diff options
Diffstat (limited to 'arch/arm/common')
| -rw-r--r-- | arch/arm/common/Kconfig | 3 | ||||
| -rw-r--r-- | arch/arm/common/Makefile | 10 | ||||
| -rw-r--r-- | arch/arm/common/bL_switcher.c | 824 | ||||
| -rw-r--r-- | arch/arm/common/bL_switcher_dummy_if.c | 71 | ||||
| -rw-r--r-- | arch/arm/common/edma.c | 1784 | ||||
| -rw-r--r-- | arch/arm/common/firmware.c | 18 | ||||
| -rw-r--r-- | arch/arm/common/it8152.c | 4 | ||||
| -rw-r--r-- | arch/arm/common/mcpm_entry.c | 296 | ||||
| -rw-r--r-- | arch/arm/common/mcpm_head.S | 233 | ||||
| -rw-r--r-- | arch/arm/common/mcpm_platsmp.c | 103 | ||||
| -rw-r--r-- | arch/arm/common/scoop.c | 3 | ||||
| -rw-r--r-- | arch/arm/common/sharpsl_param.c | 5 | ||||
| -rw-r--r-- | arch/arm/common/timer-sp.c | 155 | ||||
| -rw-r--r-- | arch/arm/common/via82c505.c | 83 | ||||
| -rw-r--r-- | arch/arm/common/vlock.S | 108 | ||||
| -rw-r--r-- | arch/arm/common/vlock.h | 29 |
16 files changed, 3618 insertions, 111 deletions
diff --git a/arch/arm/common/Kconfig b/arch/arm/common/Kconfig index 9353184d730..c3a4e9ceba3 100644 --- a/arch/arm/common/Kconfig +++ b/arch/arm/common/Kconfig @@ -17,3 +17,6 @@ config SHARP_PARAM config SHARP_SCOOP bool + +config TI_PRIV_EDMA + bool diff --git a/arch/arm/common/Makefile b/arch/arm/common/Makefile index dc8dd0de5c0..70b1eff477b 100644 --- a/arch/arm/common/Makefile +++ b/arch/arm/common/Makefile @@ -2,12 +2,20 @@ # Makefile for the linux kernel. # +obj-y += firmware.o + obj-$(CONFIG_ICST) += icst.o obj-$(CONFIG_SA1111) += sa1111.o -obj-$(CONFIG_PCI_HOST_VIA82C505) += via82c505.o obj-$(CONFIG_DMABOUNCE) += dmabounce.o obj-$(CONFIG_SHARP_LOCOMO) += locomo.o obj-$(CONFIG_SHARP_PARAM) += sharpsl_param.o obj-$(CONFIG_SHARP_SCOOP) += scoop.o obj-$(CONFIG_PCI_HOST_ITE8152) += it8152.o obj-$(CONFIG_ARM_TIMER_SP804) += timer-sp.o +obj-$(CONFIG_MCPM) += mcpm_head.o mcpm_entry.o mcpm_platsmp.o vlock.o +CFLAGS_REMOVE_mcpm_entry.o = -pg +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 00000000000..490f3dced74 --- /dev/null +++ b/arch/arm/common/bL_switcher.c @@ -0,0 +1,824 @@ +/* + * 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/atomic.h> +#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/time.h> +#include <linux/clockchips.h> +#include <linux/hrtimer.h> +#include <linux/tick.h> +#include <linux/notifier.h> +#include <linux/mm.h> +#include <linux/mutex.h> +#include <linux/smp.h> +#include <linux/spinlock.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/cputype.h> +#include <asm/suspend.h> +#include <asm/mcpm.h> +#include <asm/bL_switcher.h> + +#define CREATE_TRACE_POINTS +#include <trace/events/power_cpu_migrate.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; +} + +/* + * Get a global nanosecond time stamp for tracing. + */ +static s64 get_ns(void) +{ + struct timespec ts; + getnstimeofday(&ts); + return timespec_to_ns(&ts); +} + +/* + * bL switcher core code. + */ + +static void bL_do_switch(void *_arg) +{ + unsigned ib_mpidr, ib_cpu, ib_cluster; + long volatile handshake, **handshake_ptr = _arg; + + 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); + + /* Advertise our handshake location */ + if (handshake_ptr) { + handshake = 0; + *handshake_ptr = &handshake; + } else + handshake = -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 wait until our inbound is alive. + */ + while (!handshake) { + wfe(); + smp_mb(); + } + + /* 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 completion inbound_alive; + struct tick_device *tdev; + enum clock_event_mode tdev_mode; + long volatile *handshake_ptr; + int ipi_nr, 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); + + this_cpu = smp_processor_id(); + + /* 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); + + /* Install our "inbound alive" notifier. */ + init_completion(&inbound_alive); + ipi_nr = register_ipi_completion(&inbound_alive, this_cpu); + ipi_nr |= ((1 << 16) << bL_gic_id[ob_cpu][ob_cluster]); + mcpm_set_early_poke(ib_cpu, ib_cluster, gic_get_sgir_physaddr(), ipi_nr); + + /* + * 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; + } + + /* + * Raise a SGI on the inbound CPU to make sure it doesn't stall + * in a possible WFI, such as in bL_power_down(). + */ + gic_send_sgi(bL_gic_id[ib_cpu][ib_cluster], 0); + + /* + * Wait for the inbound to come up. This allows for other + * tasks to be scheduled in the mean time. + */ + wait_for_completion(&inbound_alive); + mcpm_set_early_poke(ib_cpu, ib_cluster, 0, 0); + + /* + * From this point we are entering the switch critical zone + * and can't take any interrupts anymore. + */ + local_irq_disable(); + local_fiq_disable(); + trace_cpu_migrate_begin(get_ns(), ob_mpidr); + + /* redirect GIC's SGIs to our counterpart */ + gic_migrate_target(bL_gic_id[ib_cpu][ib_cluster]); + + 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((unsigned long)&handshake_ptr, 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); + } + + trace_cpu_migrate_finish(get_ns(), ib_mpidr); + local_fiq_enable(); + local_irq_enable(); + + *handshake_ptr = 1; + dsb_sev(); + + if (ret) + pr_err("%s exiting with error %d\n", __func__, ret); + return ret; +} + +struct bL_thread { + spinlock_t lock; + struct task_struct *task; + wait_queue_head_t wq; + int wanted_cluster; + struct completion started; + bL_switch_completion_handler completer; + void *completer_cookie; +}; + +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; + bL_switch_completion_handler completer; + void *completer_cookie; + + 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()); + + spin_lock(&t->lock); + cluster = t->wanted_cluster; + completer = t->completer; + completer_cookie = t->completer_cookie; + t->wanted_cluster = -1; + t->completer = NULL; + spin_unlock(&t->lock); + + if (cluster != -1) { + bL_switch_to(cluster); + + if (completer) + completer(completer_cookie); + } + } 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_cb - Switch to a specific cluster for the given CPU, + * with completion notification via a callback + * + * @cpu: the CPU to switch + * @new_cluster_id: the ID of the cluster to switch to. + * @completer: switch completion callback. if non-NULL, + * @completer(@completer_cookie) will be called on completion of + * the switch, in non-atomic context. + * @completer_cookie: opaque context argument for @completer. + * + * 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. + * + * If a @completer callback function is supplied, it will be called when + * the switch is complete. This can be used to determine asynchronously + * when the switch is complete, regardless of when bL_switch_request() + * returns. When @completer is supplied, no new switch request is permitted + * for the affected CPU until after the switch is complete, and @completer + * has returned. + */ +int bL_switch_request_cb(unsigned int cpu, unsigned int new_cluster_id, + bL_switch_completion_handler completer, + void *completer_cookie) +{ + 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; + + spin_lock(&t->lock); + if (t->completer) { + spin_unlock(&t->lock); + return -EBUSY; + } + t->completer = completer; + t->completer_cookie = completer_cookie; + t->wanted_cluster = new_cluster_id; + spin_unlock(&t->lock); + wake_up(&t->wq); + return 0; +} +EXPORT_SYMBOL_GPL(bL_switch_request_cb); + +/* + * Activation and configuration code. + */ + +static DEFINE_MUTEX(bL_switcher_activation_lock); +static BLOCKING_NOTIFIER_HEAD(bL_activation_notifier); +static unsigned int bL_switcher_active; +static unsigned int bL_switcher_cpu_original_cluster[NR_CPUS]; +static cpumask_t bL_switcher_removed_logical_cpus; + +int bL_switcher_register_notifier(struct notifier_block *nb) +{ + return blocking_notifier_chain_register(&bL_activation_notifier, nb); +} +EXPORT_SYMBOL_GPL(bL_switcher_register_notifier); + +int bL_switcher_unregister_notifier(struct notifier_block *nb) +{ + return blocking_notifier_chain_unregister(&bL_activation_notifier, nb); +} +EXPORT_SYMBOL_GPL(bL_switcher_unregister_notifier); + +static int bL_activation_notify(unsigned long val) +{ + int ret; + + ret = blocking_notifier_call_chain(&bL_activation_notifier, val, NULL); + if (ret & NOTIFY_STOP_MASK) + pr_err("%s: notifier chain failed with status 0x%x\n", + __func__, ret); + return notifier_to_errno(ret); +} + +static void bL_switcher_restore_cpus(void) +{ + int i; + + for_each_cpu(i, &bL_switcher_removed_logical_cpus) { + struct device *cpu_dev = get_cpu_device(i); + int ret = device_online(cpu_dev); + if (ret) + dev_err(cpu_dev, "switcher: unable to restore CPU\n"); + } +} + +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 = device_offline(get_cpu_device(i)); + if (ret) { + bL_switcher_restore_cpus(); + return ret; + } + cpumask_set_cpu(i, &bL_switcher_removed_logical_cpus); + } + + return 0; +} + +/* Determine the logical CPU a given physical CPU is grouped on. */ +int bL_switcher_get_logical_index(u32 mpidr) +{ + int cpu; + + if (!bL_switcher_active) + return -EUNATCH; + + mpidr &= MPIDR_HWID_BITMASK; + for_each_online_cpu(cpu) { + int pairing = bL_switcher_cpu_pairing[cpu]; + if (pairing == -1) + continue; + if ((mpidr == cpu_logical_map(cpu)) || + (mpidr == cpu_logical_map(pairing))) + return cpu; + } + return -EINVAL; +} + +static void bL_switcher_trace_trigger_cpu(void *__always_unused info) +{ + trace_cpu_migrate_current(get_ns(), read_mpidr()); +} + +int bL_switcher_trace_trigger(void) +{ + int ret; + + preempt_disable(); + + bL_switcher_trace_trigger_cpu(NULL); + ret = smp_call_function(bL_switcher_trace_trigger_cpu, NULL, true); + + preempt_enable(); + + return ret; +} +EXPORT_SYMBOL_GPL(bL_switcher_trace_trigger); + +static int bL_switcher_enable(void) +{ + int cpu, ret; + + mutex_lock(&bL_switcher_activation_lock); + lock_device_hotplug(); + if (bL_switcher_active) { + unlock_device_hotplug(); + mutex_unlock(&bL_switcher_activation_lock); + return 0; + } + + pr_info("big.LITTLE switcher initializing\n"); + + ret = bL_activation_notify(BL_NOTIFY_PRE_ENABLE); + if (ret) + goto error; + + ret = bL_switcher_halve_cpus(); + if (ret) + goto error; + + bL_switcher_trace_trigger(); + + for_each_online_cpu(cpu) { + struct bL_thread *t = &bL_threads[cpu]; + spin_lock_init(&t->lock); + 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; + bL_activation_notify(BL_NOTIFY_POST_ENABLE); + pr_info("big.LITTLE switcher initialized\n"); + goto out; + +error: + pr_warn("big.LITTLE switcher initialization failed\n"); + bL_activation_notify(BL_NOTIFY_POST_DISABLE); + +out: + unlock_device_hotplug(); + mutex_unlock(&bL_switcher_activation_lock); + return ret; +} + +#ifdef CONFIG_SYSFS + +static void bL_switcher_disable(void) +{ + unsigned int cpu, cluster; + struct bL_thread *t; + struct task_struct *task; + + mutex_lock(&bL_switcher_activation_lock); + lock_device_hotplug(); + + if (!bL_switcher_active) + goto out; + + if (bL_activation_notify(BL_NOTIFY_PRE_DISABLE) != 0) { + bL_activation_notify(BL_NOTIFY_POST_ENABLE); + goto out; + } + + 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(); + bL_switcher_trace_trigger(); + + bL_activation_notify(BL_NOTIFY_POST_DISABLE); + +out: + unlock_device_hotplug(); + mutex_unlock(&bL_switcher_activation_lock); +} + +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 ssize_t bL_switcher_trace_trigger_store(struct kobject *kobj, + struct kobj_attribute *attr, const char *buf, size_t count) +{ + int ret = bL_switcher_trace_trigger(); + + return ret ? ret : count; +} + +static struct kobj_attribute bL_switcher_active_attr = + __ATTR(active, 0644, bL_switcher_active_show, bL_switcher_active_store); + +static struct kobj_attribute bL_switcher_trace_trigger_attr = + __ATTR(trace_trigger, 0200, NULL, bL_switcher_trace_trigger_store); + +static