1
linux/drivers/xen/manage.c
Ian Campbell b4606f2165 xen: explicitly create/destroy stop_machine workqueues outside suspend/resume region.
I have observed cases where the implicit stop_machine_destroy() done by
stop_machine() hangs while destroying the workqueues, specifically in
kthread_stop(). This seems to be because timer ticks are not restarted
until after stop_machine() returns.

Fortunately stop_machine provides a facility to pre-create/post-destroy
the workqueues so use this to ensure that workqueues are only destroyed
after everything is really up and running again.

I only actually observed this failure with 2.6.30. It seems that newer
kernels are somehow more robust against doing kthread_stop() without timer
interrupts (I tried some backports of some likely looking candidates but
did not track down the commit which added this robustness). However this
change seems like a reasonable belt&braces thing to do.

Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Stable Kernel <stable@kernel.org>
2009-12-03 11:14:56 -08:00

274 lines
5.6 KiB
C

/*
* Handle extern requests for shutdown, reboot and sysrq
*/
#include <linux/kernel.h>
#include <linux/err.h>
#include <linux/reboot.h>
#include <linux/sysrq.h>
#include <linux/stop_machine.h>
#include <linux/freezer.h>
#include <xen/xenbus.h>
#include <xen/grant_table.h>
#include <xen/events.h>
#include <xen/hvc-console.h>
#include <xen/xen-ops.h>
#include <asm/xen/hypercall.h>
#include <asm/xen/page.h>
enum shutdown_state {
SHUTDOWN_INVALID = -1,
SHUTDOWN_POWEROFF = 0,
SHUTDOWN_SUSPEND = 2,
/* Code 3 is SHUTDOWN_CRASH, which we don't use because the domain can only
report a crash, not be instructed to crash!
HALT is the same as POWEROFF, as far as we're concerned. The tools use
the distinction when we return the reason code to them. */
SHUTDOWN_HALT = 4,
};
/* Ignore multiple shutdown requests. */
static enum shutdown_state shutting_down = SHUTDOWN_INVALID;
#ifdef CONFIG_PM_SLEEP
static int xen_suspend(void *data)
{
int *cancelled = data;
int err;
BUG_ON(!irqs_disabled());
err = sysdev_suspend(PMSG_SUSPEND);
if (err) {
printk(KERN_ERR "xen_suspend: sysdev_suspend failed: %d\n",
err);
return err;
}
xen_mm_pin_all();
gnttab_suspend();
xen_pre_suspend();
/*
* This hypercall returns 1 if suspend was cancelled
* or the domain was merely checkpointed, and 0 if it
* is resuming in a new domain.
*/
*cancelled = HYPERVISOR_suspend(virt_to_mfn(xen_start_info));
xen_post_suspend(*cancelled);
gnttab_resume();
xen_mm_unpin_all();
if (!*cancelled) {
xen_irq_resume();
xen_console_resume();
xen_timer_resume();
}
sysdev_resume();
return 0;
}
static void do_suspend(void)
{
int err;
int cancelled = 1;
shutting_down = SHUTDOWN_SUSPEND;
err = stop_machine_create();
if (err) {
printk(KERN_ERR "xen suspend: failed to setup stop_machine %d\n", err);
goto out;
}
#ifdef CONFIG_PREEMPT
/* If the kernel is preemptible, we need to freeze all the processes
to prevent them from being in the middle of a pagetable update
during suspend. */
err = freeze_processes();
if (err) {
printk(KERN_ERR "xen suspend: freeze failed %d\n", err);
goto out_destroy_sm;
}
#endif
err = dpm_suspend_start(PMSG_SUSPEND);
if (err) {
printk(KERN_ERR "xen suspend: dpm_suspend_start %d\n", err);
