* [patch 00/47] s390 2.6.25 patch queue.
@ 2007-12-20 15:19 Martin Schwidefsky
2007-12-20 15:19 ` [patch 01/47] Cleanup in Documentation/kernel-parameters.txt Martin Schwidefsky
` (46 more replies)
0 siblings, 47 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390
This is my set of patches in the queue for the next upstream merge
window. Most of them are bug fixes and small changes. The last three
patches are big and introduce an interesting new feature: hyper PAV.
Similar to the existing PAV support in the dasd driver the new
hyper PAV support allows to start multiple channel programs on a
single dasd device. The major difference is that the old style PAV
support uses static alias devices while the hyper PAV code uses
dynamic alias devices. The shortlog:
Christian Borntraeger (2):
[S390] Change vmalloc defintions
[S390] vmemmap: allocate struct pages before 1:1 mapping
Cornelia Huck (9):
[S390] cio: Use helpers instead of container_of().
[S390] cio: css_driver: Use consistent parameters.
[S390] cio: Reset sch->driver.
[S390] cio: Add css_driver_{register,unregister}.
[S390] cio: Cleanup debug feature usage.
[S390] cio: Introduce subchannel->private.
[S390] cio: I/O subchannel specific fields.
[S390] cio: Use dev_{g,s}et_drvdata().
[S390] cio: Set driver->owner on css, ccw and ccwgroup busses.
Denis Cheng (1):
[S390] use LIST_HEAD instead of LIST_HEAD_INIT
Heiko Carstens (10):
[S390] Standby cpu activation/deactivation.
[S390] sclp: convert channel path configure code to use sync interface.
[S390] Print kernel version in dump_stack() and show_regs().
[S390] Get rid of HOLES_IN_ZONE requirement.
[S390] DEBUG_PAGEALLOC support for s390.
[S390] Remove owner_pc member from raw_spinlock_t.
[S390] Use new style spinlock initializer in __RWSEM_INITIALIZER.
[S390] Get rid of additional_cpus kernel parameter.
[S390] Remove appldata include from sysctl_check.c
[S390] Allocate and free cpu lowcores and stacks when needed/possible.
Jan Glauber (1):
[S390] crypto: move s390 Kconfig options.
Joe Perches (5):
[S390] arch/s390: Add missing "space"
[S390] drivers/s390: Add missing "space"
[S390] arch/s390/: Spelling fixes
[S390] include/asm-s390/: Spelling fixes
[S390] drivers/s390/: Spelling fixes
Martin Schwidefsky (3):
[S390] Optimize reference bit handling.
[S390] Fix tlb flushing with idte.
[S390] Move NOTES and BUG_TABLE.
Michael Ernst (1):
[S390] sclp: sysfs interface for SCLP cpi
Michael Holzheu (4):
[S390] kernel: Shutdown Actions Interface
[S390] Load disabled wait psw instead of stopping cpu on halt.
[S390] Initialize sclp_ipl_info
[S390] Use diag308 subcodes 3 and 6 for reboot and dump when possible.
Peter Oberparleiter (2):
[S390] cio: Extend adapter interrupt interface.
[S390] cio: reduce cpu utilization during device scan
Roland McGrath (1):
[S390] single-step cleanup
Sebastian Ott (3):
[S390] Cleanup in Documentation/kernel-parameters.txt.
[S390] cio: Dump ccw device information in case of timeout.
[S390] qdio: Remove double checked value.
Stefan Haberland (1):
[S390] dasd: fix return value of dasd_generic_probe()
Stefan Weinhuber (3):
[S390] dasd: add hyper PAV support to DASD device driver, part 1
[S390] dasd: add hyper PAV support to DASD device driver, part 2
[S390] dasd: add hyper PAV support to DASD device driver, part 3
Ursula Braun (1):
[S390] qdio: set QDIO_ACTIVATE_TIMEOUT to 5s
Have fun and merry xmas.
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 01/47] Cleanup in Documentation/kernel-parameters.txt.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 02/47] cio: Dump ccw device information in case of timeout Martin Schwidefsky
` (45 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Sebastian Ott, Martin Schwidefsky
[-- Attachment #1: 100-cio-docu.diff --]
[-- Type: text/plain, Size: 1339 bytes --]
From: Sebastian Ott <sebott@de.ibm.com>
Signed-off-by: Sebastian Ott <sebott@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
Documentation/kernel-parameters.txt | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
Index: quilt-2.6/Documentation/kernel-parameters.txt
===================================================================
--- quilt-2.6.orig/Documentation/kernel-parameters.txt
+++ quilt-2.6/Documentation/kernel-parameters.txt
@@ -369,8 +369,6 @@ and is between 256 and 4096 characters.
configured. Potentially dangerous and should only be
used if you are entirely sure of the consequences.
- chandev= [HW,NET] Generic channel device initialisation
-
checkreqprot [SELINUX] Set initial checkreqprot flag value.
Format: { "0" | "1" }
See security/selinux/Kconfig help text.
@@ -381,6 +379,12 @@ and is between 256 and 4096 characters.
Value can be changed at runtime via
/selinux/checkreqprot.
+ cio_ignore= [S390]
+ See Documentation/s390/CommonIO for details.
+
+ cio_msg= [S390]
+ See Documentation/s390/CommonIO for details.
+
clock= [BUGS=X86-32, HW] gettimeofday clocksource override.
[Deprecated]
Forces specified clocksource (if available) to be used
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 02/47] cio: Dump ccw device information in case of timeout.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
2007-12-20 15:19 ` [patch 01/47] Cleanup in Documentation/kernel-parameters.txt Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 03/47] cio: Use helpers instead of container_of() Martin Schwidefsky
` (44 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Sebastian Ott, Martin Schwidefsky
[-- Attachment #1: 101-cio-timeoutlog.diff --]
[-- Type: text/plain, Size: 4192 bytes --]
From: Sebastian Ott <sebott@de.ibm.com>
Information about a ccw device will be dumped in
case of a ccw timeout. This can be enabled with
the kernel parameter ccw_timeout_log.
Signed-off-by: Sebastian Ott <sebott@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
Documentation/kernel-parameters.txt | 3 ++
Documentation/s390/CommonIO | 5 +++
drivers/s390/cio/device_fsm.c | 50 ++++++++++++++++++++++++++++++++++++
3 files changed, 58 insertions(+)
Index: quilt-2.6/Documentation/kernel-parameters.txt
===================================================================
--- quilt-2.6.orig/Documentation/kernel-parameters.txt
+++ quilt-2.6/Documentation/kernel-parameters.txt
@@ -369,6 +369,9 @@ and is between 256 and 4096 characters.
configured. Potentially dangerous and should only be
used if you are entirely sure of the consequences.
+ ccw_timeout_log [S390]
+ See Documentation/s390/CommonIO for details.
+
checkreqprot [SELINUX] Set initial checkreqprot flag value.
Format: { "0" | "1" }
See security/selinux/Kconfig help text.
Index: quilt-2.6/Documentation/s390/CommonIO
===================================================================
--- quilt-2.6.orig/Documentation/s390/CommonIO
+++ quilt-2.6/Documentation/s390/CommonIO
@@ -4,6 +4,11 @@ S/390 common I/O-Layer - command line pa
Command line parameters
-----------------------
+* ccw_timeout_log
+
+ Enable logging of debug information in case of ccw device timeouts.
+
+
* cio_msg = yes | no
Determines whether information on found devices and sensed device
Index: quilt-2.6/drivers/s390/cio/device_fsm.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_fsm.c
+++ quilt-2.6/drivers/s390/cio/device_fsm.c
@@ -25,6 +25,8 @@
#include "ioasm.h"
#include "chp.h"
+static int timeout_log_enabled;
+
int
device_is_online(struct subchannel *sch)
{
@@ -82,6 +84,52 @@ int device_trigger_verify(struct subchan
return 0;
}
+static int __init ccw_timeout_log_setup(char *unused)
+{
+ timeout_log_enabled = 1;
+ return 1;
+}
+
+__setup("ccw_timeout_log", ccw_timeout_log_setup);
+
+static void ccw_timeout_log(struct ccw_device *cdev)
+{
+ struct schib schib;
+ struct subchannel *sch;
+ int cc;
+
+ sch = to_subchannel(cdev->dev.parent);
+ cc = stsch(sch->schid, &schib);
+
+ printk(KERN_WARNING "cio: ccw device timeout occurred at %llx, "
+ "device information:\n", get_clock());
+ printk(KERN_WARNING "cio: orb:\n");
+ print_hex_dump(KERN_WARNING, "cio: ", DUMP_PREFIX_NONE, 16, 1,
+ &sch->orb, sizeof(sch->orb), 0);
+ printk(KERN_WARNING "cio: ccw device bus id: %s\n", cdev->dev.bus_id);
+ printk(KERN_WARNING "cio: subchannel bus id: %s\n", sch->dev.bus_id);
+ printk(KERN_WARNING "cio: subchannel lpm: %02x, opm: %02x, "
+ "vpm: %02x\n", sch->lpm, sch->opm, sch->vpm);
+
+ if ((void *)(addr_t)sch->orb.cpa == &sch->sense_ccw ||
+ (void *)(addr_t)sch->orb.cpa == cdev->private->iccws)
+ printk(KERN_WARNING "cio: last channel program (intern):\n");
+ else
+ printk(KERN_WARNING "cio: last channel program:\n");
+
+ print_hex_dump(KERN_WARNING, "cio: ", DUMP_PREFIX_NONE, 16, 1,
+ (void *)(addr_t)sch->orb.cpa, sizeof(struct ccw1), 0);
+ printk(KERN_WARNING "cio: ccw device state: %d\n",
+ cdev->private->state);
+ printk(KERN_WARNING "cio: store subchannel returned: cc=%d\n", cc);
+ printk(KERN_WARNING "cio: schib:\n");
+ print_hex_dump(KERN_WARNING, "cio: ", DUMP_PREFIX_NONE, 16, 1,
+ &schib, sizeof(schib), 0);
+ printk(KERN_WARNING "cio: ccw device flags:\n");
+ print_hex_dump(KERN_WARNING, "cio: ", DUMP_PREFIX_NONE, 16, 1,
+ &cdev->private->flags, sizeof(cdev->private->flags), 0);
+}
+
/*
* Timeout function. It just triggers a DEV_EVENT_TIMEOUT.
*/
@@ -92,6 +140,8 @@ ccw_device_timeout(unsigned long data)
cdev = (struct ccw_device *) data;
spin_lock_irq(cdev->ccwlock);
+ if (timeout_log_enabled)
+ ccw_timeout_log(cdev);
dev_fsm_event(cdev, DEV_EVENT_TIMEOUT);
spin_unlock_irq(cdev->ccwlock);
}
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 03/47] cio: Use helpers instead of container_of().
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
2007-12-20 15:19 ` [patch 01/47] Cleanup in Documentation/kernel-parameters.txt Martin Schwidefsky
2007-12-20 15:19 ` [patch 02/47] cio: Dump ccw device information in case of timeout Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 04/47] cio: css_driver: Use consistent parameters Martin Schwidefsky
` (43 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Cornelia Huck, Martin Schwidefsky
[-- Attachment #1: 102-cio-helpers.diff --]
[-- Type: text/plain, Size: 2360 bytes --]
From: Cornelia Huck <cornelia.huck@de.ibm.com>
- Introduce to_cssdriver.
- Use to_xxx instead of container_of where possible.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/ccwgroup.c | 4 ++--
drivers/s390/cio/css.c | 6 +++---
drivers/s390/cio/css.h | 2 ++
3 files changed, 7 insertions(+), 5 deletions(-)
Index: quilt-2.6/drivers/s390/cio/ccwgroup.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/ccwgroup.c
+++ quilt-2.6/drivers/s390/cio/ccwgroup.c
@@ -35,8 +35,8 @@ ccwgroup_bus_match (struct device * dev,
struct ccwgroup_device *gdev;
struct ccwgroup_driver *gdrv;
- gdev = container_of(dev, struct ccwgroup_device, dev);
- gdrv = container_of(drv, struct ccwgroup_driver, driver);
+ gdev = to_ccwgroupdev(dev);
+ gdrv = to_ccwgroupdrv(drv);
if (gdev->creator_id == gdrv->driver_id)
return 1;
Index: quilt-2.6/drivers/s390/cio/css.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.c
+++ quilt-2.6/drivers/s390/cio/css.c
@@ -787,8 +787,8 @@ int sch_is_pseudo_sch(struct subchannel
static int
css_bus_match (struct device *dev, struct device_driver *drv)
{
- struct subchannel *sch = container_of (dev, struct subchannel, dev);
- struct css_driver *driver = container_of (drv, struct css_driver, drv);
+ struct subchannel *sch = to_subchannel(dev);
+ struct css_driver *driver = to_cssdriver(drv);
if (sch->st == driver->subchannel_type)
return 1;
@@ -802,7 +802,7 @@ css_probe (struct device *dev)
struct subchannel *sch;
sch = to_subchannel(dev);
- sch->driver = container_of (dev->driver, struct css_driver, drv);
+ sch->driver = to_cssdriver(dev->driver);
return (sch->driver->probe ? sch->driver->probe(sch) : 0);
}
Index: quilt-2.6/drivers/s390/cio/css.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.h
+++ quilt-2.6/drivers/s390/cio/css.h
@@ -134,6 +134,8 @@ struct css_driver {
void (*shutdown)(struct subchannel *);
};
+#define to_cssdriver(n) container_of(n, struct css_driver, drv)
+
/*
* all css_drivers have the css_bus_type
*/
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 04/47] cio: css_driver: Use consistent parameters.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (2 preceding siblings ...)
2007-12-20 15:19 ` [patch 03/47] cio: Use helpers instead of container_of() Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 05/47] cio: Reset sch->driver Martin Schwidefsky
` (42 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Cornelia Huck, Martin Schwidefsky
[-- Attachment #1: 103-cio-callbacks.diff --]
[-- Type: text/plain, Size: 8965 bytes --]
From: Cornelia Huck <cornelia.huck@de.ibm.com>
Make all callbacks in css_driver take a struct subchannel (and not
a struct device).
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/chsc.c | 16 ++++++++--------
drivers/s390/cio/cio.c | 4 ++--
drivers/s390/cio/css.c | 2 +-
drivers/s390/cio/css.h | 8 ++++----
drivers/s390/cio/device.c | 38 ++++++++++++++++++++++++--------------
drivers/s390/cio/device.h | 1 -
drivers/s390/cio/device_fsm.c | 19 +------------------
7 files changed, 40 insertions(+), 48 deletions(-)
Index: quilt-2.6/drivers/s390/cio/chsc.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/chsc.c
+++ quilt-2.6/drivers/s390/cio/chsc.c
@@ -132,7 +132,7 @@ static void terminate_internal_io(struct
device_set_intretry(sch);
/* Call handler. */
if (sch->driver && sch->driver->termination)
- sch->driver->termination(&sch->dev);
+ sch->driver->termination(sch);
}
static int
@@ -172,12 +172,12 @@ s390_subchannel_remove_chpid(struct devi
terminate_internal_io(sch);
/* Re-start path verification. */
if (sch->driver && sch->driver->verify)
- sch->driver->verify(&sch->dev);
+ sch->driver->verify(sch);
}
} else {
/* trigger path verification. */
if (sch->driver && sch->driver->verify)
- sch->driver->verify(&sch->dev);
+ sch->driver->verify(sch);
else if (sch->lpm == mask)
goto out_unreg;
}
@@ -279,7 +279,7 @@ __s390_process_res_acc(struct subchannel
if (!old_lpm && sch->lpm)
device_trigger_reprobe(sch);
else if (sch->driver && sch->driver->verify)
- sch->driver->verify(&sch->dev);
+ sch->driver->verify(sch);
out:
spin_unlock_irq(sch->lock);
put_device(&sch->dev);
@@ -549,7 +549,7 @@ __chp_add(struct subchannel_id schid, vo
| mask) & sch->opm;
if (sch->driver && sch->driver->verify)
- sch->driver->verify(&sch->dev);
+ sch->driver->verify(sch);
spin_unlock_irq(sch->lock);
put_device(&sch->dev);
@@ -589,7 +589,7 @@ static void __s390_subchannel_vary_chpid
if (!old_lpm)
device_trigger_reprobe(sch);
else if (sch->driver && sch->driver->verify)
- sch->driver->verify(&sch->dev);
+ sch->driver->verify(sch);
break;
}
sch->opm &= ~mask;
@@ -603,13 +603,13 @@ static void __s390_subchannel_vary_chpid
terminate_internal_io(sch);
/* Re-start path verification. */
if (sch->driver && sch->driver->verify)
- sch->driver->verify(&sch->dev);
+ sch->driver->verify(sch);
}
} else if (!sch->lpm) {
if (device_trigger_verify(sch) != 0)
css_schedule_eval(sch->schid);
} else if (sch->driver && sch->driver->verify)
- sch->driver->verify(&sch->dev);
+ sch->driver->verify(sch);
break;
}
spin_unlock_irqrestore(sch->lock, flags);
Index: quilt-2.6/drivers/s390/cio/cio.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/cio.c
+++ quilt-2.6/drivers/s390/cio/cio.c
@@ -147,7 +147,7 @@ cio_tpi(void)
spin_lock(sch->lock);
memcpy (&sch->schib.scsw, &irb->scsw, sizeof (struct scsw));
if (sch->driver && sch->driver->irq)
- sch->driver->irq(&sch->dev);
+ sch->driver->irq(sch);
spin_unlock(sch->lock);
irq_exit ();
_local_bh_enable();
@@ -680,7 +680,7 @@ do_IRQ (struct pt_regs *regs)
sizeof (irb->scsw));
/* Call interrupt handler if there is one. */
if (sch->driver && sch->driver->irq)
- sch->driver->irq(&sch->dev);
+ sch->driver->irq(sch);
}
if (sch)
spin_unlock(sch->lock);
Index: quilt-2.6/drivers/s390/cio/css.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.c
+++ quilt-2.6/drivers/s390/cio/css.c
@@ -293,7 +293,7 @@ static int css_evaluate_known_subchannel
action = UNREGISTER;
if (sch->driver && sch->driver->notify) {
spin_unlock_irqrestore(sch->lock, flags);
- ret = sch->driver->notify(&sch->dev, event);
+ ret = sch->driver->notify(sch, event);
spin_lock_irqsave(sch->lock, flags);
if (ret)
action = NONE;
Index: quilt-2.6/drivers/s390/cio/css.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.h
+++ quilt-2.6/drivers/s390/cio/css.h
@@ -125,10 +125,10 @@ struct subchannel;
struct css_driver {
unsigned int subchannel_type;
struct device_driver drv;
- void (*irq)(struct device *);
- int (*notify)(struct device *, int);
- void (*verify)(struct device *);
- void (*termination)(struct device *);
+ void (*irq)(struct subchannel *);
+ int (*notify)(struct subchannel *, int);
+ void (*verify)(struct subchannel *);
+ void (*termination)(struct subchannel *);
int (*probe)(struct subchannel *);
int (*remove)(struct subchannel *);
void (*shutdown)(struct subchannel *);
Index: quilt-2.6/drivers/s390/cio/device.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device.c
+++ quilt-2.6/drivers/s390/cio/device.c
@@ -115,11 +115,12 @@ static int ccw_uevent(struct device *dev
struct bus_type ccw_bus_type;
-static int io_subchannel_probe (struct subchannel *);
-static int io_subchannel_remove (struct subchannel *);
-static int io_subchannel_notify(struct device *, int);
-static void io_subchannel_verify(struct device *);
-static void io_subchannel_ioterm(struct device *);
+static void io_subchannel_irq(struct subchannel *);
+static int io_subchannel_probe(struct subchannel *);
+static int io_subchannel_remove(struct subchannel *);
+static int io_subchannel_notify(struct subchannel *, int);
+static void io_subchannel_verify(struct subchannel *);
+static void io_subchannel_ioterm(struct subchannel *);
static void io_subchannel_shutdown(struct subchannel *);
static struct css_driver io_subchannel_driver = {
@@ -1096,6 +1097,18 @@ out:
put_device(&cdev->dev);
}
+static void io_subchannel_irq(struct subchannel *sch)
+{
+ struct ccw_device *cdev;
+
+ cdev = sch->dev.driver_data;
+
+ CIO_TRACE_EVENT(3, "IRQ");
+ CIO_TRACE_EVENT(3, sch->dev.bus_id);
+ if (cdev)
+ dev_fsm_event(cdev, DEV_EVENT_INTERRUPT);
+}
+
static int
io_subchannel_probe (struct subchannel *sch)
{
@@ -1183,12 +1196,11 @@ io_subchannel_remove (struct subchannel
return 0;
}
-static int
-io_subchannel_notify(struct device *dev, int event)
+static int io_subchannel_notify(struct subchannel *sch, int event)
{
struct ccw_device *cdev;
- cdev = dev->driver_data;
+ cdev = sch->dev.driver_data;
if (!cdev)
return 0;
if (!cdev->drv)
@@ -1198,22 +1210,20 @@ io_subchannel_notify(struct device *dev,
return cdev->drv->notify ? cdev->drv->notify(cdev, event) : 0;
}
-static void
-io_subchannel_verify(struct device *dev)
+static void io_subchannel_verify(struct subchannel *sch)
{
struct ccw_device *cdev;
- cdev = dev->driver_data;
+ cdev = sch->dev.driver_data;
if (cdev)
dev_fsm_event(cdev, DEV_EVENT_VERIFY);
}
-static void
-io_subchannel_ioterm(struct device *dev)
+static void io_subchannel_ioterm(struct subchannel *sch)
{
struct ccw_device *cdev;
- cdev = dev->driver_data;
+ cdev = sch->dev.driver_data;
if (!cdev)
return;
/* Internal I/O will be retried by the interrupt handler. */
Index: quilt-2.6/drivers/s390/cio/device_fsm.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_fsm.c
+++ quilt-2.6/drivers/s390/cio/device_fsm.c
@@ -399,7 +399,7 @@ ccw_device_oper_notify(struct work_struc
sch = to_subchannel(cdev->dev.parent);
if (sch->driver && sch->driver->notify) {
spin_unlock_irqrestore(cdev->ccwlock, flags);
- ret = sch->driver->notify(&sch->dev, CIO_OPER);
+ ret = sch->driver->notify(sch, CIO_OPER);
spin_lock_irqsave(cdev->ccwlock, flags);
} else
ret = 0;
@@ -1273,21 +1273,4 @@ fsm_func_t *dev_jumptable[NR_DEV_STATES]
},
};
-/*
- * io_subchannel_irq is called for "real" interrupts or for status
- * pending conditions on msch.
- */
-void
-io_subchannel_irq (struct device *pdev)
-{
- struct ccw_device *cdev;
-
- cdev = to_subchannel(pdev)->dev.driver_data;
-
- CIO_TRACE_EVENT (3, "IRQ");
- CIO_TRACE_EVENT (3, pdev->bus_id);
- if (cdev)
- dev_fsm_event(cdev, DEV_EVENT_INTERRUPT);
-}
-
EXPORT_SYMBOL_GPL(ccw_device_set_timeout);
Index: quilt-2.6/drivers/s390/cio/device.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device.h
+++ quilt-2.6/drivers/s390/cio/device.h
@@ -74,7 +74,6 @@ extern struct workqueue_struct *ccw_devi
extern wait_queue_head_t ccw_device_init_wq;
extern atomic_t ccw_device_init_count;
-void io_subchannel_irq (struct device *pdev);
void io_subchannel_recog_done(struct ccw_device *cdev);
int ccw_device_cancel_halt_clear(struct ccw_device *);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 05/47] cio: Reset sch->driver.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (3 preceding siblings ...)
2007-12-20 15:19 ` [patch 04/47] cio: css_driver: Use consistent parameters Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 06/47] cio: Add css_driver_{register,unregister} Martin Schwidefsky
` (41 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Cornelia Huck, Martin Schwidefsky
[-- Attachment #1: 104-cio-reset.diff --]
[-- Type: text/plain, Size: 1695 bytes --]
From: Cornelia Huck <cornelia.huck@de.ibm.com>
sch->driver needs to be reset to NULL on failed probe and after
remove. We also need to check for sch->driver on shutdown.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/css.c | 22 +++++++++++++---------
1 file changed, 13 insertions(+), 9 deletions(-)
Index: quilt-2.6/drivers/s390/cio/css.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.c
+++ quilt-2.6/drivers/s390/cio/css.c
@@ -796,32 +796,36 @@ css_bus_match (struct device *dev, struc
return 0;
}
-static int
-css_probe (struct device *dev)
+static int css_probe(struct device *dev)
{
struct subchannel *sch;
+ int ret;
sch = to_subchannel(dev);
sch->driver = to_cssdriver(dev->driver);
- return (sch->driver->probe ? sch->driver->probe(sch) : 0);
+ ret = sch->driver->probe ? sch->driver->probe(sch) : 0;
+ if (ret)
+ sch->driver = NULL;
+ return ret;
}
-static int
-css_remove (struct device *dev)
+static int css_remove(struct device *dev)
{
struct subchannel *sch;
+ int ret;
sch = to_subchannel(dev);
- return (sch->driver->remove ? sch->driver->remove(sch) : 0);
+ ret = sch->driver->remove ? sch->driver->remove(sch) : 0;
+ sch->driver = NULL;
+ return ret;
}
-static void
-css_shutdown (struct device *dev)
+static void css_shutdown(struct device *dev)
{
struct subchannel *sch;
sch = to_subchannel(dev);
- if (sch->driver->shutdown)
+ if (sch->driver && sch->driver->shutdown)
sch->driver->shutdown(sch);
}
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 06/47] cio: Add css_driver_{register,unregister}.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (4 preceding siblings ...)
2007-12-20 15:19 ` [patch 05/47] cio: Reset sch->driver Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 07/47] cio: Cleanup debug feature usage Martin Schwidefsky
` (40 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Cornelia Huck, Martin Schwidefsky
[-- Attachment #1: 105-cio-cssdrv.diff --]
[-- Type: text/plain, Size: 3555 bytes --]
From: Cornelia Huck <cornelia.huck@de.ibm.com>
Add wrapper functions for driver_register and driver_unregister so
that css drivers don't need to muck with struct device_driver
directly.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/css.c | 27 +++++++++++++++++++++++++++
drivers/s390/cio/css.h | 4 ++++
drivers/s390/cio/device.c | 10 ++++------
3 files changed, 35 insertions(+), 6 deletions(-)
Index: quilt-2.6/drivers/s390/cio/css.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.c
+++ quilt-2.6/drivers/s390/cio/css.c
@@ -837,6 +837,33 @@ struct bus_type css_bus_type = {
.shutdown = css_shutdown,
};
+/**
+ * css_driver_register - register a css driver
+ * @cdrv: css driver to register
+ *
+ * This is mainly a wrapper around driver_register that sets name
+ * and bus_type in the embedded struct device_driver correctly.
+ */
+int css_driver_register(struct css_driver *cdrv)
+{
+ cdrv->drv.name = cdrv->name;
+ cdrv->drv.bus = &css_bus_type;
+ return driver_register(&cdrv->drv);
+}
+EXPORT_SYMBOL_GPL(css_driver_register);
+
+/**
+ * css_driver_unregister - unregister a css driver
+ * @cdrv: css driver to unregister
+ *
+ * This is a wrapper around driver_unregister.
+ */
+void css_driver_unregister(struct css_driver *cdrv)
+{
+ driver_unregister(&cdrv->drv);
+}
+EXPORT_SYMBOL_GPL(css_driver_unregister);
+
subsys_initcall(init_channel_subsystem);
MODULE_LICENSE("GPL");
Index: quilt-2.6/drivers/s390/cio/css.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.h
+++ quilt-2.6/drivers/s390/cio/css.h
@@ -132,6 +132,7 @@ struct css_driver {
int (*probe)(struct subchannel *);
int (*remove)(struct subchannel *);
void (*shutdown)(struct subchannel *);
+ const char *name;
};
#define to_cssdriver(n) container_of(n, struct css_driver, drv)
@@ -141,6 +142,9 @@ struct css_driver {
*/
extern struct bus_type css_bus_type;
+extern int css_driver_register(struct css_driver *);
+extern void css_driver_unregister(struct css_driver *);
+
extern void css_sch_device_unregister(struct subchannel *);
extern struct subchannel * get_subchannel_by_schid(struct subchannel_id);
extern int css_init_done;
Index: quilt-2.6/drivers/s390/cio/device.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device.c
+++ quilt-2.6/drivers/s390/cio/device.c
@@ -125,10 +125,7 @@ static void io_subchannel_shutdown(struc
static struct css_driver io_subchannel_driver = {
.subchannel_type = SUBCHANNEL_TYPE_IO,
- .drv = {
- .name = "io_subchannel",
- .bus = &css_bus_type,
- },
+ .name = "io_subchannel",
.irq = io_subchannel_irq,
.notify = io_subchannel_notify,
.verify = io_subchannel_verify,
@@ -167,7 +164,8 @@ init_ccw_bus_type (void)
if ((ret = bus_register (&ccw_bus_type)))
goto out_err;
- if ((ret = driver_register(&io_subchannel_driver.drv)))
+ ret = css_driver_register(&io_subchannel_driver);
+ if (ret)
goto out_err;
wait_event(ccw_device_init_wq,
@@ -187,7 +185,7 @@ out_err:
static void __exit
cleanup_ccw_bus_type (void)
{
- driver_unregister(&io_subchannel_driver.drv);
+ css_driver_unregister(&io_subchannel_driver);
bus_unregister(&ccw_bus_type);
destroy_workqueue(ccw_device_notify_work);
destroy_workqueue(ccw_device_work);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 07/47] cio: Cleanup debug feature usage.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (5 preceding siblings ...)
2007-12-20 15:19 ` [patch 06/47] cio: Add css_driver_{register,unregister} Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 08/47] cio: Introduce subchannel->private Martin Schwidefsky
` (39 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Cornelia Huck, Martin Schwidefsky
[-- Attachment #1: 106-cio-s390dbf.diff --]
[-- Type: text/plain, Size: 7064 bytes --]
From: Cornelia Huck <cornelia.huck@de.ibm.com>
Cleanup cio_debug.h.
Also make CIO_DEBUG add the "cio:" prefix to the printk string
so that it isn't needed for the debug feature.
Fix outdated comments for cio_debug_init() and clean it up.
Enlarge cio_crw to the same size as cio_msg so we may actually
find some relevant information there.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/cio.c | 38 ++++++++++++++++++--------------------
drivers/s390/cio/cio_debug.h | 22 +++++++++++-----------
drivers/s390/cio/device_fsm.c | 8 ++++----
3 files changed, 33 insertions(+), 35 deletions(-)
Index: quilt-2.6/drivers/s390/cio/cio.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/cio.c
+++ quilt-2.6/drivers/s390/cio/cio.c
@@ -56,39 +56,37 @@ __setup ("cio_msg=", cio_setup);
/*
* Function: cio_debug_init
- * Initializes three debug logs (under /proc/s390dbf) for common I/O:
- * - cio_msg logs the messages which are printk'ed when CONFIG_DEBUG_IO is on
+ * Initializes three debug logs for common I/O:
+ * - cio_msg logs generic cio messages
* - cio_trace logs the calling of different functions
- * - cio_crw logs the messages which are printk'ed when CONFIG_DEBUG_CRW is on
- * debug levels depend on CONFIG_DEBUG_IO resp. CONFIG_DEBUG_CRW
+ * - cio_crw logs machine check related cio messages
*/
-static int __init
-cio_debug_init (void)
+static int __init cio_debug_init(void)
{
- cio_debug_msg_id = debug_register ("cio_msg", 16, 4, 16*sizeof (long));
+ cio_debug_msg_id = debug_register("cio_msg", 16, 4, 16 * sizeof(long));
if (!cio_debug_msg_id)
goto out_unregister;
- debug_register_view (cio_debug_msg_id, &debug_sprintf_view);
- debug_set_level (cio_debug_msg_id, 2);
- cio_debug_trace_id = debug_register ("cio_trace", 16, 4, 16);
+ debug_register_view(cio_debug_msg_id, &debug_sprintf_view);
+ debug_set_level(cio_debug_msg_id, 2);
+ cio_debug_trace_id = debug_register("cio_trace", 16, 4, 16);
if (!cio_debug_trace_id)
goto out_unregister;
- debug_register_view (cio_debug_trace_id, &debug_hex_ascii_view);
- debug_set_level (cio_debug_trace_id, 2);
- cio_debug_crw_id = debug_register ("cio_crw", 4, 4, 16*sizeof (long));
+ debug_register_view(cio_debug_trace_id, &debug_hex_ascii_view);
+ debug_set_level(cio_debug_trace_id, 2);
+ cio_debug_crw_id = debug_register("cio_crw", 16, 4, 16 * sizeof(long));
if (!cio_debug_crw_id)
goto out_unregister;
- debug_register_view (cio_debug_crw_id, &debug_sprintf_view);
- debug_set_level (cio_debug_crw_id, 2);
+ debug_register_view(cio_debug_crw_id, &debug_sprintf_view);
+ debug_set_level(cio_debug_crw_id, 4);
return 0;
out_unregister:
if (cio_debug_msg_id)
- debug_unregister (cio_debug_msg_id);
+ debug_unregister(cio_debug_msg_id);
if (cio_debug_trace_id)
- debug_unregister (cio_debug_trace_id);
+ debug_unregister(cio_debug_trace_id);
if (cio_debug_crw_id)
- debug_unregister (cio_debug_crw_id);
+ debug_unregister(cio_debug_crw_id);
printk(KERN_WARNING"cio: could not initialize debugging\n");
return -1;
}
@@ -567,7 +565,7 @@ cio_validate_subchannel (struct subchann
*/
if (sch->st != 0) {
CIO_DEBUG(KERN_INFO, 0,
- "cio: Subchannel 0.%x.%04x reports "
+ "Subchannel 0.%x.%04x reports "
"non-I/O subchannel type %04X\n",
sch->schid.ssid, sch->schid.sch_no, sch->st);
/* We stop here for non-io subchannels. */
@@ -600,7 +598,7 @@ cio_validate_subchannel (struct subchann
sch->lpm = sch->schib.pmcw.pam & sch->opm;
CIO_DEBUG(KERN_INFO, 0,
- "cio: Detected device %04x on subchannel 0.%x.%04X"
+ "Detected device %04x on subchannel 0.%x.%04X"
" - PIM = %02X, PAM = %02X, POM = %02X\n",
sch->schib.pmcw.dev, sch->schid.ssid,
sch->schid.sch_no, sch->schib.pmcw.pim,
Index: quilt-2.6/drivers/s390/cio/cio_debug.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/cio_debug.h
+++ quilt-2.6/drivers/s390/cio/cio_debug.h
@@ -8,20 +8,19 @@ extern debug_info_t *cio_debug_msg_id;
extern debug_info_t *cio_debug_trace_id;
extern debug_info_t *cio_debug_crw_id;
-#define CIO_TRACE_EVENT(imp, txt) do { \
+#define CIO_TRACE_EVENT(imp, txt) do { \
debug_text_event(cio_debug_trace_id, imp, txt); \
} while (0)
-#define CIO_MSG_EVENT(imp, args...) do { \
- debug_sprintf_event(cio_debug_msg_id, imp , ##args); \
+#define CIO_MSG_EVENT(imp, args...) do { \
+ debug_sprintf_event(cio_debug_msg_id, imp , ##args); \
} while (0)
-#define CIO_CRW_EVENT(imp, args...) do { \
- debug_sprintf_event(cio_debug_crw_id, imp , ##args); \
+#define CIO_CRW_EVENT(imp, args...) do { \
+ debug_sprintf_event(cio_debug_crw_id, imp , ##args); \
} while (0)
-static inline void
-CIO_HEX_EVENT(int level, void *data, int length)
+static inline void CIO_HEX_EVENT(int level, void *data, int length)
{
if (unlikely(!cio_debug_trace_id))
return;
@@ -32,9 +31,10 @@ CIO_HEX_EVENT(int level, void *data, int
}
}
-#define CIO_DEBUG(printk_level,event_level,msg...) ({ \
- if (cio_show_msg) printk(printk_level msg); \
- CIO_MSG_EVENT (event_level, msg); \
-})
+#define CIO_DEBUG(printk_level, event_level, msg...) do { \
+ if (cio_show_msg) \
+ printk(printk_level "cio: " msg); \
+ CIO_MSG_EVENT(event_level, msg); \
+ } while (0)
#endif
Index: quilt-2.6/drivers/s390/cio/device_fsm.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_fsm.c
+++ quilt-2.6/drivers/s390/cio/device_fsm.c
@@ -318,7 +318,7 @@ ccw_device_recog_done(struct ccw_device
switch (state) {
case DEV_STATE_NOT_OPER:
CIO_DEBUG(KERN_WARNING, 2,
- "cio: SenseID : unknown device %04x on subchannel "
+ "SenseID : unknown device %04x on subchannel "
"0.%x.%04x\n", cdev->private->dev_id.devno,
sch->schid.ssid, sch->schid.sch_no);
break;
@@ -344,7 +344,7 @@ ccw_device_recog_done(struct ccw_device
}
/* Issue device info message. */
CIO_DEBUG(KERN_INFO, 2,
- "cio: SenseID : device 0.%x.%04x reports: "
+ "SenseID : device 0.%x.%04x reports: "
"CU Type/Mod = %04X/%02X, Dev Type/Mod = "
"%04X/%02X\n",
cdev->private->dev_id.ssid,
@@ -354,7 +354,7 @@ ccw_device_recog_done(struct ccw_device
break;
case DEV_STATE_BOXED:
CIO_DEBUG(KERN_WARNING, 2,
- "cio: SenseID : boxed device %04x on subchannel "
+ "SenseID : boxed device %04x on subchannel "
"0.%x.%04x\n", cdev->private->dev_id.devno,
sch->schid.ssid, sch->schid.sch_no);
break;
@@ -439,7 +439,7 @@ ccw_device_done(struct ccw_device *cdev,
if (state == DEV_STATE_BOXED)
CIO_DEBUG(KERN_WARNING, 2,
- "cio: Boxed device %04x on subchannel %04x\n",
+ "Boxed device %04x on subchannel %04x\n",
cdev->private->dev_id.devno, sch->schid.sch_no);
if (cdev->private->flags.donotify) {
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 08/47] cio: Introduce subchannel->private.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (6 preceding siblings ...)
2007-12-20 15:19 ` [patch 07/47] cio: Cleanup debug feature usage Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 09/47] cio: Extend adapter interrupt interface Martin Schwidefsky
` (38 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Cornelia Huck, Martin Schwidefsky
[-- Attachment #1: 107-cio-private.diff --]
[-- Type: text/plain, Size: 26034 bytes --]
From: Cornelia Huck <cornelia.huck@de.ibm.com>
Introduce a private pointer in struct subchannel to store
per-subchannel type data (cannot use dev->priv since this
is already used for something else).
Create a new header io_sch.h for I/O subchannel specific structures
and instructions.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/cio.c | 43 ++++++----
drivers/s390/cio/cio.h | 84 +++++++-------------
drivers/s390/cio/css.c | 2
drivers/s390/cio/css.h | 58 --------------
drivers/s390/cio/device.c | 16 +++
drivers/s390/cio/device.h | 2
drivers/s390/cio/device_fsm.c | 13 +--
drivers/s390/cio/device_id.c | 9 +-
drivers/s390/cio/device_pgid.c | 6 +
drivers/s390/cio/device_status.c | 13 +--
drivers/s390/cio/io_sch.h | 161 +++++++++++++++++++++++++++++++++++++++
drivers/s390/cio/ioasm.h | 66 ---------------
12 files changed, 261 insertions(+), 212 deletions(-)
Index: quilt-2.6/drivers/s390/cio/cio.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/cio.c
+++ quilt-2.6/drivers/s390/cio/cio.c
@@ -28,6 +28,7 @@
#include "css.h"
#include "chsc.h"
#include "ioasm.h"
+#include "io_sch.h"
#include "blacklist.h"
#include "cio_debug.h"
#include "chp.h"
@@ -182,33 +183,35 @@ cio_start_key (struct subchannel *sch, /
{
char dbf_txt[15];
int ccode;
+ struct orb *orb;
- CIO_TRACE_EVENT (4, "stIO");
- CIO_TRACE_EVENT (4, sch->dev.bus_id);
+ CIO_TRACE_EVENT(4, "stIO");
+ CIO_TRACE_EVENT(4, sch->dev.bus_id);
+ orb = &to_io_private(sch)->orb;
/* sch is always under 2G. */
- sch->orb.intparm = (__u32)(unsigned long)sch;
- sch->orb.fmt = 1;
+ orb->intparm = (u32)(addr_t)sch;
+ orb->fmt = 1;
- sch->orb.pfch = sch->options.prefetch == 0;
- sch->orb.spnd = sch->options.suspend;
- sch->orb.ssic = sch->options.suspend && sch->options.inter;
- sch->orb.lpm = (lpm != 0) ? lpm : sch->lpm;
+ orb->pfch = sch->options.prefetch == 0;
+ orb->spnd = sch->options.suspend;
+ orb->ssic = sch->options.suspend && sch->options.inter;
+ orb->lpm = (lpm != 0) ? lpm : sch->lpm;
#ifdef CONFIG_64BIT
/*
* for 64 bit we always support 64 bit IDAWs with 4k page size only
*/
- sch->orb.c64 = 1;
- sch->orb.i2k = 0;
+ orb->c64 = 1;
+ orb->i2k = 0;
#endif
- sch->orb.key = key >> 4;
+ orb->key = key >> 4;
/* issue "Start Subchannel" */
- sch->orb.cpa = (__u32) __pa (cpa);
- ccode = ssch (sch->schid, &sch->orb);
+ orb->cpa = (__u32) __pa(cpa);
+ ccode = ssch(sch->schid, orb);
/* process condition code */
- sprintf (dbf_txt, "ccode:%d", ccode);
- CIO_TRACE_EVENT (4, dbf_txt);
+ sprintf(dbf_txt, "ccode:%d", ccode);
+ CIO_TRACE_EVENT(4, dbf_txt);
switch (ccode) {
case 0:
@@ -423,7 +426,7 @@ cio_enable_subchannel (struct subchannel
for (retry = 5, ret = 0; retry > 0; retry--) {
sch->schib.pmcw.ena = 1;
sch->schib.pmcw.isc = isc;
- sch->schib.pmcw.intparm = (__u32)(unsigned long)sch;
+ sch->schib.pmcw.intparm = (u32)(addr_t)sch;
ret = cio_modify(sch);
if (ret == -ENODEV)
break;
@@ -696,8 +699,14 @@ do_IRQ (struct pt_regs *regs)
#ifdef CONFIG_CCW_CONSOLE
static struct subchannel console_subchannel;
+static struct io_subchannel_private console_priv;
static int console_subchannel_in_use;
+void *cio_get_console_priv(void)
+{
+ return &console_priv;
+}
+
/*
* busy wait for the next interrupt on the console
*/
@@ -802,7 +811,7 @@ cio_probe_console(void)
ctl_set_bit(6, 24);
console_subchannel.schib.pmcw.isc = 7;
console_subchannel.schib.pmcw.intparm =
- (__u32)(unsigned long)&console_subchannel;
+ (u32)(addr_t)&console_subchannel;
ret = cio_modify(&console_subchannel);
if (ret) {
console_subchannel_in_use = 0;
Index: quilt-2.6/drivers/s390/cio/cio.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/cio.h
+++ quilt-2.6/drivers/s390/cio/cio.h
@@ -11,32 +11,32 @@
* path management control word
*/
struct pmcw {
- __u32 intparm; /* interruption parameter */
- __u32 qf : 1; /* qdio facility */
- __u32 res0 : 1; /* reserved zeros */
- __u32 isc : 3; /* interruption sublass */
- __u32 res5 : 3; /* reserved zeros */
- __u32 ena : 1; /* enabled */
- __u32 lm : 2; /* limit mode */
- __u32 mme : 2; /* measurement-mode enable */
- __u32 mp : 1; /* multipath mode */
- __u32 tf : 1; /* timing facility */
- __u32 dnv : 1; /* device number valid */
- __u32 dev : 16; /* device number */
- __u8 lpm; /* logical path mask */
- __u8 pnom; /* path not operational mask */
- __u8 lpum; /* last path used mask */
- __u8 pim; /* path installed mask */
- __u16 mbi; /* measurement-block index */
- __u8 pom; /* path operational mask */
- __u8 pam; /* path available mask */
- __u8 chpid[8]; /* CHPID 0-7 (if available) */
- __u32 unused1 : 8; /* reserved zeros */
- __u32 st : 3; /* subchannel type */
- __u32 unused2 : 18; /* reserved zeros */
- __u32 mbfc : 1; /* measurement block format control */
- __u32 xmwme : 1; /* extended measurement word mode enable */
- __u32 csense : 1; /* concurrent sense; can be enabled ...*/
+ u32 intparm; /* interruption parameter */
+ u32 qf : 1; /* qdio facility */
+ u32 res0 : 1; /* reserved zeros */
+ u32 isc : 3; /* interruption sublass */
+ u32 res5 : 3; /* reserved zeros */
+ u32 ena : 1; /* enabled */
+ u32 lm : 2; /* limit mode */
+ u32 mme : 2; /* measurement-mode enable */
+ u32 mp : 1; /* multipath mode */
+ u32 tf : 1; /* timing facility */
+ u32 dnv : 1; /* device number valid */
+ u32 dev : 16; /* device number */
+ u8 lpm; /* logical path mask */
+ u8 pnom; /* path not operational mask */
+ u8 lpum; /* last path used mask */
+ u8 pim; /* path installed mask */
+ u16 mbi; /* measurement-block index */
+ u8 pom; /* path operational mask */
+ u8 pam; /* path available mask */
+ u8 chpid[8]; /* CHPID 0-7 (if available) */
+ u32 unused1 : 8; /* reserved zeros */
+ u32 st : 3; /* subchannel type */
+ u32 unused2 : 18; /* reserved zeros */
+ u32 mbfc : 1; /* measurement block format control */
+ u32 xmwme : 1; /* extended measurement word mode enable */
+ u32 csense : 1; /* concurrent sense; can be enabled ...*/
/* ... per MSCH, however, if facility */
/* ... is not installed, this results */
/* ... in an operand exception. */
@@ -52,31 +52,6 @@ struct schib {
__u8 mda[4]; /* model dependent area */
} __attribute__ ((packed,aligned(4)));
-/*
- * operation request block
- */
-struct orb {
- __u32 intparm; /* interruption parameter */
- __u32 key : 4; /* flags, like key, suspend control, etc. */
- __u32 spnd : 1; /* suspend control */
- __u32 res1 : 1; /* reserved */
- __u32 mod : 1; /* modification control */
- __u32 sync : 1; /* synchronize control */
- __u32 fmt : 1; /* format control */
- __u32 pfch : 1; /* prefetch control */
- __u32 isic : 1; /* initial-status-interruption control */
- __u32 alcc : 1; /* address-limit-checking control */
- __u32 ssic : 1; /* suppress-suspended-interr. control */
- __u32 res2 : 1; /* reserved */
- __u32 c64 : 1; /* IDAW/QDIO 64 bit control */
- __u32 i2k : 1; /* IDAW 2/4kB block size control */
- __u32 lpm : 8; /* logical path mask */
- __u32 ils : 1; /* incorrect length */
- __u32 zero : 6; /* reserved zeros */
- __u32 orbx : 1; /* ORB extension control */
- __u32 cpa; /* channel program address */
-} __attribute__ ((packed,aligned(4)));
-
/* subchannel data structure used by I/O subroutines */
struct subchannel {
struct subchannel_id schid;
@@ -99,11 +74,10 @@ struct subchannel {
__u8 lpm; /* logical path mask */
__u8 opm; /* operational path mask */
struct schib schib; /* subchannel information block */
- struct orb orb; /* operation request block */
- struct ccw1 sense_ccw; /* static ccw for sense command */
struct chsc_ssd_info ssd_info; /* subchannel description */
struct device dev; /* entry in device tree */
struct css_driver *driver;
+ void *private; /* private per subchannel type data */
} __attribute__ ((aligned(8)));
#define IO_INTERRUPT_TYPE 0 /* I/O interrupt type */
@@ -133,10 +107,12 @@ extern void cio_release_console(void);
extern int cio_is_console(struct subchannel_id);
extern struct subchannel *cio_get_console_subchannel(void);
extern spinlock_t * cio_get_console_lock(void);
+extern void *cio_get_console_priv(void);
#else
#define cio_is_console(schid) 0
#define cio_get_console_subchannel() NULL
-#define cio_get_console_lock() NULL;
+#define cio_get_console_lock() NULL
+#define cio_get_console_priv() NULL
#endif
extern int cio_show_msg;
Index: quilt-2.6/drivers/s390/cio/css.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.c
+++ quilt-2.6/drivers/s390/cio/css.c
@@ -77,7 +77,7 @@ css_alloc_subchannel(struct subchannel_i
* This is fine even on 64bit since the subchannel is always located
* under 2G.
*/
- sch->schib.pmcw.intparm = (__u32)(unsigned long)sch;
+ sch->schib.pmcw.intparm = (u32)(addr_t)sch;
ret = cio_modify(sch);
if (ret) {
kfree(sch->lock);
Index: quilt-2.6/drivers/s390/cio/css.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.h
+++ quilt-2.6/drivers/s390/cio/css.h
@@ -58,64 +58,6 @@ struct pgid {
__u32 tod_high; /* high word TOD clock */
} __attribute__ ((packed));
-#define MAX_CIWS 8
-
-/*
- * sense-id response buffer layout
- */
-struct senseid {
- /* common part */
- __u8 reserved; /* always 0x'FF' */
- __u16 cu_type; /* control unit type */
- __u8 cu_model; /* control unit model */
- __u16 dev_type; /* device type */
- __u8 dev_model; /* device model */
- __u8 unused; /* padding byte */
- /* extended part */
- struct ciw ciw[MAX_CIWS]; /* variable # of CIWs */
-} __attribute__ ((packed,aligned(4)));
-
-struct ccw_device_private {
- struct ccw_device *cdev;
- struct subchannel *sch;
- int state; /* device state */
- atomic_t onoff;
- unsigned long registered;
- struct ccw_dev_id dev_id; /* device id */
- struct subchannel_id schid; /* subchannel number */
- __u8 imask; /* lpm mask for SNID/SID/SPGID */
- int iretry; /* retry counter SNID/SID/SPGID */
- struct {
- unsigned int fast:1; /* post with "channel end" */
- unsigned int repall:1; /* report every interrupt status */
- unsigned int pgroup:1; /* do path grouping */
- unsigned int force:1; /* allow forced online */
- } __attribute__ ((packed)) options;
- struct {
- unsigned int pgid_single:1; /* use single path for Set PGID */
- unsigned int esid:1; /* Ext. SenseID supported by HW */
- unsigned int dosense:1; /* delayed SENSE required */
- unsigned int doverify:1; /* delayed path verification */
- unsigned int donotify:1; /* call notify function */
- unsigned int recog_done:1; /* dev. recog. complete */
- unsigned int fake_irb:1; /* deliver faked irb */
- unsigned int intretry:1; /* retry internal operation */
- } __attribute__((packed)) flags;
- unsigned long intparm; /* user interruption parameter */
- struct qdio_irq *qdio_data;
- struct irb irb; /* device status */
- struct senseid senseid; /* SenseID info */
- struct pgid pgid[8]; /* path group IDs per chpid*/
- struct ccw1 iccws[2]; /* ccws for SNID/SID/SPGID commands */
- struct work_struct kick_work;
- wait_queue_head_t wait_q;
- struct timer_list timer;
- void *cmb; /* measurement information */
- struct list_head cmb_list; /* list of measured devices */
- u64 cmb_start_time; /* clock value of cmb reset */
- void *cmb_wait; /* deferred cmb enable/disable */
-};
-
/*
* A css driver handles all subchannels of one type.
* Currently, we only care about I/O subchannels (type 0), these
Index: quilt-2.6/drivers/s390/cio/device.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device.c
+++ quilt-2.6/drivers/s390/cio/device.c
@@ -28,6 +28,7 @@
#include "css.h"
#include "device.h"
#include "ioasm.h"
+#include "io_sch.h"
/******************* bus type handling ***********************/
@@ -1159,10 +1160,16 @@ io_subchannel_probe (struct subchannel *
queue_work(slow_path_wq, &cdev->private->kick_work);
return 0;
}
+ /* Allocate I/O subchannel private data. */
+ sch->private = kzalloc(sizeof(struct io_subchannel_private),
+ GFP_KERNEL | GFP_DMA);
+ if (!sch->private)
+ return -ENOMEM;
cdev = io_subchannel_create_ccwdev(sch);
- if (IS_ERR(cdev))
+ if (IS_ERR(cdev)) {
+ kfree(sch->private);
return PTR_ERR(cdev);
-
+ }
rc = io_subchannel_recog(cdev, sch);
if (rc) {
spin_lock_irqsave(sch->lock, flags);
@@ -1170,6 +1177,7 @@ io_subchannel_probe (struct subchannel *
spin_unlock_irqrestore(sch->lock, flags);
if (cdev->dev.release)
cdev->dev.release(&cdev->dev);
+ kfree(sch->private);
}
return rc;
@@ -1191,6 +1199,7 @@ io_subchannel_remove (struct subchannel
spin_unlock_irqrestore(cdev->ccwlock, flags);
ccw_device_unregister(cdev);
put_device(&cdev->dev);
+ kfree(sch->private);
return 0;
}
@@ -1279,6 +1288,9 @@ ccw_device_console_enable (struct ccw_de
{
int rc;
+ /* Attach subchannel private data. */
+ sch->private = cio_get_console_priv();
+ memset(sch->private, 0, sizeof(struct io_subchannel_private));
/* Initialize the ccw_device structure. */
cdev->dev.parent= &sch->dev;
rc = io_subchannel_recog(cdev, sch);
Index: quilt-2.6/drivers/s390/cio/device_fsm.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_fsm.c
+++ quilt-2.6/drivers/s390/cio/device_fsm.c
@@ -96,29 +96,32 @@ static void ccw_timeout_log(struct ccw_d
{
struct schib schib;
struct subchannel *sch;
+ struct io_subchannel_private *private;
int cc;
sch = to_subchannel(cdev->dev.parent);
+ private = to_io_private(sch);
cc = stsch(sch->schid, &schib);
printk(KERN_WARNING "cio: ccw device timeout occurred at %llx, "
"device information:\n", get_clock());
printk(KERN_WARNING "cio: orb:\n");
print_hex_dump(KERN_WARNING, "cio: ", DUMP_PREFIX_NONE, 16, 1,
- &sch->orb, sizeof(sch->orb), 0);
+ &private->orb, sizeof(private->orb), 0);
printk(KERN_WARNING "cio: ccw device bus id: %s\n", cdev->dev.bus_id);
printk(KERN_WARNING "cio: subchannel bus id: %s\n", sch->dev.bus_id);
printk(KERN_WARNING "cio: subchannel lpm: %02x, opm: %02x, "
"vpm: %02x\n", sch->lpm, sch->opm, sch->vpm);
- if ((void *)(addr_t)sch->orb.cpa == &sch->sense_ccw ||
- (void *)(addr_t)sch->orb.cpa == cdev->private->iccws)
+ if ((void *)(addr_t)private->orb.cpa == &private->sense_ccw ||
+ (void *)(addr_t)private->orb.cpa == cdev->private->iccws)
printk(KERN_WARNING "cio: last channel program (intern):\n");
else
printk(KERN_WARNING "cio: last channel program:\n");
print_hex_dump(KERN_WARNING, "cio: ", DUMP_PREFIX_NONE, 16, 1,
- (void *)(addr_t)sch->orb.cpa, sizeof(struct ccw1), 0);
+ (void *)(addr_t)private->orb.cpa,
+ sizeof(struct ccw1), 0);
printk(KERN_WARNING "cio: ccw device state: %d\n",
cdev->private->state);
printk(KERN_WARNING "cio: store subchannel returned: cc=%d\n", cc);
@@ -1078,7 +1081,7 @@ device_trigger_reprobe(struct subchannel
sch->schib.pmcw.ena = 0;
if ((sch->lpm & (sch->lpm - 1)) != 0)
sch->schib.pmcw.mp = 1;
- sch->schib.pmcw.intparm = (__u32)(unsigned long)sch;
+ sch->schib.pmcw.intparm = (u32)(addr_t)sch;
/* We should also udate ssd info, but this has to wait. */
/* Check if this is another device which appeared on the same sch. */
if (sch->schib.pmcw.dev != cdev->private->dev_id.devno) {
Index: quilt-2.6/drivers/s390/cio/device.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device.h
+++ quilt-2.6/drivers/s390/cio/device.h
@@ -5,6 +5,8 @@
#include <asm/atomic.h>
#include <linux/wait.h>
+#include "io_sch.h"
+
/*
* states of the device statemachine
*/
Index: quilt-2.6/drivers/s390/cio/device_id.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_id.c
+++ quilt-2.6/drivers/s390/cio/device_id.c
@@ -24,6 +24,7 @@
#include "css.h"
#include "device.h"
#include "ioasm.h"
+#include "io_sch.h"
/*
* Input :
@@ -219,11 +220,13 @@ ccw_device_check_sense_id(struct ccw_dev
return -EAGAIN;
}
if (irb->scsw.cc == 3) {
- if ((sch->orb.lpm &
- sch->schib.pmcw.pim & sch->schib.pmcw.pam) != 0)
+ u8 lpm;
+
+ lpm = to_io_private(sch)->orb.lpm;
+ if ((lpm & sch->schib.pmcw.pim & sch->schib.pmcw.pam) != 0)
CIO_MSG_EVENT(2, "SenseID : path %02X for device %04x "
"on subchannel 0.%x.%04x is "
- "'not operational'\n", sch->orb.lpm,
+ "'not operational'\n", lpm,
cdev->private->dev_id.devno,
sch->schid.ssid, sch->schid.sch_no);
return -EACCES;
Index: quilt-2.6/drivers/s390/cio/device_pgid.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_pgid.c
+++ quilt-2.6/drivers/s390/cio/device_pgid.c
@@ -22,6 +22,7 @@
#include "css.h"
#include "device.h"
#include "ioasm.h"
+#include "io_sch.h"
/*
* Helper function called from interrupt context to decide whether an
@@ -155,10 +156,13 @@ __ccw_device_check_sense_pgid(struct ccw
return -EAGAIN;
}
if (irb->scsw.cc == 3) {
+ u8 lpm;
+
+ lpm = to_io_private(sch)->orb.lpm;
CIO_MSG_EVENT(2, "SNID - Device %04x on Subchannel 0.%x.%04x,"
" lpm %02X, became 'not operational'\n",
cdev->private->dev_id.devno, sch->schid.ssid,
- sch->schid.sch_no, sch->orb.lpm);
+ sch->schid.sch_no, lpm);
return -EACCES;
}
i = 8 - ffs(cdev->private->imask);
Index: quilt-2.6/drivers/s390/cio/device_status.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_status.c
+++ quilt-2.6/drivers/s390/cio/device_status.c
@@ -20,6 +20,7 @@
#include "css.h"
#include "device.h"
#include "ioasm.h"
+#include "io_sch.h"
/*
* Check for any kind of channel or interface control check but don't
@@ -310,6 +311,7 @@ int
ccw_device_do_sense(struct ccw_device *cdev, struct irb *irb)
{
struct subchannel *sch;
+ struct ccw1 *sense_ccw;
sch = to_subchannel(cdev->dev.parent);
@@ -326,15 +328,16 @@ ccw_device_do_sense(struct ccw_device *c
/*
* We have ending status but no sense information. Do a basic sense.
*/
- sch->sense_ccw.cmd_code = CCW_CMD_BASIC_SENSE;
- sch->sense_ccw.cda = (__u32) __pa(cdev->private->irb.ecw);
- sch->sense_ccw.count = SENSE_MAX_COUNT;
- sch->sense_ccw.flags = CCW_FLAG_SLI;
+ sense_ccw = &to_io_private(sch)->sense_ccw;
+ sense_ccw->cmd_code = CCW_CMD_BASIC_SENSE;
+ sense_ccw->cda = (__u32) __pa(cdev->private->irb.ecw);
+ sense_ccw->count = SENSE_MAX_COUNT;
+ sense_ccw->flags = CCW_FLAG_SLI;
/* Reset internal retry indication. */
cdev->private->flags.intretry = 0;
- return cio_start (sch, &sch->sense_ccw, 0xff);
+ return cio_start(sch, sense_ccw, 0xff);
}
/*
Index: quilt-2.6/drivers/s390/cio/ioasm.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/ioasm.h
+++ quilt-2.6/drivers/s390/cio/ioasm.h
@@ -109,72 +109,6 @@ static inline int tpi( volatile struct t
return ccode;
}
-static inline int ssch(struct subchannel_id schid,
- volatile struct orb *addr)
-{
- register struct subchannel_id reg1 asm ("1") = schid;
- int ccode;
-
- asm volatile(
- " ssch 0(%2)\n"
- " ipm %0\n"
- " srl %0,28"
- : "=d" (ccode) : "d" (reg1), "a" (addr), "m" (*addr) : "cc");
- return ccode;
-}
-
-static inline int rsch(struct subchannel_id schid)
-{
- register struct subchannel_id reg1 asm ("1") = schid;
- int ccode;
-
- asm volatile(
- " rsch\n"
- " ipm %0\n"
- " srl %0,28"
- : "=d" (ccode) : "d" (reg1) : "cc");
- return ccode;
-}
-
-static inline int csch(struct subchannel_id schid)
-{
- register struct subchannel_id reg1 asm ("1") = schid;
- int ccode;
-
- asm volatile(
- " csch\n"
- " ipm %0\n"
- " srl %0,28"
- : "=d" (ccode) : "d" (reg1) : "cc");
- return ccode;
-}
-
-static inline int hsch(struct subchannel_id schid)
-{
- register struct subchannel_id reg1 asm ("1") = schid;
- int ccode;
-
- asm volatile(
- " hsch\n"
- " ipm %0\n"
- " srl %0,28"
- : "=d" (ccode) : "d" (reg1) : "cc");
- return ccode;
-}
-
-static inline int xsch(struct subchannel_id schid)
-{
- register struct subchannel_id reg1 asm ("1") = schid;
- int ccode;
-
- asm volatile(
- " .insn rre,0xb2760000,%1,0\n"
- " ipm %0\n"
- " srl %0,28"
- : "=d" (ccode) : "d" (reg1) : "cc");
- return ccode;
-}
-
static inline int chsc(void *chsc_area)
{
typedef struct { char _[4096]; } addr_type;
Index: quilt-2.6/drivers/s390/cio/io_sch.h
===================================================================
--- /dev/null
+++ quilt-2.6/drivers/s390/cio/io_sch.h
@@ -0,0 +1,161 @@
+#ifndef S390_IO_SCH_H
+#define S390_IO_SCH_H
+
+#include "schid.h"
+
+/*
+ * operation request block
+ */
+struct orb {
+ u32 intparm; /* interruption parameter */
+ u32 key : 4; /* flags, like key, suspend control, etc. */
+ u32 spnd : 1; /* suspend control */
+ u32 res1 : 1; /* reserved */
+ u32 mod : 1; /* modification control */
+ u32 sync : 1; /* synchronize control */
+ u32 fmt : 1; /* format control */
+ u32 pfch : 1; /* prefetch control */
+ u32 isic : 1; /* initial-status-interruption control */
+ u32 alcc : 1; /* address-limit-checking control */
+ u32 ssic : 1; /* suppress-suspended-interr. control */
+ u32 res2 : 1; /* reserved */
+ u32 c64 : 1; /* IDAW/QDIO 64 bit control */
+ u32 i2k : 1; /* IDAW 2/4kB block size control */
+ u32 lpm : 8; /* logical path mask */
+ u32 ils : 1; /* incorrect length */
+ u32 zero : 6; /* reserved zeros */
+ u32 orbx : 1; /* ORB extension control */
+ u32 cpa; /* channel program address */
+} __attribute__ ((packed, aligned(4)));
+
+struct io_subchannel_private {
+ struct orb orb; /* operation request block */
+ struct ccw1 sense_ccw; /* static ccw for sense command */
+} __attribute__ ((aligned(8)));
+
+#define to_io_private(n) ((struct io_subchannel_private *)n->private)
+
+#define MAX_CIWS 8
+
+/*
+ * sense-id response buffer layout
+ */
+struct senseid {
+ /* common part */
+ u8 reserved; /* always 0x'FF' */
+ u16 cu_type; /* control unit type */
+ u8 cu_model; /* control unit model */
+ u16 dev_type; /* device type */
+ u8 dev_model; /* device model */
+ u8 unused; /* padding byte */
+ /* extended part */
+ struct ciw ciw[MAX_CIWS]; /* variable # of CIWs */
+} __attribute__ ((packed, aligned(4)));
+
+struct ccw_device_private {
+ struct ccw_device *cdev;
+ struct subchannel *sch;
+ int state; /* device state */
+ atomic_t onoff;
+ unsigned long registered;
+ struct ccw_dev_id dev_id; /* device id */
+ struct subchannel_id schid; /* subchannel number */
+ u8 imask; /* lpm mask for SNID/SID/SPGID */
+ int iretry; /* retry counter SNID/SID/SPGID */
+ struct {
+ unsigned int fast:1; /* post with "channel end" */
+ unsigned int repall:1; /* report every interrupt status */
+ unsigned int pgroup:1; /* do path grouping */
+ unsigned int force:1; /* allow forced online */
+ } __attribute__ ((packed)) options;
+ struct {
+ unsigned int pgid_single:1; /* use single path for Set PGID */
+ unsigned int esid:1; /* Ext. SenseID supported by HW */
+ unsigned int dosense:1; /* delayed SENSE required */
+ unsigned int doverify:1; /* delayed path verification */
+ unsigned int donotify:1; /* call notify function */
+ unsigned int recog_done:1; /* dev. recog. complete */
+ unsigned int fake_irb:1; /* deliver faked irb */
+ unsigned int intretry:1; /* retry internal operation */
+ } __attribute__((packed)) flags;
+ unsigned long intparm; /* user interruption parameter */
+ struct qdio_irq *qdio_data;
+ struct irb irb; /* device status */
+ struct senseid senseid; /* SenseID info */
+ struct pgid pgid[8]; /* path group IDs per chpid*/
+ struct ccw1 iccws[2]; /* ccws for SNID/SID/SPGID commands */
+ struct work_struct kick_work;
+ wait_queue_head_t wait_q;
+ struct timer_list timer;
+ void *cmb; /* measurement information */
+ struct list_head cmb_list; /* list of measured devices */
+ u64 cmb_start_time; /* clock value of cmb reset */
+ void *cmb_wait; /* deferred cmb enable/disable */
+};
+
+static inline int ssch(struct subchannel_id schid, volatile struct orb *addr)
+{
+ register struct subchannel_id reg1 asm("1") = schid;
+ int ccode;
+
+ asm volatile(
+ " ssch 0(%2)\n"
+ " ipm %0\n"
+ " srl %0,28"
+ : "=d" (ccode) : "d" (reg1), "a" (addr), "m" (*addr) : "cc");
+ return ccode;
+}
+
+static inline int rsch(struct subchannel_id schid)
+{
+ register struct subchannel_id reg1 asm("1") = schid;
+ int ccode;
+
+ asm volatile(
+ " rsch\n"
+ " ipm %0\n"
+ " srl %0,28"
+ : "=d" (ccode) : "d" (reg1) : "cc");
+ return ccode;
+}
+
+static inline int csch(struct subchannel_id schid)
+{
+ register struct subchannel_id reg1 asm("1") = schid;
+ int ccode;
+
+ asm volatile(
+ " csch\n"
+ " ipm %0\n"
+ " srl %0,28"
+ : "=d" (ccode) : "d" (reg1) : "cc");
+ return ccode;
+}
+
+static inline int hsch(struct subchannel_id schid)
+{
+ register struct subchannel_id reg1 asm("1") = schid;
+ int ccode;
+
+ asm volatile(
+ " hsch\n"
+ " ipm %0\n"
+ " srl %0,28"
+ : "=d" (ccode) : "d" (reg1) : "cc");
+ return ccode;
+}
+
+static inline int xsch(struct subchannel_id schid)
+{
+ register struct subchannel_id reg1 asm("1") = schid;
+ int ccode;
+
+ asm volatile(
+ " .insn rre,0xb2760000,%1,0\n"
+ " ipm %0\n"
+ " srl %0,28"
+ : "=d" (ccode) : "d" (reg1) : "cc");
+ return ccode;
+}
+
+#endif
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 09/47] cio: Extend adapter interrupt interface.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (7 preceding siblings ...)
2007-12-20 15:19 ` [patch 08/47] cio: Introduce subchannel->private Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 10/47] cio: I/O subchannel specific fields Martin Schwidefsky
` (37 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390
Cc: Ursula Braun, Peter Oberparleiter, Cornelia Huck,
Martin Schwidefsky
[-- Attachment #1: 108-cio-airq.diff --]
[-- Type: text/plain, Size: 12146 bytes --]
From: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
From: Cornelia Huck <cornelia.huck@de.ibm.com>
Change the adapter interrupt interface in order to allow multiple
adapter interrupt handlers to be registered. Indicators are now
allocated by cio instead of the device driver.
The qdio parts have been
Acked-by: Ursula Braun <ubraun@linux.vnet.ibm.com>
Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
Documentation/DocBook/s390-drivers.tmpl | 1
drivers/s390/cio/airq.c | 177 +++++++++++++++++++++-----------
drivers/s390/cio/airq.h | 10 -
drivers/s390/cio/cio.c | 2
drivers/s390/cio/cio.h | 1
drivers/s390/cio/qdio.c | 35 +++---
include/asm-s390/airq.h | 19 +++
7 files changed, 157 insertions(+), 88 deletions(-)
Index: quilt-2.6/Documentation/DocBook/s390-drivers.tmpl
===================================================================
--- quilt-2.6.orig/Documentation/DocBook/s390-drivers.tmpl
+++ quilt-2.6/Documentation/DocBook/s390-drivers.tmpl
@@ -116,6 +116,7 @@
!Iinclude/asm-s390/ccwdev.h
!Edrivers/s390/cio/device.c
!Edrivers/s390/cio/device_ops.c
+!Edrivers/s390/cio/airq.c
</sect1>
<sect1 id="cmf">
<title>The channel-measurement facility</title>
Index: quilt-2.6/drivers/s390/cio/airq.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/airq.c
+++ quilt-2.6/drivers/s390/cio/airq.c
@@ -1,12 +1,12 @@
/*
* drivers/s390/cio/airq.c
- * S/390 common I/O routines -- support for adapter interruptions
+ * Support for adapter interruptions
*
- * Copyright (C) 1999-2002 IBM Deutschland Entwicklung GmbH,
- * IBM Corporation
- * Author(s): Ingo Adlung (adlung@de.ibm.com)
- * Cornelia Huck (cornelia.huck@de.ibm.com)
- * Arnd Bergmann (arndb@de.ibm.com)
+ * Copyright IBM Corp. 1999,2007
+ * Author(s): Ingo Adlung <adlung@de.ibm.com>
+ * Cornelia Huck <cornelia.huck@de.ibm.com>
+ * Arnd Bergmann <arndb@de.ibm.com>
+ * Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
*/
#include <linux/init.h>
@@ -14,72 +14,131 @@
#include <linux/slab.h>
#include <linux/rcupdate.h>
-#include "cio_debug.h"
-#include "airq.h"
-
-static adapter_int_handler_t adapter_handler;
+#include <asm/airq.h>
-/*
- * register for adapter interrupts
- *
- * With HiperSockets the zSeries architecture provides for
- * means of adapter interrups, pseudo I/O interrupts that are
- * not tied to an I/O subchannel, but to an adapter. However,
- * it doesn't disclose the info how to enable/disable them, but
- * to recognize them only. Perhaps we should consider them
- * being shared interrupts, and thus build a linked list
- * of adapter handlers ... to be evaluated ...
- */
-int
-s390_register_adapter_interrupt (adapter_int_handler_t handler)
-{
- int ret;
- char dbf_txt[15];
+#include "cio.h"
+#include "cio_debug.h"
- CIO_TRACE_EVENT (4, "rgaint");
+#define NR_AIRQS 32
+#define NR_AIRQS_PER_WORD sizeof(unsigned long)
+#define NR_AIRQ_WORDS (NR_AIRQS / NR_AIRQS_PER_WORD)
+
+union indicator_t {
+ unsigned long word[NR_AIRQ_WORDS];
+ unsigned char byte[NR_AIRQS];
+} __attribute__((packed));
+
+struct airq_t {
+ adapter_int_handler_t handler;
+ void *drv_data;
+};
- if (handler == NULL)
- ret = -EINVAL;
- else
- ret = (cmpxchg(&adapter_handler, NULL, handler) ? -EBUSY : 0);
- if (!ret)
- synchronize_sched(); /* Allow interrupts to complete. */
+static union indicator_t indicators;
+static struct airq_t *airqs[NR_AIRQS];
- sprintf (dbf_txt, "ret:%d", ret);
- CIO_TRACE_EVENT (4, dbf_txt);
+static int register_airq(struct airq_t *airq)
+{
+ int i;
- return ret;
+ for (i = 0; i < NR_AIRQS; i++)
+ if (!cmpxchg(&airqs[i], NULL, airq))
+ return i;
+ return -ENOMEM;
}
-int
-s390_unregister_adapter_interrupt (adapter_int_handler_t handler)
+/**
+ * s390_register_adapter_interrupt() - register adapter interrupt handler
+ * @handler: adapter handler to be registered
+ * @drv_data: driver data passed with each call to the handler
+ *
+ * Returns:
+ * Pointer to the indicator to be used on success
+ * ERR_PTR() if registration failed
+ */
+void *s390_register_adapter_interrupt(adapter_int_handler_t handler,
+ void *drv_data)
{
+ struct airq_t *airq;
+ char dbf_txt[16];
int ret;
- char dbf_txt[15];
-
- CIO_TRACE_EVENT (4, "urgaint");
- if (handler == NULL)
- ret = -EINVAL;
- else {
- adapter_handler = NULL;
- synchronize_sched(); /* Allow interrupts to complete. */
- ret = 0;
+ airq = kmalloc(sizeof(struct airq_t), GFP_KERNEL);
+ if (!airq) {
+ ret = -ENOMEM;
+ goto out;
}
- sprintf (dbf_txt, "ret:%d", ret);
- CIO_TRACE_EVENT (4, dbf_txt);
-
- return ret;
+ airq->handler = handler;
+ airq->drv_data = drv_data;
+ ret = register_airq(airq);
+ if (ret < 0)
+ kfree(airq);
+out:
+ snprintf(dbf_txt, sizeof(dbf_txt), "rairq:%d", ret);
+ CIO_TRACE_EVENT(4, dbf_txt);
+ if (ret < 0)
+ return ERR_PTR(ret);
+ else
+ return &indicators.byte[ret];
}
+EXPORT_SYMBOL(s390_register_adapter_interrupt);
-void
-do_adapter_IO (void)
+/**
+ * s390_unregister_adapter_interrupt - unregister adapter interrupt handler
+ * @ind: indicator for which the handler is to be unregistered
+ */
+void s390_unregister_adapter_interrupt(void *ind)
{
- CIO_TRACE_EVENT (6, "doaio");
-
- if (adapter_handler)
- (*adapter_handler) ();
+ struct airq_t *airq;
+ char dbf_txt[16];
+ int i;
+
+ i = (int) ((addr_t) ind) - ((addr_t) &indicators.byte[0]);
+ snprintf(dbf_txt, sizeof(dbf_txt), "urairq:%d", i);
+ CIO_TRACE_EVENT(4, dbf_txt);
+ indicators.byte[i] = 0;
+ airq = xchg(&airqs[i], NULL);
+ /*
+ * Allow interrupts to complete. This will ensure that the airq handle
+ * is no longer referenced by any interrupt handler.
+ */
+ synchronize_sched();
+ kfree(airq);
}
+EXPORT_SYMBOL(s390_unregister_adapter_interrupt);
+
+#define INDICATOR_MASK (0xffUL << ((NR_AIRQS_PER_WORD - 1) * 8))
-EXPORT_SYMBOL (s390_register_adapter_interrupt);
-EXPORT_SYMBOL (s390_unregister_adapter_interrupt);
+void do_adapter_IO(void)
+{
+ int w;
+ int i;
+ unsigned long word;
+ struct airq_t *airq;
+
+ /*
+ * Access indicator array in word-sized chunks to minimize storage
+ * fetch operations.
+ */
+ for (w = 0; w < NR_AIRQ_WORDS; w++) {
+ word = indicators.word[w];
+ i = w * NR_AIRQS_PER_WORD;
+ /*
+ * Check bytes within word for active indicators.
+ */
+ while (word) {
+ if (word & INDICATOR_MASK) {
+ airq = airqs[i];
+ if (likely(airq))
+ airq->handler(&indicators.byte[i],
+ airq->drv_data);
+ else
+ /*
+ * Reset ill-behaved indicator.
