diff options
Diffstat (limited to 'target/linux/ubicom32/files/arch/ubicom32/kernel/smp.c')
| -rw-r--r-- | target/linux/ubicom32/files/arch/ubicom32/kernel/smp.c | 806 |
1 files changed, 806 insertions, 0 deletions
diff --git a/target/linux/ubicom32/files/arch/ubicom32/kernel/smp.c b/target/linux/ubicom32/files/arch/ubicom32/kernel/smp.c new file mode 100644 index 000000000..4aa27eb44 --- /dev/null +++ b/target/linux/ubicom32/files/arch/ubicom32/kernel/smp.c @@ -0,0 +1,806 @@ +/* + * arch/ubicom32/kernel/smp.c + * SMP implementation for Ubicom32 processors. + * + * (C) Copyright 2009, Ubicom, Inc. + * Copyright (C) 1999 Walt Drummond <drummond@valinux.com> + * Copyright (C) 1999 David Mosberger-Tang <davidm@hpl.hp.com> + * Copyright (C) 2001,2004 Grant Grundler <grundler@parisc-linux.org> + * + * This file is part of the Ubicom32 Linux Kernel Port. + * + * The Ubicom32 Linux Kernel Port 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. + * + * The Ubicom32 Linux Kernel Port is distributed in the hope that it + * will be useful, but WITHOUT ANY WARRANTY; without even the implied + * warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See + * the GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with the Ubicom32 Linux Kernel Port. If not, + * see <http://www.gnu.org/licenses/>. + * + * Ubicom32 implementation derived from (with many thanks): + * arch/m68knommu + * arch/blackfin + * arch/parisc + */ + +#include <linux/types.h> +#include <linux/spinlock.h> +#include <linux/slab.h> + +#include <linux/kernel.h> +#include <linux/bootmem.h> +#include <linux/module.h> +#include <linux/sched.h> +#include <linux/init.h> +#include <linux/interrupt.h> +#include <linux/smp.h> +#include <linux/kernel_stat.h> +#include <linux/mm.h> +#include <linux/err.h> +#include <linux/delay.h> +#include <linux/bitops.h> +#include <linux/cpu.h> +#include <linux/profile.h> +#include <linux/delay.h> +#include <linux/io.h> +#include <linux/ptrace.h> +#include <linux/unistd.h> +#include <linux/irq.h> + +#include <asm/system.h> +#include <asm/atomic.h> +#include <asm/current.h> +#include <asm/tlbflush.h> +#include <asm/timex.h> +#include <asm/cpu.h> +#include <asm/irq.h> +#include <asm/processor.h> +#include <asm/thread.h> +#include <asm/sections.h> +#include <asm/ip5000.h> + +/* + * Mask the debug printout for IPI because they are too verbose + * for regular debugging. + */ + +// #define DEBUG_SMP 1 +#if !defined(DEBUG_SMP) +#define smp_debug(lvl, ...) +#else +static unsigned int smp_debug_lvl = 50; +#define smp_debug(lvl, printargs...) \ + if (lvl >= smp_debug_lvl) { \ + printk(printargs); \ + } +#endif + +#if !defined(DEBUG_SMP) +#define DEBUG_ASSERT(cond) +#else +#define DEBUG_ASSERT(cond) \ + if (!(cond)) { \ + THREAD_STALL; \ + } +#endif + +/* + * List of IPI Commands (more than one can be set at a time). + */ +enum ipi_message_type { + IPI_NOP, + IPI_RESCHEDULE, + IPI_CALL_FUNC, + IPI_CALL_FUNC_SINGLE, + IPI_CPU_STOP, + IPI_CPU_TIMER, +}; + +/* + * We maintain a hardware thread oriented view of online threads + * and those involved or needing IPI. + */ +static volatile unsigned long smp_online_threads = 0; +static volatile unsigned long smp_needs_ipi = 0; +static volatile unsigned long smp_inside_ipi = 0; +static unsigned long