2007-07-17 21:37:06 -04:00
|
|
|
/*
|
|
|
|
* Xen SMP support
|
|
|
|
*
|
|
|
|
* This file implements the Xen versions of smp_ops. SMP under Xen is
|
|
|
|
* very straightforward. Bringing a CPU up is simply a matter of
|
|
|
|
* loading its initial context and setting it running.
|
|
|
|
*
|
|
|
|
* IPIs are handled through the Xen event mechanism.
|
|
|
|
*
|
|
|
|
* Because virtual CPUs can be scheduled onto any real CPU, there's no
|
|
|
|
* useful topology information for the kernel to make use of. As a
|
|
|
|
* result, all CPUs are treated as if they're single-core and
|
|
|
|
* single-threaded.
|
|
|
|
*
|
|
|
|
* This does not handle HOTPLUG_CPU yet.
|
|
|
|
*/
|
|
|
|
#include <linux/sched.h>
|
xen: implement Xen-specific spinlocks
The standard ticket spinlocks are very expensive in a virtual
environment, because their performance depends on Xen's scheduler
giving vcpus time in the order that they're supposed to take the
spinlock.
This implements a Xen-specific spinlock, which should be much more
efficient.
The fast-path is essentially the old Linux-x86 locks, using a single
lock byte. The locker decrements the byte; if the result is 0, then
they have the lock. If the lock is negative, then locker must spin
until the lock is positive again.
When there's contention, the locker spin for 2^16[*] iterations waiting
to get the lock. If it fails to get the lock in that time, it adds
itself to the contention count in the lock and blocks on a per-cpu
event channel.
When unlocking the spinlock, the locker looks to see if there's anyone
blocked waiting for the lock by checking for a non-zero waiter count.
If there's a waiter, it traverses the per-cpu "lock_spinners"
variable, which contains which lock each CPU is waiting on. It picks
one CPU waiting on the lock and sends it an event to wake it up.
This allows efficient fast-path spinlock operation, while allowing
spinning vcpus to give up their processor time while waiting for a
contended lock.
[*] 2^16 iterations is threshold at which 98% locks have been taken
according to Thomas Friebel's Xen Summit talk "Preventing Guests from
Spinning Around". Therefore, we'd expect the lock and unlock slow
paths will only be entered 2% of the time.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <clameter@linux-foundation.org>
Cc: Petr Tesarik <ptesarik@suse.cz>
Cc: Virtualization <virtualization@lists.linux-foundation.org>
Cc: Xen devel <xen-devel@lists.xensource.com>
Cc: Thomas Friebel <thomas.friebel@amd.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-07 15:07:53 -04:00
|
|
|
#include <linux/kernel_stat.h>
|
2007-07-17 21:37:06 -04:00
|
|
|
#include <linux/err.h>
|
|
|
|
#include <linux/smp.h>
|
|
|
|
|
|
|
|
#include <asm/paravirt.h>
|
|
|
|
#include <asm/desc.h>
|
|
|
|
#include <asm/pgtable.h>
|
|
|
|
#include <asm/cpu.h>
|
|
|
|
|
|
|
|
#include <xen/interface/xen.h>
|
|
|
|
#include <xen/interface/vcpu.h>
|
|
|
|
|
|
|
|
#include <asm/xen/interface.h>
|
|
|
|
#include <asm/xen/hypercall.h>
|
|
|
|
|
|
|
|
#include <xen/page.h>
|
|
|
|
#include <xen/events.h>
|
|
|
|
|
|
|
|
#include "xen-ops.h"
|
|
|
|
#include "mmu.h"
|
|
|
|
|
xen: implement Xen-specific spinlocks
The standard ticket spinlocks are very expensive in a virtual
environment, because their performance depends on Xen's scheduler
giving vcpus time in the order that they're supposed to take the
spinlock.
This implements a Xen-specific spinlock, which should be much more
efficient.
The fast-path is essentially the old Linux-x86 locks, using a single
lock byte. The locker decrements the byte; if the result is 0, then
they have the lock. If the lock is negative, then locker must spin
until the lock is positive again.
When there's contention, the locker spin for 2^16[*] iterations waiting
to get the lock. If it fails to get the lock in that time, it adds
itself to the contention count in the lock and blocks on a per-cpu
event channel.
When unlocking the spinlock, the locker looks to see if there's anyone
blocked waiting for the lock by checking for a non-zero waiter count.
If there's a waiter, it traverses the per-cpu "lock_spinners"
variable, which contains which lock each CPU is waiting on. It picks
one CPU waiting on the lock and sends it an event to wake it up.
This allows efficient fast-path spinlock operation, while allowing
spinning vcpus to give up their processor time while waiting for a
contended lock.