+ */
+ indicators.byte[i] = 0;
+ }
+ word <<= 8;
+ i++;
+ }
+ }
+}
Index: quilt-2.6/drivers/s390/cio/airq.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/airq.h
+++ /dev/null
@@ -1,10 +0,0 @@
-#ifndef S390_AINTERRUPT_H
-#define S390_AINTERRUPT_H
-
-typedef int (*adapter_int_handler_t)(void);
-
-extern int s390_register_adapter_interrupt(adapter_int_handler_t handler);
-extern int s390_unregister_adapter_interrupt(adapter_int_handler_t handler);
-extern void do_adapter_IO (void);
-
-#endif
Index: quilt-2.6/drivers/s390/cio/cio.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/cio.c
+++ quilt-2.6/drivers/s390/cio/cio.c
@@ -23,7 +23,7 @@
#include <asm/reset.h>
#include <asm/ipl.h>
#include <asm/chpid.h>
-#include "airq.h"
+#include <asm/airq.h>
#include "cio.h"
#include "css.h"
#include "chsc.h"
Index: quilt-2.6/drivers/s390/cio/cio.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/cio.h
+++ quilt-2.6/drivers/s390/cio/cio.h
@@ -99,6 +99,7 @@ extern int cio_get_options (struct subch
extern int cio_modify (struct subchannel *);
int cio_create_sch_lock(struct subchannel *);
+void do_adapter_IO(void);
/* Use with care. */
#ifdef CONFIG_CCW_CONSOLE
Index: quilt-2.6/drivers/s390/cio/qdio.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/qdio.c
+++ quilt-2.6/drivers/s390/cio/qdio.c
@@ -48,11 +48,11 @@
#include <asm/debug.h>
#include <asm/s390_rdev.h>
#include <asm/qdio.h>
+#include <asm/airq.h>
#include "cio.h"
#include "css.h"
#include "device.h"
-#include "airq.h"
#include "qdio.h"
#include "ioasm.h"
#include "chsc.h"
@@ -96,7 +96,7 @@ static debug_info_t *qdio_dbf_slsb_in;
static volatile struct qdio_q *tiq_list=NULL; /* volatile as it could change
during a while loop */
static DEFINE_SPINLOCK(ttiq_list_lock);
-static int register_thinint_result;
+static void *tiqdio_ind;
static void tiqdio_tl(unsigned long);
static DECLARE_TASKLET(tiqdio_tasklet,tiqdio_tl,0);
@@ -399,7 +399,7 @@ qdio_get_indicator(void)
{
int i;
- for (i=1;i<INDICATORS_PER_CACHELINE;i++)
+ for (i = 0; i < INDICATORS_PER_CACHELINE; i++)
if (!indicator_used[i]) {
indicator_used[i]=1;
return indicators+i;
@@ -1911,8 +1911,7 @@ qdio_fill_thresholds(struct qdio_irq *ir
}
}
-static int
-tiqdio_thinint_handler(void)
+static void tiqdio_thinint_handler(void *ind, void *drv_data)
{
QDIO_DBF_TEXT4(0,trace,"thin_int");
@@ -1925,7 +1924,6 @@ tiqdio_thinint_handler(void)
tiqdio_clear_global_summary();
tiqdio_inbound_checks();
- return 0;
}
static void
@@ -2445,7 +2443,7 @@ tiqdio_set_subchannel_ind(struct qdio_ir
real_addr_dev_st_chg_ind=0;
} else {
real_addr_local_summary_bit=
- virt_to_phys((volatile void *)indicators);
+ virt_to_phys((volatile void *)tiqdio_ind);
real_addr_dev_st_chg_ind=
virt_to_phys((volatile void *)irq_ptr->dev_st_chg_ind);
}
@@ -3740,23 +3738,25 @@ static void
tiqdio_register_thinints(void)
{
char dbf_text[20];
- register_thinint_result=
- s390_register_adapter_interrupt(&tiqdio_thinint_handler);
- if (register_thinint_result) {
- sprintf(dbf_text,"regthn%x",(register_thinint_result&0xff));
+
+ tiqdio_ind =
+ s390_register_adapter_interrupt(&tiqdio_thinint_handler, NULL);
+ if (IS_ERR(tiqdio_ind)) {
+ sprintf(dbf_text, "regthn%lx", PTR_ERR(tiqdio_ind));
QDIO_DBF_TEXT0(0,setup,dbf_text);
QDIO_PRINT_ERR("failed to register adapter handler " \
- "(rc=%i).\nAdapter interrupts might " \
+ "(rc=%li).\nAdapter interrupts might " \
"not work. Continuing.\n",
- register_thinint_result);
+ PTR_ERR(tiqdio_ind));
+ tiqdio_ind = NULL;
}
}
static void
tiqdio_unregister_thinints(void)
{
- if (!register_thinint_result)
- s390_unregister_adapter_interrupt(&tiqdio_thinint_handler);
+ if (tiqdio_ind)
+ s390_unregister_adapter_interrupt(tiqdio_ind);
}
static int
@@ -3768,8 +3768,8 @@ qdio_get_qdio_memory(void)
for (i=1;i<INDICATORS_PER_CACHELINE;i++)
indicator_used[i]=0;
indicators = kzalloc(sizeof(__u32)*(INDICATORS_PER_CACHELINE),
- GFP_KERNEL);
- if (!indicators)
+ GFP_KERNEL);
+ if (!indicators)
return -ENOMEM;
return 0;
}
@@ -3780,7 +3780,6 @@ qdio_release_qdio_memory(void)
kfree(indicators);
}
-
static void
qdio_unregister_dbf_views(void)
{
Index: quilt-2.6/include/asm-s390/airq.h
===================================================================
--- /dev/null
+++ quilt-2.6/include/asm-s390/airq.h
@@ -0,0 +1,19 @@
+/*
+ * include/asm-s390/airq.h
+ *
+ * Copyright IBM Corp. 2002,2007
+ * Author(s): Ingo Adlung <adlung@de.ibm.com>
+ * Cornelia Huck <cornelia.huck@de.ibm.com>
+ * Arnd Bergmann <arndb@de.ibm.com>
+ * Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
+ */
+
+#ifndef _ASM_S390_AIRQ_H
+#define _ASM_S390_AIRQ_H
+
+typedef void (*adapter_int_handler_t)(void *, void *);
+
+void *s390_register_adapter_interrupt(adapter_int_handler_t, void *);
+void s390_unregister_adapter_interrupt(void *);
+
+#endif /* _ASM_S390_AIRQ_H */
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 10/47] cio: I/O subchannel specific fields.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (8 preceding siblings ...)
2007-12-20 15:19 ` [patch 09/47] cio: Extend adapter interrupt interface Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 11/47] cio: Use dev_{g,s}et_drvdata() Martin Schwidefsky
` (36 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Cornelia Huck, Martin Schwidefsky
[-- Attachment #1: 109-cio-fields.diff --]
[-- Type: text/plain, Size: 8694 bytes --]
From: Cornelia Huck <cornelia.huck@de.ibm.com>
Some fields may be !0 only for I/O subchannels. Add some checks
where required. Also adapt cio_enable_subchannel() to make the
caller specify the intparm, which makes it more generic.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/chsc.c | 5 +++--
drivers/s390/cio/cio.c | 24 ++++++++++++------------
drivers/s390/cio/cio.h | 4 ++--
drivers/s390/cio/css.c | 18 ++++++++++++++++--
drivers/s390/cio/css.h | 2 ++
drivers/s390/cio/device_fsm.c | 9 ++++++---
drivers/s390/cio/device_ops.c | 2 +-
7 files changed, 42 insertions(+), 22 deletions(-)
Index: quilt-2.6/drivers/s390/cio/chsc.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/chsc.c
+++ quilt-2.6/drivers/s390/cio/chsc.c
@@ -89,7 +89,8 @@ int chsc_get_ssd_info(struct subchannel_
/* Copy data */
ret = 0;
memset(ssd, 0, sizeof(struct chsc_ssd_info));
- if ((ssd_area->st != 0) && (ssd_area->st != 2))
+ if ((ssd_area->st != SUBCHANNEL_TYPE_IO) &&
+ (ssd_area->st != SUBCHANNEL_TYPE_MSG))
goto out_free;
ssd->path_mask = ssd_area->path_mask;
ssd->fla_valid_mask = ssd_area->fla_valid_mask;
@@ -158,7 +159,7 @@ s390_subchannel_remove_chpid(struct devi
spin_lock_irq(sch->lock);
stsch(sch->schid, &schib);
- if (!schib.pmcw.dnv)
+ if (!css_sch_is_valid(&schib))
goto out_unreg;
memcpy(&sch->schib, &schib, sizeof(struct schib));
/* Check for single path devices. */
Index: quilt-2.6/drivers/s390/cio/cio.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/cio.c
+++ quilt-2.6/drivers/s390/cio/cio.c
@@ -406,8 +406,8 @@ cio_modify (struct subchannel *sch)
/*
* Enable subchannel.
*/
-int
-cio_enable_subchannel (struct subchannel *sch, unsigned int isc)
+int cio_enable_subchannel(struct subchannel *sch, unsigned int isc,
+ u32 intparm)
{
char dbf_txt[15];
int ccode;
@@ -426,7 +426,7 @@ cio_enable_subchannel (struct subchannel
for (retry = 5, ret = 0; retry > 0; retry--) {
sch->schib.pmcw.ena = 1;
sch->schib.pmcw.isc = isc;
- sch->schib.pmcw.intparm = (u32)(addr_t)sch;
+ sch->schib.pmcw.intparm = intparm;
ret = cio_modify(sch);
if (ret == -ENODEV)
break;
@@ -577,11 +577,8 @@ cio_validate_subchannel (struct subchann
}
/* Initialization for io subchannels. */
- if (!sch->schib.pmcw.dnv) {
- /* io subchannel but device number is invalid. */
- err = -ENODEV;
- goto out;
- }
+ if (!css_sch_is_valid(&sch->schib))
+ return -ENODEV;
/* Devno is valid. */
if (is_blacklisted (sch->schid.ssid, sch->schib.pmcw.dev)) {
/*
@@ -745,9 +742,9 @@ cio_test_for_console(struct subchannel_i
{
if (stsch_err(schid, &console_subchannel.schib) != 0)
return -ENXIO;
- if (console_subchannel.schib.pmcw.dnv &&
- console_subchannel.schib.pmcw.dev ==
- console_devno) {
+ if ((console_subchannel.schib.pmcw.st == SUBCHANNEL_TYPE_IO) &&
+ console_subchannel.schib.pmcw.dnv &&
+ (console_subchannel.schib.pmcw.dev == console_devno)) {
console_irq = schid.sch_no;
return 1; /* found */
}
@@ -765,6 +762,7 @@ cio_get_console_sch_no(void)
/* VM provided us with the irq number of the console. */
schid.sch_no = console_irq;
if (stsch(schid, &console_subchannel.schib) != 0 ||
+ (console_subchannel.schib.pmcw.st != SUBCHANNEL_TYPE_IO) ||
!console_subchannel.schib.pmcw.dnv)
return -1;
console_devno = console_subchannel.schib.pmcw.dev;
@@ -1029,7 +1027,7 @@ static int __reipl_subchannel_match(stru
if (stsch_reset(schid, &schib))
return -ENXIO;
- if (schib.pmcw.dnv &&
+ if ((schib.pmcw.st == SUBCHANNEL_TYPE_IO) && schib.pmcw.dnv &&
(schib.pmcw.dev == match_id->devid.devno) &&
(schid.ssid == match_id->devid.ssid)) {
match_id->schid = schid;
@@ -1075,6 +1073,8 @@ int __init cio_get_iplinfo(struct cio_ip
return -ENODEV;
if (stsch(schid, &schib))
return -ENODEV;
+ if (schib.pmcw.st != SUBCHANNEL_TYPE_IO)
+ return -ENODEV;
if (!schib.pmcw.dnv)
return -ENODEV;
iplinfo->devno = schib.pmcw.dev;
Index: quilt-2.6/drivers/s390/cio/cio.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/cio.h
+++ quilt-2.6/drivers/s390/cio/cio.h
@@ -60,7 +60,7 @@ struct subchannel {
enum {
SUBCHANNEL_TYPE_IO = 0,
SUBCHANNEL_TYPE_CHSC = 1,
- SUBCHANNEL_TYPE_MESSAGE = 2,
+ SUBCHANNEL_TYPE_MSG = 2,
SUBCHANNEL_TYPE_ADM = 3,
} st; /* subchannel type */
@@ -85,7 +85,7 @@ struct subchannel {
#define to_subchannel(n) container_of(n, struct subchannel, dev)
extern int cio_validate_subchannel (struct subchannel *, struct subchannel_id);
-extern int cio_enable_subchannel (struct subchannel *, unsigned int);
+extern int cio_enable_subchannel(struct subchannel *, unsigned int, u32);
extern int cio_disable_subchannel (struct subchannel *);
extern int cio_cancel (struct subchannel *);
extern int cio_clear (struct subchannel *);
Index: quilt-2.6/drivers/s390/cio/css.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.c
+++ quilt-2.6/drivers/s390/cio/css.c
@@ -237,11 +237,25 @@ get_subchannel_by_schid(struct subchanne
return dev ? to_subchannel(dev) : NULL;
}
+/**
+ * css_sch_is_valid() - check if a subchannel is valid
+ * @schib: subchannel information block for the subchannel
+ */
+int css_sch_is_valid(struct schib *schib)
+{
+ if ((schib->pmcw.st == SUBCHANNEL_TYPE_IO) && !schib->pmcw.dnv)
+ return 0;
+ return 1;
+}
+EXPORT_SYMBOL_GPL(css_sch_is_valid);
+
static int css_get_subchannel_status(struct subchannel *sch)
{
struct schib schib;
- if (stsch(sch->schid, &schib) || !schib.pmcw.dnv)
+ if (stsch(sch->schid, &schib))
+ return CIO_GONE;
+ if (!css_sch_is_valid(&schib))
return CIO_GONE;
if (sch->schib.pmcw.dnv && (schib.pmcw.dev != sch->schib.pmcw.dev))
return CIO_REVALIDATE;
@@ -349,7 +363,7 @@ static int css_evaluate_new_subchannel(s
/* Will be done on the slow path. */
return -EAGAIN;
}
- if (stsch_err(schid, &schib) || !schib.pmcw.dnv) {
+ if (stsch_err(schid, &schib) || !css_sch_is_valid(&schib)) {
/* Unusable - ignore. */
return 0;
}
Index: quilt-2.6/drivers/s390/cio/css.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.h
+++ quilt-2.6/drivers/s390/cio/css.h
@@ -136,6 +136,8 @@ void css_schedule_eval(struct subchannel
void css_schedule_eval_all(void);
int sch_is_pseudo_sch(struct subchannel *);
+struct schib;
+int css_sch_is_valid(struct schib *);
extern struct workqueue_struct *slow_path_wq;
Index: quilt-2.6/drivers/s390/cio/device_fsm.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_fsm.c
+++ quilt-2.6/drivers/s390/cio/device_fsm.c
@@ -553,7 +553,8 @@ ccw_device_recognition(struct ccw_device
(cdev->private->state != DEV_STATE_BOXED))
return -EINVAL;
sch = to_subchannel(cdev->dev.parent);
- ret = cio_enable_subchannel(sch, sch->schib.pmcw.isc);
+ ret = cio_enable_subchannel(sch, sch->schib.pmcw.isc,
+ (u32)(addr_t)sch);
if (ret != 0)
/* Couldn't enable the subchannel for i/o. Sick device. */
return ret;
@@ -663,7 +664,8 @@ ccw_device_online(struct ccw_device *cde
sch = to_subchannel(cdev->dev.parent);
if (css_init_done && !get_device(&cdev->dev))
return -ENODEV;
- ret = cio_enable_subchannel(sch, sch->schib.pmcw.isc);
+ ret = cio_enable_subchannel(sch, sch->schib.pmcw.isc,
+ (u32)(addr_t)sch);
if (ret != 0) {
/* Couldn't enable the subchannel for i/o. Sick device. */
if (ret == -ENODEV)
@@ -1043,7 +1045,8 @@ ccw_device_start_id(struct ccw_device *c
struct subchannel *sch;
sch = to_subchannel(cdev->dev.parent);
- if (cio_enable_subchannel(sch, sch->schib.pmcw.isc) != 0)
+ if (cio_enable_subchannel(sch, sch->schib.pmcw.isc,
+ (u32)(addr_t)sch) != 0)
/* Couldn't enable the subchannel for i/o. Sick device. */
return;
Index: quilt-2.6/drivers/s390/cio/device_ops.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_ops.c
+++ quilt-2.6/drivers/s390/cio/device_ops.c
@@ -501,7 +501,7 @@ ccw_device_stlck(struct ccw_device *cdev
return -ENOMEM;
}
spin_lock_irqsave(sch->lock, flags);
- ret = cio_enable_subchannel(sch, 3);
+ ret = cio_enable_subchannel(sch, 3, (u32)(addr_t)sch);
if (ret)
goto out_unlock;
/*
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 11/47] cio: Use dev_{g,s}et_drvdata().
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (9 preceding siblings ...)
2007-12-20 15:19 ` [patch 10/47] cio: I/O subchannel specific fields Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 12/47] cio: Set driver->owner on css, ccw and ccwgroup busses Martin Schwidefsky
` (35 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Cornelia Huck, Martin Schwidefsky
[-- Attachment #1: 110-cio-drvdata.diff --]
[-- Type: text/plain, Size: 9615 bytes --]
From: Cornelia Huck <cornelia.huck@de.ibm.com>
Also define helpers sch_{g,s}et_cdev() to make the intention more
clear.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/ccwgroup.c | 14 +++++++-------
drivers/s390/cio/device.c | 36 ++++++++++++++++++------------------
drivers/s390/cio/device_fsm.c | 26 +++++++++++++-------------
drivers/s390/cio/io_sch.h | 2 ++
4 files changed, 40 insertions(+), 38 deletions(-)
Index: quilt-2.6/drivers/s390/cio/ccwgroup.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/ccwgroup.c
+++ quilt-2.6/drivers/s390/cio/ccwgroup.c
@@ -111,7 +111,7 @@ ccwgroup_release (struct device *dev)
gdev = to_ccwgroupdev(dev);
for (i = 0; i < gdev->count; i++) {
- gdev->cdev[i]->dev.driver_data = NULL;
+ dev_set_drvdata(&gdev->cdev[i]->dev, NULL);
put_device(&gdev->cdev[i]->dev);
}
kfree(gdev);
@@ -196,11 +196,11 @@ int ccwgroup_create(struct device *root,
goto error;
}
/* Don't allow a device to belong to more than one group. */
- if (gdev->cdev[i]->dev.driver_data) {
+ if (dev_get_drvdata(&gdev->cdev[i]->dev)) {
rc = -EINVAL;
goto error;
}
- gdev->cdev[i]->dev.driver_data = gdev;
+ dev_set_drvdata(&gdev->cdev[i]->dev, gdev);
}
gdev->creator_id = creator_id;
@@ -234,8 +234,8 @@ int ccwgroup_create(struct device *root,
error:
for (i = 0; i < argc; i++)
if (gdev->cdev[i]) {
- if (gdev->cdev[i]->dev.driver_data == gdev)
- gdev->cdev[i]->dev.driver_data = NULL;
+ if (dev_get_drvdata(&gdev->cdev[i]->dev) == gdev)
+ dev_set_drvdata(&gdev->cdev[i]->dev, NULL);
put_device(&gdev->cdev[i]->dev);
}
mutex_unlock(&gdev->reg_mutex);
@@ -463,8 +463,8 @@ __ccwgroup_get_gdev_by_cdev(struct ccw_d
{
struct ccwgroup_device *gdev;
- if (cdev->dev.driver_data) {
- gdev = (struct ccwgroup_device *)cdev->dev.driver_data;
+ gdev = dev_get_drvdata(&cdev->dev);
+ if (gdev) {
if (get_device(&gdev->dev)) {
mutex_lock(&gdev->reg_mutex);
if (device_is_registered(&gdev->dev))
Index: quilt-2.6/drivers/s390/cio/device.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device.c
+++ quilt-2.6/drivers/s390/cio/device.c
@@ -773,7 +773,7 @@ static void sch_attach_device(struct sub
{
css_update_ssd_info(sch);
spin_lock_irq(sch->lock);
- sch->dev.driver_data = cdev;
+ sch_set_cdev(sch, cdev);
cdev->private->schid = sch->schid;
cdev->ccwlock = sch->lock;
device_trigger_reprobe(sch);
@@ -795,7 +795,7 @@ static void sch_attach_disconnected_devi
put_device(&other_sch->dev);
return;
}
- other_sch->dev.driver_data = NULL;
+ sch_set_cdev(other_sch, NULL);
/* No need to keep a subchannel without ccw device around. */
css_sch_device_unregister(other_sch);
put_device(&other_sch->dev);
@@ -831,12 +831,12 @@ static void sch_create_and_recog_new_dev
return;
}
spin_lock_irq(sch->lock);
- sch->dev.driver_data = cdev;
+ sch_set_cdev(sch, cdev);
spin_unlock_irq(sch->lock);
/* Start recognition for the new ccw device. */
if (io_subchannel_recog(cdev, sch)) {
spin_lock_irq(sch->lock);
- sch->dev.driver_data = NULL;
+ sch_set_cdev(sch, NULL);
spin_unlock_irq(sch->lock);
if (cdev->dev.release)
cdev->dev.release(&cdev->dev);
@@ -940,7 +940,7 @@ io_subchannel_register(struct work_struc
cdev->private->dev_id.devno, ret);
put_device(&cdev->dev);
spin_lock_irqsave(sch->lock, flags);
- sch->dev.driver_data = NULL;
+ sch_set_cdev(sch, NULL);
spin_unlock_irqrestore(sch->lock, flags);
kfree (cdev->private);
kfree (cdev);
@@ -1022,7 +1022,7 @@ io_subchannel_recog(struct ccw_device *c
int rc;
struct ccw_device_private *priv;
- sch->dev.driver_data = cdev;
+ sch_set_cdev(sch, cdev);
sch->driver = &io_subchannel_driver;
cdev->ccwlock = sch->lock;
@@ -1082,7 +1082,7 @@ static void ccw_device_move_to_sch(struc
}
if (former_parent) {
spin_lock_irq(former_parent->lock);
- former_parent->dev.driver_data = NULL;
+ sch_set_cdev(former_parent, NULL);
spin_unlock_irq(former_parent->lock);
css_sch_device_unregister(former_parent);
/* Reset intparm to zeroes. */
@@ -1100,7 +1100,7 @@ static void io_subchannel_irq(struct sub
{
struct ccw_device *cdev;
- cdev = sch->dev.driver_data;
+ cdev = sch_get_cdev(sch);
CIO_TRACE_EVENT(3, "IRQ");
CIO_TRACE_EVENT(3, sch->dev.bus_id);
@@ -1116,13 +1116,13 @@ io_subchannel_probe (struct subchannel *
unsigned long flags;
struct ccw_dev_id dev_id;
- if (sch->dev.driver_data) {
+ cdev = sch_get_cdev(sch);
+ if (cdev) {
/*
* This subchannel already has an associated ccw_device.
* Register it and exit. This happens for all early
* device, e.g. the console.
*/
- cdev = sch->dev.driver_data;
cdev->dev.groups = ccwdev_attr_groups;
device_initialize(&cdev->dev);
ccw_device_register(cdev);
@@ -1173,7 +1173,7 @@ io_subchannel_probe (struct subchannel *
rc = io_subchannel_recog(cdev, sch);
if (rc) {
spin_lock_irqsave(sch->lock, flags);
- sch->dev.driver_data = NULL;
+ sch_set_cdev(sch, NULL);
spin_unlock_irqrestore(sch->lock, flags);
if (cdev->dev.release)
cdev->dev.release(&cdev->dev);
@@ -1189,12 +1189,12 @@ io_subchannel_remove (struct subchannel
struct ccw_device *cdev;
unsigned long flags;
- if (!sch->dev.driver_data)
+ cdev = sch_get_cdev(sch);
+ if (!cdev)
return 0;
- cdev = sch->dev.driver_data;
/* Set ccw device to not operational and drop reference. */
spin_lock_irqsave(cdev->ccwlock, flags);
- sch->dev.driver_data = NULL;
+ sch_set_cdev(sch, NULL);
cdev->private->state = DEV_STATE_NOT_OPER;
spin_unlock_irqrestore(cdev->ccwlock, flags);
ccw_device_unregister(cdev);
@@ -1207,7 +1207,7 @@ static int io_subchannel_notify(struct s
{
struct ccw_device *cdev;
- cdev = sch->dev.driver_data;
+ cdev = sch_get_cdev(sch);
if (!cdev)
return 0;
if (!cdev->drv)
@@ -1221,7 +1221,7 @@ static void io_subchannel_verify(struct
{
struct ccw_device *cdev;
- cdev = sch->dev.driver_data;
+ cdev = sch_get_cdev(sch);
if (cdev)
dev_fsm_event(cdev, DEV_EVENT_VERIFY);
}
@@ -1230,7 +1230,7 @@ static void io_subchannel_ioterm(struct
{
struct ccw_device *cdev;
- cdev = sch->dev.driver_data;
+ cdev = sch_get_cdev(sch);
if (!cdev)
return;
/* Internal I/O will be retried by the interrupt handler. */
@@ -1248,7 +1248,7 @@ io_subchannel_shutdown(struct subchannel
struct ccw_device *cdev;
int ret;
- cdev = sch->dev.driver_data;
+ cdev = sch_get_cdev(sch);
if (cio_is_console(sch->schid))
return;
Index: quilt-2.6/drivers/s390/cio/device_fsm.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device_fsm.c
+++ quilt-2.6/drivers/s390/cio/device_fsm.c
@@ -32,9 +32,9 @@ device_is_online(struct subchannel *sch)
{
struct ccw_device *cdev;
- if (!sch->dev.driver_data)
+ cdev = sch_get_cdev(sch);
+ if (!cdev)
return 0;
- cdev = sch->dev.driver_data;
return (cdev->private->state == DEV_STATE_ONLINE);
}
@@ -43,9 +43,9 @@ device_is_disconnected(struct subchannel
{
struct ccw_device *cdev;
- if (!sch->dev.driver_data)
+ cdev = sch_get_cdev(sch);
+ if (!cdev)
return 0;
- cdev = sch->dev.driver_data;
return (cdev->private->state == DEV_STATE_DISCONNECTED ||
cdev->private->state == DEV_STATE_DISCONNECTED_SENSE_ID);
}
@@ -55,9 +55,9 @@ device_set_disconnected(struct subchanne
{
struct ccw_device *cdev;
- if (!sch->dev.driver_data)
+ cdev = sch_get_cdev(sch);
+ if (!cdev)
return;
- cdev = sch->dev.driver_data;
ccw_device_set_timeout(cdev, 0);
cdev->private->flags.fake_irb = 0;
cdev->private->state = DEV_STATE_DISCONNECTED;
@@ -67,7 +67,7 @@ void device_set_intretry(struct subchann
{
struct ccw_device *cdev;
- cdev = sch->dev.driver_data;
+ cdev = sch_get_cdev(sch);
if (!cdev)
return;
cdev->private->flags.intretry = 1;
@@ -77,7 +77,7 @@ int device_trigger_verify(struct subchan
{
struct ccw_device *cdev;
- cdev = sch->dev.driver_data;
+ cdev = sch_get_cdev(sch);
if (!cdev || !cdev->online)
return -EINVAL;
dev_fsm_event(cdev, DEV_EVENT_VERIFY);
@@ -175,9 +175,9 @@ device_kill_pending_timer(struct subchan
{
struct ccw_device *cdev;
- if (!sch->dev.driver_data)
+ cdev = sch_get_cdev(sch);
+ if (!cdev)
return;
- cdev = sch->dev.driver_data;
ccw_device_set_timeout(cdev, 0);
}
@@ -992,7 +992,7 @@ void device_kill_io(struct subchannel *s
int ret;
struct ccw_device *cdev;
- cdev = sch->dev.driver_data;
+ cdev = sch_get_cdev(sch);
ret = ccw_device_cancel_halt_clear(cdev);
if (ret == -EBUSY) {
ccw_device_set_timeout(cdev, 3*HZ);
@@ -1062,9 +1062,9 @@ device_trigger_reprobe(struct subchannel
{
struct ccw_device *cdev;
- if (!sch->dev.driver_data)
+ cdev = sch_get_cdev(sch);
+ if (!cdev)
return;
- cdev = sch->dev.driver_data;
if (cdev->private->state != DEV_STATE_DISCONNECTED)
return;
Index: quilt-2.6/drivers/s390/cio/io_sch.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/io_sch.h
+++ quilt-2.6/drivers/s390/cio/io_sch.h
@@ -34,6 +34,8 @@ struct io_subchannel_private {
} __attribute__ ((aligned(8)));
#define to_io_private(n) ((struct io_subchannel_private *)n->private)
+#define sch_get_cdev(n) (dev_get_drvdata(&n->dev))
+#define sch_set_cdev(n, c) (dev_set_drvdata(&n->dev, c))
#define MAX_CIWS 8
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 12/47] cio: Set driver->owner on css, ccw and ccwgroup busses.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (10 preceding siblings ...)
2007-12-20 15:19 ` [patch 11/47] cio: Use dev_{g,s}et_drvdata() Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 13/47] cio: reduce cpu utilization during device scan Martin Schwidefsky
` (34 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Cornelia Huck, Martin Schwidefsky
[-- Attachment #1: 111-cio-driver-owner.diff --]
[-- Type: text/plain, Size: 2522 bytes --]
From: Cornelia Huck <cornelia.huck@de.ibm.com>
Set the owner field in the embedded struct device_driver to the
value provided in the {css,ccw,ccwgroup}_driver.
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/ccwgroup.c | 1 +
drivers/s390/cio/css.c | 1 +
drivers/s390/cio/css.h | 1 +
drivers/s390/cio/device.c | 2 ++
4 files changed, 5 insertions(+)
Index: quilt-2.6/drivers/s390/cio/ccwgroup.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/ccwgroup.c
+++ quilt-2.6/drivers/s390/cio/ccwgroup.c
@@ -408,6 +408,7 @@ int ccwgroup_driver_register(struct ccwg
/* register our new driver with the core */
cdriver->driver.bus = &ccwgroup_bus_type;
cdriver->driver.name = cdriver->name;
+ cdriver->driver.owner = cdriver->owner;
return driver_register(&cdriver->driver);
}
Index: quilt-2.6/drivers/s390/cio/css.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.c
+++ quilt-2.6/drivers/s390/cio/css.c
@@ -862,6 +862,7 @@ int css_driver_register(struct css_drive
{
cdrv->drv.name = cdrv->name;
cdrv->drv.bus = &css_bus_type;
+ cdrv->drv.owner = cdrv->owner;
return driver_register(&cdrv->drv);
}
EXPORT_SYMBOL_GPL(css_driver_register);
Index: quilt-2.6/drivers/s390/cio/css.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.h
+++ quilt-2.6/drivers/s390/cio/css.h
@@ -65,6 +65,7 @@ struct pgid {
*/
struct subchannel;
struct css_driver {
+ struct module *owner;
unsigned int subchannel_type;
struct device_driver drv;
void (*irq)(struct subchannel *);
Index: quilt-2.6/drivers/s390/cio/device.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/device.c
+++ quilt-2.6/drivers/s390/cio/device.c
@@ -125,6 +125,7 @@ static void io_subchannel_ioterm(struct
static void io_subchannel_shutdown(struct subchannel *);
static struct css_driver io_subchannel_driver = {
+ .owner = THIS_MODULE,
.subchannel_type = SUBCHANNEL_TYPE_IO,
.name = "io_subchannel",
.irq = io_subchannel_irq,
@@ -1476,6 +1477,7 @@ int ccw_driver_register(struct ccw_drive
drv->bus = &ccw_bus_type;
drv->name = cdriver->name;
+ drv->owner = cdriver->owner;
return driver_register(drv);
}
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 13/47] cio: reduce cpu utilization during device scan
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (11 preceding siblings ...)
2007-12-20 15:19 ` [patch 12/47] cio: Set driver->owner on css, ccw and ccwgroup busses Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 14/47] qdio: Remove double checked value Martin Schwidefsky
` (33 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Peter Oberparleiter, Martin Schwidefsky
[-- Attachment #1: 112-cio-scan.diff --]
[-- Type: text/plain, Size: 11588 bytes --]
From: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Minimize calls to cpu intensive function get_subchannel_by_schid()
by introducing function for_each_subchannel_staged() which
temporarily caches the information about registered subchannels
in a bitmap.
Signed-off-by: Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/chsc.c | 96 ++++++++++++--------------------------
drivers/s390/cio/css.c | 120 ++++++++++++++++++++++++++++++++++++++++--------
drivers/s390/cio/css.h | 3 +
3 files changed, 136 insertions(+), 83 deletions(-)
Index: quilt-2.6/drivers/s390/cio/chsc.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/chsc.c
+++ quilt-2.6/drivers/s390/cio/chsc.c
@@ -136,17 +136,13 @@ static void terminate_internal_io(struct
sch->driver->termination(sch);
}
-static int
-s390_subchannel_remove_chpid(struct device *dev, void *data)
+static int s390_subchannel_remove_chpid(struct subchannel *sch, void *data)
{
int j;
int mask;
- struct subchannel *sch;
- struct chp_id *chpid;
+ struct chp_id *chpid = data;
struct schib schib;
- sch = to_subchannel(dev);
- chpid = data;
for (j = 0; j < 8; j++) {
mask = 0x80 >> j;
if ((sch->schib.pmcw.pim & mask) &&
@@ -202,12 +198,10 @@ void chsc_chp_offline(struct chp_id chpi
if (chp_get_status(chpid) <= 0)
return;
- bus_for_each_dev(&css_bus_type, NULL, &chpid,
- s390_subchannel_remove_chpid);
+ for_each_subchannel_staged(s390_subchannel_remove_chpid, NULL, &chpid);
}
-static int
-s390_process_res_acc_new_sch(struct subchannel_id schid)
+static int s390_process_res_acc_new_sch(struct subchannel_id schid, void *data)
{
struct schib schib;
/*
@@ -253,18 +247,10 @@ static int get_res_chpid_mask(struct chs
return 0;
}
-static int
-__s390_process_res_acc(struct subchannel_id schid, void *data)
+static int __s390_process_res_acc(struct subchannel *sch, void *data)
{
int chp_mask, old_lpm;
- struct res_acc_data *res_data;
- struct subchannel *sch;
-
- res_data = data;
- sch = get_subchannel_by_schid(schid);
- if (!sch)
- /* Check if a subchannel is newly available. */
- return s390_process_res_acc_new_sch(schid);
+ struct res_acc_data *res_data = data;
spin_lock_irq(sch->lock);
chp_mask = get_res_chpid_mask(&sch->ssd_info, res_data);
@@ -283,7 +269,7 @@ __s390_process_res_acc(struct subchannel
sch->driver->verify(sch);
out:
spin_unlock_irq(sch->lock);
- put_device(&sch->dev);
+
return 0;
}
@@ -306,7 +292,8 @@ static void s390_process_res_acc (struct
* The more information we have (info), the less scanning
* will we have to do.
*/
- for_each_subchannel(__s390_process_res_acc, res_data);
+ for_each_subchannel_staged(__s390_process_res_acc,
+ s390_process_res_acc_new_sch, res_data);
}
static int
@@ -500,8 +487,7 @@ void chsc_process_crw(void)
} while (sei_area->flags & 0x80);
}
-static int
-__chp_add_new_sch(struct subchannel_id schid)
+static int __chp_add_new_sch(struct subchannel_id schid, void *data)
{
struct schib schib;
@@ -515,35 +501,27 @@ __chp_add_new_sch(struct subchannel_id s
}
-static int
-__chp_add(struct subchannel_id schid, void *data)
+static int __chp_add(struct subchannel *sch, void *data)
{
int i, mask;
- struct chp_id *chpid;
- struct subchannel *sch;
+ struct chp_id *chpid = data;
- chpid = data;
- sch = get_subchannel_by_schid(schid);
- if (!sch)
- /* Check if the subchannel is now available. */
- return __chp_add_new_sch(schid);
spin_lock_irq(sch->lock);
for (i=0; i<8; i++) {
mask = 0x80 >> i;
if ((sch->schib.pmcw.pim & mask) &&
- (sch->schib.pmcw.chpid[i] == chpid->id)) {
- if (stsch(sch->schid, &sch->schib) != 0) {
- /* Endgame. */
- spin_unlock_irq(sch->lock);
- return -ENXIO;
- }
+ (sch->schib.pmcw.chpid[i] == chpid->id))
break;
- }
}
if (i==8) {
spin_unlock_irq(sch->lock);
return 0;
}
+ if (stsch(sch->schid, &sch->schib)) {
+ spin_unlock_irq(sch->lock);
+ css_schedule_eval(sch->schid);
+ return 0;
+ }
sch->lpm = ((sch->schib.pmcw.pim &
sch->schib.pmcw.pam &
sch->schib.pmcw.pom)
@@ -553,7 +531,7 @@ __chp_add(struct subchannel_id schid, vo
sch->driver->verify(sch);
spin_unlock_irq(sch->lock);
- put_device(&sch->dev);
+
return 0;
}
@@ -565,7 +543,8 @@ void chsc_chp_online(struct chp_id chpid
CIO_TRACE_EVENT(2, dbf_txt);
if (chp_get_status(chpid) != 0)
- for_each_subchannel(__chp_add, &chpid);
+ for_each_subchannel_staged(__chp_add, __chp_add_new_sch,
+ &chpid);
}
static void __s390_subchannel_vary_chpid(struct subchannel *sch,
@@ -616,25 +595,17 @@ static void __s390_subchannel_vary_chpid
spin_unlock_irqrestore(sch->lock, flags);
}
-static int s390_subchannel_vary_chpid_off(struct device *dev, void *data)
+static int s390_subchannel_vary_chpid_off(struct subchannel *sch, void *data)
{
- struct subchannel *sch;
- struct chp_id *chpid;
-
- sch = to_subchannel(dev);
- chpid = data;
+ struct chp_id *chpid = data;
__s390_subchannel_vary_chpid(sch, *chpid, 0);
return 0;
}
-static int s390_subchannel_vary_chpid_on(struct device *dev, void *data)
+static int s390_subchannel_vary_chpid_on(struct subchannel *sch, void *data)
{
- struct subchannel *sch;
- struct chp_id *chpid;
-
- sch = to_subchannel(dev);
- chpid = data;
+ struct chp_id *chpid = data;
__s390_subchannel_vary_chpid(sch, *chpid, 1);
return 0;
@@ -644,13 +615,7 @@ static int
__s390_vary_chpid_on(struct subchannel_id schid, void *data)
{
struct schib schib;
- struct subchannel *sch;
- sch = get_subchannel_by_schid(schid);
- if (sch) {
- put_device(&sch->dev);
- return 0;
- }
if (stsch_err(schid, &schib))
/* We're through */
return -ENXIO;
@@ -670,12 +635,13 @@ int chsc_chp_vary(struct chp_id chpid, i
* Redo PathVerification on the devices the chpid connects to
*/
- bus_for_each_dev(&css_bus_type, NULL, &chpid, on ?
- s390_subchannel_vary_chpid_on :
- s390_subchannel_vary_chpid_off);
if (on)
- /* Scan for new devices on varied on path. */
- for_each_subchannel(__s390_vary_chpid_on, NULL);
+ for_each_subchannel_staged(s390_subchannel_vary_chpid_on,
+ __s390_vary_chpid_on, &chpid);
+ else
+ for_each_subchannel_staged(s390_subchannel_vary_chpid_off,
+ NULL, &chpid);
+
return 0;
}
Index: quilt-2.6/drivers/s390/cio/css.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.c
+++ quilt-2.6/drivers/s390/cio/css.c
@@ -51,6 +51,62 @@ for_each_subchannel(int(*fn)(struct subc
return ret;
}
+struct cb_data {
+ void *data;
+ struct idset *set;
+ int (*fn_known_sch)(struct subchannel *, void *);
+ int (*fn_unknown_sch)(struct subchannel_id, void *);
+};
+
+static int call_fn_known_sch(struct device *dev, void *data)
+{
+ struct subchannel *sch = to_subchannel(dev);
+ struct cb_data *cb = data;
+ int rc = 0;
+
+ idset_sch_del(cb->set, sch->schid);
+ if (cb->fn_known_sch)
+ rc = cb->fn_known_sch(sch, cb->data);
+ return rc;
+}
+
+static int call_fn_unknown_sch(struct subchannel_id schid, void *data)
+{
+ struct cb_data *cb = data;
+ int rc = 0;
+
+ if (idset_sch_contains(cb->set, schid))
+ rc = cb->fn_unknown_sch(schid, cb->data);
+ return rc;
+}
+
+int for_each_subchannel_staged(int (*fn_known)(struct subchannel *, void *),
+ int (*fn_unknown)(struct subchannel_id,
+ void *), void *data)
+{
+ struct cb_data cb;
+ int rc;
+
+ cb.set = idset_sch_new();
+ if (!cb.set)
+ return -ENOMEM;
+ idset_fill(cb.set);
+ cb.data = data;
+ cb.fn_known_sch = fn_known;
+ cb.fn_unknown_sch = fn_unknown;
+ /* Process registered subchannels. */
+ rc = bus_for_each_dev(&css_bus_type, NULL, &cb, call_fn_known_sch);
+ if (rc)
+ goto out;
+ /* Process unregistered subchannels. */
+ if (fn_unknown)
+ rc = for_each_subchannel(call_fn_unknown_sch, &cb);
+out:
+ idset_free(cb.set);
+
+ return rc;
+}
+
static struct subchannel *
css_alloc_subchannel(struct subchannel_id schid)
{
@@ -402,20 +458,56 @@ static int __init slow_subchannel_init(v
return 0;
}
-static void css_slow_path_func(struct work_struct *unused)
+static int slow_eval_known_fn(struct subchannel *sch, void *data)
{
- struct subchannel_id schid;
+ int eval;
+ int rc;
- CIO_TRACE_EVENT(4, "slowpath");
spin_lock_irq(&slow_subchannel_lock);
- init_subchannel_id(&schid);
- while (idset_sch_get_first(slow_subchannel_set, &schid)) {
- idset_sch_del(slow_subchannel_set, schid);
- spin_unlock_irq(&slow_subchannel_lock);
- css_evaluate_subchannel(schid, 1);
- spin_lock_irq(&slow_subchannel_lock);
+ eval = idset_sch_contains(slow_subchannel_set, sch->schid);
+ idset_sch_del(slow_subchannel_set, sch->schid);
+ spin_unlock_irq(&slow_subchannel_lock);
+ if (eval) {
+ rc = css_evaluate_known_subchannel(sch, 1);
+ if (rc == -EAGAIN)
+ css_schedule_eval(sch->schid);
}
+ return 0;
+}
+
+static int slow_eval_unknown_fn(struct subchannel_id schid, void *data)
+{
+ int eval;
+ int rc = 0;
+
+ spin_lock_irq(&slow_subchannel_lock);
+ eval = idset_sch_contains(slow_subchannel_set, schid);
+ idset_sch_del(slow_subchannel_set, schid);
spin_unlock_irq(&slow_subchannel_lock);
+ if (eval) {
+ rc = css_evaluate_new_subchannel(schid, 1);
+ switch (rc) {
+ case -EAGAIN:
+ css_schedule_eval(schid);
+ rc = 0;
+ break;
+ case -ENXIO:
+ case -ENOMEM:
+ case -EIO:
+ /* These should abort looping */
+ break;
+ default:
+ rc = 0;
+ }
+ }
+ return rc;
+}
+
+static void css_slow_path_func(struct work_struct *unused)
+{
+ CIO_TRACE_EVENT(4, "slowpath");
+ for_each_subchannel_staged(slow_eval_known_fn, slow_eval_unknown_fn,
+ NULL);
}
static DECLARE_WORK(slow_path_work, css_slow_path_func);
@@ -444,7 +536,6 @@ void css_schedule_eval_all(void)
/* Reprobe subchannel if unregistered. */
static int reprobe_subchannel(struct subchannel_id schid, void *data)
{
- struct subchannel *sch;
int ret;
CIO_MSG_EVENT(6, "cio: reprobe 0.%x.%04x\n",
@@ -452,13 +543,6 @@ static int reprobe_subchannel(struct sub
if (need_reprobe)
return -EAGAIN;
- sch = get_subchannel_by_schid(schid);
- if (sch) {
- /* Already known. */
- put_device(&sch->dev);
- return 0;
- }
-
ret = css_probe_device(schid);
switch (ret) {
case 0:
@@ -486,7 +570,7 @@ static void reprobe_all(struct work_stru
/* Make sure initial subchannel scan is done. */
wait_event(ccw_device_init_wq,
atomic_read(&ccw_device_init_count) == 0);
- ret = for_each_subchannel(reprobe_subchannel, NULL);
+ ret = for_each_subchannel_staged(NULL, reprobe_subchannel, NULL);
CIO_MSG_EVENT(2, "reprobe done (rc=%d, need_reprobe=%d)\n", ret,
need_reprobe);
Index: quilt-2.6/drivers/s390/cio/css.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/css.h
+++ quilt-2.6/drivers/s390/cio/css.h
@@ -91,6 +91,9 @@ extern void css_driver_unregister(struct
extern void css_sch_device_unregister(struct subchannel *);
extern struct subchannel * get_subchannel_by_schid(struct subchannel_id);
extern int css_init_done;
+int for_each_subchannel_staged(int (*fn_known)(struct subchannel *, void *),
+ int (*fn_unknown)(struct subchannel_id,
+ void *), void *data);
extern int for_each_subchannel(int(*fn)(struct subchannel_id, void *), void *);
extern void css_process_crw(int, int);
extern void css_reiterate_subchannels(void);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 14/47] qdio: Remove double checked value.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (12 preceding siblings ...)
2007-12-20 15:19 ` [patch 13/47] cio: reduce cpu utilization during device scan Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 15/47] qdio: set QDIO_ACTIVATE_TIMEOUT to 5s Martin Schwidefsky
` (32 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Sebastian Ott, Martin Schwidefsky
[-- Attachment #1: 113-qdio-doublecheck.diff --]
[-- Type: text/plain, Size: 846 bytes --]
From: Sebastian Ott <sebott@de.ibm.com>
Signed-off-by: Sebastian Ott <sebott@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/qdio.c | 3 +--
1 file changed, 1 insertion(+), 2 deletions(-)
Index: quilt-2.6/drivers/s390/cio/qdio.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/qdio.c
+++ quilt-2.6/drivers/s390/cio/qdio.c
@@ -1408,8 +1408,7 @@ __tiqdio_inbound_processing(struct qdio_
if (q->hydra_gives_outbound_pcis) {
if (!q->siga_sync_done_on_thinints) {
SYNC_MEMORY_ALL;
- } else if ((!q->siga_sync_done_on_outb_tis)&&
- (q->hydra_gives_outbound_pcis)) {
+ } else if (!q->siga_sync_done_on_outb_tis) {
SYNC_MEMORY_ALL_OUTB;
}
} else {
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 15/47] qdio: set QDIO_ACTIVATE_TIMEOUT to 5s
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (13 preceding siblings ...)
2007-12-20 15:19 ` [patch 14/47] qdio: Remove double checked value Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 16/47] sclp: sysfs interface for SCLP cpi Martin Schwidefsky
` (31 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Ursula Braun, Martin Schwidefsky
[-- Attachment #1: 114-qdio-timeout.diff --]
[-- Type: text/plain, Size: 927 bytes --]
From: Ursula Braun <braunu@de.ibm.com>
Current definition of QDIO_ACTIVATE_TIMEOUT results in value 0.
Thus it may cause endless wait in function qdio_activate().
Signed-off-by: Ursula Braun <braunu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/cio/qdio.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Index: quilt-2.6/drivers/s390/cio/qdio.h
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/qdio.h
+++ quilt-2.6/drivers/s390/cio/qdio.h
@@ -57,7 +57,7 @@
of the queue to 0 */
#define QDIO_ESTABLISH_TIMEOUT (1*HZ)
-#define QDIO_ACTIVATE_TIMEOUT ((5*HZ)>>10)
+#define QDIO_ACTIVATE_TIMEOUT (5*HZ)
#define QDIO_CLEANUP_CLEAR_TIMEOUT (20*HZ)
#define QDIO_CLEANUP_HALT_TIMEOUT (10*HZ)
#define QDIO_FORCE_CHECK_TIMEOUT (10*HZ)
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 16/47] sclp: sysfs interface for SCLP cpi
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (14 preceding siblings ...)
2007-12-20 15:19 ` [patch 15/47] qdio: set QDIO_ACTIVATE_TIMEOUT to 5s Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 17/47] Standby cpu activation/deactivation Martin Schwidefsky
` (30 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Michael Ernst, Martin Schwidefsky
[-- Attachment #1: 115-sclp-sysfs.diff --]
[-- Type: text/plain, Size: 17645 bytes --]
From: Michael Ernst <mernst@de.ibm.com>
Signed-off-by: Michael Ernst <mernst@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/char/Makefile | 2
drivers/s390/char/sclp_cpi.c | 246 +-----------------------
drivers/s390/char/sclp_cpi_sys.c | 390 +++++++++++++++++++++++++++++++++++++++
drivers/s390/char/sclp_cpi_sys.h | 15 +
4 files changed, 422 insertions(+), 231 deletions(-)
Index: quilt-2.6/drivers/s390/char/Makefile
===================================================================
--- quilt-2.6.orig/drivers/s390/char/Makefile
+++ quilt-2.6/drivers/s390/char/Makefile
@@ -3,7 +3,7 @@
#
obj-y += ctrlchar.o keyboard.o defkeymap.o sclp.o sclp_rw.o sclp_quiesce.o \
- sclp_info.o sclp_config.o sclp_chp.o
+ sclp_info.o sclp_config.o sclp_chp.o sclp_cpi_sys.o
obj-$(CONFIG_TN3270) += raw3270.o
obj-$(CONFIG_TN3270_CONSOLE) += con3270.o
Index: quilt-2.6/drivers/s390/char/sclp_cpi.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/sclp_cpi.c
+++ quilt-2.6/drivers/s390/char/sclp_cpi.c
@@ -1,255 +1,41 @@
/*
- * Author: Martin Peschke <mpeschke@de.ibm.com>
- * Copyright (C) 2001 IBM Entwicklung GmbH, IBM Corporation
+ * drivers/s390/char/sclp_cpi.c
+ * SCLP control programm identification
*
- * SCLP Control-Program Identification.
+ * Copyright IBM Corp. 2001, 2007
+ * Author(s): Martin Peschke <mpeschke@de.ibm.com>
+ * Michael Ernst <mernst@de.ibm.com>
*/
-#include <linux/version.h>
#include <linux/kmod.h>
#include <linux/module.h>
#include <linux/moduleparam.h>
-#include <linux/init.h>
-#include <linux/timer.h>
-#include <linux/string.h>
-#include <linux/err.h>
-#include <linux/slab.h>
-#include <asm/ebcdic.h>
-#include <asm/semaphore.h>
-
-#include "sclp.h"
-#include "sclp_rw.h"
-
-#define CPI_LENGTH_SYSTEM_TYPE 8
-#define CPI_LENGTH_SYSTEM_NAME 8
-#define CPI_LENGTH_SYSPLEX_NAME 8
-
-struct cpi_evbuf {
- struct evbuf_header header;
- u8 id_format;
- u8 reserved0;
- u8 system_type[CPI_LENGTH_SYSTEM_TYPE];
- u64 reserved1;
- u8 system_name[CPI_LENGTH_SYSTEM_NAME];
- u64 reserved2;
- u64 system_level;
- u64 reserved3;
- u8 sysplex_name[CPI_LENGTH_SYSPLEX_NAME];
- u8 reserved4[16];
-} __attribute__((packed));
-
-struct cpi_sccb {
- struct sccb_header header;
- struct cpi_evbuf cpi_evbuf;
-} __attribute__((packed));
-
-/* Event type structure for write message and write priority message */
-static struct sclp_register sclp_cpi_event =
-{
- .send_mask = EVTYP_CTLPROGIDENT_MASK
-};
+#include <linux/version.h>
+#include "sclp_cpi_sys.h"
MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION("Identify this operating system instance "
+ "to the System z hardware");
+MODULE_AUTHOR("Martin Peschke <mpeschke@de.ibm.com>, "
+ "Michael Ernst <mernst@de.ibm.com>");
-MODULE_AUTHOR(
- "Martin Peschke, IBM Deutschland Entwicklung GmbH "
- "<mpeschke@de.ibm.com>");
-
-MODULE_DESCRIPTION(
- "identify this operating system instance to the S/390 "
- "or zSeries hardware");
+static char *system_name = "";
+static char *sysplex_name = "";
-static char *system_name = NULL;
module_param(system_name, charp, 0);
MODULE_PARM_DESC(system_name, "e.g. hostname - max. 8 characters");
-
-static char *sysplex_name = NULL;
-#ifdef ALLOW_SYSPLEX_NAME
module_param(sysplex_name, charp, 0);
MODULE_PARM_DESC(sysplex_name, "if applicable - max. 8 characters");
-#endif
-
-/* use default value for this field (as well as for system level) */
-static char *system_type = "LINUX";
-
-static int
-cpi_check_parms(void)
-{
- /* reject if no system type specified */
- if (!system_type) {
- printk("cpi: bug: no system type specified\n");
- return -EINVAL;
- }
-
- /* reject if system type larger than 8 characters */
- if (strlen(system_type) > CPI_LENGTH_SYSTEM_NAME) {
- printk("cpi: bug: system type has length of %li characters - "
- "only %i characters supported\n",
- strlen(system_type), CPI_LENGTH_SYSTEM_TYPE);
- return -EINVAL;
- }
-
- /* reject if no system name specified */
- if (!system_name) {
- printk("cpi: no system name specified\n");
- return -EINVAL;
- }
-
- /* reject if system name larger than 8 characters */
- if (strlen(system_name) > CPI_LENGTH_SYSTEM_NAME) {
- printk("cpi: system name has length of %li characters - "
- "only %i characters supported\n",
- strlen(system_name), CPI_LENGTH_SYSTEM_NAME);
- return -EINVAL;
- }
-
- /* reject if specified sysplex name larger than 8 characters */
- if (sysplex_name && strlen(sysplex_name) > CPI_LENGTH_SYSPLEX_NAME) {
- printk("cpi: sysplex name has length of %li characters"
- " - only %i characters supported\n",
- strlen(sysplex_name), CPI_LENGTH_SYSPLEX_NAME);
- return -EINVAL;
- }
- return 0;
-}
-
-static void
-cpi_callback(struct sclp_req *req, void *data)
-{
- struct semaphore *sem;
-
- sem = (struct semaphore *) data;
- up(sem);
-}
-
-static struct sclp_req *
-cpi_prepare_req(void)
-{
- struct sclp_req *req;
- struct cpi_sccb *sccb;
- struct cpi_evbuf *evb;
-
- req = kmalloc(sizeof(struct sclp_req), GFP_KERNEL);
- if (req == NULL)
- return ERR_PTR(-ENOMEM);
- sccb = (struct cpi_sccb *) __get_free_page(GFP_KERNEL | GFP_DMA);
- if (sccb == NULL) {
- kfree(req);
- return ERR_PTR(-ENOMEM);
- }
- memset(sccb, 0, sizeof(struct cpi_sccb));
-
- /* setup SCCB for Control-Program Identification */
- sccb->header.length = sizeof(struct cpi_sccb);
- sccb->cpi_evbuf.header.length = sizeof(struct cpi_evbuf);
- sccb->cpi_evbuf.header.type = 0x0B;
- evb = &sccb->cpi_evbuf;
-
- /* set system type */
- memset(evb->system_type, ' ', CPI_LENGTH_SYSTEM_TYPE);
- memcpy(evb->system_type, system_type, strlen(system_type));
- sclp_ascebc_str(evb->system_type, CPI_LENGTH_SYSTEM_TYPE);
- EBC_TOUPPER(evb->system_type, CPI_LENGTH_SYSTEM_TYPE);
-
- /* set system name */
- memset(evb->system_name, ' ', CPI_LENGTH_SYSTEM_NAME);
- memcpy(evb->system_name, system_name, strlen(system_name));
- sclp_ascebc_str(evb->system_name, CPI_LENGTH_SYSTEM_NAME);
- EBC_TOUPPER(evb->system_name, CPI_LENGTH_SYSTEM_NAME);
-
- /* set system level */
- evb->system_level = LINUX_VERSION_CODE;
-
- /* set sysplex name */
- if (sysplex_name) {
- memset(evb->sysplex_name, ' ', CPI_LENGTH_SYSPLEX_NAME);
- memcpy(evb->sysplex_name, sysplex_name, strlen(sysplex_name));
- sclp_ascebc_str(evb->sysplex_name, CPI_LENGTH_SYSPLEX_NAME);
- EBC_TOUPPER(evb->sysplex_name, CPI_LENGTH_SYSPLEX_NAME);
- }
-
- /* prepare request data structure presented to SCLP driver */
- req->command = SCLP_CMDW_WRITE_EVENT_DATA;
- req->sccb = sccb;
- req->status = SCLP_REQ_FILLED;
- req->callback = cpi_callback;
- return req;
-}
-
-static void
-cpi_free_req(struct sclp_req *req)
-{
- free_page((unsigned long) req->sccb);
- kfree(req);
-}
-static int __init
-cpi_module_init(void)
+static int __init cpi_module_init(void)
{
- struct semaphore sem;
- struct sclp_req *req;
- int rc;
-
- rc = cpi_check_parms();
- if (rc)
- return rc;
-
- rc = sclp_register(&sclp_cpi_event);
- if (rc) {
- /* could not register sclp event. Die. */
- printk(KERN_WARNING "cpi: could not register to hardware "
- "console.\n");
- return -EINVAL;
- }
- if (!(sclp_cpi_event.sclp_send_mask & EVTYP_CTLPROGIDENT_MASK)) {
- printk(KERN_WARNING "cpi: no control program identification "
- "support\n");
- sclp_unregister(&sclp_cpi_event);
- return -EOPNOTSUPP;
- }
-
- req = cpi_prepare_req();
- if (IS_ERR(req)) {
- printk(KERN_WARNING "cpi: couldn't allocate request\n");
- sclp_unregister(&sclp_cpi_event);
- return PTR_ERR(req);
- }
-
- /* Prepare semaphore */
- sema_init(&sem, 0);
- req->callback_data = &sem;
- /* Add request to sclp queue */
- rc = sclp_add_request(req);
- if (rc) {
- printk(KERN_WARNING "cpi: could not start request\n");
- cpi_free_req(req);
- sclp_unregister(&sclp_cpi_event);
- return rc;
- }
- /* make "insmod" sleep until callback arrives */
- down(&sem);
-
- rc = ((struct cpi_sccb *) req->sccb)->header.response_code;
- if (rc != 0x0020) {
- printk(KERN_WARNING "cpi: failed with response code 0x%x\n",
- rc);
- rc = -ECOMM;
- } else
- rc = 0;
-
- cpi_free_req(req);
- sclp_unregister(&sclp_cpi_event);
-
- return rc;
+ return sclp_cpi_set_data(system_name, sysplex_name, "LINUX",
+ LINUX_VERSION_CODE);
}
-
static void __exit cpi_module_exit(void)
{
}
-
-/* declare driver module init/cleanup functions */
module_init(cpi_module_init);
module_exit(cpi_module_exit);
-
Index: quilt-2.6/drivers/s390/char/sclp_cpi_sys.c
===================================================================
--- /dev/null
+++ quilt-2.6/drivers/s390/char/sclp_cpi_sys.c
@@ -0,0 +1,390 @@
+/*
+ * drivers/s390/char/sclp_cpi_sys.c
+ * SCLP control program identification sysfs interface
+ *
+ * Copyright IBM Corp. 2001, 2007
+ * Author(s): Martin Peschke <mpeschke@de.ibm.com>
+ * Michael Ernst <mernst@de.ibm.com>
+ */
+
+#include <linux/kernel.h>
+#include <linux/init.h>
+#include <linux/stat.h>
+#include <linux/device.h>
+#include <linux/string.h>
+#include <linux/ctype.h>
+#include <linux/kmod.h>
+#include <linux/timer.h>
+#include <linux/err.h>
+#include <linux/slab.h>
+#include <linux/completion.h>
+#include <asm/ebcdic.h>
+#include <asm/sclp.h>
+#include "sclp.h"
+#include "sclp_rw.h"
+#include "sclp_cpi_sys.h"
+
+#define CPI_LENGTH_NAME 8
+#define CPI_LENGTH_LEVEL 16
+
+struct cpi_evbuf {
+ struct evbuf_header header;
+ u8 id_format;
+ u8 reserved0;
+ u8 system_type[CPI_LENGTH_NAME];
+ u64 reserved1;
+ u8 system_name[CPI_LENGTH_NAME];
+ u64 reserved2;
+ u64 system_level;
+ u64 reserved3;
+ u8 sysplex_name[CPI_LENGTH_NAME];
+ u8 reserved4[16];
+} __attribute__((packed));
+
+struct cpi_sccb {
+ struct sccb_header header;
+ struct cpi_evbuf cpi_evbuf;
+} __attribute__((packed));
+
+static struct sclp_register sclp_cpi_event = {
+ .send_mask = EVTYP_CTLPROGIDENT_MASK,
+};
+
+static char system_name[CPI_LENGTH_NAME + 1];
+static char sysplex_name[CPI_LENGTH_NAME + 1];
+static char system_type[CPI_LENGTH_NAME + 1];
+static u64 system_level;
+
+static void set_data(char *field, char *data)
+{
+ memset(field, ' ', CPI_LENGTH_NAME);
+ memcpy(field, data, strlen(data));
+ sclp_ascebc_str(field, CPI_LENGTH_NAME);
+}
+
+static void cpi_callback(struct sclp_req *req, void *data)
+{
+ struct completion *completion = data;
+
+ complete(completion);
+}
+
+static struct sclp_req *cpi_prepare_req(void)
+{
+ struct sclp_req *req;
+ struct cpi_sccb *sccb;
+ struct cpi_evbuf *evb;
+
+ req = kzalloc(sizeof(struct sclp_req), GFP_KERNEL);
+ if (!req)
+ return ERR_PTR(-ENOMEM);
+ sccb = (struct cpi_sccb *) get_zeroed_page(GFP_KERNEL | GFP_DMA);
+ if (!sccb) {
+ kfree(req);
+ return ERR_PTR(-ENOMEM);
+ }
+
+ /* setup SCCB for Control-Program Identification */
+ sccb->header.length = sizeof(struct cpi_sccb);
+ sccb->cpi_evbuf.header.length = sizeof(struct cpi_evbuf);
+ sccb->cpi_evbuf.header.type = 0x0b;
+ evb = &sccb->cpi_evbuf;
+
+ /* set system type */
+ set_data(evb->system_type, system_type);
+
+ /* set system name */
+ set_data(evb->system_name, system_name);
+
+ /* set sytem level */
+ evb->system_level = system_level;
+
+ /* set sysplex name */
+ set_data(evb->sysplex_name, sysplex_name);
+
+ /* prepare request data structure presented to SCLP driver */
+ req->command = SCLP_CMDW_WRITE_EVENT_DATA;
+ req->sccb = sccb;
+ req->status = SCLP_REQ_FILLED;
+ req->callback = cpi_callback;
+ return req;
+}
+
+static void cpi_free_req(struct sclp_req *req)
+{
+ free_page((unsigned long) req->sccb);
+ kfree(req);
+}
+
+static int cpi_req(void)
+{
+ struct completion completion;
+ struct sclp_req *req;
+ int rc;
+ int response;
+
+ rc = sclp_register(&sclp_cpi_event);
+ if (rc) {
+ printk(KERN_WARNING "cpi: could not register "
+ "to hardware console.\n");
+ goto out;
+ }
+ if (!(sclp_cpi_event.sclp_send_mask & EVTYP_CTLPROGIDENT_MASK)) {
+ printk(KERN_WARNING "cpi: no control program "
+ "identification support\n");
+ rc = -EOPNOTSUPP;
+ goto out_unregister;
+ }
+
+ req = cpi_prepare_req();
+ if (IS_ERR(req)) {
+ printk(KERN_WARNING "cpi: could not allocate request\n");
+ rc = PTR_ERR(req);
+ goto out_unregister;
+ }
+
+ init_completion(&completion);
+ req->callback_data = &completion;
+
+ /* Add request to sclp queue */
+ rc = sclp_add_request(req);
+ if (rc) {
+ printk(KERN_WARNING "cpi: could not start request\n");
+ goto out_free_req;
+ }
+
+ wait_for_completion(&completion);
+
+ if (req->status != SCLP_REQ_DONE) {
+ printk(KERN_WARNING "cpi: request failed (status=0x%02x)\n",
+ req->status);
+ rc = -EIO;
+ goto out_free_req;
+ }
+
+ response = ((struct cpi_sccb *) req->sccb)->header.response_code;
+ if (response != 0x0020) {
+ printk(KERN_WARNING "cpi: failed with "
+ "response code 0x%x\n", response);
+ rc = -EIO;
+ }
+
+out_free_req:
+ cpi_free_req(req);
+
+out_unregister:
+ sclp_unregister(&sclp_cpi_event);
+
+out:
+ return rc;
+}
+
+static int check_string(const char *attr, const char *str)
+{
+ size_t len;
+ size_t i;
+
+ len = strlen(str);
+
+ if ((len > 0) && (str[len - 1] == '\n'))
+ len--;
+
+ if (len > CPI_LENGTH_NAME)
+ return -EINVAL;
+
+ for (i = 0; i < len ; i++) {
+ if (isalpha(str[i]) || isdigit(str[i]) ||
+ strchr("$@# ", str[i]))
+ continue;
+ return -EINVAL;
+ }
+
+ return 0;
+}
+
+static void set_string(char *attr, const char *value)
+{
+ size_t len;
+ size_t i;
+
+ len = strlen(value);
+
+ if ((len > 0) && (value[len - 1] == '\n'))
+ len--;
+
+ for (i = 0; i < CPI_LENGTH_NAME; i++) {
+ if (i < len)
+ attr[i] = toupper(value[i]);
+ else
+ attr[i] = ' ';
+ }
+}
+
+static ssize_t system_name_show(struct kset *kset, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%s\n", system_name);
+}
+
+static ssize_t system_name_store(struct kset *kset, const char *buf,
+ size_t len)
+{
+ int rc;
+
+ rc = check_string("system_name", buf);
+ if (rc)
+ return rc;
+
+ set_string(system_name, buf);
+
+ return len;
+}
+
+static struct subsys_attribute system_name_attr =
+ __ATTR(system_name, 0644, system_name_show, system_name_store);
+
+static ssize_t sysplex_name_show(struct kset *kset, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%s\n", sysplex_name);
+}
+
+static ssize_t sysplex_name_store(struct kset *kset, const char *buf,
+ size_t len)
+{
+ int rc;
+
+ rc = check_string("sysplex_name", buf);
+ if (rc)
+ return rc;
+
+ set_string(sysplex_name, buf);
+
+ return len;
+}
+
+static struct subsys_attribute sysplex_name_attr =
+ __ATTR(sysplex_name, 0644, sysplex_name_show,
+ sysplex_name_store);
+
+static ssize_t system_type_show(struct kset *kset, char *page)
+{
+ return snprintf(page, PAGE_SIZE, "%s\n", system_type);
+}
+
+static ssize_t system_type_store(struct kset *kset, const char *buf,
+ size_t len)
+{
+ int rc;
+
+ rc = check_string("system_type", buf);
+ if (rc)
+ return rc;
+
+ set_string(system_type, buf);
+
+ return len;
+}
+
+static struct subsys_attribute system_type_attr =
+ __ATTR(system_type, 0644, system_type_show, system_type_store);
+
+static ssize_t system_level_show(struct kset *kset, char *page)
+{
+ unsigned long long level = system_level;
+
+ return snprintf(page, PAGE_SIZE, "%#018llx\n", level);
+}
+
+static ssize_t system_level_store(struct kset *kset, const char *buf,
+ size_t len)
+{
+ unsigned long long level;
+ char *endp;
+
+ level = simple_strtoull(buf, &endp, 16);
+
+ if (endp == buf)
+ return -EINVAL;
+ if (*endp == '\n')
+ endp++;
+ if (*endp)
+ return -EINVAL;
+
+ system_level = level;
+
+ return len;
+}
+
+static struct subsys_attribute system_level_attr =
+ __ATTR(system_level, 0644, system_level_show,
+ system_level_store);
+
+static ssize_t set_store(struct kset *kset, const char *buf,
+ size_t len)
+{
+ int rc;
+
+ rc = cpi_req();
+ if (rc)
+ return rc;
+
+ return len;
+}
+
+static struct subsys_attribute set_attr =
+ __ATTR(set, 0200, NULL, set_store);
+
+static struct attribute *cpi_attrs[] = {
+ &system_name_attr.attr,
+ &sysplex_name_attr.attr,
+ &system_type_attr.attr,
+ &system_level_attr.attr,
+ &set_attr.attr,
+ NULL,
+};
+
+static struct attribute_group cpi_attr_group = {
+ .attrs = cpi_attrs,
+};
+
+static decl_subsys(cpi, NULL, NULL);
+
+int sclp_cpi_set_data(const char *system, const char *sysplex, const char *type,
+ const u64 level)
+{
+ int rc;
+
+ rc = check_string("system_name", system);
+ if (rc)
+ return rc;
+ rc = check_string("sysplex_name", sysplex);
+ if (rc)
+ return rc;
+ rc = check_string("system_type", type);
+ if (rc)
+ return rc;
+
+ set_string(system_name, system);
+ set_string(sysplex_name, sysplex);
+ set_string(system_type, type);
+ system_level = level;
+
+ return cpi_req();
+}
+EXPORT_SYMBOL(sclp_cpi_set_data);
+
+static int __init cpi_init(void)
+{
+ int rc;
+
+ rc = firmware_register(&cpi_subsys);
+ if (rc)
+ return rc;
+
+ rc = sysfs_create_group(&cpi_subsys.kobj, &cpi_attr_group);
+ if (rc)
+ firmware_unregister(&cpi_subsys);
+
+ return rc;
+}
+
+__initcall(cpi_init);
Index: quilt-2.6/drivers/s390/char/sclp_cpi_sys.h
===================================================================
--- /dev/null
+++ quilt-2.6/drivers/s390/char/sclp_cpi_sys.h
@@ -0,0 +1,15 @@
+/*
+ * drivers/s390/char/sclp_cpi_sys.h
+ * SCLP control program identification sysfs interface
+ *
+ * Copyright IBM Corp. 2007
+ * Author(s): Michael Ernst <mernst@de.ibm.com>
+ */
+
+#ifndef __SCLP_CPI_SYS_H__
+#define __SCLP_CPI_SYS_H__
+
+int sclp_cpi_set_data(const char *system, const char *sysplex,
+ const char *type, u64 level);
+
+#endif /* __SCLP_CPI_SYS_H__ */
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 17/47] Standby cpu activation/deactivation.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (15 preceding siblings ...)
2007-12-20 15:19 ` [patch 16/47] sclp: sysfs interface for SCLP cpi Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 18/47] sclp: convert channel path configure code to use sync interface Martin Schwidefsky
` (29 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 116-sclp-cpu.diff --]
[-- Type: text/plain, Size: 29839 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Add a new interface so that cpus can be put into standby state and
configured state.
Only offline cpus can be put into standby state or configured state.
For that the new percpu sysfs attribute "configure" must be used.
To put a cpu in standby state a "0" must be written to the attribute.
In order to switch it into configured state a "1" must be written to
the attribute.
Only cpus in configured state can be brought online.
In addition this patch introduces a static mapping of physical to
logical cpus. As a result only the sysfs directories of present cpus
will be created. To scan for new cpus the new sysfs attribute "rescan"
must be used.
Writing to /sys/devices/system/cpu/rescan will trigger a rescan of
cpus and will create directories for new cpus.
On IPL only configured cpus will be used. And on reboot/shutdown all
cpus will remain in their current state (configured/standby).
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/early.c | 3
arch/s390/kernel/setup.c | 2
arch/s390/kernel/smp.c | 401 ++++++++++++++++++++++++++++++++++--------
drivers/s390/char/Makefile | 2
drivers/s390/char/sclp.h | 4
drivers/s390/char/sclp_cmd.c | 319 +++++++++++++++++++++++++++++++++
drivers/s390/char/sclp_info.c | 116 ------------
include/asm-s390/sclp.h | 21 ++
8 files changed, 675 insertions(+), 193 deletions(-)
Index: quilt-2.6/arch/s390/kernel/early.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/early.c
+++ quilt-2.6/arch/s390/kernel/early.c
@@ -276,8 +276,9 @@ void __init startup_init(void)
create_kernel_nss();
sort_main_extable();
setup_lowcore_early();
- sclp_readinfo_early();
+ sclp_read_info_early();
sclp_facilities_detect();
+ sclp_read_cpu_info_early();
memsize = sclp_memory_detect();
#ifndef CONFIG_64BIT
/*
Index: quilt-2.6/arch/s390/kernel/setup.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/setup.c
+++ quilt-2.6/arch/s390/kernel/setup.c
@@ -920,7 +920,7 @@ setup_arch(char **cmdline_p)
void __cpuinit print_cpu_info(struct cpuinfo_S390 *cpuinfo)
{
- printk("cpu %d "
+ printk(KERN_INFO "cpu %d "
#ifdef CONFIG_SMP
"phys_idx=%d "
#endif
Index: quilt-2.6/arch/s390/kernel/smp.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/smp.c
+++ quilt-2.6/arch/s390/kernel/smp.c
@@ -42,6 +42,7 @@
#include <asm/tlbflush.h>
#include <asm/timer.h>
#include <asm/lowcore.h>
+#include <asm/sclp.h>
#include <asm/cpu.h>
/*
@@ -58,6 +59,22 @@ EXPORT_SYMBOL(cpu_possible_map);
static struct task_struct *current_set[NR_CPUS];
+static u8 smp_cpu_type;
+static int smp_use_sigp_detection;
+
+enum s390_cpu_state {
+ CPU_STATE_STANDBY,
+ CPU_STATE_CONFIGURED,
+};
+
+#ifdef CONFIG_HOTPLUG_CPU
+static DEFINE_MUTEX(smp_cpu_state_mutex);
+#endif
+static int smp_cpu_state[NR_CPUS];
+
+static DEFINE_PER_CPU(struct cpu, cpu_devices);
+DEFINE_PER_CPU(struct s390_idle_data, s390_idle);
+
static void smp_ext_bitcall(int, ec_bit_sig);
/*
@@ -355,6 +372,13 @@ void smp_ctl_clear_bit(int cr, int bit)
}
EXPORT_SYMBOL(smp_ctl_clear_bit);
+/*
+ * In early ipl state a temp. logically cpu number is needed, so the sigp
+ * functions can be used to sense other cpus. Since NR_CPUS is >= 2 on
+ * CONFIG_SMP and the ipl cpu is logical cpu 0, it must be 1.
+ */
+#define CPU_INIT_NO 1
+
#if defined(CONFIG_ZFCPDUMP) || defined(CONFIG_ZFCPDUMP_MODULE)
/*
@@ -376,8 +400,9 @@ static void __init smp_get_save_area(uns
return;
}
zfcpdump_save_areas[cpu] = alloc_bootmem(sizeof(union save_area));
- __cpu_logical_map[1] = (__u16) phy_cpu;
- while (signal_processor(1, sigp_stop_and_store_status) == sigp_busy)
+ __cpu_logical_map[CPU_INIT_NO] = (__u16) phy_cpu;
+ while (signal_processor(CPU_INIT_NO, sigp_stop_and_store_status) ==
+ sigp_busy)
cpu_relax();
memcpy(zfcpdump_save_areas[cpu],
(void *)(unsigned long) store_prefix() + SAVE_AREA_BASE,
@@ -397,32 +422,166 @@ static inline void smp_get_save_area(uns
#endif /* CONFIG_ZFCPDUMP || CONFIG_ZFCPDUMP_MODULE */
+static int cpu_stopped(int cpu)
+{
+ __u32 status;
+
+ /* Check for stopped state */
+ if (signal_processor_ps(&status, 0, cpu, sigp_sense) ==
+ sigp_status_stored) {
+ if (status & 0x40)
+ return 1;
+ }
+ return 0;
+}
+
/*
* Lets check how many CPUs we have.
*/
-static unsigned int __init smp_count_cpus(void)
+static void __init smp_count_cpus(unsigned int *configured_cpus,
+ unsigned int *standby_cpus)
{
- unsigned int cpu, num_cpus;
- __u16 boot_cpu_addr;
+ unsigned int cpu;
+ struct sclp_cpu_info *info;
+ u16 boot_cpu_addr, cpu_addr;
- /*
- * cpu 0 is the boot cpu. See smp_prepare_boot_cpu.
