_group that @cpu belongs to. + * + * Returns -1 if cpu doesn't belong to any of the groups pointed to by + * tg->thread_list. + */ +static int get_cpu_thread_group_start(int cpu, struct thread_groups *tg) +{ + int hw_cpu_id = get_hard_smp_processor_id(cpu); + int i, j; + + for (i = 0; i < tg->nr_groups; i++) { + int group_start = i * tg->threads_per_group; + + for (j = 0; j < tg->threads_per_group; j++) { + int idx = group_start + j; + + if (tg->thread_list[idx] == hw_cpu_id) + return group_start; + } + } + + return -1; +} + +static struct thread_groups *__init get_thread_groups(int cpu, + int group_property, + int *err) +{ + struct device_node *dn = of_get_cpu_node(cpu, NULL); + struct thread_groups_list *cpu_tgl = &tgl[cpu]; + struct thread_groups *tg = NULL; + int i; + *err = 0; + + if (!dn) { + *err = -ENODATA; + return NULL; + } + + if (!cpu_tgl->nr_properties) { + *err = parse_thread_groups(dn, cpu_tgl); + if (*err) + goto out; + } + + for (i = 0; i < cpu_tgl->nr_properties; i++) { + if (cpu_tgl->property_tgs[i].property == group_property) { + tg = &cpu_tgl->property_tgs[i]; + break; + } + } + + if (!tg) + *err = -EINVAL; +out: + of_node_put(dn); + return tg; +} + +static int __init update_mask_from_threadgroup(cpumask_var_t *mask, struct thread_groups *tg, + int cpu, int cpu_group_start) +{ + int first_thread = cpu_first_thread_sibling(cpu); + int i; + + zalloc_cpumask_var_node(mask, GFP_KERNEL, cpu_to_node(cpu)); + + for (i = first_thread; i < first_thread + threads_per_core; i++) { + int i_group_start = get_cpu_thread_group_start(i, tg); + + if (unlikely(i_group_start == -1)) { + WARN_ON_ONCE(1); + return -ENODATA; + } + + if (i_group_start == cpu_group_start) + cpumask_set_cpu(i, *mask); + } + + return 0; +} + +static int __init init_thread_group_cache_map(int cpu, int cache_property) + +{ + int cpu_group_start = -1, err = 0; + struct thread_groups *tg = NULL; + cpumask_var_t *mask = NULL; + + if (cache_property != THREAD_GROUP_SHARE_L1 && + cache_property != THREAD_GROUP_SHARE_L2_L3) + return -EINVAL; + + tg = get_thread_groups(cpu, cache_property, &err); + + if (!tg) + return err; + + cpu_group_start = get_cpu_thread_group_start(cpu, tg); + + if (unlikely(cpu_group_start == -1)) { + WARN_ON_ONCE(1); + return -ENODATA; + } + + if (cache_property == THREAD_GROUP_SHARE_L1) { + mask = &per_cpu(thread_group_l1_cache_map, cpu); + update_mask_from_threadgroup(mask, tg, cpu, cpu_group_start); + } + else if (cache_property == THREAD_GROUP_SHARE_L2_L3) { + mask = &per_cpu(thread_group_l2_cache_map, cpu); + update_mask_from_threadgroup(mask, tg, cpu, cpu_group_start); + mask = &per_cpu(thread_group_l3_cache_map, cpu); + update_mask_from_threadgroup(mask, tg, cpu, cpu_group_start); + } + + + return 0; +} + +static bool shared_caches __ro_after_init; + +#ifdef CONFIG_SCHED_SMT +/* cpumask of CPUs with asymmetric SMT dependency */ +static int powerpc_smt_flags(void) +{ + int flags = SD_SHARE_CPUCAPACITY | SD_SHARE_LLC; + + if (cpu_has_feature(CPU_FTR_ASYM_SMT)) { + printk_once(KERN_INFO "Enabling Asymmetric SMT scheduling\n"); + flags |= SD_ASYM_PACKING; + } + return flags; +} +#endif + +/* + * On shared processor LPARs scheduled on a big core (which has two or more + * independent thread groups per core), prefer lower numbered CPUs, so + * that workload consolidates to lesser number of cores. + */ +static __ro_after_init DEFINE_STATIC_KEY_FALSE(splpar_asym_pack); + +/* + * P9 has