struct attribute *bL_switcher_attrs[] = { + &bL_switcher_active_attr.attr, + &bL_switcher_trace_trigger_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 */ + +bool bL_switcher_get_enabled(void) +{ + mutex_lock(&bL_switcher_activation_lock); + + return bL_switcher_active; +} +EXPORT_SYMBOL_GPL(bL_switcher_get_enabled); + +void bL_switcher_put_enabled(void) +{ + mutex_unlock(&bL_switcher_activation_lock); +} +EXPORT_SYMBOL_GPL(bL_switcher_put_enabled); + +/* + * 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 (!mcpm_is_available()) + return -ENODEV; + + 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 00000000000..3f47f1203c6 --- /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/common/edma.c b/arch/arm/common/edma.c new file mode 100644 index 00000000000..485be42519b --- /dev/null +++ b/arch/arm/common/edma.c @@ -0,0 +1,1784 @@ +/* + * EDMA3 support for DaVinci + * + * Copyright (C) 2006-2009 Texas Instruments. + * + * 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. + * + * This program 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 this program; if not, write to the Free Software + * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + */ +#include <linux/err.h> +#include <linux/kernel.h> +#include <linux/init.h> +#include <linux/module.h> +#include <linux/interrupt.h> +#include <linux/platform_device.h> +#include <linux/io.h> +#include <linux/slab.h> +#include <linux/edma.h> +#include <linux/of_address.h> +#include <linux/of_device.h> +#include <linux/of_dma.h> +#include <linux/of_irq.h> +#include <linux/pm_runtime.h> + +#include <linux/platform_data/edma.h> + +/* Offsets matching "struct edmacc_param" */ +#define PARM_OPT 0x00 +#define PARM_SRC 0x04 +#define PARM_A_B_CNT 0x08 +#define PARM_DST 0x0c +#define PARM_SRC_DST_BIDX 0x10 +#define PARM_LINK_BCNTRLD 0x14 +#define PARM_SRC_DST_CIDX 0x18 +#define PARM_CCNT 0x1c + +#define PARM_SIZE 0x20 + +/* Offsets for EDMA CC global channel registers and their shadows */ +#define SH_ER 0x00 /* 64 bits */ +#define SH_ECR 0x08 /* 64 bits */ +#define SH_ESR 0x10 /* 64 bits */ +#define SH_CER 0x18 /* 64 bits */ +#define SH_EER 0x20 /* 64 bits */ +#define SH_EECR 0x28 /* 64 bits */ +#define SH_EESR 0x30 /* 64 bits */ +#define SH_SER 0x38 /* 64 bits */ +#define SH_SECR 0x40 /* 64 bits */ +#define SH_IER 0x50 /* 64 bits */ +#define SH_IECR 0x58 /* 64 bits */ +#define SH_IESR 0x60 /* 64 bits */ +#define SH_IPR 0x68 /* 64 bits */ +#define SH_ICR 0x70 /* 64 bits */ +#define SH_IEVAL 0x78 +#define SH_QER 0x80 +#define SH_QEER 0x84 +#define SH_QEECR 0x88 +#define SH_QEESR 0x8c +#define SH_QSER 0x90 +#define SH_QSECR 0x94 +#define SH_SIZE 0x200 + +/* Offsets for EDMA CC global registers */ +#define EDMA_REV 0x0000 +#define EDMA_CCCFG 0x0004 +#define EDMA_QCHMAP 0x0200 /* 8 registers */ +#define EDMA_DMAQNUM 0x0240 /* 8 registers (4 on OMAP-L1xx) */ +#define EDMA_QDMAQNUM 0x0260 +#define EDMA_QUETCMAP 0x0280 +#define EDMA_QUEPRI 0x0284 +#define EDMA_EMR 0x0300 /* 64 bits */ +#define EDMA_EMCR 0x0308 /* 64 bits */ +#define EDMA_QEMR 0x0310 +#define EDMA_QEMCR 0x0314 +#define EDMA_CCERR 0x0318 +#define EDMA_CCERRCLR 0x031c +#define EDMA_EEVAL 0x0320 +#define EDMA_DRAE 0x0340 /* 4 x 64 bits*/ +#define EDMA_QRAE 0x0380 /* 4 registers */ +#define EDMA_QUEEVTENTRY 0x0400 /* 2 x 16 registers */ +#define EDMA_QSTAT 0x0600 /* 2 registers */ +#define EDMA_QWMTHRA 0x0620 +#define EDMA_QWMTHRB 0x0624 +#define EDMA_CCSTAT 0x0640 + +#define EDMA_M 0x1000 /* global channel registers */ +#define EDMA_ECR 0x1008 +#define EDMA_ECRH 0x100C +#define EDMA_SHADOW0 0x2000 /* 4 regions shadowing global channels */ +#define EDMA_PARM 0x4000 /* 128 param entries */ + +#define PARM_OFFSET(param_no) (EDMA_PARM + ((param_no) << 5)) + +#define EDMA_DCHMAP 0x0100 /* 64 registers */ + +/* CCCFG register */ +#define GET_NUM_DMACH(x) (x & 0x7) /* bits 0-2 */ +#define GET_NUM_PAENTRY(x) ((x & 0x7000) >> 12) /* bits 12-14 */ +#define GET_NUM_EVQUE(x) ((x & 0x70000) >> 16) /* bits 16-18 */ +#define GET_NUM_REGN(x) ((x & 0x300000) >> 20) /* bits 20-21 */ +#define CHMAP_EXIST BIT(24) + +#define EDMA_MAX_DMACH 64 +#define EDMA_MAX_PARAMENTRY 512 + +/*****************************************************************************/ + +static void __iomem *edmacc_regs_base[EDMA_MAX_CC]; + +static inline unsigned int edma_read(unsigned ctlr, int offset) +{ + return (unsigned int)__raw_readl(edmacc_regs_base[ctlr] + offset); +} + +static inline void edma_write(unsigned ctlr, int offset, int val) +{ + __raw_writel(val, edmacc_regs_base[ctlr] + offset); +} +static inline void edma_modify(unsigned ctlr, int offset, unsigned and, + unsigned or) +{ + unsigned val = edma_read(ctlr, offset); + val &= and; + val |= or; + edma_write(ctlr, offset, val); +} +static inline void edma_and(unsigned ctlr, int offset, unsigned and) +{ + unsigned val = edma_read(ctlr, offset); + val &= and; + edma_write(ctlr, offset, val); +} +static inline void edma_or(unsigned ctlr, int offset, unsigned or) +{ + unsigned val = edma_read(ctlr, offset); + val |= or; + edma_write(ctlr, offset, val); +} +static inline unsigned int edma_read_array(unsigned ctlr, int offset, int i) +{ + return edma_read(ctlr, offset + (i << 2)); +} +static inline void edma_write_array(unsigned ctlr, int offset, int i, + unsigned val) +{ + edma_write(ctlr, offset + (i << 2), val); +} +static inline void edma_modify_array(unsigned ctlr, int offset, int i, + unsigned and, unsigned or) +{ + edma_modify(ctlr, offset + (i << 2), and, or); +} +static inline void edma_or_array(unsigned ctlr, int offset, int i, unsigned or) +{ + edma_or(ctlr, offset + (i << 2), or); +} +static inline void edma_or_array2(unsigned ctlr, int offset, int i, int j, + unsigned or) +{ + edma_or(ctlr, offset + ((i*2 + j) << 2), or); +} +static inline void edma_write_array2(unsigned ctlr, int offset, int i, int j, + unsigned val) +{ + edma_write(ctlr, offset + ((i*2 + j) << 2), val); +} +static inline unsigned int edma_shadow0_read(unsigned ctlr, int offset) +{ + return edma_read(ctlr, EDMA_SHADOW0 + offset); +} +static inline unsigned int edma_shadow0_read_array(unsigned ctlr, int offset, + int i) +{ + return edma_read(ctlr, EDMA_SHADOW0 + offset + (i << 2)); +} +static inline void edma_shadow0_write(unsigned ctlr, int offset, unsigned val) +{ + edma_write(ctlr, EDMA_SHADOW0 + offset, val); +} +static inline void edma_shadow0_write_array(unsigned ctlr, int offset, int i, + unsigned val) +{ + edma_write(ctlr, EDMA_SHADOW0 + offset + (i << 2), val); +} +static inline unsigned int edma_parm_read(unsigned ctlr, int offset, + int param_no) +{ + return edma_read(ctlr, EDMA_PARM + offset + (param_no << 5)); +} +static inline void edma_parm_write(unsigned ctlr, int offset, int param_no, + unsigned val) +{ + edma_write(ctlr, EDMA_PARM + offset + (param_no << 5), val); +} +static inline void edma_parm_modify(unsigned ctlr, int offset, int param_no, + unsigned and, unsigned or) +{ + edma_modify(ctlr, EDMA_PARM + offset + (param_no << 5), and, or); +} +static inline void edma_parm_and(unsigned ctlr, int offset, int param_no, + unsigned and) +{ + edma_and(ctlr, EDMA_PARM + offset + (param_no << 5), and); +} +static inline void edma_parm_or(unsigned ctlr, int offset, int param_no, + unsigned or) +{ + edma_or(ctlr, EDMA_PARM + offset + (param_no << 5), or); +} + +static inline void set_bits(int offset, int len, unsigned long *p) +{ + for (; len > 0; len--) + set_bit(offset + (len - 1), p); +} + +static inline void clear_bits(int offset, int len, unsigned long *p) +{ + for (; len > 0; len--) + clear_bit(offset + (len - 1), p); +} + +/*****************************************************************************/ + +/* actual number of DMA channels and slots on this silicon */ +struct edma { + /* how many dma resources of each type */ + unsigned num_channels; + unsigned num_region; + unsigned num_slots; + unsigned num_tc; + enum dma_event_q default_queue; + + /* list of channels with no even trigger; terminated by "-1" */ + const s8 *noevent; + + /* The edma_inuse bit for each PaRAM slot is clear unless the + * channel is in use ... by ARM or DSP, for QDMA, or whatever. + */ + DECLARE_BITMAP(edma_inuse, EDMA_MAX_PARAMENTRY); + + /* The edma_unused bit for each channel is clear unless + * it is not being used on this platform. It uses a bit + * of SOC-specific initialization code. + */ + DECLARE_BITMAP(edma_unused, EDMA_MAX_DMACH); + + unsigned irq_res_start; + unsigned irq_res_end; + + struct dma_interrupt_data { + void (*callback)(unsigned channel, unsigned short ch_status, + void *data); + void *data; + } intr_data[EDMA_MAX_DMACH]; +}; + +static struct edma *edma_cc[EDMA_MAX_CC]; +static int arch_num_cc; + +/* dummy param set used to (re)initialize parameter RAM slots */ +static const struct edmacc_param dummy_paramset = { + .link_bcntrld = 0xffff, + .ccnt = 1, +}; + +static const struct of_device_id edma_of_ids[] = { + { .compatible = "ti,edma3", }, + {} +}; + +/*****************************************************************************/ + +static void map_dmach_queue(unsigned ctlr, unsigned ch_no, + enum dma_event_q queue_no) +{ + int bit = (ch_no & 0x7) * 4; + + /* default to low priority queue */ + if (queue_no == EVENTQ_DEFAULT) + queue_no = edma_cc[ctlr]->default_queue; + + queue_no &= 7; + edma_modify_array(ctlr, EDMA_DMAQNUM, (ch_no >> 3), + ~(0x7 << bit), queue_no << bit); +} + +static void __init assign_priority_to_queue(unsigned ctlr, int queue_no, + int priority) +{ + int bit = queue_no * 4; + edma_modify(ctlr, EDMA_QUEPRI, ~(0x7 << bit), + ((priority & 0x7) << bit)); +} + +/** + * map_dmach_param - Maps channel number to param entry number + * + * This maps the dma channel number to param entry numberter. In + * other words using the DMA channel mapping registers a param entry + * can be mapped to any channel + * + * Callers are responsible for ensuring the channel mapping logic is + * included in that particular EDMA variant (Eg : dm646x) + * + */ +static void __init map_dmach_param(unsigned ctlr) +{ + int i; + for (i = 0; i < EDMA_MAX_DMACH; i++) + edma_write_array(ctlr, EDMA_DCHMAP , i , (i << 5)); +} + +static inline void +setup_dma_interrupt(unsigned lch, + void (*callback)(unsigned channel, u16 ch_status, void *data), + void *data) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(lch); + lch = EDMA_CHAN_SLOT(lch); + + if (!callback) + edma_shadow0_write_array(ctlr, SH_IECR, lch >> 5, + BIT(lch & 0x1f)); + + edma_cc[ctlr]->intr_data[lch].callback = callback; + edma_cc[ctlr]->intr_data[lch].data = data; + + if (callback) { + edma_shadow0_write_array(ctlr, SH_ICR, lch >> 5, + BIT(lch & 0x1f)); + edma_shadow0_write_array(ctlr, SH_IESR, lch >> 5, + BIT(lch & 0x1f)); + } +} + +static int irq2ctlr(int irq) +{ + if (irq >= edma_cc[0]->irq_res_start && irq <= edma_cc[0]->irq_res_end) + return 0; + else if (irq >= edma_cc[1]->irq_res_start && + irq <= edma_cc[1]->irq_res_end) + return 1; + + return -1; +} + +/****************************************************************************** + * + * DMA interrupt handler + * + *****************************************************************************/ +static irqreturn_t dma_irq_handler(int irq, void *data) +{ + int ctlr; + u32 sh_ier; + u32 sh_ipr; + u32 bank; + + ctlr = irq2ctlr(irq); + if (ctlr < 0) + return IRQ_NONE; + + dev_dbg(data, "dma_irq_handler\n"); + + sh_ipr = edma_shadow0_read_array(ctlr, SH_IPR, 0); + if (!sh_ipr) { + sh_ipr = edma_shadow0_read_array(ctlr, SH_IPR, 1); + if (!sh_ipr) + return IRQ_NONE; + sh_ier = edma_shadow0_read_array(ctlr, SH_IER, 1); + bank = 1; + } else { + sh_ier = edma_shadow0_read_array(ctlr, SH_IER, 0); + bank = 0; + } + + do { + u32 slot; + u32 