- */
boot_cpu_addr = S390_lowcore.cpu_data.cpu_addr;
current_thread_info()->cpu = 0;
- num_cpus = 1;
- for (cpu = 0; cpu <= 65535; cpu++) {
- if ((__u16) cpu == boot_cpu_addr)
+ *configured_cpus = 1;
+ *standby_cpus = 0;
+
+ info = alloc_bootmem_pages(sizeof(*info));
+ if (!info)
+ disabled_wait((unsigned long) __builtin_return_address(0));
+
+ /* Use sigp detection algorithm if sclp doesn't work. */
+ if (sclp_get_cpu_info(info)) {
+ smp_use_sigp_detection = 1;
+ for (cpu = 0; cpu <= 65535; cpu++) {
+ if (cpu == boot_cpu_addr)
+ continue;
+ __cpu_logical_map[CPU_INIT_NO] = cpu;
+ if (cpu_stopped(CPU_INIT_NO))
+ (*configured_cpus)++;
+ }
+ goto out;
+ }
+
+ if (info->has_cpu_type) {
+ for (cpu = 0; cpu < info->combined; cpu++) {
+ if (info->cpu[cpu].address == boot_cpu_addr) {
+ smp_cpu_type = info->cpu[cpu].type;
+ break;
+ }
+ }
+ }
+ /* Count cpus. */
+ for (cpu = 0; cpu < info->combined; cpu++) {
+ if (info->has_cpu_type && info->cpu[cpu].type != smp_cpu_type)
continue;
- __cpu_logical_map[1] = (__u16) cpu;
- if (signal_processor(1, sigp_sense) == sigp_not_operational)
+ cpu_addr = info->cpu[cpu].address;
+ if (cpu_addr == boot_cpu_addr)
continue;
- smp_get_save_area(num_cpus, cpu);
- num_cpus++;
+ __cpu_logical_map[CPU_INIT_NO] = cpu_addr;
+ if (!cpu_stopped(CPU_INIT_NO)) {
+ (*standby_cpus)++;
+ continue;
+ }
+ smp_get_save_area(*configured_cpus, cpu_addr);
+ (*configured_cpus)++;
}
- printk("Detected %d CPU's\n", (int) num_cpus);
- printk("Boot cpu address %2X\n", boot_cpu_addr);
- return num_cpus;
+out:
+ printk(KERN_INFO "CPUs: %d configured, %d standby\n",
+ *configured_cpus, *standby_cpus);
+ free_bootmem((unsigned long) info, sizeof(*info));
+}
+
+static int cpu_known(int cpu_id)
+{
+ int cpu;
+
+ for_each_present_cpu(cpu) {
+ if (__cpu_logical_map[cpu] == cpu_id)
+ return 1;
+ }
+ return 0;
+}
+
+static int smp_rescan_cpus_sigp(cpumask_t avail)
+{
+ int cpu_id, logical_cpu;
+
+ logical_cpu = first_cpu(avail);
+ if (logical_cpu == NR_CPUS)
+ return 0;
+ for (cpu_id = 0; cpu_id <= 65535; cpu_id++) {
+ if (cpu_known(cpu_id))
+ continue;
+ __cpu_logical_map[logical_cpu] = cpu_id;
+ if (!cpu_stopped(logical_cpu))
+ continue;
+ cpu_set(logical_cpu, cpu_present_map);
+ smp_cpu_state[logical_cpu] = CPU_STATE_CONFIGURED;
+ logical_cpu = next_cpu(logical_cpu, avail);
+ if (logical_cpu == NR_CPUS)
+ break;
+ }
+ return 0;
+}
+
+static int __init_refok smp_rescan_cpus_sclp(cpumask_t avail)
+{
+ struct sclp_cpu_info *info;
+ int cpu_id, logical_cpu, cpu;
+ int rc;
+
+ logical_cpu = first_cpu(avail);
+ if (logical_cpu == NR_CPUS)
+ return 0;
+ if (slab_is_available())
+ info = kmalloc(sizeof(*info), GFP_KERNEL);
+ else
+ info = alloc_bootmem(sizeof(*info));
+ if (!info)
+ return -ENOMEM;
+ rc = sclp_get_cpu_info(info);
+ if (rc)
+ goto out;
+ for (cpu = 0; cpu < info->combined; cpu++) {
+ if (info->has_cpu_type && info->cpu[cpu].type != smp_cpu_type)
+ continue;
+ cpu_id = info->cpu[cpu].address;
+ if (cpu_known(cpu_id))
+ continue;
+ __cpu_logical_map[logical_cpu] = cpu_id;
+ cpu_set(logical_cpu, cpu_present_map);
+ if (cpu >= info->configured)
+ smp_cpu_state[logical_cpu] = CPU_STATE_STANDBY;
+ else
+ smp_cpu_state[logical_cpu] = CPU_STATE_CONFIGURED;
+ logical_cpu = next_cpu(logical_cpu, avail);
+ if (logical_cpu == NR_CPUS)
+ break;
+ }
+out:
+ if (slab_is_available())
+ kfree(info);
+ else
+ free_bootmem((unsigned long) info, sizeof(*info));
+ return rc;
+}
+
+static int smp_rescan_cpus(void)
+{
+ cpumask_t avail;
+
+ cpus_setall(avail);
+ cpus_and(avail, avail, cpu_possible_map);
+ cpus_andnot(avail, avail, cpu_present_map);
+ if (smp_use_sigp_detection)
+ return smp_rescan_cpus_sigp(avail);
+ else
+ return smp_rescan_cpus_sclp(avail);
}
/*
@@ -453,8 +612,6 @@ int __cpuinit start_secondary(void *cpuv
return 0;
}
-DEFINE_PER_CPU(struct s390_idle_data, s390_idle);
-
static void __init smp_create_idle(unsigned int cpu)
{
struct task_struct *p;
@@ -470,37 +627,16 @@ static void __init smp_create_idle(unsig
spin_lock_init(&(&per_cpu(s390_idle, cpu))->lock);
}
-static int cpu_stopped(int cpu)
-{
- __u32 status;
-
- /* Check for stopped state */
- if (signal_processor_ps(&status, 0, cpu, sigp_sense) ==
- sigp_status_stored) {
- if (status & 0x40)
- return 1;
- }
- return 0;
-}
-
/* Upping and downing of CPUs */
-
int __cpu_up(unsigned int cpu)
{
struct task_struct *idle;
struct _lowcore *cpu_lowcore;
struct stack_frame *sf;
sigp_ccode ccode;
- int curr_cpu;
- for (curr_cpu = 0; curr_cpu <= 65535; curr_cpu++) {
- __cpu_logical_map[cpu] = (__u16) curr_cpu;
- if (cpu_stopped(cpu))
- break;
- }
-
- if (!cpu_stopped(cpu))
- return -ENODEV;
+ if (smp_cpu_state[cpu] != CPU_STATE_CONFIGURED)
+ return -EIO;
ccode = signal_processor_p((__u32)(unsigned long)(lowcore_ptr[cpu]),
cpu, sigp_set_prefix);
@@ -543,21 +679,18 @@ static unsigned int __initdata possible_
void __init smp_setup_cpu_possible_map(void)
{
- unsigned int phy_cpus, pos_cpus, cpu;
-
- phy_cpus = smp_count_cpus();
- pos_cpus = min(phy_cpus + additional_cpus, (unsigned int) NR_CPUS);
+ unsigned int pos_cpus, cpu;
+ unsigned int configured_cpus, standby_cpus;
+ smp_count_cpus(&configured_cpus, &standby_cpus);
+ pos_cpus = min(configured_cpus + standby_cpus + additional_cpus,
+ (unsigned int) NR_CPUS);
if (possible_cpus)
pos_cpus = min(possible_cpus, (unsigned int) NR_CPUS);
-
for (cpu = 0; cpu < pos_cpus; cpu++)
cpu_set(cpu, cpu_possible_map);
-
- phy_cpus = min(phy_cpus, pos_cpus);
-
- for (cpu = 0; cpu < phy_cpus; cpu++)
- cpu_set(cpu, cpu_present_map);
+ cpu_present_map = cpumask_of_cpu(0);
+ smp_rescan_cpus();
}
#ifdef CONFIG_HOTPLUG_CPU
@@ -612,7 +745,7 @@ void __cpu_die(unsigned int cpu)
/* Wait until target cpu is down */
while (!smp_cpu_not_running(cpu))
cpu_relax();
- printk("Processor %d spun down\n", cpu);
+ printk(KERN_INFO "Processor %d spun down\n", cpu);
}
void cpu_die(void)
@@ -686,12 +819,12 @@ void __init smp_prepare_boot_cpu(void)
cpu_set(0, cpu_online_map);
S390_lowcore.percpu_offset = __per_cpu_offset[0];
current_set[0] = current;
+ smp_cpu_state[0] = CPU_STATE_CONFIGURED;
spin_lock_init(&(&__get_cpu_var(s390_idle))->lock);
}
void __init smp_cpus_done(unsigned int max_cpus)
{
- cpu_present_map = cpu_possible_map;
}
/*
@@ -705,7 +838,79 @@ int setup_profiling_timer(unsigned int m
return 0;
}
-static DEFINE_PER_CPU(struct cpu, cpu_devices);
+#ifdef CONFIG_HOTPLUG_CPU
+static ssize_t cpu_configure_show(struct sys_device *dev, char *buf)
+{
+ ssize_t count;
+
+ mutex_lock(&smp_cpu_state_mutex);
+ count = sprintf(buf, "%d\n", smp_cpu_state[dev->id]);
+ mutex_unlock(&smp_cpu_state_mutex);
+ return count;
+}
+
+static ssize_t cpu_configure_store(struct sys_device *dev, const char *buf,
+ size_t count)
+{
+ int cpu = dev->id;
+ int val, rc;
+ char delim;
+
+ if (sscanf(buf, "%d %c", &val, &delim) != 1)
+ return -EINVAL;
+ if (val != 0 && val != 1)
+ return -EINVAL;
+
+ mutex_lock(&smp_cpu_state_mutex);
+ lock_cpu_hotplug();
+ rc = -EBUSY;
+ if (cpu_online(cpu))
+ goto out;
+ rc = 0;
+ switch (val) {
+ case 0:
+ if (smp_cpu_state[cpu] == CPU_STATE_CONFIGURED) {
+ rc = sclp_cpu_deconfigure(__cpu_logical_map[cpu]);
+ if (!rc)
+ smp_cpu_state[cpu] = CPU_STATE_STANDBY;
+ }
+ break;
+ case 1:
+ if (smp_cpu_state[cpu] == CPU_STATE_STANDBY) {
+ rc = sclp_cpu_configure(__cpu_logical_map[cpu]);
+ if (!rc)
+ smp_cpu_state[cpu] = CPU_STATE_CONFIGURED;
+ }
+ break;
+ default:
+ break;
+ }
+out:
+ unlock_cpu_hotplug();
+ mutex_unlock(&smp_cpu_state_mutex);
+ return rc ? rc : count;
+}
+static SYSDEV_ATTR(configure, 0644, cpu_configure_show, cpu_configure_store);
+#endif /* CONFIG_HOTPLUG_CPU */
+
+static ssize_t show_cpu_address(struct sys_device *dev, char *buf)
+{
+ return sprintf(buf, "%d\n", __cpu_logical_map[dev->id]);
+}
+static SYSDEV_ATTR(address, 0444, show_cpu_address, NULL);
+
+
+static struct attribute *cpu_common_attrs[] = {
+#ifdef CONFIG_HOTPLUG_CPU
+ &attr_configure.attr,
+#endif
+ &attr_address.attr,
+ NULL,
+};
+
+static struct attribute_group cpu_common_attr_group = {
+ .attrs = cpu_common_attrs,
+};
static ssize_t show_capability(struct sys_device *dev, char *buf)
{
@@ -750,15 +955,15 @@ static ssize_t show_idle_time(struct sys
}
static SYSDEV_ATTR(idle_time_us, 0444, show_idle_time, NULL);
-static struct attribute *cpu_attrs[] = {
+static struct attribute *cpu_online_attrs[] = {
&attr_capability.attr,
&attr_idle_count.attr,
&attr_idle_time_us.attr,
NULL,
};
-static struct attribute_group cpu_attr_group = {
- .attrs = cpu_attrs,
+static struct attribute_group cpu_online_attr_group = {
+ .attrs = cpu_online_attrs,
};
static int __cpuinit smp_cpu_notify(struct notifier_block *self,
@@ -778,12 +983,12 @@ static int __cpuinit smp_cpu_notify(stru
idle->idle_time = 0;
idle->idle_count = 0;
spin_unlock_irq(&idle->lock);
- if (sysfs_create_group(&s->kobj, &cpu_attr_group))
+ if (sysfs_create_group(&s->kobj, &cpu_online_attr_group))
return NOTIFY_BAD;
break;
case CPU_DEAD:
case CPU_DEAD_FROZEN:
- sysfs_remove_group(&s->kobj, &cpu_attr_group);
+ sysfs_remove_group(&s->kobj, &cpu_online_attr_group);
break;
}
return NOTIFY_OK;
@@ -793,6 +998,62 @@ static struct notifier_block __cpuinitda
.notifier_call = smp_cpu_notify,
};
+static int smp_add_present_cpu(int cpu)
+{
+ struct cpu *c = &per_cpu(cpu_devices, cpu);
+ struct sys_device *s = &c->sysdev;
+ int rc;
+
+ c->hotpluggable = 1;
+ rc = register_cpu(c, cpu);
+ if (rc)
+ goto out;
+ rc = sysfs_create_group(&s->kobj, &cpu_common_attr_group);
+ if (rc)
+ goto out_cpu;
+ if (!cpu_online(cpu))
+ goto out;
+ rc = sysfs_create_group(&s->kobj, &cpu_online_attr_group);
+ if (!rc)
+ return 0;
+ sysfs_remove_group(&s->kobj, &cpu_common_attr_group);
+out_cpu:
+#ifdef CONFIG_HOTPLUG_CPU
+ unregister_cpu(c);
+#endif
+out:
+ return rc;
+}
+
+#ifdef CONFIG_HOTPLUG_CPU
+static ssize_t rescan_store(struct sys_device *dev, const char *buf,
+ size_t count)
+{
+ cpumask_t newcpus;
+ int cpu;
+ int rc;
+
+ mutex_lock(&smp_cpu_state_mutex);
+ lock_cpu_hotplug();
+ newcpus = cpu_present_map;
+ rc = smp_rescan_cpus();
+ if (rc)
+ goto out;
+ cpus_andnot(newcpus, cpu_present_map, newcpus);
+ for_each_cpu_mask(cpu, newcpus) {
+ rc = smp_add_present_cpu(cpu);
+ if (rc)
+ cpu_clear(cpu, cpu_present_map);
+ }
+ rc = 0;
+out:
+ unlock_cpu_hotplug();
+ mutex_unlock(&smp_cpu_state_mutex);
+ return rc ? rc : count;
+}
+static SYSDEV_ATTR(rescan, 0200, NULL, rescan_store);
+#endif /* CONFIG_HOTPLUG_CPU */
+
static int __init topology_init(void)
{
int cpu;
@@ -800,16 +1061,14 @@ static int __init topology_init(void)
register_cpu_notifier(&smp_cpu_nb);
- for_each_possible_cpu(cpu) {
- struct cpu *c = &per_cpu(cpu_devices, cpu);
- struct sys_device *s = &c->sysdev;
-
- c->hotpluggable = 1;
- register_cpu(c, cpu);
- if (!cpu_online(cpu))
- continue;
- s = &c->sysdev;
- rc = sysfs_create_group(&s->kobj, &cpu_attr_group);
+#ifdef CONFIG_HOTPLUG_CPU
+ rc = sysfs_create_file(&cpu_sysdev_class.kset.kobj,
+ &attr_rescan.attr);
+ if (rc)
+ return rc;
+#endif
+ for_each_present_cpu(cpu) {
+ rc = smp_add_present_cpu(cpu);
if (rc)
return rc;
}
Index: quilt-2.6/drivers/s390/char/Makefile
===================================================================
--- quilt-2.6.orig/drivers/s390/char/Makefile
+++ quilt-2.6/drivers/s390/char/Makefile
@@ -3,7 +3,7 @@
#
obj-y += ctrlchar.o keyboard.o defkeymap.o sclp.o sclp_rw.o sclp_quiesce.o \
- sclp_info.o sclp_config.o sclp_chp.o sclp_cpi_sys.o
+ sclp_cmd.o sclp_config.o sclp_chp.o sclp_cpi_sys.o
obj-$(CONFIG_TN3270) += raw3270.o
obj-$(CONFIG_TN3270_CONSOLE) += con3270.o
Index: quilt-2.6/drivers/s390/char/sclp_cmd.c
===================================================================
--- /dev/null
+++ quilt-2.6/drivers/s390/char/sclp_cmd.c
@@ -0,0 +1,319 @@
+/*
+ * drivers/s390/char/sclp_cmd.c
+ *
+ * Copyright IBM Corp. 2007
+ * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>
+ */
+
+#include <linux/completion.h>
+#include <linux/init.h>
+#include <linux/errno.h>
+#include <linux/slab.h>
+#include <linux/string.h>
+#include <asm/sclp.h>
+#include "sclp.h"
+
+#define TAG "sclp_cmd: "
+
+#define SCLP_CMDW_READ_SCP_INFO 0x00020001
+#define SCLP_CMDW_READ_SCP_INFO_FORCED 0x00120001
+
+struct read_info_sccb {
+ struct sccb_header header; /* 0-7 */
+ u16 rnmax; /* 8-9 */
+ u8 rnsize; /* 10 */
+ u8 _reserved0[24 - 11]; /* 11-15 */
+ u8 loadparm[8]; /* 24-31 */
+ u8 _reserved1[48 - 32]; /* 32-47 */
+ u64 facilities; /* 48-55 */
+ u8 _reserved2[84 - 56]; /* 56-83 */
+ u8 fac84; /* 84 */
+ u8 _reserved3[91 - 85]; /* 85-90 */
+ u8 flags; /* 91 */
+ u8 _reserved4[100 - 92]; /* 92-99 */
+ u32 rnsize2; /* 100-103 */
+ u64 rnmax2; /* 104-111 */
+ u8 _reserved5[4096 - 112]; /* 112-4095 */
+} __attribute__((packed, aligned(PAGE_SIZE)));
+
+static struct read_info_sccb __initdata early_read_info_sccb;
+static int __initdata early_read_info_sccb_valid;
+
+u64 sclp_facilities;
+static u8 sclp_fac84;
+
+static int __init sclp_cmd_sync_early(sclp_cmdw_t cmd, void *sccb)
+{
+ int rc;
+
+ __ctl_set_bit(0, 9);
+ rc = sclp_service_call(cmd, sccb);
+ if (rc)
+ goto out;
+ __load_psw_mask(PSW_BASE_BITS | PSW_MASK_EXT |
+ PSW_MASK_WAIT | PSW_DEFAULT_KEY);
+ local_irq_disable();
+out:
+ /* Contents of the sccb might have changed. */
+ barrier();
+ __ctl_clear_bit(0, 9);
+ return rc;
+}
+
+void __init sclp_read_info_early(void)
+{
+ int rc;
+ int i;
+ struct read_info_sccb *sccb;
+ sclp_cmdw_t commands[] = {SCLP_CMDW_READ_SCP_INFO_FORCED,
+ SCLP_CMDW_READ_SCP_INFO};
+
+ sccb = &early_read_info_sccb;
+ for (i = 0; i < ARRAY_SIZE(commands); i++) {
+ do {
+ memset(sccb, 0, sizeof(*sccb));
+ sccb->header.length = sizeof(*sccb);
+ sccb->header.control_mask[2] = 0x80;
+ rc = sclp_cmd_sync_early(commands[i], sccb);
+ } while (rc == -EBUSY);
+
+ if (rc)
+ break;
+ if (sccb->header.response_code == 0x10) {
+ early_read_info_sccb_valid = 1;
+ break;
+ }
+ if (sccb->header.response_code != 0x1f0)
+ break;
+ }
+}
+
+void __init sclp_facilities_detect(void)
+{
+ if (!early_read_info_sccb_valid)
+ return;
+ sclp_facilities = early_read_info_sccb.facilities;
+ sclp_fac84 = early_read_info_sccb.fac84;
+}
+
+unsigned long long __init sclp_memory_detect(void)
+{
+ unsigned long long memsize;
+ struct read_info_sccb *sccb;
+
+ if (!early_read_info_sccb_valid)
+ return 0;
+ sccb = &early_read_info_sccb;
+ if (sccb->rnsize)
+ memsize = sccb->rnsize << 20;
+ else
+ memsize = sccb->rnsize2 << 20;
+ if (sccb->rnmax)
+ memsize *= sccb->rnmax;
+ else
+ memsize *= sccb->rnmax2;
+ return memsize;
+}
+
+/*
+ * This function will be called after sclp_memory_detect(), which gets called
+ * early from early.c code. Therefore the sccb should have valid contents.
+ */
+void __init sclp_get_ipl_info(struct sclp_ipl_info *info)
+{
+ struct read_info_sccb *sccb;
+
+ if (!early_read_info_sccb_valid)
+ return;
+ sccb = &early_read_info_sccb;
+ info->is_valid = 1;
+ if (sccb->flags & 0x2)
+ info->has_dump = 1;
+ memcpy(&info->loadparm, &sccb->loadparm, LOADPARM_LEN);
+}
+
+static void sclp_sync_callback(struct sclp_req *req, void *data)
+{
+ struct completion *completion = data;
+
+ complete(completion);
+}
+
+static int do_sync_request(sclp_cmdw_t cmd, void *sccb)
+{
+ struct completion completion;
+ struct sclp_req *request;
+ int rc;
+
+ request = kzalloc(sizeof(*request), GFP_KERNEL);
+ if (!request)
+ return -ENOMEM;
+ request->command = cmd;
+ request->sccb = sccb;
+ request->status = SCLP_REQ_FILLED;
+ request->callback = sclp_sync_callback;
+ request->callback_data = &completion;
+ init_completion(&completion);
+
+ /* Perform sclp request. */
+ rc = sclp_add_request(request);
+ if (rc)
+ goto out;
+ wait_for_completion(&completion);
+
+ /* Check response. */
+ if (request->status != SCLP_REQ_DONE) {
+ printk(KERN_WARNING TAG "sync request failed "
+ "(cmd=0x%08x, status=0x%02x)\n", cmd, request->status);
+ rc = -EIO;
+ }
+out:
+ kfree(request);
+ return rc;
+}
+
+/*
+ * CPU configuration related functions.
+ */
+
+#define SCLP_CMDW_READ_CPU_INFO 0x00010001
+#define SCLP_CMDW_CONFIGURE_CPU 0x00110001
+#define SCLP_CMDW_DECONFIGURE_CPU 0x00100001
+
+struct read_cpu_info_sccb {
+ struct sccb_header header;
+ u16 nr_configured;
+ u16 offset_configured;
+ u16 nr_standby;
+ u16 offset_standby;
+ u8 reserved[4096 - 16];
+} __attribute__((packed, aligned(PAGE_SIZE)));
+
+static struct read_cpu_info_sccb __initdata early_read_cpu_info_sccb;
+static struct sclp_cpu_info __initdata sclp_cpu_info;
+
+static void sclp_fill_cpu_info(struct sclp_cpu_info *info,
+ struct read_cpu_info_sccb *sccb)
+{
+ char *page = (char *) sccb;
+
+ memset(info, 0, sizeof(*info));
+ info->configured = sccb->nr_configured;
+ info->standby = sccb->nr_standby;
+ info->combined = sccb->nr_configured + sccb->nr_standby;
+ info->has_cpu_type = sclp_fac84 & 0x1;
+ memcpy(&info->cpu, page + sccb->offset_configured,
+ info->combined * sizeof(struct sclp_cpu_entry));
+}
+
+void __init sclp_read_cpu_info_early(void)
+{
+ int rc;
+ struct read_cpu_info_sccb *sccb;
+
+ if (!SCLP_HAS_CPU_INFO)
+ return;
+
+ sccb = &early_read_cpu_info_sccb;
+ do {
+ memset(sccb, 0, sizeof(*sccb));
+ sccb->header.length = sizeof(*sccb);
+ rc = sclp_cmd_sync_early(SCLP_CMDW_READ_CPU_INFO, sccb);
+ } while (rc == -EBUSY);
+
+ if (rc)
+ return;
+ if (sccb->header.response_code != 0x10)
+ return;
+ sclp_fill_cpu_info(&sclp_cpu_info, sccb);
+}
+
+static int __init sclp_get_cpu_info_early(struct sclp_cpu_info *info)
+{
+ if (!SCLP_HAS_CPU_INFO)
+ return -EOPNOTSUPP;
+ *info = sclp_cpu_info;
+ return 0;
+}
+
+static int sclp_get_cpu_info_late(struct sclp_cpu_info *info)
+{
+ int rc;
+ struct read_cpu_info_sccb *sccb;
+
+ if (!SCLP_HAS_CPU_INFO)
+ return -EOPNOTSUPP;
+ sccb = (struct read_cpu_info_sccb *) __get_free_page(GFP_KERNEL
+ | GFP_DMA);
+ if (!sccb)
+ return -ENOMEM;
+ memset(sccb, 0, sizeof(*sccb));
+ sccb->header.length = sizeof(*sccb);
+ rc = do_sync_request(SCLP_CMDW_READ_CPU_INFO, sccb);
+ if (rc)
+ goto out;
+ if (sccb->header.response_code != 0x0010) {
+ printk(KERN_WARNING TAG "readcpuinfo failed "
+ "(response=0x%04x)\n", sccb->header.response_code);
+ rc = -EIO;
+ goto out;
+ }
+ sclp_fill_cpu_info(info, sccb);
+out:
+ free_page((unsigned long) sccb);
+ return rc;
+}
+
+int __init_refok sclp_get_cpu_info(struct sclp_cpu_info *info)
+{
+ if (slab_is_available())
+ return sclp_get_cpu_info_late(info);
+ return sclp_get_cpu_info_early(info);
+}
+
+struct cpu_configure_sccb {
+ struct sccb_header header;
+} __attribute__((packed, aligned(8)));
+
+static int do_cpu_configure(sclp_cmdw_t cmd)
+{
+ struct cpu_configure_sccb *sccb;
+ int rc;
+
+ if (!SCLP_HAS_CPU_RECONFIG)
+ return -EOPNOTSUPP;
+ /*
+ * This is not going to cross a page boundary since we force
+ * kmalloc to have a minimum alignment of 8 bytes on s390.
+ */
+ sccb = kzalloc(sizeof(*sccb), GFP_KERNEL | GFP_DMA);
+ if (!sccb)
+ return -ENOMEM;
+ sccb->header.length = sizeof(*sccb);
+ rc = do_sync_request(cmd, sccb);
+ if (rc)
+ goto out;
+ switch (sccb->header.response_code) {
+ case 0x0020:
+ case 0x0120:
+ break;
+ default:
+ printk(KERN_WARNING TAG "configure cpu failed (cmd=0x%08x, "
+ "response=0x%04x)\n", cmd, sccb->header.response_code);
+ rc = -EIO;
+ break;
+ }
+out:
+ kfree(sccb);
+ return rc;
+}
+
+int sclp_cpu_configure(u8 cpu)
+{
+ return do_cpu_configure(SCLP_CMDW_CONFIGURE_CPU | cpu << 8);
+}
+
+int sclp_cpu_deconfigure(u8 cpu)
+{
+ return do_cpu_configure(SCLP_CMDW_DECONFIGURE_CPU | cpu << 8);
+}
Index: quilt-2.6/drivers/s390/char/sclp.h
===================================================================
--- quilt-2.6.orig/drivers/s390/char/sclp.h
+++ quilt-2.6/drivers/s390/char/sclp.h
@@ -56,8 +56,6 @@ typedef unsigned int sclp_cmdw_t;
#define SCLP_CMDW_READ_EVENT_DATA 0x00770005
#define SCLP_CMDW_WRITE_EVENT_DATA 0x00760005
#define SCLP_CMDW_WRITE_EVENT_MASK 0x00780005
-#define SCLP_CMDW_READ_SCP_INFO 0x00020001
-#define SCLP_CMDW_READ_SCP_INFO_FORCED 0x00120001
#define GDS_ID_MDSMU 0x1310
#define GDS_ID_MDSROUTEINFO 0x1311
@@ -83,6 +81,8 @@ extern u64 sclp_facilities;
#define SCLP_HAS_CHP_INFO (sclp_facilities & 0x8000000000000000ULL)
#define SCLP_HAS_CHP_RECONFIG (sclp_facilities & 0x2000000000000000ULL)
+#define SCLP_HAS_CPU_INFO (sclp_facilities & 0x0800000000000000ULL)
+#define SCLP_HAS_CPU_RECONFIG (sclp_facilities & 0x0400000000000000ULL)
struct gds_subvector {
u8 length;
Index: quilt-2.6/drivers/s390/char/sclp_info.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/sclp_info.c
+++ /dev/null
@@ -1,116 +0,0 @@
-/*
- * drivers/s390/char/sclp_info.c
- *
- * Copyright IBM Corp. 2007
- * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>
- */
-
-#include <linux/init.h>
-#include <linux/errno.h>
-#include <linux/string.h>
-#include <asm/sclp.h>
-#include "sclp.h"
-
-struct sclp_readinfo_sccb {
- struct sccb_header header; /* 0-7 */
- u16 rnmax; /* 8-9 */
- u8 rnsize; /* 10 */
- u8 _reserved0[24 - 11]; /* 11-23 */
- u8 loadparm[8]; /* 24-31 */
- u8 _reserved1[48 - 32]; /* 32-47 */
- u64 facilities; /* 48-55 */
- u8 _reserved2[91 - 56]; /* 56-90 */
- u8 flags; /* 91 */
- u8 _reserved3[100 - 92]; /* 92-99 */
- u32 rnsize2; /* 100-103 */
- u64 rnmax2; /* 104-111 */
- u8 _reserved4[4096 - 112]; /* 112-4095 */
-} __attribute__((packed, aligned(4096)));
-
-static struct sclp_readinfo_sccb __initdata early_readinfo_sccb;
-static int __initdata early_readinfo_sccb_valid;
-
-u64 sclp_facilities;
-
-void __init sclp_readinfo_early(void)
-{
- int ret;
- int i;
- struct sclp_readinfo_sccb *sccb;
- sclp_cmdw_t commands[] = {SCLP_CMDW_READ_SCP_INFO_FORCED,
- SCLP_CMDW_READ_SCP_INFO};
-
- /* Enable service signal subclass mask. */
- __ctl_set_bit(0, 9);
- sccb = &early_readinfo_sccb;
- for (i = 0; i < ARRAY_SIZE(commands); i++) {
- do {
- memset(sccb, 0, sizeof(*sccb));
- sccb->header.length = sizeof(*sccb);
- sccb->header.control_mask[2] = 0x80;
- ret = sclp_service_call(commands[i], sccb);
- } while (ret == -EBUSY);
-
- if (ret)
- break;
- __load_psw_mask(PSW_BASE_BITS | PSW_MASK_EXT |
- PSW_MASK_WAIT | PSW_DEFAULT_KEY);
- local_irq_disable();
- /*
- * Contents of the sccb might have changed
- * therefore a barrier is needed.
- */
- barrier();
- if (sccb->header.response_code == 0x10) {
- early_readinfo_sccb_valid = 1;
- break;
- }
- if (sccb->header.response_code != 0x1f0)
- break;
- }
- /* Disable service signal subclass mask again. */
- __ctl_clear_bit(0, 9);
-}
-
-void __init sclp_facilities_detect(void)
-{
- if (!early_readinfo_sccb_valid)
- return;
- sclp_facilities = early_readinfo_sccb.facilities;
-}
-
-unsigned long long __init sclp_memory_detect(void)
-{
- unsigned long long memsize;
- struct sclp_readinfo_sccb *sccb;
-
- if (!early_readinfo_sccb_valid)
- return 0;
- sccb = &early_readinfo_sccb;
- if (sccb->rnsize)
- memsize = sccb->rnsize << 20;
- else
- memsize = sccb->rnsize2 << 20;
- if (sccb->rnmax)
- memsize *= sccb->rnmax;
- else
- memsize *= sccb->rnmax2;
- return memsize;
-}
-
-/*
- * This function will be called after sclp_memory_detect(), which gets called
- * early from early.c code. Therefore the sccb should have valid contents.
- */
-void __init sclp_get_ipl_info(struct sclp_ipl_info *info)
-{
- struct sclp_readinfo_sccb *sccb;
-
- if (!early_readinfo_sccb_valid)
- return;
- sccb = &early_readinfo_sccb;
- info->is_valid = 1;
- if (sccb->flags & 0x2)
- info->has_dump = 1;
- memcpy(&info->loadparm, &sccb->loadparm, LOADPARM_LEN);
-}
Index: quilt-2.6/include/asm-s390/sclp.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/sclp.h
+++ quilt-2.6/include/asm-s390/sclp.h
@@ -27,7 +27,26 @@ struct sclp_ipl_info {
char loadparm[LOADPARM_LEN];
};
-void sclp_readinfo_early(void);
+struct sclp_cpu_entry {
+ u8 address;
+ u8 reserved0[13];
+ u8 type;
+ u8 reserved1;
+} __attribute__((packed));
+
+struct sclp_cpu_info {
+ unsigned int configured;
+ unsigned int standby;
+ unsigned int combined;
+ int has_cpu_type;
+ struct sclp_cpu_entry cpu[255];
+};
+
+int sclp_get_cpu_info(struct sclp_cpu_info *info);
+int sclp_cpu_configure(u8 cpu);
+int sclp_cpu_deconfigure(u8 cpu);
+void sclp_read_info_early(void);
+void sclp_read_cpu_info_early(void);
void sclp_facilities_detect(void);
unsigned long long sclp_memory_detect(void);
int sclp_sdias_blk_count(void);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 18/47] sclp: convert channel path configure code to use sync interface.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (16 preceding siblings ...)
2007-12-20 15:19 ` [patch 17/47] Standby cpu activation/deactivation Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 19/47] Optimize reference bit handling Martin Schwidefsky
` (28 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 117-sclp-chpid.diff --]
[-- Type: text/plain, Size: 10595 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/char/Makefile | 2
drivers/s390/char/sclp_chp.c | 200 -------------------------------------------
drivers/s390/char/sclp_cmd.c | 123 ++++++++++++++++++++++++++
3 files changed, 123 insertions(+), 202 deletions(-)
Index: quilt-2.6/drivers/s390/char/Makefile
===================================================================
--- quilt-2.6.orig/drivers/s390/char/Makefile
+++ quilt-2.6/drivers/s390/char/Makefile
@@ -3,7 +3,7 @@
#
obj-y += ctrlchar.o keyboard.o defkeymap.o sclp.o sclp_rw.o sclp_quiesce.o \
- sclp_cmd.o sclp_config.o sclp_chp.o sclp_cpi_sys.o
+ sclp_cmd.o sclp_config.o sclp_cpi_sys.o
obj-$(CONFIG_TN3270) += raw3270.o
obj-$(CONFIG_TN3270_CONSOLE) += con3270.o
Index: quilt-2.6/drivers/s390/char/sclp_chp.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/sclp_chp.c
+++ /dev/null
@@ -1,200 +0,0 @@
-/*
- * drivers/s390/char/sclp_chp.c
- *
- * Copyright IBM Corp. 2007
- * Author(s): Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
- */
-
-#include <linux/types.h>
-#include <linux/gfp.h>
-#include <linux/errno.h>
-#include <linux/completion.h>
-#include <asm/sclp.h>
-#include <asm/chpid.h>
-
-#include "sclp.h"
-
-#define TAG "sclp_chp: "
-
-#define SCLP_CMDW_CONFIGURE_CHANNEL_PATH 0x000f0001
-#define SCLP_CMDW_DECONFIGURE_CHANNEL_PATH 0x000e0001
-#define SCLP_CMDW_READ_CHANNEL_PATH_INFORMATION 0x00030001
-
-static inline sclp_cmdw_t get_configure_cmdw(struct chp_id chpid)
-{
- return SCLP_CMDW_CONFIGURE_CHANNEL_PATH | chpid.id << 8;
-}
-
-static inline sclp_cmdw_t get_deconfigure_cmdw(struct chp_id chpid)
-{
- return SCLP_CMDW_DECONFIGURE_CHANNEL_PATH | chpid.id << 8;
-}
-
-static void chp_callback(struct sclp_req *req, void *data)
-{
- struct completion *completion = data;
-
- complete(completion);
-}
-
-struct chp_cfg_sccb {
- struct sccb_header header;
- u8 ccm;
- u8 reserved[6];
- u8 cssid;
-} __attribute__((packed));
-
-struct chp_cfg_data {
- struct chp_cfg_sccb sccb;
- struct sclp_req req;
- struct completion completion;
-} __attribute__((packed));
-
-static int do_configure(sclp_cmdw_t cmd)
-{
- struct chp_cfg_data *data;
- int rc;
-
- if (!SCLP_HAS_CHP_RECONFIG)
- return -EOPNOTSUPP;
- /* Prepare sccb. */
- data = (struct chp_cfg_data *) get_zeroed_page(GFP_KERNEL | GFP_DMA);
- if (!data)
- return -ENOMEM;
- data->sccb.header.length = sizeof(struct chp_cfg_sccb);
- data->req.command = cmd;
- data->req.sccb = &(data->sccb);
- data->req.status = SCLP_REQ_FILLED;
- data->req.callback = chp_callback;
- data->req.callback_data = &(data->completion);
- init_completion(&data->completion);
-
- /* Perform sclp request. */
- rc = sclp_add_request(&(data->req));
- if (rc)
- goto out;
- wait_for_completion(&data->completion);
-
- /* Check response .*/
- if (data->req.status != SCLP_REQ_DONE) {
- printk(KERN_WARNING TAG "configure channel-path request failed "
- "(status=0x%02x)\n", data->req.status);
- rc = -EIO;
- goto out;
- }
- switch (data->sccb.header.response_code) {
- case 0x0020:
- case 0x0120:
- case 0x0440:
- case 0x0450:
- break;
- default:
- printk(KERN_WARNING TAG "configure channel-path failed "
- "(cmd=0x%08x, response=0x%04x)\n", cmd,
- data->sccb.header.response_code);
- rc = -EIO;
- break;
- }
-out:
- free_page((unsigned long) data);
-
- return rc;
-}
-
-/**
- * sclp_chp_configure - perform configure channel-path sclp command
- * @chpid: channel-path ID
- *
- * Perform configure channel-path command sclp command for specified chpid.
- * Return 0 after command successfully finished, non-zero otherwise.
- */
-int sclp_chp_configure(struct chp_id chpid)
-{
- return do_configure(get_configure_cmdw(chpid));
-}
-
-/**
- * sclp_chp_deconfigure - perform deconfigure channel-path sclp command
- * @chpid: channel-path ID
- *
- * Perform deconfigure channel-path command sclp command for specified chpid
- * and wait for completion. On success return 0. Return non-zero otherwise.
- */
-int sclp_chp_deconfigure(struct chp_id chpid)
-{
- return do_configure(get_deconfigure_cmdw(chpid));
-}
-
-struct chp_info_sccb {
- struct sccb_header header;
- u8 recognized[SCLP_CHP_INFO_MASK_SIZE];
- u8 standby[SCLP_CHP_INFO_MASK_SIZE];
- u8 configured[SCLP_CHP_INFO_MASK_SIZE];
- u8 ccm;
- u8 reserved[6];
- u8 cssid;
-} __attribute__((packed));
-
-struct chp_info_data {
- struct chp_info_sccb sccb;
- struct sclp_req req;
- struct completion completion;
-} __attribute__((packed));
-
-/**
- * sclp_chp_read_info - perform read channel-path information sclp command
- * @info: resulting channel-path information data
- *
- * Perform read channel-path information sclp command and wait for completion.
- * On success, store channel-path information in @info and return 0. Return
- * non-zero otherwise.
- */
-int sclp_chp_read_info(struct sclp_chp_info *info)
-{
- struct chp_info_data *data;
- int rc;
-
- if (!SCLP_HAS_CHP_INFO)
- return -EOPNOTSUPP;
- /* Prepare sccb. */
- data = (struct chp_info_data *) get_zeroed_page(GFP_KERNEL | GFP_DMA);
- if (!data)
- return -ENOMEM;
- data->sccb.header.length = sizeof(struct chp_info_sccb);
- data->req.command = SCLP_CMDW_READ_CHANNEL_PATH_INFORMATION;
- data->req.sccb = &(data->sccb);
- data->req.status = SCLP_REQ_FILLED;
- data->req.callback = chp_callback;
- data->req.callback_data = &(data->completion);
- init_completion(&data->completion);
-
- /* Perform sclp request. */
- rc = sclp_add_request(&(data->req));
- if (rc)
- goto out;
- wait_for_completion(&data->completion);
-
- /* Check response .*/
- if (data->req.status != SCLP_REQ_DONE) {
- printk(KERN_WARNING TAG "read channel-path info request failed "
- "(status=0x%02x)\n", data->req.status);
- rc = -EIO;
- goto out;
- }
- if (data->sccb.header.response_code != 0x0010) {
- printk(KERN_WARNING TAG "read channel-path info failed "
- "(response=0x%04x)\n", data->sccb.header.response_code);
- rc = -EIO;
- goto out;
- }
- memcpy(info->recognized, data->sccb.recognized,
- SCLP_CHP_INFO_MASK_SIZE);
- memcpy(info->standby, data->sccb.standby,
- SCLP_CHP_INFO_MASK_SIZE);
- memcpy(info->configured, data->sccb.configured,
- SCLP_CHP_INFO_MASK_SIZE);
-out:
- free_page((unsigned long) data);
-
- return rc;
-}
Index: quilt-2.6/drivers/s390/char/sclp_cmd.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/sclp_cmd.c
+++ quilt-2.6/drivers/s390/char/sclp_cmd.c
@@ -2,7 +2,8 @@
* drivers/s390/char/sclp_cmd.c
*
* Copyright IBM Corp. 2007
- * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>
+ * Author(s): Heiko Carstens <heiko.carstens@de.ibm.com>,
+ * Peter Oberparleiter <peter.oberparleiter@de.ibm.com>
*/
#include <linux/completion.h>
@@ -10,6 +11,7 @@
#include <linux/errno.h>
#include <linux/slab.h>
#include <linux/string.h>
+#include <asm/chpid.h>
#include <asm/sclp.h>
#include "sclp.h"
@@ -317,3 +319,122 @@ int sclp_cpu_deconfigure(u8 cpu)
{
return do_cpu_configure(SCLP_CMDW_DECONFIGURE_CPU | cpu << 8);
}
+
+/*
+ * Channel path configuration related functions.
+ */
+
+#define SCLP_CMDW_CONFIGURE_CHPATH 0x000f0001
+#define SCLP_CMDW_DECONFIGURE_CHPATH 0x000e0001
+#define SCLP_CMDW_READ_CHPATH_INFORMATION 0x00030001
+
+struct chp_cfg_sccb {
+ struct sccb_header header;
+ u8 ccm;
+ u8 reserved[6];
+ u8 cssid;
+} __attribute__((packed));
+
+static int do_chp_configure(sclp_cmdw_t cmd)
+{
+ struct chp_cfg_sccb *sccb;
+ int rc;
+
+ if (!SCLP_HAS_CHP_RECONFIG)
+ return -EOPNOTSUPP;
+ /* Prepare sccb. */
+ sccb = (struct chp_cfg_sccb *) get_zeroed_page(GFP_KERNEL | GFP_DMA);
+ if (!sccb)
+ return -ENOMEM;
+ sccb->header.length = sizeof(*sccb);
+ rc = do_sync_request(cmd, sccb);
+ if (rc)
+ goto out;
+ switch (sccb->header.response_code) {
+ case 0x0020:
+ case 0x0120:
+ case 0x0440:
+ case 0x0450:
+ break;
+ default:
+ printk(KERN_WARNING TAG "configure channel-path failed "
+ "(cmd=0x%08x, response=0x%04x)\n", cmd,
+ sccb->header.response_code);
+ rc = -EIO;
+ break;
+ }
+out:
+ free_page((unsigned long) sccb);
+ return rc;
+}
+
+/**
+ * sclp_chp_configure - perform configure channel-path sclp command
+ * @chpid: channel-path ID
+ *
+ * Perform configure channel-path command sclp command for specified chpid.
+ * Return 0 after command successfully finished, non-zero otherwise.
+ */
+int sclp_chp_configure(struct chp_id chpid)
+{
+ return do_chp_configure(SCLP_CMDW_CONFIGURE_CHPATH | chpid.id << 8);
+}
+
+/**
+ * sclp_chp_deconfigure - perform deconfigure channel-path sclp command
+ * @chpid: channel-path ID
+ *
+ * Perform deconfigure channel-path command sclp command for specified chpid
+ * and wait for completion. On success return 0. Return non-zero otherwise.
+ */
+int sclp_chp_deconfigure(struct chp_id chpid)
+{
+ return do_chp_configure(SCLP_CMDW_DECONFIGURE_CHPATH | chpid.id << 8);
+}
+
+struct chp_info_sccb {
+ struct sccb_header header;
+ u8 recognized[SCLP_CHP_INFO_MASK_SIZE];
+ u8 standby[SCLP_CHP_INFO_MASK_SIZE];
+ u8 configured[SCLP_CHP_INFO_MASK_SIZE];
+ u8 ccm;
+ u8 reserved[6];
+ u8 cssid;
+} __attribute__((packed));
+
+/**
+ * sclp_chp_read_info - perform read channel-path information sclp command
+ * @info: resulting channel-path information data
+ *
+ * Perform read channel-path information sclp command and wait for completion.
+ * On success, store channel-path information in @info and return 0. Return
+ * non-zero otherwise.
+ */
+int sclp_chp_read_info(struct sclp_chp_info *info)
+{
+ struct chp_info_sccb *sccb;
+ int rc;
+
+ if (!SCLP_HAS_CHP_INFO)
+ return -EOPNOTSUPP;
+ /* Prepare sccb. */
+ sccb = (struct chp_info_sccb *) get_zeroed_page(GFP_KERNEL | GFP_DMA);
+ if (!sccb)
+ return -ENOMEM;
+ sccb->header.length = sizeof(*sccb);
+ rc = do_sync_request(SCLP_CMDW_READ_CHPATH_INFORMATION, sccb);
+ if (rc)
+ goto out;
+ if (sccb->header.response_code != 0x0010) {
+ printk(KERN_WARNING TAG "read channel-path info failed "
+ "(response=0x%04x)\n", sccb->header.response_code);
+ rc = -EIO;
+ goto out;
+ }
+ memcpy(info->recognized, sccb->recognized, SCLP_CHP_INFO_MASK_SIZE);
+ memcpy(info->standby, sccb->standby, SCLP_CHP_INFO_MASK_SIZE);
+ memcpy(info->configured, sccb->configured, SCLP_CHP_INFO_MASK_SIZE);
+out:
+ free_page((unsigned long) sccb);
+ return rc;
+}
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 19/47] Optimize reference bit handling.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (17 preceding siblings ...)
2007-12-20 15:19 ` [patch 18/47] sclp: convert channel path configure code to use sync interface Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 20/47] Fix tlb flushing with idte Martin Schwidefsky
` (27 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Martin Schwidefsky
[-- Attachment #1: 118-mm-referenced.diff --]
[-- Type: text/plain, Size: 1878 bytes --]
From: Martin Schwidefsky <schwidefsky@de.ibm.com>
page_referenced always tests and clears the reference bit in the
storage key, even if the page is not mapped. For a page that is
only accessed with sys_read this has a negative side effect.
A page that is only read once makes two trips over the inactive
list before it is removed from the page cache. When the page is
added to the page cache it is added to the start of the inactive
list. After it went through the inactive list the reference bit
is checked with a call to page_referenced which will find the
referenced bit in the storage key set because the copy_to_user
operation will set the bit. This causes the page to be added to
the start of the inactive list again. This wastes cpu cycles in
vmscan.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
mm/rmap.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
Index: quilt-2.6/mm/rmap.c
===================================================================
--- quilt-2.6.orig/mm/rmap.c
+++ quilt-2.6/mm/rmap.c
@@ -391,9 +391,6 @@ int page_referenced(struct page *page, i
{
int referenced = 0;
- if (page_test_and_clear_young(page))
- referenced++;
-
if (TestClearPageReferenced(page))
referenced++;
@@ -409,6 +406,8 @@ int page_referenced(struct page *page, i
referenced += page_referenced_file(page);
unlock_page(page);
}
+ if (page_test_and_clear_young(page))
+ referenced++;
}
return referenced;
}
@@ -640,6 +639,8 @@ void page_remove_rmap(struct page *page,
* Leaving it set also helps swapoff to reinstate ptes
* faster for those pages still in swapcache.
*/
+ if (page_test_and_clear_young(page))
+ SetPageReferenced(page);
if (page_test_dirty(page)) {
page_clear_dirty(page);
set_page_dirty(page);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 20/47] Fix tlb flushing with idte.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (18 preceding siblings ...)
2007-12-20 15:19 ` [patch 19/47] Optimize reference bit handling Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 21/47] Change vmalloc defintions Martin Schwidefsky
` (26 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Martin Schwidefsky
[-- Attachment #1: 119-mm-idte.diff --]
[-- Type: text/plain, Size: 3966 bytes --]
From: Martin Schwidefsky <schwidefsky@de.ibm.com>
The clear-by-asce operation of the idte instruction gets an asce
(address-space-control-element) as argument to specify which TLBs
need to get flushed. The current code passes a plain pointer to
the start of the pgd without the additional bits which would make
the pointer an asce. The current machines don't mind the difference
but a future model might want to use the designation type control
bits in the asce as a filter for the TLBs to flush.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/head64.S | 2 +-
include/asm-s390/mmu_context.h | 27 ++++++++++++---------------
include/asm-s390/tlbflush.h | 12 ++++++------
3 files changed, 19 insertions(+), 22 deletions(-)
Index: quilt-2.6/arch/s390/kernel/head64.S
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/head64.S
+++ quilt-2.6/arch/s390/kernel/head64.S
@@ -157,7 +157,7 @@ startup_continue:
.long 0xb2b10000 # store facility list
tm 0xc8,0x08 # check bit for clearing-by-ASCE
bno 0f-.LPG1(%r13)
- lhi %r1,2094
+ lhi %r1,2048
lhi %r2,0
.long 0xb98e2001
oi 7(%r12),0x80 # set IDTE flag
Index: quilt-2.6/include/asm-s390/mmu_context.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/mmu_context.h
+++ quilt-2.6/include/asm-s390/mmu_context.h
@@ -12,10 +12,15 @@
#include <asm/pgalloc.h>
#include <asm-generic/mm_hooks.h>
-/*
- * get a new mmu context.. S390 don't know about contexts.
- */
-#define init_new_context(tsk,mm) 0
+static inline int init_new_context(struct task_struct *tsk,
+ struct mm_struct *mm)
+{
+ mm->context = _ASCE_TABLE_LENGTH | _ASCE_USER_BITS;
+#ifdef CONFIG_64BIT
+ mm->context |= _ASCE_TYPE_REGION3;
+#endif
+ return 0;
+}
#define destroy_context(mm) do { } while (0)
@@ -27,19 +32,11 @@
static inline void update_mm(struct mm_struct *mm, struct task_struct *tsk)
{
- pgd_t *pgd = mm->pgd;
- unsigned long asce_bits;
-
- /* Calculate asce bits from the first pgd table entry. */
- asce_bits = _ASCE_TABLE_LENGTH | _ASCE_USER_BITS;
-#ifdef CONFIG_64BIT
- asce_bits |= _ASCE_TYPE_REGION3;
-#endif
- S390_lowcore.user_asce = asce_bits | __pa(pgd);
+ S390_lowcore.user_asce = mm->context | __pa(mm->pgd);
if (switch_amode) {
/* Load primary space page table origin. */
- pgd_t *shadow_pgd = get_shadow_table(pgd) ? : pgd;
- S390_lowcore.user_exec_asce = asce_bits | __pa(shadow_pgd);
+ pgd_t *shadow_pgd = get_shadow_table(mm->pgd) ? : mm->pgd;
+ S390_lowcore.user_exec_asce = mm->context | __pa(shadow_pgd);
asm volatile(LCTL_OPCODE" 1,1,%0\n"
: : "m" (S390_lowcore.user_exec_asce) );
} else
Index: quilt-2.6/include/asm-s390/tlbflush.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/tlbflush.h
+++ quilt-2.6/include/asm-s390/tlbflush.h
@@ -42,11 +42,11 @@ static inline void __tlb_flush_global(vo
/*
* Flush all tlb entries of a page table on all cpus.
*/
-static inline void __tlb_flush_idte(pgd_t *pgd)
+static inline void __tlb_flush_idte(unsigned long asce)
{
asm volatile(
" .insn rrf,0xb98e0000,0,%0,%1,0"
- : : "a" (2048), "a" (__pa(pgd) & PAGE_MASK) : "cc" );
+ : : "a" (2048), "a" (asce) : "cc" );
}
static inline void __tlb_flush_mm(struct mm_struct * mm)
@@ -61,11 +61,11 @@ static inline void __tlb_flush_mm(struct
* only ran on the local cpu.
*/
if (MACHINE_HAS_IDTE) {
- pgd_t *shadow_pgd = get_shadow_table(mm->pgd);
+ pgd_t *shadow = get_shadow_table(mm->pgd);
- if (shadow_pgd)
- __tlb_flush_idte(shadow_pgd);
- __tlb_flush_idte(mm->pgd);
+ if (shadow)
+ __tlb_flush_idte((unsigned long) shadow | mm->context);
+ __tlb_flush_idte((unsigned long) mm->pgd | mm->context);
return;
}
preempt_disable();
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 21/47] Change vmalloc defintions
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (19 preceding siblings ...)
2007-12-20 15:19 ` [patch 20/47] Fix tlb flushing with idte Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 22/47] Print kernel version in dump_stack() and show_regs() Martin Schwidefsky
` (25 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Christian Borntraeger, Martin Schwidefsky
[-- Attachment #1: 120-vmalloc.diff --]
[-- Type: text/plain, Size: 6217 bytes --]
From: Christian Borntraeger <borntraeger@de.ibm.com>
Currently the vmalloc area starts at a dynamic address depending on
the memory size. There was also an 8MB security hole after the
physical memory to catch out-of-bounds accesses.
We can simplify the code by putting the vmalloc area explicitely at
the top of the kernel mapping and setting the vmalloc size to a fixed
value of 128MB/128GB for 31bit/64bit systems. Part of the vmalloc
area will be used for the vmem_map. This leaves an area of 96MB/1GB
for normal vmalloc allocations.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/setup.c | 6 ++---
arch/s390/mm/vmem.c | 20 ++++++-------------
include/asm-s390/pgtable.h | 46 +++++++++++++++------------------------------
3 files changed, 26 insertions(+), 46 deletions(-)
Index: quilt-2.6/arch/s390/kernel/setup.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/setup.c
+++ quilt-2.6/arch/s390/kernel/setup.c
@@ -617,7 +617,7 @@ EXPORT_SYMBOL_GPL(real_memory_size);
static void __init setup_memory_end(void)
{
unsigned long memory_size;
- unsigned long max_mem, max_phys;
+ unsigned long max_mem;
int i;
#if defined(CONFIG_ZFCPDUMP) || defined(CONFIG_ZFCPDUMP_MODULE)
@@ -625,10 +625,10 @@ static void __init setup_memory_end(void
memory_end = ZFCPDUMP_HSA_SIZE;
#endif
memory_size = 0;
- max_phys = VMALLOC_END_INIT - VMALLOC_MIN_SIZE;
memory_end &= PAGE_MASK;
- max_mem = memory_end ? min(max_phys, memory_end) : max_phys;
+ max_mem = memory_end ? min(VMALLOC_START, memory_end) : VMALLOC_START;
+ memory_end = min(max_mem, memory_end);
for (i = 0; i < MEMORY_CHUNKS; i++) {
struct mem_chunk *chunk = &memory_chunk[i];
Index: quilt-2.6/arch/s390/mm/vmem.c
===================================================================
--- quilt-2.6.orig/arch/s390/mm/vmem.c
+++ quilt-2.6/arch/s390/mm/vmem.c
@@ -15,10 +15,6 @@
#include <asm/setup.h>
#include <asm/tlbflush.h>
-unsigned long vmalloc_end;
-EXPORT_SYMBOL(vmalloc_end);
-
-static struct page *vmem_map;
static DEFINE_MUTEX(vmem_mutex);
struct memory_segment {
@@ -188,8 +184,8 @@ static int vmem_add_mem_map(unsigned lon
pte_t pte;
int ret = -ENOMEM;
- map_start = vmem_map + PFN_DOWN(start);
- map_end = vmem_map + PFN_DOWN(start + size);
+ map_start = VMEM_MAP + PFN_DOWN(start);
+ map_end = VMEM_MAP + PFN_DOWN(start + size);
start_addr = (unsigned long) map_start & PAGE_MASK;
end_addr = PFN_ALIGN((unsigned long) map_end);
@@ -254,7 +250,7 @@ static int insert_memory_segment(struct
{
struct memory_segment *tmp;
- if (PFN_DOWN(seg->start + seg->size) > max_pfn ||
+ if (seg->start + seg->size >= VMALLOC_START ||
seg->start + seg->size < seg->start)
return -ERANGE;
@@ -357,17 +353,15 @@ out:
/*
* map whole physical memory to virtual memory (identity mapping)
+ * we reserve enough space in the vmalloc area for vmemmap to hotplug
+ * additional memory segments.
*/
void __init vmem_map_init(void)
{
- unsigned long map_size;
int i;
- map_size = ALIGN(max_low_pfn, MAX_ORDER_NR_PAGES) * sizeof(struct page);
- vmalloc_end = PFN_ALIGN(VMALLOC_END_INIT) - PFN_ALIGN(map_size);
- vmem_map = (struct page *) vmalloc_end;
- NODE_DATA(0)->node_mem_map = vmem_map;
-
+ BUILD_BUG_ON((unsigned long)VMEM_MAP + VMEM_MAP_SIZE > VMEM_MAP_MAX);
+ NODE_DATA(0)->node_mem_map = VMEM_MAP;
for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++)
vmem_add_mem(memory_chunk[i].addr, memory_chunk[i].size);
}
Index: quilt-2.6/include/asm-s390/pgtable.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/pgtable.h
+++ quilt-2.6/include/asm-s390/pgtable.h
@@ -104,41 +104,27 @@ extern char empty_zero_page[PAGE_SIZE];
#ifndef __ASSEMBLY__
/*
- * Just any arbitrary offset to the start of the vmalloc VM area: the
- * current 8MB value just means that there will be a 8MB "hole" after the
- * physical memory until the kernel virtual memory starts. That means that
- * any out-of-bounds memory accesses will hopefully be caught.
- * The vmalloc() routines leaves a hole of 4kB between each vmalloced
- * area for the same reason. ;)
- * vmalloc area starts at 4GB to prevent syscall table entry exchanging
- * from modules.
- */
-extern unsigned long vmalloc_end;
-
-#ifdef CONFIG_64BIT
-#define VMALLOC_ADDR (max(0x100000000UL, (unsigned long) high_memory))
-#else
-#define VMALLOC_ADDR ((unsigned long) high_memory)
-#endif
-#define VMALLOC_OFFSET (8*1024*1024)
-#define VMALLOC_START ((VMALLOC_ADDR + VMALLOC_OFFSET) & ~(VMALLOC_OFFSET-1))
-#define VMALLOC_END vmalloc_end
-
-/*
- * We need some free virtual space to be able to do vmalloc.
- * VMALLOC_MIN_SIZE defines the minimum size of the vmalloc
- * area. On a machine with 2GB memory we make sure that we
- * have at least 128MB free space for vmalloc. On a machine
- * with 4TB we make sure we have at least 128GB.
+ * The vmalloc area will always be on the topmost area of the kernel
+ * mapping. We reserve 96MB (31bit) / 1GB (64bit) for vmalloc,
+ * which should be enough for any sane case.
+ * By putting vmalloc at the top, we maximise the gap between physical
+ * memory and vmalloc to catch misplaced memory accesses. As a side
+ * effect, this also makes sure that 64 bit module code cannot be used
+ * as system call address.
*/
#ifndef __s390x__
-#define VMALLOC_MIN_SIZE 0x8000000UL
-#define VMALLOC_END_INIT 0x80000000UL
+#define VMALLOC_START 0x78000000UL
+#define VMALLOC_END 0x7e000000UL
+#define VMEM_MAP_MAX 0x80000000UL
#else /* __s390x__ */
-#define VMALLOC_MIN_SIZE 0x2000000000UL
-#define VMALLOC_END_INIT 0x40000000000UL
+#define VMALLOC_START 0x3e000000000UL
+#define VMALLOC_END 0x3e040000000UL
+#define VMEM_MAP_MAX 0x40000000000UL
#endif /* __s390x__ */
+#define VMEM_MAP ((struct page *) VMALLOC_END)
+#define VMEM_MAP_SIZE ((VMALLOC_START / PAGE_SIZE) * sizeof(struct page))
+
/*
* A 31 bit pagetable entry of S390 has following format:
* | PFRA | | OS |
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 22/47] Print kernel version in dump_stack() and show_regs().
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (20 preceding siblings ...)
2007-12-20 15:19 ` [patch 21/47] Change vmalloc defintions Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 23/47] Get rid of HOLES_IN_ZONE requirement Martin Schwidefsky
` (24 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 121-version-printk.diff --]
[-- Type: text/plain, Size: 3282 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Also print PREEMPT and/or SMP if the kernel was configured that way.
Makes s390 look a bit more like other architectures.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/process.c | 18 ++++++++++--------
arch/s390/kernel/traps.c | 20 +++++++++++++++++---
2 files changed, 27 insertions(+), 11 deletions(-)
Index: quilt-2.6/arch/s390/kernel/process.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/process.c
+++ quilt-2.6/arch/s390/kernel/process.c
@@ -36,7 +36,7 @@
#include <linux/init.h>
#include <linux/module.h>
#include <linux/notifier.h>
-
+#include <linux/utsname.h>
#include <asm/uaccess.h>
#include <asm/pgtable.h>
#include <asm/system.h>
@@ -182,13 +182,15 @@ void cpu_idle(void)
void show_regs(struct pt_regs *regs)
{
- struct task_struct *tsk = current;
-
- printk("CPU: %d %s\n", task_thread_info(tsk)->cpu, print_tainted());
- printk("Process %s (pid: %d, task: %p, ksp: %p)\n",
- current->comm, task_pid_nr(current), (void *) tsk,
- (void *) tsk->thread.ksp);
-
+ print_modules();
+ printk("CPU: %d %s %s %.*s\n",
+ task_thread_info(current)->cpu, print_tainted(),
+ init_utsname()->release,
+ (int)strcspn(init_utsname()->version, " "),
+ init_utsname()->version);
+ printk("Process %s (pid: %d, task: %p, ksp: %p)\n",
+ current->comm, current->pid, current,
+ (void *) current->thread.ksp);
show_registers(regs);
/* Show stack backtrace if pt_regs is from kernel mode */
if (!(regs->psw.mask & PSW_MASK_PSTATE))
Index: quilt-2.6/arch/s390/kernel/traps.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/traps.c
+++ quilt-2.6/arch/s390/kernel/traps.c
@@ -31,6 +31,7 @@
#include <linux/reboot.h>
#include <linux/kprobes.h>
#include <linux/bug.h>
+#include <linux/utsname.h>
#include <asm/system.h>
#include <asm/uaccess.h>
#include <asm/io.h>
@@ -168,9 +169,16 @@ void show_stack(struct task_struct *task
*/
void dump_stack(void)
{
+ printk("CPU: %d %s %s %.*s\n",
+ task_thread_info(current)->cpu, print_tainted(),
+ init_utsname()->release,
+ (int)strcspn(init_utsname()->version, " "),
+ init_utsname()->version);
+ printk("Process %s (pid: %d, task: %p, ksp: %p)\n",
+ current->comm, current->pid, current,
+ (void *) current->thread.ksp);
show_stack(NULL, NULL);
}
-
EXPORT_SYMBOL(dump_stack);
static inline int mask_bits(struct pt_regs *regs, unsigned long bits)
@@ -258,8 +266,14 @@ void die(const char * str, struct pt_reg
console_verbose();
spin_lock_irq(&die_lock);
bust_spinlocks(1);
- printk("%s: %04lx [#%d]\n", str, err & 0xffff, ++die_counter);
- print_modules();
+ printk("%s: %04lx [#%d] ", str, err & 0xffff, ++die_counter);
+#ifdef CONFIG_PREEMPT
+ printk("PREEMPT ");
+#endif
+#ifdef CONFIG_SMP
+ printk("SMP");
+#endif
+ printk("\n");
notify_die(DIE_OOPS, str, regs, err, current->thread.trap_no, SIGSEGV);
show_regs(regs);
bust_spinlocks(0);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 23/47] Get rid of HOLES_IN_ZONE requirement.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (21 preceding siblings ...)
2007-12-20 15:19 ` [patch 22/47] Print kernel version in dump_stack() and show_regs() Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 24/47] DEBUG_PAGEALLOC support for s390 Martin Schwidefsky
` (23 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 122-zone-holes.diff --]
[-- Type: text/plain, Size: 2794 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Align everything to MAX_ORDER so we can get rid of the extra checks.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/Kconfig | 3 ---
arch/s390/kernel/setup.c | 25 ++++++++++++++++++++++++-
arch/s390/mm/vmem.c | 2 +-
3 files changed, 25 insertions(+), 5 deletions(-)
Index: quilt-2.6/arch/s390/Kconfig
===================================================================
--- quilt-2.6.orig/arch/s390/Kconfig
+++ quilt-2.6/arch/s390/Kconfig
@@ -276,9 +276,6 @@ source "kernel/Kconfig.preempt"
source "mm/Kconfig"
-config HOLES_IN_ZONE
- def_bool y
-
comment "I/O subsystem configuration"
config MACHCHK_WARNING
Index: quilt-2.6/arch/s390/kernel/setup.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/setup.c
+++ quilt-2.6/arch/s390/kernel/setup.c
@@ -559,7 +559,9 @@ setup_resources(void)
data_resource.start = (unsigned long) &_etext;
data_resource.end = (unsigned long) &_edata - 1;
- for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++) {
+ for (i = 0; i < MEMORY_CHUNKS; i++) {
+ if (!memory_chunk[i].size)
+ continue;
res = alloc_bootmem_low(sizeof(struct resource));
res->flags = IORESOURCE_BUSY | IORESOURCE_MEM;
switch (memory_chunk[i].type) {
@@ -630,6 +632,27 @@ static void __init setup_memory_end(void
max_mem = memory_end ? min(VMALLOC_START, memory_end) : VMALLOC_START;
memory_end = min(max_mem, memory_end);
+ /*
+ * Make sure all chunks are MAX_ORDER aligned so we don't need the
+ * extra checks that HOLES_IN_ZONE would require.
+ */
+ for (i = 0; i < MEMORY_CHUNKS; i++) {
+ unsigned long start, end;
+ struct mem_chunk *chunk;
+ unsigned long align;
+
+ chunk = &memory_chunk[i];
+ align = 1UL << (MAX_ORDER + PAGE_SHIFT - 1);
+ start = (chunk->addr + align - 1) & ~(align - 1);
+ end = (chunk->addr + chunk->size) & ~(align - 1);
+ if (start >= end)
+ memset(chunk, 0, sizeof(*chunk));
+ else {
+ chunk->addr = start;
+ chunk->size = end - start;
+ }
+ }
+
for (i = 0; i < MEMORY_CHUNKS; i++) {
struct mem_chunk *chunk = &memory_chunk[i];
Index: quilt-2.6/arch/s390/mm/vmem.c
===================================================================
--- quilt-2.6.orig/arch/s390/mm/vmem.c
+++ quilt-2.6/arch/s390/mm/vmem.c
@@ -376,7 +376,7 @@ static int __init vmem_convert_memory_ch
int i;
mutex_lock(&vmem_mutex);
- for (i = 0; i < MEMORY_CHUNKS && memory_chunk[i].size > 0; i++) {
+ for (i = 0; i < MEMORY_CHUNKS; i++) {
if (!memory_chunk[i].size)
continue;
seg = kzalloc(sizeof(*seg), GFP_KERNEL);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 24/47] DEBUG_PAGEALLOC support for s390.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (22 preceding siblings ...)
2007-12-20 15:19 ` [patch 23/47] Get rid of HOLES_IN_ZONE requirement Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 25/47] Remove owner_pc member from raw_spinlock_t Martin Schwidefsky
` (22 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 123-debug-pagealloc.diff --]
[-- Type: text/plain, Size: 3340 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/Kconfig.debug | 8 ++++++++
arch/s390/defconfig | 1 +
arch/s390/kernel/traps.c | 5 ++++-
arch/s390/mm/init.c | 27 +++++++++++++++++++++++++++
include/asm-s390/cacheflush.h | 4 ++++
5 files changed, 44 insertions(+), 1 deletion(-)
Index: quilt-2.6/arch/s390/defconfig
===================================================================
--- quilt-2.6.orig/arch/s390/defconfig
+++ quilt-2.6/arch/s390/defconfig
@@ -727,6 +727,7 @@ CONFIG_FORCED_INLINING=y
# CONFIG_LKDTM is not set
# CONFIG_FAULT_INJECTION is not set
CONFIG_SAMPLES=y
+# CONFIG_DEBUG_PAGEALLOC is not set
#
# Security options
Index: quilt-2.6/arch/s390/Kconfig.debug
===================================================================
--- quilt-2.6.orig/arch/s390/Kconfig.debug
+++ quilt-2.6/arch/s390/Kconfig.debug
@@ -6,4 +6,12 @@ config TRACE_IRQFLAGS_SUPPORT
source "lib/Kconfig.debug"
+config DEBUG_PAGEALLOC
+ bool "Debug page memory allocations"
+ depends on DEBUG_KERNEL
+ help
+ Unmap pages from the kernel linear mapping after free_pages().
+ This results in a slowdown, but helps to find certain types of
+ memory corruptions.
+
endmenu
Index: quilt-2.6/arch/s390/kernel/traps.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/traps.c
+++ quilt-2.6/arch/s390/kernel/traps.c
@@ -271,7 +271,10 @@ void die(const char * str, struct pt_reg
printk("PREEMPT ");
#endif
#ifdef CONFIG_SMP
- printk("SMP");
+ printk("SMP ");
+#endif
+#ifdef CONFIG_DEBUG_PAGEALLOC
+ printk("DEBUG_PAGEALLOC");
#endif
printk("\n");
notify_die(DIE_OOPS, str, regs, err, current->thread.trap_no, SIGSEGV);
Index: quilt-2.6/arch/s390/mm/init.c
===================================================================
--- quilt-2.6.orig/arch/s390/mm/init.c
+++ quilt-2.6/arch/s390/mm/init.c
@@ -167,6 +167,33 @@ void __init mem_init(void)
PFN_ALIGN((unsigned long)&_eshared) - 1);
}
+#ifdef CONFIG_DEBUG_PAGEALLOC
+void kernel_map_pages(struct page *page, int numpages, int enable)
+{
+ pgd_t *pgd;
+ pud_t *pud;
+ pmd_t *pmd;
+ pte_t *pte;
+ unsigned long address;
+ int i;
+
+ for (i = 0; i < numpages; i++) {
+ address = page_to_phys(page + i);
+ pgd = pgd_offset_k(address);
+ pud = pud_offset(pgd, address);
+ pmd = pmd_offset(pud, address);
+ pte = pte_offset_kernel(pmd, address);
+ if (!enable) {
+ ptep_invalidate(address, pte);
+ continue;
+ }
+ *pte = mk_pte_phys(address, __pgprot(_PAGE_TYPE_RW));
+ /* Flush cpu write queue. */
+ mb();
+ }
+}
+#endif
+
void free_initmem(void)
{
unsigned long addr;
Index: quilt-2.6/include/asm-s390/cacheflush.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/cacheflush.h
+++ quilt-2.6/include/asm-s390/cacheflush.h
@@ -24,4 +24,8 @@
#define copy_from_user_page(vma, page, vaddr, dst, src, len) \
memcpy(dst, src, len)
+#ifdef CONFIG_DEBUG_PAGEALLOC
+void kernel_map_pages(struct page *page, int numpages, int enable);
+#endif
+
#endif /* _S390_CACHEFLUSH_H */
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 25/47] Remove owner_pc member from raw_spinlock_t.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (23 preceding siblings ...)
2007-12-20 15:19 ` [patch 24/47] DEBUG_PAGEALLOC support for s390 Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 26/47] Use new style spinlock initializer in __RWSEM_INITIALIZER Martin Schwidefsky
` (21 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 124-spinlock-owner.diff --]
[-- Type: text/plain, Size: 3957 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Used to contain the address of the holder of the lock. But since the
spinlock code is not inlined anymore all locks contain the same address
anyway. And since in addtition nobody complained about that for ages
its obviously unused. So remove it.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/lib/spinlock.c | 12 ++++--------
include/asm-s390/spinlock.h | 19 ++++++-------------
include/asm-s390/spinlock_types.h | 1 -
3 files changed, 10 insertions(+), 22 deletions(-)
Index: quilt-2.6/arch/s390/lib/spinlock.c
===================================================================
--- quilt-2.6.orig/arch/s390/lib/spinlock.c
+++ quilt-2.6/arch/s390/lib/spinlock.c
@@ -39,7 +39,7 @@ static inline void _raw_yield_cpu(int cp
_raw_yield();
}
-void _raw_spin_lock_wait(raw_spinlock_t *lp, unsigned int pc)
+void _raw_spin_lock_wait(raw_spinlock_t *lp)
{
int count = spin_retry;
unsigned int cpu = ~smp_processor_id();
@@ -53,15 +53,13 @@ void _raw_spin_lock_wait(raw_spinlock_t
}
if (__raw_spin_is_locked(lp))
continue;
- if (_raw_compare_and_swap(&lp->owner_cpu, 0, cpu) == 0) {
- lp->owner_pc = pc;
+ if (_raw_compare_and_swap(&lp->owner_cpu, 0, cpu) == 0)
return;
- }
}
}
EXPORT_SYMBOL(_raw_spin_lock_wait);
-int _raw_spin_trylock_retry(raw_spinlock_t *lp, unsigned int pc)
+int _raw_spin_trylock_retry(raw_spinlock_t *lp)
{
unsigned int cpu = ~smp_processor_id();
int count;
@@ -69,10 +67,8 @@ int _raw_spin_trylock_retry(raw_spinlock
for (count = spin_retry; count > 0; count--) {
if (__raw_spin_is_locked(lp))
continue;
- if (_raw_compare_and_swap(&lp->owner_cpu, 0, cpu) == 0) {
- lp->owner_pc = pc;
+ if (_raw_compare_and_swap(&lp->owner_cpu, 0, cpu) == 0)
return 1;
- }
}
return 0;
}
Index: quilt-2.6/include/asm-s390/spinlock.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/spinlock.h
+++ quilt-2.6/include/asm-s390/spinlock.h
@@ -58,39 +58,32 @@ _raw_compare_and_swap(volatile unsigned
do { while (__raw_spin_is_locked(lock)) \
_raw_spin_relax(lock); } while (0)
-extern void _raw_spin_lock_wait(raw_spinlock_t *, unsigned int pc);
-extern int _raw_spin_trylock_retry(raw_spinlock_t *, unsigned int pc);
+extern void _raw_spin_lock_wait(raw_spinlock_t *);
+extern int _raw_spin_trylock_retry(raw_spinlock_t *);
extern void _raw_spin_relax(raw_spinlock_t *lock);
static inline void __raw_spin_lock(raw_spinlock_t *lp)
{
- unsigned long pc = 1 | (unsigned long) __builtin_return_address(0);
int old;
old = _raw_compare_and_swap(&lp->owner_cpu, 0, ~smp_processor_id());
- if (likely(old == 0)) {
- lp->owner_pc = pc;
+ if (likely(old == 0))
return;
- }
- _raw_spin_lock_wait(lp, pc);
+ _raw_spin_lock_wait(lp);
}
static inline int __raw_spin_trylock(raw_spinlock_t *lp)
{
- unsigned long pc = 1 | (unsigned long) __builtin_return_address(0);
int old;
old = _raw_compare_and_swap(&lp->owner_cpu, 0, ~smp_processor_id());
- if (likely(old == 0)) {
- lp->owner_pc = pc;
+ if (likely(old == 0))
return 1;
- }
- return _raw_spin_trylock_retry(lp, pc);
+ return _raw_spin_trylock_retry(lp);
}
static inline void __raw_spin_unlock(raw_spinlock_t *lp)
{
- lp->owner_pc = 0;
_raw_compare_and_swap(&lp->owner_cpu, lp->owner_cpu, 0);
}
Index: quilt-2.6/include/asm-s390/spinlock_types.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/spinlock_types.h
+++ quilt-2.6/include/asm-s390/spinlock_types.h
@@ -7,7 +7,6 @@
typedef struct {
volatile unsigned int owner_cpu;
- volatile unsigned int owner_pc;
} __attribute__ ((aligned (4))) raw_spinlock_t;
#define __RAW_SPIN_LOCK_UNLOCKED { 0 }
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 26/47] Use new style spinlock initializer in __RWSEM_INITIALIZER.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (24 preceding siblings ...)
2007-12-20 15:19 ` [patch 25/47] Remove owner_pc member from raw_spinlock_t Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 27/47] Get rid of additional_cpus kernel parameter Martin Schwidefsky
` (20 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 125-rwlock-init.diff --]
[-- Type: text/plain, Size: 950 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
include/asm-s390/rwsem.h | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
Index: quilt-2.6/include/asm-s390/rwsem.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/rwsem.h
+++ quilt-2.6/include/asm-s390/rwsem.h
@@ -91,8 +91,8 @@ struct rw_semaphore {
#endif
#define __RWSEM_INITIALIZER(name) \
-{ RWSEM_UNLOCKED_VALUE, SPIN_LOCK_UNLOCKED, LIST_HEAD_INIT((name).wait_list) \
- __RWSEM_DEP_MAP_INIT(name) }
+ { RWSEM_UNLOCKED_VALUE, __SPIN_LOCK_UNLOCKED((name).wait.lock), \
+ LIST_HEAD_INIT((name).wait_list) __RWSEM_DEP_MAP_INIT(name) }
#define DECLARE_RWSEM(name) \
struct rw_semaphore name = __RWSEM_INITIALIZER(name)
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 27/47] Get rid of additional_cpus kernel parameter.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (25 preceding siblings ...)
2007-12-20 15:19 ` [patch 26/47] Use new style spinlock initializer in __RWSEM_INITIALIZER Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 28/47] Remove appldata include from sysctl_check.c Martin Schwidefsky
` (19 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 126-additional-cpus.diff --]
[-- Type: text/plain, Size: 12852 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
It caused only a lot of confusion. From now on cpu hotplug of up to
NR_CPUS will work by default. If somebody wants to limit that then
the possible_cpus parameter can be used.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
Documentation/cpu-hotplug.txt | 2
arch/s390/kernel/early.c | 1
arch/s390/kernel/setup.c | 1
arch/s390/kernel/smp.c | 178 +++++++++++++++++-------------------------
drivers/s390/char/sclp_cmd.c | 46 ----------
include/asm-s390/sclp.h | 1
include/asm-s390/smp.h | 3
7 files changed, 78 insertions(+), 154 deletions(-)
Index: quilt-2.6/arch/s390/kernel/early.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/early.c
+++ quilt-2.6/arch/s390/kernel/early.c
@@ -278,7 +278,6 @@ void __init startup_init(void)
setup_lowcore_early();
sclp_read_info_early();
sclp_facilities_detect();
- sclp_read_cpu_info_early();
memsize = sclp_memory_detect();
#ifndef CONFIG_64BIT
/*
Index: quilt-2.6/arch/s390/kernel/setup.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/setup.c
+++ quilt-2.6/arch/s390/kernel/setup.c
@@ -922,7 +922,6 @@ setup_arch(char **cmdline_p)
cpu_init();
__cpu_logical_map[0] = S390_lowcore.cpu_data.cpu_addr;
- smp_setup_cpu_possible_map();
/*
* Setup capabilities (ELF_HWCAP & ELF_PLATFORM).
Index: quilt-2.6/arch/s390/kernel/smp.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/smp.c
+++ quilt-2.6/arch/s390/kernel/smp.c
@@ -54,7 +54,7 @@ EXPORT_SYMBOL(lowcore_ptr);
cpumask_t cpu_online_map = CPU_MASK_NONE;
EXPORT_SYMBOL(cpu_online_map);
-cpumask_t cpu_possible_map = CPU_MASK_NONE;
+cpumask_t cpu_possible_map = CPU_MASK_ALL;
EXPORT_SYMBOL(cpu_possible_map);
static struct task_struct *current_set[NR_CPUS];
@@ -399,7 +399,7 @@ static void __init smp_get_save_area(uns
"kernel was compiled with NR_CPUS=%i\n", cpu, NR_CPUS);
return;
}
- zfcpdump_save_areas[cpu] = alloc_bootmem(sizeof(union save_area));
+ zfcpdump_save_areas[cpu] = kmalloc(sizeof(union save_area), GFP_KERNEL);
__cpu_logical_map[CPU_INIT_NO] = (__u16) phy_cpu;
while (signal_processor(CPU_INIT_NO, sigp_stop_and_store_status) ==
sigp_busy)
@@ -435,67 +435,6 @@ static int cpu_stopped(int cpu)
return 0;
}
-/*
- * Lets check how many CPUs we have.
- */
-static void __init smp_count_cpus(unsigned int *configured_cpus,
- unsigned int *standby_cpus)
-{
- unsigned int cpu;
- struct sclp_cpu_info *info;
- u16 boot_cpu_addr, cpu_addr;
-
- boot_cpu_addr = S390_lowcore.cpu_data.cpu_addr;
- current_thread_info()->cpu = 0;
- *configured_cpus = 1;
- *standby_cpus = 0;
-
- info = alloc_bootmem_pages(sizeof(*info));
- if (!info)
- disabled_wait((unsigned long) __builtin_return_address(0));
-
- /* Use sigp detection algorithm if sclp doesn't work. */
- if (sclp_get_cpu_info(info)) {
- smp_use_sigp_detection = 1;
- for (cpu = 0; cpu <= 65535; cpu++) {
- if (cpu == boot_cpu_addr)
- continue;
- __cpu_logical_map[CPU_INIT_NO] = cpu;
- if (cpu_stopped(CPU_INIT_NO))
- (*configured_cpus)++;
- }
- goto out;
- }
-
- if (info->has_cpu_type) {
- for (cpu = 0; cpu < info->combined; cpu++) {
- if (info->cpu[cpu].address == boot_cpu_addr) {
- smp_cpu_type = info->cpu[cpu].type;
- break;
- }
- }
- }
- /* Count cpus. */
- for (cpu = 0; cpu < info->combined; cpu++) {
- if (info->has_cpu_type && info->cpu[cpu].type != smp_cpu_type)
- continue;
- cpu_addr = info->cpu[cpu].address;
- if (cpu_addr == boot_cpu_addr)
- continue;
- __cpu_logical_map[CPU_INIT_NO] = cpu_addr;
- if (!cpu_stopped(CPU_INIT_NO)) {
- (*standby_cpus)++;
- continue;
- }
- smp_get_save_area(*configured_cpus, cpu_addr);
- (*configured_cpus)++;
- }
-out:
- printk(KERN_INFO "CPUs: %d configured, %d standby\n",
- *configured_cpus, *standby_cpus);
- free_bootmem((unsigned long) info, sizeof(*info));
-}
-
static int cpu_known(int cpu_id)
{
int cpu;
@@ -529,7 +468,7 @@ static int smp_rescan_cpus_sigp(cpumask_
return 0;
}
-static int __init_refok smp_rescan_cpus_sclp(cpumask_t avail)
+static int smp_rescan_cpus_sclp(cpumask_t avail)
{
struct sclp_cpu_info *info;
int cpu_id, logical_cpu, cpu;
@@ -538,10 +477,7 @@ static int __init_refok smp_rescan_cpus_
logical_cpu = first_cpu(avail);
if (logical_cpu == NR_CPUS)
return 0;
- if (slab_is_available())
- info = kmalloc(sizeof(*info), GFP_KERNEL);
- else
- info = alloc_bootmem(sizeof(*info));
+ info = kmalloc(sizeof(*info), GFP_KERNEL);
if (!info)
return -ENOMEM;
rc = sclp_get_cpu_info(info);
@@ -564,10 +500,7 @@ static int __init_refok smp_rescan_cpus_
break;
}
out:
- if (slab_is_available())
- kfree(info);
- else
- free_bootmem((unsigned long) info, sizeof(*info));
+ kfree(info);
return rc;
}
@@ -575,15 +508,71 @@ static int smp_rescan_cpus(void)
{
cpumask_t avail;
- cpus_setall(avail);
- cpus_and(avail, avail, cpu_possible_map);
- cpus_andnot(avail, avail, cpu_present_map);
+ cpus_xor(avail, cpu_possible_map, cpu_present_map);
if (smp_use_sigp_detection)
return smp_rescan_cpus_sigp(avail);
else
return smp_rescan_cpus_sclp(avail);
}
+static void __init smp_detect_cpus(void)
+{
+ unsigned int cpu, c_cpus, s_cpus;
+ struct sclp_cpu_info *info;
+ u16 boot_cpu_addr, cpu_addr;
+
+ c_cpus = 1;
+ s_cpus = 0;
+ boot_cpu_addr = S390_lowcore.cpu_data.cpu_addr;
+ info = kmalloc(sizeof(*info), GFP_KERNEL);
+ if (!info)
+ panic("smp_detect_cpus failed to allocate memory\n");
+ /* Use sigp detection algorithm if sclp doesn't work. */
+ if (sclp_get_cpu_info(info)) {
+ smp_use_sigp_detection = 1;
+ for (cpu = 0; cpu <= 65535; cpu++) {
+ if (cpu == boot_cpu_addr)
+ continue;
+ __cpu_logical_map[CPU_INIT_NO] = cpu;
+ if (!cpu_stopped(CPU_INIT_NO))
+ continue;
+ smp_get_save_area(c_cpus, cpu);
+ c_cpus++;
+ }
+ goto out;
+ }
+
+ if (info->has_cpu_type) {
+ for (cpu = 0; cpu < info->combined; cpu++) {
+ if (info->cpu[cpu].address == boot_cpu_addr) {
+ smp_cpu_type = info->cpu[cpu].type;
+ break;
+ }
+ }
+ }
+
+ for (cpu = 0; cpu < info->combined; cpu++) {
+ if (info->has_cpu_type && info->cpu[cpu].type != smp_cpu_type)
+ continue;
+ cpu_addr = info->cpu[cpu].address;
+ if (cpu_addr == boot_cpu_addr)
+ continue;
+ __cpu_logical_map[CPU_INIT_NO] = cpu_addr;
+ if (!cpu_stopped(CPU_INIT_NO)) {
+ s_cpus++;
+ continue;
+ }
+ smp_get_save_area(c_cpus, cpu_addr);
+ c_cpus++;
+ }
+out:
+ kfree(info);
+ printk(KERN_INFO "CPUs: %d configured, %d standby\n", c_cpus, s_cpus);
+ lock_cpu_hotplug();
+ smp_rescan_cpus();
+ unlock_cpu_hotplug();
+}
+
/*
* Activate a secondary processor.
