diff options
author | Russell King <rmk+kernel@arm.linux.org.uk> | 2013-09-17 15:13:38 +0100 |
---|---|---|
committer | Russell King <rmk+kernel@arm.linux.org.uk> | 2013-09-17 15:14:07 +0100 |
commit | 7f63037c8fb204f666367c6469f5a9b9d6888877 (patch) | |
tree | a026c999c7166aa1004e111b2e9afd7b663acdce | |
parent | 272b98c6455f00884f0350f775c5342358ebb73f (diff) | |
parent | b22537c682671de97c932d5addb6b7d087352aa1 (diff) | |
download | op-kernel-dev-7f63037c8fb204f666367c6469f5a9b9d6888877.zip op-kernel-dev-7f63037c8fb204f666367c6469f5a9b9d6888877.tar.gz |
Merge branch 'iks_for_rmk' of git://git.linaro.org/people/nico/linux into devel-stable
Nicolas Pitre writes:
This is the first part of the patch series adding IKS (In-Kernel
Switcher) support for big.LITTLE system architectures. This consists of
the core patches only. Extra patches to come later will introduce
various optimizations and tracing support.
Those patches were posted on the list a while ago here:
http://news.gmane.org/group/gmane.linux.ports.arm.kernel/thread=253942
-rw-r--r-- | arch/arm/Kconfig | 26 | ||||
-rw-r--r-- | arch/arm/common/Makefile | 2 | ||||
-rw-r--r-- | arch/arm/common/bL_switcher.c | 606 | ||||
-rw-r--r-- | arch/arm/common/bL_switcher_dummy_if.c | 71 | ||||
-rw-r--r-- | arch/arm/include/asm/bL_switcher.h | 17 | ||||
-rw-r--r-- | arch/arm/kernel/sleep.S | 26 | ||||
-rw-r--r-- | arch/arm/kernel/suspend.c | 8 | ||||
-rw-r--r-- | drivers/irqchip/irq-gic.c | 108 | ||||
-rw-r--r-- | include/linux/irqchip/arm-gic.h | 5 |
9 files changed, 848 insertions, 21 deletions
diff --git a/arch/arm/Kconfig b/arch/arm/Kconfig index 3f7714d..b313537 100644 --- a/arch/arm/Kconfig +++ b/arch/arm/Kconfig @@ -1549,6 +1549,32 @@ config MCPM for (multi-)cluster based systems, such as big.LITTLE based systems. +config BIG_LITTLE + bool "big.LITTLE support (Experimental)" + depends on CPU_V7 && SMP + select MCPM + help + This option enables support selections for the big.LITTLE + system architecture. + +config BL_SWITCHER + bool "big.LITTLE switcher support" + depends on BIG_LITTLE && MCPM && HOTPLUG_CPU + select CPU_PM + select ARM_CPU_SUSPEND + help + The big.LITTLE "switcher" provides the core functionality to + transparently handle transition between a cluster of A15's + and a cluster of A7's in a big.LITTLE system. + +config BL_SWITCHER_DUMMY_IF + tristate "Simple big.LITTLE switcher user interface" + depends on BL_SWITCHER && DEBUG_KERNEL + help + This is a simple and dummy char dev interface to control + the big.LITTLE switcher core code. It is meant for + debugging purposes only. + choice prompt "Memory split" default VMSPLIT_3G diff --git a/arch/arm/common/Makefile b/arch/arm/common/Makefile index 8c60f47..5c8584c 100644 --- a/arch/arm/common/Makefile +++ b/arch/arm/common/Makefile @@ -17,3 +17,5 @@ obj-$(CONFIG_MCPM) += mcpm_head.o mcpm_entry.o mcpm_platsmp.o vlock.o AFLAGS_mcpm_head.o := -march=armv7-a AFLAGS_vlock.o := -march=armv7-a obj-$(CONFIG_TI_PRIV_EDMA) += edma.o +obj-$(CONFIG_BL_SWITCHER) += bL_switcher.o +obj-$(CONFIG_BL_SWITCHER_DUMMY_IF) += bL_switcher_dummy_if.o diff --git a/arch/arm/common/bL_switcher.c b/arch/arm/common/bL_switcher.c new file mode 100644 index 0000000..335ff76 --- /dev/null +++ b/arch/arm/common/bL_switcher.c @@ -0,0 +1,606 @@ +/* + * arch/arm/common/bL_switcher.c -- big.LITTLE cluster switcher core driver + * + * Created by: Nicolas Pitre, March 2012 + * Copyright: (C) 2012-2013 Linaro Limited + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include <linux/init.h> +#include <linux/kernel.h> +#include <linux/module.h> +#include <linux/sched.h> +#include <linux/interrupt.h> +#include <linux/cpu_pm.h> +#include <linux/cpu.h> +#include <linux/cpumask.h> +#include <linux/kthread.h> +#include <linux/wait.h> +#include <linux/clockchips.h> +#include <linux/hrtimer.h> +#include <linux/tick.h> +#include <linux/mm.h> +#include <linux/string.h> +#include <linux/sysfs.h> +#include <linux/irqchip/arm-gic.h> +#include <linux/moduleparam.h> + +#include <asm/smp_plat.h> +#include <asm/suspend.h> +#include <asm/mcpm.h> +#include <asm/bL_switcher.h> + + +/* + * Use our own MPIDR accessors as the generic ones in asm/cputype.h have + * __attribute_const__ and we don't want the compiler to assume any + * constness here as the value _does_ change along some code paths. + */ + +static int read_mpidr(void) +{ + unsigned int id; + asm volatile ("mrc p15, 0, %0, c0, c0, 5" : "=r" (id)); + return id & MPIDR_HWID_BITMASK; +} + +/* + * bL switcher core code. + */ + +static void bL_do_switch(void *_unused) +{ + unsigned ib_mpidr, ib_cpu, ib_cluster; + + pr_debug("%s\n", __func__); + + ib_mpidr = cpu_logical_map(smp_processor_id()); + ib_cpu = MPIDR_AFFINITY_LEVEL(ib_mpidr, 0); + ib_cluster = MPIDR_AFFINITY_LEVEL(ib_mpidr, 1); + + /* + * Our state has been saved at this point. Let's release our + * inbound CPU. + */ + mcpm_set_entry_vector(ib_cpu, ib_cluster, cpu_resume); + sev(); + + /* + * From this point, we must assume that our counterpart CPU might + * have taken over in its parallel world already, as if execution + * just returned from cpu_suspend(). It is therefore important to + * be very careful not to make any change the other guy is not + * expecting. This is why we need stack isolation. + * + * Fancy under cover tasks could be performed here. For now + * we have none. + */ + + /* Let's put ourself down. */ + mcpm_cpu_power_down(); + + /* should never get here */ + BUG(); +} + +/* + * Stack isolation. To ensure 'current' remains valid, we just use another + * piece of our thread's stack space which should be fairly lightly used. + * The selected area starts just above the thread_info structure located + * at the very bottom of the stack, aligned to a cache line, and indexed + * with the cluster number. + */ +#define STACK_SIZE 512 +extern void call_with_stack(void (*fn)(void *), void *arg, void *sp); +static int bL_switchpoint(unsigned long _arg) +{ + unsigned int mpidr = read_mpidr(); + unsigned int clusterid = MPIDR_AFFINITY_LEVEL(mpidr, 1); + void *stack = current_thread_info() + 1; + stack = PTR_ALIGN(stack, L1_CACHE_BYTES); + stack += clusterid * STACK_SIZE + STACK_SIZE; + call_with_stack(bL_do_switch, (void *)_arg, stack); + BUG(); +} + +/* + * Generic switcher interface + */ + +static unsigned int bL_gic_id[MAX_CPUS_PER_CLUSTER][MAX_NR_CLUSTERS]; +static int bL_switcher_cpu_pairing[NR_CPUS]; + +/* + * bL_switch_to - Switch to a specific cluster for the current CPU + * @new_cluster_id: the ID of the cluster to switch to. + * + * This function must be called on the CPU to be switched. + * Returns 0 on success, else a negative status code. + */ +static int bL_switch_to(unsigned int new_cluster_id) +{ + unsigned int mpidr, this_cpu, that_cpu; + unsigned int ob_mpidr, ob_cpu, ob_cluster, ib_mpidr, ib_cpu, ib_cluster; + struct tick_device *tdev; + enum clock_event_mode tdev_mode; + int ret; + + this_cpu = smp_processor_id(); + ob_mpidr = read_mpidr(); + ob_cpu = MPIDR_AFFINITY_LEVEL(ob_mpidr, 0); + ob_cluster = MPIDR_AFFINITY_LEVEL(ob_mpidr, 1); + BUG_ON(cpu_logical_map(this_cpu) != ob_mpidr); + + if (new_cluster_id == ob_cluster) + return 0; + + that_cpu = bL_switcher_cpu_pairing[this_cpu]; + ib_mpidr = cpu_logical_map(that_cpu); + ib_cpu = MPIDR_AFFINITY_LEVEL(ib_mpidr, 