smp_irq_affinity[NR_IRQS]; + +/* + * What do we need to track on a per cpu/thread basis? + */ +DEFINE_PER_CPU(struct cpuinfo_ubicom32, cpu_data); + +/* + * Each thread cpuinfo IPI information is guarded by a lock + * that is kept local to this file. + */ +DEFINE_PER_CPU(spinlock_t, ipi_lock) = SPIN_LOCK_UNLOCKED; + +/* + * The IPI(s) are based on a software IRQ through the LDSR. + */ +unsigned int smp_ipi_irq; + +/* + * Define a spinlock so that only one cpu is able to modify the + * smp_needs_ipi and to set/clear the IRQ at a time. + */ +DEFINE_SPINLOCK(smp_ipi_lock); + +/* + * smp_halt_processor() + * Halt this hardware thread. + */ +static void smp_halt_processor(void) +{ + int cpuid = thread_get_self(); + cpu_clear(smp_processor_id(), cpu_online_map); + local_irq_disable(); + printk(KERN_EMERG "cpu[%d] has halted. It is not OK to turn off power \ + until all cpu's are off.\n", cpuid); + for (;;) { + thread_suspend(); + } +} + +/* + * ipi_interrupt() + * Handle an Interprocessor Interrupt. + */ +static irqreturn_t ipi_interrupt(int irq, void *dev_id) +{ + int cpuid = smp_processor_id(); + struct cpuinfo_ubicom32 *p = &per_cpu(cpu_data, cpuid); + unsigned long ops; + + /* + * Count this now; we may make a call that never returns. + */ + p->ipi_count++; + + /* + * We are about to process all ops. If another cpu has stated + * that we need an IPI, we will have already processed it. By + * clearing our smp_needs_ipi, and processing all ops, + * we reduce the number of IPI interrupts. However, this introduces + * the possibility that smp_needs_ipi will be clear and the soft irq + * will have gone off; so we need to make the get_affinity() path + * tolerant of spurious interrupts. + */ + spin_lock(&smp_ipi_lock); + smp_needs_ipi &= ~(1 << p->tid); + spin_unlock(&smp_ipi_lock); + + for (;;) { + /* + * Read the set of IPI commands we should handle. + */ + spinlock_t *lock = &per_cpu(ipi_lock, cpuid); + spin_lock(lock); + ops = p->ipi_pending; + p->ipi_pending = 0; + spin_unlock(lock); + + /* + * If we have no IPI commands to execute, break out. + */ + if (!ops) { + break; + } + + /* + * Execute the set of commands in the ops word, one command + * at a time in no particular order. Strip of each command + * as we execute it. + */ + while (ops) { + unsigned long which = ffz(~ops); + ops &= ~(1 << which); + + BUG_ON(!irqs_disabled()); + switch (which) { + case IPI_NOP: + smp_debug(100, KERN_INFO "cpu[%d]: " + "IPI_NOP\n", cpuid); + break; + + case IPI_RESCHEDULE: + /* + * Reschedule callback. Everything to be + * done is done by the interrupt return path. + */ + smp_debug(200, KERN_INFO "cpu[%d]: " + "IPI_RESCHEDULE\n", cpuid); + break; + + case IPI_CALL_FUNC: + smp_debug(100, KERN_INFO "cpu[%d]: " + "IPI_CALL_FUNC\n", cpuid); + generic_smp_call_function_interrupt(); + break; + + case IPI_CALL_FUNC_SINGLE: + smp_debug(100, KERN_INFO "cpu[%d]: " + "IPI_CALL_FUNC_SINGLE\n", cpuid); + generic_smp_call_function_single_interrupt(); + break; + + case IPI_CPU_STOP: + smp_debug(100, KERN_INFO "cpu[%d]: " + "IPI_CPU_STOP\n", cpuid); + smp_halt_processor(); + break; + +#if !defined(CONFIG_LOCAL_TIMERS) + case IPI_CPU_TIMER: + smp_debug(100, KERN_INFO "cpu[%d]: " + "IPI_CPU_TIMER\n", cpuid); +#if defined(CONFIG_GENERIC_CLOCKEVENTS) + local_timer_interrupt(); +#else + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); +#endif +#endif + break; + + default: + printk(KERN_CRIT "cpu[%d]: " + "Unknown IPI: %lu\n", cpuid, which); + + return IRQ_NONE; + } + + /* + * Let in any pending interrupts + */ + BUG_ON(!irqs_disabled()); + local_irq_enable(); + local_irq_disable(); + } + } + return IRQ_HANDLED; +} + +/* + * ipi_send() + * Send an Interprocessor Interrupt. + */ +static void ipi_send(int cpu, enum ipi_message_type op) +{ + struct cpuinfo_ubicom32 *p = &per_cpu(cpu_data, cpu); + spinlock_t *lock = &per_cpu(ipi_lock, cpu); + unsigned long flags; + + /* + * We protect the setting of the ipi_pending field and ensure + * that the ipi delivery mechanism and interrupt are atomically + * handled. + */ + spin_lock_irqsave(lock, flags); + p->ipi_pending |= 1 << op; + spin_unlock_irqrestore(lock, flags); + + spin_lock_irqsave(&smp_ipi_lock, flags); + smp_needs_ipi |= (1 << p->tid); + ubicom32_set_interrupt(smp_ipi_irq); + spin_unlock_irqrestore(&smp_ipi_lock, flags); + smp_debug(100, KERN_INFO "cpu[%d]: send: %d\n", cpu, op); +} + +/* + * ipi_send_mask + * Send an IPI to each cpu in mask. + */ +static inline void ipi_send_mask(unsigned int op, const struct cpumask mask) +{ + int cpu; + for_each_cpu_mask(cpu, mask) { + ipi_send(cpu, op); + } +} + +/* + * ipi_send_allbutself() + * Send an IPI to all threads but ourselves. + */ +static inline void ipi_send_allbutself(unsigned int op) +{ + int self = smp_processor_id(); + struct cpumask result; + cpumask_copy(&result, &cpu_online_map); + cpu_clear(self, result); + ipi_send_mask(op, result); +} + +/* + * smp_enable_vector() + */ +static void smp_enable_vector(unsigned int irq) +{ + ubicom32_clear_interrupt(smp_ipi_irq); + ldsr_enable_vector(irq); +} + +/* + * smp_disable_vector() + * Disable the interrupt by clearing the appropriate bit in the + * LDSR Mask Register. + */ +static void smp_disable_vector(unsigned int irq) +{ + ldsr_disable_vector(irq); +} + +/* + * smp_mask_vector() + */ +static void smp_mask_vector(unsigned int irq) +{ + ldsr_mask_vector(irq); +} + +/* + * smp_unmask_vector() + */ +static void smp_unmask_vector(unsigned int irq) +{ + ldsr_unmask_vector(irq); +} + +/* + * smp_end_vector() + * Called once an interrupt is completed (reset the LDSR mask). + */ +static void smp_end_vector(unsigned int irq) +{ + struct cpuinfo_ubicom32 *p = &per_cpu(cpu_data, smp_processor_id()); + spin_lock(&smp_ipi_lock); + smp_inside_ipi &= ~(1 << p->tid); + if (smp_inside_ipi) { + spin_unlock(&smp_ipi_lock); + return; + } + spin_unlock(&smp_ipi_lock); + ldsr_unmask_vector(irq); + smp_debug(100, KERN_INFO "cpu[%d]: unamesk vector\n", smp_processor_id()); +} + +/* + * Special hanlder functions for SMP. + */ +static struct irq_chip ubicom32_smp_chip = { + .name = "UbicoIPI", + .startup = NULL, + .shutdown = NULL, + .enable = smp_enable_vector, + .disable = smp_disable_vector, + .ack = NULL, + .mask = smp_mask_vector, + .unmask = smp_unmask_vector, + .end = smp_end_vector, +}; + +/* + * smp_reset_ipi() + * None of these cpu(s) got their IPI, turn it back on. + * + * Note: This is called by the LDSR which is not a full + * Linux cpu. Thus you must use the raw form of locks + * because lock debugging will not work