*/
@@ -674,41 +663,20 @@ int __cpu_up(unsigned int cpu)
return 0;
}
-static unsigned int __initdata additional_cpus;
-static unsigned int __initdata possible_cpus;
-
-void __init smp_setup_cpu_possible_map(void)
+static int __init setup_possible_cpus(char *s)
{
- unsigned int pos_cpus, cpu;
- unsigned int configured_cpus, standby_cpus;
+ int pcpus, cpu;
- smp_count_cpus(&configured_cpus, &standby_cpus);
- pos_cpus = min(configured_cpus + standby_cpus + additional_cpus,
- (unsigned int) NR_CPUS);
- if (possible_cpus)
- pos_cpus = min(possible_cpus, (unsigned int) NR_CPUS);
- for (cpu = 0; cpu < pos_cpus; cpu++)
+ pcpus = simple_strtoul(s, NULL, 0);
+ cpu_possible_map = cpumask_of_cpu(0);
+ for (cpu = 1; cpu < pcpus && cpu < NR_CPUS; cpu++)
cpu_set(cpu, cpu_possible_map);
- cpu_present_map = cpumask_of_cpu(0);
- smp_rescan_cpus();
-}
-
-#ifdef CONFIG_HOTPLUG_CPU
-
-static int __init setup_additional_cpus(char *s)
-{
- additional_cpus = simple_strtoul(s, NULL, 0);
- return 0;
-}
-early_param("additional_cpus", setup_additional_cpus);
-
-static int __init setup_possible_cpus(char *s)
-{
- possible_cpus = simple_strtoul(s, NULL, 0);
return 0;
}
early_param("possible_cpus", setup_possible_cpus);
+#ifdef CONFIG_HOTPLUG_CPU
+
int __cpu_disable(void)
{
struct ec_creg_mask_parms cr_parms;
@@ -768,6 +736,8 @@ void __init smp_prepare_cpus(unsigned in
unsigned int cpu;
int i;
+ smp_detect_cpus();
+
/* request the 0x1201 emergency signal external interrupt */
if (register_external_interrupt(0x1201, do_ext_call_interrupt) != 0)
panic("Couldn't request external interrupt 0x1201");
@@ -816,6 +786,8 @@ void __init smp_prepare_boot_cpu(void)
{
BUG_ON(smp_processor_id() != 0);
+ current_thread_info()->cpu = 0;
+ cpu_set(0, cpu_present_map);
cpu_set(0, cpu_online_map);
S390_lowcore.percpu_offset = __per_cpu_offset[0];
current_set[0] = current;
Index: quilt-2.6/Documentation/cpu-hotplug.txt
===================================================================
--- quilt-2.6.orig/Documentation/cpu-hotplug.txt
+++ quilt-2.6/Documentation/cpu-hotplug.txt
@@ -50,7 +50,7 @@ additional_cpus=n (*) Use this to limit
cpu_possible_map = cpu_present_map + additional_cpus
(*) Option valid only for following architectures
-- x86_64, ia64, s390
+- x86_64, ia64
ia64 and x86_64 use the number of disabled local apics in ACPI tables MADT
to determine the number of potentially hot-pluggable cpus. The implementation
Index: quilt-2.6/drivers/s390/char/sclp_cmd.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/sclp_cmd.c
+++ quilt-2.6/drivers/s390/char/sclp_cmd.c
@@ -191,9 +191,6 @@ struct read_cpu_info_sccb {
u8 reserved[4096 - 16];
} __attribute__((packed, aligned(PAGE_SIZE)));
-static struct read_cpu_info_sccb __initdata early_read_cpu_info_sccb;
-static struct sclp_cpu_info __initdata sclp_cpu_info;
-
static void sclp_fill_cpu_info(struct sclp_cpu_info *info,
struct read_cpu_info_sccb *sccb)
{
@@ -208,48 +205,16 @@ static void sclp_fill_cpu_info(struct sc
info->combined * sizeof(struct sclp_cpu_entry));
}
-void __init sclp_read_cpu_info_early(void)
+int sclp_get_cpu_info(struct sclp_cpu_info *info)
{
int rc;
struct read_cpu_info_sccb *sccb;
if (!SCLP_HAS_CPU_INFO)
- return;
-
- sccb = &early_read_cpu_info_sccb;
- do {
- memset(sccb, 0, sizeof(*sccb));
- sccb->header.length = sizeof(*sccb);
- rc = sclp_cmd_sync_early(SCLP_CMDW_READ_CPU_INFO, sccb);
- } while (rc == -EBUSY);
-
- if (rc)
- return;
- if (sccb->header.response_code != 0x10)
- return;
- sclp_fill_cpu_info(&sclp_cpu_info, sccb);
-}
-
-static int __init sclp_get_cpu_info_early(struct sclp_cpu_info *info)
-{
- if (!SCLP_HAS_CPU_INFO)
return -EOPNOTSUPP;
- *info = sclp_cpu_info;
- return 0;
-}
-
-static int sclp_get_cpu_info_late(struct sclp_cpu_info *info)
-{
- int rc;
- struct read_cpu_info_sccb *sccb;
-
- if (!SCLP_HAS_CPU_INFO)
- return -EOPNOTSUPP;
- sccb = (struct read_cpu_info_sccb *) __get_free_page(GFP_KERNEL
- | GFP_DMA);
+ sccb = (void *) get_zeroed_page(GFP_KERNEL | GFP_DMA);
if (!sccb)
return -ENOMEM;
- memset(sccb, 0, sizeof(*sccb));
sccb->header.length = sizeof(*sccb);
rc = do_sync_request(SCLP_CMDW_READ_CPU_INFO, sccb);
if (rc)
@@ -266,13 +231,6 @@ out:
return rc;
}
-int __init_refok sclp_get_cpu_info(struct sclp_cpu_info *info)
-{
- if (slab_is_available())
- return sclp_get_cpu_info_late(info);
- return sclp_get_cpu_info_early(info);
-}
-
struct cpu_configure_sccb {
struct sccb_header header;
} __attribute__((packed, aligned(8)));
Index: quilt-2.6/include/asm-s390/sclp.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/sclp.h
+++ quilt-2.6/include/asm-s390/sclp.h
@@ -46,7 +46,6 @@ int sclp_get_cpu_info(struct sclp_cpu_in
int sclp_cpu_configure(u8 cpu);
int sclp_cpu_deconfigure(u8 cpu);
void sclp_read_info_early(void);
-void sclp_read_cpu_info_early(void);
void sclp_facilities_detect(void);
unsigned long long sclp_memory_detect(void);
int sclp_sdias_blk_count(void);
Index: quilt-2.6/include/asm-s390/smp.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/smp.h
+++ quilt-2.6/include/asm-s390/smp.h
@@ -35,8 +35,6 @@ extern void machine_restart_smp(char *);
extern void machine_halt_smp(void);
extern void machine_power_off_smp(void);
-extern void smp_setup_cpu_possible_map(void);
-
#define NO_PROC_ID 0xFF /* No processor magic marker */
/*
@@ -103,7 +101,6 @@ static inline void smp_send_stop(void)
#define hard_smp_processor_id() 0
#define smp_cpu_not_running(cpu) 1
-#define smp_setup_cpu_possible_map() do { } while (0)
#endif
extern union save_area *zfcpdump_save_areas[NR_CPUS + 1];
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 28/47] Remove appldata include from sysctl_check.c
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (26 preceding siblings ...)
2007-12-20 15:19 ` [patch 27/47] Get rid of additional_cpus kernel parameter Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 29/47] crypto: move s390 Kconfig options Martin Schwidefsky
` (18 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 127-appldata.diff --]
[-- Type: text/plain, Size: 780 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
Forgot to remove this when removing the appldata binary sysctls.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
kernel/sysctl_check.c | 1 -
1 file changed, 1 deletion(-)
Index: quilt-2.6/kernel/sysctl_check.c
===================================================================
--- quilt-2.6.orig/kernel/sysctl_check.c
+++ quilt-2.6/kernel/sysctl_check.c
@@ -1,6 +1,5 @@
#include <linux/stat.h>
#include <linux/sysctl.h>
-#include "../arch/s390/appldata/appldata.h"
#include "../fs/xfs/linux-2.6/xfs_sysctl.h"
#include <linux/sunrpc/debug.h>
#include <linux/string.h>
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 29/47] crypto: move s390 Kconfig options.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (27 preceding siblings ...)
2007-12-20 15:19 ` [patch 28/47] Remove appldata include from sysctl_check.c Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 30/47] dasd: fix return value of dasd_generic_probe() Martin Schwidefsky
` (17 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Jan Glauber, Martin Schwidefsky
[-- Attachment #1: 128-crypto-kconfig.diff --]
[-- Type: text/plain, Size: 5736 bytes --]
From: Jan Glauber <jan.glauber@de.ibm.com>
Move s390 crypto Kconfig options to drivers/crypto/Kconfig to have all
hardware crypto devices in one place.
This also makes messing up the kernel source tree easier for some people.
Signed-off-by: Jan Glauber <jan.glauber@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/crypto/Kconfig | 60 --------------------------------------------
drivers/crypto/Kconfig | 63 +++++++++++++++++++++++++++++++++++++++++++++--
2 files changed, 61 insertions(+), 62 deletions(-)
Index: quilt-2.6/arch/s390/crypto/Kconfig
===================================================================
--- quilt-2.6.orig/arch/s390/crypto/Kconfig
+++ /dev/null
@@ -1,60 +0,0 @@
-config CRYPTO_SHA1_S390
- tristate "SHA1 digest algorithm"
- depends on S390
- select CRYPTO_ALGAPI
- help
- This is the s390 hardware accelerated implementation of the
- SHA-1 secure hash standard (FIPS 180-1/DFIPS 180-2).
-
-config CRYPTO_SHA256_S390
- tristate "SHA256 digest algorithm"
- depends on S390
- select CRYPTO_ALGAPI
- help
- This is the s390 hardware accelerated implementation of the
- SHA256 secure hash standard (DFIPS 180-2).
-
- This version of SHA implements a 256 bit hash with 128 bits of
- security against collision attacks.
-
-config CRYPTO_DES_S390
- tristate "DES and Triple DES cipher algorithms"
- depends on S390
- select CRYPTO_ALGAPI
- select CRYPTO_BLKCIPHER
- help
- This us the s390 hardware accelerated implementation of the
- DES cipher algorithm (FIPS 46-2), and Triple DES EDE (FIPS 46-3).
-
-config CRYPTO_AES_S390
- tristate "AES cipher algorithms"
- depends on S390
- select CRYPTO_ALGAPI
- select CRYPTO_BLKCIPHER
- help
- This is the s390 hardware accelerated implementation of the
- AES cipher algorithms (FIPS-197). AES uses the Rijndael
- algorithm.
-
- Rijndael appears to be consistently a very good performer in
- both hardware and software across a wide range of computing
- environments regardless of its use in feedback or non-feedback
- modes. Its key setup time is excellent, and its key agility is
- good. Rijndael's very low memory requirements make it very well
- suited for restricted-space environments, in which it also
- demonstrates excellent performance. Rijndael's operations are
- among the easiest to defend against power and timing attacks.
-
- On s390 the System z9-109 currently only supports the key size
- of 128 bit.
-
-config S390_PRNG
- tristate "Pseudo random number generator device driver"
- depends on S390
- default "m"
- help
- Select this option if you want to use the s390 pseudo random number
- generator. The PRNG is part of the cryptographic processor functions
- and uses triple-DES to generate secure random numbers like the
- ANSI X9.17 standard. The PRNG is usable via the char device
- /dev/prandom.
Index: quilt-2.6/drivers/crypto/Kconfig
===================================================================
--- quilt-2.6.orig/drivers/crypto/Kconfig
+++ quilt-2.6/drivers/crypto/Kconfig
@@ -48,8 +48,6 @@ config CRYPTO_DEV_PADLOCK_SHA
If unsure say M. The compiled module will be
called padlock-sha.ko
-source "arch/s390/crypto/Kconfig"
-
config CRYPTO_DEV_GEODE
tristate "Support for the Geode LX AES engine"
depends on X86_32 && PCI
@@ -83,4 +81,65 @@ config ZCRYPT_MONOLITHIC
that contains all parts of the crypto device driver (ap bus,
request router and all the card drivers).
+config CRYPTO_SHA1_S390
+ tristate "SHA1 digest algorithm"
+ depends on S390
+ select CRYPTO_ALGAPI
+ help
+ This is the s390 hardware accelerated implementation of the
+ SHA-1 secure hash standard (FIPS 180-1/DFIPS 180-2).
+
+config CRYPTO_SHA256_S390
+ tristate "SHA256 digest algorithm"
+ depends on S390
+ select CRYPTO_ALGAPI
+ help
+ This is the s390 hardware accelerated implementation of the
+ SHA256 secure hash standard (DFIPS 180-2).
+
+ This version of SHA implements a 256 bit hash with 128 bits of
+ security against collision attacks.
+
+config CRYPTO_DES_S390
+ tristate "DES and Triple DES cipher algorithms"
+ depends on S390
+ select CRYPTO_ALGAPI
+ select CRYPTO_BLKCIPHER
+ help
+ This us the s390 hardware accelerated implementation of the
+ DES cipher algorithm (FIPS 46-2), and Triple DES EDE (FIPS 46-3).
+
+config CRYPTO_AES_S390
+ tristate "AES cipher algorithms"
+ depends on S390
+ select CRYPTO_ALGAPI
+ select CRYPTO_BLKCIPHER
+ help
+ This is the s390 hardware accelerated implementation of the
+ AES cipher algorithms (FIPS-197). AES uses the Rijndael
+ algorithm.
+
+ Rijndael appears to be consistently a very good performer in
+ both hardware and software across a wide range of computing
+ environments regardless of its use in feedback or non-feedback
+ modes. Its key setup time is excellent, and its key agility is
+ good. Rijndael's very low memory requirements make it very well
+ suited for restricted-space environments, in which it also
+ demonstrates excellent performance. Rijndael's operations are
+ among the easiest to defend against power and timing attacks.
+
+ On s390 the System z9-109 currently only supports the key size
+ of 128 bit.
+
+config S390_PRNG
+ tristate "Pseudo random number generator device driver"
+ depends on S390
+ default "m"
+ help
+ Select this option if you want to use the s390 pseudo random number
+ generator. The PRNG is part of the cryptographic processor functions
+ and uses triple-DES to generate secure random numbers like the
+ ANSI X9.17 standard. The PRNG is usable via the char device
+ /dev/prandom.
+
endif # CRYPTO_HW
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 30/47] dasd: fix return value of dasd_generic_probe()
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (28 preceding siblings ...)
2007-12-20 15:19 ` [patch 29/47] crypto: move s390 Kconfig options Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 31/47] arch/s390: Add missing "space" Martin Schwidefsky
` (16 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Stefan Haberland, Martin Schwidefsky
[-- Attachment #1: 129-dasd-probe.diff --]
[-- Type: text/plain, Size: 1093 bytes --]
From: Stefan Haberland <stefan.haberland@de.ibm.com>
Using the return value of ccw_device_set_online as return value for
dasd_generic_probe() causes the DASD to fail setting online
Signed-off-by: Stefan Haberland <stefan.haberland@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/block/dasd.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
Index: quilt-2.6/drivers/s390/block/dasd.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd.c
+++ quilt-2.6/drivers/s390/block/dasd.c
@@ -1969,9 +1969,10 @@ dasd_generic_probe (struct ccw_device *c
ret = ccw_device_set_online(cdev);
if (ret)
printk(KERN_WARNING
- "dasd_generic_probe: could not initially online "
- "ccw-device %s\n", cdev->dev.bus_id);
- return ret;
+ "dasd_generic_probe: could not initially "
+ "online ccw-device %s; return code: %d\n",
+ cdev->dev.bus_id, ret);
+ return 0;
}
/*
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 31/47] arch/s390: Add missing "space"
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (29 preceding siblings ...)
2007-12-20 15:19 ` [patch 30/47] dasd: fix return value of dasd_generic_probe() Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 32/47] drivers/s390: " Martin Schwidefsky
` (15 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Joe Perches, Martin Schwidefsky
[-- Attachment #1: 130-arch-whitespace.diff --]
[-- Type: text/plain, Size: 872 bytes --]
From: Joe Perches <joe@perches.com>
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/crypto/aes_s390.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Index: quilt-2.6/arch/s390/crypto/aes_s390.c
===================================================================
--- quilt-2.6.orig/arch/s390/crypto/aes_s390.c
+++ quilt-2.6/arch/s390/crypto/aes_s390.c
@@ -341,7 +341,7 @@ static int __init aes_init(void)
ecb_aes_alg.cra_u.blkcipher.max_keysize = AES_MIN_KEY_SIZE;
cbc_aes_alg.cra_u.blkcipher.max_keysize = AES_MIN_KEY_SIZE;
printk(KERN_INFO
- "aes_s390: hardware acceleration only available for"
+ "aes_s390: hardware acceleration only available for "
"128 bit keys\n");
}
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 32/47] drivers/s390: Add missing "space"
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (30 preceding siblings ...)
2007-12-20 15:19 ` [patch 31/47] arch/s390: Add missing "space" Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 33/47] kernel: Shutdown Actions Interface Martin Schwidefsky
` (14 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Joe Perches, Martin Schwidefsky
[-- Attachment #1: 131-drivers-whitespace.diff --]
[-- Type: text/plain, Size: 5507 bytes --]
From: Joe Perches <joe@perches.com>
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/char/monwriter.c | 2 +-
drivers/s390/char/vmlogrdr.c | 2 +-
drivers/s390/cio/chsc.c | 2 +-
drivers/s390/net/claw.c | 2 +-
drivers/s390/net/lcs.c | 2 +-
drivers/s390/scsi/zfcp_fsf.c | 10 +++++-----
6 files changed, 10 insertions(+), 10 deletions(-)
Index: quilt-2.6/drivers/s390/char/monwriter.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/monwriter.c
+++ quilt-2.6/drivers/s390/char/monwriter.c
@@ -295,7 +295,7 @@ module_init(mon_init);
module_exit(mon_exit);
module_param_named(max_bufs, mon_max_bufs, int, 0644);
-MODULE_PARM_DESC(max_bufs, "Maximum number of sample monitor data buffers"
+MODULE_PARM_DESC(max_bufs, "Maximum number of sample monitor data buffers "
"that can be active at one time");
MODULE_AUTHOR("Melissa Howland <Melissa.Howland@us.ibm.com>");
Index: quilt-2.6/drivers/s390/char/vmlogrdr.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/vmlogrdr.c
+++ quilt-2.6/drivers/s390/char/vmlogrdr.c
@@ -683,7 +683,7 @@ static int vmlogrdr_register_driver(void
/* Register with iucv driver */
ret = iucv_register(&vmlogrdr_iucv_handler, 1);
if (ret) {
- printk (KERN_ERR "vmlogrdr: failed to register with"
+ printk (KERN_ERR "vmlogrdr: failed to register with "
"iucv driver\n");
goto out;
}
Index: quilt-2.6/drivers/s390/cio/chsc.c
===================================================================
--- quilt-2.6.orig/drivers/s390/cio/chsc.c
+++ quilt-2.6/drivers/s390/cio/chsc.c
@@ -1042,7 +1042,7 @@ chsc_determine_css_characteristics(void)
scsc_area = (void *)get_zeroed_page(GFP_KERNEL | GFP_DMA);
if (!scsc_area) {
- CIO_MSG_EVENT(0, "Was not able to determine available"
+ CIO_MSG_EVENT(0, "Was not able to determine available "
"CHSCs due to no memory.\n");
return -ENOMEM;
}
Index: quilt-2.6/drivers/s390/net/claw.c
===================================================================
--- quilt-2.6.orig/drivers/s390/net/claw.c
+++ quilt-2.6/drivers/s390/net/claw.c
@@ -2416,7 +2416,7 @@ init_ccw_bk(struct net_device *dev)
privptr->p_buff_pages_perwrite);
#endif
if (p_buff==NULL) {
- printk(KERN_INFO "%s:%s __get_free_pages"
+ printk(KERN_INFO "%s:%s __get_free_pages "
"for writes buf failed : get is for %d pages\n",
dev->name,
__FUNCTION__,
Index: quilt-2.6/drivers/s390/net/lcs.c
===================================================================
--- quilt-2.6.orig/drivers/s390/net/lcs.c
+++ quilt-2.6/drivers/s390/net/lcs.c
@@ -1115,7 +1115,7 @@ list_modified:
rc = lcs_send_setipm(card, ipm);
spin_lock_irqsave(&card->ipm_lock, flags);
if (rc) {
- PRINT_INFO("Adding multicast address failed."
+ PRINT_INFO("Adding multicast address failed. "
"Table possibly full!\n");
/* store ipm in failed list -> will be added
* to ipm_list again, so a retry will be done
Index: quilt-2.6/drivers/s390/scsi/zfcp_fsf.c
===================================================================
--- quilt-2.6.orig/drivers/s390/scsi/zfcp_fsf.c
+++ quilt-2.6/drivers/s390/scsi/zfcp_fsf.c
@@ -502,7 +502,7 @@ zfcp_fsf_fsfstatus_qual_eval(struct zfcp
fsf_req->status |= ZFCP_STATUS_FSFREQ_ERROR;
break;
case FSF_SQ_NO_RECOM:
- ZFCP_LOG_NORMAL("bug: No recommendation could be given for a"
+ ZFCP_LOG_NORMAL("bug: No recommendation could be given for a "
"problem on the adapter %s "
"Stopping all operations on this adapter. ",
zfcp_get_busid_by_adapter(fsf_req->adapter));
@@ -813,7 +813,7 @@ zfcp_fsf_status_read_port_closed(struct
read_unlock_irqrestore(&zfcp_data.config_lock, flags);
if (!port || (port->d_id != (status_buffer->d_id & ZFCP_DID_MASK))) {
- ZFCP_LOG_NORMAL("bug: Reopen port indication received for"
+ ZFCP_LOG_NORMAL("bug: Reopen port indication received for "
"nonexisting port with d_id 0x%06x on "
"adapter %s. Ignored.\n",
status_buffer->d_id & ZFCP_DID_MASK,
@@ -2280,7 +2280,7 @@ zfcp_fsf_exchange_port_data(struct zfcp_
&lock_flags, &fsf_req);
if (retval) {
ZFCP_LOG_INFO("error: Out of resources. Could not create an "
- "exchange port data request for"
+ "exchange port data request for "
"the adapter %s.\n",
zfcp_get_busid_by_adapter(adapter));
write_unlock_irqrestore(&adapter->request_queue.queue_lock,
@@ -2339,7 +2339,7 @@ zfcp_fsf_exchange_port_data_sync(struct
0, NULL, &lock_flags, &fsf_req);
if (retval) {
ZFCP_LOG_INFO("error: Out of resources. Could not create an "
- "exchange port data request for"
+ "exchange port data request for "
"the adapter %s.\n",
zfcp_get_busid_by_adapter(adapter));
write_unlock_irqrestore(&adapter->request_queue.queue_lock,
@@ -4725,7 +4725,7 @@ zfcp_fsf_req_create(struct zfcp_adapter
/* allocate new FSF request */
fsf_req = zfcp_fsf_req_alloc(pool, req_flags);
if (unlikely(NULL == fsf_req)) {
- ZFCP_LOG_DEBUG("error: Could not put an FSF request into"
+ ZFCP_LOG_DEBUG("error: Could not put an FSF request into "
"the outbound (send) queue.\n");
ret = -ENOMEM;
goto failed_fsf_req;
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 33/47] kernel: Shutdown Actions Interface
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (31 preceding siblings ...)
2007-12-20 15:19 ` [patch 32/47] drivers/s390: " Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:19 ` [patch 34/47] Load disabled wait psw instead of stopping cpu on halt Martin Schwidefsky
` (13 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Michael Holzheu, Martin Schwidefsky
[-- Attachment #1: 132-shutdown-actions.diff --]
[-- Type: text/plain, Size: 31881 bytes --]
From: Michael Holzheu <holzheu@de.ibm.com>
In case of a kernel panic it is currently possible to specify that a dump
should be created, the system should be rebooted or stopped. Virtual sysfs
files under the directory /sys/firmware/ are used for that configuration.
In addition to that, there are kernel parameters 'vmhalt', 'vmpoff'
and 'vmpanic', which can be used to specify z/VM commands, which are
automatically executed in case of halt, power off or a kernel panic.
This patch combines both functionalities and allows to specify the z/VM CP
commands also via sysfs attributes. In addition to that, it enhances the
existing handling of shutdown triggers (e.g. halt or panic) and associated
shutdown actions (e.g. dump or reipl) and makes it more flexible.
Signed-off-by: Michael Holzheu <holzheu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/ipl.c | 923 ++++++++++++++++++++++++++++++-----------------
arch/s390/kernel/setup.c | 103 -----
arch/s390/kernel/smp.c | 27 -
include/asm-s390/ipl.h | 4
4 files changed, 595 insertions(+), 462 deletions(-)
Index: quilt-2.6/arch/s390/kernel/ipl.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/ipl.c
+++ quilt-2.6/arch/s390/kernel/ipl.c
@@ -2,7 +2,7 @@
* arch/s390/kernel/ipl.c
* ipl/reipl/dump support for Linux on s390.
*
- * Copyright (C) IBM Corp. 2005,2006
+ * Copyright IBM Corp. 2005,2007
* Author(s): Michael Holzheu <holzheu@de.ibm.com>
* Heiko Carstens <heiko.carstens@de.ibm.com>
* Volker Sameske <sameske@de.ibm.com>
@@ -31,6 +31,43 @@
#define IPL_FCP_DUMP_STR "fcp_dump"
#define IPL_NSS_STR "nss"
+#define DUMP_CCW_STR "ccw"
+#define DUMP_FCP_STR "fcp"
+#define DUMP_NONE_STR "none"
+
+/*
+ * Four shutdown trigger types are supported:
+ * - panic
+ * - halt
+ * - power off
+ * - reipl
+ */
+#define ON_PANIC_STR "on_panic"
+#define ON_HALT_STR "on_halt"
+#define ON_POFF_STR "on_poff"
+#define ON_REIPL_STR "on_reboot"
+
+struct shutdown_action;
+struct shutdown_trigger {
+ char *name;
+ struct shutdown_action *action;
+};
+
+/*
+ * Five shutdown action types are supported:
+ */
+#define SHUTDOWN_ACTION_IPL_STR "ipl"
+#define SHUTDOWN_ACTION_REIPL_STR "reipl"
+#define SHUTDOWN_ACTION_DUMP_STR "dump"
+#define SHUTDOWN_ACTION_VMCMD_STR "vmcmd"
+#define SHUTDOWN_ACTION_STOP_STR "stop"
+
+struct shutdown_action {
+ char *name;
+ void (*fn) (struct shutdown_trigger *trigger);
+ int (*init) (void);
+};
+
static char *ipl_type_str(enum ipl_type type)
{
switch (type) {
@@ -54,10 +91,6 @@ enum dump_type {
DUMP_TYPE_FCP = 4,
};
-#define DUMP_NONE_STR "none"
-#define DUMP_CCW_STR "ccw"
-#define DUMP_FCP_STR "fcp"
-
static char *dump_type_str(enum dump_type type)
{
switch (type) {
@@ -99,30 +132,6 @@ enum dump_method {
DUMP_METHOD_FCP_DIAG,
};
-enum shutdown_action {
- SHUTDOWN_REIPL,
- SHUTDOWN_DUMP,
- SHUTDOWN_STOP,
-};
-
-#define SHUTDOWN_REIPL_STR "reipl"
-#define SHUTDOWN_DUMP_STR "dump"
-#define SHUTDOWN_STOP_STR "stop"
-
-static char *shutdown_action_str(enum shutdown_action action)
-{
- switch (action) {
- case SHUTDOWN_REIPL:
- return SHUTDOWN_REIPL_STR;
- case SHUTDOWN_DUMP:
- return SHUTDOWN_DUMP_STR;
- case SHUTDOWN_STOP:
- return SHUTDOWN_STOP_STR;
- default:
- return NULL;
- }
-}
-
static int diag308_set_works = 0;
static int reipl_capabilities = IPL_TYPE_UNKNOWN;
@@ -140,8 +149,6 @@ static enum dump_method dump_method = DU
static struct ipl_parameter_block *dump_block_fcp;
static struct ipl_parameter_block *dump_block_ccw;
-static enum shutdown_action on_panic_action = SHUTDOWN_STOP;
-
static struct sclp_ipl_info sclp_ipl_info;
int diag308(unsigned long subcode, void *addr)
@@ -200,8 +207,8 @@ static ssize_t sys_##_prefix##_##_name##
static ssize_t sys_##_prefix##_##_name##_store(struct kset *kset, \
const char *buf, size_t len) \
{ \
- if (sscanf(buf, _fmt_in, _value) != 1) \
- return -EINVAL; \
+ strncpy(_value, buf, sizeof(_value) - 1); \
+ strstrip(_value); \
return len; \
} \
static struct subsys_attribute sys_##_prefix##_##_name##_attr = \
@@ -240,33 +247,6 @@ static __init enum ipl_type get_ipl_type
return IPL_TYPE_FCP;
}
-void __init setup_ipl_info(void)
-{
- ipl_info.type = get_ipl_type();
- switch (ipl_info.type) {
- case IPL_TYPE_CCW:
- ipl_info.data.ccw.dev_id.devno = ipl_devno;
- ipl_info.data.ccw.dev_id.ssid = 0;
- break;
- case IPL_TYPE_FCP:
- case IPL_TYPE_FCP_DUMP:
- ipl_info.data.fcp.dev_id.devno =
- IPL_PARMBLOCK_START->ipl_info.fcp.devno;
- ipl_info.data.fcp.dev_id.ssid = 0;
- ipl_info.data.fcp.wwpn = IPL_PARMBLOCK_START->ipl_info.fcp.wwpn;
- ipl_info.data.fcp.lun = IPL_PARMBLOCK_START->ipl_info.fcp.lun;
- break;
- case IPL_TYPE_NSS:
- strncpy(ipl_info.data.nss.name, kernel_nss_name,
- sizeof(ipl_info.data.nss.name));
- break;
- case IPL_TYPE_UNKNOWN:
- default:
- /* We have no info to copy */
- break;
- }
-}
-
struct ipl_info ipl_info;
EXPORT_SYMBOL_GPL(ipl_info);
@@ -420,8 +400,77 @@ static struct attribute_group ipl_unknow
static decl_subsys(ipl, NULL, NULL);
+static int __init ipl_register_fcp_files(void)
+{
+ int rc;
+
+ rc = sysfs_create_group(&ipl_subsys.kobj,
+ &ipl_fcp_attr_group);
+ if (rc)
+ goto out;
+ rc = sysfs_create_bin_file(&ipl_subsys.kobj,
+ &ipl_parameter_attr);
+ if (rc)
+ goto out_ipl_parm;
+ rc = sysfs_create_bin_file(&ipl_subsys.kobj,
+ &ipl_scp_data_attr);
+ if (!rc)
+ goto out;
+
+ sysfs_remove_bin_file(&ipl_subsys.kobj, &ipl_parameter_attr);
+
+out_ipl_parm:
+ sysfs_remove_group(&ipl_subsys.kobj, &ipl_fcp_attr_group);
+out:
+ return rc;
+}
+
+static void ipl_run(struct shutdown_trigger *trigger)
+{
+ diag308(DIAG308_IPL, NULL);
+ if (MACHINE_IS_VM)
+ __cpcmd("IPL", NULL, 0, NULL);
+ else if (ipl_info.type == IPL_TYPE_CCW)
+ reipl_ccw_dev(&ipl_info.data.ccw.dev_id);
+}
+
+static int ipl_init(void)
+{
+ int rc;
+
+ rc = firmware_register(&ipl_subsys);
+ if (rc)
+ goto out;
+ switch (ipl_info.type) {
+ case IPL_TYPE_CCW:
+ rc = sysfs_create_group(&ipl_subsys.kobj,
+ &ipl_ccw_attr_group);
+ break;
+ case IPL_TYPE_FCP:
+ case IPL_TYPE_FCP_DUMP:
+ rc = ipl_register_fcp_files();
+ break;
+ case IPL_TYPE_NSS:
+ rc = sysfs_create_group(&ipl_subsys.kobj,
+ &ipl_nss_attr_group);
+ break;
+ default:
+ rc = sysfs_create_group(&ipl_subsys.kobj,
+ &ipl_unknown_attr_group);
+ break;
+ }
+out:
+ if (rc)
+ panic("ipl_init failed: rc = %i\n", rc);
+
+ return 0;
+}
+
+static struct shutdown_action ipl_action = {SHUTDOWN_ACTION_IPL_STR, ipl_run,
+ ipl_init};
+
/*
- * reipl section
+ * reipl shutdown action: Reboot Linux on shutdown.
*/
/* FCP reipl device attributes */
@@ -592,135 +641,7 @@ static struct subsys_attribute reipl_typ
static decl_subsys(reipl, NULL, NULL);
-/*
- * dump section
- */
-
-/* FCP dump device attributes */
-
-DEFINE_IPL_ATTR_RW(dump_fcp, wwpn, "0x%016llx\n", "%016llx\n",
- dump_block_fcp->ipl_info.fcp.wwpn);
-DEFINE_IPL_ATTR_RW(dump_fcp, lun, "0x%016llx\n", "%016llx\n",
- dump_block_fcp->ipl_info.fcp.lun);
-DEFINE_IPL_ATTR_RW(dump_fcp, bootprog, "%lld\n", "%lld\n",
- dump_block_fcp->ipl_info.fcp.bootprog);
-DEFINE_IPL_ATTR_RW(dump_fcp, br_lba, "%lld\n", "%lld\n",
- dump_block_fcp->ipl_info.fcp.br_lba);
-DEFINE_IPL_ATTR_RW(dump_fcp, device, "0.0.%04llx\n", "0.0.%llx\n",
- dump_block_fcp->ipl_info.fcp.devno);
-
-static struct attribute *dump_fcp_attrs[] = {
- &sys_dump_fcp_device_attr.attr,
- &sys_dump_fcp_wwpn_attr.attr,
- &sys_dump_fcp_lun_attr.attr,
- &sys_dump_fcp_bootprog_attr.attr,
- &sys_dump_fcp_br_lba_attr.attr,
- NULL,
-};
-
-static struct attribute_group dump_fcp_attr_group = {
- .name = IPL_FCP_STR,
- .attrs = dump_fcp_attrs,
-};
-
-/* CCW dump device attributes */
-
-DEFINE_IPL_ATTR_RW(dump_ccw, device, "0.0.%04llx\n", "0.0.%llx\n",
- dump_block_ccw->ipl_info.ccw.devno);
-
-static struct attribute *dump_ccw_attrs[] = {
- &sys_dump_ccw_device_attr.attr,
- NULL,
-};
-
-static struct attribute_group dump_ccw_attr_group = {
- .name = IPL_CCW_STR,
- .attrs = dump_ccw_attrs,
-};
-
-/* dump type */
-
-static int dump_set_type(enum dump_type type)
-{
- if (!(dump_capabilities & type))
- return -EINVAL;
- switch(type) {
- case DUMP_TYPE_CCW:
- if (MACHINE_IS_VM)
- dump_method = DUMP_METHOD_CCW_VM;
- else if (diag308_set_works)
- dump_method = DUMP_METHOD_CCW_DIAG;
- else
- dump_method = DUMP_METHOD_CCW_CIO;
- break;
- case DUMP_TYPE_FCP:
- dump_method = DUMP_METHOD_FCP_DIAG;
- break;
- default:
- dump_method = DUMP_METHOD_NONE;
- }
- dump_type = type;
- return 0;
-}
-
-static ssize_t dump_type_show(struct kset *kset, char *page)
-{
- return sprintf(page, "%s\n", dump_type_str(dump_type));
-}
-
-static ssize_t dump_type_store(struct kset *kset, const char *buf,
- size_t len)
-{
- int rc = -EINVAL;
-
- if (strncmp(buf, DUMP_NONE_STR, strlen(DUMP_NONE_STR)) == 0)
- rc = dump_set_type(DUMP_TYPE_NONE);
- else if (strncmp(buf, DUMP_CCW_STR, strlen(DUMP_CCW_STR)) == 0)
- rc = dump_set_type(DUMP_TYPE_CCW);
- else if (strncmp(buf, DUMP_FCP_STR, strlen(DUMP_FCP_STR)) == 0)
- rc = dump_set_type(DUMP_TYPE_FCP);
- return (rc != 0) ? rc : len;
-}
-
-static struct subsys_attribute dump_type_attr =
- __ATTR(dump_type, 0644, dump_type_show, dump_type_store);
-
-static decl_subsys(dump, NULL, NULL);
-
-/*
- * Shutdown actions section
- */
-
-static decl_subsys(shutdown_actions, NULL, NULL);
-
-/* on panic */
-
-static ssize_t on_panic_show(struct kset *kset, char *page)
-{
- return sprintf(page, "%s\n", shutdown_action_str(on_panic_action));
-}
-
-static ssize_t on_panic_store(struct kset *kset, const char *buf,
- size_t len)
-{
- if (strncmp(buf, SHUTDOWN_REIPL_STR, strlen(SHUTDOWN_REIPL_STR)) == 0)
- on_panic_action = SHUTDOWN_REIPL;
- else if (strncmp(buf, SHUTDOWN_DUMP_STR,
- strlen(SHUTDOWN_DUMP_STR)) == 0)
- on_panic_action = SHUTDOWN_DUMP;
- else if (strncmp(buf, SHUTDOWN_STOP_STR,
- strlen(SHUTDOWN_STOP_STR)) == 0)
- on_panic_action = SHUTDOWN_STOP;
- else
- return -EINVAL;
-
- return len;
-}
-
-static struct subsys_attribute on_panic_attr =
- __ATTR(on_panic, 0644, on_panic_show, on_panic_store);
-
-void do_reipl(void)
+void reipl_run(struct shutdown_trigger *trigger)
{
struct ccw_dev_id devid;
static char buf[100];
@@ -771,110 +692,18 @@ void do_reipl(void)
default:
break;
}
- signal_processor(smp_processor_id(), sigp_stop_and_store_status);
}
-static void do_dump(void)
+static void __init reipl_probe(void)
{
- struct ccw_dev_id devid;
- static char buf[100];
+ void *buffer;
- switch (dump_method) {
- case DUMP_METHOD_CCW_CIO:
- smp_send_stop();
- devid.devno = dump_block_ccw->ipl_info.ccw.devno;
- devid.ssid = 0;
- reipl_ccw_dev(&devid);
- break;
- case DUMP_METHOD_CCW_VM:
- smp_send_stop();
- sprintf(buf, "STORE STATUS");
- __cpcmd(buf, NULL, 0, NULL);
- sprintf(buf, "IPL %X", dump_block_ccw->ipl_info.ccw.devno);
- __cpcmd(buf, NULL, 0, NULL);
- break;
- case DUMP_METHOD_CCW_DIAG:
- diag308(DIAG308_SET, dump_block_ccw);
- diag308(DIAG308_DUMP, NULL);
- break;
- case DUMP_METHOD_FCP_DIAG:
- diag308(DIAG308_SET, dump_block_fcp);
- diag308(DIAG308_DUMP, NULL);
- break;
- case DUMP_METHOD_NONE:
- default:
+ buffer = (void *) get_zeroed_page(GFP_KERNEL);
+ if (!buffer)
return;
- }
- printk(KERN_EMERG "Dump failed!\n");
-}
-
-/* init functions */
-
-static int __init ipl_register_fcp_files(void)
-{
- int rc;
-
- rc = sysfs_create_group(&ipl_subsys.kobj,
- &ipl_fcp_attr_group);
- if (rc)
- goto out;
- rc = sysfs_create_bin_file(&ipl_subsys.kobj,
- &ipl_parameter_attr);
- if (rc)
- goto out_ipl_parm;
- rc = sysfs_create_bin_file(&ipl_subsys.kobj,
- &ipl_scp_data_attr);
- if (!rc)
- goto out;
-
- sysfs_remove_bin_file(&ipl_subsys.kobj, &ipl_parameter_attr);
-
-out_ipl_parm:
- sysfs_remove_group(&ipl_subsys.kobj, &ipl_fcp_attr_group);
-out:
- return rc;
-}
-
-static int __init ipl_init(void)
-{
- int rc;
-
- rc = firmware_register(&ipl_subsys);
- if (rc)
- return rc;
- switch (ipl_info.type) {
- case IPL_TYPE_CCW:
- rc = sysfs_create_group(&ipl_subsys.kobj,
- &ipl_ccw_attr_group);
- break;
- case IPL_TYPE_FCP:
- case IPL_TYPE_FCP_DUMP:
- rc = ipl_register_fcp_files();
- break;
- case IPL_TYPE_NSS:
- rc = sysfs_create_group(&ipl_subsys.kobj,
- &ipl_nss_attr_group);
- break;
- default:
- rc = sysfs_create_group(&ipl_subsys.kobj,
- &ipl_unknown_attr_group);
- break;
- }
- if (rc)
- firmware_unregister(&ipl_subsys);
- return rc;
-}
-
-static void __init reipl_probe(void)
-{
- void *buffer;
-
- buffer = (void *) get_zeroed_page(GFP_KERNEL);
- if (!buffer)
- return;
- if (diag308(DIAG308_STORE, buffer) == DIAG308_RC_OK)
- diag308_set_works = 1;
- free_page((unsigned long)buffer);
+ if (diag308(DIAG308_STORE, buffer) == DIAG308_RC_OK)
+ diag308_set_works = 1;
+ free_page((unsigned long)buffer);
}
static int __init reipl_nss_init(void)
@@ -954,7 +783,7 @@ static int __init reipl_fcp_init(void)
return 0;
}
-static int __init reipl_init(void)
+static int reipl_init(void)
{
int rc;
@@ -981,6 +810,136 @@ static int __init reipl_init(void)
return 0;
}
+static struct shutdown_action reipl_action = {SHUTDOWN_ACTION_REIPL_STR,
+ reipl_run, reipl_init};
+
+/*
+ * dump shutdown action: Dump Linux on shutdown.
+ */
+
+/* FCP dump device attributes */
+
+DEFINE_IPL_ATTR_RW(dump_fcp, wwpn, "0x%016llx\n", "%016llx\n",
+ dump_block_fcp->ipl_info.fcp.wwpn);
+DEFINE_IPL_ATTR_RW(dump_fcp, lun, "0x%016llx\n", "%016llx\n",
+ dump_block_fcp->ipl_info.fcp.lun);
+DEFINE_IPL_ATTR_RW(dump_fcp, bootprog, "%lld\n", "%lld\n",
+ dump_block_fcp->ipl_info.fcp.bootprog);
+DEFINE_IPL_ATTR_RW(dump_fcp, br_lba, "%lld\n", "%lld\n",
+ dump_block_fcp->ipl_info.fcp.br_lba);
+DEFINE_IPL_ATTR_RW(dump_fcp, device, "0.0.%04llx\n", "0.0.%llx\n",
+ dump_block_fcp->ipl_info.fcp.devno);
+
+static struct attribute *dump_fcp_attrs[] = {
+ &sys_dump_fcp_device_attr.attr,
+ &sys_dump_fcp_wwpn_attr.attr,
+ &sys_dump_fcp_lun_attr.attr,
+ &sys_dump_fcp_bootprog_attr.attr,
+ &sys_dump_fcp_br_lba_attr.attr,
+ NULL,
+};
+
+static struct attribute_group dump_fcp_attr_group = {
+ .name = IPL_FCP_STR,
+ .attrs = dump_fcp_attrs,
+};
+
+/* CCW dump device attributes */
+
+DEFINE_IPL_ATTR_RW(dump_ccw, device, "0.0.%04llx\n", "0.0.%llx\n",
+ dump_block_ccw->ipl_info.ccw.devno);
+
+static struct attribute *dump_ccw_attrs[] = {
+ &sys_dump_ccw_device_attr.attr,
+ NULL,
+};
+
+static struct attribute_group dump_ccw_attr_group = {
+ .name = IPL_CCW_STR,
+ .attrs = dump_ccw_attrs,
+};
+
+/* dump type */
+
+static int dump_set_type(enum dump_type type)
+{
+ if (!(dump_capabilities & type))
+ return -EINVAL;
+ switch (type) {
+ case DUMP_TYPE_CCW:
+ if (MACHINE_IS_VM)
+ dump_method = DUMP_METHOD_CCW_VM;
+ else
+ dump_method = DUMP_METHOD_CCW_CIO;
+ break;
+ case DUMP_TYPE_FCP:
+ dump_method = DUMP_METHOD_FCP_DIAG;
+ break;
+ default:
+ dump_method = DUMP_METHOD_NONE;
+ }
+ dump_type = type;
+ return 0;
+}
+
+static ssize_t dump_type_show(struct kset *kset, char *page)
+{
+ return sprintf(page, "%s\n", dump_type_str(dump_type));
+}
+
+static ssize_t dump_type_store(struct kset *kset, const char *buf,
+ size_t len)
+{
+ int rc = -EINVAL;
+
+ if (strncmp(buf, DUMP_NONE_STR, strlen(DUMP_NONE_STR)) == 0)
+ rc = dump_set_type(DUMP_TYPE_NONE);
+ else if (strncmp(buf, DUMP_CCW_STR, strlen(DUMP_CCW_STR)) == 0)
+ rc = dump_set_type(DUMP_TYPE_CCW);
+ else if (strncmp(buf, DUMP_FCP_STR, strlen(DUMP_FCP_STR)) == 0)
+ rc = dump_set_type(DUMP_TYPE_FCP);
+ return (rc != 0) ? rc : len;
+}
+
+static struct subsys_attribute dump_type_attr =
+ __ATTR(dump_type, 0644, dump_type_show, dump_type_store);
+
+static decl_subsys(dump, NULL, NULL);
+
+static void dump_run(struct shutdown_trigger *trigger)
+{
+ struct ccw_dev_id devid;
+ static char buf[100];
+
+ switch (dump_method) {
+ case DUMP_METHOD_CCW_CIO:
+ smp_send_stop();
+ devid.devno = dump_block_ccw->ipl_info.ccw.devno;
+ devid.ssid = 0;
+ reipl_ccw_dev(&devid);
+ break;
+ case DUMP_METHOD_CCW_VM:
+ smp_send_stop();
+ sprintf(buf, "STORE STATUS");
+ __cpcmd(buf, NULL, 0, NULL);
+ sprintf(buf, "IPL %X", dump_block_ccw->ipl_info.ccw.devno);
+ __cpcmd(buf, NULL, 0, NULL);
+ break;
+ case DUMP_METHOD_CCW_DIAG:
+ diag308(DIAG308_SET, dump_block_ccw);
+ diag308(DIAG308_DUMP, NULL);
+ break;
+ case DUMP_METHOD_FCP_DIAG:
+ diag308(DIAG308_SET, dump_block_fcp);
+ diag308(DIAG308_DUMP, NULL);
+ break;
+ case DUMP_METHOD_NONE:
+ default:
+ return;
+ }
+ printk(KERN_EMERG "Dump failed!\n");
+}
+
static int __init dump_ccw_init(void)
{
int rc;
@@ -1026,24 +985,7 @@ static int __init dump_fcp_init(void)
return 0;
}
-#define SHUTDOWN_ON_PANIC_PRIO 0
-
-static int shutdown_on_panic_notify(struct notifier_block *self,
- unsigned long event, void *data)
-{
- if (on_panic_action == SHUTDOWN_DUMP)
- do_dump();
- else if (on_panic_action == SHUTDOWN_REIPL)
- do_reipl();
- return NOTIFY_OK;
-}
-
-static struct notifier_block shutdown_on_panic_nb = {
- .notifier_call = shutdown_on_panic_notify,
- .priority = SHUTDOWN_ON_PANIC_PRIO
-};
-
-static int __init dump_init(void)
+static int dump_init(void)
{
int rc;
@@ -1065,46 +1007,362 @@ static int __init dump_init(void)
return 0;
}
-static int __init shutdown_actions_init(void)
+static struct shutdown_action dump_action = {SHUTDOWN_ACTION_DUMP_STR,
+ dump_run, dump_init};
+
+/*
+ * vmcmd shutdown action: Trigger vm command on shutdown.
+ */
+
+static char vmcmd_on_reboot[128];
+static char vmcmd_on_panic[128];
+static char vmcmd_on_halt[128];
+static char vmcmd_on_poff[128];
+
+DEFINE_IPL_ATTR_STR_RW(vmcmd, on_reboot, "%s\n", "%s\n", vmcmd_on_reboot);
+DEFINE_IPL_ATTR_STR_RW(vmcmd, on_panic, "%s\n", "%s\n", vmcmd_on_panic);
+DEFINE_IPL_ATTR_STR_RW(vmcmd, on_halt, "%s\n", "%s\n", vmcmd_on_halt);
+DEFINE_IPL_ATTR_STR_RW(vmcmd, on_poff, "%s\n", "%s\n", vmcmd_on_poff);
+
+static struct attribute *vmcmd_attrs[] = {
+ &sys_vmcmd_on_reboot_attr.attr,
+ &sys_vmcmd_on_panic_attr.attr,
+ &sys_vmcmd_on_halt_attr.attr,
+ &sys_vmcmd_on_poff_attr.attr,
+ NULL,
+};
+
+static struct attribute_group vmcmd_attr_group = {
+ .attrs = vmcmd_attrs,
+};
+
+static decl_subsys(vmcmd, NULL, NULL);
+
+static void vmcmd_run(struct shutdown_trigger *trigger)
+{
+ char *cmd, *next_cmd;
+
+ if (strcmp(trigger->name, ON_REIPL_STR) == 0)
+ cmd = vmcmd_on_reboot;
+ else if (strcmp(trigger->name, ON_PANIC_STR) == 0)
+ cmd = vmcmd_on_panic;
+ else if (strcmp(trigger->name, ON_HALT_STR) == 0)
+ cmd = vmcmd_on_halt;
+ else if (strcmp(trigger->name, ON_POFF_STR) == 0)
+ cmd = vmcmd_on_poff;
+ else
+ return;
+
+ if (strlen(cmd) == 0)
+ return;
+ do {
+ next_cmd = strchr(cmd, '\n');
+ if (next_cmd) {
+ next_cmd[0] = 0;
+ next_cmd += 1;
+ }
+ __cpcmd(cmd, NULL, 0, NULL);
+ cmd = next_cmd;
+ } while (cmd != NULL);
+}
+
+static int vmcmd_init(void)
{
int rc;
- rc = firmware_register(&shutdown_actions_subsys);
+ if (!MACHINE_IS_VM)
+ return -ENOTSUPP;
+ rc = firmware_register(&vmcmd_subsys);
if (rc)
return rc;
- rc = subsys_create_file(&shutdown_actions_subsys, &on_panic_attr);
- if (rc) {
- firmware_unregister(&shutdown_actions_subsys);
- return rc;
+ return sysfs_create_group(&vmcmd_subsys.kobj, &vmcmd_attr_group);
+}
+
+static struct shutdown_action vmcmd_action = {SHUTDOWN_ACTION_VMCMD_STR,
+ vmcmd_run, vmcmd_init};
+
+/*
+ * stop shutdown action: Stop Linux on shutdown.
+ */
+
+static void stop_run(struct shutdown_trigger *trigger)
+{
+ signal_processor(smp_processor_id(), sigp_stop_and_store_status);
+ for (;;);
+}
+
+static struct shutdown_action stop_action = {SHUTDOWN_ACTION_STOP_STR,
+ stop_run, NULL};
+
+/* action list */
+
+static struct shutdown_action *shutdown_actions_list[] = {
+ &ipl_action, &reipl_action, &dump_action, &vmcmd_action, &stop_action};
+#define SHUTDOWN_ACTIONS_COUNT (sizeof(shutdown_actions_list) / sizeof(void *))
+
+/*
+ * Trigger section
+ */
+
+static decl_subsys(shutdown_actions, NULL, NULL);
+
+static int set_trigger(const char *buf, struct shutdown_trigger *trigger,
+ size_t len)
+{
+ int i;
+
+ for (i = 0; i < SHUTDOWN_ACTIONS_COUNT; i++) {
+ if (!shutdown_actions_list[i])
+ continue;
+ if (strncmp(buf, shutdown_actions_list[i]->name,
+ strlen(shutdown_actions_list[i]->name)) == 0) {
+ trigger->action = shutdown_actions_list[i];
+ return len;
+ }
}
- atomic_notifier_chain_register(&panic_notifier_list,
- &shutdown_on_panic_nb);
- return 0;
+ return -EINVAL;
}
-static int __init s390_ipl_init(void)
+/* on reipl */
+
+static struct shutdown_trigger on_reboot_trigger = {ON_REIPL_STR,
+ &reipl_action};
+
+static ssize_t on_reboot_show(struct kset *kset, char *page)
{
- int rc;
+ return sprintf(page, "%s\n", on_reboot_trigger.action->name);
+}
+
+static ssize_t on_reboot_store(struct kset *kset, const char *buf, size_t len)
+{
+ return set_trigger(buf, &on_reboot_trigger, len);
+}
+
+static struct subsys_attribute on_reboot_attr =
+ __ATTR(on_reboot, 0644, on_reboot_show, on_reboot_store);
+
+static void do_machine_restart(char *__unused)
+{
+ smp_send_stop();
+ on_reboot_trigger.action->fn(&on_reboot_trigger);
+ reipl_run(NULL);
+}
+void (*_machine_restart)(char *command) = do_machine_restart;
+
+/* on panic */
+
+static struct shutdown_trigger on_panic_trigger = {ON_PANIC_STR, &stop_action};
+
+static ssize_t on_panic_show(struct kset *kset, char *page)
+{
+ return sprintf(page, "%s\n", on_panic_trigger.action->name);
+}
+
+static ssize_t on_panic_store(struct kset *kset, const char *buf,
+ size_t len)
+{
+ return set_trigger(buf, &on_panic_trigger, len);
+}
+
+static struct subsys_attribute on_panic_attr =
+ __ATTR(on_panic, 0644, on_panic_show, on_panic_store);
+
+static void do_panic(void)
+{
+ on_panic_trigger.action->fn(&on_panic_trigger);
+ stop_run(&on_panic_trigger);
+}
+
+/* on halt */
+
+static struct shutdown_trigger on_halt_trigger = {ON_HALT_STR, &stop_action};
+
+static ssize_t on_halt_show(struct kset *kset, char *page)
+{
+ return sprintf(page, "%s\n", on_halt_trigger.action->name);
+}
+
+static ssize_t on_halt_store(struct kset *kset, const char *buf, size_t len)
+{
+ return set_trigger(buf, &on_halt_trigger, len);
+}
+
+static struct subsys_attribute on_halt_attr =
+ __ATTR(on_halt, 0644, on_halt_show, on_halt_store);
- sclp_get_ipl_info(&sclp_ipl_info);
+
+static void do_machine_halt(void)
+{
+ smp_send_stop();
+ on_halt_trigger.action->fn(&on_halt_trigger);
+ stop_run(&on_halt_trigger);
+}
+void (*_machine_halt)(void) = do_machine_halt;
+
+/* on power off */
+
+static struct shutdown_trigger on_poff_trigger = {ON_POFF_STR, &stop_action};
+
+static ssize_t on_poff_show(struct kset *kset, char *page)
+{
+ return sprintf(page, "%s\n", on_poff_trigger.action->name);
+}
+
+static ssize_t on_poff_store(struct kset *kset, const char *buf, size_t len)
+{
+ return set_trigger(buf, &on_poff_trigger, len);
+}
+
+static struct subsys_attribute on_poff_attr =
+ __ATTR(on_poff, 0644, on_poff_show, on_poff_store);
+
+
+static void do_machine_power_off(void)
+{
+ smp_send_stop();
+ on_poff_trigger.action->fn(&on_poff_trigger);
+ stop_run(&on_poff_trigger);
+}
+void (*_machine_power_off)(void) = do_machine_power_off;
+
+static void __init shutdown_triggers_init(void)
+{
+ if (firmware_register(&shutdown_actions_subsys))
+ goto fail;
+ if (subsys_create_file(&shutdown_actions_subsys, &on_reboot_attr))
+ goto fail;
+ if (subsys_create_file(&shutdown_actions_subsys, &on_panic_attr))
+ goto fail;
+ if (subsys_create_file(&shutdown_actions_subsys, &on_halt_attr))
+ goto fail;
+ if (subsys_create_file(&shutdown_actions_subsys, &on_poff_attr))
+ goto fail;
+
+ return;
+fail:
+ panic("shutdown_triggers_init failed\n");
+}
+
+static void __init shutdown_actions_init(void)
+{
+ int i;
+
+ for (i = 0; i < SHUTDOWN_ACTIONS_COUNT; i++) {
+ if (!shutdown_actions_list[i]->init)
+ continue;
+ if (shutdown_actions_list[i]->init())
+ shutdown_actions_list[i] = NULL;
+ }
+}
+
+static int __init s390_ipl_init(void)
+{
reipl_probe();
- rc = ipl_init();
- if (rc)
- return rc;
- rc = reipl_init();
- if (rc)
- return rc;
- rc = dump_init();
- if (rc)
- return rc;
- rc = shutdown_actions_init();
- if (rc)
- return rc;
+ shutdown_actions_init();
+ shutdown_triggers_init();
return 0;
}
__initcall(s390_ipl_init);
+static void __init strncpy_skip_quote(char *dst, char *src, int n)
+{
+ int sx, dx;
+
+ dx = 0;
+ for (sx = 0; src[sx] != 0; sx++) {
+ if (src[sx] == '"')
+ continue;
+ dst[dx++] = src[sx];
+ if (dx >= n)
+ break;
+ }
+}
+
+static int __init vmcmd_on_reboot_setup(char *str)
+{
+ if (!MACHINE_IS_VM)
+ return 1;
+ strncpy_skip_quote(vmcmd_on_reboot, str, 127);
+ vmcmd_on_reboot[127] = 0;
+ on_panic_trigger.action = &vmcmd_action;
+ return 1;
+}
+__setup("vmreboot=", vmcmd_on_reboot_setup);
+
+static int __init vmcmd_on_panic_setup(char *str)
+{
+ if (!MACHINE_IS_VM)
+ return 1;
+ strncpy_skip_quote(vmcmd_on_panic, str, 127);
+ vmcmd_on_panic[127] = 0;
+ on_panic_trigger.action = &vmcmd_action;
+ return 1;
+}
+__setup("vmpanic=", vmcmd_on_panic_setup);
+
+static int __init vmcmd_on_halt_setup(char *str)
+{
+ if (!MACHINE_IS_VM)
+ return 1;
+ strncpy_skip_quote(vmcmd_on_halt, str, 127);
+ vmcmd_on_halt[127] = 0;
+ on_halt_trigger.action = &vmcmd_action;
+ return 1;
+}
+__setup("vmhalt=", vmcmd_on_halt_setup);
+
+static int __init vmcmd_on_poff_setup(char *str)
+{
+ if (!MACHINE_IS_VM)
+ return 1;
+ strncpy_skip_quote(vmcmd_on_poff, str, 127);
+ vmcmd_on_poff[127] = 0;
+ on_poff_trigger.action = &vmcmd_action;
+ return 1;
+}
+__setup("vmpoff=", vmcmd_on_poff_setup);
+
+static int on_panic_notify(struct notifier_block *self,
+ unsigned long event, void *data)
+{
+ do_panic();
+ return NOTIFY_OK;
+}
+
+static struct notifier_block on_panic_nb = {
+ .notifier_call = on_panic_notify,
+ .priority = 0,
+};
+
+void __init setup_ipl(void)
+{
+ ipl_info.type = get_ipl_type();
+ switch (ipl_info.type) {
+ case IPL_TYPE_CCW:
+ ipl_info.data.ccw.dev_id.devno = ipl_devno;
+ ipl_info.data.ccw.dev_id.ssid = 0;
+ break;
+ case IPL_TYPE_FCP:
+ case IPL_TYPE_FCP_DUMP:
+ ipl_info.data.fcp.dev_id.devno =
+ IPL_PARMBLOCK_START->ipl_info.fcp.devno;
+ ipl_info.data.fcp.dev_id.ssid = 0;
+ ipl_info.data.fcp.wwpn = IPL_PARMBLOCK_START->ipl_info.fcp.wwpn;
+ ipl_info.data.fcp.lun = IPL_PARMBLOCK_START->ipl_info.fcp.lun;
+ break;
+ case IPL_TYPE_NSS:
+ strncpy(ipl_info.data.nss.name, kernel_nss_name,
+ sizeof(ipl_info.data.nss.name));
+ break;
+ case IPL_TYPE_UNKNOWN:
+ default:
+ /* We have no info to copy */
+ break;
+ }
+ atomic_notifier_chain_register(&panic_notifier_list, &on_panic_nb);
+}
+
void __init ipl_save_parameters(void)
{
struct cio_iplinfo iplinfo;
@@ -1185,3 +1443,4 @@ void s390_reset_system(void)
do_reset_calls();
}
+
Index: quilt-2.6/arch/s390/kernel/setup.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/setup.c
+++ quilt-2.6/arch/s390/kernel/setup.c
@@ -126,75 +126,6 @@ void __cpuinit cpu_init(void)
}
/*
- * VM halt and poweroff setup routines
- */
-char vmhalt_cmd[128] = "";
-char vmpoff_cmd[128] = "";
-static char vmpanic_cmd[128] = "";
-
-static void strncpy_skip_quote(char *dst, char *src, int n)
-{
- int sx, dx;
-
- dx = 0;
- for (sx = 0; src[sx] != 0; sx++) {
- if (src[sx] == '"') continue;
- dst[dx++] = src[sx];
- if (dx >= n) break;
- }
-}
-
-static int __init vmhalt_setup(char *str)
-{
- strncpy_skip_quote(vmhalt_cmd, str, 127);
- vmhalt_cmd[127] = 0;
- return 1;
-}
-
-__setup("vmhalt=", vmhalt_setup);
-
-static int __init vmpoff_setup(char *str)
-{
- strncpy_skip_quote(vmpoff_cmd, str, 127);
- vmpoff_cmd[127] = 0;
- return 1;
-}
-
-__setup("vmpoff=", vmpoff_setup);
-
-static int vmpanic_notify(struct notifier_block *self, unsigned long event,
- void *data)
-{
- if (MACHINE_IS_VM && strlen(vmpanic_cmd) > 0)
- cpcmd(vmpanic_cmd, NULL, 0, NULL);
-
- return NOTIFY_OK;
-}
-
-#define PANIC_PRI_VMPANIC 0
-
-static struct notifier_block vmpanic_nb = {
- .notifier_call = vmpanic_notify,
- .priority = PANIC_PRI_VMPANIC
-};
-
-static int __init vmpanic_setup(char *str)
-{
- static int register_done __initdata = 0;
-
- strncpy_skip_quote(vmpanic_cmd, str, 127);
- vmpanic_cmd[127] = 0;
- if (!register_done) {
- register_done = 1;
- atomic_notifier_chain_register(&panic_notifier_list,
- &vmpanic_nb);
- }
- return 1;
-}
-
-__setup("vmpanic=", vmpanic_setup);
-
-/*
* condev= and conmode= setup parameter.
*/
@@ -308,38 +239,6 @@ static void __init setup_zfcpdump(unsign
static inline void setup_zfcpdump(unsigned int console_devno) {}
#endif /* CONFIG_ZFCPDUMP */
-#ifdef CONFIG_SMP
-void (*_machine_restart)(char *command) = machine_restart_smp;
-void (*_machine_halt)(void) = machine_halt_smp;
-void (*_machine_power_off)(void) = machine_power_off_smp;
-#else
-/*
- * Reboot, halt and power_off routines for non SMP.
- */
-static void do_machine_restart_nonsmp(char * __unused)
-{
- do_reipl();
-}
-
-static void do_machine_halt_nonsmp(void)
-{
- if (MACHINE_IS_VM && strlen(vmhalt_cmd) > 0)
- __cpcmd(vmhalt_cmd, NULL, 0, NULL);
- signal_processor(smp_processor_id(), sigp_stop_and_store_status);
-}
-
-static void do_machine_power_off_nonsmp(void)
-{
- if (MACHINE_IS_VM && strlen(vmpoff_cmd) > 0)
- __cpcmd(vmpoff_cmd, NULL, 0, NULL);
- signal_processor(smp_processor_id(), sigp_stop_and_store_status);
-}
-
-void (*_machine_restart)(char *command) = do_machine_restart_nonsmp;
-void (*_machine_halt)(void) = do_machine_halt_nonsmp;
-void (*_machine_power_off)(void) = do_machine_power_off_nonsmp;
-#endif
-
/*
* Reboot, halt and power_off stubs. They just call _machine_restart,
* _machine_halt or _machine_power_off.
@@ -913,7 +812,7 @@ setup_arch(char **cmdline_p)
parse_early_param();
- setup_ipl_info();
+ setup_ipl();
setup_memory_end();
setup_addressing_mode();
setup_memory();
Index: quilt-2.6/arch/s390/kernel/smp.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/smp.c
+++ quilt-2.6/arch/s390/kernel/smp.c
@@ -234,33 +234,6 @@ void smp_send_stop(void)
}
/*
- * Reboot, halt and power_off routines for SMP.
- */
-void machine_restart_smp(char *__unused)
-{
- smp_send_stop();
- do_reipl();
-}
-
-void machine_halt_smp(void)
-{
- smp_send_stop();
- if (MACHINE_IS_VM && strlen(vmhalt_cmd) > 0)
- __cpcmd(vmhalt_cmd, NULL, 0, NULL);
- signal_processor(smp_processor_id(), sigp_stop_and_store_status);
- for (;;);
-}
-
-void machine_power_off_smp(void)
-{
- smp_send_stop();
- if (MACHINE_IS_VM && strlen(vmpoff_cmd) > 0)
- __cpcmd(vmpoff_cmd, NULL, 0, NULL);
- signal_processor(smp_processor_id(), sigp_stop_and_store_status);
- for (;;);
-}
-
-/*
* This is the main routine where commands issued by other
* cpus are handled.
*/
Index: quilt-2.6/include/asm-s390/ipl.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/ipl.h
+++ quilt-2.6/include/asm-s390/ipl.h
@@ -83,6 +83,8 @@ extern u32 dump_prefix_page;
extern unsigned int zfcpdump_prefix_array[];
extern void do_reipl(void);
+extern void do_halt(void);
+extern void do_poff(void);
extern void ipl_save_parameters(void);
enum {
@@ -118,7 +120,7 @@ struct ipl_info
};
extern struct ipl_info ipl_info;
-extern void setup_ipl_info(void);
+extern void setup_ipl(void);
/*
* DIAG 308 support
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 34/47] Load disabled wait psw instead of stopping cpu on halt.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (32 preceding siblings ...)
2007-12-20 15:19 ` [patch 33/47] kernel: Shutdown Actions Interface Martin Schwidefsky
@ 2007-12-20 15:19 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 35/47] use LIST_HEAD instead of LIST_HEAD_INIT Martin Schwidefsky
` (12 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:19 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Michael Holzheu, Martin Schwidefsky
[-- Attachment #1: 133-disabled-wait.diff --]
[-- Type: text/plain, Size: 984 bytes --]
From: Michael Holzheu <holzheu@de.ibm.com>
Signed-off-by: Michael Holzheu <holzheu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/ipl.c | 8 ++++++--
1 file changed, 6 insertions(+), 2 deletions(-)
Index: quilt-2.6/arch/s390/kernel/ipl.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/ipl.c
+++ quilt-2.6/arch/s390/kernel/ipl.c
@@ -1087,8 +1087,12 @@ static struct shutdown_action vmcmd_acti
static void stop_run(struct shutdown_trigger *trigger)
{
- signal_processor(smp_processor_id(), sigp_stop_and_store_status);
- for (;;);
+ if (strcmp(trigger->name, ON_PANIC_STR) == 0)
+ disabled_wait((unsigned long) __builtin_return_address(0));
+ else {
+ signal_processor(smp_processor_id(), sigp_stop);
+ for (;;);
+ }
}
static struct shutdown_action stop_action = {SHUTDOWN_ACTION_STOP_STR,
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 35/47] use LIST_HEAD instead of LIST_HEAD_INIT
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (33 preceding siblings ...)
2007-12-20 15:19 ` [patch 34/47] Load disabled wait psw instead of stopping cpu on halt Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 36/47] Allocate and free cpu lowcores and stacks when needed/possible Martin Schwidefsky
` (11 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Denis Cheng, Martin Schwidefsky
[-- Attachment #1: 134-list-head.diff --]
[-- Type: text/plain, Size: 3979 bytes --]
From: Denis Cheng <crquan@gmail.com>
single list_head variable initialized with LIST_HEAD_INIT could almost
always can be replaced with LIST_HEAD declaration, this shrinks the code
and looks better.
Signed-off-by: Denis Cheng <crquan@gmail.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/mm/extmem.c | 2 +-
drivers/s390/block/dcssblk.c | 2 +-
drivers/s390/char/raw3270.c | 4 ++--
drivers/s390/char/tape_core.c | 2 +-
drivers/s390/net/netiucv.c | 3 +--
drivers/s390/net/smsgiucv.c | 2 +-
6 files changed, 7 insertions(+), 8 deletions(-)
Index: quilt-2.6/arch/s390/mm/extmem.c
===================================================================
--- quilt-2.6.orig/arch/s390/mm/extmem.c
+++ quilt-2.6/arch/s390/mm/extmem.c
@@ -83,7 +83,7 @@ struct dcss_segment {
};
static DEFINE_MUTEX(dcss_lock);
-static struct list_head dcss_list = LIST_HEAD_INIT(dcss_list);
+static LIST_HEAD(dcss_list);
static char *segtype_string[] = { "SW", "EW", "SR", "ER", "SN", "EN", "SC",
"EW/EN-MIXED" };
Index: quilt-2.6/drivers/s390/block/dcssblk.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dcssblk.c
+++ quilt-2.6/drivers/s390/block/dcssblk.c
@@ -82,7 +82,7 @@ struct dcssblk_dev_info {
struct request_queue *dcssblk_queue;
};
-static struct list_head dcssblk_devices = LIST_HEAD_INIT(dcssblk_devices);
+static LIST_HEAD(dcssblk_devices);
static struct rw_semaphore dcssblk_devices_sem;
/*
Index: quilt-2.6/drivers/s390/char/raw3270.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/raw3270.c
+++ quilt-2.6/drivers/s390/char/raw3270.c
@@ -66,7 +66,7 @@ struct raw3270 {
static DEFINE_MUTEX(raw3270_mutex);
/* List of 3270 devices. */
-static struct list_head raw3270_devices = LIST_HEAD_INIT(raw3270_devices);
+static LIST_HEAD(raw3270_devices);
/*
* Flag to indicate if the driver has been registered. Some operations
@@ -1210,7 +1210,7 @@ struct raw3270_notifier {
void (*notifier)(int, int);
};
-static struct list_head raw3270_notifier = LIST_HEAD_INIT(raw3270_notifier);
+static LIST_HEAD(raw3270_notifier);
int raw3270_register_notifier(void (*notifier)(int, int))
{
Index: quilt-2.6/drivers/s390/char/tape_core.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/tape_core.c
+++ quilt-2.6/drivers/s390/char/tape_core.c
@@ -37,7 +37,7 @@ static void tape_long_busy_timeout(unsig
* we can assign the devices to minor numbers of the same major
* The list is protected by the rwlock
*/
-static struct list_head tape_device_list = LIST_HEAD_INIT(tape_device_list);
+static LIST_HEAD(tape_device_list);
static DEFINE_RWLOCK(tape_device_lock);
/*
Index: quilt-2.6/drivers/s390/net/netiucv.c
===================================================================
--- quilt-2.6.orig/drivers/s390/net/netiucv.c
+++ quilt-2.6/drivers/s390/net/netiucv.c
@@ -198,8 +198,7 @@ struct iucv_connection {
/**
* Linked list of all connection structs.
*/
-static struct list_head iucv_connection_list =
- LIST_HEAD_INIT(iucv_connection_list);
+static LIST_HEAD(iucv_connection_list);
static DEFINE_RWLOCK(iucv_connection_rwlock);
/**
Index: quilt-2.6/drivers/s390/net/smsgiucv.c
===================================================================
--- quilt-2.6.orig/drivers/s390/net/smsgiucv.c
+++ quilt-2.6/drivers/s390/net/smsgiucv.c
@@ -42,7 +42,7 @@ MODULE_DESCRIPTION ("Linux for S/390 IUC
static struct iucv_path *smsg_path;
static DEFINE_SPINLOCK(smsg_list_lock);
-static struct list_head smsg_list = LIST_HEAD_INIT(smsg_list);
+static LIST_HEAD(smsg_list);
static int smsg_path_pending(struct iucv_path *, u8 ipvmid[8], u8 ipuser[16]);
static void smsg_message_pending(struct iucv_path *, struct iucv_message *);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 36/47] Allocate and free cpu lowcores and stacks when needed/possible.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (34 preceding siblings ...)
2007-12-20 15:20 ` [patch 35/47] use LIST_HEAD instead of LIST_HEAD_INIT Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 37/47] Initialize sclp_ipl_info Martin Schwidefsky
` (10 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Heiko Carstens, Martin Schwidefsky
[-- Attachment #1: 135-cpu-lowcores.diff --]
[-- Type: text/plain, Size: 5397 bytes --]
From: Heiko Carstens <heiko.carstens@de.ibm.com>
No need to preallocate the per cpu lowcores and stacks.
Savings are 28-32k per offline cpu.
Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/smp.c | 106 +++++++++++++++++++++++++++++++++----------------
1 file changed, 72 insertions(+), 34 deletions(-)
Index: quilt-2.6/arch/s390/kernel/smp.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/smp.c
+++ quilt-2.6/arch/s390/kernel/smp.c
@@ -589,8 +589,72 @@ static void __init smp_create_idle(unsig
spin_lock_init(&(&per_cpu(s390_idle, cpu))->lock);
}
+static int __cpuinit smp_alloc_lowcore(int cpu)
+{
+ unsigned long async_stack, panic_stack;
+ struct _lowcore *lowcore;
+ int lc_order;
+
+ lc_order = sizeof(long) == 8 ? 1 : 0;
+ lowcore = (void *) __get_free_pages(GFP_KERNEL | GFP_DMA, lc_order);
+ if (!lowcore)
+ return -ENOMEM;
+ async_stack = __get_free_pages(GFP_KERNEL, ASYNC_ORDER);
+ if (!async_stack)
+ goto out_async_stack;
+ panic_stack = __get_free_page(GFP_KERNEL);
+ if (!panic_stack)
+ goto out_panic_stack;
+
+ *lowcore = S390_lowcore;
+ lowcore->async_stack = async_stack + ASYNC_SIZE;
+ lowcore->panic_stack = panic_stack + PAGE_SIZE;
+
+#ifndef CONFIG_64BIT
+ if (MACHINE_HAS_IEEE) {
+ unsigned long save_area;
+
+ save_area = get_zeroed_page(GFP_KERNEL);
+ if (!save_area)
+ goto out_save_area;
+ lowcore->extended_save_area_addr = (u32) save_area;
+ }
+#endif
+ lowcore_ptr[cpu] = lowcore;
+ return 0;
+
+#ifndef CONFIG_64BIT
+out_save_area:
+ free_page(panic_stack);
+#endif
+out_panic_stack:
+ free_pages(async_stack, ASYNC_ORDER);
+out_async_stack:
+ free_pages((unsigned long) lowcore, lc_order);
+ return -ENOMEM;
+}
+
+#ifdef CONFIG_HOTPLUG_CPU
+static void smp_free_lowcore(int cpu)
+{
+ struct _lowcore *lowcore;
+ int lc_order;
+
+ lc_order = sizeof(long) == 8 ? 1 : 0;
+ lowcore = lowcore_ptr[cpu];
+#ifndef CONFIG_64BIT
+ if (MACHINE_HAS_IEEE)
+ free_page((unsigned long) lowcore->extended_save_area_addr);
+#endif
+ free_page(lowcore->panic_stack - PAGE_SIZE);
+ free_pages(lowcore->async_stack - ASYNC_SIZE, ASYNC_ORDER);
+ free_pages((unsigned long) lowcore, lc_order);
+ lowcore_ptr[cpu] = NULL;
+}
+#endif /* CONFIG_HOTPLUG_CPU */
+
/* Upping and downing of CPUs */
-int __cpu_up(unsigned int cpu)
+int __cpuinit __cpu_up(unsigned int cpu)
{
struct task_struct *idle;
struct _lowcore *cpu_lowcore;
@@ -599,6 +663,8 @@ int __cpu_up(unsigned int cpu)
if (smp_cpu_state[cpu] != CPU_STATE_CONFIGURED)
return -EIO;
+ if (smp_alloc_lowcore(cpu))
+ return -ENOMEM;
ccode = signal_processor_p((__u32)(unsigned long)(lowcore_ptr[cpu]),
cpu, sigp_set_prefix);
@@ -613,6 +679,7 @@ int __cpu_up(unsigned int cpu)
cpu_lowcore = lowcore_ptr[cpu];
cpu_lowcore->kernel_stack = (unsigned long)
task_stack_page(idle) + THREAD_SIZE;
+ cpu_lowcore->thread_info = (unsigned long) task_thread_info(idle);
sf = (struct stack_frame *) (cpu_lowcore->kernel_stack
- sizeof(struct pt_regs)
- sizeof(struct stack_frame));
@@ -626,6 +693,8 @@ int __cpu_up(unsigned int cpu)
cpu_lowcore->percpu_offset = __per_cpu_offset[cpu];
cpu_lowcore->current_task = (unsigned long) idle;
cpu_lowcore->cpu_data.cpu_nr = cpu;
+ cpu_lowcore->softirq_pending = 0;
+ cpu_lowcore->ext_call_fast = 0;
eieio();
while (signal_processor(cpu, sigp_restart) == sigp_busy)
@@ -686,6 +755,7 @@ void __cpu_die(unsigned int cpu)
/* Wait until target cpu is down */
while (!smp_cpu_not_running(cpu))
cpu_relax();
+ smp_free_lowcore(cpu);
printk(KERN_INFO "Processor %d spun down\n", cpu);
}
@@ -699,15 +769,9 @@ void cpu_die(void)
#endif /* CONFIG_HOTPLUG_CPU */
-/*
- * Cycle through the processors and setup structures.