0); + ib_cluster = MPIDR_AFFINITY_LEVEL(ib_mpidr, 1); + + pr_debug("before switch: CPU %d MPIDR %#x -> %#x\n", + this_cpu, ob_mpidr, ib_mpidr); + + /* Close the gate for our entry vectors */ + mcpm_set_entry_vector(ob_cpu, ob_cluster, NULL); + mcpm_set_entry_vector(ib_cpu, ib_cluster, NULL); + + /* + * Let's wake up the inbound CPU now in case it requires some delay + * to come online, but leave it gated in our entry vector code. + */ + ret = mcpm_cpu_power_up(ib_cpu, ib_cluster); + if (ret) { + pr_err("%s: mcpm_cpu_power_up() returned %d\n", __func__, ret); + return ret; + } + + /* + * From this point we are entering the switch critical zone + * and can't take any interrupts anymore. + */ + local_irq_disable(); + local_fiq_disable(); + + /* redirect GIC's SGIs to our counterpart */ + gic_migrate_target(bL_gic_id[ib_cpu][ib_cluster]); + + /* + * Raise a SGI on the inbound CPU to make sure it doesn't stall + * in a possible WFI, such as in mcpm_power_down(). + */ + arch_send_wakeup_ipi_mask(cpumask_of(this_cpu)); + + tdev = tick_get_device(this_cpu); + if (tdev && !cpumask_equal(tdev->evtdev->cpumask, cpumask_of(this_cpu))) + tdev = NULL; + if (tdev) { + tdev_mode = tdev->evtdev->mode; + clockevents_set_mode(tdev->evtdev, CLOCK_EVT_MODE_SHUTDOWN); + } + + ret = cpu_pm_enter(); + + /* we can not tolerate errors at this point */ + if (ret) + panic("%s: cpu_pm_enter() returned %d\n", __func__, ret); + + /* Swap the physical CPUs in the logical map for this logical CPU. */ + cpu_logical_map(this_cpu) = ib_mpidr; + cpu_logical_map(that_cpu) = ob_mpidr; + + /* Let's do the actual CPU switch. */ + ret = cpu_suspend(0, bL_switchpoint); + if (ret > 0) + panic("%s: cpu_suspend() returned %d\n", __func__, ret); + + /* We are executing on the inbound CPU at this point */ + mpidr = read_mpidr(); + pr_debug("after switch: CPU %d MPIDR %#x\n", this_cpu, mpidr); + BUG_ON(mpidr != ib_mpidr); + + mcpm_cpu_powered_up(); + + ret = cpu_pm_exit(); + + if (tdev) { + clockevents_set_mode(tdev->evtdev, tdev_mode); + clockevents_program_event(tdev->evtdev, + tdev->evtdev->next_event, 1); + } + + local_fiq_enable(); + local_irq_enable(); + + if (ret) + pr_err("%s exiting with error %d\n", __func__, ret); + return ret; +} + +struct bL_thread { + struct task_struct *task; + wait_queue_head_t wq; + int wanted_cluster; + struct completion started; +}; + +static struct bL_thread bL_threads[NR_CPUS]; + +static int bL_switcher_thread(void *arg) +{ + struct bL_thread *t = arg; + struct sched_param param = { .sched_priority = 1 }; + int cluster; + + sched_setscheduler_nocheck(current, SCHED_FIFO, ¶m); + complete(&t->started); + + do { + if (signal_pending(current)) + flush_signals(current); + wait_event_interruptible(t->wq, + t->wanted_cluster != -1 || + kthread_should_stop()); + cluster = xchg(&t->wanted_cluster, -1); + if (cluster != -1) + bL_switch_to(cluster); + } while (!kthread_should_stop()); + + return 0; +} + +static struct task_struct *bL_switcher_thread_create(int cpu, void *arg) +{ + struct task_struct *task; + + task = kthread_create_on_node(bL_switcher_thread, arg, + cpu_to_node(cpu), "kswitcher_%d", cpu); + if (!IS_ERR(task)) { + kthread_bind(task, cpu); + wake_up_process(task); + } else + pr_err("%s failed for CPU %d\n", __func__, cpu); + return task; +} + +/* + * bL_switch_request - Switch to a specific cluster for the given CPU + * + * @cpu: the CPU to switch + * @new_cluster_id: the ID of the cluster to switch to. + * + * This function causes a cluster switch on the given CPU by waking up + * the appropriate switcher thread. This function may or may not return + * before the switch has occurred. + */ +int bL_switch_request(unsigned int cpu, unsigned int new_cluster_id) +{ + struct bL_thread *t; + + if (cpu >= ARRAY_SIZE(bL_threads)) { + pr_err("%s: cpu %d out of bounds\n", __func__, cpu); + return -EINVAL; + } + + t = &bL_threads[cpu]; + if (IS_ERR(t->task)) + return PTR_ERR(t->task); + if (!t->task) + return -ESRCH; + + t->wanted_cluster = new_cluster_id; + wake_up(&t->wq); + return 0; +} +EXPORT_SYMBOL_GPL(bL_switch_request); + +/* + * Activation and configuration code. + */ + +static unsigned int bL_switcher_active; +static unsigned int bL_switcher_cpu_original_cluster[NR_CPUS]; +static cpumask_t bL_switcher_removed_logical_cpus; + +static void bL_switcher_restore_cpus(void) +{ + int i; + + for_each_cpu(i, &bL_switcher_removed_logical_cpus) + cpu_up(i); +} + +static int bL_switcher_halve_cpus(void) +{ + int i, j, cluster_0, gic_id, ret; + unsigned int cpu, cluster, mask; + cpumask_t available_cpus; + + /* First pass to validate what we have */ + mask = 0; + for_each_online_cpu(i) { + cpu = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 0); + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 1); + if (cluster >= 2) { + pr_err("%s: only dual cluster systems are supported\n", __func__); + return -EINVAL; + } + if (WARN_ON(cpu >= MAX_CPUS_PER_CLUSTER)) + return -EINVAL; + mask |= (1 << cluster); + } + if (mask != 3) { + pr_err("%s: no CPU pairing possible\n", __func__); + return -EINVAL; + } + + /* + * Now let's do the pairing. We match each CPU with another CPU + * from a different cluster. To get a uniform scheduling behavior + * without fiddling with CPU topology and compute capacity data, + * we'll use logical CPUs initially belonging to the same cluster. + */ + memset(bL_switcher_cpu_pairing, -1, sizeof(bL_switcher_cpu_pairing)); + cpumask_copy(&available_cpus, cpu_online_mask); + cluster_0 = -1; + for_each_cpu(i, &available_cpus) { + int match = -1; + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 1); + if (cluster_0 == -1) + cluster_0 = cluster; + if (cluster != cluster_0) + continue; + cpumask_clear_cpu(i, &available_cpus); + for_each_cpu(j, &available_cpus) { + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(j), 1); + /* + * Let's remember the last match to create "odd" + * pairings on purpose in order for other code not + * to assume any relation between physical and + * logical CPU numbers. + */ + if (cluster != cluster_0) + match = j; + } + if (match != -1) { + bL_switcher_cpu_pairing[i] = match; + cpumask_clear_cpu(match, &available_cpus); + pr_info("CPU%d paired with CPU%d\n", i, match); + } + } + + /* + * Now we disable the unwanted CPUs i.e. everything that has no + * pairing information (that includes the pairing counterparts). + */ + cpumask_clear(&bL_switcher_removed_logical_cpus); + for_each_online_cpu(i) { + cpu = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 0); + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(i), 1); + + /* Let's take note of the GIC ID for this CPU */ + gic_id = gic_get_cpu_id(i); + if (gic_id < 0) { + pr_err("%s: bad GIC ID for CPU %d\n", __func__, i); + bL_switcher_restore_cpus(); + return -EINVAL; + } + bL_gic_id[cpu][cluster] = gic_id; + pr_info("GIC ID for CPU %u cluster %u is %u\n", + cpu, cluster, gic_id); + + if (bL_switcher_cpu_pairing[i] != -1) { + bL_switcher_cpu_original_cluster[i] = cluster; + continue; + } + + ret = cpu_down(i); + if (ret) { + bL_switcher_restore_cpus(); + return ret; + } + cpumask_set_cpu(i, &bL_switcher_removed_logical_cpus); + } + + return 0; +} + +static int bL_switcher_enable(void) +{ + int cpu, ret; + + cpu_hotplug_driver_lock(); + if (bL_switcher_active) { + cpu_hotplug_driver_unlock(); + return 0; + } + + pr_info("big.LITTLE switcher initializing\n"); + + ret = bL_switcher_halve_cpus(); + if (ret) { + cpu_hotplug_driver_unlock(); + return ret; + } + + for_each_online_cpu(cpu) { + struct bL_thread *t = &bL_threads[cpu]; + init_waitqueue_head(&t->wq); + init_completion(&t->started); + t->wanted_cluster = -1; + t->task = bL_switcher_thread_create(cpu, t); + } + + bL_switcher_active = 1; + cpu_hotplug_driver_unlock(); + + pr_info("big.LITTLE switcher initialized\n"); + return 0; +} + +#ifdef CONFIG_SYSFS + +static void bL_switcher_disable(void) +{ + unsigned int cpu, cluster; + struct bL_thread *t; + struct task_struct *task; + + cpu_hotplug_driver_lock(); + if (!bL_switcher_active) { + cpu_hotplug_driver_unlock(); + return; + } + bL_switcher_active = 0; + + /* + * To deactivate the switcher, we must shut down the switcher + * threads to prevent any other requests from being accepted. + * Then, if the final cluster for given logical CPU is not the + * same as the original one, we'll recreate a switcher thread + * just for the purpose of switching the CPU back without any + * possibility for interference from external requests. + */ + for_each_online_cpu(cpu) { + t = &bL_threads[cpu]; + task = t->task; + t->task = NULL; + if (!task || IS_ERR(task)) + continue; + kthread_stop(task); + /* no more switch may happen on this CPU at this point */ + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(cpu), 1); + if (cluster == bL_switcher_cpu_original_cluster[cpu]) + continue; + init_completion(&t->started); + t->wanted_cluster = bL_switcher_cpu_original_cluster[cpu]; + task = bL_switcher_thread_create(cpu, t); + if (!IS_ERR(task)) { + wait_for_completion(&t->started); + kthread_stop(task); + cluster = MPIDR_AFFINITY_LEVEL(cpu_logical_map(cpu), 1); + if (cluster == bL_switcher_cpu_original_cluster[cpu]) + continue; + } + /* If execution gets here, we're in trouble. */ + pr_crit("%s: unable to restore original cluster for CPU %d\n", + __func__, cpu); + pr_crit("%s: CPU %d can't be restored\n", + __func__, bL_switcher_cpu_pairing[cpu]); + cpumask_clear_cpu(bL_switcher_cpu_pairing[cpu], + &bL_switcher_removed_logical_cpus); + } + + bL_switcher_restore_cpus(); + cpu_hotplug_driver_unlock(); +} + +static ssize_t bL_switcher_active_show(struct kobject *kobj, + struct kobj_attribute *attr, char *buf) +{ + return sprintf(buf, "%u\n", bL_switcher_active); +} + +static ssize_t bL_switcher_active_store(struct kobject *kobj, + struct kobj_attribute *attr, const char *buf, size_t count) +{ + int ret; + + switch (buf[0]) { + case '0': + bL_switcher_disable(); + ret = 0; + break; + case '1': + ret = bL_switcher_enable(); + break; + default: + ret = -EINVAL; + } + + return (ret >= 0) ? count : ret; +} + +static struct kobj_attribute bL_switcher_active_attr = + __ATTR(active, 0644, bL_switcher_active_show, bL_switcher_active_store); + +static struct attribute *bL_switcher_attrs[] = { + &bL_switcher_active_attr.attr, + NULL, +}; + +static struct attribute_group bL_switcher_attr_group = { + .attrs = bL_switcher_attrs, +}; + +static struct kobject *bL_switcher_kobj; + +static int __init bL_switcher_sysfs_init(void) +{ + int ret; + + bL_switcher_kobj = kobject_create_and_add("bL_switcher", kernel_kobj); + if (!bL_switcher_kobj) + return -ENOMEM; + ret = sysfs_create_group(bL_switcher_kobj, &bL_switcher_attr_group); + if (ret) + kobject_put(bL_switcher_kobj); + return ret; +} + +#endif /* CONFIG_SYSFS */ + +/* + * Veto any CPU hotplug operation on those CPUs we've removed + * while the switcher is active. + * We're just not ready to deal with that given the trickery