on the partial + * cpu nature of the LDSR. + */ +void smp_reset_ipi(unsigned long mask) +{ + __raw_spin_lock(&smp_ipi_lock.raw_lock); + smp_needs_ipi |= mask; + smp_inside_ipi &= ~mask; + ubicom32_set_interrupt(smp_ipi_irq); + __raw_spin_unlock(&smp_ipi_lock.raw_lock); + smp_debug(100, KERN_INFO "smp: reset IPIs for: 0x%x\n", mask); +} + +/* + * smp_get_affinity() + * Choose the thread affinity for this interrupt. + * + * Note: This is called by the LDSR which is not a full + * Linux cpu. Thus you must use the raw form of locks + * because lock debugging will not work on the partial + * cpu nature of the LDSR. + */ +unsigned long smp_get_affinity(unsigned int irq, int *all) +{ + unsigned long mask = 0; + + /* + * Most IRQ(s) are delivered in a round robin fashion. + */ + if (irq != smp_ipi_irq) { + unsigned long result = smp_irq_affinity[irq] & smp_online_threads; + DEBUG_ASSERT(result); + *all = 0; + return result; + } + + /* + * This is an IPI request. Return all cpu(s) scheduled for an IPI. + * We also track those cpu(s) that are going to be "receiving" IPI this + * round. When all CPU(s) have called smp_end_vector(), + * we will unmask the IPI interrupt. + */ + __raw_spin_lock(&smp_ipi_lock.raw_lock); + ubicom32_clear_interrupt(smp_ipi_irq); + if (smp_needs_ipi) { + mask = smp_needs_ipi; + smp_inside_ipi |= smp_needs_ipi; + smp_needs_ipi = 0; + } + __raw_spin_unlock(&smp_ipi_lock.raw_lock); + *all = 1; + return mask; +} + +/* + * smp_set_affinity() + * Set the affinity for this irq but store the value in tid(s). + */ +void smp_set_affinity(unsigned int irq, const struct cpumask *dest) +{ + int cpuid; + unsigned long *paffinity = &smp_irq_affinity[irq]; + + /* + * If none specified, all cpus are allowed. + */ + if (cpus_empty(*dest)) { + *paffinity = 0xffffffff; + return; + } + + /* + * Make sure to clear the old value before setting up the + * list. + */ + *paffinity = 0; + for_each_cpu_mask(cpuid, *dest) { + struct cpuinfo_ubicom32 *p = &per_cpu(cpu_data, cpuid); + *paffinity |= (1 << p->tid); + } +} + +/* + * smp_send_stop() + * Send a stop request to all CPU but this one. + */ +void smp_send_stop(void) +{ + ipi_send_allbutself(IPI_CPU_STOP); +} + +/* + * smp_send_timer_all() + * Send all cpu(s) but this one, a request to update times. + */ +void smp_send_timer_all(void) +{ + ipi_send_allbutself(IPI_CPU_TIMER); +} + +/* + * smp_timer_broadcast() + * Use an IPI to broadcast a timer message + */ +void smp_timer_broadcast(const struct cpumask *mask) +{ + ipi_send_mask(IPI_CPU_TIMER, *mask); +} + +/* + * smp_send_reschedule() + * Send a reschedule request to the specified cpu. + */ +void smp_send_reschedule(int cpu) +{ + ipi_send(cpu, IPI_RESCHEDULE); +} + +/* + * arch_send_call_function_ipi() + * Cause each cpu in the mask to call the generic function handler. + */ +void arch_send_call_function_ipi_mask(const struct cpumask *mask) +{ + int cpu; + for_each_cpu_mask(cpu, *mask) { + ipi_send(cpu, IPI_CALL_FUNC); + } +} + +/* + * arch_send_call_function_single_ipi() + * Cause the specified cpu to call the generic function handler. + */ +void arch_send_call_function_single_ipi(int cpu) +{ + ipi_send(cpu, IPI_CALL_FUNC_SINGLE); +} + +/* + * setup_profiling_timer() + * Dummy function created to keep Oprofile happy in