- */
-
void __init smp_prepare_cpus(unsigned int max_cpus)
{
- unsigned long stack;
unsigned int cpu;
- int i;
smp_detect_cpus();
@@ -715,35 +779,9 @@ void __init smp_prepare_cpus(unsigned in
if (register_external_interrupt(0x1201, do_ext_call_interrupt) != 0)
panic("Couldn't request external interrupt 0x1201");
memset(lowcore_ptr, 0, sizeof(lowcore_ptr));
- /*
- * Initialize prefix pages and stacks for all possible cpus
- */
print_cpu_info(&S390_lowcore.cpu_data);
+ smp_alloc_lowcore(smp_processor_id());
- for_each_possible_cpu(i) {
- lowcore_ptr[i] = (struct _lowcore *)
- __get_free_pages(GFP_KERNEL | GFP_DMA,
- sizeof(void*) == 8 ? 1 : 0);
- stack = __get_free_pages(GFP_KERNEL, ASYNC_ORDER);
- if (!lowcore_ptr[i] || !stack)
- panic("smp_boot_cpus failed to allocate memory\n");
-
- *(lowcore_ptr[i]) = S390_lowcore;
- lowcore_ptr[i]->async_stack = stack + ASYNC_SIZE;
- stack = __get_free_pages(GFP_KERNEL, 0);
- if (!stack)
- panic("smp_boot_cpus failed to allocate memory\n");
- lowcore_ptr[i]->panic_stack = stack + PAGE_SIZE;
-#ifndef CONFIG_64BIT
- if (MACHINE_HAS_IEEE) {
- lowcore_ptr[i]->extended_save_area_addr =
- (__u32) __get_free_pages(GFP_KERNEL, 0);
- if (!lowcore_ptr[i]->extended_save_area_addr)
- panic("smp_boot_cpus failed to "
- "allocate memory\n");
- }
-#endif
- }
#ifndef CONFIG_64BIT
if (MACHINE_HAS_IEEE)
ctl_set_bit(14, 29); /* enable extended save area */
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 37/47] Initialize sclp_ipl_info
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (35 preceding siblings ...)
2007-12-20 15:20 ` [patch 36/47] Allocate and free cpu lowcores and stacks when needed/possible Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 38/47] vmemmap: allocate struct pages before 1:1 mapping Martin Schwidefsky
` (9 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Michael Holzheu, Martin Schwidefsky
[-- Attachment #1: 136-sclp-ipl-info.diff --]
[-- Type: text/plain, Size: 847 bytes --]
From: Michael Holzheu <holzheu@de.ibm.com>
The sclp ipl information has not been initialized. Therefore the ipl loadparm
and the "has_dump" flag have not been set correctly.
Signed-off-by: Michael Holzheu <holzheu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/ipl.c | 1 +
1 file changed, 1 insertion(+)
Index: quilt-2.6/arch/s390/kernel/ipl.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/ipl.c
+++ quilt-2.6/arch/s390/kernel/ipl.c
@@ -1262,6 +1262,7 @@ static void __init shutdown_actions_init
static int __init s390_ipl_init(void)
{
reipl_probe();
+ sclp_get_ipl_info(&sclp_ipl_info);
shutdown_actions_init();
shutdown_triggers_init();
return 0;
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 38/47] vmemmap: allocate struct pages before 1:1 mapping
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (36 preceding siblings ...)
2007-12-20 15:20 ` [patch 37/47] Initialize sclp_ipl_info Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 39/47] Use diag308 subcodes 3 and 6 for reboot and dump when possible Martin Schwidefsky
` (8 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Christian Borntraeger, Martin Schwidefsky
[-- Attachment #1: 137-vmalloc.diff --]
[-- Type: text/plain, Size: 1350 bytes --]
From: Christian Borntraeger <borntraeger@de.ibm.com>
We have seen an oops in an OOM situation, where show_mem tried to
access the struct page of a dcss segment. The vmemmap code has
already created the 1:1 mapping but failed allocating the struct
pages. In the OOM case, show_mem now walks the memory. It uses
pfn_valid to detect if it may access the struct page. In the case
described above, the mapping was established and pfn_valid returned
true. As the struct pages were not allocated, the kernel oopsed.
We have to ensure that we have created the struct pages, before we
add a mapping pointing to the pages.
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/mm/vmem.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
Index: quilt-2.6/arch/s390/mm/vmem.c
===================================================================
--- quilt-2.6.orig/arch/s390/mm/vmem.c
+++ quilt-2.6/arch/s390/mm/vmem.c
@@ -236,10 +236,10 @@ static int vmem_add_mem(unsigned long st
{
int ret;
- ret = vmem_add_range(start, size);
+ ret = vmem_add_mem_map(start, size);
if (ret)
return ret;
- return vmem_add_mem_map(start, size);
+ return vmem_add_range(start, size);
}
/*
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 39/47] Use diag308 subcodes 3 and 6 for reboot and dump when possible.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (37 preceding siblings ...)
2007-12-20 15:20 ` [patch 38/47] vmemmap: allocate struct pages before 1:1 mapping Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 40/47] arch/s390/: Spelling fixes Martin Schwidefsky
` (7 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Michael Holzheu, Martin Schwidefsky
[-- Attachment #1: 138-diag308.diff --]
[-- Type: text/plain, Size: 3330 bytes --]
From: Michael Holzheu <holzheu@de.ibm.com>
This patch fixes a problem with the following scenario:
1. Linux booted from DASD "A"
2. Reboot from DASD "B" using "/sys/firmware/reipl/ccw/device"
3. Reboot DASD "B"
Without this patch in step 3 on newer s390 systems under LPAR instead of
DASD "B", DASD "A" will be booted. The reason is that in step 2 we use CCW
reipl and in step 3 we use DIAG308 (subcode 3) reipl. DIAG308 does not
notice the CCW reipl and still thinks that it has to reboot DASD "A".
Before applying this fix, ensure to have MCF RJ9967101E or z9 GA3 base driver
installed.
Signed-off-by: Michael Holzheu <holzheu@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/ipl.c | 14 ++++++++------
include/asm-s390/ipl.h | 4 ++++
2 files changed, 12 insertions(+), 6 deletions(-)
Index: quilt-2.6/arch/s390/kernel/ipl.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/ipl.c
+++ quilt-2.6/arch/s390/kernel/ipl.c
@@ -588,7 +588,9 @@ static int reipl_set_type(enum ipl_type
switch(type) {
case IPL_TYPE_CCW:
- if (MACHINE_IS_VM)
+ if (diag308_set_works)
+ reipl_method = REIPL_METHOD_CCW_DIAG;
+ else if (MACHINE_IS_VM)
reipl_method = REIPL_METHOD_CCW_VM;
else
reipl_method = REIPL_METHOD_CCW_CIO;
@@ -650,8 +652,6 @@ void reipl_run(struct shutdown_trigger *
switch (reipl_method) {
case REIPL_METHOD_CCW_CIO:
devid.devno = reipl_block_ccw->ipl_info.ccw.devno;
- if (ipl_info.type == IPL_TYPE_CCW && devid.devno == ipl_devno)
- diag308(DIAG308_IPL, NULL);
devid.ssid = 0;
reipl_ccw_dev(&devid);
break;
@@ -736,6 +736,7 @@ static int __init reipl_ccw_init(void)
reipl_block_ccw->hdr.version = IPL_PARM_BLOCK_VERSION;
reipl_block_ccw->hdr.blk0_len = IPL_PARM_BLK0_CCW_LEN;
reipl_block_ccw->hdr.pbt = DIAG308_IPL_TYPE_CCW;
+ reipl_block_ccw->hdr.flags = DIAG308_FLAGS_LP_VALID;
/* check if read scp info worked and set loadparm */
if (sclp_ipl_info.is_valid)
memcpy(reipl_block_ccw->ipl_info.ccw.load_param,
@@ -744,8 +745,7 @@ static int __init reipl_ccw_init(void)
/* read scp info failed: set empty loadparm (EBCDIC blanks) */
memset(reipl_block_ccw->ipl_info.ccw.load_param, 0x40,
LOADPARM_LEN);
- /* FIXME: check for diag308_set_works when enabling diag ccw reipl */
- if (!MACHINE_IS_VM)
+ if (!MACHINE_IS_VM && !diag308_set_works)
sys_reipl_ccw_loadparm_attr.attr.mode = S_IRUGO;
if (ipl_info.type == IPL_TYPE_CCW)
reipl_block_ccw->ipl_info.ccw.devno = ipl_devno;
@@ -867,7 +867,9 @@ static int dump_set_type(enum dump_type
return -EINVAL;
switch (type) {
case DUMP_TYPE_CCW:
- if (MACHINE_IS_VM)
+ if (diag308_set_works)
+ dump_method = DUMP_METHOD_CCW_DIAG;
+ else if (MACHINE_IS_VM)
dump_method = DUMP_METHOD_CCW_VM;
else
dump_method = DUMP_METHOD_CCW_CIO;
Index: quilt-2.6/include/asm-s390/ipl.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/ipl.h
+++ quilt-2.6/include/asm-s390/ipl.h
@@ -143,6 +143,10 @@ enum diag308_opt {
DIAG308_IPL_OPT_DUMP = 0x20,
};
+enum diag308_flags {
+ DIAG308_FLAGS_LP_VALID = 0x80,
+};
+
enum diag308_rc {
DIAG308_RC_OK = 1,
};
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 40/47] arch/s390/: Spelling fixes
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (38 preceding siblings ...)
2007-12-20 15:20 ` [patch 39/47] Use diag308 subcodes 3 and 6 for reboot and dump when possible Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 41/47] include/asm-s390/: " Martin Schwidefsky
` (6 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Joe Perches, Martin Schwidefsky
[-- Attachment #1: 139-spelling-arch.diff --]
[-- Type: text/plain, Size: 805 bytes --]
From: Joe Perches <joe@perches.com>
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/crypto/prng.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
Index: quilt-2.6/arch/s390/crypto/prng.c
===================================================================
--- quilt-2.6.orig/arch/s390/crypto/prng.c
+++ quilt-2.6/arch/s390/crypto/prng.c
@@ -90,7 +90,7 @@ static ssize_t prng_read(struct file *fi
int ret = 0;
int tmp;
- /* nbytes can be arbitrary long, we spilt it into chunks */
+ /* nbytes can be arbitrary length, we split it into chunks */
while (nbytes) {
/* same as in extract_entropy_user in random.c */
if (need_resched()) {
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 41/47] include/asm-s390/: Spelling fixes
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (39 preceding siblings ...)
2007-12-20 15:20 ` [patch 40/47] arch/s390/: Spelling fixes Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 42/47] drivers/s390/: " Martin Schwidefsky
` (5 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Joe Perches, Martin Schwidefsky
[-- Attachment #1: 140-spelling-include.diff --]
[-- Type: text/plain, Size: 2711 bytes --]
From: Joe Perches <joe@perches.com>
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
include/asm-s390/cio.h | 4 ++--
include/asm-s390/dasd.h | 2 +-
include/asm-s390/qdio.h | 2 +-
include/asm-s390/zcrypt.h | 2 +-
4 files changed, 5 insertions(+), 5 deletions(-)
Index: quilt-2.6/include/asm-s390/cio.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/cio.h
+++ quilt-2.6/include/asm-s390/cio.h
@@ -24,8 +24,8 @@
* @fmt: format
* @pfch: prefetch
* @isic: initial-status interruption control
- * @alcc: adress-limit checking control
- * @ssi: supress-suspended interruption
+ * @alcc: address-limit checking control
+ * @ssi: suppress-suspended interruption
* @zcc: zero condition code
* @ectl: extended control
* @pno: path not operational
Index: quilt-2.6/include/asm-s390/dasd.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/dasd.h
+++ quilt-2.6/include/asm-s390/dasd.h
@@ -105,7 +105,7 @@ typedef struct dasd_information_t {
} dasd_information_t;
/*
- * Read Subsystem Data - Perfomance Statistics
+ * Read Subsystem Data - Performance Statistics
*/
typedef struct dasd_rssd_perf_stats_t {
unsigned char invalid:1;
Index: quilt-2.6/include/asm-s390/qdio.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/qdio.h
+++ quilt-2.6/include/asm-s390/qdio.h
@@ -184,7 +184,7 @@ struct qdr {
#endif /* QDIO_32_BIT */
unsigned long qiba; /* queue-information-block address */
unsigned int res8; /* reserved */
- unsigned int qkey : 4; /* queue-informatio-block key */
+ unsigned int qkey : 4; /* queue-information-block key */
unsigned int res9 : 28; /* reserved */
/* union _qd {*/ /* why this? */
struct qdesfmt0 qdf0[126];
Index: quilt-2.6/include/asm-s390/zcrypt.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/zcrypt.h
+++ quilt-2.6/include/asm-s390/zcrypt.h
@@ -117,7 +117,7 @@ struct CPRBX {
unsigned char padx004[16 - sizeof (char *)];
unsigned char * req_extb; /* request extension block 'addr'*/
unsigned char padx005[16 - sizeof (char *)];
- unsigned char * rpl_extb; /* reply extension block 'addres'*/
+ unsigned char * rpl_extb; /* reply extension block 'address'*/
unsigned short ccp_rtcode; /* server return code */
unsigned short ccp_rscode; /* server reason code */
unsigned int mac_data_len; /* Mac Data Length */
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 42/47] drivers/s390/: Spelling fixes
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (40 preceding siblings ...)
2007-12-20 15:20 ` [patch 41/47] include/asm-s390/: " Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 43/47] Move NOTES and BUG_TABLE Martin Schwidefsky
` (4 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Joe Perches, Martin Schwidefsky
[-- Attachment #1: 141-spelling-drivers.diff --]
[-- Type: text/plain, Size: 3955 bytes --]
From: Joe Perches <joe@perches.com>
Signed-off-by: Joe Perches <joe@perches.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/block/dasd_3990_erp.c | 2 +-
drivers/s390/block/dasd_eckd.c | 2 +-
drivers/s390/char/sclp_rw.c | 2 +-
drivers/s390/char/tape_3590.c | 2 +-
drivers/s390/scsi/zfcp_erp.c | 2 +-
drivers/s390/scsi/zfcp_qdio.c | 2 +-
6 files changed, 6 insertions(+), 6 deletions(-)
Index: quilt-2.6/drivers/s390/block/dasd_3990_erp.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_3990_erp.c
+++ quilt-2.6/drivers/s390/block/dasd_3990_erp.c
@@ -2620,7 +2620,7 @@ dasd_3990_erp_handle_match_erp(struct da
* DASD_3990_ERP_ACTION
*
* DESCRIPTION
- * controll routine for 3990 erp actions.
+ * control routine for 3990 erp actions.
* Has to be called with the queue lock (namely the s390_irq_lock) acquired.
*
* PARAMETER
Index: quilt-2.6/drivers/s390/block/dasd_eckd.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_eckd.c
+++ quilt-2.6/drivers/s390/block/dasd_eckd.c
@@ -1542,7 +1542,7 @@ dasd_eckd_performance(struct dasd_device
prssdp = (struct dasd_psf_prssd_data *) cqr->data;
memset(prssdp, 0, sizeof (struct dasd_psf_prssd_data));
prssdp->order = PSF_ORDER_PRSSD;
- prssdp->suborder = 0x01; /* Perfomance Statistics */
+ prssdp->suborder = 0x01; /* Performance Statistics */
prssdp->varies[1] = 0x01; /* Perf Statistics for the Subsystem */
ccw = cqr->cpaddr;
Index: quilt-2.6/drivers/s390/char/sclp_rw.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/sclp_rw.c
+++ quilt-2.6/drivers/s390/char/sclp_rw.c
@@ -76,7 +76,7 @@ sclp_make_buffer(void *page, unsigned sh
}
/*
- * Return a pointer to the orignal page that has been used to create
+ * Return a pointer to the original page that has been used to create
* the buffer.
*/
void *
Index: quilt-2.6/drivers/s390/char/tape_3590.c
===================================================================
--- quilt-2.6.orig/drivers/s390/char/tape_3590.c
+++ quilt-2.6/drivers/s390/char/tape_3590.c
@@ -1495,7 +1495,7 @@ tape_3590_unit_check(struct tape_device
device->cdev->dev.bus_id);
return tape_3590_erp_basic(device, request, irb, -EPERM);
case 0x8013:
- PRINT_WARN("(%s): Another host has priviliged access to the "
+ PRINT_WARN("(%s): Another host has privileged access to the "
"tape device\n", device->cdev->dev.bus_id);
PRINT_WARN("(%s): To solve the problem unload the current "
"cartridge!\n", device->cdev->dev.bus_id);
Index: quilt-2.6/drivers/s390/scsi/zfcp_erp.c
===================================================================
--- quilt-2.6.orig/drivers/s390/scsi/zfcp_erp.c
+++ quilt-2.6/drivers/s390/scsi/zfcp_erp.c
@@ -1285,7 +1285,7 @@ zfcp_erp_strategy_do_action(struct zfcp_
* note: no lock in subsequent strategy routines
* (this allows these routine to call schedule, e.g.
* kmalloc with such flags or qdio_initialize & friends)
- * Note: in case of timeout, the seperate strategies will fail
+ * Note: in case of timeout, the separate strategies will fail
* anyhow. No need for a special action. Even worse, a nameserver
* failure would not wake up waiting ports without the call.
*/
Index: quilt-2.6/drivers/s390/scsi/zfcp_qdio.c
===================================================================
--- quilt-2.6.orig/drivers/s390/scsi/zfcp_qdio.c
+++ quilt-2.6/drivers/s390/scsi/zfcp_qdio.c
@@ -529,7 +529,7 @@ zfcp_qdio_sbals_wipe(struct zfcp_fsf_req
/**
- * zfcp_qdio_sbale_fill - set address and lenght in current SBALE
+ * zfcp_qdio_sbale_fill - set address and length in current SBALE
* on request_queue
*/
static void
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 43/47] Move NOTES and BUG_TABLE.
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (41 preceding siblings ...)
2007-12-20 15:20 ` [patch 42/47] drivers/s390/: " Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 44/47] single-step cleanup Martin Schwidefsky
` (3 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Martin Schwidefsky
[-- Attachment #1: 142-linker-script.diff --]
[-- Type: text/plain, Size: 1121 bytes --]
From: Martin Schwidefsky <schwidefsky@de.ibm.com>
Move the NOTES and BUG_TABLE section in the linker script to the
read-only sections right after the text section.
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/vmlinux.lds.S | 12 +++++++++---
1 file changed, 9 insertions(+), 3 deletions(-)
Index: quilt-2.6/arch/s390/kernel/vmlinux.lds.S
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/vmlinux.lds.S
+++ quilt-2.6/arch/s390/kernel/vmlinux.lds.S
@@ -17,6 +17,12 @@ ENTRY(_start)
jiffies = jiffies_64;
#endif
+PHDRS {
+ text PT_LOAD FLAGS(5); /* R_E */
+ data PT_LOAD FLAGS(7); /* RWE */
+ note PT_NOTE FLAGS(0); /* ___ */
+}
+
SECTIONS
{
. = 0x00000000;
@@ -33,6 +39,9 @@ SECTIONS
_etext = .; /* End of text section */
+ NOTES :text :note
+ BUG_TABLE :text
+
RODATA
#ifdef CONFIG_SHARED_KERNEL
@@ -49,9 +58,6 @@ SECTIONS
__stop___ex_table = .;
}
- NOTES
- BUG_TABLE
-
.data : { /* Data */
DATA_DATA
CONSTRUCTORS
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 44/47] single-step cleanup
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (42 preceding siblings ...)
2007-12-20 15:20 ` [patch 43/47] Move NOTES and BUG_TABLE Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 45/47] dasd: add hyper PAV support to DASD device driver, part 1 Martin Schwidefsky
` (2 subsequent siblings)
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Roland McGrath, Martin Schwidefsky
[-- Attachment #1: 143-single-step.diff --]
[-- Type: text/plain, Size: 4518 bytes --]
From: Roland McGrath <roland@redhat.com>
Signed-off-by: Roland McGrath <roland@redhat.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
arch/s390/kernel/ptrace.c | 15 ++++++---------
arch/s390/kernel/signal.c | 20 +++++++++++++-------
include/asm-s390/ptrace.h | 8 ++++++++
3 files changed, 27 insertions(+), 16 deletions(-)
Index: quilt-2.6/arch/s390/kernel/ptrace.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/ptrace.c
+++ quilt-2.6/arch/s390/kernel/ptrace.c
@@ -86,13 +86,13 @@ FixPerRegisters(struct task_struct *task
per_info->control_regs.bits.storage_alt_space_ctl = 0;
}
-static void set_single_step(struct task_struct *task)
+void user_enable_single_step(struct task_struct *task)
{
task->thread.per_info.single_step = 1;
FixPerRegisters(task);
}
-static void clear_single_step(struct task_struct *task)
+void user_disable_single_step(struct task_struct *task)
{
task->thread.per_info.single_step = 0;
FixPerRegisters(task);
@@ -107,7 +107,7 @@ void
ptrace_disable(struct task_struct *child)
{
/* make sure the single step bit is not set. */
- clear_single_step(child);
+ user_disable_single_step(child);
}
#ifndef CONFIG_64BIT
@@ -651,7 +651,7 @@ do_ptrace(struct task_struct *child, lon
clear_tsk_thread_flag(child, TIF_SYSCALL_TRACE);
child->exit_code = data;
/* make sure the single step bit is not set. */
- clear_single_step(child);
+ user_disable_single_step(child);
wake_up_process(child);
return 0;
@@ -665,7 +665,7 @@ do_ptrace(struct task_struct *child, lon
return 0;
child->exit_code = SIGKILL;
/* make sure the single step bit is not set. */
- clear_single_step(child);
+ user_disable_single_step(child);
wake_up_process(child);
return 0;
@@ -675,10 +675,7 @@ do_ptrace(struct task_struct *child, lon
return -EIO;
clear_tsk_thread_flag(child, TIF_SYSCALL_TRACE);
child->exit_code = data;
- if (data)
- set_tsk_thread_flag(child, TIF_SINGLE_STEP);
- else
- set_single_step(child);
+ user_enable_single_step(child);
/* give it a chance to run. */
wake_up_process(child);
return 0;
Index: quilt-2.6/arch/s390/kernel/signal.c
===================================================================
--- quilt-2.6.orig/arch/s390/kernel/signal.c
+++ quilt-2.6/arch/s390/kernel/signal.c
@@ -471,6 +471,7 @@ void do_signal(struct pt_regs *regs)
if (signr > 0) {
/* Whee! Actually deliver the signal. */
+ int ret;
#ifdef CONFIG_COMPAT
if (test_thread_flag(TIF_31BIT)) {
extern int handle_signal32(unsigned long sig,
@@ -478,15 +479,12 @@ void do_signal(struct pt_regs *regs)
siginfo_t *info,
sigset_t *oldset,
struct pt_regs *regs);
- if (handle_signal32(
- signr, &ka, &info, oldset, regs) == 0) {
- if (test_thread_flag(TIF_RESTORE_SIGMASK))
- clear_thread_flag(TIF_RESTORE_SIGMASK);
- }
- return;
+ ret = handle_signal32(signr, &ka, &info, oldset, regs);
}
+ else
#endif
- if (handle_signal(signr, &ka, &info, oldset, regs) == 0) {
+ ret = handle_signal(signr, &ka, &info, oldset, regs);
+ if (!ret) {
/*
* A signal was successfully delivered; the saved
* sigmask will have been stored in the signal frame,
@@ -495,6 +493,14 @@ void do_signal(struct pt_regs *regs)
*/
if (test_thread_flag(TIF_RESTORE_SIGMASK))
clear_thread_flag(TIF_RESTORE_SIGMASK);
+
+ /*
+ * If we would have taken a single-step trap
+ * for a normal instruction, act like we took
+ * one for the handler setup.
+ */
+ if (current->thread.per_info.single_step)
+ set_thread_flag(TIF_SINGLE_STEP);
}
return;
}
Index: quilt-2.6/include/asm-s390/ptrace.h
===================================================================
--- quilt-2.6.orig/include/asm-s390/ptrace.h
+++ quilt-2.6/include/asm-s390/ptrace.h
@@ -465,6 +465,14 @@ struct user_regs_struct
#ifdef __KERNEL__
#define __ARCH_SYS_PTRACE 1
+/*
+ * These are defined as per linux/ptrace.h, which see.
+ */
+#define arch_has_single_step() (1)
+struct task_struct;
+extern void user_enable_single_step(struct task_struct *);
+extern void user_disable_single_step(struct task_struct *);
+
#define user_mode(regs) (((regs)->psw.mask & PSW_MASK_PSTATE) != 0)
#define instruction_pointer(regs) ((regs)->psw.addr & PSW_ADDR_INSN)
#define regs_return_value(regs)((regs)->gprs[2])
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 45/47] dasd: add hyper PAV support to DASD device driver, part 1
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (43 preceding siblings ...)
2007-12-20 15:20 ` [patch 44/47] single-step cleanup Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 46/47] dasd: add hyper PAV support to DASD device driver, part 2 Martin Schwidefsky
2007-12-20 15:20 ` [patch 47/47] dasd: add hyper PAV support to DASD device driver, part 3 Martin Schwidefsky
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Stefan Weinhuber, Martin Schwidefsky
[-- Attachment #1: 144-hyperpav-1.diff --]
[-- Type: text/plain, Size: 86868 bytes --]
From: Stefan Weinhuber <wein@de.ibm.com>
Parallel access volumes (PAV) is a storage server feature, that allows
to start multiple channel programs on the same DASD in parallel. It
defines alias devices which can be used as alternative paths to the
same disk. With the old base PAV support we only needed rudimentary
functionality in the DASD device driver. As the mapping between base
and alias devices was static, we just had to export an identifier
(uid) and could leave the combining of devices to external layers
like a device mapper multipath.
Now hyper PAV removes the requirement to dedicate alias devices to
specific base devices. Instead each alias devices can be combined with
multiple base device on a per request basis. This requires full
support by the DASD device driver as now each channel program itself
has to identify the target base device.
The changes to the dasd device driver and the ECKD discipline are:
- Separate subchannel device representation (dasd_device) from block
device representation (dasd_block). Only base devices are block
devices.
- Gather information about base and alias devices and possible
combinations.
- For each request decide which dasd_device should be used (base or
alias) and build specific channel program.
- Support summary unit checks, which allow the storage server to
upgrade / downgrade between base and hyper PAV at runtime (support
is mandatory).
Signed-off-by: Stefan Weinhuber <wein@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/block/Makefile | 4
drivers/s390/block/dasd_3370_erp.c | 84 ---
drivers/s390/block/dasd_3990_erp.c | 358 +++++---------
drivers/s390/block/dasd_9336_erp.c | 41 -
drivers/s390/block/dasd_9343_erp.c | 21
drivers/s390/block/dasd_alias.c | 903 +++++++++++++++++++++++++++++++++++++
drivers/s390/block/dasd_int.h | 158 ++++--
drivers/s390/block/dasd_ioctl.c | 165 +++---
drivers/s390/block/dasd_proc.c | 21
9 files changed, 1231 insertions(+), 524 deletions(-)
Index: quilt-2.6/drivers/s390/block/dasd_3370_erp.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_3370_erp.c
+++ /dev/null
@@ -1,84 +0,0 @@
-/*
- * File...........: linux/drivers/s390/block/dasd_3370_erp.c
- * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com>
- * Bugreports.to..: <Linux390@de.ibm.com>
- * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 2000
- *
- */
-
-#define PRINTK_HEADER "dasd_erp(3370)"
-
-#include "dasd_int.h"
-
-
-/*
- * DASD_3370_ERP_EXAMINE
- *
- * DESCRIPTION
- * Checks only for fatal/no/recover error.
- * A detailed examination of the sense data is done later outside
- * the interrupt handler.
- *
- * The logic is based on the 'IBM 3880 Storage Control Reference' manual
- * 'Chapter 7. 3370 Sense Data'.
- *
- * RETURN VALUES
- * dasd_era_none no error
- * dasd_era_fatal for all fatal (unrecoverable errors)
- * dasd_era_recover for all others.
- */
-dasd_era_t
-dasd_3370_erp_examine(struct dasd_ccw_req * cqr, struct irb * irb)
-{
- char *sense = irb->ecw;
-
- /* check for successful execution first */
- if (irb->scsw.cstat == 0x00 &&
- irb->scsw.dstat == (DEV_STAT_CHN_END | DEV_STAT_DEV_END))
- return dasd_era_none;
- if (sense[0] & 0x80) { /* CMD reject */
- return dasd_era_fatal;
- }
- if (sense[0] & 0x40) { /* Drive offline */
- return dasd_era_recover;
- }
- if (sense[0] & 0x20) { /* Bus out parity */
- return dasd_era_recover;
- }
- if (sense[0] & 0x10) { /* equipment check */
- if (sense[1] & 0x80) {
- return dasd_era_fatal;
- }
- return dasd_era_recover;
- }
- if (sense[0] & 0x08) { /* data check */
- if (sense[1] & 0x80) {
- return dasd_era_fatal;
- }
- return dasd_era_recover;
- }
- if (sense[0] & 0x04) { /* overrun */
- if (sense[1] & 0x80) {
- return dasd_era_fatal;
- }
- return dasd_era_recover;
- }
- if (sense[1] & 0x40) { /* invalid blocksize */
- return dasd_era_fatal;
- }
- if (sense[1] & 0x04) { /* file protected */
- return dasd_era_recover;
- }
- if (sense[1] & 0x01) { /* operation incomplete */
- return dasd_era_recover;
- }
- if (sense[2] & 0x80) { /* check data erroor */
- return dasd_era_recover;
- }
- if (sense[2] & 0x10) { /* Env. data present */
- return dasd_era_recover;
- }
- /* examine the 24 byte sense data */
- return dasd_era_recover;
-
-} /* END dasd_3370_erp_examine */
Index: quilt-2.6/drivers/s390/block/dasd_3990_erp.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_3990_erp.c
+++ quilt-2.6/drivers/s390/block/dasd_3990_erp.c
@@ -26,158 +26,6 @@ struct DCTL_data {
/*
*****************************************************************************
- * SECTION ERP EXAMINATION
- *****************************************************************************
- */
-
-/*
- * DASD_3990_ERP_EXAMINE_24
- *
- * DESCRIPTION
- * Checks only for fatal (unrecoverable) error.
- * A detailed examination of the sense data is done later outside
- * the interrupt handler.
- *
- * Each bit configuration leading to an action code 2 (Exit with
- * programming error or unusual condition indication)
- * are handled as fatal errors.
- *
- * All other configurations are handled as recoverable errors.
- *
- * RETURN VALUES
- * dasd_era_fatal for all fatal (unrecoverable errors)
- * dasd_era_recover for all others.
- */
-static dasd_era_t
-dasd_3990_erp_examine_24(struct dasd_ccw_req * cqr, char *sense)
-{
-
- struct dasd_device *device = cqr->device;
-
- /* check for 'Command Reject' */
- if ((sense[0] & SNS0_CMD_REJECT) &&
- (!(sense[2] & SNS2_ENV_DATA_PRESENT))) {
-
- DEV_MESSAGE(KERN_ERR, device, "%s",
- "EXAMINE 24: Command Reject detected - "
- "fatal error");
-
- return dasd_era_fatal;
- }
-
- /* check for 'Invalid Track Format' */
- if ((sense[1] & SNS1_INV_TRACK_FORMAT) &&
- (!(sense[2] & SNS2_ENV_DATA_PRESENT))) {
-
- DEV_MESSAGE(KERN_ERR, device, "%s",
- "EXAMINE 24: Invalid Track Format detected "
- "- fatal error");
-
- return dasd_era_fatal;
- }
-
- /* check for 'No Record Found' */
- if (sense[1] & SNS1_NO_REC_FOUND) {
-
- /* FIXME: fatal error ?!? */
- DEV_MESSAGE(KERN_ERR, device,
- "EXAMINE 24: No Record Found detected %s",
- device->state <= DASD_STATE_BASIC ?
- " " : "- fatal error");
-
- return dasd_era_fatal;
- }
-
- /* return recoverable for all others */
- return dasd_era_recover;
-} /* END dasd_3990_erp_examine_24 */
-
-/*
- * DASD_3990_ERP_EXAMINE_32
- *
- * DESCRIPTION
- * Checks only for fatal/no/recoverable error.
- * A detailed examination of the sense data is done later outside
- * the interrupt handler.
- *
- * RETURN VALUES
- * dasd_era_none no error
- * dasd_era_fatal for all fatal (unrecoverable errors)
- * dasd_era_recover for recoverable others.
- */
-static dasd_era_t
-dasd_3990_erp_examine_32(struct dasd_ccw_req * cqr, char *sense)
-{
-
- struct dasd_device *device = cqr->device;
-
- switch (sense[25]) {
- case 0x00:
- return dasd_era_none;
-
- case 0x01:
- DEV_MESSAGE(KERN_ERR, device, "%s", "EXAMINE 32: fatal error");
-
- return dasd_era_fatal;
-
- default:
-
- return dasd_era_recover;
- }
-
-} /* end dasd_3990_erp_examine_32 */
-
-/*
- * DASD_3990_ERP_EXAMINE
- *
- * DESCRIPTION
- * Checks only for fatal/no/recover error.
- * A detailed examination of the sense data is done later outside
- * the interrupt handler.
- *
- * The logic is based on the 'IBM 3990 Storage Control Reference' manual
- * 'Chapter 7. Error Recovery Procedures'.
- *
- * RETURN VALUES
- * dasd_era_none no error
- * dasd_era_fatal for all fatal (unrecoverable errors)
- * dasd_era_recover for all others.
- */
-dasd_era_t
-dasd_3990_erp_examine(struct dasd_ccw_req * cqr, struct irb * irb)
-{
-
- char *sense = irb->ecw;
- dasd_era_t era = dasd_era_recover;
- struct dasd_device *device = cqr->device;
-
- /* check for successful execution first */
- if (irb->scsw.cstat == 0x00 &&
- irb->scsw.dstat == (DEV_STAT_CHN_END | DEV_STAT_DEV_END))
- return dasd_era_none;
-
- /* distinguish between 24 and 32 byte sense data */
- if (sense[27] & DASD_SENSE_BIT_0) {
-
- era = dasd_3990_erp_examine_24(cqr, sense);
-
- } else {
-
- era = dasd_3990_erp_examine_32(cqr, sense);
-
- }
-
- /* log the erp chain if fatal error occurred */
- if ((era == dasd_era_fatal) && (device->state >= DASD_STATE_READY)) {
- dasd_log_sense(cqr, irb);
- }
-
- return era;
-
-} /* END dasd_3990_erp_examine */
-
-/*
- *****************************************************************************
* SECTION ERP HANDLING
*****************************************************************************
*/
@@ -206,7 +54,7 @@ dasd_3990_erp_cleanup(struct dasd_ccw_re
{
struct dasd_ccw_req *cqr = erp->refers;
- dasd_free_erp_request(erp, erp->device);
+ dasd_free_erp_request(erp, erp->memdev);
cqr->status = final_status;
return cqr;
@@ -224,15 +72,17 @@ static void
dasd_3990_erp_block_queue(struct dasd_ccw_req * erp, int expires)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
+ unsigned long flags;
DEV_MESSAGE(KERN_INFO, device,
"blocking request queue for %is", expires/HZ);
+ spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
device->stopped |= DASD_STOPPED_PENDING;
- erp->status = DASD_CQR_QUEUED;
-
- dasd_set_timer(device, expires);
+ spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+ erp->status = DASD_CQR_FILLED;
+ dasd_block_set_timer(device->block, expires);
}
/*
@@ -251,7 +101,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_int_req(struct dasd_ccw_req * erp)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
/* first time set initial retry counter and erp_function */
/* and retry once without blocking queue */
@@ -292,11 +142,14 @@ dasd_3990_erp_int_req(struct dasd_ccw_re
static void
dasd_3990_erp_alternate_path(struct dasd_ccw_req * erp)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
__u8 opm;
+ unsigned long flags;
/* try alternate valid path */
+ spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
opm = ccw_device_get_path_mask(device->cdev);
+ spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
//FIXME: start with get_opm ?
if (erp->lpm == 0)
erp->lpm = LPM_ANYPATH & ~(erp->irb.esw.esw0.sublog.lpum);
@@ -309,9 +162,8 @@ dasd_3990_erp_alternate_path(struct dasd
"try alternate lpm=%x (lpum=%x / opm=%x)",
erp->lpm, erp->irb.esw.esw0.sublog.lpum, opm);
- /* reset status to queued to handle the request again... */
- if (erp->status > DASD_CQR_QUEUED)
- erp->status = DASD_CQR_QUEUED;
+ /* reset status to submit the request again... */
+ erp->status = DASD_CQR_FILLED;
erp->retries = 1;
} else {
DEV_MESSAGE(KERN_ERR, device,
@@ -320,8 +172,7 @@ dasd_3990_erp_alternate_path(struct dasd
erp->irb.esw.esw0.sublog.lpum, opm);
/* post request with permanent error */
- if (erp->status > DASD_CQR_QUEUED)
- erp->status = DASD_CQR_FAILED;
+ erp->status = DASD_CQR_FAILED;
}
} /* end dasd_3990_erp_alternate_path */
@@ -344,14 +195,14 @@ static struct dasd_ccw_req *
dasd_3990_erp_DCTL(struct dasd_ccw_req * erp, char modifier)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
struct DCTL_data *DCTL_data;
struct ccw1 *ccw;
struct dasd_ccw_req *dctl_cqr;
dctl_cqr = dasd_alloc_erp_request((char *) &erp->magic, 1,
- sizeof (struct DCTL_data),
- erp->device);
+ sizeof(struct DCTL_data),
+ device);
if (IS_ERR(dctl_cqr)) {
DEV_MESSAGE(KERN_ERR, device, "%s",
"Unable to allocate DCTL-CQR");
@@ -365,13 +216,14 @@ dasd_3990_erp_DCTL(struct dasd_ccw_req *
DCTL_data->modifier = modifier;
ccw = dctl_cqr->cpaddr;
- memset(ccw, 0, sizeof (struct ccw1));
+ memset(ccw, 0, sizeof(struct ccw1));
ccw->cmd_code = CCW_CMD_DCTL;
ccw->count = 4;
ccw->cda = (__u32)(addr_t) DCTL_data;
dctl_cqr->function = dasd_3990_erp_DCTL;
dctl_cqr->refers = erp;
- dctl_cqr->device = erp->device;
+ dctl_cqr->startdev = device;
+ dctl_cqr->memdev = device;
dctl_cqr->magic = erp->magic;
dctl_cqr->expires = 5 * 60 * HZ;
dctl_cqr->retries = 2;
@@ -435,7 +287,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_action_4(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
/* first time set initial retry counter and erp_function */
/* and retry once without waiting for state change pending */
@@ -472,7 +324,7 @@ dasd_3990_erp_action_4(struct dasd_ccw_r
"redriving request immediately, "
"%d retries left",
erp->retries);
- erp->status = DASD_CQR_QUEUED;
+ erp->status = DASD_CQR_FILLED;
}
}
@@ -530,7 +382,7 @@ static void
dasd_3990_handle_env_data(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
char msg_format = (sense[7] & 0xF0);
char msg_no = (sense[7] & 0x0F);
@@ -1157,7 +1009,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_com_rej(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
erp->function = dasd_3990_erp_com_rej;
@@ -1198,7 +1050,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_bus_out(struct dasd_ccw_req * erp)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
/* first time set initial retry counter and erp_function */
/* and retry once without blocking queue */
@@ -1237,7 +1089,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_equip_check(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
erp->function = dasd_3990_erp_equip_check;
@@ -1279,7 +1131,6 @@ dasd_3990_erp_equip_check(struct dasd_cc
erp = dasd_3990_erp_action_5(erp);
}
-
return erp;
} /* end dasd_3990_erp_equip_check */
@@ -1299,7 +1150,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_data_check(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
erp->function = dasd_3990_erp_data_check;
@@ -1358,7 +1209,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_overrun(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
erp->function = dasd_3990_erp_overrun;
@@ -1387,7 +1238,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_inv_format(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
erp->function = dasd_3990_erp_inv_format;
@@ -1403,8 +1254,7 @@ dasd_3990_erp_inv_format(struct dasd_ccw
} else {
DEV_MESSAGE(KERN_ERR, device, "%s",
- "Invalid Track Format - Fatal error should have "
- "been handled within the interrupt handler");
+ "Invalid Track Format - Fatal error");
erp = dasd_3990_erp_cleanup(erp, DASD_CQR_FAILED);
}
@@ -1428,7 +1278,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_EOC(struct dasd_ccw_req * default_erp, char *sense)
{
- struct dasd_device *device = default_erp->device;
+ struct dasd_device *device = default_erp->startdev;
DEV_MESSAGE(KERN_ERR, device, "%s",
"End-of-Cylinder - must never happen");
@@ -1453,7 +1303,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_env_data(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
erp->function = dasd_3990_erp_env_data;
@@ -1463,11 +1313,9 @@ dasd_3990_erp_env_data(struct dasd_ccw_r
/* don't retry on disabled interface */
if (sense[7] != 0x0F) {
-
erp = dasd_3990_erp_action_4(erp, sense);
} else {
-
- erp = dasd_3990_erp_cleanup(erp, DASD_CQR_IN_IO);
+ erp->status = DASD_CQR_FILLED;
}
return erp;
@@ -1490,11 +1338,10 @@ static struct dasd_ccw_req *
dasd_3990_erp_no_rec(struct dasd_ccw_req * default_erp, char *sense)
{
- struct dasd_device *device = default_erp->device;
+ struct dasd_device *device = default_erp->startdev;
DEV_MESSAGE(KERN_ERR, device, "%s",
- "No Record Found - Fatal error should "
- "have been handled within the interrupt handler");
+ "No Record Found - Fatal error ");
return dasd_3990_erp_cleanup(default_erp, DASD_CQR_FAILED);
@@ -1517,7 +1364,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_file_prot(struct dasd_ccw_req * erp)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
DEV_MESSAGE(KERN_ERR, device, "%s", "File Protected");
@@ -1526,6 +1373,43 @@ dasd_3990_erp_file_prot(struct dasd_ccw_
} /* end dasd_3990_erp_file_prot */
/*
+ * DASD_3990_ERP_INSPECT_ALIAS
+ *
+ * DESCRIPTION
+ * Checks if the original request was started on an alias device.
+ * If yes, it modifies the original and the erp request so that
+ * the erp request can be started on a base device.
+ *
+ * PARAMETER
+ * erp pointer to the currently created default ERP
+ *
+ * RETURN VALUES
+ * erp pointer to the modified ERP, or NULL
+ */
+
+static struct dasd_ccw_req *dasd_3990_erp_inspect_alias(
+ struct dasd_ccw_req *erp)
+{
+ struct dasd_ccw_req *cqr = erp->refers;
+
+ if (cqr->block &&
+ (cqr->block->base != cqr->startdev)) {
+ if (cqr->startdev->features & DASD_FEATURE_ERPLOG) {
+ DEV_MESSAGE(KERN_ERR, cqr->startdev,
+ "ERP on alias device for request %p,"
+ " recover on base device %s", cqr,
+ cqr->block->base->cdev->dev.bus_id);
+ }
+ dasd_eckd_reset_ccw_to_base_io(cqr);
+ erp->startdev = cqr->block->base;
+ erp->function = dasd_3990_erp_inspect_alias;
+ return erp;
+ } else
+ return NULL;
+}
+
+
+/*
* DASD_3990_ERP_INSPECT_24
*
* DESCRIPTION
@@ -1623,7 +1507,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_action_10_32(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
erp->retries = 256;
erp->function = dasd_3990_erp_action_10_32;
@@ -1657,13 +1541,14 @@ static struct dasd_ccw_req *
dasd_3990_erp_action_1B_32(struct dasd_ccw_req * default_erp, char *sense)
{
- struct dasd_device *device = default_erp->device;
+ struct dasd_device *device = default_erp->startdev;
__u32 cpa = 0;
struct dasd_ccw_req *cqr;
struct dasd_ccw_req *erp;
struct DE_eckd_data *DE_data;
+ struct PFX_eckd_data *PFX_data;
char *LO_data; /* LO_eckd_data_t */
- struct ccw1 *ccw;
+ struct ccw1 *ccw, *oldccw;
DEV_MESSAGE(KERN_DEBUG, device, "%s",
"Write not finished because of unexpected condition");
@@ -1702,8 +1587,8 @@ dasd_3990_erp_action_1B_32(struct dasd_c
/* Build new ERP request including DE/LO */
erp = dasd_alloc_erp_request((char *) &cqr->magic,
2 + 1,/* DE/LO + TIC */
- sizeof (struct DE_eckd_data) +
- sizeof (struct LO_eckd_data), device);
+ sizeof(struct DE_eckd_data) +
+ sizeof(struct LO_eckd_data), device);
if (IS_ERR(erp)) {
DEV_MESSAGE(KERN_ERR, device, "%s", "Unable to allocate ERP");
@@ -1712,10 +1597,16 @@ dasd_3990_erp_action_1B_32(struct dasd_c
/* use original DE */
DE_data = erp->data;
- memcpy(DE_data, cqr->data, sizeof (struct DE_eckd_data));
+ oldccw = cqr->cpaddr;
+ if (oldccw->cmd_code == DASD_ECKD_CCW_PFX) {
+ PFX_data = cqr->data;
+ memcpy(DE_data, &PFX_data->define_extend,
+ sizeof(struct DE_eckd_data));
+ } else
+ memcpy(DE_data, cqr->data, sizeof(struct DE_eckd_data));
/* create LO */
- LO_data = erp->data + sizeof (struct DE_eckd_data);
+ LO_data = erp->data + sizeof(struct DE_eckd_data);
if ((sense[3] == 0x01) && (LO_data[1] & 0x01)) {
@@ -1748,7 +1639,7 @@ dasd_3990_erp_action_1B_32(struct dasd_c
/* create DE ccw */
ccw = erp->cpaddr;
- memset(ccw, 0, sizeof (struct ccw1));
+ memset(ccw, 0, sizeof(struct ccw1));
ccw->cmd_code = DASD_ECKD_CCW_DEFINE_EXTENT;
ccw->flags = CCW_FLAG_CC;
ccw->count = 16;
@@ -1756,7 +1647,7 @@ dasd_3990_erp_action_1B_32(struct dasd_c
/* create LO ccw */
ccw++;
- memset(ccw, 0, sizeof (struct ccw1));
+ memset(ccw, 0, sizeof(struct ccw1));
ccw->cmd_code = DASD_ECKD_CCW_LOCATE_RECORD;
ccw->flags = CCW_FLAG_CC;
ccw->count = 16;
@@ -1770,7 +1661,8 @@ dasd_3990_erp_action_1B_32(struct dasd_c
/* fill erp related fields */
erp->function = dasd_3990_erp_action_1B_32;
erp->refers = default_erp->refers;
- erp->device = device;
+ erp->startdev = device;
+ erp->memdev = device;
erp->magic = default_erp->magic;
erp->expires = 0;
erp->retries = 256;
@@ -1803,7 +1695,7 @@ static struct dasd_ccw_req *
dasd_3990_update_1B(struct dasd_ccw_req * previous_erp, char *sense)
{
- struct dasd_device *device = previous_erp->device;
+ struct dasd_device *device = previous_erp->startdev;
__u32 cpa = 0;
struct dasd_ccw_req *cqr;
struct dasd_ccw_req *erp;
@@ -1827,7 +1719,7 @@ dasd_3990_update_1B(struct dasd_ccw_req
DEV_MESSAGE(KERN_DEBUG, device, "%s",
"Imprecise ending is set - just retry");
- previous_erp->status = DASD_CQR_QUEUED;
+ previous_erp->status = DASD_CQR_FILLED;
return previous_erp;
}
@@ -1850,7 +1742,7 @@ dasd_3990_update_1B(struct dasd_ccw_req
erp = previous_erp;
/* update the LO with the new returned sense data */
- LO_data = erp->data + sizeof (struct DE_eckd_data);
+ LO_data = erp->data + sizeof(struct DE_eckd_data);
if ((sense[3] == 0x01) && (LO_data[1] & 0x01)) {
@@ -1889,7 +1781,7 @@ dasd_3990_update_1B(struct dasd_ccw_req
ccw++; /* addr of TIC ccw */
ccw->cda = cpa;
- erp->status = DASD_CQR_QUEUED;
+ erp->status = DASD_CQR_FILLED;
return erp;
@@ -1968,9 +1860,7 @@ dasd_3990_erp_compound_path(struct dasd_
* try further actions. */
erp->lpm = 0;
-
- erp->status = DASD_CQR_ERROR;
-
+ erp->status = DASD_CQR_NEED_ERP;
}
}
@@ -2047,7 +1937,7 @@ dasd_3990_erp_compound_config(struct das
if ((sense[25] & DASD_SENSE_BIT_1) && (sense[26] & DASD_SENSE_BIT_2)) {
/* set to suspended duplex state then restart */
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
DEV_MESSAGE(KERN_ERR, device, "%s",
"Set device to suspended duplex state should be "
@@ -2081,28 +1971,26 @@ dasd_3990_erp_compound(struct dasd_ccw_r
{
if ((erp->function == dasd_3990_erp_compound_retry) &&
- (erp->status == DASD_CQR_ERROR)) {
+ (erp->status == DASD_CQR_NEED_ERP)) {
dasd_3990_erp_compound_path(erp, sense);
}
if ((erp->function == dasd_3990_erp_compound_path) &&
- (erp->status == DASD_CQR_ERROR)) {
+ (erp->status == DASD_CQR_NEED_ERP)) {
erp = dasd_3990_erp_compound_code(erp, sense);
}
if ((erp->function == dasd_3990_erp_compound_code) &&
- (erp->status == DASD_CQR_ERROR)) {
+ (erp->status == DASD_CQR_NEED_ERP)) {
dasd_3990_erp_compound_config(erp, sense);
}
/* if no compound action ERP specified, the request failed */
- if (erp->status == DASD_CQR_ERROR) {
-
+ if (erp->status == DASD_CQR_NEED_ERP)
erp->status = DASD_CQR_FAILED;
- }
return erp;
@@ -2127,7 +2015,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_inspect_32(struct dasd_ccw_req * erp, char *sense)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
erp->function = dasd_3990_erp_inspect_32;
@@ -2149,8 +2037,7 @@ dasd_3990_erp_inspect_32(struct dasd_ccw
case 0x01: /* fatal error */
DEV_MESSAGE(KERN_ERR, device, "%s",
- "Fatal error should have been "
- "handled within the interrupt handler");
+ "Retry not recommended - Fatal error");
erp = dasd_3990_erp_cleanup(erp, DASD_CQR_FAILED);
break;
@@ -2253,6 +2140,11 @@ dasd_3990_erp_inspect(struct dasd_ccw_re
/* already set up new ERP ! */
char *sense = erp->refers->irb.ecw;
+ /* if this problem occured on an alias retry on base */
+ erp_new = dasd_3990_erp_inspect_alias(erp);
+ if (erp_new)
+ return erp_new;
+
/* distinguish between 24 and 32 byte sense data */
if (sense[27] & DASD_SENSE_BIT_0) {
@@ -2287,13 +2179,13 @@ static struct dasd_ccw_req *
dasd_3990_erp_add_erp(struct dasd_ccw_req * cqr)
{
- struct dasd_device *device = cqr->device;
+ struct dasd_device *device = cqr->startdev;
struct ccw1 *ccw;
/* allocate additional request block */
struct dasd_ccw_req *erp;
- erp = dasd_alloc_erp_request((char *) &cqr->magic, 2, 0, cqr->device);
+ erp = dasd_alloc_erp_request((char *) &cqr->magic, 2, 0, device);
if (IS_ERR(erp)) {
if (cqr->retries <= 0) {
DEV_MESSAGE(KERN_ERR, device, "%s",
@@ -2305,7 +2197,7 @@ dasd_3990_erp_add_erp(struct dasd_ccw_re
"Unable to allocate ERP request "
"(%i retries left)",
cqr->retries);
- dasd_set_timer(device, (HZ << 3));
+ dasd_block_set_timer(device->block, (HZ << 3));
}
return cqr;
}
@@ -2319,7 +2211,9 @@ dasd_3990_erp_add_erp(struct dasd_ccw_re
ccw->cda = (long)(cqr->cpaddr);
erp->function = dasd_3990_erp_add_erp;
erp->refers = cqr;
- erp->device = cqr->device;
+ erp->startdev = device;
+ erp->memdev = device;
+ erp->block = cqr->block;
erp->magic = cqr->magic;
erp->expires = 0;
erp->retries = 256;
@@ -2466,7 +2360,7 @@ static struct dasd_ccw_req *
dasd_3990_erp_further_erp(struct dasd_ccw_req *erp)
{
- struct dasd_device *device = erp->device;
+ struct dasd_device *device = erp->startdev;
char *sense = erp->irb.ecw;
/* check for 24 byte sense ERP */
@@ -2557,7 +2451,7 @@ dasd_3990_erp_handle_match_erp(struct da
struct dasd_ccw_req *erp)
{
- struct dasd_device *device = erp_head->device;
+ struct dasd_device *device = erp_head->startdev;
struct dasd_ccw_req *erp_done = erp_head; /* finished req */
struct dasd_ccw_req *erp_free = NULL; /* req to be freed */
@@ -2569,13 +2463,13 @@ dasd_3990_erp_handle_match_erp(struct da
"original request was lost\n");
/* remove the request from the device queue */
- list_del(&erp_done->list);
+ list_del(&erp_done->blocklist);
erp_free = erp_done;
erp_done = erp_done->refers;
/* free the finished erp request */
- dasd_free_erp_request(erp_free, erp_free->device);
+ dasd_free_erp_request(erp_free, erp_free->memdev);
} /* end while */
@@ -2603,7 +2497,7 @@ dasd_3990_erp_handle_match_erp(struct da
erp->retries, erp);
/* handle the request again... */
- erp->status = DASD_CQR_QUEUED;
+ erp->status = DASD_CQR_FILLED;
}
} else {
@@ -2636,9 +2530,8 @@ dasd_3990_erp_handle_match_erp(struct da
struct dasd_ccw_req *
dasd_3990_erp_action(struct dasd_ccw_req * cqr)
{
-
struct dasd_ccw_req *erp = NULL;
- struct dasd_device *device = cqr->device;
+ struct dasd_device *device = cqr->startdev;
struct dasd_ccw_req *temp_erp = NULL;
if (device->features & DASD_FEATURE_ERPLOG) {
@@ -2704,10 +2597,11 @@ dasd_3990_erp_action(struct dasd_ccw_req
}
}
- /* enqueue added ERP request */
- if (erp->status == DASD_CQR_FILLED) {
- erp->status = DASD_CQR_QUEUED;
- list_add(&erp->list, &device->ccw_queue);
+ /* enqueue ERP request if it's a new one */
+ if (list_empty(&erp->blocklist)) {
+ cqr->status = DASD_CQR_IN_ERP;
+ /* add erp request before the cqr */
+ list_add_tail(&erp->blocklist, &cqr->blocklist);
}
return erp;
Index: quilt-2.6/drivers/s390/block/dasd_9336_erp.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_9336_erp.c
+++ /dev/null
@@ -1,41 +0,0 @@
-/*
- * File...........: linux/drivers/s390/block/dasd_9336_erp.c
- * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com>
- * Bugreports.to..: <Linux390@de.ibm.com>
- * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 2000
- *
- */
-
-#define PRINTK_HEADER "dasd_erp(9336)"
-
-#include "dasd_int.h"
-
-
-/*
- * DASD_9336_ERP_EXAMINE
- *
- * DESCRIPTION
- * Checks only for fatal/no/recover error.
- * A detailed examination of the sense data is done later outside
- * the interrupt handler.
- *
- * The logic is based on the 'IBM 3880 Storage Control Reference' manual
- * 'Chapter 7. 9336 Sense Data'.
- *
- * RETURN VALUES
- * dasd_era_none no error
- * dasd_era_fatal for all fatal (unrecoverable errors)
- * dasd_era_recover for all others.
- */
-dasd_era_t
-dasd_9336_erp_examine(struct dasd_ccw_req * cqr, struct irb * irb)
-{
- /* check for successful execution first */
- if (irb->scsw.cstat == 0x00 &&
- irb->scsw.dstat == (DEV_STAT_CHN_END | DEV_STAT_DEV_END))
- return dasd_era_none;
-
- /* examine the 24 byte sense data */
- return dasd_era_recover;
-
-} /* END dasd_9336_erp_examine */
Index: quilt-2.6/drivers/s390/block/dasd_9343_erp.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_9343_erp.c
+++ /dev/null
@@ -1,21 +0,0 @@
-/*
- * File...........: linux/drivers/s390/block/dasd_9345_erp.c
- * Author(s)......: Holger Smolinski <Holger.Smolinski@de.ibm.com>
- * Bugreports.to..: <Linux390@de.ibm.com>
- * (C) IBM Corporation, IBM Deutschland Entwicklung GmbH, 2000
- *
- */
-
-#define PRINTK_HEADER "dasd_erp(9343)"
-
-#include "dasd_int.h"
-
-dasd_era_t
-dasd_9343_erp_examine(struct dasd_ccw_req * cqr, struct irb * irb)
-{
- if (irb->scsw.cstat == 0x00 &&
- irb->scsw.dstat == (DEV_STAT_CHN_END | DEV_STAT_DEV_END))
- return dasd_era_none;
-
- return dasd_era_recover;
-}
Index: quilt-2.6/drivers/s390/block/dasd_alias.c
===================================================================
--- /dev/null
+++ quilt-2.6/drivers/s390/block/dasd_alias.c
@@ -0,0 +1,903 @@
+/*
+ * PAV alias management for the DASD ECKD discipline
+ *
+ * Copyright IBM Corporation, IBM Deutschland Entwicklung GmbH, 2007
+ * Author(s): Stefan Weinhuber <wein@de.ibm.com>
+ */
+
+#include <linux/list.h>
+#include <asm/ebcdic.h>
+#include "dasd_int.h"
+#include "dasd_eckd.h"
+
+#ifdef PRINTK_HEADER
+#undef PRINTK_HEADER
+#endif /* PRINTK_HEADER */
+#define PRINTK_HEADER "dasd(eckd):"
+
+
+/*
+ * General concept of alias management:
+ * - PAV and DASD alias management is specific to the eckd discipline.
+ * - A device is connected to an lcu as long as the device exists.
+ * dasd_alias_make_device_known_to_lcu will be called wenn the
+ * device is checked by the eckd discipline and
+ * dasd_alias_disconnect_device_from_lcu will be called
+ * befor the device is deleted.
+ * - The dasd_alias_add_device / dasd_alias_remove_device
+ * functions mark the point when a device is 'ready for service'.
+ * - A summary unit check is a rare occasion, but it is mandatory to
+ * support it. It requires some complex recovery actions before the
+ * devices can be used again (see dasd_alias_handle_summary_unit_check).
+ * - dasd_alias_get_start_dev will find an alias device that can be used
+ * instead of the base device and does some (very simple) load balancing.
+ * This is the function that get's called for each I/O, so when improving
+ * something, this function should get faster or better, the rest has just
+ * to be correct.
+ */
+
+
+static void summary_unit_check_handling_work(struct work_struct *);
+static void lcu_update_work(struct work_struct *);
+static int _schedule_lcu_update(struct alias_lcu *, struct dasd_device *);
+
+static struct alias_root aliastree = {
+ .serverlist = LIST_HEAD_INIT(aliastree.serverlist),
+ .lock = __SPIN_LOCK_UNLOCKED(aliastree.lock),
+};
+
+static struct alias_server *_find_server(struct dasd_uid *uid)
+{
+ struct alias_server *pos;
+ list_for_each_entry(pos, &aliastree.serverlist, server) {
+ if (!strncmp(pos->uid.vendor, uid->vendor,
+ sizeof(uid->vendor))
+ && !strncmp(pos->uid.serial, uid->serial,
+ sizeof(uid->serial)))
+ return pos;
+ };
+ return NULL;
+}
+
+static struct alias_lcu *_find_lcu(struct alias_server *server,
+ struct dasd_uid *uid)
+{
+ struct alias_lcu *pos;
+ list_for_each_entry(pos, &server->lculist, lcu) {
+ if (pos->uid.ssid == uid->ssid)
+ return pos;
+ };
+ return NULL;
+}
+
+static struct alias_pav_group *_find_group(struct alias_lcu *lcu,
+ struct dasd_uid *uid)
+{
+ struct alias_pav_group *pos;
+ __u8 search_unit_addr;
+
+ /* for hyper pav there is only one group */
+ if (lcu->pav == HYPER_PAV) {
+ if (list_empty(&lcu->grouplist))
+ return NULL;
+ else
+ return list_first_entry(&lcu->grouplist,
+ struct alias_pav_group, group);
+ }
+
+ /* for base pav we have to find the group that matches the base */
+ if (uid->type == UA_BASE_DEVICE)
+ search_unit_addr = uid->real_unit_addr;
+ else
+ search_unit_addr = uid->base_unit_addr;
+ list_for_each_entry(pos, &lcu->grouplist, group) {
+ if (pos->uid.base_unit_addr == search_unit_addr)
+ return pos;
+ };
+ return NULL;
+}
+
+static struct alias_server *_allocate_server(struct dasd_uid *uid)
+{
+ struct alias_server *server;
+
+ server = kzalloc(sizeof(*server), GFP_KERNEL);
+ if (!server)
+ return ERR_PTR(-ENOMEM);
+ memcpy(server->uid.vendor, uid->vendor, sizeof(uid->vendor));
+ memcpy(server->uid.serial, uid->serial, sizeof(uid->serial));
+ INIT_LIST_HEAD(&server->server);
+ INIT_LIST_HEAD(&server->lculist);
+ return server;
+}
+
+static void _free_server(struct alias_server *server)
+{
+ kfree(server);
+}
+
+static struct alias_lcu *_allocate_lcu(struct dasd_uid *uid)
+{
+ struct alias_lcu *lcu;
+
+ lcu = kzalloc(sizeof(*lcu), GFP_KERNEL);
+ if (!lcu)
+ return ERR_PTR(-ENOMEM);
+ lcu->uac = kzalloc(sizeof(*(lcu->uac)), GFP_KERNEL | GFP_DMA);
+ if (!lcu->uac)
+ goto out_err1;
+ lcu->rsu_cqr = kzalloc(sizeof(*lcu->rsu_cqr), GFP_KERNEL | GFP_DMA);
+ if (!lcu->rsu_cqr)
+ goto out_err2;
+ lcu->rsu_cqr->cpaddr = kzalloc(sizeof(struct ccw1),
+ GFP_KERNEL | GFP_DMA);
+ if (!lcu->rsu_cqr->cpaddr)
+ goto out_err3;
+ lcu->rsu_cqr->data = kzalloc(16, GFP_KERNEL | GFP_DMA);
+ if (!lcu->rsu_cqr->data)
+ goto out_err4;
+
+ memcpy(lcu->uid.vendor, uid->vendor, sizeof(uid->vendor));
+ memcpy(lcu->uid.serial, uid->serial, sizeof(uid->serial));
+ lcu->uid.ssid = uid->ssid;
+ lcu->pav = NO_PAV;
+ lcu->flags = NEED_UAC_UPDATE | UPDATE_PENDING;
+ INIT_LIST_HEAD(&lcu->lcu);
+ INIT_LIST_HEAD(&lcu->inactive_devices);
+ INIT_LIST_HEAD(&lcu->active_devices);
+ INIT_LIST_HEAD(&lcu->grouplist);
+ INIT_WORK(&lcu->suc_data.worker, summary_unit_check_handling_work);
+ INIT_DELAYED_WORK(&lcu->ruac_data.dwork, lcu_update_work);
+ spin_lock_init(&lcu->lock);
+ return lcu;
+
+out_err4:
+ kfree(lcu->rsu_cqr->cpaddr);
+out_err3:
+ kfree(lcu->rsu_cqr);
+out_err2:
+ kfree(lcu->uac);
+out_err1:
+ kfree(lcu);
+ return ERR_PTR(-ENOMEM);
+}
+
+static void _free_lcu(struct alias_lcu *lcu)
+{
+ kfree(lcu->rsu_cqr->data);
+ kfree(lcu->rsu_cqr->cpaddr);
+ kfree(lcu->rsu_cqr);
+ kfree(lcu->uac);
+ kfree(lcu);
+}
+
+/*
+ * This is the function that will allocate all the server and lcu data,
+ * so this function must be called first for a new device.
+ * If the return value is 1, the lcu was already known before, if it
+ * is 0, this is a new lcu.
+ * Negative return code indicates that something went wrong (e.g. -ENOMEM)
+ */
+int dasd_alias_make_device_known_to_lcu(struct dasd_device *device)
+{
+ struct dasd_eckd_private *private;
+ unsigned long flags;
+ struct alias_server *server, *newserver;
+ struct alias_lcu *lcu, *newlcu;
+ int is_lcu_known;
+ struct dasd_uid *uid;
+
+ private = (struct dasd_eckd_private *) device->private;
+ uid = &private->uid;
+ spin_lock_irqsave(&aliastree.lock, flags);
+ is_lcu_known = 1;
+ server = _find_server(uid);
+ if (!server) {
+ spin_unlock_irqrestore(&aliastree.lock, flags);
+ newserver = _allocate_server(uid);
+ if (IS_ERR(newserver))
+ return PTR_ERR(newserver);
+ spin_lock_irqsave(&aliastree.lock, flags);
+ server = _find_server(uid);
+ if (!server) {
+ list_add(&newserver->server, &aliastree.serverlist);
+ server = newserver;
+ is_lcu_known = 0;
+ } else {
+ /* someone was faster */
+ _free_server(newserver);
+ }
+ }
+
+ lcu = _find_lcu(server, uid);
+ if (!lcu) {
+ spin_unlock_irqrestore(&aliastree.lock, flags);
+ newlcu = _allocate_lcu(uid);
+ if (IS_ERR(newlcu))
+ return PTR_ERR(lcu);
+ spin_lock_irqsave(&aliastree.lock, flags);
+ lcu = _find_lcu(server, uid);
+ if (!lcu) {
+ list_add(&newlcu->lcu, &server->lculist);
+ lcu = newlcu;
+ is_lcu_known = 0;
+ } else {
+ /* someone was faster */
+ _free_lcu(newlcu);
+ }
+ is_lcu_known = 0;
+ }
+ spin_lock(&lcu->lock);
+ list_add(&device->alias_list, &lcu->inactive_devices);
+ private->lcu = lcu;
+ spin_unlock(&lcu->lock);
+ spin_unlock_irqrestore(&aliastree.lock, flags);
+
+ return is_lcu_known;
+}
+
+/*
+ * This function removes a device from the scope of alias management.
+ * The complicated part is to make sure that it is not in use by
+ * any of the workers. If necessary cancel the work.
+ */
+void dasd_alias_disconnect_device_from_lcu(struct dasd_device *device)
+{
+ struct dasd_eckd_private *private;
+ unsigned long flags;
+ struct alias_lcu *lcu;
+ struct alias_server *server;
+ int was_pending;
+
+ private = (struct dasd_eckd_private *) device->private;
+ lcu = private->lcu;
+ spin_lock_irqsave(&lcu->lock, flags);
+ list_del_init(&device->alias_list);
+ /* make sure that the workers don't use this device */
+ if (device == lcu->suc_data.device) {
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ cancel_work_sync(&lcu->suc_data.worker);
+ spin_lock_irqsave(&lcu->lock, flags);
+ if (device == lcu->suc_data.device)
+ lcu->suc_data.device = NULL;
+ }
+ was_pending = 0;
+ if (device == lcu->ruac_data.device) {
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ was_pending = 1;
+ cancel_delayed_work_sync(&lcu->ruac_data.dwork);
+ spin_lock_irqsave(&lcu->lock, flags);
+ if (device == lcu->ruac_data.device)
+ lcu->ruac_data.device = NULL;
+ }
+ private->lcu = NULL;
+ spin_unlock_irqrestore(&lcu->lock, flags);
+
+ spin_lock_irqsave(&aliastree.lock, flags);
+ spin_lock(&lcu->lock);
+ if (list_empty(&lcu->grouplist) &&
+ list_empty(&lcu->active_devices) &&
+ list_empty(&lcu->inactive_devices)) {
+ list_del(&lcu->lcu);
+ spin_unlock(&lcu->lock);
+ _free_lcu(lcu);
+ lcu = NULL;
+ } else {
+ if (was_pending)
+ _schedule_lcu_update(lcu, NULL);
+ spin_unlock(&lcu->lock);
+ }
+ server = _find_server(&private->uid);
+ if (server && list_empty(&server->lculist)) {
+ list_del(&server->server);
+ _free_server(server);
+ }
+ spin_unlock_irqrestore(&aliastree.lock, flags);
+}
+
+/*
+ * This function assumes that the unit address configuration stored
+ * in the lcu is up to date and will update the device uid before
+ * adding it to a pav group.
+ */
+static int _add_device_to_lcu(struct alias_lcu *lcu,
+ struct dasd_device *device)
+{
+
+ struct dasd_eckd_private *private;
+ struct alias_pav_group *group;
+ struct dasd_uid *uid;
+
+ private = (struct dasd_eckd_private *) device->private;
+ uid = &private->uid;
+ uid->type = lcu->uac->unit[uid->real_unit_addr].ua_type;
+ uid->base_unit_addr = lcu->uac->unit[uid->real_unit_addr].base_ua;
+ dasd_set_uid(device->cdev, &private->uid);
+
+ group = _find_group(lcu, uid);
+ if (!group) {
+ group = kzalloc(sizeof(*group), GFP_ATOMIC);
+ if (!group)
+ return -ENOMEM;
+ memcpy(group->uid.vendor, uid->vendor, sizeof(uid->vendor));
+ memcpy(group->uid.serial, uid->serial, sizeof(uid->serial));
+ group->uid.ssid = uid->ssid;
+ if (uid->type == UA_BASE_DEVICE)
+ group->uid.base_unit_addr = uid->real_unit_addr;
+ else
+ group->uid.base_unit_addr = uid->base_unit_addr;
+ INIT_LIST_HEAD(&group->group);
+ INIT_LIST_HEAD(&group->baselist);
+ INIT_LIST_HEAD(&group->aliaslist);
+ list_add(&group->group, &lcu->grouplist);
+ }
+ if (uid->type == UA_BASE_DEVICE)
+ list_move(&device->alias_list, &group->baselist);
+ else
+ list_move(&device->alias_list, &group->aliaslist);
+ private->pavgroup = group;
+ return 0;
+};
+
+static void _remove_device_from_lcu(struct alias_lcu *lcu,
+ struct dasd_device *device)
+{
+ struct dasd_eckd_private *private;
+ struct alias_pav_group *group;
+
+ private = (struct dasd_eckd_private *) device->private;
+ list_move(&device->alias_list, &lcu->inactive_devices);
+ group = private->pavgroup;
+ if (!group)
+ return;
+ private->pavgroup = NULL;
+ if (list_empty(&group->baselist) && list_empty(&group->aliaslist)) {
+ list_del(&group->group);
+ kfree(group);
+ return;
+ }
+ if (group->next == device)
+ group->next = NULL;
+};
+
+static int read_unit_address_configuration(struct dasd_device *device,
+ struct alias_lcu *lcu)
+{
+ struct dasd_psf_prssd_data *prssdp;
+ struct dasd_ccw_req *cqr;
+ struct ccw1 *ccw;
+ int rc;
+ unsigned long flags;
+
+ cqr = dasd_kmalloc_request("ECKD",
+ 1 /* PSF */ + 1 /* RSSD */ ,
+ (sizeof(struct dasd_psf_prssd_data)),
+ device);
+ if (IS_ERR(cqr))
+ return PTR_ERR(cqr);
+ cqr->startdev = device;
+ cqr->memdev = device;
+ clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
+ cqr->retries = 10;
+ cqr->expires = 20 * HZ;
+
+ /* Prepare for Read Subsystem Data */
+ prssdp = (struct dasd_psf_prssd_data *) cqr->data;
+ memset(prssdp, 0, sizeof(struct dasd_psf_prssd_data));
+ prssdp->order = PSF_ORDER_PRSSD;
+ prssdp->suborder = 0x0e; /* Read unit address configuration */
+ /* all other bytes of prssdp must be zero */
+
+ ccw = cqr->cpaddr;
+ ccw->cmd_code = DASD_ECKD_CCW_PSF;
+ ccw->count = sizeof(struct dasd_psf_prssd_data);
+ ccw->flags |= CCW_FLAG_CC;
+ ccw->cda = (__u32)(addr_t) prssdp;
+
+ /* Read Subsystem Data - feature codes */
+ memset(lcu->uac, 0, sizeof(*(lcu->uac)));
+
+ ccw++;
+ ccw->cmd_code = DASD_ECKD_CCW_RSSD;
+ ccw->count = sizeof(*(lcu->uac));
+ ccw->cda = (__u32)(addr_t) lcu->uac;
+
+ cqr->buildclk = get_clock();
+ cqr->status = DASD_CQR_FILLED;
+
+ /* need to unset flag here to detect race with summary unit check */
+ spin_lock_irqsave(&lcu->lock, flags);
+ lcu->flags &= ~NEED_UAC_UPDATE;
+ spin_unlock_irqrestore(&lcu->lock, flags);
+
+ do {
+ rc = dasd_sleep_on(cqr);
+ } while (rc && (cqr->retries > 0));
+ if (rc) {
+ spin_lock_irqsave(&lcu->lock, flags);
+ lcu->flags |= NEED_UAC_UPDATE;
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ }
+ dasd_kfree_request(cqr, cqr->memdev);
+ return rc;
+}
+
+static int _lcu_update(struct dasd_device *refdev, struct alias_lcu *lcu)
+{
+ unsigned long flags;
+ struct alias_pav_group *pavgroup, *tempgroup;
+ struct dasd_device *device, *tempdev;
+ int i, rc;
+ struct dasd_eckd_private *private;
+
+ spin_lock_irqsave(&lcu->lock, flags);
+ list_for_each_entry_safe(pavgroup, tempgroup, &lcu->grouplist, group) {
+ list_for_each_entry_safe(device, tempdev, &pavgroup->baselist,
+ alias_list) {
+ list_move(&device->alias_list, &lcu->active_devices);
+ private = (struct dasd_eckd_private *) device->private;
+ private->pavgroup = NULL;
+ }
+ list_for_each_entry_safe(device, tempdev, &pavgroup->aliaslist,
+ alias_list) {
+ list_move(&device->alias_list, &lcu->active_devices);
+ private = (struct dasd_eckd_private *) device->private;
+ private->pavgroup = NULL;
+ }
+ list_del(&pavgroup->group);
+ kfree(pavgroup);
+ }
+ spin_unlock_irqrestore(&lcu->lock, flags);
+
+ rc = read_unit_address_configuration(refdev, lcu);
+ if (rc)
+ return rc;
+
+ spin_lock_irqsave(&lcu->lock, flags);
+ lcu->pav = NO_PAV;
+ for (i = 0; i < MAX_DEVICES_PER_LCU; ++i) {
+ switch (lcu->uac->unit[i].ua_type) {
+ case UA_BASE_PAV_ALIAS:
+ lcu->pav = BASE_PAV;
+ break;
+ case UA_HYPER_PAV_ALIAS:
+ lcu->pav = HYPER_PAV;
+ break;
+ }
+ if (lcu->pav != NO_PAV)
+ break;
+ }
+
+ /* if we have no PAV anyway, we don't need to bother with PAV groups */
+ if (lcu->pav == NO_PAV) {
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ return 0;
+ }
+
+ list_for_each_entry_safe(device, tempdev, &lcu->active_devices,
+ alias_list) {
+ _add_device_to_lcu(lcu, device);
+ }
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ return 0;
+}
+
+static void lcu_update_work(struct work_struct *work)
+{
+ struct alias_lcu *lcu;
+ struct read_uac_work_data *ruac_data;
+ struct dasd_device *device;
+ unsigned long flags;
+ int rc;
+
+ ruac_data = container_of(work, struct read_uac_work_data, dwork.work);
+ lcu = container_of(ruac_data, struct alias_lcu, ruac_data);
+ device = ruac_data->device;
+ rc = _lcu_update(device, lcu);
+ /*
+ * Need to check flags again, as there could have been another
+ * prepare_update or a new device a new device while we were still
+ * processing the data
+ */
+ spin_lock_irqsave(&lcu->lock, flags);
+ if (rc || (lcu->flags & NEED_UAC_UPDATE)) {
+ DEV_MESSAGE(KERN_WARNING, device, "could not update"
+ " alias data in lcu (rc = %d), retry later", rc);
+ schedule_delayed_work(&lcu->ruac_data.dwork, 30*HZ);
+ } else {
+ lcu->ruac_data.device = NULL;
+ lcu->flags &= ~UPDATE_PENDING;
+ }
+ spin_unlock_irqrestore(&lcu->lock, flags);
+}
+
+static int _schedule_lcu_update(struct alias_lcu *lcu,
+ struct dasd_device *device)
+{
+ struct dasd_device *usedev = NULL;
+ struct alias_pav_group *group;
+
+ lcu->flags |= NEED_UAC_UPDATE;
+ if (lcu->ruac_data.device) {
+ /* already scheduled or running */
+ return 0;
+ }
+ if (device && !list_empty(&device->alias_list))
+ usedev = device;
+
+ if (!usedev && !list_empty(&lcu->grouplist)) {
+ group = list_first_entry(&lcu->grouplist,
+ struct alias_pav_group, group);
+ if (!list_empty(&group->baselist))
+ usedev = list_first_entry(&group->baselist,
+ struct dasd_device,
+ alias_list);
+ else if (!list_empty(&group->aliaslist))
+ usedev = list_first_entry(&group->aliaslist,
+ struct dasd_device,
+ alias_list);
+ }
+ if (!usedev && !list_empty(&lcu->active_devices)) {
+ usedev = list_first_entry(&lcu->active_devices,
+ struct dasd_device, alias_list);
+ }
+ /*
+ * if we haven't found a proper device yet, give up for now, the next
+ * device that will be set active will trigger an lcu update
+ */
+ if (!usedev)
+ return -EINVAL;
+ lcu->ruac_data.device = usedev;
+ schedule_delayed_work(&lcu->ruac_data.dwork, 0);
+ return 0;
+}
+
+int dasd_alias_add_device(struct dasd_device *device)
+{
+ struct dasd_eckd_private *private;
+ struct alias_lcu *lcu;
+ unsigned long flags;
+ int rc;
+
+ private = (struct dasd_eckd_private *) device->private;
+ lcu = private->lcu;
+ rc = 0;
+ spin_lock_irqsave(&lcu->lock, flags);
+ if (!(lcu->flags & UPDATE_PENDING)) {
+ rc = _add_device_to_lcu(lcu, device);
+ if (rc)
+ lcu->flags |= UPDATE_PENDING;
+ }
+ if (lcu->flags & UPDATE_PENDING) {
+ list_move(&device->alias_list, &lcu->active_devices);
+ _schedule_lcu_update(lcu, device);
+ }
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ return rc;
+}
+
+int dasd_alias_remove_device(struct dasd_device *device)
+{
+ struct dasd_eckd_private *private;
+ struct alias_lcu *lcu;
+ unsigned long flags;
+
+ private = (struct dasd_eckd_private *) device->private;
+ lcu = private->lcu;
+ spin_lock_irqsave(&lcu->lock, flags);
+ _remove_device_from_lcu(lcu, device);
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ return 0;
+}
+
+struct dasd_device *dasd_alias_get_start_dev(struct dasd_device *base_device)
+{
+
+ struct dasd_device *alias_device;
+ struct alias_pav_group *group;
+ struct alias_lcu *lcu;
+ struct dasd_eckd_private *private, *alias_priv;
+ unsigned long flags;
+
+ private = (struct dasd_eckd_private *) base_device->private;
+ group = private->pavgroup;
+ lcu = private->lcu;
+ if (!group || !lcu)
+ return NULL;
+ if (lcu->pav == NO_PAV ||
+ lcu->flags & (NEED_UAC_UPDATE | UPDATE_PENDING))
+ return NULL;
+
+ spin_lock_irqsave(&lcu->lock, flags);
+ alias_device = group->next;
+ if (!alias_device) {
+ if (list_empty(&group->aliaslist)) {
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ return NULL;
+ } else {
+ alias_device = list_first_entry(&group->aliaslist,
+ struct dasd_device,
+ alias_list);
+ }
+ }
+ if (list_is_last(&alias_device->alias_list, &group->aliaslist))
+ group->next = list_first_entry(&group->aliaslist,
+ struct dasd_device, alias_list);
+ else
+ group->next = list_first_entry(&alias_device->alias_list,
+ struct dasd_device, alias_list);
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ alias_priv = (struct dasd_eckd_private *) alias_device->private;
+ if ((alias_priv->count < private->count) && !alias_device->stopped)
+ return alias_device;
+ else
+ return NULL;
+}
+
+/*
+ * Summary unit check handling depends on the way alias devices
+ * are handled so it is done here rather then in dasd_eckd.c
+ */
+static int reset_summary_unit_check(struct alias_lcu *lcu,
+ struct dasd_device *device,
+ char reason)
+{
+ struct dasd_ccw_req *cqr;
+ int rc = 0;
+
+ cqr = lcu->rsu_cqr;
+ strncpy((char *) &cqr->magic, "ECKD", 4);
+ ASCEBC((char *) &cqr->magic, 4);
+ cqr->cpaddr->cmd_code = DASD_ECKD_CCW_RSCK;
+ cqr->cpaddr->flags = 0 ;
+ cqr->cpaddr->count = 16;
+ cqr->cpaddr->cda = (__u32)(addr_t) cqr->data;
+ ((char *)cqr->data)[0] = reason;
+
+ clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
+ cqr->retries = 255; /* set retry counter to enable basic ERP */
+ cqr->startdev = device;
+ cqr->memdev = device;
+ cqr->block = NULL;
+ cqr->expires = 5 * HZ;
+ cqr->buildclk = get_clock();
+ cqr->status = DASD_CQR_FILLED;
+
+ rc = dasd_sleep_on_immediatly(cqr);
+ return rc;
+}
+
+static void _restart_all_base_devices_on_lcu(struct alias_lcu *lcu)
+{
+ struct alias_pav_group *pavgroup;
+ struct dasd_device *device;
+ struct dasd_eckd_private *private;
+
+ /* active and inactive list can contain alias as well as base devices */
+ list_for_each_entry(device, &lcu->active_devices, alias_list) {
+ private = (struct dasd_eckd_private *) device->private;
+ if (private->uid.type != UA_BASE_DEVICE)
+ continue;
+ dasd_schedule_block_bh(device->block);
+ dasd_schedule_device_bh(device);
+ }
+ list_for_each_entry(device, &lcu->inactive_devices, alias_list) {
+ private = (struct dasd_eckd_private *) device->private;
+ if (private->uid.type != UA_BASE_DEVICE)
+ continue;
+ dasd_schedule_block_bh(device->block);
+ dasd_schedule_device_bh(device);
+ }
+ list_for_each_entry(pavgroup, &lcu->grouplist, group) {
+ list_for_each_entry(device, &pavgroup->baselist, alias_list) {
+ dasd_schedule_block_bh(device->block);
+ dasd_schedule_device_bh(device);
+ }
+ }
+}
+
+static void flush_all_alias_devices_on_lcu(struct alias_lcu *lcu)
+{
+ struct alias_pav_group *pavgroup;
+ struct dasd_device *device, *temp;
+ struct dasd_eckd_private *private;
+ int rc;
+ unsigned long flags;
+ LIST_HEAD(active);
+
+ /*
+ * Problem here ist that dasd_flush_device_queue may wait
+ * for termination of a request to complete. We can't keep
+ * the lcu lock during that time, so we must assume that
+ * the lists may have changed.