involved. + */ +static int bL_switcher_hotplug_callback(struct notifier_block *nfb, + unsigned long action, void *hcpu) +{ + if (bL_switcher_active) { + int pairing = bL_switcher_cpu_pairing[(unsigned long)hcpu]; + switch (action & 0xf) { + case CPU_UP_PREPARE: + case CPU_DOWN_PREPARE: + if (pairing == -1) + return NOTIFY_BAD; + } + } + return NOTIFY_DONE; +} + +static bool no_bL_switcher; +core_param(no_bL_switcher, no_bL_switcher, bool, 0644); + +static int __init bL_switcher_init(void) +{ + int ret; + + if (MAX_NR_CLUSTERS != 2) { + pr_err("%s: only dual cluster systems are supported\n", __func__); + return -EINVAL; + } + + cpu_notifier(bL_switcher_hotplug_callback, 0); + + if (!no_bL_switcher) { + ret = bL_switcher_enable(); + if (ret) + return ret; + } + +#ifdef CONFIG_SYSFS + ret = bL_switcher_sysfs_init(); + if (ret) + pr_err("%s: unable to create sysfs entry\n", __func__); +#endif + + return 0; +} + +late_initcall(bL_switcher_init); diff --git a/arch/arm/common/bL_switcher_dummy_if.c b/arch/arm/common/bL_switcher_dummy_if.c new file mode 100644 index 0000000..3f47f12 --- /dev/null +++ b/arch/arm/common/bL_switcher_dummy_if.c @@ -0,0 +1,71 @@ +/* + * arch/arm/common/bL_switcher_dummy_if.c -- b.L switcher dummy interface + * + * Created by: Nicolas Pitre, November 2012 + * Copyright: (C) 2012-2013 Linaro Limited + * + * Dummy interface to user space for debugging purpose only. + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#include <linux/init.h> +#include <linux/module.h> +#include <linux/fs.h> +#include <linux/miscdevice.h> +#include <asm/uaccess.h> +#include <asm/bL_switcher.h> + +static ssize_t bL_switcher_write(struct file *file, const char __user *buf, + size_t len, loff_t *pos) +{ + unsigned char val[3]; + unsigned int cpu, cluster; + int ret; + + pr_debug("%s\n", __func__); + + if (len < 3) + return -EINVAL; + + if (copy_from_user(val, buf, 3)) + return -EFAULT; + + /* format: <cpu#>,<cluster#> */ + if (val[0] < '0' || val[0] > '9' || + val[1] != ',' || + val[2] < '0' || val[2] > '1') + return -EINVAL; + + cpu = val[0] - '0'; + cluster = val[2] - '0'; + ret = bL_switch_request(cpu, cluster); + + return ret ? : len; +} + +static const struct file_operations bL_switcher_fops = { + .write = bL_switcher_write, + .owner = THIS_MODULE, +}; + +static struct miscdevice bL_switcher_device = { + MISC_DYNAMIC_MINOR, + "b.L_switcher", + &bL_switcher_fops +}; + +static int __init bL_switcher_dummy_if_init(void) +{ + return misc_register(&bL_switcher_device); +} + +static void __exit bL_switcher_dummy_if_exit(void) +{ + misc_deregister(&bL_switcher_device); +} + +module_init(bL_switcher_dummy_if_init); +module_exit(bL_switcher_dummy_if_exit); diff --git a/arch/arm/include/asm/bL_switcher.h b/arch/arm/include/asm/bL_switcher.h new file mode 100644 index 0000000..e0c0bba --- /dev/null +++ b/arch/arm/include/asm/bL_switcher.h @@ -0,0 +1,17 @@ +/* + * arch/arm/include/asm/bL_switcher.h + * + * Created by: Nicolas Pitre, April 2012 + * Copyright: (C) 2012-2013 Linaro Limited + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 as + * published by the Free Software Foundation. + */ + +#ifndef ASM_BL_SWITCHER_H +#define ASM_BL_SWITCHER_H + +int bL_switch_request(unsigned int cpu, unsigned int new_cluster_id); + +#endif diff --git a/arch/arm/kernel/sleep.S b/arch/arm/kernel/sleep.S index db1536b..6224602 100644 --- a/arch/arm/kernel/sleep.S +++ b/arch/arm/kernel/sleep.S @@ -55,6 +55,7 @@ * specific registers and some other data for resume. * r0 = suspend function arg0 * r1 = suspend function + * r2 = MPIDR value the resuming CPU will use */ ENTRY(__cpu_suspend) stmfd sp!, {r4 - r11, lr} @@ -67,23 +68,18 @@ ENTRY(__cpu_suspend) mov r5, sp @ current virtual SP add r4, r4, #12 @ Space for pgd, virt sp, phys resume fn sub sp, sp, r4 @ allocate CPU state on stack - stmfd sp!, {r0, r1} @ save suspend func arg and pointer - add r0, sp, #8 @ save pointer to save block - mov r1, r4 @ size of save block - mov r2, r5 @ virtual SP ldr r3, =sleep_save_sp + stmfd sp!, {r0, r1} @ save suspend func arg and pointer ldr r3, [r3, #SLEEP_SAVE_SP_VIRT] - ALT_SMP(mrc p15, 0, r9, c0, c0, 5) - ALT_UP_B(1f) - ldr r8, =mpidr_hash - /* - * This ldmia relies on the memory layout of the mpidr_hash - * struct mpidr_hash. - */ - ldmia r8, {r4-r7} @ r4 = mpidr mask (r5,r6,r7) = l[0,1,2] shifts - compute_mpidr_hash lr, r5, r6, r7, r9, r4 - add r3, r3, lr, lsl #2 -1: + ALT_SMP(ldr r0, =mpidr_hash) + ALT_UP_B(1f) + /* This ldmia relies on the memory layout of the mpidr_hash struct */ + ldmia r0, {r1, r6-r8} @ r1 = mpidr mask (r6,r7,r8) = l[0,1,2] shifts + compute_mpidr_hash r0, r6, r7, r8, r2, r1 + add r3, r3, r0, lsl #2 +1: mov r2, r5 @ virtual SP + mov r1, r4 @ size of save block + add r0, sp, #8 @ pointer to save block bl __cpu_suspend_save adr lr, BSYM(cpu_suspend_abort) ldmfd sp!, {r0, pc} @ call suspend fn diff --git a/arch/arm/kernel/suspend.c b/arch/arm/kernel/suspend.c index 41cf3cb..2835d35 100644 --- a/arch/arm/kernel/suspend.c +++ b/arch/arm/kernel/suspend.c @@ -10,7 +10,7 @@ #include <asm/suspend.h> #include <asm/tlbflush.h> -extern int __cpu_suspend(unsigned long, int (*)(unsigned long)); +extern int __cpu_suspend(unsigned long, int (*)(unsigned long), u32 cpuid); extern void cpu_resume_mmu(void); #ifdef CONFIG_MMU @@ -21,6 +21,7 @@ extern void cpu_resume_mmu(void); int cpu_suspend(unsigned long arg, int (*fn)(unsigned long)) { struct mm_struct *mm = current->active_mm; + u32 __mpidr = cpu_logical_map(smp_processor_id()); int ret; if (!idmap_pgd) @@ -32,7 +33,7 @@ int cpu_suspend(unsigned long arg, int (*fn)(unsigned long)) * resume (indicated by a zero return code), we need to switch * back to the correct page tables. */ - ret = __cpu_suspend(arg, fn); + ret = __cpu_suspend(arg, fn, __mpidr); if (ret == 0) { cpu_switch_mm(mm->pgd, mm); local_flush_bp_all(); @@ -44,7 +45,8 @@ int cpu_suspend(unsigned long arg, int (*fn)(unsigned long)) #else int cpu_suspend(unsigned long arg, int (*fn)(unsigned long)) { - return __cpu_suspend(arg, fn); + u32 __mpidr = cpu_logical_map(smp_processor_id()); + return __cpu_suspend(arg, fn, __mpidr); } #define idmap_pgd NULL #endif diff --git a/drivers/irqchip/irq-gic.c b/drivers/irqchip/irq-gic.c index d0e9480..6365b59 100644 --- a/drivers/irqchip/irq-gic.c +++ b/drivers/irqchip/irq-gic.c @@ -253,10 +253,9 @@ static int gic_set_affinity(struct irq_data *d, const struct cpumask *mask_val, if (cpu >= NR_GIC_CPU_IF || cpu >= nr_cpu_ids) return -EINVAL; + raw_spin_lock(&irq_controller_lock); mask = 0xff << shift; bit = gic_cpu_map[cpu] << shift; - - raw_spin_lock(&irq_controller_lock); val = readl_relaxed(reg) & ~mask; writel_relaxed(val | bit, reg); raw_spin_unlock(&irq_controller_lock); @@ -652,7 +651,9 @@ static void __init gic_pm_init(struct gic_chip_data *gic) void gic_raise_softirq(const struct cpumask *mask, unsigned int irq) { int cpu; - unsigned long map = 0; + unsigned long flags, map = 0; + + raw_spin_lock_irqsave(&irq_controller_lock, flags); /* Convert our logical CPU mask into a physical one. */ for_each_cpu(cpu, mask) @@ -666,6 +667,107 @@ void gic_raise_softirq(const struct cpumask *mask, unsigned int irq) /* this always happens on GIC0 */ writel_relaxed(map << 16 | irq, gic_data_dist_base(&gic_data[0]) + GIC_DIST_SOFTINT); + + raw_spin_unlock_irqrestore(&irq_controller_lock, flags); +} +#endif + +#ifdef CONFIG_BL_SWITCHER +/* + * gic_get_cpu_id - get the CPU interface ID for the specified CPU + * + * @cpu: the logical CPU number to get the GIC ID for. + * + * Return the CPU interface ID for the given logical CPU number, + * or -1 if the CPU number is too large or the interface ID is + * unknown (more than one bit set). + */ +int gic_get_cpu_id(unsigned int cpu) +{ + unsigned int cpu_bit; + + if (cpu >= NR_GIC_CPU_IF) + return -1; + cpu_bit = gic_cpu_map[cpu]; + if (cpu_bit & (cpu_bit - 1)) + return -1; + return __ffs(cpu_bit); +} + +/* + * gic_migrate_target - migrate IRQs to another CPU interface + * + * @new_cpu_id: the CPU target ID to migrate IRQs to + * + * Migrate all peripheral interrupts with a target matching the current CPU + * to the interface corresponding to @new_cpu_id. The CPU interface mapping + * is also updated. Targets to other CPU interfaces are unchanged. + * This must be called with IRQs locally disabled. + */ +void gic_migrate_target(unsigned int new_cpu_id) +{ + unsigned int cur_cpu_id, gic_irqs, gic_nr = 0; + void __iomem *dist_base; + int i, ror_val, cpu = smp_processor_id(); + u32 val, cur_target_mask, active_mask; + + if (gic_nr >= MAX_GIC_NR) + BUG(); + + dist_base = gic_data_dist_base(&gic_data[gic_nr]); + if (!dist_base) + return; + gic_irqs = gic_data[gic_nr].gic_irqs; + + cur_cpu_id = __ffs(gic_cpu_map[cpu]); + cur_target_mask = 0x01010101 << cur_cpu_id; + ror_val = (cur_cpu_id - new_cpu_id) & 31; + + raw_spin_lock(&irq_controller_lock); + + /* Update the target interface for this logical CPU */ + gic_cpu_map[cpu] = 1 << new_cpu_id; + + /* + * Find all the peripheral interrupts targetting the current + * CPU interface and migrate them to the new CPU interface. + * We skip DIST_TARGET 0 to 7 as they are read-only. + */ + for (i = 8; i < DIV_ROUND_UP(gic_irqs, 4); i++) { + val = readl_relaxed(dist_base + GIC_DIST_TARGET + i * 4); + active_mask = val & cur_target_mask; + if (active_mask) { + val &= ~active_mask; + val |= ror32(active_mask, ror_val); + writel_relaxed(val, dist_base + GIC_DIST_TARGET + i*4); + } + } + + raw_spin_unlock(&irq_controller_lock); + + /* + * Now let's migrate and clear any potential SGIs that might be + * pending for us (cur_cpu_id). Since GIC_DIST_SGI_PENDING_SET + * is a banked register, we can only forward the SGI using + * GIC_DIST_SOFTINT. The original SGI source is lost but Linux + * doesn't use that information anyway. + * + * For the same reason we do not adjust SGI source information + * for previously sent SGIs by us to other CPUs either. + */ + for (i = 0; i < 16; i += 4) { + int j; + val = readl_relaxed(dist_base + GIC_DIST_SGI_PENDING_SET + i); + if (!val) + continue; + writel_relaxed(val, dist_base + GIC_DIST_SGI_PENDING_CLEAR + i); + for (j = i; j < i + 4; j++) { + if (val & 0xff) + writel_relaxed((1 << (new_cpu_id + 16)) | j, + dist_base + GIC_DIST_SOFTINT); + val >>= 8; + } + } } #endif diff --git a/include/linux/irqchip/arm-gic.h b/include/linux/irqchip/arm-gic.h index 0e5d9ec..46544e3 100644 --- a/include/linux/irqchip/arm-gic.h +++ b/include/linux/irqchip/arm-gic.h @@ -31,6 +31,8 @@ #define GIC_DIST_TARGET 0x800 #define GIC_DIST_CONFIG 0xc00 #define GIC_DIST_SOFTINT 0xf00 +#define GIC_DIST_SGI_PENDING_CLEAR 0xf10 +#define GIC_DIST_SGI_PENDING_SET 0xf20 #define GICH_HCR 0x0 #define GICH_VTR 0x4 @@ -74,6 +76,9 @@ static inline void gic_init(unsigned int nr, int start, gic_init_bases(nr, start, dist, cpu, 0, NULL); } +int gic_get_cpu_id(unsigned int cpu); +void gic_migrate_target(unsigned int new_cpu_id); + #endif /* __ASSEMBLY */ #endif |