channel; + + dev_dbg(data, "IPR%d %08x\n", bank, sh_ipr); + + slot = __ffs(sh_ipr); + sh_ipr &= ~(BIT(slot)); + + if (sh_ier & BIT(slot)) { + channel = (bank << 5) | slot; + /* Clear the corresponding IPR bits */ + edma_shadow0_write_array(ctlr, SH_ICR, bank, + BIT(slot)); + if (edma_cc[ctlr]->intr_data[channel].callback) + edma_cc[ctlr]->intr_data[channel].callback( + channel, EDMA_DMA_COMPLETE, + edma_cc[ctlr]->intr_data[channel].data); + } + } while (sh_ipr); + + edma_shadow0_write(ctlr, SH_IEVAL, 1); + return IRQ_HANDLED; +} + +/****************************************************************************** + * + * DMA error interrupt handler + * + *****************************************************************************/ +static irqreturn_t dma_ccerr_handler(int irq, void *data) +{ + int i; + int ctlr; + unsigned int cnt = 0; + + ctlr = irq2ctlr(irq); + if (ctlr < 0) + return IRQ_NONE; + + dev_dbg(data, "dma_ccerr_handler\n"); + + if ((edma_read_array(ctlr, EDMA_EMR, 0) == 0) && + (edma_read_array(ctlr, EDMA_EMR, 1) == 0) && + (edma_read(ctlr, EDMA_QEMR) == 0) && + (edma_read(ctlr, EDMA_CCERR) == 0)) + return IRQ_NONE; + + while (1) { + int j = -1; + if (edma_read_array(ctlr, EDMA_EMR, 0)) + j = 0; + else if (edma_read_array(ctlr, EDMA_EMR, 1)) + j = 1; + if (j >= 0) { + dev_dbg(data, "EMR%d %08x\n", j, + edma_read_array(ctlr, EDMA_EMR, j)); + for (i = 0; i < 32; i++) { + int k = (j << 5) + i; + if (edma_read_array(ctlr, EDMA_EMR, j) & + BIT(i)) { + /* Clear the corresponding EMR bits */ + edma_write_array(ctlr, EDMA_EMCR, j, + BIT(i)); + /* Clear any SER */ + edma_shadow0_write_array(ctlr, SH_SECR, + j, BIT(i)); + if (edma_cc[ctlr]->intr_data[k]. + callback) { + edma_cc[ctlr]->intr_data[k]. + callback(k, + EDMA_DMA_CC_ERROR, + edma_cc[ctlr]->intr_data + [k].data); + } + } + } + } else if (edma_read(ctlr, EDMA_QEMR)) { + dev_dbg(data, "QEMR %02x\n", + edma_read(ctlr, EDMA_QEMR)); + for (i = 0; i < 8; i++) { + if (edma_read(ctlr, EDMA_QEMR) & BIT(i)) { + /* Clear the corresponding IPR bits */ + edma_write(ctlr, EDMA_QEMCR, BIT(i)); + edma_shadow0_write(ctlr, SH_QSECR, + BIT(i)); + + /* NOTE: not reported!! */ + } + } + } else if (edma_read(ctlr, EDMA_CCERR)) { + dev_dbg(data, "CCERR %08x\n", + edma_read(ctlr, EDMA_CCERR)); + /* FIXME: CCERR.BIT(16) ignored! much better + * to just write CCERRCLR with CCERR value... + */ + for (i = 0; i < 8; i++) { + if (edma_read(ctlr, EDMA_CCERR) & BIT(i)) { + /* Clear the corresponding IPR bits */ + edma_write(ctlr, EDMA_CCERRCLR, BIT(i)); + + /* NOTE: not reported!! */ + } + } + } + if ((edma_read_array(ctlr, EDMA_EMR, 0) == 0) && + (edma_read_array(ctlr, EDMA_EMR, 1) == 0) && + (edma_read(ctlr, EDMA_QEMR) == 0) && + (edma_read(ctlr, EDMA_CCERR) == 0)) + break; + cnt++; + if (cnt > 10) + break; + } + edma_write(ctlr, EDMA_EEVAL, 1); + return IRQ_HANDLED; +} + +static int reserve_contiguous_slots(int ctlr, unsigned int id, + unsigned int num_slots, + unsigned int start_slot) +{ + int i, j; + unsigned int count = num_slots; + int stop_slot = start_slot; + DECLARE_BITMAP(tmp_inuse, EDMA_MAX_PARAMENTRY); + + for (i = start_slot; i < edma_cc[ctlr]->num_slots; ++i) { + j = EDMA_CHAN_SLOT(i); + if (!test_and_set_bit(j, edma_cc[ctlr]->edma_inuse)) { + /* Record our current beginning slot */ + if (count == num_slots) + stop_slot = i; + + count--; + set_bit(j, tmp_inuse); + + if (count == 0) + break; + } else { + clear_bit(j, tmp_inuse); + + if (id == EDMA_CONT_PARAMS_FIXED_EXACT) { + stop_slot = i; + break; + } else { + count = num_slots; + } + } + } + + /* + * We have to clear any bits that we set + * if we run out parameter RAM slots, i.e we do find a set + * of contiguous parameter RAM slots but do not find the exact number + * requested as we may reach the total number of parameter RAM slots + */ + if (i == edma_cc[ctlr]->num_slots) + stop_slot = i; + + j = start_slot; + for_each_set_bit_from(j, tmp_inuse, stop_slot) + clear_bit(j, edma_cc[ctlr]->edma_inuse); + + if (count) + return -EBUSY; + + for (j = i - num_slots + 1; j <= i; ++j) + memcpy_toio(edmacc_regs_base[ctlr] + PARM_OFFSET(j), + &dummy_paramset, PARM_SIZE); + + return EDMA_CTLR_CHAN(ctlr, i - num_slots + 1); +} + +static int prepare_unused_channel_list(struct device *dev, void *data) +{ + struct platform_device *pdev = to_platform_device(dev); + int i, count, ctlr; + struct of_phandle_args dma_spec; + + if (dev->of_node) { + count = of_property_count_strings(dev->of_node, "dma-names"); + if (count < 0) + return 0; + for (i = 0; i < count; i++) { + if (of_parse_phandle_with_args(dev->of_node, "dmas", + "#dma-cells", i, + &dma_spec)) + continue; + + if (!of_match_node(edma_of_ids, dma_spec.np)) { + of_node_put(dma_spec.np); + continue; + } + + clear_bit(EDMA_CHAN_SLOT(dma_spec.args[0]), + edma_cc[0]->edma_unused); + of_node_put(dma_spec.np); + } + return 0; + } + + /* For non-OF case */ + for (i = 0; i < pdev->num_resources; i++) { + if ((pdev->resource[i].flags & IORESOURCE_DMA) && + (int)pdev->resource[i].start >= 0) { + ctlr = EDMA_CTLR(pdev->resource[i].start); + clear_bit(EDMA_CHAN_SLOT(pdev->resource[i].start), + edma_cc[ctlr]->edma_unused); + } + } + + return 0; +} + +/*-----------------------------------------------------------------------*/ + +static bool unused_chan_list_done; + +/* Resource alloc/free: dma channels, parameter RAM slots */ + +/** + * edma_alloc_channel - allocate DMA channel and paired parameter RAM + * @channel: specific channel to allocate; negative for "any unmapped channel" + * @callback: optional; to be issued on DMA completion or errors + * @data: passed to callback + * @eventq_no: an EVENTQ_* constant, used to choose which Transfer + * Controller (TC) executes requests using this channel. Use + * EVENTQ_DEFAULT unless you really need a high priority queue. + * + * This allocates a DMA channel and its associated parameter RAM slot. + * The parameter RAM is initialized to hold a dummy transfer. + * + * Normal use is to pass a specific channel number as @channel, to make + * use of hardware events mapped to that channel. When the channel will + * be used only for software triggering or event chaining, channels not + * mapped to hardware events (or mapped to unused events) are preferable. + * + * DMA transfers start from a channel using edma_start(), or by + * chaining. When the transfer described in that channel's parameter RAM + * slot completes, that slot's data may be reloaded through a link. + * + * DMA errors are only reported to the @callback associated with the + * channel driving that transfer, but transfer completion callbacks can + * be sent to another channel under control of the TCC field in + * the option word of the transfer's parameter RAM set. Drivers must not + * use DMA transfer completion callbacks for channels they did not allocate. + * (The same applies to TCC codes used in transfer chaining.) + * + * Returns the number of the channel, else negative errno. + */ +int edma_alloc_channel(int channel, + void (*callback)(unsigned channel, u16 ch_status, void *data), + void *data, + enum dma_event_q eventq_no) +{ + unsigned i, done = 0, ctlr = 0; + int ret = 0; + + if (!unused_chan_list_done) { + /* + * Scan all the platform devices to find out the EDMA channels + * used and clear them in the unused list, making the rest + * available for ARM usage. + */ + ret = bus_for_each_dev(&platform_bus_type, NULL, NULL, + prepare_unused_channel_list); + if (ret < 0) + return ret; + + unused_chan_list_done = true; + } + + if (channel >= 0) { + ctlr = EDMA_CTLR(channel); + channel = EDMA_CHAN_SLOT(channel); + } + + if (channel < 0) { + for (i = 0; i < arch_num_cc; i++) { + channel = 0; + for (;;) { + channel = find_next_bit(edma_cc[i]->edma_unused, + edma_cc[i]->num_channels, + channel); + if (channel == edma_cc[i]->num_channels) + break; + if (!test_and_set_bit(channel, + edma_cc[i]->edma_inuse)) { + done = 1; + ctlr = i; + break; + } + channel++; + } + if (done) + break; + } + if (!done) + return -ENOMEM; + } else if (channel >= edma_cc[ctlr]->num_channels) { + return -EINVAL; + } else if (test_and_set_bit(channel, edma_cc[ctlr]->edma_inuse)) { + return -EBUSY; + } + + /* ensure access through shadow region 0 */ + edma_or_array2(ctlr, EDMA_DRAE, 0, channel >> 5, BIT(channel & 0x1f)); + + /* ensure no events are pending */ + edma_stop(EDMA_CTLR_CHAN(ctlr, channel)); + memcpy_toio(edmacc_regs_base[ctlr] + PARM_OFFSET(channel), + &dummy_paramset, PARM_SIZE); + + if (callback) + setup_dma_interrupt(EDMA_CTLR_CHAN(ctlr, channel), + callback, data); + + map_dmach_queue(ctlr, channel, eventq_no); + + return EDMA_CTLR_CHAN(ctlr, channel); +} +EXPORT_SYMBOL(edma_alloc_channel); + + +/** + * edma_free_channel - deallocate DMA channel + * @channel: dma channel returned from edma_alloc_channel() + * + * This deallocates the DMA channel and associated parameter RAM slot + * allocated by edma_alloc_channel(). + * + * Callers are responsible for ensuring the channel is inactive, and + * will not be reactivated by linking, chaining, or software calls to + * edma_start(). + */ +void edma_free_channel(unsigned channel) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(channel); + channel = EDMA_CHAN_SLOT(channel); + + if (channel >= edma_cc[ctlr]->num_channels) + return; + + setup_dma_interrupt(channel, NULL, NULL); + /* REVISIT should probably take out of shadow region 0 */ + + memcpy_toio(edmacc_regs_base[ctlr] + PARM_OFFSET(channel), + &dummy_paramset, PARM_SIZE); + clear_bit(channel, edma_cc[ctlr]->edma_inuse); +} +EXPORT_SYMBOL(edma_free_channel); + +/** + * edma_alloc_slot - allocate DMA parameter RAM + * @slot: specific slot to allocate; negative for "any unused slot" + * + * This allocates a parameter RAM slot, initializing it to hold a + * dummy transfer. Slots allocated using this routine have not been + * mapped to a hardware DMA channel, and will normally be used by + * linking to them from a slot associated with a DMA channel. + * + * Normal use is to pass EDMA_SLOT_ANY as the @slot, but specific + * slots may be allocated on behalf of DSP firmware. + * + * Returns the number of the slot, else negative errno. + */ +int edma_alloc_slot(unsigned ctlr, int slot) +{ + if (!edma_cc[ctlr]) + return -EINVAL; + + if (slot >= 0) + slot = EDMA_CHAN_SLOT(slot); + + if (slot < 0) { + slot = edma_cc[ctlr]->num_channels; + for (;;) { + slot = find_next_zero_bit(edma_cc[ctlr]->edma_inuse, + edma_cc[ctlr]->num_slots, slot); + if (slot == edma_cc[ctlr]->num_slots) + return -ENOMEM; + if (!test_and_set_bit(slot, edma_cc[ctlr]->edma_inuse)) + break; + } + } else if (slot < edma_cc[ctlr]->num_channels || + slot >= edma_cc[ctlr]->num_slots) { + return -EINVAL; + } else if (test_and_set_bit(slot, edma_cc[ctlr]->edma_inuse)) { + return -EBUSY; + } + + memcpy_toio(edmacc_regs_base[ctlr] + PARM_OFFSET(slot), + &dummy_paramset, PARM_SIZE); + + return EDMA_CTLR_CHAN(ctlr, slot); +} +EXPORT_SYMBOL(edma_alloc_slot); + +/** + * edma_free_slot - deallocate DMA parameter RAM + * @slot: parameter RAM slot returned from edma_alloc_slot() + * + * This deallocates the parameter RAM slot allocated by edma_alloc_slot(). + * Callers are responsible for ensuring the slot is inactive, and will + * not be activated. + */ +void edma_free_slot(unsigned slot) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(slot); + slot = EDMA_CHAN_SLOT(slot); + + if (slot < edma_cc[ctlr]->num_channels || + slot >= edma_cc[ctlr]->num_slots) + return; + + memcpy_toio(edmacc_regs_base[ctlr] + PARM_OFFSET(slot), + &dummy_paramset, PARM_SIZE); + clear_bit(slot, edma_cc[ctlr]->edma_inuse); +} +EXPORT_SYMBOL(edma_free_slot); + + +/** + * edma_alloc_cont_slots- alloc contiguous parameter RAM slots + * The