goto out_thaw;
}
err = dpm_suspend_noirq(PMSG_SUSPEND);
if (err) {
printk(KERN_ERR "dpm_suspend_noirq failed: %d\n", err);
goto out_resume;
}
printk(KERN_DEBUG "suspending xenstore...\n");
xs_suspend();
err = stop_machine(xen_suspend, &cancelled, cpumask_of(0));
dpm_resume_noirq(PMSG_RESUME);
if (err) {
printk(KERN_ERR "failed to start xen_suspend: %d\n", err);
cancelled = 1;
}
if (!cancelled) {
xen_arch_resume();
xs_resume();
} else
xs_suspend_cancel();
out_resume:
dpm_resume_end(PMSG_RESUME);
/* Make sure timer events get retriggered on all CPUs */
clock_was_set();
out_thaw:
#ifdef CONFIG_PREEMPT
thaw_processes();
out_destroy_sm:
#endif
stop_machine_destroy();
out:
shutting_down = SHUTDOWN_INVALID;
}
#endif /* CONFIG_PM_SLEEP */
static void shutdown_handler(struct xenbus_watch *watch,
const char **vec, unsigned int len)
{
char *str;
struct xenbus_transaction xbt;
int err;
if (shutting_down != SHUTDOWN_INVALID)
return;
again:
err = xenbus_transaction_start(&xbt);
if (err)
return;
str = (char *)xenbus_read(xbt, "control", "shutdown", NULL);
/* Ignore read errors and empty reads. */
if (XENBUS_IS_ERR_READ(str)) {
xenbus_transaction_end(xbt, 1);
return;
}
xenbus_write(xbt, "control", "shutdown", "");
err = xenbus_transaction_end(xbt, 0);
if (err == -EAGAIN) {
kfree(str);
goto again;
}
if (strcmp(str, "poweroff") == 0 ||
strcmp(str, "halt") == 0) {
shutting_down = SHUTDOWN_POWEROFF;
orderly_poweroff(false);
} else if (strcmp(str, "reboot") == 0) {
shutting_down = SHUTDOWN_POWEROFF; /* ? */
ctrl_alt_del();
#ifdef CONFIG_PM_SLEEP
} else if (strcmp(str, "suspend") == 0) {
do_suspend();
#endif
} else {
printk(KERN_INFO "Ignoring shutdown request: %s\n", str);
shutting_down = SHUTDOWN_INVALID;
}
kfree(str);
}
static void sysrq_handler(struct xenbus_watch *watch, const char **vec,
unsigned int len)
{
char sysrq_key = '\0';
struct xenbus_transaction xbt;
int err;
again:
err = xenbus_transaction_start(&xbt);
if (err)
return;
if (!xenbus_scanf(xbt, "control", "sysrq", "%c", &sysrq_key)) {
printk(KERN_ERR "Unable to read sysrq code in "
"control/sysrq\n");
xenbus_transaction_end(xbt, 1);
return;
}
if (sysrq_key != '\0')
xenbus_printf(xbt, "control", "sysrq", "%c", '\0');
err = xenbus_transaction_end(xbt, 0);
if (err == -EAGAIN)
goto again;
if (sysrq_key != '\0')
handle_sysrq(sysrq_key, NULL);
}
static struct xenbus_watch shutdown_watch = {
.node = "control/shutdown",
.callback = shutdown_handler
};
static struct xenbus_watch sysrq_watch = {
.node = "control/sysrq",
.callback = sysrq_handler
};
static int setup_shutdown_watcher(void)
{
int err;
err = register_xenbus_watch(&shutdown_watch);
if (err) {
printk(KERN_ERR "Failed to set shutdown watcher\n");
return err;
}
err = register_xenbus_watch(&sysrq_watch);
if (err) {
printk(KERN_ERR "Failed to set sysrq watcher\n");
return err;
}
return 0;
}
static int shutdown_event(struct notifier_block *notifier,
unsigned long event,
void *data)
{
setup_shutdown_watcher();
return NOTIFY_DONE;
}
static int __init setup_shutdown_event(void)
{
static struct notifier_block xenstore_notifier = {
.notifier_call = shutdown_event
};
register_xenstore_notifier(&xenstore_notifier);
return 0;
}
subsys_initcall(setup_shutdown_event);