static void
xpc_process_activate_IRQ_rcvd_uv(void)
{
unsigned long irq_flags;
short partid;
struct xpc_partition *part;
u8 act_state_req;
DBUG_ON(xpc_activate_IRQ_rcvd == 0);
spin_lock_irqsave(&xpc_activate_IRQ_rcvd_lock, irq_flags);
for (partid = 0; partid < XP_MAX_NPARTITIONS_UV; partid++) {
part = &xpc_partitions[partid];
if (part->sn.uv.act_state_req == 0)
continue;
xpc_activate_IRQ_rcvd--;
BUG_ON(xpc_activate_IRQ_rcvd < 0);
act_state_req = part->sn.uv.act_state_req;
part->sn.uv.act_state_req = 0;
spin_unlock_irqrestore(&xpc_activate_IRQ_rcvd_lock, irq_flags);
if (act_state_req == XPC_P_ASR_ACTIVATE_UV) {
if (part->act_state == XPC_P_AS_INACTIVE)
xpc_activate_partition(part);
else if (part->act_state == XPC_P_AS_DEACTIVATING)
XPC_DEACTIVATE_PARTITION(part, xpReactivating);
} else if (act_state_req == XPC_P_ASR_REACTIVATE_UV) {
if (part->act_state == XPC_P_AS_INACTIVE)
xpc_activate_partition(part);
else
XPC_DEACTIVATE_PARTITION(part, xpReactivating);
} else if (act_state_req == XPC_P_ASR_DEACTIVATE_UV) {
XPC_DEACTIVATE_PARTITION(part, part->sn.uv.reason);
} else {
BUG();
}
spin_lock_irqsave(&xpc_activate_IRQ_rcvd_lock, irq_flags);
if (xpc_activate_IRQ_rcvd == 0)
break;
}
spin_unlock_irqrestore(&xpc_activate_IRQ_rcvd_lock, irq_flags);
}
/*
* At periodic intervals, scan through all active partitions and ensure
* their heartbeat is still active. If not, the partition is deactivated.
*/
static void
xpc_check_remote_hb(void)
{
struct xpc_partition *part;
short partid;
enum xp_retval ret;
for (partid = 0; partid < xp_max_npartitions; partid++) {
if (xpc_exiting)
break;
if (partid == xp_partition_id)
continue;
part = &xpc_partitions[partid];
if (part->act_state == XPC_P_AS_INACTIVE ||
part->act_state == XPC_P_AS_DEACTIVATING) {
continue;
}
ret = xpc_get_remote_heartbeat(part);
if (ret != xpSuccess)
XPC_DEACTIVATE_PARTITION(part, ret);
}
}
static void
xpc_send_activate_IRQ_part_uv(struct xpc_partition *part, void *msg,
size_t msg_size, int msg_type)
{
enum xp_retval ret;
ret = xpc_send_activate_IRQ_uv(part, msg, msg_size, msg_type);
if (unlikely(ret != xpSuccess))
XPC_DEACTIVATE_PARTITION(part, ret);
}
static void
xpc_send_activate_IRQ_ch_uv(struct xpc_channel *ch, unsigned long *irq_flags,
void *msg, size_t msg_size, int msg_type)
{
struct xpc_partition *part = &xpc_partitions[ch->partid];
enum xp_retval ret;
ret = xpc_send_activate_IRQ_uv(part, msg, msg_size, msg_type);
if (unlikely(ret != xpSuccess)) {
if (irq_flags != NULL)
spin_unlock_irqrestore(&ch->lock, *irq_flags);
XPC_DEACTIVATE_PARTITION(part, ret);
if (irq_flags != NULL)
spin_lock_irqsave(&ch->lock, *irq_flags);
}
}
/*
* Establish first contact with the remote partititon. This involves pulling
* the XPC per partition variables from the remote partition and waiting for
* the remote partition to pull ours.