a slightly odd architecture where pairs of cores share an L2 cache. + * This topology makes it *much* cheaper to migrate tasks between adjacent cores + * since the migrated task remains cache hot. We want to take advantage of this + * at the scheduler level so an extra topology level is required. + */ +static int powerpc_shared_cache_flags(void) +{ + if (static_branch_unlikely(&splpar_asym_pack)) + return SD_SHARE_LLC | SD_ASYM_PACKING; + + return SD_SHARE_LLC; +} + +static int powerpc_shared_proc_flags(void) +{ + if (static_branch_unlikely(&splpar_asym_pack)) + return SD_ASYM_PACKING; + + return 0; +} + +/* + * We can't just pass cpu_l2_cache_mask() directly because + * returns a non-const pointer and the compiler barfs on that. + */ +static const struct cpumask *tl_cache_mask(struct sched_domain_topology_level *tl, int cpu) +{ + return per_cpu(cpu_l2_cache_map, cpu); +} + +#ifdef CONFIG_SCHED_SMT +static const struct cpumask *tl_smallcore_smt_mask(struct sched_domain_topology_level *tl, int cpu) +{ + return cpu_smallcore_mask(cpu); +} +#endif + +struct cpumask *cpu_coregroup_mask(int cpu) +{ + return per_cpu(cpu_coregroup_map, cpu); +} + +static bool has_coregroup_support(void) +{ + /* Coregroup identification not available on shared systems */ + if (is_shared_processor()) + return 0; + + return coregroup_enabled; +} + +static int __init init_big_cores(void) +{ + int cpu; + + for_each_possible_cpu(cpu) { + int err = init_thread_group_cache_map(cpu, THREAD_GROUP_SHARE_L1); + + if (err) + return err; + + zalloc_cpumask_var_node(&per_cpu(cpu_smallcore_map, cpu), + GFP_KERNEL, + cpu_to_node(cpu)); + } + + has_big_cores = true; + + for_each_possible_cpu(cpu) { + int err = init_thread_group_cache_map(cpu, THREAD_GROUP_SHARE_L2_L3); + + if (err) + return err; + } + + thread_group_shares_l2 = true; + thread_group_shares_l3 = true; + pr_debug("L2/L3 cache only shared by the threads in the small core\n"); + + return 0; +} + +/* + * die_mask and die_id are only available on systems which support + * multiple coregroups within a same package. On all other systems, die_mask + * would be same as package mask and die_id would be set to -1. + */ +const struct cpumask *cpu_die_mask(int cpu) +{ + if (has_coregroup_support()) + return per_cpu(cpu_coregroup_map, cpu); + else + return cpu_node_mask(cpu); +} +EXPORT_SYMBOL_GPL(cpu_die_mask); + +int cpu_die_id(int cpu) +{ + if (has_coregroup_support()) + return cpu_to_coregroup_id(cpu); + else + return -1; +} +EXPORT_SYMBOL_GPL(cpu_die_id); + void __init smp_prepare_cpus(unsigned int max_cpus) { - unsigned int cpu; + unsigned int cpu, num_threads; DBG("smp_prepare_cpus\n"); /* - * setup_cpu may need to be called on the boot cpu. We havent + * setup_cpu may need to be called on the boot cpu. We haven't * spun any cpus up but lets be paranoid. */ BUG_ON(boot_cpuid != smp_processor_id()); @@ -339,29 +1127,72 @@ void __init smp_prepare_cpus(unsigned int max_cpus) for_each_possible_cpu(cpu) { zalloc_cpumask_var_node(&per_cpu(cpu_sibling_map, cpu), GFP_KERNEL, cpu_to_node(cpu)); + zalloc_cpumask_var_node(&per_cpu(cpu_l2_cache_map, cpu), + GFP_KERNEL, cpu_to_node(cpu)); zalloc_cpumask_var_node(&per_cpu(cpu_core_map, cpu), GFP_KERNEL, cpu_to_node(cpu)); + if (has_coregroup_support()) + zalloc_cpumask_var_node(&per_cpu(cpu_coregroup_map, cpu), + GFP_KERNEL, cpu_to_node(cpu)); + +#ifdef CONFIG_NUMA + /* + * numa_node_id() works after this. + */ + if (cpu_present(cpu)) { + set_cpu_numa_node(cpu, numa_cpu_lookup_table[cpu]); + set_cpu_numa_mem(cpu, + local_memory_node(numa_cpu_lookup_table[cpu])); + } +#endif } + /* Init the cpumasks so the boot CPU is related to itself */ cpumask_set_cpu(boot_cpuid, cpu_sibling_mask(boot_cpuid)); + cpumask_set_cpu(boot_cpuid, cpu_l2_cache_mask(boot_cpuid)); cpumask_set_cpu(boot_cpuid, cpu_core_mask(boot_cpuid)); - if (smp_ops) - if (smp_ops->probe) - max_cpus = smp_ops->probe(); - else - max_cpus = NR_CPUS; - else - max_cpus = 1; + if (has_coregroup_support()) + cpumask_set_cpu(boot_cpuid, cpu_coregroup_mask(boot_cpuid)); + + init_big_cores(); + if (has_big_cores) { + cpumask_set_cpu(boot_cpuid, + cpu_smallcore_mask(boot_cpuid)); + } + + if (cpu_to_chip_id(boot_cpuid) != -1) { + int idx = DIV_ROUND_UP(num_possible_cpus(), threads_per_core); + + /* + * All threads of a core will all belong to the same core, + * chip_id_lookup_table will have one entry per core. + * Assumption: if boot_cpuid doesn't have a chip-id, then no + * other CPUs, will also not have chip-id. + */ + chip_id_lookup_table = kcalloc(idx, sizeof(int), GFP_KERNEL); + if (chip_id_lookup_table) + memset(chip_id_lookup_table, -1, sizeof(int) * idx); + } + + if (smp_ops && smp_ops->probe) + smp_ops->probe(); + + // Initalise the generic SMT topology support + num_threads = 1; + if (smt_enabled_at_boot) + num_threads = smt_enabled_at_boot; + cpu_smt_set_num_threads(num_threads, threads_per_core); } -void smp_prepare_boot_cpu(void) +void __init smp_prepare_boot_cpu(void) { BUG_ON(smp_processor_id() != boot_cpuid); #ifdef CONFIG_PPC64 - paca[boot_cpuid].__current = current; + paca_ptrs[boot_cpuid]->__current = current; #endif - current_set[boot_cpuid] = task_thread_info(current); + set_numa_node(numa_cpu_lookup_table[boot_cpuid]); + current_set[boot_cpuid] = current; } #ifdef CONFIG_HOTPLUG_CPU @@ -374,10 +1205,24 @@ int generic_cpu_disable(void) return -EBUSY; set_cpu_online(cpu, false); -#ifdef CONFIG_PPC64 - vdso_data->processorCount--; +#ifdef CONFIG_PPC64_PROC_SYSTEMCFG + systemcfg->processorCount--; #endif - migrate_irqs(); + /* Update affinity of all IRQs previously aimed at this CPU */ + irq_migrate_all_off_this_cpu(); + + /* + * Depending on the details of the interrupt controller, it's possible + * that one of the interrupts we just migrated away from this CPU is + * actually already pending on this CPU. If we leave it in that state + * the interrupt will never be EOI'ed, and will never fire again. So + * temporarily enable interrupts here, to allow any pending interrupt to + * be received (and EOI'ed), before we take this CPU offline. + */ + local_irq_enable(); + mdelay(1); + local_irq_disable(); + return 0; } @@ -387,27 +1232,13 @@ void generic_cpu_die(unsigned int cpu) for (i = 0; i < 100; i++) { smp_rmb(); - if (per_cpu(cpu_state, cpu) == CPU_DEAD) + if (is_cpu_dead(cpu)) return; msleep(100); } printk(KERN_ERR "CPU%d didn't die...