void
xp_exit_sn2(void)
{
BUG_ON(!is_shub());
xp_unregister_nofault_code_sn2();
}
/*
* This thread is responsible for nearly all of the partition
* activation/deactivation.
*/
static int
xpc_hb_checker(void *ignore)
{
int force_IRQ = 0;
/* this thread was marked active by xpc_hb_init() */
set_cpus_allowed_ptr(current, cpumask_of(XPC_HB_CHECK_CPU));
/* set our heartbeating to other partitions into motion */
xpc_hb_check_timeout = jiffies + (xpc_hb_check_interval * HZ);
xpc_start_hb_beater();
while (!xpc_exiting) {
dev_dbg(xpc_part, "woke up with %d ticks rem; %d IRQs have "
"been received\n",
(int)(xpc_hb_check_timeout - jiffies),
xpc_activate_IRQ_rcvd);
/* checking of remote heartbeats is skewed by IRQ handling */
if (time_is_before_eq_jiffies(xpc_hb_check_timeout)) {
xpc_hb_check_timeout = jiffies +
(xpc_hb_check_interval * HZ);
dev_dbg(xpc_part, "checking remote heartbeats\n");
xpc_check_remote_hb();
/*
* On sn2 we need to periodically recheck to ensure no
* IRQ/amo pairs have been missed.
*/
if (is_shub())
force_IRQ = 1;
}
/* check for outstanding IRQs */
if (xpc_activate_IRQ_rcvd > 0 || force_IRQ != 0) {
force_IRQ = 0;
dev_dbg(xpc_part, "processing activate IRQs "
"received\n");
xpc_process_activate_IRQ_rcvd();
}
/* wait for IRQ or timeout */
(void)wait_event_interruptible(xpc_activate_IRQ_wq,
(time_is_before_eq_jiffies(
xpc_hb_check_timeout) ||
xpc_activate_IRQ_rcvd > 0 ||
xpc_exiting));
}
xpc_stop_hb_beater();
dev_dbg(xpc_part, "heartbeat checker is exiting\n");
/* mark this thread as having exited */
complete(&xpc_hb_checker_exited);
return 0;
}
static int
xpc_hb_checker(void *ignore)
{
int force_IRQ = 0;
/* */
set_cpus_allowed_ptr(current, cpumask_of(XPC_HB_CHECK_CPU));
/* */
xpc_hb_check_timeout = jiffies + (xpc_hb_check_interval * HZ);
xpc_start_hb_beater();
while (!xpc_exiting) {
dev_dbg(xpc_part, "woke up with %d ticks rem; %d IRQs have "
"been received\n",
(int)(xpc_hb_check_timeout - jiffies),
xpc_activate_IRQ_rcvd);
/* */
if (time_is_before_eq_jiffies(xpc_hb_check_timeout)) {
xpc_hb_check_timeout = jiffies +
(xpc_hb_check_interval * HZ);
dev_dbg(xpc_part, "checking remote heartbeats\n");
xpc_check_remote_hb();
/*
*/
if (is_shub())
force_IRQ = 1;
}
/* */
if (xpc_activate_IRQ_rcvd > 0 || force_IRQ != 0) {
force_IRQ = 0;
dev_dbg(xpc_part, "processing activate IRQs "
"received\n");
xpc_arch_ops.process_activate_IRQ_rcvd();
}
/* */
(void)wait_event_interruptible(xpc_activate_IRQ_wq,
(time_is_before_eq_jiffies(
xpc_hb_check_timeout) ||
xpc_activate_IRQ_rcvd > 0 ||
xpc_exiting));
}
xpc_stop_hb_beater();
dev_dbg(xpc_part, "heartbeat checker is exiting\n");
/* */
complete(&xpc_hb_checker_exited);
return 0;
}
void __exit
xp_exit(void)
{
if (is_shub())
xp_exit_sn2();
else if (is_uv())
xp_exit_uv();
}
enum xp_retval
xp_init_sn2(void)
{
BUG_ON(!is_shub());
xp_max_npartitions = XP_MAX_NPARTITIONS_SN2;
xp_partition_id = sn_partition_id;
xp_region_size = sn_region_size;
xp_pa = xp_pa_sn2;
xp_socket_pa = xp_socket_pa_sn2;
xp_remote_memcpy = xp_remote_memcpy_sn2;
xp_cpu_to_nasid = xp_cpu_to_nasid_sn2;
xp_expand_memprotect = xp_expand_memprotect_sn2;
xp_restrict_memprotect = xp_restrict_memprotect_sn2;
return xp_register_nofault_code_sn2();
}
int __init
xp_init(void)
{
enum xp_retval ret;
int ch_number;
for (ch_number = 0; ch_number < XPC_MAX_NCHANNELS; ch_number++)
mutex_init(&xpc_registrations[ch_number].mutex);
if (is_shub())
ret = xp_init_sn2();
else if (is_uv())
ret = xp_init_uv();
else
ret = 0;
if (ret != xpSuccess)
return ret;
return 0;
}
int __init
xp_init(void)
{
enum xp_retval ret;
int ch_number;
if (is_shub())
ret = xp_init_sn2();
else if (is_uv())
ret = xp_init_uv();
else
ret = xpUnsupported;
if (ret != xpSuccess)
return -ENODEV;
/* initialize the connection registration mutex */
for (ch_number = 0; ch_number < XPC_MAX_NCHANNELS; ch_number++)
mutex_init(&xpc_registrations[ch_number].mutex);
return 0;
}
int __init
xpc_init(void)
{
int ret;
struct task_struct *kthread;
dev_set_name(xpc_part, "part");
dev_set_name(xpc_chan, "chan");
if (is_shub()) {
/*
* The ia64-sn2 architecture supports at most 64 partitions.
