[S390] dasd: Improve handling of stolen DASD reservation

If a DASD device has been reserved by a Linux system, and later
this reservation is ‘stolen’ by a second system by means of an
unconditional reserve, then the first system receives a
notification about this fact. With this patch such an event can
be either ignored, as before, or it can be used to let the device
fail all I/O request, so that the device will not block anymore.

Signed-off-by: Stefan Weinhuber <wein@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
This commit is contained in:
Stefan Weinhuber 2011-01-05 12:48:04 +01:00 committed by Martin Schwidefsky
parent a4d26c6aec
commit 5a27e60dec
7 changed files with 189 additions and 69 deletions

View File

@ -80,6 +80,7 @@ typedef struct dasd_information2_t {
#define DASD_FEATURE_INITIAL_ONLINE 0x04
#define DASD_FEATURE_ERPLOG 0x08
#define DASD_FEATURE_FAILFAST 0x10
#define DASD_FEATURE_FAILONSLCK 0x20
#define DASD_PARTN_BITS 2

View File

@ -902,6 +902,16 @@ int dasd_start_IO(struct dasd_ccw_req *cqr)
return rc;
}
device = (struct dasd_device *) cqr->startdev;
if (((cqr->block &&
test_bit(DASD_FLAG_LOCK_STOLEN, &cqr->block->base->flags)) ||
test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags)) &&
!test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) {
DBF_DEV_EVENT(DBF_DEBUG, device, "start_IO: return request %p "
"because of stolen lock", cqr);
cqr->status = DASD_CQR_ERROR;
cqr->intrc = -EPERM;
return -EPERM;
}
if (cqr->retries < 0) {
/* internal error 14 - start_IO run out of retries */
sprintf(errorstring, "14 %p", cqr);
@ -1115,16 +1125,11 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
}
now = get_clock();
/* check for unsolicited interrupts */
cqr = (struct dasd_ccw_req *) intparm;
if (!cqr || ((scsw_cc(&irb->scsw) == 1) &&
(scsw_fctl(&irb->scsw) & SCSW_FCTL_START_FUNC) &&
((scsw_stctl(&irb->scsw) == SCSW_STCTL_STATUS_PEND) ||
(scsw_stctl(&irb->scsw) == (SCSW_STCTL_STATUS_PEND |
SCSW_STCTL_ALERT_STATUS))))) {
if (cqr && cqr->status == DASD_CQR_IN_IO)
cqr->status = DASD_CQR_QUEUED;
/* check for conditions that should be handled immediately */
if (!cqr ||
!(scsw_dstat(&irb->scsw) == (DEV_STAT_CHN_END | DEV_STAT_DEV_END) &&
scsw_cstat(&irb->scsw) == 0)) {
if (cqr)
memcpy(&cqr->irb, irb, sizeof(*irb));
device = dasd_device_from_cdev_locked(cdev);
@ -1135,17 +1140,14 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
dasd_put_device(device);
return;
}
device->discipline->dump_sense_dbf(device, irb,
"unsolicited");
if ((device->features & DASD_FEATURE_ERPLOG))
device->discipline->dump_sense(device, cqr,
irb);
dasd_device_clear_timer(device);
device->discipline->handle_unsolicited_interrupt(device,
irb);
device->discipline->dump_sense_dbf(device, irb, "int");
if (device->features & DASD_FEATURE_ERPLOG)
device->discipline->dump_sense(device, cqr, irb);
device->discipline->check_for_device_change(device, cqr, irb);
dasd_put_device(device);
return;
}
if (!cqr)
return;
device = (struct dasd_device *) cqr->startdev;
if (!device ||
@ -1185,13 +1187,6 @@ void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
struct dasd_ccw_req, devlist);
}
} else { /* error */
memcpy(&cqr->irb, irb, sizeof(struct irb));
/* log sense for every failed I/O to s390 debugfeature */
dasd_log_sense_dbf(cqr, irb);
if (device->features & DASD_FEATURE_ERPLOG) {
dasd_log_sense(cqr, irb);
}
/*
* If we don't want complex ERP for this request, then just
* reset this and retry it in the fastpath
@ -1232,13 +1227,13 @@ enum uc_todo dasd_generic_uc_handler(struct ccw_device *cdev, struct irb *irb)
goto out;
if (test_bit(DASD_FLAG_OFFLINE, &device->flags) ||
device->state != device->target ||
!device->discipline->handle_unsolicited_interrupt){
!device->discipline->check_for_device_change){
dasd_put_device(device);
goto out;
}
dasd_device_clear_timer(device);
device->discipline->handle_unsolicited_interrupt(device, irb);
if (device->discipline->dump_sense_dbf)
device->discipline->dump_sense_dbf(device, irb, "uc");
device->discipline->check_for_device_change(device, NULL, irb);
dasd_put_device(device);
out:
return UC_TODO_RETRY;
@ -1659,7 +1654,12 @@ static int _dasd_sleep_on(struct dasd_ccw_req *maincqr, int interruptible)
continue;
if (cqr->status != DASD_CQR_FILLED) /* could be failed */
continue;
if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags) &&
!test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) {
cqr->status = DASD_CQR_FAILED;
cqr->intrc = -EPERM;
continue;
}
/* Non-temporary stop condition will trigger fail fast */
if (device->stopped & ~DASD_STOPPED_PENDING &&
test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) &&
@ -1667,7 +1667,6 @@ static int _dasd_sleep_on(struct dasd_ccw_req *maincqr, int interruptible)
cqr->status = DASD_CQR_FAILED;
continue;
}
/* Don't try to start requests if device is stopped */
if (interruptible) {
rc = wait_event_interruptible(
@ -1752,13 +1751,18 @@ int dasd_sleep_on_immediatly(struct dasd_ccw_req *cqr)
int rc;
device = cqr->startdev;
if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags) &&
!test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) {
cqr->status = DASD_CQR_FAILED;
cqr->intrc = -EPERM;
return -EIO;
}
spin_lock_irq(get_ccwdev_lock(device->cdev));
rc = _dasd_term_running_cqr(device);
if (rc) {
spin_unlock_irq(get_ccwdev_lock(device->cdev));
return rc;
}
cqr->callback = dasd_wakeup_cb;
cqr->callback_data = DASD_SLEEPON_START_TAG;
cqr->status = DASD_CQR_QUEUED;
@ -2062,6 +2066,13 @@ static void __dasd_block_start_head(struct dasd_block *block)
list_for_each_entry(cqr, &block->ccw_queue, blocklist) {
if (cqr->status != DASD_CQR_FILLED)
continue;
if (test_bit(DASD_FLAG_LOCK_STOLEN, &block->base->flags) &&
!test_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags)) {
cqr->status = DASD_CQR_FAILED;
cqr->intrc = -EPERM;
dasd_schedule_block_bh(block);
continue;
}
/* Non-temporary stop condition will trigger fail fast */
if (block->base->stopped & ~DASD_STOPPED_PENDING &&
test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) &&

