redonkable/alistair23-linux
redonkable/alistair23-linux
1
0
Fork 0
Code Releases Activity
alistair23-linux/arch/blackfin/mach-common/smp.c
Yi Li 73a400646b Blackfin: SMP: rewrite IPI handling to avoid memory allocation 
Currently, sending an interprocessor interrupt (IPI) requires building up
a message dynamically which means memory allocation.  But often times, we
will want to send an IPI in low level contexts where allocation is not
possible which may lead to a panic().  So create a per-cpu static array
for the message queue and use that instead.

Further, while we have two supplemental interrupts, we are currently only
using one of them.  So use the second one for the most common IPI message
of all -- smp_send_reschedule().  This avoids ugly contention for locks
which in turn would require an IPI message ...

In general, this improves SMP performance, and in some cases allows the
SMP port to work in places it wouldn't before.  Such as the PREEMPT_RT
state where the slab is protected by a per-cpu spin lock.  If the slab
kmalloc/kfree were to put the task to sleep, and that task was actually
the IPI handler, then the system falls down yet again.

After running some various stress tests on the system, the static limit
of 5 messages seems to work.  On the off chance even this overflows, we
simply panic(), and we can review that scenario to see if the limit needs
to be increased a bit more.

Signed-off-by: Yi Li <yi.li@analog.com>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
2011-01-10 07:18:15 -05:00

487 lines
11 KiB
C
Raw Blame History

/*
* IPI management based on arch/arm/kernel/smp.c (Copyright 2002 ARM Limited)
*
* Copyright 2007-2009 Analog Devices Inc.
* Philippe Gerum <rpm@xenomai.org>
*
* Licensed under the GPL-2.
*/
#include <linux/module.h>
#include <linux/delay.h>
#include <linux/init.h>
#include <linux/spinlock.h>
#include <linux/sched.h>
#include <linux/interrupt.h>
#include <linux/cache.h>
#include <linux/profile.h>
#include <linux/errno.h>
#include <linux/mm.h>
#include <linux/cpu.h>
#include <linux/smp.h>
#include <linux/cpumask.h>
#include <linux/seq_file.h>
#include <linux/irq.h>
#include <linux/slab.h>
#include <asm/atomic.h>
#include <asm/cacheflush.h>
#include <asm/mmu_context.h>
#include <asm/pgtable.h>
#include <asm/pgalloc.h>
#include <asm/processor.h>
#include <asm/ptrace.h>
#include <asm/cpu.h>
#include <asm/time.h>
#include <linux/err.h>
/*
* Anomaly notes:
* 05000120 - we always define corelock as 32-bit integer in L2
*/
struct corelock_slot corelock __attribute__ ((__section__(".l2.bss")));
void __cpuinitdata *init_retx_coreb, *init_saved_retx_coreb,
*init_saved_seqstat_coreb, *init_saved_icplb_fault_addr_coreb,
*init_saved_dcplb_fault_addr_coreb;
#define BFIN_IPI_RESCHEDULE 0
#define BFIN_IPI_CALL_FUNC 1
#define BFIN_IPI_CPU_STOP 2
struct blackfin_flush_data {
unsigned long start;
unsigned long end;
};
void *secondary_stack;
struct smp_call_struct {
void (*func)(void *info);
void *info;
int wait;
cpumask_t *waitmask;
};
static struct blackfin_flush_data smp_flush_data;
static DEFINE_SPINLOCK(stop_lock);
struct ipi_message {
unsigned long type;
struct smp_call_struct call_struct;
};
/* A magic number - stress test shows this is safe for common cases */
#define BFIN_IPI_MSGQ_LEN 5
/* Simple FIFO buffer, overflow leads to panic */
struct ipi_message_queue {
spinlock_t lock;
unsigned long count;
unsigned long head; /* head of the queue */
struct ipi_message ipi_message[BFIN_IPI_MSGQ_LEN];
};
static DEFINE_PER_CPU(struct ipi_message_queue, ipi_msg_queue);
static void ipi_cpu_stop(unsigned int cpu)
{
spin_lock(&stop_lock);
printk(KERN_CRIT "CPU%u: stopping\n", cpu);
dump_stack();
spin_unlock(&stop_lock);
cpu_clear(cpu, cpu_online_map);
local_irq_disable();
while (1)
SSYNC();
}
static void ipi_flush_icache(void *info)
{
struct blackfin_flush_data *fdata = info;
/* Invalidate the memory holding the bounds of the flushed region. */
blackfin_dcache_invalidate_range((unsigned long)fdata,
(unsigned long)fdata + sizeof(*fdata));
blackfin_icache_flush_range(fdata->start, fdata->end);
}
static void ipi_call_function(unsigned int cpu, struct ipi_message *msg)
{
int wait;
void (*func)(void *info);
void *info;
func = msg->call_struct.func;
info = msg->call_struct.info;
wait = msg->call_struct.wait;
func(info);
if (wait) {
#ifdef __ARCH_SYNC_CORE_DCACHE
/*
* 'wait' usually means synchronization between CPUs.