[*] 2^16 iterations is threshold at which 98% locks have been taken
according to Thomas Friebel's Xen Summit talk "Preventing Guests from
Spinning Around". Therefore, we'd expect the lock and unlock slow
paths will only be entered 2% of the time.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <clameter@linux-foundation.org>
Cc: Petr Tesarik <ptesarik@suse.cz>
Cc: Virtualization <virtualization@lists.linux-foundation.org>
Cc: Xen devel <xen-devel@lists.xensource.com>
Cc: Thomas Friebel <thomas.friebel@amd.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-07 15:07:53 -04:00
|
|
|
static void __cpuinit xen_init_lock_cpu(int cpu);
|
|
|
|
|
2008-05-26 18:31:27 -04:00
|
|
|
cpumask_t xen_cpu_initialized_map;
|
2007-07-17 21:37:06 -04:00
|
|
|
|
2008-06-26 05:21:54 -04:00
|
|
|
static DEFINE_PER_CPU(int, resched_irq);
|
|
|
|
static DEFINE_PER_CPU(int, callfunc_irq);
|
|
|
|
static DEFINE_PER_CPU(int, callfuncsingle_irq);
|
|
|
|
static DEFINE_PER_CPU(int, debug_irq) = -1;
|
2007-07-17 21:37:06 -04:00
|
|
|
|
|
|
|
static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id);
|
2008-06-26 05:21:54 -04:00
|
|
|
static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id);
|
2007-07-17 21:37:06 -04:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Reschedule call back. Nothing to do,
|
|
|
|
* all the work is done automatically when
|
|
|
|
* we return from the interrupt.
|
|
|
|
*/
|
|
|
|
static irqreturn_t xen_reschedule_interrupt(int irq, void *dev_id)
|
|
|
|
{
|
2008-05-26 18:31:16 -04:00
|
|
|
#ifdef CONFIG_X86_32
|
|
|
|
__get_cpu_var(irq_stat).irq_resched_count++;
|
|
|
|
#else
|
|
|
|
add_pda(irq_resched_count, 1);
|
|
|
|
#endif
|
|
|
|
|
2007-07-17 21:37:06 -04:00
|
|
|
return IRQ_HANDLED;
|
|
|
|
}
|
|
|
|
|
|
|
|
static __cpuinit void cpu_bringup_and_idle(void)
|
|
|
|
{
|
|
|
|
int cpu = smp_processor_id();
|
|
|
|
|
|
|
|
cpu_init();
|
2008-07-08 18:06:43 -04:00
|
|
|
preempt_disable();
|
|
|
|
|
2008-03-17 19:37:17 -04:00
|
|
|
xen_enable_sysenter();
|
2008-07-08 18:07:14 -04:00
|
|
|
xen_enable_syscall();
|
2007-07-17 21:37:06 -04:00
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
cpu = smp_processor_id();
|
|
|
|
smp_store_cpu_info(cpu);
|
|
|
|
cpu_data(cpu).x86_max_cores = 1;
|
|
|
|
set_cpu_sibling_map(cpu);
|
2007-07-17 21:37:06 -04:00
|
|
|
|
|
|
|
xen_setup_cpu_clockevents();
|
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
cpu_set(cpu, cpu_online_map);
|
|
|
|
x86_write_percpu(cpu_state, CPU_ONLINE);
|
|
|
|
wmb();
|
|
|
|
|
2007-07-17 21:37:06 -04:00
|
|
|
/* We can take interrupts now: we're officially "up". */
|
|
|
|
local_irq_enable();
|
|
|
|
|
|
|
|
wmb(); /* make sure everything is out */
|
|
|
|
cpu_idle();
|
|
|
|
}
|
|
|
|
|
|
|
|
static int xen_smp_intr_init(unsigned int cpu)
|
|
|
|
{
|
|
|
|
int rc;
|
2008-03-17 19:37:18 -04:00
|
|
|
const char *resched_name, *callfunc_name, *debug_name;
|
2007-07-17 21:37:06 -04:00
|
|
|
|
|
|
|
resched_name = kasprintf(GFP_KERNEL, "resched%d", cpu);
|
|
|
|
rc = bind_ipi_to_irqhandler(XEN_RESCHEDULE_VECTOR,
|
|
|
|
cpu,
|
|
|
|
xen_reschedule_interrupt,
|
|
|
|
IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
|
|
|
|
resched_name,
|
|
|
|
NULL);
|
|
|
|
if (rc < 0)
|
|
|
|
goto fail;
|
|
|
|
per_cpu(resched_irq, cpu) = rc;
|
|
|
|
|
|
|
|
callfunc_name = kasprintf(GFP_KERNEL, "callfunc%d", cpu);
|
|
|
|
rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_VECTOR,
|
|
|
|
cpu,
|
|
|
|
xen_call_function_interrupt,
|
|
|
|
IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
|
|
|
|
callfunc_name,
|
|
|
|
NULL);
|
|
|
|
if (rc < 0)
|
|
|
|
goto fail;
|
|
|
|
per_cpu(callfunc_irq, cpu) = rc;
|
|
|
|
|
2008-03-17 19:37:18 -04:00
|
|
|
debug_name = kasprintf(GFP_KERNEL, "debug%d", cpu);
|
|
