mirror of
https://mirrors.bfsu.edu.cn/git/linux.git
synced 2024-12-05 18:14:07 +08:00
0b0d1173d8
Qemu before version 2.11 does not implement the architecture correctly, and does not allow for a mask size of size different than 4. This patch introduces a compatibility mode for such systems, forcing the mask sizes to 4. Since the mask size is currently still 4 anyway, this patch should have no impact whatsoever by itself, but it will be needed when the mask size is increased to 64 bits in the next patch. Reviewed-by: Heiko Carstens <heiko.carstens@de.ibm.com> Signed-off-by: Claudio Imbrenda <imbrenda@linux.vnet.ibm.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
1260 lines
34 KiB
C
1260 lines
34 KiB
C
// SPDX-License-Identifier: GPL-2.0
|
|
/*
|
|
* core function to access sclp interface
|
|
*
|
|
* Copyright IBM Corp. 1999, 2009
|
|
*
|
|
* Author(s): Martin Peschke <mpeschke@de.ibm.com>
|
|
* Martin Schwidefsky <schwidefsky@de.ibm.com>
|
|
*/
|
|
|
|
#include <linux/kernel_stat.h>
|
|
#include <linux/module.h>
|
|
#include <linux/err.h>
|
|
#include <linux/spinlock.h>
|
|
#include <linux/interrupt.h>
|
|
#include <linux/timer.h>
|
|
#include <linux/reboot.h>
|
|
#include <linux/jiffies.h>
|
|
#include <linux/init.h>
|
|
#include <linux/suspend.h>
|
|
#include <linux/completion.h>
|
|
#include <linux/platform_device.h>
|
|
#include <asm/types.h>
|
|
#include <asm/irq.h>
|
|
|
|
#include "sclp.h"
|
|
|
|
#define SCLP_HEADER "sclp: "
|
|
|
|
/* Lock to protect internal data consistency. */
|
|
static DEFINE_SPINLOCK(sclp_lock);
|
|
|
|
/* Mask of events that we can send to the sclp interface. */
|
|
static sccb_mask_t sclp_receive_mask;
|
|
|
|
/* Mask of events that we can receive from the sclp interface. */
|
|
static sccb_mask_t sclp_send_mask;
|
|
|
|
/* List of registered event listeners and senders. */
|
|
static struct list_head sclp_reg_list;
|
|
|
|
/* List of queued requests. */
|
|
static struct list_head sclp_req_queue;
|
|
|
|
/* Data for read and and init requests. */
|
|
static struct sclp_req sclp_read_req;
|
|
static struct sclp_req sclp_init_req;
|
|
static char sclp_read_sccb[PAGE_SIZE] __attribute__((__aligned__(PAGE_SIZE)));
|
|
static char sclp_init_sccb[PAGE_SIZE] __attribute__((__aligned__(PAGE_SIZE)));
|
|
|
|
/* Suspend request */
|
|
static DECLARE_COMPLETION(sclp_request_queue_flushed);
|
|
|
|
/* Number of console pages to allocate, used by sclp_con.c and sclp_vt220.c */
|
|
int sclp_console_pages = SCLP_CONSOLE_PAGES;
|
|
/* Flag to indicate if buffer pages are dropped on buffer full condition */
|
|
int sclp_console_drop = 1;
|
|
/* Number of times the console dropped buffer pages */
|
|
unsigned long sclp_console_full;
|
|
|
|
static void sclp_suspend_req_cb(struct sclp_req *req, void *data)
|
|
{
|
|
complete(&sclp_request_queue_flushed);
|
|
}
|
|
|
|
static int __init sclp_setup_console_pages(char *str)
|
|
{
|
|
int pages, rc;
|
|
|
|
rc = kstrtoint(str, 0, &pages);
|
|
if (!rc && pages >= SCLP_CONSOLE_PAGES)
|
|
sclp_console_pages = pages;
|
|
return 1;
|
|
}
|
|
|
|
__setup("sclp_con_pages=", sclp_setup_console_pages);
|
|
|
|
static int __init sclp_setup_console_drop(char *str)
|
|
{
|
|
int drop, rc;
|
|
|
|
rc = kstrtoint(str, 0, &drop);
|
|
if (!rc)
|
|
sclp_console_drop = drop;
|
|
return 1;
|
|
}
|
|
|
|
__setup("sclp_con_drop=", sclp_setup_console_drop);
|
|
|
|
static struct sclp_req sclp_suspend_req;
|
|
|
|
/* Timer for request retries. */
|
|
static struct timer_list sclp_request_timer;
|
|
|
|
/* Timer for queued requests. */
|
|
static struct timer_list sclp_queue_timer;
|
|
|
|
/* Internal state: is a request active at the sclp? */
|
|
static volatile enum sclp_running_state_t {
|
|
sclp_running_state_idle,
|
|
sclp_running_state_running,
|
|
sclp_running_state_reset_pending
|
|
} sclp_running_state = sclp_running_state_idle;
|
|
|
|
/* Internal state: is a read request pending? */
|
|
static volatile enum sclp_reading_state_t {
|
|
sclp_reading_state_idle,
|
|
sclp_reading_state_reading
|
|
} sclp_reading_state = sclp_reading_state_idle;
|
|
|
|
/* Internal state: is the driver currently serving requests? */
|
|
static volatile enum sclp_activation_state_t {
|
|
sclp_activation_state_active,
|
|
sclp_activation_state_deactivating,
|
|
sclp_activation_state_inactive,
|
|
sclp_activation_state_activating
|
|
} sclp_activation_state = sclp_activation_state_active;
|
|
|
|
/* Internal state: is an init mask request pending? */
|
|
static volatile enum sclp_mask_state_t {
|
|
sclp_mask_state_idle,
|
|