* And the inability to unregister remote amos restricts us
* further to only support exactly 64 partitions on this
* architecture, no less.
*/
if (xp_max_npartitions != 64) {
dev_err(xpc_part, "max #of partitions not set to 64\n");
ret = -EINVAL;
} else {
ret = xpc_init_sn2();
}
} else if (is_uv()) {
ret = xpc_init_uv();
} else {
ret = -ENODEV;
}
if (ret != 0)
return ret;
ret = xpc_setup_partitions();
if (ret != 0) {
dev_err(xpc_part, "can't get memory for partition structure\n");
goto out_1;
}
xpc_sysctl = register_sysctl_table(xpc_sys_dir);
/*
* Fill the partition reserved page with the information needed by
* other partitions to discover we are alive and establish initial
* communications.
*/
ret = xpc_setup_rsvd_page();
if (ret != 0) {
dev_err(xpc_part, "can't setup our reserved page\n");
goto out_2;
}
/* add ourselves to the reboot_notifier_list */
ret = register_reboot_notifier(&xpc_reboot_notifier);
if (ret != 0)
dev_warn(xpc_part, "can't register reboot notifier\n");
/* add ourselves to the die_notifier list */
ret = register_die_notifier(&xpc_die_notifier);
if (ret != 0)
dev_warn(xpc_part, "can't register die notifier\n");
/*
* The real work-horse behind xpc. This processes incoming
* interrupts and monitors remote heartbeats.
*/
kthread = kthread_run(xpc_hb_checker, NULL, XPC_HB_CHECK_THREAD_NAME);
if (IS_ERR(kthread)) {
dev_err(xpc_part, "failed while forking hb check thread\n");
ret = -EBUSY;
goto out_3;
}
/*
* Startup a thread that will attempt to discover other partitions to
* activate based on info provided by SAL. This new thread is short
* lived and will exit once discovery is complete.
*/
kthread = kthread_run(xpc_initiate_discovery, NULL,
XPC_DISCOVERY_THREAD_NAME);
if (IS_ERR(kthread)) {
dev_err(xpc_part, "failed while forking discovery thread\n");
/* mark this new thread as a non-starter */
complete(&xpc_discovery_exited);
xpc_do_exit(xpUnloading);
return -EBUSY;
}
/* set the interface to point at XPC's functions */
xpc_set_interface(xpc_initiate_connect, xpc_initiate_disconnect,
xpc_initiate_send, xpc_initiate_send_notify,
xpc_initiate_received, xpc_initiate_partid_to_nasids);
return 0;
/* initialization was not successful */
out_3:
xpc_teardown_rsvd_page();
(void)unregister_die_notifier(&xpc_die_notifier);
(void)unregister_reboot_notifier(&xpc_reboot_notifier);
out_2:
if (xpc_sysctl)
unregister_sysctl_table(xpc_sysctl);
xpc_teardown_partitions();
out_1:
if (is_shub())
xpc_exit_sn2();
else if (is_uv())
xpc_exit_uv();
return ret;
}
static void
xpc_do_exit(enum xp_retval reason)
{
short partid;
int active_part_count, printed_waiting_msg = 0;
struct xpc_partition *part;
unsigned long printmsg_time, disengage_timeout = 0;
/* a 'rmmod XPC' and a 'reboot' cannot both end up here together */
DBUG_ON(xpc_exiting == 1);
/*
* Let the heartbeat checker thread and the discovery thread
* (if one is running) know that they should exit. Also wake up
* the heartbeat checker thread in case it's sleeping.
