Git commit ea435467 changed the
definition of atomic_t and atomic64_t for s390 by adding the volatile
modifier to the counter field. This has an unfortunate side effect
with newer versions of the gcc. The typeof operator now picks up the
volatile modifier from the expression. This causes the compiler to
think that it has to store the two temporary variable old_val and
new_val in the __CS_LOOP for the different atomic operations to the
stack as the variables are now volatile. Both stores are superfluous.

The hack to replace typeof(ptr->counter) with int in __CS_LOOP and
and long long in __CSG_LOOP avoids the two stores. A better solution
would be to drop the volatile from the counter field of the atomic_t
and atomic64_t definition. But that is a touchy subject ..

Cc: Matthew Wilcox <matthew@wil.cx>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Replace BLK with a per device mutex.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Rename tape_get_device to tape_find_device and tape_get_device_reference
to tape_get_device. The old names didn't make too much sense.

Follow the get_device()/put_device() semantic and convert tape_put_device
to a void function.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Fix a bug introduced with git commit dff59b64:
 -       if (cdev->dev.driver_data != NULL) {
 +       if (!dev_get_drvdata(&cdev->dev)) {
These two are not equivalent.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The change to f_pos in the read / write method has no effect as the
value is controlled by the VFS layer, e.g. for vfs_read:

	loff_t pos = file_pos_read(file);
	ret = vfs_read(file, buf, count, &pos);
	file_pos_write(file, pos);
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Protect the hrtimer ap_poll_timer from being scheduled at the same
time from several processes.
Signed-off-by: NFelix Beck <felix.beck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Cex3 needs a lower speed rating. Otherwise cex2 adapters will be
prefered.
Signed-off-by: NFelix Beck <felix.beck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Cards with lower speed rating are prefered. Thus change adjust the
speed rating of cex2c adapters to be prefered before pcixcc, although
they should not appear in the same machine anyway.
Signed-off-by: NFelix Beck <felix.beck@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

New definitions are added for CEX3 device types. They will be set
in the according probe functions. CEX3 device types will be handled
in the same modules as CEX2 device types. In the first step they are
the same as CEX2 types, but they can be adjusted for further
characteristics of CEX3 easily.
Signed-off-by: NFelix Beck <felix.beck@de.ibm.com>
Signed-off-by: NRalph Wuerthner <ralph.wuerthner@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

This patch renames the CEX2C2 and CEX2A2 types to CEX3 device types.
Signed-off-by: NFelix Beck <felix.beck@de.ibm.com>
Signed-off-by: NRalph Wuerthner <ralph.wuerthner@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Support for special command is implemented in the AP Bus in the NQAP
function __ap_send. This is extended for a further parameter special.
When set, the special bit, in GR0 will be set. Therefor the ap_message
struct is extended for a further bit. Thus calling functions of
__ap_send can use the special parameter in ap_message to give to
__ap_send. Affected is in the first place ap_queue_message, which is
called by the actual card driver. The second part of this support is
that the card driver for the CEX3C needs to set this special bit, when
an according CPRB is sent to the driver.
Signed-off-by: NFelix Beck <felix.beck@de.ibm.com>
Signed-off-by: NRalph Wuerthner <ralph.wuerthner@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

AP messages need to be initialized, before they will be used. Values
will be zeroized. This will be needed later when introducing support
for the special commands.
Signed-off-by: NFelix Beck <felix.beck@de.ibm.com>
Signed-off-by: NRalph Wuerthner <ralph.wuerthner@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The first DASD that is set online for a specific logical control unit
has to do certain setup steps on the storage server to make full use
of it, for example it will enable PAV.
The features and characteristics reported by the storage server will
depend on this setup, so all other devices on the same LCU will need
to wait for the setup to be finished.
Signed-off-by: NStefan Weinhuber <wein@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Remove strings from s390 debugfeature entries that could lead to a
crash when the data is read from dbf because the strings do not exist
any more.
Signed-off-by: NStefan Haberland <stefan.haberland@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Most of the error conditions reported by a FICON storage server
indicate situations which can be recovered. Sometimes the host just
needs to retry an I/O request, but sometimes the recovery
is more complex and requires the device driver to wait, choose
a different path, etc.

The DASD device driver has a fully featured error recovery
for normal block layer I/O, but not for internal I/O request which
are for example used during the device bring up.
This can lead to situations where the IPL of a system fails because
DASD devices are not properly recognized.
This patch will extend the internal I/O handling to use the existing
error recovery procedures.
Signed-off-by: NStefan Weinhuber <wein@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The DASD device driver needs to explicitly enable the prefix command
on the storage server, before it can be used. Originally we enabled
this command along with others only if we wanted to support PAV.
However, today we require this command for other features like
High Performance FICON as well, so we need to always enable prefix.
Signed-off-by: NStefan Weinhuber <wein@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

the todclk.h header file is dead code. Remove it.
Signed-off-by: NChristian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

When a DASD device is used with the DIAG discipline, the DIAG
initialization will indicate success or error with a respective
return code. So far we have interpreted a return code of 4 as error,
but it actually means that the initialization was successful, but
the device is read-only. To allow read-only devices to be used with
DIAG we need to accept a return code of 4 as success.