sclp_mask_state_initializing
|
|
} sclp_mask_state = sclp_mask_state_idle;
|
|
|
|
/* Internal state: is the driver suspended? */
|
|
static enum sclp_suspend_state_t {
|
|
sclp_suspend_state_running,
|
|
sclp_suspend_state_suspended,
|
|
} sclp_suspend_state = sclp_suspend_state_running;
|
|
|
|
/* Maximum retry counts */
|
|
#define SCLP_INIT_RETRY 3
|
|
#define SCLP_MASK_RETRY 3
|
|
|
|
/* Timeout intervals in seconds.*/
|
|
#define SCLP_BUSY_INTERVAL 10
|
|
#define SCLP_RETRY_INTERVAL 30
|
|
|
|
static void sclp_request_timeout(bool force_restart);
|
|
static void sclp_process_queue(void);
|
|
static void __sclp_make_read_req(void);
|
|
static int sclp_init_mask(int calculate);
|
|
static int sclp_init(void);
|
|
|
|
static void
|
|
__sclp_queue_read_req(void)
|
|
{
|
|
if (sclp_reading_state == sclp_reading_state_idle) {
|
|
sclp_reading_state = sclp_reading_state_reading;
|
|
__sclp_make_read_req();
|
|
/* Add request to head of queue */
|
|
list_add(&sclp_read_req.list, &sclp_req_queue);
|
|
}
|
|
}
|
|
|
|
/* Set up request retry timer. Called while sclp_lock is locked. */
|
|
static inline void
|
|
__sclp_set_request_timer(unsigned long time, void (*cb)(struct timer_list *))
|
|
{
|
|
del_timer(&sclp_request_timer);
|
|
sclp_request_timer.function = cb;
|
|
sclp_request_timer.expires = jiffies + time;
|
|
add_timer(&sclp_request_timer);
|
|
}
|
|
|
|
static void sclp_request_timeout_restart(struct timer_list *unused)
|
|
{
|
|
sclp_request_timeout(true);
|
|
}
|
|
|
|
static void sclp_request_timeout_normal(struct timer_list *unused)
|
|
{
|
|
sclp_request_timeout(false);
|
|
}
|
|
|
|
/* Request timeout handler. Restart the request queue. If force_restart,
|
|
* force restart of running request. */
|
|
static void sclp_request_timeout(bool force_restart)
|
|
{
|
|
unsigned long flags;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
if (force_restart) {
|
|
if (sclp_running_state == sclp_running_state_running) {
|
|
/* Break running state and queue NOP read event request
|
|
* to get a defined interface state. */
|
|
__sclp_queue_read_req();
|
|
sclp_running_state = sclp_running_state_idle;
|
|
}
|
|
} else {
|
|
__sclp_set_request_timer(SCLP_BUSY_INTERVAL * HZ,
|
|
sclp_request_timeout_normal);
|
|
}
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
sclp_process_queue();
|
|
}
|
|
|
|
/*
|
|
* Returns the expire value in jiffies of the next pending request timeout,
|
|
* if any. Needs to be called with sclp_lock.
|
|
*/
|
|
static unsigned long __sclp_req_queue_find_next_timeout(void)
|
|
{
|
|
unsigned long expires_next = 0;
|
|
struct sclp_req *req;
|
|
|
|
list_for_each_entry(req, &sclp_req_queue, list) {
|
|
if (!req->queue_expires)
|
|
continue;
|
|
if (!expires_next ||
|
|
(time_before(req->queue_expires, expires_next)))
|
|
expires_next = req->queue_expires;
|
|
}
|
|
return expires_next;
|
|
}
|
|
|
|
/*
|
|
* Returns expired request, if any, and removes it from the list.
|
|
*/
|
|
static struct sclp_req *__sclp_req_queue_remove_expired_req(void)
|
|
{
|
|
unsigned long flags, now;
|
|
struct sclp_req *req;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
now = jiffies;
|
|
/* Don't need list_for_each_safe because we break out after list_del */
|
|
list_for_each_entry(req, &sclp_req_queue, list) {
|
|
if (!req->queue_expires)
|
|
continue;
|
|
if (time_before_eq(req->queue_expires, now)) {
|
|
if (req->status == SCLP_REQ_QUEUED) {
|
|
req->status = SCLP_REQ_QUEUED_TIMEOUT;
|
|
list_del(&req->list);
|
|
goto out;
|
|
}
|
|
}
|
|
}
|
|
req = NULL;
|
|
out:
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return req;
|
|
}
|
|
|
|
/*
|
|
* Timeout handler for queued requests. Removes request from list and
|
|
* invokes callback. This timer can be set per request in situations where
|
|
* waiting too long would be harmful to the system, e.g. during SE reboot.
|
|
*/
|
|
static void sclp_req_queue_timeout(struct timer_list *unused)
|
|
{
|
|
unsigned long flags, expires_next;
|
|
struct sclp_req *req;
|
|
|
|
do {
|
|
req = __sclp_req_queue_remove_expired_req();
|
|
if (req && req->callback)
|
|
req->callback(req, req->callback_data);
|
|
} while (req);
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
expires_next = __sclp_req_queue_find_next_timeout();
|
|
if (expires_next)
|
|
mod_timer(&sclp_queue_timer, expires_next);
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
}
|
|
|
|
/* Try to start a request. Return zero if the request was successfully
|
|
* started or if it will be started at a later time. Return non-zero otherwise.