*/
static enum xp_retval
xpc_make_first_contact(struct xpc_partition *part)
{
enum xp_retval ret;
while ((ret = xpc_pull_remote_vars_part(part)) != xpSuccess) {
if (ret != xpRetry) {
XPC_DEACTIVATE_PARTITION(part, ret);
return ret;
}
dev_dbg(xpc_chan, "waiting to make first contact with "
"partition %d\n", XPC_PARTID(part));
/* wait a 1/4 of a second or so */
(void)msleep_interruptible(250);
if (part->act_state == XPC_P_DEACTIVATING)
return part->reason;
}
return xpc_mark_partition_active(part);
}
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();
}
static enum xp_retval
xpc_send_payload_uv(struct xpc_channel *ch, u32 flags, void *payload,
u16 payload_size, u8 notify_type, xpc_notify_func func,
void *key)
{
enum xp_retval ret = xpSuccess;
struct xpc_send_msg_slot_uv *msg_slot = NULL;
struct xpc_notify_mq_msg_uv *msg;
u8 msg_buffer[XPC_NOTIFY_MSG_SIZE_UV];
size_t msg_size;
DBUG_ON(notify_type != XPC_N_CALL);
msg_size = sizeof(struct xpc_notify_mq_msghdr_uv) + payload_size;
if (msg_size > ch->entry_size)
return xpPayloadTooBig;
xpc_msgqueue_ref(ch);
if (ch->flags & XPC_C_DISCONNECTING) {
ret = ch->reason;
goto out_1;
}
if (!(ch->flags & XPC_C_CONNECTED)) {
ret = xpNotConnected;
goto out_1;
}
ret = xpc_allocate_msg_slot_uv(ch, flags, &msg_slot);
if (ret != xpSuccess)
goto out_1;
if (func != NULL) {
atomic_inc(&ch->n_to_notify);
msg_slot->key = key;
smp_wmb(); /* a non-NULL func must hit memory after the key */
msg_slot->func = func;
if (ch->flags & XPC_C_DISCONNECTING) {
ret = ch->reason;
goto out_2;
}
}
msg = (struct xpc_notify_mq_msg_uv *)&msg_buffer;
msg->hdr.partid = xp_partition_id;
msg->hdr.ch_number = ch->number;
msg->hdr.size = msg_size;
msg->hdr.msg_slot_number = msg_slot->msg_slot_number;
memcpy(&msg->payload, payload, payload_size);
ret = xpc_send_gru_msg(ch->sn.uv.cached_notify_gru_mq_desc, msg,
msg_size);
if (ret == xpSuccess)
goto out_1;
XPC_DEACTIVATE_PARTITION(&xpc_partitions[ch->partid], ret);
out_2:
if (func != NULL) {
/*
* Try to NULL the msg_slot's func field. If we fail, then
* xpc_notify_senders_of_disconnect_uv() beat us to it, in which
* case we need to pretend we succeeded to send the message
* since the user will get a callout for the disconnect error
* by xpc_notify_senders_of_disconnect_uv(), and to also get an
* error returned here will confuse them. Additionally, since
* in this case the channel is being disconnected we don't need
* to put the the msg_slot back on the free list.