|
|
rc = bind_virq_to_irqhandler(VIRQ_DEBUG, cpu, xen_debug_interrupt,
|
|
|
|
IRQF_DISABLED | IRQF_PERCPU | IRQF_NOBALANCING,
|
|
|
|
debug_name, NULL);
|
|
|
|
if (rc < 0)
|
|
|
|
goto fail;
|
|
|
|
per_cpu(debug_irq, cpu) = rc;
|
|
|
|
|
2008-06-26 05:21:54 -04:00
|
|
|
callfunc_name = kasprintf(GFP_KERNEL, "callfuncsingle%d", cpu);
|
|
|
|
rc = bind_ipi_to_irqhandler(XEN_CALL_FUNCTION_SINGLE_VECTOR,
|
|
|
|
cpu,
|
|
|
|
xen_call_function_single_interrupt,
|
|
|
|
IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
|
|
|
|
callfunc_name,
|
|
|
|
NULL);
|
|
|
|
if (rc < 0)
|
|
|
|
goto fail;
|
|
|
|
per_cpu(callfuncsingle_irq, cpu) = rc;
|
|
|
|
|
2007-07-17 21:37:06 -04:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
fail:
|
|
|
|
if (per_cpu(resched_irq, cpu) >= 0)
|
|
|
|
unbind_from_irqhandler(per_cpu(resched_irq, cpu), NULL);
|
|
|
|
if (per_cpu(callfunc_irq, cpu) >= 0)
|
|
|
|
unbind_from_irqhandler(per_cpu(callfunc_irq, cpu), NULL);
|
2008-03-17 19:37:18 -04:00
|
|
|
if (per_cpu(debug_irq, cpu) >= 0)
|
|
|
|
unbind_from_irqhandler(per_cpu(debug_irq, cpu), NULL);
|
2008-06-26 05:21:54 -04:00
|
|
|
if (per_cpu(callfuncsingle_irq, cpu) >= 0)
|
|
|
|
unbind_from_irqhandler(per_cpu(callfuncsingle_irq, cpu), NULL);
|
|
|
|
|
2007-07-17 21:37:06 -04:00
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
static void __init xen_fill_possible_map(void)
|
2007-07-17 21:37:06 -04:00
|
|
|
{
|
|
|
|
int i, rc;
|
|
|
|
|
|
|
|
for (i = 0; i < NR_CPUS; i++) {
|
|
|
|
rc = HYPERVISOR_vcpu_op(VCPUOP_is_up, i, NULL);
|
2008-07-08 18:06:56 -04:00
|
|
|
if (rc >= 0) {
|
|
|
|
num_processors++;
|
2007-07-17 21:37:06 -04:00
|
|
|
cpu_set(i, cpu_possible_map);
|
2008-07-08 18:06:56 -04:00
|
|
|
}
|
2007-07-17 21:37:06 -04:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:41 -04:00
|
|
|
static void __init xen_smp_prepare_boot_cpu(void)
|
2007-07-17 21:37:06 -04:00
|
|
|
{
|
|
|
|
BUG_ON(smp_processor_id() != 0);
|
|
|
|
native_smp_prepare_boot_cpu();
|
|
|
|
|
|
|
|
/* We've switched to the "real" per-cpu gdt, so make sure the
|
|
|
|
old memory can be recycled */
|
2008-07-08 18:06:43 -04:00
|
|
|
make_lowmem_page_readwrite(&per_cpu_var(gdt_page));
|
2007-07-17 21:37:07 -04:00
|
|
|
|
|
|
|
xen_setup_vcpu_info_placement();
|
2007-07-17 21:37:06 -04:00
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:41 -04:00
|
|
|
static void __init xen_smp_prepare_cpus(unsigned int max_cpus)
|
2007-07-17 21:37:06 -04:00
|
|
|
{
|
|
|
|
unsigned cpu;
|
|
|
|
|
xen: implement Xen-specific spinlocks
The standard ticket spinlocks are very expensive in a virtual
environment, because their performance depends on Xen's scheduler
giving vcpus time in the order that they're supposed to take the
spinlock.
This implements a Xen-specific spinlock, which should be much more
efficient.
The fast-path is essentially the old Linux-x86 locks, using a single
lock byte. The locker decrements the byte; if the result is 0, then
they have the lock. If the lock is negative, then locker must spin
until the lock is positive again.
When there's contention, the locker spin for 2^16[*] iterations waiting
to get the lock. If it fails to get the lock in that time, it adds
itself to the contention count in the lock and blocks on a per-cpu
event channel.
When unlocking the spinlock, the locker looks to see if there's anyone
blocked waiting for the lock by checking for a non-zero waiter count.
If there's a waiter, it traverses the per-cpu "lock_spinners"
variable, which contains which lock each CPU is waiting on. It picks
one CPU waiting on the lock and sends it an event to wake it up.
This allows efficient fast-path spinlock operation, while allowing
spinning vcpus to give up their processor time while waiting for a
contended lock.