|
|
* Called while sclp_lock is locked. */
|
|
static int
|
|
__sclp_start_request(struct sclp_req *req)
|
|
{
|
|
int rc;
|
|
|
|
if (sclp_running_state != sclp_running_state_idle)
|
|
return 0;
|
|
del_timer(&sclp_request_timer);
|
|
rc = sclp_service_call(req->command, req->sccb);
|
|
req->start_count++;
|
|
|
|
if (rc == 0) {
|
|
/* Successfully started request */
|
|
req->status = SCLP_REQ_RUNNING;
|
|
sclp_running_state = sclp_running_state_running;
|
|
__sclp_set_request_timer(SCLP_RETRY_INTERVAL * HZ,
|
|
sclp_request_timeout_restart);
|
|
return 0;
|
|
} else if (rc == -EBUSY) {
|
|
/* Try again later */
|
|
__sclp_set_request_timer(SCLP_BUSY_INTERVAL * HZ,
|
|
sclp_request_timeout_normal);
|
|
return 0;
|
|
}
|
|
/* Request failed */
|
|
req->status = SCLP_REQ_FAILED;
|
|
return rc;
|
|
}
|
|
|
|
/* Try to start queued requests. */
|
|
static void
|
|
sclp_process_queue(void)
|
|
{
|
|
struct sclp_req *req;
|
|
int rc;
|
|
unsigned long flags;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
if (sclp_running_state != sclp_running_state_idle) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return;
|
|
}
|
|
del_timer(&sclp_request_timer);
|
|
while (!list_empty(&sclp_req_queue)) {
|
|
req = list_entry(sclp_req_queue.next, struct sclp_req, list);
|
|
if (!req->sccb)
|
|
goto do_post;
|
|
rc = __sclp_start_request(req);
|
|
if (rc == 0)
|
|
break;
|
|
/* Request failed */
|
|
if (req->start_count > 1) {
|
|
/* Cannot abort already submitted request - could still
|
|
* be active at the SCLP */
|
|
__sclp_set_request_timer(SCLP_BUSY_INTERVAL * HZ,
|
|
sclp_request_timeout_normal);
|
|
break;
|
|
}
|
|
do_post:
|
|
/* Post-processing for aborted request */
|
|
list_del(&req->list);
|
|
if (req->callback) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
req->callback(req, req->callback_data);
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
}
|
|
}
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
}
|
|
|
|
static int __sclp_can_add_request(struct sclp_req *req)
|
|
{
|
|
if (req == &sclp_suspend_req || req == &sclp_init_req)
|
|
return 1;
|
|
if (sclp_suspend_state != sclp_suspend_state_running)
|
|
return 0;
|
|
if (sclp_init_state != sclp_init_state_initialized)
|
|
return 0;
|
|
if (sclp_activation_state != sclp_activation_state_active)
|
|
return 0;
|
|
return 1;
|
|
}
|
|
|
|
/* Queue a new request. Return zero on success, non-zero otherwise. */
|
|
int
|
|
sclp_add_request(struct sclp_req *req)
|
|
{
|
|
unsigned long flags;
|
|
int rc;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
if (!__sclp_can_add_request(req)) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return -EIO;
|
|
}
|
|
req->status = SCLP_REQ_QUEUED;
|
|
req->start_count = 0;
|
|
list_add_tail(&req->list, &sclp_req_queue);
|
|
rc = 0;
|
|
if (req->queue_timeout) {
|
|
req->queue_expires = jiffies + req->queue_timeout * HZ;
|
|
if (!timer_pending(&sclp_queue_timer) ||
|
|
time_after(sclp_queue_timer.expires, req->queue_expires))
|
|
mod_timer(&sclp_queue_timer, req->queue_expires);
|
|
} else
|
|
req->queue_expires = 0;
|
|
/* Start if request is first in list */
|
|
if (sclp_running_state == sclp_running_state_idle &&
|
|
req->list.prev == &sclp_req_queue) {
|
|
if (!req->sccb) {
|
|
list_del(&req->list);
|
|
rc = -ENODATA;
|
|
goto out;
|
|
}
|
|
rc = __sclp_start_request(req);
|
|
if (rc)
|
|
list_del(&req->list);
|
|
}
|
|
out:
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return rc;
|
|
}
|
|
|
|
EXPORT_SYMBOL(sclp_add_request);
|
|
|
|
/* Dispatch events found in request buffer to registered listeners. Return 0
|
|
* if all events were dispatched, non-zero otherwise. */
|
|
static int
|
|
sclp_dispatch_evbufs(struct sccb_header *sccb)
|
|
{
|
|
unsigned long flags;
|
|
struct evbuf_header *evbuf;
|
|
struct list_head *l;
|
|
struct sclp_register *reg;
|
|
int offset;
|
|
int rc;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
rc = 0;
|
|
for (offset = sizeof(struct sccb_header); offset < sccb->length;
|
|
offset += evbuf->length) {
|
|
evbuf = (struct evbuf_header *) ((addr_t) sccb + offset);
|
|
/* Check for malformed hardware response */
|
|
if (evbuf->length == 0)
|
|
break;
|
|
/* Search for event handler */
|
|
reg = NULL;
|
|
list_for_each(l, &sclp_reg_list) {
|
|
reg = list_entry(l, struct sclp_register, list);
|
|
if (reg->receive_mask & SCLP_EVTYP_MASK(evbuf->type))
|
|
break;
|
|
else
|
|
reg = NULL;
|
|
}
|
|
if (reg && reg->receiver_fn) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
reg->receiver_fn(evbuf);
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
} else if (reg == NULL)
|
|
rc = -EOPNOTSUPP;
|
|
}
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return rc;
|
|
}
|
|
|
|
/* Read event data request callback. */
|
|
static void
|
|
sclp_read_cb(struct sclp_req *req, void *data)
|
|
{
|
|
unsigned long flags;
|
|
struct sccb_header *sccb;
|
|
|
|
sccb = (struct sccb_header *) req->sccb;
|
|
if (req->status == SCLP_REQ_DONE && (sccb->response_code == 0x20 ||
|
|
sccb->response_code == 0x220))
|
|
sclp_dispatch_evbufs(sccb);
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
sclp_reading_state = sclp_reading_state_idle;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
}
|
|
|
|
/* Prepare read event data request. Called while sclp_lock is locked. */
|
|
static void __sclp_make_read_req(void)
|
|
{
|
|
struct sccb_header *sccb;
|
|
|
|
sccb = (struct sccb_header *) sclp_read_sccb;
|
|
clear_page(sccb);
|
|
memset(&sclp_read_req, 0, sizeof(struct sclp_req));
|
|
sclp_read_req.command = SCLP_CMDW_READ_EVENT_DATA;
|
|
sclp_read_req.status = SCLP_REQ_QUEUED;
|
|
sclp_read_req.start_count = 0;
|
|
sclp_read_req.callback = sclp_read_cb;
|
|
sclp_read_req.sccb = sccb;
|
|
sccb->length = PAGE_SIZE;
|
|
sccb->function_code = 0;
|
|
sccb->control_mask[2] = 0x80;
|
|
}
|
|
|
|
/* Search request list for request with matching sccb. Return request if found,
|
|
* NULL otherwise. Called while sclp_lock is locked. */
|
|
static inline struct sclp_req *
|
|
__sclp_find_req(u32 sccb)
|
|
{
|
|
struct list_head *l;
|
|
struct sclp_req *req;
|
|
|
|
list_for_each(l, &sclp_req_queue) {
|
|
req = list_entry(l, struct sclp_req, list);
|
|
if (sccb == (u32) (addr_t) req->sccb)
|
|
return req;
|
|
}
|
|
return NULL;
|
|
}
|
|
|
|
/* Handler for external interruption. Perform request post-processing.