* Invalidate D cache in case shared data was changed
* by func() to ensure cache coherence.
*/
resync_core_dcache();
#endif
cpu_clear(cpu, *msg->call_struct.waitmask);
}
}
/* Use IRQ_SUPPLE_0 to request reschedule.
* When returning from interrupt to user space,
* there is chance to reschedule */
static irqreturn_t ipi_handler_int0(int irq, void *dev_instance)
{
unsigned int cpu = smp_processor_id();
platform_clear_ipi(cpu, IRQ_SUPPLE_0);
return IRQ_HANDLED;
}
static irqreturn_t ipi_handler_int1(int irq, void *dev_instance)
{
struct ipi_message *msg;
struct ipi_message_queue *msg_queue;
unsigned int cpu = smp_processor_id();
unsigned long flags;
platform_clear_ipi(cpu, IRQ_SUPPLE_1);
msg_queue = &__get_cpu_var(ipi_msg_queue);
spin_lock_irqsave(&msg_queue->lock, flags);
while (msg_queue->count) {
msg = &msg_queue->ipi_message[msg_queue->head];
switch (msg->type) {
case BFIN_IPI_CALL_FUNC:
spin_unlock_irqrestore(&msg_queue->lock, flags);
ipi_call_function(cpu, msg);
spin_lock_irqsave(&msg_queue->lock, flags);
break;
case BFIN_IPI_CPU_STOP:
spin_unlock_irqrestore(&msg_queue->lock, flags);
ipi_cpu_stop(cpu);
spin_lock_irqsave(&msg_queue->lock, flags);
break;
default:
printk(KERN_CRIT "CPU%u: Unknown IPI message 0x%lx\n",
cpu, msg->type);
break;
}
msg_queue->head++;
msg_queue->head %= BFIN_IPI_MSGQ_LEN;
msg_queue->count--;
}
spin_unlock_irqrestore(&msg_queue->lock, flags);
return IRQ_HANDLED;
}
static void ipi_queue_init(void)
{
unsigned int cpu;
struct ipi_message_queue *msg_queue;
for_each_possible_cpu(cpu) {
msg_queue = &per_cpu(ipi_msg_queue, cpu);
spin_lock_init(&msg_queue->lock);
msg_queue->count = 0;
msg_queue->head = 0;
}
}
static inline void smp_send_message(cpumask_t callmap, unsigned long type,
void (*func) (void *info), void *info, int wait)
{
unsigned int cpu;
struct ipi_message_queue *msg_queue;
struct ipi_message *msg;
unsigned long flags, next_msg;
cpumask_t waitmask = callmap; /* waitmask is shared by all cpus */
for_each_cpu_mask(cpu, callmap) {
msg_queue = &per_cpu(ipi_msg_queue, cpu);
spin_lock_irqsave(&msg_queue->lock, flags);
if (msg_queue->count < BFIN_IPI_MSGQ_LEN) {
next_msg = (msg_queue->head + msg_queue->count)
% BFIN_IPI_MSGQ_LEN;
msg = &msg_queue->ipi_message[next_msg];
msg->type = type;
if (type == BFIN_IPI_CALL_FUNC) {
msg->call_struct.func = func;
msg->call_struct.info = info;
msg->call_struct.wait = wait;
msg->call_struct.waitmask = &waitmask;
}
msg_queue->count++;
} else
panic("IPI message queue overflow\n");
spin_unlock_irqrestore(&msg_queue->lock, flags);
platform_send_ipi_cpu(cpu, IRQ_SUPPLE_1);
}
if (wait) {
while (!cpus_empty(waitmask))
blackfin_dcache_invalidate_range(
(unsigned long)(&waitmask),
(unsigned long)(&waitmask));
#ifdef __ARCH_SYNC_CORE_DCACHE
/*
* Invalidate D cache in case shared data was changed by
* other processors to ensure cache coherence.