[*] 2^16 iterations is threshold at which 98% locks have been taken
according to Thomas Friebel's Xen Summit talk "Preventing Guests from
Spinning Around". Therefore, we'd expect the lock and unlock slow
paths will only be entered 2% of the time.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <clameter@linux-foundation.org>
Cc: Petr Tesarik <ptesarik@suse.cz>
Cc: Virtualization <virtualization@lists.linux-foundation.org>
Cc: Xen devel <xen-devel@lists.xensource.com>
Cc: Thomas Friebel <thomas.friebel@amd.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-07 15:07:53 -04:00
|
|
|
xen_init_lock_cpu(0);
|
|
|
|
|
2007-07-17 21:37:06 -04:00
|
|
|
smp_store_cpu_info(0);
|
2008-07-08 18:06:43 -04:00
|
|
|
cpu_data(0).x86_max_cores = 1;
|
2007-07-17 21:37:06 -04:00
|
|
|
set_cpu_sibling_map(0);
|
|
|
|
|
|
|
|
if (xen_smp_intr_init(0))
|
|
|
|
BUG();
|
|
|
|
|
2008-03-27 13:06:03 -04:00
|
|
|
xen_cpu_initialized_map = cpumask_of_cpu(0);
|
2007-07-17 21:37:06 -04:00
|
|
|
|
|
|
|
/* Restrict the possible_map according to max_cpus. */
|
|
|
|
while ((num_possible_cpus() > 1) && (num_possible_cpus() > max_cpus)) {
|
2008-04-19 10:55:17 -04:00
|
|
|
for (cpu = NR_CPUS - 1; !cpu_possible(cpu); cpu--)
|
2007-07-17 21:37:06 -04:00
|
|
|
continue;
|
|
|
|
cpu_clear(cpu, cpu_possible_map);
|
|
|
|
}
|
|
|
|
|
|
|
|
for_each_possible_cpu (cpu) {
|
|
|
|
struct task_struct *idle;
|
|
|
|
|
|
|
|
if (cpu == 0)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
idle = fork_idle(cpu);
|
|
|
|
if (IS_ERR(idle))
|
|
|
|
panic("failed fork for CPU %d", cpu);
|
|
|
|
|
|
|
|
cpu_set(cpu, cpu_present_map);
|
|
|
|
}
|
|
|
|
|
|
|
|
//init_xenbus_allowed_cpumask();
|
|
|
|
}
|
|
|
|
|
|
|
|
static __cpuinit int
|
|
|
|
cpu_initialize_context(unsigned int cpu, struct task_struct *idle)
|
|
|
|
{
|
|
|
|
struct vcpu_guest_context *ctxt;
|
2008-07-08 18:06:43 -04:00
|
|
|
struct desc_struct *gdt;
|
2007-07-17 21:37:06 -04:00
|
|
|
|
2008-03-27 13:06:03 -04:00
|
|
|
if (cpu_test_and_set(cpu, xen_cpu_initialized_map))
|
2007-07-17 21:37:06 -04:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
ctxt = kzalloc(sizeof(*ctxt), GFP_KERNEL);
|
|
|
|
if (ctxt == NULL)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
gdt = get_cpu_gdt_table(cpu);
|
|
|
|
|
2007-07-17 21:37:06 -04:00
|
|
|
ctxt->flags = VGCF_IN_KERNEL;
|
|
|
|
ctxt->user_regs.ds = __USER_DS;
|
|
|
|
ctxt->user_regs.es = __USER_DS;
|
|
|
|
ctxt->user_regs.ss = __KERNEL_DS;
|
2008-07-08 18:06:43 -04:00
|
|
|
#ifdef CONFIG_X86_32
|
|
|
|
ctxt->user_regs.fs = __KERNEL_PERCPU;
|
|
|
|
#endif
|
2007-07-17 21:37:06 -04:00
|
|
|
ctxt->user_regs.eip = (unsigned long)cpu_bringup_and_idle;
|
|
|
|
ctxt->user_regs.eflags = 0x1000; /* IOPL_RING1 */
|
|
|
|
|
|
|
|
memset(&ctxt->fpu_ctxt, 0, sizeof(ctxt->fpu_ctxt));
|
|
|
|
|
|
|
|
xen_copy_trap_info(ctxt->trap_ctxt);
|
|
|
|
|
|
|
|
ctxt->ldt_ents = 0;
|
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
BUG_ON((unsigned long)gdt & ~PAGE_MASK);
|
|
|
|
make_lowmem_page_readonly(gdt);
|
2007-07-17 21:37:06 -04:00
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
ctxt->gdt_frames[0] = virt_to_mfn(gdt);
|
|
|
|
ctxt->gdt_ents = GDT_ENTRIES;
|
2007-07-17 21:37:06 -04:00
|
|
|
|
|
|
|
ctxt->user_regs.cs = __KERNEL_CS;
|
2008-01-30 07:31:02 -05:00
|
|
|
ctxt->user_regs.esp = idle->thread.sp0 - sizeof(struct pt_regs);
|
2007-07-17 21:37:06 -04:00
|
|
|
|
|
|
|
ctxt->kernel_ss = __KERNEL_DS;
|
2008-01-30 07:31:02 -05:00
|
|
|
ctxt->kernel_sp = idle->thread.sp0;
|
2007-07-17 21:37:06 -04:00
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
#ifdef CONFIG_X86_32
|
2007-07-17 21:37:06 -04:00
|
|
|
ctxt->event_callback_cs = __KERNEL_CS;
|
|
|
|
ctxt->failsafe_callback_cs = __KERNEL_CS;
|
2008-07-08 18:06:43 -04:00
|
|
|
#endif
|
|
|
|
ctxt->event_callback_eip = (unsigned long)xen_hypervisor_callback;
|
2007-07-17 21:37:06 -04:00
|
|
|
ctxt->failsafe_callback_eip = (unsigned long)xen_failsafe_callback;
|
|
|
|
|
|
|
|
per_cpu(xen_cr3, cpu) = __pa(swapper_pg_dir);
|
|
|
|
ctxt->ctrlreg[3] = xen_pfn_to_cr3(virt_to_mfn(swapper_pg_dir));
|
|
|
|
|
|
|
|
if (HYPERVISOR_vcpu_op(VCPUOP_initialise, cpu, ctxt))
|
|
|
|
BUG();
|
|
|
|
|
|
|
|
kfree(ctxt);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:41 -04:00
|
|
|
static int __cpuinit xen_cpu_up(unsigned int cpu)
|
2007-07-17 21:37:06 -04:00
|
|
|
{
|
|
|
|
struct task_struct *idle = idle_task(cpu);
|
|
|
|
int rc;
|
|
|
|
|
|
|
|
#if 0
|
|
|
|
rc = cpu_up_check(cpu);
|
|
|
|
if (rc)
|
|
|
|
return rc;
|
|
|
|
#endif
|
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
#ifdef CONFIG_X86_64
|
|
|
|
/* Allocate node local memory for AP pdas */
|
|
|
|
WARN_ON(cpu == 0);
|
|
|
|
if (cpu > 0) {
|
|
|
|
rc = get_local_pda(cpu);
|
|
|
|
if (rc)
|
|
|
|
return rc;
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
|
|
|
|
#ifdef CONFIG_X86_32
|
2007-07-17 21:37:06 -04:00
|
|
|
init_gdt(cpu);
|
|
|
|
per_cpu(current_task, cpu) = idle;
|
|
|
|
irq_ctx_init(cpu);
|
2008-07-08 18:06:43 -04:00
|
|
|
#else
|
|
|
|
cpu_pda(cpu)->pcurrent = idle;
|
|
|
|
clear_tsk_thread_flag(idle, TIF_FORK);
|
|
|
|
#endif
|
2007-07-17 21:37:06 -04:00
|
|
|
xen_setup_timer(cpu);
|
xen: implement Xen-specific spinlocks
The standard ticket spinlocks are very expensive in a virtual
environment, because their performance depends on Xen's scheduler
giving vcpus time in the order that they're supposed to take the
spinlock.