|
|
* Prepare read event data request if necessary. Start processing of next
|
|
* request on queue. */
|
|
static void sclp_interrupt_handler(struct ext_code ext_code,
|
|
unsigned int param32, unsigned long param64)
|
|
{
|
|
struct sclp_req *req;
|
|
u32 finished_sccb;
|
|
u32 evbuf_pending;
|
|
|
|
inc_irq_stat(IRQEXT_SCP);
|
|
spin_lock(&sclp_lock);
|
|
finished_sccb = param32 & 0xfffffff8;
|
|
evbuf_pending = param32 & 0x3;
|
|
if (finished_sccb) {
|
|
del_timer(&sclp_request_timer);
|
|
sclp_running_state = sclp_running_state_reset_pending;
|
|
req = __sclp_find_req(finished_sccb);
|
|
if (req) {
|
|
/* Request post-processing */
|
|
list_del(&req->list);
|
|
req->status = SCLP_REQ_DONE;
|
|
if (req->callback) {
|
|
spin_unlock(&sclp_lock);
|
|
req->callback(req, req->callback_data);
|
|
spin_lock(&sclp_lock);
|
|
}
|
|
}
|
|
sclp_running_state = sclp_running_state_idle;
|
|
}
|
|
if (evbuf_pending &&
|
|
sclp_activation_state == sclp_activation_state_active)
|
|
__sclp_queue_read_req();
|
|
spin_unlock(&sclp_lock);
|
|
sclp_process_queue();
|
|
}
|
|
|
|
/* Convert interval in jiffies to TOD ticks. */
|
|
static inline u64
|
|
sclp_tod_from_jiffies(unsigned long jiffies)
|
|
{
|
|
return (u64) (jiffies / HZ) << 32;
|
|
}
|
|
|
|
/* Wait until a currently running request finished. Note: while this function
|
|
* is running, no timers are served on the calling CPU. */
|
|
void
|
|
sclp_sync_wait(void)
|
|
{
|
|
unsigned long long old_tick;
|
|
unsigned long flags;
|
|
unsigned long cr0, cr0_sync;
|
|
u64 timeout;
|
|
int irq_context;
|
|
|
|
/* We'll be disabling timer interrupts, so we need a custom timeout
|
|
* mechanism */
|
|
timeout = 0;
|
|
if (timer_pending(&sclp_request_timer)) {
|
|
/* Get timeout TOD value */
|
|
timeout = get_tod_clock_fast() +
|
|
sclp_tod_from_jiffies(sclp_request_timer.expires -
|
|
jiffies);
|
|
}
|
|
local_irq_save(flags);
|
|
/* Prevent bottom half from executing once we force interrupts open */
|
|
irq_context = in_interrupt();
|
|
if (!irq_context)
|
|
local_bh_disable();
|
|
/* Enable service-signal interruption, disable timer interrupts */
|
|
old_tick = local_tick_disable();
|
|
trace_hardirqs_on();
|
|
__ctl_store(cr0, 0, 0);
|
|
cr0_sync = cr0 & ~CR0_IRQ_SUBCLASS_MASK;
|
|
cr0_sync |= 1UL << (63 - 54);
|
|
__ctl_load(cr0_sync, 0, 0);
|
|
__arch_local_irq_stosm(0x01);
|
|
/* Loop until driver state indicates finished request */
|
|
while (sclp_running_state != sclp_running_state_idle) {
|
|
/* Check for expired request timer */
|
|
if (timer_pending(&sclp_request_timer) &&
|
|
get_tod_clock_fast() > timeout &&
|
|
del_timer(&sclp_request_timer))
|
|
sclp_request_timer.function(&sclp_request_timer);
|
|
cpu_relax();
|
|
}
|
|
local_irq_disable();
|
|
__ctl_load(cr0, 0, 0);
|
|
if (!irq_context)
|
|
_local_bh_enable();
|
|
local_tick_enable(old_tick);
|
|
local_irq_restore(flags);
|
|
}
|
|
EXPORT_SYMBOL(sclp_sync_wait);
|
|
|
|
/* Dispatch changes in send and receive mask to registered listeners. */
|
|
static void
|
|
sclp_dispatch_state_change(void)
|
|
{
|
|
struct list_head *l;
|
|
struct sclp_register *reg;
|
|
unsigned long flags;
|
|
sccb_mask_t receive_mask;
|
|
sccb_mask_t send_mask;
|
|
|
|
do {
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
reg = NULL;
|
|
list_for_each(l, &sclp_reg_list) {
|
|
reg = list_entry(l, struct sclp_register, list);
|
|
receive_mask = reg->send_mask & sclp_receive_mask;
|
|
send_mask = reg->receive_mask & sclp_send_mask;
|
|
if (reg->sclp_receive_mask != receive_mask ||
|
|
reg->sclp_send_mask != send_mask) {
|
|
reg->sclp_receive_mask = receive_mask;
|
|
reg->sclp_send_mask = send_mask;
|
|
break;
|
|
} else
|
|
reg = NULL;
|
|
}
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
if (reg && reg->state_change_fn)
|
|
reg->state_change_fn(reg);
|
|
} while (reg);
|
|
}
|
|
|
|
struct sclp_statechangebuf {
|
|
struct evbuf_header header;
|
|
u8 validity_sclp_active_facility_mask : 1;
|
|
u8 validity_sclp_receive_mask : 1;
|
|
u8 validity_sclp_send_mask : 1;
|
|
u8 validity_read_data_function_mask : 1;
|
|
u16 _zeros : 12;
|
|
u16 mask_length;
|
|
u64 sclp_active_facility_mask;
|
|
u8 masks[2 * 1021 + 4]; /* variable length */
|
|
/*
|
|
* u8 sclp_receive_mask[mask_length];
|
|
* u8 sclp_send_mask[mask_length];
|
|
* u32 read_data_function_mask;
|
|
*/
|
|
} __attribute__((packed));
|
|
|
|
|
|
/* State change event callback. Inform listeners of changes. */
|
|
static void
|
|
sclp_state_change_cb(struct evbuf_header *evbuf)
|
|
{
|
|
unsigned long flags;
|
|
struct sclp_statechangebuf *scbuf;
|
|
|
|
BUILD_BUG_ON(sizeof(struct sclp_statechangebuf) > PAGE_SIZE);
|
|
|
|
scbuf = (struct sclp_statechangebuf *) evbuf;
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
if (scbuf->validity_sclp_receive_mask)