View File

@ -1127,6 +1127,103 @@ dasd_expires_store(struct device *dev, struct device_attribute *attr,
static DEVICE_ATTR(expires, 0644, dasd_expires_show, dasd_expires_store);
static ssize_t dasd_reservation_policy_show(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct dasd_devmap *devmap;
int rc = 0;
devmap = dasd_find_busid(dev_name(dev));
if (IS_ERR(devmap)) {
rc = snprintf(buf, PAGE_SIZE, "ignore\n");
} else {
spin_lock(&dasd_devmap_lock);
if (devmap->features & DASD_FEATURE_FAILONSLCK)
rc = snprintf(buf, PAGE_SIZE, "fail\n");
else
rc = snprintf(buf, PAGE_SIZE, "ignore\n");
spin_unlock(&dasd_devmap_lock);
}
return rc;
}
static ssize_t dasd_reservation_policy_store(struct device *dev,
struct device_attribute *attr,
const char *buf, size_t count)
{
struct dasd_devmap *devmap;
int rc;
devmap = dasd_devmap_from_cdev(to_ccwdev(dev));
if (IS_ERR(devmap))
return PTR_ERR(devmap);
rc = 0;
spin_lock(&dasd_devmap_lock);
if (sysfs_streq("ignore", buf))
devmap->features &= ~DASD_FEATURE_FAILONSLCK;
else if (sysfs_streq("fail", buf))
devmap->features |= DASD_FEATURE_FAILONSLCK;
else
rc = -EINVAL;
if (devmap->device)
devmap->device->features = devmap->features;
spin_unlock(&dasd_devmap_lock);
if (rc)
return rc;
else
return count;
}
static DEVICE_ATTR(reservation_policy, 0644,
dasd_reservation_policy_show, dasd_reservation_policy_store);
static ssize_t dasd_reservation_state_show(struct device *dev,
struct device_attribute *attr,
char *buf)
{
struct dasd_device *device;
int rc = 0;
device = dasd_device_from_cdev(to_ccwdev(dev));
if (IS_ERR(device))
return snprintf(buf, PAGE_SIZE, "none\n");
if (test_bit(DASD_FLAG_IS_RESERVED, &device->flags))
rc = snprintf(buf, PAGE_SIZE, "reserved\n");
else if (test_bit(DASD_FLAG_LOCK_STOLEN, &device->flags))
rc = snprintf(buf, PAGE_SIZE, "lost\n");
else
rc = snprintf(buf, PAGE_SIZE, "none\n");
dasd_put_device(device);
return rc;
}
static ssize_t dasd_reservation_state_store(struct device *dev,
struct device_attribute *attr,
const char *buf, size_t count)
{
struct dasd_device *device;
int rc = 0;
device = dasd_device_from_cdev(to_ccwdev(dev));
if (IS_ERR(device))
return -ENODEV;
if (sysfs_streq("reset", buf))
clear_bit(DASD_FLAG_LOCK_STOLEN, &device->flags);
else
rc = -EINVAL;
dasd_put_device(device);
if (rc)
return rc;
else
return count;
}
static DEVICE_ATTR(last_known_reservation_state, 0644,
dasd_reservation_state_show, dasd_reservation_state_store);
static struct attribute * dasd_attrs[] = {
&dev_attr_readonly.attr,
&dev_attr_discipline.attr,
@ -1139,6 +1236,8 @@ static struct attribute * dasd_attrs[] = {
&dev_attr_erplog.attr,
&dev_attr_failfast.attr,
&dev_attr_expires.attr,
&dev_attr_reservation_policy.attr,
&dev_attr_last_known_reservation_state.attr,
NULL,
};