This implements a Xen-specific spinlock, which should be much more
efficient.
The fast-path is essentially the old Linux-x86 locks, using a single
lock byte. The locker decrements the byte; if the result is 0, then
they have the lock. If the lock is negative, then locker must spin
until the lock is positive again.
When there's contention, the locker spin for 2^16[*] iterations waiting
to get the lock. If it fails to get the lock in that time, it adds
itself to the contention count in the lock and blocks on a per-cpu
event channel.
When unlocking the spinlock, the locker looks to see if there's anyone
blocked waiting for the lock by checking for a non-zero waiter count.
If there's a waiter, it traverses the per-cpu "lock_spinners"
variable, which contains which lock each CPU is waiting on. It picks
one CPU waiting on the lock and sends it an event to wake it up.
This allows efficient fast-path spinlock operation, while allowing
spinning vcpus to give up their processor time while waiting for a
contended lock.
[*] 2^16 iterations is threshold at which 98% locks have been taken
according to Thomas Friebel's Xen Summit talk "Preventing Guests from
Spinning Around". Therefore, we'd expect the lock and unlock slow
paths will only be entered 2% of the time.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <clameter@linux-foundation.org>
Cc: Petr Tesarik <ptesarik@suse.cz>
Cc: Virtualization <virtualization@lists.linux-foundation.org>
Cc: Xen devel <xen-devel@lists.xensource.com>
Cc: Thomas Friebel <thomas.friebel@amd.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-07 15:07:53 -04:00
|
|
|
xen_init_lock_cpu(cpu);
|
2007-07-17 21:37:06 -04:00
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
per_cpu(cpu_state, cpu) = CPU_UP_PREPARE;
|
|
|
|
|
2007-07-17 21:37:06 -04:00
|
|
|
/* make sure interrupts start blocked */
|
|
|
|
per_cpu(xen_vcpu, cpu)->evtchn_upcall_mask = 1;
|
|
|
|
|
|
|
|
rc = cpu_initialize_context(cpu, idle);
|
|
|
|
if (rc)
|
|
|
|
return rc;
|
|
|
|
|
|
|
|
if (num_online_cpus() == 1)
|
|
|
|
alternatives_smp_switch(1);
|
|
|
|
|
|
|
|
rc = xen_smp_intr_init(cpu);
|
|
|
|
if (rc)
|
|
|
|
return rc;
|
|
|
|
|
|
|
|
rc = HYPERVISOR_vcpu_op(VCPUOP_up, cpu, NULL);
|
|
|
|
BUG_ON(rc);
|
|
|
|
|
2008-07-08 18:06:43 -04:00
|
|
|
while(per_cpu(cpu_state, cpu) != CPU_ONLINE) {
|
|
|
|
HYPERVISOR_sched_op(SCHEDOP_yield, 0);
|
|
|
|
barrier();
|
|
|
|
}
|
|
|
|
|
2007-07-17 21:37:06 -04:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:41 -04:00
|
|
|
static void xen_smp_cpus_done(unsigned int max_cpus)
|
2007-07-17 21:37:06 -04:00
|
|
|
{
|
|
|
|
}
|
|
|
|
|
|
|
|
static void stop_self(void *v)
|
|
|
|
{
|
|
|
|
int cpu = smp_processor_id();
|
|
|
|
|
|
|
|
/* make sure we're not pinning something down */
|
|
|
|
load_cr3(swapper_pg_dir);
|
|
|
|
/* should set up a minimal gdt */
|
|
|
|
|
|
|
|
HYPERVISOR_vcpu_op(VCPUOP_down, cpu, NULL);
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:41 -04:00
|
|
|
static void xen_smp_send_stop(void)
|
2007-07-17 21:37:06 -04:00
|
|
|
{
|
2008-06-06 05:18:06 -04:00
|
|
|
smp_call_function(stop_self, NULL, 0);
|
2007-07-17 21:37:06 -04:00
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:41 -04:00
|
|
|
static void xen_smp_send_reschedule(int cpu)
|
2007-07-17 21:37:06 -04:00
|
|
|
{
|
|
|
|
xen_send_IPI_one(cpu, XEN_RESCHEDULE_VECTOR);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void xen_send_IPI_mask(cpumask_t mask, enum ipi_vector vector)
|
|
|
|