+ * Idea: first gather all active alias devices in a separate list,
+ * then flush the first element of this list unlocked, and afterwards
+ * check if it is still on the list before moving it to the
+ * active_devices list.
+ */
+
+ spin_lock_irqsave(&lcu->lock, flags);
+ list_for_each_entry_safe(device, temp, &lcu->active_devices,
+ alias_list) {
+ private = (struct dasd_eckd_private *) device->private;
+ if (private->uid.type == UA_BASE_DEVICE)
+ continue;
+ list_move(&device->alias_list, &active);
+ }
+
+ list_for_each_entry(pavgroup, &lcu->grouplist, group) {
+ list_splice_init(&pavgroup->aliaslist, &active);
+ }
+ while (!list_empty(&active)) {
+ device = list_first_entry(&active, struct dasd_device,
+ alias_list);
+ spin_unlock_irqrestore(&lcu->lock, flags);
+ rc = dasd_flush_device_queue(device);
+ spin_lock_irqsave(&lcu->lock, flags);
+ /*
+ * only move device around if it wasn't moved away while we
+ * were waiting for the flush
+ */
+ if (device == list_first_entry(&active,
+ struct dasd_device, alias_list))
+ list_move(&device->alias_list, &lcu->active_devices);
+ }
+ spin_unlock_irqrestore(&lcu->lock, flags);
+}
+
+/*
+ * This function is called in interrupt context, so the
+ * cdev lock for device is already locked!
+ */
+static void _stop_all_devices_on_lcu(struct alias_lcu *lcu,
+ struct dasd_device *device)
+{
+ struct alias_pav_group *pavgroup;
+ struct dasd_device *pos;
+
+ list_for_each_entry(pos, &lcu->active_devices, alias_list) {
+ if (pos != device)
+ spin_lock(get_ccwdev_lock(pos->cdev));
+ pos->stopped |= DASD_STOPPED_SU;
+ if (pos != device)
+ spin_unlock(get_ccwdev_lock(pos->cdev));
+ }
+ list_for_each_entry(pos, &lcu->inactive_devices, alias_list) {
+ if (pos != device)
+ spin_lock(get_ccwdev_lock(pos->cdev));
+ pos->stopped |= DASD_STOPPED_SU;
+ if (pos != device)
+ spin_unlock(get_ccwdev_lock(pos->cdev));
+ }
+ list_for_each_entry(pavgroup, &lcu->grouplist, group) {
+ list_for_each_entry(pos, &pavgroup->baselist, alias_list) {
+ if (pos != device)
+ spin_lock(get_ccwdev_lock(pos->cdev));
+ pos->stopped |= DASD_STOPPED_SU;
+ if (pos != device)
+ spin_unlock(get_ccwdev_lock(pos->cdev));
+ }
+ list_for_each_entry(pos, &pavgroup->aliaslist, alias_list) {
+ if (pos != device)
+ spin_lock(get_ccwdev_lock(pos->cdev));
+ pos->stopped |= DASD_STOPPED_SU;
+ if (pos != device)
+ spin_unlock(get_ccwdev_lock(pos->cdev));
+ }
+ }
+}
+
+static void _unstop_all_devices_on_lcu(struct alias_lcu *lcu)
+{
+ struct alias_pav_group *pavgroup;
+ struct dasd_device *device;
+ unsigned long flags;
+
+ list_for_each_entry(device, &lcu->active_devices, alias_list) {
+ spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
+ device->stopped &= ~DASD_STOPPED_SU;
+ spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+ }
+
+ list_for_each_entry(device, &lcu->inactive_devices, alias_list) {
+ spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
+ device->stopped &= ~DASD_STOPPED_SU;
+ spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+ }
+
+ list_for_each_entry(pavgroup, &lcu->grouplist, group) {
+ list_for_each_entry(device, &pavgroup->baselist, alias_list) {
+ spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
+ device->stopped &= ~DASD_STOPPED_SU;
+ spin_unlock_irqrestore(get_ccwdev_lock(device->cdev),
+ flags);
+ }
+ list_for_each_entry(device, &pavgroup->aliaslist, alias_list) {
+ spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
+ device->stopped &= ~DASD_STOPPED_SU;
+ spin_unlock_irqrestore(get_ccwdev_lock(device->cdev),
+ flags);
+ }
+ }
+}
+
+static void summary_unit_check_handling_work(struct work_struct *work)
+{
+ struct alias_lcu *lcu;
+ struct summary_unit_check_work_data *suc_data;
+ unsigned long flags;
+ struct dasd_device *device;
+
+ suc_data = container_of(work, struct summary_unit_check_work_data,
+ worker);
+ lcu = container_of(suc_data, struct alias_lcu, suc_data);
+ device = suc_data->device;
+
+ /* 1. flush alias devices */
+ flush_all_alias_devices_on_lcu(lcu);
+
+ /* 2. reset summary unit check */
+ spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
+ device->stopped &= ~(DASD_STOPPED_SU | DASD_STOPPED_PENDING);
+ spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+ reset_summary_unit_check(lcu, device, suc_data->reason);
+
+ spin_lock_irqsave(&lcu->lock, flags);
+ _unstop_all_devices_on_lcu(lcu);
+ _restart_all_base_devices_on_lcu(lcu);
+ /* 3. read new alias configuration */
+ _schedule_lcu_update(lcu, device);
+ lcu->suc_data.device = NULL;
+ spin_unlock_irqrestore(&lcu->lock, flags);
+}
+
+/*
+ * note: this will be called from int handler context (cdev locked)
+ */
+void dasd_alias_handle_summary_unit_check(struct dasd_device *device,
+ struct irb *irb)
+{
+ struct alias_lcu *lcu;
+ char reason;
+ struct dasd_eckd_private *private;
+
+ private = (struct dasd_eckd_private *) device->private;
+
+ reason = irb->ecw[8];
+ DEV_MESSAGE(KERN_WARNING, device, "%s %x",
+ "eckd handle summary unit check: reason", reason);
+
+ lcu = private->lcu;
+ if (!lcu) {
+ DEV_MESSAGE(KERN_WARNING, device, "%s",
+ "device not ready to handle summary"
+ " unit check (no lcu structure)");
+ return;
+ }
+ spin_lock(&lcu->lock);
+ _stop_all_devices_on_lcu(lcu, device);
+ /* prepare for lcu_update */
+ private->lcu->flags |= NEED_UAC_UPDATE | UPDATE_PENDING;
+ /* If this device is about to be removed just return and wait for
+ * the next interrupt on a different device
+ */
+ if (list_empty(&device->alias_list)) {
+ DEV_MESSAGE(KERN_WARNING, device, "%s",
+ "device is in offline processing,"
+ " don't do summary unit check handling");
+ spin_unlock(&lcu->lock);
+ return;
+ }
+ if (lcu->suc_data.device) {
+ /* already scheduled or running */
+ DEV_MESSAGE(KERN_WARNING, device, "%s",
+ "previous instance of summary unit check worker"
+ " still pending");
+ spin_unlock(&lcu->lock);
+ return ;
+ }
+ lcu->suc_data.reason = reason;
+ lcu->suc_data.device = device;
+ spin_unlock(&lcu->lock);
+ schedule_work(&lcu->suc_data.worker);
+};
Index: quilt-2.6/drivers/s390/block/dasd_int.h
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_int.h
+++ quilt-2.6/drivers/s390/block/dasd_int.h
@@ -64,13 +64,7 @@
* SECTION: Type definitions
*/
struct dasd_device;
-
-typedef enum {
- dasd_era_fatal = -1, /* no chance to recover */
- dasd_era_none = 0, /* don't recover, everything alright */
- dasd_era_msg = 1, /* don't recover, just report... */
- dasd_era_recover = 2 /* recovery action recommended */
-} dasd_era_t;
+struct dasd_block;
/* BIT DEFINITIONS FOR SENSE DATA */
#define DASD_SENSE_BIT_0 0x80
@@ -151,19 +145,22 @@ do { \
struct dasd_ccw_req {
unsigned int magic; /* Eye catcher */
- struct list_head list; /* list_head for request queueing. */
+ struct list_head devlist; /* for dasd_device request queue */
+ struct list_head blocklist; /* for dasd_block request queue */
/* Where to execute what... */
- struct dasd_device *device; /* device the request is for */
+ struct dasd_block *block; /* the originating block device */
+ struct dasd_device *memdev; /* the device used to allocate this */
+ struct dasd_device *startdev; /* device the request is started on */
struct ccw1 *cpaddr; /* address of channel program */
- char status; /* status of this request */
+ char status; /* status of this request */
short retries; /* A retry counter */
unsigned long flags; /* flags of this request */
/* ... and how */
unsigned long starttime; /* jiffies time of request start */
int expires; /* expiration period in jiffies */
- char lpm; /* logical path mask */
+ char lpm; /* logical path mask */
void *data; /* pointer to data area */
/* these are important for recovering erroneous requests */
@@ -178,20 +175,27 @@ struct dasd_ccw_req {
unsigned long long endclk; /* TOD-clock of request termination */
/* Callback that is called after reaching final status. */
- void (*callback)(struct dasd_ccw_req *, void *data);
- void *callback_data;
+ void (*callback)(struct dasd_ccw_req *, void *data);
+ void *callback_data;
};
/*
* dasd_ccw_req -> status can be:
*/
-#define DASD_CQR_FILLED 0x00 /* request is ready to be processed */
-#define DASD_CQR_QUEUED 0x01 /* request is queued to be processed */
-#define DASD_CQR_IN_IO 0x02 /* request is currently in IO */
-#define DASD_CQR_DONE 0x03 /* request is completed successfully */
-#define DASD_CQR_ERROR 0x04 /* request is completed with error */
-#define DASD_CQR_FAILED 0x05 /* request is finally failed */
-#define DASD_CQR_CLEAR 0x06 /* request is clear pending */
+#define DASD_CQR_FILLED 0x00 /* request is ready to be processed */
+#define DASD_CQR_DONE 0x01 /* request is completed successfully */
+#define DASD_CQR_NEED_ERP 0x02 /* request needs recovery action */
+#define DASD_CQR_IN_ERP 0x03 /* request is in recovery */
+#define DASD_CQR_FAILED 0x04 /* request is finally failed */
+#define DASD_CQR_TERMINATED 0x05 /* request was successfull */
+
+#define DASD_CQR_QUEUED 0x80 /* request is queued to be processed */
+#define DASD_CQR_IN_IO 0x81 /* request is currently in IO */
+#define DASD_CQR_ERROR 0x82 /* request is completed with error */
+#define DASD_CQR_CLEAR_PENDING 0x83 /* request is clear pending */
+#define DASD_CQR_CLEARED 0x84 /* request was cleared */
+#define DASD_CQR_SUCCESS 0x85 /* request was successfull */
+
/* per dasd_ccw_req flags */
#define DASD_CQR_FLAGS_USE_ERP 0 /* use ERP for this request */
@@ -218,29 +222,46 @@ struct dasd_discipline {
* Device recognition functions. check_device is used to verify
* the sense data and the information returned by read device
* characteristics. It returns 0 if the discipline can be used
- * for the device in question.
+ * for the device in question. uncheck_device is called during
+ * device shutdown to deregister a device from it's discipline.
+ */
+ int (*check_device)(struct dasd_device *);
+ void (*uncheck_device)(struct dasd_device *);
+
+ /*
* do_analysis is used in the step from device state "basic" to
* state "accept". It returns 0 if the device can be made ready,
* it returns -EMEDIUMTYPE if the device can't be made ready or
* -EAGAIN if do_analysis started a ccw that needs to complete
* before the analysis may be repeated.
*/
- int (*check_device)(struct dasd_device *);
- int (*do_analysis) (struct dasd_device *);
+ int (*do_analysis) (struct dasd_block *);
+
+ /*
+ * Last things to do when a device is set online, and first things
+ * when it is set offline.
+ */
+ int (*ready_to_online) (struct dasd_device *);
+ int (*online_to_ready) (struct dasd_device *);
/*
* Device operation functions. build_cp creates a ccw chain for
* a block device request, start_io starts the request and
* term_IO cancels it (e.g. in case of a timeout). format_device
* returns a ccw chain to be used to format the device.
+ * handle_terminated_request allows to examine a cqr and prepare
+ * it for retry.
*/
struct dasd_ccw_req *(*build_cp) (struct dasd_device *,
+ struct dasd_block *,
struct request *);
int (*start_IO) (struct dasd_ccw_req *);
int (*term_IO) (struct dasd_ccw_req *);
+ void (*handle_terminated_request) (struct dasd_ccw_req *);
struct dasd_ccw_req *(*format_device) (struct dasd_device *,
struct format_data_t *);
int (*free_cp) (struct dasd_ccw_req *, struct request *);
+
/*
* Error recovery functions. examine_error() returns a value that
* indicates what to do for an error condition. If examine_error()
@@ -250,16 +271,17 @@ struct dasd_discipline {
* is called for every error condition to print the sense data
* to the console.
*/
- dasd_era_t(*examine_error) (struct dasd_ccw_req *, struct irb *);
dasd_erp_fn_t(*erp_action) (struct dasd_ccw_req *);
dasd_erp_fn_t(*erp_postaction) (struct dasd_ccw_req *);
void (*dump_sense) (struct dasd_device *, struct dasd_ccw_req *,
struct irb *);
+ void (*handle_unsolicited_interrupt) (struct dasd_device *, struct irb *);
+
/* i/o control functions. */
- int (*fill_geometry) (struct dasd_device *, struct hd_geometry *);
+ int (*fill_geometry) (struct dasd_block *, struct hd_geometry *);
int (*fill_info) (struct dasd_device *, struct dasd_information2_t *);
- int (*ioctl) (struct dasd_device *, unsigned int, void __user *);
+ int (*ioctl) (struct dasd_block *, unsigned int, void __user *);
};
extern struct dasd_discipline *dasd_diag_discipline_pointer;
@@ -267,12 +289,18 @@ extern struct dasd_discipline *dasd_diag
/*
* Unique identifier for dasd device.
*/
+#define UA_NOT_CONFIGURED 0x00
+#define UA_BASE_DEVICE 0x01
+#define UA_BASE_PAV_ALIAS 0x02
+#define UA_HYPER_PAV_ALIAS 0x03
+
struct dasd_uid {
- __u8 alias;
+ __u8 type;
char vendor[4];
char serial[15];
__u16 ssid;
- __u8 unit_addr;
+ __u8 real_unit_addr;
+ __u8 base_unit_addr;
};
/*
@@ -293,14 +321,9 @@ struct dasd_uid {
struct dasd_device {
/* Block device stuff. */
- struct gendisk *gdp;
- struct request_queue *request_queue;
- spinlock_t request_queue_lock;
- struct block_device *bdev;
+ struct dasd_block *block;
+
unsigned int devindex;
- unsigned long blocks; /* size of volume in blocks */
- unsigned int bp_block; /* bytes per block */
- unsigned int s2b_shift; /* log2 (bp_block/512) */
unsigned long flags; /* per device flags */
unsigned short features; /* copy of devmap-features (read-only!) */
@@ -316,9 +339,8 @@ struct dasd_device {
int state, target;
int stopped; /* device (ccw_device_start) was stopped */
- /* Open and reference count. */
+ /* reference count. */
atomic_t ref_count;
- atomic_t open_count;
/* ccw queue and memory for static ccw/erp buffers. */
struct list_head ccw_queue;
@@ -337,20 +359,45 @@ struct dasd_device {
struct ccw_device *cdev;
+ /* hook for alias management */
+ struct list_head alias_list;
+};
+
+struct dasd_block {
+ /* Block device stuff. */
+ struct gendisk *gdp;
+ struct request_queue *request_queue;
+ spinlock_t request_queue_lock;
+ struct block_device *bdev;
+ atomic_t open_count;
+
+ unsigned long blocks; /* size of volume in blocks */
+ unsigned int bp_block; /* bytes per block */
+ unsigned int s2b_shift; /* log2 (bp_block/512) */
+
+ struct dasd_device *base;
+ struct list_head ccw_queue;
+ spinlock_t queue_lock;
+
+ atomic_t tasklet_scheduled;
+ struct tasklet_struct tasklet;
+ struct timer_list timer;
+
#ifdef CONFIG_DASD_PROFILE
struct dasd_profile_info_t profile;
#endif
};
+
+
/* reasons why device (ccw_device_start) was stopped */
#define DASD_STOPPED_NOT_ACC 1 /* not accessible */
#define DASD_STOPPED_QUIESCE 2 /* Quiesced */
#define DASD_STOPPED_PENDING 4 /* long busy */
#define DASD_STOPPED_DC_WAIT 8 /* disconnected, wait */
-#define DASD_STOPPED_DC_EIO 16 /* disconnected, return -EIO */
+#define DASD_STOPPED_SU 16 /* summary unit check handling */
/* per device flags */
-#define DASD_FLAG_DSC_ERROR 2 /* return -EIO when disconnected */
#define DASD_FLAG_OFFLINE 3 /* device is in offline processing */
#define DASD_FLAG_EER_SNSS 4 /* A SNSS is required */
#define DASD_FLAG_EER_IN_USE 5 /* A SNSS request is running */
@@ -489,6 +536,9 @@ dasd_kmalloc_set_cda(struct ccw1 *ccw, v
struct dasd_device *dasd_alloc_device(void);
void dasd_free_device(struct dasd_device *);
+struct dasd_block *dasd_alloc_block(void);
+void dasd_free_block(struct dasd_block *);
+
void dasd_enable_device(struct dasd_device *);
void dasd_set_target_state(struct dasd_device *, int);
void dasd_kick_device(struct dasd_device *);
@@ -497,18 +547,23 @@ void dasd_add_request_head(struct dasd_c
void dasd_add_request_tail(struct dasd_ccw_req *);
int dasd_start_IO(struct dasd_ccw_req *);
int dasd_term_IO(struct dasd_ccw_req *);
-void dasd_schedule_bh(struct dasd_device *);
+void dasd_schedule_device_bh(struct dasd_device *);
+void dasd_schedule_block_bh(struct dasd_block *);
int dasd_sleep_on(struct dasd_ccw_req *);
int dasd_sleep_on_immediatly(struct dasd_ccw_req *);
int dasd_sleep_on_interruptible(struct dasd_ccw_req *);
-void dasd_set_timer(struct dasd_device *, int);
-void dasd_clear_timer(struct dasd_device *);
+void dasd_device_set_timer(struct dasd_device *, int);
+void dasd_device_clear_timer(struct dasd_device *);
+void dasd_block_set_timer(struct dasd_block *, int);
+void dasd_block_clear_timer(struct dasd_block *);
int dasd_cancel_req(struct dasd_ccw_req *);
+int dasd_flush_device_queue(struct dasd_device *);
int dasd_generic_probe (struct ccw_device *, struct dasd_discipline *);
void dasd_generic_remove (struct ccw_device *cdev);
int dasd_generic_set_online(struct ccw_device *, struct dasd_discipline *);
int dasd_generic_set_offline (struct ccw_device *cdev);
int dasd_generic_notify(struct ccw_device *, int);
+void dasd_generic_handle_state_change(struct dasd_device *);
int dasd_generic_read_dev_chars(struct dasd_device *, char *, void **, int);
@@ -542,10 +597,10 @@ int dasd_busid_known(char *);
/* externals in dasd_gendisk.c */
int dasd_gendisk_init(void);
void dasd_gendisk_exit(void);
-int dasd_gendisk_alloc(struct dasd_device *);
-void dasd_gendisk_free(struct dasd_device *);
-int dasd_scan_partitions(struct dasd_device *);
-void dasd_destroy_partitions(struct dasd_device *);
+int dasd_gendisk_alloc(struct dasd_block *);
+void dasd_gendisk_free(struct dasd_block *);
+int dasd_scan_partitions(struct dasd_block *);
+void dasd_destroy_partitions(struct dasd_block *);
/* externals in dasd_ioctl.c */
int dasd_ioctl(struct inode *, struct file *, unsigned int, unsigned long);
@@ -563,20 +618,9 @@ struct dasd_ccw_req *dasd_alloc_erp_requ
void dasd_free_erp_request(struct dasd_ccw_req *, struct dasd_device *);
void dasd_log_sense(struct dasd_ccw_req *, struct irb *);
-/* externals in dasd_3370_erp.c */
-dasd_era_t dasd_3370_erp_examine(struct dasd_ccw_req *, struct irb *);
-
/* externals in dasd_3990_erp.c */
-dasd_era_t dasd_3990_erp_examine(struct dasd_ccw_req *, struct irb *);
struct dasd_ccw_req *dasd_3990_erp_action(struct dasd_ccw_req *);
-/* externals in dasd_9336_erp.c */
-dasd_era_t dasd_9336_erp_examine(struct dasd_ccw_req *, struct irb *);
-
-/* externals in dasd_9336_erp.c */
-dasd_era_t dasd_9343_erp_examine(struct dasd_ccw_req *, struct irb *);
-struct dasd_ccw_req *dasd_9343_erp_action(struct dasd_ccw_req *);
-
/* externals in dasd_eer.c */
#ifdef CONFIG_DASD_EER
int dasd_eer_init(void);
Index: quilt-2.6/drivers/s390/block/dasd_ioctl.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_ioctl.c
+++ quilt-2.6/drivers/s390/block/dasd_ioctl.c
@@ -38,15 +38,15 @@ dasd_ioctl_api_version(void __user *argp
static int
dasd_ioctl_enable(struct block_device *bdev)
{
- struct dasd_device *device = bdev->bd_disk->private_data;
+ struct dasd_block *block = bdev->bd_disk->private_data;
if (!capable(CAP_SYS_ADMIN))
return -EACCES;
- dasd_enable_device(device);
+ dasd_enable_device(block->base);
/* Formatting the dasd device can change the capacity. */
mutex_lock(&bdev->bd_mutex);
- i_size_write(bdev->bd_inode, (loff_t)get_capacity(device->gdp) << 9);
+ i_size_write(bdev->bd_inode, (loff_t)get_capacity(block->gdp) << 9);
mutex_unlock(&bdev->bd_mutex);
return 0;
}
@@ -58,7 +58,7 @@ dasd_ioctl_enable(struct block_device *b
static int
dasd_ioctl_disable(struct block_device *bdev)
{
- struct dasd_device *device = bdev->bd_disk->private_data;
+ struct dasd_block *block = bdev->bd_disk->private_data;
if (!capable(CAP_SYS_ADMIN))
return -EACCES;
@@ -71,7 +71,7 @@ dasd_ioctl_disable(struct block_device *
* using the BIODASDFMT ioctl. Therefore the correct state for the
* device is DASD_STATE_BASIC that allows to do basic i/o.
*/
- dasd_set_target_state(device, DASD_STATE_BASIC);
+ dasd_set_target_state(block->base, DASD_STATE_BASIC);
/*
* Set i_size to zero, since read, write, etc. check against this
* value.
@@ -86,18 +86,20 @@ dasd_ioctl_disable(struct block_device *
* Quiesce device.
*/
static int
-dasd_ioctl_quiesce(struct dasd_device *device)
+dasd_ioctl_quiesce(struct dasd_block *block)
{
unsigned long flags;
+ struct dasd_device *base;
+ base = block->base;
if (!capable (CAP_SYS_ADMIN))
return -EACCES;
- DEV_MESSAGE (KERN_DEBUG, device, "%s",
- "Quiesce IO on device");
- spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
- device->stopped |= DASD_STOPPED_QUIESCE;
- spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+ DEV_MESSAGE(KERN_DEBUG, base, "%s",
+ "Quiesce IO on device");
+ spin_lock_irqsave(get_ccwdev_lock(base->cdev), flags);
+ base->stopped |= DASD_STOPPED_QUIESCE;
+ spin_unlock_irqrestore(get_ccwdev_lock(base->cdev), flags);
return 0;
}
@@ -106,21 +108,23 @@ dasd_ioctl_quiesce(struct dasd_device *d
* Quiesce device.
*/
static int
-dasd_ioctl_resume(struct dasd_device *device)
+dasd_ioctl_resume(struct dasd_block *block)
{
unsigned long flags;
+ struct dasd_device *base;
+ base = block->base;
if (!capable (CAP_SYS_ADMIN))
return -EACCES;
- DEV_MESSAGE (KERN_DEBUG, device, "%s",
- "resume IO on device");
+ DEV_MESSAGE(KERN_DEBUG, base, "%s",
+ "resume IO on device");
- spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
- device->stopped &= ~DASD_STOPPED_QUIESCE;
- spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+ spin_lock_irqsave(get_ccwdev_lock(base->cdev), flags);
+ base->stopped &= ~DASD_STOPPED_QUIESCE;
+ spin_unlock_irqrestore(get_ccwdev_lock(base->cdev), flags);
- dasd_schedule_bh (device);
+ dasd_schedule_block_bh(block);
return 0;
}
@@ -131,21 +135,23 @@ dasd_ioctl_resume(struct dasd_device *de
* devices this means CCWs are generated to format a single track.
*/
static int
-dasd_format(struct dasd_device * device, struct format_data_t * fdata)
+dasd_format(struct dasd_block *block, struct format_data_t *fdata)
{
struct dasd_ccw_req *cqr;
+ struct dasd_device *base;
int rc;
- if (device->discipline->format_device == NULL)
+ base = block->base;
+ if (base->discipline->format_device == NULL)
return -EPERM;
- if (device->state != DASD_STATE_BASIC) {
- DEV_MESSAGE(KERN_WARNING, device, "%s",
+ if (base->state != DASD_STATE_BASIC) {
+ DEV_MESSAGE(KERN_WARNING, base, "%s",
"dasd_format: device is not disabled! ");
return -EBUSY;
}
- DBF_DEV_EVENT(DBF_NOTICE, device,
+ DBF_DEV_EVENT(DBF_NOTICE, base,
"formatting units %d to %d (%d B blocks) flags %d",
fdata->start_unit,
fdata->stop_unit, fdata->blksize, fdata->intensity);
@@ -156,20 +162,20 @@ dasd_format(struct dasd_device * device,
* enabling the device later.
*/
if (fdata->start_unit == 0) {
- struct block_device *bdev = bdget_disk(device->gdp, 0);
+ struct block_device *bdev = bdget_disk(block->gdp, 0);
bdev->bd_inode->i_blkbits = blksize_bits(fdata->blksize);
bdput(bdev);
}
while (fdata->start_unit <= fdata->stop_unit) {
- cqr = device->discipline->format_device(device, fdata);
+ cqr = base->discipline->format_device(base, fdata);
if (IS_ERR(cqr))
return PTR_ERR(cqr);
rc = dasd_sleep_on_interruptible(cqr);
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
if (rc) {
if (rc != -ERESTARTSYS)
- DEV_MESSAGE(KERN_ERR, device,
+ DEV_MESSAGE(KERN_ERR, base,
" Formatting of unit %d failed "
"with rc = %d",
fdata->start_unit, rc);
@@ -186,7 +192,7 @@ dasd_format(struct dasd_device * device,
static int
dasd_ioctl_format(struct block_device *bdev, void __user *argp)
{
- struct dasd_device *device = bdev->bd_disk->private_data;
+ struct dasd_block *block = bdev->bd_disk->private_data;
struct format_data_t fdata;
if (!capable(CAP_SYS_ADMIN))
@@ -194,16 +200,16 @@ dasd_ioctl_format(struct block_device *b
if (!argp)
return -EINVAL;
- if (device->features & DASD_FEATURE_READONLY)
+ if (block->base->features & DASD_FEATURE_READONLY)
return -EROFS;
if (copy_from_user(&fdata, argp, sizeof(struct format_data_t)))
return -EFAULT;
if (bdev != bdev->bd_contains) {
- DEV_MESSAGE(KERN_WARNING, device, "%s",
+ DEV_MESSAGE(KERN_WARNING, block->base, "%s",
"Cannot low-level format a partition");
return -EINVAL;
}
- return dasd_format(device, &fdata);
+ return dasd_format(block, &fdata);
}
#ifdef CONFIG_DASD_PROFILE
@@ -211,9 +217,9 @@ dasd_ioctl_format(struct block_device *b
* Reset device profile information
*/
static int
-dasd_ioctl_reset_profile(struct dasd_device *device)
+dasd_ioctl_reset_profile(struct dasd_block *block)
{
- memset(&device->profile, 0, sizeof (struct dasd_profile_info_t));
+ memset(&block->profile, 0, sizeof(struct dasd_profile_info_t));
return 0;
}
@@ -221,24 +227,24 @@ dasd_ioctl_reset_profile(struct dasd_dev
* Return device profile information
*/
static int
-dasd_ioctl_read_profile(struct dasd_device *device, void __user *argp)
+dasd_ioctl_read_profile(struct dasd_block *block, void __user *argp)
{
if (dasd_profile_level == DASD_PROFILE_OFF)
return -EIO;
- if (copy_to_user(argp, &device->profile,
- sizeof (struct dasd_profile_info_t)))
+ if (copy_to_user(argp, &block->profile,
+ sizeof(struct dasd_profile_info_t)))
return -EFAULT;
return 0;
}
#else
static int
-dasd_ioctl_reset_profile(struct dasd_device *device)
+dasd_ioctl_reset_profile(struct dasd_block *block)
{
return -ENOSYS;
}
static int
-dasd_ioctl_read_profile(struct dasd_device *device, void __user *argp)
+dasd_ioctl_read_profile(struct dasd_block *block, void __user *argp)
{
return -ENOSYS;
}
@@ -247,87 +253,88 @@ dasd_ioctl_read_profile(struct dasd_devi
/*
* Return dasd information. Used for BIODASDINFO and BIODASDINFO2.
*/
-static int
-dasd_ioctl_information(struct dasd_device *device,
- unsigned int cmd, void __user *argp)
+static int dasd_ioctl_information(struct dasd_block *block,
+ unsigned int cmd, void __user *argp)
{
struct dasd_information2_t *dasd_info;
unsigned long flags;
int rc;
+ struct dasd_device *base;
struct ccw_device *cdev;
struct ccw_dev_id dev_id;
- if (!device->discipline->fill_info)
+ base = block->base;
+ if (!base->discipline->fill_info)
return -EINVAL;
dasd_info = kzalloc(sizeof(struct dasd_information2_t), GFP_KERNEL);
if (dasd_info == NULL)
return -ENOMEM;
- rc = device->discipline->fill_info(device, dasd_info);
+ rc = base->discipline->fill_info(base, dasd_info);
if (rc) {
kfree(dasd_info);
return rc;
}
- cdev = device->cdev;
+ cdev = base->cdev;
ccw_device_get_id(cdev, &dev_id);
dasd_info->devno = dev_id.devno;
- dasd_info->schid = _ccw_device_get_subchannel_number(device->cdev);
+ dasd_info->schid = _ccw_device_get_subchannel_number(base->cdev);
dasd_info->cu_type = cdev->id.cu_type;
dasd_info->cu_model = cdev->id.cu_model;
dasd_info->dev_type = cdev->id.dev_type;
dasd_info->dev_model = cdev->id.dev_model;
- dasd_info->status = device->state;
+ dasd_info->status = base->state;
/*
* The open_count is increased for every opener, that includes
* the blkdev_get in dasd_scan_partitions.
* This must be hidden from user-space.
*/
- dasd_info->open_count = atomic_read(&device->open_count);
- if (!device->bdev)
+ dasd_info->open_count = atomic_read(&block->open_count);
+ if (!block->bdev)
dasd_info->open_count++;
/*
* check if device is really formatted
* LDL / CDL was returned by 'fill_info'
*/
- if ((device->state < DASD_STATE_READY) ||
- (dasd_check_blocksize(device->bp_block)))
+ if ((base->state < DASD_STATE_READY) ||
+ (dasd_check_blocksize(block->bp_block)))
dasd_info->format = DASD_FORMAT_NONE;
dasd_info->features |=
- ((device->features & DASD_FEATURE_READONLY) != 0);
+ ((base->features & DASD_FEATURE_READONLY) != 0);
- if (device->discipline)
- memcpy(dasd_info->type, device->discipline->name, 4);
+ if (base->discipline)
+ memcpy(dasd_info->type, base->discipline->name, 4);
else
memcpy(dasd_info->type, "none", 4);
- if (device->request_queue->request_fn) {
+ if (block->request_queue->request_fn) {
struct list_head *l;
#ifdef DASD_EXTENDED_PROFILING
{
struct list_head *l;
- spin_lock_irqsave(&device->lock, flags);
- list_for_each(l, &device->request_queue->queue_head)
+ spin_lock_irqsave(&block->lock, flags);
+ list_for_each(l, &block->request_queue->queue_head)
dasd_info->req_queue_len++;
- spin_unlock_irqrestore(&device->lock, flags);
+ spin_unlock_irqrestore(&block->lock, flags);
}
#endif /* DASD_EXTENDED_PROFILING */
- spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
- list_for_each(l, &device->ccw_queue)
+ spin_lock_irqsave(get_ccwdev_lock(base->cdev), flags);
+ list_for_each(l, &base->ccw_queue)
dasd_info->chanq_len++;
- spin_unlock_irqrestore(get_ccwdev_lock(device->cdev),
+ spin_unlock_irqrestore(get_ccwdev_lock(base->cdev),
flags);
}
rc = 0;
if (copy_to_user(argp, dasd_info,
((cmd == (unsigned int) BIODASDINFO2) ?
- sizeof (struct dasd_information2_t) :
- sizeof (struct dasd_information_t))))
+ sizeof(struct dasd_information2_t) :
+ sizeof(struct dasd_information_t))))
rc = -EFAULT;
kfree(dasd_info);
return rc;
@@ -339,7 +346,7 @@ dasd_ioctl_information(struct dasd_devic
static int
dasd_ioctl_set_ro(struct block_device *bdev, void __user *argp)
{
- struct dasd_device *device = bdev->bd_disk->private_data;
+ struct dasd_block *block = bdev->bd_disk->private_data;
int intval;
if (!capable(CAP_SYS_ADMIN))
@@ -351,11 +358,11 @@ dasd_ioctl_set_ro(struct block_device *b
return -EFAULT;
set_disk_ro(bdev->bd_disk, intval);
- return dasd_set_feature(device->cdev, DASD_FEATURE_READONLY, intval);
+ return dasd_set_feature(block->base->cdev, DASD_FEATURE_READONLY, intval);
}
static int
-dasd_ioctl_readall_cmb(struct dasd_device *device, unsigned int cmd,
+dasd_ioctl_readall_cmb(struct dasd_block *block, unsigned int cmd,
unsigned long arg)
{
struct cmbdata __user *argp = (void __user *) arg;
@@ -363,7 +370,7 @@ dasd_ioctl_readall_cmb(struct dasd_devic
struct cmbdata data;
int ret;
- ret = cmf_readall(device->cdev, &data);
+ ret = cmf_readall(block->base->cdev, &data);
if (!ret && copy_to_user(argp, &data, min(size, sizeof(*argp))))
return -EFAULT;
return ret;
@@ -374,10 +381,10 @@ dasd_ioctl(struct inode *inode, struct f
unsigned int cmd, unsigned long arg)
{
struct block_device *bdev = inode->i_bdev;
- struct dasd_device *device = bdev->bd_disk->private_data;
+ struct dasd_block *block = bdev->bd_disk->private_data;
void __user *argp = (void __user *)arg;
- if (!device)
+ if (!block)
return -ENODEV;
if ((_IOC_DIR(cmd) != _IOC_NONE) && !arg) {
@@ -391,33 +398,33 @@ dasd_ioctl(struct inode *inode, struct f
case BIODASDENABLE:
return dasd_ioctl_enable(bdev);
case BIODASDQUIESCE:
- return dasd_ioctl_quiesce(device);
+ return dasd_ioctl_quiesce(block);
case BIODASDRESUME:
- return dasd_ioctl_resume(device);
+ return dasd_ioctl_resume(block);
case BIODASDFMT:
return dasd_ioctl_format(bdev, argp);
case BIODASDINFO:
- return dasd_ioctl_information(device, cmd, argp);
+ return dasd_ioctl_information(block, cmd, argp);
case BIODASDINFO2:
- return dasd_ioctl_information(device, cmd, argp);
+ return dasd_ioctl_information(block, cmd, argp);
case BIODASDPRRD:
- return dasd_ioctl_read_profile(device, argp);
+ return dasd_ioctl_read_profile(block, argp);
case BIODASDPRRST:
- return dasd_ioctl_reset_profile(device);
+ return dasd_ioctl_reset_profile(block);
case BLKROSET:
return dasd_ioctl_set_ro(bdev, argp);
case DASDAPIVER:
return dasd_ioctl_api_version(argp);
case BIODASDCMFENABLE:
- return enable_cmf(device->cdev);
+ return enable_cmf(block->base->cdev);
case BIODASDCMFDISABLE:
- return disable_cmf(device->cdev);
+ return disable_cmf(block->base->cdev);
case BIODASDREADALLCMB:
- return dasd_ioctl_readall_cmb(device, cmd, arg);
+ return dasd_ioctl_readall_cmb(block, cmd, arg);
default:
/* if the discipline has an ioctl method try it. */
- if (device->discipline->ioctl) {
- int rval = device->discipline->ioctl(device, cmd, argp);
+ if (block->base->discipline->ioctl) {
+ int rval = block->base->discipline->ioctl(block, cmd, argp);
if (rval != -ENOIOCTLCMD)
return rval;
}
Index: quilt-2.6/drivers/s390/block/dasd_proc.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_proc.c
+++ quilt-2.6/drivers/s390/block/dasd_proc.c
@@ -54,11 +54,16 @@ static int
dasd_devices_show(struct seq_file *m, void *v)
{
struct dasd_device *device;
+ struct dasd_block *block;
char *substr;
device = dasd_device_from_devindex((unsigned long) v - 1);
if (IS_ERR(device))
return 0;
+ if (device->block)
+ block = device->block;
+ else
+ return 0;
/* Print device number. */
seq_printf(m, "%s", device->cdev->dev.bus_id);
/* Print discipline string. */
@@ -67,14 +72,14 @@ dasd_devices_show(struct seq_file *m, vo
else
seq_printf(m, "(none)");
/* Print kdev. */
- if (device->gdp)
+ if (block->gdp)
seq_printf(m, " at (%3d:%6d)",
- device->gdp->major, device->gdp->first_minor);
+ block->gdp->major, block->gdp->first_minor);
else
seq_printf(m, " at (???:??????)");
/* Print device name. */
- if (device->gdp)
- seq_printf(m, " is %-8s", device->gdp->disk_name);
+ if (block->gdp)
+ seq_printf(m, " is %-8s", block->gdp->disk_name);
else
seq_printf(m, " is ????????");
/* Print devices features. */
@@ -100,14 +105,14 @@ dasd_devices_show(struct seq_file *m, vo
case DASD_STATE_READY:
case DASD_STATE_ONLINE:
seq_printf(m, "active ");
- if (dasd_check_blocksize(device->bp_block))
+ if (dasd_check_blocksize(block->bp_block))
seq_printf(m, "n/f ");
else
seq_printf(m,
"at blocksize: %d, %ld blocks, %ld MB",
- device->bp_block, device->blocks,
- ((device->bp_block >> 9) *
- device->blocks) >> 11);
+ block->bp_block, block->blocks,
+ ((block->bp_block >> 9) *
+ block->blocks) >> 11);
break;
default:
seq_printf(m, "no stat");
Index: quilt-2.6/drivers/s390/block/Makefile
===================================================================
--- quilt-2.6.orig/drivers/s390/block/Makefile
+++ quilt-2.6/drivers/s390/block/Makefile
@@ -2,8 +2,8 @@
# S/390 block devices
#
-dasd_eckd_mod-objs := dasd_eckd.o dasd_3990_erp.o dasd_9343_erp.o
-dasd_fba_mod-objs := dasd_fba.o dasd_3370_erp.o dasd_9336_erp.o
+dasd_eckd_mod-objs := dasd_eckd.o dasd_3990_erp.o dasd_alias.o
+dasd_fba_mod-objs := dasd_fba.o
dasd_diag_mod-objs := dasd_diag.o
dasd_mod-objs := dasd.o dasd_ioctl.o dasd_proc.o dasd_devmap.o \
dasd_genhd.o dasd_erp.o
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 46/47] dasd: add hyper PAV support to DASD device driver, part 2
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (44 preceding siblings ...)
2007-12-20 15:20 ` [patch 45/47] dasd: add hyper PAV support to DASD device driver, part 1 Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
2007-12-20 15:20 ` [patch 47/47] dasd: add hyper PAV support to DASD device driver, part 3 Martin Schwidefsky
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Stefan Weinhuber, Martin Schwidefsky
[-- Attachment #1: 145-hyperpav-2.diff --]
[-- Type: text/plain, Size: 79278 bytes --]
From: Stefan Weinhuber <wein@de.ibm.com>
Signed-off-by: Stefan Weinhuber <wein@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/block/dasd.c | 1661 +++++++++++++++++++++++++---------------------
1 file changed, 938 insertions(+), 723 deletions(-)
Index: quilt-2.6/drivers/s390/block/dasd.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd.c
+++ quilt-2.6/drivers/s390/block/dasd.c
@@ -48,13 +48,15 @@ MODULE_LICENSE("GPL");
/*
* SECTION: prototypes for static functions of dasd.c
*/
-static int dasd_alloc_queue(struct dasd_device * device);
-static void dasd_setup_queue(struct dasd_device * device);
-static void dasd_free_queue(struct dasd_device * device);
-static void dasd_flush_request_queue(struct dasd_device *);
-static int dasd_flush_ccw_queue(struct dasd_device *, int);
-static void dasd_tasklet(struct dasd_device *);
+static int dasd_alloc_queue(struct dasd_block *);
+static void dasd_setup_queue(struct dasd_block *);
+static void dasd_free_queue(struct dasd_block *);
+static void dasd_flush_request_queue(struct dasd_block *);
+static int dasd_flush_block_queue(struct dasd_block *);
+static void dasd_device_tasklet(struct dasd_device *);
+static void dasd_block_tasklet(struct dasd_block *);
static void do_kick_device(struct work_struct *);
+static void dasd_return_cqr_cb(struct dasd_ccw_req *, void *);
/*
* SECTION: Operations on the device structure.
@@ -65,16 +67,13 @@ static wait_queue_head_t dasd_flush_wq;
/*
* Allocate memory for a new device structure.
*/
-struct dasd_device *
-dasd_alloc_device(void)
+struct dasd_device *dasd_alloc_device(void)
{
struct dasd_device *device;
- device = kzalloc(sizeof (struct dasd_device), GFP_ATOMIC);
+ device = kzalloc(sizeof(struct dasd_device), GFP_ATOMIC);
if (device == NULL)
return ERR_PTR(-ENOMEM);
- /* open_count = 0 means device online but not in use */
- atomic_set(&device->open_count, -1);
/* Get two pages for normal block device operations. */
device->ccw_mem = (void *) __get_free_pages(GFP_ATOMIC | GFP_DMA, 1);
@@ -93,10 +92,9 @@ dasd_alloc_device(void)
dasd_init_chunklist(&device->ccw_chunks, device->ccw_mem, PAGE_SIZE*2);
dasd_init_chunklist(&device->erp_chunks, device->erp_mem, PAGE_SIZE);
spin_lock_init(&device->mem_lock);
- spin_lock_init(&device->request_queue_lock);
- atomic_set (&device->tasklet_scheduled, 0);
+ atomic_set(&device->tasklet_scheduled, 0);
tasklet_init(&device->tasklet,
- (void (*)(unsigned long)) dasd_tasklet,
+ (void (*)(unsigned long)) dasd_device_tasklet,
(unsigned long) device);
INIT_LIST_HEAD(&device->ccw_queue);
init_timer(&device->timer);
@@ -110,8 +108,7 @@ dasd_alloc_device(void)
/*
* Free memory of a device structure.
*/
-void
-dasd_free_device(struct dasd_device *device)
+void dasd_free_device(struct dasd_device *device)
{
kfree(device->private);
free_page((unsigned long) device->erp_mem);
@@ -120,10 +117,42 @@ dasd_free_device(struct dasd_device *dev
}
/*
+ * Allocate memory for a new device structure.
+ */
+struct dasd_block *dasd_alloc_block(void)
+{
+ struct dasd_block *block;
+
+ block = kzalloc(sizeof(*block), GFP_ATOMIC);
+ if (block == NULL)
+ return ERR_PTR(-ENOMEM);
+ /* open_count = 0 means device online but not in use */
+ atomic_set(&block->open_count, -1);
+
+ spin_lock_init(&block->request_queue_lock);
+ atomic_set (&block->tasklet_scheduled, 0);
+ tasklet_init(&block->tasklet,
+ (void (*)(unsigned long)) dasd_block_tasklet,
+ (unsigned long) block);
+ INIT_LIST_HEAD(&block->ccw_queue);
+ spin_lock_init(&block->queue_lock);
+ init_timer(&block->timer);
+
+ return block;
+}
+
+/*
+ * Free memory of a device structure.
+ */
+void dasd_free_block(struct dasd_block *block)
+{
+ kfree(block);
+}
+
+/*
* Make a new device known to the system.
*/
-static int
-dasd_state_new_to_known(struct dasd_device *device)
+static int dasd_state_new_to_known(struct dasd_device *device)
{
int rc;
@@ -133,12 +162,13 @@ dasd_state_new_to_known(struct dasd_devi
*/
dasd_get_device(device);
- rc = dasd_alloc_queue(device);
- if (rc) {
- dasd_put_device(device);
- return rc;
+ if (device->block) {
+ rc = dasd_alloc_queue(device->block);
+ if (rc) {
+ dasd_put_device(device);
+ return rc;
+ }
}
-
device->state = DASD_STATE_KNOWN;
return 0;
}
@@ -146,21 +176,24 @@ dasd_state_new_to_known(struct dasd_devi
/*
* Let the system forget about a device.
*/
-static int
-dasd_state_known_to_new(struct dasd_device * device)
+static int dasd_state_known_to_new(struct dasd_device *device)
{
/* Disable extended error reporting for this device. */
dasd_eer_disable(device);
/* Forget the discipline information. */
- if (device->discipline)
+ if (device->discipline) {
+ if (device->discipline->uncheck_device)
+ device->discipline->uncheck_device(device);
module_put(device->discipline->owner);
+ }
device->discipline = NULL;
if (device->base_discipline)
module_put(device->base_discipline->owner);
device->base_discipline = NULL;
device->state = DASD_STATE_NEW;
- dasd_free_queue(device);
+ if (device->block)
+ dasd_free_queue(device->block);
/* Give up reference we took in dasd_state_new_to_known. */
dasd_put_device(device);
@@ -170,19 +203,19 @@ dasd_state_known_to_new(struct dasd_devi
/*
* Request the irq line for the device.
*/
-static int
-dasd_state_known_to_basic(struct dasd_device * device)
+static int dasd_state_known_to_basic(struct dasd_device *device)
{
int rc;
/* Allocate and register gendisk structure. */
- rc = dasd_gendisk_alloc(device);
- if (rc)
- return rc;
-
+ if (device->block) {
+ rc = dasd_gendisk_alloc(device->block);
+ if (rc)
+ return rc;
+ }
/* register 'device' debug area, used for all DBF_DEV_XXX calls */
device->debug_area = debug_register(device->cdev->dev.bus_id, 1, 2,
- 8 * sizeof (long));
+ 8 * sizeof(long));
debug_register_view(device->debug_area, &debug_sprintf_view);
debug_set_level(device->debug_area, DBF_WARNING);
DBF_DEV_EVENT(DBF_EMERG, device, "%s", "debug area created");
@@ -194,16 +227,17 @@ dasd_state_known_to_basic(struct dasd_de
/*
* Release the irq line for the device. Terminate any running i/o.
*/
-static int
-dasd_state_basic_to_known(struct dasd_device * device)
+static int dasd_state_basic_to_known(struct dasd_device *device)
{
int rc;
-
- dasd_gendisk_free(device);
- rc = dasd_flush_ccw_queue(device, 1);
+ if (device->block) {
+ dasd_gendisk_free(device->block);
+ dasd_block_clear_timer(device->block);
+ }
+ rc = dasd_flush_device_queue(device);
if (rc)
return rc;
- dasd_clear_timer(device);
+ dasd_device_clear_timer(device);
DBF_DEV_EVENT(DBF_EMERG, device, "%p debug area deleted", device);
if (device->debug_area != NULL) {
@@ -228,26 +262,32 @@ dasd_state_basic_to_known(struct dasd_de
* In case the analysis returns an error, the device setup is stopped
* (a fake disk was already added to allow formatting).
*/
-static int
-dasd_state_basic_to_ready(struct dasd_device * device)
+static int dasd_state_basic_to_ready(struct dasd_device *device)
{
int rc;
+ struct dasd_block *block;
rc = 0;
- if (device->discipline->do_analysis != NULL)
- rc = device->discipline->do_analysis(device);
- if (rc) {
- if (rc != -EAGAIN)
- device->state = DASD_STATE_UNFMT;
- return rc;
- }
+ block = device->block;
/* make disk known with correct capacity */
- dasd_setup_queue(device);
- set_capacity(device->gdp, device->blocks << device->s2b_shift);
- device->state = DASD_STATE_READY;
- rc = dasd_scan_partitions(device);
- if (rc)
- device->state = DASD_STATE_BASIC;
+ if (block) {
+ if (block->base->discipline->do_analysis != NULL)
+ rc = block->base->discipline->do_analysis(block);
+ if (rc) {
+ if (rc != -EAGAIN)
+ device->state = DASD_STATE_UNFMT;
+ return rc;
+ }
+ dasd_setup_queue(block);
+ set_capacity(block->gdp,
+ block->blocks << block->s2b_shift);
+ device->state = DASD_STATE_READY;
+ rc = dasd_scan_partitions(block);
+ if (rc)
+ device->state = DASD_STATE_BASIC;
+ } else {
+ device->state = DASD_STATE_READY;
+ }
return rc;
}
@@ -256,28 +296,31 @@ dasd_state_basic_to_ready(struct dasd_de
* Forget format information. Check if the target level is basic
* and if it is create fake disk for formatting.
*/
-static int
-dasd_state_ready_to_basic(struct dasd_device * device)
+static int dasd_state_ready_to_basic(struct dasd_device *device)
{
int rc;
- rc = dasd_flush_ccw_queue(device, 0);
- if (rc)
- return rc;
- dasd_destroy_partitions(device);
- dasd_flush_request_queue(device);
- device->blocks = 0;
- device->bp_block = 0;
- device->s2b_shift = 0;
device->state = DASD_STATE_BASIC;
+ if (device->block) {
+ struct dasd_block *block = device->block;
+ rc = dasd_flush_block_queue(block);
+ if (rc) {
+ device->state = DASD_STATE_READY;
+ return rc;
+ }
+ dasd_destroy_partitions(block);
+ dasd_flush_request_queue(block);
+ block->blocks = 0;
+ block->bp_block = 0;
+ block->s2b_shift = 0;
+ }
return 0;
}
/*
* Back to basic.
*/
-static int
-dasd_state_unfmt_to_basic(struct dasd_device * device)
+static int dasd_state_unfmt_to_basic(struct dasd_device *device)
{
device->state = DASD_STATE_BASIC;
return 0;
@@ -291,17 +334,31 @@ dasd_state_unfmt_to_basic(struct dasd_de
static int
dasd_state_ready_to_online(struct dasd_device * device)
{
+ int rc;
+
+ if (device->discipline->ready_to_online) {
+ rc = device->discipline->ready_to_online(device);
+ if (rc)
+ return rc;
+ }
device->state = DASD_STATE_ONLINE;
- dasd_schedule_bh(device);
+ if (device->block)
+ dasd_schedule_block_bh(device->block);
return 0;
}
/*
* Stop the requeueing of requests again.
*/
-static int
-dasd_state_online_to_ready(struct dasd_device * device)
+static int dasd_state_online_to_ready(struct dasd_device *device)
{
+ int rc;
+
+ if (device->discipline->online_to_ready) {
+ rc = device->discipline->online_to_ready(device);
+ if (rc)
+ return rc;
+ }
device->state = DASD_STATE_READY;
return 0;
}
@@ -309,8 +366,7 @@ dasd_state_online_to_ready(struct dasd_d
/*
* Device startup state changes.
*/
-static int
-dasd_increase_state(struct dasd_device *device)
+static int dasd_increase_state(struct dasd_device *device)
{
int rc;
@@ -345,8 +401,7 @@ dasd_increase_state(struct dasd_device *
/*
* Device shutdown state changes.
*/
-static int
-dasd_decrease_state(struct dasd_device *device)
+static int dasd_decrease_state(struct dasd_device *device)
{
int rc;
@@ -381,8 +436,7 @@ dasd_decrease_state(struct dasd_device *
/*
* This is the main startup/shutdown routine.
*/
-static void
-dasd_change_state(struct dasd_device *device)
+static void dasd_change_state(struct dasd_device *device)
{
int rc;
@@ -409,17 +463,15 @@ dasd_change_state(struct dasd_device *de
* dasd_kick_device will schedule a call do do_kick_device to the kernel
* event daemon.
*/
-static void
-do_kick_device(struct work_struct *work)
+static void do_kick_device(struct work_struct *work)
{
struct dasd_device *device = container_of(work, struct dasd_device, kick_work);
dasd_change_state(device);
- dasd_schedule_bh(device);
+ dasd_schedule_device_bh(device);
dasd_put_device(device);
}
-void
-dasd_kick_device(struct dasd_device *device)
+void dasd_kick_device(struct dasd_device *device)
{
dasd_get_device(device);
/* queue call to dasd_kick_device to the kernel event daemon. */
@@ -429,8 +481,7 @@ dasd_kick_device(struct dasd_device *dev
/*
* Set the target state for a device and starts the state change.
*/
-void
-dasd_set_target_state(struct dasd_device *device, int target)
+void dasd_set_target_state(struct dasd_device *device, int target)
{
/* If we are in probeonly mode stop at DASD_STATE_READY. */
if (dasd_probeonly && target > DASD_STATE_READY)
@@ -447,14 +498,12 @@ dasd_set_target_state(struct dasd_device
/*
* Enable devices with device numbers in [from..to].
*/
-static inline int
-_wait_for_device(struct dasd_device *device)
+static inline int _wait_for_device(struct dasd_device *device)
{
return (device->state == device->target);
}
-void
-dasd_enable_device(struct dasd_device *device)
+void dasd_enable_device(struct dasd_device *device)
{
dasd_set_target_state(device, DASD_STATE_ONLINE);
if (device->state <= DASD_STATE_KNOWN)
@@ -475,20 +524,20 @@ unsigned int dasd_profile_level = DASD_P
/*
* Increments counter in global and local profiling structures.
*/
-#define dasd_profile_counter(value, counter, device) \
+#define dasd_profile_counter(value, counter, block) \
{ \
int index; \
for (index = 0; index < 31 && value >> (2+index); index++); \
dasd_global_profile.counter[index]++; \
- device->profile.counter[index]++; \
+ block->profile.counter[index]++; \
}
/*
* Add profiling information for cqr before execution.
*/
-static void
-dasd_profile_start(struct dasd_device *device, struct dasd_ccw_req * cqr,
- struct request *req)
+static void dasd_profile_start(struct dasd_block *block,
+ struct dasd_ccw_req *cqr,
+ struct request *req)
{
struct list_head *l;
unsigned int counter;
@@ -498,19 +547,19 @@ dasd_profile_start(struct dasd_device *d
/* count the length of the chanq for statistics */
counter = 0;
- list_for_each(l, &device->ccw_queue)
+ list_for_each(l, &block->ccw_queue)
if (++counter >= 31)
break;
dasd_global_profile.dasd_io_nr_req[counter]++;
- device->profile.dasd_io_nr_req[counter]++;
+ block->profile.dasd_io_nr_req[counter]++;
}
/*
* Add profiling information for cqr after execution.
*/
-static void
-dasd_profile_end(struct dasd_device *device, struct dasd_ccw_req * cqr,
- struct request *req)
+static void dasd_profile_end(struct dasd_block *block,
+ struct dasd_ccw_req *cqr,
+ struct request *req)
{
long strtime, irqtime, endtime, tottime; /* in microseconds */
long tottimeps, sectors;
@@ -532,27 +581,27 @@ dasd_profile_end(struct dasd_device *dev
if (!dasd_global_profile.dasd_io_reqs)
memset(&dasd_global_profile, 0,
- sizeof (struct dasd_profile_info_t));
+ sizeof(struct dasd_profile_info_t));
dasd_global_profile.dasd_io_reqs++;
dasd_global_profile.dasd_io_sects += sectors;
- if (!device->profile.dasd_io_reqs)
- memset(&device->profile, 0,
- sizeof (struct dasd_profile_info_t));
- device->profile.dasd_io_reqs++;
- device->profile.dasd_io_sects += sectors;
-
- dasd_profile_counter(sectors, dasd_io_secs, device);
- dasd_profile_counter(tottime, dasd_io_times, device);
- dasd_profile_counter(tottimeps, dasd_io_timps, device);
- dasd_profile_counter(strtime, dasd_io_time1, device);
- dasd_profile_counter(irqtime, dasd_io_time2, device);
- dasd_profile_counter(irqtime / sectors, dasd_io_time2ps, device);
- dasd_profile_counter(endtime, dasd_io_time3, device);
+ if (!block->profile.dasd_io_reqs)
+ memset(&block->profile, 0,
+ sizeof(struct dasd_profile_info_t));
+ block->profile.dasd_io_reqs++;
+ block->profile.dasd_io_sects += sectors;
+
+ dasd_profile_counter(sectors, dasd_io_secs, block);
+ dasd_profile_counter(tottime, dasd_io_times, block);
+ dasd_profile_counter(tottimeps, dasd_io_timps, block);
+ dasd_profile_counter(strtime, dasd_io_time1, block);
+ dasd_profile_counter(irqtime, dasd_io_time2, block);
+ dasd_profile_counter(irqtime / sectors, dasd_io_time2ps, block);
+ dasd_profile_counter(endtime, dasd_io_time3, block);
}
#else
-#define dasd_profile_start(device, cqr, req) do {} while (0)
-#define dasd_profile_end(device, cqr, req) do {} while (0)
+#define dasd_profile_start(block, cqr, req) do {} while (0)
+#define dasd_profile_end(block, cqr, req) do {} while (0)
#endif /* CONFIG_DASD_PROFILE */
/*
@@ -562,9 +611,9 @@ dasd_profile_end(struct dasd_device *dev
* memory and 2) dasd_smalloc_request uses the static ccw memory
* that gets allocated for each device.
*/
-struct dasd_ccw_req *
-dasd_kmalloc_request(char *magic, int cplength, int datasize,
- struct dasd_device * device)
+struct dasd_ccw_req *dasd_kmalloc_request(char *magic, int cplength,
+ int datasize,
+ struct dasd_device *device)
{
struct dasd_ccw_req *cqr;
@@ -600,9 +649,9 @@ dasd_kmalloc_request(char *magic, int cp
return cqr;
}
-struct dasd_ccw_req *
-dasd_smalloc_request(char *magic, int cplength, int datasize,
- struct dasd_device * device)
+struct dasd_ccw_req *dasd_smalloc_request(char *magic, int cplength,
+ int datasize,
+ struct dasd_device *device)
{
unsigned long flags;
struct dasd_ccw_req *cqr;
@@ -649,8 +698,7 @@ dasd_smalloc_request(char *magic, int cp
* idal lists that might have been created by dasd_set_cda and the
* struct dasd_ccw_req itself.
*/
-void
-dasd_kfree_request(struct dasd_ccw_req * cqr, struct dasd_device * device)
+void dasd_kfree_request(struct dasd_ccw_req *cqr, struct dasd_device *device)
{
#ifdef CONFIG_64BIT
struct ccw1 *ccw;
@@ -667,8 +715,7 @@ dasd_kfree_request(struct dasd_ccw_req *
dasd_put_device(device);
}
-void
-dasd_sfree_request(struct dasd_ccw_req * cqr, struct dasd_device * device)
+void dasd_sfree_request(struct dasd_ccw_req *cqr, struct dasd_device *device)
{
unsigned long flags;
@@ -681,14 +728,13 @@ dasd_sfree_request(struct dasd_ccw_req *
/*
* Check discipline magic in cqr.
*/
-static inline int
-dasd_check_cqr(struct dasd_ccw_req *cqr)
+static inline int dasd_check_cqr(struct dasd_ccw_req *cqr)
{
struct dasd_device *device;
if (cqr == NULL)
return -EINVAL;
- device = cqr->device;
+ device = cqr->startdev;
if (strncmp((char *) &cqr->magic, device->discipline->ebcname, 4)) {
DEV_MESSAGE(KERN_WARNING, device,
" dasd_ccw_req 0x%08x magic doesn't match"
@@ -706,8 +752,7 @@ dasd_check_cqr(struct dasd_ccw_req *cqr)
* ccw_device_clear can fail if the i/o subsystem
* is in a bad mood.
*/
-int
-dasd_term_IO(struct dasd_ccw_req * cqr)
+int dasd_term_IO(struct dasd_ccw_req *cqr)
{
struct dasd_device *device;
int retries, rc;
@@ -717,13 +762,13 @@ dasd_term_IO(struct dasd_ccw_req * cqr)
if (rc)
return rc;
retries = 0;
- device = (struct dasd_device *) cqr->device;
+ device = (struct dasd_device *) cqr->startdev;
while ((retries < 5) && (cqr->status == DASD_CQR_IN_IO)) {
rc = ccw_device_clear(device->cdev, (long) cqr);
switch (rc) {
case 0: /* termination successful */
cqr->retries--;
- cqr->status = DASD_CQR_CLEAR;
+ cqr->status = DASD_CQR_CLEAR_PENDING;
cqr->stopclk = get_clock();
cqr->starttime = 0;
DBF_DEV_EVENT(DBF_DEBUG, device,
@@ -753,7 +798,7 @@ dasd_term_IO(struct dasd_ccw_req * cqr)
}
retries++;
}
- dasd_schedule_bh(device);
+ dasd_schedule_device_bh(device);
return rc;
}
@@ -761,8 +806,7 @@ dasd_term_IO(struct dasd_ccw_req * cqr)
* Start the i/o. This start_IO can fail if the channel is really busy.
* In that case set up a timer to start the request later.
*/
-int
-dasd_start_IO(struct dasd_ccw_req * cqr)
+int dasd_start_IO(struct dasd_ccw_req *cqr)
{
struct dasd_device *device;
int rc;
@@ -771,12 +815,12 @@ dasd_start_IO(struct dasd_ccw_req * cqr)
rc = dasd_check_cqr(cqr);
if (rc)
return rc;
- device = (struct dasd_device *) cqr->device;
+ device = (struct dasd_device *) cqr->startdev;
if (cqr->retries < 0) {
DEV_MESSAGE(KERN_DEBUG, device,
"start_IO: request %p (%02x/%i) - no retry left.",
cqr, cqr->status, cqr->retries);
- cqr->status = DASD_CQR_FAILED;
+ cqr->status = DASD_CQR_ERROR;
return -EIO;
}
cqr->startclk = get_clock();
@@ -833,8 +877,7 @@ dasd_start_IO(struct dasd_ccw_req * cqr)
* The head of the ccw queue will have status DASD_CQR_IN_IO for 1),
* DASD_CQR_QUEUED for 2) and 3).
*/
-static void
-dasd_timeout_device(unsigned long ptr)
+static void dasd_device_timeout(unsigned long ptr)
{
unsigned long flags;
struct dasd_device *device;
@@ -844,14 +887,13 @@ dasd_timeout_device(unsigned long ptr)
/* re-activate request queue */
device->stopped &= ~DASD_STOPPED_PENDING;
spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
- dasd_schedule_bh(device);
+ dasd_schedule_device_bh(device);
}
/*
* Setup timeout for a device in jiffies.
*/
-void
-dasd_set_timer(struct dasd_device *device, int expires)
+void dasd_device_set_timer(struct dasd_device *device, int expires)
{
if (expires == 0) {
if (timer_pending(&device->timer))
@@ -862,7 +904,7 @@ dasd_set_timer(struct dasd_device *devic
if (mod_timer(&device->timer, jiffies + expires))
return;
}
- device->timer.function = dasd_timeout_device;
+ device->timer.function = dasd_device_timeout;
device->timer.data = (unsigned long) device;
device->timer.expires = jiffies + expires;
add_timer(&device->timer);
@@ -871,15 +913,14 @@ dasd_set_timer(struct dasd_device *devic
/*
* Clear timeout for a device.
*/
-void
-dasd_clear_timer(struct dasd_device *device)
+void dasd_device_clear_timer(struct dasd_device *device)
{
if (timer_pending(&device->timer))
del_timer(&device->timer);
}
-static void
-dasd_handle_killed_request(struct ccw_device *cdev, unsigned long intparm)
+static void dasd_handle_killed_request(struct ccw_device *cdev,
+ unsigned long intparm)
{
struct dasd_ccw_req *cqr;
struct dasd_device *device;
@@ -893,7 +934,7 @@ dasd_handle_killed_request(struct ccw_de
return;
}
- device = (struct dasd_device *) cqr->device;
+ device = (struct dasd_device *) cqr->startdev;
if (device == NULL ||
device != dasd_device_from_cdev_locked(cdev) ||
strncmp(device->discipline->ebcname, (char *) &cqr->magic, 4)) {
@@ -905,13 +946,12 @@ dasd_handle_killed_request(struct ccw_de
/* Schedule request to be retried. */
cqr->status = DASD_CQR_QUEUED;
- dasd_clear_timer(device);
- dasd_schedule_bh(device);
+ dasd_device_clear_timer(device);
+ dasd_schedule_device_bh(device);
dasd_put_device(device);
}
-static void
-dasd_handle_state_change_pending(struct dasd_device *device)
+void dasd_generic_handle_state_change(struct dasd_device *device)
{
struct dasd_ccw_req *cqr;
struct list_head *l, *n;
@@ -923,28 +963,27 @@ dasd_handle_state_change_pending(struct
/* restart all 'running' IO on queue */
list_for_each_safe(l, n, &device->ccw_queue) {
- cqr = list_entry(l, struct dasd_ccw_req, list);
+ cqr = list_entry(l, struct dasd_ccw_req, devlist);
if (cqr->status == DASD_CQR_IN_IO) {
cqr->status = DASD_CQR_QUEUED;
}
}
- dasd_clear_timer(device);
- dasd_schedule_bh(device);
+ dasd_device_clear_timer(device);
+ dasd_schedule_device_bh(device);
+ if (device->block)
+ dasd_schedule_block_bh(device->block);
}
/*
* Interrupt handler for "normal" ssch-io based dasd devices.
*/
-void
-dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
- struct irb *irb)
+void dasd_int_handler(struct ccw_device *cdev, unsigned long intparm,
+ struct irb *irb)
{
struct dasd_ccw_req *cqr, *next;
struct dasd_device *device;
unsigned long long now;
int expires;
- dasd_era_t era;
- char mask;
if (IS_ERR(irb)) {
switch (PTR_ERR(irb)) {
@@ -969,28 +1008,19 @@ dasd_int_handler(struct ccw_device *cdev
cdev->dev.bus_id, ((irb->scsw.cstat<<8)|irb->scsw.dstat),
(unsigned int) intparm);
- /* first of all check for state change pending interrupt */
- mask = DEV_STAT_ATTENTION | DEV_STAT_DEV_END | DEV_STAT_UNIT_EXCEP;
- if ((irb->scsw.dstat & mask) == mask) {
+ /* check for unsolicited interrupts */
+ cqr = (struct dasd_ccw_req *) intparm;
+ if (cqr == NULL) {
device = dasd_device_from_cdev_locked(cdev);
if (!IS_ERR(device)) {
- dasd_handle_state_change_pending(device);
+ device->discipline->handle_unsolicited_interrupt(device,
+ irb);
dasd_put_device(device);
}
return;
}
- cqr = (struct dasd_ccw_req *) intparm;
-
- /* check for unsolicited interrupts */
- if (cqr == NULL) {
- MESSAGE(KERN_DEBUG,
- "unsolicited interrupt received: bus_id %s",
- cdev->dev.bus_id);
- return;
- }
-
- device = (struct dasd_device *) cqr->device;
+ device = (struct dasd_device *) cqr->startdev;
if (device == NULL ||
strncmp(device->discipline->ebcname, (char *) &cqr->magic, 4)) {
MESSAGE(KERN_DEBUG, "invalid device in request: bus_id %s",
@@ -999,12 +1029,12 @@ dasd_int_handler(struct ccw_device *cdev
}
/* Check for clear pending */
- if (cqr->status == DASD_CQR_CLEAR &&
+ if (cqr->status == DASD_CQR_CLEAR_PENDING &&
irb->scsw.fctl & SCSW_FCTL_CLEAR_FUNC) {
- cqr->status = DASD_CQR_QUEUED;
- dasd_clear_timer(device);
+ cqr->status = DASD_CQR_CLEARED;
+ dasd_device_clear_timer(device);
wake_up(&dasd_flush_wq);
- dasd_schedule_bh(device);
+ dasd_schedule_device_bh(device);
return;
}
@@ -1017,272 +1047,164 @@ dasd_int_handler(struct ccw_device *cdev
}
DBF_DEV_EVENT(DBF_DEBUG, device, "Int: CS/DS 0x%04x for cqr %p",
((irb->scsw.cstat << 8) | irb->scsw.dstat), cqr);
-
- /* Find out the appropriate era_action. */
- if (irb->scsw.fctl & SCSW_FCTL_HALT_FUNC)
- era = dasd_era_fatal;
- else if (irb->scsw.dstat == (DEV_STAT_CHN_END | DEV_STAT_DEV_END) &&
- irb->scsw.cstat == 0 &&
- !irb->esw.esw0.erw.cons)
- era = dasd_era_none;
- else if (irb->esw.esw0.erw.cons)
- era = device->discipline->examine_error(cqr, irb);
- else
- era = dasd_era_recover;
-
- DBF_DEV_EVENT(DBF_DEBUG, device, "era_code %d", era);
+ next = NULL;
expires = 0;
- if (era == dasd_era_none) {
- cqr->status = DASD_CQR_DONE;
+ if (irb->scsw.dstat == (DEV_STAT_CHN_END | DEV_STAT_DEV_END) &&
+ irb->scsw.cstat == 0 && !irb->esw.esw0.erw.cons) {
+ /* request was completed successfully */
+ cqr->status = DASD_CQR_SUCCESS;
cqr->stopclk = now;
/* Start first request on queue if possible -> fast_io. */
- if (cqr->list.next != &device->ccw_queue) {
- next = list_entry(cqr->list.next,
- struct dasd_ccw_req, list);
- if ((next->status == DASD_CQR_QUEUED) &&
- (!device->stopped)) {
- if (device->discipline->start_IO(next) == 0)
- expires = next->expires;
- else
- DEV_MESSAGE(KERN_DEBUG, device, "%s",
- "Interrupt fastpath "
- "failed!");
- }
+ if (cqr->devlist.next != &device->ccw_queue) {
+ next = list_entry(cqr->devlist.next,
+ struct dasd_ccw_req, devlist);
}
- } else { /* error */
- memcpy(&cqr->irb, irb, sizeof (struct irb));
+ } else { /* error */
+ memcpy(&cqr->irb, irb, sizeof(struct irb));
if (device->features & DASD_FEATURE_ERPLOG) {
- /* dump sense data */
dasd_log_sense(cqr, irb);
}
- switch (era) {
- case dasd_era_fatal:
- cqr->status = DASD_CQR_FAILED;
- cqr->stopclk = now;
- break;
- case dasd_era_recover:
+ /* If we have no sense data, or we just don't want complex ERP
+ * for this request, but if we have retries left, then just
+ * reset this request and retry it in the fastpath
+ */
+ if (!(cqr->irb.esw.esw0.erw.cons &&
+ test_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags)) &&
+ cqr->retries > 0) {
+ DEV_MESSAGE(KERN_DEBUG, device,
+ "default ERP in fastpath (%i retries left)",
+ cqr->retries);
+ cqr->lpm = LPM_ANYPATH;
+ cqr->status = DASD_CQR_QUEUED;
+ next = cqr;
+ } else
cqr->status = DASD_CQR_ERROR;
- break;
- default:
- BUG();
- }
+ }
+ if (next && (next->status == DASD_CQR_QUEUED) &&
+ (!device->stopped)) {
+ if (device->discipline->start_IO(next) == 0)
+ expires = next->expires;
+ else
+ DEV_MESSAGE(KERN_DEBUG, device, "%s",
+ "Interrupt fastpath "
+ "failed!");
}
if (expires != 0)
- dasd_set_timer(device, expires);
+ dasd_device_set_timer(device, expires);
else
- dasd_clear_timer(device);
- dasd_schedule_bh(device);
+ dasd_device_clear_timer(device);
+ dasd_schedule_device_bh(device);
}
/*
- * posts the buffer_cache about a finalized request
+ * If we have an error on a dasd_block layer request then we cancel
+ * and return all further requests from the same dasd_block as well.
*/
-static inline void
-dasd_end_request(struct request *req, int uptodate)
+static void __dasd_device_recovery(struct dasd_device *device,
+ struct dasd_ccw_req *ref_cqr)
{
- if (end_that_request_first(req, uptodate, req->hard_nr_sectors))
- BUG();
- add_disk_randomness(req->rq_disk);
- end_that_request_last(req, uptodate);
-}
+ struct list_head *l, *n;
+ struct dasd_ccw_req *cqr;
-/*
- * Process finished error recovery ccw.
- */
-static inline void
-__dasd_process_erp(struct dasd_device *device, struct dasd_ccw_req *cqr)
-{
- dasd_erp_fn_t erp_fn;
+ /*
+ * only requeue request that came from the dasd_block layer
+ */
+ if (!ref_cqr->block)
+ return;
- if (cqr->status == DASD_CQR_DONE)
- DBF_DEV_EVENT(DBF_NOTICE, device, "%s", "ERP successful");
- else
- DEV_MESSAGE(KERN_ERR, device, "%s", "ERP unsuccessful");
- erp_fn = device->discipline->erp_postaction(cqr);
- erp_fn(cqr);
-}
+ list_for_each_safe(l, n, &device->ccw_queue) {
+ cqr = list_entry(l, struct dasd_ccw_req, devlist);
+ if (cqr->status == DASD_CQR_QUEUED &&
+ ref_cqr->block == cqr->block) {
+ cqr->status = DASD_CQR_CLEARED;
+ }
+ }
+};
/*
- * Process ccw request queue.