|
|
sclp_receive_mask = sccb_get_recv_mask(scbuf);
|
|
if (scbuf->validity_sclp_send_mask)
|
|
sclp_send_mask = sccb_get_send_mask(scbuf);
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
if (scbuf->validity_sclp_active_facility_mask)
|
|
sclp.facilities = scbuf->sclp_active_facility_mask;
|
|
sclp_dispatch_state_change();
|
|
}
|
|
|
|
static struct sclp_register sclp_state_change_event = {
|
|
.receive_mask = EVTYP_STATECHANGE_MASK,
|
|
.receiver_fn = sclp_state_change_cb
|
|
};
|
|
|
|
/* Calculate receive and send mask of currently registered listeners.
|
|
* Called while sclp_lock is locked. */
|
|
static inline void
|
|
__sclp_get_mask(sccb_mask_t *receive_mask, sccb_mask_t *send_mask)
|
|
{
|
|
struct list_head *l;
|
|
struct sclp_register *t;
|
|
|
|
*receive_mask = 0;
|
|
*send_mask = 0;
|
|
list_for_each(l, &sclp_reg_list) {
|
|
t = list_entry(l, struct sclp_register, list);
|
|
*receive_mask |= t->receive_mask;
|
|
*send_mask |= t->send_mask;
|
|
}
|
|
}
|
|
|
|
/* Register event listener. Return 0 on success, non-zero otherwise. */
|
|
int
|
|
sclp_register(struct sclp_register *reg)
|
|
{
|
|
unsigned long flags;
|
|
sccb_mask_t receive_mask;
|
|
sccb_mask_t send_mask;
|
|
int rc;
|
|
|
|
rc = sclp_init();
|
|
if (rc)
|
|
return rc;
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
/* Check event mask for collisions */
|
|
__sclp_get_mask(&receive_mask, &send_mask);
|
|
if (reg->receive_mask & receive_mask || reg->send_mask & send_mask) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return -EBUSY;
|
|
}
|
|
/* Trigger initial state change callback */
|
|
reg->sclp_receive_mask = 0;
|
|
reg->sclp_send_mask = 0;
|
|
reg->pm_event_posted = 0;
|
|
list_add(®->list, &sclp_reg_list);
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
rc = sclp_init_mask(1);
|
|
if (rc) {
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
list_del(®->list);
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
}
|
|
return rc;
|
|
}
|
|
|
|
EXPORT_SYMBOL(sclp_register);
|
|
|
|
/* Unregister event listener. */
|
|
void
|
|
sclp_unregister(struct sclp_register *reg)
|
|
{
|
|
unsigned long flags;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
list_del(®->list);
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
sclp_init_mask(1);
|
|
}
|
|
|
|
EXPORT_SYMBOL(sclp_unregister);
|
|
|
|
/* Remove event buffers which are marked processed. Return the number of
|
|
* remaining event buffers. */
|
|
int
|
|
sclp_remove_processed(struct sccb_header *sccb)
|
|
{
|
|
struct evbuf_header *evbuf;
|
|
int unprocessed;
|
|
u16 remaining;
|
|
|
|
evbuf = (struct evbuf_header *) (sccb + 1);
|
|
unprocessed = 0;
|
|
remaining = sccb->length - sizeof(struct sccb_header);
|
|
while (remaining > 0) {
|
|
remaining -= evbuf->length;
|
|
if (evbuf->flags & 0x80) {
|
|
sccb->length -= evbuf->length;
|
|
memcpy(evbuf, (void *) ((addr_t) evbuf + evbuf->length),
|
|
remaining);
|
|
} else {
|
|
unprocessed++;
|
|
evbuf = (struct evbuf_header *)
|
|
((addr_t) evbuf + evbuf->length);
|
|
}
|
|
}
|
|
return unprocessed;
|
|
}
|
|
|
|
EXPORT_SYMBOL(sclp_remove_processed);
|
|
|
|
/* Prepare init mask request. Called while sclp_lock is locked. */
|
|
static inline void
|
|
__sclp_make_init_req(sccb_mask_t receive_mask, sccb_mask_t send_mask)
|
|
{
|
|
struct init_sccb *sccb;
|
|
|
|
sccb = (struct init_sccb *) sclp_init_sccb;
|
|
clear_page(sccb);
|
|
memset(&sclp_init_req, 0, sizeof(struct sclp_req));
|
|
sclp_init_req.command = SCLP_CMDW_WRITE_EVENT_MASK;
|
|
sclp_init_req.status = SCLP_REQ_FILLED;
|
|
sclp_init_req.start_count = 0;
|
|
sclp_init_req.callback = NULL;
|
|
sclp_init_req.callback_data = NULL;
|
|
sclp_init_req.sccb = sccb;
|
|
sccb->header.length = sizeof(*sccb);
|
|
if (sclp_mask_compat_mode)
|
|
sccb->mask_length = SCLP_MASK_SIZE_COMPAT;
|
|
else
|
|
sccb->mask_length = sizeof(sccb_mask_t);
|
|
sccb_set_recv_mask(sccb, receive_mask);
|
|
sccb_set_send_mask(sccb, send_mask);
|
|
sccb_set_sclp_recv_mask(sccb, 0);
|
|
sccb_set_sclp_send_mask(sccb, 0);
|
|
}
|
|
|
|
/* Start init mask request. If calculate is non-zero, calculate the mask as
|
|
* requested by registered listeners. Use zero mask otherwise. Return 0 on
|
|
* success, non-zero otherwise. */
|
|
static int
|
|
sclp_init_mask(int calculate)
|
|
{
|
|
unsigned long flags;
|
|
struct init_sccb *sccb = (struct init_sccb *) sclp_init_sccb;
|
|
sccb_mask_t receive_mask;
|
|
sccb_mask_t send_mask;
|
|
int retry;
|
|
int rc;
|
|
unsigned long wait;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
/* Check if interface is in appropriate state */
|
|
if (sclp_mask_state != sclp_mask_state_idle) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return -EBUSY;
|
|
}
|
|