{
|
|
|
|
unsigned cpu;
|
|
|
|
|
|
|
|
cpus_and(mask, mask, cpu_online_map);
|
|
|
|
|
2008-05-12 15:21:13 -04:00
|
|
|
for_each_cpu_mask_nr(cpu, mask)
|
2007-07-17 21:37:06 -04:00
|
|
|
xen_send_IPI_one(cpu, vector);
|
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:41 -04:00
|
|
|
static void xen_smp_send_call_function_ipi(cpumask_t mask)
|
2008-06-26 05:21:54 -04:00
|
|
|
{
|
|
|
|
int cpu;
|
|
|
|
|
|
|
|
xen_send_IPI_mask(mask, XEN_CALL_FUNCTION_VECTOR);
|
|
|
|
|
|
|
|
/* Make sure other vcpus get a chance to run if they need to. */
|
2008-07-15 18:29:07 -04:00
|
|
|
for_each_cpu_mask_nr(cpu, mask) {
|
2008-06-26 05:21:54 -04:00
|
|
|
if (xen_vcpu_stolen(cpu)) {
|
|
|
|
HYPERVISOR_sched_op(SCHEDOP_yield, 0);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:41 -04:00
|
|
|
static void xen_smp_send_call_function_single_ipi(int cpu)
|
2008-06-26 05:21:54 -04:00
|
|
|
{
|
|
|
|
xen_send_IPI_mask(cpumask_of_cpu(cpu), XEN_CALL_FUNCTION_SINGLE_VECTOR);
|
|
|
|
}
|
|
|
|
|
2007-07-17 21:37:06 -04:00
|
|
|
static irqreturn_t xen_call_function_interrupt(int irq, void *dev_id)
|
|
|
|
{
|
|
|
|
irq_enter();
|
2008-06-26 05:21:54 -04:00
|
|
|
generic_smp_call_function_interrupt();
|
2008-07-08 18:06:43 -04:00
|
|
|
#ifdef CONFIG_X86_32
|
2007-10-17 12:04:40 -04:00
|
|
|
__get_cpu_var(irq_stat).irq_call_count++;
|
2008-07-08 18:06:43 -04:00
|
|
|
#else
|
|
|
|
add_pda(irq_call_count, 1);
|
|
|
|
#endif
|
2007-07-17 21:37:06 -04:00
|
|
|
irq_exit();
|
|
|
|
|
|
|
|
return IRQ_HANDLED;
|
|
|
|
}
|
|
|
|
|
2008-06-26 05:21:54 -04:00
|
|
|
static irqreturn_t xen_call_function_single_interrupt(int irq, void *dev_id)
|
2007-07-17 21:37:06 -04:00
|
|
|
{
|
2008-06-26 05:21:54 -04:00
|
|
|
irq_enter();
|
|
|
|
generic_smp_call_function_single_interrupt();
|
2008-07-08 18:06:43 -04:00
|
|
|
#ifdef CONFIG_X86_32
|
2008-06-26 05:21:54 -04:00
|
|
|
__get_cpu_var(irq_stat).irq_call_count++;
|
2008-07-08 18:06:43 -04:00
|
|
|
#else
|
|
|
|
add_pda(irq_call_count, 1);
|
|
|
|
#endif
|
2008-06-26 05:21:54 -04:00
|
|
|
irq_exit();
|
2007-07-17 21:37:06 -04:00
|
|
|
|
2008-06-26 05:21:54 -04:00
|
|
|
return IRQ_HANDLED;
|
2007-07-17 21:37:06 -04:00
|
|
|
}
|
2008-07-08 18:06:41 -04:00
|
|
|
|
xen: implement Xen-specific spinlocks
The standard ticket spinlocks are very expensive in a virtual
environment, because their performance depends on Xen's scheduler
giving vcpus time in the order that they're supposed to take the
spinlock.
This implements a Xen-specific spinlock, which should be much more
efficient.
The fast-path is essentially the old Linux-x86 locks, using a single
lock byte. The locker decrements the byte; if the result is 0, then
they have the lock. If the lock is negative, then locker must spin
until the lock is positive again.
When there's contention, the locker spin for 2^16[*] iterations waiting
to get the lock. If it fails to get the lock in that time, it adds
itself to the contention count in the lock and blocks on a per-cpu
event channel.
When unlocking the spinlock, the locker looks to see if there's anyone
blocked waiting for the lock by checking for a non-zero waiter count.
If there's a waiter, it traverses the per-cpu "lock_spinners"
variable, which contains which lock each CPU is waiting on. It picks
one CPU waiting on the lock and sends it an event to wake it up.
This allows efficient fast-path spinlock operation, while allowing
spinning vcpus to give up their processor time while waiting for a
contended lock.