+ * Remove those ccw requests from the queue that need to be returned
+ * to the upper layer.
*/
-static void
-__dasd_process_ccw_queue(struct dasd_device * device,
- struct list_head *final_queue)
+static void __dasd_device_process_ccw_queue(struct dasd_device *device,
+ struct list_head *final_queue)
{
struct list_head *l, *n;
struct dasd_ccw_req *cqr;
- dasd_erp_fn_t erp_fn;
-restart:
/* Process request with final status. */
list_for_each_safe(l, n, &device->ccw_queue) {
- cqr = list_entry(l, struct dasd_ccw_req, list);
+ cqr = list_entry(l, struct dasd_ccw_req, devlist);
+
/* Stop list processing at the first non-final request. */
- if (cqr->status != DASD_CQR_DONE &&
- cqr->status != DASD_CQR_FAILED &&
- cqr->status != DASD_CQR_ERROR)
+ if (cqr->status == DASD_CQR_QUEUED ||
+ cqr->status == DASD_CQR_IN_IO ||
+ cqr->status == DASD_CQR_CLEAR_PENDING)
break;
- /* Process requests with DASD_CQR_ERROR */
if (cqr->status == DASD_CQR_ERROR) {
- if (cqr->irb.scsw.fctl & SCSW_FCTL_HALT_FUNC) {
- cqr->status = DASD_CQR_FAILED;
- cqr->stopclk = get_clock();
- } else {
- if (cqr->irb.esw.esw0.erw.cons &&
- test_bit(DASD_CQR_FLAGS_USE_ERP,
- &cqr->flags)) {
- erp_fn = device->discipline->
- erp_action(cqr);
- erp_fn(cqr);
- } else
- dasd_default_erp_action(cqr);
- }
- goto restart;
- }
-
- /* First of all call extended error reporting. */
- if (dasd_eer_enabled(device) &&
- cqr->status == DASD_CQR_FAILED) {
- dasd_eer_write(device, cqr, DASD_EER_FATALERROR);
-
- /* restart request */
- cqr->status = DASD_CQR_QUEUED;
- cqr->retries = 255;
- device->stopped |= DASD_STOPPED_QUIESCE;
- goto restart;
- }
-
- /* Process finished ERP request. */
- if (cqr->refers) {
- __dasd_process_erp(device, cqr);
- goto restart;
+ __dasd_device_recovery(device, cqr);
}
-
/* Rechain finished requests to final queue */
- cqr->endclk = get_clock();
- list_move_tail(&cqr->list, final_queue);
+ list_move_tail(&cqr->devlist, final_queue);
}
}
-static void
-dasd_end_request_cb(struct dasd_ccw_req * cqr, void *data)
-{
- struct request *req;
- struct dasd_device *device;
- int status;
-
- req = (struct request *) data;
- device = cqr->device;
- dasd_profile_end(device, cqr, req);
- status = cqr->device->discipline->free_cp(cqr,req);
- spin_lock_irq(&device->request_queue_lock);
- dasd_end_request(req, status);
- spin_unlock_irq(&device->request_queue_lock);
-}
-
-
/*
- * Fetch requests from the block device queue.
+ * the cqrs from the final queue are returned to the upper layer
+ * by setting a dasd_block state and calling the callback function
*/
-static void
-__dasd_process_blk_queue(struct dasd_device * device)
+static void __dasd_device_process_final_queue(struct dasd_device *device,
+ struct list_head *final_queue)
{
- struct request_queue *queue;
- struct request *req;
+ struct list_head *l, *n;
struct dasd_ccw_req *cqr;
- int nr_queued;
-
- queue = device->request_queue;
- /* No queue ? Then there is nothing to do. */
- if (queue == NULL)
- return;
-
- /*
- * We requeue request from the block device queue to the ccw
- * queue only in two states. In state DASD_STATE_READY the
- * partition detection is done and we need to requeue requests
- * for that. State DASD_STATE_ONLINE is normal block device
- * operation.
- */
- if (device->state != DASD_STATE_READY &&
- device->state != DASD_STATE_ONLINE)
- return;
- nr_queued = 0;
- /* Now we try to fetch requests from the request queue */
- list_for_each_entry(cqr, &device->ccw_queue, list)
- if (cqr->status == DASD_CQR_QUEUED)
- nr_queued++;
- while (!blk_queue_plugged(queue) &&
- elv_next_request(queue) &&
- nr_queued < DASD_CHANQ_MAX_SIZE) {
- req = elv_next_request(queue);
- if (device->features & DASD_FEATURE_READONLY &&
- rq_data_dir(req) == WRITE) {
- DBF_DEV_EVENT(DBF_ERR, device,
- "Rejecting write request %p",
- req);
- blkdev_dequeue_request(req);
- dasd_end_request(req, 0);
- continue;
- }
- if (device->stopped & DASD_STOPPED_DC_EIO) {
- blkdev_dequeue_request(req);
- dasd_end_request(req, 0);
- continue;
- }
- cqr = device->discipline->build_cp(device, req);
- if (IS_ERR(cqr)) {
- if (PTR_ERR(cqr) == -ENOMEM)
- break; /* terminate request queue loop */
- if (PTR_ERR(cqr) == -EAGAIN) {
- /*
- * The current request cannot be build right
- * now, we have to try later. If this request
- * is the head-of-queue we stop the device
- * for 1/2 second.
- */
- if (!list_empty(&device->ccw_queue))
- break;
- device->stopped |= DASD_STOPPED_PENDING;
- dasd_set_timer(device, HZ/2);
- break;
- }
- DBF_DEV_EVENT(DBF_ERR, device,
- "CCW creation failed (rc=%ld) "
- "on request %p",
- PTR_ERR(cqr), req);
- blkdev_dequeue_request(req);
- dasd_end_request(req, 0);
- continue;
+ list_for_each_safe(l, n, final_queue) {
+ cqr = list_entry(l, struct dasd_ccw_req, devlist);
+ list_del_init(&cqr->devlist);
+ if (cqr->block)
+ spin_lock_bh(&cqr->block->queue_lock);
+ switch (cqr->status) {
+ case DASD_CQR_SUCCESS:
+ cqr->status = DASD_CQR_DONE;
+ break;
+ case DASD_CQR_ERROR:
+ cqr->status = DASD_CQR_NEED_ERP;
+ break;
+ case DASD_CQR_CLEARED:
+ cqr->status = DASD_CQR_TERMINATED;
+ break;
+ default:
+ DEV_MESSAGE(KERN_ERR, device,
+ "wrong cqr status in __dasd_process_final_queue "
+ "for cqr %p, status %x",
+ cqr, cqr->status);
+ BUG();
}
- cqr->callback = dasd_end_request_cb;
- cqr->callback_data = (void *) req;
- cqr->status = DASD_CQR_QUEUED;
- blkdev_dequeue_request(req);
- list_add_tail(&cqr->list, &device->ccw_queue);
- dasd_profile_start(device, cqr, req);
- nr_queued++;
+ if (cqr->block)
+ spin_unlock_bh(&cqr->block->queue_lock);
+ if (cqr->callback != NULL)
+ (cqr->callback)(cqr, cqr->callback_data);
}
}
+
+
/*
* Take a look at the first request on the ccw queue and check
* if it reached its expire time. If so, terminate the IO.
*/
-static void
-__dasd_check_expire(struct dasd_device * device)
+static void __dasd_device_check_expire(struct dasd_device *device)
{
struct dasd_ccw_req *cqr;
if (list_empty(&device->ccw_queue))
return;
- cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, list);
+ cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, devlist);
if ((cqr->status == DASD_CQR_IN_IO && cqr->expires != 0) &&
(time_after_eq(jiffies, cqr->expires + cqr->starttime))) {
if (device->discipline->term_IO(cqr) != 0) {
/* Hmpf, try again in 5 sec */
- dasd_set_timer(device, 5*HZ);
+ dasd_device_set_timer(device, 5*HZ);
DEV_MESSAGE(KERN_ERR, device,
"internal error - timeout (%is) expired "
"for cqr %p, termination failed, "
@@ -1301,77 +1223,53 @@ __dasd_check_expire(struct dasd_device *
* Take a look at the first request on the ccw queue and check
* if it needs to be started.
*/
-static void
-__dasd_start_head(struct dasd_device * device)
+static void __dasd_device_start_head(struct dasd_device *device)
{
struct dasd_ccw_req *cqr;
int rc;
if (list_empty(&device->ccw_queue))
return;
- cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, list);
+ cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, devlist);
if (cqr->status != DASD_CQR_QUEUED)
return;
- /* Non-temporary stop condition will trigger fail fast */
- if (device->stopped & ~DASD_STOPPED_PENDING &&
- test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) &&
- (!dasd_eer_enabled(device))) {
- cqr->status = DASD_CQR_FAILED;
- dasd_schedule_bh(device);
+ /* when device is stopped, return request to previous layer */
+ if (device->stopped) {
+ cqr->status = DASD_CQR_CLEARED;
+ dasd_schedule_device_bh(device);
return;
}
- /* Don't try to start requests if device is stopped */
- if (device->stopped)
- return;
rc = device->discipline->start_IO(cqr);
if (rc == 0)
- dasd_set_timer(device, cqr->expires);
+ dasd_device_set_timer(device, cqr->expires);
else if (rc == -EACCES) {
- dasd_schedule_bh(device);
+ dasd_schedule_device_bh(device);
} else
/* Hmpf, try again in 1/2 sec */
- dasd_set_timer(device, 50);
-}
-
-static inline int
-_wait_for_clear(struct dasd_ccw_req *cqr)
-{
- return (cqr->status == DASD_CQR_QUEUED);
+ dasd_device_set_timer(device, 50);
}
/*
- * Remove all requests from the ccw queue (all = '1') or only block device
- * requests in case all = '0'.
- * Take care of the erp-chain (chained via cqr->refers) and remove either
- * the whole erp-chain or none of the erp-requests.
- * If a request is currently running, term_IO is called and the request
- * is re-queued. Prior to removing the terminated request we need to wait
- * for the clear-interrupt.
- * In case termination is not possible we stop processing and just finishing
- * the already moved requests.
+ * Go through all request on the dasd_device request queue,
+ * terminate them on the cdev if necessary, and return them to the
+ * submitting layer via callback.
+ * Note:
+ * Make sure that all 'submitting layers' still exist when
+ * this function is called!. In other words, when 'device' is a base
+ * device then all block layer requests must have been removed before
+ * via dasd_flush_block_queue.
*/
-static int
-dasd_flush_ccw_queue(struct dasd_device * device, int all)
+int dasd_flush_device_queue(struct dasd_device *device)
{
- struct dasd_ccw_req *cqr, *orig, *n;
- int rc, i;
-
+ struct dasd_ccw_req *cqr, *n;
+ int rc;
struct list_head flush_queue;
INIT_LIST_HEAD(&flush_queue);
spin_lock_irq(get_ccwdev_lock(device->cdev));
rc = 0;
-restart:
- list_for_each_entry_safe(cqr, n, &device->ccw_queue, list) {
- /* get original request of erp request-chain */
- for (orig = cqr; orig->refers != NULL; orig = orig->refers);
-
- /* Flush all request or only block device requests? */
- if (all == 0 && cqr->callback != dasd_end_request_cb &&
- orig->callback != dasd_end_request_cb) {
- continue;
- }
+ list_for_each_entry_safe(cqr, n, &device->ccw_queue, devlist) {
/* Check status and move request to flush_queue */
switch (cqr->status) {
case DASD_CQR_IN_IO:
@@ -1387,90 +1285,60 @@ restart:
}
break;
case DASD_CQR_QUEUED:
- case DASD_CQR_ERROR:
- /* set request to FAILED */
cqr->stopclk = get_clock();
- cqr->status = DASD_CQR_FAILED;
+ cqr->status = DASD_CQR_CLEARED;
break;
- default: /* do not touch the others */
+ default: /* no need to modify the others */
break;
}
- /* Rechain request (including erp chain) */
- for (i = 0; cqr != NULL; cqr = cqr->refers, i++) {
- cqr->endclk = get_clock();
- list_move_tail(&cqr->list, &flush_queue);
- }
- if (i > 1)
- /* moved more than one request - need to restart */
- goto restart;
+ list_move_tail(&cqr->devlist, &flush_queue);
}
-
finished:
spin_unlock_irq(get_ccwdev_lock(device->cdev));
- /* Now call the callback function of flushed requests */
-restart_cb:
- list_for_each_entry_safe(cqr, n, &flush_queue, list) {
- if (cqr->status == DASD_CQR_CLEAR) {
- /* wait for clear interrupt! */
- wait_event(dasd_flush_wq, _wait_for_clear(cqr));
- cqr->status = DASD_CQR_FAILED;
- }
- /* Process finished ERP request. */
- if (cqr->refers) {
- __dasd_process_erp(device, cqr);
- /* restart list_for_xx loop since dasd_process_erp
- * might remove multiple elements */
- goto restart_cb;
- }
- /* call the callback function */
- cqr->endclk = get_clock();
- if (cqr->callback != NULL)
- (cqr->callback)(cqr, cqr->callback_data);
- }
+ /*
+ * After this point all requests must be in state CLEAR_PENDING,
+ * CLEARED, SUCCESS or ERROR. Now wait for CLEAR_PENDING to become
+ * one of the others.
+ */
+ list_for_each_entry_safe(cqr, n, &flush_queue, devlist)
+ wait_event(dasd_flush_wq,
+ (cqr->status != DASD_CQR_CLEAR_PENDING));
+ /*
+ * Now set each request back to TERMINATED, DONE or NEED_ERP
+ * and call the callback function of flushed requests
+ */
+ __dasd_device_process_final_queue(device, &flush_queue);
return rc;
}
/*
* Acquire the device lock and process queues for the device.
*/
-static void
-dasd_tasklet(struct dasd_device * device)
+static void dasd_device_tasklet(struct dasd_device *device)
{
struct list_head final_queue;
- struct list_head *l, *n;
- struct dasd_ccw_req *cqr;
atomic_set (&device->tasklet_scheduled, 0);
INIT_LIST_HEAD(&final_queue);
spin_lock_irq(get_ccwdev_lock(device->cdev));
/* Check expire time of first request on the ccw queue. */
- __dasd_check_expire(device);
- /* Finish off requests on ccw queue */
- __dasd_process_ccw_queue(device, &final_queue);
+ __dasd_device_check_expire(device);
+ /* find final requests on ccw queue */
+ __dasd_device_process_ccw_queue(device, &final_queue);
spin_unlock_irq(get_ccwdev_lock(device->cdev));
/* Now call the callback function of requests with final status */
- list_for_each_safe(l, n, &final_queue) {
- cqr = list_entry(l, struct dasd_ccw_req, list);
- list_del_init(&cqr->list);
- if (cqr->callback != NULL)
- (cqr->callback)(cqr, cqr->callback_data);
- }
- spin_lock_irq(&device->request_queue_lock);
- spin_lock(get_ccwdev_lock(device->cdev));
- /* Get new request from the block device request queue */
- __dasd_process_blk_queue(device);
+ __dasd_device_process_final_queue(device, &final_queue);
+ spin_lock_irq(get_ccwdev_lock(device->cdev));
/* Now check if the head of the ccw queue needs to be started. */
- __dasd_start_head(device);
- spin_unlock(get_ccwdev_lock(device->cdev));
- spin_unlock_irq(&device->request_queue_lock);
+ __dasd_device_start_head(device);
+ spin_unlock_irq(get_ccwdev_lock(device->cdev));
dasd_put_device(device);
}
/*
* Schedules a call to dasd_tasklet over the device tasklet.
*/
-void
-dasd_schedule_bh(struct dasd_device * device)
+void dasd_schedule_device_bh(struct dasd_device *device)
{
/* Protect against rescheduling. */
if (atomic_cmpxchg (&device->tasklet_scheduled, 0, 1) != 0)
@@ -1480,160 +1348,109 @@ dasd_schedule_bh(struct dasd_device * de
}
/*
- * Queue a request to the head of the ccw_queue. Start the I/O if
- * possible.
+ * Queue a request to the head of the device ccw_queue.
+ * Start the I/O if possible.
*/
-void
-dasd_add_request_head(struct dasd_ccw_req *req)
+void dasd_add_request_head(struct dasd_ccw_req *cqr)
{
struct dasd_device *device;
unsigned long flags;
- device = req->device;
+ device = cqr->startdev;
spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
- req->status = DASD_CQR_QUEUED;
- req->device = device;
- list_add(&req->list, &device->ccw_queue);
+ cqr->status = DASD_CQR_QUEUED;
+ list_add(&cqr->devlist, &device->ccw_queue);
/* let the bh start the request to keep them in order */
- dasd_schedule_bh(device);
+ dasd_schedule_device_bh(device);
spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
}
/*
- * Queue a request to the tail of the ccw_queue. Start the I/O if
- * possible.
+ * Queue a request to the tail of the device ccw_queue.
+ * Start the I/O if possible.
*/
-void
-dasd_add_request_tail(struct dasd_ccw_req *req)
+void dasd_add_request_tail(struct dasd_ccw_req *cqr)
{
struct dasd_device *device;
unsigned long flags;
- device = req->device;
+ device = cqr->startdev;
spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
- req->status = DASD_CQR_QUEUED;
- req->device = device;
- list_add_tail(&req->list, &device->ccw_queue);
+ cqr->status = DASD_CQR_QUEUED;
+ list_add_tail(&cqr->devlist, &device->ccw_queue);
/* let the bh start the request to keep them in order */
- dasd_schedule_bh(device);
+ dasd_schedule_device_bh(device);
spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
}
/*
- * Wakeup callback.
+ * Wakeup helper for the 'sleep_on' functions.
*/
-static void
-dasd_wakeup_cb(struct dasd_ccw_req *cqr, void *data)
+static void dasd_wakeup_cb(struct dasd_ccw_req *cqr, void *data)
{
wake_up((wait_queue_head_t *) data);
}
-static inline int
-_wait_for_wakeup(struct dasd_ccw_req *cqr)
+static inline int _wait_for_wakeup(struct dasd_ccw_req *cqr)
{
struct dasd_device *device;
int rc;
- device = cqr->device;
+ device = cqr->startdev;
spin_lock_irq(get_ccwdev_lock(device->cdev));
rc = ((cqr->status == DASD_CQR_DONE ||
- cqr->status == DASD_CQR_FAILED) &&
- list_empty(&cqr->list));
+ cqr->status == DASD_CQR_NEED_ERP ||
+ cqr->status == DASD_CQR_TERMINATED) &&
+ list_empty(&cqr->devlist));
spin_unlock_irq(get_ccwdev_lock(device->cdev));
return rc;
}
/*
- * Attempts to start a special ccw queue and waits for its completion.
+ * Queue a request to the tail of the device ccw_queue and wait for
+ * it's completion.
*/
-int
-dasd_sleep_on(struct dasd_ccw_req * cqr)
+int dasd_sleep_on(struct dasd_ccw_req *cqr)
{
wait_queue_head_t wait_q;
struct dasd_device *device;
int rc;
- device = cqr->device;
- spin_lock_irq(get_ccwdev_lock(device->cdev));
+ device = cqr->startdev;
init_waitqueue_head (&wait_q);
cqr->callback = dasd_wakeup_cb;
cqr->callback_data = (void *) &wait_q;
- cqr->status = DASD_CQR_QUEUED;
- list_add_tail(&cqr->list, &device->ccw_queue);
-
- /* let the bh start the request to keep them in order */
- dasd_schedule_bh(device);
-
- spin_unlock_irq(get_ccwdev_lock(device->cdev));
-
+ dasd_add_request_tail(cqr);
wait_event(wait_q, _wait_for_wakeup(cqr));
/* Request status is either done or failed. */
- rc = (cqr->status == DASD_CQR_FAILED) ? -EIO : 0;
+ rc = (cqr->status == DASD_CQR_DONE) ? 0 : -EIO;
return rc;
}
/*
- * Attempts to start a special ccw queue and wait interruptible
- * for its completion.
+ * Queue a request to the tail of the device ccw_queue and wait
+ * interruptible for it's completion.
*/
-int
-dasd_sleep_on_interruptible(struct dasd_ccw_req * cqr)
+int dasd_sleep_on_interruptible(struct dasd_ccw_req *cqr)
{
wait_queue_head_t wait_q;
struct dasd_device *device;
- int rc, finished;
-
- device = cqr->device;
- spin_lock_irq(get_ccwdev_lock(device->cdev));
+ int rc;
+ device = cqr->startdev;
init_waitqueue_head (&wait_q);
cqr->callback = dasd_wakeup_cb;
cqr->callback_data = (void *) &wait_q;
- cqr->status = DASD_CQR_QUEUED;
- list_add_tail(&cqr->list, &device->ccw_queue);
-
- /* let the bh start the request to keep them in order */
- dasd_schedule_bh(device);
- spin_unlock_irq(get_ccwdev_lock(device->cdev));
-
- finished = 0;
- while (!finished) {
- rc = wait_event_interruptible(wait_q, _wait_for_wakeup(cqr));
- if (rc != -ERESTARTSYS) {
- /* Request is final (done or failed) */
- rc = (cqr->status == DASD_CQR_DONE) ? 0 : -EIO;
- break;
- }
- spin_lock_irq(get_ccwdev_lock(device->cdev));
- switch (cqr->status) {
- case DASD_CQR_IN_IO:
- /* terminate runnig cqr */
- if (device->discipline->term_IO) {
- cqr->retries = -1;
- device->discipline->term_IO(cqr);
- /* wait (non-interruptible) for final status
- * because signal ist still pending */
- spin_unlock_irq(get_ccwdev_lock(device->cdev));
- wait_event(wait_q, _wait_for_wakeup(cqr));
- spin_lock_irq(get_ccwdev_lock(device->cdev));
- rc = (cqr->status == DASD_CQR_DONE) ? 0 : -EIO;
- finished = 1;
- }
- break;
- case DASD_CQR_QUEUED:
- /* request */
- list_del_init(&cqr->list);
- rc = -EIO;
- finished = 1;
- break;
- default:
- /* cqr with 'non-interruptable' status - just wait */
- break;
- }
- spin_unlock_irq(get_ccwdev_lock(device->cdev));
+ dasd_add_request_tail(cqr);
+ rc = wait_event_interruptible(wait_q, _wait_for_wakeup(cqr));
+ if (rc == -ERESTARTSYS) {
+ dasd_cancel_req(cqr);
+ /* wait (non-interruptible) for final status */
+ wait_event(wait_q, _wait_for_wakeup(cqr));
}
+ rc = (cqr->status == DASD_CQR_DONE) ? 0 : -EIO;
return rc;
}
@@ -1643,25 +1460,23 @@ dasd_sleep_on_interruptible(struct dasd_
* and be put back to status queued, before the special request is added
* to the head of the queue. Then the special request is waited on normally.
*/
-static inline int
-_dasd_term_running_cqr(struct dasd_device *device)
+static inline int _dasd_term_running_cqr(struct dasd_device *device)
{
struct dasd_ccw_req *cqr;
if (list_empty(&device->ccw_queue))
return 0;
- cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, list);
+ cqr = list_entry(device->ccw_queue.next, struct dasd_ccw_req, devlist);
return device->discipline->term_IO(cqr);
}
-int
-dasd_sleep_on_immediatly(struct dasd_ccw_req * cqr)
+int dasd_sleep_on_immediatly(struct dasd_ccw_req *cqr)
{
wait_queue_head_t wait_q;
struct dasd_device *device;
int rc;
- device = cqr->device;
+ device = cqr->startdev;
spin_lock_irq(get_ccwdev_lock(device->cdev));
rc = _dasd_term_running_cqr(device);
if (rc) {
@@ -1673,17 +1488,17 @@ dasd_sleep_on_immediatly(struct dasd_ccw
cqr->callback = dasd_wakeup_cb;
cqr->callback_data = (void *) &wait_q;
cqr->status = DASD_CQR_QUEUED;
- list_add(&cqr->list, &device->ccw_queue);
+ list_add(&cqr->devlist, &device->ccw_queue);
/* let the bh start the request to keep them in order */
- dasd_schedule_bh(device);
+ dasd_schedule_device_bh(device);
spin_unlock_irq(get_ccwdev_lock(device->cdev));
wait_event(wait_q, _wait_for_wakeup(cqr));
/* Request status is either done or failed. */
- rc = (cqr->status == DASD_CQR_FAILED) ? -EIO : 0;
+ rc = (cqr->status == DASD_CQR_DONE) ? 0 : -EIO;
return rc;
}
@@ -1692,11 +1507,14 @@ dasd_sleep_on_immediatly(struct dasd_ccw
* This is useful to timeout requests. The request will be
* terminated if it is currently in i/o.
* Returns 1 if the request has been terminated.
+ * 0 if there was no need to terminate the request (not started yet)
+ * negative error code if termination failed
+ * Cancellation of a request is an asynchronous operation! The calling
+ * function has to wait until the request is properly returned via callback.
*/
-int
-dasd_cancel_req(struct dasd_ccw_req *cqr)
+int dasd_cancel_req(struct dasd_ccw_req *cqr)
{
- struct dasd_device *device = cqr->device;
+ struct dasd_device *device = cqr->startdev;
unsigned long flags;
int rc;
@@ -1704,74 +1522,453 @@ dasd_cancel_req(struct dasd_ccw_req *cqr
spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
switch (cqr->status) {
case DASD_CQR_QUEUED:
- /* request was not started - just set to failed */
- cqr->status = DASD_CQR_FAILED;
+ /* request was not started - just set to cleared */
+ cqr->status = DASD_CQR_CLEARED;
break;
case DASD_CQR_IN_IO:
/* request in IO - terminate IO and release again */
- if (device->discipline->term_IO(cqr) != 0)
- /* what to do if unable to terminate ??????
- e.g. not _IN_IO */
- cqr->status = DASD_CQR_FAILED;
- cqr->stopclk = get_clock();
- rc = 1;
+ rc = device->discipline->term_IO(cqr);
+ if (rc) {
+ DEV_MESSAGE(KERN_ERR, device,
+ "dasd_cancel_req is unable "
+ " to terminate request %p, rc = %d",
+ cqr, rc);
+ } else {
+ cqr->stopclk = get_clock();
+ rc = 1;
+ }
break;
- case DASD_CQR_DONE:
- case DASD_CQR_FAILED:
- /* already finished - do nothing */
+ default: /* already finished or clear pending - do nothing */
break;
- default:
- DEV_MESSAGE(KERN_ALERT, device,
- "invalid status %02x in request",
- cqr->status);
+ }
+ spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
+ dasd_schedule_device_bh(device);
+ return rc;
+}
+
+
+/*
+ * SECTION: Operations of the dasd_block layer.
+ */
+
+/*
+ * Timeout function for dasd_block. This is used when the block layer
+ * is waiting for something that may not come reliably, (e.g. a state
+ * change interrupt)
+ */
+static void dasd_block_timeout(unsigned long ptr)
+{
+ unsigned long flags;
+ struct dasd_block *block;
+
+ block = (struct dasd_block *) ptr;
+ spin_lock_irqsave(get_ccwdev_lock(block->base->cdev), flags);
+ /* re-activate request queue */
+ block->base->stopped &= ~DASD_STOPPED_PENDING;
+ spin_unlock_irqrestore(get_ccwdev_lock(block->base->cdev), flags);
+ dasd_schedule_block_bh(block);
+}
+
+/*
+ * Setup timeout for a dasd_block in jiffies.
+ */
+void dasd_block_set_timer(struct dasd_block *block, int expires)
+{
+ if (expires == 0) {
+ if (timer_pending(&block->timer))
+ del_timer(&block->timer);
+ return;
+ }
+ if (timer_pending(&block->timer)) {
+ if (mod_timer(&block->timer, jiffies + expires))
+ return;
+ }
+ block->timer.function = dasd_block_timeout;
+ block->timer.data = (unsigned long) block;
+ block->timer.expires = jiffies + expires;
+ add_timer(&block->timer);
+}
+
+/*
+ * Clear timeout for a dasd_block.
+ */
+void dasd_block_clear_timer(struct dasd_block *block)
+{
+ if (timer_pending(&block->timer))
+ del_timer(&block->timer);
+}
+
+/*
+ * posts the buffer_cache about a finalized request
+ */
+static inline void dasd_end_request(struct request *req, int uptodate)
+{
+ if (end_that_request_first(req, uptodate, req->hard_nr_sectors))
BUG();
+ add_disk_randomness(req->rq_disk);
+ end_that_request_last(req, uptodate);
+}
+
+/*
+ * Process finished error recovery ccw.
+ */
+static inline void __dasd_block_process_erp(struct dasd_block *block,
+ struct dasd_ccw_req *cqr)
+{
+ dasd_erp_fn_t erp_fn;
+ struct dasd_device *device = block->base;
+
+ if (cqr->status == DASD_CQR_DONE)
+ DBF_DEV_EVENT(DBF_NOTICE, device, "%s", "ERP successful");
+ else
+ DEV_MESSAGE(KERN_ERR, device, "%s", "ERP unsuccessful");
+ erp_fn = device->discipline->erp_postaction(cqr);
+ erp_fn(cqr);
+}
+
+/*
+ * Fetch requests from the block device queue.
+ */
+static void __dasd_process_request_queue(struct dasd_block *block)
+{
+ struct request_queue *queue;
+ struct request *req;
+ struct dasd_ccw_req *cqr;
+ struct dasd_device *basedev;
+ unsigned long flags;
+ queue = block->request_queue;
+ basedev = block->base;
+ /* No queue ? Then there is nothing to do. */
+ if (queue == NULL)
+ return;
+
+ /*
+ * We requeue request from the block device queue to the ccw
+ * queue only in two states. In state DASD_STATE_READY the
+ * partition detection is done and we need to requeue requests
+ * for that. State DASD_STATE_ONLINE is normal block device
+ * operation.
+ */
+ if (basedev->state < DASD_STATE_READY)
+ return;
+ /* Now we try to fetch requests from the request queue */
+ while (!blk_queue_plugged(queue) &&
+ elv_next_request(queue)) {
+
+ req = elv_next_request(queue);
+
+ if (basedev->features & DASD_FEATURE_READONLY &&
+ rq_data_dir(req) == WRITE) {
+ DBF_DEV_EVENT(DBF_ERR, basedev,
+ "Rejecting write request %p",
+ req);
+ blkdev_dequeue_request(req);
+ dasd_end_request(req, 0);
+ continue;
+ }
+ cqr = basedev->discipline->build_cp(basedev, block, req);
+ if (IS_ERR(cqr)) {
+ if (PTR_ERR(cqr) == -EBUSY)
+ break; /* normal end condition */
+ if (PTR_ERR(cqr) == -ENOMEM)
+ break; /* terminate request queue loop */
+ if (PTR_ERR(cqr) == -EAGAIN) {
+ /*
+ * The current request cannot be build right
+ * now, we have to try later. If this request
+ * is the head-of-queue we stop the device
+ * for 1/2 second.
+ */
+ if (!list_empty(&block->ccw_queue))
+ break;
+ spin_lock_irqsave(get_ccwdev_lock(basedev->cdev), flags);
+ basedev->stopped |= DASD_STOPPED_PENDING;
+ spin_unlock_irqrestore(get_ccwdev_lock(basedev->cdev), flags);
+ dasd_block_set_timer(block, HZ/2);
+ break;
+ }
+ DBF_DEV_EVENT(DBF_ERR, basedev,
+ "CCW creation failed (rc=%ld) "
+ "on request %p",
+ PTR_ERR(cqr), req);
+ blkdev_dequeue_request(req);
+ dasd_end_request(req, 0);
+ continue;
+ }
+ /*
+ * Note: callback is set to dasd_return_cqr_cb in
+ * __dasd_block_start_head to cover erp requests as well
+ */
+ cqr->callback_data = (void *) req;
+ cqr->status = DASD_CQR_FILLED;
+ blkdev_dequeue_request(req);
+ list_add_tail(&cqr->blocklist, &block->ccw_queue);
+ dasd_profile_start(block, cqr, req);
+ }
+}
+
+static void __dasd_cleanup_cqr(struct dasd_ccw_req *cqr)
+{
+ struct request *req;
+ int status;
+
+ req = (struct request *) cqr->callback_data;
+ dasd_profile_end(cqr->block, cqr, req);
+ status = cqr->memdev->discipline->free_cp(cqr, req);
+ dasd_end_request(req, status);
+}
+
+/*
+ * Process ccw request queue.
+ */
+static void __dasd_process_block_ccw_queue(struct dasd_block *block,
+ struct list_head *final_queue)
+{
+ struct list_head *l, *n;
+ struct dasd_ccw_req *cqr;
+ dasd_erp_fn_t erp_fn;
+ unsigned long flags;
+ struct dasd_device *base = block->base;
+
+restart:
+ /* Process request with final status. */
+ list_for_each_safe(l, n, &block->ccw_queue) {
+ cqr = list_entry(l, struct dasd_ccw_req, blocklist);
+ if (cqr->status != DASD_CQR_DONE &&
+ cqr->status != DASD_CQR_FAILED &&
+ cqr->status != DASD_CQR_NEED_ERP &&
+ cqr->status != DASD_CQR_TERMINATED)
+ continue;
+
+ if (cqr->status == DASD_CQR_TERMINATED) {
+ base->discipline->handle_terminated_request(cqr);
+ goto restart;
+ }
+
+ /* Process requests that may be recovered */
+ if (cqr->status == DASD_CQR_NEED_ERP) {
+ if (cqr->irb.esw.esw0.erw.cons &&
+ test_bit(DASD_CQR_FLAGS_USE_ERP,
+ &cqr->flags)) {
+ erp_fn = base->discipline->erp_action(cqr);
+ erp_fn(cqr);
+ }
+ goto restart;
+ }
+
+ /* First of all call extended error reporting. */
+ if (dasd_eer_enabled(base) &&
+ cqr->status == DASD_CQR_FAILED) {
+ dasd_eer_write(base, cqr, DASD_EER_FATALERROR);
+
+ /* restart request */
+ cqr->status = DASD_CQR_FILLED;
+ cqr->retries = 255;
+ spin_lock_irqsave(get_ccwdev_lock(base->cdev), flags);
+ base->stopped |= DASD_STOPPED_QUIESCE;
+ spin_unlock_irqrestore(get_ccwdev_lock(base->cdev),
+ flags);
+ goto restart;
+ }
+ /* Process finished ERP request. */
+ if (cqr->refers) {
+ __dasd_block_process_erp(block, cqr);
+ goto restart;
+ }
+
+ /* Rechain finished requests to final queue */
+ cqr->endclk = get_clock();
+ list_move_tail(&cqr->blocklist, final_queue);
+ }
+}
+
+static void dasd_return_cqr_cb(struct dasd_ccw_req *cqr, void *data)
+{
+ dasd_schedule_block_bh(cqr->block);
+}
+
+static void __dasd_block_start_head(struct dasd_block *block)
+{
+ struct dasd_ccw_req *cqr;
+
+ if (list_empty(&block->ccw_queue))
+ return;
+ /* We allways begin with the first requests on the queue, as some
+ * of previously started requests have to be enqueued on a
+ * dasd_device again for error recovery.
+ */
+ list_for_each_entry(cqr, &block->ccw_queue, blocklist) {
+ if (cqr->status != DASD_CQR_FILLED)
+ continue;
+ /* Non-temporary stop condition will trigger fail fast */
+ if (block->base->stopped & ~DASD_STOPPED_PENDING &&
+ test_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags) &&
+ (!dasd_eer_enabled(block->base))) {
+ cqr->status = DASD_CQR_FAILED;
+ dasd_schedule_block_bh(block);
+ continue;
+ }
+ /* Don't try to start requests if device is stopped */
+ if (block->base->stopped)
+ return;
+
+ /* just a fail safe check, should not happen */
+ if (!cqr->startdev)
+ cqr->startdev = block->base;
+
+ /* make sure that the requests we submit find their way back */
+ cqr->callback = dasd_return_cqr_cb;
+
+ dasd_add_request_tail(cqr);
+ }
+}
+
+/*
+ * Central dasd_block layer routine. Takes requests from the generic
+ * block layer request queue, creates ccw requests, enqueues them on
+ * a dasd_device and processes ccw requests that have been returned.
+ */
+static void dasd_block_tasklet(struct dasd_block *block)
+{
+ struct list_head final_queue;
+ struct list_head *l, *n;
+ struct dasd_ccw_req *cqr;
+
+ atomic_set(&block->tasklet_scheduled, 0);
+ INIT_LIST_HEAD(&final_queue);
+ spin_lock(&block->queue_lock);
+ /* Finish off requests on ccw queue */
+ __dasd_process_block_ccw_queue(block, &final_queue);
+ spin_unlock(&block->queue_lock);
+ /* Now call the callback function of requests with final status */
+ spin_lock_irq(&block->request_queue_lock);
+ list_for_each_safe(l, n, &final_queue) {
+ cqr = list_entry(l, struct dasd_ccw_req, blocklist);
+ list_del_init(&cqr->blocklist);
+ __dasd_cleanup_cqr(cqr);
+ }
+ spin_lock(&block->queue_lock);
+ /* Get new request from the block device request queue */
+ __dasd_process_request_queue(block);
+ /* Now check if the head of the ccw queue needs to be started. */
+ __dasd_block_start_head(block);
+ spin_unlock(&block->queue_lock);
+ spin_unlock_irq(&block->request_queue_lock);
+ dasd_put_device(block->base);
+}
+
+static void _dasd_wake_block_flush_cb(struct dasd_ccw_req *cqr, void *data)
+{
+ wake_up(&dasd_flush_wq);
+}
+
+/*
+ * Go through all request on the dasd_block request queue, cancel them
+ * on the respective dasd_device, and return them to the generic
+ * block layer.
+ */
+static int dasd_flush_block_queue(struct dasd_block *block)
+{
+ struct dasd_ccw_req *cqr, *n;
+ int rc, i;
+ struct list_head flush_queue;
+
+ INIT_LIST_HEAD(&flush_queue);
+ spin_lock_bh(&block->queue_lock);
+ rc = 0;
+restart:
+ list_for_each_entry_safe(cqr, n, &block->ccw_queue, blocklist) {
+ /* if this request currently owned by a dasd_device cancel it */
+ if (cqr->status >= DASD_CQR_QUEUED)
+ rc = dasd_cancel_req(cqr);
+ if (rc < 0)
+ break;
+ /* Rechain request (including erp chain) so it won't be
+ * touched by the dasd_block_tasklet anymore.
+ * Replace the callback so we notice when the request
+ * is returned from the dasd_device layer.
+ */
+ cqr->callback = _dasd_wake_block_flush_cb;
+ for (i = 0; cqr != NULL; cqr = cqr->refers, i++)
+ list_move_tail(&cqr->blocklist, &flush_queue);
+ if (i > 1)
+ /* moved more than one request - need to restart */
+ goto restart;
+ }
+ spin_unlock_bh(&block->queue_lock);
+ /* Now call the callback function of flushed requests */
+restart_cb:
+ list_for_each_entry_safe(cqr, n, &flush_queue, blocklist) {
+ wait_event(dasd_flush_wq, (cqr->status < DASD_CQR_QUEUED));
+ /* Process finished ERP request. */
+ if (cqr->refers) {
+ __dasd_block_process_erp(block, cqr);
+ /* restart list_for_xx loop since dasd_process_erp
+ * might remove multiple elements */
+ goto restart_cb;
+ }
+ /* call the callback function */
+ cqr->endclk = get_clock();
+ list_del_init(&cqr->blocklist);
+ __dasd_cleanup_cqr(cqr);
}
- spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
- dasd_schedule_bh(device);
return rc;
}
/*
- * SECTION: Block device operations (request queue, partitions, open, release).
+ * Schedules a call to dasd_tasklet over the device tasklet.
+ */
+void dasd_schedule_block_bh(struct dasd_block *block)
+{
+ /* Protect against rescheduling. */
+ if (atomic_cmpxchg(&block->tasklet_scheduled, 0, 1) != 0)
+ return;
+ /* life cycle of block is bound to it's base device */
+ dasd_get_device(block->base);
+ tasklet_hi_schedule(&block->tasklet);
+}
+
+
+/*
+ * SECTION: external block device operations
+ * (request queue handling, open, release, etc.)
*/
/*
* Dasd request queue function. Called from ll_rw_blk.c
*/
-static void
-do_dasd_request(struct request_queue * queue)
+static void do_dasd_request(struct request_queue *queue)
{
- struct dasd_device *device;
+ struct dasd_block *block;
- device = (struct dasd_device *) queue->queuedata;
- spin_lock(get_ccwdev_lock(device->cdev));
+ block = queue->queuedata;
+ spin_lock(&block->queue_lock);
/* Get new request from the block device request queue */
- __dasd_process_blk_queue(device);
+ __dasd_process_request_queue(block);
/* Now check if the head of the ccw queue needs to be started. */
- __dasd_start_head(device);
- spin_unlock(get_ccwdev_lock(device->cdev));
+ __dasd_block_start_head(block);
+ spin_unlock(&block->queue_lock);
}
/*
* Allocate and initialize request queue and default I/O scheduler.
*/
-static int
-dasd_alloc_queue(struct dasd_device * device)
+static int dasd_alloc_queue(struct dasd_block *block)
{
int rc;
- device->request_queue = blk_init_queue(do_dasd_request,
- &device->request_queue_lock);
- if (device->request_queue == NULL)
+ block->request_queue = blk_init_queue(do_dasd_request,
+ &block->request_queue_lock);
+ if (block->request_queue == NULL)
return -ENOMEM;
- device->request_queue->queuedata = device;
+ block->request_queue->queuedata = block;
- elevator_exit(device->request_queue->elevator);
- rc = elevator_init(device->request_queue, "deadline");
+ elevator_exit(block->request_queue->elevator);
+ rc = elevator_init(block->request_queue, "deadline");
if (rc) {
- blk_cleanup_queue(device->request_queue);
+ blk_cleanup_queue(block->request_queue);
return rc;
}
return 0;
@@ -1780,79 +1977,76 @@ dasd_alloc_queue(struct dasd_device * de
/*
* Allocate and initialize request queue.
*/
-static void
-dasd_setup_queue(struct dasd_device * device)
+static void dasd_setup_queue(struct dasd_block *block)
{
int max;
- blk_queue_hardsect_size(device->request_queue, device->bp_block);
- max = device->discipline->max_blocks << device->s2b_shift;
- blk_queue_max_sectors(device->request_queue, max);
- blk_queue_max_phys_segments(device->request_queue, -1L);
- blk_queue_max_hw_segments(device->request_queue, -1L);
- blk_queue_max_segment_size(device->request_queue, -1L);
- blk_queue_segment_boundary(device->request_queue, -1L);
- blk_queue_ordered(device->request_queue, QUEUE_ORDERED_TAG, NULL);
+ blk_queue_hardsect_size(block->request_queue, block->bp_block);
+ max = block->base->discipline->max_blocks << block->s2b_shift;
+ blk_queue_max_sectors(block->request_queue, max);
+ blk_queue_max_phys_segments(block->request_queue, -1L);
+ blk_queue_max_hw_segments(block->request_queue, -1L);
+ blk_queue_max_segment_size(block->request_queue, -1L);
+ blk_queue_segment_boundary(block->request_queue, -1L);
+ blk_queue_ordered(block->request_queue, QUEUE_ORDERED_DRAIN, NULL);
}
/*
* Deactivate and free request queue.
*/
-static void
-dasd_free_queue(struct dasd_device * device)
+static void dasd_free_queue(struct dasd_block *block)
{
- if (device->request_queue) {
- blk_cleanup_queue(device->request_queue);
- device->request_queue = NULL;
+ if (block->request_queue) {
+ blk_cleanup_queue(block->request_queue);
+ block->request_queue = NULL;
}
}
/*
* Flush request on the request queue.
*/
-static void
-dasd_flush_request_queue(struct dasd_device * device)
+static void dasd_flush_request_queue(struct dasd_block *block)
{
struct request *req;
- if (!device->request_queue)
+ if (!block->request_queue)
return;
- spin_lock_irq(&device->request_queue_lock);
- while ((req = elv_next_request(device->request_queue))) {
+ spin_lock_irq(&block->request_queue_lock);
+ while ((req = elv_next_request(block->request_queue))) {
blkdev_dequeue_request(req);
dasd_end_request(req, 0);
}
- spin_unlock_irq(&device->request_queue_lock);
+ spin_unlock_irq(&block->request_queue_lock);
}
-static int
-dasd_open(struct inode *inp, struct file *filp)
+static int dasd_open(struct inode *inp, struct file *filp)
{
struct gendisk *disk = inp->i_bdev->bd_disk;
- struct dasd_device *device = disk->private_data;
+ struct dasd_block *block = disk->private_data;
+ struct dasd_device *base = block->base;
int rc;
- atomic_inc(&device->open_count);
- if (test_bit(DASD_FLAG_OFFLINE, &device->flags)) {
+ atomic_inc(&block->open_count);
+ if (test_bit(DASD_FLAG_OFFLINE, &base->flags)) {
rc = -ENODEV;
goto unlock;
}
- if (!try_module_get(device->discipline->owner)) {
+ if (!try_module_get(base->discipline->owner)) {
rc = -EINVAL;
goto unlock;
}
if (dasd_probeonly) {
- DEV_MESSAGE(KERN_INFO, device, "%s",
+ DEV_MESSAGE(KERN_INFO, base, "%s",
"No access to device due to probeonly mode");
rc = -EPERM;
goto out;
}
- if (device->state <= DASD_STATE_BASIC) {
- DBF_DEV_EVENT(DBF_ERR, device, " %s",
+ if (base->state <= DASD_STATE_BASIC) {
+ DBF_DEV_EVENT(DBF_ERR, base, " %s",
" Cannot open unrecognized device");
rc = -ENODEV;
goto out;
@@ -1861,41 +2055,41 @@ dasd_open(struct inode *inp, struct file
return 0;
out:
- module_put(device->discipline->owner);
+ module_put(base->discipline->owner);
unlock:
- atomic_dec(&device->open_count);
+ atomic_dec(&block->open_count);
return rc;
}
-static int
-dasd_release(struct inode *inp, struct file *filp)
+static int dasd_release(struct inode *inp, struct file *filp)
{
struct gendisk *disk = inp->i_bdev->bd_disk;
- struct dasd_device *device = disk->private_data;
+ struct dasd_block *block = disk->private_data;
- atomic_dec(&device->open_count);
- module_put(device->discipline->owner);
+ atomic_dec(&block->open_count);
+ module_put(block->base->discipline->owner);
return 0;
}
/*
* Return disk geometry.
*/
-static int
-dasd_getgeo(struct block_device *bdev, struct hd_geometry *geo)
+static int dasd_getgeo(struct block_device *bdev, struct hd_geometry *geo)
{
- struct dasd_device *device;
+ struct dasd_block *block;
+ struct dasd_device *base;
- device = bdev->bd_disk->private_data;
- if (!device)
+ block = bdev->bd_disk->private_data;
+ base = block->base;
+ if (!block)
return -ENODEV;
- if (!device->discipline ||
- !device->discipline->fill_geometry)
+ if (!base->discipline ||
+ !base->discipline->fill_geometry)
return -EINVAL;
- device->discipline->fill_geometry(device, geo);
- geo->start = get_start_sect(bdev) >> device->s2b_shift;
+ base->discipline->fill_geometry(block, geo);
+ geo->start = get_start_sect(bdev) >> block->s2b_shift;
return 0;
}
@@ -1909,6 +2103,9 @@ dasd_device_operations = {
.getgeo = dasd_getgeo,
};
+/*******************************************************************************
+ * end of block device operations
+ */
static void
dasd_exit(void)
@@ -1937,9 +2134,8 @@ dasd_exit(void)
* Initial attempt at a probe function. this can be simplified once
* the other detection code is gone.
*/
-int
-dasd_generic_probe (struct ccw_device *cdev,
- struct dasd_discipline *discipline)
+int dasd_generic_probe(struct ccw_device *cdev,
+ struct dasd_discipline *discipline)
{
int ret;
@@ -1979,10 +2175,10 @@ dasd_generic_probe (struct ccw_device *c
* This will one day be called from a global not_oper handler.
* It is also used by driver_unregister during module unload.
*/
-void
-dasd_generic_remove (struct ccw_device *cdev)
+void dasd_generic_remove(struct ccw_device *cdev)
{
struct dasd_device *device;
+ struct dasd_block *block;
cdev->handler = NULL;
@@ -2002,7 +2198,15 @@ dasd_generic_remove (struct ccw_device *
*/
dasd_set_target_state(device, DASD_STATE_NEW);
/* dasd_delete_device destroys the device reference. */
+ block = device->block;
+ device->block = NULL;
dasd_delete_device(device);
+ /*
+ * life cycle of block is bound to device, so delete it after
+ * device was safely removed
+ */
+ if (block)
+ dasd_free_block(block);
}
/*
@@ -2010,10 +2214,8 @@ dasd_generic_remove (struct ccw_device *
* the device is detected for the first time and is supposed to be used
* or the user has started activation through sysfs.
*/
-int
-dasd_generic_set_online (struct ccw_device *cdev,
- struct dasd_discipline *base_discipline)
-
+int dasd_generic_set_online(struct ccw_device *cdev,
+ struct dasd_discipline *base_discipline)
{
struct dasd_discipline *discipline;
struct dasd_device *device;
@@ -2049,6 +2251,7 @@ dasd_generic_set_online (struct ccw_devi
device->base_discipline = base_discipline;
device->discipline = discipline;
+ /* check_device will allocate block device if necessary */
rc = discipline->check_device(device);
if (rc) {
printk (KERN_WARNING
@@ -2068,6 +2271,8 @@ dasd_generic_set_online (struct ccw_devi
cdev->dev.bus_id);
rc = -ENODEV;
dasd_set_target_state(device, DASD_STATE_NEW);
+ if (device->block)
+ dasd_free_block(device->block);
dasd_delete_device(device);
} else
pr_debug("dasd_generic device %s found\n",
@@ -2082,10 +2287,10 @@ dasd_generic_set_online (struct ccw_devi
return rc;
}
-int
-dasd_generic_set_offline (struct ccw_device *cdev)
+int dasd_generic_set_offline(struct ccw_device *cdev)
{
struct dasd_device *device;
+ struct dasd_block *block;
int max_count, open_count;
device = dasd_device_from_cdev(cdev);
@@ -2102,30 +2307,39 @@ dasd_generic_set_offline (struct ccw_dev
* the blkdev_get in dasd_scan_partitions. We are only interested
* in the other openers.
*/
- max_count = device->bdev ? 0 : -1;
- open_count = (int) atomic_read(&device->open_count);
- if (open_count > max_count) {
- if (open_count > 0)
- printk (KERN_WARNING "Can't offline dasd device with "
- "open count = %i.\n",
- open_count);
- else
- printk (KERN_WARNING "%s",
- "Can't offline dasd device due to internal "
- "use\n");
- clear_bit(DASD_FLAG_OFFLINE, &device->flags);
- dasd_put_device(device);
- return -EBUSY;
+ if (device->block) {
+ struct dasd_block *block = device->block;
+ max_count = block->bdev ? 0 : -1;
+ open_count = (int) atomic_read(&block->open_count);
+ if (open_count > max_count) {
+ if (open_count > 0)
+ printk(KERN_WARNING "Can't offline dasd "
+ "device with open count = %i.\n",
+ open_count);
+ else
+ printk(KERN_WARNING "%s",
+ "Can't offline dasd device due "
+ "to internal use\n");
+ clear_bit(DASD_FLAG_OFFLINE, &device->flags);
+ dasd_put_device(device);
+ return -EBUSY;
+ }
}
dasd_set_target_state(device, DASD_STATE_NEW);
/* dasd_delete_device destroys the device reference. */
+ block = device->block;
+ device->block = NULL;
dasd_delete_device(device);
-
+ /*
+ * life cycle of block is bound to device, so delete it after
+ * device was safely removed
+ */
+ if (block)
+ dasd_free_block(block);
return 0;
}
-int
-dasd_generic_notify(struct ccw_device *cdev, int event)
+int dasd_generic_notify(struct ccw_device *cdev, int event)
{
struct dasd_device *device;
struct dasd_ccw_req *cqr;
@@ -2146,27 +2360,22 @@ dasd_generic_notify(struct ccw_device *c
if (device->state < DASD_STATE_BASIC)
break;
/* Device is active. We want to keep it. */
- if (test_bit(DASD_FLAG_DSC_ERROR, &device->flags)) {
- list_for_each_entry(cqr, &device->ccw_queue, list)
- if (cqr->status == DASD_CQR_IN_IO)
- cqr->status = DASD_CQR_FAILED;
- device->stopped |= DASD_STOPPED_DC_EIO;
- } else {
- list_for_each_entry(cqr, &device->ccw_queue, list)
- if (cqr->status == DASD_CQR_IN_IO) {
- cqr->status = DASD_CQR_QUEUED;
- cqr->retries++;
- }
- device->stopped |= DASD_STOPPED_DC_WAIT;
- dasd_set_timer(device, 0);
- }
- dasd_schedule_bh(device);
+ list_for_each_entry(cqr, &device->ccw_queue, devlist)
+ if (cqr->status == DASD_CQR_IN_IO) {
+ cqr->status = DASD_CQR_QUEUED;
+ cqr->retries++;
+ }
+ device->stopped |= DASD_STOPPED_DC_WAIT;
+ dasd_device_clear_timer(device);
+ dasd_schedule_device_bh(device);
ret = 1;
break;
case CIO_OPER:
/* FIXME: add a sanity check. */
- device->stopped &= ~(DASD_STOPPED_DC_WAIT|DASD_STOPPED_DC_EIO);
- dasd_schedule_bh(device);
+ device->stopped &= ~DASD_STOPPED_DC_WAIT;
+ dasd_schedule_device_bh(device);
+ if (device->block)
+ dasd_schedule_block_bh(device->block);
ret = 1;
break;
}
@@ -2196,7 +2405,8 @@ static struct dasd_ccw_req *dasd_generic
ccw->cda = (__u32)(addr_t)rdc_buffer;
ccw->count = rdc_buffer_size;
- cqr->device = device;
+ cqr->startdev = device;
+ cqr->memdev = device;
cqr->expires = 10*HZ;
clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
cqr->retries = 2;
@@ -2218,13 +2428,12 @@ int dasd_generic_read_dev_chars(struct d
return PTR_ERR(cqr);
ret = dasd_sleep_on(cqr);
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
return ret;
}
EXPORT_SYMBOL_GPL(dasd_generic_read_dev_chars);
-static int __init
-dasd_init(void)
+static int __init dasd_init(void)
{
int rc;
@@ -2232,7 +2441,7 @@ dasd_init(void)
init_waitqueue_head(&dasd_flush_wq);
/* register 'common' DASD debug area, used for all DBF_XXX calls */
- dasd_debug_area = debug_register("dasd", 1, 2, 8 * sizeof (long));
+ dasd_debug_area = debug_register("dasd", 1, 2, 8 * sizeof(long));
if (dasd_debug_area == NULL) {
rc = -ENOMEM;
goto failed;
@@ -2278,15 +2487,18 @@ EXPORT_SYMBOL(dasd_diag_discipline_point
EXPORT_SYMBOL(dasd_add_request_head);
EXPORT_SYMBOL(dasd_add_request_tail);
EXPORT_SYMBOL(dasd_cancel_req);
-EXPORT_SYMBOL(dasd_clear_timer);
+EXPORT_SYMBOL(dasd_device_clear_timer);
+EXPORT_SYMBOL(dasd_block_clear_timer);
EXPORT_SYMBOL(dasd_enable_device);
EXPORT_SYMBOL(dasd_int_handler);
EXPORT_SYMBOL(dasd_kfree_request);
EXPORT_SYMBOL(dasd_kick_device);
EXPORT_SYMBOL(dasd_kmalloc_request);
-EXPORT_SYMBOL(dasd_schedule_bh);
+EXPORT_SYMBOL(dasd_schedule_device_bh);
+EXPORT_SYMBOL(dasd_schedule_block_bh);
EXPORT_SYMBOL(dasd_set_target_state);
-EXPORT_SYMBOL(dasd_set_timer);
+EXPORT_SYMBOL(dasd_device_set_timer);
+EXPORT_SYMBOL(dasd_block_set_timer);
EXPORT_SYMBOL(dasd_sfree_request);
EXPORT_SYMBOL(dasd_sleep_on);
EXPORT_SYMBOL(dasd_sleep_on_immediatly);
@@ -2300,4 +2512,7 @@ EXPORT_SYMBOL_GPL(dasd_generic_remove);
EXPORT_SYMBOL_GPL(dasd_generic_notify);
EXPORT_SYMBOL_GPL(dasd_generic_set_online);
EXPORT_SYMBOL_GPL(dasd_generic_set_offline);
-
+EXPORT_SYMBOL_GPL(dasd_generic_handle_state_change);
+EXPORT_SYMBOL_GPL(dasd_flush_device_queue);
+EXPORT_SYMBOL_GPL(dasd_alloc_block);
+EXPORT_SYMBOL_GPL(dasd_free_block);
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
* [patch 47/47] dasd: add hyper PAV support to DASD device driver, part 3
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
` (45 preceding siblings ...)
2007-12-20 15:20 ` [patch 46/47] dasd: add hyper PAV support to DASD device driver, part 2 Martin Schwidefsky
@ 2007-12-20 15:20 ` Martin Schwidefsky
46 siblings, 0 replies; 48+ messages in thread
From: Martin Schwidefsky @ 2007-12-20 15:20 UTC (permalink / raw)
To: linux-kernel, linux-s390; +Cc: Stefan Weinhuber, Martin Schwidefsky
[-- Attachment #1: 146-hyperpav-3.diff --]
[-- Type: text/plain, Size: 78467 bytes --]
From: Stefan Weinhuber <wein@de.ibm.com>
Signed-off-by: Stefan Weinhuber <wein@de.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
---
drivers/s390/block/dasd_devmap.c | 94 +---
drivers/s390/block/dasd_diag.c | 107 +++--
drivers/s390/block/dasd_eckd.c | 780 ++++++++++++++++++++++++++++-----------
drivers/s390/block/dasd_eckd.h | 125 ++++++
drivers/s390/block/dasd_eer.c | 11
drivers/s390/block/dasd_erp.c | 25 -
drivers/s390/block/dasd_fba.c | 114 +++--
drivers/s390/block/dasd_genhd.c | 76 +--
8 files changed, 912 insertions(+), 420 deletions(-)
Index: quilt-2.6/drivers/s390/block/dasd_devmap.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_devmap.c
+++ quilt-2.6/drivers/s390/block/dasd_devmap.c
@@ -49,22 +49,6 @@ struct dasd_devmap {
};
/*
- * dasd_server_ssid_map contains a globally unique storage server subsystem ID.
- * dasd_server_ssid_list contains the list of all subsystem IDs accessed by
- * the DASD device driver.
- */
-struct dasd_server_ssid_map {
- struct list_head list;
- struct system_id {
- char vendor[4];
- char serial[15];
- __u16 ssid;
- } sid;
-};
-
-static struct list_head dasd_server_ssid_list;
-
-/*
* Parameter parsing functions for dasd= parameter. The syntax is:
* <devno> : (0x)?[0-9a-fA-F]+
* <busid> : [0-0a-f]\.[0-9a-f]\.(0x)?[0-9a-fA-F]+
@@ -721,8 +705,9 @@ dasd_ro_store(struct device *dev, struct
devmap->features &= ~DASD_FEATURE_READONLY;
if (devmap->device)
devmap->device->features = devmap->features;
- if (devmap->device && devmap->device->gdp)
- set_disk_ro(devmap->device->gdp, val);
+ if (devmap->device && devmap->device->block
+ && devmap->device->block->gdp)
+ set_disk_ro(devmap->device->block->gdp, val);
spin_unlock(&dasd_devmap_lock);
return count;
}
@@ -893,12 +878,16 @@ dasd_alias_show(struct device *dev, stru
devmap = dasd_find_busid(dev->bus_id);
spin_lock(&dasd_devmap_lock);
- if (!IS_ERR(devmap))
- alias = devmap->uid.alias;
+ if (IS_ERR(devmap) || strlen(devmap->uid.vendor) == 0) {
+ spin_unlock(&dasd_devmap_lock);
+ return sprintf(buf, "0\n");
+ }
+ if (devmap->uid.type == UA_BASE_PAV_ALIAS ||
+ devmap->uid.type == UA_HYPER_PAV_ALIAS)
+ alias = 1;
else
alias = 0;
spin_unlock(&dasd_devmap_lock);
-
return sprintf(buf, alias ? "1\n" : "0\n");
}
@@ -930,19 +919,36 @@ static ssize_t
dasd_uid_show(struct device *dev, struct device_attribute *attr, char *buf)
{
struct dasd_devmap *devmap;
- char uid[UID_STRLEN];
+ char uid_string[UID_STRLEN];
+ char ua_string[3];
+ struct dasd_uid *uid;
devmap = dasd_find_busid(dev->bus_id);
spin_lock(&dasd_devmap_lock);
- if (!IS_ERR(devmap) && strlen(devmap->uid.vendor) > 0)
- snprintf(uid, sizeof(uid), "%s.%s.%04x.%02x",
- devmap->uid.vendor, devmap->uid.serial,
- devmap->uid.ssid, devmap->uid.unit_addr);
- else
- uid[0] = 0;
+ if (IS_ERR(devmap) || strlen(devmap->uid.vendor) == 0) {
+ spin_unlock(&dasd_devmap_lock);
+ return sprintf(buf, "\n");
+ }
+ uid = &devmap->uid;
+ switch (uid->type) {
+ case UA_BASE_DEVICE:
+ sprintf(ua_string, "%02x", uid->real_unit_addr);
+ break;
+ case UA_BASE_PAV_ALIAS:
+ sprintf(ua_string, "%02x", uid->base_unit_addr);
+ break;
+ case UA_HYPER_PAV_ALIAS:
+ sprintf(ua_string, "xx");
+ break;
+ default:
+ /* should not happen, treat like base device */
+ sprintf(ua_string, "%02x", uid->real_unit_addr);
+ break;
+ }
+ snprintf(uid_string, sizeof(uid_string), "%s.%s.%04x.%s",
+ uid->vendor, uid->serial, uid->ssid, ua_string);
spin_unlock(&dasd_devmap_lock);
-
- return snprintf(buf, PAGE_SIZE, "%s\n", uid);
+ return snprintf(buf, PAGE_SIZE, "%s\n", uid_string);
}
static DEVICE_ATTR(uid, 0444, dasd_uid_show, NULL);
@@ -1040,39 +1046,16 @@ int
dasd_set_uid(struct ccw_device *cdev, struct dasd_uid *uid)
{
struct dasd_devmap *devmap;
- struct dasd_server_ssid_map *srv, *tmp;
devmap = dasd_find_busid(cdev->dev.bus_id);
if (IS_ERR(devmap))
return PTR_ERR(devmap);
- /* generate entry for server_ssid_map */
- srv = (struct dasd_server_ssid_map *)
- kzalloc(sizeof(struct dasd_server_ssid_map), GFP_KERNEL);
- if (!srv)
- return -ENOMEM;
- strncpy(srv->sid.vendor, uid->vendor, sizeof(srv->sid.vendor) - 1);
- strncpy(srv->sid.serial, uid->serial, sizeof(srv->sid.serial) - 1);
- srv->sid.ssid = uid->ssid;
-
- /* server is already contained ? */
spin_lock(&dasd_devmap_lock);
devmap->uid = *uid;
- list_for_each_entry(tmp, &dasd_server_ssid_list, list) {
- if (!memcmp(&srv->sid, &tmp->sid,
- sizeof(struct system_id))) {
- kfree(srv);
- srv = NULL;
- break;
- }
- }
-
- /* add servermap to serverlist */
- if (srv)
- list_add(&srv->list, &dasd_server_ssid_list);
spin_unlock(&dasd_devmap_lock);
- return (srv ? 1 : 0);
+ return 0;
}
EXPORT_SYMBOL_GPL(dasd_set_uid);
@@ -1138,9 +1121,6 @@ dasd_devmap_init(void)
dasd_max_devindex = 0;
for (i = 0; i < 256; i++)
INIT_LIST_HEAD(&dasd_hashlists[i]);
-
- /* Initialize servermap structure. */
- INIT_LIST_HEAD(&dasd_server_ssid_list);
return 0;
}
Index: quilt-2.6/drivers/s390/block/dasd_diag.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_diag.c
+++ quilt-2.6/drivers/s390/block/dasd_diag.c
@@ -142,7 +142,7 @@ dasd_diag_erp(struct dasd_device *device
int rc;
mdsk_term_io(device);
- rc = mdsk_init_io(device, device->bp_block, 0, NULL);
+ rc = mdsk_init_io(device, device->block->bp_block, 0, NULL);
if (rc)
DEV_MESSAGE(KERN_WARNING, device, "DIAG ERP unsuccessful, "
"rc=%d", rc);
@@ -158,11 +158,11 @@ dasd_start_diag(struct dasd_ccw_req * cq
struct dasd_diag_req *dreq;
int rc;
- device = cqr->device;
+ device = cqr->startdev;
if (cqr->retries < 0) {
DEV_MESSAGE(KERN_WARNING, device, "DIAG start_IO: request %p "
"- no retry left)", cqr);
- cqr->status = DASD_CQR_FAILED;
+ cqr->status = DASD_CQR_ERROR;
return -EIO;
}
private = (struct dasd_diag_private *) device->private;
@@ -184,7 +184,7 @@ dasd_start_diag(struct dasd_ccw_req * cq
switch (rc) {
case 0: /* Synchronous I/O finished successfully */
cqr->stopclk = get_clock();
- cqr->status = DASD_CQR_DONE;
+ cqr->status = DASD_CQR_SUCCESS;
/* Indicate to calling function that only a dasd_schedule_bh()
and no timer is needed */
rc = -EACCES;
@@ -209,12 +209,12 @@ dasd_diag_term_IO(struct dasd_ccw_req *
{
struct dasd_device *device;
- device = cqr->device;
+ device = cqr->startdev;
mdsk_term_io(device);
- mdsk_init_io(device, device->bp_block, 0, NULL);
- cqr->status = DASD_CQR_CLEAR;
+ mdsk_init_io(device, device->block->bp_block, 0, NULL);
+ cqr->status = DASD_CQR_CLEAR_PENDING;
cqr->stopclk = get_clock();
- dasd_schedule_bh(device);
+ dasd_schedule_device_bh(device);
return 0;
}
@@ -247,7 +247,7 @@ dasd_ext_handler(__u16 code)
return;
}
cqr = (struct dasd_ccw_req *) ip;
- device = (struct dasd_device *) cqr->device;
+ device = (struct dasd_device *) cqr->startdev;
if (strncmp(device->discipline->ebcname, (char *) &cqr->magic, 4)) {
DEV_MESSAGE(KERN_WARNING, device,
" magic number of dasd_ccw_req 0x%08X doesn't"
@@ -260,10 +260,10 @@ dasd_ext_handler(__u16 code)
spin_lock_irqsave(get_ccwdev_lock(device->cdev), flags);
/* Check for a pending clear operation */
- if (cqr->status == DASD_CQR_CLEAR) {
- cqr->status = DASD_CQR_QUEUED;
- dasd_clear_timer(device);
- dasd_schedule_bh(device);
+ if (cqr->status == DASD_CQR_CLEAR_PENDING) {
+ cqr->status = DASD_CQR_CLEARED;
+ dasd_device_clear_timer(device);
+ dasd_schedule_device_bh(device);
spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
return;
}
@@ -272,11 +272,11 @@ dasd_ext_handler(__u16 code)
expires = 0;
if (status == 0) {
- cqr->status = DASD_CQR_DONE;
+ cqr->status = DASD_CQR_SUCCESS;
/* Start first request on queue if possible -> fast_io. */
if (!list_empty(&device->ccw_queue)) {
next = list_entry(device->ccw_queue.next,
- struct dasd_ccw_req, list);
+ struct dasd_ccw_req, devlist);
if (next->status == DASD_CQR_QUEUED) {
rc = dasd_start_diag(next);
if (rc == 0)
@@ -296,10 +296,10 @@ dasd_ext_handler(__u16 code)
}
if (expires != 0)
- dasd_set_timer(device, expires);
+ dasd_device_set_timer(device, expires);
else
- dasd_clear_timer(device);
- dasd_schedule_bh(device);
+ dasd_device_clear_timer(device);
+ dasd_schedule_device_bh(device);
spin_unlock_irqrestore(get_ccwdev_lock(device->cdev), flags);
}
@@ -309,6 +309,7 @@ dasd_ext_handler(__u16 code)
static int
dasd_diag_check_device(struct dasd_device *device)
{
+ struct dasd_block *block;
struct dasd_diag_private *private;
struct dasd_diag_characteristics *rdc_data;
struct dasd_diag_bio bio;
@@ -328,6 +329,16 @@ dasd_diag_check_device(struct dasd_devic
ccw_device_get_id(device->cdev, &private->dev_id);
device->private = (void *) private;
}
+ block = dasd_alloc_block();
+ if (IS_ERR(block)) {
+ DEV_MESSAGE(KERN_WARNING, device, "%s",
+ "could not allocate dasd block structure");
+ kfree(device->private);
+ return PTR_ERR(block);
+ }
+ device->block = block;
+ block->base = device;
+
/* Read Device Characteristics */
rdc_data = (void *) &(private->rdc_data);
rdc_data->dev_nr = private->dev_id.devno;
@@ -409,14 +420,14 @@ dasd_diag_check_device(struct dasd_devic
sizeof(DASD_DIAG_CMS1)) == 0) {
/* get formatted blocksize from label block */
bsize = (unsigned int) label->block_size;
- device->blocks = (unsigned long) label->block_count;
+ block->blocks = (unsigned long) label->block_count;
} else
- device->blocks = end_block;
- device->bp_block = bsize;
- device->s2b_shift = 0; /* bits to shift 512 to get a block */
+ block->blocks = end_block;
+ block->bp_block = bsize;
+ block->s2b_shift = 0; /* bits to shift 512 to get a block */
for (sb = 512; sb < bsize; sb = sb << 1)
- device->s2b_shift++;
- rc = mdsk_init_io(device, device->bp_block, 0, NULL);
+ block->s2b_shift++;
+ rc = mdsk_init_io(device, block->bp_block, 0, NULL);
if (rc) {
DEV_MESSAGE(KERN_WARNING, device, "DIAG initialization "
"failed (rc=%d)", rc);
@@ -424,9 +435,9 @@ dasd_diag_check_device(struct dasd_devic
} else {
DEV_MESSAGE(KERN_INFO, device,
"(%ld B/blk): %ldkB",
- (unsigned long) device->bp_block,
- (unsigned long) (device->blocks <<
- device->s2b_shift) >> 1);
+ (unsigned long) block->bp_block,
+ (unsigned long) (block->blocks <<
+ block->s2b_shift) >> 1);
}
out:
free_page((long) label);
@@ -436,22 +447,16 @@ out:
/* Fill in virtual disk geometry for device. Return zero on success, non-zero
* otherwise. */
static int
-dasd_diag_fill_geometry(struct dasd_device *device, struct hd_geometry *geo)
+dasd_diag_fill_geometry(struct dasd_block *block, struct hd_geometry *geo)
{
- if (dasd_check_blocksize(device->bp_block) != 0)
+ if (dasd_check_blocksize(block->bp_block) != 0)
return -EINVAL;
- geo->cylinders = (device->blocks << device->s2b_shift) >> 10;
+ geo->cylinders = (block->blocks << block->s2b_shift) >> 10;
geo->heads = 16;
- geo->sectors = 128 >> device->s2b_shift;
+ geo->sectors = 128 >> block->s2b_shift;
return 0;
}
-static dasd_era_t
-dasd_diag_examine_error(struct dasd_ccw_req * cqr, struct irb * stat)
-{
- return dasd_era_fatal;
-}
-
static dasd_erp_fn_t
dasd_diag_erp_action(struct dasd_ccw_req * cqr)
{
@@ -466,8 +471,9 @@ dasd_diag_erp_postaction(struct dasd_ccw
/* Create DASD request from block device request. Return pointer to new
* request on success, ERR_PTR otherwise. */
-static struct dasd_ccw_req *
-dasd_diag_build_cp(struct dasd_device * device, struct request *req)
+static struct dasd_ccw_req *dasd_diag_build_cp(struct dasd_device *memdev,
+ struct dasd_block *block,
+ struct request *req)
{
struct dasd_ccw_req *cqr;
struct dasd_diag_req *dreq;
@@ -486,17 +492,17 @@ dasd_diag_build_cp(struct dasd_device *
rw_cmd = MDSK_WRITE_REQ;
else
return ERR_PTR(-EINVAL);
- blksize = device->bp_block;
+ blksize = block->bp_block;
/* Calculate record id of first and last block. */
- first_rec = req->sector >> device->s2b_shift;
- last_rec = (req->sector + req->nr_sectors - 1) >> device->s2b_shift;
+ first_rec = req->sector >> block->s2b_shift;
+ last_rec = (req->sector + req->nr_sectors - 1) >> block->s2b_shift;
/* Check struct bio and count the number of blocks for the request. */
count = 0;
rq_for_each_segment(bv, req, iter) {
if (bv->bv_len & (blksize - 1))
/* Fba can only do full blocks. */
return ERR_PTR(-EINVAL);
- count += bv->bv_len >> (device->s2b_shift + 9);
+ count += bv->bv_len >> (block->s2b_shift + 9);
}
/* Paranoia. */
if (count != last_rec - first_rec + 1)
@@ -505,7 +511,7 @@ dasd_diag_build_cp(struct dasd_device *
datasize = sizeof(struct dasd_diag_req) +
count*sizeof(struct dasd_diag_bio);
cqr = dasd_smalloc_request(dasd_diag_discipline.name, 0,
- datasize, device);
+ datasize, memdev);
if (IS_ERR(cqr))
return cqr;
@@ -529,7 +535,9 @@ dasd_diag_build_cp(struct dasd_device *
cqr->buildclk = get_clock();
if (req->cmd_flags & REQ_FAILFAST)
set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
- cqr->device = device;
+ cqr->startdev = memdev;
+ cqr->memdev = memdev;
+ cqr->block = block;
cqr->expires = DIAG_TIMEOUT;
cqr->status = DASD_CQR_FILLED;
return cqr;
@@ -543,10 +551,15 @@ dasd_diag_free_cp(struct dasd_ccw_req *c
int status;
status = cqr->status == DASD_CQR_DONE;
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
return status;
}
+static void dasd_diag_handle_terminated_request(struct dasd_ccw_req *cqr)
+{
+ cqr->status = DASD_CQR_FILLED;
+};
+
/* Fill in IOCTL data for device. */
static int
dasd_diag_fill_info(struct dasd_device * device,
@@ -583,7 +596,7 @@ static struct dasd_discipline dasd_diag_
.fill_geometry = dasd_diag_fill_geometry,
.start_IO = dasd_start_diag,
.term_IO = dasd_diag_term_IO,
- .examine_error = dasd_diag_examine_error,
+ .handle_terminated_request = dasd_diag_handle_terminated_request,
.erp_action = dasd_diag_erp_action,
.erp_postaction = dasd_diag_erp_postaction,
.build_cp = dasd_diag_build_cp,
Index: quilt-2.6/drivers/s390/block/dasd_eckd.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_eckd.c
+++ quilt-2.6/drivers/s390/block/dasd_eckd.c
@@ -52,16 +52,6 @@ MODULE_LICENSE("GPL");
static struct dasd_discipline dasd_eckd_discipline;
-struct dasd_eckd_private {
- struct dasd_eckd_characteristics rdc_data;
- struct dasd_eckd_confdata conf_data;
- struct dasd_eckd_path path_data;
- struct eckd_count count_area[5];
- int init_cqr_status;
- int uses_cdl;
- struct attrib_data_t attrib; /* e.g. cache operations */
-};
-
/* The ccw bus type uses this table to find devices that it sends to
* dasd_eckd_probe */
static struct ccw_device_id dasd_eckd_ids[] = {
@@ -188,7 +178,7 @@ check_XRC (struct ccw1 *de_ccw,
if (rc == -ENOSYS || rc == -EACCES)
rc = 0;
- de_ccw->count = sizeof (struct DE_eckd_data);
+ de_ccw->count = sizeof(struct DE_eckd_data);
de_ccw->flags |= CCW_FLAG_SLI;
return rc;
}
@@ -208,7 +198,7 @@ define_extent(struct ccw1 * ccw, struct
ccw->count = 16;
ccw->cda = (__u32) __pa(data);
- memset(data, 0, sizeof (struct DE_eckd_data));
+ memset(data, 0, sizeof(struct DE_eckd_data));
switch (cmd) {
case DASD_ECKD_CCW_READ_HOME_ADDRESS:
case DASD_ECKD_CCW_READ_RECORD_ZERO:
@@ -280,6 +270,132 @@ define_extent(struct ccw1 * ccw, struct
return rc;
}
+static int check_XRC_on_prefix(struct PFX_eckd_data *pfxdata,
+ struct dasd_device *device)
+{
+ struct dasd_eckd_private *private;
+ int rc;
+
+ private = (struct dasd_eckd_private *) device->private;
+ if (!private->rdc_data.facilities.XRC_supported)
+ return 0;
+
+ /* switch on System Time Stamp - needed for XRC Support */
+ pfxdata->define_extend.ga_extended |= 0x08; /* 'Time Stamp Valid' */
+ pfxdata->define_extend.ga_extended |= 0x02; /* 'Extended Parameter' */
+ pfxdata->validity.time_stamp = 1; /* 'Time Stamp Valid' */
+
+ rc = get_sync_clock(&pfxdata->define_extend.ep_sys_time);
+ /* Ignore return code if sync clock is switched off. */
+ if (rc == -ENOSYS || rc == -EACCES)
+ rc = 0;
+ return rc;
+}
+
+static int prefix(struct ccw1 *ccw, struct PFX_eckd_data *pfxdata, int trk,
+ int totrk, int cmd, struct dasd_device *basedev,
+ struct dasd_device *startdev)
+{
+ struct dasd_eckd_private *basepriv, *startpriv;
+ struct DE_eckd_data *data;
+ struct ch_t geo, beg, end;
+ int rc = 0;
+
+ basepriv = (struct dasd_eckd_private *) basedev->private;
+ startpriv = (struct dasd_eckd_private *) startdev->private;
+ data = &pfxdata->define_extend;
+
+ ccw->cmd_code = DASD_ECKD_CCW_PFX;
+ ccw->flags = 0;
+ ccw->count = sizeof(*pfxdata);
+ ccw->cda = (__u32) __pa(pfxdata);
+
+ memset(pfxdata, 0, sizeof(*pfxdata));
+ /* prefix data */
+ pfxdata->format = 0;
+ pfxdata->base_address = basepriv->conf_data.ned1.unit_addr;
+ pfxdata->base_lss = basepriv->conf_data.ned1.ID;
+ pfxdata->validity.define_extend = 1;
+
+ /* private uid is kept up to date, conf_data may be outdated */
+ if (startpriv->uid.type != UA_BASE_DEVICE) {
+ pfxdata->validity.verify_base = 1;
+ if (startpriv->uid.type == UA_HYPER_PAV_ALIAS)
+ pfxdata->validity.hyper_pav = 1;
+ }
+
+ /* define extend data (mostly)*/
+ switch (cmd) {
+ case DASD_ECKD_CCW_READ_HOME_ADDRESS:
+ case DASD_ECKD_CCW_READ_RECORD_ZERO:
+ case DASD_ECKD_CCW_READ:
+ case DASD_ECKD_CCW_READ_MT:
+ case DASD_ECKD_CCW_READ_CKD:
+ case DASD_ECKD_CCW_READ_CKD_MT:
+ case DASD_ECKD_CCW_READ_KD:
+ case DASD_ECKD_CCW_READ_KD_MT:
+ case DASD_ECKD_CCW_READ_COUNT:
+ data->mask.perm = 0x1;
+ data->attributes.operation = basepriv->attrib.operation;
+ break;
+ case DASD_ECKD_CCW_WRITE:
+ case DASD_ECKD_CCW_WRITE_MT:
+ case DASD_ECKD_CCW_WRITE_KD:
+ case DASD_ECKD_CCW_WRITE_KD_MT:
+ data->mask.perm = 0x02;
+ data->attributes.operation = basepriv->attrib.operation;
+ rc = check_XRC_on_prefix(pfxdata, basedev);
+ break;
+ case DASD_ECKD_CCW_WRITE_CKD:
+ case DASD_ECKD_CCW_WRITE_CKD_MT:
+ data->attributes.operation = DASD_BYPASS_CACHE;
+ rc = check_XRC_on_prefix(pfxdata, basedev);
+ break;
+ case DASD_ECKD_CCW_ERASE:
+ case DASD_ECKD_CCW_WRITE_HOME_ADDRESS:
+ case DASD_ECKD_CCW_WRITE_RECORD_ZERO:
+ data->mask.perm = 0x3;
+ data->mask.auth = 0x1;
+ data->attributes.operation = DASD_BYPASS_CACHE;
+ rc = check_XRC_on_prefix(pfxdata, basedev);
+ break;
+ default:
+ DEV_MESSAGE(KERN_ERR, basedev, "unknown opcode 0x%x", cmd);
+ break;
+ }
+
+ data->attributes.mode = 0x3; /* ECKD */
+
+ if ((basepriv->rdc_data.cu_type == 0x2105 ||
+ basepriv->rdc_data.cu_type == 0x2107 ||
+ basepriv->rdc_data.cu_type == 0x1750)
+ && !(basepriv->uses_cdl && trk < 2))
+ data->ga_extended |= 0x40; /* Regular Data Format Mode */
+
+ geo.cyl = basepriv->rdc_data.no_cyl;
+ geo.head = basepriv->rdc_data.trk_per_cyl;
+ beg.cyl = trk / geo.head;
+ beg.head = trk % geo.head;
+ end.cyl = totrk / geo.head;
+ end.head = totrk % geo.head;
+
+ /* check for sequential prestage - enhance cylinder range */
+ if (data->attributes.operation == DASD_SEQ_PRESTAGE ||
+ data->attributes.operation == DASD_SEQ_ACCESS) {
+
+ if (end.cyl + basepriv->attrib.nr_cyl < geo.cyl)
+ end.cyl += basepriv->attrib.nr_cyl;
+ else
+ end.cyl = (geo.cyl - 1);
+ }
+
+ data->beg_ext.cyl = beg.cyl;
+ data->beg_ext.head = beg.head;
+ data->end_ext.cyl = end.cyl;
+ data->end_ext.head = end.head;
+ return rc;
+}
+
static void
locate_record(struct ccw1 *ccw, struct LO_eckd_data *data, int trk,
int rec_on_trk, int no_rec, int cmd,
@@ -300,7 +416,7 @@ locate_record(struct ccw1 *ccw, struct L
ccw->count = 16;
ccw->cda = (__u32) __pa(data);
- memset(data, 0, sizeof (struct LO_eckd_data));
+ memset(data, 0, sizeof(struct LO_eckd_data));
sector = 0;
if (rec_on_trk) {
switch (private->rdc_data.dev_type) {
@@ -441,12 +557,15 @@ dasd_eckd_generate_uid(struct dasd_devic
sizeof(uid->serial) - 1);
EBCASC(uid->serial, sizeof(uid->serial) - 1);
uid->ssid = confdata->neq.subsystemID;
- if (confdata->ned2.sneq.flags == 0x40) {
- uid->alias = 1;
- uid->unit_addr = confdata->ned2.sneq.base_unit_addr;
- } else
- uid->unit_addr = confdata->ned1.unit_addr;
-
+ uid->real_unit_addr = confdata->ned1.unit_addr;
+ if (confdata->ned2.sneq.flags == 0x40 &&
+ confdata->ned2.sneq.format == 0x0001) {
+ uid->type = confdata->ned2.sneq.sua_flags;
+ if (uid->type == UA_BASE_PAV_ALIAS)
+ uid->base_unit_addr = confdata->ned2.sneq.base_unit_addr;
+ } else {
+ uid->type = UA_BASE_DEVICE;
+ }
return 0;
}
@@ -470,7 +589,9 @@ static struct dasd_ccw_req *dasd_eckd_bu
ccw->cda = (__u32)(addr_t)rcd_buffer;
ccw->count = ciw->count;
- cqr->device = device;
+ cqr->startdev = device;
+ cqr->memdev = device;
+ cqr->block = NULL;
cqr->expires = 10*HZ;
cqr->lpm = lpm;
clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
@@ -511,7 +632,7 @@ static int dasd_eckd_read_conf_lpm(struc
/*
* on success we update the user input parms
*/
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
if (ret)
goto out_error;
@@ -557,19 +678,19 @@ dasd_eckd_read_conf(struct dasd_device *
"data retrieved");
continue; /* no error */
}
- if (conf_len != sizeof (struct dasd_eckd_confdata)) {
+ if (conf_len != sizeof(struct dasd_eckd_confdata)) {
MESSAGE(KERN_WARNING,
"sizes of configuration data mismatch"
"%d (read) vs %ld (expected)",
conf_len,
- sizeof (struct dasd_eckd_confdata));
+ sizeof(struct dasd_eckd_confdata));
kfree(conf_data);
continue; /* no error */
}
/* save first valid configuration data */
if (!conf_data_saved){
memcpy(&private->conf_data, conf_data,
- sizeof (struct dasd_eckd_confdata));
+ sizeof(struct dasd_eckd_confdata));
conf_data_saved++;
}
switch (((char *)conf_data)[242] & 0x07){
@@ -586,39 +707,106 @@ dasd_eckd_read_conf(struct dasd_device *
return 0;
}
+static int
+dasd_eckd_read_features(struct dasd_device *device)
+{
+ struct dasd_psf_prssd_data *prssdp;
+ struct dasd_rssd_features *features;
+ struct dasd_ccw_req *cqr;
+ struct ccw1 *ccw;
+ int rc;
+ struct dasd_eckd_private *private;
+
+ private = (struct dasd_eckd_private *) device->private;
+ cqr = dasd_smalloc_request(dasd_eckd_discipline.name,
+ 1 /* PSF */ + 1 /* RSSD */ ,
+ (sizeof(struct dasd_psf_prssd_data) +
+ sizeof(struct dasd_rssd_features)),
+ device);
+ if (IS_ERR(cqr)) {
+ DEV_MESSAGE(KERN_WARNING, device, "%s",
+ "Could not allocate initialization request");
+ return PTR_ERR(cqr);
+ }
+ cqr->startdev = device;
+ cqr->memdev = device;
+ cqr->block = NULL;
+ clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
+ cqr->retries = 5;
+ cqr->expires = 10 * HZ;
+
+ /* Prepare for Read Subsystem Data */
+ prssdp = (struct dasd_psf_prssd_data *) cqr->data;
+ memset(prssdp, 0, sizeof(struct dasd_psf_prssd_data));
+ prssdp->order = PSF_ORDER_PRSSD;
+ prssdp->suborder = 0x41; /* Read Feature Codes */
+ /* all other bytes of prssdp must be zero */
+
+ ccw = cqr->cpaddr;
+ ccw->cmd_code = DASD_ECKD_CCW_PSF;
+ ccw->count = sizeof(struct dasd_psf_prssd_data);
+ ccw->flags |= CCW_FLAG_CC;
+ ccw->cda = (__u32)(addr_t) prssdp;
+
+ /* Read Subsystem Data - feature codes */
+ features = (struct dasd_rssd_features *) (prssdp + 1);
+ memset(features, 0, sizeof(struct dasd_rssd_features));
+
+ ccw++;
+ ccw->cmd_code = DASD_ECKD_CCW_RSSD;
+ ccw->count = sizeof(struct dasd_rssd_features);
+ ccw->cda = (__u32)(addr_t) features;
+
+ cqr->buildclk = get_clock();
+ cqr->status = DASD_CQR_FILLED;
+ rc = dasd_sleep_on(cqr);
+ if (rc == 0) {
+ /* Prepare for Read Subsystem Data */
+ prssdp = (struct dasd_psf_prssd_data *) cqr->data;
+ features = (struct dasd_rssd_features *) (prssdp + 1);
+ memcpy(&private->features, features,
+ sizeof(struct dasd_rssd_features));
+ }
+ dasd_sfree_request(cqr, cqr->memdev);
+ return rc;
+}
+
+
/*
* Build CP for Perform Subsystem Function - SSC.