*/
xpc_exiting = 1;
wake_up_interruptible(&xpc_activate_IRQ_wq);
/* wait for the discovery thread to exit */
wait_for_completion(&xpc_discovery_exited);
/* wait for the heartbeat checker thread to exit */
wait_for_completion(&xpc_hb_checker_exited);
/* sleep for a 1/3 of a second or so */
(void)msleep_interruptible(300);
/* wait for all partitions to become inactive */
printmsg_time = jiffies + (XPC_DEACTIVATE_PRINTMSG_INTERVAL * HZ);
xpc_disengage_timedout = 0;
do {
active_part_count = 0;
for (partid = 0; partid < xp_max_npartitions; partid++) {
part = &xpc_partitions[partid];
if (xpc_partition_disengaged(part) &&
part->act_state == XPC_P_AS_INACTIVE) {
continue;
}
active_part_count++;
XPC_DEACTIVATE_PARTITION(part, reason);
if (part->disengage_timeout > disengage_timeout)
disengage_timeout = part->disengage_timeout;
}
if (xpc_any_partition_engaged()) {
if (time_is_before_jiffies(printmsg_time)) {
dev_info(xpc_part, "waiting for remote "
"partitions to deactivate, timeout in "
"%ld seconds\n", (disengage_timeout -
jiffies) / HZ);
printmsg_time = jiffies +
(XPC_DEACTIVATE_PRINTMSG_INTERVAL * HZ);
printed_waiting_msg = 1;
}
} else if (active_part_count > 0) {
if (printed_waiting_msg) {
dev_info(xpc_part, "waiting for local partition"
" to deactivate\n");
printed_waiting_msg = 0;
}
} else {
if (!xpc_disengage_timedout) {
dev_info(xpc_part, "all partitions have "
"deactivated\n");
}
break;
}
/* sleep for a 1/3 of a second or so */
(void)msleep_interruptible(300);
} while (1);
DBUG_ON(xpc_any_partition_engaged());
DBUG_ON(xpc_any_hbs_allowed() != 0);
xpc_teardown_rsvd_page();
if (reason == xpUnloading) {
(void)unregister_die_notifier(&xpc_die_notifier);
(void)unregister_reboot_notifier(&xpc_reboot_notifier);
}
/* clear the interface to XPC's functions */
xpc_clear_interface();
if (xpc_sysctl)
unregister_sysctl_table(xpc_sysctl);
xpc_teardown_partitions();
if (is_shub())
xpc_exit_sn2();
else if (is_uv())
xpc_exit_uv();
}
static void
xpc_do_exit(enum xp_retval reason)
{
short partid;
int active_part_count, printed_waiting_msg = 0;
struct xpc_partition *part;
unsigned long printmsg_time, disengage_timeout = 0;
/* */
DBUG_ON(xpc_exiting == 1);
/*
*/
xpc_exiting = 1;
wake_up_interruptible(&xpc_activate_IRQ_wq);
/* */
wait_for_completion(&xpc_discovery_exited);
/* */
wait_for_completion(&xpc_hb_checker_exited);
/* */
(void)msleep_interruptible(300);
/* */
printmsg_time = jiffies + (XPC_DEACTIVATE_PRINTMSG_INTERVAL * HZ);
xpc_disengage_timedout = 0;
do {
active_part_count = 0;
for (partid = 0; partid < xp_max_npartitions; partid++) {
part = &xpc_partitions[partid];
if (xpc_partition_disengaged(part) &&
part->act_state == XPC_P_AS_INACTIVE) {
continue;
}
active_part_count++;
XPC_DEACTIVATE_PARTITION(part, reason);
if (part->disengage_timeout > disengage_timeout)
disengage_timeout = part->disengage_timeout;
}
if (xpc_arch_ops.any_partition_engaged()) {
if (time_is_before_jiffies(printmsg_time)) {
dev_info(xpc_part, "waiting for remote "
"partitions to deactivate, timeout in "
"%ld seconds\n", (disengage_timeout -
jiffies) / HZ);
printmsg_time = jiffies +
(XPC_DEACTIVATE_PRINTMSG_INTERVAL * HZ);
printed_waiting_msg = 1;
}
} else if (active_part_count > 0) {
if (printed_waiting_msg) {
dev_info(xpc_part, "waiting for local partition"
" to deactivate\n");
printed_waiting_msg = 0;
}
} else {
if (!xpc_disengage_timedout) {
dev_info(xpc_part, "all partitions have "
"deactivated\n");
}
break;
}
/* */
(void)msleep_interruptible(300);
} while (1);
DBUG_ON(xpc_arch_ops.any_partition_engaged());
xpc_teardown_rsvd_page();
if (reason == xpUnloading) {
(void)unregister_die_notifier(&xpc_die_notifier);
(void)unregister_reboot_notifier(&xpc_reboot_notifier);