Re-initialization of the DIAG access is also part of the DIAG error
recovery. If we find that the access mode of a device has been
changed from writable to read-only while the device was in use,
we print an error message.
Signed-off-by: NStefan Weinhuber <wein@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The pagetable walk usercopy functions have used a modified copy of the
do_exception() function for fault handling. This lead to inconsistencies
with recent changes to do_exception(), e.g. performance counters. This
patch changes the pagetable walk usercopy code to call do_exception()
directly, eliminating the redundancy. A new parameter is added to
do_exception() to specify the fault address.
Signed-off-by: NGerald Schaefer <gerald.schaefer@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Simplify the check of the vma->flags in do_exception for the
different fault types.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Slim down the do_exception function to handle only the fast path of a
fault and move the exceptional cases into a new function. That slightly
increases the performance of the fault handling.

Build fix for !CONFIG_COMPAT by
Kamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

notify_page_fault does a preempt_disable/preempt_enable for each
fault generated by a kernel access to user space. If kprobes
is not active that is unnecessary since the interrupts are not
reenabled yet. To play safe repeat the kprobe_running check after
preempt_disable().
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Introduce user_mode to replace the two variables switch_amode and
s390_noexec. There are three valid combinations of the old values:
  1) switch_amode == 0 && s390_noexec == 0
  2) switch_amode == 1 && s390_noexec == 0
  3) switch_amode == 1 && s390_noexec == 1
They get replaced by
  1) user_mode == HOME_SPACE_MODE
  2) user_mode == PRIMARY_SPACE_MODE
  3) user_mode == SECONDARY_SPACE_MODE
The new kernel parameter user_mode=[primary,secondary,home] lets
you choose the address space mode the user space processes should
use. In addition the CONFIG_S390_SWITCH_AMODE config option
is removed.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

A data access in access-register mode always is a user mode access,
the code to inspect the access-registers can be removed. The second
change is to use a different test to check for no-execute fault.
The third change is to pass the translation exception identification
as parameter, in theory the trans_exc_code in the lowcore could have
been overwritten by the time the call to check_space from do_no_context
is done.
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

If we detect a busy subchannel after the driver's set_offline
callback returned in ccw_device_set_offline, the current behavior
is to unregister the device, which may lead to undesired
consequences. Change this to just quiesce the subchannel and go on
with the offline processing.

Note: This is no excuse for not fixing these drivers -
after the set_offline callback they should have no running IO!
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Improve error recovery for internal I/Os by repeating each I/O
256 times per path to cope with long-running non-permanent error
conditions. Also retry each path twice to cope with link flapping,
i.e. single paths becoming unavailable in the order in which they
are tried.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

IO subchannels are always unregistered in process context, so use
spin_lock_irq in the corresponding remove callback.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Ensure that there will be no more interrupts for an
unregistered device by using the same quiesce and disable loop
as in io_subchannel_shutdown.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Try to disable the old subchannel before we ask the driver core
to move the attached device to a new parent. This way we can use
the QUIESCE state during shutdown which prevents a possible use
after free situation in some error cases.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Handle a failing cio_disable_subchannel at the end of our device
recognition as if the recognition itself failed. This way
subsequent registration steps do not need to handle enabled
subchannels.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

DEV_STATE_QUIESCE is used to stop all IO on a busy subchannel.
This patch fixes the following problems related to the QUIESCE
state:

* Fix a potential race condition which could occur when the
resulting state was DEV_STATE_OFFLINE.

* Add missing locking around cio_disable_subchannel,
ccw_device_cancel_halt_clear and the cdev's handler.

* Loop until we know for sure that the subchannel is disabled.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

The function ccw_device_unregister has to ensure to remove
all references obtained by device_add and device_initialize.
Unfortunately it gets called for devices which are
1) uninitialized, 2) initialized but unregistered, and
3) registered devices. To distinguish 1) and 2) this patch
introduces a new flag "initialized", which is 1 as long as we
hold the initial device reference.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

We used to maintain a "registered" flag in our ccw_device_private
structure. This patch removes the "registered" flag and converts
all users of it to device_is_registered which has the exact same
meaning.

Note: The usage the atomic operation test_and_clear_bit is replaced
by the non-atomic if (device_is_registered()) device_del(). This
will not do harm, since we serialize calls to ccw_device_unregister
with a single-threaded workqueue.
Signed-off-by: NSebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

An Unconditional Reserve + Release operation (steal lock) for a
boxed device may fail when encountering special error cases
(e.g. unit checks or path errors). Fix this by using the more
robust ccw_request infrastructure for performing the steal lock
CCW program.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Set-pgid operations fail for some device types under z/VM for which
the hypervisor has already set the pgid. Also reserved devices or
changed pgids are not correctly recognized. Fix these problems by
using a combination of sense-pgid and set-pgid and by also accepting
pre-defined pgid settings.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Split setting (driver wants feature enabled) and status (feature
setup was successful) for PGID related ccw device features so that
setup errors can be detected. Previously, incorrectly handled setup
errors could in rare cases lead to erratic I/O behavior and
permanently unusuable devices.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

After changing all internal I/O functions to use the newly introduced
ccw request infrastructure, retries are handled automatically after a
clear operation. Therefore remove the internal retry flag and
associated code.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Accept a request for setting a not-operational device offline.
This way, users can remove devices from Linux which would otherwise
remain unusable until reboot.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Use the newly introduced ccw request infrastructure to implement
pgid related operations: sense pgid, set pgid and disband pg.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>

Use the newly introduced ccw request infrastructure to implement
the sense id operation.
Signed-off-by: NPeter Oberparleiter <peter.oberparleiter@de.ibm.com>
Signed-off-by: NMartin Schwidefsky <schwidefsky@de.ibm.com>