\n", cpu); } -void generic_mach_cpu_die(void) -{ - unsigned int cpu; - - local_irq_disable(); - idle_task_exit(); - cpu = smp_processor_id(); - printk(KERN_DEBUG "CPU%d offline\n", cpu); - __get_cpu_var(cpu_state) = CPU_DEAD; - smp_wmb(); - while (__get_cpu_var(cpu_state) != CPU_UP_PREPARE) - cpu_relax(); -} - void generic_set_cpu_dead(unsigned int cpu) { per_cpu(cpu_state, cpu) = CPU_DEAD; @@ -428,38 +1259,14 @@ int generic_check_cpu_restart(unsigned int cpu) return per_cpu(cpu_state, cpu) == CPU_UP_PREPARE; } -static atomic_t secondary_inhibit_count; - -/* - * Don't allow secondary CPU threads to come online - */ -void inhibit_secondary_onlining(void) +int is_cpu_dead(unsigned int cpu) { - /* - * This makes secondary_inhibit_count stable during cpu - * online/offline operations. - */ - get_online_cpus(); - - atomic_inc(&secondary_inhibit_count); - put_online_cpus(); + return per_cpu(cpu_state, cpu) == CPU_DEAD; } -EXPORT_SYMBOL_GPL(inhibit_secondary_onlining); -/* - * Allow secondary CPU threads to come online again - */ -void uninhibit_secondary_onlining(void) +static bool secondaries_inhibited(void) { - get_online_cpus(); - atomic_dec(&secondary_inhibit_count); - put_online_cpus(); -} -EXPORT_SYMBOL_GPL(uninhibit_secondary_onlining); - -static int secondaries_inhibited(void) -{ - return atomic_read(&secondary_inhibit_count); + return kvm_hv_mode_active(); } #else /* HOTPLUG_CPU */ @@ -470,25 +1277,29 @@ static int secondaries_inhibited(void) static void cpu_idle_thread_init(unsigned int cpu, struct task_struct *idle) { - struct thread_info *ti = task_thread_info(idle); - #ifdef CONFIG_PPC64 - paca[cpu].__current = idle; - paca[cpu].kstack = (unsigned long)ti + THREAD_SIZE - STACK_FRAME_OVERHEAD; + paca_ptrs[cpu]->__current = idle; + paca_ptrs[cpu]->kstack = (unsigned long)task_stack_page(idle) + + THREAD_SIZE - STACK_FRAME_MIN_SIZE; #endif - ti->cpu = cpu; - secondary_ti = current_set[cpu] = ti; + task_thread_info(idle)->cpu = cpu; + secondary_current = current_set[cpu] = idle; } int __cpu_up(unsigned int cpu, struct task_struct *tidle) { - int rc, c; + const unsigned long boot_spin_ms = 5 * MSEC_PER_SEC; + const bool booting = system_state < SYSTEM_RUNNING; + const unsigned long hp_spin_ms = 1; + unsigned long deadline; + int rc; + const unsigned long spin_wait_ms = booting ? boot_spin_ms : hp_spin_ms; /* * Don't allow secondary threads to come online if inhibited */ if (threads_per_core > 1 && secondaries_inhibited() && - cpu % threads_per_core != 0) + cpu_thread_in_subcore(cpu)) return -EBUSY; if (smp_ops == NULL || @@ -497,6 +1308,16 @@ int __cpu_up(unsigned int cpu, struct task_struct *tidle) cpu_idle_thread_init(cpu, tidle); + /* + * The platform might need to allocate resources prior to bringing + * up the CPU + */ + if (smp_ops->prepare_cpu) { + rc = smp_ops->prepare_cpu(cpu); + if (rc) + return rc; + } + /* Make sure callin-map entry is 0 (can be leftover a CPU * hotplug */ @@ -517,22 +1338,23 @@ int __cpu_up(unsigned int cpu, struct task_struct *tidle) } /* - * wait to see if the cpu made a callin (is actually up). - * use this value that I found through experimentation. - * -- Cort + * At boot time, simply spin on the callin word until the + * deadline passes. + * + * At run time, spin for an optimistic amount of time to avoid + * sleeping in the common case. */ - if (system_state < SYSTEM_RUNNING) - for (c = 50000; c && !cpu_callin_map[cpu]; c--) - udelay(100); -#ifdef CONFIG_HOTPLUG_CPU - else - /* - * CPUs can take much longer to come up in the - * hotplug case. Wait five seconds. - */ - for (c = 5000; c && !cpu_callin_map[cpu]; c--) - msleep(1); -#endif + deadline = jiffies + msecs_to_jiffies(spin_wait_ms); + spin_until_cond(cpu_callin_map[cpu] || time_is_before_jiffies(deadline)); + + if (!cpu_callin_map[cpu] && system_state >= SYSTEM_RUNNING) { + const unsigned long sleep_interval_us = 10 * USEC_PER_MSEC; + const unsigned long sleep_wait_ms = 100 * MSEC_PER_SEC; + + deadline = jiffies + msecs_to_jiffies(sleep_wait_ms); + while (!cpu_callin_map[cpu] && time_is_after_jiffies(deadline)) + fsleep(sleep_interval_us); + } if (!cpu_callin_map[cpu]) { printk(KERN_ERR "Processor %u is stuck.