View File

@ -817,6 +817,7 @@ static int dasd_eckd_read_conf_immediately(struct dasd_device *device,
dasd_eckd_fill_rcd_cqr(device, cqr, rcd_buffer, lpm);
clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags);
cqr->retries = 5;
rc = dasd_sleep_on_immediatly(cqr);
return rc;
@ -1947,9 +1948,9 @@ dasd_eckd_erp_postaction(struct dasd_ccw_req * cqr)
return dasd_default_erp_postaction;
}
static void dasd_eckd_handle_unsolicited_interrupt(struct dasd_device *device,
struct irb *irb)
static void dasd_eckd_check_for_device_change(struct dasd_device *device,
struct dasd_ccw_req *cqr,
struct irb *irb)
{
char mask;
char *sense = NULL;
@ -1973,40 +1974,41 @@ static void dasd_eckd_handle_unsolicited_interrupt(struct dasd_device *device,
/* schedule worker to reload device */
dasd_reload_device(device);
}
dasd_generic_handle_state_change(device);
return;
}
/* summary unit check */
sense = dasd_get_sense(irb);
if (sense && (sense[7] == 0x0D) &&
if (!sense)
return;
/* summary unit check */
if ((sense[7] == 0x0D) &&
(scsw_dstat(&irb->scsw) & DEV_STAT_UNIT_CHECK)) {
dasd_alias_handle_summary_unit_check(device, irb);
return;
}
/* service information message SIM */
if (sense && !(sense[27] & DASD_SENSE_BIT_0) &&
if (!cqr && !(sense[27] & DASD_SENSE_BIT_0) &&
((sense[6] & DASD_SIM_SENSE) == DASD_SIM_SENSE)) {
dasd_3990_erp_handle_sim(device, sense);
dasd_schedule_device_bh(device);
return;
}
if ((scsw_cc(&irb->scsw) == 1) && !sense &&
(scsw_fctl(&irb->scsw) == SCSW_FCTL_START_FUNC) &&
(scsw_actl(&irb->scsw) == SCSW_ACTL_START_PEND) &&
(scsw_stctl(&irb->scsw) == SCSW_STCTL_STATUS_PEND)) {
/* fake irb do nothing, they are handled elsewhere */
dasd_schedule_device_bh(device);
return;
/* loss of device reservation is handled via base devices only
* as alias devices may be used with several bases
*/
if (device->block && (sense[7] == 0x3F) &&
(scsw_dstat(&irb->scsw) & DEV_STAT_UNIT_CHECK) &&
test_bit(DASD_FLAG_IS_RESERVED, &device->flags)) {
if (device->features & DASD_FEATURE_FAILONSLCK)
set_bit(DASD_FLAG_LOCK_STOLEN, &device->flags);
clear_bit(DASD_FLAG_IS_RESERVED, &device->flags);
dev_err(&device->cdev->dev,
"The device reservation was lost\n");
}
dasd_schedule_device_bh(device);
return;
};
}
static struct dasd_ccw_req *dasd_eckd_build_cp_cmd_single(
struct dasd_device *startdev,
@ -2931,6 +2933,8 @@ dasd_eckd_release(struct dasd_device *device)
cqr->status = DASD_CQR_FILLED;
rc = dasd_sleep_on_immediatly(cqr);
if (!rc)
clear_bit(DASD_FLAG_IS_RESERVED, &device->flags);
if (useglobal)
mutex_unlock(&dasd_reserve_mutex);
@ -2984,6 +2988,8 @@ dasd_eckd_reserve(struct dasd_device *device)
cqr->status = DASD_CQR_FILLED;
rc = dasd_sleep_on_immediatly(cqr);
if (!rc)
set_bit(DASD_FLAG_IS_RESERVED, &device->flags);
if (useglobal)
mutex_unlock(&dasd_reserve_mutex);
@ -3036,6 +3042,8 @@ dasd_eckd_steal_lock(struct dasd_device *device)
cqr->status = DASD_CQR_FILLED;
rc = dasd_sleep_on_immediatly(cqr);
if (!rc)
set_bit(DASD_FLAG_IS_RESERVED, &device->flags);
if (useglobal)
mutex_unlock(&dasd_reserve_mutex);
@ -3088,6 +3096,7 @@ static int dasd_eckd_snid(struct dasd_device *device,
cqr->memdev = device;
clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags);
cqr->retries = 5;
cqr->expires = 10 * HZ;
cqr->buildclk = get_clock();
@ -3832,7 +3841,7 @@ static struct dasd_discipline dasd_eckd_discipline = {
.format_device = dasd_eckd_format_device,
.erp_action = dasd_eckd_erp_action,
.erp_postaction = dasd_eckd_erp_postaction,
.handle_unsolicited_interrupt = dasd_eckd_handle_unsolicited_interrupt,
.check_for_device_change = dasd_eckd_check_for_device_change,
.build_cp = dasd_eckd_build_alias_cp,
.free_cp = dasd_eckd_free_alias_cp,
.dump_sense = dasd_eckd_dump_sense,