API will return the starting point of a set of + * contiguous parameter RAM slots that have been requested + * + * @id: can only be EDMA_CONT_PARAMS_ANY or EDMA_CONT_PARAMS_FIXED_EXACT + * or EDMA_CONT_PARAMS_FIXED_NOT_EXACT + * @count: number of contiguous Paramter RAM slots + * @slot - the start value of Parameter RAM slot that should be passed if id + * is EDMA_CONT_PARAMS_FIXED_EXACT or EDMA_CONT_PARAMS_FIXED_NOT_EXACT + * + * If id is EDMA_CONT_PARAMS_ANY then the API starts looking for a set of + * contiguous Parameter RAM slots from parameter RAM 64 in the case of + * DaVinci SOCs and 32 in the case of DA8xx SOCs. + * + * If id is EDMA_CONT_PARAMS_FIXED_EXACT then the API starts looking for a + * set of contiguous parameter RAM slots from the "slot" that is passed as an + * argument to the API. + * + * If id is EDMA_CONT_PARAMS_FIXED_NOT_EXACT then the API initially tries + * starts looking for a set of contiguous parameter RAMs from the "slot" + * that is passed as an argument to the API. On failure the API will try to + * find a set of contiguous Parameter RAM slots from the remaining Parameter + * RAM slots + */ +int edma_alloc_cont_slots(unsigned ctlr, unsigned int id, int slot, int count) +{ + /* + * The start slot requested should be greater than + * the number of channels and lesser than the total number + * of slots + */ + if ((id != EDMA_CONT_PARAMS_ANY) && + (slot < edma_cc[ctlr]->num_channels || + slot >= edma_cc[ctlr]->num_slots)) + return -EINVAL; + + /* + * The number of parameter RAM slots requested cannot be less than 1 + * and cannot be more than the number of slots minus the number of + * channels + */ + if (count < 1 || count > + (edma_cc[ctlr]->num_slots - edma_cc[ctlr]->num_channels)) + return -EINVAL; + + switch (id) { + case EDMA_CONT_PARAMS_ANY: + return reserve_contiguous_slots(ctlr, id, count, + edma_cc[ctlr]->num_channels); + case EDMA_CONT_PARAMS_FIXED_EXACT: + case EDMA_CONT_PARAMS_FIXED_NOT_EXACT: + return reserve_contiguous_slots(ctlr, id, count, slot); + default: + return -EINVAL; + } + +} +EXPORT_SYMBOL(edma_alloc_cont_slots); + +/** + * edma_free_cont_slots - deallocate DMA parameter RAM slots + * @slot: first parameter RAM of a set of parameter RAM slots to be freed + * @count: the number of contiguous parameter RAM slots to be freed + * + * This deallocates the parameter RAM slots allocated by + * edma_alloc_cont_slots. + * Callers/applications need to keep track of sets of contiguous + * parameter RAM slots that have been allocated using the edma_alloc_cont_slots + * API. + * Callers are responsible for ensuring the slots are inactive, and will + * not be activated. + */ +int edma_free_cont_slots(unsigned slot, int count) +{ + unsigned ctlr, slot_to_free; + int i; + + ctlr = EDMA_CTLR(slot); + slot = EDMA_CHAN_SLOT(slot); + + if (slot < edma_cc[ctlr]->num_channels || + slot >= edma_cc[ctlr]->num_slots || + count < 1) + return -EINVAL; + + for (i = slot; i < slot + count; ++i) { + ctlr = EDMA_CTLR(i); + slot_to_free = EDMA_CHAN_SLOT(i); + + memcpy_toio(edmacc_regs_base[ctlr] + PARM_OFFSET(slot_to_free), + &dummy_paramset, PARM_SIZE); + clear_bit(slot_to_free, edma_cc[ctlr]->edma_inuse); + } + + return 0; +} +EXPORT_SYMBOL(edma_free_cont_slots); + +/*-----------------------------------------------------------------------*/ + +/* Parameter RAM operations (i) -- read/write partial slots */ + +/** + * edma_set_src - set initial DMA source address in parameter RAM slot + * @slot: parameter RAM slot being configured + * @src_port: physical address of source (memory, controller FIFO, etc) + * @addressMode: INCR, except in very rare cases + * @fifoWidth: ignored unless @addressMode is FIFO, else specifies the + * width to use when addressing the fifo (e.g. W8BIT, W32BIT) + * + * Note that the source address is modified during the DMA transfer + * according to edma_set_src_index(). + */ +void edma_set_src(unsigned slot, dma_addr_t src_port, + enum address_mode mode, enum fifo_width width) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(slot); + slot = EDMA_CHAN_SLOT(slot); + + if (slot < edma_cc[ctlr]->num_slots) { + unsigned int i = edma_parm_read(ctlr, PARM_OPT, slot); + + if (mode) { + /* set SAM and program FWID */ + i = (i & ~(EDMA_FWID)) | (SAM | ((width & 0x7) << 8)); + } else { + /* clear SAM */ + i &= ~SAM; + } + edma_parm_write(ctlr, PARM_OPT, slot, i); + + /* set the source port address + in source register of param structure */ + edma_parm_write(ctlr, PARM_SRC, slot, src_port); + } +} +EXPORT_SYMBOL(edma_set_src); + +/** + * edma_set_dest - set initial DMA destination address in parameter RAM slot + * @slot: parameter RAM slot being configured + * @dest_port: physical address of destination (memory, controller FIFO, etc) + * @addressMode: INCR, except in very rare cases + * @fifoWidth: ignored unless @addressMode is FIFO, else specifies the + * width to use when addressing the fifo (e.g. W8BIT, W32BIT) + * + * Note that the destination address is modified during the DMA transfer + * according to edma_set_dest_index(). + */ +void edma_set_dest(unsigned slot, dma_addr_t dest_port, + enum address_mode mode, enum fifo_width width) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(slot); + slot = EDMA_CHAN_SLOT(slot); + + if (slot < edma_cc[ctlr]->num_slots) { + unsigned int i = edma_parm_read(ctlr, PARM_OPT, slot); + + if (mode) { + /* set DAM and program FWID */ + i = (i & ~(EDMA_FWID)) | (DAM | ((width & 0x7) << 8)); + } else { + /* clear DAM */ + i &= ~DAM; + } + edma_parm_write(ctlr, PARM_OPT, slot, i); + /* set the destination port address + in dest register of param structure */ + edma_parm_write(ctlr, PARM_DST, slot, dest_port); + } +} +EXPORT_SYMBOL(edma_set_dest); + +/** + * edma_get_position - returns the current transfer point + * @slot: parameter RAM slot being examined + * @dst: true selects the dest position, false the source + * + * Returns the position of the current active slot + */ +dma_addr_t edma_get_position(unsigned slot, bool dst) +{ + u32 offs, ctlr = EDMA_CTLR(slot); + + slot = EDMA_CHAN_SLOT(slot); + + offs = PARM_OFFSET(slot); + offs += dst ? PARM_DST : PARM_SRC; + + return edma_read(ctlr, offs); +} + +/** + * edma_set_src_index - configure DMA source address indexing + * @slot: parameter RAM slot being configured + * @src_bidx: byte offset between source arrays in a frame + * @src_cidx: byte offset between source frames in a block + * + * Offsets are specified to support either contiguous or discontiguous + * memory transfers, or repeated access to a hardware register, as needed. + * When accessing hardware registers, both offsets are normally zero. + */ +void edma_set_src_index(unsigned slot, s16 src_bidx, s16 src_cidx) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(slot); + slot = EDMA_CHAN_SLOT(slot); + + if (slot < edma_cc[ctlr]->num_slots) { + edma_parm_modify(ctlr, PARM_SRC_DST_BIDX, slot, + 0xffff0000, src_bidx); + edma_parm_modify(ctlr, PARM_SRC_DST_CIDX, slot, + 0xffff0000, src_cidx); + } +} +EXPORT_SYMBOL(edma_set_src_index); + +/** + * edma_set_dest_index - configure DMA destination address indexing + * @slot: parameter RAM slot being configured + * @dest_bidx: byte offset between destination arrays in a frame + * @dest_cidx: byte offset between destination frames in a block + * + * Offsets are specified to support either contiguous or discontiguous + * memory transfers, or repeated access to a hardware register, as needed. + * When accessing hardware registers, both offsets are normally zero. + */ +void edma_set_dest_index(unsigned slot, s16 dest_bidx, s16 dest_cidx) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(slot); + slot = EDMA_CHAN_SLOT(slot); + + if (slot < edma_cc[ctlr]->num_slots) { + edma_parm_modify(ctlr, PARM_SRC_DST_BIDX, slot, + 0x0000ffff, dest_bidx << 16); + edma_parm_modify(ctlr, PARM_SRC_DST_CIDX, slot, + 0x0000ffff, dest_cidx << 16); + } +} +EXPORT_SYMBOL(edma_set_dest_index); + +/** + * edma_set_transfer_params - configure DMA transfer parameters + * @slot: parameter RAM slot being configured + * @acnt: how many bytes per array (at least one) + * @bcnt: how many arrays per frame (at least one) + * @ccnt: how many frames per block (at least one) + * @bcnt_rld: used only for A-Synchronized transfers; this specifies + * the value to reload into bcnt when it decrements to zero + * @sync_mode: ASYNC or ABSYNC + * + * See the EDMA3 documentation to understand how to configure and link + * transfers using the fields in PaRAM slots. If you are not doing it + * all at once with edma_write_slot(), you will use this routine + * plus two calls each for source and destination, setting the initial + * address and saying how to index that address. + * + * An example of an A-Synchronized transfer is a serial link using a + * single word shift register. In that case, @acnt would be equal to + * that word size; the serial controller issues a DMA synchronization + * event to transfer each word, and memory access by the DMA transfer + * controller will be word-at-a-time. + * + * An example of an AB-Synchronized transfer is a device using a FIFO. + * In that case, @acnt equals the FIFO width and @bcnt equals its depth. + * The controller with the FIFO issues DMA synchronization events when + * the FIFO threshold is reached, and the DMA transfer controller will + * transfer one frame to (or from) the FIFO. It will probably use + * efficient burst modes to access memory. + */ +void edma_set_transfer_params(unsigned slot, + u16 acnt, u16 bcnt, u16 ccnt, + u16 bcnt_rld, enum sync_dimension sync_mode) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(slot); + slot = EDMA_CHAN_SLOT(slot); + + if (slot < edma_cc[ctlr]->num_slots) { + edma_parm_modify(ctlr, PARM_LINK_BCNTRLD, slot, + 0x0000ffff, bcnt_rld << 16); + if (sync_mode == ASYNC) + edma_parm_and(ctlr, PARM_OPT, slot, ~SYNCDIM); + else + edma_parm_or(ctlr, PARM_OPT, slot, SYNCDIM); + /* Set the acount, bcount, ccount registers */ + edma_parm_write(ctlr, PARM_A_B_CNT, slot, (bcnt << 16) | acnt); + edma_parm_write(ctlr, PARM_CCNT, slot, ccnt); + } +} +EXPORT_SYMBOL(edma_set_transfer_params); + +/** + * edma_link - link one parameter RAM slot to another + * @from: parameter RAM slot originating the link + * @to: parameter RAM slot which is the link target + * + * The originating slot should not be part of any active DMA transfer. + */ +void edma_link(unsigned from, unsigned to) +{ + unsigned ctlr_from, ctlr_to; + + ctlr_from = EDMA_CTLR(from); + from = EDMA_CHAN_SLOT(from); + ctlr_to = EDMA_CTLR(to); + to = EDMA_CHAN_SLOT(to); + + if (from >= edma_cc[ctlr_from]->num_slots) + return; + if (to >= edma_cc[ctlr_to]->num_slots) + return; + edma_parm_modify(ctlr_from, PARM_LINK_BCNTRLD, from, 0xffff0000, + PARM_OFFSET(to)); +} +EXPORT_SYMBOL(edma_link); + +/** + * edma_unlink - cut link from one parameter RAM slot + * @from: parameter RAM slot originating the link + * + * The originating slot should not be part of any active DMA transfer. + * Its link is set to 0xffff. + */ +void edma_unlink(unsigned from) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(from); + from = EDMA_CHAN_SLOT(from); + + if (from >= edma_cc[ctlr]->num_slots) + return; + edma_parm_or(ctlr, PARM_LINK_BCNTRLD, from, 0xffff); +} +EXPORT_SYMBOL(edma_unlink); + +/*-----------------------------------------------------------------------*/ + +/* Parameter RAM operations (ii) -- read/write whole parameter sets */ + +/** + * edma_write_slot - write parameter RAM data for slot + * @slot: number of parameter RAM slot being modified + * @param: data to be written into parameter RAM slot + * + * Use this to assign all parameters of a transfer at once. This + * allows more