\n", cpu); @@ -544,9 +1366,8 @@ int __cpu_up(unsigned int cpu, struct task_struct *tidle) if (smp_ops->give_timebase) smp_ops->give_timebase(); - /* Wait until cpu puts itself in the online map */ - while (!cpu_online(cpu)) - cpu_relax(); + /* Wait until cpu puts itself in the online & active maps */ + spin_until_cond(cpu_online(cpu)); return 0; } @@ -557,22 +1378,18 @@ int __cpu_up(unsigned int cpu, struct task_struct *tidle) int cpu_to_core_id(int cpu) { struct device_node *np; - const int *reg; int id = -1; np = of_get_cpu_node(cpu, NULL); if (!np) goto out; - reg = of_get_property(np, "reg", NULL); - if (!reg) - goto out; - - id = *reg; + id = of_get_cpu_hwid(np, 0); out: of_node_put(np); return id; } +EXPORT_SYMBOL_GPL(cpu_to_core_id); /* Helper routines for cpu to core mapping */ int cpu_core_index_of_thread(int cpu) @@ -609,19 +1426,232 @@ static struct device_node *cpu_to_l2cache(int cpu) return cache; } +static bool update_mask_by_l2(int cpu, cpumask_var_t *mask) +{ + struct cpumask *(*submask_fn)(int) = cpu_sibling_mask; + struct device_node *l2_cache, *np; + int i; + + if (has_big_cores) + submask_fn = cpu_smallcore_mask; + + /* + * If the threads in a thread-group share L2 cache, then the + * L2-mask can be obtained from thread_group_l2_cache_map. + */ + if (thread_group_shares_l2) { + cpumask_set_cpu(cpu, cpu_l2_cache_mask(cpu)); + + for_each_cpu(i, per_cpu(thread_group_l2_cache_map, cpu)) { + if (cpu_online(i)) + set_cpus_related(i, cpu, cpu_l2_cache_mask); + } + + /* Verify that L1-cache siblings are a subset of L2 cache-siblings */ + if (!cpumask_equal(submask_fn(cpu), cpu_l2_cache_mask(cpu)) && + !cpumask_subset(submask_fn(cpu), cpu_l2_cache_mask(cpu))) { + pr_warn_once("CPU %d : Inconsistent L1 and L2 cache siblings\n", + cpu); + } + + return true; + } + + l2_cache = cpu_to_l2cache(cpu); + if (!l2_cache || !*mask) { + /* Assume only core siblings share cache with this CPU */ + for_each_cpu(i, cpu_sibling_mask(cpu)) + set_cpus_related(cpu, i, cpu_l2_cache_mask); + + return false; + } + + cpumask_and(*mask, cpu_online_mask, cpu_node_mask(cpu)); + + /* Update l2-cache mask with all the CPUs that are part of submask */ + or_cpumasks_related(cpu, cpu, submask_fn, cpu_l2_cache_mask); + + /* Skip all CPUs already part of current CPU l2-cache mask */ + cpumask_andnot(*mask, *mask, cpu_l2_cache_mask(cpu)); + + for_each_cpu(i, *mask) { + /* + * when updating the marks the current CPU has not been marked + * online, but we need to update the cache masks + */ + np = cpu_to_l2cache(i); + + /* Skip all CPUs already part of current CPU l2-cache */ + if (np == l2_cache) { + or_cpumasks_related(cpu, i, submask_fn, cpu_l2_cache_mask); + cpumask_andnot(*mask, *mask, submask_fn(i)); + } else { + cpumask_andnot(*mask, *mask, cpu_l2_cache_mask(i)); + } + + of_node_put(np); + } + of_node_put(l2_cache); + + return true; +} + +#ifdef CONFIG_HOTPLUG_CPU +static void remove_cpu_from_masks(int cpu) +{ + struct cpumask *(*mask_fn)(int) = cpu_sibling_mask; + int i; + + unmap_cpu_from_node(cpu); + + if (shared_caches) + mask_fn = cpu_l2_cache_mask; + + for_each_cpu(i, mask_fn(cpu)) { + set_cpus_unrelated(cpu, i, cpu_l2_cache_mask); + set_cpus_unrelated(cpu, i, cpu_sibling_mask); + if (has_big_cores) + set_cpus_unrelated(cpu, i, cpu_smallcore_mask); + } + + for_each_cpu(i, cpu_core_mask(cpu)) + set_cpus_unrelated(cpu, i, cpu_core_mask); + + if (has_coregroup_support()) { + for_each_cpu(i, cpu_coregroup_mask(cpu)) + set_cpus_unrelated(cpu, i, cpu_coregroup_mask); + } +} +#endif + +static inline void add_cpu_to_smallcore_masks(int cpu) +{ + int i; + + if (!has_big_cores) + return; + + cpumask_set_cpu(cpu, cpu_smallcore_mask(cpu)); + + for_each_cpu(i, per_cpu(thread_group_l1_cache_map, cpu)) { + if (cpu_online(i)) + set_cpus_related(i, cpu, cpu_smallcore_mask); + } +} + +static void update_coregroup_mask(int cpu, cpumask_var_t *mask) +{ + struct cpumask *(*submask_fn)(int) = cpu_sibling_mask; + int coregroup_id = cpu_to_coregroup_id(cpu); + int i; + + if (shared_caches) + submask_fn = cpu_l2_cache_mask; + + if (!*mask) { + /* Assume only siblings are part of this CPU's coregroup */ + for_each_cpu(i, submask_fn(cpu)) + set_cpus_related(cpu, i, cpu_coregroup_mask); + + return; + } + + cpumask_and(*mask, cpu_online_mask, cpu_node_mask(cpu)); + + /* Update coregroup mask with all the CPUs that are part of submask */ + or_cpumasks_related(cpu, cpu, submask_fn, cpu_coregroup_mask); + + /* Skip all CPUs already part of coregroup mask */ + cpumask_andnot(*mask, *mask, cpu_coregroup_mask(cpu)); + + for_each_cpu(i, *mask) { + /* Skip all CPUs not part of this coregroup */ + if (coregroup_id == cpu_to_coregroup_id(i)) { + or_cpumasks_related(cpu, i, submask_fn, cpu_coregroup_mask); + cpumask_andnot(*mask, *mask, submask_fn(i)); + } else { + cpumask_andnot(*mask, *mask, cpu_coregroup_mask(i)); + } + } +} + +static void add_cpu_to_masks(int cpu) +{ + struct cpumask *(*submask_fn)(int) = cpu_sibling_mask; + int first_thread = cpu_first_thread_sibling(cpu); + cpumask_var_t mask; + int chip_id = -1; + bool ret; + int i; + + /* + * This CPU will not be in the online mask yet so we need to manually + * add it to its own thread sibling mask. + */ + map_cpu_to_node(cpu, cpu_to_node(cpu)); + cpumask_set_cpu(cpu, cpu_sibling_mask(cpu)); + cpumask_set_cpu(cpu, cpu_core_mask(cpu)); + + for (i = first_thread; i < first_thread + threads_per_core; i++) + if (cpu_online(i)) + set_cpus_related(i, cpu, cpu_sibling_mask); + + add_cpu_to_smallcore_masks(cpu); + + /* In CPU-hotplug path, hence use GFP_ATOMIC */ + ret = alloc_cpumask_var_node(&mask, GFP_ATOMIC, cpu_to_node(cpu)); + update_mask_by_l2(cpu, &mask); + + if (has_coregroup_support()) + update_coregroup_mask(cpu, &mask); + + if (chip_id_lookup_table && ret) + chip_id = cpu_to_chip_id(cpu); + + if (shared_caches) + submask_fn = cpu_l2_cache_mask; + + /* Update core_mask with all the CPUs that are part of submask */ + or_cpumasks_related(cpu, cpu, submask_fn, cpu_core_mask); + + /* Skip all CPUs already part of current CPU core mask */ + cpumask_andnot(mask, cpu_online_mask, cpu_core_mask(cpu)); + + /* If chip_id is -1; limit the cpu_core_mask to within PKG */ + if (chip_id == -1) + cpumask_and(mask, mask, cpu_node_mask(cpu)); + + for_each_cpu(i, mask) { + if (chip_id == cpu_to_chip_id(i)) { + or_cpumasks_related(cpu, i, submask_fn, cpu_core_mask); + cpumask_andnot(mask, mask, submask_fn(i)); + } else { + cpumask_andnot(mask, mask, cpu_core_mask(i)); + } + } + + free_cpumask_var(mask); +} + /* Activate a secondary processor. */ +__no_stack_protector void start_secondary(void *unused) { - unsigned int cpu = smp_processor_id(); - struct device_node *l2_cache; - int i, base; + unsigned int cpu = raw_smp_processor_id(); + + /* PPC64 calls setup_kup() in early_setup_secondary() */ + if (IS_ENABLED(CONFIG_PPC32)) + setup_kup(); - atomic_inc(&init_mm.mm_count); + mmgrab_lazy_tlb(&init_mm); current->active_mm = &init_mm; + VM_WARN_ON(cpumask_test_cpu(smp_processor_id(), mm_cpumask(&init_mm))); + cpumask_set_cpu(cpu, mm_cpumask(&init_mm)); + inc_mm_active_cpus(&init_mm); smp_store_cpu_info(cpu); set_dec(tb_ticks_per_jiffy); - preempt_disable(); + rcutree_report_cpu_starting(cpu); cpu_callin_map[cpu] = 1; if (smp_ops->setup_cpu) @@ -631,154 +1661,160 @@ void start_secondary(void *unused) secondary_cpu_time_init(); -#ifdef CONFIG_PPC64 +#ifdef CONFIG_PPC64_PROC_SYSTEMCFG if (system_state == SYSTEM_RUNNING) - vdso_data->processorCount++; + systemcfg->processorCount++; +#endif +#ifdef CONFIG_PPC64 vdso_getcpu_init(); #endif - /* Update sibling maps */ - base = cpu_first_thread_sibling(cpu); - for (i = 0; i < threads_per_core; i++) { - if (cpu_is_offline(base + i) && (cpu != base + i)) - continue; - cpumask_set_cpu(cpu, cpu_sibling_mask(base + i)); - cpumask_set_cpu(base + i, cpu_sibling_mask(cpu)); - - /* cpu_core_map should be a superset of - * cpu_sibling_map even if we don't have cache - * information, so update the former here, too. - */ - cpumask_set_cpu(cpu, cpu_core_mask(base + i)); - cpumask_set_cpu(base + i, cpu_core_mask(cpu)); - } - l2_cache = cpu_to_l2cache(cpu); - for_each_online_cpu(i) { - struct device_node *np = cpu_to_l2cache(i); - if (!np) - continue; - if (np == l2_cache) { - cpumask_set_cpu(cpu, cpu_core_mask(i)); - cpumask_set_cpu(i, cpu_core_mask(cpu)); - } - of_node_put(np); + set_numa_node(numa_cpu_lookup_table[cpu]); + set_numa_mem(local_memory_node(numa_cpu_lookup_table[cpu])); + + /* Update topology CPU masks */ + add_cpu_to_masks(cpu); + + /* + * Check for any shared caches. Note that this must be done on a + * per-core basis because one core in the pair might be disabled. + */ + if (!shared_caches) { + struct cpumask *(*sibling_mask)(int) = cpu_sibling_mask; + struct cpumask *mask = cpu_l2_cache_mask(cpu); + + if (has_big_cores) + sibling_mask = cpu_smallcore_mask; + + if (cpumask_weight(mask) > cpumask_weight(sibling_mask(cpu))) + shared_caches = true; } - of_node_put(l2_cache); smp_wmb(); notify_cpu_starting(cpu); set_cpu_online(cpu, true); + boot_init_stack_canary(); + local_irq_enable(); - cpu_startup_entry(CPUHP_ONLINE); + /* We can enable ftrace for secondary cpus now */ + this_cpu_enable_ftrace(); + + cpu_startup_entry(CPUHP_AP_ONLINE_IDLE); BUG(); } -int setup_profiling_timer(unsigned int multiplier) +static struct sched_domain_topology_level powerpc_topology[6]; + +static void __init build_sched_topology(void) { - return 0; + int i = 0; + + if (is_shared_processor() && has_big_cores) + static_branch_enable(&splpar_asym_pack); + +#ifdef CONFIG_SCHED_SMT + if (has_big_cores) { + pr_info("Big cores detected but using small core scheduling\n"); + powerpc_topology[i++] = + SDTL_INIT(tl_smallcore_smt_mask, powerpc_smt_flags, SMT); + } else { + powerpc_topology[i++] = SDTL_INIT(tl_smt_mask, powerpc_smt_flags, SMT); + } +#endif + if (shared_caches) { + powerpc_topology[i++] = + SDTL_INIT(tl_cache_mask, powerpc_shared_cache_flags, CACHE); + } + + if (has_coregroup_support()) { + powerpc_topology[i++] = + SDTL_INIT(tl_mc_mask, powerpc_shared_proc_flags, MC); + } + + powerpc_topology[i++] = SDTL_INIT(tl_pkg_mask, powerpc_shared_proc_flags, PKG); + + /* There