if (sclp_activation_state == sclp_activation_state_inactive) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return -EINVAL;
|
|
}
|
|
sclp_mask_state = sclp_mask_state_initializing;
|
|
/* Determine mask */
|
|
if (calculate)
|
|
__sclp_get_mask(&receive_mask, &send_mask);
|
|
else {
|
|
receive_mask = 0;
|
|
send_mask = 0;
|
|
}
|
|
rc = -EIO;
|
|
for (retry = 0; retry <= SCLP_MASK_RETRY; retry++) {
|
|
/* Prepare request */
|
|
__sclp_make_init_req(receive_mask, send_mask);
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
if (sclp_add_request(&sclp_init_req)) {
|
|
/* Try again later */
|
|
wait = jiffies + SCLP_BUSY_INTERVAL * HZ;
|
|
while (time_before(jiffies, wait))
|
|
sclp_sync_wait();
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
continue;
|
|
}
|
|
while (sclp_init_req.status != SCLP_REQ_DONE &&
|
|
sclp_init_req.status != SCLP_REQ_FAILED)
|
|
sclp_sync_wait();
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
if (sclp_init_req.status == SCLP_REQ_DONE &&
|
|
sccb->header.response_code == 0x20) {
|
|
/* Successful request */
|
|
if (calculate) {
|
|
sclp_receive_mask = sccb_get_sclp_recv_mask(sccb);
|
|
sclp_send_mask = sccb_get_sclp_send_mask(sccb);
|
|
} else {
|
|
sclp_receive_mask = 0;
|
|
sclp_send_mask = 0;
|
|
}
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
sclp_dispatch_state_change();
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
rc = 0;
|
|
break;
|
|
}
|
|
}
|
|
sclp_mask_state = sclp_mask_state_idle;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return rc;
|
|
}
|
|
|
|
/* Deactivate SCLP interface. On success, new requests will be rejected,
|
|
* events will no longer be dispatched. Return 0 on success, non-zero
|
|
* otherwise. */
|
|
int
|
|
sclp_deactivate(void)
|
|
{
|
|
unsigned long flags;
|
|
int rc;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
/* Deactivate can only be called when active */
|
|
if (sclp_activation_state != sclp_activation_state_active) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return -EINVAL;
|
|
}
|
|
sclp_activation_state = sclp_activation_state_deactivating;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
rc = sclp_init_mask(0);
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
if (rc == 0)
|
|
sclp_activation_state = sclp_activation_state_inactive;
|
|
else
|
|
sclp_activation_state = sclp_activation_state_active;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return rc;
|
|
}
|
|
|
|
EXPORT_SYMBOL(sclp_deactivate);
|
|
|
|
/* Reactivate SCLP interface after sclp_deactivate. On success, new
|
|
* requests will be accepted, events will be dispatched again. Return 0 on
|
|
* success, non-zero otherwise. */
|
|
int
|
|
sclp_reactivate(void)
|
|
{
|
|
unsigned long flags;
|
|
int rc;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
/* Reactivate can only be called when inactive */
|
|
if (sclp_activation_state != sclp_activation_state_inactive) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return -EINVAL;
|
|
}
|
|
sclp_activation_state = sclp_activation_state_activating;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
rc = sclp_init_mask(1);
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
if (rc == 0)
|
|
sclp_activation_state = sclp_activation_state_active;
|
|
else
|
|
sclp_activation_state = sclp_activation_state_inactive;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return rc;
|
|
}
|
|
|
|
EXPORT_SYMBOL(sclp_reactivate);
|
|
|
|
/* Handler for external interruption used during initialization. Modify
|
|
* request state to done. */
|
|
static void sclp_check_handler(struct ext_code ext_code,
|
|
unsigned int param32, unsigned long param64)
|
|
{
|
|
u32 finished_sccb;
|
|
|
|
inc_irq_stat(IRQEXT_SCP);
|
|
finished_sccb = param32 & 0xfffffff8;
|
|
/* Is this the interrupt we are waiting for? */
|
|
if (finished_sccb == 0)
|
|
return;
|
|
if (finished_sccb != (u32) (addr_t) sclp_init_sccb)
|
|
panic("sclp: unsolicited interrupt for buffer at 0x%x\n",
|
|
finished_sccb);
|
|
spin_lock(&sclp_lock);
|
|
if (sclp_running_state == sclp_running_state_running) {
|
|
sclp_init_req.status = SCLP_REQ_DONE;
|
|
sclp_running_state = sclp_running_state_idle;
|
|
}
|
|
spin_unlock(&sclp_lock);
|
|
}
|
|
|
|
/* Initial init mask request timed out. Modify request state to failed. */
|
|
static void
|
|
sclp_check_timeout(struct timer_list *unused)
|
|
{
|
|
unsigned long flags;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
if (sclp_running_state == sclp_running_state_running) {
|
|
sclp_init_req.status = SCLP_REQ_FAILED;
|
|
sclp_running_state = sclp_running_state_idle;
|
|
}
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
}
|
|
|
|
/* Perform a check of the SCLP interface. Return zero if the interface is
|
|
* available and there are no pending requests from a previous instance.