efficient setup of transfers than issuing multiple + * calls to set up those parameters in small pieces, and provides + * complete control over all transfer options. + */ +void edma_write_slot(unsigned slot, const struct edmacc_param *param) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(slot); + slot = EDMA_CHAN_SLOT(slot); + + if (slot >= edma_cc[ctlr]->num_slots) + return; + memcpy_toio(edmacc_regs_base[ctlr] + PARM_OFFSET(slot), param, + PARM_SIZE); +} +EXPORT_SYMBOL(edma_write_slot); + +/** + * edma_read_slot - read parameter RAM data from slot + * @slot: number of parameter RAM slot being copied + * @param: where to store copy of parameter RAM data + * + * Use this to read data from a parameter RAM slot, perhaps to + * save them as a template for later reuse. + */ +void edma_read_slot(unsigned slot, struct edmacc_param *param) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(slot); + slot = EDMA_CHAN_SLOT(slot); + + if (slot >= edma_cc[ctlr]->num_slots) + return; + memcpy_fromio(param, edmacc_regs_base[ctlr] + PARM_OFFSET(slot), + PARM_SIZE); +} +EXPORT_SYMBOL(edma_read_slot); + +/*-----------------------------------------------------------------------*/ + +/* Various EDMA channel control operations */ + +/** + * edma_pause - pause dma on a channel + * @channel: on which edma_start() has been called + * + * This temporarily disables EDMA hardware events on the specified channel, + * preventing them from triggering new transfers on its behalf + */ +void edma_pause(unsigned channel) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(channel); + channel = EDMA_CHAN_SLOT(channel); + + if (channel < edma_cc[ctlr]->num_channels) { + unsigned int mask = BIT(channel & 0x1f); + + edma_shadow0_write_array(ctlr, SH_EECR, channel >> 5, mask); + } +} +EXPORT_SYMBOL(edma_pause); + +/** + * edma_resume - resumes dma on a paused channel + * @channel: on which edma_pause() has been called + * + * This re-enables EDMA hardware events on the specified channel. + */ +void edma_resume(unsigned channel) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(channel); + channel = EDMA_CHAN_SLOT(channel); + + if (channel < edma_cc[ctlr]->num_channels) { + unsigned int mask = BIT(channel & 0x1f); + + edma_shadow0_write_array(ctlr, SH_EESR, channel >> 5, mask); + } +} +EXPORT_SYMBOL(edma_resume); + +int edma_trigger_channel(unsigned channel) +{ + unsigned ctlr; + unsigned int mask; + + ctlr = EDMA_CTLR(channel); + channel = EDMA_CHAN_SLOT(channel); + mask = BIT(channel & 0x1f); + + edma_shadow0_write_array(ctlr, SH_ESR, (channel >> 5), mask); + + pr_debug("EDMA: ESR%d %08x\n", (channel >> 5), + edma_shadow0_read_array(ctlr, SH_ESR, (channel >> 5))); + return 0; +} +EXPORT_SYMBOL(edma_trigger_channel); + +/** + * edma_start - start dma on a channel + * @channel: channel being activated + * + * Channels with event associations will be triggered by their hardware + * events, and channels without such associations will be triggered by + * software. (At this writing there is no interface for using software + * triggers except with channels that don't support hardware triggers.) + * + * Returns zero on success, else negative errno. + */ +int edma_start(unsigned channel) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(channel); + channel = EDMA_CHAN_SLOT(channel); + + if (channel < edma_cc[ctlr]->num_channels) { + int j = channel >> 5; + unsigned int mask = BIT(channel & 0x1f); + + /* EDMA channels without event association */ + if (test_bit(channel, edma_cc[ctlr]->edma_unused)) { + pr_debug("EDMA: ESR%d %08x\n", j, + edma_shadow0_read_array(ctlr, SH_ESR, j)); + edma_shadow0_write_array(ctlr, SH_ESR, j, mask); + return 0; + } + + /* EDMA channel with event association */ + pr_debug("EDMA: ER%d %08x\n", j, + edma_shadow0_read_array(ctlr, SH_ER, j)); + /* Clear any pending event or error */ + edma_write_array(ctlr, EDMA_ECR, j, mask); + edma_write_array(ctlr, EDMA_EMCR, j, mask); + /* Clear any SER */ + edma_shadow0_write_array(ctlr, SH_SECR, j, mask); + edma_shadow0_write_array(ctlr, SH_EESR, j, mask); + pr_debug("EDMA: EER%d %08x\n", j, + edma_shadow0_read_array(ctlr, SH_EER, j)); + return 0; + } + + return -EINVAL; +} +EXPORT_SYMBOL(edma_start); + +/** + * edma_stop - stops dma on the channel passed + * @channel: channel being deactivated + * + * When @lch is a channel, any active transfer is paused and + * all pending hardware events are cleared. The current transfer + * may not be resumed, and the channel's Parameter RAM should be + * reinitialized before being reused. + */ +void edma_stop(unsigned channel) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(channel); + channel = EDMA_CHAN_SLOT(channel); + + if (channel < edma_cc[ctlr]->num_channels) { + int j = channel >> 5; + unsigned int mask = BIT(channel & 0x1f); + + edma_shadow0_write_array(ctlr, SH_EECR, j, mask); + edma_shadow0_write_array(ctlr, SH_ECR, j, mask); + edma_shadow0_write_array(ctlr, SH_SECR, j, mask); + edma_write_array(ctlr, EDMA_EMCR, j, mask); + + pr_debug("EDMA: EER%d %08x\n", j, + edma_shadow0_read_array(ctlr, SH_EER, j)); + + /* REVISIT: consider guarding against inappropriate event + * chaining by overwriting with dummy_paramset. + */ + } +} +EXPORT_SYMBOL(edma_stop); + +/****************************************************************************** + * + * It cleans ParamEntry qand bring back EDMA to initial state if media has + * been removed before EDMA has finished.It is usedful for removable media. + * Arguments: + * ch_no - channel no + * + * Return: zero on success, or corresponding error no on failure + * + * FIXME this should not be needed ... edma_stop() should suffice. + * + *****************************************************************************/ + +void edma_clean_channel(unsigned channel) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(channel); + channel = EDMA_CHAN_SLOT(channel); + + if (channel < edma_cc[ctlr]->num_channels) { + int j = (channel >> 5); + unsigned int mask = BIT(channel & 0x1f); + + pr_debug("EDMA: EMR%d %08x\n", j, + edma_read_array(ctlr, EDMA_EMR, j)); + edma_shadow0_write_array(ctlr, SH_ECR, j, mask); + /* Clear the corresponding EMR bits */ + edma_write_array(ctlr, EDMA_EMCR, j, mask); + /* Clear any SER */ + edma_shadow0_write_array(ctlr, SH_SECR, j, mask); + edma_write(ctlr, EDMA_CCERRCLR, BIT(16) | BIT(1) | BIT(0)); + } +} +EXPORT_SYMBOL(edma_clean_channel); + +/* + * edma_clear_event - clear an outstanding event on the DMA channel + * Arguments: + * channel - channel number + */ +void edma_clear_event(unsigned channel) +{ + unsigned ctlr; + + ctlr = EDMA_CTLR(channel); + channel = EDMA_CHAN_SLOT(channel); + + if (channel >= edma_cc[ctlr]->num_channels) + return; + if (channel < 32) + edma_write(ctlr, EDMA_ECR, BIT(channel)); + else + edma_write(ctlr, EDMA_ECRH, BIT(channel - 32)); +} +EXPORT_SYMBOL(edma_clear_event); + +static int edma_setup_from_hw(struct device *dev, struct edma_soc_info *pdata, + struct edma *edma_cc) +{ + int i; + u32 value, cccfg; + s8 (*queue_priority_map)[2]; + + /* Decode the eDMA3 configuration from CCCFG register */ + cccfg = edma_read(0, EDMA_CCCFG); + + value = GET_NUM_REGN(cccfg); + edma_cc->num_region = BIT(value); + + value = GET_NUM_DMACH(cccfg); + edma_cc->num_channels = BIT(value + 1); + + value = GET_NUM_PAENTRY(cccfg); + edma_cc->num_slots = BIT(value + 4); + + value = GET_NUM_EVQUE(cccfg); + edma_cc->num_tc = value + 1; + + dev_dbg(dev, "eDMA3 HW configuration (cccfg: 0x%08x):\n", cccfg); + dev_dbg(dev, "num_region: %u\n", edma_cc->num_region); + dev_dbg(dev, "num_channel: %u\n", edma_cc->num_channels); + dev_dbg(dev, "num_slot: %u\n", edma_cc->num_slots); + dev_dbg(dev, "num_tc: %u\n", edma_cc->num_tc); + + /* Nothing need to be done if queue priority is provided */ + if (pdata->queue_priority_mapping) + return 0; + + /* + * Configure TC/queue priority as follows: + * Q0 - priority 0 + * Q1 - priority 1 + * Q2 - priority 2 + * ... + * The meaning of priority numbers: 0 highest priority, 7 lowest + * priority. So Q0 is the highest priority queue and the last queue has + * the lowest priority. + */ + queue_priority_map = devm_kzalloc(dev, + (edma_cc->num_tc + 1) * sizeof(s8), + GFP_KERNEL); + if (!queue_priority_map) + return -ENOMEM; + + for (i = 0; i < edma_cc->num_tc; i++) { + queue_priority_map[i][0] = i; + queue_priority_map[i][1] = i; + } + queue_priority_map[i][0] = -1; + queue_priority_map[i][1] = -1; + + pdata->queue_priority_mapping = queue_priority_map; + pdata->default_queue = 0; + + return 0; +} + +#if IS_ENABLED(CONFIG_OF) && IS_ENABLED(CONFIG_DMADEVICES) + +static int edma_xbar_event_map(struct device *dev, struct device_node *node, + struct edma_soc_info *pdata, size_t sz) +{ + const char pname[] = "ti,edma-xbar-event-map"; + struct resource res; + void __iomem *xbar; + s16 (*xbar_chans)[2]; + size_t nelm = sz / sizeof(s16); + u32 shift, offset, mux; + int ret, i; + + xbar_chans = devm_kzalloc(dev, (nelm + 2) * sizeof(s16), GFP_KERNEL); + if (!xbar_chans) + return -ENOMEM; + + ret = of_address_to_resource(node, 1, &res); + if (ret) + return -ENOMEM; + + xbar = devm_ioremap(dev, res.start, resource_size(&res)); + if (!xbar) + return -ENOMEM; + + ret = of_property_read_u16_array(node, pname, (u16 *)xbar_chans, nelm); + if (ret) + return -EIO; + + /* Invalidate last entry for the other user of this mess */ + nelm >>= 1; + xbar_chans[nelm][0] = xbar_chans[nelm][1] = -1; + + for (i = 0; i < nelm; i++) { + shift = (xbar_chans[i][1] & 0x03) << 3; + offset = xbar_chans[i][1] & 0xfffffffc; + mux = readl(xbar + offset); + mux &= ~(0xff << shift); + mux |= xbar_chans[i][0] << shift; + writel(mux, (xbar + offset)); + } + + pdata->xbar_chans = (const s16 (*)[2]) xbar_chans; + return 0; +} + +static int edma_of_parse_dt(struct device *dev, + struct device_node *node, + struct edma_soc_info *pdata) +{ + int ret = 0; + struct property *prop; + size_t sz; + struct edma_rsv_info *rsv_info; + + rsv_info = devm_kzalloc(dev, sizeof(struct edma_rsv_info), GFP_KERNEL); + if (!rsv_info) + return -ENOMEM; + pdata->rsv = rsv_info; + + prop = of_find_property(node, "ti,edma-xbar-event-map", &sz); + if (prop) + ret = edma_xbar_event_map(dev, node, pdata, sz); + + return ret; +} + +static struct of_dma_filter_info edma_filter_info = { + .filter_fn = edma_filter_fn, +}; + +static struct edma_soc_info *edma_setup_info_from_dt(struct device *dev, + struct device_node *node) +{ + struct edma_soc_info *info; + int ret; + + info = devm_kzalloc(dev, sizeof(struct edma_soc_info), GFP_KERNEL); + if (!info) + return ERR_PTR(-ENOMEM); + + ret = edma_of_parse_dt(dev, node, info); + if (ret) + return ERR_PTR(ret); + + dma_cap_set(DMA_SLAVE, edma_filter_info.dma_cap); + dma_cap_set(DMA_CYCLIC, edma_filter_info.dma_cap); + of_dma_controller_register(dev->of_node, of_dma_simple_xlate, + &edma_filter_info); + + return info; +} +#else +static struct edma_soc_info *edma_setup_info_from_dt(struct device *dev, + struct device_node *node) +{ + return ERR_PTR(-ENOSYS); +} +#endif + +static int edma_probe(struct platform_device *pdev) +{ + struct edma_soc_info **info = pdev->dev.platform_data; + struct edma_soc_info *ninfo[EDMA_MAX_CC] = {NULL}; + s8 (*queue_priority_mapping)[2]; + int i, j, off, ln, found = 0; + int status = -1; + const s16 (*rsv_chans)[2]; + const s16 (*rsv_slots)[2]; + const s16 (*xbar_chans)[2]; + int irq[EDMA_MAX_CC] = {0, 0}; + int err_irq[EDMA_MAX_CC] = {0, 0}; + struct resource *r[EDMA_MAX_CC] = {NULL}; + struct resource res[EDMA_MAX_CC]; + char res_name[10]; + struct device_node *node = pdev->dev.of_node; + struct device *dev = &pdev->dev; + int ret; + + if (node) { + /* Check if this is a second instance registered */ + if (arch_num_cc) { + dev_err(dev, "only one EDMA instance is supported via DT\n"); + return -ENODEV; + } + + ninfo[0] = edma_setup_info_from_dt(dev, node); + if (IS_ERR(ninfo[0])) { + dev_err(dev, "failed to get DT data\n"); + return PTR_ERR(ninfo[0]); + } + + info = ninfo; + } + + if (!info) + return -ENODEV; + + pm_runtime_enable(dev); + ret = pm_runtime_get_sync(dev); + if (ret < 0) { + dev_err(dev, "pm_runtime_get_sync() failed\n"); + return ret; + } + + for (j = 0; j < EDMA_MAX_CC; j++) { + if (!info[j]) { + if (!found) + return -ENODEV; + break; + } + if (node) { + ret = of_address_to_resource(node, j, &res[j]); + if (!ret) + r[j] = &res[j]; + } else { + sprintf(res_name, "edma_cc%d", j); + r[j] = platform_get_resource_byname(pdev, + IORESOURCE_MEM, + res_name); + } + if (!r[j]) { + if (found) + break; + else + return -ENODEV; + } else { + found = 1; + } + + edmacc_regs_base[j] = devm_ioremap_resource(&pdev->dev, r[j]); + if (IS_ERR(edmacc_regs_base[j])) + return PTR_ERR(edmacc_regs_base[j]); + + edma_cc[j] = devm_kzalloc(&pdev->dev, sizeof(struct edma), + GFP_KERNEL); + if (!edma_cc[j]) + return -ENOMEM; + + /* Get eDMA3 configuration from IP */ + ret = edma_setup_from_hw(dev, info[j], edma_cc[j]); + if (ret) + return ret; + + edma_cc[j]->default_queue = info[j]->default_queue; + + dev_dbg(&pdev->dev, "DMA REG BASE ADDR=%p\n", + edmacc_regs_base[j]); + + for (i = 0; i < edma_cc[j]->num_slots; i++) + memcpy_toio(edmacc_regs_base[j] + PARM_OFFSET(i), + &dummy_paramset, PARM_SIZE); + + /* Mark all channels as unused */ + memset(edma_cc[j]->edma_unused, 0xff, + sizeof(edma_cc[j]->edma_unused)); + + if (info[j]->rsv) { + + /* Clear the reserved channels in unused list */ + rsv_chans = info[j]->rsv->rsv_chans; + if (rsv_chans) { + for (i = 0; rsv_chans[i][0] != -1; i++) { + off = rsv_chans[i][0]; + ln = rsv_chans[i][1]; + clear_bits(off, ln, + edma_cc[j]->edma_unused); + } + } + + /* Set the reserved slots in inuse list */ + rsv_slots = info[j]->rsv->rsv_slots; + if (rsv_slots) { + for (i = 0; rsv_slots[i][0] != -1; i++) { + off = rsv_slots[i][0]; + ln = rsv_slots[i][1]; + set_bits(off, ln, + edma_cc[j]->edma_inuse); + } + } + } + + /* Clear the xbar mapped channels in unused list */ + xbar_chans = info[j]->xbar_chans; + if (xbar_chans) { + for (i = 0; xbar_chans[i][1] != -1; i++) { + off = xbar_chans[i][1]; + clear_bits(off, 1, + edma_cc[j]->edma_unused); + } + } + + if (node) { + irq[j] = irq_of_parse_and_map(node, 0); + err_irq[j] = irq_of_parse_and_map(node, 2); + } else { + char irq_name[10]; + + sprintf(irq_name, "edma%d", j); + irq[j] = platform_get_irq_byname(pdev, irq_name); + + sprintf(irq_name, "edma%d_err", j); + err_irq[j] = platform_get_irq_byname(pdev, irq_name); + } + edma_cc[j]->irq_res_start = irq[j]; + edma_cc[j]->irq_res_end = err_irq[j]; + + status = devm_request_irq(dev, irq[j], dma_irq_handler, 0, + "edma", dev); + if (status < 0) { + dev_dbg(&pdev->dev, + "devm_request_irq %d failed --> %d\n", + irq[j], status); + return status; + } + + status = devm_request_irq(dev, err_irq[j], dma_ccerr_handler, 0, + "edma_error", dev); + if (status < 0) { + dev_dbg(&pdev->dev, + "devm_request_irq %d failed --> %d\n", + err_irq[j], status); + return status; + } + + for (i = 0; i < edma_cc[j]->num_channels; i++) + map_dmach_queue(j, i, info[j]->default_queue); + + queue_priority_mapping = info[j]->queue_priority_mapping; + + /* Event queue priority mapping */ + for (i = 0; queue_priority_mapping[i][0] != -1; i++) + assign_priority_to_queue(j, + queue_priority_mapping[i][0], + queue_priority_mapping[i][1]); + + /* Map the channel to param entry if channel mapping logic + * exist + */ + if (edma_read(j, EDMA_CCCFG) & CHMAP_EXIST) + map_dmach_param(j); + + for (i = 0; i < edma_cc[j]->num_region; i++) { + edma_write_array2(j, EDMA_DRAE, i, 0, 0x0); + edma_write_array2(j, EDMA_DRAE, i, 1, 0x0); + edma_write_array(j, EDMA_QRAE, i, 0x0); + } + arch_num_cc++; + } + + return 0; +} + +static struct platform_driver edma_driver = { + .driver = { + .name = "edma", + .of_match_table = edma_of_ids, + }, + .probe = edma_probe, +}; + +static int __init edma_init(void) +{ + return platform_driver_probe(&edma_driver, edma_probe); +} +arch_initcall(edma_init); + diff --git a/arch/arm/common/firmware.c b/arch/arm/common/firmware.c new file mode 100644 index 00000000000..27ddccb1131 --- /dev/null +++ b/arch/arm/common/firmware.c @@ -0,0 +1,18 @@ +/* + * Copyright (C) 2012 Samsung Electronics. + * Kyungmin Park <kyungmin.park@samsung.com> + * Tomasz Figa <t.figa@samsung.com> + * + * 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/kernel.h> +#include <linux/suspend.h> + +#include <asm/firmware.h> + +static const struct firmware_ops default_firmware_ops; + +const struct firmware_ops *firmware_ops = &default_firmware_ops; diff --git a/arch/arm/common/it8152.c b/arch/arm/common/it8152.c index 001f4913799..5114b68e99d 100644 --- a/arch/arm/common/it8152.c +++ b/arch/arm/common/it8152.c @@ -257,7 +257,7 @@ static int it8152_needs_bounce(struct device *dev, dma_addr_t dma_addr, size_t s */ static int it8152_pci_platform_notify(struct device *dev) { - if (dev->bus == &pci_bus_type) { + if (dev_is_pci(dev)) { if (dev->dma_mask) *dev->dma_mask = (SZ_64M - 1) | PHYS_OFFSET; dev->coherent_dma_mask = (SZ_64M - 1) | PHYS_OFFSET; @@ -268,7 +268,7 @@ static int it8152_pci_platform_notify(struct device *dev) static int it8152_pci_platform_notify_remove(struct device *dev) { - if (dev->bus == &pci_bus_type) + if (dev_is_pci(dev)) dmabounce_unregister_dev(dev); return 0; diff --git a/arch/arm/common/mcpm_entry.c b/arch/arm/common/mcpm_entry.c new file mode 100644 index 00000000000..f91136ab447 --- /dev/null +++ b/arch/arm/common/mcpm_entry.c @@ -0,0 +1,296 @@ +/* + * arch/arm/common/mcpm_entry.c -- entry point for multi-cluster PM + * + * 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/kernel.h> +#include <linux/init.h> +#include <linux/irqflags.h> + +#include <asm/mcpm.h> +#include <asm/cacheflush.h> +#include <asm/idmap.h> +#include <asm/cputype.h> + +extern unsigned long mcpm_entry_vectors[MAX_NR_CLUSTERS][MAX_CPUS_PER_CLUSTER]; + +void mcpm_set_entry_vector(unsigned cpu, unsigned cluster, void *ptr) +{ + unsigned long val = ptr ? virt_to_phys(ptr) : 0; + mcpm_entry_vectors[cluster][cpu] = val; + sync_cache_w(&mcpm_entry_vectors[cluster][cpu]); +} + +extern unsigned long mcpm_entry_early_pokes[MAX_NR_CLUSTERS][MAX_CPUS_PER_CLUSTER][2]; + +void mcpm_set_early_poke(unsigned cpu, unsigned cluster, + unsigned long poke_phys_addr, unsigned long poke_val) +{ + unsigned long *poke = &mcpm_entry_early_pokes[cluster][cpu][0]; + poke[0] = poke_phys_addr; + poke[1] = poke_val; + __sync_cache_range_w(poke, 2 * sizeof(*poke)); +} + +static const struct mcpm_platform_ops *platform_ops; + +int __init mcpm_platform_register(const struct mcpm_platform_ops *ops) +{ + if (platform_ops) + return -EBUSY; + platform_ops = ops; + return 0; +} + +bool mcpm_is_available(void) +{ + return (platform_ops) ? true : false; +} + +int mcpm_cpu_power_up(unsigned int cpu, unsigned int cluster) +{ + if (!platform_ops) + return -EUNATCH; /* try not to shadow power_up errors */ + might_sleep(); + return platform_ops->power_up(cpu, cluster); +} + +typedef void (*phys_reset_t)(unsigned long); + +void mcpm_cpu_power_down(void) +{ + phys_reset_t phys_reset; + + if (WARN_ON_ONCE(!platform_ops || !platform_ops->power_down)) + return; + BUG_ON(!irqs_disabled()); + + /* + * Do this before calling into the power_down method, + * as it might not always be safe to do afterwards. + */ + setup_mm_for_reboot(); + + platform_ops->power_down(); + + /* + * It is possible for a power_up request to happen concurrently + * with a power_down request for the same CPU. In this case the + * power_down method might not be able to actually enter a + * powered down state with the WFI instruction if the power_up + * method has removed the required reset condition. The + * power_down method is then allowed to return. We must perform + * a re-entry in the kernel as if the power_up method just had + * deasserted reset on the CPU. + * + * To simplify race issues, the platform specific implementation + * must accommodate for the possibility of unordered calls to + * power_down and power_up with a usage count. Therefore, if a + * call to power_up is issued for a CPU that is not down, then + * the next call to power_down must not attempt a full shutdown + * but only do the minimum (normally disabling L1 cache and CPU + * coherency) and return just as if a concurrent power_up request + * had happened as described above. + */ + + phys_reset = (phys_reset_t)(unsigned long)virt_to_phys(cpu_reset); + phys_reset(virt_to_phys(mcpm_entry_point)); + + /* should never get here */ + BUG(); +} + +int mcpm_wait_for_cpu_powerdown(unsigned int cpu, unsigned int cluster) +{ + int ret; + + if (WARN_ON_ONCE(!platform_ops || !platform_ops->wait_for_powerdown)) + return -EUNATCH; + + ret = platform_ops->wait_for_powerdown(cpu, cluster); + if (ret) + pr_warn("%s: cpu %u, cluster %u failed to power down (%d)\n", + __func__, cpu, cluster, ret); + + return ret; +} + +void mcpm_cpu_suspend(u64 expected_residency) +{ + phys_reset_t phys_reset; + + if (WARN_ON_ONCE(!platform_ops || !platform_ops->suspend)) + return; + BUG_ON(!irqs_disabled()); + + /* Very similar to mcpm_cpu_power_down() */ + setup_mm_for_reboot(); + platform_ops->suspend(expected_residency); + phys_reset = (phys_reset_t)(unsigned long)virt_to_phys(cpu_reset); + phys_reset(virt_to_phys(mcpm_entry_point)); + BUG(); +} + +int mcpm_cpu_powered_up(void) +{ + if (!platform_ops) + return -EUNATCH; + if (platform_ops->powered_up) + platform_ops->powered_up(); + return 0; +} + +struct sync_struct mcpm_sync; + +/* + * __mcpm_cpu_going_down: Indicates that the cpu is being torn down. + * This must be called at the point of committing to teardown of a CPU. + * The CPU cache (SCTRL.C bit) is expected to still be active. + */ +void __mcpm_cpu_going_down(unsigned int cpu, unsigned int cluster) +{ + mcpm_sync.clusters[cluster].cpus[cpu].cpu = CPU_GOING_DOWN; + sync_cache_w(&mcpm_sync.clusters[cluster].cpus[cpu].cpu); +} + +/* + * __mcpm_cpu_down: Indicates that cpu teardown is complete and that the + * cluster can be torn down without disrupting this CPU. + * To avoid deadlocks, this must be called before a CPU is powered down. + * The CPU cache (SCTRL.C bit) is expected to be off. + * However L2 cache might or might not be active. + */ +void __mcpm_cpu_down(unsigned int cpu, unsigned int cluster) +{ + dmb(); + mcpm_sync.clusters[cluster].cpus[cpu].cpu = CPU_DOWN; + sync_cache_w(&mcpm_sync.clusters[cluster].cpus[cpu].cpu); + sev(); +} + +/* + * __mcpm_outbound_leave_critical: Leave the cluster teardown critical section. + * @state: the final state of the cluster: + * CLUSTER_UP: no destructive teardown was done and the cluster has been + * restored to the previous state (CPU cache still active); or + * CLUSTER_DOWN: the cluster has been torn-down, ready for power-off + * (CPU cache disabled, L2 cache either enabled or disabled). + */ +void __mcpm_outbound_leave_critical(unsigned int cluster, int state) +{ + dmb(); + mcpm_sync.clusters[cluster].cluster = state; + sync_cache_w(&mcpm_sync.clusters[cluster].cluster); + sev(); +} + +/* + * __mcpm_outbound_enter_critical: Enter the cluster teardown critical section. + * This function should be called by the last man, after local CPU teardown + * is complete. CPU cache expected to be active. + * + * Returns: + * false: the critical section was not entered because an inbound CPU was + * observed, or the cluster is already being set up; + * true: the critical section was entered: it is now safe to tear down