*/
-static struct dasd_ccw_req *
-dasd_eckd_build_psf_ssc(struct dasd_device *device)
+static struct dasd_ccw_req *dasd_eckd_build_psf_ssc(struct dasd_device *device)
{
- struct dasd_ccw_req *cqr;
- struct dasd_psf_ssc_data *psf_ssc_data;
- struct ccw1 *ccw;
+ struct dasd_ccw_req *cqr;
+ struct dasd_psf_ssc_data *psf_ssc_data;
+ struct ccw1 *ccw;
- cqr = dasd_smalloc_request("ECKD", 1 /* PSF */ ,
+ cqr = dasd_smalloc_request("ECKD", 1 /* PSF */ ,
sizeof(struct dasd_psf_ssc_data),
device);
- if (IS_ERR(cqr)) {
- DEV_MESSAGE(KERN_WARNING, device, "%s",
+ if (IS_ERR(cqr)) {
+ DEV_MESSAGE(KERN_WARNING, device, "%s",
"Could not allocate PSF-SSC request");
- return cqr;
- }
- psf_ssc_data = (struct dasd_psf_ssc_data *)cqr->data;
- psf_ssc_data->order = PSF_ORDER_SSC;
- psf_ssc_data->suborder = 0x08;
-
- ccw = cqr->cpaddr;
- ccw->cmd_code = DASD_ECKD_CCW_PSF;
- ccw->cda = (__u32)(addr_t)psf_ssc_data;
- ccw->count = 66;
-
- cqr->device = device;
- cqr->expires = 10*HZ;
- cqr->buildclk = get_clock();
- cqr->status = DASD_CQR_FILLED;
- return cqr;
+ return cqr;
+ }
+ psf_ssc_data = (struct dasd_psf_ssc_data *)cqr->data;
+ psf_ssc_data->order = PSF_ORDER_SSC;
+ psf_ssc_data->suborder = 0x88;
+ psf_ssc_data->reserved[0] = 0x88;
+
+ ccw = cqr->cpaddr;
+ ccw->cmd_code = DASD_ECKD_CCW_PSF;
+ ccw->cda = (__u32)(addr_t)psf_ssc_data;
+ ccw->count = 66;
+
+ cqr->startdev = device;
+ cqr->memdev = device;
+ cqr->block = NULL;
+ cqr->expires = 10*HZ;
+ cqr->buildclk = get_clock();
+ cqr->status = DASD_CQR_FILLED;
+ return cqr;
}
/*
@@ -629,28 +817,29 @@ dasd_eckd_build_psf_ssc(struct dasd_devi
static int
dasd_eckd_psf_ssc(struct dasd_device *device)
{
- struct dasd_ccw_req *cqr;
- int rc;
+ struct dasd_ccw_req *cqr;
+ int rc;
- cqr = dasd_eckd_build_psf_ssc(device);
- if (IS_ERR(cqr))
- return PTR_ERR(cqr);
-
- rc = dasd_sleep_on(cqr);
- if (!rc)
- /* trigger CIO to reprobe devices */
- css_schedule_reprobe();
- dasd_sfree_request(cqr, cqr->device);
- return rc;
+ cqr = dasd_eckd_build_psf_ssc(device);
+ if (IS_ERR(cqr))
+ return PTR_ERR(cqr);
+
+ rc = dasd_sleep_on(cqr);
+ if (!rc)
+ /* trigger CIO to reprobe devices */
+ css_schedule_reprobe();
+ dasd_sfree_request(cqr, cqr->memdev);
+ return rc;
}
/*
* Valide storage server of current device.
*/
static int
-dasd_eckd_validate_server(struct dasd_device *device, struct dasd_uid *uid)
+dasd_eckd_validate_server(struct dasd_device *device)
{
int rc;
+ struct dasd_eckd_private *private;
/* Currently PAV is the only reason to 'validate' server on LPAR */
if (dasd_nopav || MACHINE_IS_VM)
@@ -659,9 +848,11 @@ dasd_eckd_validate_server(struct dasd_de
rc = dasd_eckd_psf_ssc(device);
/* may be requested feature is not available on server,
* therefore just report error and go ahead */
+ private = (struct dasd_eckd_private *) device->private;
DEV_MESSAGE(KERN_INFO, device,
"PSF-SSC on storage subsystem %s.%s.%04x returned rc=%d",
- uid->vendor, uid->serial, uid->ssid, rc);
+ private->uid.vendor, private->uid.serial,
+ private->uid.ssid, rc);
/* RE-Read Configuration Data */
return dasd_eckd_read_conf(device);
}
@@ -674,9 +865,9 @@ static int
dasd_eckd_check_characteristics(struct dasd_device *device)
{
struct dasd_eckd_private *private;
- struct dasd_uid uid;
+ struct dasd_block *block;
void *rdc_data;
- int rc;
+ int is_known, rc;
private = (struct dasd_eckd_private *) device->private;
if (private == NULL) {
@@ -699,27 +890,52 @@ dasd_eckd_check_characteristics(struct d
/* Read Configuration Data */
rc = dasd_eckd_read_conf(device);
if (rc)
- return rc;
+ goto out_err1;
/* Generate device unique id and register in devmap */
- rc = dasd_eckd_generate_uid(device, &uid);
+ rc = dasd_eckd_generate_uid(device, &private->uid);
if (rc)
- return rc;
- rc = dasd_set_uid(device->cdev, &uid);
- if (rc == 1) /* new server found */
- rc = dasd_eckd_validate_server(device, &uid);
+ goto out_err1;
+ dasd_set_uid(device->cdev, &private->uid);
+ /* register lcu with alias handling, enable PAV if this is a new lcu */
+ is_known = dasd_alias_make_device_known_to_lcu(device);
+ if (is_known < 0) {
+ rc = is_known;
+ goto out_err1;
+ }
+ if (!is_known) {
+ /* new lcu found */
+ rc = dasd_eckd_validate_server(device); /* will switch pav on */
+ if (rc)
+ goto out_err2;
+ }
+ if (private->uid.type == UA_BASE_DEVICE) {
+ block = dasd_alloc_block();
+ if (IS_ERR(block)) {
+ DEV_MESSAGE(KERN_WARNING, device, "%s",
+ "could not allocate dasd block structure");
+ rc = PTR_ERR(block);
+ goto out_err2;
+ }
+ device->block = block;
+ block->base = device;
+ }
+
+ /* Read Feature Codes */
+ rc = dasd_eckd_read_features(device);
if (rc)
- return rc;
+ goto out_err3;
/* Read Device Characteristics */
rdc_data = (void *) &(private->rdc_data);
memset(rdc_data, 0, sizeof(rdc_data));
rc = dasd_generic_read_dev_chars(device, "ECKD", &rdc_data, 64);
- if (rc)
+ if (rc) {
DEV_MESSAGE(KERN_WARNING, device,
"Read device characteristics returned "
"rc=%d", rc);
-
+ goto out_err3;
+ }
DEV_MESSAGE(KERN_INFO, device,
"%04X/%02X(CU:%04X/%02X) Cyl:%d Head:%d Sec:%d",
private->rdc_data.dev_type,
@@ -729,9 +945,24 @@ dasd_eckd_check_characteristics(struct d
private->rdc_data.no_cyl,
private->rdc_data.trk_per_cyl,
private->rdc_data.sec_per_trk);
+ return 0;
+
+out_err3:
+ dasd_free_block(device->block);
+ device->block = NULL;
+out_err2:
+ dasd_alias_disconnect_device_from_lcu(device);
+out_err1:
+ kfree(device->private);
+ device->private = NULL;
return rc;
}
+static void dasd_eckd_uncheck_device(struct dasd_device *device)
+{
+ dasd_alias_disconnect_device_from_lcu(device);
+}
+
static struct dasd_ccw_req *
dasd_eckd_analysis_ccw(struct dasd_device *device)
{
@@ -755,7 +986,7 @@ dasd_eckd_analysis_ccw(struct dasd_devic
/* Define extent for the first 3 tracks. */
define_extent(ccw++, cqr->data, 0, 2,
DASD_ECKD_CCW_READ_COUNT, device);
- LO_data = cqr->data + sizeof (struct DE_eckd_data);
+ LO_data = cqr->data + sizeof(struct DE_eckd_data);
/* Locate record for the first 4 records on track 0. */
ccw[-1].flags |= CCW_FLAG_CC;
locate_record(ccw++, LO_data++, 0, 0, 4,
@@ -783,7 +1014,9 @@ dasd_eckd_analysis_ccw(struct dasd_devic
ccw->count = 8;
ccw->cda = (__u32)(addr_t) count_data;
- cqr->device = device;
+ cqr->block = NULL;
+ cqr->startdev = device;
+ cqr->memdev = device;
cqr->retries = 0;
cqr->buildclk = get_clock();
cqr->status = DASD_CQR_FILLED;
@@ -803,7 +1036,7 @@ dasd_eckd_analysis_callback(struct dasd_
struct dasd_eckd_private *private;
struct dasd_device *device;
- device = init_cqr->device;
+ device = init_cqr->startdev;
private = (struct dasd_eckd_private *) device->private;
private->init_cqr_status = init_cqr->status;
dasd_sfree_request(init_cqr, device);
@@ -811,13 +1044,13 @@ dasd_eckd_analysis_callback(struct dasd_
}
static int
-dasd_eckd_start_analysis(struct dasd_device *device)
+dasd_eckd_start_analysis(struct dasd_block *block)
{
struct dasd_eckd_private *private;
struct dasd_ccw_req *init_cqr;
- private = (struct dasd_eckd_private *) device->private;
- init_cqr = dasd_eckd_analysis_ccw(device);
+ private = (struct dasd_eckd_private *) block->base->private;
+ init_cqr = dasd_eckd_analysis_ccw(block->base);
if (IS_ERR(init_cqr))
return PTR_ERR(init_cqr);
init_cqr->callback = dasd_eckd_analysis_callback;
@@ -828,13 +1061,15 @@ dasd_eckd_start_analysis(struct dasd_dev
}
static int
-dasd_eckd_end_analysis(struct dasd_device *device)
+dasd_eckd_end_analysis(struct dasd_block *block)
{
+ struct dasd_device *device;
struct dasd_eckd_private *private;
struct eckd_count *count_area;
unsigned int sb, blk_per_trk;
int status, i;
+ device = block->base;
private = (struct dasd_eckd_private *) device->private;
status = private->init_cqr_status;
private->init_cqr_status = -1;
@@ -846,7 +1081,7 @@ dasd_eckd_end_analysis(struct dasd_devic
private->uses_cdl = 1;
/* Calculate number of blocks/records per track. */
- blk_per_trk = recs_per_track(&private->rdc_data, 0, device->bp_block);
+ blk_per_trk = recs_per_track(&private->rdc_data, 0, block->bp_block);
/* Check Track 0 for Compatible Disk Layout */
count_area = NULL;
for (i = 0; i < 3; i++) {
@@ -876,29 +1111,29 @@ dasd_eckd_end_analysis(struct dasd_devic
if (count_area != NULL && count_area->kl == 0) {
/* we found notthing violating our disk layout */
if (dasd_check_blocksize(count_area->dl) == 0)
- device->bp_block = count_area->dl;
+ block->bp_block = count_area->dl;
}
- if (device->bp_block == 0) {
+ if (block->bp_block == 0) {
DEV_MESSAGE(KERN_WARNING, device, "%s",
"Volume has incompatible disk layout");
return -EMEDIUMTYPE;
}
- device->s2b_shift = 0; /* bits to shift 512 to get a block */
- for (sb = 512; sb < device->bp_block; sb = sb << 1)
- device->s2b_shift++;
+ block->s2b_shift = 0; /* bits to shift 512 to get a block */
+ for (sb = 512; sb < block->bp_block; sb = sb << 1)
+ block->s2b_shift++;
- blk_per_trk = recs_per_track(&private->rdc_data, 0, device->bp_block);
- device->blocks = (private->rdc_data.no_cyl *
+ blk_per_trk = recs_per_track(&private->rdc_data, 0, block->bp_block);
+ block->blocks = (private->rdc_data.no_cyl *
private->rdc_data.trk_per_cyl *
blk_per_trk);
DEV_MESSAGE(KERN_INFO, device,
"(%dkB blks): %dkB at %dkB/trk %s",
- (device->bp_block >> 10),
+ (block->bp_block >> 10),
((private->rdc_data.no_cyl *
private->rdc_data.trk_per_cyl *
- blk_per_trk * (device->bp_block >> 9)) >> 1),
- ((blk_per_trk * device->bp_block) >> 10),
+ blk_per_trk * (block->bp_block >> 9)) >> 1),
+ ((blk_per_trk * block->bp_block) >> 10),
private->uses_cdl ?
"compatible disk layout" : "linux disk layout");
@@ -906,26 +1141,36 @@ dasd_eckd_end_analysis(struct dasd_devic
}
static int
-dasd_eckd_do_analysis(struct dasd_device *device)
+dasd_eckd_do_analysis(struct dasd_block *block)
{
struct dasd_eckd_private *private;
- private = (struct dasd_eckd_private *) device->private;
+ private = (struct dasd_eckd_private *) block->base->private;
if (private->init_cqr_status < 0)
- return dasd_eckd_start_analysis(device);
+ return dasd_eckd_start_analysis(block);
else
- return dasd_eckd_end_analysis(device);
+ return dasd_eckd_end_analysis(block);
}
+static int dasd_eckd_ready_to_online(struct dasd_device *device)
+{
+ return dasd_alias_add_device(device);
+};
+
+static int dasd_eckd_online_to_ready(struct dasd_device *device)
+{
+ return dasd_alias_remove_device(device);
+};
+
static int
-dasd_eckd_fill_geometry(struct dasd_device *device, struct hd_geometry *geo)
+dasd_eckd_fill_geometry(struct dasd_block *block, struct hd_geometry *geo)
{
struct dasd_eckd_private *private;
- private = (struct dasd_eckd_private *) device->private;
- if (dasd_check_blocksize(device->bp_block) == 0) {
+ private = (struct dasd_eckd_private *) block->base->private;
+ if (dasd_check_blocksize(block->bp_block) == 0) {
geo->sectors = recs_per_track(&private->rdc_data,
- 0, device->bp_block);
+ 0, block->bp_block);
}
geo->cylinders = private->rdc_data.no_cyl;
geo->heads = private->rdc_data.trk_per_cyl;
@@ -1037,7 +1282,7 @@ dasd_eckd_format_device(struct dasd_devi
locate_record(ccw++, (struct LO_eckd_data *) data,
fdata->start_unit, 0, rpt + 1,
DASD_ECKD_CCW_WRITE_RECORD_ZERO, device,
- device->bp_block);
+ device->block->bp_block);
data += sizeof(struct LO_eckd_data);
break;
case 0x04: /* Invalidate track. */
@@ -1110,43 +1355,28 @@ dasd_eckd_format_device(struct dasd_devi
ccw++;
}
}
- fcp->device = device;
- fcp->retries = 2; /* set retry counter to enable ERP */
+ fcp->startdev = device;
+ fcp->memdev = device;
+ clear_bit(DASD_CQR_FLAGS_USE_ERP, &fcp->flags);
+ fcp->retries = 5; /* set retry counter to enable default ERP */
fcp->buildclk = get_clock();
fcp->status = DASD_CQR_FILLED;
return fcp;
}
-static dasd_era_t
-dasd_eckd_examine_error(struct dasd_ccw_req * cqr, struct irb * irb)
+static void dasd_eckd_handle_terminated_request(struct dasd_ccw_req *cqr)
{
- struct dasd_device *device = (struct dasd_device *) cqr->device;
- struct ccw_device *cdev = device->cdev;
-
- if (irb->scsw.cstat == 0x00 &&
- irb->scsw.dstat == (DEV_STAT_CHN_END | DEV_STAT_DEV_END))
- return dasd_era_none;
-
- switch (cdev->id.cu_type) {
- case 0x3990:
- case 0x2105:
- case 0x2107:
- case 0x1750:
- return dasd_3990_erp_examine(cqr, irb);
- case 0x9343:
- return dasd_9343_erp_examine(cqr, irb);
- case 0x3880:
- default:
- DEV_MESSAGE(KERN_WARNING, device, "%s",
- "default (unknown CU type) - RECOVERABLE return");
- return dasd_era_recover;
+ cqr->status = DASD_CQR_FILLED;
+ if (cqr->block && (cqr->startdev != cqr->block->base)) {
+ dasd_eckd_reset_ccw_to_base_io(cqr);
+ cqr->startdev = cqr->block->base;
}
-}
+};
static dasd_erp_fn_t
dasd_eckd_erp_action(struct dasd_ccw_req * cqr)
{
- struct dasd_device *device = (struct dasd_device *) cqr->device;
+ struct dasd_device *device = (struct dasd_device *) cqr->startdev;
struct ccw_device *cdev = device->cdev;
switch (cdev->id.cu_type) {
@@ -1168,8 +1398,35 @@ dasd_eckd_erp_postaction(struct dasd_ccw
return dasd_default_erp_postaction;
}
-static struct dasd_ccw_req *
-dasd_eckd_build_cp(struct dasd_device * device, struct request *req)
+
+static void dasd_eckd_handle_unsolicited_interrupt(struct dasd_device *device,
+ 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.dstat & mask) == mask) {
+ dasd_generic_handle_state_change(device);
+ return;
+ }
+
+ /* summary unit check */
+ if ((irb->scsw.dstat & DEV_STAT_UNIT_CHECK) && irb->ecw[7] == 0x0D) {
+ dasd_alias_handle_summary_unit_check(device, irb);
+ return;
+ }
+
+ /* just report other unsolicited interrupts */
+ DEV_MESSAGE(KERN_DEBUG, device, "%s",
+ "unsolicited interrupt received");
+ device->discipline->dump_sense(device, NULL, irb);
+ return;
+};
+
+static struct dasd_ccw_req *dasd_eckd_build_cp(struct dasd_device *startdev,
+ struct dasd_block *block,
+ struct request *req)
{
struct dasd_eckd_private *private;
unsigned long *idaws;
@@ -1185,8 +1442,11 @@ dasd_eckd_build_cp(struct dasd_device *
sector_t first_trk, last_trk;
unsigned int first_offs, last_offs;
unsigned char cmd, rcmd;
+ int use_prefix;
+ struct dasd_device *basedev;
- private = (struct dasd_eckd_private *) device->private;
+ basedev = block->base;
+ private = (struct dasd_eckd_private *) basedev->private;
if (rq_data_dir(req) == READ)
cmd = DASD_ECKD_CCW_READ_MT;
else if (rq_data_dir(req) == WRITE)
@@ -1194,13 +1454,13 @@ dasd_eckd_build_cp(struct dasd_device *
else
return ERR_PTR(-EINVAL);
/* Calculate number of blocks/records per track. */
- blksize = device->bp_block;
+ blksize = block->bp_block;
blk_per_trk = recs_per_track(&private->rdc_data, 0, blksize);
/* Calculate record id of first and last block. */
- first_rec = first_trk = req->sector >> device->s2b_shift;
+ first_rec = first_trk = req->sector >> block->s2b_shift;
first_offs = sector_div(first_trk, blk_per_trk);
last_rec = last_trk =
- (req->sector + req->nr_sectors - 1) >> device->s2b_shift;
+ (req->sector + req->nr_sectors - 1) >> block->s2b_shift;
last_offs = sector_div(last_trk, blk_per_trk);
/* Check struct bio and count the number of blocks for the request. */
count = 0;
@@ -1209,20 +1469,33 @@ dasd_eckd_build_cp(struct dasd_device *
if (bv->bv_len & (blksize - 1))
/* Eckd can only do full blocks. */
return ERR_PTR(-EINVAL);
- count += bv->bv_len >> (device->s2b_shift + 9);
+ count += bv->bv_len >> (block->s2b_shift + 9);
#if defined(CONFIG_64BIT)
if (idal_is_needed (page_address(bv->bv_page), bv->bv_len))
- cidaw += bv->bv_len >> (device->s2b_shift + 9);
+ cidaw += bv->bv_len >> (block->s2b_shift + 9);
#endif
}
/* Paranoia. */
if (count != last_rec - first_rec + 1)
return ERR_PTR(-EINVAL);
- /* 1x define extent + 1x locate record + number of blocks */
- cplength = 2 + count;
- /* 1x define extent + 1x locate record + cidaws*sizeof(long) */
- datasize = sizeof(struct DE_eckd_data) + sizeof(struct LO_eckd_data) +
- cidaw * sizeof(unsigned long);
+
+ /* use the prefix command if available */
+ use_prefix = private->features.feature[8] & 0x01;
+ if (use_prefix) {
+ /* 1x prefix + number of blocks */
+ cplength = 2 + count;
+ /* 1x prefix + cidaws*sizeof(long) */
+ datasize = sizeof(struct PFX_eckd_data) +
+ sizeof(struct LO_eckd_data) +
+ cidaw * sizeof(unsigned long);
+ } else {
+ /* 1x define extent + 1x locate record + number of blocks */
+ cplength = 2 + count;
+ /* 1x define extent + 1x locate record + cidaws*sizeof(long) */
+ datasize = sizeof(struct DE_eckd_data) +
+ sizeof(struct LO_eckd_data) +
+ cidaw * sizeof(unsigned long);
+ }
/* Find out the number of additional locate record ccws for cdl. */
if (private->uses_cdl && first_rec < 2*blk_per_trk) {
if (last_rec >= 2*blk_per_trk)
@@ -1232,26 +1505,42 @@ dasd_eckd_build_cp(struct dasd_device *
}
/* Allocate the ccw request. */
cqr = dasd_smalloc_request(dasd_eckd_discipline.name,
- cplength, datasize, device);
+ cplength, datasize, startdev);
if (IS_ERR(cqr))
return cqr;
ccw = cqr->cpaddr;
- /* First ccw is define extent. */
- if (define_extent(ccw++, cqr->data, first_trk,
- last_trk, cmd, device) == -EAGAIN) {
- /* Clock not in sync and XRC is enabled. Try again later. */
- dasd_sfree_request(cqr, device);
- return ERR_PTR(-EAGAIN);
+ /* First ccw is define extent or prefix. */
+ if (use_prefix) {
+ if (prefix(ccw++, cqr->data, first_trk,
+ last_trk, cmd, basedev, startdev) == -EAGAIN) {
+ /* Clock not in sync and XRC is enabled.
+ * Try again later.
+ */
+ dasd_sfree_request(cqr, startdev);
+ return ERR_PTR(-EAGAIN);
+ }
+ idaws = (unsigned long *) (cqr->data +
+ sizeof(struct PFX_eckd_data));
+ } else {
+ if (define_extent(ccw++, cqr->data, first_trk,
+ last_trk, cmd, startdev) == -EAGAIN) {
+ /* Clock not in sync and XRC is enabled.
+ * Try again later.
+ */
+ dasd_sfree_request(cqr, startdev);
+ return ERR_PTR(-EAGAIN);
+ }
+ idaws = (unsigned long *) (cqr->data +
+ sizeof(struct DE_eckd_data));
}
/* Build locate_record+read/write/ccws. */
- idaws = (unsigned long *) (cqr->data + sizeof(struct DE_eckd_data));
LO_data = (struct LO_eckd_data *) (idaws + cidaw);
recid = first_rec;
if (private->uses_cdl == 0 || recid > 2*blk_per_trk) {
/* Only standard blocks so there is just one locate record. */
ccw[-1].flags |= CCW_FLAG_CC;
locate_record(ccw++, LO_data++, first_trk, first_offs + 1,
- last_rec - recid + 1, cmd, device, blksize);
+ last_rec - recid + 1, cmd, basedev, blksize);
}
rq_for_each_segment(bv, req, iter) {
dst = page_address(bv->bv_page) + bv->bv_offset;
@@ -1281,7 +1570,7 @@ dasd_eckd_build_cp(struct dasd_device *
ccw[-1].flags |= CCW_FLAG_CC;
locate_record(ccw++, LO_data++,
trkid, recoffs + 1,
- 1, rcmd, device, count);
+ 1, rcmd, basedev, count);
}
/* Locate record for standard blocks ? */
if (private->uses_cdl && recid == 2*blk_per_trk) {
@@ -1289,7 +1578,7 @@ dasd_eckd_build_cp(struct dasd_device *
locate_record(ccw++, LO_data++,
trkid, recoffs + 1,
last_rec - recid + 1,
- cmd, device, count);
+ cmd, basedev, count);
}
/* Read/write ccw. */
ccw[-1].flags |= CCW_FLAG_CC;
@@ -1310,7 +1599,9 @@ dasd_eckd_build_cp(struct dasd_device *
}
if (req->cmd_flags & REQ_FAILFAST)
set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
- cqr->device = device;
+ cqr->startdev = startdev;
+ cqr->memdev = startdev;
+ cqr->block = block;
cqr->expires = 5 * 60 * HZ; /* 5 minutes */
cqr->lpm = private->path_data.ppm;
cqr->retries = 256;
@@ -1333,10 +1624,10 @@ dasd_eckd_free_cp(struct dasd_ccw_req *c
if (!dasd_page_cache)
goto out;
- private = (struct dasd_eckd_private *) cqr->device->private;
- blksize = cqr->device->bp_block;
+ private = (struct dasd_eckd_private *) cqr->block->base->private;
+ blksize = cqr->block->bp_block;
blk_per_trk = recs_per_track(&private->rdc_data, 0, blksize);
- recid = req->sector >> cqr->device->s2b_shift;
+ recid = req->sector >> cqr->block->s2b_shift;
ccw = cqr->cpaddr;
/* Skip over define extent & locate record. */
ccw++;
@@ -1367,10 +1658,71 @@ dasd_eckd_free_cp(struct dasd_ccw_req *c
}
out:
status = cqr->status == DASD_CQR_DONE;
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
return status;
}
+/*
+ * Modify ccw chain in cqr so it can be started on a base device.
+ *
+ * Note that this is not enough to restart the cqr!
+ * Either reset cqr->startdev as well (summary unit check handling)
+ * or restart via separate cqr (as in ERP handling).
+ */
+void dasd_eckd_reset_ccw_to_base_io(struct dasd_ccw_req *cqr)
+{
+ struct ccw1 *ccw;
+ struct PFX_eckd_data *pfxdata;
+
+ ccw = cqr->cpaddr;
+ pfxdata = cqr->data;
+
+ if (ccw->cmd_code == DASD_ECKD_CCW_PFX) {
+ pfxdata->validity.verify_base = 0;
+ pfxdata->validity.hyper_pav = 0;
+ }
+}
+
+#define DASD_ECKD_CHANQ_MAX_SIZE 4
+
+static struct dasd_ccw_req *dasd_eckd_build_alias_cp(struct dasd_device *base,
+ struct dasd_block *block,
+ struct request *req)
+{
+ struct dasd_eckd_private *private;
+ struct dasd_device *startdev;
+ unsigned long flags;
+ struct dasd_ccw_req *cqr;
+
+ startdev = dasd_alias_get_start_dev(base);
+ if (!startdev)
+ startdev = base;
+ private = (struct dasd_eckd_private *) startdev->private;
+ if (private->count >= DASD_ECKD_CHANQ_MAX_SIZE)
+ return ERR_PTR(-EBUSY);
+
+ spin_lock_irqsave(get_ccwdev_lock(startdev->cdev), flags);
+ private->count++;
+ cqr = dasd_eckd_build_cp(startdev, block, req);
+ if (IS_ERR(cqr))
+ private->count--;
+ spin_unlock_irqrestore(get_ccwdev_lock(startdev->cdev), flags);
+ return cqr;
+}
+
+static int dasd_eckd_free_alias_cp(struct dasd_ccw_req *cqr,
+ struct request *req)
+{
+ struct dasd_eckd_private *private;
+ unsigned long flags;
+
+ spin_lock_irqsave(get_ccwdev_lock(cqr->memdev->cdev), flags);
+ private = (struct dasd_eckd_private *) cqr->memdev->private;
+ private->count--;
+ spin_unlock_irqrestore(get_ccwdev_lock(cqr->memdev->cdev), flags);
+ return dasd_eckd_free_cp(cqr, req);
+}
+
static int
dasd_eckd_fill_info(struct dasd_device * device,
struct dasd_information2_t * info)
@@ -1384,9 +1736,9 @@ dasd_eckd_fill_info(struct dasd_device *
info->characteristics_size = sizeof(struct dasd_eckd_characteristics);
memcpy(info->characteristics, &private->rdc_data,
sizeof(struct dasd_eckd_characteristics));
- info->confdata_size = sizeof (struct dasd_eckd_confdata);
+ info->confdata_size = sizeof(struct dasd_eckd_confdata);
memcpy(info->configuration_data, &private->conf_data,
- sizeof (struct dasd_eckd_confdata));
+ sizeof(struct dasd_eckd_confdata));
return 0;
}
@@ -1419,7 +1771,8 @@ dasd_eckd_release(struct dasd_device *de
cqr->cpaddr->flags |= CCW_FLAG_SLI;
cqr->cpaddr->count = 32;
cqr->cpaddr->cda = (__u32)(addr_t) cqr->data;
- cqr->device = device;
+ cqr->startdev = device;
+ cqr->memdev = device;
clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
cqr->retries = 2; /* set retry counter to enable basic ERP */
@@ -1429,7 +1782,7 @@ dasd_eckd_release(struct dasd_device *de
rc = dasd_sleep_on_immediatly(cqr);
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
return rc;
}
@@ -1459,7 +1812,8 @@ dasd_eckd_reserve(struct dasd_device *de
cqr->cpaddr->flags |= CCW_FLAG_SLI;
cqr->cpaddr->count = 32;
cqr->cpaddr->cda = (__u32)(addr_t) cqr->data;
- cqr->device = device;
+ cqr->startdev = device;
+ cqr->memdev = device;
clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
cqr->retries = 2; /* set retry counter to enable basic ERP */
@@ -1469,7 +1823,7 @@ dasd_eckd_reserve(struct dasd_device *de
rc = dasd_sleep_on_immediatly(cqr);
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
return rc;
}
@@ -1498,7 +1852,8 @@ dasd_eckd_steal_lock(struct dasd_device
cqr->cpaddr->flags |= CCW_FLAG_SLI;
cqr->cpaddr->count = 32;
cqr->cpaddr->cda = (__u32)(addr_t) cqr->data;
- cqr->device = device;
+ cqr->startdev = device;
+ cqr->memdev = device;
clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
cqr->retries = 2; /* set retry counter to enable basic ERP */
@@ -1508,7 +1863,7 @@ dasd_eckd_steal_lock(struct dasd_device
rc = dasd_sleep_on_immediatly(cqr);
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
return rc;
}
@@ -1526,38 +1881,39 @@ dasd_eckd_performance(struct dasd_device
cqr = dasd_smalloc_request(dasd_eckd_discipline.name,
1 /* PSF */ + 1 /* RSSD */ ,
- (sizeof (struct dasd_psf_prssd_data) +
- sizeof (struct dasd_rssd_perf_stats_t)),
+ (sizeof(struct dasd_psf_prssd_data) +
+ sizeof(struct dasd_rssd_perf_stats_t)),
device);
if (IS_ERR(cqr)) {
DEV_MESSAGE(KERN_WARNING, device, "%s",
"Could not allocate initialization request");
return PTR_ERR(cqr);
}
- cqr->device = device;
+ cqr->startdev = device;
+ cqr->memdev = device;
cqr->retries = 0;
cqr->expires = 10 * HZ;
/* Prepare for Read Subsystem Data */
prssdp = (struct dasd_psf_prssd_data *) cqr->data;
- memset(prssdp, 0, sizeof (struct dasd_psf_prssd_data));
+ memset(prssdp, 0, sizeof(struct dasd_psf_prssd_data));
prssdp->order = PSF_ORDER_PRSSD;
prssdp->suborder = 0x01; /* Performance Statistics */
prssdp->varies[1] = 0x01; /* Perf Statistics for the Subsystem */
ccw = cqr->cpaddr;
ccw->cmd_code = DASD_ECKD_CCW_PSF;
- ccw->count = sizeof (struct dasd_psf_prssd_data);
+ ccw->count = sizeof(struct dasd_psf_prssd_data);
ccw->flags |= CCW_FLAG_CC;
ccw->cda = (__u32)(addr_t) prssdp;
/* Read Subsystem Data - Performance Statistics */
stats = (struct dasd_rssd_perf_stats_t *) (prssdp + 1);
- memset(stats, 0, sizeof (struct dasd_rssd_perf_stats_t));
+ memset(stats, 0, sizeof(struct dasd_rssd_perf_stats_t));
ccw++;
ccw->cmd_code = DASD_ECKD_CCW_RSSD;
- ccw->count = sizeof (struct dasd_rssd_perf_stats_t);
+ ccw->count = sizeof(struct dasd_rssd_perf_stats_t);
ccw->cda = (__u32)(addr_t) stats;
cqr->buildclk = get_clock();
@@ -1571,7 +1927,7 @@ dasd_eckd_performance(struct dasd_device
sizeof(struct dasd_rssd_perf_stats_t)))
rc = -EFAULT;
}
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
return rc;
}
@@ -1594,7 +1950,7 @@ dasd_eckd_get_attrib(struct dasd_device
rc = 0;
if (copy_to_user(argp, (long *) &attrib,
- sizeof (struct attrib_data_t)))
+ sizeof(struct attrib_data_t)))
rc = -EFAULT;
return rc;
@@ -1627,8 +1983,10 @@ dasd_eckd_set_attrib(struct dasd_device
}
static int
-dasd_eckd_ioctl(struct dasd_device *device, unsigned int cmd, void __user *argp)
+dasd_eckd_ioctl(struct dasd_block *block, unsigned int cmd, void __user *argp)
{
+ struct dasd_device *device = block->base;
+
switch (cmd) {
case BIODASDGATTR:
return dasd_eckd_get_attrib(device, argp);
@@ -1685,9 +2043,8 @@ dasd_eckd_dump_ccw_range(struct ccw1 *fr
* Print sense data and related channel program.
* Parts are printed because printk buffer is only 1024 bytes.
*/
-static void
-dasd_eckd_dump_sense(struct dasd_device *device, struct dasd_ccw_req * req,
- struct irb *irb)
+static void dasd_eckd_dump_sense(struct dasd_device *device,
+ struct dasd_ccw_req *req, struct irb *irb)
{
char *page;
struct ccw1 *first, *last, *fail, *from, *to;
@@ -1743,37 +2100,40 @@ dasd_eckd_dump_sense(struct dasd_device
}
printk("%s", page);
- /* dump the Channel Program (max 140 Bytes per line) */
- /* Count CCW and print first CCWs (maximum 1024 % 140 = 7) */
- first = req->cpaddr;
- for (last = first; last->flags & (CCW_FLAG_CC | CCW_FLAG_DC); last++);
- to = min(first + 6, last);
- len = sprintf(page, KERN_ERR PRINTK_HEADER
- " Related CP in req: %p\n", req);
- dasd_eckd_dump_ccw_range(first, to, page + len);
- printk("%s", page);
+ if (req) {
+ /* req == NULL for unsolicited interrupts */
+ /* dump the Channel Program (max 140 Bytes per line) */
+ /* Count CCW and print first CCWs (maximum 1024 % 140 = 7) */
+ first = req->cpaddr;
+ for (last = first; last->flags & (CCW_FLAG_CC | CCW_FLAG_DC); last++);
+ to = min(first + 6, last);
+ len = sprintf(page, KERN_ERR PRINTK_HEADER
+ " Related CP in req: %p\n", req);
+ dasd_eckd_dump_ccw_range(first, to, page + len);
+ printk("%s", page);
- /* print failing CCW area (maximum 4) */
- /* scsw->cda is either valid or zero */
- len = 0;
- from = ++to;
- fail = (struct ccw1 *)(addr_t) irb->scsw.cpa; /* failing CCW */
- if (from < fail - 2) {
- from = fail - 2; /* there is a gap - print header */
- len += sprintf(page, KERN_ERR PRINTK_HEADER "......\n");
- }
- to = min(fail + 1, last);
- len += dasd_eckd_dump_ccw_range(from, to, page + len);
-
- /* print last CCWs (maximum 2) */
- from = max(from, ++to);
- if (from < last - 1) {
- from = last - 1; /* there is a gap - print header */
- len += sprintf(page + len, KERN_ERR PRINTK_HEADER "......\n");
+ /* print failing CCW area (maximum 4) */
+ /* scsw->cda is either valid or zero */
+ len = 0;
+ from = ++to;
+ fail = (struct ccw1 *)(addr_t) irb->scsw.cpa; /* failing CCW */
+ if (from < fail - 2) {
+ from = fail - 2; /* there is a gap - print header */
+ len += sprintf(page, KERN_ERR PRINTK_HEADER "......\n");
+ }
+ to = min(fail + 1, last);
+ len += dasd_eckd_dump_ccw_range(from, to, page + len);
+
+ /* print last CCWs (maximum 2) */
+ from = max(from, ++to);
+ if (from < last - 1) {
+ from = last - 1; /* there is a gap - print header */
+ len += sprintf(page + len, KERN_ERR PRINTK_HEADER "......\n");
+ }
+ len += dasd_eckd_dump_ccw_range(from, last, page + len);
+ if (len > 0)
+ printk("%s", page);
}
- len += dasd_eckd_dump_ccw_range(from, last, page + len);
- if (len > 0)
- printk("%s", page);
free_page((unsigned long) page);
}
@@ -1796,16 +2156,20 @@ static struct dasd_discipline dasd_eckd_
.ebcname = "ECKD",
.max_blocks = 240,
.check_device = dasd_eckd_check_characteristics,
+ .uncheck_device = dasd_eckd_uncheck_device,
.do_analysis = dasd_eckd_do_analysis,
+ .ready_to_online = dasd_eckd_ready_to_online,
+ .online_to_ready = dasd_eckd_online_to_ready,
.fill_geometry = dasd_eckd_fill_geometry,
.start_IO = dasd_start_IO,
.term_IO = dasd_term_IO,
+ .handle_terminated_request = dasd_eckd_handle_terminated_request,
.format_device = dasd_eckd_format_device,
- .examine_error = dasd_eckd_examine_error,
.erp_action = dasd_eckd_erp_action,
.erp_postaction = dasd_eckd_erp_postaction,
- .build_cp = dasd_eckd_build_cp,
- .free_cp = dasd_eckd_free_cp,
+ .handle_unsolicited_interrupt = dasd_eckd_handle_unsolicited_interrupt,
+ .build_cp = dasd_eckd_build_alias_cp,
+ .free_cp = dasd_eckd_free_alias_cp,
.dump_sense = dasd_eckd_dump_sense,
.fill_info = dasd_eckd_fill_info,
.ioctl = dasd_eckd_ioctl,
Index: quilt-2.6/drivers/s390/block/dasd_eckd.h
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_eckd.h
+++ quilt-2.6/drivers/s390/block/dasd_eckd.h
@@ -39,6 +39,8 @@
#define DASD_ECKD_CCW_READ_CKD_MT 0x9e
#define DASD_ECKD_CCW_WRITE_CKD_MT 0x9d
#define DASD_ECKD_CCW_RESERVE 0xB4
+#define DASD_ECKD_CCW_PFX 0xE7
+#define DASD_ECKD_CCW_RSCK 0xF9
/*
* Perform Subsystem Function / Sub-Orders
@@ -137,6 +139,25 @@ struct LO_eckd_data {
__u16 length;
} __attribute__ ((packed));
+/* Prefix data for format 0x00 and 0x01 */
+struct PFX_eckd_data {
+ unsigned char format;
+ struct {
+ unsigned char define_extend:1;
+ unsigned char time_stamp:1;
+ unsigned char verify_base:1;
+ unsigned char hyper_pav:1;
+ unsigned char reserved:4;
+ } __attribute__ ((packed)) validity;
+ __u8 base_address;
+ __u8 aux;
+ __u8 base_lss;
+ __u8 reserved[7];
+ struct DE_eckd_data define_extend;
+ struct LO_eckd_data locate_record;
+ __u8 LO_extended_data[4];
+} __attribute__ ((packed));
+
struct dasd_eckd_characteristics {
__u16 cu_type;
struct {
@@ -254,7 +275,9 @@ struct dasd_eckd_confdata {
} __attribute__ ((packed)) ned;
struct {
unsigned char flags; /* byte 0 */
- unsigned char res2[7]; /* byte 1- 7 */
+ unsigned char res1; /* byte 1 */
+ __u16 format; /* byte 2-3 */
+ unsigned char res2[4]; /* byte 4-7 */
unsigned char sua_flags; /* byte 8 */
__u8 base_unit_addr; /* byte 9 */
unsigned char res3[22]; /* byte 10-31 */
@@ -343,6 +366,11 @@ struct dasd_eckd_path {
__u8 npm;
};
+struct dasd_rssd_features {
+ char feature[256];
+} __attribute__((packed));
+
+
/*
* Perform Subsystem Function - Prepare for Read Subsystem Data
*/
@@ -365,4 +393,99 @@ struct dasd_psf_ssc_data {
unsigned char reserved[59];
} __attribute__((packed));
+
+/*
+ * some structures and definitions for alias handling
+ */
+struct dasd_unit_address_configuration {
+ struct {
+ char ua_type;
+ char base_ua;
+ } unit[256];
+} __attribute__((packed));
+
+
+#define MAX_DEVICES_PER_LCU 256
+
+/* flags on the LCU */
+#define NEED_UAC_UPDATE 0x01
+#define UPDATE_PENDING 0x02
+
+enum pavtype {NO_PAV, BASE_PAV, HYPER_PAV};
+
+
+struct alias_root {
+ struct list_head serverlist;
+ spinlock_t lock;
+};
+
+struct alias_server {
+ struct list_head server;
+ struct dasd_uid uid;
+ struct list_head lculist;
+};
+
+struct summary_unit_check_work_data {
+ char reason;
+ struct dasd_device *device;
+ struct work_struct worker;
+};
+
+struct read_uac_work_data {
+ struct dasd_device *device;
+ struct delayed_work dwork;
+};
+
+struct alias_lcu {
+ struct list_head lcu;
+ struct dasd_uid uid;
+ enum pavtype pav;
+ char flags;
+ spinlock_t lock;
+ struct list_head grouplist;
+ struct list_head active_devices;
+ struct list_head inactive_devices;
+ struct dasd_unit_address_configuration *uac;
+ struct summary_unit_check_work_data suc_data;
+ struct read_uac_work_data ruac_data;
+ struct dasd_ccw_req *rsu_cqr;
+};
+
+struct alias_pav_group {
+ struct list_head group;
+ struct dasd_uid uid;
+ struct alias_lcu *lcu;
+ struct list_head baselist;
+ struct list_head aliaslist;
+ struct dasd_device *next;
+};
+
+
+struct dasd_eckd_private {
+ struct dasd_eckd_characteristics rdc_data;
+ struct dasd_eckd_confdata conf_data;
+ struct dasd_eckd_path path_data;
+ struct eckd_count count_area[5];
+ int init_cqr_status;
+ int uses_cdl;
+ struct attrib_data_t attrib; /* e.g. cache operations */
+ struct dasd_rssd_features features;
+
+ /* alias managemnet */
+ struct dasd_uid uid;
+ struct alias_pav_group *pavgroup;
+ struct alias_lcu *lcu;
+ int count;
+};
+
+
+
+int dasd_alias_make_device_known_to_lcu(struct dasd_device *);
+void dasd_alias_disconnect_device_from_lcu(struct dasd_device *);
+int dasd_alias_add_device(struct dasd_device *);
+int dasd_alias_remove_device(struct dasd_device *);
+struct dasd_device *dasd_alias_get_start_dev(struct dasd_device *);
+void dasd_alias_handle_summary_unit_check(struct dasd_device *, struct irb *);
+void dasd_eckd_reset_ccw_to_base_io(struct dasd_ccw_req *);
+
#endif /* DASD_ECKD_H */
Index: quilt-2.6/drivers/s390/block/dasd_eer.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_eer.c
+++ quilt-2.6/drivers/s390/block/dasd_eer.c
@@ -336,7 +336,7 @@ static void dasd_eer_write_snss_trigger(
unsigned long flags;
struct eerbuffer *eerb;
- snss_rc = (cqr->status == DASD_CQR_FAILED) ? -EIO : 0;
+ snss_rc = (cqr->status == DASD_CQR_DONE) ? 0 : -EIO;
if (snss_rc)
data_size = 0;
else
@@ -404,10 +404,11 @@ void dasd_eer_snss(struct dasd_device *d
set_bit(DASD_FLAG_EER_SNSS, &device->flags);
return;
}
+ /* cdev is already locked, can't use dasd_add_request_head */
clear_bit(DASD_FLAG_EER_SNSS, &device->flags);
cqr->status = DASD_CQR_QUEUED;
- list_add(&cqr->list, &device->ccw_queue);
- dasd_schedule_bh(device);
+ list_add(&cqr->devlist, &device->ccw_queue);
+ dasd_schedule_device_bh(device);
}
/*
@@ -415,7 +416,7 @@ void dasd_eer_snss(struct dasd_device *d
*/
static void dasd_eer_snss_cb(struct dasd_ccw_req *cqr, void *data)
{
- struct dasd_device *device = cqr->device;
+ struct dasd_device *device = cqr->startdev;
unsigned long flags;
dasd_eer_write(device, cqr, DASD_EER_STATECHANGE);
@@ -458,7 +459,7 @@ int dasd_eer_enable(struct dasd_device *
if (!cqr)
return -ENOMEM;
- cqr->device = device;
+ cqr->startdev = device;
cqr->retries = 255;
cqr->expires = 10 * HZ;
clear_bit(DASD_CQR_FLAGS_USE_ERP, &cqr->flags);
Index: quilt-2.6/drivers/s390/block/dasd_erp.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_erp.c
+++ quilt-2.6/drivers/s390/block/dasd_erp.c
@@ -46,6 +46,8 @@ dasd_alloc_erp_request(char *magic, int
if (cqr == NULL)
return ERR_PTR(-ENOMEM);
memset(cqr, 0, sizeof(struct dasd_ccw_req));
+ INIT_LIST_HEAD(&cqr->devlist);
+ INIT_LIST_HEAD(&cqr->blocklist);
data = (char *) cqr + ((sizeof(struct dasd_ccw_req) + 7L) & -8L);
cqr->cpaddr = NULL;
if (cplength > 0) {
@@ -66,7 +68,7 @@ dasd_alloc_erp_request(char *magic, int
}
void
-dasd_free_erp_request(struct dasd_ccw_req * cqr, struct dasd_device * device)
+dasd_free_erp_request(struct dasd_ccw_req *cqr, struct dasd_device * device)
{
unsigned long flags;
@@ -81,11 +83,11 @@ dasd_free_erp_request(struct dasd_ccw_re
* dasd_default_erp_action just retries the current cqr
*/
struct dasd_ccw_req *
-dasd_default_erp_action(struct dasd_ccw_req * cqr)
+dasd_default_erp_action(struct dasd_ccw_req *cqr)
{
struct dasd_device *device;
- device = cqr->device;
+ device = cqr->startdev;
/* just retry - there is nothing to save ... I got no sense data.... */
if (cqr->retries > 0) {
@@ -93,12 +95,12 @@ dasd_default_erp_action(struct dasd_ccw_
"default ERP called (%i retries left)",
cqr->retries);
cqr->lpm = LPM_ANYPATH;
- cqr->status = DASD_CQR_QUEUED;
+ cqr->status = DASD_CQR_FILLED;
} else {
DEV_MESSAGE (KERN_WARNING, device, "%s",
"default ERP called (NO retry left)");
cqr->status = DASD_CQR_FAILED;
- cqr->stopclk = get_clock ();
+ cqr->stopclk = get_clock();
}
return cqr;
} /* end dasd_default_erp_action */
@@ -117,15 +119,12 @@ dasd_default_erp_action(struct dasd_ccw_
* RETURN VALUES
* cqr pointer to the original CQR
*/
-struct dasd_ccw_req *
-dasd_default_erp_postaction(struct dasd_ccw_req * cqr)
+struct dasd_ccw_req *dasd_default_erp_postaction(struct dasd_ccw_req *cqr)
{
- struct dasd_device *device;
int success;
BUG_ON(cqr->refers == NULL || cqr->function == NULL);
- device = cqr->device;
success = cqr->status == DASD_CQR_DONE;
/* free all ERPs - but NOT the original cqr */
@@ -133,10 +132,10 @@ dasd_default_erp_postaction(struct dasd_
struct dasd_ccw_req *refers;
refers = cqr->refers;
- /* remove the request from the device queue */
- list_del(&cqr->list);
+ /* remove the request from the block queue */
+ list_del(&cqr->blocklist);
/* free the finished erp request */
- dasd_free_erp_request(cqr, device);
+ dasd_free_erp_request(cqr, cqr->memdev);
cqr = refers;
}
@@ -157,7 +156,7 @@ dasd_log_sense(struct dasd_ccw_req *cqr,
{
struct dasd_device *device;
- device = cqr->device;
+ device = cqr->startdev;
/* dump sense data */
if (device->discipline && device->discipline->dump_sense)
device->discipline->dump_sense(device, cqr, irb);
Index: quilt-2.6/drivers/s390/block/dasd_fba.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_fba.c
+++ quilt-2.6/drivers/s390/block/dasd_fba.c
@@ -117,6 +117,7 @@ locate_record(struct ccw1 * ccw, struct
static int
dasd_fba_check_characteristics(struct dasd_device *device)
{
+ struct dasd_block *block;
struct dasd_fba_private *private;
struct ccw_device *cdev = device->cdev;
void *rdc_data;
@@ -133,6 +134,16 @@ dasd_fba_check_characteristics(struct da
}
device->private = (void *) private;
}
+ block = dasd_alloc_block();
+ if (IS_ERR(block)) {
+ DEV_MESSAGE(KERN_WARNING, device, "%s",
+ "could not allocate dasd block structure");
+ kfree(device->private);
+ return PTR_ERR(block);
+ }
+ device->block = block;
+ block->base = device;
+
/* Read Device Characteristics */
rdc_data = (void *) &(private->rdc_data);
rc = dasd_generic_read_dev_chars(device, "FBA ", &rdc_data, 32);
@@ -156,59 +167,37 @@ dasd_fba_check_characteristics(struct da
}
static int
-dasd_fba_do_analysis(struct dasd_device *device)
+dasd_fba_do_analysis(struct dasd_block *block)
{
struct dasd_fba_private *private;
int sb, rc;
- private = (struct dasd_fba_private *) device->private;
+ private = (struct dasd_fba_private *) block->base->private;
rc = dasd_check_blocksize(private->rdc_data.blk_size);
if (rc) {
- DEV_MESSAGE(KERN_INFO, device, "unknown blocksize %d",
+ DEV_MESSAGE(KERN_INFO, block->base, "unknown blocksize %d",
private->rdc_data.blk_size);
return rc;
}
- device->blocks = private->rdc_data.blk_bdsa;
- device->bp_block = private->rdc_data.blk_size;
- device->s2b_shift = 0; /* bits to shift 512 to get a block */
+ block->blocks = private->rdc_data.blk_bdsa;
+ block->bp_block = private->rdc_data.blk_size;
+ block->s2b_shift = 0; /* bits to shift 512 to get a block */
for (sb = 512; sb < private->rdc_data.blk_size; sb = sb << 1)
- device->s2b_shift++;
+ block->s2b_shift++;
return 0;
}
static int
-dasd_fba_fill_geometry(struct dasd_device *device, struct hd_geometry *geo)
+dasd_fba_fill_geometry(struct dasd_block *block, struct hd_geometry *geo)
{
- if (dasd_check_blocksize(device->bp_block) != 0)
+ if (dasd_check_blocksize(block->bp_block) != 0)
return -EINVAL;
- geo->cylinders = (device->blocks << device->s2b_shift) >> 10;
+ geo->cylinders = (block->blocks << block->s2b_shift) >> 10;
geo->heads = 16;
- geo->sectors = 128 >> device->s2b_shift;
+ geo->sectors = 128 >> block->s2b_shift;
return 0;
}
-static dasd_era_t
-dasd_fba_examine_error(struct dasd_ccw_req * cqr, struct irb * irb)
-{
- struct dasd_device *device;
- struct ccw_device *cdev;
-
- device = (struct dasd_device *) cqr->device;
- if (irb->scsw.cstat == 0x00 &&
- irb->scsw.dstat == (DEV_STAT_CHN_END | DEV_STAT_DEV_END))
- return dasd_era_none;
-
- cdev = device->cdev;
- switch (cdev->id.dev_type) {
- case 0x3370:
- return dasd_3370_erp_examine(cqr, irb);
- case 0x9336:
- return dasd_9336_erp_examine(cqr, irb);
- default:
- return dasd_era_recover;
- }
-}
-
static dasd_erp_fn_t
dasd_fba_erp_action(struct dasd_ccw_req * cqr)
{
@@ -221,13 +210,32 @@ dasd_fba_erp_postaction(struct dasd_ccw_
if (cqr->function == dasd_default_erp_action)
return dasd_default_erp_postaction;
- DEV_MESSAGE(KERN_WARNING, cqr->device, "unknown ERP action %p",
+ DEV_MESSAGE(KERN_WARNING, cqr->startdev, "unknown ERP action %p",
cqr->function);
return NULL;
}
-static struct dasd_ccw_req *
-dasd_fba_build_cp(struct dasd_device * device, struct request *req)
+static void dasd_fba_handle_unsolicited_interrupt(struct dasd_device *device,
+ 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.dstat & mask) == mask) {
+ dasd_generic_handle_state_change(device);
+ return;
+ }
+
+ /* check for unsolicited interrupts */
+ DEV_MESSAGE(KERN_DEBUG, device, "%s",
+ "unsolicited interrupt received");
+ device->discipline->dump_sense(device, NULL, irb);
+ return;
+};
+
+static struct dasd_ccw_req *dasd_fba_build_cp(struct dasd_device * memdev,
+ struct dasd_block *block,
+ struct request *req)
{
struct dasd_fba_private *private;
unsigned long *idaws;
@@ -242,17 +250,17 @@ dasd_fba_build_cp(struct dasd_device * d
unsigned int blksize, off;
unsigned char cmd;
- private = (struct dasd_fba_private *) device->private;
+ private = (struct dasd_fba_private *) block->base->private;
if (rq_data_dir(req) == READ) {
cmd = DASD_FBA_CCW_READ;
} else if (rq_data_dir(req) == WRITE) {
cmd = DASD_FBA_CCW_WRITE;
} else
return ERR_PTR(-EINVAL);
- blksize = device->bp_block;
+ blksize = block->bp_block;
/* Calculate record id of first and last block. */
- first_rec = req->sector >> device->s2b_shift;
- last_rec = (req->sector + req->nr_sectors - 1) >> device->s2b_shift;
+ first_rec = req->sector >> block->s2b_shift;
+ last_rec = (req->sector + req->nr_sectors - 1) >> block->s2b_shift;
/* Check struct bio and count the number of blocks for the request. */
count = 0;
cidaw = 0;
@@ -260,7 +268,7 @@ dasd_fba_build_cp(struct dasd_device * d
if (bv->bv_len & (blksize - 1))
/* Fba can only do full blocks. */
return ERR_PTR(-EINVAL);
- count += bv->bv_len >> (device->s2b_shift + 9);
+ count += bv->bv_len >> (block->s2b_shift + 9);
#if defined(CONFIG_64BIT)
if (idal_is_needed (page_address(bv->bv_page), bv->bv_len))
cidaw += bv->bv_len / blksize;
@@ -284,13 +292,13 @@ dasd_fba_build_cp(struct dasd_device * d
}
/* Allocate the ccw request. */
cqr = dasd_smalloc_request(dasd_fba_discipline.name,
- cplength, datasize, device);
+ cplength, datasize, memdev);
if (IS_ERR(cqr))
return cqr;
ccw = cqr->cpaddr;
/* First ccw is define extent. */
define_extent(ccw++, cqr->data, rq_data_dir(req),
- device->bp_block, req->sector, req->nr_sectors);
+ block->bp_block, req->sector, req->nr_sectors);
/* Build locate_record + read/write ccws. */
idaws = (unsigned long *) (cqr->data + sizeof(struct DE_fba_data));
LO_data = (struct LO_fba_data *) (idaws + cidaw);
@@ -326,7 +334,7 @@ dasd_fba_build_cp(struct dasd_device * d
ccw[-1].flags |= CCW_FLAG_CC;
}
ccw->cmd_code = cmd;
- ccw->count = device->bp_block;
+ ccw->count = block->bp_block;
if (idal_is_needed(dst, blksize)) {
ccw->cda = (__u32)(addr_t) idaws;
ccw->flags = CCW_FLAG_IDA;
@@ -342,7 +350,9 @@ dasd_fba_build_cp(struct dasd_device * d
}
if (req->cmd_flags & REQ_FAILFAST)
set_bit(DASD_CQR_FLAGS_FAILFAST, &cqr->flags);
- cqr->device = device;
+ cqr->startdev = memdev;
+ cqr->memdev = memdev;
+ cqr->block = block;
cqr->expires = 5 * 60 * HZ; /* 5 minutes */
cqr->retries = 32;
cqr->buildclk = get_clock();
@@ -363,8 +373,8 @@ dasd_fba_free_cp(struct dasd_ccw_req *cq
if (!dasd_page_cache)
goto out;
- private = (struct dasd_fba_private *) cqr->device->private;
- blksize = cqr->device->bp_block;
+ private = (struct dasd_fba_private *) cqr->block->base->private;
+ blksize = cqr->block->bp_block;
ccw = cqr->cpaddr;
/* Skip over define extent & locate record. */
ccw++;
@@ -394,10 +404,15 @@ dasd_fba_free_cp(struct dasd_ccw_req *cq
}
out:
status = cqr->status == DASD_CQR_DONE;
- dasd_sfree_request(cqr, cqr->device);
+ dasd_sfree_request(cqr, cqr->memdev);
return status;
}
+static void dasd_fba_handle_terminated_request(struct dasd_ccw_req *cqr)
+{
+ cqr->status = DASD_CQR_FILLED;
+};
+
static int
dasd_fba_fill_info(struct dasd_device * device,
struct dasd_information2_t * info)
@@ -546,9 +561,10 @@ static struct dasd_discipline dasd_fba_d
.fill_geometry = dasd_fba_fill_geometry,
.start_IO = dasd_start_IO,
.term_IO = dasd_term_IO,
- .examine_error = dasd_fba_examine_error,
+ .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,
.build_cp = dasd_fba_build_cp,
.free_cp = dasd_fba_free_cp,
.dump_sense = dasd_fba_dump_sense,
Index: quilt-2.6/drivers/s390/block/dasd_genhd.c
===================================================================
--- quilt-2.6.orig/drivers/s390/block/dasd_genhd.c
+++ quilt-2.6/drivers/s390/block/dasd_genhd.c
@@ -25,14 +25,15 @@
/*
* Allocate and register gendisk structure for device.
*/
-int
-dasd_gendisk_alloc(struct dasd_device *device)
+int dasd_gendisk_alloc(struct dasd_block *block)
{
struct gendisk *gdp;
+ struct dasd_device *base;
int len;
/* Make sure the minor for this device exists. */
- if (device->devindex >= DASD_PER_MAJOR)
+ base = block->base;
+ if (base->devindex >= DASD_PER_MAJOR)
return -EBUSY;
gdp = alloc_disk(1 << DASD_PARTN_BITS);
@@ -41,9 +42,9 @@ dasd_gendisk_alloc(struct dasd_device *d
/* Initialize gendisk structure. */
gdp->major = DASD_MAJOR;
- gdp->first_minor = device->devindex << DASD_PARTN_BITS;
+ gdp->first_minor = base->devindex << DASD_PARTN_BITS;
gdp->fops = &dasd_device_operations;
- gdp->driverfs_dev = &device->cdev->dev;
+ gdp->driverfs_dev = &base->cdev->dev;
/*
* Set device name.
@@ -53,53 +54,51 @@ dasd_gendisk_alloc(struct dasd_device *d
* dasdaaaa - dasdzzzz : 456976 devices, added up = 475252
*/
len = sprintf(gdp->disk_name, "dasd");
- if (device->devindex > 25) {
- if (device->devindex > 701) {
- if (device->devindex > 18277)
+ if (base->devindex > 25) {
+ if (base->devindex > 701) {
+ if (base->devindex > 18277)
len += sprintf(gdp->disk_name + len, "%c",
- 'a'+(((device->devindex-18278)
+ 'a'+(((base->devindex-18278)
/17576)%26));
len += sprintf(gdp->disk_name + len, "%c",
- 'a'+(((device->devindex-702)/676)%26));
+ 'a'+(((base->devindex-702)/676)%26));
}
len += sprintf(gdp->disk_name + len, "%c",
- 'a'+(((device->devindex-26)/26)%26));
+ 'a'+(((base->devindex-26)/26)%26));
}
- len += sprintf(gdp->disk_name + len, "%c", 'a'+(device->devindex%26));
+ len += sprintf(gdp->disk_name + len, "%c", 'a'+(base->devindex%26));
- if (device->features & DASD_FEATURE_READONLY)
+ if (block->base->features & DASD_FEATURE_READONLY)
set_disk_ro(gdp, 1);
- gdp->private_data = device;
- gdp->queue = device->request_queue;
- device->gdp = gdp;
- set_capacity(device->gdp, 0);
- add_disk(device->gdp);
+ gdp->private_data = block;
+ gdp->queue = block->request_queue;
+ block->gdp = gdp;
+ set_capacity(block->gdp, 0);
+ add_disk(block->gdp);
return 0;
}
/*
* Unregister and free gendisk structure for device.
*/
-void
-dasd_gendisk_free(struct dasd_device *device)
+void dasd_gendisk_free(struct dasd_block *block)
{
- if (device->gdp) {
- del_gendisk(device->gdp);
- device->gdp->queue = NULL;
- put_disk(device->gdp);
- device->gdp = NULL;
+ if (block->gdp) {
+ del_gendisk(block->gdp);
+ block->gdp->queue = NULL;
+ put_disk(block->gdp);
+ block->gdp = NULL;
}
}
/*
* Trigger a partition detection.
*/
-int
-dasd_scan_partitions(struct dasd_device * device)
+int dasd_scan_partitions(struct dasd_block *block)
{
struct block_device *bdev;
- bdev = bdget_disk(device->gdp, 0);
+ bdev = bdget_disk(block->gdp, 0);
if (!bdev || blkdev_get(bdev, FMODE_READ, 1) < 0)
return -ENODEV;
/*
@@ -117,7 +116,7 @@ dasd_scan_partitions(struct dasd_device
* is why the assignment to device->bdev is done AFTER
* the BLKRRPART ioctl.
*/
- device->bdev = bdev;
+ block->bdev = bdev;
return 0;
}
@@ -125,8 +124,7 @@ dasd_scan_partitions(struct dasd_device
* Remove all inodes in the system for a device, delete the
* partitions and make device unusable by setting its size to zero.
*/
-void
-dasd_destroy_partitions(struct dasd_device * device)
+void dasd_destroy_partitions(struct dasd_block *block)
{
/* The two structs have 168/176 byte on 31/64 bit. */
struct blkpg_partition bpart;
@@ -137,8 +135,8 @@ dasd_destroy_partitions(struct dasd_devi
* Get the bdev pointer from the device structure and clear
* device->bdev to lower the offline open_count limit again.
*/
- bdev = device->bdev;
- device->bdev = NULL;
+ bdev = block->bdev;
+ block->bdev = NULL;
/*
* See fs/partition/check.c:delete_partition
@@ -149,17 +147,16 @@ dasd_destroy_partitions(struct dasd_devi
memset(&barg, 0, sizeof(struct blkpg_ioctl_arg));
barg.data = (void __force __user *) &bpart;
barg.op = BLKPG_DEL_PARTITION;
- for (bpart.pno = device->gdp->minors - 1; bpart.pno > 0; bpart.pno--)
+ for (bpart.pno = block->gdp->minors - 1; bpart.pno > 0; bpart.pno--)
ioctl_by_bdev(bdev, BLKPG, (unsigned long) &barg);
- invalidate_partition(device->gdp, 0);
+ invalidate_partition(block->gdp, 0);
/* Matching blkdev_put to the blkdev_get in dasd_scan_partitions. */
blkdev_put(bdev);
- set_capacity(device->gdp, 0);
+ set_capacity(block->gdp, 0);
}
-int
-dasd_gendisk_init(void)
+int dasd_gendisk_init(void)
{
int rc;
@@ -174,8 +171,7 @@ dasd_gendisk_init(void)
return 0;
}
-void
-dasd_gendisk_exit(void)
+void dasd_gendisk_exit(void)
{
unregister_blkdev(DASD_MAJOR, "dasd");
}
--
blue skies,
Martin.
"Reality continues to ruin my life." - Calvin.
^ permalink raw reply [flat|nested] 48+ messages in thread
end of thread, other threads:[~2007-12-20 15:20 UTC | newest]
Thread overview: 48+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2007-12-20 15:19 [patch 00/47] s390 2.6.25 patch queue Martin Schwidefsky
2007-12-20 15:19 ` [patch 01/47] Cleanup in Documentation/kernel-parameters.txt Martin Schwidefsky
2007-12-20 15:19 ` [patch 02/47] cio: Dump ccw device information in case of timeout Martin Schwidefsky
2007-12-20 15:19 ` [patch 03/47] cio: Use helpers instead of container_of() Martin Schwidefsky
2007-12-20 15:19 ` [patch 04/47] cio: css_driver: Use consistent parameters Martin Schwidefsky
2007-12-20 15:19 ` [patch 05/47] cio: Reset sch->driver Martin Schwidefsky
2007-12-20 15:19 ` [patch 06/47] cio: Add css_driver_{register,unregister} Martin Schwidefsky
2007-12-20 15:19 ` [patch 07/47] cio: Cleanup debug feature usage Martin Schwidefsky
2007-12-20 15:19 ` [patch 08/47] cio: Introduce subchannel->private Martin Schwidefsky
2007-12-20 15:19 ` [patch 09/47] cio: Extend adapter interrupt interface Martin Schwidefsky
2007-12-20 15:19 ` [patch 10/47] cio: I/O subchannel specific fields Martin Schwidefsky
2007-12-20 15:19 ` [patch 11/47] cio: Use dev_{g,s}et_drvdata() Martin Schwidefsky
2007-12-20 15:19 ` [patch 12/47] cio: Set driver->owner on css, ccw and ccwgroup busses Martin Schwidefsky
2007-12-20 15:19 ` [patch 13/47] cio: reduce cpu utilization during device scan Martin Schwidefsky
2007-12-20 15:19 ` [patch 14/47] qdio: Remove double checked value Martin Schwidefsky
2007-12-20 15:19 ` [patch 15/47] qdio: set QDIO_ACTIVATE_TIMEOUT to 5s Martin Schwidefsky
2007-12-20 15:19 ` [patch 16/47] sclp: sysfs interface for SCLP cpi Martin Schwidefsky
2007-12-20 15:19 ` [patch 17/47] Standby cpu activation/deactivation Martin Schwidefsky
2007-12-20 15:19 ` [patch 18/47] sclp: convert channel path configure code to use sync interface Martin Schwidefsky
2007-12-20 15:19 ` [patch 19/47] Optimize reference bit handling Martin Schwidefsky
2007-12-20 15:19 ` [patch 20/47] Fix tlb flushing with idte Martin Schwidefsky
2007-12-20 15:19 ` [patch 21/47] Change vmalloc defintions Martin Schwidefsky
2007-12-20 15:19 ` [patch 22/47] Print kernel version in dump_stack() and show_regs() Martin Schwidefsky
2007-12-20 15:19 ` [patch 23/47] Get rid of HOLES_IN_ZONE requirement Martin Schwidefsky
2007-12-20 15:19 ` [patch 24/47] DEBUG_PAGEALLOC support for s390 Martin Schwidefsky
2007-12-20 15:19 ` [patch 25/47] Remove owner_pc member from raw_spinlock_t Martin Schwidefsky
2007-12-20 15:19 ` [patch 26/47] Use new style spinlock initializer in __RWSEM_INITIALIZER Martin Schwidefsky
2007-12-20 15:19 ` [patch 27/47] Get rid of additional_cpus kernel parameter Martin Schwidefsky
2007-12-20 15:19 ` [patch 28/47] Remove appldata include from sysctl_check.c Martin Schwidefsky
2007-12-20 15:19 ` [patch 29/47] crypto: move s390 Kconfig options Martin Schwidefsky
2007-12-20 15:19 ` [patch 30/47] dasd: fix return value of dasd_generic_probe() Martin Schwidefsky
2007-12-20 15:19 ` [patch 31/47] arch/s390: Add missing "space" Martin Schwidefsky
2007-12-20 15:19 ` [patch 32/47] drivers/s390: " Martin Schwidefsky
2007-12-20 15:19 ` [patch 33/47] kernel: Shutdown Actions Interface Martin Schwidefsky
2007-12-20 15:19 ` [patch 34/47] Load disabled wait psw instead of stopping cpu on halt Martin Schwidefsky
2007-12-20 15:20 ` [patch 35/47] use LIST_HEAD instead of LIST_HEAD_INIT Martin Schwidefsky
2007-12-20 15:20 ` [patch 36/47] Allocate and free cpu lowcores and stacks when needed/possible Martin Schwidefsky
2007-12-20 15:20 ` [patch 37/47] Initialize sclp_ipl_info Martin Schwidefsky
2007-12-20 15:20 ` [patch 38/47] vmemmap: allocate struct pages before 1:1 mapping Martin Schwidefsky
2007-12-20 15:20 ` [patch 39/47] Use diag308 subcodes 3 and 6 for reboot and dump when possible Martin Schwidefsky
2007-12-20 15:20 ` [patch 40/47] arch/s390/: Spelling fixes Martin Schwidefsky
2007-12-20 15:20 ` [patch 41/47] include/asm-s390/: " Martin Schwidefsky
2007-12-20 15:20 ` [patch 42/47] drivers/s390/: " Martin Schwidefsky
2007-12-20 15:20 ` [patch 43/47] Move NOTES and BUG_TABLE Martin Schwidefsky
2007-12-20 15:20 ` [patch 44/47] single-step cleanup Martin Schwidefsky
2007-12-20 15:20 ` [patch 45/47] dasd: add hyper PAV support to DASD device driver, part 1 Martin Schwidefsky
2007-12-20 15:20 ` [patch 46/47] dasd: add hyper PAV support to DASD device driver, part 2 Martin Schwidefsky
2007-12-20 15:20 ` [patch 47/47] dasd: add hyper PAV support to DASD device driver, part 3 Martin Schwidefsky
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).