|
|
* Return non-zero otherwise. */
|
|
static int
|
|
sclp_check_interface(void)
|
|
{
|
|
struct init_sccb *sccb;
|
|
unsigned long flags;
|
|
int retry;
|
|
int rc;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
/* Prepare init mask command */
|
|
rc = register_external_irq(EXT_IRQ_SERVICE_SIG, sclp_check_handler);
|
|
if (rc) {
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return rc;
|
|
}
|
|
for (retry = 0; retry <= SCLP_INIT_RETRY; retry++) {
|
|
__sclp_make_init_req(0, 0);
|
|
sccb = (struct init_sccb *) sclp_init_req.sccb;
|
|
rc = sclp_service_call(sclp_init_req.command, sccb);
|
|
if (rc == -EIO)
|
|
break;
|
|
sclp_init_req.status = SCLP_REQ_RUNNING;
|
|
sclp_running_state = sclp_running_state_running;
|
|
__sclp_set_request_timer(SCLP_RETRY_INTERVAL * HZ,
|
|
sclp_check_timeout);
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
/* Enable service-signal interruption - needs to happen
|
|
* with IRQs enabled. */
|
|
irq_subclass_register(IRQ_SUBCLASS_SERVICE_SIGNAL);
|
|
/* Wait for signal from interrupt or timeout */
|
|
sclp_sync_wait();
|
|
/* Disable service-signal interruption - needs to happen
|
|
* with IRQs enabled. */
|
|
irq_subclass_unregister(IRQ_SUBCLASS_SERVICE_SIGNAL);
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
del_timer(&sclp_request_timer);
|
|
rc = -EBUSY;
|
|
if (sclp_init_req.status == SCLP_REQ_DONE) {
|
|
if (sccb->header.response_code == 0x20) {
|
|
rc = 0;
|
|
break;
|
|
} else if (sccb->header.response_code == 0x74f0) {
|
|
if (!sclp_mask_compat_mode) {
|
|
sclp_mask_compat_mode = true;
|
|
retry = 0;
|
|
}
|
|
}
|
|
}
|
|
}
|
|
unregister_external_irq(EXT_IRQ_SERVICE_SIG, sclp_check_handler);
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return rc;
|
|
}
|
|
|
|
/* Reboot event handler. Reset send and receive mask to prevent pending SCLP
|
|
* events from interfering with rebooted system. */
|
|
static int
|
|
sclp_reboot_event(struct notifier_block *this, unsigned long event, void *ptr)
|
|
{
|
|
sclp_deactivate();
|
|
return NOTIFY_DONE;
|
|
}
|
|
|
|
static struct notifier_block sclp_reboot_notifier = {
|
|
.notifier_call = sclp_reboot_event
|
|
};
|
|
|
|
/*
|
|
* Suspend/resume SCLP notifier implementation
|
|
*/
|
|
|
|
static void sclp_pm_event(enum sclp_pm_event sclp_pm_event, int rollback)
|
|
{
|
|
struct sclp_register *reg;
|
|
unsigned long flags;
|
|
|
|
if (!rollback) {
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
list_for_each_entry(reg, &sclp_reg_list, list)
|
|
reg->pm_event_posted = 0;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
}
|
|
do {
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
list_for_each_entry(reg, &sclp_reg_list, list) {
|
|
if (rollback && reg->pm_event_posted)
|
|
goto found;
|
|
if (!rollback && !reg->pm_event_posted)
|
|
goto found;
|
|
}
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return;
|
|
found:
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
if (reg->pm_event_fn)
|
|
reg->pm_event_fn(reg, sclp_pm_event);
|
|
reg->pm_event_posted = rollback ? 0 : 1;
|
|
} while (1);
|
|
}
|
|
|
|
/*
|
|
* Susend/resume callbacks for platform device
|
|
*/
|
|
|
|
static int sclp_freeze(struct device *dev)
|
|
{
|
|
unsigned long flags;
|
|
int rc;
|
|
|
|
sclp_pm_event(SCLP_PM_EVENT_FREEZE, 0);
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
sclp_suspend_state = sclp_suspend_state_suspended;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
|
|
/* Init supend data */
|
|
memset(&sclp_suspend_req, 0, sizeof(sclp_suspend_req));
|
|
sclp_suspend_req.callback = sclp_suspend_req_cb;
|
|
sclp_suspend_req.status = SCLP_REQ_FILLED;
|
|
init_completion(&sclp_request_queue_flushed);
|
|
|
|
rc = sclp_add_request(&sclp_suspend_req);
|
|
if (rc == 0)
|
|
wait_for_completion(&sclp_request_queue_flushed);
|
|
else if (rc != -ENODATA)
|
|
goto fail_thaw;
|
|
|
|
rc = sclp_deactivate();
|
|
if (rc)
|
|
goto fail_thaw;
|
|
return 0;
|
|
|
|
fail_thaw:
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
sclp_suspend_state = sclp_suspend_state_running;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
sclp_pm_event(SCLP_PM_EVENT_THAW, 1);
|
|
return rc;
|
|
}
|
|
|
|
static int sclp_undo_suspend(enum sclp_pm_event event)
|
|
{
|
|
unsigned long flags;
|
|
int rc;
|
|
|
|
rc = sclp_reactivate();
|
|
if (rc)
|
|
return rc;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
sclp_suspend_state = sclp_suspend_state_running;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
|
|
sclp_pm_event(event, 0);