[*] 2^16 iterations is threshold at which 98% locks have been taken
according to Thomas Friebel's Xen Summit talk "Preventing Guests from
Spinning Around". Therefore, we'd expect the lock and unlock slow
paths will only be entered 2% of the time.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <clameter@linux-foundation.org>
Cc: Petr Tesarik <ptesarik@suse.cz>
Cc: Virtualization <virtualization@lists.linux-foundation.org>
Cc: Xen devel <xen-devel@lists.xensource.com>
Cc: Thomas Friebel <thomas.friebel@amd.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-07 15:07:53 -04:00
|
|
|
struct xen_spinlock {
|
|
|
|
unsigned char lock; /* 0 -> free; 1 -> locked */
|
|
|
|
unsigned short spinners; /* count of waiting cpus */
|
|
|
|
};
|
|
|
|
|
|
|
|
static int xen_spin_is_locked(struct raw_spinlock *lock)
|
|
|
|
{
|
|
|
|
struct xen_spinlock *xl = (struct xen_spinlock *)lock;
|
|
|
|
|
|
|
|
return xl->lock != 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int xen_spin_is_contended(struct raw_spinlock *lock)
|
|
|
|
{
|
|
|
|
struct xen_spinlock *xl = (struct xen_spinlock *)lock;
|
|
|
|
|
|
|
|
/* Not strictly true; this is only the count of contended
|
|
|
|
lock-takers entering the slow path. */
|
|
|
|
return xl->spinners != 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int xen_spin_trylock(struct raw_spinlock *lock)
|
|
|
|
{
|
|
|
|
struct xen_spinlock *xl = (struct xen_spinlock *)lock;
|
|
|
|
u8 old = 1;
|
|
|
|
|
|
|
|
asm("xchgb %b0,%1"
|
|
|
|
: "+q" (old), "+m" (xl->lock) : : "memory");
|
|
|
|
|
|
|
|
return old == 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static DEFINE_PER_CPU(int, lock_kicker_irq) = -1;
|
|
|
|
static DEFINE_PER_CPU(struct xen_spinlock *, lock_spinners);
|
|
|
|
|
|
|
|
static inline void spinning_lock(struct xen_spinlock *xl)
|
|
|
|
{
|
|
|
|
__get_cpu_var(lock_spinners) = xl;
|
|
|
|
wmb(); /* set lock of interest before count */
|
|
|
|
asm(LOCK_PREFIX " incw %0"
|
|
|
|
: "+m" (xl->spinners) : : "memory");
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void unspinning_lock(struct xen_spinlock *xl)
|
|
|
|
{
|
|
|
|
asm(LOCK_PREFIX " decw %0"
|
|
|
|
: "+m" (xl->spinners) : : "memory");
|
|
|
|
wmb(); /* decrement count before clearing lock */
|
|
|
|
__get_cpu_var(lock_spinners) = NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
static noinline int xen_spin_lock_slow(struct raw_spinlock *lock)
|
|
|
|
{
|
|
|
|
struct xen_spinlock *xl = (struct xen_spinlock *)lock;
|
|
|
|
int irq = __get_cpu_var(lock_kicker_irq);
|
|
|
|
int ret;
|
|
|
|
|
|
|
|
/* If kicker interrupts not initialized yet, just spin */
|
|
|
|
if (irq == -1)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
/* announce we're spinning */
|
|
|
|
spinning_lock(xl);
|
|
|
|
|
|
|
|
/* clear pending */
|
|
|
|
xen_clear_irq_pending(irq);
|
|
|
|
|
|
|
|
/* check again make sure it didn't become free while
|
|
|
|
we weren't looking */
|
|
|
|
ret = xen_spin_trylock(lock);
|
|
|
|
if (ret)
|
|
|
|
goto out;
|
|
|
|
|
|
|
|
/* block until irq becomes pending */
|
|
|
|
xen_poll_irq(irq);
|
|
|
|
kstat_this_cpu.irqs[irq]++;
|
|
|
|
|
|
|
|
out:
|
|
|
|
unspinning_lock(xl);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void xen_spin_lock(struct raw_spinlock *lock)
|
|
|
|
{
|
|
|
|
struct xen_spinlock *xl = (struct xen_spinlock *)lock;
|
|
|
|
int timeout;
|
|
|
|
u8 oldval;
|
|
|
|
|
|
|
|
do {
|
|
|
|
timeout = 1 << 10;
|
|
|
|
|
|
|
|
asm("1: xchgb %1,%0\n"
|
|
|
|
" testb %1,%1\n"
|
|
|
|
" jz 3f\n"
|
|
|
|
"2: rep;nop\n"
|
|
|
|
" cmpb $0,%0\n"
|
|
|
|
" je 1b\n"
|
|
|
|
" dec %2\n"
|
|
|
|
" jnz 2b\n"
|
|
|
|
"3:\n"
|
|
|
|
: "+m" (xl->lock), "=q" (oldval), "+r" (timeout)
|
|
|
|
: "1" (1)
|
|
|
|
: "memory");
|
|
|
|
|
|
|
|
} while (unlikely(oldval != 0 && !xen_spin_lock_slow(lock)));
|
|
|
|
}
|
|
|
|
|
|
|
|
static noinline void xen_spin_unlock_slow(struct xen_spinlock *xl)
|
|
|
|
{
|
|
|
|
int cpu;
|
|
|
|
|
|
|
|
for_each_online_cpu(cpu) {
|
|
|
|
/* XXX should mix up next cpu selection */
|
|
|
|
if (per_cpu(lock_spinners, cpu) == xl) {
|
|
|
|