the SMP case. + */ +int setup_profiling_timer(unsigned int multiplier) +{ + return 0; +} + +/* + * smp_mainline_start() + * Start a slave thread executing a mainline Linux context. + */ +static void __init smp_mainline_start(void *arg) +{ + int cpuid = smp_processor_id(); + struct cpuinfo_ubicom32 *p = &per_cpu(cpu_data, cpuid); + + BUG_ON(p->tid != thread_get_self()); + + /* + * Well, support 2.4 linux scheme as well. + */ + if (cpu_test_and_set(cpuid, cpu_online_map)) { + printk(KERN_CRIT "cpu[%d]: already initialized!\n", cpuid); + smp_halt_processor(); + return; + } + + /* + * Initialise the idle task for this CPU + */ + atomic_inc(&init_mm.mm_count); + current->active_mm = &init_mm; + if (current->mm) { + printk(KERN_CRIT "cpu[%d]: idle task already has memory " + "management\n", cpuid); + smp_halt_processor(); + return; + } + + /* + * TODO: X86 does this prior to calling notify, try to understand why? + */ + preempt_disable(); + +#if defined(CONFIG_GENERIC_CLOCKEVENTS) + /* + * Setup a local timer event so that this cpu will get timer interrupts + */ + if (local_timer_setup(cpuid) == -1) { + printk(KERN_CRIT "cpu[%d]: timer alloc failed\n", cpuid); + smp_halt_processor(); + return; + } +#endif + + /* + * Notify those interested that we are up and alive. This must + * be done before interrupts are enabled. It must also be completed + * before the bootstrap cpu returns from __cpu_up() (see comment + * above cpu_set() of the cpu_online_map). + */ + notify_cpu_starting(cpuid); + + /* + * Indicate that this thread is now online and present. Setting + * cpu_online_map has the side effect of allowing the bootstrap + * cpu to continue along; so anything that MUST be done prior to the + * bootstrap cpu returning from __cpu_up() needs to go above here. + */ + cpu_set(cpuid, cpu_online_map); + cpu_set(cpuid, cpu_present_map); + + /* + * Maintain a thread mapping in addition to the cpu mapping. + */ + smp_online_threads |= (1 << p->tid); + + /* + * Enable interrupts for this thread. + */ + local_irq_enable(); + + /* + * Enter the idle loop and wait for a timer to schedule some work. + */ + printk(KERN_INFO "cpu[%d]: entering cpu_idle()\n", cpuid); + cpu_idle(); + + /* Not Reached */ +} + +/* + * smp_cpus_done() + * Called once the kernel_init() has brought up all cpu(s). + */ +void smp_cpus_done(unsigned int cpu_max) +{ + /* Do Nothing */ +} + +/* + * __cpu_up() + * Called to startup a sepcific cpu. + */ +int __cpuinit __cpu_up(unsigned int cpu) +{ + struct task_struct *idle; + unsigned int *stack; + long timeout; + struct cpuinfo_ubicom32 *p = &per_cpu(cpu_data, cpu); + + /* + * Create an idle task for this CPU. + */ + idle = fork_idle(cpu); + if (IS_ERR(idle)) { + panic("cpu[%d]: fork failed\n", cpu); + return -ENOSYS; + } + task_thread_info(idle)->cpu = cpu; + + /* + * Setup the sw_ksp[] to point to this new task. + */ + sw_ksp[p->tid] = (unsigned int)idle->stack; + stack = (unsigned int *)(sw_ksp[p->tid] + PAGE_SIZE - 8); + + /* + * Cause the specified thread to execute our smp_mainline_start + * function as a TYPE_NORMAL thread. + */ + printk(KERN_INFO "cpu[%d]: launching mainline Linux thread\n", cpu); + if (thread_start(p->tid, smp_mainline_start, (void *)NULL, stack, + THREAD_TYPE_NORMAL) == -1) { + printk(KERN_WARNING "cpu[%d]: failed thread_start\n", cpu); + return -ENOSYS; + } + + /* + * Wait for the thread to start up. The thread will set + * the online bit when it is running. Our caller execpts the + * cpu to be online if we return 0. + */ + for (timeout = 0; timeout < 10000; timeout++) { + if (cpu_online(cpu)) { + break; + } + + udelay(100); + barrier(); + continue; + } + + if (!cpu_online(cpu)) { + printk(KERN_CRIT "cpu[%d]: failed to live after %ld us\n", + cpu, timeout * 100); + return -ENOSYS; + } + + printk(KERN_INFO "cpu[%d]: came alive after %ld us\n", + cpu, timeout * 100); + return 0; +} + +/* + * Data used by setup_irq for the IPI. + */ +static struct irqaction ipi_irq = { + .name = "ipi", + .flags = IRQF_DISABLED | IRQF_PERCPU, + .handler = ipi_interrupt, +}; + +/* + * smp_prepare_cpus() + * Mark threads that are available to Linux as possible cpus(s). + */ +void __init smp_prepare_cpus(unsigned int max_cpus) +{ + int i; + + /* + * We will need a software IRQ to send IPI(s). We will use + * a single software IRQ for all IPI(s). + */ + if (irq_soft_alloc(&smp_ipi_irq) < 0) { + panic("no software IRQ is available\n"); + return; + } + + /* + * For the IPI interrupt, we want to use our own chip definition. + * This allows us to define what happens in SMP IPI without affecting + * the performance of the other interrupts. + * + * Next, Register the IPI interrupt function against the soft IRQ. + */ + set_irq_chip(smp_ipi_irq, &ubicom32_smp_chip); + setup_irq(smp_ipi_irq, &ipi_irq); + + /* + * We use the device tree node to determine how many + * free cpus we will have (up to NR_CPUS) and we indicate + * that those cpus are present. + * + * We need to do this very early in the SMP case + * because the Linux init code uses the cpu_present_map. + */ + for_each_possible_cpu(i) { + thread_t tid; + struct cpuinfo_ubicom32 *p = &per_cpu(cpu_data, i); + + /* + * Skip the bootstrap cpu + */ + if (i == 0) { + continue; + } + + /* + * If we have a free thread left in the mask, + * indicate that the cpu is present. + */ + tid = thread_alloc(); + if (tid == (thread_t)-1) { + break; + } + + /* + * Save the hardware thread id for this cpu. + */ + p->tid = tid; + cpu_set(i, cpu_present_map); + printk(KERN_INFO "cpu[%d]: added to cpu_present_map - tid: %d\n", i, tid); + } +} + +/* + * smp_prepare_boot_cpu() + * Copy the per_cpu data into the appropriate spot for the bootstrap cpu. + * + * The code in boot_cpu_init() has already set the boot cpu's + * state in the possible, present, and online maps. + */ +void __devinit smp_prepare_boot_cpu(void) +{ + struct cpuinfo_ubicom32 *p = &per_cpu(cpu_data, 0); + + smp_online_threads |= (1 << p->tid); + printk(KERN_INFO "cpu[%d]: bootstrap CPU online - tid: %ld\n", + current_thread_info()->cpu, p->tid); +} + +/* + * smp_setup_processor_id() + * Set the current_thread_info() structure cpu value. + * + * We set the value to the true hardware thread value that we are running on. + * NOTE: this function overrides the weak alias function in main.c + */ +void __init smp_setup_processor_id(void) +{ + struct cpuinfo_ubicom32 *p = &per_cpu(cpu_data, 0); + int i; + for_each_cpu_mask(i, CPU_MASK_ALL) + set_cpu_possible(i, true); + + current_thread_info()->cpu = 0; + p->tid = thread_get_self(); +} |