*/
resync_core_dcache();
#endif
}
}
int smp_call_function(void (*func)(void *info), void *info, int wait)
{
cpumask_t callmap;
callmap = cpu_online_map;
cpu_clear(smp_processor_id(), callmap);
if (cpus_empty(callmap))
return 0;
smp_send_message(callmap, BFIN_IPI_CALL_FUNC, func, info, wait);
return 0;
}
EXPORT_SYMBOL_GPL(smp_call_function);
int smp_call_function_single(int cpuid, void (*func) (void *info), void *info,
int wait)
{
unsigned int cpu = cpuid;
cpumask_t callmap;
if (cpu_is_offline(cpu))
return 0;
cpus_clear(callmap);
cpu_set(cpu, callmap);
smp_send_message(callmap, BFIN_IPI_CALL_FUNC, func, info, wait);
return 0;
}
EXPORT_SYMBOL_GPL(smp_call_function_single);
void smp_send_reschedule(int cpu)
{
/* simply trigger an ipi */
if (cpu_is_offline(cpu))
return;
platform_send_ipi_cpu(cpu, IRQ_SUPPLE_0);
return;
}
void smp_send_stop(void)
{
cpumask_t callmap;
callmap = cpu_online_map;
cpu_clear(smp_processor_id(), callmap);
if (cpus_empty(callmap))
return;
smp_send_message(callmap, BFIN_IPI_CPU_STOP, NULL, NULL, 0);
return;
}
int __cpuinit __cpu_up(unsigned int cpu)
{
int ret;
static struct task_struct *idle;
if (idle)
free_task(idle);
idle = fork_idle(cpu);
if (IS_ERR(idle)) {
printk(KERN_ERR "CPU%u: fork() failed\n", cpu);
return PTR_ERR(idle);
}
secondary_stack = task_stack_page(idle) + THREAD_SIZE;
ret = platform_boot_secondary(cpu, idle);
secondary_stack = NULL;
return ret;
}
static void __cpuinit setup_secondary(unsigned int cpu)
{
unsigned long ilat;
bfin_write_IMASK(0);
CSYNC();
ilat = bfin_read_ILAT();
CSYNC();
bfin_write_ILAT(ilat);
CSYNC();
/* Enable interrupt levels IVG7-15. IARs have been already
* programmed by the boot CPU. */
bfin_irq_flags |= IMASK_IVG15 |
IMASK_IVG14 | IMASK_IVG13 | IMASK_IVG12 | IMASK_IVG11 |
IMASK_IVG10 | IMASK_IVG9 | IMASK_IVG8 | IMASK_IVG7 | IMASK_IVGHW;
}
void __cpuinit secondary_start_kernel(void)
{
unsigned int cpu = smp_processor_id();
struct mm_struct *mm = &init_mm;
if (_bfin_swrst & SWRST_DBL_FAULT_B) {
printk(KERN_EMERG "CoreB Recovering from DOUBLE FAULT event\n");
#ifdef CONFIG_DEBUG_DOUBLEFAULT
printk(KERN_EMERG " While handling exception (EXCAUSE = 0x%x) at %pF\n",
(int)init_saved_seqstat_coreb & SEQSTAT_EXCAUSE, init_saved_retx_coreb);
printk(KERN_NOTICE " DCPLB_FAULT_ADDR: %pF\n", init_saved_dcplb_fault_addr_coreb);
printk(KERN_NOTICE " ICPLB_FAULT_ADDR: %pF\n", init_saved_icplb_fault_addr_coreb);
#endif
printk(KERN_NOTICE " The instruction at %pF caused a double exception\n",
init_retx_coreb);
}
/*
* We want the D-cache to be enabled early, in case the atomic
* support code emulates cache coherence (see
* __ARCH_SYNC_CORE_DCACHE).