the + * cluster. + */ +bool __mcpm_outbound_enter_critical(unsigned int cpu, unsigned int cluster) +{ + unsigned int i; + struct mcpm_sync_struct *c = &mcpm_sync.clusters[cluster]; + + /* Warn inbound CPUs that the cluster is being torn down: */ + c->cluster = CLUSTER_GOING_DOWN; + sync_cache_w(&c->cluster); + + /* Back out if the inbound cluster is already in the critical region: */ + sync_cache_r(&c->inbound); + if (c->inbound == INBOUND_COMING_UP) + goto abort; + + /* + * Wait for all CPUs to get out of the GOING_DOWN state, so that local + * teardown is complete on each CPU before tearing down the cluster. + * + * If any CPU has been woken up again from the DOWN state, then we + * shouldn't be taking the cluster down at all: abort in that case. + */ + sync_cache_r(&c->cpus); + for (i = 0; i < MAX_CPUS_PER_CLUSTER; i++) { + int cpustate; + + if (i == cpu) + continue; + + while (1) { + cpustate = c->cpus[i].cpu; + if (cpustate != CPU_GOING_DOWN) + break; + + wfe(); + sync_cache_r(&c->cpus[i].cpu); + } + + switch (cpustate) { + case CPU_DOWN: + continue; + + default: + goto abort; + } + } + + return true; + +abort: + __mcpm_outbound_leave_critical(cluster, CLUSTER_UP); + return false; +} + +int __mcpm_cluster_state(unsigned int cluster) +{ + sync_cache_r(&mcpm_sync.clusters[cluster].cluster); + return mcpm_sync.clusters[cluster].cluster; +} + +extern unsigned long mcpm_power_up_setup_phys; + +int __init mcpm_sync_init( + void (*power_up_setup)(unsigned int affinity_level)) +{ + unsigned int i, j, mpidr, this_cluster; + + BUILD_BUG_ON(MCPM_SYNC_CLUSTER_SIZE * MAX_NR_CLUSTERS != sizeof mcpm_sync); + BUG_ON((unsigned long)&mcpm_sync & (__CACHE_WRITEBACK_GRANULE - 1)); + + /* + * Set initial CPU and cluster states. + * Only one cluster is assumed to be active at this point. + */ + for (i = 0; i < MAX_NR_CLUSTERS; i++) { + mcpm_sync.clusters[i].cluster = CLUSTER_DOWN; + mcpm_sync.clusters[i].inbound = INBOUND_NOT_COMING_UP; + for (j = 0; j < MAX_CPUS_PER_CLUSTER; j++) + mcpm_sync.clusters[i].cpus[j].cpu = CPU_DOWN; + } + mpidr = read_cpuid_mpidr(); + this_cluster = MPIDR_AFFINITY_LEVEL(mpidr, 1); + for_each_online_cpu(i) + mcpm_sync.clusters[this_cluster].cpus[i].cpu = CPU_UP; + mcpm_sync.clusters[this_cluster].cluster = CLUSTER_UP; + sync_cache_w(&mcpm_sync); + + if (power_up_setup) { + mcpm_power_up_setup_phys = virt_to_phys(power_up_setup); + sync_cache_w(&mcpm_power_up_setup_phys); + } + + return 0; +} diff --git a/arch/arm/common/mcpm_head.S b/arch/arm/common/mcpm_head.S new file mode 100644 index 00000000000..e02db4b81a6 --- /dev/null +++ b/arch/arm/common/mcpm_head.S @@ -0,0 +1,233 @@ +/* + * arch/arm/common/mcpm_head.S -- kernel entry point for multi-cluster PM + * + * 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. + * + * + * Refer to Documentation/arm/cluster-pm-race-avoidance.txt + * for details of the synchronisation algorithms used here. + */ + +#include <linux/linkage.h> +#include <asm/mcpm.h> +#include <asm/assembler.h> + +#include "vlock.h" + +.if MCPM_SYNC_CLUSTER_CPUS +.error "cpus must be the first member of struct mcpm_sync_struct" +.endif + + .macro pr_dbg string +#if defined(CONFIG_DEBUG_LL) && defined(DEBUG) + b 1901f +1902: .asciz "CPU" +1903: .asciz " cluster" +1904: .asciz ": \string" + .align +1901: adr r0, 1902b + bl printascii + mov r0, r9 + bl printhex2 + adr r0, 1903b + bl printascii + mov r0, r10 + bl printhex2 + adr r0, 1904b + bl printascii +#endif + .endm + + .arm + .align + +ENTRY(mcpm_entry_point) + + ARM_BE8(setend be) + THUMB( adr r12, BSYM(1f) ) + THUMB( bx r12 ) + THUMB( .thumb ) +1: + mrc p15, 0, r0, c0, c0, 5 @ MPIDR + ubfx r9, r0, #0, #8 @ r9 = cpu + ubfx r10, r0, #8, #8 @ r10 = cluster + mov r3, #MAX_CPUS_PER_CLUSTER + mla r4, r3, r10, r9 @ r4 = canonical CPU index + cmp r4, #(MAX_CPUS_PER_CLUSTER * MAX_NR_CLUSTERS) + blo 2f + + /* We didn't expect this CPU. Try to cheaply make it quiet. */ +1: wfi + wfe + b 1b + +2: pr_dbg "kernel mcpm_entry_point\n" + + /* + * MMU is off so we need to get to various variables in a + * position independent way. + */ + adr r5, 3f + ldmia r5, {r0, r6, r7, r8, r11} + add r0, r5, r0 @ r0 = mcpm_entry_early_pokes + add r6, r5, r6 @ r6 = mcpm_entry_vectors + ldr r7, [r5, r7] @ r7 = mcpm_power_up_setup_phys + add r8, r5, r8 @ r8 = mcpm_sync + add r11, r5, r11 @ r11 = first_man_locks + + @ Perform an early poke, if any + add r0, r0, r4, lsl #3 + ldmia r0, {r0, r1} + teq r0, #0 + strne r1, [r0] + + mov r0, #MCPM_SYNC_CLUSTER_SIZE + mla r8, r0, r10, r8 @ r8 = sync cluster base + + @ Signal that this CPU is coming UP: + mov r0, #CPU_COMING_UP + mov r5, #MCPM_SYNC_CPU_SIZE + mla r5, r9, r5, r8 @ r5 = sync cpu address + strb r0, [r5] + + @ At this point, the cluster cannot unexpectedly enter the GOING_DOWN + @ state, because there is at least one active CPU (this CPU). + + mov r0, #VLOCK_SIZE + mla r11, r0, r10, r11 @ r11 = cluster first man lock + mov r0, r11 + mov r1, r9 @ cpu + bl vlock_trylock @ implies DMB + + cmp r0, #0 @ failed to get the lock? + bne mcpm_setup_wait @ wait for cluster setup if so + + ldrb r0, [r8, #MCPM_SYNC_CLUSTER_CLUSTER] + cmp r0, #CLUSTER_UP @ cluster already up? + bne mcpm_setup @ if not, set up the cluster + + @ Otherwise, release the first man lock and skip setup: + mov r0, r11 + bl vlock_unlock + b mcpm_setup_complete + +mcpm_setup: + @ Control dependency implies strb not observable before previous ldrb. + + @ Signal that the cluster is being brought up: + mov r0, #INBOUND_COMING_UP + strb r0, [r8, #MCPM_SYNC_CLUSTER_INBOUND] + dmb + + @ Any CPU trying to take the cluster into CLUSTER_GOING_DOWN from this + @ point onwards will observe INBOUND_COMING_UP and abort. + + @ Wait for any previously-pending cluster teardown operations to abort + @ or complete: +mcpm_teardown_wait: + ldrb r0, [r8, #MCPM_SYNC_CLUSTER_CLUSTER] + cmp r0, #CLUSTER_GOING_DOWN + bne first_man_setup + wfe + b mcpm_teardown_wait + +first_man_setup: + dmb + + @ If the outbound gave up before teardown started, skip cluster setup: + + cmp r0, #CLUSTER_UP + beq mcpm_setup_leave + + @ power_up_setup is now responsible for setting up the cluster: + + cmp r7, #0 + mov r0, #1 @ second (cluster) affinity level + blxne r7 @ Call power_up_setup if defined + dmb + + mov r0, #CLUSTER_UP + strb r0, [r8, #MCPM_SYNC_CLUSTER_CLUSTER] + dmb + +mcpm_setup_leave: + @ Leave the cluster setup critical section: + + mov r0, #INBOUND_NOT_COMING_UP + strb r0, [r8, #MCPM_SYNC_CLUSTER_INBOUND] + dsb st + sev + + mov r0, r11 + bl vlock_unlock @ implies DMB + b mcpm_setup_complete + + @ In the contended case, non-first men wait here for cluster setup + @ to complete: +mcpm_setup_wait: + ldrb r0, [r8, #MCPM_SYNC_CLUSTER_CLUSTER] + cmp r0, #CLUSTER_UP + wfene + bne mcpm_setup_wait + dmb + +mcpm_setup_complete: + @ If a platform-specific CPU setup hook is needed, it is + @ called from here. + + cmp r7, #0 + mov r0, #0 @ first (CPU) affinity level + blxne r7 @ Call power_up_setup if defined + dmb + + @ Mark the CPU as up: + + mov r0, #CPU_UP + strb r0, [r5] + + @ Observability order of CPU_UP and opening of the gate does not matter. + +mcpm_entry_gated: + ldr r5, [r6, r4, lsl #2] @ r5 = CPU entry vector + cmp r5, #0 + wfeeq + beq mcpm_entry_gated + dmb + + pr_dbg "released\n" + bx r5 + + .align 2 + +3: .word mcpm_entry_early_pokes - . + .word mcpm_entry_vectors - 3b + .word mcpm_power_up_setup_phys - 3b + .word mcpm_sync - 3b + .word first_man_locks - 3b + +ENDPROC(mcpm_entry_point) + + .bss + + .align CACHE_WRITEBACK_ORDER + .type first_man_locks, #object +first_man_locks: + .space VLOCK_SIZE * MAX_NR_CLUSTERS + .align CACHE_WRITEBACK_ORDER + + .type mcpm_entry_vectors, #object +ENTRY(mcpm_entry_vectors) + .space 4 * MAX_NR_CLUSTERS * MAX_CPUS_PER_CLUSTER + + .type mcpm_entry_early_pokes, #object +ENTRY(mcpm_entry_early_pokes) + .space 8 * MAX_NR_CLUSTERS * MAX_CPUS_PER_CLUSTER + + .type mcpm_power_up_setup_phys, #object +ENTRY(mcpm_power_up_setup_phys) + .space 4 @ set by mcpm_sync_init() diff --git a/arch/arm/common/mcpm_platsmp.c b/arch/arm/common/mcpm_platsmp.c new file mode 100644 index 00000000000..92e54d7c6f4 --- /dev/null +++ b/arch/arm/common/mcpm_platsmp.c @@ -0,0 +1,103 @@ +/* + * linux/arch/arm/mach-vexpress/mcpm_platsmp.c + * + * Created by: Nicolas Pitre, November 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. + * + * Code to handle secondary CPU bringup and hotplug for the cluster power API. + */ + +#include <linux/init.h> +#include <linux/smp.h> +#include <linux/spinlock.h> + +#include <asm/mcpm.h> +#include <asm/smp.h> +#include <asm/smp_plat.h> + +static void cpu_to_pcpu(unsigned int cpu, + unsigned int *pcpu, unsigned int *pcluster) +{ + unsigned int mpidr; + + mpidr = cpu_logical_map(cpu); + *pcpu = MPIDR_AFFINITY_LEVEL(mpidr, 0); + *pcluster = MPIDR_AFFINITY_LEVEL(mpidr, 1); +} + +static int mcpm_boot_secondary(unsigned int cpu, struct task_struct *idle) +{ + unsigned int pcpu, pcluster, ret; + extern void secondary_startup(void); + + cpu_to_pcpu(cpu, &pcpu, &pcluster); + + pr_debug("%s: logical CPU %d is physical CPU %d cluster %d\n", + __func__, cpu, pcpu, pcluster); + + mcpm_set_entry_vector(pcpu, pcluster, NULL); + ret = mcpm_cpu_power_up(pcpu, pcluster); + if (ret) + return ret; + mcpm_set_entry_vector(pcpu, pcluster, secondary_startup); + arch_send_wakeup_ipi_mask(cpumask_of(cpu)); + dsb_sev(); + return 0; +} + +static void mcpm_secondary_init(unsigned int cpu) +{ + mcpm_cpu_powered_up(); +} + +#ifdef CONFIG_HOTPLUG_CPU + +static int mcpm_cpu_kill(unsigned int cpu) +{ + unsigned int pcpu, pcluster; + + cpu_to_pcpu(cpu, &pcpu, &pcluster); + + return !mcpm_wait_for_cpu_powerdown(pcpu, pcluster); +} + +static int mcpm_cpu_disable(unsigned int cpu) +{ + /* + * We assume all CPUs may be shut down. + * This would be the hook to use for eventual Secure + * OS migration requests as described in the PSCI spec. + */ + return 0; +} + +static void mcpm_cpu_die(unsigned int cpu) +{ + unsigned int mpidr, pcpu, pcluster; + mpidr = read_cpuid_mpidr(); + pcpu = MPIDR_AFFINITY_LEVEL(mpidr, 0); + pcluster = MPIDR_AFFINITY_LEVEL(mpidr, 1); + mcpm_set_entry_vector(pcpu, pcluster, NULL); + mcpm_cpu_power_down(); +} + +#endif + +static struct smp_operations __initdata mcpm_smp_ops = { + .smp_boot_secondary = mcpm_boot_secondary, + .smp_secondary_init = mcpm_secondary_init, +#ifdef CONFIG_HOTPLUG_CPU + .cpu_kill = mcpm_cpu_kill, + .cpu_disable = mcpm_cpu_disable, + .cpu_die = mcpm_cpu_die, +#endif +}; + +void __init mcpm_smp_set_ops(void) +{ + smp_set_ops(&mcpm_smp_ops); +} diff --git a/arch/arm/common/scoop.c b/arch/arm/common/scoop.c index a5c3dc38aa1..a20fa80776d 100644 --- a/arch/arm/common/scoop.c +++ b/arch/arm/common/scoop.c @@ -182,7 +182,6 @@ static int scoop_probe(struct platform_device *pdev) struct scoop_config *inf; struct resource *mem = platform_get_resource(pdev, IORESOURCE_MEM, 0); int ret; - int temp; if (!mem) return -EINVAL; @@ -232,8 +231,6 @@ static int scoop_probe(struct platform_device *pdev) return 0; - if (devptr->gpio.base != -1) - temp = gpiochip_remove(&devptr->gpio); err_gpio: platform_set_drvdata(pdev, NULL); err_ioremap: diff --git a/arch/arm/common/sharpsl_param.c b/arch/arm/common/sharpsl_param.c index d56c932580e..025f6ce3859 100644 --- a/arch/arm/common/sharpsl_param.c +++ b/arch/arm/common/sharpsl_param.c @@ -15,6 +15,7 @@ #include <linux/module.h> #include <linux/string.h> #include <asm/mach/sharpsl_param.h> +#include <asm/memory.h> /* * Certain hardware parameters determined at the time of device manufacture, @@ -25,8 +26,10 @@ */ #ifdef CONFIG_ARCH_SA1100 #define PARAM_BASE 0xe8ffc000 +#define param_start(x) (void *)(x) #else #define PARAM_BASE 0xa0000a00 +#define param_start(x) __va(x) #endif #define MAGIC_CHG(a,b,c,d) ( ( d << 24 ) | ( c << 16 ) | ( b << 8 ) | a ) @@ -41,7 +44,7 @@ EXPORT_SYMBOL(sharpsl_param); void sharpsl_save_param(void) { - memcpy(&sharpsl_param, (void *)PARAM_BASE, sizeof(struct sharpsl_param_info)); + memcpy(&sharpsl_param, param_start(PARAM_BASE), sizeof(struct sharpsl_param_info)); if (sharpsl_param.comadj_keyword != COMADJ_MAGIC) sharpsl_param.comadj=-1; diff --git a/arch/arm/common/timer-sp.c b/arch/arm/common/timer-sp.c index 9d2d3ba339f..fd6bff0c5b9 100644 --- a/arch/arm/common/timer-sp.c +++ b/arch/arm/common/timer-sp.c @@ -25,33 +25,29 @@ #include <linux/interrupt.h> #include <linux/irq.h> #include <linux/io.h> +#include <linux/of.h> +#include <linux/of_address.h> +#include <linux/of_irq.h> +#include <linux/sched_clock.h> -#include <asm/sched_clock.h> #include <asm/hardware/arm_timer.h> +#include <asm/hardware/timer-sp.h> -static long __init sp804_get_clock_rate(const char *name) +static long __init sp804_get_clock_rate(struct clk *clk) { - struct clk *clk; long rate; int err; - clk = clk_get_sys("sp804", name); - if (IS_ERR(clk)) { - pr_err("sp804: %s clock not found: %d\n", name, - (int)PTR_ERR(clk)); - return PTR_ERR(clk); - } - err = clk_prepare(clk); if (err) { - pr_err("sp804: %s clock failed to prepare: %d\n", name, err); + pr_err("sp804: clock failed to prepare: %d\n", err); clk_put(clk); return err; } err = clk_enable(clk); if (err) { - pr_err("sp804: %s clock failed to enable: %d\n", name, err); + pr_err("sp804: clock failed to enable: %d\n", err); clk_unprepare(clk); clk_put(clk); return err; @@ -59,7 +55,7 @@ static long __init sp804_get_clock_rate(const char *name) rate = clk_get_rate(clk); if (rate < 0) { - pr_err("sp804: %s clock failed to get rate: %ld\n", name, rate); + pr_err("sp804: clock failed to get rate: %ld\n", rate); clk_disable(clk); clk_unprepare(clk); clk_put(clk); @@ -70,16 +66,28 @@ static long __init sp804_get_clock_rate(const char *name) static void __iomem *sched_clock_base; -static u32 sp804_read(void) +static u64 notrace sp804_read(void) { return ~readl_relaxed(sched_clock_base + TIMER_VALUE); } void __init __sp804_clocksource_and_sched_clock_init(void __iomem *base, const char *name, + struct clk *clk, int use_sched_clock) { - long rate = sp804_get_clock_rate(name); + long rate; + + if (!clk) { + clk = clk_get_sys("sp804", name); + if (IS_ERR(clk)) { + pr_err("sp804: clock not found: %d\n", + (int)PTR_ERR(clk)); + return; + } + } + + rate = sp804_get_clock_rate(clk); if (rate < 0) return; @@ -96,7 +104,7 @@ void __init __sp804_clocksource_and_sched_clock_init(void __iomem *base, if (use_sched_clock) { sched_clock_base = base; - setup_sched_clock(sp804_read, 32, rate); + sched_clock_register(sp804_read, 32, rate); } } @@ -158,7 +166,8 @@ static int sp804_set_next_event(unsigned long next, } static struct clock_event_device sp804_clockevent = { - .features = CLOCK_EVT_FEAT_PERIODIC | CLOCK_EVT_FEAT_ONESHOT, + .features = CLOCK_EVT_FEAT_PERIODIC | CLOCK_EVT_FEAT_ONESHOT | + CLOCK_EVT_FEAT_DYNIRQ, .set_mode = sp804_set_mode, .set_next_event = sp804_set_next_event, .rating = 300, @@ -166,17 +175,25 @@ static struct clock_event_device sp804_clockevent = { static struct irqaction sp804_timer_irq = { .name = "timer", - .flags = IRQF_DISABLED | IRQF_TIMER | IRQF_IRQPOLL, + .flags = IRQF_TIMER | IRQF_IRQPOLL, .handler = sp804_timer_interrupt, .dev_id = &sp804_clockevent, }; -void __init sp804_clockevents_init(void __iomem *base, unsigned int irq, - const char *name) +void __init __sp804_clockevents_init(void __iomem *base, unsigned int irq, struct clk *clk, const char *name) { struct clock_event_device *evt = &sp804_clockevent; - long rate = sp804_get_clock_rate(name); + long rate; + if (!clk) + clk = clk_get_sys("sp804", name); + if (IS_ERR(clk)) { + pr_err("sp804: %s clock not found: %d\n", name, + (int)PTR_ERR(clk)); + return; + } + + rate = sp804_get_clock_rate(clk); if (rate < 0) return; @@ -186,6 +203,102 @@ void __init sp804_clockevents_init(void __iomem *base, unsigned int irq, evt->irq = irq; evt->cpumask = cpu_possible_mask; + writel(0, base + TIMER_CTRL); + setup_irq(irq, &sp804_timer_irq); clockevents_config_and_register(evt, rate, 0xf, 0xffffffff); } + +static void __init sp804_of_init(struct device_node *np) +{ + static bool initialized = false; + void __iomem *base; + int irq; + u32 irq_num = 0; + struct clk *clk1, *clk2; + const char *name = of_get_property(np, "compatible", NULL); + + base = of_iomap(np, 0); + if (WARN_ON(!base)) + return; + + /* Ensure timers are disabled */ + writel(0, base + TIMER_CTRL); + writel(0, base + TIMER_2_BASE + TIMER_CTRL); + + if (initialized || !of_device_is_available(np)) + goto err; + + clk1 = of_clk_get(np, 0); + if (IS_ERR(clk1)) + clk1 = NULL; + + /* Get the 2nd clock if the timer has 2 timer clocks */ + if (of_count_phandle_with_args(np, "clocks", "#clock-cells") == 3) { + clk2 = of_clk_get(np, 1); + if (IS_ERR(clk2)) { + pr_err("sp804: %s clock not found: %d\n", np->name, + (int)PTR_ERR(clk2)); + goto err; + } + } else + clk2 = clk1; + + irq = irq_of_parse_and_map(np, 0); + if (irq <= 0) + goto err; + + of_property_read_u32(np, "arm,sp804-has-irq", &irq_num); + if (irq_num == 2) { + __sp804_clockevents_init(base + TIMER_2_BASE, irq, clk2, name); + __sp804_clocksource_and_sched_clock_init(base, name, clk1, 1); + } else { + __sp804_clockevents_init(base, irq, clk1 , name); + __sp804_clocksource_and_sched_clock_init(base + TIMER_2_BASE, + name, clk2, 1); + } + initialized = true; + + return; +err: + iounmap(base); +} +CLOCKSOURCE_OF_DECLARE(sp804, "arm,sp804", sp804_of_init); + +static void __init integrator_cp_of_init(struct device_node *np) +{ + static int init_count = 0; + void __iomem *base; + int irq; + const char *name = of_get_property(np, "compatible", NULL); + struct clk *clk; + + base = of_iomap(np, 0); + if (WARN_ON(!base)) + return; + clk = of_clk_get(np, 0); + if (WARN_ON(IS_ERR(clk))) + return; + + /* Ensure timer is disabled */ + writel(0, base + TIMER_CTRL); + + if (init_count == 2 || !of_device_is_available(np)) + goto err; + + if (!init_count) + __sp804_clocksource_and_sched_clock_init(base, name, clk, 0); + else { + irq = irq_of_parse_and_map(np, 0); + if (irq <= 0) + goto err; + + __sp804_clockevents_init(base, irq, clk, name); + } + + init_count++; + return; +err: + iounmap(base); +} +CLOCKSOURCE_OF_DECLARE(intcp, "arm,integrator-cp-timer", integrator_cp_of_init); diff --git a/arch/arm/common/via82c505.c b/arch/arm/common/via82c505.c deleted file mode 100644 index 6cb362e56d2..00000000000 --- a/arch/arm/common/via82c505.c +++ /dev/null @@ -1,83 +0,0 @@ -#include <linux/kernel.h> -#include <linux/pci.h> -#include <linux/interrupt.h> -#include <linux/mm.h> -#include <linux/init.h> -#include <linux/ioport.h> -#include <linux/io.h> - - -#include <asm/mach/pci.h> - -#define MAX_SLOTS 7 - -#define CONFIG_CMD(bus, devfn, where) (0x80000000 | (bus->number << 16) | (devfn << 8) | (where & ~3)) - -static int -via82c505_read_config(struct pci_bus *bus, unsigned int devfn, int where, - int size, u32 *value) -{ - outl(CONFIG_CMD(bus,devfn,where),0xCF8); - switch (size) { - case 1: - *value=inb(0xCFC + (where&3)); - break; - case 2: - *value=inw(0xCFC + (where&2)); - break; - case 4: - *value=inl(0xCFC); - break; - } - return PCIBIOS_SUCCESSFUL; -} - -static int -via82c505_write_config(struct pci_bus *bus, unsigned int devfn, int where, - int size, u32 value) -{ - outl(CONFIG_CMD(bus,devfn,where),0xCF8); - switch (size) { - case 1: - outb(value, 0xCFC + (where&3)); - break; - case 2: - outw(value, 0xCFC + (where&2)); - break; - case 4: - outl(value, 0xCFC); - break; - } - return PCIBIOS_SUCCESSFUL; -} - -struct pci_ops via82c505_ops = { - .read = via82c505_read_config, - .write = via82c505_write_config, -}; - -void __init via82c505_preinit(void) -{ - printk(KERN_DEBUG "PCI: VIA 82c505\n"); - if (!request_region(0xA8,2,"via config")) { - printk(KERN_WARNING"VIA 82c505: Unable to request region 0xA8\n"); - return; - } - if (!request_region(0xCF8,8,"pci config")) { - printk(KERN_WARNING"VIA 82c505: Unable to request region 0xCF8\n"); - release_region(0xA8, 2); - return; - } - - /* Enable compatible Mode */ - outb(0x96,0xA8); - outb(0x18,0xA9); - outb(0x93,0xA8); - outb(0xd0,0xA9); - -} - -int __init via82c505_setup(int nr, struct pci_sys_data *sys) -{ - return (nr == 0); -} diff --git a/arch/arm/common/vlock.S b/arch/arm/common/vlock.S new file mode 100644 index 00000000000..8b7df283fed --- /dev/null +++ b/arch/arm/common/vlock.S @@ -0,0 +1,108 @@ +/* + * vlock.S - simple voting lock implementation for ARM + * + * Created by: Dave Martin, 2012-08-16 + * 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. + * + * This program 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. + * + * + * This algorithm is described in more detail in + * Documentation/arm/vlocks.txt. + */ + +#include <linux/linkage.h> +#include "vlock.h" + +/* Select different code if voting flags can fit in a single word. */ +#if VLOCK_VOTING_SIZE > 4 +#define FEW(x...) +#define MANY(x...) x +#else +#define FEW(x...) x +#define MANY(x...) +#endif + +@ voting lock for first-man coordination + +.macro voting_begin rbase:req, rcpu:req, rscratch:req + mov \rscratch, #1 + strb \rscratch, [\rbase, \rcpu] + dmb +.endm + +.macro voting_end rbase:req, rcpu:req, rscratch:req + dmb + mov \rscratch, #0 + strb \rscratch, [\rbase, \rcpu] + dsb st + sev +.endm + +/* + * The vlock structure must reside in Strongly-Ordered or Device memory. + * This implementation deliberately eliminates most of the barriers which + * would be required for other memory types, and assumes that independent + * writes to neighbouring locations within a cacheline do not interfere + * with one another. + */ + +@ r0: lock structure base +@ r1: CPU ID (0-based index within cluster) +ENTRY(vlock_trylock) + add r1, r1, #VLOCK_VOTING_OFFSET + + voting_begin r0, r1, r2 + + ldrb r2, [r0, #VLOCK_OWNER_OFFSET] @ check whether lock is held + cmp r2, #VLOCK_OWNER_NONE + bne trylock_fail @ fail if so + + @ Control dependency implies strb not observable before previous ldrb. + + strb r1, [r0, #VLOCK_OWNER_OFFSET] @ submit my vote + + voting_end r0, r1, r2 @ implies DMB + + @ Wait for the current round of voting to finish: + + MANY( mov r3, #VLOCK_VOTING_OFFSET ) +0: + MANY( ldr r2, [r0, r3] ) + FEW( ldr r2, [r0, #VLOCK_VOTING_OFFSET] ) + cmp r2, #0 + wfene + bne 0b + MANY( add r3, r3, #4 ) + MANY( cmp r3, #VLOCK_VOTING_OFFSET + VLOCK_VOTING_SIZE ) + MANY( bne 0b ) + + @ Check who won: + + dmb + ldrb r2, [r0, #VLOCK_OWNER_OFFSET] + eor r0, r1, r2 @ zero if I won, else nonzero + bx lr + +trylock_fail: + voting_end r0, r1, r2 + mov r0, #1 @ nonzero indicates that I lost + bx lr +ENDPROC(vlock_trylock) + +@ r0: lock structure base +ENTRY(vlock_unlock) + dmb + mov r1, #VLOCK_OWNER_NONE + strb r1, [r0, #VLOCK_OWNER_OFFSET] + dsb st + sev + bx lr +ENDPROC(vlock_unlock) diff --git a/arch/arm/common/vlock.h b/arch/arm/common/vlock.h new file mode 100644 index 00000000000..3b441475a59 --- /dev/null +++ b/arch/arm/common/vlock.h @@ -0,0 +1,29 @@ +/* + * vlock.h - simple voting lock implementation + * + * Created by: Dave Martin, 2012-08-16 + * 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. + * + * This program 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. + */ + +#ifndef __VLOCK_H +#define __VLOCK_H + +#include <asm/mcpm.h> + +/* Offsets and sizes are rounded to a word (4 bytes) */ +#define VLOCK_OWNER_OFFSET 0 +#define VLOCK_VOTING_OFFSET 4 +#define VLOCK_VOTING_SIZE ((MAX_CPUS_PER_CLUSTER + 3) / 4 * 4) +#define VLOCK_SIZE (VLOCK_VOTING_OFFSET + VLOCK_VOTING_SIZE) +#define VLOCK_OWNER_NONE 0 + +#endif /* ! __VLOCK_H */ |