must be one trailing NULL entry left. */ + BUG_ON(i >= ARRAY_SIZE(powerpc_topology) - 1); + + set_sched_topology(powerpc_topology); } void __init smp_cpus_done(unsigned int max_cpus) { - cpumask_var_t old_mask; - - /* We want the setup_cpu() here to be called from CPU 0, but our - * init thread may have been "borrowed" by another CPU in the meantime - * se we pin us down to CPU 0 for a short while + /* + * We are running pinned to the boot CPU, see rest_init(). */ - alloc_cpumask_var(&old_mask, GFP_NOWAIT); - cpumask_copy(old_mask, tsk_cpus_allowed(current)); - set_cpus_allowed_ptr(current, cpumask_of(boot_cpuid)); - if (smp_ops && smp_ops->setup_cpu) smp_ops->setup_cpu(boot_cpuid); - set_cpus_allowed_ptr(current, old_mask); - - free_cpumask_var(old_mask); - if (smp_ops && smp_ops->bringup_done) smp_ops->bringup_done(); dump_numa_cpu_topology(); - + build_sched_topology(); } -int arch_sd_sibling_asym_packing(void) +/* + * For asym packing, by default lower numbered CPU has higher priority. + * On shared processors, pack to lower numbered core. However avoid moving + * between thread_groups within the same core. + */ +int arch_asym_cpu_priority(int cpu) { - if (cpu_has_feature(CPU_FTR_ASYM_SMT)) { - printk_once(KERN_INFO "Enabling Asymmetric SMT scheduling\n"); - return SD_ASYM_PACKING; - } - return 0; + if (static_branch_unlikely(&splpar_asym_pack)) + return -cpu / threads_per_core; + + return -cpu; } #ifdef CONFIG_HOTPLUG_CPU int __cpu_disable(void) { - struct device_node *l2_cache; int cpu = smp_processor_id(); - int base, i; int err; if (!smp_ops->cpu_disable) return -ENOSYS; + this_cpu_disable_ftrace(); + err = smp_ops->cpu_disable(); if (err) return err; /* Update sibling maps */ - base = cpu_first_thread_sibling(cpu); - for (i = 0; i < threads_per_core; i++) { - cpumask_clear_cpu(cpu, cpu_sibling_mask(base + i)); - cpumask_clear_cpu(base + i, cpu_sibling_mask(cpu)); - cpumask_clear_cpu(cpu, cpu_core_mask(base + i)); - cpumask_clear_cpu(base + i, cpu_core_mask(cpu)); - } - - l2_cache = cpu_to_l2cache(cpu); - for_each_present_cpu(i) { - struct device_node *np = cpu_to_l2cache(i); - if (!np) - continue; - if (np == l2_cache) { - cpumask_clear_cpu(cpu, cpu_core_mask(i)); - cpumask_clear_cpu(i, cpu_core_mask(cpu)); - } - of_node_put(np); - } - of_node_put(l2_cache); - + remove_cpu_from_masks(cpu); return 0; } void __cpu_die(unsigned int cpu) { + /* + * This could perhaps be a generic call in idlea_task_dead(), but + * that requires testing from all archs, so first put it here to + */ + VM_WARN_ON_ONCE(!cpumask_test_cpu(cpu, mm_cpumask(&init_mm))); + dec_mm_active_cpus(&init_mm); + cpumask_clear_cpu(cpu, mm_cpumask(&init_mm)); + if (smp_ops->cpu_die) smp_ops->cpu_die(cpu); } -static DEFINE_MUTEX(powerpc_cpu_hotplug_driver_mutex); - -void cpu_hotplug_driver_lock() -{ - mutex_lock(&powerpc_cpu_hotplug_driver_mutex); -} - -void cpu_hotplug_driver_unlock() +void __noreturn arch_cpu_idle_dead(void) { - mutex_unlock(&powerpc_cpu_hotplug_driver_mutex); -} + /* + * Disable on the down path. This will be re-enabled by + * start_secondary() via start_secondary_resume() below + */ + this_cpu_disable_ftrace(); -void cpu_die(void) -{ - if (ppc_md.cpu_die) - ppc_md.cpu_die(); + if (smp_ops->cpu_offline_self) + smp_ops->cpu_offline_self(); /* If we return, we re-enter start_secondary */ start_secondary_resume(); |