*/
init_exception_vectors();
bfin_setup_caches(cpu);
local_irq_disable();
/* Attach the new idle task to the global mm. */
atomic_inc(&mm->mm_users);
atomic_inc(&mm->mm_count);
current->active_mm = mm;
preempt_disable();
setup_secondary(cpu);
platform_secondary_init(cpu);
/* setup local core timer */
bfin_local_timer_setup();
local_irq_enable();
/*
* Calibrate loops per jiffy value.
* IRQs need to be enabled here - D-cache can be invalidated
* in timer irq handler, so core B can read correct jiffies.
*/
calibrate_delay();
cpu_idle();
}
void __init smp_prepare_boot_cpu(void)
{
}
void __init smp_prepare_cpus(unsigned int max_cpus)
{
platform_prepare_cpus(max_cpus);
ipi_queue_init();
platform_request_ipi(IRQ_SUPPLE_0, ipi_handler_int0);
platform_request_ipi(IRQ_SUPPLE_1, ipi_handler_int1);
}
void __init smp_cpus_done(unsigned int max_cpus)
{
unsigned long bogosum = 0;
unsigned int cpu;
for_each_online_cpu(cpu)
bogosum += loops_per_jiffy;
printk(KERN_INFO "SMP: Total of %d processors activated "
"(%lu.%02lu BogoMIPS).\n",
num_online_cpus(),
bogosum / (500000/HZ),
(bogosum / (5000/HZ)) % 100);
}
void smp_icache_flush_range_others(unsigned long start, unsigned long end)
{
smp_flush_data.start = start;
smp_flush_data.end = end;
if (smp_call_function(&ipi_flush_icache, &smp_flush_data, 0))
printk(KERN_WARNING "SMP: failed to run I-cache flush request on other CPUs\n");
}
EXPORT_SYMBOL_GPL(smp_icache_flush_range_others);
#ifdef __ARCH_SYNC_CORE_ICACHE
unsigned long icache_invld_count[NR_CPUS];
void resync_core_icache(void)
{
unsigned int cpu = get_cpu();
blackfin_invalidate_entire_icache();
icache_invld_count[cpu]++;
put_cpu();
}
EXPORT_SYMBOL(resync_core_icache);
#endif
#ifdef __ARCH_SYNC_CORE_DCACHE
unsigned long dcache_invld_count[NR_CPUS];
unsigned long barrier_mask __attribute__ ((__section__(".l2.bss")));
void resync_core_dcache(void)
{
unsigned int cpu = get_cpu();
blackfin_invalidate_entire_dcache();
dcache_invld_count[cpu]++;
put_cpu();
}
EXPORT_SYMBOL(resync_core_dcache);
#endif
#ifdef CONFIG_HOTPLUG_CPU
int __cpuexit __cpu_disable(void)
{
unsigned int cpu = smp_processor_id();
if (cpu == 0)
return -EPERM;
set_cpu_online(cpu, false);
return 0;
}
static DECLARE_COMPLETION(cpu_killed);
int __cpuexit __cpu_die(unsigned int cpu)
{
return wait_for_completion_timeout(&cpu_killed, 5000);
}
void cpu_die(void)
{
complete(&cpu_killed);
atomic_dec(&init_mm.mm_users);
atomic_dec(&init_mm.mm_count);
local_irq_disable();
platform_cpu_die();
}
#endif
View git blame Copy permalink