|
|
return 0;
|
|
}
|
|
|
|
static int sclp_thaw(struct device *dev)
|
|
{
|
|
return sclp_undo_suspend(SCLP_PM_EVENT_THAW);
|
|
}
|
|
|
|
static int sclp_restore(struct device *dev)
|
|
{
|
|
return sclp_undo_suspend(SCLP_PM_EVENT_RESTORE);
|
|
}
|
|
|
|
static const struct dev_pm_ops sclp_pm_ops = {
|
|
.freeze = sclp_freeze,
|
|
.thaw = sclp_thaw,
|
|
.restore = sclp_restore,
|
|
};
|
|
|
|
static ssize_t con_pages_show(struct device_driver *dev, char *buf)
|
|
{
|
|
return sprintf(buf, "%i\n", sclp_console_pages);
|
|
}
|
|
|
|
static DRIVER_ATTR_RO(con_pages);
|
|
|
|
static ssize_t con_drop_show(struct device_driver *dev, char *buf)
|
|
{
|
|
return sprintf(buf, "%i\n", sclp_console_drop);
|
|
}
|
|
|
|
static DRIVER_ATTR_RO(con_drop);
|
|
|
|
static ssize_t con_full_show(struct device_driver *dev, char *buf)
|
|
{
|
|
return sprintf(buf, "%lu\n", sclp_console_full);
|
|
}
|
|
|
|
static DRIVER_ATTR_RO(con_full);
|
|
|
|
static struct attribute *sclp_drv_attrs[] = {
|
|
&driver_attr_con_pages.attr,
|
|
&driver_attr_con_drop.attr,
|
|
&driver_attr_con_full.attr,
|
|
NULL,
|
|
};
|
|
static struct attribute_group sclp_drv_attr_group = {
|
|
.attrs = sclp_drv_attrs,
|
|
};
|
|
static const struct attribute_group *sclp_drv_attr_groups[] = {
|
|
&sclp_drv_attr_group,
|
|
NULL,
|
|
};
|
|
|
|
static struct platform_driver sclp_pdrv = {
|
|
.driver = {
|
|
.name = "sclp",
|
|
.pm = &sclp_pm_ops,
|
|
.groups = sclp_drv_attr_groups,
|
|
},
|
|
};
|
|
|
|
static struct platform_device *sclp_pdev;
|
|
|
|
/* Initialize SCLP driver. Return zero if driver is operational, non-zero
|
|
* otherwise. */
|
|
static int
|
|
sclp_init(void)
|
|
{
|
|
unsigned long flags;
|
|
int rc = 0;
|
|
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
/* Check for previous or running initialization */
|
|
if (sclp_init_state != sclp_init_state_uninitialized)
|
|
goto fail_unlock;
|
|
sclp_init_state = sclp_init_state_initializing;
|
|
/* Set up variables */
|
|
INIT_LIST_HEAD(&sclp_req_queue);
|
|
INIT_LIST_HEAD(&sclp_reg_list);
|
|
list_add(&sclp_state_change_event.list, &sclp_reg_list);
|
|
timer_setup(&sclp_request_timer, NULL, 0);
|
|
timer_setup(&sclp_queue_timer, sclp_req_queue_timeout, 0);
|
|
/* Check interface */
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
rc = sclp_check_interface();
|
|
spin_lock_irqsave(&sclp_lock, flags);
|
|
if (rc)
|
|
goto fail_init_state_uninitialized;
|
|
/* Register reboot handler */
|
|
rc = register_reboot_notifier(&sclp_reboot_notifier);
|
|
if (rc)
|
|
goto fail_init_state_uninitialized;
|
|
/* Register interrupt handler */
|
|
rc = register_external_irq(EXT_IRQ_SERVICE_SIG, sclp_interrupt_handler);
|
|
if (rc)
|
|
goto fail_unregister_reboot_notifier;
|
|
sclp_init_state = sclp_init_state_initialized;
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
/* Enable service-signal external interruption - needs to happen with
|
|
* IRQs enabled. */
|
|
irq_subclass_register(IRQ_SUBCLASS_SERVICE_SIGNAL);
|
|
sclp_init_mask(1);
|
|
return 0;
|
|
|
|
fail_unregister_reboot_notifier:
|
|
unregister_reboot_notifier(&sclp_reboot_notifier);
|
|
fail_init_state_uninitialized:
|
|
sclp_init_state = sclp_init_state_uninitialized;
|
|
fail_unlock:
|
|
spin_unlock_irqrestore(&sclp_lock, flags);
|
|
return rc;
|
|
}
|
|
|
|
/*
|
|
* SCLP panic notifier: If we are suspended, we thaw SCLP in order to be able
|
|
* to print the panic message.
|
|
*/
|
|
static int sclp_panic_notify(struct notifier_block *self,
|
|
unsigned long event, void *data)
|
|
{
|
|
if (sclp_suspend_state == sclp_suspend_state_suspended)
|
|
sclp_undo_suspend(SCLP_PM_EVENT_THAW);
|
|
return NOTIFY_OK;
|
|
}
|
|
|
|
static struct notifier_block sclp_on_panic_nb = {
|
|
.notifier_call = sclp_panic_notify,
|
|
.priority = SCLP_PANIC_PRIO,
|
|
};
|
|
|
|
static __init int sclp_initcall(void)
|
|
{
|
|
int rc;
|
|
|
|
rc = platform_driver_register(&sclp_pdrv);
|
|
if (rc)
|
|
return rc;
|
|
|
|
sclp_pdev = platform_device_register_simple("sclp", -1, NULL, 0);
|
|
rc = PTR_ERR_OR_ZERO(sclp_pdev);
|
|
if (rc)
|
|
goto fail_platform_driver_unregister;
|
|
|
|
rc = atomic_notifier_chain_register(&panic_notifier_list,
|
|
&sclp_on_panic_nb);
|
|
if (rc)
|
|
goto fail_platform_device_unregister;
|
|
|
|
return sclp_init();
|
|
|
|
fail_platform_device_unregister:
|
|
platform_device_unregister(sclp_pdev);
|
|
fail_platform_driver_unregister:
|
|
platform_driver_unregister(&sclp_pdrv);
|
|
return rc;
|
|
}
|
|
|
|
arch_initcall(sclp_initcall);
|