xen_send_IPI_one(cpu, XEN_SPIN_UNLOCK_VECTOR);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void xen_spin_unlock(struct raw_spinlock *lock)
|
|
|
|
{
|
|
|
|
struct xen_spinlock *xl = (struct xen_spinlock *)lock;
|
|
|
|
|
|
|
|
smp_wmb(); /* make sure no writes get moved after unlock */
|
|
|
|
xl->lock = 0; /* release lock */
|
|
|
|
|
|
|
|
/* make sure unlock happens before kick */
|
|
|
|
barrier();
|
|
|
|
|
|
|
|
if (unlikely(xl->spinners))
|
|
|
|
xen_spin_unlock_slow(xl);
|
|
|
|
}
|
|
|
|
|
|
|
|
static __cpuinit void xen_init_lock_cpu(int cpu)
|
|
|
|
{
|
|
|
|
int irq;
|
|
|
|
const char *name;
|
|
|
|
|
|
|
|
name = kasprintf(GFP_KERNEL, "spinlock%d", cpu);
|
|
|
|
irq = bind_ipi_to_irqhandler(XEN_SPIN_UNLOCK_VECTOR,
|
|
|
|
cpu,
|
|
|
|
xen_reschedule_interrupt,
|
|
|
|
IRQF_DISABLED|IRQF_PERCPU|IRQF_NOBALANCING,
|
|
|
|
name,
|
|
|
|
NULL);
|
|
|
|
|
|
|
|
if (irq >= 0) {
|
|
|
|
disable_irq(irq); /* make sure it's never delivered */
|
|
|
|
per_cpu(lock_kicker_irq, cpu) = irq;
|
|
|
|
}
|
|
|
|
|
|
|
|
printk("cpu %d spinlock event irq %d\n", cpu, irq);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __init xen_init_spinlocks(void)
|
|
|
|
{
|
|
|
|
pv_lock_ops.spin_is_locked = xen_spin_is_locked;
|
|
|
|
pv_lock_ops.spin_is_contended = xen_spin_is_contended;
|
|
|
|
pv_lock_ops.spin_lock = xen_spin_lock;
|
|
|
|
pv_lock_ops.spin_trylock = xen_spin_trylock;
|
|
|
|
pv_lock_ops.spin_unlock = xen_spin_unlock;
|
|
|
|
}
|
|
|
|
|
2008-07-08 18:06:41 -04:00
|
|
|
static const struct smp_ops xen_smp_ops __initdata = {
|
|
|
|
.smp_prepare_boot_cpu = xen_smp_prepare_boot_cpu,
|
|
|
|
.smp_prepare_cpus = xen_smp_prepare_cpus,
|
|
|
|
.cpu_up = xen_cpu_up,
|
|
|
|
.smp_cpus_done = xen_smp_cpus_done,
|
|
|
|
|
|
|
|
.smp_send_stop = xen_smp_send_stop,
|
|
|
|
.smp_send_reschedule = xen_smp_send_reschedule,
|
|
|
|
|
|
|
|
.send_call_func_ipi = xen_smp_send_call_function_ipi,
|
|
|
|
.send_call_func_single_ipi = xen_smp_send_call_function_single_ipi,
|
|
|
|
};
|
|
|
|
|
|
|
|
void __init xen_smp_init(void)
|
|
|
|
{
|
|
|
|
smp_ops = xen_smp_ops;
|
2008-07-08 18:06:43 -04:00
|
|
|
xen_fill_possible_map();
|
xen: implement Xen-specific spinlocks
The standard ticket spinlocks are very expensive in a virtual
environment, because their performance depends on Xen's scheduler
giving vcpus time in the order that they're supposed to take the
spinlock.
This implements a Xen-specific spinlock, which should be much more
efficient.
The fast-path is essentially the old Linux-x86 locks, using a single
lock byte. The locker decrements the byte; if the result is 0, then
they have the lock. If the lock is negative, then locker must spin
until the lock is positive again.
When there's contention, the locker spin for 2^16[*] iterations waiting
to get the lock. If it fails to get the lock in that time, it adds
itself to the contention count in the lock and blocks on a per-cpu
event channel.
When unlocking the spinlock, the locker looks to see if there's anyone
blocked waiting for the lock by checking for a non-zero waiter count.
If there's a waiter, it traverses the per-cpu "lock_spinners"
variable, which contains which lock each CPU is waiting on. It picks
one CPU waiting on the lock and sends it an event to wake it up.
This allows efficient fast-path spinlock operation, while allowing
spinning vcpus to give up their processor time while waiting for a
contended lock.
[*] 2^16 iterations is threshold at which 98% locks have been taken
according to Thomas Friebel's Xen Summit talk "Preventing Guests from
Spinning Around". Therefore, we'd expect the lock and unlock slow
paths will only be entered 2% of the time.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Christoph Lameter <clameter@linux-foundation.org>
Cc: Petr Tesarik <ptesarik@suse.cz>
Cc: Virtualization <virtualization@lists.linux-foundation.org>
Cc: Xen devel <xen-devel@lists.xensource.com>
Cc: Thomas Friebel <thomas.friebel@amd.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2008-07-07 15:07:53 -04:00
|
|
|
xen_init_spinlocks();
|
2008-07-08 18:06:41 -04:00
|
|
|
}
|