}
/* */
xpc_clear_interface();
if (xpc_sysctl)
unregister_sysctl_table(xpc_sysctl);
xpc_teardown_partitions();
if (is_shub())
xpc_exit_sn2();
else if (is_uv())
xpc_exit_uv();
}
int __init
xpc_init(void)
{
int ret;
struct task_struct *kthread;
dev_set_name(xpc_part, "part");
dev_set_name(xpc_chan, "chan");
if (is_shub()) {
/*
*/
if (xp_max_npartitions != 64) {
dev_err(xpc_part, "max #of partitions not set to 64\n");
ret = -EINVAL;
} else {
ret = xpc_init_sn2();
}
} else if (is_uv()) {
ret = xpc_init_uv();
} else {
ret = -ENODEV;
}
if (ret != 0)
return ret;
ret = xpc_setup_partitions();
if (ret != 0) {
dev_err(xpc_part, "can't get memory for partition structure\n");
goto out_1;
}
xpc_sysctl = register_sysctl_table(xpc_sys_dir);
/*
*/
ret = xpc_setup_rsvd_page();
if (ret != 0) {
dev_err(xpc_part, "can't setup our reserved page\n");
goto out_2;
}
/* */
ret = register_reboot_notifier(&xpc_reboot_notifier);
if (ret != 0)
dev_warn(xpc_part, "can't register reboot notifier\n");
/* */
ret = register_die_notifier(&xpc_die_notifier);
if (ret != 0)
dev_warn(xpc_part, "can't register die notifier\n");
/*
*/
kthread = kthread_run(xpc_hb_checker, NULL, XPC_HB_CHECK_THREAD_NAME);
if (IS_ERR(kthread)) {
dev_err(xpc_part, "failed while forking hb check thread\n");
ret = -EBUSY;
goto out_3;
}
/*
*/
kthread = kthread_run(xpc_initiate_discovery, NULL,
XPC_DISCOVERY_THREAD_NAME);
if (IS_ERR(kthread)) {
dev_err(xpc_part, "failed while forking discovery thread\n");
/* */
complete(&xpc_discovery_exited);
xpc_do_exit(xpUnloading);
return -EBUSY;
}
/* */
xpc_set_interface(xpc_initiate_connect, xpc_initiate_disconnect,
xpc_initiate_send, xpc_initiate_send_notify,
xpc_initiate_received, xpc_initiate_partid_to_nasids);
return 0;
/* */
out_3:
xpc_teardown_rsvd_page();
(void)unregister_die_notifier(&xpc_die_notifier);
(void)unregister_reboot_notifier(&xpc_reboot_notifier);
out_2:
if (xpc_sysctl)
unregister_sysctl_table(xpc_sysctl);
xpc_teardown_partitions();
out_1:
if (is_shub())
xpc_exit_sn2();
else if (is_uv())
xpc_exit_uv();
return ret;
}
/*
* Given a nasid, get the physical address of the partition's reserved page
* for that nasid. This function returns 0 on any error.
*/
static unsigned long
xpc_get_rsvd_page_pa(int nasid)
{
enum xp_retval ret;
u64 cookie = 0;
unsigned long rp_pa = nasid; /* seed with nasid */
size_t len = 0;
size_t buf_len = 0;
void *buf = buf;
void *buf_base = NULL;
enum xp_retval (*get_partition_rsvd_page_pa)
(void *, u64 *, unsigned long *, size_t *) =
xpc_arch_ops.get_partition_rsvd_page_pa;
while (1) {
/* !!! rp_pa will need to be _gpa on UV.
* ??? So do we save it into the architecture specific parts
* ??? of the xpc_partition structure? Do we rename this
* ??? function or have two versions? Rename rp_pa for UV to
* ??? rp_gpa?
*/
ret = get_partition_rsvd_page_pa(buf, &cookie, &rp_pa, &len);
dev_dbg(xpc_part, "SAL returned with ret=%d, cookie=0x%016lx, "
"address=0x%016lx, len=0x%016lx\n", ret,
(unsigned long)cookie, rp_pa, len);
if (ret != xpNeedMoreInfo)
break;
/* !!! L1_CACHE_ALIGN() is only a sn2-bte_copy requirement */
if (is_shub())
len = L1_CACHE_ALIGN(len);
if (len > buf_len) {
if (buf_base != NULL)
kfree(buf_base);
buf_len = L1_CACHE_ALIGN(len);
buf = xpc_kmalloc_cacheline_aligned(buf_len, GFP_KERNEL,
&buf_base);
if (buf_base == NULL) {
dev_err(xpc_part, "unable to kmalloc "
"len=0x%016lx\n", buf_len);
ret = xpNoMemory;
break;
}
}
ret = xp_remote_memcpy(xp_pa(buf), rp_pa, len);
if (ret != xpSuccess) {
dev_dbg(xpc_part, "xp_remote_memcpy failed %d\n", ret);
break;
}
}
kfree(buf_base);
if (ret != xpSuccess)
rp_pa = 0;
dev_dbg(xpc_part, "reserved page at phys address 0x%016lx\n", rp_pa);
return rp_pa;
}