void __exit
xp_exit(void)
{
if (is_shub())
xp_exit_sn2();
else if (is_uv())
xp_exit_uv();
}
enum xp_retval
xp_init_uv(void)
{
BUG_ON(!is_uv());
xp_max_npartitions = XP_MAX_NPARTITIONS_UV;
xp_partition_id = sn_partition_id;
xp_region_size = sn_region_size;
xp_pa = xp_pa_uv;
xp_socket_pa = xp_socket_pa_uv;
xp_remote_memcpy = xp_remote_memcpy_uv;
xp_cpu_to_nasid = xp_cpu_to_nasid_uv;
xp_expand_memprotect = xp_expand_memprotect_uv;
xp_restrict_memprotect = xp_restrict_memprotect_uv;
return xpSuccess;
}
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();
}
void
xp_exit_uv(void)
{
BUG_ON(!is_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;
}
/*
* SAL has provided a partition and machine mask. The partition mask
* contains a bit for each even nasid in our partition. The machine
* mask contains a bit for each even nasid in the entire machine.
*
* Using those two bit arrays, we can determine which nasids are
* known in the machine. Each should also have a reserved page
* initialized if they are available for partitioning.
*/
void
xpc_discovery(void)
{
void *remote_rp_base;
struct xpc_rsvd_page *remote_rp;
unsigned long remote_rp_pa;
int region;
int region_size;
int max_regions;
int nasid;
struct xpc_rsvd_page *rp;
unsigned long *discovered_nasids;
enum xp_retval ret;
remote_rp = xpc_kmalloc_cacheline_aligned(XPC_RP_HEADER_SIZE +
xpc_nasid_mask_nbytes,
GFP_KERNEL, &remote_rp_base);
if (remote_rp == NULL)
return;
discovered_nasids = kzalloc(sizeof(long) * xpc_nasid_mask_nlongs,
GFP_KERNEL);
if (discovered_nasids == NULL) {
kfree(remote_rp_base);
return;
}
rp = (struct xpc_rsvd_page *)xpc_rsvd_page;
/*
* The term 'region' in this context refers to the minimum number of
* nodes that can comprise an access protection grouping. The access
* protection is in regards to memory, IOI and IPI.
*/
region_size = xp_region_size;
if (is_uv())
max_regions = 256;
else {
max_regions = 64;
switch (region_size) {
case 128:
max_regions *= 2;
case 64:
max_regions *= 2;
case 32:
max_regions *= 2;
region_size = 16;
DBUG_ON(!is_shub2());
}
}
for (region = 0; region < max_regions; region++) {
if (xpc_exiting)
break;
dev_dbg(xpc_part, "searching region %d\n", region);
for (nasid = (region * region_size * 2);
nasid < ((region + 1) * region_size * 2); nasid += 2) {
if (xpc_exiting)
break;
dev_dbg(xpc_part, "checking nasid %d\n", nasid);
if (test_bit(nasid / 2, xpc_part_nasids)) {
dev_dbg(xpc_part, "PROM indicates Nasid %d is "
"part of the local partition; skipping "
"region\n", nasid);
break;
}
if (!(test_bit(nasid / 2, xpc_mach_nasids))) {
dev_dbg(xpc_part, "PROM indicates Nasid %d was "
"not on Numa-Link network at reset\n",
nasid);
continue;
}
if (test_bit(nasid / 2, discovered_nasids)) {
dev_dbg(xpc_part, "Nasid %d is part of a "
"partition which was previously "
"discovered\n", nasid);
continue;
}
/* pull over the rsvd page header & part_nasids mask */
ret = xpc_get_remote_rp(nasid, discovered_nasids,
remote_rp, &remote_rp_pa);
if (ret != xpSuccess) {
dev_dbg(xpc_part, "unable to get reserved page "
"from nasid %d, reason=%d\n", nasid,
ret);
if (ret == xpLocalPartid)
break;
continue;
}
xpc_arch_ops.request_partition_activation(remote_rp,
remote_rp_pa, nasid);
}
}
kfree(discovered_nasids);
kfree(remote_rp_base);
}