*/
if (cmpxchg(&msg_slot->func, func, NULL) != func) {
ret = xpSuccess;
goto out_1;
}
msg_slot->key = NULL;
atomic_dec(&ch->n_to_notify);
}
xpc_free_msg_slot_uv(ch, msg_slot);
out_1:
xpc_msgqueue_deref(ch);
return ret;
}
static enum xp_retval
xpc_send_payload_uv(struct xpc_channel *ch, u32 flags, void *payload,
u16 payload_size, u8 notify_type, xpc_notify_func func,
void *key)
{
enum xp_retval ret = xpSuccess;
struct xpc_send_msg_slot_uv *msg_slot = NULL;
struct xpc_notify_mq_msg_uv *msg;
u8 msg_buffer[XPC_NOTIFY_MSG_SIZE_UV];
size_t msg_size;
DBUG_ON(notify_type != XPC_N_CALL);
msg_size = sizeof(struct xpc_notify_mq_msghdr_uv) + payload_size;
if (msg_size > ch->entry_size)
return xpPayloadTooBig;
xpc_msgqueue_ref(ch);
if (ch->flags & XPC_C_DISCONNECTING) {
ret = ch->reason;
goto out_1;
}
if (!(ch->flags & XPC_C_CONNECTED)) {
ret = xpNotConnected;
goto out_1;
}
ret = xpc_allocate_msg_slot_uv(ch, flags, &msg_slot);
if (ret != xpSuccess)
goto out_1;
if (func != NULL) {
atomic_inc(&ch->n_to_notify);
msg_slot->key = key;
smp_wmb();
msg_slot->func = func;
if (ch->flags & XPC_C_DISCONNECTING) {
ret = ch->reason;
goto out_2;
}
}
msg = (struct xpc_notify_mq_msg_uv *)&msg_buffer;
msg->hdr.partid = xp_partition_id;
msg->hdr.ch_number = ch->number;
msg->hdr.size = msg_size;
msg->hdr.msg_slot_number = msg_slot->msg_slot_number;
memcpy(&msg->payload, payload, payload_size);
ret = xpc_send_gru_msg(ch->sn.uv.cached_notify_gru_mq_desc, msg,
msg_size);
if (ret == xpSuccess)
goto out_1;
XPC_DEACTIVATE_PARTITION(&xpc_partitions[ch->partid], ret);
out_2:
if (func != NULL) {
if (cmpxchg(&msg_slot->func, func, NULL) != func) {
ret = xpSuccess;
goto out_1;
}
msg_slot->key = NULL;
atomic_dec(&ch->n_to_notify);
}
xpc_free_msg_slot_uv(ch, msg_slot);
out_1:
xpc_msgqueue_deref(ch);
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_request_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_act_IRQ_wq);
/* ignore all incoming interrupts */
free_irq(SGI_XPC_ACTIVATE, NULL);
/* 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_DISENGAGE_PRINTMSG_INTERVAL * HZ);
xpc_disengage_request_timedout = 0;
do {
active_part_count = 0;
for (partid = 1; partid < XP_MAX_PARTITIONS; partid++) {
part = &xpc_partitions[partid];
if (xpc_partition_disengaged(part) &&
part->act_state == XPC_P_INACTIVE) {
continue;
}
active_part_count++;
XPC_DEACTIVATE_PARTITION(part, reason);
if (part->disengage_request_timeout >
disengage_request_timeout) {
disengage_request_timeout =
part->disengage_request_timeout;
}
}
if (xpc_partition_engaged(-1UL)) {
if (time_after(jiffies, printmsg_time)) {
dev_info(xpc_part, "waiting for remote "
"partitions to disengage, timeout in "
"%ld seconds\n",
(disengage_request_timeout - jiffies)
/ HZ);
printmsg_time = jiffies +
(XPC_DISENGAGE_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 disengage\n");
printed_waiting_msg = 0;
}
} else {
if (!xpc_disengage_request_timedout) {
dev_info(xpc_part, "all partitions have "
"disengaged\n");
}
break;
}
/* sleep for a 1/3 of a second or so */
(void)msleep_interruptible(300);
} while (1);
DBUG_ON(xpc_partition_engaged(-1UL));
/* indicate to others that our reserved page is uninitialized */
xpc_rsvd_page->vars_pa = 0;
/* now it's time to eliminate our heartbeat */
del_timer_sync(&xpc_hb_timer);
DBUG_ON(xpc_vars->heartbeating_to_mask != 0);
if (reason == xpUnloading) {
/* take ourselves off of the reboot_notifier_list */
(void)unregister_reboot_notifier(&xpc_reboot_notifier);
/* take ourselves off of the die_notifier list */
(void)unregister_die_notifier(&xpc_die_notifier);
}
/* close down protections for IPI operations */
xpc_restrict_IPI_ops();
/* clear the interface to XPC's functions */
xpc_clear_interface();
if (xpc_sysctl)
unregister_sysctl_table(xpc_sysctl);
kfree(xpc_remote_copy_buffer_base);
}