View File

@ -473,6 +473,7 @@ int dasd_eer_enable(struct dasd_device *device)
cqr->retries = 255;
cqr->expires = 10 * HZ;
clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
set_bit(DASD_CQR_ALLOW_SLOCK, &cqr->flags);
ccw = cqr->cpaddr;
ccw->cmd_code = DASD_ECKD_CCW_SNSS;

View File

@ -233,24 +233,16 @@ dasd_fba_erp_postaction(struct dasd_ccw_req * cqr)
return NULL;
}
static void dasd_fba_handle_unsolicited_interrupt(struct dasd_device *device,
struct irb *irb)
static void dasd_fba_check_for_device_change(struct dasd_device *device,
struct dasd_ccw_req *cqr,
struct irb *irb)
{
char mask;
/* first of all check for state change pending interrupt */
mask = DEV_STAT_ATTENTION | DEV_STAT_DEV_END | DEV_STAT_UNIT_EXCEP;
if ((irb->scsw.cmd.dstat & mask) == mask) {
if ((irb->scsw.cmd.dstat & mask) == mask)
dasd_generic_handle_state_change(device);
return;
}
/* check for unsolicited interrupts */
DBF_DEV_EVENT(DBF_WARNING, device, "%s",
"unsolicited interrupt received");
device->discipline->dump_sense_dbf(device, irb, "unsolicited");
dasd_schedule_device_bh(device);
return;
};
static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev,
@ -605,7 +597,7 @@ static struct dasd_discipline dasd_fba_discipline = {
.handle_terminated_request = dasd_fba_handle_terminated_request,
.erp_action = dasd_fba_erp_action,
.erp_postaction = dasd_fba_erp_postaction,
.handle_unsolicited_interrupt = dasd_fba_handle_unsolicited_interrupt,
.check_for_device_change = dasd_fba_check_for_device_change,
.build_cp = dasd_fba_build_cp,
.free_cp = dasd_fba_free_cp,
.dump_sense = dasd_fba_dump_sense,

View File

@ -232,6 +232,10 @@ struct dasd_ccw_req {
#define DASD_CQR_FLAGS_USE_ERP 0 /* use ERP for this request */
#define DASD_CQR_FLAGS_FAILFAST 1 /* FAILFAST */
#define DASD_CQR_VERIFY_PATH 2 /* path verification request */
#define DASD_CQR_ALLOW_SLOCK 3 /* Try this request even when lock was
* stolen. Should not be combined with
* DASD_CQR_FLAGS_USE_ERP
*/
/* Signature for error recovery functions. */
typedef struct dasd_ccw_req *(*dasd_erp_fn_t) (struct dasd_ccw_req *);
@ -334,9 +338,9 @@ struct dasd_discipline {
void (*dump_sense) (struct dasd_device *, struct dasd_ccw_req *,
struct irb *);
void (*dump_sense_dbf) (struct dasd_device *, struct irb *, char *);
void (*handle_unsolicited_interrupt) (struct dasd_device *,
struct irb *);
void (*check_for_device_change) (struct dasd_device *,
struct dasd_ccw_req *,
struct irb *);
/* i/o control functions. */
int (*fill_geometry) (struct dasd_block *, struct hd_geometry *);
@ -473,6 +477,9 @@ struct dasd_block {
* confuse this with the user specified
* read-only feature.
*/
#define DASD_FLAG_IS_RESERVED 7 /* The device is reserved */
#define DASD_FLAG_LOCK_STOLEN 8 /* The device lock was stolen */
void dasd_put_device_wake(struct dasd_device *);